From e09b41010ba33a20a87472ee821fa407a5b8da36 Mon Sep 17 00:00:00 2001
From: José Pekkarinen <jose.pekkarinen@nokia.com>
Date: Mon, 11 Apr 2016 10:41:07 +0300
Subject: These changes are the raw update to linux-4.4.6-rt14. Kernel sources
 are taken from kernel.org, and rt patch from the rt wiki download page.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

During the rebasing, the following patch collided:

Force tick interrupt and get rid of softirq magic(I70131fb85).

Collisions have been removed because its logic was found on the
source already.

Change-Id: I7f57a4081d9deaa0d9ccfc41a6c8daccdee3b769
Signed-off-by: José Pekkarinen <jose.pekkarinen@nokia.com>
---
 kernel/drivers/gpu/drm/amd/amdgpu/Kconfig          |   17 +
 kernel/drivers/gpu/drm/amd/amdgpu/Makefile         |  100 +
 kernel/drivers/gpu/drm/amd/amdgpu/ObjectID.h       |  736 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu.h         | 2401 +++++++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c    |  767 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h    |  445 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c    |  105 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c  |  269 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h  |   65 +
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c      |  670 ++
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c      |  543 ++
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c   | 1602 +++++
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h   |  206 +
 .../gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c       |  572 ++
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c  |  223 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c    |  362 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c |  271 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c     |  800 +++
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c | 1912 ++++++
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h |   42 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c      | 1007 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c     |  309 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c  | 2049 ++++++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c |  910 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c     |  955 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h     |   85 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c     |  579 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h     |   48 +
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c   |  245 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c      |  421 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c   |  823 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c    |  372 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h     |   72 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c     |  740 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c     |   72 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h     |   30 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c     |  395 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h     |   44 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c      |  348 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c      |  219 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h      |   63 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c   |   47 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c     |  463 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h     |   93 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c     |  752 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c      |  323 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h    |  596 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c  |  667 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h  |  204 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c     |  350 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h     |   38 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c      |  815 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h      |   35 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c   |  128 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c    |  562 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c      |  451 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c   |  108 +
 .../drivers/gpu/drm/amd/amdgpu/amdgpu_semaphore.c  |  102 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c    |  388 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c    |  555 ++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h   |  291 +
 .../gpu/drm/amd/amdgpu/amdgpu_trace_points.c       |    9 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c     | 1254 ++++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c   |  317 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h   |  176 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c     | 1044 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h     |   39 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c     |  880 +++
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h     |   46 +
 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c      | 1356 ++++
 kernel/drivers/gpu/drm/amd/amdgpu/atom.c           | 1457 +++++
 kernel/drivers/gpu/drm/amd/amdgpu/atom.h           |  159 +
 kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c  |  807 +++
 kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.h  |   58 +
 kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.c    |  776 +++
 kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.h    |   42 +
 .../drivers/gpu/drm/amd/amdgpu/atombios_encoders.c | 2065 ++++++
 .../drivers/gpu/drm/amd/amdgpu/atombios_encoders.h |   73 +
 kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c   |  158 +
 kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.h   |   31 +
 kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.c         | 6701 ++++++++++++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.h         |  348 +
 kernel/drivers/gpu/drm/amd/amdgpu/ci_smc.c         |  279 +
 kernel/drivers/gpu/drm/amd/amdgpu/cik.c            | 2521 ++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/cik.h            |   33 +
 kernel/drivers/gpu/drm/amd/amdgpu/cik_dpm.h        |   30 +
 kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.c         |  471 ++
 kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.h         |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.c       | 1412 +++++
 kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/cikd.h           |  568 ++
 kernel/drivers/gpu/drm/amd/amdgpu/clearstate_ci.h  |  944 +++
 .../drivers/gpu/drm/amd/amdgpu/clearstate_defs.h   |   44 +
 kernel/drivers/gpu/drm/amd/amdgpu/clearstate_vi.h  |  944 +++
 kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.c         | 2002 ++++++
 kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.h         |  237 +
 kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.c          |  452 ++
 kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.h          |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h       |  185 +
 kernel/drivers/gpu/drm/amd/amdgpu/cz_smc.c         | 1000 +++
 kernel/drivers/gpu/drm/amd/amdgpu/cz_smumgr.h      |   94 +
 kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c      | 3831 +++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h      |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c      | 3824 +++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h      |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c       | 3761 +++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c       |  181 +
 kernel/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h     |  182 +
 kernel/drivers/gpu/drm/amd/amdgpu/fiji_smc.c       |  857 +++
 kernel/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h    |   42 +
 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c       | 5691 +++++++++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h       |   37 +
 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c       | 5171 +++++++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h       |   33 +
 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c       | 1409 ++++
 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h       |   36 +
 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c       | 1348 ++++
 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h       |   36 +
 kernel/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c    |  195 +
 kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.c     |  450 ++
 kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.h     |   29 +
 .../gpu/drm/amd/amdgpu/iceland_sdma_pkt_open.h     | 2172 +++++++
 kernel/drivers/gpu/drm/amd/amdgpu/iceland_smc.c    |  671 ++
 kernel/drivers/gpu/drm/amd/amdgpu/iceland_smumgr.h |   41 +
 kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.c         | 3349 ++++++++++
 kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.h         |  229 +
 kernel/drivers/gpu/drm/amd/amdgpu/kv_smc.c         |  219 +
 kernel/drivers/gpu/drm/amd/amdgpu/ppsmc.h          |  196 +
 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c      | 1418 +++++
 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.h      |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c      | 1579 +++++
 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.h      |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/smu7.h           |  170 +
 kernel/drivers/gpu/drm/amd/amdgpu/smu7_discrete.h  |  514 ++
 kernel/drivers/gpu/drm/amd/amdgpu/smu7_fusion.h    |  300 +
 kernel/drivers/gpu/drm/amd/amdgpu/smu8.h           |   72 +
 kernel/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h    |  127 +
 .../drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h |  147 +
 .../drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_vi.h |  100 +
 kernel/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c      |  181 +
 kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.c       |  473 ++
 kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h    |  198 +
 .../gpu/drm/amd/amdgpu/tonga_sdma_pkt_open.h       | 2245 +++++++
 kernel/drivers/gpu/drm/amd/amdgpu/tonga_smc.c      |  856 +++
 kernel/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h   |   42 +
 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c       |  905 +++
 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c       |  844 +++
 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c       |  828 +++
 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c       |  663 ++
 kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c       |  679 ++
 kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.h       |   29 +
 kernel/drivers/gpu/drm/amd/amdgpu/vi.c             | 1547 +++++
 kernel/drivers/gpu/drm/amd/amdgpu/vi.h             |   33 +
 kernel/drivers/gpu/drm/amd/amdgpu/vi_dpm.h         |   36 +
 kernel/drivers/gpu/drm/amd/amdgpu/vid.h            |  373 ++
 161 files changed, 106600 insertions(+)
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/Kconfig
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/Makefile
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/ObjectID.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_semaphore.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atom.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atom.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/ci_smc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik_dpm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cikd.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/clearstate_ci.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/clearstate_defs.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/clearstate_vi.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_smc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/cz_smumgr.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/fiji_smc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/iceland_sdma_pkt_open.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/iceland_smc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/iceland_smumgr.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/kv_smc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/ppsmc.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu7.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu7_discrete.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu7_fusion.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu8.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_vi.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_sdma_pkt_open.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_smc.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vi.c
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vi.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vi_dpm.h
 create mode 100644 kernel/drivers/gpu/drm/amd/amdgpu/vid.h

(limited to 'kernel/drivers/gpu/drm/amd/amdgpu')

diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/Kconfig b/kernel/drivers/gpu/drm/amd/amdgpu/Kconfig
new file mode 100644
index 000000000..b30fcfa4b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/Kconfig
@@ -0,0 +1,17 @@
+config DRM_AMDGPU_CIK
+	bool "Enable amdgpu support for CIK parts"
+	depends on DRM_AMDGPU
+	help
+	  Choose this option if you want to enable experimental support
+	  for CIK asics.
+
+	  CIK is already supported in radeon.  CIK support in amdgpu
+	  is for experimentation and testing.
+
+config DRM_AMDGPU_USERPTR
+	bool "Always enable userptr write support"
+	depends on DRM_AMDGPU
+	select MMU_NOTIFIER
+	help
+	  This option selects CONFIG_MMU_NOTIFIER if it isn't already
+	  selected to enabled full userptr support.
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/Makefile b/kernel/drivers/gpu/drm/amd/amdgpu/Makefile
new file mode 100644
index 000000000..ca066018e
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -0,0 +1,100 @@
+#
+# Makefile for the drm device driver.  This driver provides support for the
+# Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher.
+
+ccflags-y := -Iinclude/drm -Idrivers/gpu/drm/amd/include/asic_reg \
+	-Idrivers/gpu/drm/amd/include \
+	-Idrivers/gpu/drm/amd/amdgpu \
+	-Idrivers/gpu/drm/amd/scheduler
+
+amdgpu-y := amdgpu_drv.o
+
+# add KMS driver
+amdgpu-y += amdgpu_device.o amdgpu_kms.o \
+	amdgpu_atombios.o atombios_crtc.o amdgpu_connectors.o \
+	atom.o amdgpu_fence.o amdgpu_ttm.o amdgpu_object.o amdgpu_gart.o \
+	amdgpu_encoders.o amdgpu_display.o amdgpu_i2c.o \
+	amdgpu_fb.o amdgpu_gem.o amdgpu_ring.o \
+	amdgpu_cs.o amdgpu_bios.o amdgpu_benchmark.o amdgpu_test.o \
+	amdgpu_pm.o atombios_dp.o amdgpu_afmt.o amdgpu_trace_points.o \
+	atombios_encoders.o amdgpu_semaphore.o amdgpu_sa.o atombios_i2c.o \
+	amdgpu_prime.o amdgpu_vm.o amdgpu_ib.o amdgpu_pll.o \
+	amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o
+
+# add asic specific block
+amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
+	ci_smc.o ci_dpm.o dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o \
+	amdgpu_amdkfd_gfx_v7.o
+
+amdgpu-y += \
+	vi.o
+
+# add GMC block
+amdgpu-y += \
+	gmc_v7_0.o \
+	gmc_v8_0.o
+
+# add IH block
+amdgpu-y += \
+	amdgpu_irq.o \
+	amdgpu_ih.o \
+	iceland_ih.o \
+	tonga_ih.o \
+	cz_ih.o
+
+# add SMC block
+amdgpu-y += \
+	amdgpu_dpm.o \
+	cz_smc.o cz_dpm.o \
+	tonga_smc.o tonga_dpm.o \
+	fiji_smc.o fiji_dpm.o \
+	iceland_smc.o iceland_dpm.o
+
+# add DCE block
+amdgpu-y += \
+	dce_v10_0.o \
+	dce_v11_0.o
+
+# add GFX block
+amdgpu-y += \
+	amdgpu_gfx.o \
+	gfx_v8_0.o
+
+# add async DMA block
+amdgpu-y += \
+	sdma_v2_4.o \
+	sdma_v3_0.o
+
+# add UVD block
+amdgpu-y += \
+	amdgpu_uvd.o \
+	uvd_v5_0.o \
+	uvd_v6_0.o
+
+# add VCE block
+amdgpu-y += \
+	amdgpu_vce.o \
+	vce_v3_0.o
+
+# add amdkfd interfaces
+amdgpu-y += \
+	 amdgpu_amdkfd.o \
+	 amdgpu_amdkfd_gfx_v8.o
+
+# add cgs
+amdgpu-y += amdgpu_cgs.o
+
+# GPU scheduler
+amdgpu-y += \
+	../scheduler/gpu_scheduler.o \
+	../scheduler/sched_fence.o \
+	amdgpu_sched.o
+
+amdgpu-$(CONFIG_COMPAT) += amdgpu_ioc32.o
+amdgpu-$(CONFIG_VGA_SWITCHEROO) += amdgpu_atpx_handler.o
+amdgpu-$(CONFIG_ACPI) += amdgpu_acpi.o
+amdgpu-$(CONFIG_MMU_NOTIFIER) += amdgpu_mn.o
+
+obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o
+
+CFLAGS_amdgpu_trace_points.o := -I$(src)
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/ObjectID.h b/kernel/drivers/gpu/drm/amd/amdgpu/ObjectID.h
new file mode 100644
index 000000000..06192698b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/ObjectID.h
@@ -0,0 +1,736 @@
+/*
+* Copyright 2006-2007 Advanced Micro Devices, Inc.  
+*
+* Permission is hereby granted, free of charge, to any person obtaining a
+* copy of this software and associated documentation files (the "Software"),
+* to deal in the Software without restriction, including without limitation
+* the rights to use, copy, modify, merge, publish, distribute, sublicense,
+* and/or sell copies of the Software, and to permit persons to whom the
+* Software is furnished to do so, subject to the following conditions:
+*
+* The above copyright notice and this permission notice shall be included in
+* all copies or substantial portions of the Software.
+*
+* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+* OTHER DEALINGS IN THE SOFTWARE.
+*/
+/* based on stg/asic_reg/drivers/inc/asic_reg/ObjectID.h ver 23 */
+
+#ifndef _OBJECTID_H
+#define _OBJECTID_H
+
+#if defined(_X86_)
+#pragma pack(1)
+#endif
+
+/****************************************************/
+/* Graphics Object Type Definition                  */
+/****************************************************/
+#define GRAPH_OBJECT_TYPE_NONE                    0x0
+#define GRAPH_OBJECT_TYPE_GPU                     0x1
+#define GRAPH_OBJECT_TYPE_ENCODER                 0x2
+#define GRAPH_OBJECT_TYPE_CONNECTOR               0x3
+#define GRAPH_OBJECT_TYPE_ROUTER                  0x4
+/* deleted */
+#define GRAPH_OBJECT_TYPE_DISPLAY_PATH            0x6  
+#define GRAPH_OBJECT_TYPE_GENERIC                 0x7
+
+/****************************************************/
+/* Encoder Object ID Definition                     */
+/****************************************************/
+#define ENCODER_OBJECT_ID_NONE                    0x00 
+
+/* Radeon Class Display Hardware */
+#define ENCODER_OBJECT_ID_INTERNAL_LVDS           0x01
+#define ENCODER_OBJECT_ID_INTERNAL_TMDS1          0x02
+#define ENCODER_OBJECT_ID_INTERNAL_TMDS2          0x03
+#define ENCODER_OBJECT_ID_INTERNAL_DAC1           0x04
+#define ENCODER_OBJECT_ID_INTERNAL_DAC2           0x05     /* TV/CV DAC */
+#define ENCODER_OBJECT_ID_INTERNAL_SDVOA          0x06
+#define ENCODER_OBJECT_ID_INTERNAL_SDVOB          0x07
+
+/* External Third Party Encoders */
+#define ENCODER_OBJECT_ID_SI170B                  0x08
+#define ENCODER_OBJECT_ID_CH7303                  0x09
+#define ENCODER_OBJECT_ID_CH7301                  0x0A
+#define ENCODER_OBJECT_ID_INTERNAL_DVO1           0x0B    /* This belongs to Radeon Class Display Hardware */
+#define ENCODER_OBJECT_ID_EXTERNAL_SDVOA          0x0C
+#define ENCODER_OBJECT_ID_EXTERNAL_SDVOB          0x0D
+#define ENCODER_OBJECT_ID_TITFP513                0x0E
+#define ENCODER_OBJECT_ID_INTERNAL_LVTM1          0x0F    /* not used for Radeon */
+#define ENCODER_OBJECT_ID_VT1623                  0x10
+#define ENCODER_OBJECT_ID_HDMI_SI1930             0x11
+#define ENCODER_OBJECT_ID_HDMI_INTERNAL           0x12
+#define ENCODER_OBJECT_ID_ALMOND                  0x22
+#define ENCODER_OBJECT_ID_TRAVIS                  0x23
+#define ENCODER_OBJECT_ID_NUTMEG                  0x22
+#define ENCODER_OBJECT_ID_HDMI_ANX9805            0x26
+
+/* Kaleidoscope (KLDSCP) Class Display Hardware (internal) */
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1   0x13
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1    0x14
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1    0x15
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2    0x16  /* Shared with CV/TV and CRT */
+#define ENCODER_OBJECT_ID_SI178                   0X17  /* External TMDS (dual link, no HDCP.) */
+#define ENCODER_OBJECT_ID_MVPU_FPGA               0x18  /* MVPU FPGA chip */
+#define ENCODER_OBJECT_ID_INTERNAL_DDI            0x19
+#define ENCODER_OBJECT_ID_VT1625                  0x1A
+#define ENCODER_OBJECT_ID_HDMI_SI1932             0x1B
+#define ENCODER_OBJECT_ID_DP_AN9801               0x1C
+#define ENCODER_OBJECT_ID_DP_DP501                0x1D
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY         0x1E
+#define ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA   0x1F
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY1        0x20
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY2        0x21
+#define ENCODER_OBJECT_ID_INTERNAL_VCE            0x24
+#define ENCODER_OBJECT_ID_INTERNAL_UNIPHY3        0x25
+#define ENCODER_OBJECT_ID_INTERNAL_AMCLK          0x27
+
+#define ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO    0xFF
+
+/****************************************************/
+/* Connector Object ID Definition                   */
+/****************************************************/
+#define CONNECTOR_OBJECT_ID_NONE                  0x00 
+#define CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I     0x01
+#define CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I       0x02
+#define CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D     0x03
+#define CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D       0x04
+#define CONNECTOR_OBJECT_ID_VGA                   0x05
+#define CONNECTOR_OBJECT_ID_COMPOSITE             0x06
+#define CONNECTOR_OBJECT_ID_SVIDEO                0x07
+#define CONNECTOR_OBJECT_ID_YPbPr                 0x08
+#define CONNECTOR_OBJECT_ID_D_CONNECTOR           0x09
+#define CONNECTOR_OBJECT_ID_9PIN_DIN              0x0A  /* Supports both CV & TV */
+#define CONNECTOR_OBJECT_ID_SCART                 0x0B
+#define CONNECTOR_OBJECT_ID_HDMI_TYPE_A           0x0C
+#define CONNECTOR_OBJECT_ID_HDMI_TYPE_B           0x0D
+#define CONNECTOR_OBJECT_ID_LVDS                  0x0E
+#define CONNECTOR_OBJECT_ID_7PIN_DIN              0x0F
+#define CONNECTOR_OBJECT_ID_PCIE_CONNECTOR        0x10
+#define CONNECTOR_OBJECT_ID_CROSSFIRE             0x11
+#define CONNECTOR_OBJECT_ID_HARDCODE_DVI          0x12
+#define CONNECTOR_OBJECT_ID_DISPLAYPORT           0x13
+#define CONNECTOR_OBJECT_ID_eDP                   0x14
+#define CONNECTOR_OBJECT_ID_MXM                   0x15
+#define CONNECTOR_OBJECT_ID_LVDS_eDP              0x16
+
+/* deleted */
+
+/****************************************************/
+/* Router Object ID Definition                      */
+/****************************************************/
+#define ROUTER_OBJECT_ID_NONE											0x00
+#define ROUTER_OBJECT_ID_I2C_EXTENDER_CNTL				0x01
+
+/****************************************************/
+/* Generic Object ID Definition                     */
+/****************************************************/
+#define GENERIC_OBJECT_ID_NONE                    0x00
+#define GENERIC_OBJECT_ID_GLSYNC                  0x01
+#define GENERIC_OBJECT_ID_PX2_NON_DRIVABLE        0x02
+#define GENERIC_OBJECT_ID_MXM_OPM                 0x03
+#define GENERIC_OBJECT_ID_STEREO_PIN              0x04        //This object could show up from Misc Object table, it follows ATOM_OBJECT format, and contains one ATOM_OBJECT_GPIO_CNTL_RECORD for the stereo pin
+
+/****************************************************/
+/* Graphics Object ENUM ID Definition               */
+/****************************************************/
+#define GRAPH_OBJECT_ENUM_ID1                     0x01
+#define GRAPH_OBJECT_ENUM_ID2                     0x02
+#define GRAPH_OBJECT_ENUM_ID3                     0x03
+#define GRAPH_OBJECT_ENUM_ID4                     0x04
+#define GRAPH_OBJECT_ENUM_ID5                     0x05
+#define GRAPH_OBJECT_ENUM_ID6                     0x06
+#define GRAPH_OBJECT_ENUM_ID7                     0x07
+
+/****************************************************/
+/* Graphics Object ID Bit definition                */
+/****************************************************/
+#define OBJECT_ID_MASK                            0x00FF
+#define ENUM_ID_MASK                              0x0700
+#define RESERVED1_ID_MASK                         0x0800
+#define OBJECT_TYPE_MASK                          0x7000
+#define RESERVED2_ID_MASK                         0x8000
+                                                  
+#define OBJECT_ID_SHIFT                           0x00
+#define ENUM_ID_SHIFT                             0x08
+#define OBJECT_TYPE_SHIFT                         0x0C
+
+
+/****************************************************/
+/* Graphics Object family definition                */
+/****************************************************/
+#define CONSTRUCTOBJECTFAMILYID(GRAPHICS_OBJECT_TYPE, GRAPHICS_OBJECT_ID) (GRAPHICS_OBJECT_TYPE << OBJECT_TYPE_SHIFT | \
+                                                                           GRAPHICS_OBJECT_ID   << OBJECT_ID_SHIFT)
+/****************************************************/
+/* GPU Object ID definition - Shared with BIOS      */
+/****************************************************/
+#define GPU_ENUM_ID1                            ( GRAPH_OBJECT_TYPE_GPU << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT)
+
+/****************************************************/
+/* Encoder Object ID definition - Shared with BIOS  */
+/****************************************************/
+/*
+#define ENCODER_INTERNAL_LVDS_ENUM_ID1        0x2101      
+#define ENCODER_INTERNAL_TMDS1_ENUM_ID1       0x2102
+#define ENCODER_INTERNAL_TMDS2_ENUM_ID1       0x2103
+#define ENCODER_INTERNAL_DAC1_ENUM_ID1        0x2104
+#define ENCODER_INTERNAL_DAC2_ENUM_ID1        0x2105
+#define ENCODER_INTERNAL_SDVOA_ENUM_ID1       0x2106
+#define ENCODER_INTERNAL_SDVOB_ENUM_ID1       0x2107
+#define ENCODER_SIL170B_ENUM_ID1              0x2108  
+#define ENCODER_CH7303_ENUM_ID1               0x2109
+#define ENCODER_CH7301_ENUM_ID1               0x210A
+#define ENCODER_INTERNAL_DVO1_ENUM_ID1        0x210B
+#define ENCODER_EXTERNAL_SDVOA_ENUM_ID1       0x210C
+#define ENCODER_EXTERNAL_SDVOB_ENUM_ID1       0x210D
+#define ENCODER_TITFP513_ENUM_ID1             0x210E
+#define ENCODER_INTERNAL_LVTM1_ENUM_ID1       0x210F
+#define ENCODER_VT1623_ENUM_ID1               0x2110
+#define ENCODER_HDMI_SI1930_ENUM_ID1          0x2111
+#define ENCODER_HDMI_INTERNAL_ENUM_ID1        0x2112
+#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1   0x2113
+#define ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1    0x2114
+#define ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1    0x2115
+#define ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1    0x2116  
+#define ENCODER_SI178_ENUM_ID1                   0x2117 
+#define ENCODER_MVPU_FPGA_ENUM_ID1               0x2118
+#define ENCODER_INTERNAL_DDI_ENUM_ID1            0x2119
+#define ENCODER_VT1625_ENUM_ID1                  0x211A
+#define ENCODER_HDMI_SI1932_ENUM_ID1             0x211B
+#define ENCODER_ENCODER_DP_AN9801_ENUM_ID1       0x211C
+#define ENCODER_DP_DP501_ENUM_ID1                0x211D
+#define ENCODER_INTERNAL_UNIPHY_ENUM_ID1         0x211E
+*/
+#define ENCODER_INTERNAL_LVDS_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_LVDS << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_TMDS1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_TMDS1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_TMDS2_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_TMDS2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DAC1_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DAC1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DAC2_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DAC2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_SDVOA_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_SDVOA_ENUM_ID2    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_SDVOB_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_SDVOB << OBJECT_ID_SHIFT)
+
+#define ENCODER_SIL170B_ENUM_ID1           ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_SI170B << OBJECT_ID_SHIFT)
+
+#define ENCODER_CH7303_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_CH7303 << OBJECT_ID_SHIFT)
+
+#define ENCODER_CH7301_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_CH7301 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DVO1_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DVO1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_EXTERNAL_SDVOA_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_EXTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+#define ENCODER_EXTERNAL_SDVOA_ENUM_ID2    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_EXTERNAL_SDVOA << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_EXTERNAL_SDVOB_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_EXTERNAL_SDVOB << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_TITFP513_ENUM_ID1          ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_TITFP513 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_LVTM1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_LVTM1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_VT1623_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_VT1623 << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_SI1930_ENUM_ID1       ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_HDMI_SI1930 << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_INTERNAL_ENUM_ID1     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_HDMI_INTERNAL << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID1   ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_INTERNAL_KLDSCP_TMDS1_ENUM_ID2   ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1 << OBJECT_ID_SHIFT)
+
+
+#define ENCODER_INTERNAL_KLDSCP_DVO1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_DAC1_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_DAC2_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2 << OBJECT_ID_SHIFT)  // Shared with CV/TV and CRT
+
+#define ENCODER_SI178_ENUM_ID1                    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_SI178 << OBJECT_ID_SHIFT)  
+
+#define ENCODER_MVPU_FPGA_ENUM_ID1                ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                   GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                   ENCODER_OBJECT_ID_MVPU_FPGA << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_DDI_ENUM_ID1     (  GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_INTERNAL_DDI << OBJECT_ID_SHIFT) 
+
+#define ENCODER_VT1625_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_VT1625 << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_SI1932_ENUM_ID1       ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_HDMI_SI1932 << OBJECT_ID_SHIFT)
+
+#define ENCODER_DP_DP501_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_DP_DP501 << OBJECT_ID_SHIFT)
+
+#define ENCODER_DP_AN9801_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                             GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                             ENCODER_OBJECT_ID_DP_AN9801 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_KLDSCP_LVTMA_ENUM_ID1   ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA << OBJECT_ID_SHIFT)  
+
+#define ENCODER_INTERNAL_UNIPHY1_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY1_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY1 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY2_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY2_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY2 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY3_ENUM_ID1         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY3 << OBJECT_ID_SHIFT)
+
+#define ENCODER_INTERNAL_UNIPHY3_ENUM_ID2         ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 ENCODER_OBJECT_ID_INTERNAL_UNIPHY3 << OBJECT_ID_SHIFT)
+
+#define ENCODER_GENERAL_EXTERNAL_DVO_ENUM_ID1    ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO << OBJECT_ID_SHIFT)
+
+#define ENCODER_ALMOND_ENUM_ID1                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_ALMOND << OBJECT_ID_SHIFT)
+
+#define ENCODER_ALMOND_ENUM_ID2                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_ALMOND << OBJECT_ID_SHIFT)
+
+#define ENCODER_TRAVIS_ENUM_ID1                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_TRAVIS << OBJECT_ID_SHIFT)
+
+#define ENCODER_TRAVIS_ENUM_ID2                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_TRAVIS << OBJECT_ID_SHIFT)
+
+#define ENCODER_NUTMEG_ENUM_ID1                  ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_NUTMEG << OBJECT_ID_SHIFT)
+
+#define ENCODER_VCE_ENUM_ID1                     ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_INTERNAL_VCE << OBJECT_ID_SHIFT)
+
+#define ENCODER_HDMI_ANX9805_ENUM_ID1            ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\
+                                                  GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                  ENCODER_OBJECT_ID_HDMI_ANX9805 << OBJECT_ID_SHIFT)
+
+/****************************************************/
+/* Connector Object ID definition - Shared with BIOS */
+/****************************************************/
+/*
+#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID1        0x3101
+#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID1          0x3102
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID1        0x3103
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID1          0x3104
+#define CONNECTOR_VGA_ENUM_ID1                      0x3105
+#define CONNECTOR_COMPOSITE_ENUM_ID1                0x3106
+#define CONNECTOR_SVIDEO_ENUM_ID1                   0x3107
+#define CONNECTOR_YPbPr_ENUM_ID1                    0x3108
+#define CONNECTOR_D_CONNECTORE_ENUM_ID1             0x3109
+#define CONNECTOR_9PIN_DIN_ENUM_ID1                 0x310A
+#define CONNECTOR_SCART_ENUM_ID1                    0x310B
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID1              0x310C
+#define CONNECTOR_HDMI_TYPE_B_ENUM_ID1              0x310D
+#define CONNECTOR_LVDS_ENUM_ID1                     0x310E
+#define CONNECTOR_7PIN_DIN_ENUM_ID1                 0x310F
+#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID1           0x3110
+*/
+#define CONNECTOR_LVDS_ENUM_ID1                ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_LVDS_ENUM_ID2                ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_eDP_ENUM_ID1                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_eDP << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_eDP_ENUM_ID2                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_eDP << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID1   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_I_ENUM_ID2   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID1     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_I_ENUM_ID2     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID1   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID2   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID3   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID4   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID5   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SINGLE_LINK_DVI_D_ENUM_ID6   ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SINGLE_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID1     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID2     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID3     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DUAL_LINK_DVI_D_ENUM_ID4     ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_VGA_ENUM_ID1                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_VGA << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_VGA_ENUM_ID2                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_VGA << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_COMPOSITE_ENUM_ID1           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_COMPOSITE << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_COMPOSITE_ENUM_ID2           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_COMPOSITE << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SVIDEO_ENUM_ID1              ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SVIDEO << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SVIDEO_ENUM_ID2              ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SVIDEO << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_YPbPr_ENUM_ID1               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_YPbPr << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_YPbPr_ENUM_ID2               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_YPbPr << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_D_CONNECTOR_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_D_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_D_CONNECTOR_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_D_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_9PIN_DIN_ENUM_ID1            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_9PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_9PIN_DIN_ENUM_ID2            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_9PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SCART_ENUM_ID1               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SCART << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_SCART_ENUM_ID2               ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_SCART << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID3         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID4         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID5         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_A_ENUM_ID6         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_A << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_B_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_B << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HDMI_TYPE_B_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HDMI_TYPE_B << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_7PIN_DIN_ENUM_ID1            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_7PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_7PIN_DIN_ENUM_ID2            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_7PIN_DIN << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID1      ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_PCIE_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_PCIE_CONNECTOR_ENUM_ID2      ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_PCIE_CONNECTOR << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_CROSSFIRE_ENUM_ID1           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_CROSSFIRE << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_CROSSFIRE_ENUM_ID2           ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_CROSSFIRE << OBJECT_ID_SHIFT)
+
+
+#define CONNECTOR_HARDCODE_DVI_ENUM_ID1        ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HARDCODE_DVI << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_HARDCODE_DVI_ENUM_ID2        ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_HARDCODE_DVI << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID1         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID2         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID3         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID4         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID5         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_DISPLAYPORT_ENUM_ID6         ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_DISPLAYPORT << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_MXM_ENUM_ID1                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_A
+
+#define CONNECTOR_MXM_ENUM_ID2                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_B
+
+#define CONNECTOR_MXM_ENUM_ID3                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID3 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_C
+
+#define CONNECTOR_MXM_ENUM_ID4                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID4 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DP_D
+
+#define CONNECTOR_MXM_ENUM_ID5                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID5 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_LVDS_TXxx
+
+#define CONNECTOR_MXM_ENUM_ID6                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID6 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_LVDS_UXxx
+
+#define CONNECTOR_MXM_ENUM_ID7                 ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID7 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_MXM << OBJECT_ID_SHIFT)          //Mapping to MXM_DAC
+
+#define CONNECTOR_LVDS_eDP_ENUM_ID1            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS_eDP << OBJECT_ID_SHIFT)
+
+#define CONNECTOR_LVDS_eDP_ENUM_ID2            ( GRAPH_OBJECT_TYPE_CONNECTOR << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 CONNECTOR_OBJECT_ID_LVDS_eDP << OBJECT_ID_SHIFT)
+
+/****************************************************/
+/* Router Object ID definition - Shared with BIOS   */
+/****************************************************/
+#define ROUTER_I2C_EXTENDER_CNTL_ENUM_ID1      ( GRAPH_OBJECT_TYPE_ROUTER << OBJECT_TYPE_SHIFT |\
+                                                GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                ROUTER_OBJECT_ID_I2C_EXTENDER_CNTL << OBJECT_ID_SHIFT)
+
+/* deleted */
+
+/****************************************************/
+/* Generic Object ID definition - Shared with BIOS  */
+/****************************************************/
+#define GENERICOBJECT_GLSYNC_ENUM_ID1           (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_GLSYNC << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_PX2_NON_DRIVABLE_ID1       (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_PX2_NON_DRIVABLE<< OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_PX2_NON_DRIVABLE_ID2       (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID2 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_PX2_NON_DRIVABLE<< OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_MXM_OPM_ENUM_ID1           (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_MXM_OPM << OBJECT_ID_SHIFT)
+
+#define GENERICOBJECT_STEREO_PIN_ENUM_ID1        (GRAPH_OBJECT_TYPE_GENERIC << OBJECT_TYPE_SHIFT |\
+                                                 GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\
+                                                 GENERIC_OBJECT_ID_STEREO_PIN << OBJECT_ID_SHIFT)
+
+/****************************************************/
+/* Object Cap definition - Shared with BIOS         */
+/****************************************************/
+#define GRAPHICS_OBJECT_CAP_I2C                 0x00000001L
+#define GRAPHICS_OBJECT_CAP_TABLE_ID            0x00000002L
+
+
+#define GRAPHICS_OBJECT_I2CCOMMAND_TABLE_ID                   0x01
+#define GRAPHICS_OBJECT_HOTPLUGDETECTIONINTERUPT_TABLE_ID     0x02
+#define GRAPHICS_OBJECT_ENCODER_OUTPUT_PROTECTION_TABLE_ID    0x03
+
+#if defined(_X86_)
+#pragma pack()
+#endif
+
+#endif  /*GRAPHICTYPE */
+
+
+
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu.h
new file mode 100644
index 000000000..bb1099c54
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -0,0 +1,2401 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#ifndef __AMDGPU_H__
+#define __AMDGPU_H__
+
+#include <linux/atomic.h>
+#include <linux/wait.h>
+#include <linux/list.h>
+#include <linux/kref.h>
+#include <linux/interval_tree.h>
+#include <linux/hashtable.h>
+#include <linux/fence.h>
+
+#include <ttm/ttm_bo_api.h>
+#include <ttm/ttm_bo_driver.h>
+#include <ttm/ttm_placement.h>
+#include <ttm/ttm_module.h>
+#include <ttm/ttm_execbuf_util.h>
+
+#include <drm/drmP.h>
+#include <drm/drm_gem.h>
+#include <drm/amdgpu_drm.h>
+
+#include "amd_shared.h"
+#include "amdgpu_mode.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_irq.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_gds.h"
+
+#include "gpu_scheduler.h"
+
+/*
+ * Modules parameters.
+ */
+extern int amdgpu_modeset;
+extern int amdgpu_vram_limit;
+extern int amdgpu_gart_size;
+extern int amdgpu_benchmarking;
+extern int amdgpu_testing;
+extern int amdgpu_audio;
+extern int amdgpu_disp_priority;
+extern int amdgpu_hw_i2c;
+extern int amdgpu_pcie_gen2;
+extern int amdgpu_msi;
+extern int amdgpu_lockup_timeout;
+extern int amdgpu_dpm;
+extern int amdgpu_smc_load_fw;
+extern int amdgpu_aspm;
+extern int amdgpu_runtime_pm;
+extern int amdgpu_hard_reset;
+extern unsigned amdgpu_ip_block_mask;
+extern int amdgpu_bapm;
+extern int amdgpu_deep_color;
+extern int amdgpu_vm_size;
+extern int amdgpu_vm_block_size;
+extern int amdgpu_vm_fault_stop;
+extern int amdgpu_vm_debug;
+extern int amdgpu_enable_scheduler;
+extern int amdgpu_sched_jobs;
+extern int amdgpu_sched_hw_submission;
+extern int amdgpu_enable_semaphores;
+
+#define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
+#define AMDGPU_MAX_USEC_TIMEOUT			100000	/* 100 ms */
+#define AMDGPU_FENCE_JIFFIES_TIMEOUT		(HZ / 2)
+/* AMDGPU_IB_POOL_SIZE must be a power of 2 */
+#define AMDGPU_IB_POOL_SIZE			16
+#define AMDGPU_DEBUGFS_MAX_COMPONENTS		32
+#define AMDGPUFB_CONN_LIMIT			4
+#define AMDGPU_BIOS_NUM_SCRATCH			8
+
+/* max number of rings */
+#define AMDGPU_MAX_RINGS			16
+#define AMDGPU_MAX_GFX_RINGS			1
+#define AMDGPU_MAX_COMPUTE_RINGS		8
+#define AMDGPU_MAX_VCE_RINGS			2
+
+/* max number of IP instances */
+#define AMDGPU_MAX_SDMA_INSTANCES		2
+
+/* number of hw syncs before falling back on blocking */
+#define AMDGPU_NUM_SYNCS			4
+
+/* hardcode that limit for now */
+#define AMDGPU_VA_RESERVED_SIZE			(8 << 20)
+
+/* hard reset data */
+#define AMDGPU_ASIC_RESET_DATA                  0x39d5e86b
+
+/* reset flags */
+#define AMDGPU_RESET_GFX			(1 << 0)
+#define AMDGPU_RESET_COMPUTE			(1 << 1)
+#define AMDGPU_RESET_DMA			(1 << 2)
+#define AMDGPU_RESET_CP				(1 << 3)
+#define AMDGPU_RESET_GRBM			(1 << 4)
+#define AMDGPU_RESET_DMA1			(1 << 5)
+#define AMDGPU_RESET_RLC			(1 << 6)
+#define AMDGPU_RESET_SEM			(1 << 7)
+#define AMDGPU_RESET_IH				(1 << 8)
+#define AMDGPU_RESET_VMC			(1 << 9)
+#define AMDGPU_RESET_MC				(1 << 10)
+#define AMDGPU_RESET_DISPLAY			(1 << 11)
+#define AMDGPU_RESET_UVD			(1 << 12)
+#define AMDGPU_RESET_VCE			(1 << 13)
+#define AMDGPU_RESET_VCE1			(1 << 14)
+
+/* CG block flags */
+#define AMDGPU_CG_BLOCK_GFX			(1 << 0)
+#define AMDGPU_CG_BLOCK_MC			(1 << 1)
+#define AMDGPU_CG_BLOCK_SDMA			(1 << 2)
+#define AMDGPU_CG_BLOCK_UVD			(1 << 3)
+#define AMDGPU_CG_BLOCK_VCE			(1 << 4)
+#define AMDGPU_CG_BLOCK_HDP			(1 << 5)
+#define AMDGPU_CG_BLOCK_BIF			(1 << 6)
+
+/* CG flags */
+#define AMDGPU_CG_SUPPORT_GFX_MGCG		(1 << 0)
+#define AMDGPU_CG_SUPPORT_GFX_MGLS		(1 << 1)
+#define AMDGPU_CG_SUPPORT_GFX_CGCG		(1 << 2)
+#define AMDGPU_CG_SUPPORT_GFX_CGLS		(1 << 3)
+#define AMDGPU_CG_SUPPORT_GFX_CGTS		(1 << 4)
+#define AMDGPU_CG_SUPPORT_GFX_CGTS_LS		(1 << 5)
+#define AMDGPU_CG_SUPPORT_GFX_CP_LS		(1 << 6)
+#define AMDGPU_CG_SUPPORT_GFX_RLC_LS		(1 << 7)
+#define AMDGPU_CG_SUPPORT_MC_LS			(1 << 8)
+#define AMDGPU_CG_SUPPORT_MC_MGCG		(1 << 9)
+#define AMDGPU_CG_SUPPORT_SDMA_LS		(1 << 10)
+#define AMDGPU_CG_SUPPORT_SDMA_MGCG		(1 << 11)
+#define AMDGPU_CG_SUPPORT_BIF_LS		(1 << 12)
+#define AMDGPU_CG_SUPPORT_UVD_MGCG		(1 << 13)
+#define AMDGPU_CG_SUPPORT_VCE_MGCG		(1 << 14)
+#define AMDGPU_CG_SUPPORT_HDP_LS		(1 << 15)
+#define AMDGPU_CG_SUPPORT_HDP_MGCG		(1 << 16)
+
+/* PG flags */
+#define AMDGPU_PG_SUPPORT_GFX_PG		(1 << 0)
+#define AMDGPU_PG_SUPPORT_GFX_SMG		(1 << 1)
+#define AMDGPU_PG_SUPPORT_GFX_DMG		(1 << 2)
+#define AMDGPU_PG_SUPPORT_UVD			(1 << 3)
+#define AMDGPU_PG_SUPPORT_VCE			(1 << 4)
+#define AMDGPU_PG_SUPPORT_CP			(1 << 5)
+#define AMDGPU_PG_SUPPORT_GDS			(1 << 6)
+#define AMDGPU_PG_SUPPORT_RLC_SMU_HS		(1 << 7)
+#define AMDGPU_PG_SUPPORT_SDMA			(1 << 8)
+#define AMDGPU_PG_SUPPORT_ACP			(1 << 9)
+#define AMDGPU_PG_SUPPORT_SAMU			(1 << 10)
+
+/* GFX current status */
+#define AMDGPU_GFX_NORMAL_MODE			0x00000000L
+#define AMDGPU_GFX_SAFE_MODE			0x00000001L
+#define AMDGPU_GFX_PG_DISABLED_MODE		0x00000002L
+#define AMDGPU_GFX_CG_DISABLED_MODE		0x00000004L
+#define AMDGPU_GFX_LBPW_DISABLED_MODE		0x00000008L
+
+/* max cursor sizes (in pixels) */
+#define CIK_CURSOR_WIDTH 128
+#define CIK_CURSOR_HEIGHT 128
+
+struct amdgpu_device;
+struct amdgpu_fence;
+struct amdgpu_ib;
+struct amdgpu_vm;
+struct amdgpu_ring;
+struct amdgpu_semaphore;
+struct amdgpu_cs_parser;
+struct amdgpu_job;
+struct amdgpu_irq_src;
+struct amdgpu_fpriv;
+
+enum amdgpu_cp_irq {
+	AMDGPU_CP_IRQ_GFX_EOP = 0,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP,
+	AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP,
+
+	AMDGPU_CP_IRQ_LAST
+};
+
+enum amdgpu_sdma_irq {
+	AMDGPU_SDMA_IRQ_TRAP0 = 0,
+	AMDGPU_SDMA_IRQ_TRAP1,
+
+	AMDGPU_SDMA_IRQ_LAST
+};
+
+enum amdgpu_thermal_irq {
+	AMDGPU_THERMAL_IRQ_LOW_TO_HIGH = 0,
+	AMDGPU_THERMAL_IRQ_HIGH_TO_LOW,
+
+	AMDGPU_THERMAL_IRQ_LAST
+};
+
+int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
+				  enum amd_ip_block_type block_type,
+				  enum amd_clockgating_state state);
+int amdgpu_set_powergating_state(struct amdgpu_device *adev,
+				  enum amd_ip_block_type block_type,
+				  enum amd_powergating_state state);
+
+struct amdgpu_ip_block_version {
+	enum amd_ip_block_type type;
+	u32 major;
+	u32 minor;
+	u32 rev;
+	const struct amd_ip_funcs *funcs;
+};
+
+int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev,
+				enum amd_ip_block_type type,
+				u32 major, u32 minor);
+
+const struct amdgpu_ip_block_version * amdgpu_get_ip_block(
+					struct amdgpu_device *adev,
+					enum amd_ip_block_type type);
+
+/* provided by hw blocks that can move/clear data.  e.g., gfx or sdma */
+struct amdgpu_buffer_funcs {
+	/* maximum bytes in a single operation */
+	uint32_t	copy_max_bytes;
+
+	/* number of dw to reserve per operation */
+	unsigned	copy_num_dw;
+
+	/* used for buffer migration */
+	void (*emit_copy_buffer)(struct amdgpu_ib *ib,
+				 /* src addr in bytes */
+				 uint64_t src_offset,
+				 /* dst addr in bytes */
+				 uint64_t dst_offset,
+				 /* number of byte to transfer */
+				 uint32_t byte_count);
+
+	/* maximum bytes in a single operation */
+	uint32_t	fill_max_bytes;
+
+	/* number of dw to reserve per operation */
+	unsigned	fill_num_dw;
+
+	/* used for buffer clearing */
+	void (*emit_fill_buffer)(struct amdgpu_ib *ib,
+				 /* value to write to memory */
+				 uint32_t src_data,
+				 /* dst addr in bytes */
+				 uint64_t dst_offset,
+				 /* number of byte to fill */
+				 uint32_t byte_count);
+};
+
+/* provided by hw blocks that can write ptes, e.g., sdma */
+struct amdgpu_vm_pte_funcs {
+	/* copy pte entries from GART */
+	void (*copy_pte)(struct amdgpu_ib *ib,
+			 uint64_t pe, uint64_t src,
+			 unsigned count);
+	/* write pte one entry at a time with addr mapping */
+	void (*write_pte)(struct amdgpu_ib *ib,
+			  uint64_t pe,
+			  uint64_t addr, unsigned count,
+			  uint32_t incr, uint32_t flags);
+	/* for linear pte/pde updates without addr mapping */
+	void (*set_pte_pde)(struct amdgpu_ib *ib,
+			    uint64_t pe,
+			    uint64_t addr, unsigned count,
+			    uint32_t incr, uint32_t flags);
+	/* pad the indirect buffer to the necessary number of dw */
+	void (*pad_ib)(struct amdgpu_ib *ib);
+};
+
+/* provided by the gmc block */
+struct amdgpu_gart_funcs {
+	/* flush the vm tlb via mmio */
+	void (*flush_gpu_tlb)(struct amdgpu_device *adev,
+			      uint32_t vmid);
+	/* write pte/pde updates using the cpu */
+	int (*set_pte_pde)(struct amdgpu_device *adev,
+			   void *cpu_pt_addr, /* cpu addr of page table */
+			   uint32_t gpu_page_idx, /* pte/pde to update */
+			   uint64_t addr, /* addr to write into pte/pde */
+			   uint32_t flags); /* access flags */
+};
+
+/* provided by the ih block */
+struct amdgpu_ih_funcs {
+	/* ring read/write ptr handling, called from interrupt context */
+	u32 (*get_wptr)(struct amdgpu_device *adev);
+	void (*decode_iv)(struct amdgpu_device *adev,
+			  struct amdgpu_iv_entry *entry);
+	void (*set_rptr)(struct amdgpu_device *adev);
+};
+
+/* provided by hw blocks that expose a ring buffer for commands */
+struct amdgpu_ring_funcs {
+	/* ring read/write ptr handling */
+	u32 (*get_rptr)(struct amdgpu_ring *ring);
+	u32 (*get_wptr)(struct amdgpu_ring *ring);
+	void (*set_wptr)(struct amdgpu_ring *ring);
+	/* validating and patching of IBs */
+	int (*parse_cs)(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+	/* command emit functions */
+	void (*emit_ib)(struct amdgpu_ring *ring,
+			struct amdgpu_ib *ib);
+	void (*emit_fence)(struct amdgpu_ring *ring, uint64_t addr,
+			   uint64_t seq, unsigned flags);
+	bool (*emit_semaphore)(struct amdgpu_ring *ring,
+			       struct amdgpu_semaphore *semaphore,
+			       bool emit_wait);
+	void (*emit_vm_flush)(struct amdgpu_ring *ring, unsigned vm_id,
+			      uint64_t pd_addr);
+	void (*emit_hdp_flush)(struct amdgpu_ring *ring);
+	void (*emit_gds_switch)(struct amdgpu_ring *ring, uint32_t vmid,
+				uint32_t gds_base, uint32_t gds_size,
+				uint32_t gws_base, uint32_t gws_size,
+				uint32_t oa_base, uint32_t oa_size);
+	/* testing functions */
+	int (*test_ring)(struct amdgpu_ring *ring);
+	int (*test_ib)(struct amdgpu_ring *ring);
+	/* insert NOP packets */
+	void (*insert_nop)(struct amdgpu_ring *ring, uint32_t count);
+};
+
+/*
+ * BIOS.
+ */
+bool amdgpu_get_bios(struct amdgpu_device *adev);
+bool amdgpu_read_bios(struct amdgpu_device *adev);
+
+/*
+ * Dummy page
+ */
+struct amdgpu_dummy_page {
+	struct page	*page;
+	dma_addr_t	addr;
+};
+int amdgpu_dummy_page_init(struct amdgpu_device *adev);
+void amdgpu_dummy_page_fini(struct amdgpu_device *adev);
+
+
+/*
+ * Clocks
+ */
+
+#define AMDGPU_MAX_PPLL 3
+
+struct amdgpu_clock {
+	struct amdgpu_pll ppll[AMDGPU_MAX_PPLL];
+	struct amdgpu_pll spll;
+	struct amdgpu_pll mpll;
+	/* 10 Khz units */
+	uint32_t default_mclk;
+	uint32_t default_sclk;
+	uint32_t default_dispclk;
+	uint32_t current_dispclk;
+	uint32_t dp_extclk;
+	uint32_t max_pixel_clock;
+};
+
+/*
+ * Fences.
+ */
+struct amdgpu_fence_driver {
+	uint64_t			gpu_addr;
+	volatile uint32_t		*cpu_addr;
+	/* sync_seq is protected by ring emission lock */
+	uint64_t			sync_seq[AMDGPU_MAX_RINGS];
+	atomic64_t			last_seq;
+	bool				initialized;
+	struct amdgpu_irq_src		*irq_src;
+	unsigned			irq_type;
+	struct timer_list		fallback_timer;
+	wait_queue_head_t		fence_queue;
+};
+
+/* some special values for the owner field */
+#define AMDGPU_FENCE_OWNER_UNDEFINED	((void*)0ul)
+#define AMDGPU_FENCE_OWNER_VM		((void*)1ul)
+
+#define AMDGPU_FENCE_FLAG_64BIT         (1 << 0)
+#define AMDGPU_FENCE_FLAG_INT           (1 << 1)
+
+struct amdgpu_fence {
+	struct fence base;
+
+	/* RB, DMA, etc. */
+	struct amdgpu_ring		*ring;
+	uint64_t			seq;
+
+	/* filp or special value for fence creator */
+	void				*owner;
+
+	wait_queue_t			fence_wake;
+};
+
+struct amdgpu_user_fence {
+	/* write-back bo */
+	struct amdgpu_bo 	*bo;
+	/* write-back address offset to bo start */
+	uint32_t                offset;
+};
+
+int amdgpu_fence_driver_init(struct amdgpu_device *adev);
+void amdgpu_fence_driver_fini(struct amdgpu_device *adev);
+void amdgpu_fence_driver_force_completion(struct amdgpu_device *adev);
+
+int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring);
+int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
+				   struct amdgpu_irq_src *irq_src,
+				   unsigned irq_type);
+void amdgpu_fence_driver_suspend(struct amdgpu_device *adev);
+void amdgpu_fence_driver_resume(struct amdgpu_device *adev);
+int amdgpu_fence_emit(struct amdgpu_ring *ring, void *owner,
+		      struct amdgpu_fence **fence);
+void amdgpu_fence_process(struct amdgpu_ring *ring);
+int amdgpu_fence_wait_next(struct amdgpu_ring *ring);
+int amdgpu_fence_wait_empty(struct amdgpu_ring *ring);
+unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring);
+
+bool amdgpu_fence_need_sync(struct amdgpu_fence *fence,
+			    struct amdgpu_ring *ring);
+void amdgpu_fence_note_sync(struct amdgpu_fence *fence,
+			    struct amdgpu_ring *ring);
+
+/*
+ * TTM.
+ */
+struct amdgpu_mman {
+	struct ttm_bo_global_ref        bo_global_ref;
+	struct drm_global_reference	mem_global_ref;
+	struct ttm_bo_device		bdev;
+	bool				mem_global_referenced;
+	bool				initialized;
+
+#if defined(CONFIG_DEBUG_FS)
+	struct dentry			*vram;
+	struct dentry			*gtt;
+#endif
+
+	/* buffer handling */
+	const struct amdgpu_buffer_funcs	*buffer_funcs;
+	struct amdgpu_ring			*buffer_funcs_ring;
+};
+
+int amdgpu_copy_buffer(struct amdgpu_ring *ring,
+		       uint64_t src_offset,
+		       uint64_t dst_offset,
+		       uint32_t byte_count,
+		       struct reservation_object *resv,
+		       struct fence **fence);
+int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma);
+
+struct amdgpu_bo_list_entry {
+	struct amdgpu_bo		*robj;
+	struct ttm_validate_buffer	tv;
+	struct amdgpu_bo_va		*bo_va;
+	unsigned			prefered_domains;
+	unsigned			allowed_domains;
+	uint32_t			priority;
+};
+
+struct amdgpu_bo_va_mapping {
+	struct list_head		list;
+	struct interval_tree_node	it;
+	uint64_t			offset;
+	uint32_t			flags;
+};
+
+/* bo virtual addresses in a specific vm */
+struct amdgpu_bo_va {
+	struct mutex		        mutex;
+	/* protected by bo being reserved */
+	struct list_head		bo_list;
+	struct fence		        *last_pt_update;
+	unsigned			ref_count;
+
+	/* protected by vm mutex and spinlock */
+	struct list_head		vm_status;
+
+	/* mappings for this bo_va */
+	struct list_head		invalids;
+	struct list_head		valids;
+
+	/* constant after initialization */
+	struct amdgpu_vm		*vm;
+	struct amdgpu_bo		*bo;
+};
+
+#define AMDGPU_GEM_DOMAIN_MAX		0x3
+
+struct amdgpu_bo {
+	/* Protected by gem.mutex */
+	struct list_head		list;
+	/* Protected by tbo.reserved */
+	u32				initial_domain;
+	struct ttm_place		placements[AMDGPU_GEM_DOMAIN_MAX + 1];
+	struct ttm_placement		placement;
+	struct ttm_buffer_object	tbo;
+	struct ttm_bo_kmap_obj		kmap;
+	u64				flags;
+	unsigned			pin_count;
+	void				*kptr;
+	u64				tiling_flags;
+	u64				metadata_flags;
+	void				*metadata;
+	u32				metadata_size;
+	/* list of all virtual address to which this bo
+	 * is associated to
+	 */
+	struct list_head		va;
+	/* Constant after initialization */
+	struct amdgpu_device		*adev;
+	struct drm_gem_object		gem_base;
+	struct amdgpu_bo		*parent;
+
+	struct ttm_bo_kmap_obj		dma_buf_vmap;
+	pid_t				pid;
+	struct amdgpu_mn		*mn;
+	struct list_head		mn_list;
+};
+#define gem_to_amdgpu_bo(gobj) container_of((gobj), struct amdgpu_bo, gem_base)
+
+void amdgpu_gem_object_free(struct drm_gem_object *obj);
+int amdgpu_gem_object_open(struct drm_gem_object *obj,
+				struct drm_file *file_priv);
+void amdgpu_gem_object_close(struct drm_gem_object *obj,
+				struct drm_file *file_priv);
+unsigned long amdgpu_gem_timeout(uint64_t timeout_ns);
+struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj);
+struct drm_gem_object *amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
+							struct dma_buf_attachment *attach,
+							struct sg_table *sg);
+struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
+					struct drm_gem_object *gobj,
+					int flags);
+int amdgpu_gem_prime_pin(struct drm_gem_object *obj);
+void amdgpu_gem_prime_unpin(struct drm_gem_object *obj);
+struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *);
+void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj);
+void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr);
+int amdgpu_gem_debugfs_init(struct amdgpu_device *adev);
+
+/* sub-allocation manager, it has to be protected by another lock.
+ * By conception this is an helper for other part of the driver
+ * like the indirect buffer or semaphore, which both have their
+ * locking.
+ *
+ * Principe is simple, we keep a list of sub allocation in offset
+ * order (first entry has offset == 0, last entry has the highest
+ * offset).
+ *
+ * When allocating new object we first check if there is room at
+ * the end total_size - (last_object_offset + last_object_size) >=
+ * alloc_size. If so we allocate new object there.
+ *
+ * When there is not enough room at the end, we start waiting for
+ * each sub object until we reach object_offset+object_size >=
+ * alloc_size, this object then become the sub object we return.
+ *
+ * Alignment can't be bigger than page size.
+ *
+ * Hole are not considered for allocation to keep things simple.
+ * Assumption is that there won't be hole (all object on same
+ * alignment).
+ */
+struct amdgpu_sa_manager {
+	wait_queue_head_t	wq;
+	struct amdgpu_bo	*bo;
+	struct list_head	*hole;
+	struct list_head	flist[AMDGPU_MAX_RINGS];
+	struct list_head	olist;
+	unsigned		size;
+	uint64_t		gpu_addr;
+	void			*cpu_ptr;
+	uint32_t		domain;
+	uint32_t		align;
+};
+
+/* sub-allocation buffer */
+struct amdgpu_sa_bo {
+	struct list_head		olist;
+	struct list_head		flist;
+	struct amdgpu_sa_manager	*manager;
+	unsigned			soffset;
+	unsigned			eoffset;
+	struct fence		        *fence;
+};
+
+/*
+ * GEM objects.
+ */
+struct amdgpu_gem {
+	struct mutex		mutex;
+	struct list_head	objects;
+};
+
+int amdgpu_gem_init(struct amdgpu_device *adev);
+void amdgpu_gem_fini(struct amdgpu_device *adev);
+int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
+				int alignment, u32 initial_domain,
+				u64 flags, bool kernel,
+				struct drm_gem_object **obj);
+
+int amdgpu_mode_dumb_create(struct drm_file *file_priv,
+			    struct drm_device *dev,
+			    struct drm_mode_create_dumb *args);
+int amdgpu_mode_dumb_mmap(struct drm_file *filp,
+			  struct drm_device *dev,
+			  uint32_t handle, uint64_t *offset_p);
+
+/*
+ * Semaphores.
+ */
+struct amdgpu_semaphore {
+	struct amdgpu_sa_bo	*sa_bo;
+	signed			waiters;
+	uint64_t		gpu_addr;
+};
+
+int amdgpu_semaphore_create(struct amdgpu_device *adev,
+			    struct amdgpu_semaphore **semaphore);
+bool amdgpu_semaphore_emit_signal(struct amdgpu_ring *ring,
+				  struct amdgpu_semaphore *semaphore);
+bool amdgpu_semaphore_emit_wait(struct amdgpu_ring *ring,
+				struct amdgpu_semaphore *semaphore);
+void amdgpu_semaphore_free(struct amdgpu_device *adev,
+			   struct amdgpu_semaphore **semaphore,
+			   struct fence *fence);
+
+/*
+ * Synchronization
+ */
+struct amdgpu_sync {
+	struct amdgpu_semaphore *semaphores[AMDGPU_NUM_SYNCS];
+	struct fence		*sync_to[AMDGPU_MAX_RINGS];
+	DECLARE_HASHTABLE(fences, 4);
+	struct fence	        *last_vm_update;
+};
+
+void amdgpu_sync_create(struct amdgpu_sync *sync);
+int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		      struct fence *f);
+int amdgpu_sync_resv(struct amdgpu_device *adev,
+		     struct amdgpu_sync *sync,
+		     struct reservation_object *resv,
+		     void *owner);
+int amdgpu_sync_rings(struct amdgpu_sync *sync,
+		      struct amdgpu_ring *ring);
+struct fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync);
+int amdgpu_sync_wait(struct amdgpu_sync *sync);
+void amdgpu_sync_free(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		      struct fence *fence);
+
+/*
+ * GART structures, functions & helpers
+ */
+struct amdgpu_mc;
+
+#define AMDGPU_GPU_PAGE_SIZE 4096
+#define AMDGPU_GPU_PAGE_MASK (AMDGPU_GPU_PAGE_SIZE - 1)
+#define AMDGPU_GPU_PAGE_SHIFT 12
+#define AMDGPU_GPU_PAGE_ALIGN(a) (((a) + AMDGPU_GPU_PAGE_MASK) & ~AMDGPU_GPU_PAGE_MASK)
+
+struct amdgpu_gart {
+	dma_addr_t			table_addr;
+	struct amdgpu_bo		*robj;
+	void				*ptr;
+	unsigned			num_gpu_pages;
+	unsigned			num_cpu_pages;
+	unsigned			table_size;
+	struct page			**pages;
+	dma_addr_t			*pages_addr;
+	bool				ready;
+	const struct amdgpu_gart_funcs *gart_funcs;
+};
+
+int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev);
+void amdgpu_gart_table_ram_free(struct amdgpu_device *adev);
+int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev);
+void amdgpu_gart_table_vram_free(struct amdgpu_device *adev);
+int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev);
+void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev);
+int amdgpu_gart_init(struct amdgpu_device *adev);
+void amdgpu_gart_fini(struct amdgpu_device *adev);
+void amdgpu_gart_unbind(struct amdgpu_device *adev, unsigned offset,
+			int pages);
+int amdgpu_gart_bind(struct amdgpu_device *adev, unsigned offset,
+		     int pages, struct page **pagelist,
+		     dma_addr_t *dma_addr, uint32_t flags);
+
+/*
+ * GPU MC structures, functions & helpers
+ */
+struct amdgpu_mc {
+	resource_size_t		aper_size;
+	resource_size_t		aper_base;
+	resource_size_t		agp_base;
+	/* for some chips with <= 32MB we need to lie
+	 * about vram size near mc fb location */
+	u64			mc_vram_size;
+	u64			visible_vram_size;
+	u64			gtt_size;
+	u64			gtt_start;
+	u64			gtt_end;
+	u64			vram_start;
+	u64			vram_end;
+	unsigned		vram_width;
+	u64			real_vram_size;
+	int			vram_mtrr;
+	u64                     gtt_base_align;
+	u64                     mc_mask;
+	const struct firmware   *fw;	/* MC firmware */
+	uint32_t                fw_version;
+	struct amdgpu_irq_src	vm_fault;
+	uint32_t		vram_type;
+};
+
+/*
+ * GPU doorbell structures, functions & helpers
+ */
+typedef enum _AMDGPU_DOORBELL_ASSIGNMENT
+{
+	AMDGPU_DOORBELL_KIQ                     = 0x000,
+	AMDGPU_DOORBELL_HIQ                     = 0x001,
+	AMDGPU_DOORBELL_DIQ                     = 0x002,
+	AMDGPU_DOORBELL_MEC_RING0               = 0x010,
+	AMDGPU_DOORBELL_MEC_RING1               = 0x011,
+	AMDGPU_DOORBELL_MEC_RING2               = 0x012,
+	AMDGPU_DOORBELL_MEC_RING3               = 0x013,
+	AMDGPU_DOORBELL_MEC_RING4               = 0x014,
+	AMDGPU_DOORBELL_MEC_RING5               = 0x015,
+	AMDGPU_DOORBELL_MEC_RING6               = 0x016,
+	AMDGPU_DOORBELL_MEC_RING7               = 0x017,
+	AMDGPU_DOORBELL_GFX_RING0               = 0x020,
+	AMDGPU_DOORBELL_sDMA_ENGINE0            = 0x1E0,
+	AMDGPU_DOORBELL_sDMA_ENGINE1            = 0x1E1,
+	AMDGPU_DOORBELL_IH                      = 0x1E8,
+	AMDGPU_DOORBELL_MAX_ASSIGNMENT          = 0x3FF,
+	AMDGPU_DOORBELL_INVALID                 = 0xFFFF
+} AMDGPU_DOORBELL_ASSIGNMENT;
+
+struct amdgpu_doorbell {
+	/* doorbell mmio */
+	resource_size_t		base;
+	resource_size_t		size;
+	u32 __iomem		*ptr;
+	u32			num_doorbells;	/* Number of doorbells actually reserved for amdgpu. */
+};
+
+void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
+				phys_addr_t *aperture_base,
+				size_t *aperture_size,
+				size_t *start_offset);
+
+/*
+ * IRQS.
+ */
+
+struct amdgpu_flip_work {
+	struct work_struct		flip_work;
+	struct work_struct		unpin_work;
+	struct amdgpu_device		*adev;
+	int				crtc_id;
+	uint64_t			base;
+	struct drm_pending_vblank_event *event;
+	struct amdgpu_bo		*old_rbo;
+	struct fence			*excl;
+	unsigned			shared_count;
+	struct fence			**shared;
+};
+
+
+/*
+ * CP & rings.
+ */
+
+struct amdgpu_ib {
+	struct amdgpu_sa_bo		*sa_bo;
+	uint32_t			length_dw;
+	uint64_t			gpu_addr;
+	uint32_t			*ptr;
+	struct amdgpu_ring		*ring;
+	struct amdgpu_fence		*fence;
+	struct amdgpu_user_fence        *user;
+	struct amdgpu_vm		*vm;
+	struct amdgpu_ctx		*ctx;
+	struct amdgpu_sync		sync;
+	uint32_t			gds_base, gds_size;
+	uint32_t			gws_base, gws_size;
+	uint32_t			oa_base, oa_size;
+	uint32_t			flags;
+	/* resulting sequence number */
+	uint64_t			sequence;
+};
+
+enum amdgpu_ring_type {
+	AMDGPU_RING_TYPE_GFX,
+	AMDGPU_RING_TYPE_COMPUTE,
+	AMDGPU_RING_TYPE_SDMA,
+	AMDGPU_RING_TYPE_UVD,
+	AMDGPU_RING_TYPE_VCE
+};
+
+extern struct amd_sched_backend_ops amdgpu_sched_ops;
+
+int amdgpu_sched_ib_submit_kernel_helper(struct amdgpu_device *adev,
+					 struct amdgpu_ring *ring,
+					 struct amdgpu_ib *ibs,
+					 unsigned num_ibs,
+					 int (*free_job)(struct amdgpu_job *),
+					 void *owner,
+					 struct fence **fence);
+
+struct amdgpu_ring {
+	struct amdgpu_device		*adev;
+	const struct amdgpu_ring_funcs	*funcs;
+	struct amdgpu_fence_driver	fence_drv;
+	struct amd_gpu_scheduler 	sched;
+
+	spinlock_t              fence_lock;
+	struct mutex		*ring_lock;
+	struct amdgpu_bo	*ring_obj;
+	volatile uint32_t	*ring;
+	unsigned		rptr_offs;
+	u64			next_rptr_gpu_addr;
+	volatile u32		*next_rptr_cpu_addr;
+	unsigned		wptr;
+	unsigned		wptr_old;
+	unsigned		ring_size;
+	unsigned		ring_free_dw;
+	int			count_dw;
+	uint64_t		gpu_addr;
+	uint32_t		align_mask;
+	uint32_t		ptr_mask;
+	bool			ready;
+	u32			nop;
+	u32			idx;
+	u64			last_semaphore_signal_addr;
+	u64			last_semaphore_wait_addr;
+	u32			me;
+	u32			pipe;
+	u32			queue;
+	struct amdgpu_bo	*mqd_obj;
+	u32			doorbell_index;
+	bool			use_doorbell;
+	unsigned		wptr_offs;
+	unsigned		next_rptr_offs;
+	unsigned		fence_offs;
+	struct amdgpu_ctx	*current_ctx;
+	enum amdgpu_ring_type	type;
+	char			name[16];
+	bool                    is_pte_ring;
+};
+
+/*
+ * VM
+ */
+
+/* maximum number of VMIDs */
+#define AMDGPU_NUM_VM	16
+
+/* number of entries in page table */
+#define AMDGPU_VM_PTE_COUNT (1 << amdgpu_vm_block_size)
+
+/* PTBs (Page Table Blocks) need to be aligned to 32K */
+#define AMDGPU_VM_PTB_ALIGN_SIZE   32768
+#define AMDGPU_VM_PTB_ALIGN_MASK (AMDGPU_VM_PTB_ALIGN_SIZE - 1)
+#define AMDGPU_VM_PTB_ALIGN(a) (((a) + AMDGPU_VM_PTB_ALIGN_MASK) & ~AMDGPU_VM_PTB_ALIGN_MASK)
+
+#define AMDGPU_PTE_VALID	(1 << 0)
+#define AMDGPU_PTE_SYSTEM	(1 << 1)
+#define AMDGPU_PTE_SNOOPED	(1 << 2)
+
+/* VI only */
+#define AMDGPU_PTE_EXECUTABLE	(1 << 4)
+
+#define AMDGPU_PTE_READABLE	(1 << 5)
+#define AMDGPU_PTE_WRITEABLE	(1 << 6)
+
+/* PTE (Page Table Entry) fragment field for different page sizes */
+#define AMDGPU_PTE_FRAG_4KB	(0 << 7)
+#define AMDGPU_PTE_FRAG_64KB	(4 << 7)
+#define AMDGPU_LOG2_PAGES_PER_FRAG 4
+
+/* How to programm VM fault handling */
+#define AMDGPU_VM_FAULT_STOP_NEVER	0
+#define AMDGPU_VM_FAULT_STOP_FIRST	1
+#define AMDGPU_VM_FAULT_STOP_ALWAYS	2
+
+struct amdgpu_vm_pt {
+	struct amdgpu_bo	*bo;
+	uint64_t		addr;
+};
+
+struct amdgpu_vm_id {
+	unsigned		id;
+	uint64_t		pd_gpu_addr;
+	/* last flushed PD/PT update */
+	struct fence	        *flushed_updates;
+};
+
+struct amdgpu_vm {
+	struct rb_root		va;
+
+	/* protecting invalidated */
+	spinlock_t		status_lock;
+
+	/* BOs moved, but not yet updated in the PT */
+	struct list_head	invalidated;
+
+	/* BOs cleared in the PT because of a move */
+	struct list_head	cleared;
+
+	/* BO mappings freed, but not yet updated in the PT */
+	struct list_head	freed;
+
+	/* contains the page directory */
+	struct amdgpu_bo	*page_directory;
+	unsigned		max_pde_used;
+	struct fence		*page_directory_fence;
+
+	/* array of page tables, one for each page directory entry */
+	struct amdgpu_vm_pt	*page_tables;
+
+	/* for id and flush management per ring */
+	struct amdgpu_vm_id	ids[AMDGPU_MAX_RINGS];
+	/* for interval tree */
+	spinlock_t		it_lock;
+	/* protecting freed */
+	spinlock_t		freed_lock;
+};
+
+struct amdgpu_vm_manager {
+	struct {
+		struct fence	*active;
+		atomic_long_t	owner;
+	} ids[AMDGPU_NUM_VM];
+
+	uint32_t				max_pfn;
+	/* number of VMIDs */
+	unsigned				nvm;
+	/* vram base address for page table entry  */
+	u64					vram_base_offset;
+	/* is vm enabled? */
+	bool					enabled;
+	/* vm pte handling */
+	const struct amdgpu_vm_pte_funcs        *vm_pte_funcs;
+	struct amdgpu_ring                      *vm_pte_funcs_ring;
+};
+
+void amdgpu_vm_manager_fini(struct amdgpu_device *adev);
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev,
+					       struct amdgpu_vm *vm,
+					       struct list_head *head);
+int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		      struct amdgpu_sync *sync);
+void amdgpu_vm_flush(struct amdgpu_ring *ring,
+		     struct amdgpu_vm *vm,
+		     struct fence *updates);
+void amdgpu_vm_fence(struct amdgpu_device *adev,
+		     struct amdgpu_vm *vm,
+		     struct fence *fence);
+uint64_t amdgpu_vm_map_gart(struct amdgpu_device *adev, uint64_t addr);
+int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm);
+int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
+			  struct amdgpu_vm *vm);
+int amdgpu_vm_clear_invalids(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			     struct amdgpu_sync *sync);
+int amdgpu_vm_bo_update(struct amdgpu_device *adev,
+			struct amdgpu_bo_va *bo_va,
+			struct ttm_mem_reg *mem);
+void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
+			     struct amdgpu_bo *bo);
+struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
+				       struct amdgpu_bo *bo);
+struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
+				      struct amdgpu_vm *vm,
+				      struct amdgpu_bo *bo);
+int amdgpu_vm_bo_map(struct amdgpu_device *adev,
+		     struct amdgpu_bo_va *bo_va,
+		     uint64_t addr, uint64_t offset,
+		     uint64_t size, uint32_t flags);
+int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
+		       struct amdgpu_bo_va *bo_va,
+		       uint64_t addr);
+void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
+		      struct amdgpu_bo_va *bo_va);
+int amdgpu_vm_free_job(struct amdgpu_job *job);
+
+/*
+ * context related structures
+ */
+
+#define AMDGPU_CTX_MAX_CS_PENDING	16
+
+struct amdgpu_ctx_ring {
+	uint64_t		sequence;
+	struct fence		*fences[AMDGPU_CTX_MAX_CS_PENDING];
+	struct amd_sched_entity	entity;
+};
+
+struct amdgpu_ctx {
+	struct kref		refcount;
+	struct amdgpu_device    *adev;
+	unsigned		reset_counter;
+	spinlock_t		ring_lock;
+	struct amdgpu_ctx_ring	rings[AMDGPU_MAX_RINGS];
+};
+
+struct amdgpu_ctx_mgr {
+	struct amdgpu_device	*adev;
+	struct mutex		lock;
+	/* protected by lock */
+	struct idr		ctx_handles;
+};
+
+int amdgpu_ctx_init(struct amdgpu_device *adev, bool kernel,
+		    struct amdgpu_ctx *ctx);
+void amdgpu_ctx_fini(struct amdgpu_ctx *ctx);
+
+struct amdgpu_ctx *amdgpu_ctx_get(struct amdgpu_fpriv *fpriv, uint32_t id);
+int amdgpu_ctx_put(struct amdgpu_ctx *ctx);
+
+uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring,
+			      struct fence *fence);
+struct fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
+				   struct amdgpu_ring *ring, uint64_t seq);
+
+int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
+		     struct drm_file *filp);
+
+void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr);
+void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
+
+/*
+ * file private structure
+ */
+
+struct amdgpu_fpriv {
+	struct amdgpu_vm	vm;
+	struct mutex		bo_list_lock;
+	struct idr		bo_list_handles;
+	struct amdgpu_ctx_mgr	ctx_mgr;
+};
+
+/*
+ * residency list
+ */
+
+struct amdgpu_bo_list {
+	struct mutex lock;
+	struct amdgpu_bo *gds_obj;
+	struct amdgpu_bo *gws_obj;
+	struct amdgpu_bo *oa_obj;
+	bool has_userptr;
+	unsigned num_entries;
+	struct amdgpu_bo_list_entry *array;
+};
+
+struct amdgpu_bo_list *
+amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id);
+void amdgpu_bo_list_put(struct amdgpu_bo_list *list);
+void amdgpu_bo_list_free(struct amdgpu_bo_list *list);
+
+/*
+ * GFX stuff
+ */
+#include "clearstate_defs.h"
+
+struct amdgpu_rlc {
+	/* for power gating */
+	struct amdgpu_bo	*save_restore_obj;
+	uint64_t		save_restore_gpu_addr;
+	volatile uint32_t	*sr_ptr;
+	const u32               *reg_list;
+	u32                     reg_list_size;
+	/* for clear state */
+	struct amdgpu_bo	*clear_state_obj;
+	uint64_t		clear_state_gpu_addr;
+	volatile uint32_t	*cs_ptr;
+	const struct cs_section_def   *cs_data;
+	u32                     clear_state_size;
+	/* for cp tables */
+	struct amdgpu_bo	*cp_table_obj;
+	uint64_t		cp_table_gpu_addr;
+	volatile uint32_t	*cp_table_ptr;
+	u32                     cp_table_size;
+};
+
+struct amdgpu_mec {
+	struct amdgpu_bo	*hpd_eop_obj;
+	u64			hpd_eop_gpu_addr;
+	u32 num_pipe;
+	u32 num_mec;
+	u32 num_queue;
+};
+
+/*
+ * GPU scratch registers structures, functions & helpers
+ */
+struct amdgpu_scratch {
+	unsigned		num_reg;
+	uint32_t                reg_base;
+	bool			free[32];
+	uint32_t		reg[32];
+};
+
+/*
+ * GFX configurations
+ */
+struct amdgpu_gca_config {
+	unsigned max_shader_engines;
+	unsigned max_tile_pipes;
+	unsigned max_cu_per_sh;
+	unsigned max_sh_per_se;
+	unsigned max_backends_per_se;
+	unsigned max_texture_channel_caches;
+	unsigned max_gprs;
+	unsigned max_gs_threads;
+	unsigned max_hw_contexts;
+	unsigned sc_prim_fifo_size_frontend;
+	unsigned sc_prim_fifo_size_backend;
+	unsigned sc_hiz_tile_fifo_size;
+	unsigned sc_earlyz_tile_fifo_size;
+
+	unsigned num_tile_pipes;
+	unsigned backend_enable_mask;
+	unsigned mem_max_burst_length_bytes;
+	unsigned mem_row_size_in_kb;
+	unsigned shader_engine_tile_size;
+	unsigned num_gpus;
+	unsigned multi_gpu_tile_size;
+	unsigned mc_arb_ramcfg;
+	unsigned gb_addr_config;
+
+	uint32_t tile_mode_array[32];
+	uint32_t macrotile_mode_array[16];
+};
+
+struct amdgpu_gfx {
+	struct mutex			gpu_clock_mutex;
+	struct amdgpu_gca_config	config;
+	struct amdgpu_rlc		rlc;
+	struct amdgpu_mec		mec;
+	struct amdgpu_scratch		scratch;
+	const struct firmware		*me_fw;	/* ME firmware */
+	uint32_t			me_fw_version;
+	const struct firmware		*pfp_fw; /* PFP firmware */
+	uint32_t			pfp_fw_version;
+	const struct firmware		*ce_fw;	/* CE firmware */
+	uint32_t			ce_fw_version;
+	const struct firmware		*rlc_fw; /* RLC firmware */
+	uint32_t			rlc_fw_version;
+	const struct firmware		*mec_fw; /* MEC firmware */
+	uint32_t			mec_fw_version;
+	const struct firmware		*mec2_fw; /* MEC2 firmware */
+	uint32_t			mec2_fw_version;
+	uint32_t			me_feature_version;
+	uint32_t			ce_feature_version;
+	uint32_t			pfp_feature_version;
+	uint32_t			rlc_feature_version;
+	uint32_t			mec_feature_version;
+	uint32_t			mec2_feature_version;
+	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
+	unsigned			num_gfx_rings;
+	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS];
+	unsigned			num_compute_rings;
+	struct amdgpu_irq_src		eop_irq;
+	struct amdgpu_irq_src		priv_reg_irq;
+	struct amdgpu_irq_src		priv_inst_irq;
+	/* gfx status */
+	uint32_t gfx_current_status;
+	/* ce ram size*/
+	unsigned ce_ram_size;
+};
+
+int amdgpu_ib_get(struct amdgpu_ring *ring, struct amdgpu_vm *vm,
+		  unsigned size, struct amdgpu_ib *ib);
+void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib);
+int amdgpu_ib_schedule(struct amdgpu_device *adev, unsigned num_ibs,
+		       struct amdgpu_ib *ib, void *owner);
+int amdgpu_ib_pool_init(struct amdgpu_device *adev);
+void amdgpu_ib_pool_fini(struct amdgpu_device *adev);
+int amdgpu_ib_ring_tests(struct amdgpu_device *adev);
+/* Ring access between begin & end cannot sleep */
+void amdgpu_ring_free_size(struct amdgpu_ring *ring);
+int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
+int amdgpu_ring_lock(struct amdgpu_ring *ring, unsigned ndw);
+void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
+void amdgpu_ring_commit(struct amdgpu_ring *ring);
+void amdgpu_ring_unlock_commit(struct amdgpu_ring *ring);
+void amdgpu_ring_undo(struct amdgpu_ring *ring);
+void amdgpu_ring_unlock_undo(struct amdgpu_ring *ring);
+unsigned amdgpu_ring_backup(struct amdgpu_ring *ring,
+			    uint32_t **data);
+int amdgpu_ring_restore(struct amdgpu_ring *ring,
+			unsigned size, uint32_t *data);
+int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
+		     unsigned ring_size, u32 nop, u32 align_mask,
+		     struct amdgpu_irq_src *irq_src, unsigned irq_type,
+		     enum amdgpu_ring_type ring_type);
+void amdgpu_ring_fini(struct amdgpu_ring *ring);
+struct amdgpu_ring *amdgpu_ring_from_fence(struct fence *f);
+
+/*
+ * CS.
+ */
+struct amdgpu_cs_chunk {
+	uint32_t		chunk_id;
+	uint32_t		length_dw;
+	uint32_t		*kdata;
+	void __user		*user_ptr;
+};
+
+struct amdgpu_cs_parser {
+	struct amdgpu_device	*adev;
+	struct drm_file		*filp;
+	struct amdgpu_ctx	*ctx;
+	struct amdgpu_bo_list *bo_list;
+	/* chunks */
+	unsigned		nchunks;
+	struct amdgpu_cs_chunk	*chunks;
+	/* relocations */
+	struct amdgpu_bo_list_entry	*vm_bos;
+	struct list_head	validated;
+	struct fence		*fence;
+
+	struct amdgpu_ib	*ibs;
+	uint32_t		num_ibs;
+
+	struct ww_acquire_ctx	ticket;
+
+	/* user fence */
+	struct amdgpu_user_fence	uf;
+	struct amdgpu_bo_list_entry	uf_entry;
+};
+
+struct amdgpu_job {
+	struct amd_sched_job    base;
+	struct amdgpu_device	*adev;
+	struct amdgpu_ib	*ibs;
+	uint32_t		num_ibs;
+	void			*owner;
+	struct amdgpu_user_fence uf;
+	int (*free_job)(struct amdgpu_job *job);
+};
+#define to_amdgpu_job(sched_job)		\
+		container_of((sched_job), struct amdgpu_job, base)
+
+static inline u32 amdgpu_get_ib_value(struct amdgpu_cs_parser *p, uint32_t ib_idx, int idx)
+{
+	return p->ibs[ib_idx].ptr[idx];
+}
+
+/*
+ * Writeback
+ */
+#define AMDGPU_MAX_WB 1024	/* Reserve at most 1024 WB slots for amdgpu-owned rings. */
+
+struct amdgpu_wb {
+	struct amdgpu_bo	*wb_obj;
+	volatile uint32_t	*wb;
+	uint64_t		gpu_addr;
+	u32			num_wb;	/* Number of wb slots actually reserved for amdgpu. */
+	unsigned long		used[DIV_ROUND_UP(AMDGPU_MAX_WB, BITS_PER_LONG)];
+};
+
+int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb);
+void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb);
+
+/**
+ * struct amdgpu_pm - power management datas
+ * It keeps track of various data needed to take powermanagement decision.
+ */
+
+enum amdgpu_pm_state_type {
+	/* not used for dpm */
+	POWER_STATE_TYPE_DEFAULT,
+	POWER_STATE_TYPE_POWERSAVE,
+	/* user selectable states */
+	POWER_STATE_TYPE_BATTERY,
+	POWER_STATE_TYPE_BALANCED,
+	POWER_STATE_TYPE_PERFORMANCE,
+	/* internal states */
+	POWER_STATE_TYPE_INTERNAL_UVD,
+	POWER_STATE_TYPE_INTERNAL_UVD_SD,
+	POWER_STATE_TYPE_INTERNAL_UVD_HD,
+	POWER_STATE_TYPE_INTERNAL_UVD_HD2,
+	POWER_STATE_TYPE_INTERNAL_UVD_MVC,
+	POWER_STATE_TYPE_INTERNAL_BOOT,
+	POWER_STATE_TYPE_INTERNAL_THERMAL,
+	POWER_STATE_TYPE_INTERNAL_ACPI,
+	POWER_STATE_TYPE_INTERNAL_ULV,
+	POWER_STATE_TYPE_INTERNAL_3DPERF,
+};
+
+enum amdgpu_int_thermal_type {
+	THERMAL_TYPE_NONE,
+	THERMAL_TYPE_EXTERNAL,
+	THERMAL_TYPE_EXTERNAL_GPIO,
+	THERMAL_TYPE_RV6XX,
+	THERMAL_TYPE_RV770,
+	THERMAL_TYPE_ADT7473_WITH_INTERNAL,
+	THERMAL_TYPE_EVERGREEN,
+	THERMAL_TYPE_SUMO,
+	THERMAL_TYPE_NI,
+	THERMAL_TYPE_SI,
+	THERMAL_TYPE_EMC2103_WITH_INTERNAL,
+	THERMAL_TYPE_CI,
+	THERMAL_TYPE_KV,
+};
+
+enum amdgpu_dpm_auto_throttle_src {
+	AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL,
+	AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL
+};
+
+enum amdgpu_dpm_event_src {
+	AMDGPU_DPM_EVENT_SRC_ANALOG = 0,
+	AMDGPU_DPM_EVENT_SRC_EXTERNAL = 1,
+	AMDGPU_DPM_EVENT_SRC_DIGITAL = 2,
+	AMDGPU_DPM_EVENT_SRC_ANALOG_OR_EXTERNAL = 3,
+	AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL = 4
+};
+
+#define AMDGPU_MAX_VCE_LEVELS 6
+
+enum amdgpu_vce_level {
+	AMDGPU_VCE_LEVEL_AC_ALL = 0,     /* AC, All cases */
+	AMDGPU_VCE_LEVEL_DC_EE = 1,      /* DC, entropy encoding */
+	AMDGPU_VCE_LEVEL_DC_LL_LOW = 2,  /* DC, low latency queue, res <= 720 */
+	AMDGPU_VCE_LEVEL_DC_LL_HIGH = 3, /* DC, low latency queue, 1080 >= res > 720 */
+	AMDGPU_VCE_LEVEL_DC_GP_LOW = 4,  /* DC, general purpose queue, res <= 720 */
+	AMDGPU_VCE_LEVEL_DC_GP_HIGH = 5, /* DC, general purpose queue, 1080 >= res > 720 */
+};
+
+struct amdgpu_ps {
+	u32 caps; /* vbios flags */
+	u32 class; /* vbios flags */
+	u32 class2; /* vbios flags */
+	/* UVD clocks */
+	u32 vclk;
+	u32 dclk;
+	/* VCE clocks */
+	u32 evclk;
+	u32 ecclk;
+	bool vce_active;
+	enum amdgpu_vce_level vce_level;
+	/* asic priv */
+	void *ps_priv;
+};
+
+struct amdgpu_dpm_thermal {
+	/* thermal interrupt work */
+	struct work_struct work;
+	/* low temperature threshold */
+	int                min_temp;
+	/* high temperature threshold */
+	int                max_temp;
+	/* was last interrupt low to high or high to low */
+	bool               high_to_low;
+	/* interrupt source */
+	struct amdgpu_irq_src	irq;
+};
+
+enum amdgpu_clk_action
+{
+	AMDGPU_SCLK_UP = 1,
+	AMDGPU_SCLK_DOWN
+};
+
+struct amdgpu_blacklist_clocks
+{
+	u32 sclk;
+	u32 mclk;
+	enum amdgpu_clk_action action;
+};
+
+struct amdgpu_clock_and_voltage_limits {
+	u32 sclk;
+	u32 mclk;
+	u16 vddc;
+	u16 vddci;
+};
+
+struct amdgpu_clock_array {
+	u32 count;
+	u32 *values;
+};
+
+struct amdgpu_clock_voltage_dependency_entry {
+	u32 clk;
+	u16 v;
+};
+
+struct amdgpu_clock_voltage_dependency_table {
+	u32 count;
+	struct amdgpu_clock_voltage_dependency_entry *entries;
+};
+
+union amdgpu_cac_leakage_entry {
+	struct {
+		u16 vddc;
+		u32 leakage;
+	};
+	struct {
+		u16 vddc1;
+		u16 vddc2;
+		u16 vddc3;
+	};
+};
+
+struct amdgpu_cac_leakage_table {
+	u32 count;
+	union amdgpu_cac_leakage_entry *entries;
+};
+
+struct amdgpu_phase_shedding_limits_entry {
+	u16 voltage;
+	u32 sclk;
+	u32 mclk;
+};
+
+struct amdgpu_phase_shedding_limits_table {
+	u32 count;
+	struct amdgpu_phase_shedding_limits_entry *entries;
+};
+
+struct amdgpu_uvd_clock_voltage_dependency_entry {
+	u32 vclk;
+	u32 dclk;
+	u16 v;
+};
+
+struct amdgpu_uvd_clock_voltage_dependency_table {
+	u8 count;
+	struct amdgpu_uvd_clock_voltage_dependency_entry *entries;
+};
+
+struct amdgpu_vce_clock_voltage_dependency_entry {
+	u32 ecclk;
+	u32 evclk;
+	u16 v;
+};
+
+struct amdgpu_vce_clock_voltage_dependency_table {
+	u8 count;
+	struct amdgpu_vce_clock_voltage_dependency_entry *entries;
+};
+
+struct amdgpu_ppm_table {
+	u8 ppm_design;
+	u16 cpu_core_number;
+	u32 platform_tdp;
+	u32 small_ac_platform_tdp;
+	u32 platform_tdc;
+	u32 small_ac_platform_tdc;
+	u32 apu_tdp;
+	u32 dgpu_tdp;
+	u32 dgpu_ulv_power;
+	u32 tj_max;
+};
+
+struct amdgpu_cac_tdp_table {
+	u16 tdp;
+	u16 configurable_tdp;
+	u16 tdc;
+	u16 battery_power_limit;
+	u16 small_power_limit;
+	u16 low_cac_leakage;
+	u16 high_cac_leakage;
+	u16 maximum_power_delivery_limit;
+};
+
+struct amdgpu_dpm_dynamic_state {
+	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table vddci_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table mvdd_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table vddc_dependency_on_dispclk;
+	struct amdgpu_uvd_clock_voltage_dependency_table uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table vce_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table samu_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table acp_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table vddgfx_dependency_on_sclk;
+	struct amdgpu_clock_array valid_sclk_values;
+	struct amdgpu_clock_array valid_mclk_values;
+	struct amdgpu_clock_and_voltage_limits max_clock_voltage_on_dc;
+	struct amdgpu_clock_and_voltage_limits max_clock_voltage_on_ac;
+	u32 mclk_sclk_ratio;
+	u32 sclk_mclk_delta;
+	u16 vddc_vddci_delta;
+	u16 min_vddc_for_pcie_gen2;
+	struct amdgpu_cac_leakage_table cac_leakage_table;
+	struct amdgpu_phase_shedding_limits_table phase_shedding_limits_table;
+	struct amdgpu_ppm_table *ppm_table;
+	struct amdgpu_cac_tdp_table *cac_tdp_table;
+};
+
+struct amdgpu_dpm_fan {
+	u16 t_min;
+	u16 t_med;
+	u16 t_high;
+	u16 pwm_min;
+	u16 pwm_med;
+	u16 pwm_high;
+	u8 t_hyst;
+	u32 cycle_delay;
+	u16 t_max;
+	u8 control_mode;
+	u16 default_max_fan_pwm;
+	u16 default_fan_output_sensitivity;
+	u16 fan_output_sensitivity;
+	bool ucode_fan_control;
+};
+
+enum amdgpu_pcie_gen {
+	AMDGPU_PCIE_GEN1 = 0,
+	AMDGPU_PCIE_GEN2 = 1,
+	AMDGPU_PCIE_GEN3 = 2,
+	AMDGPU_PCIE_GEN_INVALID = 0xffff
+};
+
+enum amdgpu_dpm_forced_level {
+	AMDGPU_DPM_FORCED_LEVEL_AUTO = 0,
+	AMDGPU_DPM_FORCED_LEVEL_LOW = 1,
+	AMDGPU_DPM_FORCED_LEVEL_HIGH = 2,
+};
+
+struct amdgpu_vce_state {
+	/* vce clocks */
+	u32 evclk;
+	u32 ecclk;
+	/* gpu clocks */
+	u32 sclk;
+	u32 mclk;
+	u8 clk_idx;
+	u8 pstate;
+};
+
+struct amdgpu_dpm_funcs {
+	int (*get_temperature)(struct amdgpu_device *adev);
+	int (*pre_set_power_state)(struct amdgpu_device *adev);
+	int (*set_power_state)(struct amdgpu_device *adev);
+	void (*post_set_power_state)(struct amdgpu_device *adev);
+	void (*display_configuration_changed)(struct amdgpu_device *adev);
+	u32 (*get_sclk)(struct amdgpu_device *adev, bool low);
+	u32 (*get_mclk)(struct amdgpu_device *adev, bool low);
+	void (*print_power_state)(struct amdgpu_device *adev, struct amdgpu_ps *ps);
+	void (*debugfs_print_current_performance_level)(struct amdgpu_device *adev, struct seq_file *m);
+	int (*force_performance_level)(struct amdgpu_device *adev, enum amdgpu_dpm_forced_level level);
+	bool (*vblank_too_short)(struct amdgpu_device *adev);
+	void (*powergate_uvd)(struct amdgpu_device *adev, bool gate);
+	void (*powergate_vce)(struct amdgpu_device *adev, bool gate);
+	void (*enable_bapm)(struct amdgpu_device *adev, bool enable);
+	void (*set_fan_control_mode)(struct amdgpu_device *adev, u32 mode);
+	u32 (*get_fan_control_mode)(struct amdgpu_device *adev);
+	int (*set_fan_speed_percent)(struct amdgpu_device *adev, u32 speed);
+	int (*get_fan_speed_percent)(struct amdgpu_device *adev, u32 *speed);
+};
+
+struct amdgpu_dpm {
+	struct amdgpu_ps        *ps;
+	/* number of valid power states */
+	int                     num_ps;
+	/* current power state that is active */
+	struct amdgpu_ps        *current_ps;
+	/* requested power state */
+	struct amdgpu_ps        *requested_ps;
+	/* boot up power state */
+	struct amdgpu_ps        *boot_ps;
+	/* default uvd power state */
+	struct amdgpu_ps        *uvd_ps;
+	/* vce requirements */
+	struct amdgpu_vce_state vce_states[AMDGPU_MAX_VCE_LEVELS];
+	enum amdgpu_vce_level vce_level;
+	enum amdgpu_pm_state_type state;
+	enum amdgpu_pm_state_type user_state;
+	u32                     platform_caps;
+	u32                     voltage_response_time;
+	u32                     backbias_response_time;
+	void                    *priv;
+	u32			new_active_crtcs;
+	int			new_active_crtc_count;
+	u32			current_active_crtcs;
+	int			current_active_crtc_count;
+	struct amdgpu_dpm_dynamic_state dyn_state;
+	struct amdgpu_dpm_fan fan;
+	u32 tdp_limit;
+	u32 near_tdp_limit;
+	u32 near_tdp_limit_adjusted;
+	u32 sq_ramping_threshold;
+	u32 cac_leakage;
+	u16 tdp_od_limit;
+	u32 tdp_adjustment;
+	u16 load_line_slope;
+	bool power_control;
+	bool ac_power;
+	/* special states active */
+	bool                    thermal_active;
+	bool                    uvd_active;
+	bool                    vce_active;
+	/* thermal handling */
+	struct amdgpu_dpm_thermal thermal;
+	/* forced levels */
+	enum amdgpu_dpm_forced_level forced_level;
+};
+
+struct amdgpu_pm {
+	struct mutex		mutex;
+	u32                     current_sclk;
+	u32                     current_mclk;
+	u32                     default_sclk;
+	u32                     default_mclk;
+	struct amdgpu_i2c_chan *i2c_bus;
+	/* internal thermal controller on rv6xx+ */
+	enum amdgpu_int_thermal_type int_thermal_type;
+	struct device	        *int_hwmon_dev;
+	/* fan control parameters */
+	bool                    no_fan;
+	u8                      fan_pulses_per_revolution;
+	u8                      fan_min_rpm;
+	u8                      fan_max_rpm;
+	/* dpm */
+	bool                    dpm_enabled;
+	bool                    sysfs_initialized;
+	struct amdgpu_dpm       dpm;
+	const struct firmware	*fw;	/* SMC firmware */
+	uint32_t                fw_version;
+	const struct amdgpu_dpm_funcs *funcs;
+};
+
+/*
+ * UVD
+ */
+#define AMDGPU_MAX_UVD_HANDLES	10
+#define AMDGPU_UVD_STACK_SIZE	(1024*1024)
+#define AMDGPU_UVD_HEAP_SIZE	(1024*1024)
+#define AMDGPU_UVD_FIRMWARE_OFFSET 256
+
+struct amdgpu_uvd {
+	struct amdgpu_bo	*vcpu_bo;
+	void			*cpu_addr;
+	uint64_t		gpu_addr;
+	atomic_t		handles[AMDGPU_MAX_UVD_HANDLES];
+	struct drm_file		*filp[AMDGPU_MAX_UVD_HANDLES];
+	struct delayed_work	idle_work;
+	const struct firmware	*fw;	/* UVD firmware */
+	struct amdgpu_ring	ring;
+	struct amdgpu_irq_src	irq;
+	bool			address_64_bit;
+};
+
+/*
+ * VCE
+ */
+#define AMDGPU_MAX_VCE_HANDLES	16
+#define AMDGPU_VCE_FIRMWARE_OFFSET 256
+
+#define AMDGPU_VCE_HARVEST_VCE0 (1 << 0)
+#define AMDGPU_VCE_HARVEST_VCE1 (1 << 1)
+
+struct amdgpu_vce {
+	struct amdgpu_bo	*vcpu_bo;
+	uint64_t		gpu_addr;
+	unsigned		fw_version;
+	unsigned		fb_version;
+	atomic_t		handles[AMDGPU_MAX_VCE_HANDLES];
+	struct drm_file		*filp[AMDGPU_MAX_VCE_HANDLES];
+	uint32_t		img_size[AMDGPU_MAX_VCE_HANDLES];
+	struct delayed_work	idle_work;
+	const struct firmware	*fw;	/* VCE firmware */
+	struct amdgpu_ring	ring[AMDGPU_MAX_VCE_RINGS];
+	struct amdgpu_irq_src	irq;
+	unsigned		harvest_config;
+};
+
+/*
+ * SDMA
+ */
+struct amdgpu_sdma_instance {
+	/* SDMA firmware */
+	const struct firmware	*fw;
+	uint32_t		fw_version;
+	uint32_t		feature_version;
+
+	struct amdgpu_ring	ring;
+	bool			burst_nop;
+};
+
+struct amdgpu_sdma {
+	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
+	struct amdgpu_irq_src	trap_irq;
+	struct amdgpu_irq_src	illegal_inst_irq;
+	int 			num_instances;
+};
+
+/*
+ * Firmware
+ */
+struct amdgpu_firmware {
+	struct amdgpu_firmware_info ucode[AMDGPU_UCODE_ID_MAXIMUM];
+	bool smu_load;
+	struct amdgpu_bo *fw_buf;
+	unsigned int fw_size;
+};
+
+/*
+ * Benchmarking
+ */
+void amdgpu_benchmark(struct amdgpu_device *adev, int test_number);
+
+
+/*
+ * Testing
+ */
+void amdgpu_test_moves(struct amdgpu_device *adev);
+void amdgpu_test_ring_sync(struct amdgpu_device *adev,
+			   struct amdgpu_ring *cpA,
+			   struct amdgpu_ring *cpB);
+void amdgpu_test_syncing(struct amdgpu_device *adev);
+
+/*
+ * MMU Notifier
+ */
+#if defined(CONFIG_MMU_NOTIFIER)
+int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr);
+void amdgpu_mn_unregister(struct amdgpu_bo *bo);
+#else
+static inline int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
+{
+	return -ENODEV;
+}
+static inline void amdgpu_mn_unregister(struct amdgpu_bo *bo) {}
+#endif
+
+/*
+ * Debugfs
+ */
+struct amdgpu_debugfs {
+	struct drm_info_list	*files;
+	unsigned		num_files;
+};
+
+int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
+			     struct drm_info_list *files,
+			     unsigned nfiles);
+int amdgpu_debugfs_fence_init(struct amdgpu_device *adev);
+
+#if defined(CONFIG_DEBUG_FS)
+int amdgpu_debugfs_init(struct drm_minor *minor);
+void amdgpu_debugfs_cleanup(struct drm_minor *minor);
+#endif
+
+/*
+ * amdgpu smumgr functions
+ */
+struct amdgpu_smumgr_funcs {
+	int (*check_fw_load_finish)(struct amdgpu_device *adev, uint32_t fwtype);
+	int (*request_smu_load_fw)(struct amdgpu_device *adev);
+	int (*request_smu_specific_fw)(struct amdgpu_device *adev, uint32_t fwtype);
+};
+
+/*
+ * amdgpu smumgr
+ */
+struct amdgpu_smumgr {
+	struct amdgpu_bo *toc_buf;
+	struct amdgpu_bo *smu_buf;
+	/* asic priv smu data */
+	void *priv;
+	spinlock_t smu_lock;
+	/* smumgr functions */
+	const struct amdgpu_smumgr_funcs *smumgr_funcs;
+	/* ucode loading complete flag */
+	uint32_t fw_flags;
+};
+
+/*
+ * ASIC specific register table accessible by UMD
+ */
+struct amdgpu_allowed_register_entry {
+	uint32_t reg_offset;
+	bool untouched;
+	bool grbm_indexed;
+};
+
+struct amdgpu_cu_info {
+	uint32_t number; /* total active CU number */
+	uint32_t ao_cu_mask;
+	uint32_t bitmap[4][4];
+};
+
+
+/*
+ * ASIC specific functions.
+ */
+struct amdgpu_asic_funcs {
+	bool (*read_disabled_bios)(struct amdgpu_device *adev);
+	int (*read_register)(struct amdgpu_device *adev, u32 se_num,
+			     u32 sh_num, u32 reg_offset, u32 *value);
+	void (*set_vga_state)(struct amdgpu_device *adev, bool state);
+	int (*reset)(struct amdgpu_device *adev);
+	/* wait for mc_idle */
+	int (*wait_for_mc_idle)(struct amdgpu_device *adev);
+	/* get the reference clock */
+	u32 (*get_xclk)(struct amdgpu_device *adev);
+	/* get the gpu clock counter */
+	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
+	int (*get_cu_info)(struct amdgpu_device *adev, struct amdgpu_cu_info *info);
+	/* MM block clocks */
+	int (*set_uvd_clocks)(struct amdgpu_device *adev, u32 vclk, u32 dclk);
+	int (*set_vce_clocks)(struct amdgpu_device *adev, u32 evclk, u32 ecclk);
+};
+
+/*
+ * IOCTL.
+ */
+int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp);
+int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp);
+
+int amdgpu_gem_info_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp);
+int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp);
+int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp);
+int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp);
+int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
+
+int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp);
+
+/* VRAM scratch page for HDP bug, default vram page */
+struct amdgpu_vram_scratch {
+	struct amdgpu_bo		*robj;
+	volatile uint32_t		*ptr;
+	u64				gpu_addr;
+};
+
+/*
+ * ACPI
+ */
+struct amdgpu_atif_notification_cfg {
+	bool enabled;
+	int command_code;
+};
+
+struct amdgpu_atif_notifications {
+	bool display_switch;
+	bool expansion_mode_change;
+	bool thermal_state;
+	bool forced_power_state;
+	bool system_power_state;
+	bool display_conf_change;
+	bool px_gfx_switch;
+	bool brightness_change;
+	bool dgpu_display_event;
+};
+
+struct amdgpu_atif_functions {
+	bool system_params;
+	bool sbios_requests;
+	bool select_active_disp;
+	bool lid_state;
+	bool get_tv_standard;
+	bool set_tv_standard;
+	bool get_panel_expansion_mode;
+	bool set_panel_expansion_mode;
+	bool temperature_change;
+	bool graphics_device_types;
+};
+
+struct amdgpu_atif {
+	struct amdgpu_atif_notifications notifications;
+	struct amdgpu_atif_functions functions;
+	struct amdgpu_atif_notification_cfg notification_cfg;
+	struct amdgpu_encoder *encoder_for_bl;
+};
+
+struct amdgpu_atcs_functions {
+	bool get_ext_state;
+	bool pcie_perf_req;
+	bool pcie_dev_rdy;
+	bool pcie_bus_width;
+};
+
+struct amdgpu_atcs {
+	struct amdgpu_atcs_functions functions;
+};
+
+/*
+ * CGS
+ */
+void *amdgpu_cgs_create_device(struct amdgpu_device *adev);
+void amdgpu_cgs_destroy_device(void *cgs_device);
+
+
+/*
+ * Core structure, functions and helpers.
+ */
+typedef uint32_t (*amdgpu_rreg_t)(struct amdgpu_device*, uint32_t);
+typedef void (*amdgpu_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t);
+
+typedef uint32_t (*amdgpu_block_rreg_t)(struct amdgpu_device*, uint32_t, uint32_t);
+typedef void (*amdgpu_block_wreg_t)(struct amdgpu_device*, uint32_t, uint32_t, uint32_t);
+
+struct amdgpu_ip_block_status {
+	bool valid;
+	bool sw;
+	bool hw;
+};
+
+struct amdgpu_device {
+	struct device			*dev;
+	struct drm_device		*ddev;
+	struct pci_dev			*pdev;
+
+	/* ASIC */
+	enum amd_asic_type		asic_type;
+	uint32_t			family;
+	uint32_t			rev_id;
+	uint32_t			external_rev_id;
+	unsigned long			flags;
+	int				usec_timeout;
+	const struct amdgpu_asic_funcs	*asic_funcs;
+	bool				shutdown;
+	bool				suspend;
+	bool				need_dma32;
+	bool				accel_working;
+	struct work_struct 		reset_work;
+	struct notifier_block		acpi_nb;
+	struct amdgpu_i2c_chan		*i2c_bus[AMDGPU_MAX_I2C_BUS];
+	struct amdgpu_debugfs		debugfs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
+	unsigned 			debugfs_count;
+#if defined(CONFIG_DEBUG_FS)
+	struct dentry			*debugfs_regs;
+#endif
+	struct amdgpu_atif		atif;
+	struct amdgpu_atcs		atcs;
+	struct mutex			srbm_mutex;
+	/* GRBM index mutex. Protects concurrent access to GRBM index */
+	struct mutex                    grbm_idx_mutex;
+	struct dev_pm_domain		vga_pm_domain;
+	bool				have_disp_power_ref;
+
+	/* BIOS */
+	uint8_t				*bios;
+	bool				is_atom_bios;
+	uint16_t			bios_header_start;
+	struct amdgpu_bo		*stollen_vga_memory;
+	uint32_t			bios_scratch[AMDGPU_BIOS_NUM_SCRATCH];
+
+	/* Register/doorbell mmio */
+	resource_size_t			rmmio_base;
+	resource_size_t			rmmio_size;
+	void __iomem			*rmmio;
+	/* protects concurrent MM_INDEX/DATA based register access */
+	spinlock_t mmio_idx_lock;
+	/* protects concurrent SMC based register access */
+	spinlock_t smc_idx_lock;
+	amdgpu_rreg_t			smc_rreg;
+	amdgpu_wreg_t			smc_wreg;
+	/* protects concurrent PCIE register access */
+	spinlock_t pcie_idx_lock;
+	amdgpu_rreg_t			pcie_rreg;
+	amdgpu_wreg_t			pcie_wreg;
+	/* protects concurrent UVD register access */
+	spinlock_t uvd_ctx_idx_lock;
+	amdgpu_rreg_t			uvd_ctx_rreg;
+	amdgpu_wreg_t			uvd_ctx_wreg;
+	/* protects concurrent DIDT register access */
+	spinlock_t didt_idx_lock;
+	amdgpu_rreg_t			didt_rreg;
+	amdgpu_wreg_t			didt_wreg;
+	/* protects concurrent ENDPOINT (audio) register access */
+	spinlock_t audio_endpt_idx_lock;
+	amdgpu_block_rreg_t		audio_endpt_rreg;
+	amdgpu_block_wreg_t		audio_endpt_wreg;
+	void __iomem                    *rio_mem;
+	resource_size_t			rio_mem_size;
+	struct amdgpu_doorbell		doorbell;
+
+	/* clock/pll info */
+	struct amdgpu_clock            clock;
+
+	/* MC */
+	struct amdgpu_mc		mc;
+	struct amdgpu_gart		gart;
+	struct amdgpu_dummy_page	dummy_page;
+	struct amdgpu_vm_manager	vm_manager;
+
+	/* memory management */
+	struct amdgpu_mman		mman;
+	struct amdgpu_gem		gem;
+	struct amdgpu_vram_scratch	vram_scratch;
+	struct amdgpu_wb		wb;
+	atomic64_t			vram_usage;
+	atomic64_t			vram_vis_usage;
+	atomic64_t			gtt_usage;
+	atomic64_t			num_bytes_moved;
+	atomic_t			gpu_reset_counter;
+
+	/* display */
+	struct amdgpu_mode_info		mode_info;
+	struct work_struct		hotplug_work;
+	struct amdgpu_irq_src		crtc_irq;
+	struct amdgpu_irq_src		pageflip_irq;
+	struct amdgpu_irq_src		hpd_irq;
+
+	/* rings */
+	unsigned			fence_context;
+	struct mutex			ring_lock;
+	unsigned			num_rings;
+	struct amdgpu_ring		*rings[AMDGPU_MAX_RINGS];
+	bool				ib_pool_ready;
+	struct amdgpu_sa_manager	ring_tmp_bo;
+
+	/* interrupts */
+	struct amdgpu_irq		irq;
+
+	/* dpm */
+	struct amdgpu_pm		pm;
+	u32				cg_flags;
+	u32				pg_flags;
+
+	/* amdgpu smumgr */
+	struct amdgpu_smumgr smu;
+
+	/* gfx */
+	struct amdgpu_gfx		gfx;
+
+	/* sdma */
+	struct amdgpu_sdma		sdma;
+
+	/* uvd */
+	bool				has_uvd;
+	struct amdgpu_uvd		uvd;
+
+	/* vce */
+	struct amdgpu_vce		vce;
+
+	/* firmwares */
+	struct amdgpu_firmware		firmware;
+
+	/* GDS */
+	struct amdgpu_gds		gds;
+
+	const struct amdgpu_ip_block_version *ip_blocks;
+	int				num_ip_blocks;
+	struct amdgpu_ip_block_status	*ip_block_status;
+	struct mutex	mn_lock;
+	DECLARE_HASHTABLE(mn_hash, 7);
+
+	/* tracking pinned memory */
+	u64 vram_pin_size;
+	u64 gart_pin_size;
+
+	/* amdkfd interface */
+	struct kfd_dev          *kfd;
+
+	/* kernel conext for IB submission */
+	struct amdgpu_ctx	kernel_ctx;
+};
+
+bool amdgpu_device_is_px(struct drm_device *dev);
+int amdgpu_device_init(struct amdgpu_device *adev,
+		       struct drm_device *ddev,
+		       struct pci_dev *pdev,
+		       uint32_t flags);
+void amdgpu_device_fini(struct amdgpu_device *adev);
+int amdgpu_gpu_wait_for_idle(struct amdgpu_device *adev);
+
+uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
+			bool always_indirect);
+void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
+		    bool always_indirect);
+u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg);
+void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v);
+
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index);
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
+
+/*
+ * Cast helper
+ */
+extern const struct fence_ops amdgpu_fence_ops;
+static inline struct amdgpu_fence *to_amdgpu_fence(struct fence *f)
+{
+	struct amdgpu_fence *__f = container_of(f, struct amdgpu_fence, base);
+
+	if (__f->base.ops == &amdgpu_fence_ops)
+		return __f;
+
+	return NULL;
+}
+
+/*
+ * Registers read & write functions.
+ */
+#define RREG32(reg) amdgpu_mm_rreg(adev, (reg), false)
+#define RREG32_IDX(reg) amdgpu_mm_rreg(adev, (reg), true)
+#define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", amdgpu_mm_rreg(adev, (reg), false))
+#define WREG32(reg, v) amdgpu_mm_wreg(adev, (reg), (v), false)
+#define WREG32_IDX(reg, v) amdgpu_mm_wreg(adev, (reg), (v), true)
+#define REG_SET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
+#define REG_GET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
+#define RREG32_PCIE(reg) adev->pcie_rreg(adev, (reg))
+#define WREG32_PCIE(reg, v) adev->pcie_wreg(adev, (reg), (v))
+#define RREG32_SMC(reg) adev->smc_rreg(adev, (reg))
+#define WREG32_SMC(reg, v) adev->smc_wreg(adev, (reg), (v))
+#define RREG32_UVD_CTX(reg) adev->uvd_ctx_rreg(adev, (reg))
+#define WREG32_UVD_CTX(reg, v) adev->uvd_ctx_wreg(adev, (reg), (v))
+#define RREG32_DIDT(reg) adev->didt_rreg(adev, (reg))
+#define WREG32_DIDT(reg, v) adev->didt_wreg(adev, (reg), (v))
+#define RREG32_AUDIO_ENDPT(block, reg) adev->audio_endpt_rreg(adev, (block), (reg))
+#define WREG32_AUDIO_ENDPT(block, reg, v) adev->audio_endpt_wreg(adev, (block), (reg), (v))
+#define WREG32_P(reg, val, mask)				\
+	do {							\
+		uint32_t tmp_ = RREG32(reg);			\
+		tmp_ &= (mask);					\
+		tmp_ |= ((val) & ~(mask));			\
+		WREG32(reg, tmp_);				\
+	} while (0)
+#define WREG32_AND(reg, and) WREG32_P(reg, 0, and)
+#define WREG32_OR(reg, or) WREG32_P(reg, or, ~(or))
+#define WREG32_PLL_P(reg, val, mask)				\
+	do {							\
+		uint32_t tmp_ = RREG32_PLL(reg);		\
+		tmp_ &= (mask);					\
+		tmp_ |= ((val) & ~(mask));			\
+		WREG32_PLL(reg, tmp_);				\
+	} while (0)
+#define DREG32_SYS(sqf, adev, reg) seq_printf((sqf), #reg " : 0x%08X\n", amdgpu_mm_rreg((adev), (reg), false))
+#define RREG32_IO(reg) amdgpu_io_rreg(adev, (reg))
+#define WREG32_IO(reg, v) amdgpu_io_wreg(adev, (reg), (v))
+
+#define RDOORBELL32(index) amdgpu_mm_rdoorbell(adev, (index))
+#define WDOORBELL32(index, v) amdgpu_mm_wdoorbell(adev, (index), (v))
+
+#define REG_FIELD_SHIFT(reg, field) reg##__##field##__SHIFT
+#define REG_FIELD_MASK(reg, field) reg##__##field##_MASK
+
+#define REG_SET_FIELD(orig_val, reg, field, field_val)			\
+	(((orig_val) & ~REG_FIELD_MASK(reg, field)) |			\
+	 (REG_FIELD_MASK(reg, field) & ((field_val) << REG_FIELD_SHIFT(reg, field))))
+
+#define REG_GET_FIELD(value, reg, field)				\
+	(((value) & REG_FIELD_MASK(reg, field)) >> REG_FIELD_SHIFT(reg, field))
+
+/*
+ * BIOS helpers.
+ */
+#define RBIOS8(i) (adev->bios[i])
+#define RBIOS16(i) (RBIOS8(i) | (RBIOS8((i)+1) << 8))
+#define RBIOS32(i) ((RBIOS16(i)) | (RBIOS16((i)+2) << 16))
+
+/*
+ * RING helpers.
+ */
+static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
+{
+	if (ring->count_dw <= 0)
+		DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
+	ring->ring[ring->wptr++] = v;
+	ring->wptr &= ring->ptr_mask;
+	ring->count_dw--;
+	ring->ring_free_dw--;
+}
+
+static inline struct amdgpu_sdma_instance *
+amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		if (&adev->sdma.instance[i].ring == ring)
+			break;
+
+	if (i < AMDGPU_MAX_SDMA_INSTANCES)
+		return &adev->sdma.instance[i];
+	else
+		return NULL;
+}
+
+/*
+ * ASICs macro.
+ */
+#define amdgpu_asic_set_vga_state(adev, state) (adev)->asic_funcs->set_vga_state((adev), (state))
+#define amdgpu_asic_reset(adev) (adev)->asic_funcs->reset((adev))
+#define amdgpu_asic_wait_for_mc_idle(adev) (adev)->asic_funcs->wait_for_mc_idle((adev))
+#define amdgpu_asic_get_xclk(adev) (adev)->asic_funcs->get_xclk((adev))
+#define amdgpu_asic_set_uvd_clocks(adev, v, d) (adev)->asic_funcs->set_uvd_clocks((adev), (v), (d))
+#define amdgpu_asic_set_vce_clocks(adev, ev, ec) (adev)->asic_funcs->set_vce_clocks((adev), (ev), (ec))
+#define amdgpu_asic_get_gpu_clock_counter(adev) (adev)->asic_funcs->get_gpu_clock_counter((adev))
+#define amdgpu_asic_read_disabled_bios(adev) (adev)->asic_funcs->read_disabled_bios((adev))
+#define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v)))
+#define amdgpu_asic_get_cu_info(adev, info) (adev)->asic_funcs->get_cu_info((adev), (info))
+#define amdgpu_gart_flush_gpu_tlb(adev, vmid) (adev)->gart.gart_funcs->flush_gpu_tlb((adev), (vmid))
+#define amdgpu_gart_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gart.gart_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags))
+#define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count)))
+#define amdgpu_vm_write_pte(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (addr), (count), (incr), (flags)))
+#define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
+#define amdgpu_vm_pad_ib(adev, ib) ((adev)->vm_manager.vm_pte_funcs->pad_ib((ib)))
+#define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
+#define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
+#define amdgpu_ring_test_ib(r) (r)->funcs->test_ib((r))
+#define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
+#define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
+#define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
+#define amdgpu_ring_emit_ib(r, ib) (r)->funcs->emit_ib((r), (ib))
+#define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr))
+#define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
+#define amdgpu_ring_emit_semaphore(r, semaphore, emit_wait) (r)->funcs->emit_semaphore((r), (semaphore), (emit_wait))
+#define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
+#define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
+#define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
+#define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
+#define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
+#define amdgpu_display_set_vga_render_state(adev, r) (adev)->mode_info.funcs->set_vga_render_state((adev), (r))
+#define amdgpu_display_vblank_get_counter(adev, crtc) (adev)->mode_info.funcs->vblank_get_counter((adev), (crtc))
+#define amdgpu_display_vblank_wait(adev, crtc) (adev)->mode_info.funcs->vblank_wait((adev), (crtc))
+#define amdgpu_display_is_display_hung(adev) (adev)->mode_info.funcs->is_display_hung((adev))
+#define amdgpu_display_backlight_set_level(adev, e, l) (adev)->mode_info.funcs->backlight_set_level((e), (l))
+#define amdgpu_display_backlight_get_level(adev, e) (adev)->mode_info.funcs->backlight_get_level((e))
+#define amdgpu_display_hpd_sense(adev, h) (adev)->mode_info.funcs->hpd_sense((adev), (h))
+#define amdgpu_display_hpd_set_polarity(adev, h) (adev)->mode_info.funcs->hpd_set_polarity((adev), (h))
+#define amdgpu_display_hpd_get_gpio_reg(adev) (adev)->mode_info.funcs->hpd_get_gpio_reg((adev))
+#define amdgpu_display_bandwidth_update(adev) (adev)->mode_info.funcs->bandwidth_update((adev))
+#define amdgpu_display_page_flip(adev, crtc, base) (adev)->mode_info.funcs->page_flip((adev), (crtc), (base))
+#define amdgpu_display_page_flip_get_scanoutpos(adev, crtc, vbl, pos) (adev)->mode_info.funcs->page_flip_get_scanoutpos((adev), (crtc), (vbl), (pos))
+#define amdgpu_display_add_encoder(adev, e, s, c) (adev)->mode_info.funcs->add_encoder((adev), (e), (s), (c))
+#define amdgpu_display_add_connector(adev, ci, sd, ct, ib, coi, h, r) (adev)->mode_info.funcs->add_connector((adev), (ci), (sd), (ct), (ib), (coi), (h), (r))
+#define amdgpu_display_stop_mc_access(adev, s) (adev)->mode_info.funcs->stop_mc_access((adev), (s))
+#define amdgpu_display_resume_mc_access(adev, s) (adev)->mode_info.funcs->resume_mc_access((adev), (s))
+#define amdgpu_emit_copy_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_copy_buffer((ib),  (s), (d), (b))
+#define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b))
+#define amdgpu_dpm_get_temperature(adev) (adev)->pm.funcs->get_temperature((adev))
+#define amdgpu_dpm_pre_set_power_state(adev) (adev)->pm.funcs->pre_set_power_state((adev))
+#define amdgpu_dpm_set_power_state(adev) (adev)->pm.funcs->set_power_state((adev))
+#define amdgpu_dpm_post_set_power_state(adev) (adev)->pm.funcs->post_set_power_state((adev))
+#define amdgpu_dpm_display_configuration_changed(adev) (adev)->pm.funcs->display_configuration_changed((adev))
+#define amdgpu_dpm_get_sclk(adev, l) (adev)->pm.funcs->get_sclk((adev), (l))
+#define amdgpu_dpm_get_mclk(adev, l) (adev)->pm.funcs->get_mclk((adev), (l))
+#define amdgpu_dpm_print_power_state(adev, ps) (adev)->pm.funcs->print_power_state((adev), (ps))
+#define amdgpu_dpm_debugfs_print_current_performance_level(adev, m) (adev)->pm.funcs->debugfs_print_current_performance_level((adev), (m))
+#define amdgpu_dpm_force_performance_level(adev, l) (adev)->pm.funcs->force_performance_level((adev), (l))
+#define amdgpu_dpm_vblank_too_short(adev) (adev)->pm.funcs->vblank_too_short((adev))
+#define amdgpu_dpm_powergate_uvd(adev, g) (adev)->pm.funcs->powergate_uvd((adev), (g))
+#define amdgpu_dpm_powergate_vce(adev, g) (adev)->pm.funcs->powergate_vce((adev), (g))
+#define amdgpu_dpm_enable_bapm(adev, e) (adev)->pm.funcs->enable_bapm((adev), (e))
+#define amdgpu_dpm_set_fan_control_mode(adev, m) (adev)->pm.funcs->set_fan_control_mode((adev), (m))
+#define amdgpu_dpm_get_fan_control_mode(adev) (adev)->pm.funcs->get_fan_control_mode((adev))
+#define amdgpu_dpm_set_fan_speed_percent(adev, s) (adev)->pm.funcs->set_fan_speed_percent((adev), (s))
+#define amdgpu_dpm_get_fan_speed_percent(adev, s) (adev)->pm.funcs->get_fan_speed_percent((adev), (s))
+
+#define amdgpu_gds_switch(adev, r, v, d, w, a) (adev)->gds.funcs->patch_gds_switch((r), (v), (d), (w), (a))
+
+/* Common functions */
+int amdgpu_gpu_reset(struct amdgpu_device *adev);
+void amdgpu_pci_config_reset(struct amdgpu_device *adev);
+bool amdgpu_card_posted(struct amdgpu_device *adev);
+void amdgpu_update_display_priority(struct amdgpu_device *adev);
+bool amdgpu_boot_test_post_card(struct amdgpu_device *adev);
+
+int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data);
+int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type,
+		       u32 ip_instance, u32 ring,
+		       struct amdgpu_ring **out_ring);
+void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *rbo, u32 domain);
+bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
+int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
+				     uint32_t flags);
+bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm);
+bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
+				  unsigned long end);
+bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
+uint32_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
+				 struct ttm_mem_reg *mem);
+void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64 base);
+void amdgpu_gtt_location(struct amdgpu_device *adev, struct amdgpu_mc *mc);
+void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size);
+void amdgpu_program_register_sequence(struct amdgpu_device *adev,
+					     const u32 *registers,
+					     const u32 array_size);
+
+bool amdgpu_device_is_px(struct drm_device *dev);
+/* atpx handler */
+#if defined(CONFIG_VGA_SWITCHEROO)
+void amdgpu_register_atpx_handler(void);
+void amdgpu_unregister_atpx_handler(void);
+#else
+static inline void amdgpu_register_atpx_handler(void) {}
+static inline void amdgpu_unregister_atpx_handler(void) {}
+#endif
+
+/*
+ * KMS
+ */
+extern const struct drm_ioctl_desc amdgpu_ioctls_kms[];
+extern int amdgpu_max_kms_ioctl;
+
+int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags);
+int amdgpu_driver_unload_kms(struct drm_device *dev);
+void amdgpu_driver_lastclose_kms(struct drm_device *dev);
+int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv);
+void amdgpu_driver_postclose_kms(struct drm_device *dev,
+				 struct drm_file *file_priv);
+void amdgpu_driver_preclose_kms(struct drm_device *dev,
+				struct drm_file *file_priv);
+int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon);
+int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon);
+u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe);
+int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe);
+void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe);
+int amdgpu_get_vblank_timestamp_kms(struct drm_device *dev, unsigned int pipe,
+				    int *max_error,
+				    struct timeval *vblank_time,
+				    unsigned flags);
+long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd,
+			     unsigned long arg);
+
+/*
+ * functions used by amdgpu_encoder.c
+ */
+struct amdgpu_afmt_acr {
+	u32 clock;
+
+	int n_32khz;
+	int cts_32khz;
+
+	int n_44_1khz;
+	int cts_44_1khz;
+
+	int n_48khz;
+	int cts_48khz;
+
+};
+
+struct amdgpu_afmt_acr amdgpu_afmt_acr(uint32_t clock);
+
+/* amdgpu_acpi.c */
+#if defined(CONFIG_ACPI)
+int amdgpu_acpi_init(struct amdgpu_device *adev);
+void amdgpu_acpi_fini(struct amdgpu_device *adev);
+bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *adev);
+int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
+						u8 perf_req, bool advertise);
+int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev);
+#else
+static inline int amdgpu_acpi_init(struct amdgpu_device *adev) { return 0; }
+static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { }
+#endif
+
+struct amdgpu_bo_va_mapping *
+amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
+		       uint64_t addr, struct amdgpu_bo **bo);
+
+#include "amdgpu_object.h"
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
new file mode 100644
index 000000000..a142d5ae1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -0,0 +1,767 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+#include <linux/acpi.h>
+#include <linux/slab.h>
+#include <linux/power_supply.h>
+#include <acpi/video.h>
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include "amdgpu.h"
+#include "amdgpu_acpi.h"
+#include "atom.h"
+
+#define ACPI_AC_CLASS           "ac_adapter"
+
+extern void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
+
+struct atif_verify_interface {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u16 version;		/* version */
+	u32 notification_mask;	/* supported notifications mask */
+	u32 function_bits;	/* supported functions bit vector */
+} __packed;
+
+struct atif_system_params {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u32 valid_mask;		/* valid flags mask */
+	u32 flags;		/* flags */
+	u8 command_code;	/* notify command code */
+} __packed;
+
+struct atif_sbios_requests {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u32 pending;		/* pending sbios requests */
+	u8 panel_exp_mode;	/* panel expansion mode */
+	u8 thermal_gfx;		/* thermal state: target gfx controller */
+	u8 thermal_state;	/* thermal state: state id (0: exit state, non-0: state) */
+	u8 forced_power_gfx;	/* forced power state: target gfx controller */
+	u8 forced_power_state;	/* forced power state: state id */
+	u8 system_power_src;	/* system power source */
+	u8 backlight_level;	/* panel backlight level (0-255) */
+} __packed;
+
+#define ATIF_NOTIFY_MASK	0x3
+#define ATIF_NOTIFY_NONE	0
+#define ATIF_NOTIFY_81		1
+#define ATIF_NOTIFY_N		2
+
+struct atcs_verify_interface {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u16 version;		/* version */
+	u32 function_bits;	/* supported functions bit vector */
+} __packed;
+
+#define ATCS_VALID_FLAGS_MASK	0x3
+
+struct atcs_pref_req_input {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u16 client_id;		/* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */
+	u16 valid_flags_mask;	/* valid flags mask */
+	u16 flags;		/* flags */
+	u8 req_type;		/* request type */
+	u8 perf_req;		/* performance request */
+} __packed;
+
+struct atcs_pref_req_output {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u8 ret_val;		/* return value */
+} __packed;
+
+/* Call the ATIF method
+ */
+/**
+ * amdgpu_atif_call - call an ATIF method
+ *
+ * @handle: acpi handle
+ * @function: the ATIF function to execute
+ * @params: ATIF function params
+ *
+ * Executes the requested ATIF function (all asics).
+ * Returns a pointer to the acpi output buffer.
+ */
+static union acpi_object *amdgpu_atif_call(acpi_handle handle, int function,
+		struct acpi_buffer *params)
+{
+	acpi_status status;
+	union acpi_object atif_arg_elements[2];
+	struct acpi_object_list atif_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
+
+	atif_arg.count = 2;
+	atif_arg.pointer = &atif_arg_elements[0];
+
+	atif_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atif_arg_elements[0].integer.value = function;
+
+	if (params) {
+		atif_arg_elements[1].type = ACPI_TYPE_BUFFER;
+		atif_arg_elements[1].buffer.length = params->length;
+		atif_arg_elements[1].buffer.pointer = params->pointer;
+	} else {
+		/* We need a second fake parameter */
+		atif_arg_elements[1].type = ACPI_TYPE_INTEGER;
+		atif_arg_elements[1].integer.value = 0;
+	}
+
+	status = acpi_evaluate_object(handle, "ATIF", &atif_arg, &buffer);
+
+	/* Fail only if calling the method fails and ATIF is supported */
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		DRM_DEBUG_DRIVER("failed to evaluate ATIF got %s\n",
+				 acpi_format_exception(status));
+		kfree(buffer.pointer);
+		return NULL;
+	}
+
+	return buffer.pointer;
+}
+
+/**
+ * amdgpu_atif_parse_notification - parse supported notifications
+ *
+ * @n: supported notifications struct
+ * @mask: supported notifications mask from ATIF
+ *
+ * Use the supported notifications mask from ATIF function
+ * ATIF_FUNCTION_VERIFY_INTERFACE to determine what notifications
+ * are supported (all asics).
+ */
+static void amdgpu_atif_parse_notification(struct amdgpu_atif_notifications *n, u32 mask)
+{
+	n->display_switch = mask & ATIF_DISPLAY_SWITCH_REQUEST_SUPPORTED;
+	n->expansion_mode_change = mask & ATIF_EXPANSION_MODE_CHANGE_REQUEST_SUPPORTED;
+	n->thermal_state = mask & ATIF_THERMAL_STATE_CHANGE_REQUEST_SUPPORTED;
+	n->forced_power_state = mask & ATIF_FORCED_POWER_STATE_CHANGE_REQUEST_SUPPORTED;
+	n->system_power_state = mask & ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST_SUPPORTED;
+	n->display_conf_change = mask & ATIF_DISPLAY_CONF_CHANGE_REQUEST_SUPPORTED;
+	n->px_gfx_switch = mask & ATIF_PX_GFX_SWITCH_REQUEST_SUPPORTED;
+	n->brightness_change = mask & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST_SUPPORTED;
+	n->dgpu_display_event = mask & ATIF_DGPU_DISPLAY_EVENT_SUPPORTED;
+}
+
+/**
+ * amdgpu_atif_parse_functions - parse supported functions
+ *
+ * @f: supported functions struct
+ * @mask: supported functions mask from ATIF
+ *
+ * Use the supported functions mask from ATIF function
+ * ATIF_FUNCTION_VERIFY_INTERFACE to determine what functions
+ * are supported (all asics).
+ */
+static void amdgpu_atif_parse_functions(struct amdgpu_atif_functions *f, u32 mask)
+{
+	f->system_params = mask & ATIF_GET_SYSTEM_PARAMETERS_SUPPORTED;
+	f->sbios_requests = mask & ATIF_GET_SYSTEM_BIOS_REQUESTS_SUPPORTED;
+	f->select_active_disp = mask & ATIF_SELECT_ACTIVE_DISPLAYS_SUPPORTED;
+	f->lid_state = mask & ATIF_GET_LID_STATE_SUPPORTED;
+	f->get_tv_standard = mask & ATIF_GET_TV_STANDARD_FROM_CMOS_SUPPORTED;
+	f->set_tv_standard = mask & ATIF_SET_TV_STANDARD_IN_CMOS_SUPPORTED;
+	f->get_panel_expansion_mode = mask & ATIF_GET_PANEL_EXPANSION_MODE_FROM_CMOS_SUPPORTED;
+	f->set_panel_expansion_mode = mask & ATIF_SET_PANEL_EXPANSION_MODE_IN_CMOS_SUPPORTED;
+	f->temperature_change = mask & ATIF_TEMPERATURE_CHANGE_NOTIFICATION_SUPPORTED;
+	f->graphics_device_types = mask & ATIF_GET_GRAPHICS_DEVICE_TYPES_SUPPORTED;
+}
+
+/**
+ * amdgpu_atif_verify_interface - verify ATIF
+ *
+ * @handle: acpi handle
+ * @atif: amdgpu atif struct
+ *
+ * Execute the ATIF_FUNCTION_VERIFY_INTERFACE ATIF function
+ * to initialize ATIF and determine what features are supported
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_verify_interface(acpi_handle handle,
+		struct amdgpu_atif *atif)
+{
+	union acpi_object *info;
+	struct atif_verify_interface output;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atif_call(handle, ATIF_FUNCTION_VERIFY_INTERFACE, NULL);
+	if (!info)
+		return -EIO;
+
+	memset(&output, 0, sizeof(output));
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 12) {
+		DRM_INFO("ATIF buffer is too small: %zu\n", size);
+		err = -EINVAL;
+		goto out;
+	}
+	size = min(sizeof(output), size);
+
+	memcpy(&output, info->buffer.pointer, size);
+
+	/* TODO: check version? */
+	DRM_DEBUG_DRIVER("ATIF version %u\n", output.version);
+
+	amdgpu_atif_parse_notification(&atif->notifications, output.notification_mask);
+	amdgpu_atif_parse_functions(&atif->functions, output.function_bits);
+
+out:
+	kfree(info);
+	return err;
+}
+
+/**
+ * amdgpu_atif_get_notification_params - determine notify configuration
+ *
+ * @handle: acpi handle
+ * @n: atif notification configuration struct
+ *
+ * Execute the ATIF_FUNCTION_GET_SYSTEM_PARAMETERS ATIF function
+ * to determine if a notifier is used and if so which one
+ * (all asics).  This is either Notify(VGA, 0x81) or Notify(VGA, n)
+ * where n is specified in the result if a notifier is used.
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_get_notification_params(acpi_handle handle,
+		struct amdgpu_atif_notification_cfg *n)
+{
+	union acpi_object *info;
+	struct atif_system_params params;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atif_call(handle, ATIF_FUNCTION_GET_SYSTEM_PARAMETERS, NULL);
+	if (!info) {
+		err = -EIO;
+		goto out;
+	}
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 10) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	memset(&params, 0, sizeof(params));
+	size = min(sizeof(params), size);
+	memcpy(&params, info->buffer.pointer, size);
+
+	DRM_DEBUG_DRIVER("SYSTEM_PARAMS: mask = %#x, flags = %#x\n",
+			params.flags, params.valid_mask);
+	params.flags = params.flags & params.valid_mask;
+
+	if ((params.flags & ATIF_NOTIFY_MASK) == ATIF_NOTIFY_NONE) {
+		n->enabled = false;
+		n->command_code = 0;
+	} else if ((params.flags & ATIF_NOTIFY_MASK) == ATIF_NOTIFY_81) {
+		n->enabled = true;
+		n->command_code = 0x81;
+	} else {
+		if (size < 11) {
+			err = -EINVAL;
+			goto out;
+		}
+		n->enabled = true;
+		n->command_code = params.command_code;
+	}
+
+out:
+	DRM_DEBUG_DRIVER("Notification %s, command code = %#x\n",
+			(n->enabled ? "enabled" : "disabled"),
+			n->command_code);
+	kfree(info);
+	return err;
+}
+
+/**
+ * amdgpu_atif_get_sbios_requests - get requested sbios event
+ *
+ * @handle: acpi handle
+ * @req: atif sbios request struct
+ *
+ * Execute the ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS ATIF function
+ * to determine what requests the sbios is making to the driver
+ * (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atif_get_sbios_requests(acpi_handle handle,
+		struct atif_sbios_requests *req)
+{
+	union acpi_object *info;
+	size_t size;
+	int count = 0;
+
+	info = amdgpu_atif_call(handle, ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS, NULL);
+	if (!info)
+		return -EIO;
+
+	size = *(u16 *)info->buffer.pointer;
+	if (size < 0xd) {
+		count = -EINVAL;
+		goto out;
+	}
+	memset(req, 0, sizeof(*req));
+
+	size = min(sizeof(*req), size);
+	memcpy(req, info->buffer.pointer, size);
+	DRM_DEBUG_DRIVER("SBIOS pending requests: %#x\n", req->pending);
+
+	count = hweight32(req->pending);
+
+out:
+	kfree(info);
+	return count;
+}
+
+/**
+ * amdgpu_atif_handler - handle ATIF notify requests
+ *
+ * @adev: amdgpu_device pointer
+ * @event: atif sbios request struct
+ *
+ * Checks the acpi event and if it matches an atif event,
+ * handles it.
+ * Returns NOTIFY code
+ */
+int amdgpu_atif_handler(struct amdgpu_device *adev,
+			struct acpi_bus_event *event)
+{
+	struct amdgpu_atif *atif = &adev->atif;
+	struct atif_sbios_requests req;
+	acpi_handle handle;
+	int count;
+
+	DRM_DEBUG_DRIVER("event, device_class = %s, type = %#x\n",
+			event->device_class, event->type);
+
+	if (strcmp(event->device_class, ACPI_VIDEO_CLASS) != 0)
+		return NOTIFY_DONE;
+
+	if (!atif->notification_cfg.enabled ||
+	    event->type != atif->notification_cfg.command_code)
+		/* Not our event */
+		return NOTIFY_DONE;
+
+	/* Check pending SBIOS requests */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+	count = amdgpu_atif_get_sbios_requests(handle, &req);
+
+	if (count <= 0)
+		return NOTIFY_DONE;
+
+	DRM_DEBUG_DRIVER("ATIF: %d pending SBIOS requests\n", count);
+
+	if (req.pending & ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST) {
+		struct amdgpu_encoder *enc = atif->encoder_for_bl;
+
+		if (enc) {
+			struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
+
+			DRM_DEBUG_DRIVER("Changing brightness to %d\n",
+					req.backlight_level);
+
+			amdgpu_display_backlight_set_level(adev, enc, req.backlight_level);
+
+#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
+			backlight_force_update(dig->bl_dev,
+					       BACKLIGHT_UPDATE_HOTKEY);
+#endif
+		}
+	}
+	/* TODO: check other events */
+
+	/* We've handled the event, stop the notifier chain. The ACPI interface
+	 * overloads ACPI_VIDEO_NOTIFY_PROBE, we don't want to send that to
+	 * userspace if the event was generated only to signal a SBIOS
+	 * request.
+	 */
+	return NOTIFY_BAD;
+}
+
+/* Call the ATCS method
+ */
+/**
+ * amdgpu_atcs_call - call an ATCS method
+ *
+ * @handle: acpi handle
+ * @function: the ATCS function to execute
+ * @params: ATCS function params
+ *
+ * Executes the requested ATCS function (all asics).
+ * Returns a pointer to the acpi output buffer.
+ */
+static union acpi_object *amdgpu_atcs_call(acpi_handle handle, int function,
+					   struct acpi_buffer *params)
+{
+	acpi_status status;
+	union acpi_object atcs_arg_elements[2];
+	struct acpi_object_list atcs_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
+
+	atcs_arg.count = 2;
+	atcs_arg.pointer = &atcs_arg_elements[0];
+
+	atcs_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atcs_arg_elements[0].integer.value = function;
+
+	if (params) {
+		atcs_arg_elements[1].type = ACPI_TYPE_BUFFER;
+		atcs_arg_elements[1].buffer.length = params->length;
+		atcs_arg_elements[1].buffer.pointer = params->pointer;
+	} else {
+		/* We need a second fake parameter */
+		atcs_arg_elements[1].type = ACPI_TYPE_INTEGER;
+		atcs_arg_elements[1].integer.value = 0;
+	}
+
+	status = acpi_evaluate_object(handle, "ATCS", &atcs_arg, &buffer);
+
+	/* Fail only if calling the method fails and ATIF is supported */
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		DRM_DEBUG_DRIVER("failed to evaluate ATCS got %s\n",
+				 acpi_format_exception(status));
+		kfree(buffer.pointer);
+		return NULL;
+	}
+
+	return buffer.pointer;
+}
+
+/**
+ * amdgpu_atcs_parse_functions - parse supported functions
+ *
+ * @f: supported functions struct
+ * @mask: supported functions mask from ATCS
+ *
+ * Use the supported functions mask from ATCS function
+ * ATCS_FUNCTION_VERIFY_INTERFACE to determine what functions
+ * are supported (all asics).
+ */
+static void amdgpu_atcs_parse_functions(struct amdgpu_atcs_functions *f, u32 mask)
+{
+	f->get_ext_state = mask & ATCS_GET_EXTERNAL_STATE_SUPPORTED;
+	f->pcie_perf_req = mask & ATCS_PCIE_PERFORMANCE_REQUEST_SUPPORTED;
+	f->pcie_dev_rdy = mask & ATCS_PCIE_DEVICE_READY_NOTIFICATION_SUPPORTED;
+	f->pcie_bus_width = mask & ATCS_SET_PCIE_BUS_WIDTH_SUPPORTED;
+}
+
+/**
+ * amdgpu_atcs_verify_interface - verify ATCS
+ *
+ * @handle: acpi handle
+ * @atcs: amdgpu atcs struct
+ *
+ * Execute the ATCS_FUNCTION_VERIFY_INTERFACE ATCS function
+ * to initialize ATCS and determine what features are supported
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+static int amdgpu_atcs_verify_interface(acpi_handle handle,
+					struct amdgpu_atcs *atcs)
+{
+	union acpi_object *info;
+	struct atcs_verify_interface output;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atcs_call(handle, ATCS_FUNCTION_VERIFY_INTERFACE, NULL);
+	if (!info)
+		return -EIO;
+
+	memset(&output, 0, sizeof(output));
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 8) {
+		DRM_INFO("ATCS buffer is too small: %zu\n", size);
+		err = -EINVAL;
+		goto out;
+	}
+	size = min(sizeof(output), size);
+
+	memcpy(&output, info->buffer.pointer, size);
+
+	/* TODO: check version? */
+	DRM_DEBUG_DRIVER("ATCS version %u\n", output.version);
+
+	amdgpu_atcs_parse_functions(&atcs->functions, output.function_bits);
+
+out:
+	kfree(info);
+	return err;
+}
+
+/**
+ * amdgpu_acpi_is_pcie_performance_request_supported
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check if the ATCS pcie_perf_req and pcie_dev_rdy methods
+ * are supported (all asics).
+ * returns true if supported, false if not.
+ */
+bool amdgpu_acpi_is_pcie_performance_request_supported(struct amdgpu_device *adev)
+{
+	struct amdgpu_atcs *atcs = &adev->atcs;
+
+	if (atcs->functions.pcie_perf_req && atcs->functions.pcie_dev_rdy)
+		return true;
+
+	return false;
+}
+
+/**
+ * amdgpu_acpi_pcie_notify_device_ready
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Executes the PCIE_DEVICE_READY_NOTIFICATION method
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_pcie_notify_device_ready(struct amdgpu_device *adev)
+{
+	acpi_handle handle;
+	union acpi_object *info;
+	struct amdgpu_atcs *atcs = &adev->atcs;
+
+	/* Get the device handle */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+	if (!handle)
+		return -EINVAL;
+
+	if (!atcs->functions.pcie_dev_rdy)
+		return -EINVAL;
+
+	info = amdgpu_atcs_call(handle, ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION, NULL);
+	if (!info)
+		return -EIO;
+
+	kfree(info);
+
+	return 0;
+}
+
+/**
+ * amdgpu_acpi_pcie_performance_request
+ *
+ * @adev: amdgpu_device pointer
+ * @perf_req: requested perf level (pcie gen speed)
+ * @advertise: set advertise caps flag if set
+ *
+ * Executes the PCIE_PERFORMANCE_REQUEST method to
+ * change the pcie gen speed (all asics).
+ * returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_pcie_performance_request(struct amdgpu_device *adev,
+					 u8 perf_req, bool advertise)
+{
+	acpi_handle handle;
+	union acpi_object *info;
+	struct amdgpu_atcs *atcs = &adev->atcs;
+	struct atcs_pref_req_input atcs_input;
+	struct atcs_pref_req_output atcs_output;
+	struct acpi_buffer params;
+	size_t size;
+	u32 retry = 3;
+
+	/* Get the device handle */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+	if (!handle)
+		return -EINVAL;
+
+	if (!atcs->functions.pcie_perf_req)
+		return -EINVAL;
+
+	atcs_input.size = sizeof(struct atcs_pref_req_input);
+	/* client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num) */
+	atcs_input.client_id = adev->pdev->devfn | (adev->pdev->bus->number << 8);
+	atcs_input.valid_flags_mask = ATCS_VALID_FLAGS_MASK;
+	atcs_input.flags = ATCS_WAIT_FOR_COMPLETION;
+	if (advertise)
+		atcs_input.flags |= ATCS_ADVERTISE_CAPS;
+	atcs_input.req_type = ATCS_PCIE_LINK_SPEED;
+	atcs_input.perf_req = perf_req;
+
+	params.length = sizeof(struct atcs_pref_req_input);
+	params.pointer = &atcs_input;
+
+	while (retry--) {
+		info = amdgpu_atcs_call(handle, ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST, &params);
+		if (!info)
+			return -EIO;
+
+		memset(&atcs_output, 0, sizeof(atcs_output));
+
+		size = *(u16 *) info->buffer.pointer;
+		if (size < 3) {
+			DRM_INFO("ATCS buffer is too small: %zu\n", size);
+			kfree(info);
+			return -EINVAL;
+		}
+		size = min(sizeof(atcs_output), size);
+
+		memcpy(&atcs_output, info->buffer.pointer, size);
+
+		kfree(info);
+
+		switch (atcs_output.ret_val) {
+		case ATCS_REQUEST_REFUSED:
+		default:
+			return -EINVAL;
+		case ATCS_REQUEST_COMPLETE:
+			return 0;
+		case ATCS_REQUEST_IN_PROGRESS:
+			udelay(10);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_acpi_event - handle notify events
+ *
+ * @nb: notifier block
+ * @val: val
+ * @data: acpi event
+ *
+ * Calls relevant amdgpu functions in response to various
+ * acpi events.
+ * Returns NOTIFY code
+ */
+static int amdgpu_acpi_event(struct notifier_block *nb,
+			     unsigned long val,
+			     void *data)
+{
+	struct amdgpu_device *adev = container_of(nb, struct amdgpu_device, acpi_nb);
+	struct acpi_bus_event *entry = (struct acpi_bus_event *)data;
+
+	if (strcmp(entry->device_class, ACPI_AC_CLASS) == 0) {
+		if (power_supply_is_system_supplied() > 0)
+			DRM_DEBUG_DRIVER("pm: AC\n");
+		else
+			DRM_DEBUG_DRIVER("pm: DC\n");
+
+		amdgpu_pm_acpi_event_handler(adev);
+	}
+
+	/* Check for pending SBIOS requests */
+	return amdgpu_atif_handler(adev, entry);
+}
+
+/* Call all ACPI methods here */
+/**
+ * amdgpu_acpi_init - init driver acpi support
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Verifies the AMD ACPI interfaces and registers with the acpi
+ * notifier chain (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_acpi_init(struct amdgpu_device *adev)
+{
+	acpi_handle handle;
+	struct amdgpu_atif *atif = &adev->atif;
+	struct amdgpu_atcs *atcs = &adev->atcs;
+	int ret;
+
+	/* Get the device handle */
+	handle = ACPI_HANDLE(&adev->pdev->dev);
+
+	if (!adev->bios || !handle)
+		return 0;
+
+	/* Call the ATCS method */
+	ret = amdgpu_atcs_verify_interface(handle, atcs);
+	if (ret) {
+		DRM_DEBUG_DRIVER("Call to ATCS verify_interface failed: %d\n", ret);
+	}
+
+	/* Call the ATIF method */
+	ret = amdgpu_atif_verify_interface(handle, atif);
+	if (ret) {
+		DRM_DEBUG_DRIVER("Call to ATIF verify_interface failed: %d\n", ret);
+		goto out;
+	}
+
+	if (atif->notifications.brightness_change) {
+		struct drm_encoder *tmp;
+
+		/* Find the encoder controlling the brightness */
+		list_for_each_entry(tmp, &adev->ddev->mode_config.encoder_list,
+				head) {
+			struct amdgpu_encoder *enc = to_amdgpu_encoder(tmp);
+
+			if ((enc->devices & (ATOM_DEVICE_LCD_SUPPORT)) &&
+			    enc->enc_priv) {
+				if (adev->is_atom_bios) {
+					struct amdgpu_encoder_atom_dig *dig = enc->enc_priv;
+					if (dig->bl_dev) {
+						atif->encoder_for_bl = enc;
+						break;
+					}
+				}
+			}
+		}
+	}
+
+	if (atif->functions.sbios_requests && !atif->functions.system_params) {
+		/* XXX check this workraround, if sbios request function is
+		 * present we have to see how it's configured in the system
+		 * params
+		 */
+		atif->functions.system_params = true;
+	}
+
+	if (atif->functions.system_params) {
+		ret = amdgpu_atif_get_notification_params(handle,
+				&atif->notification_cfg);
+		if (ret) {
+			DRM_DEBUG_DRIVER("Call to GET_SYSTEM_PARAMS failed: %d\n",
+					ret);
+			/* Disable notification */
+			atif->notification_cfg.enabled = false;
+		}
+	}
+
+out:
+	adev->acpi_nb.notifier_call = amdgpu_acpi_event;
+	register_acpi_notifier(&adev->acpi_nb);
+
+	return ret;
+}
+
+/**
+ * amdgpu_acpi_fini - tear down driver acpi support
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unregisters with the acpi notifier chain (all asics).
+ */
+void amdgpu_acpi_fini(struct amdgpu_device *adev)
+{
+	unregister_acpi_notifier(&adev->acpi_nb);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h
new file mode 100644
index 000000000..01a29c3d7
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.h
@@ -0,0 +1,445 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef AMDGPU_ACPI_H
+#define AMDGPU_ACPI_H
+
+struct amdgpu_device;
+struct acpi_bus_event;
+
+int amdgpu_atif_handler(struct amdgpu_device *adev,
+		struct acpi_bus_event *event);
+
+/* AMD hw uses four ACPI control methods:
+ * 1. ATIF
+ * ARG0: (ACPI_INTEGER) function code
+ * ARG1: (ACPI_BUFFER) parameter buffer, 256 bytes
+ * OUTPUT: (ACPI_BUFFER) output buffer, 256 bytes
+ * ATIF provides an entry point for the gfx driver to interact with the sbios.
+ * The AMD ACPI notification mechanism uses Notify (VGA, 0x81) or a custom
+ * notification. Which notification is used as indicated by the ATIF Control
+ * Method GET_SYSTEM_PARAMETERS. When the driver receives Notify (VGA, 0x81) or
+ * a custom notification it invokes ATIF Control Method GET_SYSTEM_BIOS_REQUESTS
+ * to identify pending System BIOS requests and associated parameters. For
+ * example, if one of the pending requests is DISPLAY_SWITCH_REQUEST, the driver
+ * will perform display device detection and invoke ATIF Control Method
+ * SELECT_ACTIVE_DISPLAYS.
+ *
+ * 2. ATPX
+ * ARG0: (ACPI_INTEGER) function code
+ * ARG1: (ACPI_BUFFER) parameter buffer, 256 bytes
+ * OUTPUT: (ACPI_BUFFER) output buffer, 256 bytes
+ * ATPX methods are used on PowerXpress systems to handle mux switching and
+ * discrete GPU power control.
+ *
+ * 3. ATRM
+ * ARG0: (ACPI_INTEGER) offset of vbios rom data
+ * ARG1: (ACPI_BUFFER) size of the buffer to fill (up to 4K).
+ * OUTPUT: (ACPI_BUFFER) output buffer
+ * ATRM provides an interfacess to access the discrete GPU vbios image on
+ * PowerXpress systems with multiple GPUs.
+ *
+ * 4. ATCS
+ * ARG0: (ACPI_INTEGER) function code
+ * ARG1: (ACPI_BUFFER) parameter buffer, 256 bytes
+ * OUTPUT: (ACPI_BUFFER) output buffer, 256 bytes
+ * ATCS provides an interface to AMD chipset specific functionality.
+ *
+ */
+/* ATIF */
+#define ATIF_FUNCTION_VERIFY_INTERFACE                             0x0
+/* ARG0: ATIF_FUNCTION_VERIFY_INTERFACE
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - version
+ * DWORD - supported notifications mask
+ * DWORD - supported functions bit vector
+ */
+/* Notifications mask */
+#       define ATIF_DISPLAY_SWITCH_REQUEST_SUPPORTED               (1 << 0)
+#       define ATIF_EXPANSION_MODE_CHANGE_REQUEST_SUPPORTED        (1 << 1)
+#       define ATIF_THERMAL_STATE_CHANGE_REQUEST_SUPPORTED         (1 << 2)
+#       define ATIF_FORCED_POWER_STATE_CHANGE_REQUEST_SUPPORTED    (1 << 3)
+#       define ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST_SUPPORTED   (1 << 4)
+#       define ATIF_DISPLAY_CONF_CHANGE_REQUEST_SUPPORTED          (1 << 5)
+#       define ATIF_PX_GFX_SWITCH_REQUEST_SUPPORTED                (1 << 6)
+#       define ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST_SUPPORTED      (1 << 7)
+#       define ATIF_DGPU_DISPLAY_EVENT_SUPPORTED                   (1 << 8)
+/* supported functions vector */
+#       define ATIF_GET_SYSTEM_PARAMETERS_SUPPORTED               (1 << 0)
+#       define ATIF_GET_SYSTEM_BIOS_REQUESTS_SUPPORTED            (1 << 1)
+#       define ATIF_SELECT_ACTIVE_DISPLAYS_SUPPORTED              (1 << 2)
+#       define ATIF_GET_LID_STATE_SUPPORTED                       (1 << 3)
+#       define ATIF_GET_TV_STANDARD_FROM_CMOS_SUPPORTED           (1 << 4)
+#       define ATIF_SET_TV_STANDARD_IN_CMOS_SUPPORTED             (1 << 5)
+#       define ATIF_GET_PANEL_EXPANSION_MODE_FROM_CMOS_SUPPORTED  (1 << 6)
+#       define ATIF_SET_PANEL_EXPANSION_MODE_IN_CMOS_SUPPORTED    (1 << 7)
+#       define ATIF_TEMPERATURE_CHANGE_NOTIFICATION_SUPPORTED     (1 << 12)
+#       define ATIF_GET_GRAPHICS_DEVICE_TYPES_SUPPORTED           (1 << 14)
+#define ATIF_FUNCTION_GET_SYSTEM_PARAMETERS                        0x1
+/* ARG0: ATIF_FUNCTION_GET_SYSTEM_PARAMETERS
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * DWORD - valid flags mask
+ * DWORD - flags
+ *
+ * OR
+ *
+ * WORD  - structure size in bytes (includes size field)
+ * DWORD - valid flags mask
+ * DWORD - flags
+ * BYTE  - notify command code
+ *
+ * flags
+ * bits 1:0:
+ * 0 - Notify(VGA, 0x81) is not used for notification
+ * 1 - Notify(VGA, 0x81) is used for notification
+ * 2 - Notify(VGA, n) is used for notification where
+ * n (0xd0-0xd9) is specified in notify command code.
+ * bit 2:
+ * 1 - lid changes not reported though int10
+ */
+#define ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS                     0x2
+/* ARG0: ATIF_FUNCTION_GET_SYSTEM_BIOS_REQUESTS
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * DWORD - pending sbios requests
+ * BYTE  - panel expansion mode
+ * BYTE  - thermal state: target gfx controller
+ * BYTE  - thermal state: state id (0: exit state, non-0: state)
+ * BYTE  - forced power state: target gfx controller
+ * BYTE  - forced power state: state id
+ * BYTE  - system power source
+ * BYTE  - panel backlight level (0-255)
+ */
+/* pending sbios requests */
+#       define ATIF_DISPLAY_SWITCH_REQUEST                         (1 << 0)
+#       define ATIF_EXPANSION_MODE_CHANGE_REQUEST                  (1 << 1)
+#       define ATIF_THERMAL_STATE_CHANGE_REQUEST                   (1 << 2)
+#       define ATIF_FORCED_POWER_STATE_CHANGE_REQUEST              (1 << 3)
+#       define ATIF_SYSTEM_POWER_SOURCE_CHANGE_REQUEST             (1 << 4)
+#       define ATIF_DISPLAY_CONF_CHANGE_REQUEST                    (1 << 5)
+#       define ATIF_PX_GFX_SWITCH_REQUEST                          (1 << 6)
+#       define ATIF_PANEL_BRIGHTNESS_CHANGE_REQUEST                (1 << 7)
+#       define ATIF_DGPU_DISPLAY_EVENT                             (1 << 8)
+/* panel expansion mode */
+#       define ATIF_PANEL_EXPANSION_DISABLE                        0
+#       define ATIF_PANEL_EXPANSION_FULL                           1
+#       define ATIF_PANEL_EXPANSION_ASPECT                         2
+/* target gfx controller */
+#       define ATIF_TARGET_GFX_SINGLE                              0
+#       define ATIF_TARGET_GFX_PX_IGPU                             1
+#       define ATIF_TARGET_GFX_PX_DGPU                             2
+/* system power source */
+#       define ATIF_POWER_SOURCE_AC                                1
+#       define ATIF_POWER_SOURCE_DC                                2
+#       define ATIF_POWER_SOURCE_RESTRICTED_AC_1                   3
+#       define ATIF_POWER_SOURCE_RESTRICTED_AC_2                   4
+#define ATIF_FUNCTION_SELECT_ACTIVE_DISPLAYS                       0x3
+/* ARG0: ATIF_FUNCTION_SELECT_ACTIVE_DISPLAYS
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - selected displays
+ * WORD  - connected displays
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - selected displays
+ */
+#       define ATIF_LCD1                                           (1 << 0)
+#       define ATIF_CRT1                                           (1 << 1)
+#       define ATIF_TV                                             (1 << 2)
+#       define ATIF_DFP1                                           (1 << 3)
+#       define ATIF_CRT2                                           (1 << 4)
+#       define ATIF_LCD2                                           (1 << 5)
+#       define ATIF_DFP2                                           (1 << 7)
+#       define ATIF_CV                                             (1 << 8)
+#       define ATIF_DFP3                                           (1 << 9)
+#       define ATIF_DFP4                                           (1 << 10)
+#       define ATIF_DFP5                                           (1 << 11)
+#       define ATIF_DFP6                                           (1 << 12)
+#define ATIF_FUNCTION_GET_LID_STATE                                0x4
+/* ARG0: ATIF_FUNCTION_GET_LID_STATE
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - lid state (0: open, 1: closed)
+ *
+ * GET_LID_STATE only works at boot and resume, for general lid
+ * status, use the kernel provided status
+ */
+#define ATIF_FUNCTION_GET_TV_STANDARD_FROM_CMOS                    0x5
+/* ARG0: ATIF_FUNCTION_GET_TV_STANDARD_FROM_CMOS
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - 0
+ * BYTE  - TV standard
+ */
+#       define ATIF_TV_STD_NTSC                                    0
+#       define ATIF_TV_STD_PAL                                     1
+#       define ATIF_TV_STD_PALM                                    2
+#       define ATIF_TV_STD_PAL60                                   3
+#       define ATIF_TV_STD_NTSCJ                                   4
+#       define ATIF_TV_STD_PALCN                                   5
+#       define ATIF_TV_STD_PALN                                    6
+#       define ATIF_TV_STD_SCART_RGB                               9
+#define ATIF_FUNCTION_SET_TV_STANDARD_IN_CMOS                      0x6
+/* ARG0: ATIF_FUNCTION_SET_TV_STANDARD_IN_CMOS
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - 0
+ * BYTE  - TV standard
+ * OUTPUT: none
+ */
+#define ATIF_FUNCTION_GET_PANEL_EXPANSION_MODE_FROM_CMOS           0x7
+/* ARG0: ATIF_FUNCTION_GET_PANEL_EXPANSION_MODE_FROM_CMOS
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - panel expansion mode
+ */
+#define ATIF_FUNCTION_SET_PANEL_EXPANSION_MODE_IN_CMOS             0x8
+/* ARG0: ATIF_FUNCTION_SET_PANEL_EXPANSION_MODE_IN_CMOS
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - panel expansion mode
+ * OUTPUT: none
+ */
+#define ATIF_FUNCTION_TEMPERATURE_CHANGE_NOTIFICATION              0xD
+/* ARG0: ATIF_FUNCTION_TEMPERATURE_CHANGE_NOTIFICATION
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - gfx controller id
+ * BYTE  - current temperature (degress Celsius)
+ * OUTPUT: none
+ */
+#define ATIF_FUNCTION_GET_GRAPHICS_DEVICE_TYPES                    0xF
+/* ARG0: ATIF_FUNCTION_GET_GRAPHICS_DEVICE_TYPES
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - number of gfx devices
+ * WORD  - device structure size in bytes (excludes device size field)
+ * DWORD - flags         \
+ * WORD  - bus number     } repeated structure
+ * WORD  - device number /
+ */
+/* flags */
+#       define ATIF_PX_REMOVABLE_GRAPHICS_DEVICE                   (1 << 0)
+#       define ATIF_XGP_PORT                                       (1 << 1)
+#       define ATIF_VGA_ENABLED_GRAPHICS_DEVICE                    (1 << 2)
+#       define ATIF_XGP_PORT_IN_DOCK                               (1 << 3)
+
+/* ATPX */
+#define ATPX_FUNCTION_VERIFY_INTERFACE                             0x0
+/* ARG0: ATPX_FUNCTION_VERIFY_INTERFACE
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - version
+ * DWORD - supported functions bit vector
+ */
+/* supported functions vector */
+#       define ATPX_GET_PX_PARAMETERS_SUPPORTED                    (1 << 0)
+#       define ATPX_POWER_CONTROL_SUPPORTED                        (1 << 1)
+#       define ATPX_DISPLAY_MUX_CONTROL_SUPPORTED                  (1 << 2)
+#       define ATPX_I2C_MUX_CONTROL_SUPPORTED                      (1 << 3)
+#       define ATPX_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION_SUPPORTED (1 << 4)
+#       define ATPX_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION_SUPPORTED   (1 << 5)
+#       define ATPX_GET_DISPLAY_CONNECTORS_MAPPING_SUPPORTED       (1 << 7)
+#       define ATPX_GET_DISPLAY_DETECTION_PORTS_SUPPORTED          (1 << 8)
+#define ATPX_FUNCTION_GET_PX_PARAMETERS                            0x1
+/* ARG0: ATPX_FUNCTION_GET_PX_PARAMETERS
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * DWORD - valid flags mask
+ * DWORD - flags
+ */
+/* flags */
+#       define ATPX_LVDS_I2C_AVAILABLE_TO_BOTH_GPUS                (1 << 0)
+#       define ATPX_CRT1_I2C_AVAILABLE_TO_BOTH_GPUS                (1 << 1)
+#       define ATPX_DVI1_I2C_AVAILABLE_TO_BOTH_GPUS                (1 << 2)
+#       define ATPX_CRT1_RGB_SIGNAL_MUXED                          (1 << 3)
+#       define ATPX_TV_SIGNAL_MUXED                                (1 << 4)
+#       define ATPX_DFP_SIGNAL_MUXED                               (1 << 5)
+#       define ATPX_SEPARATE_MUX_FOR_I2C                           (1 << 6)
+#       define ATPX_DYNAMIC_PX_SUPPORTED                           (1 << 7)
+#       define ATPX_ACF_NOT_SUPPORTED                              (1 << 8)
+#       define ATPX_FIXED_NOT_SUPPORTED                            (1 << 9)
+#       define ATPX_DYNAMIC_DGPU_POWER_OFF_SUPPORTED               (1 << 10)
+#       define ATPX_DGPU_REQ_POWER_FOR_DISPLAYS                    (1 << 11)
+#define ATPX_FUNCTION_POWER_CONTROL                                0x2
+/* ARG0: ATPX_FUNCTION_POWER_CONTROL
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - dGPU power state (0: power off, 1: power on)
+ * OUTPUT: none
+ */
+#define ATPX_FUNCTION_DISPLAY_MUX_CONTROL                          0x3
+/* ARG0: ATPX_FUNCTION_DISPLAY_MUX_CONTROL
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - display mux control (0: iGPU, 1: dGPU)
+ * OUTPUT: none
+ */
+#       define ATPX_INTEGRATED_GPU                                 0
+#       define ATPX_DISCRETE_GPU                                   1
+#define ATPX_FUNCTION_I2C_MUX_CONTROL                              0x4
+/* ARG0: ATPX_FUNCTION_I2C_MUX_CONTROL
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - i2c/aux/hpd mux control (0: iGPU, 1: dGPU)
+ * OUTPUT: none
+ */
+#define ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION    0x5
+/* ARG0: ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - target gpu (0: iGPU, 1: dGPU)
+ * OUTPUT: none
+ */
+#define ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION      0x6
+/* ARG0: ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - target gpu (0: iGPU, 1: dGPU)
+ * OUTPUT: none
+ */
+#define ATPX_FUNCTION_GET_DISPLAY_CONNECTORS_MAPPING               0x8
+/* ARG0: ATPX_FUNCTION_GET_DISPLAY_CONNECTORS_MAPPING
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - number of display connectors
+ * WORD  - connector structure size in bytes (excludes connector size field)
+ * BYTE  - flags                                                     \
+ * BYTE  - ATIF display vector bit position                           } repeated
+ * BYTE  - adapter id (0: iGPU, 1-n: dGPU ordered by pcie bus number) } structure
+ * WORD  - connector ACPI id                                         /
+ */
+/* flags */
+#       define ATPX_DISPLAY_OUTPUT_SUPPORTED_BY_ADAPTER_ID_DEVICE  (1 << 0)
+#       define ATPX_DISPLAY_HPD_SUPPORTED_BY_ADAPTER_ID_DEVICE     (1 << 1)
+#       define ATPX_DISPLAY_I2C_SUPPORTED_BY_ADAPTER_ID_DEVICE     (1 << 2)
+#define ATPX_FUNCTION_GET_DISPLAY_DETECTION_PORTS                  0x9
+/* ARG0: ATPX_FUNCTION_GET_DISPLAY_DETECTION_PORTS
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - number of HPD/DDC ports
+ * WORD  - port structure size in bytes (excludes port size field)
+ * BYTE  - ATIF display vector bit position \
+ * BYTE  - hpd id                            } reapeated structure
+ * BYTE  - ddc id                           /
+ *
+ * available on A+A systems only
+ */
+/* hpd id */
+#       define ATPX_HPD_NONE                                       0
+#       define ATPX_HPD1                                           1
+#       define ATPX_HPD2                                           2
+#       define ATPX_HPD3                                           3
+#       define ATPX_HPD4                                           4
+#       define ATPX_HPD5                                           5
+#       define ATPX_HPD6                                           6
+/* ddc id */
+#       define ATPX_DDC_NONE                                       0
+#       define ATPX_DDC1                                           1
+#       define ATPX_DDC2                                           2
+#       define ATPX_DDC3                                           3
+#       define ATPX_DDC4                                           4
+#       define ATPX_DDC5                                           5
+#       define ATPX_DDC6                                           6
+#       define ATPX_DDC7                                           7
+#       define ATPX_DDC8                                           8
+
+/* ATCS */
+#define ATCS_FUNCTION_VERIFY_INTERFACE                             0x0
+/* ARG0: ATCS_FUNCTION_VERIFY_INTERFACE
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - version
+ * DWORD - supported functions bit vector
+ */
+/* supported functions vector */
+#       define ATCS_GET_EXTERNAL_STATE_SUPPORTED                   (1 << 0)
+#       define ATCS_PCIE_PERFORMANCE_REQUEST_SUPPORTED             (1 << 1)
+#       define ATCS_PCIE_DEVICE_READY_NOTIFICATION_SUPPORTED       (1 << 2)
+#       define ATCS_SET_PCIE_BUS_WIDTH_SUPPORTED                   (1 << 3)
+#define ATCS_FUNCTION_GET_EXTERNAL_STATE                           0x1
+/* ARG0: ATCS_FUNCTION_GET_EXTERNAL_STATE
+ * ARG1: none
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * DWORD - valid flags mask
+ * DWORD - flags (0: undocked, 1: docked)
+ */
+/* flags */
+#       define ATCS_DOCKED                                         (1 << 0)
+#define ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST                     0x2
+/* ARG0: ATCS_FUNCTION_PCIE_PERFORMANCE_REQUEST
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num)
+ * WORD  - valid flags mask
+ * WORD  - flags
+ * BYTE  - request type
+ * BYTE  - performance request
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - return value
+ */
+/* flags */
+#       define ATCS_ADVERTISE_CAPS                                 (1 << 0)
+#       define ATCS_WAIT_FOR_COMPLETION                            (1 << 1)
+/* request type */
+#       define ATCS_PCIE_LINK_SPEED                                1
+/* performance request */
+#       define ATCS_REMOVE                                         0
+#       define ATCS_FORCE_LOW_POWER                                1
+#       define ATCS_PERF_LEVEL_1                                   2 /* PCIE Gen 1 */
+#       define ATCS_PERF_LEVEL_2                                   3 /* PCIE Gen 2 */
+#       define ATCS_PERF_LEVEL_3                                   4 /* PCIE Gen 3 */
+/* return value */
+#       define ATCS_REQUEST_REFUSED                                1
+#       define ATCS_REQUEST_COMPLETE                               2
+#       define ATCS_REQUEST_IN_PROGRESS                            3
+#define ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION               0x3
+/* ARG0: ATCS_FUNCTION_PCIE_DEVICE_READY_NOTIFICATION
+ * ARG1: none
+ * OUTPUT: none
+ */
+#define ATCS_FUNCTION_SET_PCIE_BUS_WIDTH                           0x4
+/* ARG0: ATCS_FUNCTION_SET_PCIE_BUS_WIDTH
+ * ARG1:
+ * WORD  - structure size in bytes (includes size field)
+ * WORD  - client id (bit 2-0: func num, 7-3: dev num, 15-8: bus num)
+ * BYTE  - number of active lanes
+ * OUTPUT:
+ * WORD  - structure size in bytes (includes size field)
+ * BYTE  - number of active lanes
+ */
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c
new file mode 100644
index 000000000..857ba0897
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_afmt.c
@@ -0,0 +1,105 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Christian König.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König
+ */
+#include <linux/hdmi.h>
+#include <linux/gcd.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+static const struct amdgpu_afmt_acr amdgpu_afmt_predefined_acr[] = {
+    /*	     32kHz	  44.1kHz	48kHz    */
+    /* Clock      N     CTS      N     CTS      N     CTS */
+    {  25175,  4096,  25175, 28224, 125875,  6144,  25175 }, /*  25,20/1.001 MHz */
+    {  25200,  4096,  25200,  6272,  28000,  6144,  25200 }, /*  25.20       MHz */
+    {  27000,  4096,  27000,  6272,  30000,  6144,  27000 }, /*  27.00       MHz */
+    {  27027,  4096,  27027,  6272,  30030,  6144,  27027 }, /*  27.00*1.001 MHz */
+    {  54000,  4096,  54000,  6272,  60000,  6144,  54000 }, /*  54.00       MHz */
+    {  54054,  4096,  54054,  6272,  60060,  6144,  54054 }, /*  54.00*1.001 MHz */
+    {  74176,  4096,  74176,  5733,  75335,  6144,  74176 }, /*  74.25/1.001 MHz */
+    {  74250,  4096,  74250,  6272,  82500,  6144,  74250 }, /*  74.25       MHz */
+    { 148352,  4096, 148352,  5733, 150670,  6144, 148352 }, /* 148.50/1.001 MHz */
+    { 148500,  4096, 148500,  6272, 165000,  6144, 148500 }, /* 148.50       MHz */
+};
+
+
+/*
+ * calculate CTS and N values if they are not found in the table
+ */
+static void amdgpu_afmt_calc_cts(uint32_t clock, int *CTS, int *N, int freq)
+{
+	int n, cts;
+	unsigned long div, mul;
+
+	/* Safe, but overly large values */
+	n = 128 * freq;
+	cts = clock * 1000;
+
+	/* Smallest valid fraction */
+	div = gcd(n, cts);
+
+	n /= div;
+	cts /= div;
+
+	/*
+	 * The optimal N is 128*freq/1000. Calculate the closest larger
+	 * value that doesn't truncate any bits.
+	 */
+	mul = ((128*freq/1000) + (n-1))/n;
+
+	n *= mul;
+	cts *= mul;
+
+	/* Check that we are in spec (not always possible) */
+	if (n < (128*freq/1500))
+		printk(KERN_WARNING "Calculated ACR N value is too small. You may experience audio problems.\n");
+	if (n > (128*freq/300))
+		printk(KERN_WARNING "Calculated ACR N value is too large. You may experience audio problems.\n");
+
+	*N = n;
+	*CTS = cts;
+
+	DRM_DEBUG("Calculated ACR timing N=%d CTS=%d for frequency %d\n",
+		  *N, *CTS, freq);
+}
+
+struct amdgpu_afmt_acr amdgpu_afmt_acr(uint32_t clock)
+{
+	struct amdgpu_afmt_acr res;
+	u8 i;
+
+	/* Precalculated values for common clocks */
+	for (i = 0; i < ARRAY_SIZE(amdgpu_afmt_predefined_acr); i++) {
+		if (amdgpu_afmt_predefined_acr[i].clock == clock)
+			return amdgpu_afmt_predefined_acr[i];
+	}
+
+	/* And odd clocks get manually calculated */
+	amdgpu_afmt_calc_cts(clock, &res.cts_32khz, &res.n_32khz, 32000);
+	amdgpu_afmt_calc_cts(clock, &res.cts_44_1khz, &res.n_44_1khz, 44100);
+	amdgpu_afmt_calc_cts(clock, &res.cts_48khz, &res.n_48khz, 48000);
+
+	return res;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
new file mode 100644
index 000000000..84d68d658
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -0,0 +1,269 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu_amdkfd.h"
+#include "amd_shared.h"
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include <linux/module.h>
+
+const struct kfd2kgd_calls *kfd2kgd;
+const struct kgd2kfd_calls *kgd2kfd;
+bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+
+bool amdgpu_amdkfd_init(void)
+{
+#if defined(CONFIG_HSA_AMD_MODULE)
+	bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+
+	kgd2kfd_init_p = symbol_request(kgd2kfd_init);
+
+	if (kgd2kfd_init_p == NULL)
+		return false;
+#endif
+	return true;
+}
+
+bool amdgpu_amdkfd_load_interface(struct amdgpu_device *rdev)
+{
+#if defined(CONFIG_HSA_AMD_MODULE)
+	bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+#endif
+
+	switch (rdev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_KAVERI:
+		kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
+		break;
+#endif
+	case CHIP_CARRIZO:
+		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
+		break;
+	default:
+		return false;
+	}
+
+#if defined(CONFIG_HSA_AMD_MODULE)
+	kgd2kfd_init_p = symbol_request(kgd2kfd_init);
+
+	if (kgd2kfd_init_p == NULL) {
+		kfd2kgd = NULL;
+		return false;
+	}
+
+	if (!kgd2kfd_init_p(KFD_INTERFACE_VERSION, &kgd2kfd)) {
+		symbol_put(kgd2kfd_init);
+		kfd2kgd = NULL;
+		kgd2kfd = NULL;
+
+		return false;
+	}
+
+	return true;
+#elif defined(CONFIG_HSA_AMD)
+	if (!kgd2kfd_init(KFD_INTERFACE_VERSION, &kgd2kfd)) {
+		kfd2kgd = NULL;
+		kgd2kfd = NULL;
+		return false;
+	}
+
+	return true;
+#else
+	kfd2kgd = NULL;
+	return false;
+#endif
+}
+
+void amdgpu_amdkfd_fini(void)
+{
+	if (kgd2kfd) {
+		kgd2kfd->exit();
+		symbol_put(kgd2kfd_init);
+	}
+}
+
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *rdev)
+{
+	if (kgd2kfd)
+		rdev->kfd = kgd2kfd->probe((struct kgd_dev *)rdev,
+					rdev->pdev, kfd2kgd);
+}
+
+void amdgpu_amdkfd_device_init(struct amdgpu_device *rdev)
+{
+	if (rdev->kfd) {
+		struct kgd2kfd_shared_resources gpu_resources = {
+			.compute_vmid_bitmap = 0xFF00,
+
+			.first_compute_pipe = 1,
+			.compute_pipe_count = 4 - 1,
+		};
+
+		amdgpu_doorbell_get_kfd_info(rdev,
+				&gpu_resources.doorbell_physical_address,
+				&gpu_resources.doorbell_aperture_size,
+				&gpu_resources.doorbell_start_offset);
+
+		kgd2kfd->device_init(rdev->kfd, &gpu_resources);
+	}
+}
+
+void amdgpu_amdkfd_device_fini(struct amdgpu_device *rdev)
+{
+	if (rdev->kfd) {
+		kgd2kfd->device_exit(rdev->kfd);
+		rdev->kfd = NULL;
+	}
+}
+
+void amdgpu_amdkfd_interrupt(struct amdgpu_device *rdev,
+		const void *ih_ring_entry)
+{
+	if (rdev->kfd)
+		kgd2kfd->interrupt(rdev->kfd, ih_ring_entry);
+}
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *rdev)
+{
+	if (rdev->kfd)
+		kgd2kfd->suspend(rdev->kfd);
+}
+
+int amdgpu_amdkfd_resume(struct amdgpu_device *rdev)
+{
+	int r = 0;
+
+	if (rdev->kfd)
+		r = kgd2kfd->resume(rdev->kfd);
+
+	return r;
+}
+
+u32 pool_to_domain(enum kgd_memory_pool p)
+{
+	switch (p) {
+	case KGD_POOL_FRAMEBUFFER: return AMDGPU_GEM_DOMAIN_VRAM;
+	default: return AMDGPU_GEM_DOMAIN_GTT;
+	}
+}
+
+int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
+			void **mem_obj, uint64_t *gpu_addr,
+			void **cpu_ptr)
+{
+	struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
+	struct kgd_mem **mem = (struct kgd_mem **) mem_obj;
+	int r;
+
+	BUG_ON(kgd == NULL);
+	BUG_ON(gpu_addr == NULL);
+	BUG_ON(cpu_ptr == NULL);
+
+	*mem = kmalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+	if ((*mem) == NULL)
+		return -ENOMEM;
+
+	r = amdgpu_bo_create(rdev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
+			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, &(*mem)->bo);
+	if (r) {
+		dev_err(rdev->dev,
+			"failed to allocate BO for amdkfd (%d)\n", r);
+		return r;
+	}
+
+	/* map the buffer */
+	r = amdgpu_bo_reserve((*mem)->bo, true);
+	if (r) {
+		dev_err(rdev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
+		goto allocate_mem_reserve_bo_failed;
+	}
+
+	r = amdgpu_bo_pin((*mem)->bo, AMDGPU_GEM_DOMAIN_GTT,
+				&(*mem)->gpu_addr);
+	if (r) {
+		dev_err(rdev->dev, "(%d) failed to pin bo for amdkfd\n", r);
+		goto allocate_mem_pin_bo_failed;
+	}
+	*gpu_addr = (*mem)->gpu_addr;
+
+	r = amdgpu_bo_kmap((*mem)->bo, &(*mem)->cpu_ptr);
+	if (r) {
+		dev_err(rdev->dev,
+			"(%d) failed to map bo to kernel for amdkfd\n", r);
+		goto allocate_mem_kmap_bo_failed;
+	}
+	*cpu_ptr = (*mem)->cpu_ptr;
+
+	amdgpu_bo_unreserve((*mem)->bo);
+
+	return 0;
+
+allocate_mem_kmap_bo_failed:
+	amdgpu_bo_unpin((*mem)->bo);
+allocate_mem_pin_bo_failed:
+	amdgpu_bo_unreserve((*mem)->bo);
+allocate_mem_reserve_bo_failed:
+	amdgpu_bo_unref(&(*mem)->bo);
+
+	return r;
+}
+
+void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
+{
+	struct kgd_mem *mem = (struct kgd_mem *) mem_obj;
+
+	BUG_ON(mem == NULL);
+
+	amdgpu_bo_reserve(mem->bo, true);
+	amdgpu_bo_kunmap(mem->bo);
+	amdgpu_bo_unpin(mem->bo);
+	amdgpu_bo_unreserve(mem->bo);
+	amdgpu_bo_unref(&(mem->bo));
+	kfree(mem);
+}
+
+uint64_t get_vmem_size(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *rdev =
+		(struct amdgpu_device *)kgd;
+
+	BUG_ON(kgd == NULL);
+
+	return rdev->mc.real_vram_size;
+}
+
+uint64_t get_gpu_clock_counter(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
+
+	if (rdev->asic_funcs->get_gpu_clock_counter)
+		return rdev->asic_funcs->get_gpu_clock_counter(rdev);
+	return 0;
+}
+
+uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
+
+	/* The sclk is in quantas of 10kHz */
+	return rdev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk / 100;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
new file mode 100644
index 000000000..a8be76554
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -0,0 +1,65 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* amdgpu_amdkfd.h defines the private interface between amdgpu and amdkfd. */
+
+#ifndef AMDGPU_AMDKFD_H_INCLUDED
+#define AMDGPU_AMDKFD_H_INCLUDED
+
+#include <linux/types.h>
+#include <kgd_kfd_interface.h>
+
+struct amdgpu_device;
+
+struct kgd_mem {
+	struct amdgpu_bo *bo;
+	uint64_t gpu_addr;
+	void *cpu_ptr;
+};
+
+bool amdgpu_amdkfd_init(void);
+void amdgpu_amdkfd_fini(void);
+
+bool amdgpu_amdkfd_load_interface(struct amdgpu_device *rdev);
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *rdev);
+int amdgpu_amdkfd_resume(struct amdgpu_device *rdev);
+void amdgpu_amdkfd_interrupt(struct amdgpu_device *rdev,
+			const void *ih_ring_entry);
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *rdev);
+void amdgpu_amdkfd_device_init(struct amdgpu_device *rdev);
+void amdgpu_amdkfd_device_fini(struct amdgpu_device *rdev);
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void);
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void);
+
+/* Shared API */
+int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
+			void **mem_obj, uint64_t *gpu_addr,
+			void **cpu_ptr);
+void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj);
+uint64_t get_vmem_size(struct kgd_dev *kgd);
+uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
+
+uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
+
+#endif /* AMDGPU_AMDKFD_H_INCLUDED */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
new file mode 100644
index 000000000..0e1376317
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -0,0 +1,670 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/fdtable.h>
+#include <linux/uaccess.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "cikd.h"
+#include "cik_sdma.h"
+#include "amdgpu_ucode.h"
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+#include "cik_structs.h"
+
+#define CIK_PIPE_PER_MEC	(4)
+
+enum {
+	MAX_TRAPID = 8,		/* 3 bits in the bitfield. */
+	MAX_WATCH_ADDRESSES = 4
+};
+
+enum {
+	ADDRESS_WATCH_REG_ADDR_HI = 0,
+	ADDRESS_WATCH_REG_ADDR_LO,
+	ADDRESS_WATCH_REG_CNTL,
+	ADDRESS_WATCH_REG_MAX
+};
+
+/*  not defined in the CI/KV reg file  */
+enum {
+	ADDRESS_WATCH_REG_CNTL_ATC_BIT = 0x10000000UL,
+	ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK = 0x00FFFFFF,
+	ADDRESS_WATCH_REG_ADDLOW_MASK_EXTENSION = 0x03000000,
+	/* extend the mask to 26 bits to match the low address field */
+	ADDRESS_WATCH_REG_ADDLOW_SHIFT = 6,
+	ADDRESS_WATCH_REG_ADDHIGH_MASK = 0xFFFF
+};
+
+static const uint32_t watchRegs[MAX_WATCH_ADDRESSES * ADDRESS_WATCH_REG_MAX] = {
+	mmTCP_WATCH0_ADDR_H, mmTCP_WATCH0_ADDR_L, mmTCP_WATCH0_CNTL,
+	mmTCP_WATCH1_ADDR_H, mmTCP_WATCH1_ADDR_L, mmTCP_WATCH1_CNTL,
+	mmTCP_WATCH2_ADDR_H, mmTCP_WATCH2_ADDR_L, mmTCP_WATCH2_CNTL,
+	mmTCP_WATCH3_ADDR_H, mmTCP_WATCH3_ADDR_L, mmTCP_WATCH3_CNTL
+};
+
+union TCP_WATCH_CNTL_BITS {
+	struct {
+		uint32_t mask:24;
+		uint32_t vmid:4;
+		uint32_t atc:1;
+		uint32_t mode:2;
+		uint32_t valid:1;
+	} bitfields, bits;
+	uint32_t u32All;
+	signed int i32All;
+	float f32All;
+};
+
+/*
+ * Register access functions
+ */
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t sh_mem_config,	uint32_t sh_mem_ape1_base,
+		uint32_t sh_mem_ape1_limit, uint32_t sh_mem_bases);
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid);
+
+static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t hpd_size, uint64_t hpd_gpu_addr);
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr);
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd);
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id);
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+				unsigned int timeout, uint32_t pipe_id,
+				uint32_t queue_id);
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int timeout);
+static int kgd_address_watch_disable(struct kgd_dev *kgd);
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo);
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd);
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset);
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd, uint8_t vmid);
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+							uint8_t vmid);
+static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+
+static const struct kfd2kgd_calls kfd2kgd = {
+	.init_gtt_mem_allocation = alloc_gtt_mem,
+	.free_gtt_mem = free_gtt_mem,
+	.get_vmem_size = get_vmem_size,
+	.get_gpu_clock_counter = get_gpu_clock_counter,
+	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_pipeline = kgd_init_pipeline,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.address_watch_disable = kgd_address_watch_disable,
+	.address_watch_execute = kgd_address_watch_execute,
+	.wave_control_execute = kgd_wave_control_execute,
+	.address_watch_get_offset = kgd_address_watch_get_offset,
+	.get_atc_vmid_pasid_mapping_pasid = get_atc_vmid_pasid_mapping_pasid,
+	.get_atc_vmid_pasid_mapping_valid = get_atc_vmid_pasid_mapping_valid,
+	.write_vmid_invalidate_request = write_vmid_invalidate_request,
+	.get_fw_version = get_fw_version
+};
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions()
+{
+	return (struct kfd2kgd_calls *)&kfd2kgd;
+}
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue);
+
+	mutex_lock(&adev->srbm_mutex);
+	WREG32(mmSRBM_GFX_CNTL, value);
+}
+
+static void unlock_srbm(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	WREG32(mmSRBM_GFX_CNTL, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	uint32_t mec = (++pipe_id / CIK_PIPE_PER_MEC) + 1;
+	uint32_t pipe = (pipe_id % CIK_PIPE_PER_MEC);
+
+	lock_srbm(kgd, mec, pipe, queue_id, 0);
+}
+
+static void release_queue(struct kgd_dev *kgd)
+{
+	unlock_srbm(kgd);
+}
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+
+	WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32(mmSH_MEM_APE1_BASE, sh_mem_ape1_base);
+	WREG32(mmSH_MEM_APE1_LIMIT, sh_mem_ape1_limit);
+	WREG32(mmSH_MEM_BASES, sh_mem_bases);
+
+	unlock_srbm(kgd);
+}
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished and the
+	 * SW cleared it. So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	WREG32(mmATC_VMID0_PASID_MAPPING + vmid, pasid_mapping);
+
+	while (!(RREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS) & (1U << vmid)))
+		cpu_relax();
+	WREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS, 1U << vmid);
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(mmIH_VMID_0_LUT + vmid, pasid_mapping);
+
+	return 0;
+}
+
+static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t hpd_size, uint64_t hpd_gpu_addr)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	uint32_t mec = (++pipe_id / CIK_PIPE_PER_MEC) + 1;
+	uint32_t pipe = (pipe_id % CIK_PIPE_PER_MEC);
+
+	lock_srbm(kgd, mec, pipe, 0, 0);
+	WREG32(mmCP_HPD_EOP_BASE_ADDR, lower_32_bits(hpd_gpu_addr >> 8));
+	WREG32(mmCP_HPD_EOP_BASE_ADDR_HI, upper_32_bits(hpd_gpu_addr >> 8));
+	WREG32(mmCP_HPD_EOP_VMID, 0);
+	WREG32(mmCP_HPD_EOP_CONTROL, hpd_size);
+	unlock_srbm(kgd);
+
+	return 0;
+}
+
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (pipe_id / CIK_PIPE_PER_MEC) + 1;
+	pipe = (pipe_id % CIK_PIPE_PER_MEC);
+
+	lock_srbm(kgd, mec, pipe, 0, 0);
+
+	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
+			CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
+
+	unlock_srbm(kgd);
+
+	return 0;
+}
+
+static inline uint32_t get_sdma_base_addr(struct cik_sdma_rlc_registers *m)
+{
+	uint32_t retval;
+
+	retval = m->sdma_engine_id * SDMA1_REGISTER_OFFSET +
+			m->sdma_queue_id * KFD_CIK_SDMA_QUEUE_OFFSET;
+
+	pr_debug("kfd: sdma base address: 0x%x\n", retval);
+
+	return retval;
+}
+
+static inline struct cik_mqd *get_mqd(void *mqd)
+{
+	return (struct cik_mqd *)mqd;
+}
+
+static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
+{
+	return (struct cik_sdma_rlc_registers *)mqd;
+}
+
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t wptr_shadow, is_wptr_shadow_valid;
+	struct cik_mqd *m;
+
+	m = get_mqd(mqd);
+
+	is_wptr_shadow_valid = !get_user(wptr_shadow, wptr);
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	WREG32(mmCP_MQD_BASE_ADDR, m->cp_mqd_base_addr_lo);
+	WREG32(mmCP_MQD_BASE_ADDR_HI, m->cp_mqd_base_addr_hi);
+	WREG32(mmCP_MQD_CONTROL, m->cp_mqd_control);
+
+	WREG32(mmCP_HQD_PQ_BASE, m->cp_hqd_pq_base_lo);
+	WREG32(mmCP_HQD_PQ_BASE_HI, m->cp_hqd_pq_base_hi);
+	WREG32(mmCP_HQD_PQ_CONTROL, m->cp_hqd_pq_control);
+
+	WREG32(mmCP_HQD_IB_CONTROL, m->cp_hqd_ib_control);
+	WREG32(mmCP_HQD_IB_BASE_ADDR, m->cp_hqd_ib_base_addr_lo);
+	WREG32(mmCP_HQD_IB_BASE_ADDR_HI, m->cp_hqd_ib_base_addr_hi);
+
+	WREG32(mmCP_HQD_IB_RPTR, m->cp_hqd_ib_rptr);
+
+	WREG32(mmCP_HQD_PERSISTENT_STATE, m->cp_hqd_persistent_state);
+	WREG32(mmCP_HQD_SEMA_CMD, m->cp_hqd_sema_cmd);
+	WREG32(mmCP_HQD_MSG_TYPE, m->cp_hqd_msg_type);
+
+	WREG32(mmCP_HQD_ATOMIC0_PREOP_LO, m->cp_hqd_atomic0_preop_lo);
+	WREG32(mmCP_HQD_ATOMIC0_PREOP_HI, m->cp_hqd_atomic0_preop_hi);
+	WREG32(mmCP_HQD_ATOMIC1_PREOP_LO, m->cp_hqd_atomic1_preop_lo);
+	WREG32(mmCP_HQD_ATOMIC1_PREOP_HI, m->cp_hqd_atomic1_preop_hi);
+
+	WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR, m->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+			m->cp_hqd_pq_rptr_report_addr_hi);
+
+	WREG32(mmCP_HQD_PQ_RPTR, m->cp_hqd_pq_rptr);
+
+	WREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR, m->cp_hqd_pq_wptr_poll_addr_lo);
+	WREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR_HI, m->cp_hqd_pq_wptr_poll_addr_hi);
+
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, m->cp_hqd_pq_doorbell_control);
+
+	WREG32(mmCP_HQD_VMID, m->cp_hqd_vmid);
+
+	WREG32(mmCP_HQD_QUANTUM, m->cp_hqd_quantum);
+
+	WREG32(mmCP_HQD_PIPE_PRIORITY, m->cp_hqd_pipe_priority);
+	WREG32(mmCP_HQD_QUEUE_PRIORITY, m->cp_hqd_queue_priority);
+
+	WREG32(mmCP_HQD_IQ_RPTR, m->cp_hqd_iq_rptr);
+
+	if (is_wptr_shadow_valid)
+		WREG32(mmCP_HQD_PQ_WPTR, wptr_shadow);
+
+	WREG32(mmCP_HQD_ACTIVE, m->cp_hqd_active);
+	release_queue(kgd);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_VIRTUAL_ADDR,
+			m->sdma_rlc_virtual_addr);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE,
+			m->sdma_rlc_rb_base);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE_HI,
+			m->sdma_rlc_rb_base_hi);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
+			m->sdma_rlc_rb_rptr_addr_lo);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
+			m->sdma_rlc_rb_rptr_addr_hi);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL,
+			m->sdma_rlc_doorbell);
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
+			m->sdma_rlc_rb_cntl);
+
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	act = RREG32(mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32(mmCP_HQD_PQ_BASE) &&
+				high == RREG32(mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	release_queue(kgd);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+				unsigned int timeout, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t temp;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, reset_type);
+
+	while (true) {
+		temp = RREG32(mmCP_HQD_ACTIVE);
+		if (temp & CP_HQD_ACTIVE__ACTIVE_MASK)
+			break;
+		if (timeout == 0) {
+			pr_err("kfd: cp queue preemption time out (%dms)\n",
+				temp);
+			release_queue(kgd);
+			return -ETIME;
+		}
+		msleep(20);
+		timeout -= 20;
+	}
+
+	release_queue(kgd);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int timeout)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+	uint32_t temp;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_STATUS_REG__RB_CMD_IDLE__SHIFT)
+			break;
+		if (timeout == 0)
+			return -ETIME;
+		msleep(20);
+		timeout -= 20;
+	}
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, 0);
+
+	return 0;
+}
+
+static int kgd_address_watch_disable(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	union TCP_WATCH_CNTL_BITS cntl;
+	unsigned int i;
+
+	cntl.u32All = 0;
+
+	cntl.bitfields.valid = 0;
+	cntl.bitfields.mask = ADDRESS_WATCH_REG_CNTL_DEFAULT_MASK;
+	cntl.bitfields.atc = 1;
+
+	/* Turning off this address until we set all the registers */
+	for (i = 0; i < MAX_WATCH_ADDRESSES; i++)
+		WREG32(watchRegs[i * ADDRESS_WATCH_REG_MAX +
+			ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+
+	return 0;
+}
+
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	union TCP_WATCH_CNTL_BITS cntl;
+
+	cntl.u32All = cntl_val;
+
+	/* Turning off this watch point until we set all the registers */
+	cntl.bitfields.valid = 0;
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_ADDR_HI], addr_hi);
+
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_ADDR_LO], addr_lo);
+
+	/* Enable the watch point */
+	cntl.bitfields.valid = 1;
+
+	WREG32(watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX +
+		ADDRESS_WATCH_REG_CNTL], cntl.u32All);
+
+	return 0;
+}
+
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t data;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32(mmSQ_CMD, sq_cmd);
+
+	/*  Restore the GRBM_GFX_INDEX register  */
+
+	data = GRBM_GFX_INDEX__INSTANCE_BROADCAST_WRITES_MASK |
+		GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+		GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK;
+
+	WREG32(mmGRBM_GFX_INDEX, data);
+
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset)
+{
+	return watchRegs[watch_point_id * ADDRESS_WATCH_REG_MAX + reg_offset];
+}
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+							uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+								uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	const union amdgpu_firmware_header *hdr;
+
+	BUG_ON(kgd == NULL);
+
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.pfp_fw->data;
+		break;
+
+	case KGD_ENGINE_ME:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.me_fw->data;
+		break;
+
+	case KGD_ENGINE_CE:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.ce_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC1:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.mec_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC2:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.mec2_fw->data;
+		break;
+
+	case KGD_ENGINE_RLC:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.rlc_fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA1:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->sdma.instance[0].fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA2:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->sdma.instance[1].fw->data;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (hdr == NULL)
+		return 0;
+
+	/* Only 12 bit in use*/
+	return hdr->common.ucode_version;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
new file mode 100644
index 000000000..79fa5c7de
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -0,0 +1,543 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/fdtable.h>
+#include <linux/uaccess.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_ucode.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "oss/oss_3_0_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "vi_structs.h"
+#include "vid.h"
+
+#define VI_PIPE_PER_MEC	(4)
+
+struct cik_sdma_rlc_registers;
+
+/*
+ * Register access functions
+ */
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+		uint32_t sh_mem_config,
+		uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit,
+		uint32_t sh_mem_bases);
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+		unsigned int vmid);
+static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
+		uint32_t hpd_size, uint64_t hpd_gpu_addr);
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+		uint32_t queue_id, uint32_t __user *wptr);
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd);
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+		uint32_t pipe_id, uint32_t queue_id);
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
+static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+				unsigned int timeout, uint32_t pipe_id,
+				uint32_t queue_id);
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int timeout);
+static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
+static int kgd_address_watch_disable(struct kgd_dev *kgd);
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo);
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd);
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset);
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+		uint8_t vmid);
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+		uint8_t vmid);
+static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+
+static const struct kfd2kgd_calls kfd2kgd = {
+	.init_gtt_mem_allocation = alloc_gtt_mem,
+	.free_gtt_mem = free_gtt_mem,
+	.get_vmem_size = get_vmem_size,
+	.get_gpu_clock_counter = get_gpu_clock_counter,
+	.get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz,
+	.program_sh_mem_settings = kgd_program_sh_mem_settings,
+	.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
+	.init_pipeline = kgd_init_pipeline,
+	.init_interrupts = kgd_init_interrupts,
+	.hqd_load = kgd_hqd_load,
+	.hqd_sdma_load = kgd_hqd_sdma_load,
+	.hqd_is_occupied = kgd_hqd_is_occupied,
+	.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
+	.hqd_destroy = kgd_hqd_destroy,
+	.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
+	.address_watch_disable = kgd_address_watch_disable,
+	.address_watch_execute = kgd_address_watch_execute,
+	.wave_control_execute = kgd_wave_control_execute,
+	.address_watch_get_offset = kgd_address_watch_get_offset,
+	.get_atc_vmid_pasid_mapping_pasid =
+			get_atc_vmid_pasid_mapping_pasid,
+	.get_atc_vmid_pasid_mapping_valid =
+			get_atc_vmid_pasid_mapping_valid,
+	.write_vmid_invalidate_request = write_vmid_invalidate_request,
+	.get_fw_version = get_fw_version
+};
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions()
+{
+	return (struct kfd2kgd_calls *)&kfd2kgd;
+}
+
+static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
+{
+	return (struct amdgpu_device *)kgd;
+}
+
+static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
+			uint32_t queue, uint32_t vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue);
+
+	mutex_lock(&adev->srbm_mutex);
+	WREG32(mmSRBM_GFX_CNTL, value);
+}
+
+static void unlock_srbm(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	WREG32(mmSRBM_GFX_CNTL, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	uint32_t mec = (++pipe_id / VI_PIPE_PER_MEC) + 1;
+	uint32_t pipe = (pipe_id % VI_PIPE_PER_MEC);
+
+	lock_srbm(kgd, mec, pipe, queue_id, 0);
+}
+
+static void release_queue(struct kgd_dev *kgd)
+{
+	unlock_srbm(kgd);
+}
+
+static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
+					uint32_t sh_mem_config,
+					uint32_t sh_mem_ape1_base,
+					uint32_t sh_mem_ape1_limit,
+					uint32_t sh_mem_bases)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+
+	WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+	WREG32(mmSH_MEM_APE1_BASE, sh_mem_ape1_base);
+	WREG32(mmSH_MEM_APE1_LIMIT, sh_mem_ape1_limit);
+	WREG32(mmSH_MEM_BASES, sh_mem_bases);
+
+	unlock_srbm(kgd);
+}
+
+static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
+					unsigned int vmid)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	/*
+	 * We have to assume that there is no outstanding mapping.
+	 * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
+	 * a mapping is in progress or because a mapping finished
+	 * and the SW cleared it.
+	 * So the protocol is to always wait & clear.
+	 */
+	uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
+			ATC_VMID0_PASID_MAPPING__VALID_MASK;
+
+	WREG32(mmATC_VMID0_PASID_MAPPING + vmid, pasid_mapping);
+
+	while (!(RREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS) & (1U << vmid)))
+		cpu_relax();
+	WREG32(mmATC_VMID_PASID_MAPPING_UPDATE_STATUS, 1U << vmid);
+
+	/* Mapping vmid to pasid also for IH block */
+	WREG32(mmIH_VMID_0_LUT + vmid, pasid_mapping);
+
+	return 0;
+}
+
+static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
+				uint32_t hpd_size, uint64_t hpd_gpu_addr)
+{
+	return 0;
+}
+
+static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t mec;
+	uint32_t pipe;
+
+	mec = (++pipe_id / VI_PIPE_PER_MEC) + 1;
+	pipe = (pipe_id % VI_PIPE_PER_MEC);
+
+	lock_srbm(kgd, mec, pipe, 0, 0);
+
+	WREG32(mmCPC_INT_CNTL, CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK);
+
+	unlock_srbm(kgd);
+
+	return 0;
+}
+
+static inline uint32_t get_sdma_base_addr(struct cik_sdma_rlc_registers *m)
+{
+	return 0;
+}
+
+static inline struct vi_mqd *get_mqd(void *mqd)
+{
+	return (struct vi_mqd *)mqd;
+}
+
+static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
+{
+	return (struct cik_sdma_rlc_registers *)mqd;
+}
+
+static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
+			uint32_t queue_id, uint32_t __user *wptr)
+{
+	struct vi_mqd *m;
+	uint32_t shadow_wptr, valid_wptr;
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+
+	m = get_mqd(mqd);
+
+	valid_wptr = copy_from_user(&shadow_wptr, wptr, sizeof(shadow_wptr));
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	WREG32(mmCP_MQD_CONTROL, m->cp_mqd_control);
+	WREG32(mmCP_MQD_BASE_ADDR, m->cp_mqd_base_addr_lo);
+	WREG32(mmCP_MQD_BASE_ADDR_HI, m->cp_mqd_base_addr_hi);
+
+	WREG32(mmCP_HQD_VMID, m->cp_hqd_vmid);
+	WREG32(mmCP_HQD_PERSISTENT_STATE, m->cp_hqd_persistent_state);
+	WREG32(mmCP_HQD_PIPE_PRIORITY, m->cp_hqd_pipe_priority);
+	WREG32(mmCP_HQD_QUEUE_PRIORITY, m->cp_hqd_queue_priority);
+	WREG32(mmCP_HQD_QUANTUM, m->cp_hqd_quantum);
+	WREG32(mmCP_HQD_PQ_BASE, m->cp_hqd_pq_base_lo);
+	WREG32(mmCP_HQD_PQ_BASE_HI, m->cp_hqd_pq_base_hi);
+	WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR, m->cp_hqd_pq_rptr_report_addr_lo);
+	WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+			m->cp_hqd_pq_rptr_report_addr_hi);
+
+	if (valid_wptr > 0)
+		WREG32(mmCP_HQD_PQ_WPTR, shadow_wptr);
+
+	WREG32(mmCP_HQD_PQ_CONTROL, m->cp_hqd_pq_control);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, m->cp_hqd_pq_doorbell_control);
+
+	WREG32(mmCP_HQD_EOP_BASE_ADDR, m->cp_hqd_eop_base_addr_lo);
+	WREG32(mmCP_HQD_EOP_BASE_ADDR_HI, m->cp_hqd_eop_base_addr_hi);
+	WREG32(mmCP_HQD_EOP_CONTROL, m->cp_hqd_eop_control);
+	WREG32(mmCP_HQD_EOP_RPTR, m->cp_hqd_eop_rptr);
+	WREG32(mmCP_HQD_EOP_WPTR, m->cp_hqd_eop_wptr);
+	WREG32(mmCP_HQD_EOP_EVENTS, m->cp_hqd_eop_done_events);
+
+	WREG32(mmCP_HQD_CTX_SAVE_BASE_ADDR_LO, m->cp_hqd_ctx_save_base_addr_lo);
+	WREG32(mmCP_HQD_CTX_SAVE_BASE_ADDR_HI, m->cp_hqd_ctx_save_base_addr_hi);
+	WREG32(mmCP_HQD_CTX_SAVE_CONTROL, m->cp_hqd_ctx_save_control);
+	WREG32(mmCP_HQD_CNTL_STACK_OFFSET, m->cp_hqd_cntl_stack_offset);
+	WREG32(mmCP_HQD_CNTL_STACK_SIZE, m->cp_hqd_cntl_stack_size);
+	WREG32(mmCP_HQD_WG_STATE_OFFSET, m->cp_hqd_wg_state_offset);
+	WREG32(mmCP_HQD_CTX_SAVE_SIZE, m->cp_hqd_ctx_save_size);
+
+	WREG32(mmCP_HQD_IB_CONTROL, m->cp_hqd_ib_control);
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, m->cp_hqd_dequeue_request);
+	WREG32(mmCP_HQD_ERROR, m->cp_hqd_error);
+	WREG32(mmCP_HQD_EOP_WPTR_MEM, m->cp_hqd_eop_wptr_mem);
+	WREG32(mmCP_HQD_EOP_DONES, m->cp_hqd_eop_dones);
+
+	WREG32(mmCP_HQD_ACTIVE, m->cp_hqd_active);
+
+	release_queue(kgd);
+
+	return 0;
+}
+
+static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd)
+{
+	return 0;
+}
+
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
+				uint32_t pipe_id, uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t act;
+	bool retval = false;
+	uint32_t low, high;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+	act = RREG32(mmCP_HQD_ACTIVE);
+	if (act) {
+		low = lower_32_bits(queue_address >> 8);
+		high = upper_32_bits(queue_address >> 8);
+
+		if (low == RREG32(mmCP_HQD_PQ_BASE) &&
+				high == RREG32(mmCP_HQD_PQ_BASE_HI))
+			retval = true;
+	}
+	release_queue(kgd);
+	return retval;
+}
+
+static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+	uint32_t sdma_rlc_rb_cntl;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+
+	if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
+		return true;
+
+	return false;
+}
+
+static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+				unsigned int timeout, uint32_t pipe_id,
+				uint32_t queue_id)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t temp;
+
+	acquire_queue(kgd, pipe_id, queue_id);
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, reset_type);
+
+	while (true) {
+		temp = RREG32(mmCP_HQD_ACTIVE);
+		if (temp & CP_HQD_ACTIVE__ACTIVE_MASK)
+			break;
+		if (timeout == 0) {
+			pr_err("kfd: cp queue preemption time out (%dms)\n",
+				temp);
+			release_queue(kgd);
+			return -ETIME;
+		}
+		msleep(20);
+		timeout -= 20;
+	}
+
+	release_queue(kgd);
+	return 0;
+}
+
+static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
+				unsigned int timeout)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct cik_sdma_rlc_registers *m;
+	uint32_t sdma_base_addr;
+	uint32_t temp;
+
+	m = get_sdma_mqd(mqd);
+	sdma_base_addr = get_sdma_base_addr(m);
+
+	temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
+	temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
+
+	while (true) {
+		temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
+		if (temp & SDMA0_STATUS_REG__RB_CMD_IDLE__SHIFT)
+			break;
+		if (timeout == 0)
+			return -ETIME;
+		msleep(20);
+		timeout -= 20;
+	}
+
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR, 0);
+	WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, 0);
+
+	return 0;
+}
+
+static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
+							uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
+								uint8_t vmid)
+{
+	uint32_t reg;
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	reg = RREG32(mmATC_VMID0_PASID_MAPPING + vmid);
+	return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
+}
+
+static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+static int kgd_address_watch_disable(struct kgd_dev *kgd)
+{
+	return 0;
+}
+
+static int kgd_address_watch_execute(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					uint32_t cntl_val,
+					uint32_t addr_hi,
+					uint32_t addr_lo)
+{
+	return 0;
+}
+
+static int kgd_wave_control_execute(struct kgd_dev *kgd,
+					uint32_t gfx_index_val,
+					uint32_t sq_cmd)
+{
+	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	uint32_t data = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	WREG32(mmGRBM_GFX_INDEX, gfx_index_val);
+	WREG32(mmSQ_CMD, sq_cmd);
+
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		INSTANCE_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SH_BROADCAST_WRITES, 1);
+	data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
+		SE_BROADCAST_WRITES, 1);
+
+	WREG32(mmGRBM_GFX_INDEX, data);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	return 0;
+}
+
+static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
+					unsigned int watch_point_id,
+					unsigned int reg_offset)
+{
+	return 0;
+}
+
+static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+	const union amdgpu_firmware_header *hdr;
+
+	BUG_ON(kgd == NULL);
+
+	switch (type) {
+	case KGD_ENGINE_PFP:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.pfp_fw->data;
+		break;
+
+	case KGD_ENGINE_ME:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.me_fw->data;
+		break;
+
+	case KGD_ENGINE_CE:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.ce_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC1:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.mec_fw->data;
+		break;
+
+	case KGD_ENGINE_MEC2:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.mec2_fw->data;
+		break;
+
+	case KGD_ENGINE_RLC:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->gfx.rlc_fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA1:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->sdma.instance[0].fw->data;
+		break;
+
+	case KGD_ENGINE_SDMA2:
+		hdr = (const union amdgpu_firmware_header *)
+							adev->sdma.instance[1].fw->data;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (hdr == NULL)
+		return 0;
+
+	/* Only 12 bit in use*/
+	return hdr->common.ucode_version;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
new file mode 100644
index 000000000..9416e0f5c
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
@@ -0,0 +1,1602 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_i2c.h"
+
+#include "atom.h"
+#include "atom-bits.h"
+#include "atombios_encoders.h"
+#include "bif/bif_4_1_d.h"
+
+static void amdgpu_atombios_lookup_i2c_gpio_quirks(struct amdgpu_device *adev,
+					  ATOM_GPIO_I2C_ASSIGMENT *gpio,
+					  u8 index)
+{
+
+}
+
+static struct amdgpu_i2c_bus_rec amdgpu_atombios_get_bus_rec_for_i2c_gpio(ATOM_GPIO_I2C_ASSIGMENT *gpio)
+{
+	struct amdgpu_i2c_bus_rec i2c;
+
+	memset(&i2c, 0, sizeof(struct amdgpu_i2c_bus_rec));
+
+	i2c.mask_clk_reg = le16_to_cpu(gpio->usClkMaskRegisterIndex);
+	i2c.mask_data_reg = le16_to_cpu(gpio->usDataMaskRegisterIndex);
+	i2c.en_clk_reg = le16_to_cpu(gpio->usClkEnRegisterIndex);
+	i2c.en_data_reg = le16_to_cpu(gpio->usDataEnRegisterIndex);
+	i2c.y_clk_reg = le16_to_cpu(gpio->usClkY_RegisterIndex);
+	i2c.y_data_reg = le16_to_cpu(gpio->usDataY_RegisterIndex);
+	i2c.a_clk_reg = le16_to_cpu(gpio->usClkA_RegisterIndex);
+	i2c.a_data_reg = le16_to_cpu(gpio->usDataA_RegisterIndex);
+	i2c.mask_clk_mask = (1 << gpio->ucClkMaskShift);
+	i2c.mask_data_mask = (1 << gpio->ucDataMaskShift);
+	i2c.en_clk_mask = (1 << gpio->ucClkEnShift);
+	i2c.en_data_mask = (1 << gpio->ucDataEnShift);
+	i2c.y_clk_mask = (1 << gpio->ucClkY_Shift);
+	i2c.y_data_mask = (1 << gpio->ucDataY_Shift);
+	i2c.a_clk_mask = (1 << gpio->ucClkA_Shift);
+	i2c.a_data_mask = (1 << gpio->ucDataA_Shift);
+
+	if (gpio->sucI2cId.sbfAccess.bfHW_Capable)
+		i2c.hw_capable = true;
+	else
+		i2c.hw_capable = false;
+
+	if (gpio->sucI2cId.ucAccess == 0xa0)
+		i2c.mm_i2c = true;
+	else
+		i2c.mm_i2c = false;
+
+	i2c.i2c_id = gpio->sucI2cId.ucAccess;
+
+	if (i2c.mask_clk_reg)
+		i2c.valid = true;
+	else
+		i2c.valid = false;
+
+	return i2c;
+}
+
+struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *adev,
+							  uint8_t id)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	ATOM_GPIO_I2C_ASSIGMENT *gpio;
+	struct amdgpu_i2c_bus_rec i2c;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info);
+	struct _ATOM_GPIO_I2C_INFO *i2c_info;
+	uint16_t data_offset, size;
+	int i, num_indices;
+
+	memset(&i2c, 0, sizeof(struct amdgpu_i2c_bus_rec));
+	i2c.valid = false;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+
+		gpio = &i2c_info->asGPIO_Info[0];
+		for (i = 0; i < num_indices; i++) {
+
+			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
+
+			if (gpio->sucI2cId.ucAccess == id) {
+				i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
+				break;
+			}
+			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
+				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
+		}
+	}
+
+	return i2c;
+}
+
+void amdgpu_atombios_i2c_init(struct amdgpu_device *adev)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	ATOM_GPIO_I2C_ASSIGMENT *gpio;
+	struct amdgpu_i2c_bus_rec i2c;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_I2C_Info);
+	struct _ATOM_GPIO_I2C_INFO *i2c_info;
+	uint16_t data_offset, size;
+	int i, num_indices;
+	char stmp[32];
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		i2c_info = (struct _ATOM_GPIO_I2C_INFO *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_I2C_ASSIGMENT);
+
+		gpio = &i2c_info->asGPIO_Info[0];
+		for (i = 0; i < num_indices; i++) {
+			amdgpu_atombios_lookup_i2c_gpio_quirks(adev, gpio, i);
+
+			i2c = amdgpu_atombios_get_bus_rec_for_i2c_gpio(gpio);
+
+			if (i2c.valid) {
+				sprintf(stmp, "0x%x", i2c.i2c_id);
+				adev->i2c_bus[i] = amdgpu_i2c_create(adev->ddev, &i2c, stmp);
+			}
+			gpio = (ATOM_GPIO_I2C_ASSIGMENT *)
+				((u8 *)gpio + sizeof(ATOM_GPIO_I2C_ASSIGMENT));
+		}
+	}
+}
+
+struct amdgpu_gpio_rec
+amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
+			    u8 id)
+{
+	struct atom_context *ctx = adev->mode_info.atom_context;
+	struct amdgpu_gpio_rec gpio;
+	int index = GetIndexIntoMasterTable(DATA, GPIO_Pin_LUT);
+	struct _ATOM_GPIO_PIN_LUT *gpio_info;
+	ATOM_GPIO_PIN_ASSIGNMENT *pin;
+	u16 data_offset, size;
+	int i, num_indices;
+
+	memset(&gpio, 0, sizeof(struct amdgpu_gpio_rec));
+	gpio.valid = false;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, &size, NULL, NULL, &data_offset)) {
+		gpio_info = (struct _ATOM_GPIO_PIN_LUT *)(ctx->bios + data_offset);
+
+		num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+			sizeof(ATOM_GPIO_PIN_ASSIGNMENT);
+
+		pin = gpio_info->asGPIO_Pin;
+		for (i = 0; i < num_indices; i++) {
+			if (id == pin->ucGPIO_ID) {
+				gpio.id = pin->ucGPIO_ID;
+				gpio.reg = le16_to_cpu(pin->usGpioPin_AIndex);
+				gpio.shift = pin->ucGpioPinBitShift;
+				gpio.mask = (1 << pin->ucGpioPinBitShift);
+				gpio.valid = true;
+				break;
+			}
+			pin = (ATOM_GPIO_PIN_ASSIGNMENT *)
+				((u8 *)pin + sizeof(ATOM_GPIO_PIN_ASSIGNMENT));
+		}
+	}
+
+	return gpio;
+}
+
+static struct amdgpu_hpd
+amdgpu_atombios_get_hpd_info_from_gpio(struct amdgpu_device *adev,
+				       struct amdgpu_gpio_rec *gpio)
+{
+	struct amdgpu_hpd hpd;
+	u32 reg;
+
+	memset(&hpd, 0, sizeof(struct amdgpu_hpd));
+
+	reg = amdgpu_display_hpd_get_gpio_reg(adev);
+
+	hpd.gpio = *gpio;
+	if (gpio->reg == reg) {
+		switch(gpio->mask) {
+		case (1 << 0):
+			hpd.hpd = AMDGPU_HPD_1;
+			break;
+		case (1 << 8):
+			hpd.hpd = AMDGPU_HPD_2;
+			break;
+		case (1 << 16):
+			hpd.hpd = AMDGPU_HPD_3;
+			break;
+		case (1 << 24):
+			hpd.hpd = AMDGPU_HPD_4;
+			break;
+		case (1 << 26):
+			hpd.hpd = AMDGPU_HPD_5;
+			break;
+		case (1 << 28):
+			hpd.hpd = AMDGPU_HPD_6;
+			break;
+		default:
+			hpd.hpd = AMDGPU_HPD_NONE;
+			break;
+		}
+	} else
+		hpd.hpd = AMDGPU_HPD_NONE;
+	return hpd;
+}
+
+static bool amdgpu_atombios_apply_quirks(struct amdgpu_device *adev,
+					 uint32_t supported_device,
+					 int *connector_type,
+					 struct amdgpu_i2c_bus_rec *i2c_bus,
+					 uint16_t *line_mux,
+					 struct amdgpu_hpd *hpd)
+{
+	return true;
+}
+
+static const int object_connector_convert[] = {
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_DVII,
+	DRM_MODE_CONNECTOR_DVII,
+	DRM_MODE_CONNECTOR_DVID,
+	DRM_MODE_CONNECTOR_DVID,
+	DRM_MODE_CONNECTOR_VGA,
+	DRM_MODE_CONNECTOR_Composite,
+	DRM_MODE_CONNECTOR_SVIDEO,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_9PinDIN,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_HDMIA,
+	DRM_MODE_CONNECTOR_HDMIB,
+	DRM_MODE_CONNECTOR_LVDS,
+	DRM_MODE_CONNECTOR_9PinDIN,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_Unknown,
+	DRM_MODE_CONNECTOR_DisplayPort,
+	DRM_MODE_CONNECTOR_eDP,
+	DRM_MODE_CONNECTOR_Unknown
+};
+
+bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct atom_context *ctx = mode_info->atom_context;
+	int index = GetIndexIntoMasterTable(DATA, Object_Header);
+	u16 size, data_offset;
+	u8 frev, crev;
+	ATOM_CONNECTOR_OBJECT_TABLE *con_obj;
+	ATOM_ENCODER_OBJECT_TABLE *enc_obj;
+	ATOM_OBJECT_TABLE *router_obj;
+	ATOM_DISPLAY_OBJECT_PATH_TABLE *path_obj;
+	ATOM_OBJECT_HEADER *obj_header;
+	int i, j, k, path_size, device_support;
+	int connector_type;
+	u16 conn_id, connector_object_id;
+	struct amdgpu_i2c_bus_rec ddc_bus;
+	struct amdgpu_router router;
+	struct amdgpu_gpio_rec gpio;
+	struct amdgpu_hpd hpd;
+
+	if (!amdgpu_atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset))
+		return false;
+
+	if (crev < 2)
+		return false;
+
+	obj_header = (ATOM_OBJECT_HEADER *) (ctx->bios + data_offset);
+	path_obj = (ATOM_DISPLAY_OBJECT_PATH_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usDisplayPathTableOffset));
+	con_obj = (ATOM_CONNECTOR_OBJECT_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usConnectorObjectTableOffset));
+	enc_obj = (ATOM_ENCODER_OBJECT_TABLE *)
+	    (ctx->bios + data_offset +
+	     le16_to_cpu(obj_header->usEncoderObjectTableOffset));
+	router_obj = (ATOM_OBJECT_TABLE *)
+		(ctx->bios + data_offset +
+		 le16_to_cpu(obj_header->usRouterObjectTableOffset));
+	device_support = le16_to_cpu(obj_header->usDeviceSupport);
+
+	path_size = 0;
+	for (i = 0; i < path_obj->ucNumOfDispPath; i++) {
+		uint8_t *addr = (uint8_t *) path_obj->asDispPath;
+		ATOM_DISPLAY_OBJECT_PATH *path;
+		addr += path_size;
+		path = (ATOM_DISPLAY_OBJECT_PATH *) addr;
+		path_size += le16_to_cpu(path->usSize);
+
+		if (device_support & le16_to_cpu(path->usDeviceTag)) {
+			uint8_t con_obj_id, con_obj_num, con_obj_type;
+
+			con_obj_id =
+			    (le16_to_cpu(path->usConnObjectId) & OBJECT_ID_MASK)
+			    >> OBJECT_ID_SHIFT;
+			con_obj_num =
+			    (le16_to_cpu(path->usConnObjectId) & ENUM_ID_MASK)
+			    >> ENUM_ID_SHIFT;
+			con_obj_type =
+			    (le16_to_cpu(path->usConnObjectId) &
+			     OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT;
+
+			connector_type =
+				object_connector_convert[con_obj_id];
+			connector_object_id = con_obj_id;
+
+			if (connector_type == DRM_MODE_CONNECTOR_Unknown)
+				continue;
+
+			router.ddc_valid = false;
+			router.cd_valid = false;
+			for (j = 0; j < ((le16_to_cpu(path->usSize) - 8) / 2); j++) {
+				uint8_t grph_obj_id, grph_obj_num, grph_obj_type;
+
+				grph_obj_id =
+				    (le16_to_cpu(path->usGraphicObjIds[j]) &
+				     OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+				grph_obj_num =
+				    (le16_to_cpu(path->usGraphicObjIds[j]) &
+				     ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+				grph_obj_type =
+				    (le16_to_cpu(path->usGraphicObjIds[j]) &
+				     OBJECT_TYPE_MASK) >> OBJECT_TYPE_SHIFT;
+
+				if (grph_obj_type == GRAPH_OBJECT_TYPE_ENCODER) {
+					for (k = 0; k < enc_obj->ucNumberOfObjects; k++) {
+						u16 encoder_obj = le16_to_cpu(enc_obj->asObjects[k].usObjectID);
+						if (le16_to_cpu(path->usGraphicObjIds[j]) == encoder_obj) {
+							ATOM_COMMON_RECORD_HEADER *record = (ATOM_COMMON_RECORD_HEADER *)
+								(ctx->bios + data_offset +
+								 le16_to_cpu(enc_obj->asObjects[k].usRecordOffset));
+							ATOM_ENCODER_CAP_RECORD *cap_record;
+							u16 caps = 0;
+
+							while (record->ucRecordSize > 0 &&
+							       record->ucRecordType > 0 &&
+							       record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) {
+								switch (record->ucRecordType) {
+								case ATOM_ENCODER_CAP_RECORD_TYPE:
+									cap_record =(ATOM_ENCODER_CAP_RECORD *)
+										record;
+									caps = le16_to_cpu(cap_record->usEncoderCap);
+									break;
+								}
+								record = (ATOM_COMMON_RECORD_HEADER *)
+									((char *)record + record->ucRecordSize);
+							}
+							amdgpu_display_add_encoder(adev, encoder_obj,
+										    le16_to_cpu(path->usDeviceTag),
+										    caps);
+						}
+					}
+				} else if (grph_obj_type == GRAPH_OBJECT_TYPE_ROUTER) {
+					for (k = 0; k < router_obj->ucNumberOfObjects; k++) {
+						u16 router_obj_id = le16_to_cpu(router_obj->asObjects[k].usObjectID);
+						if (le16_to_cpu(path->usGraphicObjIds[j]) == router_obj_id) {
+							ATOM_COMMON_RECORD_HEADER *record = (ATOM_COMMON_RECORD_HEADER *)
+								(ctx->bios + data_offset +
+								 le16_to_cpu(router_obj->asObjects[k].usRecordOffset));
+							ATOM_I2C_RECORD *i2c_record;
+							ATOM_I2C_ID_CONFIG_ACCESS *i2c_config;
+							ATOM_ROUTER_DDC_PATH_SELECT_RECORD *ddc_path;
+							ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD *cd_path;
+							ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT *router_src_dst_table =
+								(ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT *)
+								(ctx->bios + data_offset +
+								 le16_to_cpu(router_obj->asObjects[k].usSrcDstTableOffset));
+							u8 *num_dst_objs = (u8 *)
+								((u8 *)router_src_dst_table + 1 +
+								 (router_src_dst_table->ucNumberOfSrc * 2));
+							u16 *dst_objs = (u16 *)(num_dst_objs + 1);
+							int enum_id;
+
+							router.router_id = router_obj_id;
+							for (enum_id = 0; enum_id < (*num_dst_objs); enum_id++) {
+								if (le16_to_cpu(path->usConnObjectId) ==
+								    le16_to_cpu(dst_objs[enum_id]))
+									break;
+							}
+
+							while (record->ucRecordSize > 0 &&
+							       record->ucRecordType > 0 &&
+							       record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) {
+								switch (record->ucRecordType) {
+								case ATOM_I2C_RECORD_TYPE:
+									i2c_record =
+										(ATOM_I2C_RECORD *)
+										record;
+									i2c_config =
+										(ATOM_I2C_ID_CONFIG_ACCESS *)
+										&i2c_record->sucI2cId;
+									router.i2c_info =
+										amdgpu_atombios_lookup_i2c_gpio(adev,
+												       i2c_config->
+												       ucAccess);
+									router.i2c_addr = i2c_record->ucI2CAddr >> 1;
+									break;
+								case ATOM_ROUTER_DDC_PATH_SELECT_RECORD_TYPE:
+									ddc_path = (ATOM_ROUTER_DDC_PATH_SELECT_RECORD *)
+										record;
+									router.ddc_valid = true;
+									router.ddc_mux_type = ddc_path->ucMuxType;
+									router.ddc_mux_control_pin = ddc_path->ucMuxControlPin;
+									router.ddc_mux_state = ddc_path->ucMuxState[enum_id];
+									break;
+								case ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD_TYPE:
+									cd_path = (ATOM_ROUTER_DATA_CLOCK_PATH_SELECT_RECORD *)
+										record;
+									router.cd_valid = true;
+									router.cd_mux_type = cd_path->ucMuxType;
+									router.cd_mux_control_pin = cd_path->ucMuxControlPin;
+									router.cd_mux_state = cd_path->ucMuxState[enum_id];
+									break;
+								}
+								record = (ATOM_COMMON_RECORD_HEADER *)
+									((char *)record + record->ucRecordSize);
+							}
+						}
+					}
+				}
+			}
+
+			/* look up gpio for ddc, hpd */
+			ddc_bus.valid = false;
+			hpd.hpd = AMDGPU_HPD_NONE;
+			if ((le16_to_cpu(path->usDeviceTag) &
+			     (ATOM_DEVICE_TV_SUPPORT | ATOM_DEVICE_CV_SUPPORT)) == 0) {
+				for (j = 0; j < con_obj->ucNumberOfObjects; j++) {
+					if (le16_to_cpu(path->usConnObjectId) ==
+					    le16_to_cpu(con_obj->asObjects[j].
+							usObjectID)) {
+						ATOM_COMMON_RECORD_HEADER
+						    *record =
+						    (ATOM_COMMON_RECORD_HEADER
+						     *)
+						    (ctx->bios + data_offset +
+						     le16_to_cpu(con_obj->
+								 asObjects[j].
+								 usRecordOffset));
+						ATOM_I2C_RECORD *i2c_record;
+						ATOM_HPD_INT_RECORD *hpd_record;
+						ATOM_I2C_ID_CONFIG_ACCESS *i2c_config;
+
+						while (record->ucRecordSize > 0 &&
+						       record->ucRecordType > 0 &&
+						       record->ucRecordType <= ATOM_MAX_OBJECT_RECORD_NUMBER) {
+							switch (record->ucRecordType) {
+							case ATOM_I2C_RECORD_TYPE:
+								i2c_record =
+								    (ATOM_I2C_RECORD *)
+									record;
+								i2c_config =
+									(ATOM_I2C_ID_CONFIG_ACCESS *)
+									&i2c_record->sucI2cId;
+								ddc_bus = amdgpu_atombios_lookup_i2c_gpio(adev,
+												 i2c_config->
+												 ucAccess);
+								break;
+							case ATOM_HPD_INT_RECORD_TYPE:
+								hpd_record =
+									(ATOM_HPD_INT_RECORD *)
+									record;
+								gpio = amdgpu_atombios_lookup_gpio(adev,
+											  hpd_record->ucHPDIntGPIOID);
+								hpd = amdgpu_atombios_get_hpd_info_from_gpio(adev, &gpio);
+								hpd.plugged_state = hpd_record->ucPlugged_PinState;
+								break;
+							}
+							record =
+							    (ATOM_COMMON_RECORD_HEADER
+							     *) ((char *)record
+								 +
+								 record->
+								 ucRecordSize);
+						}
+						break;
+					}
+				}
+			}
+
+			/* needed for aux chan transactions */
+			ddc_bus.hpd = hpd.hpd;
+
+			conn_id = le16_to_cpu(path->usConnObjectId);
+
+			if (!amdgpu_atombios_apply_quirks
+			    (adev, le16_to_cpu(path->usDeviceTag), &connector_type,
+			     &ddc_bus, &conn_id, &hpd))
+				continue;
+
+			amdgpu_display_add_connector(adev,
+						      conn_id,
+						      le16_to_cpu(path->usDeviceTag),
+						      connector_type, &ddc_bus,
+						      connector_object_id,
+						      &hpd,
+						      &router);
+
+		}
+	}
+
+	amdgpu_link_encoder_connector(adev->ddev);
+
+	return true;
+}
+
+union firmware_info {
+	ATOM_FIRMWARE_INFO info;
+	ATOM_FIRMWARE_INFO_V1_2 info_12;
+	ATOM_FIRMWARE_INFO_V1_3 info_13;
+	ATOM_FIRMWARE_INFO_V1_4 info_14;
+	ATOM_FIRMWARE_INFO_V2_1 info_21;
+	ATOM_FIRMWARE_INFO_V2_2 info_22;
+};
+
+int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
+	uint8_t frev, crev;
+	uint16_t data_offset;
+	int ret = -EINVAL;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		int i;
+		struct amdgpu_pll *ppll = &adev->clock.ppll[0];
+		struct amdgpu_pll *spll = &adev->clock.spll;
+		struct amdgpu_pll *mpll = &adev->clock.mpll;
+		union firmware_info *firmware_info =
+			(union firmware_info *)(mode_info->atom_context->bios +
+						data_offset);
+		/* pixel clocks */
+		ppll->reference_freq =
+		    le16_to_cpu(firmware_info->info.usReferenceClock);
+		ppll->reference_div = 0;
+
+		if (crev < 2)
+			ppll->pll_out_min =
+				le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Output);
+		else
+			ppll->pll_out_min =
+				le32_to_cpu(firmware_info->info_12.ulMinPixelClockPLL_Output);
+		ppll->pll_out_max =
+		    le32_to_cpu(firmware_info->info.ulMaxPixelClockPLL_Output);
+
+		if (crev >= 4) {
+			ppll->lcd_pll_out_min =
+				le16_to_cpu(firmware_info->info_14.usLcdMinPixelClockPLL_Output) * 100;
+			if (ppll->lcd_pll_out_min == 0)
+				ppll->lcd_pll_out_min = ppll->pll_out_min;
+			ppll->lcd_pll_out_max =
+				le16_to_cpu(firmware_info->info_14.usLcdMaxPixelClockPLL_Output) * 100;
+			if (ppll->lcd_pll_out_max == 0)
+				ppll->lcd_pll_out_max = ppll->pll_out_max;
+		} else {
+			ppll->lcd_pll_out_min = ppll->pll_out_min;
+			ppll->lcd_pll_out_max = ppll->pll_out_max;
+		}
+
+		if (ppll->pll_out_min == 0)
+			ppll->pll_out_min = 64800;
+
+		ppll->pll_in_min =
+		    le16_to_cpu(firmware_info->info.usMinPixelClockPLL_Input);
+		ppll->pll_in_max =
+		    le16_to_cpu(firmware_info->info.usMaxPixelClockPLL_Input);
+
+		ppll->min_post_div = 2;
+		ppll->max_post_div = 0x7f;
+		ppll->min_frac_feedback_div = 0;
+		ppll->max_frac_feedback_div = 9;
+		ppll->min_ref_div = 2;
+		ppll->max_ref_div = 0x3ff;
+		ppll->min_feedback_div = 4;
+		ppll->max_feedback_div = 0xfff;
+		ppll->best_vco = 0;
+
+		for (i = 1; i < AMDGPU_MAX_PPLL; i++)
+			adev->clock.ppll[i] = *ppll;
+
+		/* system clock */
+		spll->reference_freq =
+			le16_to_cpu(firmware_info->info_21.usCoreReferenceClock);
+		spll->reference_div = 0;
+
+		spll->pll_out_min =
+		    le16_to_cpu(firmware_info->info.usMinEngineClockPLL_Output);
+		spll->pll_out_max =
+		    le32_to_cpu(firmware_info->info.ulMaxEngineClockPLL_Output);
+
+		/* ??? */
+		if (spll->pll_out_min == 0)
+			spll->pll_out_min = 64800;
+
+		spll->pll_in_min =
+		    le16_to_cpu(firmware_info->info.usMinEngineClockPLL_Input);
+		spll->pll_in_max =
+		    le16_to_cpu(firmware_info->info.usMaxEngineClockPLL_Input);
+
+		spll->min_post_div = 1;
+		spll->max_post_div = 1;
+		spll->min_ref_div = 2;
+		spll->max_ref_div = 0xff;
+		spll->min_feedback_div = 4;
+		spll->max_feedback_div = 0xff;
+		spll->best_vco = 0;
+
+		/* memory clock */
+		mpll->reference_freq =
+			le16_to_cpu(firmware_info->info_21.usMemoryReferenceClock);
+		mpll->reference_div = 0;
+
+		mpll->pll_out_min =
+		    le16_to_cpu(firmware_info->info.usMinMemoryClockPLL_Output);
+		mpll->pll_out_max =
+		    le32_to_cpu(firmware_info->info.ulMaxMemoryClockPLL_Output);
+
+		/* ??? */
+		if (mpll->pll_out_min == 0)
+			mpll->pll_out_min = 64800;
+
+		mpll->pll_in_min =
+		    le16_to_cpu(firmware_info->info.usMinMemoryClockPLL_Input);
+		mpll->pll_in_max =
+		    le16_to_cpu(firmware_info->info.usMaxMemoryClockPLL_Input);
+
+		adev->clock.default_sclk =
+		    le32_to_cpu(firmware_info->info.ulDefaultEngineClock);
+		adev->clock.default_mclk =
+		    le32_to_cpu(firmware_info->info.ulDefaultMemoryClock);
+
+		mpll->min_post_div = 1;
+		mpll->max_post_div = 1;
+		mpll->min_ref_div = 2;
+		mpll->max_ref_div = 0xff;
+		mpll->min_feedback_div = 4;
+		mpll->max_feedback_div = 0xff;
+		mpll->best_vco = 0;
+
+		/* disp clock */
+		adev->clock.default_dispclk =
+			le32_to_cpu(firmware_info->info_21.ulDefaultDispEngineClkFreq);
+		/* set a reasonable default for DP */
+		if (adev->clock.default_dispclk < 53900) {
+			DRM_INFO("Changing default dispclk from %dMhz to 600Mhz\n",
+				 adev->clock.default_dispclk / 100);
+			adev->clock.default_dispclk = 60000;
+		}
+		adev->clock.dp_extclk =
+			le16_to_cpu(firmware_info->info_21.usUniphyDPModeExtClkFreq);
+		adev->clock.current_dispclk = adev->clock.default_dispclk;
+
+		adev->clock.max_pixel_clock = le16_to_cpu(firmware_info->info.usMaxPixelClock);
+		if (adev->clock.max_pixel_clock == 0)
+			adev->clock.max_pixel_clock = 40000;
+
+		/* not technically a clock, but... */
+		adev->mode_info.firmware_flags =
+			le16_to_cpu(firmware_info->info.usFirmwareCapability.susAccess);
+
+		ret = 0;
+	}
+
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+
+	return ret;
+}
+
+union igp_info {
+	struct _ATOM_INTEGRATED_SYSTEM_INFO info;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 info_2;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 info_6;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 info_7;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_8 info_8;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_9 info_9;
+};
+
+static void amdgpu_atombios_get_igp_ss_overrides(struct amdgpu_device *adev,
+						 struct amdgpu_atom_ss *ss,
+						 int id)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo);
+	u16 data_offset, size;
+	union igp_info *igp_info;
+	u8 frev, crev;
+	u16 percentage = 0, rate = 0;
+
+	/* get any igp specific overrides */
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)
+			(mode_info->atom_context->bios + data_offset);
+		switch (crev) {
+		case 6:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_6.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_6.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_6.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_6.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_6.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_6.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		case 7:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_7.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_7.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_7.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_7.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_7.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_7.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		case 8:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_8.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_8.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_8.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_8.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_8.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_8.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		case 9:
+			switch (id) {
+			case ASIC_INTERNAL_SS_ON_TMDS:
+				percentage = le16_to_cpu(igp_info->info_9.usDVISSPercentage);
+				rate = le16_to_cpu(igp_info->info_9.usDVISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_HDMI:
+				percentage = le16_to_cpu(igp_info->info_9.usHDMISSPercentage);
+				rate = le16_to_cpu(igp_info->info_9.usHDMISSpreadRateIn10Hz);
+				break;
+			case ASIC_INTERNAL_SS_ON_LVDS:
+				percentage = le16_to_cpu(igp_info->info_9.usLvdsSSPercentage);
+				rate = le16_to_cpu(igp_info->info_9.usLvdsSSpreadRateIn10Hz);
+				break;
+			}
+			break;
+		default:
+			DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev);
+			break;
+		}
+		if (percentage)
+			ss->percentage = percentage;
+		if (rate)
+			ss->rate = rate;
+	}
+}
+
+union asic_ss_info {
+	struct _ATOM_ASIC_INTERNAL_SS_INFO info;
+	struct _ATOM_ASIC_INTERNAL_SS_INFO_V2 info_2;
+	struct _ATOM_ASIC_INTERNAL_SS_INFO_V3 info_3;
+};
+
+union asic_ss_assignment {
+	struct _ATOM_ASIC_SS_ASSIGNMENT v1;
+	struct _ATOM_ASIC_SS_ASSIGNMENT_V2 v2;
+	struct _ATOM_ASIC_SS_ASSIGNMENT_V3 v3;
+};
+
+bool amdgpu_atombios_get_asic_ss_info(struct amdgpu_device *adev,
+				      struct amdgpu_atom_ss *ss,
+				      int id, u32 clock)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info);
+	uint16_t data_offset, size;
+	union asic_ss_info *ss_info;
+	union asic_ss_assignment *ss_assign;
+	uint8_t frev, crev;
+	int i, num_indices;
+
+	if (id == ASIC_INTERNAL_MEMORY_SS) {
+		if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_MEMORY_CLOCK_SS_SUPPORT))
+			return false;
+	}
+	if (id == ASIC_INTERNAL_ENGINE_SS) {
+		if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_ENGINE_CLOCK_SS_SUPPORT))
+			return false;
+	}
+
+	memset(ss, 0, sizeof(struct amdgpu_atom_ss));
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+
+		ss_info =
+			(union asic_ss_info *)(mode_info->atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 1:
+			num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+				sizeof(ATOM_ASIC_SS_ASSIGNMENT);
+
+			ss_assign = (union asic_ss_assignment *)((u8 *)&ss_info->info.asSpreadSpectrum[0]);
+			for (i = 0; i < num_indices; i++) {
+				if ((ss_assign->v1.ucClockIndication == id) &&
+				    (clock <= le32_to_cpu(ss_assign->v1.ulTargetClockRange))) {
+					ss->percentage =
+						le16_to_cpu(ss_assign->v1.usSpreadSpectrumPercentage);
+					ss->type = ss_assign->v1.ucSpreadSpectrumMode;
+					ss->rate = le16_to_cpu(ss_assign->v1.usSpreadRateInKhz);
+					ss->percentage_divider = 100;
+					return true;
+				}
+				ss_assign = (union asic_ss_assignment *)
+					((u8 *)ss_assign + sizeof(ATOM_ASIC_SS_ASSIGNMENT));
+			}
+			break;
+		case 2:
+			num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+				sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2);
+			ss_assign = (union asic_ss_assignment *)((u8 *)&ss_info->info_2.asSpreadSpectrum[0]);
+			for (i = 0; i < num_indices; i++) {
+				if ((ss_assign->v2.ucClockIndication == id) &&
+				    (clock <= le32_to_cpu(ss_assign->v2.ulTargetClockRange))) {
+					ss->percentage =
+						le16_to_cpu(ss_assign->v2.usSpreadSpectrumPercentage);
+					ss->type = ss_assign->v2.ucSpreadSpectrumMode;
+					ss->rate = le16_to_cpu(ss_assign->v2.usSpreadRateIn10Hz);
+					ss->percentage_divider = 100;
+					if ((crev == 2) &&
+					    ((id == ASIC_INTERNAL_ENGINE_SS) ||
+					     (id == ASIC_INTERNAL_MEMORY_SS)))
+						ss->rate /= 100;
+					return true;
+				}
+				ss_assign = (union asic_ss_assignment *)
+					((u8 *)ss_assign + sizeof(ATOM_ASIC_SS_ASSIGNMENT_V2));
+			}
+			break;
+		case 3:
+			num_indices = (size - sizeof(ATOM_COMMON_TABLE_HEADER)) /
+				sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3);
+			ss_assign = (union asic_ss_assignment *)((u8 *)&ss_info->info_3.asSpreadSpectrum[0]);
+			for (i = 0; i < num_indices; i++) {
+				if ((ss_assign->v3.ucClockIndication == id) &&
+				    (clock <= le32_to_cpu(ss_assign->v3.ulTargetClockRange))) {
+					ss->percentage =
+						le16_to_cpu(ss_assign->v3.usSpreadSpectrumPercentage);
+					ss->type = ss_assign->v3.ucSpreadSpectrumMode;
+					ss->rate = le16_to_cpu(ss_assign->v3.usSpreadRateIn10Hz);
+					if (ss_assign->v3.ucSpreadSpectrumMode &
+					    SS_MODE_V3_PERCENTAGE_DIV_BY_1000_MASK)
+						ss->percentage_divider = 1000;
+					else
+						ss->percentage_divider = 100;
+					if ((id == ASIC_INTERNAL_ENGINE_SS) ||
+					    (id == ASIC_INTERNAL_MEMORY_SS))
+						ss->rate /= 100;
+					if (adev->flags & AMD_IS_APU)
+						amdgpu_atombios_get_igp_ss_overrides(adev, ss, id);
+					return true;
+				}
+				ss_assign = (union asic_ss_assignment *)
+					((u8 *)ss_assign + sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3));
+			}
+			break;
+		default:
+			DRM_ERROR("Unsupported ASIC_InternalSS_Info table: %d %d\n", frev, crev);
+			break;
+		}
+
+	}
+	return false;
+}
+
+union get_clock_dividers {
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS v1;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V2 v2;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 v3;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V4 v4;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 v5;
+	struct _COMPUTE_GPU_CLOCK_INPUT_PARAMETERS_V1_6 v6_in;
+	struct _COMPUTE_GPU_CLOCK_OUTPUT_PARAMETERS_V1_6 v6_out;
+};
+
+int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
+				       u8 clock_type,
+				       u32 clock,
+				       bool strobe_mode,
+				       struct atom_clock_dividers *dividers)
+{
+	union get_clock_dividers args;
+	int index = GetIndexIntoMasterTable(COMMAND, ComputeMemoryEnginePLL);
+	u8 frev, crev;
+
+	memset(&args, 0, sizeof(args));
+	memset(dividers, 0, sizeof(struct atom_clock_dividers));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 4:
+		/* fusion */
+		args.v4.ulClock = cpu_to_le32(clock);	/* 10 khz */
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		dividers->post_divider = dividers->post_div = args.v4.ucPostDiv;
+		dividers->real_clock = le32_to_cpu(args.v4.ulClock);
+		break;
+	case 6:
+		/* CI */
+		/* COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK, COMPUTE_GPUCLK_INPUT_FLAG_SCLK */
+		args.v6_in.ulClock.ulComputeClockFlag = clock_type;
+		args.v6_in.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		dividers->whole_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDiv);
+		dividers->frac_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDivFrac);
+		dividers->ref_div = args.v6_out.ucPllRefDiv;
+		dividers->post_div = args.v6_out.ucPllPostDiv;
+		dividers->flags = args.v6_out.ucPllCntlFlag;
+		dividers->real_clock = le32_to_cpu(args.v6_out.ulClock.ulClock);
+		dividers->post_divider = args.v6_out.ulClock.ucPostDiv;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
+					    u32 clock,
+					    bool strobe_mode,
+					    struct atom_mpll_param *mpll_param)
+{
+	COMPUTE_MEMORY_CLOCK_PARAM_PARAMETERS_V2_1 args;
+	int index = GetIndexIntoMasterTable(COMMAND, ComputeMemoryClockParam);
+	u8 frev, crev;
+
+	memset(&args, 0, sizeof(args));
+	memset(mpll_param, 0, sizeof(struct atom_mpll_param));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (frev) {
+	case 2:
+		switch (crev) {
+		case 1:
+			/* SI */
+			args.ulClock = cpu_to_le32(clock);	/* 10 khz */
+			args.ucInputFlag = 0;
+			if (strobe_mode)
+				args.ucInputFlag |= MPLL_INPUT_FLAG_STROBE_MODE_EN;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			mpll_param->clkfrac = le16_to_cpu(args.ulFbDiv.usFbDivFrac);
+			mpll_param->clkf = le16_to_cpu(args.ulFbDiv.usFbDiv);
+			mpll_param->post_div = args.ucPostDiv;
+			mpll_param->dll_speed = args.ucDllSpeed;
+			mpll_param->bwcntl = args.ucBWCntl;
+			mpll_param->vco_mode =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_VCO_MODE_MASK);
+			mpll_param->yclk_sel =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_BYPASS_DQ_PLL) ? 1 : 0;
+			mpll_param->qdr =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_QDR_ENABLE) ? 1 : 0;
+			mpll_param->half_rate =
+				(args.ucPllCntlFlag & MPLL_CNTL_FLAG_AD_HALF_RATE) ? 1 : 0;
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+uint32_t amdgpu_atombios_get_engine_clock(struct amdgpu_device *adev)
+{
+	GET_ENGINE_CLOCK_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, GetEngineClock);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	return le32_to_cpu(args.ulReturnEngineClock);
+}
+
+uint32_t amdgpu_atombios_get_memory_clock(struct amdgpu_device *adev)
+{
+	GET_MEMORY_CLOCK_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, GetMemoryClock);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	return le32_to_cpu(args.ulReturnMemoryClock);
+}
+
+void amdgpu_atombios_set_engine_clock(struct amdgpu_device *adev,
+				      uint32_t eng_clock)
+{
+	SET_ENGINE_CLOCK_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetEngineClock);
+
+	args.ulTargetEngineClock = cpu_to_le32(eng_clock);	/* 10 khz */
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_set_memory_clock(struct amdgpu_device *adev,
+				      uint32_t mem_clock)
+{
+	SET_MEMORY_CLOCK_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetMemoryClock);
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	args.ulTargetMemoryClock = cpu_to_le32(mem_clock);	/* 10 khz */
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
+					     u32 eng_clock, u32 mem_clock)
+{
+	SET_ENGINE_CLOCK_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, DynamicMemorySettings);
+	u32 tmp;
+
+	memset(&args, 0, sizeof(args));
+
+	tmp = eng_clock & SET_CLOCK_FREQ_MASK;
+	tmp |= (COMPUTE_ENGINE_PLL_PARAM << 24);
+
+	args.ulTargetEngineClock = cpu_to_le32(tmp);
+	if (mem_clock)
+		args.sReserved.ulClock = cpu_to_le32(mem_clock & SET_CLOCK_FREQ_MASK);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union set_voltage {
+	struct _SET_VOLTAGE_PS_ALLOCATION alloc;
+	struct _SET_VOLTAGE_PARAMETERS v1;
+	struct _SET_VOLTAGE_PARAMETERS_V2 v2;
+	struct _SET_VOLTAGE_PARAMETERS_V1_3 v3;
+};
+
+void amdgpu_atombios_set_voltage(struct amdgpu_device *adev,
+				 u16 voltage_level,
+				 u8 voltage_type)
+{
+	union set_voltage args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetVoltage);
+	u8 frev, crev, volt_index = voltage_level;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	/* 0xff01 is a flag rather then an actual voltage */
+	if (voltage_level == 0xff01)
+		return;
+
+	switch (crev) {
+	case 1:
+		args.v1.ucVoltageType = voltage_type;
+		args.v1.ucVoltageMode = SET_ASIC_VOLTAGE_MODE_ALL_SOURCE;
+		args.v1.ucVoltageIndex = volt_index;
+		break;
+	case 2:
+		args.v2.ucVoltageType = voltage_type;
+		args.v2.ucVoltageMode = SET_ASIC_VOLTAGE_MODE_SET_VOLTAGE;
+		args.v2.usVoltageLevel = cpu_to_le16(voltage_level);
+		break;
+	case 3:
+		args.v3.ucVoltageType = voltage_type;
+		args.v3.ucVoltageMode = ATOM_SET_VOLTAGE;
+		args.v3.usVoltageLevel = cpu_to_le16(voltage_level);
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev,
+					      u16 *leakage_id)
+{
+	union set_voltage args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetVoltage);
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 3:
+	case 4:
+		args.v3.ucVoltageType = 0;
+		args.v3.ucVoltageMode = ATOM_GET_LEAKAGE_ID;
+		args.v3.usVoltageLevel = 0;
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		*leakage_id = le16_to_cpu(args.v3.usVoltageLevel);
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev,
+							     u16 *vddc, u16 *vddci,
+							     u16 virtual_voltage_id,
+							     u16 vbios_voltage_id)
+{
+	int index = GetIndexIntoMasterTable(DATA, ASIC_ProfilingInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	int i, j;
+	ATOM_ASIC_PROFILING_INFO_V2_1 *profile;
+	u16 *leakage_bin, *vddc_id_buf, *vddc_buf, *vddci_id_buf, *vddci_buf;
+
+	*vddc = 0;
+	*vddci = 0;
+
+	if (!amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				    &frev, &crev, &data_offset))
+		return -EINVAL;
+
+	profile = (ATOM_ASIC_PROFILING_INFO_V2_1 *)
+		(adev->mode_info.atom_context->bios + data_offset);
+
+	switch (frev) {
+	case 1:
+		return -EINVAL;
+	case 2:
+		switch (crev) {
+		case 1:
+			if (size < sizeof(ATOM_ASIC_PROFILING_INFO_V2_1))
+				return -EINVAL;
+			leakage_bin = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usLeakageBinArrayOffset));
+			vddc_id_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDC_IdArrayOffset));
+			vddc_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDC_LevelArrayOffset));
+			vddci_id_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDCI_IdArrayOffset));
+			vddci_buf = (u16 *)
+				(adev->mode_info.atom_context->bios + data_offset +
+				 le16_to_cpu(profile->usElbVDDCI_LevelArrayOffset));
+
+			if (profile->ucElbVDDC_Num > 0) {
+				for (i = 0; i < profile->ucElbVDDC_Num; i++) {
+					if (vddc_id_buf[i] == virtual_voltage_id) {
+						for (j = 0; j < profile->ucLeakageBinNum; j++) {
+							if (vbios_voltage_id <= leakage_bin[j]) {
+								*vddc = vddc_buf[j * profile->ucElbVDDC_Num + i];
+								break;
+							}
+						}
+						break;
+					}
+				}
+			}
+			if (profile->ucElbVDDCI_Num > 0) {
+				for (i = 0; i < profile->ucElbVDDCI_Num; i++) {
+					if (vddci_id_buf[i] == virtual_voltage_id) {
+						for (j = 0; j < profile->ucLeakageBinNum; j++) {
+							if (vbios_voltage_id <= leakage_bin[j]) {
+								*vddci = vddci_buf[j * profile->ucElbVDDCI_Num + i];
+								break;
+							}
+						}
+						break;
+					}
+				}
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			return -EINVAL;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+union get_voltage_info {
+	struct _GET_VOLTAGE_INFO_INPUT_PARAMETER_V1_2 in;
+	struct _GET_EVV_VOLTAGE_INFO_OUTPUT_PARAMETER_V1_2 evv_out;
+};
+
+int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev,
+				    u16 virtual_voltage_id,
+				    u16 *voltage)
+{
+	int index = GetIndexIntoMasterTable(COMMAND, GetVoltageInfo);
+	u32 entry_id;
+	u32 count = adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count;
+	union get_voltage_info args;
+
+	for (entry_id = 0; entry_id < count; entry_id++) {
+		if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].v ==
+		    virtual_voltage_id)
+			break;
+	}
+
+	if (entry_id >= count)
+		return -EINVAL;
+
+	args.in.ucVoltageType = VOLTAGE_TYPE_VDDC;
+	args.in.ucVoltageMode = ATOM_GET_VOLTAGE_EVV_VOLTAGE;
+	args.in.usVoltageLevel = cpu_to_le16(virtual_voltage_id);
+	args.in.ulSCLKFreq =
+		cpu_to_le32(adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[entry_id].clk);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	*voltage = le16_to_cpu(args.evv_out.usVoltageLevel);
+
+	return 0;
+}
+
+union voltage_object_info {
+	struct _ATOM_VOLTAGE_OBJECT_INFO v1;
+	struct _ATOM_VOLTAGE_OBJECT_INFO_V2 v2;
+	struct _ATOM_VOLTAGE_OBJECT_INFO_V3_1 v3;
+};
+
+union voltage_object {
+	struct _ATOM_VOLTAGE_OBJECT v1;
+	struct _ATOM_VOLTAGE_OBJECT_V2 v2;
+	union _ATOM_VOLTAGE_OBJECT_V3 v3;
+};
+
+
+static ATOM_VOLTAGE_OBJECT_V3 *amdgpu_atombios_lookup_voltage_object_v3(ATOM_VOLTAGE_OBJECT_INFO_V3_1 *v3,
+									u8 voltage_type, u8 voltage_mode)
+{
+	u32 size = le16_to_cpu(v3->sHeader.usStructureSize);
+	u32 offset = offsetof(ATOM_VOLTAGE_OBJECT_INFO_V3_1, asVoltageObj[0]);
+	u8 *start = (u8*)v3;
+
+	while (offset < size) {
+		ATOM_VOLTAGE_OBJECT_V3 *vo = (ATOM_VOLTAGE_OBJECT_V3 *)(start + offset);
+		if ((vo->asGpioVoltageObj.sHeader.ucVoltageType == voltage_type) &&
+		    (vo->asGpioVoltageObj.sHeader.ucVoltageMode == voltage_mode))
+			return vo;
+		offset += le16_to_cpu(vo->asGpioVoltageObj.sHeader.usSize);
+	}
+	return NULL;
+}
+
+bool
+amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev,
+				u8 voltage_type, u8 voltage_mode)
+{
+	int index = GetIndexIntoMasterTable(DATA, VoltageObjectInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	union voltage_object_info *voltage_info;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		voltage_info = (union voltage_object_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 3:
+			switch (crev) {
+			case 1:
+				if (amdgpu_atombios_lookup_voltage_object_v3(&voltage_info->v3,
+								  voltage_type, voltage_mode))
+					return true;
+				break;
+			default:
+				DRM_ERROR("unknown voltage object table\n");
+				return false;
+			}
+			break;
+		default:
+			DRM_ERROR("unknown voltage object table\n");
+			return false;
+		}
+
+	}
+	return false;
+}
+
+int amdgpu_atombios_get_voltage_table(struct amdgpu_device *adev,
+				      u8 voltage_type, u8 voltage_mode,
+				      struct atom_voltage_table *voltage_table)
+{
+	int index = GetIndexIntoMasterTable(DATA, VoltageObjectInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	int i;
+	union voltage_object_info *voltage_info;
+	union voltage_object *voltage_object = NULL;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		voltage_info = (union voltage_object_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 3:
+			switch (crev) {
+			case 1:
+				voltage_object = (union voltage_object *)
+					amdgpu_atombios_lookup_voltage_object_v3(&voltage_info->v3,
+								      voltage_type, voltage_mode);
+				if (voltage_object) {
+					ATOM_GPIO_VOLTAGE_OBJECT_V3 *gpio =
+						&voltage_object->v3.asGpioVoltageObj;
+					VOLTAGE_LUT_ENTRY_V2 *lut;
+					if (gpio->ucGpioEntryNum > MAX_VOLTAGE_ENTRIES)
+						return -EINVAL;
+					lut = &gpio->asVolGpioLut[0];
+					for (i = 0; i < gpio->ucGpioEntryNum; i++) {
+						voltage_table->entries[i].value =
+							le16_to_cpu(lut->usVoltageValue);
+						voltage_table->entries[i].smio_low =
+							le32_to_cpu(lut->ulVoltageId);
+						lut = (VOLTAGE_LUT_ENTRY_V2 *)
+							((u8 *)lut + sizeof(VOLTAGE_LUT_ENTRY_V2));
+					}
+					voltage_table->mask_low = le32_to_cpu(gpio->ulGpioMaskVal);
+					voltage_table->count = gpio->ucGpioEntryNum;
+					voltage_table->phase_delay = gpio->ucPhaseDelay;
+					return 0;
+				}
+				break;
+			default:
+				DRM_ERROR("unknown voltage object table\n");
+				return -EINVAL;
+			}
+			break;
+		default:
+			DRM_ERROR("unknown voltage object table\n");
+			return -EINVAL;
+		}
+	}
+	return -EINVAL;
+}
+
+union vram_info {
+	struct _ATOM_VRAM_INFO_V3 v1_3;
+	struct _ATOM_VRAM_INFO_V4 v1_4;
+	struct _ATOM_VRAM_INFO_HEADER_V2_1 v2_1;
+};
+
+#define MEM_ID_MASK           0xff000000
+#define MEM_ID_SHIFT          24
+#define CLOCK_RANGE_MASK      0x00ffffff
+#define CLOCK_RANGE_SHIFT     0
+#define LOW_NIBBLE_MASK       0xf
+#define DATA_EQU_PREV         0
+#define DATA_FROM_TABLE       4
+
+int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
+				      u8 module_index,
+				      struct atom_mc_reg_table *reg_table)
+{
+	int index = GetIndexIntoMasterTable(DATA, VRAM_Info);
+	u8 frev, crev, num_entries, t_mem_id, num_ranges = 0;
+	u32 i = 0, j;
+	u16 data_offset, size;
+	union vram_info *vram_info;
+
+	memset(reg_table, 0, sizeof(struct atom_mc_reg_table));
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		vram_info = (union vram_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+		switch (frev) {
+		case 1:
+			DRM_ERROR("old table version %d, %d\n", frev, crev);
+			return -EINVAL;
+		case 2:
+			switch (crev) {
+			case 1:
+				if (module_index < vram_info->v2_1.ucNumOfVRAMModule) {
+					ATOM_INIT_REG_BLOCK *reg_block =
+						(ATOM_INIT_REG_BLOCK *)
+						((u8 *)vram_info + le16_to_cpu(vram_info->v2_1.usMemClkPatchTblOffset));
+					ATOM_MEMORY_SETTING_DATA_BLOCK *reg_data =
+						(ATOM_MEMORY_SETTING_DATA_BLOCK *)
+						((u8 *)reg_block + (2 * sizeof(u16)) +
+						 le16_to_cpu(reg_block->usRegIndexTblSize));
+					ATOM_INIT_REG_INDEX_FORMAT *format = &reg_block->asRegIndexBuf[0];
+					num_entries = (u8)((le16_to_cpu(reg_block->usRegIndexTblSize)) /
+							   sizeof(ATOM_INIT_REG_INDEX_FORMAT)) - 1;
+					if (num_entries > VBIOS_MC_REGISTER_ARRAY_SIZE)
+						return -EINVAL;
+					while (i < num_entries) {
+						if (format->ucPreRegDataLength & ACCESS_PLACEHOLDER)
+							break;
+						reg_table->mc_reg_address[i].s1 =
+							(u16)(le16_to_cpu(format->usRegIndex));
+						reg_table->mc_reg_address[i].pre_reg_data =
+							(u8)(format->ucPreRegDataLength);
+						i++;
+						format = (ATOM_INIT_REG_INDEX_FORMAT *)
+							((u8 *)format + sizeof(ATOM_INIT_REG_INDEX_FORMAT));
+					}
+					reg_table->last = i;
+					while ((le32_to_cpu(*(u32 *)reg_data) != END_OF_REG_DATA_BLOCK) &&
+					       (num_ranges < VBIOS_MAX_AC_TIMING_ENTRIES)) {
+						t_mem_id = (u8)((le32_to_cpu(*(u32 *)reg_data) & MEM_ID_MASK)
+								>> MEM_ID_SHIFT);
+						if (module_index == t_mem_id) {
+							reg_table->mc_reg_table_entry[num_ranges].mclk_max =
+								(u32)((le32_to_cpu(*(u32 *)reg_data) & CLOCK_RANGE_MASK)
+								      >> CLOCK_RANGE_SHIFT);
+							for (i = 0, j = 1; i < reg_table->last; i++) {
+								if ((reg_table->mc_reg_address[i].pre_reg_data & LOW_NIBBLE_MASK) == DATA_FROM_TABLE) {
+									reg_table->mc_reg_table_entry[num_ranges].mc_data[i] =
+										(u32)le32_to_cpu(*((u32 *)reg_data + j));
+									j++;
+								} else if ((reg_table->mc_reg_address[i].pre_reg_data & LOW_NIBBLE_MASK) == DATA_EQU_PREV) {
+									reg_table->mc_reg_table_entry[num_ranges].mc_data[i] =
+										reg_table->mc_reg_table_entry[num_ranges].mc_data[i - 1];
+								}
+							}
+							num_ranges++;
+						}
+						reg_data = (ATOM_MEMORY_SETTING_DATA_BLOCK *)
+							((u8 *)reg_data + le16_to_cpu(reg_block->usRegDataBlkSize));
+					}
+					if (le32_to_cpu(*(u32 *)reg_data) != END_OF_REG_DATA_BLOCK)
+						return -EINVAL;
+					reg_table->num_entries = num_ranges;
+				} else
+					return -EINVAL;
+				break;
+			default:
+				DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+				return -EINVAL;
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			return -EINVAL;
+		}
+		return 0;
+	}
+	return -EINVAL;
+}
+
+void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock)
+{
+	uint32_t bios_6_scratch;
+
+	bios_6_scratch = RREG32(mmBIOS_SCRATCH_6);
+
+	if (lock) {
+		bios_6_scratch |= ATOM_S6_CRITICAL_STATE;
+		bios_6_scratch &= ~ATOM_S6_ACC_MODE;
+	} else {
+		bios_6_scratch &= ~ATOM_S6_CRITICAL_STATE;
+		bios_6_scratch |= ATOM_S6_ACC_MODE;
+	}
+
+	WREG32(mmBIOS_SCRATCH_6, bios_6_scratch);
+}
+
+void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
+{
+	uint32_t bios_2_scratch, bios_6_scratch;
+
+	bios_2_scratch = RREG32(mmBIOS_SCRATCH_2);
+	bios_6_scratch = RREG32(mmBIOS_SCRATCH_6);
+
+	/* let the bios control the backlight */
+	bios_2_scratch &= ~ATOM_S2_VRI_BRIGHT_ENABLE;
+
+	/* tell the bios not to handle mode switching */
+	bios_6_scratch |= ATOM_S6_ACC_BLOCK_DISPLAY_SWITCH;
+
+	/* clear the vbios dpms state */
+	bios_2_scratch &= ~ATOM_S2_DEVICE_DPMS_STATE;
+
+	WREG32(mmBIOS_SCRATCH_2, bios_2_scratch);
+	WREG32(mmBIOS_SCRATCH_6, bios_6_scratch);
+}
+
+void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
+		adev->bios_scratch[i] = RREG32(mmBIOS_SCRATCH_0 + i);
+}
+
+void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
+		WREG32(mmBIOS_SCRATCH_0 + i, adev->bios_scratch[i]);
+}
+
+/* Atom needs data in little endian format
+ * so swap as appropriate when copying data to
+ * or from atom. Note that atom operates on
+ * dw units.
+ */
+void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le)
+{
+#ifdef __BIG_ENDIAN
+	u8 src_tmp[20], dst_tmp[20]; /* used for byteswapping */
+	u32 *dst32, *src32;
+	int i;
+
+	memcpy(src_tmp, src, num_bytes);
+	src32 = (u32 *)src_tmp;
+	dst32 = (u32 *)dst_tmp;
+	if (to_le) {
+		for (i = 0; i < ((num_bytes + 3) / 4); i++)
+			dst32[i] = cpu_to_le32(src32[i]);
+		memcpy(dst, dst_tmp, num_bytes);
+	} else {
+		u8 dws = num_bytes & ~3;
+		for (i = 0; i < ((num_bytes + 3) / 4); i++)
+			dst32[i] = le32_to_cpu(src32[i]);
+		memcpy(dst, dst_tmp, dws);
+		if (num_bytes % 4) {
+			for (i = 0; i < (num_bytes % 4); i++)
+				dst[dws+i] = dst_tmp[dws+i];
+		}
+	}
+#else
+	memcpy(dst, src, num_bytes);
+#endif
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
new file mode 100644
index 000000000..0ebb959ea
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
@@ -0,0 +1,206 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_ATOMBIOS_H__
+#define __AMDGPU_ATOMBIOS_H__
+
+struct atom_clock_dividers {
+	u32 post_div;
+	union {
+		struct {
+#ifdef __BIG_ENDIAN
+			u32 reserved : 6;
+			u32 whole_fb_div : 12;
+			u32 frac_fb_div : 14;
+#else
+			u32 frac_fb_div : 14;
+			u32 whole_fb_div : 12;
+			u32 reserved : 6;
+#endif
+		};
+		u32 fb_div;
+	};
+	u32 ref_div;
+	bool enable_post_div;
+	bool enable_dithen;
+	u32 vco_mode;
+	u32 real_clock;
+	/* added for CI */
+	u32 post_divider;
+	u32 flags;
+};
+
+struct atom_mpll_param {
+	union {
+		struct {
+#ifdef __BIG_ENDIAN
+			u32 reserved : 8;
+			u32 clkfrac : 12;
+			u32 clkf : 12;
+#else
+			u32 clkf : 12;
+			u32 clkfrac : 12;
+			u32 reserved : 8;
+#endif
+		};
+		u32 fb_div;
+	};
+	u32 post_div;
+	u32 bwcntl;
+	u32 dll_speed;
+	u32 vco_mode;
+	u32 yclk_sel;
+	u32 qdr;
+	u32 half_rate;
+};
+
+#define MEM_TYPE_GDDR5  0x50
+#define MEM_TYPE_GDDR4  0x40
+#define MEM_TYPE_GDDR3  0x30
+#define MEM_TYPE_DDR2   0x20
+#define MEM_TYPE_GDDR1  0x10
+#define MEM_TYPE_DDR3   0xb0
+#define MEM_TYPE_MASK   0xf0
+
+struct atom_memory_info {
+	u8 mem_vendor;
+	u8 mem_type;
+};
+
+#define MAX_AC_TIMING_ENTRIES 16
+
+struct atom_memory_clock_range_table
+{
+	u8 num_entries;
+	u8 rsv[3];
+	u32 mclk[MAX_AC_TIMING_ENTRIES];
+};
+
+#define VBIOS_MC_REGISTER_ARRAY_SIZE 32
+#define VBIOS_MAX_AC_TIMING_ENTRIES 20
+
+struct atom_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[VBIOS_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct atom_mc_register_address {
+	u16 s1;
+	u8 pre_reg_data;
+};
+
+struct atom_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	struct atom_mc_reg_entry mc_reg_table_entry[VBIOS_MAX_AC_TIMING_ENTRIES];
+	struct atom_mc_register_address mc_reg_address[VBIOS_MC_REGISTER_ARRAY_SIZE];
+};
+
+#define MAX_VOLTAGE_ENTRIES 32
+
+struct atom_voltage_table_entry
+{
+	u16 value;
+	u32 smio_low;
+};
+
+struct atom_voltage_table
+{
+	u32 count;
+	u32 mask_low;
+	u32 phase_delay;
+	struct atom_voltage_table_entry entries[MAX_VOLTAGE_ENTRIES];
+};
+
+struct amdgpu_gpio_rec
+amdgpu_atombios_lookup_gpio(struct amdgpu_device *adev,
+			    u8 id);
+
+struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device *adev,
+							  uint8_t id);
+void amdgpu_atombios_i2c_init(struct amdgpu_device *adev);
+
+bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *adev);
+
+int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev);
+
+bool amdgpu_atombios_get_asic_ss_info(struct amdgpu_device *adev,
+				      struct amdgpu_atom_ss *ss,
+				      int id, u32 clock);
+
+int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
+				       u8 clock_type,
+				       u32 clock,
+				       bool strobe_mode,
+				       struct atom_clock_dividers *dividers);
+
+int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
+					    u32 clock,
+					    bool strobe_mode,
+					    struct atom_mpll_param *mpll_param);
+
+uint32_t amdgpu_atombios_get_engine_clock(struct amdgpu_device *adev);
+uint32_t amdgpu_atombios_get_memory_clock(struct amdgpu_device *adev);
+void amdgpu_atombios_set_engine_clock(struct amdgpu_device *adev,
+				      uint32_t eng_clock);
+void amdgpu_atombios_set_memory_clock(struct amdgpu_device *adev,
+				      uint32_t mem_clock);
+void amdgpu_atombios_set_voltage(struct amdgpu_device *adev,
+				 u16 voltage_level,
+				 u8 voltage_type);
+
+void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
+					     u32 eng_clock, u32 mem_clock);
+
+int amdgpu_atombios_get_leakage_id_from_vbios(struct amdgpu_device *adev,
+					      u16 *leakage_id);
+
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(struct amdgpu_device *adev,
+							     u16 *vddc, u16 *vddci,
+							     u16 virtual_voltage_id,
+							     u16 vbios_voltage_id);
+
+int amdgpu_atombios_get_voltage_evv(struct amdgpu_device *adev,
+				    u16 virtual_voltage_id,
+				    u16 *voltage);
+
+bool
+amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev,
+				u8 voltage_type, u8 voltage_mode);
+
+int amdgpu_atombios_get_voltage_table(struct amdgpu_device *adev,
+				      u8 voltage_type, u8 voltage_mode,
+				      struct atom_voltage_table *voltage_table);
+
+int amdgpu_atombios_init_mc_reg_table(struct amdgpu_device *adev,
+				      u8 module_index,
+				      struct atom_mc_reg_table *reg_table);
+
+void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock);
+void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev);
+void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev);
+void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev);
+
+void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
new file mode 100644
index 000000000..5a8fbadbd
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
@@ -0,0 +1,572 @@
+/*
+ * Copyright (c) 2010 Red Hat Inc.
+ * Author : Dave Airlie <airlied@redhat.com>
+ *
+ * Licensed under GPLv2
+ *
+ * ATPX support for both Intel/ATI
+ */
+#include <linux/vga_switcheroo.h>
+#include <linux/slab.h>
+#include <linux/acpi.h>
+#include <linux/pci.h>
+
+#include "amdgpu_acpi.h"
+
+struct amdgpu_atpx_functions {
+	bool px_params;
+	bool power_cntl;
+	bool disp_mux_cntl;
+	bool i2c_mux_cntl;
+	bool switch_start;
+	bool switch_end;
+	bool disp_connectors_mapping;
+	bool disp_detetion_ports;
+};
+
+struct amdgpu_atpx {
+	acpi_handle handle;
+	struct amdgpu_atpx_functions functions;
+};
+
+static struct amdgpu_atpx_priv {
+	bool atpx_detected;
+	/* handle for device - and atpx */
+	acpi_handle dhandle;
+	acpi_handle other_handle;
+	struct amdgpu_atpx atpx;
+} amdgpu_atpx_priv;
+
+struct atpx_verify_interface {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u16 version;		/* version */
+	u32 function_bits;	/* supported functions bit vector */
+} __packed;
+
+struct atpx_px_params {
+	u16 size;		/* structure size in bytes (includes size field) */
+	u32 valid_flags;	/* which flags are valid */
+	u32 flags;		/* flags */
+} __packed;
+
+struct atpx_power_control {
+	u16 size;
+	u8 dgpu_state;
+} __packed;
+
+struct atpx_mux {
+	u16 size;
+	u16 mux;
+} __packed;
+
+bool amdgpu_has_atpx(void) {
+	return amdgpu_atpx_priv.atpx_detected;
+}
+
+/**
+ * amdgpu_atpx_call - call an ATPX method
+ *
+ * @handle: acpi handle
+ * @function: the ATPX function to execute
+ * @params: ATPX function params
+ *
+ * Executes the requested ATPX function (all asics).
+ * Returns a pointer to the acpi output buffer.
+ */
+static union acpi_object *amdgpu_atpx_call(acpi_handle handle, int function,
+					   struct acpi_buffer *params)
+{
+	acpi_status status;
+	union acpi_object atpx_arg_elements[2];
+	struct acpi_object_list atpx_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
+
+	atpx_arg.count = 2;
+	atpx_arg.pointer = &atpx_arg_elements[0];
+
+	atpx_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atpx_arg_elements[0].integer.value = function;
+
+	if (params) {
+		atpx_arg_elements[1].type = ACPI_TYPE_BUFFER;
+		atpx_arg_elements[1].buffer.length = params->length;
+		atpx_arg_elements[1].buffer.pointer = params->pointer;
+	} else {
+		/* We need a second fake parameter */
+		atpx_arg_elements[1].type = ACPI_TYPE_INTEGER;
+		atpx_arg_elements[1].integer.value = 0;
+	}
+
+	status = acpi_evaluate_object(handle, NULL, &atpx_arg, &buffer);
+
+	/* Fail only if calling the method fails and ATPX is supported */
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		printk("failed to evaluate ATPX got %s\n",
+		       acpi_format_exception(status));
+		kfree(buffer.pointer);
+		return NULL;
+	}
+
+	return buffer.pointer;
+}
+
+/**
+ * amdgpu_atpx_parse_functions - parse supported functions
+ *
+ * @f: supported functions struct
+ * @mask: supported functions mask from ATPX
+ *
+ * Use the supported functions mask from ATPX function
+ * ATPX_FUNCTION_VERIFY_INTERFACE to determine what functions
+ * are supported (all asics).
+ */
+static void amdgpu_atpx_parse_functions(struct amdgpu_atpx_functions *f, u32 mask)
+{
+	f->px_params = mask & ATPX_GET_PX_PARAMETERS_SUPPORTED;
+	f->power_cntl = mask & ATPX_POWER_CONTROL_SUPPORTED;
+	f->disp_mux_cntl = mask & ATPX_DISPLAY_MUX_CONTROL_SUPPORTED;
+	f->i2c_mux_cntl = mask & ATPX_I2C_MUX_CONTROL_SUPPORTED;
+	f->switch_start = mask & ATPX_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION_SUPPORTED;
+	f->switch_end = mask & ATPX_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION_SUPPORTED;
+	f->disp_connectors_mapping = mask & ATPX_GET_DISPLAY_CONNECTORS_MAPPING_SUPPORTED;
+	f->disp_detetion_ports = mask & ATPX_GET_DISPLAY_DETECTION_PORTS_SUPPORTED;
+}
+
+/**
+ * amdgpu_atpx_validate_functions - validate ATPX functions
+ *
+ * @atpx: amdgpu atpx struct
+ *
+ * Validate that required functions are enabled (all asics).
+ * returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_validate(struct amdgpu_atpx *atpx)
+{
+	/* make sure required functions are enabled */
+	/* dGPU power control is required */
+	atpx->functions.power_cntl = true;
+
+	if (atpx->functions.px_params) {
+		union acpi_object *info;
+		struct atpx_px_params output;
+		size_t size;
+		u32 valid_bits;
+
+		info = amdgpu_atpx_call(atpx->handle, ATPX_FUNCTION_GET_PX_PARAMETERS, NULL);
+		if (!info)
+			return -EIO;
+
+		memset(&output, 0, sizeof(output));
+
+		size = *(u16 *) info->buffer.pointer;
+		if (size < 10) {
+			printk("ATPX buffer is too small: %zu\n", size);
+			kfree(info);
+			return -EINVAL;
+		}
+		size = min(sizeof(output), size);
+
+		memcpy(&output, info->buffer.pointer, size);
+
+		valid_bits = output.flags & output.valid_flags;
+		/* if separate mux flag is set, mux controls are required */
+		if (valid_bits & ATPX_SEPARATE_MUX_FOR_I2C) {
+			atpx->functions.i2c_mux_cntl = true;
+			atpx->functions.disp_mux_cntl = true;
+		}
+		/* if any outputs are muxed, mux controls are required */
+		if (valid_bits & (ATPX_CRT1_RGB_SIGNAL_MUXED |
+				  ATPX_TV_SIGNAL_MUXED |
+				  ATPX_DFP_SIGNAL_MUXED))
+			atpx->functions.disp_mux_cntl = true;
+
+		kfree(info);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_verify_interface - verify ATPX
+ *
+ * @atpx: amdgpu atpx struct
+ *
+ * Execute the ATPX_FUNCTION_VERIFY_INTERFACE ATPX function
+ * to initialize ATPX and determine what features are supported
+ * (all asics).
+ * returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_verify_interface(struct amdgpu_atpx *atpx)
+{
+	union acpi_object *info;
+	struct atpx_verify_interface output;
+	size_t size;
+	int err = 0;
+
+	info = amdgpu_atpx_call(atpx->handle, ATPX_FUNCTION_VERIFY_INTERFACE, NULL);
+	if (!info)
+		return -EIO;
+
+	memset(&output, 0, sizeof(output));
+
+	size = *(u16 *) info->buffer.pointer;
+	if (size < 8) {
+		printk("ATPX buffer is too small: %zu\n", size);
+		err = -EINVAL;
+		goto out;
+	}
+	size = min(sizeof(output), size);
+
+	memcpy(&output, info->buffer.pointer, size);
+
+	/* TODO: check version? */
+	printk("ATPX version %u, functions 0x%08x\n",
+	       output.version, output.function_bits);
+
+	amdgpu_atpx_parse_functions(&atpx->functions, output.function_bits);
+
+out:
+	kfree(info);
+	return err;
+}
+
+/**
+ * amdgpu_atpx_set_discrete_state - power up/down discrete GPU
+ *
+ * @atpx: atpx info struct
+ * @state: discrete GPU state (0 = power down, 1 = power up)
+ *
+ * Execute the ATPX_FUNCTION_POWER_CONTROL ATPX function to
+ * power down/up the discrete GPU (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_set_discrete_state(struct amdgpu_atpx *atpx, u8 state)
+{
+	struct acpi_buffer params;
+	union acpi_object *info;
+	struct atpx_power_control input;
+
+	if (atpx->functions.power_cntl) {
+		input.size = 3;
+		input.dgpu_state = state;
+		params.length = input.size;
+		params.pointer = &input;
+		info = amdgpu_atpx_call(atpx->handle,
+					ATPX_FUNCTION_POWER_CONTROL,
+					&params);
+		if (!info)
+			return -EIO;
+		kfree(info);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_switch_disp_mux - switch display mux
+ *
+ * @atpx: atpx info struct
+ * @mux_id: mux state (0 = integrated GPU, 1 = discrete GPU)
+ *
+ * Execute the ATPX_FUNCTION_DISPLAY_MUX_CONTROL ATPX function to
+ * switch the display mux between the discrete GPU and integrated GPU
+ * (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_switch_disp_mux(struct amdgpu_atpx *atpx, u16 mux_id)
+{
+	struct acpi_buffer params;
+	union acpi_object *info;
+	struct atpx_mux input;
+
+	if (atpx->functions.disp_mux_cntl) {
+		input.size = 4;
+		input.mux = mux_id;
+		params.length = input.size;
+		params.pointer = &input;
+		info = amdgpu_atpx_call(atpx->handle,
+					ATPX_FUNCTION_DISPLAY_MUX_CONTROL,
+					&params);
+		if (!info)
+			return -EIO;
+		kfree(info);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_switch_i2c_mux - switch i2c/hpd mux
+ *
+ * @atpx: atpx info struct
+ * @mux_id: mux state (0 = integrated GPU, 1 = discrete GPU)
+ *
+ * Execute the ATPX_FUNCTION_I2C_MUX_CONTROL ATPX function to
+ * switch the i2c/hpd mux between the discrete GPU and integrated GPU
+ * (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_switch_i2c_mux(struct amdgpu_atpx *atpx, u16 mux_id)
+{
+	struct acpi_buffer params;
+	union acpi_object *info;
+	struct atpx_mux input;
+
+	if (atpx->functions.i2c_mux_cntl) {
+		input.size = 4;
+		input.mux = mux_id;
+		params.length = input.size;
+		params.pointer = &input;
+		info = amdgpu_atpx_call(atpx->handle,
+					ATPX_FUNCTION_I2C_MUX_CONTROL,
+					&params);
+		if (!info)
+			return -EIO;
+		kfree(info);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_switch_start - notify the sbios of a GPU switch
+ *
+ * @atpx: atpx info struct
+ * @mux_id: mux state (0 = integrated GPU, 1 = discrete GPU)
+ *
+ * Execute the ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION ATPX
+ * function to notify the sbios that a switch between the discrete GPU and
+ * integrated GPU has begun (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_switch_start(struct amdgpu_atpx *atpx, u16 mux_id)
+{
+	struct acpi_buffer params;
+	union acpi_object *info;
+	struct atpx_mux input;
+
+	if (atpx->functions.switch_start) {
+		input.size = 4;
+		input.mux = mux_id;
+		params.length = input.size;
+		params.pointer = &input;
+		info = amdgpu_atpx_call(atpx->handle,
+					ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_START_NOTIFICATION,
+					&params);
+		if (!info)
+			return -EIO;
+		kfree(info);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_switch_end - notify the sbios of a GPU switch
+ *
+ * @atpx: atpx info struct
+ * @mux_id: mux state (0 = integrated GPU, 1 = discrete GPU)
+ *
+ * Execute the ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION ATPX
+ * function to notify the sbios that a switch between the discrete GPU and
+ * integrated GPU has ended (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_switch_end(struct amdgpu_atpx *atpx, u16 mux_id)
+{
+	struct acpi_buffer params;
+	union acpi_object *info;
+	struct atpx_mux input;
+
+	if (atpx->functions.switch_end) {
+		input.size = 4;
+		input.mux = mux_id;
+		params.length = input.size;
+		params.pointer = &input;
+		info = amdgpu_atpx_call(atpx->handle,
+					ATPX_FUNCTION_GRAPHICS_DEVICE_SWITCH_END_NOTIFICATION,
+					&params);
+		if (!info)
+			return -EIO;
+		kfree(info);
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_switchto - switch to the requested GPU
+ *
+ * @id: GPU to switch to
+ *
+ * Execute the necessary ATPX functions to switch between the discrete GPU and
+ * integrated GPU (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_switchto(enum vga_switcheroo_client_id id)
+{
+	u16 gpu_id;
+
+	if (id == VGA_SWITCHEROO_IGD)
+		gpu_id = ATPX_INTEGRATED_GPU;
+	else
+		gpu_id = ATPX_DISCRETE_GPU;
+
+	amdgpu_atpx_switch_start(&amdgpu_atpx_priv.atpx, gpu_id);
+	amdgpu_atpx_switch_disp_mux(&amdgpu_atpx_priv.atpx, gpu_id);
+	amdgpu_atpx_switch_i2c_mux(&amdgpu_atpx_priv.atpx, gpu_id);
+	amdgpu_atpx_switch_end(&amdgpu_atpx_priv.atpx, gpu_id);
+
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_power_state - power down/up the requested GPU
+ *
+ * @id: GPU to power down/up
+ * @state: requested power state (0 = off, 1 = on)
+ *
+ * Execute the necessary ATPX function to power down/up the discrete GPU
+ * (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_power_state(enum vga_switcheroo_client_id id,
+				   enum vga_switcheroo_state state)
+{
+	/* on w500 ACPI can't change intel gpu state */
+	if (id == VGA_SWITCHEROO_IGD)
+		return 0;
+
+	amdgpu_atpx_set_discrete_state(&amdgpu_atpx_priv.atpx, state);
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_pci_probe_handle - look up the ATPX handle
+ *
+ * @pdev: pci device
+ *
+ * Look up the ATPX handles (all asics).
+ * Returns true if the handles are found, false if not.
+ */
+static bool amdgpu_atpx_pci_probe_handle(struct pci_dev *pdev)
+{
+	acpi_handle dhandle, atpx_handle;
+	acpi_status status;
+
+	dhandle = ACPI_HANDLE(&pdev->dev);
+	if (!dhandle)
+		return false;
+
+	status = acpi_get_handle(dhandle, "ATPX", &atpx_handle);
+	if (ACPI_FAILURE(status)) {
+		amdgpu_atpx_priv.other_handle = dhandle;
+		return false;
+	}
+	amdgpu_atpx_priv.dhandle = dhandle;
+	amdgpu_atpx_priv.atpx.handle = atpx_handle;
+	return true;
+}
+
+/**
+ * amdgpu_atpx_init - verify the ATPX interface
+ *
+ * Verify the ATPX interface (all asics).
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_atpx_init(void)
+{
+	int r;
+
+	/* set up the ATPX handle */
+	r = amdgpu_atpx_verify_interface(&amdgpu_atpx_priv.atpx);
+	if (r)
+		return r;
+
+	/* validate the atpx setup */
+	r = amdgpu_atpx_validate(&amdgpu_atpx_priv.atpx);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * amdgpu_atpx_get_client_id - get the client id
+ *
+ * @pdev: pci device
+ *
+ * look up whether we are the integrated or discrete GPU (all asics).
+ * Returns the client id.
+ */
+static int amdgpu_atpx_get_client_id(struct pci_dev *pdev)
+{
+	if (amdgpu_atpx_priv.dhandle == ACPI_HANDLE(&pdev->dev))
+		return VGA_SWITCHEROO_IGD;
+	else
+		return VGA_SWITCHEROO_DIS;
+}
+
+static const struct vga_switcheroo_handler amdgpu_atpx_handler = {
+	.switchto = amdgpu_atpx_switchto,
+	.power_state = amdgpu_atpx_power_state,
+	.init = amdgpu_atpx_init,
+	.get_client_id = amdgpu_atpx_get_client_id,
+};
+
+/**
+ * amdgpu_atpx_detect - detect whether we have PX
+ *
+ * Check if we have a PX system (all asics).
+ * Returns true if we have a PX system, false if not.
+ */
+static bool amdgpu_atpx_detect(void)
+{
+	char acpi_method_name[255] = { 0 };
+	struct acpi_buffer buffer = {sizeof(acpi_method_name), acpi_method_name};
+	struct pci_dev *pdev = NULL;
+	bool has_atpx = false;
+	int vga_count = 0;
+
+	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
+		vga_count++;
+
+		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+	}
+
+	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
+		vga_count++;
+
+		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+	}
+
+	if (has_atpx && vga_count == 2) {
+		acpi_get_name(amdgpu_atpx_priv.atpx.handle, ACPI_FULL_PATHNAME, &buffer);
+		printk(KERN_INFO "vga_switcheroo: detected switching method %s handle\n",
+		       acpi_method_name);
+		amdgpu_atpx_priv.atpx_detected = true;
+		return true;
+	}
+	return false;
+}
+
+/**
+ * amdgpu_register_atpx_handler - register with vga_switcheroo
+ *
+ * Register the PX callbacks with vga_switcheroo (all asics).
+ */
+void amdgpu_register_atpx_handler(void)
+{
+	bool r;
+
+	/* detect if we have any ATPX + 2 VGA in the system */
+	r = amdgpu_atpx_detect();
+	if (!r)
+		return;
+
+	vga_switcheroo_register_handler(&amdgpu_atpx_handler);
+}
+
+/**
+ * amdgpu_unregister_atpx_handler - unregister with vga_switcheroo
+ *
+ * Unregister the PX callbacks with vga_switcheroo (all asics).
+ */
+void amdgpu_unregister_atpx_handler(void)
+{
+	vga_switcheroo_unregister_handler();
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
new file mode 100644
index 000000000..cd639c362
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
@@ -0,0 +1,223 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#define AMDGPU_BENCHMARK_ITERATIONS 1024
+#define AMDGPU_BENCHMARK_COMMON_MODES_N 17
+
+static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size,
+				    uint64_t saddr, uint64_t daddr, int n)
+{
+	unsigned long start_jiffies;
+	unsigned long end_jiffies;
+	struct fence *fence = NULL;
+	int i, r;
+
+	start_jiffies = jiffies;
+	for (i = 0; i < n; i++) {
+		struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+		r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence);
+		if (r)
+			goto exit_do_move;
+		r = fence_wait(fence, false);
+		if (r)
+			goto exit_do_move;
+		fence_put(fence);
+	}
+	end_jiffies = jiffies;
+	r = jiffies_to_msecs(end_jiffies - start_jiffies);
+
+exit_do_move:
+	if (fence)
+		fence_put(fence);
+	return r;
+}
+
+
+static void amdgpu_benchmark_log_results(int n, unsigned size,
+					 unsigned int time,
+					 unsigned sdomain, unsigned ddomain,
+					 char *kind)
+{
+	unsigned int throughput = (n * (size >> 10)) / time;
+	DRM_INFO("amdgpu: %s %u bo moves of %u kB from"
+		 " %d to %d in %u ms, throughput: %u Mb/s or %u MB/s\n",
+		 kind, n, size >> 10, sdomain, ddomain, time,
+		 throughput * 8, throughput);
+}
+
+static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
+				  unsigned sdomain, unsigned ddomain)
+{
+	struct amdgpu_bo *dobj = NULL;
+	struct amdgpu_bo *sobj = NULL;
+	uint64_t saddr, daddr;
+	int r, n;
+	int time;
+
+	n = AMDGPU_BENCHMARK_ITERATIONS;
+	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, sdomain, 0, NULL,
+			     NULL, &sobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_reserve(sobj, false);
+	if (unlikely(r != 0))
+		goto out_cleanup;
+	r = amdgpu_bo_pin(sobj, sdomain, &saddr);
+	amdgpu_bo_unreserve(sobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, ddomain, 0, NULL,
+			     NULL, &dobj);
+	if (r) {
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_reserve(dobj, false);
+	if (unlikely(r != 0))
+		goto out_cleanup;
+	r = amdgpu_bo_pin(dobj, ddomain, &daddr);
+	amdgpu_bo_unreserve(dobj);
+	if (r) {
+		goto out_cleanup;
+	}
+
+	if (adev->mman.buffer_funcs) {
+		time = amdgpu_benchmark_do_move(adev, size, saddr, daddr, n);
+		if (time < 0)
+			goto out_cleanup;
+		if (time > 0)
+			amdgpu_benchmark_log_results(n, size, time,
+						     sdomain, ddomain, "dma");
+	}
+
+out_cleanup:
+	if (sobj) {
+		r = amdgpu_bo_reserve(sobj, false);
+		if (likely(r == 0)) {
+			amdgpu_bo_unpin(sobj);
+			amdgpu_bo_unreserve(sobj);
+		}
+		amdgpu_bo_unref(&sobj);
+	}
+	if (dobj) {
+		r = amdgpu_bo_reserve(dobj, false);
+		if (likely(r == 0)) {
+			amdgpu_bo_unpin(dobj);
+			amdgpu_bo_unreserve(dobj);
+		}
+		amdgpu_bo_unref(&dobj);
+	}
+
+	if (r) {
+		DRM_ERROR("Error while benchmarking BO move.\n");
+	}
+}
+
+void amdgpu_benchmark(struct amdgpu_device *adev, int test_number)
+{
+	int i;
+	int common_modes[AMDGPU_BENCHMARK_COMMON_MODES_N] = {
+		640 * 480 * 4,
+		720 * 480 * 4,
+		800 * 600 * 4,
+		848 * 480 * 4,
+		1024 * 768 * 4,
+		1152 * 768 * 4,
+		1280 * 720 * 4,
+		1280 * 800 * 4,
+		1280 * 854 * 4,
+		1280 * 960 * 4,
+		1280 * 1024 * 4,
+		1440 * 900 * 4,
+		1400 * 1050 * 4,
+		1680 * 1050 * 4,
+		1600 * 1200 * 4,
+		1920 * 1080 * 4,
+		1920 * 1200 * 4
+	};
+
+	switch (test_number) {
+	case 1:
+		/* simple test, VRAM to GTT and GTT to VRAM */
+		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_GTT,
+				      AMDGPU_GEM_DOMAIN_VRAM);
+		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case 2:
+		/* simple test, VRAM to VRAM */
+		amdgpu_benchmark_move(adev, 1024*1024, AMDGPU_GEM_DOMAIN_VRAM,
+				      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 3:
+		/* GTT to VRAM, buffer size sweep, powers of 2 */
+		for (i = 1; i <= 16384; i <<= 1)
+			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 4:
+		/* VRAM to GTT, buffer size sweep, powers of 2 */
+		for (i = 1; i <= 16384; i <<= 1)
+			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case 5:
+		/* VRAM to VRAM, buffer size sweep, powers of 2 */
+		for (i = 1; i <= 16384; i <<= 1)
+			amdgpu_benchmark_move(adev, i * AMDGPU_GPU_PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 6:
+		/* GTT to VRAM, buffer size sweep, common modes */
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
+			amdgpu_benchmark_move(adev, common_modes[i],
+					      AMDGPU_GEM_DOMAIN_GTT,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+	case 7:
+		/* VRAM to GTT, buffer size sweep, common modes */
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
+			amdgpu_benchmark_move(adev, common_modes[i],
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case 8:
+		/* VRAM to VRAM, buffer size sweep, common modes */
+		for (i = 0; i < AMDGPU_BENCHMARK_COMMON_MODES_N; i++)
+			amdgpu_benchmark_move(adev, common_modes[i],
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      AMDGPU_GEM_DOMAIN_VRAM);
+		break;
+
+	default:
+		DRM_ERROR("Unknown benchmark\n");
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
new file mode 100644
index 000000000..c44c0c6af
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
@@ -0,0 +1,362 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "atom.h"
+
+#include <linux/slab.h>
+#include <linux/acpi.h>
+/*
+ * BIOS.
+ */
+
+/* If you boot an IGP board with a discrete card as the primary,
+ * the IGP rom is not accessible via the rom bar as the IGP rom is
+ * part of the system bios.  On boot, the system bios puts a
+ * copy of the igp rom at the start of vram if a discrete card is
+ * present.
+ */
+static bool igp_read_bios_from_vram(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	resource_size_t vram_base;
+	resource_size_t size = 256 * 1024; /* ??? */
+
+	if (!(adev->flags & AMD_IS_APU))
+		if (!amdgpu_card_posted(adev))
+			return false;
+
+	adev->bios = NULL;
+	vram_base = pci_resource_start(adev->pdev, 0);
+	bios = ioremap(vram_base, size);
+	if (!bios) {
+		return false;
+	}
+
+	if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) {
+		iounmap(bios);
+		return false;
+	}
+	adev->bios = kmalloc(size, GFP_KERNEL);
+	if (adev->bios == NULL) {
+		iounmap(bios);
+		return false;
+	}
+	memcpy_fromio(adev->bios, bios, size);
+	iounmap(bios);
+	return true;
+}
+
+bool amdgpu_read_bios(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios, val1, val2;
+	size_t size;
+
+	adev->bios = NULL;
+	/* XXX: some cards may return 0 for rom size? ddx has a workaround */
+	bios = pci_map_rom(adev->pdev, &size);
+	if (!bios) {
+		return false;
+	}
+
+	val1 = readb(&bios[0]);
+	val2 = readb(&bios[1]);
+
+	if (size == 0 || val1 != 0x55 || val2 != 0xaa) {
+		pci_unmap_rom(adev->pdev, bios);
+		return false;
+	}
+	adev->bios = kzalloc(size, GFP_KERNEL);
+	if (adev->bios == NULL) {
+		pci_unmap_rom(adev->pdev, bios);
+		return false;
+	}
+	memcpy_fromio(adev->bios, bios, size);
+	pci_unmap_rom(adev->pdev, bios);
+	return true;
+}
+
+static bool amdgpu_read_platform_bios(struct amdgpu_device *adev)
+{
+	uint8_t __iomem *bios;
+	size_t size;
+
+	adev->bios = NULL;
+
+	bios = pci_platform_rom(adev->pdev, &size);
+	if (!bios) {
+		return false;
+	}
+
+	if (size == 0 || bios[0] != 0x55 || bios[1] != 0xaa) {
+		return false;
+	}
+	adev->bios = kmemdup(bios, size, GFP_KERNEL);
+	if (adev->bios == NULL) {
+		return false;
+	}
+
+	return true;
+}
+
+#ifdef CONFIG_ACPI
+/* ATRM is used to get the BIOS on the discrete cards in
+ * dual-gpu systems.
+ */
+/* retrieve the ROM in 4k blocks */
+#define ATRM_BIOS_PAGE 4096
+/**
+ * amdgpu_atrm_call - fetch a chunk of the vbios
+ *
+ * @atrm_handle: acpi ATRM handle
+ * @bios: vbios image pointer
+ * @offset: offset of vbios image data to fetch
+ * @len: length of vbios image data to fetch
+ *
+ * Executes ATRM to fetch a chunk of the discrete
+ * vbios image on PX systems (all asics).
+ * Returns the length of the buffer fetched.
+ */
+static int amdgpu_atrm_call(acpi_handle atrm_handle, uint8_t *bios,
+			    int offset, int len)
+{
+	acpi_status status;
+	union acpi_object atrm_arg_elements[2], *obj;
+	struct acpi_object_list atrm_arg;
+	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL};
+
+	atrm_arg.count = 2;
+	atrm_arg.pointer = &atrm_arg_elements[0];
+
+	atrm_arg_elements[0].type = ACPI_TYPE_INTEGER;
+	atrm_arg_elements[0].integer.value = offset;
+
+	atrm_arg_elements[1].type = ACPI_TYPE_INTEGER;
+	atrm_arg_elements[1].integer.value = len;
+
+	status = acpi_evaluate_object(atrm_handle, NULL, &atrm_arg, &buffer);
+	if (ACPI_FAILURE(status)) {
+		printk("failed to evaluate ATRM got %s\n", acpi_format_exception(status));
+		return -ENODEV;
+	}
+
+	obj = (union acpi_object *)buffer.pointer;
+	memcpy(bios+offset, obj->buffer.pointer, obj->buffer.length);
+	len = obj->buffer.length;
+	kfree(buffer.pointer);
+	return len;
+}
+
+static bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
+{
+	int ret;
+	int size = 256 * 1024;
+	int i;
+	struct pci_dev *pdev = NULL;
+	acpi_handle dhandle, atrm_handle;
+	acpi_status status;
+	bool found = false;
+
+	/* ATRM is for the discrete card only */
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
+		dhandle = ACPI_HANDLE(&pdev->dev);
+		if (!dhandle)
+			continue;
+
+		status = acpi_get_handle(dhandle, "ATRM", &atrm_handle);
+		if (!ACPI_FAILURE(status)) {
+			found = true;
+			break;
+		}
+	}
+
+	if (!found) {
+		while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
+			dhandle = ACPI_HANDLE(&pdev->dev);
+			if (!dhandle)
+				continue;
+
+			status = acpi_get_handle(dhandle, "ATRM", &atrm_handle);
+			if (!ACPI_FAILURE(status)) {
+				found = true;
+				break;
+			}
+		}
+	}
+
+	if (!found)
+		return false;
+
+	adev->bios = kmalloc(size, GFP_KERNEL);
+	if (!adev->bios) {
+		DRM_ERROR("Unable to allocate bios\n");
+		return false;
+	}
+
+	for (i = 0; i < size / ATRM_BIOS_PAGE; i++) {
+		ret = amdgpu_atrm_call(atrm_handle,
+				       adev->bios,
+				       (i * ATRM_BIOS_PAGE),
+				       ATRM_BIOS_PAGE);
+		if (ret < ATRM_BIOS_PAGE)
+			break;
+	}
+
+	if (i == 0 || adev->bios[0] != 0x55 || adev->bios[1] != 0xaa) {
+		kfree(adev->bios);
+		return false;
+	}
+	return true;
+}
+#else
+static inline bool amdgpu_atrm_get_bios(struct amdgpu_device *adev)
+{
+	return false;
+}
+#endif
+
+static bool amdgpu_read_disabled_bios(struct amdgpu_device *adev)
+{
+	if (adev->flags & AMD_IS_APU)
+		return igp_read_bios_from_vram(adev);
+	else
+		return amdgpu_asic_read_disabled_bios(adev);
+}
+
+#ifdef CONFIG_ACPI
+static bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
+{
+	bool ret = false;
+	struct acpi_table_header *hdr;
+	acpi_size tbl_size;
+	UEFI_ACPI_VFCT *vfct;
+	GOP_VBIOS_CONTENT *vbios;
+	VFCT_IMAGE_HEADER *vhdr;
+
+	if (!ACPI_SUCCESS(acpi_get_table_with_size("VFCT", 1, &hdr, &tbl_size)))
+		return false;
+	if (tbl_size < sizeof(UEFI_ACPI_VFCT)) {
+		DRM_ERROR("ACPI VFCT table present but broken (too short #1)\n");
+		goto out_unmap;
+	}
+
+	vfct = (UEFI_ACPI_VFCT *)hdr;
+	if (vfct->VBIOSImageOffset + sizeof(VFCT_IMAGE_HEADER) > tbl_size) {
+		DRM_ERROR("ACPI VFCT table present but broken (too short #2)\n");
+		goto out_unmap;
+	}
+
+	vbios = (GOP_VBIOS_CONTENT *)((char *)hdr + vfct->VBIOSImageOffset);
+	vhdr = &vbios->VbiosHeader;
+	DRM_INFO("ACPI VFCT contains a BIOS for %02x:%02x.%d %04x:%04x, size %d\n",
+			vhdr->PCIBus, vhdr->PCIDevice, vhdr->PCIFunction,
+			vhdr->VendorID, vhdr->DeviceID, vhdr->ImageLength);
+
+	if (vhdr->PCIBus != adev->pdev->bus->number ||
+	    vhdr->PCIDevice != PCI_SLOT(adev->pdev->devfn) ||
+	    vhdr->PCIFunction != PCI_FUNC(adev->pdev->devfn) ||
+	    vhdr->VendorID != adev->pdev->vendor ||
+	    vhdr->DeviceID != adev->pdev->device) {
+		DRM_INFO("ACPI VFCT table is not for this card\n");
+		goto out_unmap;
+	}
+
+	if (vfct->VBIOSImageOffset + sizeof(VFCT_IMAGE_HEADER) + vhdr->ImageLength > tbl_size) {
+		DRM_ERROR("ACPI VFCT image truncated\n");
+		goto out_unmap;
+	}
+
+	adev->bios = kmemdup(&vbios->VbiosContent, vhdr->ImageLength, GFP_KERNEL);
+	ret = !!adev->bios;
+
+out_unmap:
+	return ret;
+}
+#else
+static inline bool amdgpu_acpi_vfct_bios(struct amdgpu_device *adev)
+{
+	return false;
+}
+#endif
+
+bool amdgpu_get_bios(struct amdgpu_device *adev)
+{
+	bool r;
+	uint16_t tmp;
+
+	r = amdgpu_atrm_get_bios(adev);
+	if (r == false)
+		r = amdgpu_acpi_vfct_bios(adev);
+	if (r == false)
+		r = igp_read_bios_from_vram(adev);
+	if (r == false)
+		r = amdgpu_read_bios(adev);
+	if (r == false) {
+		r = amdgpu_read_disabled_bios(adev);
+	}
+	if (r == false) {
+		r = amdgpu_read_platform_bios(adev);
+	}
+	if (r == false || adev->bios == NULL) {
+		DRM_ERROR("Unable to locate a BIOS ROM\n");
+		adev->bios = NULL;
+		return false;
+	}
+	if (adev->bios[0] != 0x55 || adev->bios[1] != 0xaa) {
+		printk("BIOS signature incorrect %x %x\n", adev->bios[0], adev->bios[1]);
+		goto free_bios;
+	}
+
+	tmp = RBIOS16(0x18);
+	if (RBIOS8(tmp + 0x14) != 0x0) {
+		DRM_INFO("Not an x86 BIOS ROM, not using.\n");
+		goto free_bios;
+	}
+
+	adev->bios_header_start = RBIOS16(0x48);
+	if (!adev->bios_header_start) {
+		goto free_bios;
+	}
+	tmp = adev->bios_header_start + 4;
+	if (!memcmp(adev->bios + tmp, "ATOM", 4) ||
+	    !memcmp(adev->bios + tmp, "MOTA", 4)) {
+		adev->is_atom_bios = true;
+	} else {
+		adev->is_atom_bios = false;
+	}
+
+	DRM_DEBUG("%sBIOS detected\n", adev->is_atom_bios ? "ATOM" : "COM");
+	return true;
+free_bios:
+	kfree(adev->bios);
+	adev->bios = NULL;
+	return false;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
new file mode 100644
index 000000000..f82a2dd83
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -0,0 +1,271 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+static int amdgpu_bo_list_create(struct amdgpu_fpriv *fpriv,
+				 struct amdgpu_bo_list **result,
+				 int *id)
+{
+	int r;
+
+	*result = kzalloc(sizeof(struct amdgpu_bo_list), GFP_KERNEL);
+	if (!*result)
+		return -ENOMEM;
+
+	mutex_lock(&fpriv->bo_list_lock);
+	r = idr_alloc(&fpriv->bo_list_handles, *result,
+		      1, 0, GFP_KERNEL);
+	if (r < 0) {
+		mutex_unlock(&fpriv->bo_list_lock);
+		kfree(*result);
+		return r;
+	}
+	*id = r;
+
+	mutex_init(&(*result)->lock);
+	(*result)->num_entries = 0;
+	(*result)->array = NULL;
+
+	mutex_lock(&(*result)->lock);
+	mutex_unlock(&fpriv->bo_list_lock);
+
+	return 0;
+}
+
+static void amdgpu_bo_list_destroy(struct amdgpu_fpriv *fpriv, int id)
+{
+	struct amdgpu_bo_list *list;
+
+	mutex_lock(&fpriv->bo_list_lock);
+	list = idr_find(&fpriv->bo_list_handles, id);
+	if (list) {
+		mutex_lock(&list->lock);
+		idr_remove(&fpriv->bo_list_handles, id);
+		mutex_unlock(&list->lock);
+		amdgpu_bo_list_free(list);
+	}
+	mutex_unlock(&fpriv->bo_list_lock);
+}
+
+static int amdgpu_bo_list_set(struct amdgpu_device *adev,
+				     struct drm_file *filp,
+				     struct amdgpu_bo_list *list,
+				     struct drm_amdgpu_bo_list_entry *info,
+				     unsigned num_entries)
+{
+	struct amdgpu_bo_list_entry *array;
+	struct amdgpu_bo *gds_obj = adev->gds.gds_gfx_bo;
+	struct amdgpu_bo *gws_obj = adev->gds.gws_gfx_bo;
+	struct amdgpu_bo *oa_obj = adev->gds.oa_gfx_bo;
+
+	bool has_userptr = false;
+	unsigned i;
+
+	array = drm_malloc_ab(num_entries, sizeof(struct amdgpu_bo_list_entry));
+	if (!array)
+		return -ENOMEM;
+	memset(array, 0, num_entries * sizeof(struct amdgpu_bo_list_entry));
+
+	for (i = 0; i < num_entries; ++i) {
+		struct amdgpu_bo_list_entry *entry = &array[i];
+		struct drm_gem_object *gobj;
+
+		gobj = drm_gem_object_lookup(adev->ddev, filp, info[i].bo_handle);
+		if (!gobj)
+			goto error_free;
+
+		entry->robj = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+		drm_gem_object_unreference_unlocked(gobj);
+		entry->priority = info[i].bo_priority;
+		entry->prefered_domains = entry->robj->initial_domain;
+		entry->allowed_domains = entry->prefered_domains;
+		if (entry->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
+			entry->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
+		if (amdgpu_ttm_tt_has_userptr(entry->robj->tbo.ttm)) {
+			has_userptr = true;
+			entry->prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
+			entry->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+		}
+		entry->tv.bo = &entry->robj->tbo;
+		entry->tv.shared = true;
+
+		if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_GDS)
+			gds_obj = entry->robj;
+		if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_GWS)
+			gws_obj = entry->robj;
+		if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_OA)
+			oa_obj = entry->robj;
+
+		trace_amdgpu_bo_list_set(list, entry->robj);
+	}
+
+	for (i = 0; i < list->num_entries; ++i)
+		amdgpu_bo_unref(&list->array[i].robj);
+
+	drm_free_large(list->array);
+
+	list->gds_obj = gds_obj;
+	list->gws_obj = gws_obj;
+	list->oa_obj = oa_obj;
+	list->has_userptr = has_userptr;
+	list->array = array;
+	list->num_entries = num_entries;
+
+	return 0;
+
+error_free:
+	drm_free_large(array);
+	return -ENOENT;
+}
+
+struct amdgpu_bo_list *
+amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id)
+{
+	struct amdgpu_bo_list *result;
+
+	mutex_lock(&fpriv->bo_list_lock);
+	result = idr_find(&fpriv->bo_list_handles, id);
+	if (result)
+		mutex_lock(&result->lock);
+	mutex_unlock(&fpriv->bo_list_lock);
+	return result;
+}
+
+void amdgpu_bo_list_put(struct amdgpu_bo_list *list)
+{
+	mutex_unlock(&list->lock);
+}
+
+void amdgpu_bo_list_free(struct amdgpu_bo_list *list)
+{
+	unsigned i;
+
+	for (i = 0; i < list->num_entries; ++i)
+		amdgpu_bo_unref(&list->array[i].robj);
+
+	mutex_destroy(&list->lock);
+	drm_free_large(list->array);
+	kfree(list);
+}
+
+int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp)
+{
+	const uint32_t info_size = sizeof(struct drm_amdgpu_bo_list_entry);
+
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	union drm_amdgpu_bo_list *args = data;
+	uint32_t handle = args->in.list_handle;
+	const void __user *uptr = (const void*)(long)args->in.bo_info_ptr;
+
+	struct drm_amdgpu_bo_list_entry *info;
+	struct amdgpu_bo_list *list;
+
+	int r;
+
+	info = drm_malloc_ab(args->in.bo_number,
+			     sizeof(struct drm_amdgpu_bo_list_entry));
+	if (!info)
+		return -ENOMEM;
+
+	/* copy the handle array from userspace to a kernel buffer */
+	r = -EFAULT;
+	if (likely(info_size == args->in.bo_info_size)) {
+		unsigned long bytes = args->in.bo_number *
+			args->in.bo_info_size;
+
+		if (copy_from_user(info, uptr, bytes))
+			goto error_free;
+
+	} else {
+		unsigned long bytes = min(args->in.bo_info_size, info_size);
+		unsigned i;
+
+		memset(info, 0, args->in.bo_number * info_size);
+		for (i = 0; i < args->in.bo_number; ++i) {
+			if (copy_from_user(&info[i], uptr, bytes))
+				goto error_free;
+			
+			uptr += args->in.bo_info_size;
+		}
+	}
+
+	switch (args->in.operation) {
+	case AMDGPU_BO_LIST_OP_CREATE:
+		r = amdgpu_bo_list_create(fpriv, &list, &handle);
+		if (r) 
+			goto error_free;
+
+		r = amdgpu_bo_list_set(adev, filp, list, info,
+					      args->in.bo_number);
+		amdgpu_bo_list_put(list);
+		if (r)
+			goto error_free;
+
+		break;
+		
+	case AMDGPU_BO_LIST_OP_DESTROY:
+		amdgpu_bo_list_destroy(fpriv, handle);
+		handle = 0;
+		break;
+
+	case AMDGPU_BO_LIST_OP_UPDATE:
+		r = -ENOENT;
+		list = amdgpu_bo_list_get(fpriv, handle);
+		if (!list)
+			goto error_free;
+
+		r = amdgpu_bo_list_set(adev, filp, list, info,
+					      args->in.bo_number);
+		amdgpu_bo_list_put(list);
+		if (r)
+			goto error_free;
+
+		break;
+
+	default:
+		r = -EINVAL;
+		goto error_free;
+	}
+
+	memset(args, 0, sizeof(*args));
+	args->out.list_handle = handle;
+	drm_free_large(info);
+
+	return 0;
+
+error_free:
+	drm_free_large(info);
+	return r;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
new file mode 100644
index 000000000..8e995148f
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
@@ -0,0 +1,800 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+#include <drm/drmP.h>
+#include <linux/firmware.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "cgs_linux.h"
+#include "atom.h"
+#include "amdgpu_ucode.h"
+
+
+struct amdgpu_cgs_device {
+	struct cgs_device base;
+	struct amdgpu_device *adev;
+};
+
+#define CGS_FUNC_ADEV							\
+	struct amdgpu_device *adev =					\
+		((struct amdgpu_cgs_device *)cgs_device)->adev
+
+static int amdgpu_cgs_gpu_mem_info(void *cgs_device, enum cgs_gpu_mem_type type,
+				   uint64_t *mc_start, uint64_t *mc_size,
+				   uint64_t *mem_size)
+{
+	CGS_FUNC_ADEV;
+	switch(type) {
+	case CGS_GPU_MEM_TYPE__VISIBLE_CONTIG_FB:
+	case CGS_GPU_MEM_TYPE__VISIBLE_FB:
+		*mc_start = 0;
+		*mc_size = adev->mc.visible_vram_size;
+		*mem_size = adev->mc.visible_vram_size - adev->vram_pin_size;
+		break;
+	case CGS_GPU_MEM_TYPE__INVISIBLE_CONTIG_FB:
+	case CGS_GPU_MEM_TYPE__INVISIBLE_FB:
+		*mc_start = adev->mc.visible_vram_size;
+		*mc_size = adev->mc.real_vram_size - adev->mc.visible_vram_size;
+		*mem_size = *mc_size;
+		break;
+	case CGS_GPU_MEM_TYPE__GART_CACHEABLE:
+	case CGS_GPU_MEM_TYPE__GART_WRITECOMBINE:
+		*mc_start = adev->mc.gtt_start;
+		*mc_size = adev->mc.gtt_size;
+		*mem_size = adev->mc.gtt_size - adev->gart_pin_size;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cgs_gmap_kmem(void *cgs_device, void *kmem,
+				uint64_t size,
+				uint64_t min_offset, uint64_t max_offset,
+				cgs_handle_t *kmem_handle, uint64_t *mcaddr)
+{
+	CGS_FUNC_ADEV;
+	int ret;
+	struct amdgpu_bo *bo;
+	struct page *kmem_page = vmalloc_to_page(kmem);
+	int npages = ALIGN(size, PAGE_SIZE) >> PAGE_SHIFT;
+
+	struct sg_table *sg = drm_prime_pages_to_sg(&kmem_page, npages);
+	ret = amdgpu_bo_create(adev, size, PAGE_SIZE, false,
+			       AMDGPU_GEM_DOMAIN_GTT, 0, sg, NULL, &bo);
+	if (ret)
+		return ret;
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret != 0))
+		return ret;
+
+	/* pin buffer into GTT */
+	ret = amdgpu_bo_pin_restricted(bo, AMDGPU_GEM_DOMAIN_GTT,
+				       min_offset, max_offset, mcaddr);
+	amdgpu_bo_unreserve(bo);
+
+	*kmem_handle = (cgs_handle_t)bo;
+	return ret;
+}
+
+static int amdgpu_cgs_gunmap_kmem(void *cgs_device, cgs_handle_t kmem_handle)
+{
+	struct amdgpu_bo *obj = (struct amdgpu_bo *)kmem_handle;
+
+	if (obj) {
+		int r = amdgpu_bo_reserve(obj, false);
+		if (likely(r == 0)) {
+			amdgpu_bo_unpin(obj);
+			amdgpu_bo_unreserve(obj);
+		}
+		amdgpu_bo_unref(&obj);
+
+	}
+	return 0;
+}
+
+static int amdgpu_cgs_alloc_gpu_mem(void *cgs_device,
+				    enum cgs_gpu_mem_type type,
+				    uint64_t size, uint64_t align,
+				    uint64_t min_offset, uint64_t max_offset,
+				    cgs_handle_t *handle)
+{
+	CGS_FUNC_ADEV;
+	uint16_t flags = 0;
+	int ret = 0;
+	uint32_t domain = 0;
+	struct amdgpu_bo *obj;
+	struct ttm_placement placement;
+	struct ttm_place place;
+
+	if (min_offset > max_offset) {
+		BUG_ON(1);
+		return -EINVAL;
+	}
+
+	/* fail if the alignment is not a power of 2 */
+	if (((align != 1) && (align & (align - 1)))
+	    || size == 0 || align == 0)
+		return -EINVAL;
+
+
+	switch(type) {
+	case CGS_GPU_MEM_TYPE__VISIBLE_CONTIG_FB:
+	case CGS_GPU_MEM_TYPE__VISIBLE_FB:
+		flags = AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
+		domain = AMDGPU_GEM_DOMAIN_VRAM;
+		if (max_offset > adev->mc.real_vram_size)
+			return -EINVAL;
+		place.fpfn = min_offset >> PAGE_SHIFT;
+		place.lpfn = max_offset >> PAGE_SHIFT;
+		place.flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
+			TTM_PL_FLAG_VRAM;
+		break;
+	case CGS_GPU_MEM_TYPE__INVISIBLE_CONTIG_FB:
+	case CGS_GPU_MEM_TYPE__INVISIBLE_FB:
+		flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
+		domain = AMDGPU_GEM_DOMAIN_VRAM;
+		if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
+			place.fpfn =
+				max(min_offset, adev->mc.visible_vram_size) >> PAGE_SHIFT;
+			place.lpfn =
+				min(max_offset, adev->mc.real_vram_size) >> PAGE_SHIFT;
+			place.flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
+				TTM_PL_FLAG_VRAM;
+		}
+
+		break;
+	case CGS_GPU_MEM_TYPE__GART_CACHEABLE:
+		domain = AMDGPU_GEM_DOMAIN_GTT;
+		place.fpfn = min_offset >> PAGE_SHIFT;
+		place.lpfn = max_offset >> PAGE_SHIFT;
+		place.flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT;
+		break;
+	case CGS_GPU_MEM_TYPE__GART_WRITECOMBINE:
+		flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+		domain = AMDGPU_GEM_DOMAIN_GTT;
+		place.fpfn = min_offset >> PAGE_SHIFT;
+		place.lpfn = max_offset >> PAGE_SHIFT;
+		place.flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_TT |
+			TTM_PL_FLAG_UNCACHED;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+
+	*handle = 0;
+
+	placement.placement = &place;
+	placement.num_placement = 1;
+	placement.busy_placement = &place;
+	placement.num_busy_placement = 1;
+
+	ret = amdgpu_bo_create_restricted(adev, size, PAGE_SIZE,
+					  true, domain, flags,
+					  NULL, &placement, NULL,
+					  &obj);
+	if (ret) {
+		DRM_ERROR("(%d) bo create failed\n", ret);
+		return ret;
+	}
+	*handle = (cgs_handle_t)obj;
+
+	return ret;
+}
+
+static int amdgpu_cgs_free_gpu_mem(void *cgs_device, cgs_handle_t handle)
+{
+	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
+
+	if (obj) {
+		int r = amdgpu_bo_reserve(obj, false);
+		if (likely(r == 0)) {
+			amdgpu_bo_kunmap(obj);
+			amdgpu_bo_unpin(obj);
+			amdgpu_bo_unreserve(obj);
+		}
+		amdgpu_bo_unref(&obj);
+
+	}
+	return 0;
+}
+
+static int amdgpu_cgs_gmap_gpu_mem(void *cgs_device, cgs_handle_t handle,
+				   uint64_t *mcaddr)
+{
+	int r;
+	u64 min_offset, max_offset;
+	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
+
+	WARN_ON_ONCE(obj->placement.num_placement > 1);
+
+	min_offset = obj->placements[0].fpfn << PAGE_SHIFT;
+	max_offset = obj->placements[0].lpfn << PAGE_SHIFT;
+
+	r = amdgpu_bo_reserve(obj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = amdgpu_bo_pin_restricted(obj, AMDGPU_GEM_DOMAIN_GTT,
+				     min_offset, max_offset, mcaddr);
+	amdgpu_bo_unreserve(obj);
+	return r;
+}
+
+static int amdgpu_cgs_gunmap_gpu_mem(void *cgs_device, cgs_handle_t handle)
+{
+	int r;
+	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
+	r = amdgpu_bo_reserve(obj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = amdgpu_bo_unpin(obj);
+	amdgpu_bo_unreserve(obj);
+	return r;
+}
+
+static int amdgpu_cgs_kmap_gpu_mem(void *cgs_device, cgs_handle_t handle,
+				   void **map)
+{
+	int r;
+	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
+	r = amdgpu_bo_reserve(obj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = amdgpu_bo_kmap(obj, map);
+	amdgpu_bo_unreserve(obj);
+	return r;
+}
+
+static int amdgpu_cgs_kunmap_gpu_mem(void *cgs_device, cgs_handle_t handle)
+{
+	int r;
+	struct amdgpu_bo *obj = (struct amdgpu_bo *)handle;
+	r = amdgpu_bo_reserve(obj, false);
+	if (unlikely(r != 0))
+		return r;
+	amdgpu_bo_kunmap(obj);
+	amdgpu_bo_unreserve(obj);
+	return r;
+}
+
+static uint32_t amdgpu_cgs_read_register(void *cgs_device, unsigned offset)
+{
+	CGS_FUNC_ADEV;
+	return RREG32(offset);
+}
+
+static void amdgpu_cgs_write_register(void *cgs_device, unsigned offset,
+				      uint32_t value)
+{
+	CGS_FUNC_ADEV;
+	WREG32(offset, value);
+}
+
+static uint32_t amdgpu_cgs_read_ind_register(void *cgs_device,
+					     enum cgs_ind_reg space,
+					     unsigned index)
+{
+	CGS_FUNC_ADEV;
+	switch (space) {
+	case CGS_IND_REG__MMIO:
+		return RREG32_IDX(index);
+	case CGS_IND_REG__PCIE:
+		return RREG32_PCIE(index);
+	case CGS_IND_REG__SMC:
+		return RREG32_SMC(index);
+	case CGS_IND_REG__UVD_CTX:
+		return RREG32_UVD_CTX(index);
+	case CGS_IND_REG__DIDT:
+		return RREG32_DIDT(index);
+	case CGS_IND_REG__AUDIO_ENDPT:
+		DRM_ERROR("audio endpt register access not implemented.\n");
+		return 0;
+	}
+	WARN(1, "Invalid indirect register space");
+	return 0;
+}
+
+static void amdgpu_cgs_write_ind_register(void *cgs_device,
+					  enum cgs_ind_reg space,
+					  unsigned index, uint32_t value)
+{
+	CGS_FUNC_ADEV;
+	switch (space) {
+	case CGS_IND_REG__MMIO:
+		return WREG32_IDX(index, value);
+	case CGS_IND_REG__PCIE:
+		return WREG32_PCIE(index, value);
+	case CGS_IND_REG__SMC:
+		return WREG32_SMC(index, value);
+	case CGS_IND_REG__UVD_CTX:
+		return WREG32_UVD_CTX(index, value);
+	case CGS_IND_REG__DIDT:
+		return WREG32_DIDT(index, value);
+	case CGS_IND_REG__AUDIO_ENDPT:
+		DRM_ERROR("audio endpt register access not implemented.\n");
+		return;
+	}
+	WARN(1, "Invalid indirect register space");
+}
+
+static uint8_t amdgpu_cgs_read_pci_config_byte(void *cgs_device, unsigned addr)
+{
+	CGS_FUNC_ADEV;
+	uint8_t val;
+	int ret = pci_read_config_byte(adev->pdev, addr, &val);
+	if (WARN(ret, "pci_read_config_byte error"))
+		return 0;
+	return val;
+}
+
+static uint16_t amdgpu_cgs_read_pci_config_word(void *cgs_device, unsigned addr)
+{
+	CGS_FUNC_ADEV;
+	uint16_t val;
+	int ret = pci_read_config_word(adev->pdev, addr, &val);
+	if (WARN(ret, "pci_read_config_word error"))
+		return 0;
+	return val;
+}
+
+static uint32_t amdgpu_cgs_read_pci_config_dword(void *cgs_device,
+						 unsigned addr)
+{
+	CGS_FUNC_ADEV;
+	uint32_t val;
+	int ret = pci_read_config_dword(adev->pdev, addr, &val);
+	if (WARN(ret, "pci_read_config_dword error"))
+		return 0;
+	return val;
+}
+
+static void amdgpu_cgs_write_pci_config_byte(void *cgs_device, unsigned addr,
+					     uint8_t value)
+{
+	CGS_FUNC_ADEV;
+	int ret = pci_write_config_byte(adev->pdev, addr, value);
+	WARN(ret, "pci_write_config_byte error");
+}
+
+static void amdgpu_cgs_write_pci_config_word(void *cgs_device, unsigned addr,
+					     uint16_t value)
+{
+	CGS_FUNC_ADEV;
+	int ret = pci_write_config_word(adev->pdev, addr, value);
+	WARN(ret, "pci_write_config_word error");
+}
+
+static void amdgpu_cgs_write_pci_config_dword(void *cgs_device, unsigned addr,
+					      uint32_t value)
+{
+	CGS_FUNC_ADEV;
+	int ret = pci_write_config_dword(adev->pdev, addr, value);
+	WARN(ret, "pci_write_config_dword error");
+}
+
+static const void *amdgpu_cgs_atom_get_data_table(void *cgs_device,
+						  unsigned table, uint16_t *size,
+						  uint8_t *frev, uint8_t *crev)
+{
+	CGS_FUNC_ADEV;
+	uint16_t data_start;
+
+	if (amdgpu_atom_parse_data_header(
+		    adev->mode_info.atom_context, table, size,
+		    frev, crev, &data_start))
+		return (uint8_t*)adev->mode_info.atom_context->bios +
+			data_start;
+
+	return NULL;
+}
+
+static int amdgpu_cgs_atom_get_cmd_table_revs(void *cgs_device, unsigned table,
+					      uint8_t *frev, uint8_t *crev)
+{
+	CGS_FUNC_ADEV;
+
+	if (amdgpu_atom_parse_cmd_header(
+		    adev->mode_info.atom_context, table,
+		    frev, crev))
+		return 0;
+
+	return -EINVAL;
+}
+
+static int amdgpu_cgs_atom_exec_cmd_table(void *cgs_device, unsigned table,
+					  void *args)
+{
+	CGS_FUNC_ADEV;
+
+	return amdgpu_atom_execute_table(
+		adev->mode_info.atom_context, table, args);
+}
+
+static int amdgpu_cgs_create_pm_request(void *cgs_device, cgs_handle_t *request)
+{
+	/* TODO */
+	return 0;
+}
+
+static int amdgpu_cgs_destroy_pm_request(void *cgs_device, cgs_handle_t request)
+{
+	/* TODO */
+	return 0;
+}
+
+static int amdgpu_cgs_set_pm_request(void *cgs_device, cgs_handle_t request,
+				     int active)
+{
+	/* TODO */
+	return 0;
+}
+
+static int amdgpu_cgs_pm_request_clock(void *cgs_device, cgs_handle_t request,
+				       enum cgs_clock clock, unsigned freq)
+{
+	/* TODO */
+	return 0;
+}
+
+static int amdgpu_cgs_pm_request_engine(void *cgs_device, cgs_handle_t request,
+					enum cgs_engine engine, int powered)
+{
+	/* TODO */
+	return 0;
+}
+
+
+
+static int amdgpu_cgs_pm_query_clock_limits(void *cgs_device,
+					    enum cgs_clock clock,
+					    struct cgs_clock_limits *limits)
+{
+	/* TODO */
+	return 0;
+}
+
+static int amdgpu_cgs_set_camera_voltages(void *cgs_device, uint32_t mask,
+					  const uint32_t *voltages)
+{
+	DRM_ERROR("not implemented");
+	return -EPERM;
+}
+
+struct cgs_irq_params {
+	unsigned src_id;
+	cgs_irq_source_set_func_t set;
+	cgs_irq_handler_func_t handler;
+	void *private_data;
+};
+
+static int cgs_set_irq_state(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *src,
+			     unsigned type,
+			     enum amdgpu_interrupt_state state)
+{
+	struct cgs_irq_params *irq_params =
+		(struct cgs_irq_params *)src->data;
+	if (!irq_params)
+		return -EINVAL;
+	if (!irq_params->set)
+		return -EINVAL;
+	return irq_params->set(irq_params->private_data,
+			       irq_params->src_id,
+			       type,
+			       (int)state);
+}
+
+static int cgs_process_irq(struct amdgpu_device *adev,
+			   struct amdgpu_irq_src *source,
+			   struct amdgpu_iv_entry *entry)
+{
+	struct cgs_irq_params *irq_params =
+		(struct cgs_irq_params *)source->data;
+	if (!irq_params)
+		return -EINVAL;
+	if (!irq_params->handler)
+		return -EINVAL;
+	return irq_params->handler(irq_params->private_data,
+				   irq_params->src_id,
+				   entry->iv_entry);
+}
+
+static const struct amdgpu_irq_src_funcs cgs_irq_funcs = {
+	.set = cgs_set_irq_state,
+	.process = cgs_process_irq,
+};
+
+static int amdgpu_cgs_add_irq_source(void *cgs_device, unsigned src_id,
+				     unsigned num_types,
+				     cgs_irq_source_set_func_t set,
+				     cgs_irq_handler_func_t handler,
+				     void *private_data)
+{
+	CGS_FUNC_ADEV;
+	int ret = 0;
+	struct cgs_irq_params *irq_params;
+	struct amdgpu_irq_src *source =
+		kzalloc(sizeof(struct amdgpu_irq_src), GFP_KERNEL);
+	if (!source)
+		return -ENOMEM;
+	irq_params =
+		kzalloc(sizeof(struct cgs_irq_params), GFP_KERNEL);
+	if (!irq_params) {
+		kfree(source);
+		return -ENOMEM;
+	}
+	source->num_types = num_types;
+	source->funcs = &cgs_irq_funcs;
+	irq_params->src_id = src_id;
+	irq_params->set = set;
+	irq_params->handler = handler;
+	irq_params->private_data = private_data;
+	source->data = (void *)irq_params;
+	ret = amdgpu_irq_add_id(adev, src_id, source);
+	if (ret) {
+		kfree(irq_params);
+		kfree(source);
+	}
+
+	return ret;
+}
+
+static int amdgpu_cgs_irq_get(void *cgs_device, unsigned src_id, unsigned type)
+{
+	CGS_FUNC_ADEV;
+	return amdgpu_irq_get(adev, adev->irq.sources[src_id], type);
+}
+
+static int amdgpu_cgs_irq_put(void *cgs_device, unsigned src_id, unsigned type)
+{
+	CGS_FUNC_ADEV;
+	return amdgpu_irq_put(adev, adev->irq.sources[src_id], type);
+}
+
+int amdgpu_cgs_set_clockgating_state(void *cgs_device,
+				  enum amd_ip_block_type block_type,
+				  enum amd_clockgating_state state)
+{
+	CGS_FUNC_ADEV;
+	int i, r = -1;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+
+		if (adev->ip_blocks[i].type == block_type) {
+			r = adev->ip_blocks[i].funcs->set_clockgating_state(
+								(void *)adev,
+									state);
+			break;
+		}
+	}
+	return r;
+}
+
+int amdgpu_cgs_set_powergating_state(void *cgs_device,
+				  enum amd_ip_block_type block_type,
+				  enum amd_powergating_state state)
+{
+	CGS_FUNC_ADEV;
+	int i, r = -1;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+
+		if (adev->ip_blocks[i].type == block_type) {
+			r = adev->ip_blocks[i].funcs->set_powergating_state(
+								(void *)adev,
+									state);
+			break;
+		}
+	}
+	return r;
+}
+
+
+static uint32_t fw_type_convert(void *cgs_device, uint32_t fw_type)
+{
+	CGS_FUNC_ADEV;
+	enum AMDGPU_UCODE_ID result = AMDGPU_UCODE_ID_MAXIMUM;
+
+	switch (fw_type) {
+	case CGS_UCODE_ID_SDMA0:
+		result = AMDGPU_UCODE_ID_SDMA0;
+		break;
+	case CGS_UCODE_ID_SDMA1:
+		result = AMDGPU_UCODE_ID_SDMA1;
+		break;
+	case CGS_UCODE_ID_CP_CE:
+		result = AMDGPU_UCODE_ID_CP_CE;
+		break;
+	case CGS_UCODE_ID_CP_PFP:
+		result = AMDGPU_UCODE_ID_CP_PFP;
+		break;
+	case CGS_UCODE_ID_CP_ME:
+		result = AMDGPU_UCODE_ID_CP_ME;
+		break;
+	case CGS_UCODE_ID_CP_MEC:
+	case CGS_UCODE_ID_CP_MEC_JT1:
+		result = AMDGPU_UCODE_ID_CP_MEC1;
+		break;
+	case CGS_UCODE_ID_CP_MEC_JT2:
+		if (adev->asic_type == CHIP_TONGA)
+			result = AMDGPU_UCODE_ID_CP_MEC2;
+		else if (adev->asic_type == CHIP_CARRIZO)
+			result = AMDGPU_UCODE_ID_CP_MEC1;
+		break;
+	case CGS_UCODE_ID_RLC_G:
+		result = AMDGPU_UCODE_ID_RLC_G;
+		break;
+	default:
+		DRM_ERROR("Firmware type not supported\n");
+	}
+	return result;
+}
+
+static int amdgpu_cgs_get_firmware_info(void *cgs_device,
+					enum cgs_ucode_id type,
+					struct cgs_firmware_info *info)
+{
+	CGS_FUNC_ADEV;
+
+	if (CGS_UCODE_ID_SMU != type) {
+		uint64_t gpu_addr;
+		uint32_t data_size;
+		const struct gfx_firmware_header_v1_0 *header;
+		enum AMDGPU_UCODE_ID id;
+		struct amdgpu_firmware_info *ucode;
+
+		id = fw_type_convert(cgs_device, type);
+		ucode = &adev->firmware.ucode[id];
+		if (ucode->fw == NULL)
+			return -EINVAL;
+
+		gpu_addr  = ucode->mc_addr;
+		header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+		data_size = le32_to_cpu(header->header.ucode_size_bytes);
+
+		if ((type == CGS_UCODE_ID_CP_MEC_JT1) ||
+		    (type == CGS_UCODE_ID_CP_MEC_JT2)) {
+			gpu_addr += le32_to_cpu(header->jt_offset) << 2;
+			data_size = le32_to_cpu(header->jt_size) << 2;
+		}
+		info->mc_addr = gpu_addr;
+		info->image_size = data_size;
+		info->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
+		info->feature_version = (uint16_t)le32_to_cpu(header->ucode_feature_version);
+	} else {
+		char fw_name[30] = {0};
+		int err = 0;
+		uint32_t ucode_size;
+		uint32_t ucode_start_address;
+		const uint8_t *src;
+		const struct smc_firmware_header_v1_0 *hdr;
+
+		switch (adev->asic_type) {
+		case CHIP_TONGA:
+			strcpy(fw_name, "amdgpu/tonga_smc.bin");
+			break;
+		default:
+			DRM_ERROR("SMC firmware not supported\n");
+			return -EINVAL;
+		}
+
+		err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+		if (err) {
+			DRM_ERROR("Failed to request firmware\n");
+			return err;
+		}
+
+		err = amdgpu_ucode_validate(adev->pm.fw);
+		if (err) {
+			DRM_ERROR("Failed to load firmware \"%s\"", fw_name);
+			release_firmware(adev->pm.fw);
+			adev->pm.fw = NULL;
+			return err;
+		}
+
+		hdr = (const struct smc_firmware_header_v1_0 *)	adev->pm.fw->data;
+		adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+		ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+		ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+		src = (const uint8_t *)(adev->pm.fw->data +
+		       le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+		info->version = adev->pm.fw_version;
+		info->image_size = ucode_size;
+		info->kptr = (void *)src;
+	}
+	return 0;
+}
+
+static const struct cgs_ops amdgpu_cgs_ops = {
+	amdgpu_cgs_gpu_mem_info,
+	amdgpu_cgs_gmap_kmem,
+	amdgpu_cgs_gunmap_kmem,
+	amdgpu_cgs_alloc_gpu_mem,
+	amdgpu_cgs_free_gpu_mem,
+	amdgpu_cgs_gmap_gpu_mem,
+	amdgpu_cgs_gunmap_gpu_mem,
+	amdgpu_cgs_kmap_gpu_mem,
+	amdgpu_cgs_kunmap_gpu_mem,
+	amdgpu_cgs_read_register,
+	amdgpu_cgs_write_register,
+	amdgpu_cgs_read_ind_register,
+	amdgpu_cgs_write_ind_register,
+	amdgpu_cgs_read_pci_config_byte,
+	amdgpu_cgs_read_pci_config_word,
+	amdgpu_cgs_read_pci_config_dword,
+	amdgpu_cgs_write_pci_config_byte,
+	amdgpu_cgs_write_pci_config_word,
+	amdgpu_cgs_write_pci_config_dword,
+	amdgpu_cgs_atom_get_data_table,
+	amdgpu_cgs_atom_get_cmd_table_revs,
+	amdgpu_cgs_atom_exec_cmd_table,
+	amdgpu_cgs_create_pm_request,
+	amdgpu_cgs_destroy_pm_request,
+	amdgpu_cgs_set_pm_request,
+	amdgpu_cgs_pm_request_clock,
+	amdgpu_cgs_pm_request_engine,
+	amdgpu_cgs_pm_query_clock_limits,
+	amdgpu_cgs_set_camera_voltages,
+	amdgpu_cgs_get_firmware_info,
+	amdgpu_cgs_set_powergating_state,
+	amdgpu_cgs_set_clockgating_state
+};
+
+static const struct cgs_os_ops amdgpu_cgs_os_ops = {
+	amdgpu_cgs_add_irq_source,
+	amdgpu_cgs_irq_get,
+	amdgpu_cgs_irq_put
+};
+
+void *amdgpu_cgs_create_device(struct amdgpu_device *adev)
+{
+	struct amdgpu_cgs_device *cgs_device =
+		kmalloc(sizeof(*cgs_device), GFP_KERNEL);
+
+	if (!cgs_device) {
+		DRM_ERROR("Couldn't allocate CGS device structure\n");
+		return NULL;
+	}
+
+	cgs_device->base.ops = &amdgpu_cgs_ops;
+	cgs_device->base.os_ops = &amdgpu_cgs_os_ops;
+	cgs_device->adev = adev;
+
+	return cgs_device;
+}
+
+void amdgpu_cgs_destroy_device(void *cgs_device)
+{
+	kfree(cgs_device);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
new file mode 100644
index 000000000..119cdc2c4
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
@@ -0,0 +1,1912 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_fb_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+#include "atombios_dp.h"
+#include "amdgpu_connectors.h"
+#include "amdgpu_i2c.h"
+
+#include <linux/pm_runtime.h>
+
+void amdgpu_connector_hotplug(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	/* bail if the connector does not have hpd pin, e.g.,
+	 * VGA, TV, etc.
+	 */
+	if (amdgpu_connector->hpd.hpd == AMDGPU_HPD_NONE)
+		return;
+
+	amdgpu_display_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+
+	/* if the connector is already off, don't turn it back on */
+	if (connector->dpms != DRM_MODE_DPMS_ON)
+		return;
+
+	/* just deal with DP (not eDP) here. */
+	if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) {
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		/* if existing sink type was not DP no need to retrain */
+		if (dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_DISPLAYPORT)
+			return;
+
+		/* first get sink type as it may be reset after (un)plug */
+		dig_connector->dp_sink_type = amdgpu_atombios_dp_get_sinktype(amdgpu_connector);
+		/* don't do anything if sink is not display port, i.e.,
+		 * passive dp->(dvi|hdmi) adaptor
+		 */
+		if (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) {
+			int saved_dpms = connector->dpms;
+			/* Only turn off the display if it's physically disconnected */
+			if (!amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
+				drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
+			} else if (amdgpu_atombios_dp_needs_link_train(amdgpu_connector)) {
+				/* Don't try to start link training before we
+				 * have the dpcd */
+				if (amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+					return;
+
+				/* set it to OFF so that drm_helper_connector_dpms()
+				 * won't return immediately since the current state
+				 * is ON at this point.
+				 */
+				connector->dpms = DRM_MODE_DPMS_OFF;
+				drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
+			}
+			connector->dpms = saved_dpms;
+		}
+	}
+}
+
+static void amdgpu_connector_property_change_mode(struct drm_encoder *encoder)
+{
+	struct drm_crtc *crtc = encoder->crtc;
+
+	if (crtc && crtc->enabled) {
+		drm_crtc_helper_set_mode(crtc, &crtc->mode,
+					 crtc->x, crtc->y, crtc->primary->fb);
+	}
+}
+
+int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	int bpc = 8;
+	unsigned mode_clock, max_tmds_clock;
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_DVII:
+	case DRM_MODE_CONNECTOR_HDMIB:
+		if (amdgpu_connector->use_digital) {
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				if (connector->display_info.bpc)
+					bpc = connector->display_info.bpc;
+			}
+		}
+		break;
+	case DRM_MODE_CONNECTOR_DVID:
+	case DRM_MODE_CONNECTOR_HDMIA:
+		if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.bpc)
+				bpc = connector->display_info.bpc;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		dig_connector = amdgpu_connector->con_priv;
+		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) ||
+		    drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			if (connector->display_info.bpc)
+				bpc = connector->display_info.bpc;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_eDP:
+	case DRM_MODE_CONNECTOR_LVDS:
+		if (connector->display_info.bpc)
+			bpc = connector->display_info.bpc;
+		else {
+			const struct drm_connector_helper_funcs *connector_funcs =
+				connector->helper_private;
+			struct drm_encoder *encoder = connector_funcs->best_encoder(connector);
+			struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+			struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+			if (dig->lcd_misc & ATOM_PANEL_MISC_V13_6BIT_PER_COLOR)
+				bpc = 6;
+			else if (dig->lcd_misc & ATOM_PANEL_MISC_V13_8BIT_PER_COLOR)
+				bpc = 8;
+		}
+		break;
+	}
+
+	if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+		/*
+		 * Pre DCE-8 hw can't handle > 12 bpc, and more than 12 bpc doesn't make
+		 * much sense without support for > 12 bpc framebuffers. RGB 4:4:4 at
+		 * 12 bpc is always supported on hdmi deep color sinks, as this is
+		 * required by the HDMI-1.3 spec. Clamp to a safe 12 bpc maximum.
+		 */
+		if (bpc > 12) {
+			DRM_DEBUG("%s: HDMI deep color %d bpc unsupported. Using 12 bpc.\n",
+				  connector->name, bpc);
+			bpc = 12;
+		}
+
+		/* Any defined maximum tmds clock limit we must not exceed? */
+		if (connector->max_tmds_clock > 0) {
+			/* mode_clock is clock in kHz for mode to be modeset on this connector */
+			mode_clock = amdgpu_connector->pixelclock_for_modeset;
+
+			/* Maximum allowable input clock in kHz */
+			max_tmds_clock = connector->max_tmds_clock * 1000;
+
+			DRM_DEBUG("%s: hdmi mode dotclock %d kHz, max tmds input clock %d kHz.\n",
+				  connector->name, mode_clock, max_tmds_clock);
+
+			/* Check if bpc is within clock limit. Try to degrade gracefully otherwise */
+			if ((bpc == 12) && (mode_clock * 3/2 > max_tmds_clock)) {
+				if ((connector->display_info.edid_hdmi_dc_modes & DRM_EDID_HDMI_DC_30) &&
+				    (mode_clock * 5/4 <= max_tmds_clock))
+					bpc = 10;
+				else
+					bpc = 8;
+
+				DRM_DEBUG("%s: HDMI deep color 12 bpc exceeds max tmds clock. Using %d bpc.\n",
+					  connector->name, bpc);
+			}
+
+			if ((bpc == 10) && (mode_clock * 5/4 > max_tmds_clock)) {
+				bpc = 8;
+				DRM_DEBUG("%s: HDMI deep color 10 bpc exceeds max tmds clock. Using %d bpc.\n",
+					  connector->name, bpc);
+			} else if (bpc > 8) {
+				/* max_tmds_clock missing, but hdmi spec mandates it for deep color. */
+				DRM_DEBUG("%s: Required max tmds clock for HDMI deep color missing. Using 8 bpc.\n",
+					  connector->name);
+				bpc = 8;
+			}
+		}
+	}
+
+	if ((amdgpu_deep_color == 0) && (bpc > 8)) {
+		DRM_DEBUG("%s: Deep color disabled. Set amdgpu module param deep_color=1 to enable.\n",
+			  connector->name);
+		bpc = 8;
+	}
+
+	DRM_DEBUG("%s: Display bpc=%d, returned bpc=%d\n",
+		  connector->name, connector->display_info.bpc, bpc);
+
+	return bpc;
+}
+
+static void
+amdgpu_connector_update_scratch_regs(struct drm_connector *connector,
+				      enum drm_connector_status status)
+{
+	struct drm_encoder *best_encoder = NULL;
+	struct drm_encoder *encoder = NULL;
+	const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+	bool connected;
+	int i;
+
+	best_encoder = connector_funcs->best_encoder(connector);
+
+	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
+		if (connector->encoder_ids[i] == 0)
+			break;
+
+		encoder = drm_encoder_find(connector->dev,
+					connector->encoder_ids[i]);
+		if (!encoder)
+			continue;
+
+		if ((encoder == best_encoder) && (status == connector_status_connected))
+			connected = true;
+		else
+			connected = false;
+
+		amdgpu_atombios_encoder_set_bios_scratch_regs(connector, encoder, connected);
+
+	}
+}
+
+static struct drm_encoder *
+amdgpu_connector_find_encoder(struct drm_connector *connector,
+			       int encoder_type)
+{
+	struct drm_encoder *encoder;
+	int i;
+
+	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
+		if (connector->encoder_ids[i] == 0)
+			break;
+		encoder = drm_encoder_find(connector->dev,
+					connector->encoder_ids[i]);
+		if (!encoder)
+			continue;
+
+		if (encoder->encoder_type == encoder_type)
+			return encoder;
+	}
+	return NULL;
+}
+
+struct edid *amdgpu_connector_edid(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_property_blob *edid_blob = connector->edid_blob_ptr;
+
+	if (amdgpu_connector->edid) {
+		return amdgpu_connector->edid;
+	} else if (edid_blob) {
+		struct edid *edid = kmemdup(edid_blob->data, edid_blob->length, GFP_KERNEL);
+		if (edid)
+			amdgpu_connector->edid = edid;
+	}
+	return amdgpu_connector->edid;
+}
+
+static struct edid *
+amdgpu_connector_get_hardcoded_edid(struct amdgpu_device *adev)
+{
+	struct edid *edid;
+
+	if (adev->mode_info.bios_hardcoded_edid) {
+		edid = kmalloc(adev->mode_info.bios_hardcoded_edid_size, GFP_KERNEL);
+		if (edid) {
+			memcpy((unsigned char *)edid,
+			       (unsigned char *)adev->mode_info.bios_hardcoded_edid,
+			       adev->mode_info.bios_hardcoded_edid_size);
+			return edid;
+		}
+	}
+	return NULL;
+}
+
+static void amdgpu_connector_get_edid(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_connector->edid)
+		return;
+
+	/* on hw with routers, select right port */
+	if (amdgpu_connector->router.ddc_valid)
+		amdgpu_i2c_router_select_ddc_port(amdgpu_connector);
+
+	if ((amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+	     ENCODER_OBJECT_ID_NONE) &&
+	    amdgpu_connector->ddc_bus->has_aux) {
+		amdgpu_connector->edid = drm_get_edid(connector,
+						      &amdgpu_connector->ddc_bus->aux.ddc);
+	} else if ((connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) ||
+		   (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) {
+		struct amdgpu_connector_atom_dig *dig = amdgpu_connector->con_priv;
+
+		if ((dig->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT ||
+		     dig->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) &&
+		    amdgpu_connector->ddc_bus->has_aux)
+			amdgpu_connector->edid = drm_get_edid(connector,
+							      &amdgpu_connector->ddc_bus->aux.ddc);
+		else if (amdgpu_connector->ddc_bus)
+			amdgpu_connector->edid = drm_get_edid(connector,
+							      &amdgpu_connector->ddc_bus->adapter);
+	} else if (amdgpu_connector->ddc_bus) {
+		amdgpu_connector->edid = drm_get_edid(connector,
+						      &amdgpu_connector->ddc_bus->adapter);
+	}
+
+	if (!amdgpu_connector->edid) {
+		/* some laptops provide a hardcoded edid in rom for LCDs */
+		if (((connector->connector_type == DRM_MODE_CONNECTOR_LVDS) ||
+		     (connector->connector_type == DRM_MODE_CONNECTOR_eDP)))
+			amdgpu_connector->edid = amdgpu_connector_get_hardcoded_edid(adev);
+	}
+}
+
+static void amdgpu_connector_free_edid(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_connector->edid) {
+		kfree(amdgpu_connector->edid);
+		amdgpu_connector->edid = NULL;
+	}
+}
+
+static int amdgpu_connector_ddc_get_modes(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	int ret;
+
+	if (amdgpu_connector->edid) {
+		drm_mode_connector_update_edid_property(connector, amdgpu_connector->edid);
+		ret = drm_add_edid_modes(connector, amdgpu_connector->edid);
+		drm_edid_to_eld(connector, amdgpu_connector->edid);
+		return ret;
+	}
+	drm_mode_connector_update_edid_property(connector, NULL);
+	return 0;
+}
+
+static struct drm_encoder *
+amdgpu_connector_best_single_encoder(struct drm_connector *connector)
+{
+	int enc_id = connector->encoder_ids[0];
+
+	/* pick the encoder ids */
+	if (enc_id)
+		return drm_encoder_find(connector->dev, enc_id);
+	return NULL;
+}
+
+static void amdgpu_get_native_mode(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	if (encoder == NULL)
+		return;
+
+	amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	if (!list_empty(&connector->probed_modes)) {
+		struct drm_display_mode *preferred_mode =
+			list_first_entry(&connector->probed_modes,
+					 struct drm_display_mode, head);
+
+		amdgpu_encoder->native_mode = *preferred_mode;
+	} else {
+		amdgpu_encoder->native_mode.clock = 0;
+	}
+}
+
+static struct drm_display_mode *
+amdgpu_connector_lcd_native_mode(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+	if (native_mode->hdisplay != 0 &&
+	    native_mode->vdisplay != 0 &&
+	    native_mode->clock != 0) {
+		mode = drm_mode_duplicate(dev, native_mode);
+		mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER;
+		drm_mode_set_name(mode);
+
+		DRM_DEBUG_KMS("Adding native panel mode %s\n", mode->name);
+	} else if (native_mode->hdisplay != 0 &&
+		   native_mode->vdisplay != 0) {
+		/* mac laptops without an edid */
+		/* Note that this is not necessarily the exact panel mode,
+		 * but an approximation based on the cvt formula.  For these
+		 * systems we should ideally read the mode info out of the
+		 * registers or add a mode table, but this works and is much
+		 * simpler.
+		 */
+		mode = drm_cvt_mode(dev, native_mode->hdisplay, native_mode->vdisplay, 60, true, false, false);
+		mode->type = DRM_MODE_TYPE_PREFERRED | DRM_MODE_TYPE_DRIVER;
+		DRM_DEBUG_KMS("Adding cvt approximation of native panel mode %s\n", mode->name);
+	}
+	return mode;
+}
+
+static void amdgpu_connector_add_common_modes(struct drm_encoder *encoder,
+					       struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	int i;
+	struct mode_size {
+		int w;
+		int h;
+	} common_modes[17] = {
+		{ 640,  480},
+		{ 720,  480},
+		{ 800,  600},
+		{ 848,  480},
+		{1024,  768},
+		{1152,  768},
+		{1280,  720},
+		{1280,  800},
+		{1280,  854},
+		{1280,  960},
+		{1280, 1024},
+		{1440,  900},
+		{1400, 1050},
+		{1680, 1050},
+		{1600, 1200},
+		{1920, 1080},
+		{1920, 1200}
+	};
+
+	for (i = 0; i < 17; i++) {
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_TV_SUPPORT)) {
+			if (common_modes[i].w > 1024 ||
+			    common_modes[i].h > 768)
+				continue;
+		}
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			if (common_modes[i].w > native_mode->hdisplay ||
+			    common_modes[i].h > native_mode->vdisplay ||
+			    (common_modes[i].w == native_mode->hdisplay &&
+			     common_modes[i].h == native_mode->vdisplay))
+				continue;
+		}
+		if (common_modes[i].w < 320 || common_modes[i].h < 200)
+			continue;
+
+		mode = drm_cvt_mode(dev, common_modes[i].w, common_modes[i].h, 60, false, false, false);
+		drm_mode_probed_add(connector, mode);
+	}
+}
+
+static int amdgpu_connector_set_property(struct drm_connector *connector,
+					  struct drm_property *property,
+					  uint64_t val)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	if (property == adev->mode_info.coherent_mode_property) {
+		struct amdgpu_encoder_atom_dig *dig;
+		bool new_coherent_mode;
+
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (!amdgpu_encoder->enc_priv)
+			return 0;
+
+		dig = amdgpu_encoder->enc_priv;
+		new_coherent_mode = val ? true : false;
+		if (dig->coherent_mode != new_coherent_mode) {
+			dig->coherent_mode = new_coherent_mode;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.audio_property) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_connector->audio != val) {
+			amdgpu_connector->audio = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.dither_property) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_connector->dither != val) {
+			amdgpu_connector->dither = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.underscan_property) {
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_encoder->underscan_type != val) {
+			amdgpu_encoder->underscan_type = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.underscan_hborder_property) {
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_encoder->underscan_hborder != val) {
+			amdgpu_encoder->underscan_hborder = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.underscan_vborder_property) {
+		/* need to find digital encoder on connector */
+		encoder = amdgpu_connector_find_encoder(connector, DRM_MODE_ENCODER_TMDS);
+		if (!encoder)
+			return 0;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		if (amdgpu_encoder->underscan_vborder != val) {
+			amdgpu_encoder->underscan_vborder = val;
+			amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+		}
+	}
+
+	if (property == adev->mode_info.load_detect_property) {
+		struct amdgpu_connector *amdgpu_connector =
+			to_amdgpu_connector(connector);
+
+		if (val == 0)
+			amdgpu_connector->dac_load_detect = false;
+		else
+			amdgpu_connector->dac_load_detect = true;
+	}
+
+	if (property == dev->mode_config.scaling_mode_property) {
+		enum amdgpu_rmx_type rmx_type;
+
+		if (connector->encoder) {
+			amdgpu_encoder = to_amdgpu_encoder(connector->encoder);
+		} else {
+			const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+			amdgpu_encoder = to_amdgpu_encoder(connector_funcs->best_encoder(connector));
+		}
+
+		switch (val) {
+		default:
+		case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break;
+		case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break;
+		case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break;
+		case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break;
+		}
+		if (amdgpu_encoder->rmx_type == rmx_type)
+			return 0;
+
+		if ((rmx_type != DRM_MODE_SCALE_NONE) &&
+		    (amdgpu_encoder->native_mode.clock == 0))
+			return 0;
+
+		amdgpu_encoder->rmx_type = rmx_type;
+
+		amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+	}
+
+	return 0;
+}
+
+static void
+amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder,
+					struct drm_connector *connector)
+{
+	struct amdgpu_encoder *amdgpu_encoder =	to_amdgpu_encoder(encoder);
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	struct drm_display_mode *t, *mode;
+
+	/* If the EDID preferred mode doesn't match the native mode, use it */
+	list_for_each_entry_safe(mode, t, &connector->probed_modes, head) {
+		if (mode->type & DRM_MODE_TYPE_PREFERRED) {
+			if (mode->hdisplay != native_mode->hdisplay ||
+			    mode->vdisplay != native_mode->vdisplay)
+				memcpy(native_mode, mode, sizeof(*mode));
+		}
+	}
+
+	/* Try to get native mode details from EDID if necessary */
+	if (!native_mode->clock) {
+		list_for_each_entry_safe(mode, t, &connector->probed_modes, head) {
+			if (mode->hdisplay == native_mode->hdisplay &&
+			    mode->vdisplay == native_mode->vdisplay) {
+				*native_mode = *mode;
+				drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V);
+				DRM_DEBUG_KMS("Determined LVDS native mode details from EDID\n");
+				break;
+			}
+		}
+	}
+
+	if (!native_mode->clock) {
+		DRM_DEBUG_KMS("No LVDS native mode details, disabling RMX\n");
+		amdgpu_encoder->rmx_type = RMX_OFF;
+	}
+}
+
+static int amdgpu_connector_lvds_get_modes(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	int ret = 0;
+	struct drm_display_mode *mode;
+
+	amdgpu_connector_get_edid(connector);
+	ret = amdgpu_connector_ddc_get_modes(connector);
+	if (ret > 0) {
+		encoder = amdgpu_connector_best_single_encoder(connector);
+		if (encoder) {
+			amdgpu_connector_fixup_lcd_native_mode(encoder, connector);
+			/* add scaled modes */
+			amdgpu_connector_add_common_modes(encoder, connector);
+		}
+		return ret;
+	}
+
+	encoder = amdgpu_connector_best_single_encoder(connector);
+	if (!encoder)
+		return 0;
+
+	/* we have no EDID modes */
+	mode = amdgpu_connector_lcd_native_mode(encoder);
+	if (mode) {
+		ret = 1;
+		drm_mode_probed_add(connector, mode);
+		/* add the width/height from vbios tables if available */
+		connector->display_info.width_mm = mode->width_mm;
+		connector->display_info.height_mm = mode->height_mm;
+		/* add scaled modes */
+		amdgpu_connector_add_common_modes(encoder, connector);
+	}
+
+	return ret;
+}
+
+static int amdgpu_connector_lvds_mode_valid(struct drm_connector *connector,
+					     struct drm_display_mode *mode)
+{
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+
+	if ((mode->hdisplay < 320) || (mode->vdisplay < 240))
+		return MODE_PANEL;
+
+	if (encoder) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+		/* AVIVO hardware supports downscaling modes larger than the panel
+		 * to the panel size, but I'm not sure this is desirable.
+		 */
+		if ((mode->hdisplay > native_mode->hdisplay) ||
+		    (mode->vdisplay > native_mode->vdisplay))
+			return MODE_PANEL;
+
+		/* if scaling is disabled, block non-native modes */
+		if (amdgpu_encoder->rmx_type == RMX_OFF) {
+			if ((mode->hdisplay != native_mode->hdisplay) ||
+			    (mode->vdisplay != native_mode->vdisplay))
+				return MODE_PANEL;
+		}
+	}
+
+	return MODE_OK;
+}
+
+static enum drm_connector_status
+amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	enum drm_connector_status ret = connector_status_disconnected;
+	int r;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	if (encoder) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+		/* check if panel is valid */
+		if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240)
+			ret = connector_status_connected;
+
+	}
+
+	/* check for edid as well */
+	amdgpu_connector_get_edid(connector);
+	if (amdgpu_connector->edid)
+		ret = connector_status_connected;
+	/* check acpi lid status ??? */
+
+	amdgpu_connector_update_scratch_regs(connector, ret);
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+	return ret;
+}
+
+static void amdgpu_connector_destroy(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_connector->ddc_bus->has_aux)
+		drm_dp_aux_unregister(&amdgpu_connector->ddc_bus->aux);
+	amdgpu_connector_free_edid(connector);
+	kfree(amdgpu_connector->con_priv);
+	drm_connector_unregister(connector);
+	drm_connector_cleanup(connector);
+	kfree(connector);
+}
+
+static int amdgpu_connector_set_lcd_property(struct drm_connector *connector,
+					      struct drm_property *property,
+					      uint64_t value)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_encoder *amdgpu_encoder;
+	enum amdgpu_rmx_type rmx_type;
+
+	DRM_DEBUG_KMS("\n");
+	if (property != dev->mode_config.scaling_mode_property)
+		return 0;
+
+	if (connector->encoder)
+		amdgpu_encoder = to_amdgpu_encoder(connector->encoder);
+	else {
+		const struct drm_connector_helper_funcs *connector_funcs = connector->helper_private;
+		amdgpu_encoder = to_amdgpu_encoder(connector_funcs->best_encoder(connector));
+	}
+
+	switch (value) {
+	case DRM_MODE_SCALE_NONE: rmx_type = RMX_OFF; break;
+	case DRM_MODE_SCALE_CENTER: rmx_type = RMX_CENTER; break;
+	case DRM_MODE_SCALE_ASPECT: rmx_type = RMX_ASPECT; break;
+	default:
+	case DRM_MODE_SCALE_FULLSCREEN: rmx_type = RMX_FULL; break;
+	}
+	if (amdgpu_encoder->rmx_type == rmx_type)
+		return 0;
+
+	amdgpu_encoder->rmx_type = rmx_type;
+
+	amdgpu_connector_property_change_mode(&amdgpu_encoder->base);
+	return 0;
+}
+
+
+static const struct drm_connector_helper_funcs amdgpu_connector_lvds_helper_funcs = {
+	.get_modes = amdgpu_connector_lvds_get_modes,
+	.mode_valid = amdgpu_connector_lvds_mode_valid,
+	.best_encoder = amdgpu_connector_best_single_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_lvds_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_lvds_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = amdgpu_connector_destroy,
+	.set_property = amdgpu_connector_set_lcd_property,
+};
+
+static int amdgpu_connector_vga_get_modes(struct drm_connector *connector)
+{
+	int ret;
+
+	amdgpu_connector_get_edid(connector);
+	ret = amdgpu_connector_ddc_get_modes(connector);
+
+	return ret;
+}
+
+static int amdgpu_connector_vga_mode_valid(struct drm_connector *connector,
+					    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* XXX check mode bandwidth */
+
+	if ((mode->clock / 10) > adev->clock.max_pixel_clock)
+		return MODE_CLOCK_HIGH;
+
+	return MODE_OK;
+}
+
+static enum drm_connector_status
+amdgpu_connector_vga_detect(struct drm_connector *connector, bool force)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder;
+	const struct drm_encoder_helper_funcs *encoder_funcs;
+	bool dret = false;
+	enum drm_connector_status ret = connector_status_disconnected;
+	int r;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	encoder = amdgpu_connector_best_single_encoder(connector);
+	if (!encoder)
+		ret = connector_status_disconnected;
+
+	if (amdgpu_connector->ddc_bus)
+		dret = amdgpu_ddc_probe(amdgpu_connector, false);
+	if (dret) {
+		amdgpu_connector->detected_by_load = false;
+		amdgpu_connector_free_edid(connector);
+		amdgpu_connector_get_edid(connector);
+
+		if (!amdgpu_connector->edid) {
+			DRM_ERROR("%s: probed a monitor but no|invalid EDID\n",
+					connector->name);
+			ret = connector_status_connected;
+		} else {
+			amdgpu_connector->use_digital =
+				!!(amdgpu_connector->edid->input & DRM_EDID_INPUT_DIGITAL);
+
+			/* some oems have boards with separate digital and analog connectors
+			 * with a shared ddc line (often vga + hdmi)
+			 */
+			if (amdgpu_connector->use_digital && amdgpu_connector->shared_ddc) {
+				amdgpu_connector_free_edid(connector);
+				ret = connector_status_disconnected;
+			} else {
+				ret = connector_status_connected;
+			}
+		}
+	} else {
+
+		/* if we aren't forcing don't do destructive polling */
+		if (!force) {
+			/* only return the previous status if we last
+			 * detected a monitor via load.
+			 */
+			if (amdgpu_connector->detected_by_load)
+				ret = connector->status;
+			goto out;
+		}
+
+		if (amdgpu_connector->dac_load_detect && encoder) {
+			encoder_funcs = encoder->helper_private;
+			ret = encoder_funcs->detect(encoder, connector);
+			if (ret != connector_status_disconnected)
+				amdgpu_connector->detected_by_load = true;
+		}
+	}
+
+	amdgpu_connector_update_scratch_regs(connector, ret);
+
+out:
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+static const struct drm_connector_helper_funcs amdgpu_connector_vga_helper_funcs = {
+	.get_modes = amdgpu_connector_vga_get_modes,
+	.mode_valid = amdgpu_connector_vga_mode_valid,
+	.best_encoder = amdgpu_connector_best_single_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_vga_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_vga_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.destroy = amdgpu_connector_destroy,
+	.set_property = amdgpu_connector_set_property,
+};
+
+static bool
+amdgpu_connector_check_hpd_status_unchanged(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	enum drm_connector_status status;
+
+	if (amdgpu_connector->hpd.hpd != AMDGPU_HPD_NONE) {
+		if (amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd))
+			status = connector_status_connected;
+		else
+			status = connector_status_disconnected;
+		if (connector->status == status)
+			return true;
+	}
+
+	return false;
+}
+
+/*
+ * DVI is complicated
+ * Do a DDC probe, if DDC probe passes, get the full EDID so
+ * we can do analog/digital monitor detection at this point.
+ * If the monitor is an analog monitor or we got no DDC,
+ * we need to find the DAC encoder object for this connector.
+ * If we got no DDC, we do load detection on the DAC encoder object.
+ * If we got analog DDC or load detection passes on the DAC encoder
+ * we have to check if this analog encoder is shared with anyone else (TV)
+ * if its shared we have to set the other connector to disconnected.
+ */
+static enum drm_connector_status
+amdgpu_connector_dvi_detect(struct drm_connector *connector, bool force)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder = NULL;
+	const struct drm_encoder_helper_funcs *encoder_funcs;
+	int i, r;
+	enum drm_connector_status ret = connector_status_disconnected;
+	bool dret = false, broken_edid = false;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	if (!force && amdgpu_connector_check_hpd_status_unchanged(connector)) {
+		ret = connector->status;
+		goto exit;
+	}
+
+	if (amdgpu_connector->ddc_bus)
+		dret = amdgpu_ddc_probe(amdgpu_connector, false);
+	if (dret) {
+		amdgpu_connector->detected_by_load = false;
+		amdgpu_connector_free_edid(connector);
+		amdgpu_connector_get_edid(connector);
+
+		if (!amdgpu_connector->edid) {
+			DRM_ERROR("%s: probed a monitor but no|invalid EDID\n",
+					connector->name);
+			ret = connector_status_connected;
+			broken_edid = true; /* defer use_digital to later */
+		} else {
+			amdgpu_connector->use_digital =
+				!!(amdgpu_connector->edid->input & DRM_EDID_INPUT_DIGITAL);
+
+			/* some oems have boards with separate digital and analog connectors
+			 * with a shared ddc line (often vga + hdmi)
+			 */
+			if ((!amdgpu_connector->use_digital) && amdgpu_connector->shared_ddc) {
+				amdgpu_connector_free_edid(connector);
+				ret = connector_status_disconnected;
+			} else {
+				ret = connector_status_connected;
+			}
+
+			/* This gets complicated.  We have boards with VGA + HDMI with a
+			 * shared DDC line and we have boards with DVI-D + HDMI with a shared
+			 * DDC line.  The latter is more complex because with DVI<->HDMI adapters
+			 * you don't really know what's connected to which port as both are digital.
+			 */
+			if (amdgpu_connector->shared_ddc && (ret == connector_status_connected)) {
+				struct drm_connector *list_connector;
+				struct amdgpu_connector *list_amdgpu_connector;
+				list_for_each_entry(list_connector, &dev->mode_config.connector_list, head) {
+					if (connector == list_connector)
+						continue;
+					list_amdgpu_connector = to_amdgpu_connector(list_connector);
+					if (list_amdgpu_connector->shared_ddc &&
+					    (list_amdgpu_connector->ddc_bus->rec.i2c_id ==
+					     amdgpu_connector->ddc_bus->rec.i2c_id)) {
+						/* cases where both connectors are digital */
+						if (list_connector->connector_type != DRM_MODE_CONNECTOR_VGA) {
+							/* hpd is our only option in this case */
+							if (!amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
+								amdgpu_connector_free_edid(connector);
+								ret = connector_status_disconnected;
+							}
+						}
+					}
+				}
+			}
+		}
+	}
+
+	if ((ret == connector_status_connected) && (amdgpu_connector->use_digital == true))
+		goto out;
+
+	/* DVI-D and HDMI-A are digital only */
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_DVID) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_HDMIA))
+		goto out;
+
+	/* if we aren't forcing don't do destructive polling */
+	if (!force) {
+		/* only return the previous status if we last
+		 * detected a monitor via load.
+		 */
+		if (amdgpu_connector->detected_by_load)
+			ret = connector->status;
+		goto out;
+	}
+
+	/* find analog encoder */
+	if (amdgpu_connector->dac_load_detect) {
+		for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
+			if (connector->encoder_ids[i] == 0)
+				break;
+
+			encoder = drm_encoder_find(connector->dev, connector->encoder_ids[i]);
+			if (!encoder)
+				continue;
+
+			if (encoder->encoder_type != DRM_MODE_ENCODER_DAC &&
+			    encoder->encoder_type != DRM_MODE_ENCODER_TVDAC)
+				continue;
+
+			encoder_funcs = encoder->helper_private;
+			if (encoder_funcs->detect) {
+				if (!broken_edid) {
+					if (ret != connector_status_connected) {
+						/* deal with analog monitors without DDC */
+						ret = encoder_funcs->detect(encoder, connector);
+						if (ret == connector_status_connected) {
+							amdgpu_connector->use_digital = false;
+						}
+						if (ret != connector_status_disconnected)
+							amdgpu_connector->detected_by_load = true;
+					}
+				} else {
+					enum drm_connector_status lret;
+					/* assume digital unless load detected otherwise */
+					amdgpu_connector->use_digital = true;
+					lret = encoder_funcs->detect(encoder, connector);
+					DRM_DEBUG_KMS("load_detect %x returned: %x\n",encoder->encoder_type,lret);
+					if (lret == connector_status_connected)
+						amdgpu_connector->use_digital = false;
+				}
+				break;
+			}
+		}
+	}
+
+out:
+	/* updated in get modes as well since we need to know if it's analog or digital */
+	amdgpu_connector_update_scratch_regs(connector, ret);
+
+exit:
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+/* okay need to be smart in here about which encoder to pick */
+static struct drm_encoder *
+amdgpu_connector_dvi_encoder(struct drm_connector *connector)
+{
+	int enc_id = connector->encoder_ids[0];
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *encoder;
+	int i;
+	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
+		if (connector->encoder_ids[i] == 0)
+			break;
+
+		encoder = drm_encoder_find(connector->dev, connector->encoder_ids[i]);
+		if (!encoder)
+			continue;
+
+		if (amdgpu_connector->use_digital == true) {
+			if (encoder->encoder_type == DRM_MODE_ENCODER_TMDS)
+				return encoder;
+		} else {
+			if (encoder->encoder_type == DRM_MODE_ENCODER_DAC ||
+			    encoder->encoder_type == DRM_MODE_ENCODER_TVDAC)
+				return encoder;
+		}
+	}
+
+	/* see if we have a default encoder  TODO */
+
+	/* then check use digitial */
+	/* pick the first one */
+	if (enc_id)
+		return drm_encoder_find(connector->dev, enc_id);
+	return NULL;
+}
+
+static void amdgpu_connector_dvi_force(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	if (connector->force == DRM_FORCE_ON)
+		amdgpu_connector->use_digital = false;
+	if (connector->force == DRM_FORCE_ON_DIGITAL)
+		amdgpu_connector->use_digital = true;
+}
+
+static int amdgpu_connector_dvi_mode_valid(struct drm_connector *connector,
+					    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	/* XXX check mode bandwidth */
+
+	if (amdgpu_connector->use_digital && (mode->clock > 165000)) {
+		if ((amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_I) ||
+		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_DUAL_LINK_DVI_D) ||
+		    (amdgpu_connector->connector_object_id == CONNECTOR_OBJECT_ID_HDMI_TYPE_B)) {
+			return MODE_OK;
+		} else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+			/* HDMI 1.3+ supports max clock of 340 Mhz */
+			if (mode->clock > 340000)
+				return MODE_CLOCK_HIGH;
+			else
+				return MODE_OK;
+		} else {
+			return MODE_CLOCK_HIGH;
+		}
+	}
+
+	/* check against the max pixel clock */
+	if ((mode->clock / 10) > adev->clock.max_pixel_clock)
+		return MODE_CLOCK_HIGH;
+
+	return MODE_OK;
+}
+
+static const struct drm_connector_helper_funcs amdgpu_connector_dvi_helper_funcs = {
+	.get_modes = amdgpu_connector_vga_get_modes,
+	.mode_valid = amdgpu_connector_dvi_mode_valid,
+	.best_encoder = amdgpu_connector_dvi_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_dvi_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_dvi_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = amdgpu_connector_set_property,
+	.destroy = amdgpu_connector_destroy,
+	.force = amdgpu_connector_dvi_force,
+};
+
+static int amdgpu_connector_dp_get_modes(struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	int ret;
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		struct drm_display_mode *mode;
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+			if (!amdgpu_dig_connector->edp_on)
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_ON);
+			amdgpu_connector_get_edid(connector);
+			ret = amdgpu_connector_ddc_get_modes(connector);
+			if (!amdgpu_dig_connector->edp_on)
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+		} else {
+			/* need to setup ddc on the bridge */
+			if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+			    ENCODER_OBJECT_ID_NONE) {
+				if (encoder)
+					amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
+			}
+			amdgpu_connector_get_edid(connector);
+			ret = amdgpu_connector_ddc_get_modes(connector);
+		}
+
+		if (ret > 0) {
+			if (encoder) {
+				amdgpu_connector_fixup_lcd_native_mode(encoder, connector);
+				/* add scaled modes */
+				amdgpu_connector_add_common_modes(encoder, connector);
+			}
+			return ret;
+		}
+
+		if (!encoder)
+			return 0;
+
+		/* we have no EDID modes */
+		mode = amdgpu_connector_lcd_native_mode(encoder);
+		if (mode) {
+			ret = 1;
+			drm_mode_probed_add(connector, mode);
+			/* add the width/height from vbios tables if available */
+			connector->display_info.width_mm = mode->width_mm;
+			connector->display_info.height_mm = mode->height_mm;
+			/* add scaled modes */
+			amdgpu_connector_add_common_modes(encoder, connector);
+		}
+	} else {
+		/* need to setup ddc on the bridge */
+		if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+			ENCODER_OBJECT_ID_NONE) {
+			if (encoder)
+				amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
+		}
+		amdgpu_connector_get_edid(connector);
+		ret = amdgpu_connector_ddc_get_modes(connector);
+
+		amdgpu_get_native_mode(connector);
+	}
+
+	return ret;
+}
+
+u16 amdgpu_connector_encoder_get_dp_bridge_encoder_id(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	int i;
+
+	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
+		if (connector->encoder_ids[i] == 0)
+			break;
+
+		encoder = drm_encoder_find(connector->dev,
+					connector->encoder_ids[i]);
+		if (!encoder)
+			continue;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_TRAVIS:
+		case ENCODER_OBJECT_ID_NUTMEG:
+			return amdgpu_encoder->encoder_id;
+		default:
+			break;
+		}
+	}
+
+	return ENCODER_OBJECT_ID_NONE;
+}
+
+static bool amdgpu_connector_encoder_is_hbr2(struct drm_connector *connector)
+{
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	int i;
+	bool found = false;
+
+	for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) {
+		if (connector->encoder_ids[i] == 0)
+			break;
+		encoder = drm_encoder_find(connector->dev,
+					connector->encoder_ids[i]);
+		if (!encoder)
+			continue;
+
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->caps & ATOM_ENCODER_CAP_RECORD_HBR2)
+			found = true;
+	}
+
+	return found;
+}
+
+bool amdgpu_connector_is_dp12_capable(struct drm_connector *connector)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if ((adev->clock.default_dispclk >= 53900) &&
+	    amdgpu_connector_encoder_is_hbr2(connector)) {
+		return true;
+	}
+
+	return false;
+}
+
+static enum drm_connector_status
+amdgpu_connector_dp_detect(struct drm_connector *connector, bool force)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	enum drm_connector_status ret = connector_status_disconnected;
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
+	struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+	int r;
+
+	r = pm_runtime_get_sync(connector->dev->dev);
+	if (r < 0)
+		return connector_status_disconnected;
+
+	if (!force && amdgpu_connector_check_hpd_status_unchanged(connector)) {
+		ret = connector->status;
+		goto out;
+	}
+
+	amdgpu_connector_free_edid(connector);
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		if (encoder) {
+			struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+			struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+			/* check if panel is valid */
+			if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240)
+				ret = connector_status_connected;
+		}
+		/* eDP is always DP */
+		amdgpu_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT;
+		if (!amdgpu_dig_connector->edp_on)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+		if (!amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+			ret = connector_status_connected;
+		if (!amdgpu_dig_connector->edp_on)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+	} else if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) !=
+		   ENCODER_OBJECT_ID_NONE) {
+		/* DP bridges are always DP */
+		amdgpu_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT;
+		/* get the DPCD from the bridge */
+		amdgpu_atombios_dp_get_dpcd(amdgpu_connector);
+
+		if (encoder) {
+			/* setup ddc on the bridge */
+			amdgpu_atombios_encoder_setup_ext_encoder_ddc(encoder);
+			/* bridge chips are always aux */
+			if (amdgpu_ddc_probe(amdgpu_connector, true)) /* try DDC */
+				ret = connector_status_connected;
+			else if (amdgpu_connector->dac_load_detect) { /* try load detection */
+				const struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private;
+				ret = encoder_funcs->detect(encoder, connector);
+			}
+		}
+	} else {
+		amdgpu_dig_connector->dp_sink_type =
+			amdgpu_atombios_dp_get_sinktype(amdgpu_connector);
+		if (amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd)) {
+			ret = connector_status_connected;
+			if (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT)
+				amdgpu_atombios_dp_get_dpcd(amdgpu_connector);
+		} else {
+			if (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) {
+				if (!amdgpu_atombios_dp_get_dpcd(amdgpu_connector))
+					ret = connector_status_connected;
+			} else {
+				/* try non-aux ddc (DP to DVI/HDMI/etc. adapter) */
+				if (amdgpu_ddc_probe(amdgpu_connector, false))
+					ret = connector_status_connected;
+			}
+		}
+	}
+
+	amdgpu_connector_update_scratch_regs(connector, ret);
+out:
+	pm_runtime_mark_last_busy(connector->dev->dev);
+	pm_runtime_put_autosuspend(connector->dev->dev);
+
+	return ret;
+}
+
+static int amdgpu_connector_dp_mode_valid(struct drm_connector *connector,
+					   struct drm_display_mode *mode)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = amdgpu_connector->con_priv;
+
+	/* XXX check mode bandwidth */
+
+	if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) ||
+	    (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) {
+		struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector);
+
+		if ((mode->hdisplay < 320) || (mode->vdisplay < 240))
+			return MODE_PANEL;
+
+		if (encoder) {
+			struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+			struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+
+			/* AVIVO hardware supports downscaling modes larger than the panel
+			 * to the panel size, but I'm not sure this is desirable.
+			 */
+			if ((mode->hdisplay > native_mode->hdisplay) ||
+			    (mode->vdisplay > native_mode->vdisplay))
+				return MODE_PANEL;
+
+			/* if scaling is disabled, block non-native modes */
+			if (amdgpu_encoder->rmx_type == RMX_OFF) {
+				if ((mode->hdisplay != native_mode->hdisplay) ||
+				    (mode->vdisplay != native_mode->vdisplay))
+					return MODE_PANEL;
+			}
+		}
+		return MODE_OK;
+	} else {
+		if ((amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (amdgpu_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
+			return amdgpu_atombios_dp_mode_valid_helper(connector, mode);
+		} else {
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				/* HDMI 1.3+ supports max clock of 340 Mhz */
+				if (mode->clock > 340000)
+					return MODE_CLOCK_HIGH;
+			} else {
+				if (mode->clock > 165000)
+					return MODE_CLOCK_HIGH;
+			}
+		}
+	}
+
+	return MODE_OK;
+}
+
+static const struct drm_connector_helper_funcs amdgpu_connector_dp_helper_funcs = {
+	.get_modes = amdgpu_connector_dp_get_modes,
+	.mode_valid = amdgpu_connector_dp_mode_valid,
+	.best_encoder = amdgpu_connector_dvi_encoder,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_dp_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_dp_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = amdgpu_connector_set_property,
+	.destroy = amdgpu_connector_destroy,
+	.force = amdgpu_connector_dvi_force,
+};
+
+static const struct drm_connector_funcs amdgpu_connector_edp_funcs = {
+	.dpms = drm_helper_connector_dpms,
+	.detect = amdgpu_connector_dp_detect,
+	.fill_modes = drm_helper_probe_single_connector_modes,
+	.set_property = amdgpu_connector_set_lcd_property,
+	.destroy = amdgpu_connector_destroy,
+	.force = amdgpu_connector_dvi_force,
+};
+
+void
+amdgpu_connector_add(struct amdgpu_device *adev,
+		      uint32_t connector_id,
+		      uint32_t supported_device,
+		      int connector_type,
+		      struct amdgpu_i2c_bus_rec *i2c_bus,
+		      uint16_t connector_object_id,
+		      struct amdgpu_hpd *hpd,
+		      struct amdgpu_router *router)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	uint32_t subpixel_order = SubPixelNone;
+	bool shared_ddc = false;
+	bool is_dp_bridge = false;
+	bool has_aux = false;
+
+	if (connector_type == DRM_MODE_CONNECTOR_Unknown)
+		return;
+
+	/* see if we already added it */
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		if (amdgpu_connector->connector_id == connector_id) {
+			amdgpu_connector->devices |= supported_device;
+			return;
+		}
+		if (amdgpu_connector->ddc_bus && i2c_bus->valid) {
+			if (amdgpu_connector->ddc_bus->rec.i2c_id == i2c_bus->i2c_id) {
+				amdgpu_connector->shared_ddc = true;
+				shared_ddc = true;
+			}
+			if (amdgpu_connector->router_bus && router->ddc_valid &&
+			    (amdgpu_connector->router.router_id == router->router_id)) {
+				amdgpu_connector->shared_ddc = false;
+				shared_ddc = false;
+			}
+		}
+	}
+
+	/* check if it's a dp bridge */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->devices & supported_device) {
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_TRAVIS:
+			case ENCODER_OBJECT_ID_NUTMEG:
+				is_dp_bridge = true;
+				break;
+			default:
+				break;
+			}
+		}
+	}
+
+	amdgpu_connector = kzalloc(sizeof(struct amdgpu_connector), GFP_KERNEL);
+	if (!amdgpu_connector)
+		return;
+
+	connector = &amdgpu_connector->base;
+
+	amdgpu_connector->connector_id = connector_id;
+	amdgpu_connector->devices = supported_device;
+	amdgpu_connector->shared_ddc = shared_ddc;
+	amdgpu_connector->connector_object_id = connector_object_id;
+	amdgpu_connector->hpd = *hpd;
+
+	amdgpu_connector->router = *router;
+	if (router->ddc_valid || router->cd_valid) {
+		amdgpu_connector->router_bus = amdgpu_i2c_lookup(adev, &router->i2c_info);
+		if (!amdgpu_connector->router_bus)
+			DRM_ERROR("Failed to assign router i2c bus! Check dmesg for i2c errors.\n");
+	}
+
+	if (is_dp_bridge) {
+		amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+		if (!amdgpu_dig_connector)
+			goto failed;
+		amdgpu_connector->con_priv = amdgpu_dig_connector;
+		if (i2c_bus->valid) {
+			amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+			if (amdgpu_connector->ddc_bus)
+				has_aux = true;
+			else
+				DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+		}
+		switch (connector_type) {
+		case DRM_MODE_CONNECTOR_VGA:
+		case DRM_MODE_CONNECTOR_DVIA:
+		default:
+			drm_connector_init(dev, &amdgpu_connector->base,
+					   &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base,
+						 &amdgpu_connector_dp_helper_funcs);
+			connector->interlace_allowed = true;
+			connector->doublescan_allowed = true;
+			amdgpu_connector->dac_load_detect = true;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.load_detect_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			break;
+		case DRM_MODE_CONNECTOR_DVII:
+		case DRM_MODE_CONNECTOR_DVID:
+		case DRM_MODE_CONNECTOR_HDMIA:
+		case DRM_MODE_CONNECTOR_HDMIB:
+		case DRM_MODE_CONNECTOR_DisplayPort:
+			drm_connector_init(dev, &amdgpu_connector->base,
+					   &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base,
+						 &amdgpu_connector_dp_helper_funcs);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.underscan_property,
+						      UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.underscan_hborder_property,
+						      0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.underscan_vborder_property,
+						      0);
+
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+
+			if (amdgpu_audio != 0)
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = true;
+			if (connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				connector->doublescan_allowed = true;
+			else
+				connector->doublescan_allowed = false;
+			if (connector_type == DRM_MODE_CONNECTOR_DVII) {
+				amdgpu_connector->dac_load_detect = true;
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							      adev->mode_info.load_detect_property,
+							      1);
+			}
+			break;
+		case DRM_MODE_CONNECTOR_LVDS:
+		case DRM_MODE_CONNECTOR_eDP:
+			drm_connector_init(dev, &amdgpu_connector->base,
+					   &amdgpu_connector_edp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base,
+						 &amdgpu_connector_dp_helper_funcs);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      dev->mode_config.scaling_mode_property,
+						      DRM_MODE_SCALE_FULLSCREEN);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = false;
+			connector->doublescan_allowed = false;
+			break;
+		}
+	} else {
+		switch (connector_type) {
+		case DRM_MODE_CONNECTOR_VGA:
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_vga_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			amdgpu_connector->dac_load_detect = true;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.load_detect_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			/* no HPD on analog connectors */
+			amdgpu_connector->hpd.hpd = AMDGPU_HPD_NONE;
+			connector->polled = DRM_CONNECTOR_POLL_CONNECT;
+			connector->interlace_allowed = true;
+			connector->doublescan_allowed = true;
+			break;
+		case DRM_MODE_CONNECTOR_DVIA:
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_vga_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_vga_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			amdgpu_connector->dac_load_detect = true;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.load_detect_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			/* no HPD on analog connectors */
+			amdgpu_connector->hpd.hpd = AMDGPU_HPD_NONE;
+			connector->interlace_allowed = true;
+			connector->doublescan_allowed = true;
+			break;
+		case DRM_MODE_CONNECTOR_DVII:
+		case DRM_MODE_CONNECTOR_DVID:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dvi_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			subpixel_order = SubPixelHorizontalRGB;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.coherent_mode_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_property,
+						   UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_hborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_vborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+
+			if (amdgpu_audio != 0) {
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+			if (connector_type == DRM_MODE_CONNECTOR_DVII) {
+				amdgpu_connector->dac_load_detect = true;
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.load_detect_property,
+							   1);
+			}
+			connector->interlace_allowed = true;
+			if (connector_type == DRM_MODE_CONNECTOR_DVII)
+				connector->doublescan_allowed = true;
+			else
+				connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_HDMIA:
+		case DRM_MODE_CONNECTOR_HDMIB:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dvi_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dvi_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("HDMI: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.coherent_mode_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_property,
+						   UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_hborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_vborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			if (amdgpu_audio != 0) {
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = true;
+			if (connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				connector->doublescan_allowed = true;
+			else
+				connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_DisplayPort:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_dp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (amdgpu_connector->ddc_bus)
+					has_aux = true;
+				else
+					DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			subpixel_order = SubPixelHorizontalRGB;
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      adev->mode_info.coherent_mode_property,
+						      1);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_property,
+						   UNDERSCAN_OFF);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_hborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.underscan_vborder_property,
+						   0);
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   dev->mode_config.scaling_mode_property,
+						   DRM_MODE_SCALE_NONE);
+			if (amdgpu_audio != 0) {
+				drm_object_attach_property(&amdgpu_connector->base.base,
+							   adev->mode_info.audio_property,
+							   AMDGPU_AUDIO_AUTO);
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						   adev->mode_info.dither_property,
+						   AMDGPU_FMT_DITHER_DISABLE);
+			connector->interlace_allowed = true;
+			/* in theory with a DP to VGA converter... */
+			connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_eDP:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_edp_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_dp_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (amdgpu_connector->ddc_bus)
+					has_aux = true;
+				else
+					DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      dev->mode_config.scaling_mode_property,
+						      DRM_MODE_SCALE_FULLSCREEN);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = false;
+			connector->doublescan_allowed = false;
+			break;
+		case DRM_MODE_CONNECTOR_LVDS:
+			amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL);
+			if (!amdgpu_dig_connector)
+				goto failed;
+			amdgpu_connector->con_priv = amdgpu_dig_connector;
+			drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_lvds_funcs, connector_type);
+			drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_lvds_helper_funcs);
+			if (i2c_bus->valid) {
+				amdgpu_connector->ddc_bus = amdgpu_i2c_lookup(adev, i2c_bus);
+				if (!amdgpu_connector->ddc_bus)
+					DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n");
+			}
+			drm_object_attach_property(&amdgpu_connector->base.base,
+						      dev->mode_config.scaling_mode_property,
+						      DRM_MODE_SCALE_FULLSCREEN);
+			subpixel_order = SubPixelHorizontalRGB;
+			connector->interlace_allowed = false;
+			connector->doublescan_allowed = false;
+			break;
+		}
+	}
+
+	if (amdgpu_connector->hpd.hpd == AMDGPU_HPD_NONE) {
+		if (i2c_bus->valid)
+			connector->polled = DRM_CONNECTOR_POLL_CONNECT;
+	} else
+		connector->polled = DRM_CONNECTOR_POLL_HPD;
+
+	connector->display_info.subpixel_order = subpixel_order;
+	drm_connector_register(connector);
+
+	if (has_aux)
+		amdgpu_atombios_dp_aux_init(amdgpu_connector);
+
+	return;
+
+failed:
+	drm_connector_cleanup(connector);
+	kfree(connector);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h
new file mode 100644
index 000000000..61fcef15a
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_CONNECTORS_H__
+#define __AMDGPU_CONNECTORS_H__
+
+struct edid *amdgpu_connector_edid(struct drm_connector *connector);
+void amdgpu_connector_hotplug(struct drm_connector *connector);
+int amdgpu_connector_get_monitor_bpc(struct drm_connector *connector);
+u16 amdgpu_connector_encoder_get_dp_bridge_encoder_id(struct drm_connector *connector);
+bool amdgpu_connector_is_dp12_capable(struct drm_connector *connector);
+void
+amdgpu_connector_add(struct amdgpu_device *adev,
+		      uint32_t connector_id,
+		      uint32_t supported_device,
+		      int connector_type,
+		      struct amdgpu_i2c_bus_rec *i2c_bus,
+		      uint16_t connector_object_id,
+		      struct amdgpu_hpd *hpd,
+		      struct amdgpu_router *router);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
new file mode 100644
index 000000000..25a3e2485
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -0,0 +1,1007 @@
+/*
+ * Copyright 2008 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ */
+#include <linux/list_sort.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+#define AMDGPU_CS_MAX_PRIORITY		32u
+#define AMDGPU_CS_NUM_BUCKETS		(AMDGPU_CS_MAX_PRIORITY + 1)
+
+/* This is based on the bucket sort with O(n) time complexity.
+ * An item with priority "i" is added to bucket[i]. The lists are then
+ * concatenated in descending order.
+ */
+struct amdgpu_cs_buckets {
+	struct list_head bucket[AMDGPU_CS_NUM_BUCKETS];
+};
+
+static void amdgpu_cs_buckets_init(struct amdgpu_cs_buckets *b)
+{
+	unsigned i;
+
+	for (i = 0; i < AMDGPU_CS_NUM_BUCKETS; i++)
+		INIT_LIST_HEAD(&b->bucket[i]);
+}
+
+static void amdgpu_cs_buckets_add(struct amdgpu_cs_buckets *b,
+				  struct list_head *item, unsigned priority)
+{
+	/* Since buffers which appear sooner in the relocation list are
+	 * likely to be used more often than buffers which appear later
+	 * in the list, the sort mustn't change the ordering of buffers
+	 * with the same priority, i.e. it must be stable.
+	 */
+	list_add_tail(item, &b->bucket[min(priority, AMDGPU_CS_MAX_PRIORITY)]);
+}
+
+static void amdgpu_cs_buckets_get_list(struct amdgpu_cs_buckets *b,
+				       struct list_head *out_list)
+{
+	unsigned i;
+
+	/* Connect the sorted buckets in the output list. */
+	for (i = 0; i < AMDGPU_CS_NUM_BUCKETS; i++) {
+		list_splice(&b->bucket[i], out_list);
+	}
+}
+
+int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type,
+		       u32 ip_instance, u32 ring,
+		       struct amdgpu_ring **out_ring)
+{
+	/* Right now all IPs have only one instance - multiple rings. */
+	if (ip_instance != 0) {
+		DRM_ERROR("invalid ip instance: %d\n", ip_instance);
+		return -EINVAL;
+	}
+
+	switch (ip_type) {
+	default:
+		DRM_ERROR("unknown ip type: %d\n", ip_type);
+		return -EINVAL;
+	case AMDGPU_HW_IP_GFX:
+		if (ring < adev->gfx.num_gfx_rings) {
+			*out_ring = &adev->gfx.gfx_ring[ring];
+		} else {
+			DRM_ERROR("only %d gfx rings are supported now\n",
+				  adev->gfx.num_gfx_rings);
+			return -EINVAL;
+		}
+		break;
+	case AMDGPU_HW_IP_COMPUTE:
+		if (ring < adev->gfx.num_compute_rings) {
+			*out_ring = &adev->gfx.compute_ring[ring];
+		} else {
+			DRM_ERROR("only %d compute rings are supported now\n",
+				  adev->gfx.num_compute_rings);
+			return -EINVAL;
+		}
+		break;
+	case AMDGPU_HW_IP_DMA:
+		if (ring < adev->sdma.num_instances) {
+			*out_ring = &adev->sdma.instance[ring].ring;
+		} else {
+			DRM_ERROR("only %d SDMA rings are supported\n",
+				  adev->sdma.num_instances);
+			return -EINVAL;
+		}
+		break;
+	case AMDGPU_HW_IP_UVD:
+		*out_ring = &adev->uvd.ring;
+		break;
+	case AMDGPU_HW_IP_VCE:
+		if (ring < 2){
+			*out_ring = &adev->vce.ring[ring];
+		} else {
+			DRM_ERROR("only two VCE rings are supported\n");
+			return -EINVAL;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
+				      struct drm_amdgpu_cs_chunk_fence *fence_data)
+{
+	struct drm_gem_object *gobj;
+	uint32_t handle;
+
+	handle = fence_data->handle;
+	gobj = drm_gem_object_lookup(p->adev->ddev, p->filp,
+				     fence_data->handle);
+	if (gobj == NULL)
+		return -EINVAL;
+
+	p->uf.bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
+	p->uf.offset = fence_data->offset;
+
+	if (amdgpu_ttm_tt_has_userptr(p->uf.bo->tbo.ttm)) {
+		drm_gem_object_unreference_unlocked(gobj);
+		return -EINVAL;
+	}
+
+	p->uf_entry.robj = amdgpu_bo_ref(p->uf.bo);
+	p->uf_entry.prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
+	p->uf_entry.allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
+	p->uf_entry.priority = 0;
+	p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
+	p->uf_entry.tv.shared = true;
+
+	drm_gem_object_unreference_unlocked(gobj);
+	return 0;
+}
+
+int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
+{
+	union drm_amdgpu_cs *cs = data;
+	uint64_t *chunk_array_user;
+	uint64_t *chunk_array;
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	unsigned size;
+	int i;
+	int ret;
+
+	if (cs->in.num_chunks == 0)
+		return 0;
+
+	chunk_array = kmalloc_array(cs->in.num_chunks, sizeof(uint64_t), GFP_KERNEL);
+	if (!chunk_array)
+		return -ENOMEM;
+
+	p->ctx = amdgpu_ctx_get(fpriv, cs->in.ctx_id);
+	if (!p->ctx) {
+		ret = -EINVAL;
+		goto free_chunk;
+	}
+
+	p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
+
+	/* get chunks */
+	INIT_LIST_HEAD(&p->validated);
+	chunk_array_user = (uint64_t __user *)(unsigned long)(cs->in.chunks);
+	if (copy_from_user(chunk_array, chunk_array_user,
+			   sizeof(uint64_t)*cs->in.num_chunks)) {
+		ret = -EFAULT;
+		goto put_bo_list;
+	}
+
+	p->nchunks = cs->in.num_chunks;
+	p->chunks = kmalloc_array(p->nchunks, sizeof(struct amdgpu_cs_chunk),
+			    GFP_KERNEL);
+	if (!p->chunks) {
+		ret = -ENOMEM;
+		goto put_bo_list;
+	}
+
+	for (i = 0; i < p->nchunks; i++) {
+		struct drm_amdgpu_cs_chunk __user **chunk_ptr = NULL;
+		struct drm_amdgpu_cs_chunk user_chunk;
+		uint32_t __user *cdata;
+
+		chunk_ptr = (void __user *)(unsigned long)chunk_array[i];
+		if (copy_from_user(&user_chunk, chunk_ptr,
+				       sizeof(struct drm_amdgpu_cs_chunk))) {
+			ret = -EFAULT;
+			i--;
+			goto free_partial_kdata;
+		}
+		p->chunks[i].chunk_id = user_chunk.chunk_id;
+		p->chunks[i].length_dw = user_chunk.length_dw;
+
+		size = p->chunks[i].length_dw;
+		cdata = (void __user *)(unsigned long)user_chunk.chunk_data;
+		p->chunks[i].user_ptr = cdata;
+
+		p->chunks[i].kdata = drm_malloc_ab(size, sizeof(uint32_t));
+		if (p->chunks[i].kdata == NULL) {
+			ret = -ENOMEM;
+			i--;
+			goto free_partial_kdata;
+		}
+		size *= sizeof(uint32_t);
+		if (copy_from_user(p->chunks[i].kdata, cdata, size)) {
+			ret = -EFAULT;
+			goto free_partial_kdata;
+		}
+
+		switch (p->chunks[i].chunk_id) {
+		case AMDGPU_CHUNK_ID_IB:
+			p->num_ibs++;
+			break;
+
+		case AMDGPU_CHUNK_ID_FENCE:
+			size = sizeof(struct drm_amdgpu_cs_chunk_fence);
+			if (p->chunks[i].length_dw * sizeof(uint32_t) < size) {
+				ret = -EINVAL;
+				goto free_partial_kdata;
+			}
+
+			ret = amdgpu_cs_user_fence_chunk(p, (void *)p->chunks[i].kdata);
+			if (ret)
+				goto free_partial_kdata;
+
+			break;
+
+		case AMDGPU_CHUNK_ID_DEPENDENCIES:
+			break;
+
+		default:
+			ret = -EINVAL;
+			goto free_partial_kdata;
+		}
+	}
+
+
+	p->ibs = kcalloc(p->num_ibs, sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!p->ibs) {
+		ret = -ENOMEM;
+		goto free_all_kdata;
+	}
+
+	kfree(chunk_array);
+	return 0;
+
+free_all_kdata:
+	i = p->nchunks - 1;
+free_partial_kdata:
+	for (; i >= 0; i--)
+		drm_free_large(p->chunks[i].kdata);
+	kfree(p->chunks);
+put_bo_list:
+	if (p->bo_list)
+		amdgpu_bo_list_put(p->bo_list);
+	amdgpu_ctx_put(p->ctx);
+free_chunk:
+	kfree(chunk_array);
+
+	return ret;
+}
+
+/* Returns how many bytes TTM can move per IB.
+ */
+static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
+{
+	u64 real_vram_size = adev->mc.real_vram_size;
+	u64 vram_usage = atomic64_read(&adev->vram_usage);
+
+	/* This function is based on the current VRAM usage.
+	 *
+	 * - If all of VRAM is free, allow relocating the number of bytes that
+	 *   is equal to 1/4 of the size of VRAM for this IB.
+
+	 * - If more than one half of VRAM is occupied, only allow relocating
+	 *   1 MB of data for this IB.
+	 *
+	 * - From 0 to one half of used VRAM, the threshold decreases
+	 *   linearly.
+	 *         __________________
+	 * 1/4 of -|\               |
+	 * VRAM    | \              |
+	 *         |  \             |
+	 *         |   \            |
+	 *         |    \           |
+	 *         |     \          |
+	 *         |      \         |
+	 *         |       \________|1 MB
+	 *         |----------------|
+	 *    VRAM 0 %             100 %
+	 *         used            used
+	 *
+	 * Note: It's a threshold, not a limit. The threshold must be crossed
+	 * for buffer relocations to stop, so any buffer of an arbitrary size
+	 * can be moved as long as the threshold isn't crossed before
+	 * the relocation takes place. We don't want to disable buffer
+	 * relocations completely.
+	 *
+	 * The idea is that buffers should be placed in VRAM at creation time
+	 * and TTM should only do a minimum number of relocations during
+	 * command submission. In practice, you need to submit at least
+	 * a dozen IBs to move all buffers to VRAM if they are in GTT.
+	 *
+	 * Also, things can get pretty crazy under memory pressure and actual
+	 * VRAM usage can change a lot, so playing safe even at 50% does
+	 * consistently increase performance.
+	 */
+
+	u64 half_vram = real_vram_size >> 1;
+	u64 half_free_vram = vram_usage >= half_vram ? 0 : half_vram - vram_usage;
+	u64 bytes_moved_threshold = half_free_vram >> 1;
+	return max(bytes_moved_threshold, 1024*1024ull);
+}
+
+int amdgpu_cs_list_validate(struct amdgpu_device *adev,
+			    struct amdgpu_vm *vm,
+			    struct list_head *validated)
+{
+	struct amdgpu_bo_list_entry *lobj;
+	struct amdgpu_bo *bo;
+	u64 bytes_moved = 0, initial_bytes_moved;
+	u64 bytes_moved_threshold = amdgpu_cs_get_threshold_for_moves(adev);
+	int r;
+
+	list_for_each_entry(lobj, validated, tv.head) {
+		bo = lobj->robj;
+		if (!bo->pin_count) {
+			u32 domain = lobj->prefered_domains;
+			u32 current_domain =
+				amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
+
+			/* Check if this buffer will be moved and don't move it
+			 * if we have moved too many buffers for this IB already.
+			 *
+			 * Note that this allows moving at least one buffer of
+			 * any size, because it doesn't take the current "bo"
+			 * into account. We don't want to disallow buffer moves
+			 * completely.
+			 */
+			if ((lobj->allowed_domains & current_domain) != 0 &&
+			    (domain & current_domain) == 0 && /* will be moved */
+			    bytes_moved > bytes_moved_threshold) {
+				/* don't move it */
+				domain = current_domain;
+			}
+
+		retry:
+			amdgpu_ttm_placement_from_domain(bo, domain);
+			initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
+			r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+			bytes_moved += atomic64_read(&adev->num_bytes_moved) -
+				       initial_bytes_moved;
+
+			if (unlikely(r)) {
+				if (r != -ERESTARTSYS && domain != lobj->allowed_domains) {
+					domain = lobj->allowed_domains;
+					goto retry;
+				}
+				return r;
+			}
+		}
+		lobj->bo_va = amdgpu_vm_bo_find(vm, bo);
+	}
+	return 0;
+}
+
+static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_cs_buckets buckets;
+	struct list_head duplicates;
+	bool need_mmap_lock = false;
+	int i, r;
+
+	if (p->bo_list) {
+		need_mmap_lock = p->bo_list->has_userptr;
+		amdgpu_cs_buckets_init(&buckets);
+		for (i = 0; i < p->bo_list->num_entries; i++)
+			amdgpu_cs_buckets_add(&buckets, &p->bo_list->array[i].tv.head,
+								  p->bo_list->array[i].priority);
+
+		amdgpu_cs_buckets_get_list(&buckets, &p->validated);
+	}
+
+	p->vm_bos = amdgpu_vm_get_bos(p->adev, &fpriv->vm,
+				      &p->validated);
+
+	if (p->uf.bo)
+		list_add(&p->uf_entry.tv.head, &p->validated);
+
+	if (need_mmap_lock)
+		down_read(&current->mm->mmap_sem);
+
+	INIT_LIST_HEAD(&duplicates);
+	r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true, &duplicates);
+	if (unlikely(r != 0))
+		goto error_reserve;
+
+	r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &p->validated);
+	if (r)
+		goto error_validate;
+
+	r = amdgpu_cs_list_validate(p->adev, &fpriv->vm, &duplicates);
+
+error_validate:
+	if (r)
+		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+
+error_reserve:
+	if (need_mmap_lock)
+		up_read(&current->mm->mmap_sem);
+
+	return r;
+}
+
+static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_bo_list_entry *e;
+	int r;
+
+	list_for_each_entry(e, &p->validated, tv.head) {
+		struct reservation_object *resv = e->robj->tbo.resv;
+		r = amdgpu_sync_resv(p->adev, &p->ibs[0].sync, resv, p->filp);
+
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+static int cmp_size_smaller_first(void *priv, struct list_head *a,
+				  struct list_head *b)
+{
+	struct amdgpu_bo_list_entry *la = list_entry(a, struct amdgpu_bo_list_entry, tv.head);
+	struct amdgpu_bo_list_entry *lb = list_entry(b, struct amdgpu_bo_list_entry, tv.head);
+
+	/* Sort A before B if A is smaller. */
+	return (int)la->robj->tbo.num_pages - (int)lb->robj->tbo.num_pages;
+}
+
+/**
+ * cs_parser_fini() - clean parser states
+ * @parser:	parser structure holding parsing context.
+ * @error:	error number
+ *
+ * If error is set than unvalidate buffer, otherwise just free memory
+ * used by parsing context.
+ **/
+static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bool backoff)
+{
+	unsigned i;
+
+	if (!error) {
+		/* Sort the buffer list from the smallest to largest buffer,
+		 * which affects the order of buffers in the LRU list.
+		 * This assures that the smallest buffers are added first
+		 * to the LRU list, so they are likely to be later evicted
+		 * first, instead of large buffers whose eviction is more
+		 * expensive.
+		 *
+		 * This slightly lowers the number of bytes moved by TTM
+		 * per frame under memory pressure.
+		 */
+		list_sort(NULL, &parser->validated, cmp_size_smaller_first);
+
+		ttm_eu_fence_buffer_objects(&parser->ticket,
+					    &parser->validated,
+					    parser->fence);
+	} else if (backoff) {
+		ttm_eu_backoff_reservation(&parser->ticket,
+					   &parser->validated);
+	}
+	fence_put(parser->fence);
+
+	if (parser->ctx)
+		amdgpu_ctx_put(parser->ctx);
+	if (parser->bo_list)
+		amdgpu_bo_list_put(parser->bo_list);
+
+	drm_free_large(parser->vm_bos);
+	for (i = 0; i < parser->nchunks; i++)
+		drm_free_large(parser->chunks[i].kdata);
+	kfree(parser->chunks);
+	if (parser->ibs)
+		for (i = 0; i < parser->num_ibs; i++)
+			amdgpu_ib_free(parser->adev, &parser->ibs[i]);
+	kfree(parser->ibs);
+	amdgpu_bo_unref(&parser->uf.bo);
+	amdgpu_bo_unref(&parser->uf_entry.robj);
+}
+
+static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p,
+				   struct amdgpu_vm *vm)
+{
+	struct amdgpu_device *adev = p->adev;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	int i, r;
+
+	r = amdgpu_vm_update_page_directory(adev, vm);
+	if (r)
+		return r;
+
+	r = amdgpu_sync_fence(adev, &p->ibs[0].sync, vm->page_directory_fence);
+	if (r)
+		return r;
+
+	r = amdgpu_vm_clear_freed(adev, vm);
+	if (r)
+		return r;
+
+	if (p->bo_list) {
+		for (i = 0; i < p->bo_list->num_entries; i++) {
+			struct fence *f;
+
+			/* ignore duplicates */
+			bo = p->bo_list->array[i].robj;
+			if (!bo)
+				continue;
+
+			bo_va = p->bo_list->array[i].bo_va;
+			if (bo_va == NULL)
+				continue;
+
+			r = amdgpu_vm_bo_update(adev, bo_va, &bo->tbo.mem);
+			if (r)
+				return r;
+
+			f = bo_va->last_pt_update;
+			r = amdgpu_sync_fence(adev, &p->ibs[0].sync, f);
+			if (r)
+				return r;
+		}
+
+	}
+
+	r = amdgpu_vm_clear_invalids(adev, vm, &p->ibs[0].sync);
+
+	if (amdgpu_vm_debug && p->bo_list) {
+		/* Invalidate all BOs to test for userspace bugs */
+		for (i = 0; i < p->bo_list->num_entries; i++) {
+			/* ignore duplicates */
+			bo = p->bo_list->array[i].robj;
+			if (!bo)
+				continue;
+
+			amdgpu_vm_bo_invalidate(adev, bo);
+		}
+	}
+
+	return r;
+}
+
+static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
+				 struct amdgpu_cs_parser *parser)
+{
+	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_ring *ring;
+	int i, r;
+
+	if (parser->num_ibs == 0)
+		return 0;
+
+	/* Only for UVD/VCE VM emulation */
+	for (i = 0; i < parser->num_ibs; i++) {
+		ring = parser->ibs[i].ring;
+		if (ring->funcs->parse_cs) {
+			r = amdgpu_ring_parse_cs(ring, parser, i);
+			if (r)
+				return r;
+		}
+	}
+
+	r = amdgpu_bo_vm_update_pte(parser, vm);
+	if (!r)
+		amdgpu_cs_sync_rings(parser);
+
+	return r;
+}
+
+static int amdgpu_cs_handle_lockup(struct amdgpu_device *adev, int r)
+{
+	if (r == -EDEADLK) {
+		r = amdgpu_gpu_reset(adev);
+		if (!r)
+			r = -EAGAIN;
+	}
+	return r;
+}
+
+static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
+			     struct amdgpu_cs_parser *parser)
+{
+	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	int i, j;
+	int r;
+
+	for (i = 0, j = 0; i < parser->nchunks && j < parser->num_ibs; i++) {
+		struct amdgpu_cs_chunk *chunk;
+		struct amdgpu_ib *ib;
+		struct drm_amdgpu_cs_chunk_ib *chunk_ib;
+		struct amdgpu_ring *ring;
+
+		chunk = &parser->chunks[i];
+		ib = &parser->ibs[j];
+		chunk_ib = (struct drm_amdgpu_cs_chunk_ib *)chunk->kdata;
+
+		if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
+			continue;
+
+		r = amdgpu_cs_get_ring(adev, chunk_ib->ip_type,
+				       chunk_ib->ip_instance, chunk_ib->ring,
+				       &ring);
+		if (r)
+			return r;
+
+		if (ring->funcs->parse_cs) {
+			struct amdgpu_bo_va_mapping *m;
+			struct amdgpu_bo *aobj = NULL;
+			uint64_t offset;
+			uint8_t *kptr;
+
+			m = amdgpu_cs_find_mapping(parser, chunk_ib->va_start,
+						   &aobj);
+			if (!aobj) {
+				DRM_ERROR("IB va_start is invalid\n");
+				return -EINVAL;
+			}
+
+			if ((chunk_ib->va_start + chunk_ib->ib_bytes) >
+			    (m->it.last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+				DRM_ERROR("IB va_start+ib_bytes is invalid\n");
+				return -EINVAL;
+			}
+
+			/* the IB should be reserved at this point */
+			r = amdgpu_bo_kmap(aobj, (void **)&kptr);
+			if (r) {
+				return r;
+			}
+
+			offset = ((uint64_t)m->it.start) * AMDGPU_GPU_PAGE_SIZE;
+			kptr += chunk_ib->va_start - offset;
+
+			r =  amdgpu_ib_get(ring, NULL, chunk_ib->ib_bytes, ib);
+			if (r) {
+				DRM_ERROR("Failed to get ib !\n");
+				return r;
+			}
+
+			memcpy(ib->ptr, kptr, chunk_ib->ib_bytes);
+			amdgpu_bo_kunmap(aobj);
+		} else {
+			r =  amdgpu_ib_get(ring, vm, 0, ib);
+			if (r) {
+				DRM_ERROR("Failed to get ib !\n");
+				return r;
+			}
+
+			ib->gpu_addr = chunk_ib->va_start;
+		}
+
+		ib->length_dw = chunk_ib->ib_bytes / 4;
+		ib->flags = chunk_ib->flags;
+		ib->ctx = parser->ctx;
+		j++;
+	}
+
+	if (!parser->num_ibs)
+		return 0;
+
+	/* add GDS resources to first IB */
+	if (parser->bo_list) {
+		struct amdgpu_bo *gds = parser->bo_list->gds_obj;
+		struct amdgpu_bo *gws = parser->bo_list->gws_obj;
+		struct amdgpu_bo *oa = parser->bo_list->oa_obj;
+		struct amdgpu_ib *ib = &parser->ibs[0];
+
+		if (gds) {
+			ib->gds_base = amdgpu_bo_gpu_offset(gds);
+			ib->gds_size = amdgpu_bo_size(gds);
+		}
+		if (gws) {
+			ib->gws_base = amdgpu_bo_gpu_offset(gws);
+			ib->gws_size = amdgpu_bo_size(gws);
+		}
+		if (oa) {
+			ib->oa_base = amdgpu_bo_gpu_offset(oa);
+			ib->oa_size = amdgpu_bo_size(oa);
+		}
+	}
+	/* wrap the last IB with user fence */
+	if (parser->uf.bo) {
+		struct amdgpu_ib *ib = &parser->ibs[parser->num_ibs - 1];
+
+		/* UVD & VCE fw doesn't support user fences */
+		if (ib->ring->type == AMDGPU_RING_TYPE_UVD ||
+		    ib->ring->type == AMDGPU_RING_TYPE_VCE)
+			return -EINVAL;
+
+		ib->user = &parser->uf;
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
+				  struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_ib *ib;
+	int i, j, r;
+
+	if (!p->num_ibs)
+		return 0;
+
+	/* Add dependencies to first IB */
+	ib = &p->ibs[0];
+	for (i = 0; i < p->nchunks; ++i) {
+		struct drm_amdgpu_cs_chunk_dep *deps;
+		struct amdgpu_cs_chunk *chunk;
+		unsigned num_deps;
+
+		chunk = &p->chunks[i];
+
+		if (chunk->chunk_id != AMDGPU_CHUNK_ID_DEPENDENCIES)
+			continue;
+
+		deps = (struct drm_amdgpu_cs_chunk_dep *)chunk->kdata;
+		num_deps = chunk->length_dw * 4 /
+			sizeof(struct drm_amdgpu_cs_chunk_dep);
+
+		for (j = 0; j < num_deps; ++j) {
+			struct amdgpu_ring *ring;
+			struct amdgpu_ctx *ctx;
+			struct fence *fence;
+
+			r = amdgpu_cs_get_ring(adev, deps[j].ip_type,
+					       deps[j].ip_instance,
+					       deps[j].ring, &ring);
+			if (r)
+				return r;
+
+			ctx = amdgpu_ctx_get(fpriv, deps[j].ctx_id);
+			if (ctx == NULL)
+				return -EINVAL;
+
+			fence = amdgpu_ctx_get_fence(ctx, ring,
+						     deps[j].handle);
+			if (IS_ERR(fence)) {
+				r = PTR_ERR(fence);
+				amdgpu_ctx_put(ctx);
+				return r;
+
+			} else if (fence) {
+				r = amdgpu_sync_fence(adev, &ib->sync, fence);
+				fence_put(fence);
+				amdgpu_ctx_put(ctx);
+				if (r)
+					return r;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_cs_free_job(struct amdgpu_job *job)
+{
+	int i;
+	if (job->ibs)
+		for (i = 0; i < job->num_ibs; i++)
+			amdgpu_ib_free(job->adev, &job->ibs[i]);
+	kfree(job->ibs);
+	if (job->uf.bo)
+		amdgpu_bo_unref(&job->uf.bo);
+	return 0;
+}
+
+int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	union drm_amdgpu_cs *cs = data;
+	struct amdgpu_cs_parser parser = {};
+	bool reserved_buffers = false;
+	int i, r;
+
+	if (!adev->accel_working)
+		return -EBUSY;
+
+	parser.adev = adev;
+	parser.filp = filp;
+
+	r = amdgpu_cs_parser_init(&parser, data);
+	if (r) {
+		DRM_ERROR("Failed to initialize parser !\n");
+		amdgpu_cs_parser_fini(&parser, r, false);
+		r = amdgpu_cs_handle_lockup(adev, r);
+		return r;
+	}
+	r = amdgpu_cs_parser_relocs(&parser);
+	if (r == -ENOMEM)
+		DRM_ERROR("Not enough memory for command submission!\n");
+	else if (r && r != -ERESTARTSYS)
+		DRM_ERROR("Failed to process the buffer list %d!\n", r);
+	else if (!r) {
+		reserved_buffers = true;
+		r = amdgpu_cs_ib_fill(adev, &parser);
+	}
+
+	if (!r) {
+		r = amdgpu_cs_dependencies(adev, &parser);
+		if (r)
+			DRM_ERROR("Failed in the dependencies handling %d!\n", r);
+	}
+
+	if (r)
+		goto out;
+
+	for (i = 0; i < parser.num_ibs; i++)
+		trace_amdgpu_cs(&parser, i);
+
+	r = amdgpu_cs_ib_vm_chunk(adev, &parser);
+	if (r)
+		goto out;
+
+	if (amdgpu_enable_scheduler && parser.num_ibs) {
+		struct amdgpu_ring * ring = parser.ibs->ring;
+		struct amd_sched_fence *fence;
+		struct amdgpu_job *job;
+
+		job = kzalloc(sizeof(struct amdgpu_job), GFP_KERNEL);
+		if (!job) {
+			r = -ENOMEM;
+			goto out;
+		}
+
+		job->base.sched = &ring->sched;
+		job->base.s_entity = &parser.ctx->rings[ring->idx].entity;
+		job->adev = parser.adev;
+		job->owner = parser.filp;
+		job->free_job = amdgpu_cs_free_job;
+
+		job->ibs = parser.ibs;
+		job->num_ibs = parser.num_ibs;
+		parser.ibs = NULL;
+		parser.num_ibs = 0;
+
+		if (job->ibs[job->num_ibs - 1].user) {
+			job->uf = parser.uf;
+			job->ibs[job->num_ibs - 1].user = &job->uf;
+			parser.uf.bo = NULL;
+		}
+
+		fence = amd_sched_fence_create(job->base.s_entity,
+					       parser.filp);
+		if (!fence) {
+			r = -ENOMEM;
+			amdgpu_cs_free_job(job);
+			kfree(job);
+			goto out;
+		}
+		job->base.s_fence = fence;
+		parser.fence = fence_get(&fence->base);
+
+		cs->out.handle = amdgpu_ctx_add_fence(parser.ctx, ring,
+						      &fence->base);
+		job->ibs[job->num_ibs - 1].sequence = cs->out.handle;
+
+		trace_amdgpu_cs_ioctl(job);
+		amd_sched_entity_push_job(&job->base);
+
+	} else {
+		struct amdgpu_fence *fence;
+
+		r = amdgpu_ib_schedule(adev, parser.num_ibs, parser.ibs,
+				       parser.filp);
+		fence = parser.ibs[parser.num_ibs - 1].fence;
+		parser.fence = fence_get(&fence->base);
+		cs->out.handle = parser.ibs[parser.num_ibs - 1].sequence;
+	}
+
+out:
+	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
+	r = amdgpu_cs_handle_lockup(adev, r);
+	return r;
+}
+
+/**
+ * amdgpu_cs_wait_ioctl - wait for a command submission to finish
+ *
+ * @dev: drm device
+ * @data: data from userspace
+ * @filp: file private
+ *
+ * Wait for the command submission identified by handle to finish.
+ */
+int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
+			 struct drm_file *filp)
+{
+	union drm_amdgpu_wait_cs *wait = data;
+	struct amdgpu_device *adev = dev->dev_private;
+	unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
+	struct amdgpu_ring *ring = NULL;
+	struct amdgpu_ctx *ctx;
+	struct fence *fence;
+	long r;
+
+	r = amdgpu_cs_get_ring(adev, wait->in.ip_type, wait->in.ip_instance,
+			       wait->in.ring, &ring);
+	if (r)
+		return r;
+
+	ctx = amdgpu_ctx_get(filp->driver_priv, wait->in.ctx_id);
+	if (ctx == NULL)
+		return -EINVAL;
+
+	fence = amdgpu_ctx_get_fence(ctx, ring, wait->in.handle);
+	if (IS_ERR(fence))
+		r = PTR_ERR(fence);
+	else if (fence) {
+		r = fence_wait_timeout(fence, true, timeout);
+		fence_put(fence);
+	} else
+		r = 1;
+
+	amdgpu_ctx_put(ctx);
+	if (r < 0)
+		return r;
+
+	memset(wait, 0, sizeof(*wait));
+	wait->out.status = (r == 0);
+
+	return 0;
+}
+
+/**
+ * amdgpu_cs_find_bo_va - find bo_va for VM address
+ *
+ * @parser: command submission parser context
+ * @addr: VM address
+ * @bo: resulting BO of the mapping found
+ *
+ * Search the buffer objects in the command submission context for a certain
+ * virtual memory address. Returns allocation structure when found, NULL
+ * otherwise.
+ */
+struct amdgpu_bo_va_mapping *
+amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
+		       uint64_t addr, struct amdgpu_bo **bo)
+{
+	struct amdgpu_bo_list_entry *reloc;
+	struct amdgpu_bo_va_mapping *mapping;
+
+	addr /= AMDGPU_GPU_PAGE_SIZE;
+
+	list_for_each_entry(reloc, &parser->validated, tv.head) {
+		if (!reloc->bo_va)
+			continue;
+
+		list_for_each_entry(mapping, &reloc->bo_va->valids, list) {
+			if (mapping->it.start > addr ||
+			    addr > mapping->it.last)
+				continue;
+
+			*bo = reloc->bo_va->bo;
+			return mapping;
+		}
+
+		list_for_each_entry(mapping, &reloc->bo_va->invalids, list) {
+			if (mapping->it.start > addr ||
+			    addr > mapping->it.last)
+				continue;
+
+			*bo = reloc->bo_va->bo;
+			return mapping;
+		}
+	}
+
+	return NULL;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
new file mode 100644
index 000000000..fec65f01c
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
@@ -0,0 +1,309 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: monk liu <monk.liu@amd.com>
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+
+int amdgpu_ctx_init(struct amdgpu_device *adev, bool kernel,
+		    struct amdgpu_ctx *ctx)
+{
+	unsigned i, j;
+	int r;
+
+	memset(ctx, 0, sizeof(*ctx));
+	ctx->adev = adev;
+	kref_init(&ctx->refcount);
+	spin_lock_init(&ctx->ring_lock);
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		ctx->rings[i].sequence = 1;
+
+	if (amdgpu_enable_scheduler) {
+		/* create context entity for each ring */
+		for (i = 0; i < adev->num_rings; i++) {
+			struct amd_sched_rq *rq;
+			if (kernel)
+				rq = &adev->rings[i]->sched.kernel_rq;
+			else
+				rq = &adev->rings[i]->sched.sched_rq;
+			r = amd_sched_entity_init(&adev->rings[i]->sched,
+						  &ctx->rings[i].entity,
+						  rq, amdgpu_sched_jobs);
+			if (r)
+				break;
+		}
+
+		if (i < adev->num_rings) {
+			for (j = 0; j < i; j++)
+				amd_sched_entity_fini(&adev->rings[j]->sched,
+						      &ctx->rings[j].entity);
+			kfree(ctx);
+			return r;
+		}
+	}
+	return 0;
+}
+
+void amdgpu_ctx_fini(struct amdgpu_ctx *ctx)
+{
+	struct amdgpu_device *adev = ctx->adev;
+	unsigned i, j;
+
+	if (!adev)
+		return;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		for (j = 0; j < AMDGPU_CTX_MAX_CS_PENDING; ++j)
+			fence_put(ctx->rings[i].fences[j]);
+
+	if (amdgpu_enable_scheduler) {
+		for (i = 0; i < adev->num_rings; i++)
+			amd_sched_entity_fini(&adev->rings[i]->sched,
+					      &ctx->rings[i].entity);
+	}
+}
+
+static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
+			    struct amdgpu_fpriv *fpriv,
+			    uint32_t *id)
+{
+	struct amdgpu_ctx_mgr *mgr = &fpriv->ctx_mgr;
+	struct amdgpu_ctx *ctx;
+	int r;
+
+	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+
+	mutex_lock(&mgr->lock);
+	r = idr_alloc(&mgr->ctx_handles, ctx, 1, 0, GFP_KERNEL);
+	if (r < 0) {
+		mutex_unlock(&mgr->lock);
+		kfree(ctx);
+		return r;
+	}
+	*id = (uint32_t)r;
+	r = amdgpu_ctx_init(adev, false, ctx);
+	mutex_unlock(&mgr->lock);
+
+	return r;
+}
+
+static void amdgpu_ctx_do_release(struct kref *ref)
+{
+	struct amdgpu_ctx *ctx;
+
+	ctx = container_of(ref, struct amdgpu_ctx, refcount);
+
+	amdgpu_ctx_fini(ctx);
+
+	kfree(ctx);
+}
+
+static int amdgpu_ctx_free(struct amdgpu_fpriv *fpriv, uint32_t id)
+{
+	struct amdgpu_ctx_mgr *mgr = &fpriv->ctx_mgr;
+	struct amdgpu_ctx *ctx;
+
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (ctx) {
+		idr_remove(&mgr->ctx_handles, id);
+		kref_put(&ctx->refcount, amdgpu_ctx_do_release);
+		mutex_unlock(&mgr->lock);
+		return 0;
+	}
+	mutex_unlock(&mgr->lock);
+	return -EINVAL;
+}
+
+static int amdgpu_ctx_query(struct amdgpu_device *adev,
+			    struct amdgpu_fpriv *fpriv, uint32_t id,
+			    union drm_amdgpu_ctx_out *out)
+{
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+	unsigned reset_counter;
+
+	if (!fpriv)
+		return -EINVAL;
+
+	mgr = &fpriv->ctx_mgr;
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (!ctx) {
+		mutex_unlock(&mgr->lock);
+		return -EINVAL;
+	}
+
+	/* TODO: these two are always zero */
+	out->state.flags = 0x0;
+	out->state.hangs = 0x0;
+
+	/* determine if a GPU reset has occured since the last call */
+	reset_counter = atomic_read(&adev->gpu_reset_counter);
+	/* TODO: this should ideally return NO, GUILTY, or INNOCENT. */
+	if (ctx->reset_counter == reset_counter)
+		out->state.reset_status = AMDGPU_CTX_NO_RESET;
+	else
+		out->state.reset_status = AMDGPU_CTX_UNKNOWN_RESET;
+	ctx->reset_counter = reset_counter;
+
+	mutex_unlock(&mgr->lock);
+	return 0;
+}
+
+int amdgpu_ctx_ioctl(struct drm_device *dev, void *data,
+		     struct drm_file *filp)
+{
+	int r;
+	uint32_t id;
+
+	union drm_amdgpu_ctx *args = data;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+
+	r = 0;
+	id = args->in.ctx_id;
+
+	switch (args->in.op) {
+		case AMDGPU_CTX_OP_ALLOC_CTX:
+			r = amdgpu_ctx_alloc(adev, fpriv, &id);
+			args->out.alloc.ctx_id = id;
+			break;
+		case AMDGPU_CTX_OP_FREE_CTX:
+			r = amdgpu_ctx_free(fpriv, id);
+			break;
+		case AMDGPU_CTX_OP_QUERY_STATE:
+			r = amdgpu_ctx_query(adev, fpriv, id, &args->out);
+			break;
+		default:
+			return -EINVAL;
+	}
+
+	return r;
+}
+
+struct amdgpu_ctx *amdgpu_ctx_get(struct amdgpu_fpriv *fpriv, uint32_t id)
+{
+	struct amdgpu_ctx *ctx;
+	struct amdgpu_ctx_mgr *mgr;
+
+	if (!fpriv)
+		return NULL;
+
+	mgr = &fpriv->ctx_mgr;
+
+	mutex_lock(&mgr->lock);
+	ctx = idr_find(&mgr->ctx_handles, id);
+	if (ctx)
+		kref_get(&ctx->refcount);
+	mutex_unlock(&mgr->lock);
+	return ctx;
+}
+
+int amdgpu_ctx_put(struct amdgpu_ctx *ctx)
+{
+	if (ctx == NULL)
+		return -EINVAL;
+
+	kref_put(&ctx->refcount, amdgpu_ctx_do_release);
+	return 0;
+}
+
+uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, struct amdgpu_ring *ring,
+			      struct fence *fence)
+{
+	struct amdgpu_ctx_ring *cring = & ctx->rings[ring->idx];
+	uint64_t seq = cring->sequence;
+	unsigned idx = 0;
+	struct fence *other = NULL;
+
+	idx = seq % AMDGPU_CTX_MAX_CS_PENDING;
+	other = cring->fences[idx];
+	if (other) {
+		signed long r;
+		r = fence_wait_timeout(other, false, MAX_SCHEDULE_TIMEOUT);
+		if (r < 0)
+			DRM_ERROR("Error (%ld) waiting for fence!\n", r);
+	}
+
+	fence_get(fence);
+
+	spin_lock(&ctx->ring_lock);
+	cring->fences[idx] = fence;
+	cring->sequence++;
+	spin_unlock(&ctx->ring_lock);
+
+	fence_put(other);
+
+	return seq;
+}
+
+struct fence *amdgpu_ctx_get_fence(struct amdgpu_ctx *ctx,
+				   struct amdgpu_ring *ring, uint64_t seq)
+{
+	struct amdgpu_ctx_ring *cring = & ctx->rings[ring->idx];
+	struct fence *fence;
+
+	spin_lock(&ctx->ring_lock);
+
+	if (seq >= cring->sequence) {
+		spin_unlock(&ctx->ring_lock);
+		return ERR_PTR(-EINVAL);
+	}
+
+
+	if (seq + AMDGPU_CTX_MAX_CS_PENDING < cring->sequence) {
+		spin_unlock(&ctx->ring_lock);
+		return NULL;
+	}
+
+	fence = fence_get(cring->fences[seq % AMDGPU_CTX_MAX_CS_PENDING]);
+	spin_unlock(&ctx->ring_lock);
+
+	return fence;
+}
+
+void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr)
+{
+	mutex_init(&mgr->lock);
+	idr_init(&mgr->ctx_handles);
+}
+
+void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr)
+{
+	struct amdgpu_ctx *ctx;
+	struct idr *idp;
+	uint32_t id;
+
+	idp = &mgr->ctx_handles;
+
+	idr_for_each_entry(idp, ctx, id) {
+		if (kref_put(&ctx->refcount, amdgpu_ctx_do_release) != 1)
+			DRM_ERROR("ctx %p is still alive\n", ctx);
+	}
+
+	idr_destroy(&mgr->ctx_handles);
+	mutex_destroy(&mgr->lock);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
new file mode 100644
index 000000000..c961fe093
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -0,0 +1,2049 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <linux/console.h>
+#include <linux/slab.h>
+#include <linux/debugfs.h>
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include <linux/vgaarb.h>
+#include <linux/vga_switcheroo.h>
+#include <linux/efi.h>
+#include "amdgpu.h"
+#include "amdgpu_i2c.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#include "cik.h"
+#endif
+#include "vi.h"
+#include "bif/bif_4_1_d.h"
+
+static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
+static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
+
+static const char *amdgpu_asic_name[] = {
+	"BONAIRE",
+	"KAVERI",
+	"KABINI",
+	"HAWAII",
+	"MULLINS",
+	"TOPAZ",
+	"TONGA",
+	"FIJI",
+	"CARRIZO",
+	"STONEY",
+	"LAST",
+};
+
+bool amdgpu_device_is_px(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (adev->flags & AMD_IS_PX)
+		return true;
+	return false;
+}
+
+/*
+ * MMIO register access helper functions.
+ */
+uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
+			bool always_indirect)
+{
+	if ((reg * 4) < adev->rmmio_size && !always_indirect)
+		return readl(((void __iomem *)adev->rmmio) + (reg * 4));
+	else {
+		unsigned long flags;
+		uint32_t ret;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4));
+		ret = readl(((void __iomem *)adev->rmmio) + (mmMM_DATA * 4));
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+
+		return ret;
+	}
+}
+
+void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
+		    bool always_indirect)
+{
+	if ((reg * 4) < adev->rmmio_size && !always_indirect)
+		writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
+	else {
+		unsigned long flags;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4));
+		writel(v, ((void __iomem *)adev->rmmio) + (mmMM_DATA * 4));
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+	}
+}
+
+u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	if ((reg * 4) < adev->rio_mem_size)
+		return ioread32(adev->rio_mem + (reg * 4));
+	else {
+		iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4));
+		return ioread32(adev->rio_mem + (mmMM_DATA * 4));
+	}
+}
+
+void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+
+	if ((reg * 4) < adev->rio_mem_size)
+		iowrite32(v, adev->rio_mem + (reg * 4));
+	else {
+		iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4));
+		iowrite32(v, adev->rio_mem + (mmMM_DATA * 4));
+	}
+}
+
+/**
+ * amdgpu_mm_rdoorbell - read a doorbell dword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ *
+ * Returns the value in the doorbell aperture at the
+ * requested doorbell index (CIK).
+ */
+u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
+{
+	if (index < adev->doorbell.num_doorbells) {
+		return readl(adev->doorbell.ptr + index);
+	} else {
+		DRM_ERROR("reading beyond doorbell aperture: 0x%08x!\n", index);
+		return 0;
+	}
+}
+
+/**
+ * amdgpu_mm_wdoorbell - write a doorbell dword
+ *
+ * @adev: amdgpu_device pointer
+ * @index: doorbell index
+ * @v: value to write
+ *
+ * Writes @v to the doorbell aperture at the
+ * requested doorbell index (CIK).
+ */
+void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
+{
+	if (index < adev->doorbell.num_doorbells) {
+		writel(v, adev->doorbell.ptr + index);
+	} else {
+		DRM_ERROR("writing beyond doorbell aperture: 0x%08x!\n", index);
+	}
+}
+
+/**
+ * amdgpu_invalid_rreg - dummy reg read function
+ *
+ * @adev: amdgpu device pointer
+ * @reg: offset of register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ * Returns the value in the register.
+ */
+static uint32_t amdgpu_invalid_rreg(struct amdgpu_device *adev, uint32_t reg)
+{
+	DRM_ERROR("Invalid callback to read register 0x%04X\n", reg);
+	BUG();
+	return 0;
+}
+
+/**
+ * amdgpu_invalid_wreg - dummy reg write function
+ *
+ * @adev: amdgpu device pointer
+ * @reg: offset of register
+ * @v: value to write to the register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ */
+static void amdgpu_invalid_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
+{
+	DRM_ERROR("Invalid callback to write register 0x%04X with 0x%08X\n",
+		  reg, v);
+	BUG();
+}
+
+/**
+ * amdgpu_block_invalid_rreg - dummy reg read function
+ *
+ * @adev: amdgpu device pointer
+ * @block: offset of instance
+ * @reg: offset of register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ * Returns the value in the register.
+ */
+static uint32_t amdgpu_block_invalid_rreg(struct amdgpu_device *adev,
+					  uint32_t block, uint32_t reg)
+{
+	DRM_ERROR("Invalid callback to read register 0x%04X in block 0x%04X\n",
+		  reg, block);
+	BUG();
+	return 0;
+}
+
+/**
+ * amdgpu_block_invalid_wreg - dummy reg write function
+ *
+ * @adev: amdgpu device pointer
+ * @block: offset of instance
+ * @reg: offset of register
+ * @v: value to write to the register
+ *
+ * Dummy register read function.  Used for register blocks
+ * that certain asics don't have (all asics).
+ */
+static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
+				      uint32_t block,
+				      uint32_t reg, uint32_t v)
+{
+	DRM_ERROR("Invalid block callback to write register 0x%04X in block 0x%04X with 0x%08X\n",
+		  reg, block, v);
+	BUG();
+}
+
+static int amdgpu_vram_scratch_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->vram_scratch.robj == NULL) {
+		r = amdgpu_bo_create(adev, AMDGPU_GPU_PAGE_SIZE,
+				     PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
+				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+				     NULL, NULL, &adev->vram_scratch.robj);
+		if (r) {
+			return r;
+		}
+	}
+
+	r = amdgpu_bo_reserve(adev->vram_scratch.robj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = amdgpu_bo_pin(adev->vram_scratch.robj,
+			  AMDGPU_GEM_DOMAIN_VRAM, &adev->vram_scratch.gpu_addr);
+	if (r) {
+		amdgpu_bo_unreserve(adev->vram_scratch.robj);
+		return r;
+	}
+	r = amdgpu_bo_kmap(adev->vram_scratch.robj,
+				(void **)&adev->vram_scratch.ptr);
+	if (r)
+		amdgpu_bo_unpin(adev->vram_scratch.robj);
+	amdgpu_bo_unreserve(adev->vram_scratch.robj);
+
+	return r;
+}
+
+static void amdgpu_vram_scratch_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->vram_scratch.robj == NULL) {
+		return;
+	}
+	r = amdgpu_bo_reserve(adev->vram_scratch.robj, false);
+	if (likely(r == 0)) {
+		amdgpu_bo_kunmap(adev->vram_scratch.robj);
+		amdgpu_bo_unpin(adev->vram_scratch.robj);
+		amdgpu_bo_unreserve(adev->vram_scratch.robj);
+	}
+	amdgpu_bo_unref(&adev->vram_scratch.robj);
+}
+
+/**
+ * amdgpu_program_register_sequence - program an array of registers.
+ *
+ * @adev: amdgpu_device pointer
+ * @registers: pointer to the register array
+ * @array_size: size of the register array
+ *
+ * Programs an array or registers with and and or masks.
+ * This is a helper for setting golden registers.
+ */
+void amdgpu_program_register_sequence(struct amdgpu_device *adev,
+				      const u32 *registers,
+				      const u32 array_size)
+{
+	u32 tmp, reg, and_mask, or_mask;
+	int i;
+
+	if (array_size % 3)
+		return;
+
+	for (i = 0; i < array_size; i +=3) {
+		reg = registers[i + 0];
+		and_mask = registers[i + 1];
+		or_mask = registers[i + 2];
+
+		if (and_mask == 0xffffffff) {
+			tmp = or_mask;
+		} else {
+			tmp = RREG32(reg);
+			tmp &= ~and_mask;
+			tmp |= or_mask;
+		}
+		WREG32(reg, tmp);
+	}
+}
+
+void amdgpu_pci_config_reset(struct amdgpu_device *adev)
+{
+	pci_write_config_dword(adev->pdev, 0x7c, AMDGPU_ASIC_RESET_DATA);
+}
+
+/*
+ * GPU doorbell aperture helpers function.
+ */
+/**
+ * amdgpu_doorbell_init - Init doorbell driver information.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Init doorbell driver information (CIK)
+ * Returns 0 on success, error on failure.
+ */
+static int amdgpu_doorbell_init(struct amdgpu_device *adev)
+{
+	/* doorbell bar mapping */
+	adev->doorbell.base = pci_resource_start(adev->pdev, 2);
+	adev->doorbell.size = pci_resource_len(adev->pdev, 2);
+
+	adev->doorbell.num_doorbells = min_t(u32, adev->doorbell.size / sizeof(u32), 
+					     AMDGPU_DOORBELL_MAX_ASSIGNMENT+1);
+	if (adev->doorbell.num_doorbells == 0)
+		return -EINVAL;
+
+	adev->doorbell.ptr = ioremap(adev->doorbell.base, adev->doorbell.num_doorbells * sizeof(u32));
+	if (adev->doorbell.ptr == NULL) {
+		return -ENOMEM;
+	}
+	DRM_INFO("doorbell mmio base: 0x%08X\n", (uint32_t)adev->doorbell.base);
+	DRM_INFO("doorbell mmio size: %u\n", (unsigned)adev->doorbell.size);
+
+	return 0;
+}
+
+/**
+ * amdgpu_doorbell_fini - Tear down doorbell driver information.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down doorbell driver information (CIK)
+ */
+static void amdgpu_doorbell_fini(struct amdgpu_device *adev)
+{
+	iounmap(adev->doorbell.ptr);
+	adev->doorbell.ptr = NULL;
+}
+
+/**
+ * amdgpu_doorbell_get_kfd_info - Report doorbell configuration required to
+ *                                setup amdkfd
+ *
+ * @adev: amdgpu_device pointer
+ * @aperture_base: output returning doorbell aperture base physical address
+ * @aperture_size: output returning doorbell aperture size in bytes
+ * @start_offset: output returning # of doorbell bytes reserved for amdgpu.
+ *
+ * amdgpu and amdkfd share the doorbell aperture. amdgpu sets it up,
+ * takes doorbells required for its own rings and reports the setup to amdkfd.
+ * amdgpu reserved doorbells are at the start of the doorbell aperture.
+ */
+void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
+				phys_addr_t *aperture_base,
+				size_t *aperture_size,
+				size_t *start_offset)
+{
+	/*
+	 * The first num_doorbells are used by amdgpu.
+	 * amdkfd takes whatever's left in the aperture.
+	 */
+	if (adev->doorbell.size > adev->doorbell.num_doorbells * sizeof(u32)) {
+		*aperture_base = adev->doorbell.base;
+		*aperture_size = adev->doorbell.size;
+		*start_offset = adev->doorbell.num_doorbells * sizeof(u32);
+	} else {
+		*aperture_base = 0;
+		*aperture_size = 0;
+		*start_offset = 0;
+	}
+}
+
+/*
+ * amdgpu_wb_*()
+ * Writeback is the the method by which the the GPU updates special pages
+ * in memory with the status of certain GPU events (fences, ring pointers,
+ * etc.).
+ */
+
+/**
+ * amdgpu_wb_fini - Disable Writeback and free memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disables Writeback and frees the Writeback memory (all asics).
+ * Used at driver shutdown.
+ */
+static void amdgpu_wb_fini(struct amdgpu_device *adev)
+{
+	if (adev->wb.wb_obj) {
+		if (!amdgpu_bo_reserve(adev->wb.wb_obj, false)) {
+			amdgpu_bo_kunmap(adev->wb.wb_obj);
+			amdgpu_bo_unpin(adev->wb.wb_obj);
+			amdgpu_bo_unreserve(adev->wb.wb_obj);
+		}
+		amdgpu_bo_unref(&adev->wb.wb_obj);
+		adev->wb.wb = NULL;
+		adev->wb.wb_obj = NULL;
+	}
+}
+
+/**
+ * amdgpu_wb_init- Init Writeback driver info and allocate memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disables Writeback and frees the Writeback memory (all asics).
+ * Used at driver startup.
+ * Returns 0 on success or an -error on failure.
+ */
+static int amdgpu_wb_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->wb.wb_obj == NULL) {
+		r = amdgpu_bo_create(adev, AMDGPU_MAX_WB * 4, PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_GTT, 0,  NULL, NULL,
+				     &adev->wb.wb_obj);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create WB bo failed\n", r);
+			return r;
+		}
+		r = amdgpu_bo_reserve(adev->wb.wb_obj, false);
+		if (unlikely(r != 0)) {
+			amdgpu_wb_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_pin(adev->wb.wb_obj, AMDGPU_GEM_DOMAIN_GTT,
+				&adev->wb.gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(adev->wb.wb_obj);
+			dev_warn(adev->dev, "(%d) pin WB bo failed\n", r);
+			amdgpu_wb_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_kmap(adev->wb.wb_obj, (void **)&adev->wb.wb);
+		amdgpu_bo_unreserve(adev->wb.wb_obj);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map WB bo failed\n", r);
+			amdgpu_wb_fini(adev);
+			return r;
+		}
+
+		adev->wb.num_wb = AMDGPU_MAX_WB;
+		memset(&adev->wb.used, 0, sizeof(adev->wb.used));
+
+		/* clear wb memory */
+		memset((char *)adev->wb.wb, 0, AMDGPU_GPU_PAGE_SIZE);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_wb_get - Allocate a wb entry
+ *
+ * @adev: amdgpu_device pointer
+ * @wb: wb index
+ *
+ * Allocate a wb slot for use by the driver (all asics).
+ * Returns 0 on success or -EINVAL on failure.
+ */
+int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb)
+{
+	unsigned long offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb);
+	if (offset < adev->wb.num_wb) {
+		__set_bit(offset, adev->wb.used);
+		*wb = offset;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+/**
+ * amdgpu_wb_free - Free a wb entry
+ *
+ * @adev: amdgpu_device pointer
+ * @wb: wb index
+ *
+ * Free a wb slot allocated for use by the driver (all asics)
+ */
+void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb)
+{
+	if (wb < adev->wb.num_wb)
+		__clear_bit(wb, adev->wb.used);
+}
+
+/**
+ * amdgpu_vram_location - try to find VRAM location
+ * @adev: amdgpu device structure holding all necessary informations
+ * @mc: memory controller structure holding memory informations
+ * @base: base address at which to put VRAM
+ *
+ * Function will place try to place VRAM at base address provided
+ * as parameter (which is so far either PCI aperture address or
+ * for IGP TOM base address).
+ *
+ * If there is not enough space to fit the unvisible VRAM in the 32bits
+ * address space then we limit the VRAM size to the aperture.
+ *
+ * Note: We don't explicitly enforce VRAM start to be aligned on VRAM size,
+ * this shouldn't be a problem as we are using the PCI aperture as a reference.
+ * Otherwise this would be needed for rv280, all r3xx, and all r4xx, but
+ * not IGP.
+ *
+ * Note: we use mc_vram_size as on some board we need to program the mc to
+ * cover the whole aperture even if VRAM size is inferior to aperture size
+ * Novell bug 204882 + along with lots of ubuntu ones
+ *
+ * Note: when limiting vram it's safe to overwritte real_vram_size because
+ * we are not in case where real_vram_size is inferior to mc_vram_size (ie
+ * note afected by bogus hw of Novell bug 204882 + along with lots of ubuntu
+ * ones)
+ *
+ * Note: IGP TOM addr should be the same as the aperture addr, we don't
+ * explicitly check for that thought.
+ *
+ * FIXME: when reducing VRAM size align new size on power of 2.
+ */
+void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64 base)
+{
+	uint64_t limit = (uint64_t)amdgpu_vram_limit << 20;
+
+	mc->vram_start = base;
+	if (mc->mc_vram_size > (adev->mc.mc_mask - base + 1)) {
+		dev_warn(adev->dev, "limiting VRAM to PCI aperture size\n");
+		mc->real_vram_size = mc->aper_size;
+		mc->mc_vram_size = mc->aper_size;
+	}
+	mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
+	if (limit && limit < mc->real_vram_size)
+		mc->real_vram_size = limit;
+	dev_info(adev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
+			mc->mc_vram_size >> 20, mc->vram_start,
+			mc->vram_end, mc->real_vram_size >> 20);
+}
+
+/**
+ * amdgpu_gtt_location - try to find GTT location
+ * @adev: amdgpu device structure holding all necessary informations
+ * @mc: memory controller structure holding memory informations
+ *
+ * Function will place try to place GTT before or after VRAM.
+ *
+ * If GTT size is bigger than space left then we ajust GTT size.
+ * Thus function will never fails.
+ *
+ * FIXME: when reducing GTT size align new size on power of 2.
+ */
+void amdgpu_gtt_location(struct amdgpu_device *adev, struct amdgpu_mc *mc)
+{
+	u64 size_af, size_bf;
+
+	size_af = ((adev->mc.mc_mask - mc->vram_end) + mc->gtt_base_align) & ~mc->gtt_base_align;
+	size_bf = mc->vram_start & ~mc->gtt_base_align;
+	if (size_bf > size_af) {
+		if (mc->gtt_size > size_bf) {
+			dev_warn(adev->dev, "limiting GTT\n");
+			mc->gtt_size = size_bf;
+		}
+		mc->gtt_start = (mc->vram_start & ~mc->gtt_base_align) - mc->gtt_size;
+	} else {
+		if (mc->gtt_size > size_af) {
+			dev_warn(adev->dev, "limiting GTT\n");
+			mc->gtt_size = size_af;
+		}
+		mc->gtt_start = (mc->vram_end + 1 + mc->gtt_base_align) & ~mc->gtt_base_align;
+	}
+	mc->gtt_end = mc->gtt_start + mc->gtt_size - 1;
+	dev_info(adev->dev, "GTT: %lluM 0x%016llX - 0x%016llX\n",
+			mc->gtt_size >> 20, mc->gtt_start, mc->gtt_end);
+}
+
+/*
+ * GPU helpers function.
+ */
+/**
+ * amdgpu_card_posted - check if the hw has already been initialized
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check if the asic has been initialized (all asics).
+ * Used at driver startup.
+ * Returns true if initialized or false if not.
+ */
+bool amdgpu_card_posted(struct amdgpu_device *adev)
+{
+	uint32_t reg;
+
+	/* then check MEM_SIZE, in case the crtcs are off */
+	reg = RREG32(mmCONFIG_MEMSIZE);
+
+	if (reg)
+		return true;
+
+	return false;
+
+}
+
+/**
+ * amdgpu_boot_test_post_card - check and possibly initialize the hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check if the asic is initialized and if not, attempt to initialize
+ * it (all asics).
+ * Returns true if initialized or false if not.
+ */
+bool amdgpu_boot_test_post_card(struct amdgpu_device *adev)
+{
+	if (amdgpu_card_posted(adev))
+		return true;
+
+	if (adev->bios) {
+		DRM_INFO("GPU not posted. posting now...\n");
+		if (adev->is_atom_bios)
+			amdgpu_atom_asic_init(adev->mode_info.atom_context);
+		return true;
+	} else {
+		dev_err(adev->dev, "Card not posted and no BIOS - ignoring\n");
+		return false;
+	}
+}
+
+/**
+ * amdgpu_dummy_page_init - init dummy page used by the driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate the dummy page used by the driver (all asics).
+ * This dummy page is used by the driver as a filler for gart entries
+ * when pages are taken out of the GART
+ * Returns 0 on sucess, -ENOMEM on failure.
+ */
+int amdgpu_dummy_page_init(struct amdgpu_device *adev)
+{
+	if (adev->dummy_page.page)
+		return 0;
+	adev->dummy_page.page = alloc_page(GFP_DMA32 | GFP_KERNEL | __GFP_ZERO);
+	if (adev->dummy_page.page == NULL)
+		return -ENOMEM;
+	adev->dummy_page.addr = pci_map_page(adev->pdev, adev->dummy_page.page,
+					0, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+	if (pci_dma_mapping_error(adev->pdev, adev->dummy_page.addr)) {
+		dev_err(&adev->pdev->dev, "Failed to DMA MAP the dummy page\n");
+		__free_page(adev->dummy_page.page);
+		adev->dummy_page.page = NULL;
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_dummy_page_fini - free dummy page used by the driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Frees the dummy page used by the driver (all asics).
+ */
+void amdgpu_dummy_page_fini(struct amdgpu_device *adev)
+{
+	if (adev->dummy_page.page == NULL)
+		return;
+	pci_unmap_page(adev->pdev, adev->dummy_page.addr,
+			PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+	__free_page(adev->dummy_page.page);
+	adev->dummy_page.page = NULL;
+}
+
+
+/* ATOM accessor methods */
+/*
+ * ATOM is an interpreted byte code stored in tables in the vbios.  The
+ * driver registers callbacks to access registers and the interpreter
+ * in the driver parses the tables and executes then to program specific
+ * actions (set display modes, asic init, etc.).  See amdgpu_atombios.c,
+ * atombios.h, and atom.c
+ */
+
+/**
+ * cail_pll_read - read PLL register
+ *
+ * @info: atom card_info pointer
+ * @reg: PLL register offset
+ *
+ * Provides a PLL register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the PLL register.
+ */
+static uint32_t cail_pll_read(struct card_info *info, uint32_t reg)
+{
+	return 0;
+}
+
+/**
+ * cail_pll_write - write PLL register
+ *
+ * @info: atom card_info pointer
+ * @reg: PLL register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a PLL register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_pll_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+
+}
+
+/**
+ * cail_mc_read - read MC (Memory Controller) register
+ *
+ * @info: atom card_info pointer
+ * @reg: MC register offset
+ *
+ * Provides an MC register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the MC register.
+ */
+static uint32_t cail_mc_read(struct card_info *info, uint32_t reg)
+{
+	return 0;
+}
+
+/**
+ * cail_mc_write - write MC (Memory Controller) register
+ *
+ * @info: atom card_info pointer
+ * @reg: MC register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a MC register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_mc_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+
+}
+
+/**
+ * cail_reg_write - write MMIO register
+ *
+ * @info: atom card_info pointer
+ * @reg: MMIO register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a MMIO register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_reg_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+
+	WREG32(reg, val);
+}
+
+/**
+ * cail_reg_read - read MMIO register
+ *
+ * @info: atom card_info pointer
+ * @reg: MMIO register offset
+ *
+ * Provides an MMIO register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the MMIO register.
+ */
+static uint32_t cail_reg_read(struct card_info *info, uint32_t reg)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+	uint32_t r;
+
+	r = RREG32(reg);
+	return r;
+}
+
+/**
+ * cail_ioreg_write - write IO register
+ *
+ * @info: atom card_info pointer
+ * @reg: IO register offset
+ * @val: value to write to the pll register
+ *
+ * Provides a IO register accessor for the atom interpreter (r4xx+).
+ */
+static void cail_ioreg_write(struct card_info *info, uint32_t reg, uint32_t val)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+
+	WREG32_IO(reg, val);
+}
+
+/**
+ * cail_ioreg_read - read IO register
+ *
+ * @info: atom card_info pointer
+ * @reg: IO register offset
+ *
+ * Provides an IO register accessor for the atom interpreter (r4xx+).
+ * Returns the value of the IO register.
+ */
+static uint32_t cail_ioreg_read(struct card_info *info, uint32_t reg)
+{
+	struct amdgpu_device *adev = info->dev->dev_private;
+	uint32_t r;
+
+	r = RREG32_IO(reg);
+	return r;
+}
+
+/**
+ * amdgpu_atombios_fini - free the driver info and callbacks for atombios
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Frees the driver info and register access callbacks for the ATOM
+ * interpreter (r4xx+).
+ * Called at driver shutdown.
+ */
+static void amdgpu_atombios_fini(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.atom_context)
+		kfree(adev->mode_info.atom_context->scratch);
+	kfree(adev->mode_info.atom_context);
+	adev->mode_info.atom_context = NULL;
+	kfree(adev->mode_info.atom_card_info);
+	adev->mode_info.atom_card_info = NULL;
+}
+
+/**
+ * amdgpu_atombios_init - init the driver info and callbacks for atombios
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initializes the driver info and register access callbacks for the
+ * ATOM interpreter (r4xx+).
+ * Returns 0 on sucess, -ENOMEM on failure.
+ * Called at driver startup.
+ */
+static int amdgpu_atombios_init(struct amdgpu_device *adev)
+{
+	struct card_info *atom_card_info =
+	    kzalloc(sizeof(struct card_info), GFP_KERNEL);
+
+	if (!atom_card_info)
+		return -ENOMEM;
+
+	adev->mode_info.atom_card_info = atom_card_info;
+	atom_card_info->dev = adev->ddev;
+	atom_card_info->reg_read = cail_reg_read;
+	atom_card_info->reg_write = cail_reg_write;
+	/* needed for iio ops */
+	if (adev->rio_mem) {
+		atom_card_info->ioreg_read = cail_ioreg_read;
+		atom_card_info->ioreg_write = cail_ioreg_write;
+	} else {
+		DRM_ERROR("Unable to find PCI I/O BAR; using MMIO for ATOM IIO\n");
+		atom_card_info->ioreg_read = cail_reg_read;
+		atom_card_info->ioreg_write = cail_reg_write;
+	}
+	atom_card_info->mc_read = cail_mc_read;
+	atom_card_info->mc_write = cail_mc_write;
+	atom_card_info->pll_read = cail_pll_read;
+	atom_card_info->pll_write = cail_pll_write;
+
+	adev->mode_info.atom_context = amdgpu_atom_parse(atom_card_info, adev->bios);
+	if (!adev->mode_info.atom_context) {
+		amdgpu_atombios_fini(adev);
+		return -ENOMEM;
+	}
+
+	mutex_init(&adev->mode_info.atom_context->mutex);
+	amdgpu_atombios_scratch_regs_init(adev);
+	amdgpu_atom_allocate_fb_scratch(adev->mode_info.atom_context);
+	return 0;
+}
+
+/* if we get transitioned to only one device, take VGA back */
+/**
+ * amdgpu_vga_set_decode - enable/disable vga decode
+ *
+ * @cookie: amdgpu_device pointer
+ * @state: enable/disable vga decode
+ *
+ * Enable/disable vga decode (all asics).
+ * Returns VGA resource flags.
+ */
+static unsigned int amdgpu_vga_set_decode(void *cookie, bool state)
+{
+	struct amdgpu_device *adev = cookie;
+	amdgpu_asic_set_vga_state(adev, state);
+	if (state)
+		return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM |
+		       VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
+	else
+		return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
+}
+
+/**
+ * amdgpu_check_pot_argument - check that argument is a power of two
+ *
+ * @arg: value to check
+ *
+ * Validates that a certain argument is a power of two (all asics).
+ * Returns true if argument is valid.
+ */
+static bool amdgpu_check_pot_argument(int arg)
+{
+	return (arg & (arg - 1)) == 0;
+}
+
+/**
+ * amdgpu_check_arguments - validate module params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Validates certain module parameters and updates
+ * the associated values used by the driver (all asics).
+ */
+static void amdgpu_check_arguments(struct amdgpu_device *adev)
+{
+	/* vramlimit must be a power of two */
+	if (!amdgpu_check_pot_argument(amdgpu_vram_limit)) {
+		dev_warn(adev->dev, "vram limit (%d) must be a power of 2\n",
+				amdgpu_vram_limit);
+		amdgpu_vram_limit = 0;
+	}
+
+	if (amdgpu_gart_size != -1) {
+		/* gtt size must be power of two and greater or equal to 32M */
+		if (amdgpu_gart_size < 32) {
+			dev_warn(adev->dev, "gart size (%d) too small\n",
+				 amdgpu_gart_size);
+			amdgpu_gart_size = -1;
+		} else if (!amdgpu_check_pot_argument(amdgpu_gart_size)) {
+			dev_warn(adev->dev, "gart size (%d) must be a power of 2\n",
+				 amdgpu_gart_size);
+			amdgpu_gart_size = -1;
+		}
+	}
+
+	if (!amdgpu_check_pot_argument(amdgpu_vm_size)) {
+		dev_warn(adev->dev, "VM size (%d) must be a power of 2\n",
+			 amdgpu_vm_size);
+		amdgpu_vm_size = 8;
+	}
+
+	if (amdgpu_vm_size < 1) {
+		dev_warn(adev->dev, "VM size (%d) too small, min is 1GB\n",
+			 amdgpu_vm_size);
+		amdgpu_vm_size = 8;
+	}
+
+	/*
+	 * Max GPUVM size for Cayman, SI and CI are 40 bits.
+	 */
+	if (amdgpu_vm_size > 1024) {
+		dev_warn(adev->dev, "VM size (%d) too large, max is 1TB\n",
+			 amdgpu_vm_size);
+		amdgpu_vm_size = 8;
+	}
+
+	/* defines number of bits in page table versus page directory,
+	 * a page is 4KB so we have 12 bits offset, minimum 9 bits in the
+	 * page table and the remaining bits are in the page directory */
+	if (amdgpu_vm_block_size == -1) {
+
+		/* Total bits covered by PD + PTs */
+		unsigned bits = ilog2(amdgpu_vm_size) + 18;
+
+		/* Make sure the PD is 4K in size up to 8GB address space.
+		   Above that split equal between PD and PTs */
+		if (amdgpu_vm_size <= 8)
+			amdgpu_vm_block_size = bits - 9;
+		else
+			amdgpu_vm_block_size = (bits + 3) / 2;
+
+	} else if (amdgpu_vm_block_size < 9) {
+		dev_warn(adev->dev, "VM page table size (%d) too small\n",
+			 amdgpu_vm_block_size);
+		amdgpu_vm_block_size = 9;
+	}
+
+	if (amdgpu_vm_block_size > 24 ||
+	    (amdgpu_vm_size * 1024) < (1ull << amdgpu_vm_block_size)) {
+		dev_warn(adev->dev, "VM page table size (%d) too large\n",
+			 amdgpu_vm_block_size);
+		amdgpu_vm_block_size = 9;
+	}
+}
+
+/**
+ * amdgpu_switcheroo_set_state - set switcheroo state
+ *
+ * @pdev: pci dev pointer
+ * @state: vga_switcheroo state
+ *
+ * Callback for the switcheroo driver.  Suspends or resumes the
+ * the asics before or after it is powered up using ACPI methods.
+ */
+static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_state state)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	if (amdgpu_device_is_px(dev) && state == VGA_SWITCHEROO_OFF)
+		return;
+
+	if (state == VGA_SWITCHEROO_ON) {
+		unsigned d3_delay = dev->pdev->d3_delay;
+
+		printk(KERN_INFO "amdgpu: switched on\n");
+		/* don't suspend or resume card normally */
+		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+
+		amdgpu_resume_kms(dev, true, true);
+
+		dev->pdev->d3_delay = d3_delay;
+
+		dev->switch_power_state = DRM_SWITCH_POWER_ON;
+		drm_kms_helper_poll_enable(dev);
+	} else {
+		printk(KERN_INFO "amdgpu: switched off\n");
+		drm_kms_helper_poll_disable(dev);
+		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+		amdgpu_suspend_kms(dev, true, true);
+		dev->switch_power_state = DRM_SWITCH_POWER_OFF;
+	}
+}
+
+/**
+ * amdgpu_switcheroo_can_switch - see if switcheroo state can change
+ *
+ * @pdev: pci dev pointer
+ *
+ * Callback for the switcheroo driver.  Check of the switcheroo
+ * state can be changed.
+ * Returns true if the state can be changed, false if not.
+ */
+static bool amdgpu_switcheroo_can_switch(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	/*
+	* FIXME: open_count is protected by drm_global_mutex but that would lead to
+	* locking inversion with the driver load path. And the access here is
+	* completely racy anyway. So don't bother with locking for now.
+	*/
+	return dev->open_count == 0;
+}
+
+static const struct vga_switcheroo_client_ops amdgpu_switcheroo_ops = {
+	.set_gpu_state = amdgpu_switcheroo_set_state,
+	.reprobe = NULL,
+	.can_switch = amdgpu_switcheroo_can_switch,
+};
+
+int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
+				  enum amd_ip_block_type block_type,
+				  enum amd_clockgating_state state)
+{
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (adev->ip_blocks[i].type == block_type) {
+			r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev,
+									    state);
+			if (r)
+				return r;
+		}
+	}
+	return r;
+}
+
+int amdgpu_set_powergating_state(struct amdgpu_device *adev,
+				  enum amd_ip_block_type block_type,
+				  enum amd_powergating_state state)
+{
+	int i, r = 0;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (adev->ip_blocks[i].type == block_type) {
+			r = adev->ip_blocks[i].funcs->set_powergating_state((void *)adev,
+									    state);
+			if (r)
+				return r;
+		}
+	}
+	return r;
+}
+
+const struct amdgpu_ip_block_version * amdgpu_get_ip_block(
+					struct amdgpu_device *adev,
+					enum amd_ip_block_type type)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++)
+		if (adev->ip_blocks[i].type == type)
+			return &adev->ip_blocks[i];
+
+	return NULL;
+}
+
+/**
+ * amdgpu_ip_block_version_cmp
+ *
+ * @adev: amdgpu_device pointer
+ * @type: enum amd_ip_block_type
+ * @major: major version
+ * @minor: minor version
+ *
+ * return 0 if equal or greater
+ * return 1 if smaller or the ip_block doesn't exist
+ */
+int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev,
+				enum amd_ip_block_type type,
+				u32 major, u32 minor)
+{
+	const struct amdgpu_ip_block_version *ip_block;
+	ip_block = amdgpu_get_ip_block(adev, type);
+
+	if (ip_block && ((ip_block->major > major) ||
+			((ip_block->major == major) &&
+			(ip_block->minor >= minor))))
+		return 0;
+
+	return 1;
+}
+
+static int amdgpu_early_init(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+	case CHIP_TONGA:
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY)
+			adev->family = AMDGPU_FAMILY_CZ;
+		else
+			adev->family = AMDGPU_FAMILY_VI;
+
+		r = vi_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		if ((adev->asic_type == CHIP_BONAIRE) || (adev->asic_type == CHIP_HAWAII))
+			adev->family = AMDGPU_FAMILY_CI;
+		else
+			adev->family = AMDGPU_FAMILY_KV;
+
+		r = cik_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+#endif
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	adev->ip_block_status = kcalloc(adev->num_ip_blocks,
+					sizeof(struct amdgpu_ip_block_status), GFP_KERNEL);
+	if (adev->ip_block_status == NULL)
+		return -ENOMEM;
+
+	if (adev->ip_blocks == NULL) {
+		DRM_ERROR("No IP blocks found!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if ((amdgpu_ip_block_mask & (1 << i)) == 0) {
+			DRM_ERROR("disabled ip block: %d\n", i);
+			adev->ip_block_status[i].valid = false;
+		} else {
+			if (adev->ip_blocks[i].funcs->early_init) {
+				r = adev->ip_blocks[i].funcs->early_init((void *)adev);
+				if (r == -ENOENT)
+					adev->ip_block_status[i].valid = false;
+				else if (r)
+					return r;
+				else
+					adev->ip_block_status[i].valid = true;
+			} else {
+				adev->ip_block_status[i].valid = true;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_init(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+		r = adev->ip_blocks[i].funcs->sw_init((void *)adev);
+		if (r)
+			return r;
+		adev->ip_block_status[i].sw = true;
+		/* need to do gmc hw init early so we can allocate gpu mem */
+		if (adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_GMC) {
+			r = amdgpu_vram_scratch_init(adev);
+			if (r)
+				return r;
+			r = adev->ip_blocks[i].funcs->hw_init((void *)adev);
+			if (r)
+				return r;
+			r = amdgpu_wb_init(adev);
+			if (r)
+				return r;
+			adev->ip_block_status[i].hw = true;
+		}
+	}
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].sw)
+			continue;
+		/* gmc hw init is done early */
+		if (adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_GMC)
+			continue;
+		r = adev->ip_blocks[i].funcs->hw_init((void *)adev);
+		if (r)
+			return r;
+		adev->ip_block_status[i].hw = true;
+	}
+
+	return 0;
+}
+
+static int amdgpu_late_init(struct amdgpu_device *adev)
+{
+	int i = 0, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+		/* enable clockgating to save power */
+		r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev,
+								    AMD_CG_STATE_GATE);
+		if (r)
+			return r;
+		if (adev->ip_blocks[i].funcs->late_init) {
+			r = adev->ip_blocks[i].funcs->late_init((void *)adev);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
+static int amdgpu_fini(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_block_status[i].hw)
+			continue;
+		if (adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_GMC) {
+			amdgpu_wb_fini(adev);
+			amdgpu_vram_scratch_fini(adev);
+		}
+		/* ungate blocks before hw fini so that we can shutdown the blocks safely */
+		r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev,
+								    AMD_CG_STATE_UNGATE);
+		if (r)
+			return r;
+		r = adev->ip_blocks[i].funcs->hw_fini((void *)adev);
+		/* XXX handle errors */
+		adev->ip_block_status[i].hw = false;
+	}
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_block_status[i].sw)
+			continue;
+		r = adev->ip_blocks[i].funcs->sw_fini((void *)adev);
+		/* XXX handle errors */
+		adev->ip_block_status[i].sw = false;
+		adev->ip_block_status[i].valid = false;
+	}
+
+	return 0;
+}
+
+static int amdgpu_suspend(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+		/* ungate blocks so that suspend can properly shut them down */
+		r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev,
+								    AMD_CG_STATE_UNGATE);
+		/* XXX handle errors */
+		r = adev->ip_blocks[i].funcs->suspend(adev);
+		/* XXX handle errors */
+	}
+
+	return 0;
+}
+
+static int amdgpu_resume(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+		r = adev->ip_blocks[i].funcs->resume(adev);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_device_init - initialize the driver
+ *
+ * @adev: amdgpu_device pointer
+ * @pdev: drm dev pointer
+ * @pdev: pci dev pointer
+ * @flags: driver flags
+ *
+ * Initializes the driver info and hw (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver startup.
+ */
+int amdgpu_device_init(struct amdgpu_device *adev,
+		       struct drm_device *ddev,
+		       struct pci_dev *pdev,
+		       uint32_t flags)
+{
+	int r, i;
+	bool runtime = false;
+
+	adev->shutdown = false;
+	adev->dev = &pdev->dev;
+	adev->ddev = ddev;
+	adev->pdev = pdev;
+	adev->flags = flags;
+	adev->asic_type = flags & AMD_ASIC_MASK;
+	adev->is_atom_bios = false;
+	adev->usec_timeout = AMDGPU_MAX_USEC_TIMEOUT;
+	adev->mc.gtt_size = 512 * 1024 * 1024;
+	adev->accel_working = false;
+	adev->num_rings = 0;
+	adev->mman.buffer_funcs = NULL;
+	adev->mman.buffer_funcs_ring = NULL;
+	adev->vm_manager.vm_pte_funcs = NULL;
+	adev->vm_manager.vm_pte_funcs_ring = NULL;
+	adev->gart.gart_funcs = NULL;
+	adev->fence_context = fence_context_alloc(AMDGPU_MAX_RINGS);
+
+	adev->smc_rreg = &amdgpu_invalid_rreg;
+	adev->smc_wreg = &amdgpu_invalid_wreg;
+	adev->pcie_rreg = &amdgpu_invalid_rreg;
+	adev->pcie_wreg = &amdgpu_invalid_wreg;
+	adev->uvd_ctx_rreg = &amdgpu_invalid_rreg;
+	adev->uvd_ctx_wreg = &amdgpu_invalid_wreg;
+	adev->didt_rreg = &amdgpu_invalid_rreg;
+	adev->didt_wreg = &amdgpu_invalid_wreg;
+	adev->audio_endpt_rreg = &amdgpu_block_invalid_rreg;
+	adev->audio_endpt_wreg = &amdgpu_block_invalid_wreg;
+
+	DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
+		 amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
+		 pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
+
+	/* mutex initialization are all done here so we
+	 * can recall function without having locking issues */
+	mutex_init(&adev->ring_lock);
+	atomic_set(&adev->irq.ih.lock, 0);
+	mutex_init(&adev->gem.mutex);
+	mutex_init(&adev->pm.mutex);
+	mutex_init(&adev->gfx.gpu_clock_mutex);
+	mutex_init(&adev->srbm_mutex);
+	mutex_init(&adev->grbm_idx_mutex);
+	mutex_init(&adev->mn_lock);
+	hash_init(adev->mn_hash);
+
+	amdgpu_check_arguments(adev);
+
+	/* Registers mapping */
+	/* TODO: block userspace mapping of io register */
+	spin_lock_init(&adev->mmio_idx_lock);
+	spin_lock_init(&adev->smc_idx_lock);
+	spin_lock_init(&adev->pcie_idx_lock);
+	spin_lock_init(&adev->uvd_ctx_idx_lock);
+	spin_lock_init(&adev->didt_idx_lock);
+	spin_lock_init(&adev->audio_endpt_idx_lock);
+
+	adev->rmmio_base = pci_resource_start(adev->pdev, 5);
+	adev->rmmio_size = pci_resource_len(adev->pdev, 5);
+	adev->rmmio = ioremap(adev->rmmio_base, adev->rmmio_size);
+	if (adev->rmmio == NULL) {
+		return -ENOMEM;
+	}
+	DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base);
+	DRM_INFO("register mmio size: %u\n", (unsigned)adev->rmmio_size);
+
+	/* doorbell bar mapping */
+	amdgpu_doorbell_init(adev);
+
+	/* io port mapping */
+	for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
+		if (pci_resource_flags(adev->pdev, i) & IORESOURCE_IO) {
+			adev->rio_mem_size = pci_resource_len(adev->pdev, i);
+			adev->rio_mem = pci_iomap(adev->pdev, i, adev->rio_mem_size);
+			break;
+		}
+	}
+	if (adev->rio_mem == NULL)
+		DRM_ERROR("Unable to find PCI I/O BAR\n");
+
+	/* early init functions */
+	r = amdgpu_early_init(adev);
+	if (r)
+		return r;
+
+	/* if we have > 1 VGA cards, then disable the amdgpu VGA resources */
+	/* this will fail for cards that aren't VGA class devices, just
+	 * ignore it */
+	vga_client_register(adev->pdev, adev, NULL, amdgpu_vga_set_decode);
+
+	if (amdgpu_runtime_pm == 1)
+		runtime = true;
+	if (amdgpu_device_is_px(ddev))
+		runtime = true;
+	vga_switcheroo_register_client(adev->pdev, &amdgpu_switcheroo_ops, runtime);
+	if (runtime)
+		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
+
+	/* Read BIOS */
+	if (!amdgpu_get_bios(adev))
+		return -EINVAL;
+	/* Must be an ATOMBIOS */
+	if (!adev->is_atom_bios) {
+		dev_err(adev->dev, "Expecting atombios for GPU\n");
+		return -EINVAL;
+	}
+	r = amdgpu_atombios_init(adev);
+	if (r)
+		return r;
+
+	/* Post card if necessary */
+	if (!amdgpu_card_posted(adev)) {
+		if (!adev->bios) {
+			dev_err(adev->dev, "Card not posted and no BIOS - ignoring\n");
+			return -EINVAL;
+		}
+		DRM_INFO("GPU not posted. posting now...\n");
+		amdgpu_atom_asic_init(adev->mode_info.atom_context);
+	}
+
+	/* Initialize clocks */
+	r = amdgpu_atombios_get_clock_info(adev);
+	if (r)
+		return r;
+	/* init i2c buses */
+	amdgpu_atombios_i2c_init(adev);
+
+	/* Fence driver */
+	r = amdgpu_fence_driver_init(adev);
+	if (r)
+		return r;
+
+	/* init the mode config */
+	drm_mode_config_init(adev->ddev);
+
+	r = amdgpu_init(adev);
+	if (r) {
+		amdgpu_fini(adev);
+		return r;
+	}
+
+	adev->accel_working = true;
+
+	amdgpu_fbdev_init(adev);
+
+	r = amdgpu_ib_pool_init(adev);
+	if (r) {
+		dev_err(adev->dev, "IB initialization failed (%d).\n", r);
+		return r;
+	}
+
+	r = amdgpu_ctx_init(adev, true, &adev->kernel_ctx);
+	if (r) {
+		dev_err(adev->dev, "failed to create kernel context (%d).\n", r);
+		return r;
+	}
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		DRM_ERROR("ib ring test failed (%d).\n", r);
+
+	r = amdgpu_gem_debugfs_init(adev);
+	if (r) {
+		DRM_ERROR("registering gem debugfs failed (%d).\n", r);
+	}
+
+	r = amdgpu_debugfs_regs_init(adev);
+	if (r) {
+		DRM_ERROR("registering register debugfs failed (%d).\n", r);
+	}
+
+	if ((amdgpu_testing & 1)) {
+		if (adev->accel_working)
+			amdgpu_test_moves(adev);
+		else
+			DRM_INFO("amdgpu: acceleration disabled, skipping move tests\n");
+	}
+	if ((amdgpu_testing & 2)) {
+		if (adev->accel_working)
+			amdgpu_test_syncing(adev);
+		else
+			DRM_INFO("amdgpu: acceleration disabled, skipping sync tests\n");
+	}
+	if (amdgpu_benchmarking) {
+		if (adev->accel_working)
+			amdgpu_benchmark(adev, amdgpu_benchmarking);
+		else
+			DRM_INFO("amdgpu: acceleration disabled, skipping benchmarks\n");
+	}
+
+	/* enable clockgating, etc. after ib tests, etc. since some blocks require
+	 * explicit gating rather than handling it automatically.
+	 */
+	r = amdgpu_late_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void amdgpu_debugfs_remove_files(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_device_fini - tear down the driver
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the driver info (all asics).
+ * Called at driver shutdown.
+ */
+void amdgpu_device_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	DRM_INFO("amdgpu: finishing device.\n");
+	adev->shutdown = true;
+	/* evict vram memory */
+	amdgpu_bo_evict_vram(adev);
+	amdgpu_ctx_fini(&adev->kernel_ctx);
+	amdgpu_ib_pool_fini(adev);
+	amdgpu_fence_driver_fini(adev);
+	amdgpu_fbdev_fini(adev);
+	r = amdgpu_fini(adev);
+	kfree(adev->ip_block_status);
+	adev->ip_block_status = NULL;
+	adev->accel_working = false;
+	/* free i2c buses */
+	amdgpu_i2c_fini(adev);
+	amdgpu_atombios_fini(adev);
+	kfree(adev->bios);
+	adev->bios = NULL;
+	vga_switcheroo_unregister_client(adev->pdev);
+	vga_client_register(adev->pdev, NULL, NULL, NULL);
+	if (adev->rio_mem)
+		pci_iounmap(adev->pdev, adev->rio_mem);
+	adev->rio_mem = NULL;
+	iounmap(adev->rmmio);
+	adev->rmmio = NULL;
+	amdgpu_doorbell_fini(adev);
+	amdgpu_debugfs_regs_cleanup(adev);
+	amdgpu_debugfs_remove_files(adev);
+}
+
+
+/*
+ * Suspend & resume.
+ */
+/**
+ * amdgpu_suspend_kms - initiate device suspend
+ *
+ * @pdev: drm dev pointer
+ * @state: suspend state
+ *
+ * Puts the hw in the suspend state (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver suspend.
+ */
+int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
+{
+	struct amdgpu_device *adev;
+	struct drm_crtc *crtc;
+	struct drm_connector *connector;
+	int r;
+
+	if (dev == NULL || dev->dev_private == NULL) {
+		return -ENODEV;
+	}
+
+	adev = dev->dev_private;
+
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+		return 0;
+
+	drm_kms_helper_poll_disable(dev);
+
+	/* turn off display hw */
+	drm_modeset_lock_all(dev);
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF);
+	}
+	drm_modeset_unlock_all(dev);
+
+	/* unpin the front buffers and cursors */
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+		struct amdgpu_framebuffer *rfb = to_amdgpu_framebuffer(crtc->primary->fb);
+		struct amdgpu_bo *robj;
+
+		if (amdgpu_crtc->cursor_bo) {
+			struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+			r = amdgpu_bo_reserve(aobj, false);
+			if (r == 0) {
+				amdgpu_bo_unpin(aobj);
+				amdgpu_bo_unreserve(aobj);
+			}
+		}
+
+		if (rfb == NULL || rfb->obj == NULL) {
+			continue;
+		}
+		robj = gem_to_amdgpu_bo(rfb->obj);
+		/* don't unpin kernel fb objects */
+		if (!amdgpu_fbdev_robj_is_fb(adev, robj)) {
+			r = amdgpu_bo_reserve(robj, false);
+			if (r == 0) {
+				amdgpu_bo_unpin(robj);
+				amdgpu_bo_unreserve(robj);
+			}
+		}
+	}
+	/* evict vram memory */
+	amdgpu_bo_evict_vram(adev);
+
+	amdgpu_fence_driver_suspend(adev);
+
+	r = amdgpu_suspend(adev);
+
+	/* evict remaining vram memory */
+	amdgpu_bo_evict_vram(adev);
+
+	pci_save_state(dev->pdev);
+	if (suspend) {
+		/* Shut down the device */
+		pci_disable_device(dev->pdev);
+		pci_set_power_state(dev->pdev, PCI_D3hot);
+	}
+
+	if (fbcon) {
+		console_lock();
+		amdgpu_fbdev_set_suspend(adev, 1);
+		console_unlock();
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_resume_kms - initiate device resume
+ *
+ * @pdev: drm dev pointer
+ *
+ * Bring the hw back to operating state (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver resume.
+ */
+int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon)
+{
+	struct drm_connector *connector;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_crtc *crtc;
+	int r;
+
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+		return 0;
+
+	if (fbcon) {
+		console_lock();
+	}
+	if (resume) {
+		pci_set_power_state(dev->pdev, PCI_D0);
+		pci_restore_state(dev->pdev);
+		if (pci_enable_device(dev->pdev)) {
+			if (fbcon)
+				console_unlock();
+			return -1;
+		}
+	}
+
+	/* post card */
+	if (!amdgpu_card_posted(adev))
+		amdgpu_atom_asic_init(adev->mode_info.atom_context);
+
+	r = amdgpu_resume(adev);
+	if (r)
+		DRM_ERROR("amdgpu_resume failed (%d).\n", r);
+
+	amdgpu_fence_driver_resume(adev);
+
+	if (resume) {
+		r = amdgpu_ib_ring_tests(adev);
+		if (r)
+			DRM_ERROR("ib ring test failed (%d).\n", r);
+	}
+
+	r = amdgpu_late_init(adev);
+	if (r)
+		return r;
+
+	/* pin cursors */
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+		if (amdgpu_crtc->cursor_bo) {
+			struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+			r = amdgpu_bo_reserve(aobj, false);
+			if (r == 0) {
+				r = amdgpu_bo_pin(aobj,
+						  AMDGPU_GEM_DOMAIN_VRAM,
+						  &amdgpu_crtc->cursor_addr);
+				if (r != 0)
+					DRM_ERROR("Failed to pin cursor BO (%d)\n", r);
+				amdgpu_bo_unreserve(aobj);
+			}
+		}
+	}
+
+	/* blat the mode back in */
+	if (fbcon) {
+		drm_helper_resume_force_mode(dev);
+		/* turn on display hw */
+		drm_modeset_lock_all(dev);
+		list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+			drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON);
+		}
+		drm_modeset_unlock_all(dev);
+	}
+
+	drm_kms_helper_poll_enable(dev);
+	drm_helper_hpd_irq_event(dev);
+
+	if (fbcon) {
+		amdgpu_fbdev_set_suspend(adev, 0);
+		console_unlock();
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_gpu_reset - reset the asic
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Attempt the reset the GPU if it has hung (all asics).
+ * Returns 0 for success or an error on failure.
+ */
+int amdgpu_gpu_reset(struct amdgpu_device *adev)
+{
+	unsigned ring_sizes[AMDGPU_MAX_RINGS];
+	uint32_t *ring_data[AMDGPU_MAX_RINGS];
+
+	bool saved = false;
+
+	int i, r;
+	int resched;
+
+	atomic_inc(&adev->gpu_reset_counter);
+
+	/* block TTM */
+	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
+
+	r = amdgpu_suspend(adev);
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring)
+			continue;
+
+		ring_sizes[i] = amdgpu_ring_backup(ring, &ring_data[i]);
+		if (ring_sizes[i]) {
+			saved = true;
+			dev_info(adev->dev, "Saved %d dwords of commands "
+				 "on ring %d.\n", ring_sizes[i], i);
+		}
+	}
+
+retry:
+	r = amdgpu_asic_reset(adev);
+	if (!r) {
+		dev_info(adev->dev, "GPU reset succeeded, trying to resume\n");
+		r = amdgpu_resume(adev);
+	}
+
+	if (!r) {
+		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+			struct amdgpu_ring *ring = adev->rings[i];
+			if (!ring)
+				continue;
+
+			amdgpu_ring_restore(ring, ring_sizes[i], ring_data[i]);
+			ring_sizes[i] = 0;
+			ring_data[i] = NULL;
+		}
+
+		r = amdgpu_ib_ring_tests(adev);
+		if (r) {
+			dev_err(adev->dev, "ib ring test failed (%d).\n", r);
+			if (saved) {
+				saved = false;
+				r = amdgpu_suspend(adev);
+				goto retry;
+			}
+		}
+	} else {
+		amdgpu_fence_driver_force_completion(adev);
+		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+			if (adev->rings[i])
+				kfree(ring_data[i]);
+		}
+	}
+
+	drm_helper_resume_force_mode(adev->ddev);
+
+	ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
+	if (r) {
+		/* bad news, how to tell it to userspace ? */
+		dev_info(adev->dev, "GPU reset failed\n");
+	}
+
+	return r;
+}
+
+
+/*
+ * Debugfs
+ */
+int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
+			     struct drm_info_list *files,
+			     unsigned nfiles)
+{
+	unsigned i;
+
+	for (i = 0; i < adev->debugfs_count; i++) {
+		if (adev->debugfs[i].files == files) {
+			/* Already registered */
+			return 0;
+		}
+	}
+
+	i = adev->debugfs_count + 1;
+	if (i > AMDGPU_DEBUGFS_MAX_COMPONENTS) {
+		DRM_ERROR("Reached maximum number of debugfs components.\n");
+		DRM_ERROR("Report so we increase "
+			  "AMDGPU_DEBUGFS_MAX_COMPONENTS.\n");
+		return -EINVAL;
+	}
+	adev->debugfs[adev->debugfs_count].files = files;
+	adev->debugfs[adev->debugfs_count].num_files = nfiles;
+	adev->debugfs_count = i;
+#if defined(CONFIG_DEBUG_FS)
+	drm_debugfs_create_files(files, nfiles,
+				 adev->ddev->control->debugfs_root,
+				 adev->ddev->control);
+	drm_debugfs_create_files(files, nfiles,
+				 adev->ddev->primary->debugfs_root,
+				 adev->ddev->primary);
+#endif
+	return 0;
+}
+
+static void amdgpu_debugfs_remove_files(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	unsigned i;
+
+	for (i = 0; i < adev->debugfs_count; i++) {
+		drm_debugfs_remove_files(adev->debugfs[i].files,
+					 adev->debugfs[i].num_files,
+					 adev->ddev->control);
+		drm_debugfs_remove_files(adev->debugfs[i].files,
+					 adev->debugfs[i].num_files,
+					 adev->ddev->primary);
+	}
+#endif
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		if (*pos > adev->rmmio_size)
+			return result;
+
+		value = RREG32(*pos >> 2);
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+static ssize_t amdgpu_debugfs_regs_write(struct file *f, const char __user *buf,
+					 size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		uint32_t value;
+
+		if (*pos > adev->rmmio_size)
+			return result;
+
+		r = get_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		WREG32(*pos >> 2, value);
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_debugfs_regs_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_regs_read,
+	.write = amdgpu_debugfs_regs_write,
+	.llseek = default_llseek
+};
+
+static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+{
+	struct drm_minor *minor = adev->ddev->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+
+	ent = debugfs_create_file("amdgpu_regs", S_IFREG | S_IRUGO, root,
+				  adev, &amdgpu_debugfs_regs_fops);
+	if (IS_ERR(ent))
+		return PTR_ERR(ent);
+	i_size_write(ent->d_inode, adev->rmmio_size);
+	adev->debugfs_regs = ent;
+
+	return 0;
+}
+
+static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev)
+{
+	debugfs_remove(adev->debugfs_regs);
+	adev->debugfs_regs = NULL;
+}
+
+int amdgpu_debugfs_init(struct drm_minor *minor)
+{
+	return 0;
+}
+
+void amdgpu_debugfs_cleanup(struct drm_minor *minor)
+{
+}
+#else
+static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
+{
+	return 0;
+}
+static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev) { }
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
new file mode 100644
index 000000000..82903ca78
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -0,0 +1,910 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_i2c.h"
+#include "atom.h"
+#include "amdgpu_connectors.h"
+#include <asm/div64.h>
+
+#include <linux/pm_runtime.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_edid.h>
+
+static void amdgpu_flip_wait_fence(struct amdgpu_device *adev,
+				   struct fence **f)
+{
+	struct amdgpu_fence *fence;
+	long r;
+
+	if (*f == NULL)
+		return;
+
+	fence = to_amdgpu_fence(*f);
+	if (fence) {
+		r = fence_wait(&fence->base, false);
+		if (r == -EDEADLK)
+			r = amdgpu_gpu_reset(adev);
+	} else
+		r = fence_wait(*f, false);
+
+	if (r)
+		DRM_ERROR("failed to wait on page flip fence (%ld)!\n", r);
+
+	/* We continue with the page flip even if we failed to wait on
+	 * the fence, otherwise the DRM core and userspace will be
+	 * confused about which BO the CRTC is scanning out
+	 */
+	fence_put(*f);
+	*f = NULL;
+}
+
+static void amdgpu_flip_work_func(struct work_struct *__work)
+{
+	struct amdgpu_flip_work *work =
+		container_of(__work, struct amdgpu_flip_work, flip_work);
+	struct amdgpu_device *adev = work->adev;
+	struct amdgpu_crtc *amdgpuCrtc = adev->mode_info.crtcs[work->crtc_id];
+
+	struct drm_crtc *crtc = &amdgpuCrtc->base;
+	unsigned long flags;
+	unsigned i, repcnt = 4;
+	int vpos, hpos, stat, min_udelay = 0;
+	struct drm_vblank_crtc *vblank = &crtc->dev->vblank[work->crtc_id];
+
+	amdgpu_flip_wait_fence(adev, &work->excl);
+	for (i = 0; i < work->shared_count; ++i)
+		amdgpu_flip_wait_fence(adev, &work->shared[i]);
+
+	/* We borrow the event spin lock for protecting flip_status */
+	spin_lock_irqsave(&crtc->dev->event_lock, flags);
+
+	/* If this happens to execute within the "virtually extended" vblank
+	 * interval before the start of the real vblank interval then it needs
+	 * to delay programming the mmio flip until the real vblank is entered.
+	 * This prevents completing a flip too early due to the way we fudge
+	 * our vblank counter and vblank timestamps in order to work around the
+	 * problem that the hw fires vblank interrupts before actual start of
+	 * vblank (when line buffer refilling is done for a frame). It
+	 * complements the fudging logic in amdgpu_get_crtc_scanoutpos() for
+	 * timestamping and amdgpu_get_vblank_counter_kms() for vblank counts.
+	 *
+	 * In practice this won't execute very often unless on very fast
+	 * machines because the time window for this to happen is very small.
+	 */
+	while (amdgpuCrtc->enabled && --repcnt) {
+		/* GET_DISTANCE_TO_VBLANKSTART returns distance to real vblank
+		 * start in hpos, and to the "fudged earlier" vblank start in
+		 * vpos.
+		 */
+		stat = amdgpu_get_crtc_scanoutpos(adev->ddev, work->crtc_id,
+						  GET_DISTANCE_TO_VBLANKSTART,
+						  &vpos, &hpos, NULL, NULL,
+						  &crtc->hwmode);
+
+		if ((stat & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE)) !=
+		    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE) ||
+		    !(vpos >= 0 && hpos <= 0))
+			break;
+
+		/* Sleep at least until estimated real start of hw vblank */
+		min_udelay = (-hpos + 1) * max(vblank->linedur_ns / 1000, 5);
+		if (min_udelay > vblank->framedur_ns / 2000) {
+			/* Don't wait ridiculously long - something is wrong */
+			repcnt = 0;
+			break;
+		}
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+		usleep_range(min_udelay, 2 * min_udelay);
+		spin_lock_irqsave(&crtc->dev->event_lock, flags);
+	};
+
+	if (!repcnt)
+		DRM_DEBUG_DRIVER("Delay problem on crtc %d: min_udelay %d, "
+				 "framedur %d, linedur %d, stat %d, vpos %d, "
+				 "hpos %d\n", work->crtc_id, min_udelay,
+				 vblank->framedur_ns / 1000,
+				 vblank->linedur_ns / 1000, stat, vpos, hpos);
+
+	/* do the flip (mmio) */
+	adev->mode_info.funcs->page_flip(adev, work->crtc_id, work->base);
+	/* set the flip status */
+	amdgpuCrtc->pflip_status = AMDGPU_FLIP_SUBMITTED;
+
+	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+}
+
+/*
+ * Handle unpin events outside the interrupt handler proper.
+ */
+static void amdgpu_unpin_work_func(struct work_struct *__work)
+{
+	struct amdgpu_flip_work *work =
+		container_of(__work, struct amdgpu_flip_work, unpin_work);
+	int r;
+
+	/* unpin of the old buffer */
+	r = amdgpu_bo_reserve(work->old_rbo, false);
+	if (likely(r == 0)) {
+		r = amdgpu_bo_unpin(work->old_rbo);
+		if (unlikely(r != 0)) {
+			DRM_ERROR("failed to unpin buffer after flip\n");
+		}
+		amdgpu_bo_unreserve(work->old_rbo);
+	} else
+		DRM_ERROR("failed to reserve buffer after flip\n");
+
+	amdgpu_bo_unref(&work->old_rbo);
+	kfree(work->shared);
+	kfree(work);
+}
+
+int amdgpu_crtc_page_flip(struct drm_crtc *crtc,
+			  struct drm_framebuffer *fb,
+			  struct drm_pending_vblank_event *event,
+			  uint32_t page_flip_flags)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_framebuffer *old_amdgpu_fb;
+	struct amdgpu_framebuffer *new_amdgpu_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_flip_work *work;
+	struct amdgpu_bo *new_rbo;
+	unsigned long flags;
+	u64 tiling_flags;
+	u64 base;
+	int i, r;
+
+	work = kzalloc(sizeof *work, GFP_KERNEL);
+	if (work == NULL)
+		return -ENOMEM;
+
+	INIT_WORK(&work->flip_work, amdgpu_flip_work_func);
+	INIT_WORK(&work->unpin_work, amdgpu_unpin_work_func);
+
+	work->event = event;
+	work->adev = adev;
+	work->crtc_id = amdgpu_crtc->crtc_id;
+
+	/* schedule unpin of the old buffer */
+	old_amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+	obj = old_amdgpu_fb->obj;
+
+	/* take a reference to the old object */
+	work->old_rbo = gem_to_amdgpu_bo(obj);
+	amdgpu_bo_ref(work->old_rbo);
+
+	new_amdgpu_fb = to_amdgpu_framebuffer(fb);
+	obj = new_amdgpu_fb->obj;
+	new_rbo = gem_to_amdgpu_bo(obj);
+
+	/* pin the new buffer */
+	r = amdgpu_bo_reserve(new_rbo, false);
+	if (unlikely(r != 0)) {
+		DRM_ERROR("failed to reserve new rbo buffer before flip\n");
+		goto cleanup;
+	}
+
+	r = amdgpu_bo_pin_restricted(new_rbo, AMDGPU_GEM_DOMAIN_VRAM, 0, 0, &base);
+	if (unlikely(r != 0)) {
+		amdgpu_bo_unreserve(new_rbo);
+		r = -EINVAL;
+		DRM_ERROR("failed to pin new rbo buffer before flip\n");
+		goto cleanup;
+	}
+
+	r = reservation_object_get_fences_rcu(new_rbo->tbo.resv, &work->excl,
+					      &work->shared_count,
+					      &work->shared);
+	if (unlikely(r != 0)) {
+		amdgpu_bo_unreserve(new_rbo);
+		DRM_ERROR("failed to get fences for buffer\n");
+		goto cleanup;
+	}
+
+	amdgpu_bo_get_tiling_flags(new_rbo, &tiling_flags);
+	amdgpu_bo_unreserve(new_rbo);
+
+	work->base = base;
+
+	r = drm_vblank_get(crtc->dev, amdgpu_crtc->crtc_id);
+	if (r) {
+		DRM_ERROR("failed to get vblank before flip\n");
+		goto pflip_cleanup;
+	}
+
+	/* we borrow the event spin lock for protecting flip_wrok */
+	spin_lock_irqsave(&crtc->dev->event_lock, flags);
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_NONE) {
+		DRM_DEBUG_DRIVER("flip queue: crtc already busy\n");
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+		r = -EBUSY;
+		goto vblank_cleanup;
+	}
+
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_PENDING;
+	amdgpu_crtc->pflip_works = work;
+
+	/* update crtc fb */
+	crtc->primary->fb = fb;
+	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+	queue_work(amdgpu_crtc->pflip_queue, &work->flip_work);
+	return 0;
+
+vblank_cleanup:
+	drm_vblank_put(crtc->dev, amdgpu_crtc->crtc_id);
+
+pflip_cleanup:
+	if (unlikely(amdgpu_bo_reserve(new_rbo, false) != 0)) {
+		DRM_ERROR("failed to reserve new rbo in error path\n");
+		goto cleanup;
+	}
+	if (unlikely(amdgpu_bo_unpin(new_rbo) != 0)) {
+		DRM_ERROR("failed to unpin new rbo in error path\n");
+	}
+	amdgpu_bo_unreserve(new_rbo);
+
+cleanup:
+	amdgpu_bo_unref(&work->old_rbo);
+	fence_put(work->excl);
+	for (i = 0; i < work->shared_count; ++i)
+		fence_put(work->shared[i]);
+	kfree(work->shared);
+	kfree(work);
+
+	return r;
+}
+
+int amdgpu_crtc_set_config(struct drm_mode_set *set)
+{
+	struct drm_device *dev;
+	struct amdgpu_device *adev;
+	struct drm_crtc *crtc;
+	bool active = false;
+	int ret;
+
+	if (!set || !set->crtc)
+		return -EINVAL;
+
+	dev = set->crtc->dev;
+
+	ret = pm_runtime_get_sync(dev->dev);
+	if (ret < 0)
+		return ret;
+
+	ret = drm_crtc_helper_set_config(set);
+
+	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head)
+		if (crtc->enabled)
+			active = true;
+
+	pm_runtime_mark_last_busy(dev->dev);
+
+	adev = dev->dev_private;
+	/* if we have active crtcs and we don't have a power ref,
+	   take the current one */
+	if (active && !adev->have_disp_power_ref) {
+		adev->have_disp_power_ref = true;
+		return ret;
+	}
+	/* if we have no active crtcs, then drop the power ref
+	   we got before */
+	if (!active && adev->have_disp_power_ref) {
+		pm_runtime_put_autosuspend(dev->dev);
+		adev->have_disp_power_ref = false;
+	}
+
+	/* drop the power reference we got coming in here */
+	pm_runtime_put_autosuspend(dev->dev);
+	return ret;
+}
+
+static const char *encoder_names[38] = {
+	"NONE",
+	"INTERNAL_LVDS",
+	"INTERNAL_TMDS1",
+	"INTERNAL_TMDS2",
+	"INTERNAL_DAC1",
+	"INTERNAL_DAC2",
+	"INTERNAL_SDVOA",
+	"INTERNAL_SDVOB",
+	"SI170B",
+	"CH7303",
+	"CH7301",
+	"INTERNAL_DVO1",
+	"EXTERNAL_SDVOA",
+	"EXTERNAL_SDVOB",
+	"TITFP513",
+	"INTERNAL_LVTM1",
+	"VT1623",
+	"HDMI_SI1930",
+	"HDMI_INTERNAL",
+	"INTERNAL_KLDSCP_TMDS1",
+	"INTERNAL_KLDSCP_DVO1",
+	"INTERNAL_KLDSCP_DAC1",
+	"INTERNAL_KLDSCP_DAC2",
+	"SI178",
+	"MVPU_FPGA",
+	"INTERNAL_DDI",
+	"VT1625",
+	"HDMI_SI1932",
+	"DP_AN9801",
+	"DP_DP501",
+	"INTERNAL_UNIPHY",
+	"INTERNAL_KLDSCP_LVTMA",
+	"INTERNAL_UNIPHY1",
+	"INTERNAL_UNIPHY2",
+	"NUTMEG",
+	"TRAVIS",
+	"INTERNAL_VCE",
+	"INTERNAL_UNIPHY3",
+};
+
+static const char *hpd_names[6] = {
+	"HPD1",
+	"HPD2",
+	"HPD3",
+	"HPD4",
+	"HPD5",
+	"HPD6",
+};
+
+void amdgpu_print_display_setup(struct drm_device *dev)
+{
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+	uint32_t devices;
+	int i = 0;
+
+	DRM_INFO("AMDGPU Display Connectors\n");
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		DRM_INFO("Connector %d:\n", i);
+		DRM_INFO("  %s\n", connector->name);
+		if (amdgpu_connector->hpd.hpd != AMDGPU_HPD_NONE)
+			DRM_INFO("  %s\n", hpd_names[amdgpu_connector->hpd.hpd]);
+		if (amdgpu_connector->ddc_bus) {
+			DRM_INFO("  DDC: 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x\n",
+				 amdgpu_connector->ddc_bus->rec.mask_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.mask_data_reg,
+				 amdgpu_connector->ddc_bus->rec.a_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.a_data_reg,
+				 amdgpu_connector->ddc_bus->rec.en_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.en_data_reg,
+				 amdgpu_connector->ddc_bus->rec.y_clk_reg,
+				 amdgpu_connector->ddc_bus->rec.y_data_reg);
+			if (amdgpu_connector->router.ddc_valid)
+				DRM_INFO("  DDC Router 0x%x/0x%x\n",
+					 amdgpu_connector->router.ddc_mux_control_pin,
+					 amdgpu_connector->router.ddc_mux_state);
+			if (amdgpu_connector->router.cd_valid)
+				DRM_INFO("  Clock/Data Router 0x%x/0x%x\n",
+					 amdgpu_connector->router.cd_mux_control_pin,
+					 amdgpu_connector->router.cd_mux_state);
+		} else {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_VGA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVII ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVID ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVIA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_HDMIA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				DRM_INFO("  DDC: no ddc bus - possible BIOS bug - please report to xorg-driver-ati@lists.x.org\n");
+		}
+		DRM_INFO("  Encoders:\n");
+		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+			amdgpu_encoder = to_amdgpu_encoder(encoder);
+			devices = amdgpu_encoder->devices & amdgpu_connector->devices;
+			if (devices) {
+				if (devices & ATOM_DEVICE_CRT1_SUPPORT)
+					DRM_INFO("    CRT1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_CRT2_SUPPORT)
+					DRM_INFO("    CRT2: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_LCD1_SUPPORT)
+					DRM_INFO("    LCD1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP1_SUPPORT)
+					DRM_INFO("    DFP1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP2_SUPPORT)
+					DRM_INFO("    DFP2: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP3_SUPPORT)
+					DRM_INFO("    DFP3: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP4_SUPPORT)
+					DRM_INFO("    DFP4: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP5_SUPPORT)
+					DRM_INFO("    DFP5: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_DFP6_SUPPORT)
+					DRM_INFO("    DFP6: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_TV1_SUPPORT)
+					DRM_INFO("    TV1: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+				if (devices & ATOM_DEVICE_CV_SUPPORT)
+					DRM_INFO("    CV: %s\n", encoder_names[amdgpu_encoder->encoder_id]);
+			}
+		}
+		i++;
+	}
+}
+
+/**
+ * amdgpu_ddc_probe
+ *
+ */
+bool amdgpu_ddc_probe(struct amdgpu_connector *amdgpu_connector,
+		       bool use_aux)
+{
+	u8 out = 0x0;
+	u8 buf[8];
+	int ret;
+	struct i2c_msg msgs[] = {
+		{
+			.addr = DDC_ADDR,
+			.flags = 0,
+			.len = 1,
+			.buf = &out,
+		},
+		{
+			.addr = DDC_ADDR,
+			.flags = I2C_M_RD,
+			.len = 8,
+			.buf = buf,
+		}
+	};
+
+	/* on hw with routers, select right port */
+	if (amdgpu_connector->router.ddc_valid)
+		amdgpu_i2c_router_select_ddc_port(amdgpu_connector);
+
+	if (use_aux) {
+		ret = i2c_transfer(&amdgpu_connector->ddc_bus->aux.ddc, msgs, 2);
+	} else {
+		ret = i2c_transfer(&amdgpu_connector->ddc_bus->adapter, msgs, 2);
+	}
+
+	if (ret != 2)
+		/* Couldn't find an accessible DDC on this connector */
+		return false;
+	/* Probe also for valid EDID header
+	 * EDID header starts with:
+	 * 0x00,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0x00.
+	 * Only the first 6 bytes must be valid as
+	 * drm_edid_block_valid() can fix the last 2 bytes */
+	if (drm_edid_header_is_valid(buf) < 6) {
+		/* Couldn't find an accessible EDID on this
+		 * connector */
+		return false;
+	}
+	return true;
+}
+
+static void amdgpu_user_framebuffer_destroy(struct drm_framebuffer *fb)
+{
+	struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
+
+	if (amdgpu_fb->obj) {
+		drm_gem_object_unreference_unlocked(amdgpu_fb->obj);
+	}
+	drm_framebuffer_cleanup(fb);
+	kfree(amdgpu_fb);
+}
+
+static int amdgpu_user_framebuffer_create_handle(struct drm_framebuffer *fb,
+						  struct drm_file *file_priv,
+						  unsigned int *handle)
+{
+	struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
+
+	return drm_gem_handle_create(file_priv, amdgpu_fb->obj, handle);
+}
+
+static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
+	.destroy = amdgpu_user_framebuffer_destroy,
+	.create_handle = amdgpu_user_framebuffer_create_handle,
+};
+
+int
+amdgpu_framebuffer_init(struct drm_device *dev,
+			struct amdgpu_framebuffer *rfb,
+			struct drm_mode_fb_cmd2 *mode_cmd,
+			struct drm_gem_object *obj)
+{
+	int ret;
+	rfb->obj = obj;
+	drm_helper_mode_fill_fb_struct(&rfb->base, mode_cmd);
+	ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
+	if (ret) {
+		rfb->obj = NULL;
+		return ret;
+	}
+	return 0;
+}
+
+static struct drm_framebuffer *
+amdgpu_user_framebuffer_create(struct drm_device *dev,
+			       struct drm_file *file_priv,
+			       struct drm_mode_fb_cmd2 *mode_cmd)
+{
+	struct drm_gem_object *obj;
+	struct amdgpu_framebuffer *amdgpu_fb;
+	int ret;
+
+	obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handles[0]);
+	if (obj ==  NULL) {
+		dev_err(&dev->pdev->dev, "No GEM object associated to handle 0x%08X, "
+			"can't create framebuffer\n", mode_cmd->handles[0]);
+		return ERR_PTR(-ENOENT);
+	}
+
+	amdgpu_fb = kzalloc(sizeof(*amdgpu_fb), GFP_KERNEL);
+	if (amdgpu_fb == NULL) {
+		drm_gem_object_unreference_unlocked(obj);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	ret = amdgpu_framebuffer_init(dev, amdgpu_fb, mode_cmd, obj);
+	if (ret) {
+		kfree(amdgpu_fb);
+		drm_gem_object_unreference_unlocked(obj);
+		return ERR_PTR(ret);
+	}
+
+	return &amdgpu_fb->base;
+}
+
+static void amdgpu_output_poll_changed(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	amdgpu_fb_output_poll_changed(adev);
+}
+
+const struct drm_mode_config_funcs amdgpu_mode_funcs = {
+	.fb_create = amdgpu_user_framebuffer_create,
+	.output_poll_changed = amdgpu_output_poll_changed
+};
+
+static struct drm_prop_enum_list amdgpu_underscan_enum_list[] =
+{	{ UNDERSCAN_OFF, "off" },
+	{ UNDERSCAN_ON, "on" },
+	{ UNDERSCAN_AUTO, "auto" },
+};
+
+static struct drm_prop_enum_list amdgpu_audio_enum_list[] =
+{	{ AMDGPU_AUDIO_DISABLE, "off" },
+	{ AMDGPU_AUDIO_ENABLE, "on" },
+	{ AMDGPU_AUDIO_AUTO, "auto" },
+};
+
+/* XXX support different dither options? spatial, temporal, both, etc. */
+static struct drm_prop_enum_list amdgpu_dither_enum_list[] =
+{	{ AMDGPU_FMT_DITHER_DISABLE, "off" },
+	{ AMDGPU_FMT_DITHER_ENABLE, "on" },
+};
+
+int amdgpu_modeset_create_props(struct amdgpu_device *adev)
+{
+	int sz;
+
+	if (adev->is_atom_bios) {
+		adev->mode_info.coherent_mode_property =
+			drm_property_create_range(adev->ddev, 0 , "coherent", 0, 1);
+		if (!adev->mode_info.coherent_mode_property)
+			return -ENOMEM;
+	}
+
+	adev->mode_info.load_detect_property =
+		drm_property_create_range(adev->ddev, 0, "load detection", 0, 1);
+	if (!adev->mode_info.load_detect_property)
+		return -ENOMEM;
+
+	drm_mode_create_scaling_mode_property(adev->ddev);
+
+	sz = ARRAY_SIZE(amdgpu_underscan_enum_list);
+	adev->mode_info.underscan_property =
+		drm_property_create_enum(adev->ddev, 0,
+				    "underscan",
+				    amdgpu_underscan_enum_list, sz);
+
+	adev->mode_info.underscan_hborder_property =
+		drm_property_create_range(adev->ddev, 0,
+					"underscan hborder", 0, 128);
+	if (!adev->mode_info.underscan_hborder_property)
+		return -ENOMEM;
+
+	adev->mode_info.underscan_vborder_property =
+		drm_property_create_range(adev->ddev, 0,
+					"underscan vborder", 0, 128);
+	if (!adev->mode_info.underscan_vborder_property)
+		return -ENOMEM;
+
+	sz = ARRAY_SIZE(amdgpu_audio_enum_list);
+	adev->mode_info.audio_property =
+		drm_property_create_enum(adev->ddev, 0,
+					 "audio",
+					 amdgpu_audio_enum_list, sz);
+
+	sz = ARRAY_SIZE(amdgpu_dither_enum_list);
+	adev->mode_info.dither_property =
+		drm_property_create_enum(adev->ddev, 0,
+					 "dither",
+					 amdgpu_dither_enum_list, sz);
+
+	return 0;
+}
+
+void amdgpu_update_display_priority(struct amdgpu_device *adev)
+{
+	/* adjustment options for the display watermarks */
+	if ((amdgpu_disp_priority == 0) || (amdgpu_disp_priority > 2))
+		adev->mode_info.disp_priority = 0;
+	else
+		adev->mode_info.disp_priority = amdgpu_disp_priority;
+
+}
+
+static bool is_hdtv_mode(const struct drm_display_mode *mode)
+{
+	/* try and guess if this is a tv or a monitor */
+	if ((mode->vdisplay == 480 && mode->hdisplay == 720) || /* 480p */
+	    (mode->vdisplay == 576) || /* 576p */
+	    (mode->vdisplay == 720) || /* 720p */
+	    (mode->vdisplay == 1080)) /* 1080p */
+		return true;
+	else
+		return false;
+}
+
+bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
+				    const struct drm_display_mode *mode,
+				    struct drm_display_mode *adjusted_mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_encoder *amdgpu_encoder;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	u32 src_v = 1, dst_v = 1;
+	u32 src_h = 1, dst_h = 1;
+
+	amdgpu_crtc->h_border = 0;
+	amdgpu_crtc->v_border = 0;
+
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc != crtc)
+			continue;
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		connector = amdgpu_get_connector_for_encoder(encoder);
+		amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* set scaling */
+		if (amdgpu_encoder->rmx_type == RMX_OFF)
+			amdgpu_crtc->rmx_type = RMX_OFF;
+		else if (mode->hdisplay < amdgpu_encoder->native_mode.hdisplay ||
+			 mode->vdisplay < amdgpu_encoder->native_mode.vdisplay)
+			amdgpu_crtc->rmx_type = amdgpu_encoder->rmx_type;
+		else
+			amdgpu_crtc->rmx_type = RMX_OFF;
+		/* copy native mode */
+		memcpy(&amdgpu_crtc->native_mode,
+		       &amdgpu_encoder->native_mode,
+		       sizeof(struct drm_display_mode));
+		src_v = crtc->mode.vdisplay;
+		dst_v = amdgpu_crtc->native_mode.vdisplay;
+		src_h = crtc->mode.hdisplay;
+		dst_h = amdgpu_crtc->native_mode.hdisplay;
+
+		/* fix up for overscan on hdmi */
+		if ((!(mode->flags & DRM_MODE_FLAG_INTERLACE)) &&
+		    ((amdgpu_encoder->underscan_type == UNDERSCAN_ON) ||
+		     ((amdgpu_encoder->underscan_type == UNDERSCAN_AUTO) &&
+		      drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+		      is_hdtv_mode(mode)))) {
+			if (amdgpu_encoder->underscan_hborder != 0)
+				amdgpu_crtc->h_border = amdgpu_encoder->underscan_hborder;
+			else
+				amdgpu_crtc->h_border = (mode->hdisplay >> 5) + 16;
+			if (amdgpu_encoder->underscan_vborder != 0)
+				amdgpu_crtc->v_border = amdgpu_encoder->underscan_vborder;
+			else
+				amdgpu_crtc->v_border = (mode->vdisplay >> 5) + 16;
+			amdgpu_crtc->rmx_type = RMX_FULL;
+			src_v = crtc->mode.vdisplay;
+			dst_v = crtc->mode.vdisplay - (amdgpu_crtc->v_border * 2);
+			src_h = crtc->mode.hdisplay;
+			dst_h = crtc->mode.hdisplay - (amdgpu_crtc->h_border * 2);
+		}
+	}
+	if (amdgpu_crtc->rmx_type != RMX_OFF) {
+		fixed20_12 a, b;
+		a.full = dfixed_const(src_v);
+		b.full = dfixed_const(dst_v);
+		amdgpu_crtc->vsc.full = dfixed_div(a, b);
+		a.full = dfixed_const(src_h);
+		b.full = dfixed_const(dst_h);
+		amdgpu_crtc->hsc.full = dfixed_div(a, b);
+	} else {
+		amdgpu_crtc->vsc.full = dfixed_const(1);
+		amdgpu_crtc->hsc.full = dfixed_const(1);
+	}
+	return true;
+}
+
+/*
+ * Retrieve current video scanout position of crtc on a given gpu, and
+ * an optional accurate timestamp of when query happened.
+ *
+ * \param dev Device to query.
+ * \param pipe Crtc to query.
+ * \param flags Flags from caller (DRM_CALLED_FROM_VBLIRQ or 0).
+ *              For driver internal use only also supports these flags:
+ *
+ *              USE_REAL_VBLANKSTART to use the real start of vblank instead
+ *              of a fudged earlier start of vblank.
+ *
+ *              GET_DISTANCE_TO_VBLANKSTART to return distance to the
+ *              fudged earlier start of vblank in *vpos and the distance
+ *              to true start of vblank in *hpos.
+ *
+ * \param *vpos Location where vertical scanout position should be stored.
+ * \param *hpos Location where horizontal scanout position should go.
+ * \param *stime Target location for timestamp taken immediately before
+ *               scanout position query. Can be NULL to skip timestamp.
+ * \param *etime Target location for timestamp taken immediately after
+ *               scanout position query. Can be NULL to skip timestamp.
+ *
+ * Returns vpos as a positive number while in active scanout area.
+ * Returns vpos as a negative number inside vblank, counting the number
+ * of scanlines to go until end of vblank, e.g., -1 means "one scanline
+ * until start of active scanout / end of vblank."
+ *
+ * \return Flags, or'ed together as follows:
+ *
+ * DRM_SCANOUTPOS_VALID = Query successful.
+ * DRM_SCANOUTPOS_INVBL = Inside vblank.
+ * DRM_SCANOUTPOS_ACCURATE = Returned position is accurate. A lack of
+ * this flag means that returned position may be offset by a constant but
+ * unknown small number of scanlines wrt. real scanout position.
+ *
+ */
+int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
+			       unsigned int flags, int *vpos, int *hpos,
+			       ktime_t *stime, ktime_t *etime,
+			       const struct drm_display_mode *mode)
+{
+	u32 vbl = 0, position = 0;
+	int vbl_start, vbl_end, vtotal, ret = 0;
+	bool in_vbl = true;
+
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* preempt_disable_rt() should go right here in PREEMPT_RT patchset. */
+
+	/* Get optional system timestamp before query. */
+	if (stime)
+		*stime = ktime_get();
+
+	if (amdgpu_display_page_flip_get_scanoutpos(adev, pipe, &vbl, &position) == 0)
+		ret |= DRM_SCANOUTPOS_VALID;
+
+	/* Get optional system timestamp after query. */
+	if (etime)
+		*etime = ktime_get();
+
+	/* preempt_enable_rt() should go right here in PREEMPT_RT patchset. */
+
+	/* Decode into vertical and horizontal scanout position. */
+	*vpos = position & 0x1fff;
+	*hpos = (position >> 16) & 0x1fff;
+
+	/* Valid vblank area boundaries from gpu retrieved? */
+	if (vbl > 0) {
+		/* Yes: Decode. */
+		ret |= DRM_SCANOUTPOS_ACCURATE;
+		vbl_start = vbl & 0x1fff;
+		vbl_end = (vbl >> 16) & 0x1fff;
+	}
+	else {
+		/* No: Fake something reasonable which gives at least ok results. */
+		vbl_start = mode->crtc_vdisplay;
+		vbl_end = 0;
+	}
+
+	/* Called from driver internal vblank counter query code? */
+	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
+	    /* Caller wants distance from real vbl_start in *hpos */
+	    *hpos = *vpos - vbl_start;
+	}
+
+	/* Fudge vblank to start a few scanlines earlier to handle the
+	 * problem that vblank irqs fire a few scanlines before start
+	 * of vblank. Some driver internal callers need the true vblank
+	 * start to be used and signal this via the USE_REAL_VBLANKSTART flag.
+	 *
+	 * The cause of the "early" vblank irq is that the irq is triggered
+	 * by the line buffer logic when the line buffer read position enters
+	 * the vblank, whereas our crtc scanout position naturally lags the
+	 * line buffer read position.
+	 */
+	if (!(flags & USE_REAL_VBLANKSTART))
+		vbl_start -= adev->mode_info.crtcs[pipe]->lb_vblank_lead_lines;
+
+	/* Test scanout position against vblank region. */
+	if ((*vpos < vbl_start) && (*vpos >= vbl_end))
+		in_vbl = false;
+
+	/* In vblank? */
+	if (in_vbl)
+	    ret |= DRM_SCANOUTPOS_IN_VBLANK;
+
+	/* Called from driver internal vblank counter query code? */
+	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
+		/* Caller wants distance from fudged earlier vbl_start */
+		*vpos -= vbl_start;
+		return ret;
+	}
+
+	/* Check if inside vblank area and apply corrective offsets:
+	 * vpos will then be >=0 in video scanout area, but negative
+	 * within vblank area, counting down the number of lines until
+	 * start of scanout.
+	 */
+
+	/* Inside "upper part" of vblank area? Apply corrective offset if so: */
+	if (in_vbl && (*vpos >= vbl_start)) {
+		vtotal = mode->crtc_vtotal;
+		*vpos = *vpos - vtotal;
+	}
+
+	/* Correct for shifted end of vbl at vbl_end. */
+	*vpos = *vpos - vbl_end;
+
+	return ret;
+}
+
+int amdgpu_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc < 0 || crtc >= adev->mode_info.num_crtc)
+		return AMDGPU_CRTC_IRQ_NONE;
+
+	switch (crtc) {
+	case 0:
+		return AMDGPU_CRTC_IRQ_VBLANK1;
+	case 1:
+		return AMDGPU_CRTC_IRQ_VBLANK2;
+	case 2:
+		return AMDGPU_CRTC_IRQ_VBLANK3;
+	case 3:
+		return AMDGPU_CRTC_IRQ_VBLANK4;
+	case 4:
+		return AMDGPU_CRTC_IRQ_VBLANK5;
+	case 5:
+		return AMDGPU_CRTC_IRQ_VBLANK6;
+	default:
+		return AMDGPU_CRTC_IRQ_NONE;
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c
new file mode 100644
index 000000000..7b7f4aba6
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c
@@ -0,0 +1,955 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_i2c.h"
+#include "amdgpu_dpm.h"
+#include "atom.h"
+
+void amdgpu_dpm_print_class_info(u32 class, u32 class2)
+{
+	printk("\tui class: ");
+	switch (class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) {
+	case ATOM_PPLIB_CLASSIFICATION_UI_NONE:
+	default:
+		printk("none\n");
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY:
+		printk("battery\n");
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_BALANCED:
+		printk("balanced\n");
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE:
+		printk("performance\n");
+		break;
+	}
+	printk("\tinternal class: ");
+	if (((class & ~ATOM_PPLIB_CLASSIFICATION_UI_MASK) == 0) &&
+	    (class2 == 0))
+		printk("none");
+	else {
+		if (class & ATOM_PPLIB_CLASSIFICATION_BOOT)
+			printk("boot ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_THERMAL)
+			printk("thermal ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_LIMITEDPOWERSOURCE)
+			printk("limited_pwr ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_REST)
+			printk("rest ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_FORCED)
+			printk("forced ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE)
+			printk("3d_perf ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_OVERDRIVETEMPLATE)
+			printk("ovrdrv ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+			printk("uvd ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_3DLOW)
+			printk("3d_low ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_ACPI)
+			printk("acpi ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+			printk("uvd_hd2 ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+			printk("uvd_hd ");
+		if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+			printk("uvd_sd ");
+		if (class2 & ATOM_PPLIB_CLASSIFICATION2_LIMITEDPOWERSOURCE_2)
+			printk("limited_pwr2 ");
+		if (class2 & ATOM_PPLIB_CLASSIFICATION2_ULV)
+			printk("ulv ");
+		if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+			printk("uvd_mvc ");
+	}
+	printk("\n");
+}
+
+void amdgpu_dpm_print_cap_info(u32 caps)
+{
+	printk("\tcaps: ");
+	if (caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY)
+		printk("single_disp ");
+	if (caps & ATOM_PPLIB_SUPPORTS_VIDEO_PLAYBACK)
+		printk("video ");
+	if (caps & ATOM_PPLIB_DISALLOW_ON_DC)
+		printk("no_dc ");
+	printk("\n");
+}
+
+void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev,
+				struct amdgpu_ps *rps)
+{
+	printk("\tstatus: ");
+	if (rps == adev->pm.dpm.current_ps)
+		printk("c ");
+	if (rps == adev->pm.dpm.requested_ps)
+		printk("r ");
+	if (rps == adev->pm.dpm.boot_ps)
+		printk("b ");
+	printk("\n");
+}
+
+u32 amdgpu_dpm_get_vblank_time(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+	u32 line_time_us, vblank_lines;
+	u32 vblank_time_us = 0xffffffff; /* if the displays are off, vblank time is max */
+
+	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
+		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+			amdgpu_crtc = to_amdgpu_crtc(crtc);
+			if (crtc->enabled && amdgpu_crtc->enabled && amdgpu_crtc->hw_mode.clock) {
+				line_time_us = (amdgpu_crtc->hw_mode.crtc_htotal * 1000) /
+					amdgpu_crtc->hw_mode.clock;
+				vblank_lines = amdgpu_crtc->hw_mode.crtc_vblank_end -
+					amdgpu_crtc->hw_mode.crtc_vdisplay +
+					(amdgpu_crtc->v_border * 2);
+				vblank_time_us = vblank_lines * line_time_us;
+				break;
+			}
+		}
+	}
+
+	return vblank_time_us;
+}
+
+u32 amdgpu_dpm_get_vrefresh(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+	u32 vrefresh = 0;
+
+	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
+		list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
+			amdgpu_crtc = to_amdgpu_crtc(crtc);
+			if (crtc->enabled && amdgpu_crtc->enabled && amdgpu_crtc->hw_mode.clock) {
+				vrefresh = amdgpu_crtc->hw_mode.vrefresh;
+				break;
+			}
+		}
+	}
+
+	return vrefresh;
+}
+
+void amdgpu_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
+			      u32 *p, u32 *u)
+{
+	u32 b_c = 0;
+	u32 i_c;
+	u32 tmp;
+
+	i_c = (i * r_c) / 100;
+	tmp = i_c >> p_b;
+
+	while (tmp) {
+		b_c++;
+		tmp >>= 1;
+	}
+
+	*u = (b_c + 1) / 2;
+	*p = i_c / (1 << (2 * (*u)));
+}
+
+int amdgpu_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th)
+{
+	u32 k, a, ah, al;
+	u32 t1;
+
+	if ((fl == 0) || (fh == 0) || (fl > fh))
+		return -EINVAL;
+
+	k = (100 * fh) / fl;
+	t1 = (t * (k - 100));
+	a = (1000 * (100 * h + t1)) / (10000 + (t1 / 100));
+	a = (a + 5) / 10;
+	ah = ((a * t) + 5000) / 10000;
+	al = a - ah;
+
+	*th = t - ah;
+	*tl = t + al;
+
+	return 0;
+}
+
+bool amdgpu_is_uvd_state(u32 class, u32 class2)
+{
+	if (class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+		return true;
+	if (class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+		return true;
+	if (class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+		return true;
+	return false;
+}
+
+bool amdgpu_is_internal_thermal_sensor(enum amdgpu_int_thermal_type sensor)
+{
+	switch (sensor) {
+	case THERMAL_TYPE_RV6XX:
+	case THERMAL_TYPE_RV770:
+	case THERMAL_TYPE_EVERGREEN:
+	case THERMAL_TYPE_SUMO:
+	case THERMAL_TYPE_NI:
+	case THERMAL_TYPE_SI:
+	case THERMAL_TYPE_CI:
+	case THERMAL_TYPE_KV:
+		return true;
+	case THERMAL_TYPE_ADT7473_WITH_INTERNAL:
+	case THERMAL_TYPE_EMC2103_WITH_INTERNAL:
+		return false; /* need special handling */
+	case THERMAL_TYPE_NONE:
+	case THERMAL_TYPE_EXTERNAL:
+	case THERMAL_TYPE_EXTERNAL_GPIO:
+	default:
+		return false;
+	}
+}
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE4 pplib4;
+	struct _ATOM_PPLIB_POWERPLAYTABLE5 pplib5;
+};
+
+union fan_info {
+	struct _ATOM_PPLIB_FANTABLE fan;
+	struct _ATOM_PPLIB_FANTABLE2 fan2;
+	struct _ATOM_PPLIB_FANTABLE3 fan3;
+};
+
+static int amdgpu_parse_clk_voltage_dep_table(struct amdgpu_clock_voltage_dependency_table *amdgpu_table,
+					      ATOM_PPLIB_Clock_Voltage_Dependency_Table *atom_table)
+{
+	u32 size = atom_table->ucNumEntries *
+		sizeof(struct amdgpu_clock_voltage_dependency_entry);
+	int i;
+	ATOM_PPLIB_Clock_Voltage_Dependency_Record *entry;
+
+	amdgpu_table->entries = kzalloc(size, GFP_KERNEL);
+	if (!amdgpu_table->entries)
+		return -ENOMEM;
+
+	entry = &atom_table->entries[0];
+	for (i = 0; i < atom_table->ucNumEntries; i++) {
+		amdgpu_table->entries[i].clk = le16_to_cpu(entry->usClockLow) |
+			(entry->ucClockHigh << 16);
+		amdgpu_table->entries[i].v = le16_to_cpu(entry->usVoltage);
+		entry = (ATOM_PPLIB_Clock_Voltage_Dependency_Record *)
+			((u8 *)entry + sizeof(ATOM_PPLIB_Clock_Voltage_Dependency_Record));
+	}
+	amdgpu_table->count = atom_table->ucNumEntries;
+
+	return 0;
+}
+
+int amdgpu_get_platform_caps(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	adev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps);
+	adev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime);
+	adev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime);
+
+	return 0;
+}
+
+/* sizeof(ATOM_PPLIB_EXTENDEDHEADER) */
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2 12
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3 14
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4 16
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5 18
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6 20
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7 22
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V8 24
+#define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V9 26
+
+int amdgpu_parse_extended_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	union power_info *power_info;
+	union fan_info *fan_info;
+	ATOM_PPLIB_Clock_Voltage_Dependency_Table *dep_table;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	int ret, i;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	/* fan table */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE3)) {
+		if (power_info->pplib3.usFanTableOffset) {
+			fan_info = (union fan_info *)(mode_info->atom_context->bios + data_offset +
+						      le16_to_cpu(power_info->pplib3.usFanTableOffset));
+			adev->pm.dpm.fan.t_hyst = fan_info->fan.ucTHyst;
+			adev->pm.dpm.fan.t_min = le16_to_cpu(fan_info->fan.usTMin);
+			adev->pm.dpm.fan.t_med = le16_to_cpu(fan_info->fan.usTMed);
+			adev->pm.dpm.fan.t_high = le16_to_cpu(fan_info->fan.usTHigh);
+			adev->pm.dpm.fan.pwm_min = le16_to_cpu(fan_info->fan.usPWMMin);
+			adev->pm.dpm.fan.pwm_med = le16_to_cpu(fan_info->fan.usPWMMed);
+			adev->pm.dpm.fan.pwm_high = le16_to_cpu(fan_info->fan.usPWMHigh);
+			if (fan_info->fan.ucFanTableFormat >= 2)
+				adev->pm.dpm.fan.t_max = le16_to_cpu(fan_info->fan2.usTMax);
+			else
+				adev->pm.dpm.fan.t_max = 10900;
+			adev->pm.dpm.fan.cycle_delay = 100000;
+			if (fan_info->fan.ucFanTableFormat >= 3) {
+				adev->pm.dpm.fan.control_mode = fan_info->fan3.ucFanControlMode;
+				adev->pm.dpm.fan.default_max_fan_pwm =
+					le16_to_cpu(fan_info->fan3.usFanPWMMax);
+				adev->pm.dpm.fan.default_fan_output_sensitivity = 4836;
+				adev->pm.dpm.fan.fan_output_sensitivity =
+					le16_to_cpu(fan_info->fan3.usFanOutputSensitivity);
+			}
+			adev->pm.dpm.fan.ucode_fan_control = true;
+		}
+	}
+
+	/* clock dependancy tables, shedding tables */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE4)) {
+		if (power_info->pplib4.usVddcDependencyOnSCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddcDependencyOnSCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usVddciDependencyOnMCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddciDependencyOnMCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usVddcDependencyOnMCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddcDependencyOnMCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usMvddDependencyOnMCLKOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usMvddDependencyOnMCLKOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(&adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
+								 dep_table);
+			if (ret) {
+				amdgpu_free_extended_power_table(adev);
+				return ret;
+			}
+		}
+		if (power_info->pplib4.usMaxClockVoltageOnDCOffset) {
+			ATOM_PPLIB_Clock_Voltage_Limit_Table *clk_v =
+				(ATOM_PPLIB_Clock_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usMaxClockVoltageOnDCOffset));
+			if (clk_v->ucNumEntries) {
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk =
+					le16_to_cpu(clk_v->entries[0].usSclkLow) |
+					(clk_v->entries[0].ucSclkHigh << 16);
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk =
+					le16_to_cpu(clk_v->entries[0].usMclkLow) |
+					(clk_v->entries[0].ucMclkHigh << 16);
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddc =
+					le16_to_cpu(clk_v->entries[0].usVddc);
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.vddci =
+					le16_to_cpu(clk_v->entries[0].usVddci);
+			}
+		}
+		if (power_info->pplib4.usVddcPhaseShedLimitsTableOffset) {
+			ATOM_PPLIB_PhaseSheddingLimits_Table *psl =
+				(ATOM_PPLIB_PhaseSheddingLimits_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib4.usVddcPhaseShedLimitsTableOffset));
+			ATOM_PPLIB_PhaseSheddingLimits_Record *entry;
+
+			adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries =
+				kzalloc(psl->ucNumEntries *
+					sizeof(struct amdgpu_phase_shedding_limits_entry),
+					GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+
+			entry = &psl->entries[0];
+			for (i = 0; i < psl->ucNumEntries; i++) {
+				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].sclk =
+					le16_to_cpu(entry->usSclkLow) | (entry->ucSclkHigh << 16);
+				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].mclk =
+					le16_to_cpu(entry->usMclkLow) | (entry->ucMclkHigh << 16);
+				adev->pm.dpm.dyn_state.phase_shedding_limits_table.entries[i].voltage =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_PhaseSheddingLimits_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_PhaseSheddingLimits_Record));
+			}
+			adev->pm.dpm.dyn_state.phase_shedding_limits_table.count =
+				psl->ucNumEntries;
+		}
+	}
+
+	/* cac data */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE5)) {
+		adev->pm.dpm.tdp_limit = le32_to_cpu(power_info->pplib5.ulTDPLimit);
+		adev->pm.dpm.near_tdp_limit = le32_to_cpu(power_info->pplib5.ulNearTDPLimit);
+		adev->pm.dpm.near_tdp_limit_adjusted = adev->pm.dpm.near_tdp_limit;
+		adev->pm.dpm.tdp_od_limit = le16_to_cpu(power_info->pplib5.usTDPODLimit);
+		if (adev->pm.dpm.tdp_od_limit)
+			adev->pm.dpm.power_control = true;
+		else
+			adev->pm.dpm.power_control = false;
+		adev->pm.dpm.tdp_adjustment = 0;
+		adev->pm.dpm.sq_ramping_threshold = le32_to_cpu(power_info->pplib5.ulSQRampingThreshold);
+		adev->pm.dpm.cac_leakage = le32_to_cpu(power_info->pplib5.ulCACLeakage);
+		adev->pm.dpm.load_line_slope = le16_to_cpu(power_info->pplib5.usLoadLineSlope);
+		if (power_info->pplib5.usCACLeakageTableOffset) {
+			ATOM_PPLIB_CAC_Leakage_Table *cac_table =
+				(ATOM_PPLIB_CAC_Leakage_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(power_info->pplib5.usCACLeakageTableOffset));
+			ATOM_PPLIB_CAC_Leakage_Record *entry;
+			u32 size = cac_table->ucNumEntries * sizeof(struct amdgpu_cac_leakage_table);
+			adev->pm.dpm.dyn_state.cac_leakage_table.entries = kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			entry = &cac_table->entries[0];
+			for (i = 0; i < cac_table->ucNumEntries; i++) {
+				if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc1 =
+						le16_to_cpu(entry->usVddc1);
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc2 =
+						le16_to_cpu(entry->usVddc2);
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc3 =
+						le16_to_cpu(entry->usVddc3);
+				} else {
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc =
+						le16_to_cpu(entry->usVddc);
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].leakage =
+						le32_to_cpu(entry->ulLeakageValue);
+				}
+				entry = (ATOM_PPLIB_CAC_Leakage_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_CAC_Leakage_Record));
+			}
+			adev->pm.dpm.dyn_state.cac_leakage_table.count = cac_table->ucNumEntries;
+		}
+	}
+
+	/* ext tables */
+	if (le16_to_cpu(power_info->pplib.usTableSize) >=
+	    sizeof(struct _ATOM_PPLIB_POWERPLAYTABLE3)) {
+		ATOM_PPLIB_EXTENDEDHEADER *ext_hdr = (ATOM_PPLIB_EXTENDEDHEADER *)
+			(mode_info->atom_context->bios + data_offset +
+			 le16_to_cpu(power_info->pplib3.usExtendendedHeaderOffset));
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2) &&
+			ext_hdr->usVCETableOffset) {
+			VCEClockInfoArray *array = (VCEClockInfoArray *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1);
+			ATOM_PPLIB_VCE_Clock_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1 +
+				 1 + array->ucNumEntries * sizeof(VCEClockInfo));
+			ATOM_PPLIB_VCE_State_Table *states =
+				(ATOM_PPLIB_VCE_State_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usVCETableOffset) + 1 +
+				 1 + (array->ucNumEntries * sizeof (VCEClockInfo)) +
+				 1 + (limits->numEntries * sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record)));
+			ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *entry;
+			ATOM_PPLIB_VCE_State_Record *state_entry;
+			VCEClockInfo *vce_clk;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_vce_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			state_entry = &states->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				vce_clk = (VCEClockInfo *)
+					((u8 *)&array->entries[0] +
+					 (entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo)));
+				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].evclk =
+					le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16);
+				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].ecclk =
+					le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16);
+				adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record));
+			}
+			for (i = 0; i < states->numEntries; i++) {
+				if (i >= AMDGPU_MAX_VCE_LEVELS)
+					break;
+				vce_clk = (VCEClockInfo *)
+					((u8 *)&array->entries[0] +
+					 (state_entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo)));
+				adev->pm.dpm.vce_states[i].evclk =
+					le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16);
+				adev->pm.dpm.vce_states[i].ecclk =
+					le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16);
+				adev->pm.dpm.vce_states[i].clk_idx =
+					state_entry->ucClockInfoIndex & 0x3f;
+				adev->pm.dpm.vce_states[i].pstate =
+					(state_entry->ucClockInfoIndex & 0xc0) >> 6;
+				state_entry = (ATOM_PPLIB_VCE_State_Record *)
+					((u8 *)state_entry + sizeof(ATOM_PPLIB_VCE_State_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3) &&
+			ext_hdr->usUVDTableOffset) {
+			UVDClockInfoArray *array = (UVDClockInfoArray *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usUVDTableOffset) + 1);
+			ATOM_PPLIB_UVD_Clock_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_UVD_Clock_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usUVDTableOffset) + 1 +
+				 1 + (array->ucNumEntries * sizeof (UVDClockInfo)));
+			ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record *entry;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_uvd_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				UVDClockInfo *uvd_clk = (UVDClockInfo *)
+					((u8 *)&array->entries[0] +
+					 (entry->ucUVDClockInfoIndex * sizeof(UVDClockInfo)));
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].vclk =
+					le16_to_cpu(uvd_clk->usVClkLow) | (uvd_clk->ucVClkHigh << 16);
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].dclk =
+					le16_to_cpu(uvd_clk->usDClkLow) | (uvd_clk->ucDClkHigh << 16);
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_UVD_Clock_Voltage_Limit_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V4) &&
+			ext_hdr->usSAMUTableOffset) {
+			ATOM_PPLIB_SAMClk_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_SAMClk_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usSAMUTableOffset) + 1);
+			ATOM_PPLIB_SAMClk_Voltage_Limit_Record *entry;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].clk =
+					le16_to_cpu(entry->usSAMClockLow) | (entry->ucSAMClockHigh << 16);
+				adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_SAMClk_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_SAMClk_Voltage_Limit_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V5) &&
+		    ext_hdr->usPPMTableOffset) {
+			ATOM_PPLIB_PPM_Table *ppm = (ATOM_PPLIB_PPM_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usPPMTableOffset));
+			adev->pm.dpm.dyn_state.ppm_table =
+				kzalloc(sizeof(struct amdgpu_ppm_table), GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.ppm_table) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.ppm_table->ppm_design = ppm->ucPpmDesign;
+			adev->pm.dpm.dyn_state.ppm_table->cpu_core_number =
+				le16_to_cpu(ppm->usCpuCoreNumber);
+			adev->pm.dpm.dyn_state.ppm_table->platform_tdp =
+				le32_to_cpu(ppm->ulPlatformTDP);
+			adev->pm.dpm.dyn_state.ppm_table->small_ac_platform_tdp =
+				le32_to_cpu(ppm->ulSmallACPlatformTDP);
+			adev->pm.dpm.dyn_state.ppm_table->platform_tdc =
+				le32_to_cpu(ppm->ulPlatformTDC);
+			adev->pm.dpm.dyn_state.ppm_table->small_ac_platform_tdc =
+				le32_to_cpu(ppm->ulSmallACPlatformTDC);
+			adev->pm.dpm.dyn_state.ppm_table->apu_tdp =
+				le32_to_cpu(ppm->ulApuTDP);
+			adev->pm.dpm.dyn_state.ppm_table->dgpu_tdp =
+				le32_to_cpu(ppm->ulDGpuTDP);
+			adev->pm.dpm.dyn_state.ppm_table->dgpu_ulv_power =
+				le32_to_cpu(ppm->ulDGpuUlvPower);
+			adev->pm.dpm.dyn_state.ppm_table->tj_max =
+				le32_to_cpu(ppm->ulTjmax);
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V6) &&
+			ext_hdr->usACPTableOffset) {
+			ATOM_PPLIB_ACPClk_Voltage_Limit_Table *limits =
+				(ATOM_PPLIB_ACPClk_Voltage_Limit_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usACPTableOffset) + 1);
+			ATOM_PPLIB_ACPClk_Voltage_Limit_Record *entry;
+			u32 size = limits->numEntries *
+				sizeof(struct amdgpu_clock_voltage_dependency_entry);
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries =
+				kzalloc(size, GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count =
+				limits->numEntries;
+			entry = &limits->entries[0];
+			for (i = 0; i < limits->numEntries; i++) {
+				adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].clk =
+					le16_to_cpu(entry->usACPClockLow) | (entry->ucACPClockHigh << 16);
+				adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].v =
+					le16_to_cpu(entry->usVoltage);
+				entry = (ATOM_PPLIB_ACPClk_Voltage_Limit_Record *)
+					((u8 *)entry + sizeof(ATOM_PPLIB_ACPClk_Voltage_Limit_Record));
+			}
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V7) &&
+			ext_hdr->usPowerTuneTableOffset) {
+			u8 rev = *(u8 *)(mode_info->atom_context->bios + data_offset +
+					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
+			ATOM_PowerTune_Table *pt;
+			adev->pm.dpm.dyn_state.cac_tdp_table =
+				kzalloc(sizeof(struct amdgpu_cac_tdp_table), GFP_KERNEL);
+			if (!adev->pm.dpm.dyn_state.cac_tdp_table) {
+				amdgpu_free_extended_power_table(adev);
+				return -ENOMEM;
+			}
+			if (rev > 0) {
+				ATOM_PPLIB_POWERTUNE_Table_V1 *ppt = (ATOM_PPLIB_POWERTUNE_Table_V1 *)
+					(mode_info->atom_context->bios + data_offset +
+					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
+				adev->pm.dpm.dyn_state.cac_tdp_table->maximum_power_delivery_limit =
+					ppt->usMaximumPowerDeliveryLimit;
+				pt = &ppt->power_tune_table;
+			} else {
+				ATOM_PPLIB_POWERTUNE_Table *ppt = (ATOM_PPLIB_POWERTUNE_Table *)
+					(mode_info->atom_context->bios + data_offset +
+					 le16_to_cpu(ext_hdr->usPowerTuneTableOffset));
+				adev->pm.dpm.dyn_state.cac_tdp_table->maximum_power_delivery_limit = 255;
+				pt = &ppt->power_tune_table;
+			}
+			adev->pm.dpm.dyn_state.cac_tdp_table->tdp = le16_to_cpu(pt->usTDP);
+			adev->pm.dpm.dyn_state.cac_tdp_table->configurable_tdp =
+				le16_to_cpu(pt->usConfigurableTDP);
+			adev->pm.dpm.dyn_state.cac_tdp_table->tdc = le16_to_cpu(pt->usTDC);
+			adev->pm.dpm.dyn_state.cac_tdp_table->battery_power_limit =
+				le16_to_cpu(pt->usBatteryPowerLimit);
+			adev->pm.dpm.dyn_state.cac_tdp_table->small_power_limit =
+				le16_to_cpu(pt->usSmallPowerLimit);
+			adev->pm.dpm.dyn_state.cac_tdp_table->low_cac_leakage =
+				le16_to_cpu(pt->usLowCACLeakage);
+			adev->pm.dpm.dyn_state.cac_tdp_table->high_cac_leakage =
+				le16_to_cpu(pt->usHighCACLeakage);
+		}
+		if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V8) &&
+				ext_hdr->usSclkVddgfxTableOffset) {
+			dep_table = (ATOM_PPLIB_Clock_Voltage_Dependency_Table *)
+				(mode_info->atom_context->bios + data_offset +
+				 le16_to_cpu(ext_hdr->usSclkVddgfxTableOffset));
+			ret = amdgpu_parse_clk_voltage_dep_table(
+					&adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk,
+					dep_table);
+			if (ret) {
+				kfree(adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk.entries);
+				return ret;
+			}
+		}
+	}
+
+	return 0;
+}
+
+void amdgpu_free_extended_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_dpm_dynamic_state *dyn_state = &adev->pm.dpm.dyn_state;
+
+	kfree(dyn_state->vddc_dependency_on_sclk.entries);
+	kfree(dyn_state->vddci_dependency_on_mclk.entries);
+	kfree(dyn_state->vddc_dependency_on_mclk.entries);
+	kfree(dyn_state->mvdd_dependency_on_mclk.entries);
+	kfree(dyn_state->cac_leakage_table.entries);
+	kfree(dyn_state->phase_shedding_limits_table.entries);
+	kfree(dyn_state->ppm_table);
+	kfree(dyn_state->cac_tdp_table);
+	kfree(dyn_state->vce_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->uvd_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->samu_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->acp_clock_voltage_dependency_table.entries);
+	kfree(dyn_state->vddgfx_dependency_on_sclk.entries);
+}
+
+static const char *pp_lib_thermal_controller_names[] = {
+	"NONE",
+	"lm63",
+	"adm1032",
+	"adm1030",
+	"max6649",
+	"lm64",
+	"f75375",
+	"RV6xx",
+	"RV770",
+	"adt7473",
+	"NONE",
+	"External GPIO",
+	"Evergreen",
+	"emc2103",
+	"Sumo",
+	"Northern Islands",
+	"Southern Islands",
+	"lm96163",
+	"Sea Islands",
+	"Kaveri/Kabini",
+};
+
+void amdgpu_add_thermal_controller(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	ATOM_PPLIB_POWERPLAYTABLE *power_table;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	ATOM_PPLIB_THERMALCONTROLLER *controller;
+	struct amdgpu_i2c_bus_rec i2c_bus;
+	u16 data_offset;
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return;
+	power_table = (ATOM_PPLIB_POWERPLAYTABLE *)
+		(mode_info->atom_context->bios + data_offset);
+	controller = &power_table->sThermalController;
+
+	/* add the i2c bus for thermal/fan chip */
+	if (controller->ucType > 0) {
+		if (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN)
+			adev->pm.no_fan = true;
+		adev->pm.fan_pulses_per_revolution =
+			controller->ucFanParameters & ATOM_PP_FANPARAMETERS_TACHOMETER_PULSES_PER_REVOLUTION_MASK;
+		if (adev->pm.fan_pulses_per_revolution) {
+			adev->pm.fan_min_rpm = controller->ucFanMinRPM;
+			adev->pm.fan_max_rpm = controller->ucFanMaxRPM;
+		}
+		if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV6xx) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_RV6XX;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV770) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_RV770;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EVERGREEN) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EVERGREEN;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SUMO) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_SUMO;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_NISLANDS) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_NI;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_SISLANDS) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_SI;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_CISLANDS) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_CI;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_KAVERI) {
+			DRM_INFO("Internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_KV;
+		} else if (controller->ucType == ATOM_PP_THERMALCONTROLLER_EXTERNAL_GPIO) {
+			DRM_INFO("External GPIO thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL_GPIO;
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_ADT7473_WITH_INTERNAL) {
+			DRM_INFO("ADT7473 with internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_ADT7473_WITH_INTERNAL;
+		} else if (controller->ucType ==
+			   ATOM_PP_THERMALCONTROLLER_EMC2103_WITH_INTERNAL) {
+			DRM_INFO("EMC2103 with internal thermal controller %s fan control\n",
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EMC2103_WITH_INTERNAL;
+		} else if (controller->ucType < ARRAY_SIZE(pp_lib_thermal_controller_names)) {
+			DRM_INFO("Possible %s thermal controller at 0x%02x %s fan control\n",
+				 pp_lib_thermal_controller_names[controller->ucType],
+				 controller->ucI2cAddress >> 1,
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+			adev->pm.int_thermal_type = THERMAL_TYPE_EXTERNAL;
+			i2c_bus = amdgpu_atombios_lookup_i2c_gpio(adev, controller->ucI2cLine);
+			adev->pm.i2c_bus = amdgpu_i2c_lookup(adev, &i2c_bus);
+			if (adev->pm.i2c_bus) {
+				struct i2c_board_info info = { };
+				const char *name = pp_lib_thermal_controller_names[controller->ucType];
+				info.addr = controller->ucI2cAddress >> 1;
+				strlcpy(info.type, name, sizeof(info.type));
+				i2c_new_device(&adev->pm.i2c_bus->adapter, &info);
+			}
+		} else {
+			DRM_INFO("Unknown thermal controller type %d at 0x%02x %s fan control\n",
+				 controller->ucType,
+				 controller->ucI2cAddress >> 1,
+				 (controller->ucFanParameters &
+				  ATOM_PP_FANPARAMETERS_NOFAN) ? "without" : "with");
+		}
+	}
+}
+
+enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
+						 u32 sys_mask,
+						 enum amdgpu_pcie_gen asic_gen,
+						 enum amdgpu_pcie_gen default_gen)
+{
+	switch (asic_gen) {
+	case AMDGPU_PCIE_GEN1:
+		return AMDGPU_PCIE_GEN1;
+	case AMDGPU_PCIE_GEN2:
+		return AMDGPU_PCIE_GEN2;
+	case AMDGPU_PCIE_GEN3:
+		return AMDGPU_PCIE_GEN3;
+	default:
+		if ((sys_mask & DRM_PCIE_SPEED_80) && (default_gen == AMDGPU_PCIE_GEN3))
+			return AMDGPU_PCIE_GEN3;
+		else if ((sys_mask & DRM_PCIE_SPEED_50) && (default_gen == AMDGPU_PCIE_GEN2))
+			return AMDGPU_PCIE_GEN2;
+		else
+			return AMDGPU_PCIE_GEN1;
+	}
+	return AMDGPU_PCIE_GEN1;
+}
+
+u16 amdgpu_get_pcie_lane_support(struct amdgpu_device *adev,
+				 u16 asic_lanes,
+				 u16 default_lanes)
+{
+	switch (asic_lanes) {
+	case 0:
+	default:
+		return default_lanes;
+	case 1:
+		return 1;
+	case 2:
+		return 2;
+	case 4:
+		return 4;
+	case 8:
+		return 8;
+	case 12:
+		return 12;
+	case 16:
+		return 16;
+	}
+}
+
+u8 amdgpu_encode_pci_lane_width(u32 lanes)
+{
+	u8 encoded_lanes[] = { 0, 1, 2, 0, 3, 0, 0, 0, 4, 0, 0, 0, 5, 0, 0, 0, 6 };
+
+	if (lanes > 16)
+		return 0;
+
+	return encoded_lanes[lanes];
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
new file mode 100644
index 000000000..3738a96c2
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
@@ -0,0 +1,85 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_DPM_H__
+#define __AMDGPU_DPM_H__
+
+#define R600_SSTU_DFLT                               0
+#define R600_SST_DFLT                                0x00C8
+
+/* XXX are these ok? */
+#define R600_TEMP_RANGE_MIN (90 * 1000)
+#define R600_TEMP_RANGE_MAX (120 * 1000)
+
+#define FDO_PWM_MODE_STATIC  1
+#define FDO_PWM_MODE_STATIC_RPM 5
+
+enum amdgpu_td {
+	AMDGPU_TD_AUTO,
+	AMDGPU_TD_UP,
+	AMDGPU_TD_DOWN,
+};
+
+enum amdgpu_display_watermark {
+	AMDGPU_DISPLAY_WATERMARK_LOW = 0,
+	AMDGPU_DISPLAY_WATERMARK_HIGH = 1,
+};
+
+enum amdgpu_display_gap
+{
+    AMDGPU_PM_DISPLAY_GAP_VBLANK_OR_WM = 0,
+    AMDGPU_PM_DISPLAY_GAP_VBLANK       = 1,
+    AMDGPU_PM_DISPLAY_GAP_WATERMARK    = 2,
+    AMDGPU_PM_DISPLAY_GAP_IGNORE       = 3,
+};
+
+void amdgpu_dpm_print_class_info(u32 class, u32 class2);
+void amdgpu_dpm_print_cap_info(u32 caps);
+void amdgpu_dpm_print_ps_status(struct amdgpu_device *adev,
+				struct amdgpu_ps *rps);
+u32 amdgpu_dpm_get_vblank_time(struct amdgpu_device *adev);
+u32 amdgpu_dpm_get_vrefresh(struct amdgpu_device *adev);
+bool amdgpu_is_uvd_state(u32 class, u32 class2);
+void amdgpu_calculate_u_and_p(u32 i, u32 r_c, u32 p_b,
+			      u32 *p, u32 *u);
+int amdgpu_calculate_at(u32 t, u32 h, u32 fh, u32 fl, u32 *tl, u32 *th);
+
+bool amdgpu_is_internal_thermal_sensor(enum amdgpu_int_thermal_type sensor);
+
+int amdgpu_get_platform_caps(struct amdgpu_device *adev);
+
+int amdgpu_parse_extended_power_table(struct amdgpu_device *adev);
+void amdgpu_free_extended_power_table(struct amdgpu_device *adev);
+
+void amdgpu_add_thermal_controller(struct amdgpu_device *adev);
+
+enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
+						 u32 sys_mask,
+						 enum amdgpu_pcie_gen asic_gen,
+						 enum amdgpu_pcie_gen default_gen);
+
+u16 amdgpu_get_pcie_lane_support(struct amdgpu_device *adev,
+				 u16 asic_lanes,
+				 u16 default_lanes);
+u8 amdgpu_encode_pci_lane_width(u32 lanes);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
new file mode 100644
index 000000000..8d6668ced
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -0,0 +1,579 @@
+/**
+ * \file amdgpu_drv.c
+ * AMD Amdgpu driver
+ *
+ * \author Gareth Hughes <gareth@valinux.com>
+ */
+
+/*
+ * Copyright 2000 VA Linux Systems, Inc., Sunnyvale, California.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_gem.h>
+#include "amdgpu_drv.h"
+
+#include <drm/drm_pciids.h>
+#include <linux/console.h>
+#include <linux/module.h>
+#include <linux/pm_runtime.h>
+#include <linux/vga_switcheroo.h>
+#include "drm_crtc_helper.h"
+
+#include "amdgpu.h"
+#include "amdgpu_irq.h"
+
+#include "amdgpu_amdkfd.h"
+
+/*
+ * KMS wrapper.
+ * - 3.0.0 - initial driver
+ * - 3.1.0 - allow reading more status registers (GRBM, SRBM, SDMA, CP)
+ */
+#define KMS_DRIVER_MAJOR	3
+#define KMS_DRIVER_MINOR	1
+#define KMS_DRIVER_PATCHLEVEL	0
+
+int amdgpu_vram_limit = 0;
+int amdgpu_gart_size = -1; /* auto */
+int amdgpu_benchmarking = 0;
+int amdgpu_testing = 0;
+int amdgpu_audio = -1;
+int amdgpu_disp_priority = 0;
+int amdgpu_hw_i2c = 0;
+int amdgpu_pcie_gen2 = -1;
+int amdgpu_msi = -1;
+int amdgpu_lockup_timeout = 0;
+int amdgpu_dpm = -1;
+int amdgpu_smc_load_fw = 1;
+int amdgpu_aspm = -1;
+int amdgpu_runtime_pm = -1;
+int amdgpu_hard_reset = 0;
+unsigned amdgpu_ip_block_mask = 0xffffffff;
+int amdgpu_bapm = -1;
+int amdgpu_deep_color = 0;
+int amdgpu_vm_size = 64;
+int amdgpu_vm_block_size = -1;
+int amdgpu_vm_fault_stop = 0;
+int amdgpu_vm_debug = 0;
+int amdgpu_exp_hw_support = 0;
+int amdgpu_enable_scheduler = 1;
+int amdgpu_sched_jobs = 16;
+int amdgpu_sched_hw_submission = 2;
+int amdgpu_enable_semaphores = 0;
+
+MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
+module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
+
+MODULE_PARM_DESC(gartsize, "Size of PCIE/IGP gart to setup in megabytes (32, 64, etc., -1 = auto)");
+module_param_named(gartsize, amdgpu_gart_size, int, 0600);
+
+MODULE_PARM_DESC(benchmark, "Run benchmark");
+module_param_named(benchmark, amdgpu_benchmarking, int, 0444);
+
+MODULE_PARM_DESC(test, "Run tests");
+module_param_named(test, amdgpu_testing, int, 0444);
+
+MODULE_PARM_DESC(audio, "Audio enable (-1 = auto, 0 = disable, 1 = enable)");
+module_param_named(audio, amdgpu_audio, int, 0444);
+
+MODULE_PARM_DESC(disp_priority, "Display Priority (0 = auto, 1 = normal, 2 = high)");
+module_param_named(disp_priority, amdgpu_disp_priority, int, 0444);
+
+MODULE_PARM_DESC(hw_i2c, "hw i2c engine enable (0 = disable)");
+module_param_named(hw_i2c, amdgpu_hw_i2c, int, 0444);
+
+MODULE_PARM_DESC(pcie_gen2, "PCIE Gen2 mode (-1 = auto, 0 = disable, 1 = enable)");
+module_param_named(pcie_gen2, amdgpu_pcie_gen2, int, 0444);
+
+MODULE_PARM_DESC(msi, "MSI support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(msi, amdgpu_msi, int, 0444);
+
+MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms (default 0 = disable)");
+module_param_named(lockup_timeout, amdgpu_lockup_timeout, int, 0444);
+
+MODULE_PARM_DESC(dpm, "DPM support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(dpm, amdgpu_dpm, int, 0444);
+
+MODULE_PARM_DESC(smc_load_fw, "SMC firmware loading(1 = enable, 0 = disable)");
+module_param_named(smc_load_fw, amdgpu_smc_load_fw, int, 0444);
+
+MODULE_PARM_DESC(aspm, "ASPM support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(aspm, amdgpu_aspm, int, 0444);
+
+MODULE_PARM_DESC(runpm, "PX runtime pm (1 = force enable, 0 = disable, -1 = PX only default)");
+module_param_named(runpm, amdgpu_runtime_pm, int, 0444);
+
+MODULE_PARM_DESC(hard_reset, "PCI config reset (1 = force enable, 0 = disable (default))");
+module_param_named(hard_reset, amdgpu_hard_reset, int, 0444);
+
+MODULE_PARM_DESC(ip_block_mask, "IP Block Mask (all blocks enabled (default))");
+module_param_named(ip_block_mask, amdgpu_ip_block_mask, uint, 0444);
+
+MODULE_PARM_DESC(bapm, "BAPM support (1 = enable, 0 = disable, -1 = auto)");
+module_param_named(bapm, amdgpu_bapm, int, 0444);
+
+MODULE_PARM_DESC(deep_color, "Deep Color support (1 = enable, 0 = disable (default))");
+module_param_named(deep_color, amdgpu_deep_color, int, 0444);
+
+MODULE_PARM_DESC(vm_size, "VM address space size in gigabytes (default 64GB)");
+module_param_named(vm_size, amdgpu_vm_size, int, 0444);
+
+MODULE_PARM_DESC(vm_block_size, "VM page table size in bits (default depending on vm_size)");
+module_param_named(vm_block_size, amdgpu_vm_block_size, int, 0444);
+
+MODULE_PARM_DESC(vm_fault_stop, "Stop on VM fault (0 = never (default), 1 = print first, 2 = always)");
+module_param_named(vm_fault_stop, amdgpu_vm_fault_stop, int, 0444);
+
+MODULE_PARM_DESC(vm_debug, "Debug VM handling (0 = disabled (default), 1 = enabled)");
+module_param_named(vm_debug, amdgpu_vm_debug, int, 0644);
+
+MODULE_PARM_DESC(exp_hw_support, "experimental hw support (1 = enable, 0 = disable (default))");
+module_param_named(exp_hw_support, amdgpu_exp_hw_support, int, 0444);
+
+MODULE_PARM_DESC(enable_scheduler, "enable SW GPU scheduler (1 = enable (default), 0 = disable)");
+module_param_named(enable_scheduler, amdgpu_enable_scheduler, int, 0444);
+
+MODULE_PARM_DESC(sched_jobs, "the max number of jobs supported in the sw queue (default 16)");
+module_param_named(sched_jobs, amdgpu_sched_jobs, int, 0444);
+
+MODULE_PARM_DESC(sched_hw_submission, "the max number of HW submissions (default 2)");
+module_param_named(sched_hw_submission, amdgpu_sched_hw_submission, int, 0444);
+
+MODULE_PARM_DESC(enable_semaphores, "Enable semaphores (1 = enable, 0 = disable (default))");
+module_param_named(enable_semaphores, amdgpu_enable_semaphores, int, 0644);
+
+static struct pci_device_id pciidlist[] = {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	/* Kaveri */
+	{0x1002, 0x1304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x1305, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1306, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x1307, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1309, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x130F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1310, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1311, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1312, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1313, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1315, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1316, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x1317, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x1318, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x131B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x131C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	{0x1002, 0x131D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_APU},
+	/* Bonaire */
+	{0x1002, 0x6640, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6641, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6646, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6647, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE|AMD_IS_MOBILITY},
+	{0x1002, 0x6649, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x6650, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x6651, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x6658, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x665c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x665d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	{0x1002, 0x665f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_BONAIRE},
+	/* Hawaii */
+	{0x1002, 0x67A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67A9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67AA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B1, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67B9, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67BA, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	{0x1002, 0x67BE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAWAII},
+	/* Kabini */
+	{0x1002, 0x9830, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9831, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9832, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9833, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9834, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9835, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9836, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9837, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x9838, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9839, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x983a, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983b, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x983c, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983d, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983e, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	{0x1002, 0x983f, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KABINI|AMD_IS_APU},
+	/* mullins */
+	{0x1002, 0x9850, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9851, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9852, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9853, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9854, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9855, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9856, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9857, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9858, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x9859, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+	{0x1002, 0x985F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_MULLINS|AMD_IS_MOBILITY|AMD_IS_APU},
+#endif
+	/* topaz */
+	{0x1002, 0x6900, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6901, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6902, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6903, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	{0x1002, 0x6907, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TOPAZ},
+	/* tonga */
+	{0x1002, 0x6920, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6921, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6928, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6929, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x692B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x692F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6930, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6938, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	{0x1002, 0x6939, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TONGA},
+	/* fiji */
+	{0x1002, 0x7300, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_FIJI},
+	/* carrizo */
+	{0x1002, 0x9870, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9874, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9875, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9876, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	{0x1002, 0x9877, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_CARRIZO|AMD_IS_APU},
+	/* stoney */
+	{0x1002, 0x98E4, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_STONEY|AMD_IS_APU},
+
+	{0, 0, 0}
+};
+
+MODULE_DEVICE_TABLE(pci, pciidlist);
+
+static struct drm_driver kms_driver;
+
+static int amdgpu_kick_out_firmware_fb(struct pci_dev *pdev)
+{
+	struct apertures_struct *ap;
+	bool primary = false;
+
+	ap = alloc_apertures(1);
+	if (!ap)
+		return -ENOMEM;
+
+	ap->ranges[0].base = pci_resource_start(pdev, 0);
+	ap->ranges[0].size = pci_resource_len(pdev, 0);
+
+#ifdef CONFIG_X86
+	primary = pdev->resource[PCI_ROM_RESOURCE].flags & IORESOURCE_ROM_SHADOW;
+#endif
+	remove_conflicting_framebuffers(ap, "amdgpudrmfb", primary);
+	kfree(ap);
+
+	return 0;
+}
+
+static int amdgpu_pci_probe(struct pci_dev *pdev,
+			    const struct pci_device_id *ent)
+{
+	unsigned long flags = ent->driver_data;
+	int ret;
+
+	if ((flags & AMD_EXP_HW_SUPPORT) && !amdgpu_exp_hw_support) {
+		DRM_INFO("This hardware requires experimental hardware support.\n"
+			 "See modparam exp_hw_support\n");
+		return -ENODEV;
+	}
+
+	/* Get rid of things like offb */
+	ret = amdgpu_kick_out_firmware_fb(pdev);
+	if (ret)
+		return ret;
+
+	return drm_get_pci_dev(pdev, ent, &kms_driver);
+}
+
+static void
+amdgpu_pci_remove(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+
+	drm_put_dev(dev);
+}
+
+static int amdgpu_pmops_suspend(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_suspend_kms(drm_dev, true, true);
+}
+
+static int amdgpu_pmops_resume(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_resume_kms(drm_dev, true, true);
+}
+
+static int amdgpu_pmops_freeze(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_suspend_kms(drm_dev, false, true);
+}
+
+static int amdgpu_pmops_thaw(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_resume_kms(drm_dev, false, true);
+}
+
+static int amdgpu_pmops_runtime_suspend(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	int ret;
+
+	if (!amdgpu_device_is_px(drm_dev)) {
+		pm_runtime_forbid(dev);
+		return -EBUSY;
+	}
+
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+	drm_kms_helper_poll_disable(drm_dev);
+	vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_OFF);
+
+	ret = amdgpu_suspend_kms(drm_dev, false, false);
+	pci_save_state(pdev);
+	pci_disable_device(pdev);
+	pci_ignore_hotplug(pdev);
+	pci_set_power_state(pdev, PCI_D3cold);
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF;
+
+	return 0;
+}
+
+static int amdgpu_pmops_runtime_resume(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	int ret;
+
+	if (!amdgpu_device_is_px(drm_dev))
+		return -EINVAL;
+
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+
+	pci_set_power_state(pdev, PCI_D0);
+	pci_restore_state(pdev);
+	ret = pci_enable_device(pdev);
+	if (ret)
+		return ret;
+	pci_set_master(pdev);
+
+	ret = amdgpu_resume_kms(drm_dev, false, false);
+	drm_kms_helper_poll_enable(drm_dev);
+	vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_ON);
+	drm_dev->switch_power_state = DRM_SWITCH_POWER_ON;
+	return 0;
+}
+
+static int amdgpu_pmops_runtime_idle(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	struct drm_crtc *crtc;
+
+	if (!amdgpu_device_is_px(drm_dev)) {
+		pm_runtime_forbid(dev);
+		return -EBUSY;
+	}
+
+	list_for_each_entry(crtc, &drm_dev->mode_config.crtc_list, head) {
+		if (crtc->enabled) {
+			DRM_DEBUG_DRIVER("failing to power off - crtc active\n");
+			return -EBUSY;
+		}
+	}
+
+	pm_runtime_mark_last_busy(dev);
+	pm_runtime_autosuspend(dev);
+	/* we don't want the main rpm_idle to call suspend - we want to autosuspend */
+	return 1;
+}
+
+long amdgpu_drm_ioctl(struct file *filp,
+		      unsigned int cmd, unsigned long arg)
+{
+	struct drm_file *file_priv = filp->private_data;
+	struct drm_device *dev;
+	long ret;
+	dev = file_priv->minor->dev;
+	ret = pm_runtime_get_sync(dev->dev);
+	if (ret < 0)
+		return ret;
+
+	ret = drm_ioctl(filp, cmd, arg);
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+	return ret;
+}
+
+static const struct dev_pm_ops amdgpu_pm_ops = {
+	.suspend = amdgpu_pmops_suspend,
+	.resume = amdgpu_pmops_resume,
+	.freeze = amdgpu_pmops_freeze,
+	.thaw = amdgpu_pmops_thaw,
+	.poweroff = amdgpu_pmops_freeze,
+	.restore = amdgpu_pmops_resume,
+	.runtime_suspend = amdgpu_pmops_runtime_suspend,
+	.runtime_resume = amdgpu_pmops_runtime_resume,
+	.runtime_idle = amdgpu_pmops_runtime_idle,
+};
+
+static const struct file_operations amdgpu_driver_kms_fops = {
+	.owner = THIS_MODULE,
+	.open = drm_open,
+	.release = drm_release,
+	.unlocked_ioctl = amdgpu_drm_ioctl,
+	.mmap = amdgpu_mmap,
+	.poll = drm_poll,
+	.read = drm_read,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl = amdgpu_kms_compat_ioctl,
+#endif
+};
+
+static struct drm_driver kms_driver = {
+	.driver_features =
+	    DRIVER_USE_AGP |
+	    DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED | DRIVER_GEM |
+	    DRIVER_PRIME | DRIVER_RENDER,
+	.dev_priv_size = 0,
+	.load = amdgpu_driver_load_kms,
+	.open = amdgpu_driver_open_kms,
+	.preclose = amdgpu_driver_preclose_kms,
+	.postclose = amdgpu_driver_postclose_kms,
+	.lastclose = amdgpu_driver_lastclose_kms,
+	.set_busid = drm_pci_set_busid,
+	.unload = amdgpu_driver_unload_kms,
+	.get_vblank_counter = amdgpu_get_vblank_counter_kms,
+	.enable_vblank = amdgpu_enable_vblank_kms,
+	.disable_vblank = amdgpu_disable_vblank_kms,
+	.get_vblank_timestamp = amdgpu_get_vblank_timestamp_kms,
+	.get_scanout_position = amdgpu_get_crtc_scanoutpos,
+#if defined(CONFIG_DEBUG_FS)
+	.debugfs_init = amdgpu_debugfs_init,
+	.debugfs_cleanup = amdgpu_debugfs_cleanup,
+#endif
+	.irq_preinstall = amdgpu_irq_preinstall,
+	.irq_postinstall = amdgpu_irq_postinstall,
+	.irq_uninstall = amdgpu_irq_uninstall,
+	.irq_handler = amdgpu_irq_handler,
+	.ioctls = amdgpu_ioctls_kms,
+	.gem_free_object = amdgpu_gem_object_free,
+	.gem_open_object = amdgpu_gem_object_open,
+	.gem_close_object = amdgpu_gem_object_close,
+	.dumb_create = amdgpu_mode_dumb_create,
+	.dumb_map_offset = amdgpu_mode_dumb_mmap,
+	.dumb_destroy = drm_gem_dumb_destroy,
+	.fops = &amdgpu_driver_kms_fops,
+
+	.prime_handle_to_fd = drm_gem_prime_handle_to_fd,
+	.prime_fd_to_handle = drm_gem_prime_fd_to_handle,
+	.gem_prime_export = amdgpu_gem_prime_export,
+	.gem_prime_import = drm_gem_prime_import,
+	.gem_prime_pin = amdgpu_gem_prime_pin,
+	.gem_prime_unpin = amdgpu_gem_prime_unpin,
+	.gem_prime_res_obj = amdgpu_gem_prime_res_obj,
+	.gem_prime_get_sg_table = amdgpu_gem_prime_get_sg_table,
+	.gem_prime_import_sg_table = amdgpu_gem_prime_import_sg_table,
+	.gem_prime_vmap = amdgpu_gem_prime_vmap,
+	.gem_prime_vunmap = amdgpu_gem_prime_vunmap,
+
+	.name = DRIVER_NAME,
+	.desc = DRIVER_DESC,
+	.date = DRIVER_DATE,
+	.major = KMS_DRIVER_MAJOR,
+	.minor = KMS_DRIVER_MINOR,
+	.patchlevel = KMS_DRIVER_PATCHLEVEL,
+};
+
+static struct drm_driver *driver;
+static struct pci_driver *pdriver;
+
+static struct pci_driver amdgpu_kms_pci_driver = {
+	.name = DRIVER_NAME,
+	.id_table = pciidlist,
+	.probe = amdgpu_pci_probe,
+	.remove = amdgpu_pci_remove,
+	.driver.pm = &amdgpu_pm_ops,
+};
+
+static int __init amdgpu_init(void)
+{
+#ifdef CONFIG_VGA_CONSOLE
+	if (vgacon_text_force()) {
+		DRM_ERROR("VGACON disables amdgpu kernel modesetting.\n");
+		return -EINVAL;
+	}
+#endif
+	DRM_INFO("amdgpu kernel modesetting enabled.\n");
+	driver = &kms_driver;
+	pdriver = &amdgpu_kms_pci_driver;
+	driver->driver_features |= DRIVER_MODESET;
+	driver->num_ioctls = amdgpu_max_kms_ioctl;
+	amdgpu_register_atpx_handler();
+
+	amdgpu_amdkfd_init();
+
+	/* let modprobe override vga console setting */
+	return drm_pci_init(driver, pdriver);
+}
+
+static void __exit amdgpu_exit(void)
+{
+	amdgpu_amdkfd_fini();
+	drm_pci_exit(driver, pdriver);
+	amdgpu_unregister_atpx_handler();
+}
+
+module_init(amdgpu_init);
+module_exit(amdgpu_exit);
+
+MODULE_AUTHOR(DRIVER_AUTHOR);
+MODULE_DESCRIPTION(DRIVER_DESC);
+MODULE_LICENSE("GPL and additional rights");
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h
new file mode 100644
index 000000000..e3a4f7048
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.h
@@ -0,0 +1,48 @@
+/* amdgpu_drv.h -- Private header for amdgpu driver -*- linux-c -*-
+ *
+ * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas.
+ * Copyright 2000 VA Linux Systems, Inc., Fremont, California.
+ * All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_DRV_H__
+#define __AMDGPU_DRV_H__
+
+#include <linux/firmware.h>
+#include <linux/platform_device.h>
+
+#include "amd_shared.h"
+
+/* General customization:
+ */
+
+#define DRIVER_AUTHOR		"AMD linux driver team"
+
+#define DRIVER_NAME		"amdgpu"
+#define DRIVER_DESC		"AMD GPU"
+#define DRIVER_DATE		"20150101"
+
+long amdgpu_drm_ioctl(struct file *filp,
+		      unsigned int cmd, unsigned long arg);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
new file mode 100644
index 000000000..94138abe0
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_encoders.c
@@ -0,0 +1,245 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_connectors.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+
+void
+amdgpu_link_encoder_connector(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* walk the list and link encoders to connectors */
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+			amdgpu_encoder = to_amdgpu_encoder(encoder);
+			if (amdgpu_encoder->devices & amdgpu_connector->devices) {
+				drm_mode_connector_attach_encoder(connector, encoder);
+				if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+					amdgpu_atombios_encoder_init_backlight(amdgpu_encoder, connector);
+					adev->mode_info.bl_encoder = amdgpu_encoder;
+				}
+			}
+		}
+	}
+}
+
+void amdgpu_encoder_set_active_device(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+			amdgpu_encoder->active_device = amdgpu_encoder->devices & amdgpu_connector->devices;
+			DRM_DEBUG_KMS("setting active device to %08x from %08x %08x for encoder %d\n",
+				  amdgpu_encoder->active_device, amdgpu_encoder->devices,
+				  amdgpu_connector->devices, encoder->encoder_type);
+		}
+	}
+}
+
+struct drm_connector *
+amdgpu_get_connector_for_encoder(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		if (amdgpu_encoder->active_device & amdgpu_connector->devices)
+			return connector;
+	}
+	return NULL;
+}
+
+struct drm_connector *
+amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		if (amdgpu_encoder->devices & amdgpu_connector->devices)
+			return connector;
+	}
+	return NULL;
+}
+
+struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_encoder *other_encoder;
+	struct amdgpu_encoder *other_amdgpu_encoder;
+
+	if (amdgpu_encoder->is_ext_encoder)
+		return NULL;
+
+	list_for_each_entry(other_encoder, &dev->mode_config.encoder_list, head) {
+		if (other_encoder == encoder)
+			continue;
+		other_amdgpu_encoder = to_amdgpu_encoder(other_encoder);
+		if (other_amdgpu_encoder->is_ext_encoder &&
+		    (amdgpu_encoder->devices & other_amdgpu_encoder->devices))
+			return other_encoder;
+	}
+	return NULL;
+}
+
+u16 amdgpu_encoder_get_dp_bridge_encoder_id(struct drm_encoder *encoder)
+{
+	struct drm_encoder *other_encoder = amdgpu_get_external_encoder(encoder);
+
+	if (other_encoder) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(other_encoder);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_TRAVIS:
+		case ENCODER_OBJECT_ID_NUTMEG:
+			return amdgpu_encoder->encoder_id;
+		default:
+			return ENCODER_OBJECT_ID_NONE;
+		}
+	}
+	return ENCODER_OBJECT_ID_NONE;
+}
+
+void amdgpu_panel_mode_fixup(struct drm_encoder *encoder,
+			     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_display_mode *native_mode = &amdgpu_encoder->native_mode;
+	unsigned hblank = native_mode->htotal - native_mode->hdisplay;
+	unsigned vblank = native_mode->vtotal - native_mode->vdisplay;
+	unsigned hover = native_mode->hsync_start - native_mode->hdisplay;
+	unsigned vover = native_mode->vsync_start - native_mode->vdisplay;
+	unsigned hsync_width = native_mode->hsync_end - native_mode->hsync_start;
+	unsigned vsync_width = native_mode->vsync_end - native_mode->vsync_start;
+
+	adjusted_mode->clock = native_mode->clock;
+	adjusted_mode->flags = native_mode->flags;
+
+	adjusted_mode->hdisplay = native_mode->hdisplay;
+	adjusted_mode->vdisplay = native_mode->vdisplay;
+
+	adjusted_mode->htotal = native_mode->hdisplay + hblank;
+	adjusted_mode->hsync_start = native_mode->hdisplay + hover;
+	adjusted_mode->hsync_end = adjusted_mode->hsync_start + hsync_width;
+
+	adjusted_mode->vtotal = native_mode->vdisplay + vblank;
+	adjusted_mode->vsync_start = native_mode->vdisplay + vover;
+	adjusted_mode->vsync_end = adjusted_mode->vsync_start + vsync_width;
+
+	drm_mode_set_crtcinfo(adjusted_mode, CRTC_INTERLACE_HALVE_V);
+
+	adjusted_mode->crtc_hdisplay = native_mode->hdisplay;
+	adjusted_mode->crtc_vdisplay = native_mode->vdisplay;
+
+	adjusted_mode->crtc_htotal = adjusted_mode->crtc_hdisplay + hblank;
+	adjusted_mode->crtc_hsync_start = adjusted_mode->crtc_hdisplay + hover;
+	adjusted_mode->crtc_hsync_end = adjusted_mode->crtc_hsync_start + hsync_width;
+
+	adjusted_mode->crtc_vtotal = adjusted_mode->crtc_vdisplay + vblank;
+	adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + vover;
+	adjusted_mode->crtc_vsync_end = adjusted_mode->crtc_vsync_start + vsync_width;
+
+}
+
+bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
+				    u32 pixel_clock)
+{
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	connector = amdgpu_get_connector_for_encoder(encoder);
+	/* if we don't have an active device yet, just use one of
+	 * the connectors tied to the encoder.
+	 */
+	if (!connector)
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+	amdgpu_connector = to_amdgpu_connector(connector);
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_DVII:
+	case DRM_MODE_CONNECTOR_HDMIB:
+		if (amdgpu_connector->use_digital) {
+			/* HDMI 1.3 supports up to 340 Mhz over single link */
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				if (pixel_clock > 340000)
+					return true;
+				else
+					return false;
+			} else {
+				if (pixel_clock > 165000)
+					return true;
+				else
+					return false;
+			}
+		} else
+			return false;
+	case DRM_MODE_CONNECTOR_DVID:
+	case DRM_MODE_CONNECTOR_HDMIA:
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		dig_connector = amdgpu_connector->con_priv;
+		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP))
+			return false;
+		else {
+			/* HDMI 1.3 supports up to 340 Mhz over single link */
+			if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector))) {
+				if (pixel_clock > 340000)
+					return true;
+				else
+					return false;
+			} else {
+				if (pixel_clock > 165000)
+					return true;
+				else
+					return false;
+			}
+		}
+	default:
+		return false;
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
new file mode 100644
index 000000000..093a8c618
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
@@ -0,0 +1,421 @@
+/*
+ * Copyright © 2007 David Airlie
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *     David Airlie
+ */
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/fb.h>
+
+#include <drm/drmP.h>
+#include <drm/drm_crtc.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "cikd.h"
+
+#include <drm/drm_fb_helper.h>
+
+#include <linux/vga_switcheroo.h>
+
+/* object hierarchy -
+   this contains a helper + a amdgpu fb
+   the helper contains a pointer to amdgpu framebuffer baseclass.
+*/
+struct amdgpu_fbdev {
+	struct drm_fb_helper helper;
+	struct amdgpu_framebuffer rfb;
+	struct list_head fbdev_list;
+	struct amdgpu_device *adev;
+};
+
+static struct fb_ops amdgpufb_ops = {
+	.owner = THIS_MODULE,
+	.fb_check_var = drm_fb_helper_check_var,
+	.fb_set_par = drm_fb_helper_set_par,
+	.fb_fillrect = drm_fb_helper_cfb_fillrect,
+	.fb_copyarea = drm_fb_helper_cfb_copyarea,
+	.fb_imageblit = drm_fb_helper_cfb_imageblit,
+	.fb_pan_display = drm_fb_helper_pan_display,
+	.fb_blank = drm_fb_helper_blank,
+	.fb_setcmap = drm_fb_helper_setcmap,
+	.fb_debug_enter = drm_fb_helper_debug_enter,
+	.fb_debug_leave = drm_fb_helper_debug_leave,
+};
+
+
+int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int bpp, bool tiled)
+{
+	int aligned = width;
+	int pitch_mask = 0;
+
+	switch (bpp / 8) {
+	case 1:
+		pitch_mask = 255;
+		break;
+	case 2:
+		pitch_mask = 127;
+		break;
+	case 3:
+	case 4:
+		pitch_mask = 63;
+		break;
+	}
+
+	aligned += pitch_mask;
+	aligned &= ~pitch_mask;
+	return aligned;
+}
+
+static void amdgpufb_destroy_pinned_object(struct drm_gem_object *gobj)
+{
+	struct amdgpu_bo *rbo = gem_to_amdgpu_bo(gobj);
+	int ret;
+
+	ret = amdgpu_bo_reserve(rbo, false);
+	if (likely(ret == 0)) {
+		amdgpu_bo_kunmap(rbo);
+		amdgpu_bo_unpin(rbo);
+		amdgpu_bo_unreserve(rbo);
+	}
+	drm_gem_object_unreference_unlocked(gobj);
+}
+
+static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
+					 struct drm_mode_fb_cmd2 *mode_cmd,
+					 struct drm_gem_object **gobj_p)
+{
+	struct amdgpu_device *adev = rfbdev->adev;
+	struct drm_gem_object *gobj = NULL;
+	struct amdgpu_bo *rbo = NULL;
+	bool fb_tiled = false; /* useful for testing */
+	u32 tiling_flags = 0;
+	int ret;
+	int aligned_size, size;
+	int height = mode_cmd->height;
+	u32 bpp, depth;
+
+	drm_fb_get_bpp_depth(mode_cmd->pixel_format, &depth, &bpp);
+
+	/* need to align pitch with crtc limits */
+	mode_cmd->pitches[0] = amdgpu_align_pitch(adev, mode_cmd->width, bpp,
+						  fb_tiled) * ((bpp + 1) / 8);
+
+	height = ALIGN(mode_cmd->height, 8);
+	size = mode_cmd->pitches[0] * height;
+	aligned_size = ALIGN(size, PAGE_SIZE);
+	ret = amdgpu_gem_object_create(adev, aligned_size, 0,
+				       AMDGPU_GEM_DOMAIN_VRAM,
+				       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+				       true, &gobj);
+	if (ret) {
+		printk(KERN_ERR "failed to allocate framebuffer (%d)\n",
+		       aligned_size);
+		return -ENOMEM;
+	}
+	rbo = gem_to_amdgpu_bo(gobj);
+
+	if (fb_tiled)
+		tiling_flags = AMDGPU_TILING_SET(ARRAY_MODE, GRPH_ARRAY_2D_TILED_THIN1);
+
+	ret = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(ret != 0))
+		goto out_unref;
+
+	if (tiling_flags) {
+		ret = amdgpu_bo_set_tiling_flags(rbo,
+						 tiling_flags);
+		if (ret)
+			dev_err(adev->dev, "FB failed to set tiling flags\n");
+	}
+
+
+	ret = amdgpu_bo_pin_restricted(rbo, AMDGPU_GEM_DOMAIN_VRAM, 0, 0, NULL);
+	if (ret) {
+		amdgpu_bo_unreserve(rbo);
+		goto out_unref;
+	}
+	ret = amdgpu_bo_kmap(rbo, NULL);
+	amdgpu_bo_unreserve(rbo);
+	if (ret) {
+		goto out_unref;
+	}
+
+	*gobj_p = gobj;
+	return 0;
+out_unref:
+	amdgpufb_destroy_pinned_object(gobj);
+	*gobj_p = NULL;
+	return ret;
+}
+
+static int amdgpufb_create(struct drm_fb_helper *helper,
+			   struct drm_fb_helper_surface_size *sizes)
+{
+	struct amdgpu_fbdev *rfbdev = (struct amdgpu_fbdev *)helper;
+	struct amdgpu_device *adev = rfbdev->adev;
+	struct fb_info *info;
+	struct drm_framebuffer *fb = NULL;
+	struct drm_mode_fb_cmd2 mode_cmd;
+	struct drm_gem_object *gobj = NULL;
+	struct amdgpu_bo *rbo = NULL;
+	int ret;
+	unsigned long tmp;
+
+	mode_cmd.width = sizes->surface_width;
+	mode_cmd.height = sizes->surface_height;
+
+	if (sizes->surface_bpp == 24)
+		sizes->surface_bpp = 32;
+
+	mode_cmd.pixel_format = drm_mode_legacy_fb_format(sizes->surface_bpp,
+							  sizes->surface_depth);
+
+	ret = amdgpufb_create_pinned_object(rfbdev, &mode_cmd, &gobj);
+	if (ret) {
+		DRM_ERROR("failed to create fbcon object %d\n", ret);
+		return ret;
+	}
+
+	rbo = gem_to_amdgpu_bo(gobj);
+
+	/* okay we have an object now allocate the framebuffer */
+	info = drm_fb_helper_alloc_fbi(helper);
+	if (IS_ERR(info)) {
+		ret = PTR_ERR(info);
+		goto out_unref;
+	}
+
+	info->par = rfbdev;
+	info->skip_vt_switch = true;
+
+	ret = amdgpu_framebuffer_init(adev->ddev, &rfbdev->rfb, &mode_cmd, gobj);
+	if (ret) {
+		DRM_ERROR("failed to initialize framebuffer %d\n", ret);
+		goto out_destroy_fbi;
+	}
+
+	fb = &rfbdev->rfb.base;
+
+	/* setup helper */
+	rfbdev->helper.fb = fb;
+
+	memset_io(rbo->kptr, 0x0, amdgpu_bo_size(rbo));
+
+	strcpy(info->fix.id, "amdgpudrmfb");
+
+	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->depth);
+
+	info->flags = FBINFO_DEFAULT | FBINFO_CAN_FORCE_OUTPUT;
+	info->fbops = &amdgpufb_ops;
+
+	tmp = amdgpu_bo_gpu_offset(rbo) - adev->mc.vram_start;
+	info->fix.smem_start = adev->mc.aper_base + tmp;
+	info->fix.smem_len = amdgpu_bo_size(rbo);
+	info->screen_base = rbo->kptr;
+	info->screen_size = amdgpu_bo_size(rbo);
+
+	drm_fb_helper_fill_var(info, &rfbdev->helper, sizes->fb_width, sizes->fb_height);
+
+	/* setup aperture base/size for vesafb takeover */
+	info->apertures->ranges[0].base = adev->ddev->mode_config.fb_base;
+	info->apertures->ranges[0].size = adev->mc.aper_size;
+
+	/* Use default scratch pixmap (info->pixmap.flags = FB_PIXMAP_SYSTEM) */
+
+	if (info->screen_base == NULL) {
+		ret = -ENOSPC;
+		goto out_destroy_fbi;
+	}
+
+	DRM_INFO("fb mappable at 0x%lX\n",  info->fix.smem_start);
+	DRM_INFO("vram apper at 0x%lX\n",  (unsigned long)adev->mc.aper_base);
+	DRM_INFO("size %lu\n", (unsigned long)amdgpu_bo_size(rbo));
+	DRM_INFO("fb depth is %d\n", fb->depth);
+	DRM_INFO("   pitch is %d\n", fb->pitches[0]);
+
+	vga_switcheroo_client_fb_set(adev->ddev->pdev, info);
+	return 0;
+
+out_destroy_fbi:
+	drm_fb_helper_release_fbi(helper);
+out_unref:
+	if (rbo) {
+
+	}
+	if (fb && ret) {
+		drm_gem_object_unreference(gobj);
+		drm_framebuffer_unregister_private(fb);
+		drm_framebuffer_cleanup(fb);
+		kfree(fb);
+	}
+	return ret;
+}
+
+void amdgpu_fb_output_poll_changed(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.rfbdev)
+		drm_fb_helper_hotplug_event(&adev->mode_info.rfbdev->helper);
+}
+
+static int amdgpu_fbdev_destroy(struct drm_device *dev, struct amdgpu_fbdev *rfbdev)
+{
+	struct amdgpu_framebuffer *rfb = &rfbdev->rfb;
+
+	drm_fb_helper_unregister_fbi(&rfbdev->helper);
+	drm_fb_helper_release_fbi(&rfbdev->helper);
+
+	if (rfb->obj) {
+		amdgpufb_destroy_pinned_object(rfb->obj);
+		rfb->obj = NULL;
+	}
+	drm_fb_helper_fini(&rfbdev->helper);
+	drm_framebuffer_unregister_private(&rfb->base);
+	drm_framebuffer_cleanup(&rfb->base);
+
+	return 0;
+}
+
+/** Sets the color ramps on behalf of fbcon */
+static void amdgpu_crtc_fb_gamma_set(struct drm_crtc *crtc, u16 red, u16 green,
+				      u16 blue, int regno)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	amdgpu_crtc->lut_r[regno] = red >> 6;
+	amdgpu_crtc->lut_g[regno] = green >> 6;
+	amdgpu_crtc->lut_b[regno] = blue >> 6;
+}
+
+/** Gets the color ramps on behalf of fbcon */
+static void amdgpu_crtc_fb_gamma_get(struct drm_crtc *crtc, u16 *red, u16 *green,
+				      u16 *blue, int regno)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	*red = amdgpu_crtc->lut_r[regno] << 6;
+	*green = amdgpu_crtc->lut_g[regno] << 6;
+	*blue = amdgpu_crtc->lut_b[regno] << 6;
+}
+
+static const struct drm_fb_helper_funcs amdgpu_fb_helper_funcs = {
+	.gamma_set = amdgpu_crtc_fb_gamma_set,
+	.gamma_get = amdgpu_crtc_fb_gamma_get,
+	.fb_probe = amdgpufb_create,
+};
+
+int amdgpu_fbdev_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_fbdev *rfbdev;
+	int bpp_sel = 32;
+	int ret;
+
+	/* don't init fbdev on hw without DCE */
+	if (!adev->mode_info.mode_config_initialized)
+		return 0;
+
+	/* select 8 bpp console on low vram cards */
+	if (adev->mc.real_vram_size <= (32*1024*1024))
+		bpp_sel = 8;
+
+	rfbdev = kzalloc(sizeof(struct amdgpu_fbdev), GFP_KERNEL);
+	if (!rfbdev)
+		return -ENOMEM;
+
+	rfbdev->adev = adev;
+	adev->mode_info.rfbdev = rfbdev;
+
+	drm_fb_helper_prepare(adev->ddev, &rfbdev->helper,
+			&amdgpu_fb_helper_funcs);
+
+	ret = drm_fb_helper_init(adev->ddev, &rfbdev->helper,
+				 adev->mode_info.num_crtc,
+				 AMDGPUFB_CONN_LIMIT);
+	if (ret) {
+		kfree(rfbdev);
+		return ret;
+	}
+
+	drm_fb_helper_single_add_all_connectors(&rfbdev->helper);
+
+	/* disable all the possible outputs/crtcs before entering KMS mode */
+	drm_helper_disable_unused_functions(adev->ddev);
+
+	drm_fb_helper_initial_config(&rfbdev->helper, bpp_sel);
+	return 0;
+}
+
+void amdgpu_fbdev_fini(struct amdgpu_device *adev)
+{
+	if (!adev->mode_info.rfbdev)
+		return;
+
+	amdgpu_fbdev_destroy(adev->ddev, adev->mode_info.rfbdev);
+	kfree(adev->mode_info.rfbdev);
+	adev->mode_info.rfbdev = NULL;
+}
+
+void amdgpu_fbdev_set_suspend(struct amdgpu_device *adev, int state)
+{
+	if (adev->mode_info.rfbdev)
+		drm_fb_helper_set_suspend(&adev->mode_info.rfbdev->helper,
+			state);
+}
+
+int amdgpu_fbdev_total_size(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo *robj;
+	int size = 0;
+
+	if (!adev->mode_info.rfbdev)
+		return 0;
+
+	robj = gem_to_amdgpu_bo(adev->mode_info.rfbdev->rfb.obj);
+	size += amdgpu_bo_size(robj);
+	return size;
+}
+
+bool amdgpu_fbdev_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj)
+{
+	if (!adev->mode_info.rfbdev)
+		return false;
+	if (robj == gem_to_amdgpu_bo(adev->mode_info.rfbdev->rfb.obj))
+		return true;
+	return false;
+}
+
+void amdgpu_fbdev_restore_mode(struct amdgpu_device *adev)
+{
+	struct amdgpu_fbdev *afbdev = adev->mode_info.rfbdev;
+	struct drm_fb_helper *fb_helper;
+	int ret;
+
+	if (!afbdev)
+		return;
+
+	fb_helper = &afbdev->helper;
+
+	ret = drm_fb_helper_restore_fbdev_mode_unlocked(fb_helper);
+	if (ret)
+		DRM_DEBUG("failed to restore crtc mode\n");
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
new file mode 100644
index 000000000..3671f9f22
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -0,0 +1,823 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ *    Dave Airlie
+ */
+#include <linux/seq_file.h>
+#include <linux/atomic.h>
+#include <linux/wait.h>
+#include <linux/kref.h>
+#include <linux/slab.h>
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+/*
+ * Fences
+ * Fences mark an event in the GPUs pipeline and are used
+ * for GPU/CPU synchronization.  When the fence is written,
+ * it is expected that all buffers associated with that fence
+ * are no longer in use by the associated ring on the GPU and
+ * that the the relevant GPU caches have been flushed.
+ */
+
+static struct kmem_cache *amdgpu_fence_slab;
+static atomic_t amdgpu_fence_slab_ref = ATOMIC_INIT(0);
+
+/**
+ * amdgpu_fence_write - write a fence value
+ *
+ * @ring: ring the fence is associated with
+ * @seq: sequence number to write
+ *
+ * Writes a fence value to memory (all asics).
+ */
+static void amdgpu_fence_write(struct amdgpu_ring *ring, u32 seq)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+
+	if (drv->cpu_addr)
+		*drv->cpu_addr = cpu_to_le32(seq);
+}
+
+/**
+ * amdgpu_fence_read - read a fence value
+ *
+ * @ring: ring the fence is associated with
+ *
+ * Reads a fence value from memory (all asics).
+ * Returns the value of the fence read from memory.
+ */
+static u32 amdgpu_fence_read(struct amdgpu_ring *ring)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	u32 seq = 0;
+
+	if (drv->cpu_addr)
+		seq = le32_to_cpu(*drv->cpu_addr);
+	else
+		seq = lower_32_bits(atomic64_read(&drv->last_seq));
+
+	return seq;
+}
+
+/**
+ * amdgpu_fence_emit - emit a fence on the requested ring
+ *
+ * @ring: ring the fence is associated with
+ * @owner: creator of the fence
+ * @fence: amdgpu fence object
+ *
+ * Emits a fence command on the requested ring (all asics).
+ * Returns 0 on success, -ENOMEM on failure.
+ */
+int amdgpu_fence_emit(struct amdgpu_ring *ring, void *owner,
+		      struct amdgpu_fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* we are protected by the ring emission mutex */
+	*fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_KERNEL);
+	if ((*fence) == NULL) {
+		return -ENOMEM;
+	}
+	(*fence)->seq = ++ring->fence_drv.sync_seq[ring->idx];
+	(*fence)->ring = ring;
+	(*fence)->owner = owner;
+	fence_init(&(*fence)->base, &amdgpu_fence_ops,
+		&ring->fence_drv.fence_queue.lock,
+		adev->fence_context + ring->idx,
+		(*fence)->seq);
+	amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr,
+			       (*fence)->seq,
+			       AMDGPU_FENCE_FLAG_INT);
+	return 0;
+}
+
+/**
+ * amdgpu_fence_schedule_fallback - schedule fallback check
+ *
+ * @ring: pointer to struct amdgpu_ring
+ *
+ * Start a timer as fallback to our interrupts.
+ */
+static void amdgpu_fence_schedule_fallback(struct amdgpu_ring *ring)
+{
+	mod_timer(&ring->fence_drv.fallback_timer,
+		  jiffies + AMDGPU_FENCE_JIFFIES_TIMEOUT);
+}
+
+/**
+ * amdgpu_fence_activity - check for fence activity
+ *
+ * @ring: pointer to struct amdgpu_ring
+ *
+ * Checks the current fence value and calculates the last
+ * signalled fence value. Returns true if activity occured
+ * on the ring, and the fence_queue should be waken up.
+ */
+static bool amdgpu_fence_activity(struct amdgpu_ring *ring)
+{
+	uint64_t seq, last_seq, last_emitted;
+	unsigned count_loop = 0;
+	bool wake = false;
+
+	/* Note there is a scenario here for an infinite loop but it's
+	 * very unlikely to happen. For it to happen, the current polling
+	 * process need to be interrupted by another process and another
+	 * process needs to update the last_seq btw the atomic read and
+	 * xchg of the current process.
+	 *
+	 * More over for this to go in infinite loop there need to be
+	 * continuously new fence signaled ie amdgpu_fence_read needs
+	 * to return a different value each time for both the currently
+	 * polling process and the other process that xchg the last_seq
+	 * btw atomic read and xchg of the current process. And the
+	 * value the other process set as last seq must be higher than
+	 * the seq value we just read. Which means that current process
+	 * need to be interrupted after amdgpu_fence_read and before
+	 * atomic xchg.
+	 *
+	 * To be even more safe we count the number of time we loop and
+	 * we bail after 10 loop just accepting the fact that we might
+	 * have temporarly set the last_seq not to the true real last
+	 * seq but to an older one.
+	 */
+	last_seq = atomic64_read(&ring->fence_drv.last_seq);
+	do {
+		last_emitted = ring->fence_drv.sync_seq[ring->idx];
+		seq = amdgpu_fence_read(ring);
+		seq |= last_seq & 0xffffffff00000000LL;
+		if (seq < last_seq) {
+			seq &= 0xffffffff;
+			seq |= last_emitted & 0xffffffff00000000LL;
+		}
+
+		if (seq <= last_seq || seq > last_emitted) {
+			break;
+		}
+		/* If we loop over we don't want to return without
+		 * checking if a fence is signaled as it means that the
+		 * seq we just read is different from the previous on.
+		 */
+		wake = true;
+		last_seq = seq;
+		if ((count_loop++) > 10) {
+			/* We looped over too many time leave with the
+			 * fact that we might have set an older fence
+			 * seq then the current real last seq as signaled
+			 * by the hw.
+			 */
+			break;
+		}
+	} while (atomic64_xchg(&ring->fence_drv.last_seq, seq) > seq);
+
+	if (seq < last_emitted)
+		amdgpu_fence_schedule_fallback(ring);
+
+	return wake;
+}
+
+/**
+ * amdgpu_fence_process - process a fence
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring index the fence is associated with
+ *
+ * Checks the current fence value and wakes the fence queue
+ * if the sequence number has increased (all asics).
+ */
+void amdgpu_fence_process(struct amdgpu_ring *ring)
+{
+	if (amdgpu_fence_activity(ring))
+		wake_up_all(&ring->fence_drv.fence_queue);
+}
+
+/**
+ * amdgpu_fence_fallback - fallback for hardware interrupts
+ *
+ * @work: delayed work item
+ *
+ * Checks for fence activity.
+ */
+static void amdgpu_fence_fallback(unsigned long arg)
+{
+	struct amdgpu_ring *ring = (void *)arg;
+
+	amdgpu_fence_process(ring);
+}
+
+/**
+ * amdgpu_fence_seq_signaled - check if a fence sequence number has signaled
+ *
+ * @ring: ring the fence is associated with
+ * @seq: sequence number
+ *
+ * Check if the last signaled fence sequnce number is >= the requested
+ * sequence number (all asics).
+ * Returns true if the fence has signaled (current fence value
+ * is >= requested value) or false if it has not (current fence
+ * value is < the requested value.  Helper function for
+ * amdgpu_fence_signaled().
+ */
+static bool amdgpu_fence_seq_signaled(struct amdgpu_ring *ring, u64 seq)
+{
+	if (atomic64_read(&ring->fence_drv.last_seq) >= seq)
+		return true;
+
+	/* poll new last sequence at least once */
+	amdgpu_fence_process(ring);
+	if (atomic64_read(&ring->fence_drv.last_seq) >= seq)
+		return true;
+
+	return false;
+}
+
+/*
+ * amdgpu_ring_wait_seq_timeout - wait for seq of the specific ring to signal
+ * @ring: ring to wait on for the seq number
+ * @seq: seq number wait for
+ *
+ * return value:
+ * 0: seq signaled, and gpu not hang
+ * -EDEADL: GPU hang detected
+ * -EINVAL: some paramter is not valid
+ */
+static int amdgpu_fence_ring_wait_seq(struct amdgpu_ring *ring, uint64_t seq)
+{
+	bool signaled = false;
+
+	BUG_ON(!ring);
+	if (seq > ring->fence_drv.sync_seq[ring->idx])
+		return -EINVAL;
+
+	if (atomic64_read(&ring->fence_drv.last_seq) >= seq)
+		return 0;
+
+	amdgpu_fence_schedule_fallback(ring);
+	wait_event(ring->fence_drv.fence_queue, (
+		   (signaled = amdgpu_fence_seq_signaled(ring, seq))));
+
+	if (signaled)
+		return 0;
+	else
+		return -EDEADLK;
+}
+
+/**
+ * amdgpu_fence_wait_next - wait for the next fence to signal
+ *
+ * @adev: amdgpu device pointer
+ * @ring: ring index the fence is associated with
+ *
+ * Wait for the next fence on the requested ring to signal (all asics).
+ * Returns 0 if the next fence has passed, error for all other cases.
+ * Caller must hold ring lock.
+ */
+int amdgpu_fence_wait_next(struct amdgpu_ring *ring)
+{
+	uint64_t seq = atomic64_read(&ring->fence_drv.last_seq) + 1ULL;
+
+	if (seq >= ring->fence_drv.sync_seq[ring->idx])
+		return -ENOENT;
+
+	return amdgpu_fence_ring_wait_seq(ring, seq);
+}
+
+/**
+ * amdgpu_fence_wait_empty - wait for all fences to signal
+ *
+ * @adev: amdgpu device pointer
+ * @ring: ring index the fence is associated with
+ *
+ * Wait for all fences on the requested ring to signal (all asics).
+ * Returns 0 if the fences have passed, error for all other cases.
+ * Caller must hold ring lock.
+ */
+int amdgpu_fence_wait_empty(struct amdgpu_ring *ring)
+{
+	uint64_t seq = ring->fence_drv.sync_seq[ring->idx];
+
+	if (!seq)
+		return 0;
+
+	return amdgpu_fence_ring_wait_seq(ring, seq);
+}
+
+/**
+ * amdgpu_fence_count_emitted - get the count of emitted fences
+ *
+ * @ring: ring the fence is associated with
+ *
+ * Get the number of fences emitted on the requested ring (all asics).
+ * Returns the number of emitted fences on the ring.  Used by the
+ * dynpm code to ring track activity.
+ */
+unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring)
+{
+	uint64_t emitted;
+
+	/* We are not protected by ring lock when reading the last sequence
+	 * but it's ok to report slightly wrong fence count here.
+	 */
+	amdgpu_fence_process(ring);
+	emitted = ring->fence_drv.sync_seq[ring->idx]
+		- atomic64_read(&ring->fence_drv.last_seq);
+	/* to avoid 32bits warp around */
+	if (emitted > 0x10000000)
+		emitted = 0x10000000;
+
+	return (unsigned)emitted;
+}
+
+/**
+ * amdgpu_fence_need_sync - do we need a semaphore
+ *
+ * @fence: amdgpu fence object
+ * @dst_ring: which ring to check against
+ *
+ * Check if the fence needs to be synced against another ring
+ * (all asics).  If so, we need to emit a semaphore.
+ * Returns true if we need to sync with another ring, false if
+ * not.
+ */
+bool amdgpu_fence_need_sync(struct amdgpu_fence *fence,
+			    struct amdgpu_ring *dst_ring)
+{
+	struct amdgpu_fence_driver *fdrv;
+
+	if (!fence)
+		return false;
+
+	if (fence->ring == dst_ring)
+		return false;
+
+	/* we are protected by the ring mutex */
+	fdrv = &dst_ring->fence_drv;
+	if (fence->seq <= fdrv->sync_seq[fence->ring->idx])
+		return false;
+
+	return true;
+}
+
+/**
+ * amdgpu_fence_note_sync - record the sync point
+ *
+ * @fence: amdgpu fence object
+ * @dst_ring: which ring to check against
+ *
+ * Note the sequence number at which point the fence will
+ * be synced with the requested ring (all asics).
+ */
+void amdgpu_fence_note_sync(struct amdgpu_fence *fence,
+			    struct amdgpu_ring *dst_ring)
+{
+	struct amdgpu_fence_driver *dst, *src;
+	unsigned i;
+
+	if (!fence)
+		return;
+
+	if (fence->ring == dst_ring)
+		return;
+
+	/* we are protected by the ring mutex */
+	src = &fence->ring->fence_drv;
+	dst = &dst_ring->fence_drv;
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		if (i == dst_ring->idx)
+			continue;
+
+		dst->sync_seq[i] = max(dst->sync_seq[i], src->sync_seq[i]);
+	}
+}
+
+/**
+ * amdgpu_fence_driver_start_ring - make the fence driver
+ * ready for use on the requested ring.
+ *
+ * @ring: ring to start the fence driver on
+ * @irq_src: interrupt source to use for this ring
+ * @irq_type: interrupt type to use for this ring
+ *
+ * Make the fence driver ready for processing (all asics).
+ * Not all asics have all rings, so each asic will only
+ * start the fence driver on the rings it has.
+ * Returns 0 for success, errors for failure.
+ */
+int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
+				   struct amdgpu_irq_src *irq_src,
+				   unsigned irq_type)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint64_t index;
+
+	if (ring != &adev->uvd.ring) {
+		ring->fence_drv.cpu_addr = &adev->wb.wb[ring->fence_offs];
+		ring->fence_drv.gpu_addr = adev->wb.gpu_addr + (ring->fence_offs * 4);
+	} else {
+		/* put fence directly behind firmware */
+		index = ALIGN(adev->uvd.fw->size, 8);
+		ring->fence_drv.cpu_addr = adev->uvd.cpu_addr + index;
+		ring->fence_drv.gpu_addr = adev->uvd.gpu_addr + index;
+	}
+	amdgpu_fence_write(ring, atomic64_read(&ring->fence_drv.last_seq));
+	amdgpu_irq_get(adev, irq_src, irq_type);
+
+	ring->fence_drv.irq_src = irq_src;
+	ring->fence_drv.irq_type = irq_type;
+	ring->fence_drv.initialized = true;
+
+	dev_info(adev->dev, "fence driver on ring %d use gpu addr 0x%016llx, "
+		 "cpu addr 0x%p\n", ring->idx,
+		 ring->fence_drv.gpu_addr, ring->fence_drv.cpu_addr);
+	return 0;
+}
+
+/**
+ * amdgpu_fence_driver_init_ring - init the fence driver
+ * for the requested ring.
+ *
+ * @ring: ring to init the fence driver on
+ *
+ * Init the fence driver for the requested ring (all asics).
+ * Helper function for amdgpu_fence_driver_init().
+ */
+int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring)
+{
+	int i, r;
+
+	ring->fence_drv.cpu_addr = NULL;
+	ring->fence_drv.gpu_addr = 0;
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		ring->fence_drv.sync_seq[i] = 0;
+
+	atomic64_set(&ring->fence_drv.last_seq, 0);
+	ring->fence_drv.initialized = false;
+
+	setup_timer(&ring->fence_drv.fallback_timer, amdgpu_fence_fallback,
+		    (unsigned long)ring);
+
+	init_waitqueue_head(&ring->fence_drv.fence_queue);
+
+	if (amdgpu_enable_scheduler) {
+		long timeout = msecs_to_jiffies(amdgpu_lockup_timeout);
+		if (timeout == 0) {
+			/*
+			 * FIXME:
+			 * Delayed workqueue cannot use it directly,
+			 * so the scheduler will not use delayed workqueue if
+			 * MAX_SCHEDULE_TIMEOUT is set.
+			 * Currently keep it simple and silly.
+			 */
+			timeout = MAX_SCHEDULE_TIMEOUT;
+		}
+		r = amd_sched_init(&ring->sched, &amdgpu_sched_ops,
+				   amdgpu_sched_hw_submission,
+				   timeout, ring->name);
+		if (r) {
+			DRM_ERROR("Failed to create scheduler on ring %s.\n",
+				  ring->name);
+			return r;
+		}
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_fence_driver_init - init the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Init the fence driver for all possible rings (all asics).
+ * Not all asics have all rings, so each asic will only
+ * start the fence driver on the rings it has using
+ * amdgpu_fence_driver_start_ring().
+ * Returns 0 for success.
+ */
+int amdgpu_fence_driver_init(struct amdgpu_device *adev)
+{
+	if (atomic_inc_return(&amdgpu_fence_slab_ref) == 1) {
+		amdgpu_fence_slab = kmem_cache_create(
+			"amdgpu_fence", sizeof(struct amdgpu_fence), 0,
+			SLAB_HWCACHE_ALIGN, NULL);
+		if (!amdgpu_fence_slab)
+			return -ENOMEM;
+	}
+	if (amdgpu_debugfs_fence_init(adev))
+		dev_err(adev->dev, "fence debugfs file creation failed\n");
+
+	return 0;
+}
+
+/**
+ * amdgpu_fence_driver_fini - tear down the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Tear down the fence driver for all possible rings (all asics).
+ */
+void amdgpu_fence_driver_fini(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	if (atomic_dec_and_test(&amdgpu_fence_slab_ref))
+		kmem_cache_destroy(amdgpu_fence_slab);
+	mutex_lock(&adev->ring_lock);
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+		r = amdgpu_fence_wait_empty(ring);
+		if (r) {
+			/* no need to trigger GPU reset as we are unloading */
+			amdgpu_fence_driver_force_completion(adev);
+		}
+		wake_up_all(&ring->fence_drv.fence_queue);
+		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
+			       ring->fence_drv.irq_type);
+		amd_sched_fini(&ring->sched);
+		del_timer_sync(&ring->fence_drv.fallback_timer);
+		ring->fence_drv.initialized = false;
+	}
+	mutex_unlock(&adev->ring_lock);
+}
+
+/**
+ * amdgpu_fence_driver_suspend - suspend the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Suspend the fence driver for all possible rings (all asics).
+ */
+void amdgpu_fence_driver_suspend(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	mutex_lock(&adev->ring_lock);
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		/* wait for gpu to finish processing current batch */
+		r = amdgpu_fence_wait_empty(ring);
+		if (r) {
+			/* delay GPU reset to resume */
+			amdgpu_fence_driver_force_completion(adev);
+		}
+
+		/* disable the interrupt */
+		amdgpu_irq_put(adev, ring->fence_drv.irq_src,
+			       ring->fence_drv.irq_type);
+	}
+	mutex_unlock(&adev->ring_lock);
+}
+
+/**
+ * amdgpu_fence_driver_resume - resume the fence driver
+ * for all possible rings.
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Resume the fence driver for all possible rings (all asics).
+ * Not all asics have all rings, so each asic will only
+ * start the fence driver on the rings it has using
+ * amdgpu_fence_driver_start_ring().
+ * Returns 0 for success.
+ */
+void amdgpu_fence_driver_resume(struct amdgpu_device *adev)
+{
+	int i;
+
+	mutex_lock(&adev->ring_lock);
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		/* enable the interrupt */
+		amdgpu_irq_get(adev, ring->fence_drv.irq_src,
+			       ring->fence_drv.irq_type);
+	}
+	mutex_unlock(&adev->ring_lock);
+}
+
+/**
+ * amdgpu_fence_driver_force_completion - force all fence waiter to complete
+ *
+ * @adev: amdgpu device pointer
+ *
+ * In case of GPU reset failure make sure no process keep waiting on fence
+ * that will never complete.
+ */
+void amdgpu_fence_driver_force_completion(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		amdgpu_fence_write(ring, ring->fence_drv.sync_seq[i]);
+	}
+}
+
+/*
+ * Common fence implementation
+ */
+
+static const char *amdgpu_fence_get_driver_name(struct fence *fence)
+{
+	return "amdgpu";
+}
+
+static const char *amdgpu_fence_get_timeline_name(struct fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	return (const char *)fence->ring->name;
+}
+
+/**
+ * amdgpu_fence_is_signaled - test if fence is signaled
+ *
+ * @f: fence to test
+ *
+ * Test the fence sequence number if it is already signaled. If it isn't
+ * signaled start fence processing. Returns True if the fence is signaled.
+ */
+static bool amdgpu_fence_is_signaled(struct fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	struct amdgpu_ring *ring = fence->ring;
+
+	if (atomic64_read(&ring->fence_drv.last_seq) >= fence->seq)
+		return true;
+
+	amdgpu_fence_process(ring);
+
+	if (atomic64_read(&ring->fence_drv.last_seq) >= fence->seq)
+		return true;
+
+	return false;
+}
+
+/**
+ * amdgpu_fence_check_signaled - callback from fence_queue
+ *
+ * this function is called with fence_queue lock held, which is also used
+ * for the fence locking itself, so unlocked variants are used for
+ * fence_signal, and remove_wait_queue.
+ */
+static int amdgpu_fence_check_signaled(wait_queue_t *wait, unsigned mode, int flags, void *key)
+{
+	struct amdgpu_fence *fence;
+	struct amdgpu_device *adev;
+	u64 seq;
+	int ret;
+
+	fence = container_of(wait, struct amdgpu_fence, fence_wake);
+	adev = fence->ring->adev;
+
+	/*
+	 * We cannot use amdgpu_fence_process here because we're already
+	 * in the waitqueue, in a call from wake_up_all.
+	 */
+	seq = atomic64_read(&fence->ring->fence_drv.last_seq);
+	if (seq >= fence->seq) {
+		ret = fence_signal_locked(&fence->base);
+		if (!ret)
+			FENCE_TRACE(&fence->base, "signaled from irq context\n");
+		else
+			FENCE_TRACE(&fence->base, "was already signaled\n");
+
+		__remove_wait_queue(&fence->ring->fence_drv.fence_queue, &fence->fence_wake);
+		fence_put(&fence->base);
+	} else
+		FENCE_TRACE(&fence->base, "pending\n");
+	return 0;
+}
+
+/**
+ * amdgpu_fence_enable_signaling - enable signalling on fence
+ * @fence: fence
+ *
+ * This function is called with fence_queue lock held, and adds a callback
+ * to fence_queue that checks if this fence is signaled, and if so it
+ * signals the fence and removes itself.
+ */
+static bool amdgpu_fence_enable_signaling(struct fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	struct amdgpu_ring *ring = fence->ring;
+
+	if (atomic64_read(&ring->fence_drv.last_seq) >= fence->seq)
+		return false;
+
+	fence->fence_wake.flags = 0;
+	fence->fence_wake.private = NULL;
+	fence->fence_wake.func = amdgpu_fence_check_signaled;
+	__add_wait_queue(&ring->fence_drv.fence_queue, &fence->fence_wake);
+	fence_get(f);
+	if (!timer_pending(&ring->fence_drv.fallback_timer))
+		amdgpu_fence_schedule_fallback(ring);
+	FENCE_TRACE(&fence->base, "armed on ring %i!\n", ring->idx);
+	return true;
+}
+
+static void amdgpu_fence_release(struct fence *f)
+{
+	struct amdgpu_fence *fence = to_amdgpu_fence(f);
+	kmem_cache_free(amdgpu_fence_slab, fence);
+}
+
+const struct fence_ops amdgpu_fence_ops = {
+	.get_driver_name = amdgpu_fence_get_driver_name,
+	.get_timeline_name = amdgpu_fence_get_timeline_name,
+	.enable_signaling = amdgpu_fence_enable_signaling,
+	.signaled = amdgpu_fence_is_signaled,
+	.wait = fence_default_wait,
+	.release = amdgpu_fence_release,
+};
+
+/*
+ * Fence debugfs
+ */
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int i, j;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (!ring || !ring->fence_drv.initialized)
+			continue;
+
+		amdgpu_fence_process(ring);
+
+		seq_printf(m, "--- ring %d (%s) ---\n", i, ring->name);
+		seq_printf(m, "Last signaled fence 0x%016llx\n",
+			   (unsigned long long)atomic64_read(&ring->fence_drv.last_seq));
+		seq_printf(m, "Last emitted        0x%016llx\n",
+			   ring->fence_drv.sync_seq[i]);
+
+		for (j = 0; j < AMDGPU_MAX_RINGS; ++j) {
+			struct amdgpu_ring *other = adev->rings[j];
+			if (i != j && other && other->fence_drv.initialized &&
+			    ring->fence_drv.sync_seq[j])
+				seq_printf(m, "Last sync to ring %d 0x%016llx\n",
+					   j, ring->fence_drv.sync_seq[j]);
+		}
+	}
+	return 0;
+}
+
+static struct drm_info_list amdgpu_debugfs_fence_list[] = {
+	{"amdgpu_fence_info", &amdgpu_debugfs_fence_info, 0, NULL},
+};
+#endif
+
+int amdgpu_debugfs_fence_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_fence_list, 1);
+#else
+	return 0;
+#endif
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
new file mode 100644
index 000000000..7312d729d
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
@@ -0,0 +1,372 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+/*
+ * GART
+ * The GART (Graphics Aperture Remapping Table) is an aperture
+ * in the GPU's address space.  System pages can be mapped into
+ * the aperture and look like contiguous pages from the GPU's
+ * perspective.  A page table maps the pages in the aperture
+ * to the actual backing pages in system memory.
+ *
+ * Radeon GPUs support both an internal GART, as described above,
+ * and AGP.  AGP works similarly, but the GART table is configured
+ * and maintained by the northbridge rather than the driver.
+ * Radeon hw has a separate AGP aperture that is programmed to
+ * point to the AGP aperture provided by the northbridge and the
+ * requests are passed through to the northbridge aperture.
+ * Both AGP and internal GART can be used at the same time, however
+ * that is not currently supported by the driver.
+ *
+ * This file handles the common internal GART management.
+ */
+
+/*
+ * Common GART table functions.
+ */
+/**
+ * amdgpu_gart_table_ram_alloc - allocate system ram for gart page table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate system memory for GART page table
+ * (r1xx-r3xx, non-pcie r4xx, rs400).  These asics require the
+ * gart table to be in system memory.
+ * Returns 0 for success, -ENOMEM for failure.
+ */
+int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev)
+{
+	void *ptr;
+
+	ptr = pci_alloc_consistent(adev->pdev, adev->gart.table_size,
+				   &adev->gart.table_addr);
+	if (ptr == NULL) {
+		return -ENOMEM;
+	}
+#ifdef CONFIG_X86
+	if (0) {
+		set_memory_uc((unsigned long)ptr,
+			      adev->gart.table_size >> PAGE_SHIFT);
+	}
+#endif
+	adev->gart.ptr = ptr;
+	memset((void *)adev->gart.ptr, 0, adev->gart.table_size);
+	return 0;
+}
+
+/**
+ * amdgpu_gart_table_ram_free - free system ram for gart page table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Free system memory for GART page table
+ * (r1xx-r3xx, non-pcie r4xx, rs400).  These asics require the
+ * gart table to be in system memory.
+ */
+void amdgpu_gart_table_ram_free(struct amdgpu_device *adev)
+{
+	if (adev->gart.ptr == NULL) {
+		return;
+	}
+#ifdef CONFIG_X86
+	if (0) {
+		set_memory_wb((unsigned long)adev->gart.ptr,
+			      adev->gart.table_size >> PAGE_SHIFT);
+	}
+#endif
+	pci_free_consistent(adev->pdev, adev->gart.table_size,
+			    (void *)adev->gart.ptr,
+			    adev->gart.table_addr);
+	adev->gart.ptr = NULL;
+	adev->gart.table_addr = 0;
+}
+
+/**
+ * amdgpu_gart_table_vram_alloc - allocate vram for gart page table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate video memory for GART page table
+ * (pcie r4xx, r5xx+).  These asics require the
+ * gart table to be in video memory.
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj == NULL) {
+		r = amdgpu_bo_create(adev, adev->gart.table_size,
+				     PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
+				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+				     NULL, NULL, &adev->gart.robj);
+		if (r) {
+			return r;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_gart_table_vram_pin - pin gart page table in vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Pin the GART page table in vram so it will not be moved
+ * by the memory manager (pcie r4xx, r5xx+).  These asics require the
+ * gart table to be in video memory.
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev)
+{
+	uint64_t gpu_addr;
+	int r;
+
+	r = amdgpu_bo_reserve(adev->gart.robj, false);
+	if (unlikely(r != 0))
+		return r;
+	r = amdgpu_bo_pin(adev->gart.robj,
+				AMDGPU_GEM_DOMAIN_VRAM, &gpu_addr);
+	if (r) {
+		amdgpu_bo_unreserve(adev->gart.robj);
+		return r;
+	}
+	r = amdgpu_bo_kmap(adev->gart.robj, &adev->gart.ptr);
+	if (r)
+		amdgpu_bo_unpin(adev->gart.robj);
+	amdgpu_bo_unreserve(adev->gart.robj);
+	adev->gart.table_addr = gpu_addr;
+	return r;
+}
+
+/**
+ * amdgpu_gart_table_vram_unpin - unpin gart page table in vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unpin the GART page table in vram (pcie r4xx, r5xx+).
+ * These asics require the gart table to be in video memory.
+ */
+void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj == NULL) {
+		return;
+	}
+	r = amdgpu_bo_reserve(adev->gart.robj, false);
+	if (likely(r == 0)) {
+		amdgpu_bo_kunmap(adev->gart.robj);
+		amdgpu_bo_unpin(adev->gart.robj);
+		amdgpu_bo_unreserve(adev->gart.robj);
+		adev->gart.ptr = NULL;
+	}
+}
+
+/**
+ * amdgpu_gart_table_vram_free - free gart page table vram
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Free the video memory used for the GART page table
+ * (pcie r4xx, r5xx+).  These asics require the gart table to
+ * be in video memory.
+ */
+void amdgpu_gart_table_vram_free(struct amdgpu_device *adev)
+{
+	if (adev->gart.robj == NULL) {
+		return;
+	}
+	amdgpu_bo_unref(&adev->gart.robj);
+}
+
+/*
+ * Common gart functions.
+ */
+/**
+ * amdgpu_gart_unbind - unbind pages from the gart page table
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: offset into the GPU's gart aperture
+ * @pages: number of pages to unbind
+ *
+ * Unbinds the requested pages from the gart page table and
+ * replaces them with the dummy page (all asics).
+ */
+void amdgpu_gart_unbind(struct amdgpu_device *adev, unsigned offset,
+			int pages)
+{
+	unsigned t;
+	unsigned p;
+	int i, j;
+	u64 page_base;
+	uint32_t flags = AMDGPU_PTE_SYSTEM;
+
+	if (!adev->gart.ready) {
+		WARN(1, "trying to unbind memory from uninitialized GART !\n");
+		return;
+	}
+
+	t = offset / AMDGPU_GPU_PAGE_SIZE;
+	p = t / (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE);
+	for (i = 0; i < pages; i++, p++) {
+		if (adev->gart.pages[p]) {
+			adev->gart.pages[p] = NULL;
+			adev->gart.pages_addr[p] = adev->dummy_page.addr;
+			page_base = adev->gart.pages_addr[p];
+			if (!adev->gart.ptr)
+				continue;
+
+			for (j = 0; j < (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE); j++, t++) {
+				amdgpu_gart_set_pte_pde(adev, adev->gart.ptr,
+							t, page_base, flags);
+				page_base += AMDGPU_GPU_PAGE_SIZE;
+			}
+		}
+	}
+	mb();
+	amdgpu_gart_flush_gpu_tlb(adev, 0);
+}
+
+/**
+ * amdgpu_gart_bind - bind pages into the gart page table
+ *
+ * @adev: amdgpu_device pointer
+ * @offset: offset into the GPU's gart aperture
+ * @pages: number of pages to bind
+ * @pagelist: pages to bind
+ * @dma_addr: DMA addresses of pages
+ *
+ * Binds the requested pages to the gart page table
+ * (all asics).
+ * Returns 0 for success, -EINVAL for failure.
+ */
+int amdgpu_gart_bind(struct amdgpu_device *adev, unsigned offset,
+		     int pages, struct page **pagelist, dma_addr_t *dma_addr,
+		     uint32_t flags)
+{
+	unsigned t;
+	unsigned p;
+	uint64_t page_base;
+	int i, j;
+
+	if (!adev->gart.ready) {
+		WARN(1, "trying to bind memory to uninitialized GART !\n");
+		return -EINVAL;
+	}
+
+	t = offset / AMDGPU_GPU_PAGE_SIZE;
+	p = t / (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE);
+
+	for (i = 0; i < pages; i++, p++) {
+		adev->gart.pages_addr[p] = dma_addr[i];
+		adev->gart.pages[p] = pagelist[i];
+		if (adev->gart.ptr) {
+			page_base = adev->gart.pages_addr[p];
+			for (j = 0; j < (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE); j++, t++) {
+				amdgpu_gart_set_pte_pde(adev, adev->gart.ptr, t, page_base, flags);
+				page_base += AMDGPU_GPU_PAGE_SIZE;
+			}
+		}
+	}
+	mb();
+	amdgpu_gart_flush_gpu_tlb(adev, 0);
+	return 0;
+}
+
+/**
+ * amdgpu_gart_init - init the driver info for managing the gart
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate the dummy page and init the gart driver info (all asics).
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_gart_init(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	if (adev->gart.pages) {
+		return 0;
+	}
+	/* We need PAGE_SIZE >= AMDGPU_GPU_PAGE_SIZE */
+	if (PAGE_SIZE < AMDGPU_GPU_PAGE_SIZE) {
+		DRM_ERROR("Page size is smaller than GPU page size!\n");
+		return -EINVAL;
+	}
+	r = amdgpu_dummy_page_init(adev);
+	if (r)
+		return r;
+	/* Compute table size */
+	adev->gart.num_cpu_pages = adev->mc.gtt_size / PAGE_SIZE;
+	adev->gart.num_gpu_pages = adev->mc.gtt_size / AMDGPU_GPU_PAGE_SIZE;
+	DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n",
+		 adev->gart.num_cpu_pages, adev->gart.num_gpu_pages);
+	/* Allocate pages table */
+	adev->gart.pages = vzalloc(sizeof(void *) * adev->gart.num_cpu_pages);
+	if (adev->gart.pages == NULL) {
+		amdgpu_gart_fini(adev);
+		return -ENOMEM;
+	}
+	adev->gart.pages_addr = vzalloc(sizeof(dma_addr_t) *
+					adev->gart.num_cpu_pages);
+	if (adev->gart.pages_addr == NULL) {
+		amdgpu_gart_fini(adev);
+		return -ENOMEM;
+	}
+	/* set GART entry to point to the dummy page by default */
+	for (i = 0; i < adev->gart.num_cpu_pages; i++) {
+		adev->gart.pages_addr[i] = adev->dummy_page.addr;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_gart_fini - tear down the driver info for managing the gart
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the gart driver info and free the dummy page (all asics).
+ */
+void amdgpu_gart_fini(struct amdgpu_device *adev)
+{
+	if (adev->gart.pages && adev->gart.pages_addr && adev->gart.ready) {
+		/* unbind pages */
+		amdgpu_gart_unbind(adev, 0, adev->gart.num_cpu_pages);
+	}
+	adev->gart.ready = false;
+	vfree(adev->gart.pages);
+	vfree(adev->gart.pages_addr);
+	adev->gart.pages = NULL;
+	adev->gart.pages_addr = NULL;
+
+	amdgpu_dummy_page_fini(adev);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
new file mode 100644
index 000000000..c3f4e8559
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
@@ -0,0 +1,72 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_GDS_H__
+#define __AMDGPU_GDS_H__
+
+/* Because TTM request that alloacted buffer should be PAGE_SIZE aligned,
+ * we should report GDS/GWS/OA size as PAGE_SIZE aligned
+ * */
+#define AMDGPU_GDS_SHIFT	2
+#define AMDGPU_GWS_SHIFT	PAGE_SHIFT
+#define AMDGPU_OA_SHIFT		PAGE_SHIFT
+
+#define AMDGPU_PL_GDS		TTM_PL_PRIV0
+#define AMDGPU_PL_GWS		TTM_PL_PRIV1
+#define AMDGPU_PL_OA		TTM_PL_PRIV2
+
+#define AMDGPU_PL_FLAG_GDS		TTM_PL_FLAG_PRIV0
+#define AMDGPU_PL_FLAG_GWS		TTM_PL_FLAG_PRIV1
+#define AMDGPU_PL_FLAG_OA		TTM_PL_FLAG_PRIV2
+
+struct amdgpu_ring;
+struct amdgpu_bo;
+
+struct amdgpu_gds_asic_info {
+	uint32_t 	total_size;
+	uint32_t	gfx_partition_size;
+	uint32_t	cs_partition_size;
+};
+
+struct amdgpu_gds {
+	struct amdgpu_gds_asic_info	mem;
+	struct amdgpu_gds_asic_info	gws;
+	struct amdgpu_gds_asic_info	oa;
+	/* At present, GDS, GWS and OA resources for gfx (graphics) 
+	 * is always pre-allocated and available for graphics operation. 
+	 * Such resource is shared between all gfx clients.
+	 * TODO: move this operation to user space
+	 * */
+	struct amdgpu_bo*		gds_gfx_bo;
+	struct amdgpu_bo*		gws_gfx_bo;
+	struct amdgpu_bo*		oa_gfx_bo;
+};
+
+struct amdgpu_gds_reg_offset {
+	uint32_t	mem_base;
+	uint32_t	mem_size;
+	uint32_t	gws;
+	uint32_t	oa;
+};
+
+#endif /* __AMDGPU_GDS_H__ */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
new file mode 100644
index 000000000..9c253c535
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -0,0 +1,740 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <linux/ktime.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+void amdgpu_gem_object_free(struct drm_gem_object *gobj)
+{
+	struct amdgpu_bo *robj = gem_to_amdgpu_bo(gobj);
+
+	if (robj) {
+		if (robj->gem_base.import_attach)
+			drm_prime_gem_destroy(&robj->gem_base, robj->tbo.sg);
+		amdgpu_mn_unregister(robj);
+		amdgpu_bo_unref(&robj);
+	}
+}
+
+int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
+				int alignment, u32 initial_domain,
+				u64 flags, bool kernel,
+				struct drm_gem_object **obj)
+{
+	struct amdgpu_bo *robj;
+	unsigned long max_size;
+	int r;
+
+	*obj = NULL;
+	/* At least align on page size */
+	if (alignment < PAGE_SIZE) {
+		alignment = PAGE_SIZE;
+	}
+
+	if (!(initial_domain & (AMDGPU_GEM_DOMAIN_GDS | AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA))) {
+		/* Maximum bo size is the unpinned gtt size since we use the gtt to
+		 * handle vram to system pool migrations.
+		 */
+		max_size = adev->mc.gtt_size - adev->gart_pin_size;
+		if (size > max_size) {
+			DRM_DEBUG("Allocation size %ldMb bigger than %ldMb limit\n",
+				  size >> 20, max_size >> 20);
+			return -ENOMEM;
+		}
+	}
+retry:
+	r = amdgpu_bo_create(adev, size, alignment, kernel, initial_domain,
+			     flags, NULL, NULL, &robj);
+	if (r) {
+		if (r != -ERESTARTSYS) {
+			if (initial_domain == AMDGPU_GEM_DOMAIN_VRAM) {
+				initial_domain |= AMDGPU_GEM_DOMAIN_GTT;
+				goto retry;
+			}
+			DRM_ERROR("Failed to allocate GEM object (%ld, %d, %u, %d)\n",
+				  size, initial_domain, alignment, r);
+		}
+		return r;
+	}
+	*obj = &robj->gem_base;
+	robj->pid = task_pid_nr(current);
+
+	mutex_lock(&adev->gem.mutex);
+	list_add_tail(&robj->list, &adev->gem.objects);
+	mutex_unlock(&adev->gem.mutex);
+
+	return 0;
+}
+
+int amdgpu_gem_init(struct amdgpu_device *adev)
+{
+	INIT_LIST_HEAD(&adev->gem.objects);
+	return 0;
+}
+
+void amdgpu_gem_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_force_delete(adev);
+}
+
+/*
+ * Call from drm_gem_handle_create which appear in both new and open ioctl
+ * case.
+ */
+int amdgpu_gem_object_open(struct drm_gem_object *obj, struct drm_file *file_priv)
+{
+	struct amdgpu_bo *rbo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = rbo->adev;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_va *bo_va;
+	int r;
+	r = amdgpu_bo_reserve(rbo, false);
+	if (r)
+		return r;
+
+	bo_va = amdgpu_vm_bo_find(vm, rbo);
+	if (!bo_va) {
+		bo_va = amdgpu_vm_bo_add(adev, vm, rbo);
+	} else {
+		++bo_va->ref_count;
+	}
+	amdgpu_bo_unreserve(rbo);
+	return 0;
+}
+
+void amdgpu_gem_object_close(struct drm_gem_object *obj,
+			     struct drm_file *file_priv)
+{
+	struct amdgpu_bo *rbo = gem_to_amdgpu_bo(obj);
+	struct amdgpu_device *adev = rbo->adev;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_va *bo_va;
+	int r;
+	r = amdgpu_bo_reserve(rbo, true);
+	if (r) {
+		dev_err(adev->dev, "leaking bo va because "
+			"we fail to reserve bo (%d)\n", r);
+		return;
+	}
+	bo_va = amdgpu_vm_bo_find(vm, rbo);
+	if (bo_va) {
+		if (--bo_va->ref_count == 0) {
+			amdgpu_vm_bo_rmv(adev, bo_va);
+		}
+	}
+	amdgpu_bo_unreserve(rbo);
+}
+
+static int amdgpu_gem_handle_lockup(struct amdgpu_device *adev, int r)
+{
+	if (r == -EDEADLK) {
+		r = amdgpu_gpu_reset(adev);
+		if (!r)
+			r = -EAGAIN;
+	}
+	return r;
+}
+
+/*
+ * GEM ioctls.
+ */
+int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
+			    struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	union drm_amdgpu_gem_create *args = data;
+	uint64_t size = args->in.bo_size;
+	struct drm_gem_object *gobj;
+	uint32_t handle;
+	bool kernel = false;
+	int r;
+
+	/* create a gem object to contain this object in */
+	if (args->in.domains & (AMDGPU_GEM_DOMAIN_GDS |
+	    AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
+		kernel = true;
+		if (args->in.domains == AMDGPU_GEM_DOMAIN_GDS)
+			size = size << AMDGPU_GDS_SHIFT;
+		else if (args->in.domains == AMDGPU_GEM_DOMAIN_GWS)
+			size = size << AMDGPU_GWS_SHIFT;
+		else if (args->in.domains == AMDGPU_GEM_DOMAIN_OA)
+			size = size << AMDGPU_OA_SHIFT;
+		else {
+			r = -EINVAL;
+			goto error_unlock;
+		}
+	}
+	size = roundup(size, PAGE_SIZE);
+
+	r = amdgpu_gem_object_create(adev, size, args->in.alignment,
+				     (u32)(0xffffffff & args->in.domains),
+				     args->in.domain_flags,
+				     kernel, &gobj);
+	if (r)
+		goto error_unlock;
+
+	r = drm_gem_handle_create(filp, gobj, &handle);
+	/* drop reference from allocate - handle holds it now */
+	drm_gem_object_unreference_unlocked(gobj);
+	if (r)
+		goto error_unlock;
+
+	memset(args, 0, sizeof(*args));
+	args->out.handle = handle;
+	return 0;
+
+error_unlock:
+	r = amdgpu_gem_handle_lockup(adev, r);
+	return r;
+}
+
+int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
+			     struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_gem_userptr *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *bo;
+	uint32_t handle;
+	int r;
+
+	if (offset_in_page(args->addr | args->size))
+		return -EINVAL;
+
+	/* reject unknown flag values */
+	if (args->flags & ~(AMDGPU_GEM_USERPTR_READONLY |
+	    AMDGPU_GEM_USERPTR_ANONONLY | AMDGPU_GEM_USERPTR_VALIDATE |
+	    AMDGPU_GEM_USERPTR_REGISTER))
+		return -EINVAL;
+
+	if (!(args->flags & AMDGPU_GEM_USERPTR_READONLY) && (
+	     !(args->flags & AMDGPU_GEM_USERPTR_ANONONLY) ||
+	     !(args->flags & AMDGPU_GEM_USERPTR_REGISTER))) {
+
+		/* if we want to write to it we must require anonymous
+		   memory and install a MMU notifier */
+		return -EACCES;
+	}
+
+	/* create a gem object to contain this object in */
+	r = amdgpu_gem_object_create(adev, args->size, 0,
+				     AMDGPU_GEM_DOMAIN_CPU, 0,
+				     0, &gobj);
+	if (r)
+		goto handle_lockup;
+
+	bo = gem_to_amdgpu_bo(gobj);
+	r = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
+	if (r)
+		goto release_object;
+
+	if (args->flags & AMDGPU_GEM_USERPTR_REGISTER) {
+		r = amdgpu_mn_register(bo, args->addr);
+		if (r)
+			goto release_object;
+	}
+
+	if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) {
+		down_read(&current->mm->mmap_sem);
+		r = amdgpu_bo_reserve(bo, true);
+		if (r) {
+			up_read(&current->mm->mmap_sem);
+			goto release_object;
+		}
+
+		amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+		amdgpu_bo_unreserve(bo);
+		up_read(&current->mm->mmap_sem);
+		if (r)
+			goto release_object;
+	}
+
+	r = drm_gem_handle_create(filp, gobj, &handle);
+	/* drop reference from allocate - handle holds it now */
+	drm_gem_object_unreference_unlocked(gobj);
+	if (r)
+		goto handle_lockup;
+
+	args->handle = handle;
+	return 0;
+
+release_object:
+	drm_gem_object_unreference_unlocked(gobj);
+
+handle_lockup:
+	r = amdgpu_gem_handle_lockup(adev, r);
+
+	return r;
+}
+
+int amdgpu_mode_dumb_mmap(struct drm_file *filp,
+			  struct drm_device *dev,
+			  uint32_t handle, uint64_t *offset_p)
+{
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+
+	gobj = drm_gem_object_lookup(dev, filp, handle);
+	if (gobj == NULL) {
+		return -ENOENT;
+	}
+	robj = gem_to_amdgpu_bo(gobj);
+	if (amdgpu_ttm_tt_has_userptr(robj->tbo.ttm) ||
+	    (robj->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
+		drm_gem_object_unreference_unlocked(gobj);
+		return -EPERM;
+	}
+	*offset_p = amdgpu_bo_mmap_offset(robj);
+	drm_gem_object_unreference_unlocked(gobj);
+	return 0;
+}
+
+int amdgpu_gem_mmap_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp)
+{
+	union drm_amdgpu_gem_mmap *args = data;
+	uint32_t handle = args->in.handle;
+	memset(args, 0, sizeof(*args));
+	return amdgpu_mode_dumb_mmap(filp, dev, handle, &args->out.addr_ptr);
+}
+
+/**
+ * amdgpu_gem_timeout - calculate jiffies timeout from absolute value
+ *
+ * @timeout_ns: timeout in ns
+ *
+ * Calculate the timeout in jiffies from an absolute timeout in ns.
+ */
+unsigned long amdgpu_gem_timeout(uint64_t timeout_ns)
+{
+	unsigned long timeout_jiffies;
+	ktime_t timeout;
+
+	/* clamp timeout if it's to large */
+	if (((int64_t)timeout_ns) < 0)
+		return MAX_SCHEDULE_TIMEOUT;
+
+	timeout = ktime_sub(ns_to_ktime(timeout_ns), ktime_get());
+	if (ktime_to_ns(timeout) < 0)
+		return 0;
+
+	timeout_jiffies = nsecs_to_jiffies(ktime_to_ns(timeout));
+	/*  clamp timeout to avoid unsigned-> signed overflow */
+	if (timeout_jiffies > MAX_SCHEDULE_TIMEOUT )
+		return MAX_SCHEDULE_TIMEOUT - 1;
+
+	return timeout_jiffies;
+}
+
+int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
+			      struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	union drm_amdgpu_gem_wait_idle *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+	uint32_t handle = args->in.handle;
+	unsigned long timeout = amdgpu_gem_timeout(args->in.timeout);
+	int r = 0;
+	long ret;
+
+	gobj = drm_gem_object_lookup(dev, filp, handle);
+	if (gobj == NULL) {
+		return -ENOENT;
+	}
+	robj = gem_to_amdgpu_bo(gobj);
+	if (timeout == 0)
+		ret = reservation_object_test_signaled_rcu(robj->tbo.resv, true);
+	else
+		ret = reservation_object_wait_timeout_rcu(robj->tbo.resv, true, true, timeout);
+
+	/* ret == 0 means not signaled,
+	 * ret > 0 means signaled
+	 * ret < 0 means interrupted before timeout
+	 */
+	if (ret >= 0) {
+		memset(args, 0, sizeof(*args));
+		args->out.status = (ret == 0);
+	} else
+		r = ret;
+
+	drm_gem_object_unreference_unlocked(gobj);
+	r = amdgpu_gem_handle_lockup(adev, r);
+	return r;
+}
+
+int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
+				struct drm_file *filp)
+{
+	struct drm_amdgpu_gem_metadata *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+	int r = -1;
+
+	DRM_DEBUG("%d \n", args->handle);
+	gobj = drm_gem_object_lookup(dev, filp, args->handle);
+	if (gobj == NULL)
+		return -ENOENT;
+	robj = gem_to_amdgpu_bo(gobj);
+
+	r = amdgpu_bo_reserve(robj, false);
+	if (unlikely(r != 0))
+		goto out;
+
+	if (args->op == AMDGPU_GEM_METADATA_OP_GET_METADATA) {
+		amdgpu_bo_get_tiling_flags(robj, &args->data.tiling_info);
+		r = amdgpu_bo_get_metadata(robj, args->data.data,
+					   sizeof(args->data.data),
+					   &args->data.data_size_bytes,
+					   &args->data.flags);
+	} else if (args->op == AMDGPU_GEM_METADATA_OP_SET_METADATA) {
+		if (args->data.data_size_bytes > sizeof(args->data.data)) {
+			r = -EINVAL;
+			goto unreserve;
+		}
+		r = amdgpu_bo_set_tiling_flags(robj, args->data.tiling_info);
+		if (!r)
+			r = amdgpu_bo_set_metadata(robj, args->data.data,
+						   args->data.data_size_bytes,
+						   args->data.flags);
+	}
+
+unreserve:
+	amdgpu_bo_unreserve(robj);
+out:
+	drm_gem_object_unreference_unlocked(gobj);
+	return r;
+}
+
+/**
+ * amdgpu_gem_va_update_vm -update the bo_va in its VM
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to update
+ *
+ * Update the bo_va directly after setting it's address. Errors are not
+ * vital here, so they are not reported back to userspace.
+ */
+static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
+				    struct amdgpu_bo_va *bo_va, uint32_t operation)
+{
+	struct ttm_validate_buffer tv, *entry;
+	struct amdgpu_bo_list_entry *vm_bos;
+	struct ww_acquire_ctx ticket;
+	struct list_head list, duplicates;
+	unsigned domain;
+	int r;
+
+	INIT_LIST_HEAD(&list);
+	INIT_LIST_HEAD(&duplicates);
+
+	tv.bo = &bo_va->bo->tbo;
+	tv.shared = true;
+	list_add(&tv.head, &list);
+
+	vm_bos = amdgpu_vm_get_bos(adev, bo_va->vm, &list);
+	if (!vm_bos)
+		return;
+
+	/* Provide duplicates to avoid -EALREADY */
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates);
+	if (r)
+		goto error_free;
+
+	list_for_each_entry(entry, &list, head) {
+		domain = amdgpu_mem_type_to_domain(entry->bo->mem.mem_type);
+		/* if anything is swapped out don't swap it in here,
+		   just abort and wait for the next CS */
+		if (domain == AMDGPU_GEM_DOMAIN_CPU)
+			goto error_unreserve;
+	}
+	list_for_each_entry(entry, &duplicates, head) {
+		domain = amdgpu_mem_type_to_domain(entry->bo->mem.mem_type);
+		/* if anything is swapped out don't swap it in here,
+		   just abort and wait for the next CS */
+		if (domain == AMDGPU_GEM_DOMAIN_CPU)
+			goto error_unreserve;
+	}
+
+	r = amdgpu_vm_update_page_directory(adev, bo_va->vm);
+	if (r)
+		goto error_unreserve;
+
+	r = amdgpu_vm_clear_freed(adev, bo_va->vm);
+	if (r)
+		goto error_unreserve;
+
+	if (operation == AMDGPU_VA_OP_MAP)
+		r = amdgpu_vm_bo_update(adev, bo_va, &bo_va->bo->tbo.mem);
+
+error_unreserve:
+	ttm_eu_backoff_reservation(&ticket, &list);
+
+error_free:
+	drm_free_large(vm_bos);
+
+	if (r && r != -ERESTARTSYS)
+		DRM_ERROR("Couldn't update BO_VA (%d)\n", r);
+}
+
+
+
+int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *filp)
+{
+	struct drm_amdgpu_gem_va *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = filp->driver_priv;
+	struct amdgpu_bo *rbo;
+	struct amdgpu_bo_va *bo_va;
+	struct ttm_validate_buffer tv, tv_pd;
+	struct ww_acquire_ctx ticket;
+	struct list_head list, duplicates;
+	uint32_t invalid_flags, va_flags = 0;
+	int r = 0;
+
+	if (!adev->vm_manager.enabled)
+		return -ENOTTY;
+
+	if (args->va_address < AMDGPU_VA_RESERVED_SIZE) {
+		dev_err(&dev->pdev->dev,
+			"va_address 0x%lX is in reserved area 0x%X\n",
+			(unsigned long)args->va_address,
+			AMDGPU_VA_RESERVED_SIZE);
+		return -EINVAL;
+	}
+
+	invalid_flags = ~(AMDGPU_VM_DELAY_UPDATE | AMDGPU_VM_PAGE_READABLE |
+			AMDGPU_VM_PAGE_WRITEABLE | AMDGPU_VM_PAGE_EXECUTABLE);
+	if ((args->flags & invalid_flags)) {
+		dev_err(&dev->pdev->dev, "invalid flags 0x%08X vs 0x%08X\n",
+			args->flags, invalid_flags);
+		return -EINVAL;
+	}
+
+	switch (args->operation) {
+	case AMDGPU_VA_OP_MAP:
+	case AMDGPU_VA_OP_UNMAP:
+		break;
+	default:
+		dev_err(&dev->pdev->dev, "unsupported operation %d\n",
+			args->operation);
+		return -EINVAL;
+	}
+
+	gobj = drm_gem_object_lookup(dev, filp, args->handle);
+	if (gobj == NULL)
+		return -ENOENT;
+	rbo = gem_to_amdgpu_bo(gobj);
+	INIT_LIST_HEAD(&list);
+	INIT_LIST_HEAD(&duplicates);
+	tv.bo = &rbo->tbo;
+	tv.shared = true;
+	list_add(&tv.head, &list);
+
+	if (args->operation == AMDGPU_VA_OP_MAP) {
+		tv_pd.bo = &fpriv->vm.page_directory->tbo;
+		tv_pd.shared = true;
+		list_add(&tv_pd.head, &list);
+	}
+	r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates);
+	if (r) {
+		drm_gem_object_unreference_unlocked(gobj);
+		return r;
+	}
+
+	bo_va = amdgpu_vm_bo_find(&fpriv->vm, rbo);
+	if (!bo_va) {
+		ttm_eu_backoff_reservation(&ticket, &list);
+		drm_gem_object_unreference_unlocked(gobj);
+		return -ENOENT;
+	}
+
+	switch (args->operation) {
+	case AMDGPU_VA_OP_MAP:
+		if (args->flags & AMDGPU_VM_PAGE_READABLE)
+			va_flags |= AMDGPU_PTE_READABLE;
+		if (args->flags & AMDGPU_VM_PAGE_WRITEABLE)
+			va_flags |= AMDGPU_PTE_WRITEABLE;
+		if (args->flags & AMDGPU_VM_PAGE_EXECUTABLE)
+			va_flags |= AMDGPU_PTE_EXECUTABLE;
+		r = amdgpu_vm_bo_map(adev, bo_va, args->va_address,
+				     args->offset_in_bo, args->map_size,
+				     va_flags);
+		break;
+	case AMDGPU_VA_OP_UNMAP:
+		r = amdgpu_vm_bo_unmap(adev, bo_va, args->va_address);
+		break;
+	default:
+		break;
+	}
+	ttm_eu_backoff_reservation(&ticket, &list);
+	if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE))
+		amdgpu_gem_va_update_vm(adev, bo_va, args->operation);
+
+	drm_gem_object_unreference_unlocked(gobj);
+	return r;
+}
+
+int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
+			struct drm_file *filp)
+{
+	struct drm_amdgpu_gem_op *args = data;
+	struct drm_gem_object *gobj;
+	struct amdgpu_bo *robj;
+	int r;
+
+	gobj = drm_gem_object_lookup(dev, filp, args->handle);
+	if (gobj == NULL) {
+		return -ENOENT;
+	}
+	robj = gem_to_amdgpu_bo(gobj);
+
+	r = amdgpu_bo_reserve(robj, false);
+	if (unlikely(r))
+		goto out;
+
+	switch (args->op) {
+	case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
+		struct drm_amdgpu_gem_create_in info;
+		void __user *out = (void __user *)(long)args->value;
+
+		info.bo_size = robj->gem_base.size;
+		info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
+		info.domains = robj->initial_domain;
+		info.domain_flags = robj->flags;
+		amdgpu_bo_unreserve(robj);
+		if (copy_to_user(out, &info, sizeof(info)))
+			r = -EFAULT;
+		break;
+	}
+	case AMDGPU_GEM_OP_SET_PLACEMENT:
+		if (amdgpu_ttm_tt_has_userptr(robj->tbo.ttm)) {
+			r = -EPERM;
+			amdgpu_bo_unreserve(robj);
+			break;
+		}
+		robj->initial_domain = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
+						      AMDGPU_GEM_DOMAIN_GTT |
+						      AMDGPU_GEM_DOMAIN_CPU);
+		amdgpu_bo_unreserve(robj);
+		break;
+	default:
+		amdgpu_bo_unreserve(robj);
+		r = -EINVAL;
+	}
+
+out:
+	drm_gem_object_unreference_unlocked(gobj);
+	return r;
+}
+
+int amdgpu_mode_dumb_create(struct drm_file *file_priv,
+			    struct drm_device *dev,
+			    struct drm_mode_create_dumb *args)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_gem_object *gobj;
+	uint32_t handle;
+	int r;
+
+	args->pitch = amdgpu_align_pitch(adev, args->width, args->bpp, 0) * ((args->bpp + 1) / 8);
+	args->size = (u64)args->pitch * args->height;
+	args->size = ALIGN(args->size, PAGE_SIZE);
+
+	r = amdgpu_gem_object_create(adev, args->size, 0,
+				     AMDGPU_GEM_DOMAIN_VRAM,
+				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+				     ttm_bo_type_device,
+				     &gobj);
+	if (r)
+		return -ENOMEM;
+
+	r = drm_gem_handle_create(file_priv, gobj, &handle);
+	/* drop reference from allocate - handle holds it now */
+	drm_gem_object_unreference_unlocked(gobj);
+	if (r) {
+		return r;
+	}
+	args->handle = handle;
+	return 0;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_debugfs_gem_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_bo *rbo;
+	unsigned i = 0;
+
+	mutex_lock(&adev->gem.mutex);
+	list_for_each_entry(rbo, &adev->gem.objects, list) {
+		unsigned domain;
+		const char *placement;
+
+		domain = amdgpu_mem_type_to_domain(rbo->tbo.mem.mem_type);
+		switch (domain) {
+		case AMDGPU_GEM_DOMAIN_VRAM:
+			placement = "VRAM";
+			break;
+		case AMDGPU_GEM_DOMAIN_GTT:
+			placement = " GTT";
+			break;
+		case AMDGPU_GEM_DOMAIN_CPU:
+		default:
+			placement = " CPU";
+			break;
+		}
+		seq_printf(m, "bo[0x%08x] %8ldkB %8ldMB %s pid %8ld\n",
+			   i, amdgpu_bo_size(rbo) >> 10, amdgpu_bo_size(rbo) >> 20,
+			   placement, (unsigned long)rbo->pid);
+		i++;
+	}
+	mutex_unlock(&adev->gem.mutex);
+	return 0;
+}
+
+static struct drm_info_list amdgpu_debugfs_gem_list[] = {
+	{"amdgpu_gem_info", &amdgpu_debugfs_gem_info, 0, NULL},
+};
+#endif
+
+int amdgpu_gem_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_gem_list, 1);
+#endif
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
new file mode 100644
index 000000000..9f95da4f0
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -0,0 +1,72 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+
+/*
+ * GPU scratch registers helpers function.
+ */
+/**
+ * amdgpu_gfx_scratch_get - Allocate a scratch register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: scratch register mmio offset
+ *
+ * Allocate a CP scratch register for use by the driver (all asics).
+ * Returns 0 on success or -EINVAL on failure.
+ */
+int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.scratch.num_reg; i++) {
+		if (adev->gfx.scratch.free[i]) {
+			adev->gfx.scratch.free[i] = false;
+			*reg = adev->gfx.scratch.reg[i];
+			return 0;
+		}
+	}
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_gfx_scratch_free - Free a scratch register
+ *
+ * @adev: amdgpu_device pointer
+ * @reg: scratch register mmio offset
+ *
+ * Free a CP scratch register allocated for use by the driver (all asics)
+ */
+void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.scratch.num_reg; i++) {
+		if (adev->gfx.scratch.reg[i] == reg) {
+			adev->gfx.scratch.free[i] = true;
+			return;
+		}
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
new file mode 100644
index 000000000..dc06cbda7
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_GFX_H__
+#define __AMDGPU_GFX_H__
+
+int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg);
+void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
new file mode 100644
index 000000000..31a676376
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c
@@ -0,0 +1,395 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <linux/export.h>
+
+#include <drm/drmP.h>
+#include <drm/drm_edid.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_i2c.h"
+#include "amdgpu_atombios.h"
+#include "atom.h"
+#include "atombios_dp.h"
+#include "atombios_i2c.h"
+
+/* bit banging i2c */
+static int amdgpu_i2c_pre_xfer(struct i2c_adapter *i2c_adap)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t temp;
+
+	mutex_lock(&i2c->mutex);
+
+	/* switch the pads to ddc mode */
+	if (rec->hw_capable) {
+		temp = RREG32(rec->mask_clk_reg);
+		temp &= ~(1 << 16);
+		WREG32(rec->mask_clk_reg, temp);
+	}
+
+	/* clear the output pin values */
+	temp = RREG32(rec->a_clk_reg) & ~rec->a_clk_mask;
+	WREG32(rec->a_clk_reg, temp);
+
+	temp = RREG32(rec->a_data_reg) & ~rec->a_data_mask;
+	WREG32(rec->a_data_reg, temp);
+
+	/* set the pins to input */
+	temp = RREG32(rec->en_clk_reg) & ~rec->en_clk_mask;
+	WREG32(rec->en_clk_reg, temp);
+
+	temp = RREG32(rec->en_data_reg) & ~rec->en_data_mask;
+	WREG32(rec->en_data_reg, temp);
+
+	/* mask the gpio pins for software use */
+	temp = RREG32(rec->mask_clk_reg) | rec->mask_clk_mask;
+	WREG32(rec->mask_clk_reg, temp);
+	temp = RREG32(rec->mask_clk_reg);
+
+	temp = RREG32(rec->mask_data_reg) | rec->mask_data_mask;
+	WREG32(rec->mask_data_reg, temp);
+	temp = RREG32(rec->mask_data_reg);
+
+	return 0;
+}
+
+static void amdgpu_i2c_post_xfer(struct i2c_adapter *i2c_adap)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t temp;
+
+	/* unmask the gpio pins for software use */
+	temp = RREG32(rec->mask_clk_reg) & ~rec->mask_clk_mask;
+	WREG32(rec->mask_clk_reg, temp);
+	temp = RREG32(rec->mask_clk_reg);
+
+	temp = RREG32(rec->mask_data_reg) & ~rec->mask_data_mask;
+	WREG32(rec->mask_data_reg, temp);
+	temp = RREG32(rec->mask_data_reg);
+
+	mutex_unlock(&i2c->mutex);
+}
+
+static int amdgpu_i2c_get_clock(void *i2c_priv)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* read the value off the pin */
+	val = RREG32(rec->y_clk_reg);
+	val &= rec->y_clk_mask;
+
+	return (val != 0);
+}
+
+
+static int amdgpu_i2c_get_data(void *i2c_priv)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* read the value off the pin */
+	val = RREG32(rec->y_data_reg);
+	val &= rec->y_data_mask;
+
+	return (val != 0);
+}
+
+static void amdgpu_i2c_set_clock(void *i2c_priv, int clock)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* set pin direction */
+	val = RREG32(rec->en_clk_reg) & ~rec->en_clk_mask;
+	val |= clock ? 0 : rec->en_clk_mask;
+	WREG32(rec->en_clk_reg, val);
+}
+
+static void amdgpu_i2c_set_data(void *i2c_priv, int data)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_priv;
+	struct amdgpu_device *adev = i2c->dev->dev_private;
+	struct amdgpu_i2c_bus_rec *rec = &i2c->rec;
+	uint32_t val;
+
+	/* set pin direction */
+	val = RREG32(rec->en_data_reg) & ~rec->en_data_mask;
+	val |= data ? 0 : rec->en_data_mask;
+	WREG32(rec->en_data_reg, val);
+}
+
+static const struct i2c_algorithm amdgpu_atombios_i2c_algo = {
+	.master_xfer = amdgpu_atombios_i2c_xfer,
+	.functionality = amdgpu_atombios_i2c_func,
+};
+
+struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
+					    struct amdgpu_i2c_bus_rec *rec,
+					    const char *name)
+{
+	struct amdgpu_i2c_chan *i2c;
+	int ret;
+
+	/* don't add the mm_i2c bus unless hw_i2c is enabled */
+	if (rec->mm_i2c && (amdgpu_hw_i2c == 0))
+		return NULL;
+
+	i2c = kzalloc(sizeof(struct amdgpu_i2c_chan), GFP_KERNEL);
+	if (i2c == NULL)
+		return NULL;
+
+	i2c->rec = *rec;
+	i2c->adapter.owner = THIS_MODULE;
+	i2c->adapter.class = I2C_CLASS_DDC;
+	i2c->adapter.dev.parent = &dev->pdev->dev;
+	i2c->dev = dev;
+	i2c_set_adapdata(&i2c->adapter, i2c);
+	mutex_init(&i2c->mutex);
+	if (rec->hw_capable &&
+	    amdgpu_hw_i2c) {
+		/* hw i2c using atom */
+		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+			 "AMDGPU i2c hw bus %s", name);
+		i2c->adapter.algo = &amdgpu_atombios_i2c_algo;
+		ret = i2c_add_adapter(&i2c->adapter);
+		if (ret) {
+			DRM_ERROR("Failed to register hw i2c %s\n", name);
+			goto out_free;
+		}
+	} else {
+		/* set the amdgpu bit adapter */
+		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+			 "AMDGPU i2c bit bus %s", name);
+		i2c->adapter.algo_data = &i2c->bit;
+		i2c->bit.pre_xfer = amdgpu_i2c_pre_xfer;
+		i2c->bit.post_xfer = amdgpu_i2c_post_xfer;
+		i2c->bit.setsda = amdgpu_i2c_set_data;
+		i2c->bit.setscl = amdgpu_i2c_set_clock;
+		i2c->bit.getsda = amdgpu_i2c_get_data;
+		i2c->bit.getscl = amdgpu_i2c_get_clock;
+		i2c->bit.udelay = 10;
+		i2c->bit.timeout = usecs_to_jiffies(2200);	/* from VESA */
+		i2c->bit.data = i2c;
+		ret = i2c_bit_add_bus(&i2c->adapter);
+		if (ret) {
+			DRM_ERROR("Failed to register bit i2c %s\n", name);
+			goto out_free;
+		}
+	}
+
+	return i2c;
+out_free:
+	kfree(i2c);
+	return NULL;
+
+}
+
+void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c)
+{
+	if (!i2c)
+		return;
+	i2c_del_adapter(&i2c->adapter);
+	kfree(i2c);
+}
+
+/* Add the default buses */
+void amdgpu_i2c_init(struct amdgpu_device *adev)
+{
+	if (amdgpu_hw_i2c)
+		DRM_INFO("hw_i2c forced on, you may experience display detection problems!\n");
+
+	if (adev->is_atom_bios)
+		amdgpu_atombios_i2c_init(adev);
+}
+
+/* remove all the buses */
+void amdgpu_i2c_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
+		if (adev->i2c_bus[i]) {
+			amdgpu_i2c_destroy(adev->i2c_bus[i]);
+			adev->i2c_bus[i] = NULL;
+		}
+	}
+}
+
+/* Add additional buses */
+void amdgpu_i2c_add(struct amdgpu_device *adev,
+		     struct amdgpu_i2c_bus_rec *rec,
+		     const char *name)
+{
+	struct drm_device *dev = adev->ddev;
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
+		if (!adev->i2c_bus[i]) {
+			adev->i2c_bus[i] = amdgpu_i2c_create(dev, rec, name);
+			return;
+		}
+	}
+}
+
+/* looks up bus based on id */
+struct amdgpu_i2c_chan *
+amdgpu_i2c_lookup(struct amdgpu_device *adev,
+		   struct amdgpu_i2c_bus_rec *i2c_bus)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_I2C_BUS; i++) {
+		if (adev->i2c_bus[i] &&
+		    (adev->i2c_bus[i]->rec.i2c_id == i2c_bus->i2c_id)) {
+			return adev->i2c_bus[i];
+		}
+	}
+	return NULL;
+}
+
+static void amdgpu_i2c_get_byte(struct amdgpu_i2c_chan *i2c_bus,
+				 u8 slave_addr,
+				 u8 addr,
+				 u8 *val)
+{
+	u8 out_buf[2];
+	u8 in_buf[2];
+	struct i2c_msg msgs[] = {
+		{
+			.addr = slave_addr,
+			.flags = 0,
+			.len = 1,
+			.buf = out_buf,
+		},
+		{
+			.addr = slave_addr,
+			.flags = I2C_M_RD,
+			.len = 1,
+			.buf = in_buf,
+		}
+	};
+
+	out_buf[0] = addr;
+	out_buf[1] = 0;
+
+	if (i2c_transfer(&i2c_bus->adapter, msgs, 2) == 2) {
+		*val = in_buf[0];
+		DRM_DEBUG("val = 0x%02x\n", *val);
+	} else {
+		DRM_DEBUG("i2c 0x%02x 0x%02x read failed\n",
+			  addr, *val);
+	}
+}
+
+static void amdgpu_i2c_put_byte(struct amdgpu_i2c_chan *i2c_bus,
+				 u8 slave_addr,
+				 u8 addr,
+				 u8 val)
+{
+	uint8_t out_buf[2];
+	struct i2c_msg msg = {
+		.addr = slave_addr,
+		.flags = 0,
+		.len = 2,
+		.buf = out_buf,
+	};
+
+	out_buf[0] = addr;
+	out_buf[1] = val;
+
+	if (i2c_transfer(&i2c_bus->adapter, &msg, 1) != 1)
+		DRM_DEBUG("i2c 0x%02x 0x%02x write failed\n",
+			  addr, val);
+}
+
+/* ddc router switching */
+void
+amdgpu_i2c_router_select_ddc_port(struct amdgpu_connector *amdgpu_connector)
+{
+	u8 val;
+
+	if (!amdgpu_connector->router.ddc_valid)
+		return;
+
+	if (!amdgpu_connector->router_bus)
+		return;
+
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, &val);
+	val &= ~amdgpu_connector->router.ddc_mux_control_pin;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, val);
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, &val);
+	val &= ~amdgpu_connector->router.ddc_mux_control_pin;
+	val |= amdgpu_connector->router.ddc_mux_state;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, val);
+}
+
+/* clock/data router switching */
+void
+amdgpu_i2c_router_select_cd_port(struct amdgpu_connector *amdgpu_connector)
+{
+	u8 val;
+
+	if (!amdgpu_connector->router.cd_valid)
+		return;
+
+	if (!amdgpu_connector->router_bus)
+		return;
+
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, &val);
+	val &= ~amdgpu_connector->router.cd_mux_control_pin;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x3, val);
+	amdgpu_i2c_get_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, &val);
+	val &= ~amdgpu_connector->router.cd_mux_control_pin;
+	val |= amdgpu_connector->router.cd_mux_state;
+	amdgpu_i2c_put_byte(amdgpu_connector->router_bus,
+			    amdgpu_connector->router.i2c_addr,
+			    0x1, val);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
new file mode 100644
index 000000000..d81e19b53
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_I2C_H__
+#define __AMDGPU_I2C_H__
+
+struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev,
+					    struct amdgpu_i2c_bus_rec *rec,
+					    const char *name);
+void amdgpu_i2c_destroy(struct amdgpu_i2c_chan *i2c);
+void amdgpu_i2c_init(struct amdgpu_device *adev);
+void amdgpu_i2c_fini(struct amdgpu_device *adev);
+void amdgpu_i2c_add(struct amdgpu_device *adev,
+		     struct amdgpu_i2c_bus_rec *rec,
+		     const char *name);
+struct amdgpu_i2c_chan *
+amdgpu_i2c_lookup(struct amdgpu_device *adev,
+		   struct amdgpu_i2c_bus_rec *i2c_bus);
+void
+amdgpu_i2c_router_select_ddc_port(struct amdgpu_connector *amdgpu_connector);
+void
+amdgpu_i2c_router_select_cd_port(struct amdgpu_connector *amdgpu_connector);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
new file mode 100644
index 000000000..9e25edafa
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -0,0 +1,348 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ *          Christian König
+ */
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+
+/*
+ * IB
+ * IBs (Indirect Buffers) and areas of GPU accessible memory where
+ * commands are stored.  You can put a pointer to the IB in the
+ * command ring and the hw will fetch the commands from the IB
+ * and execute them.  Generally userspace acceleration drivers
+ * produce command buffers which are send to the kernel and
+ * put in IBs for execution by the requested ring.
+ */
+static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev);
+
+/**
+ * amdgpu_ib_get - request an IB (Indirect Buffer)
+ *
+ * @ring: ring index the IB is associated with
+ * @size: requested IB size
+ * @ib: IB object returned
+ *
+ * Request an IB (all asics).  IBs are allocated using the
+ * suballocator.
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ib_get(struct amdgpu_ring *ring, struct amdgpu_vm *vm,
+		  unsigned size, struct amdgpu_ib *ib)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int r;
+
+	if (size) {
+		r = amdgpu_sa_bo_new(&adev->ring_tmp_bo,
+				      &ib->sa_bo, size, 256);
+		if (r) {
+			dev_err(adev->dev, "failed to get a new IB (%d)\n", r);
+			return r;
+		}
+
+		ib->ptr = amdgpu_sa_bo_cpu_addr(ib->sa_bo);
+
+		if (!vm)
+			ib->gpu_addr = amdgpu_sa_bo_gpu_addr(ib->sa_bo);
+	}
+
+	amdgpu_sync_create(&ib->sync);
+
+	ib->ring = ring;
+	ib->vm = vm;
+
+	return 0;
+}
+
+/**
+ * amdgpu_ib_free - free an IB (Indirect Buffer)
+ *
+ * @adev: amdgpu_device pointer
+ * @ib: IB object to free
+ *
+ * Free an IB (all asics).
+ */
+void amdgpu_ib_free(struct amdgpu_device *adev, struct amdgpu_ib *ib)
+{
+	amdgpu_sync_free(adev, &ib->sync, &ib->fence->base);
+	amdgpu_sa_bo_free(adev, &ib->sa_bo, &ib->fence->base);
+	if (ib->fence)
+		fence_put(&ib->fence->base);
+}
+
+/**
+ * amdgpu_ib_schedule - schedule an IB (Indirect Buffer) on the ring
+ *
+ * @adev: amdgpu_device pointer
+ * @num_ibs: number of IBs to schedule
+ * @ibs: IB objects to schedule
+ * @owner: owner for creating the fences
+ *
+ * Schedule an IB on the associated ring (all asics).
+ * Returns 0 on success, error on failure.
+ *
+ * On SI, there are two parallel engines fed from the primary ring,
+ * the CE (Constant Engine) and the DE (Drawing Engine).  Since
+ * resource descriptors have moved to memory, the CE allows you to
+ * prime the caches while the DE is updating register state so that
+ * the resource descriptors will be already in cache when the draw is
+ * processed.  To accomplish this, the userspace driver submits two
+ * IBs, one for the CE and one for the DE.  If there is a CE IB (called
+ * a CONST_IB), it will be put on the ring prior to the DE IB.  Prior
+ * to SI there was just a DE IB.
+ */
+int amdgpu_ib_schedule(struct amdgpu_device *adev, unsigned num_ibs,
+		       struct amdgpu_ib *ibs, void *owner)
+{
+	struct amdgpu_ib *ib = &ibs[0];
+	struct amdgpu_ring *ring;
+	struct amdgpu_ctx *ctx, *old_ctx;
+	struct amdgpu_vm *vm;
+	unsigned i;
+	int r = 0;
+
+	if (num_ibs == 0)
+		return -EINVAL;
+
+	ring = ibs->ring;
+	ctx = ibs->ctx;
+	vm = ibs->vm;
+
+	if (!ring->ready) {
+		dev_err(adev->dev, "couldn't schedule ib\n");
+		return -EINVAL;
+	}
+	r = amdgpu_sync_wait(&ibs->sync);
+	if (r) {
+		dev_err(adev->dev, "IB sync failed (%d).\n", r);
+		return r;
+	}
+	r = amdgpu_ring_lock(ring, (256 + AMDGPU_NUM_SYNCS * 8) * num_ibs);
+	if (r) {
+		dev_err(adev->dev, "scheduling IB failed (%d).\n", r);
+		return r;
+	}
+
+	if (vm) {
+		/* grab a vm id if necessary */
+		r = amdgpu_vm_grab_id(ibs->vm, ibs->ring, &ibs->sync);
+		if (r) {
+			amdgpu_ring_unlock_undo(ring);
+			return r;
+		}
+	}
+
+	r = amdgpu_sync_rings(&ibs->sync, ring);
+	if (r) {
+		amdgpu_ring_unlock_undo(ring);
+		dev_err(adev->dev, "failed to sync rings (%d)\n", r);
+		return r;
+	}
+
+	if (vm) {
+		/* do context switch */
+		amdgpu_vm_flush(ring, vm, ib->sync.last_vm_update);
+
+		if (ring->funcs->emit_gds_switch)
+			amdgpu_ring_emit_gds_switch(ring, ib->vm->ids[ring->idx].id,
+						    ib->gds_base, ib->gds_size,
+						    ib->gws_base, ib->gws_size,
+						    ib->oa_base, ib->oa_size);
+
+		if (ring->funcs->emit_hdp_flush)
+			amdgpu_ring_emit_hdp_flush(ring);
+	}
+
+	old_ctx = ring->current_ctx;
+	for (i = 0; i < num_ibs; ++i) {
+		ib = &ibs[i];
+
+		if (ib->ring != ring || ib->ctx != ctx || ib->vm != vm) {
+			ring->current_ctx = old_ctx;
+			amdgpu_ring_unlock_undo(ring);
+			return -EINVAL;
+		}
+		amdgpu_ring_emit_ib(ring, ib);
+		ring->current_ctx = ctx;
+	}
+
+	r = amdgpu_fence_emit(ring, owner, &ib->fence);
+	if (r) {
+		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
+		ring->current_ctx = old_ctx;
+		amdgpu_ring_unlock_undo(ring);
+		return r;
+	}
+
+	if (!amdgpu_enable_scheduler && ib->ctx)
+		ib->sequence = amdgpu_ctx_add_fence(ib->ctx, ring,
+						    &ib->fence->base);
+
+	/* wrap the last IB with fence */
+	if (ib->user) {
+		uint64_t addr = amdgpu_bo_gpu_offset(ib->user->bo);
+		addr += ib->user->offset;
+		amdgpu_ring_emit_fence(ring, addr, ib->sequence,
+				       AMDGPU_FENCE_FLAG_64BIT);
+	}
+
+	if (ib->vm)
+		amdgpu_vm_fence(adev, ib->vm, &ib->fence->base);
+
+	amdgpu_ring_unlock_commit(ring);
+	return 0;
+}
+
+/**
+ * amdgpu_ib_pool_init - Init the IB (Indirect Buffer) pool
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the suballocator to manage a pool of memory
+ * for use as IBs (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ib_pool_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->ib_pool_ready) {
+		return 0;
+	}
+	r = amdgpu_sa_bo_manager_init(adev, &adev->ring_tmp_bo,
+				      AMDGPU_IB_POOL_SIZE*64*1024,
+				      AMDGPU_GPU_PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT);
+	if (r) {
+		return r;
+	}
+
+	r = amdgpu_sa_bo_manager_start(adev, &adev->ring_tmp_bo);
+	if (r) {
+		return r;
+	}
+
+	adev->ib_pool_ready = true;
+	if (amdgpu_debugfs_sa_init(adev)) {
+		dev_err(adev->dev, "failed to register debugfs file for SA\n");
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_ib_pool_fini - Free the IB (Indirect Buffer) pool
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the suballocator managing the pool of memory
+ * for use as IBs (all asics).
+ */
+void amdgpu_ib_pool_fini(struct amdgpu_device *adev)
+{
+	if (adev->ib_pool_ready) {
+		amdgpu_sa_bo_manager_suspend(adev, &adev->ring_tmp_bo);
+		amdgpu_sa_bo_manager_fini(adev, &adev->ring_tmp_bo);
+		adev->ib_pool_ready = false;
+	}
+}
+
+/**
+ * amdgpu_ib_ring_tests - test IBs on the rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Test an IB (Indirect Buffer) on each ring.
+ * If the test fails, disable the ring.
+ * Returns 0 on success, error if the primary GFX ring
+ * IB test fails.
+ */
+int amdgpu_ib_ring_tests(struct amdgpu_device *adev)
+{
+	unsigned i;
+	int r;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->ready)
+			continue;
+
+		r = amdgpu_ring_test_ib(ring);
+		if (r) {
+			ring->ready = false;
+
+			if (ring == &adev->gfx.gfx_ring[0]) {
+				/* oh, oh, that's really bad */
+				DRM_ERROR("amdgpu: failed testing IB on GFX ring (%d).\n", r);
+				adev->accel_working = false;
+				return r;
+
+			} else {
+				/* still not good, but we can live with it */
+				DRM_ERROR("amdgpu: failed testing IB on ring %d (%d).\n", i, r);
+			}
+		}
+	}
+	return 0;
+}
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_sa_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	amdgpu_sa_bo_dump_debug_info(&adev->ring_tmp_bo, m);
+
+	return 0;
+
+}
+
+static struct drm_info_list amdgpu_debugfs_sa_list[] = {
+	{"amdgpu_sa_info", &amdgpu_debugfs_sa_info, 0, NULL},
+};
+
+#endif
+
+static int amdgpu_debugfs_sa_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_sa_list, 1);
+#else
+	return 0;
+#endif
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
new file mode 100644
index 000000000..534fc04e8
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
@@ -0,0 +1,219 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_amdkfd.h"
+
+/**
+ * amdgpu_ih_ring_alloc - allocate memory for the IH ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller.
+ * Returns 0 for success, errors for failure.
+ */
+static int amdgpu_ih_ring_alloc(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* Allocate ring buffer */
+	if (adev->irq.ih.ring_obj == NULL) {
+		r = amdgpu_bo_create(adev, adev->irq.ih.ring_size,
+				     PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_GTT, 0,
+				     NULL, NULL, &adev->irq.ih.ring_obj);
+		if (r) {
+			DRM_ERROR("amdgpu: failed to create ih ring buffer (%d).\n", r);
+			return r;
+		}
+		r = amdgpu_bo_reserve(adev->irq.ih.ring_obj, false);
+		if (unlikely(r != 0))
+			return r;
+		r = amdgpu_bo_pin(adev->irq.ih.ring_obj,
+				  AMDGPU_GEM_DOMAIN_GTT,
+				  &adev->irq.ih.gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(adev->irq.ih.ring_obj);
+			DRM_ERROR("amdgpu: failed to pin ih ring buffer (%d).\n", r);
+			return r;
+		}
+		r = amdgpu_bo_kmap(adev->irq.ih.ring_obj,
+				   (void **)&adev->irq.ih.ring);
+		amdgpu_bo_unreserve(adev->irq.ih.ring_obj);
+		if (r) {
+			DRM_ERROR("amdgpu: failed to map ih ring buffer (%d).\n", r);
+			return r;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_ih_ring_init - initialize the IH state
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initializes the IH state and allocates a buffer
+ * for the IH ring buffer.
+ * Returns 0 for success, errors for failure.
+ */
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
+			bool use_bus_addr)
+{
+	u32 rb_bufsz;
+	int r;
+
+	/* Align ring size */
+	rb_bufsz = order_base_2(ring_size / 4);
+	ring_size = (1 << rb_bufsz) * 4;
+	adev->irq.ih.ring_size = ring_size;
+	adev->irq.ih.ptr_mask = adev->irq.ih.ring_size - 1;
+	adev->irq.ih.rptr = 0;
+	adev->irq.ih.use_bus_addr = use_bus_addr;
+
+	if (adev->irq.ih.use_bus_addr) {
+		if (!adev->irq.ih.ring) {
+			/* add 8 bytes for the rptr/wptr shadows and
+			 * add them to the end of the ring allocation.
+			 */
+			adev->irq.ih.ring = pci_alloc_consistent(adev->pdev,
+								 adev->irq.ih.ring_size + 8,
+								 &adev->irq.ih.rb_dma_addr);
+			if (adev->irq.ih.ring == NULL)
+				return -ENOMEM;
+			memset((void *)adev->irq.ih.ring, 0, adev->irq.ih.ring_size + 8);
+			adev->irq.ih.wptr_offs = (adev->irq.ih.ring_size / 4) + 0;
+			adev->irq.ih.rptr_offs = (adev->irq.ih.ring_size / 4) + 1;
+		}
+		return 0;
+	} else {
+		r = amdgpu_wb_get(adev, &adev->irq.ih.wptr_offs);
+		if (r) {
+			dev_err(adev->dev, "(%d) ih wptr_offs wb alloc failed\n", r);
+			return r;
+		}
+
+		r = amdgpu_wb_get(adev, &adev->irq.ih.rptr_offs);
+		if (r) {
+			amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
+			dev_err(adev->dev, "(%d) ih rptr_offs wb alloc failed\n", r);
+			return r;
+		}
+
+		return amdgpu_ih_ring_alloc(adev);
+	}
+}
+
+/**
+ * amdgpu_ih_ring_fini - tear down the IH state
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the IH state and frees buffer
+ * used for the IH ring buffer.
+ */
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->irq.ih.use_bus_addr) {
+		if (adev->irq.ih.ring) {
+			/* add 8 bytes for the rptr/wptr shadows and
+			 * add them to the end of the ring allocation.
+			 */
+			pci_free_consistent(adev->pdev, adev->irq.ih.ring_size + 8,
+					    (void *)adev->irq.ih.ring,
+					    adev->irq.ih.rb_dma_addr);
+			adev->irq.ih.ring = NULL;
+		}
+	} else {
+		if (adev->irq.ih.ring_obj) {
+			r = amdgpu_bo_reserve(adev->irq.ih.ring_obj, false);
+			if (likely(r == 0)) {
+				amdgpu_bo_kunmap(adev->irq.ih.ring_obj);
+				amdgpu_bo_unpin(adev->irq.ih.ring_obj);
+				amdgpu_bo_unreserve(adev->irq.ih.ring_obj);
+			}
+			amdgpu_bo_unref(&adev->irq.ih.ring_obj);
+			adev->irq.ih.ring = NULL;
+			adev->irq.ih.ring_obj = NULL;
+		}
+		amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
+		amdgpu_wb_free(adev, adev->irq.ih.rptr_offs);
+	}
+}
+
+/**
+ * amdgpu_ih_process - interrupt handler
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Interrupt hander (VI), walk the IH ring.
+ * Returns irq process return code.
+ */
+int amdgpu_ih_process(struct amdgpu_device *adev)
+{
+	struct amdgpu_iv_entry entry;
+	u32 wptr;
+
+	if (!adev->irq.ih.enabled || adev->shutdown)
+		return IRQ_NONE;
+
+	wptr = amdgpu_ih_get_wptr(adev);
+
+restart_ih:
+	/* is somebody else already processing irqs? */
+	if (atomic_xchg(&adev->irq.ih.lock, 1))
+		return IRQ_NONE;
+
+	DRM_DEBUG("%s: rptr %d, wptr %d\n", __func__, adev->irq.ih.rptr, wptr);
+
+	/* Order reading of wptr vs. reading of IH ring data */
+	rmb();
+
+	while (adev->irq.ih.rptr != wptr) {
+		u32 ring_index = adev->irq.ih.rptr >> 2;
+
+		/* Before dispatching irq to IP blocks, send it to amdkfd */
+		amdgpu_amdkfd_interrupt(adev,
+				(const void *) &adev->irq.ih.ring[ring_index]);
+
+		entry.iv_entry = (const uint32_t *)
+			&adev->irq.ih.ring[ring_index];
+		amdgpu_ih_decode_iv(adev, &entry);
+		adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
+
+		amdgpu_irq_dispatch(adev, &entry);
+	}
+	amdgpu_ih_set_rptr(adev);
+	atomic_set(&adev->irq.ih.lock, 0);
+
+	/* make sure wptr hasn't changed while processing */
+	wptr = amdgpu_ih_get_wptr(adev);
+	if (wptr != adev->irq.ih.rptr)
+		goto restart_ih;
+
+	return IRQ_HANDLED;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
new file mode 100644
index 000000000..ba38ae6a1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
@@ -0,0 +1,63 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_IH_H__
+#define __AMDGPU_IH_H__
+
+struct amdgpu_device;
+
+/*
+ * R6xx+ IH ring
+ */
+struct amdgpu_ih_ring {
+	struct amdgpu_bo	*ring_obj;
+	volatile uint32_t	*ring;
+	unsigned		rptr;
+	unsigned		ring_size;
+	uint64_t		gpu_addr;
+	uint32_t		ptr_mask;
+	atomic_t		lock;
+	bool                    enabled;
+	unsigned		wptr_offs;
+	unsigned		rptr_offs;
+	u32			doorbell_index;
+	bool			use_doorbell;
+	bool			use_bus_addr;
+	dma_addr_t		rb_dma_addr; /* only used when use_bus_addr = true */
+};
+
+struct amdgpu_iv_entry {
+	unsigned src_id;
+	unsigned src_data;
+	unsigned ring_id;
+	unsigned vm_id;
+	unsigned pas_id;
+	const uint32_t *iv_entry;
+};
+
+int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
+			bool use_bus_addr);
+void amdgpu_ih_ring_fini(struct amdgpu_device *adev);
+int amdgpu_ih_process(struct amdgpu_device *adev);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c
new file mode 100644
index 000000000..26482914d
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ioc32.c
@@ -0,0 +1,47 @@
+/**
+ * \file amdgpu_ioc32.c
+ *
+ * 32-bit ioctl compatibility routines for the AMDGPU DRM.
+ *
+ * \author Paul Mackerras <paulus@samba.org>
+ *
+ * Copyright (C) Paul Mackerras 2005
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+#include <linux/compat.h>
+
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu_drv.h"
+
+long amdgpu_kms_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
+{
+	unsigned int nr = DRM_IOCTL_NR(cmd);
+	int ret;
+
+	if (nr < DRM_COMMAND_BASE)
+		return drm_compat_ioctl(filp, cmd, arg);
+
+	ret = amdgpu_drm_ioctl(filp, cmd, arg);
+
+	return ret;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
new file mode 100644
index 000000000..7c42ff670
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -0,0 +1,463 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "atom.h"
+#include "amdgpu_connectors.h"
+
+#include <linux/pm_runtime.h>
+
+#define AMDGPU_WAIT_IDLE_TIMEOUT 200
+
+/*
+ * Handle hotplug events outside the interrupt handler proper.
+ */
+/**
+ * amdgpu_hotplug_work_func - display hotplug work handler
+ *
+ * @work: work struct
+ *
+ * This is the hot plug event work handler (all asics).
+ * The work gets scheduled from the irq handler if there
+ * was a hot plug interrupt.  It walks the connector table
+ * and calls the hotplug handler for each one, then sends
+ * a drm hotplug event to alert userspace.
+ */
+static void amdgpu_hotplug_work_func(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  hotplug_work);
+	struct drm_device *dev = adev->ddev;
+	struct drm_mode_config *mode_config = &dev->mode_config;
+	struct drm_connector *connector;
+
+	mutex_lock(&mode_config->mutex);
+	if (mode_config->num_connector) {
+		list_for_each_entry(connector, &mode_config->connector_list, head)
+			amdgpu_connector_hotplug(connector);
+	}
+	mutex_unlock(&mode_config->mutex);
+	/* Just fire off a uevent and let userspace tell us what to do */
+	drm_helper_hpd_irq_event(dev);
+}
+
+/**
+ * amdgpu_irq_reset_work_func - execute gpu reset
+ *
+ * @work: work struct
+ *
+ * Execute scheduled gpu reset (cayman+).
+ * This function is called when the irq handler
+ * thinks we need a gpu reset.
+ */
+static void amdgpu_irq_reset_work_func(struct work_struct *work)
+{
+	struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
+						  reset_work);
+
+	amdgpu_gpu_reset(adev);
+}
+
+/* Disable *all* interrupts */
+static void amdgpu_irq_disable_all(struct amdgpu_device *adev)
+{
+	unsigned long irqflags;
+	unsigned i, j;
+	int r;
+
+	spin_lock_irqsave(&adev->irq.lock, irqflags);
+	for (i = 0; i < AMDGPU_MAX_IRQ_SRC_ID; ++i) {
+		struct amdgpu_irq_src *src = adev->irq.sources[i];
+
+		if (!src || !src->funcs->set || !src->num_types)
+			continue;
+
+		for (j = 0; j < src->num_types; ++j) {
+			atomic_set(&src->enabled_types[j], 0);
+			r = src->funcs->set(adev, src, j,
+					    AMDGPU_IRQ_STATE_DISABLE);
+			if (r)
+				DRM_ERROR("error disabling interrupt (%d)\n",
+					  r);
+		}
+	}
+	spin_unlock_irqrestore(&adev->irq.lock, irqflags);
+}
+
+/**
+ * amdgpu_irq_preinstall - drm irq preinstall callback
+ *
+ * @dev: drm dev pointer
+ *
+ * Gets the hw ready to enable irqs (all asics).
+ * This function disables all interrupt sources on the GPU.
+ */
+void amdgpu_irq_preinstall(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* Disable *all* interrupts */
+	amdgpu_irq_disable_all(adev);
+	/* Clear bits */
+	amdgpu_ih_process(adev);
+}
+
+/**
+ * amdgpu_irq_postinstall - drm irq preinstall callback
+ *
+ * @dev: drm dev pointer
+ *
+ * Handles stuff to be done after enabling irqs (all asics).
+ * Returns 0 on success.
+ */
+int amdgpu_irq_postinstall(struct drm_device *dev)
+{
+	dev->max_vblank_count = 0x00ffffff;
+	return 0;
+}
+
+/**
+ * amdgpu_irq_uninstall - drm irq uninstall callback
+ *
+ * @dev: drm dev pointer
+ *
+ * This function disables all interrupt sources on the GPU (all asics).
+ */
+void amdgpu_irq_uninstall(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (adev == NULL) {
+		return;
+	}
+	amdgpu_irq_disable_all(adev);
+}
+
+/**
+ * amdgpu_irq_handler - irq handler
+ *
+ * @int irq, void *arg: args
+ *
+ * This is the irq handler for the amdgpu driver (all asics).
+ */
+irqreturn_t amdgpu_irq_handler(int irq, void *arg)
+{
+	struct drm_device *dev = (struct drm_device *) arg;
+	struct amdgpu_device *adev = dev->dev_private;
+	irqreturn_t ret;
+
+	ret = amdgpu_ih_process(adev);
+	if (ret == IRQ_HANDLED)
+		pm_runtime_mark_last_busy(dev->dev);
+	return ret;
+}
+
+/**
+ * amdgpu_msi_ok - asic specific msi checks
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Handles asic specific MSI checks to determine if
+ * MSIs should be enabled on a particular chip (all asics).
+ * Returns true if MSIs should be enabled, false if MSIs
+ * should not be enabled.
+ */
+static bool amdgpu_msi_ok(struct amdgpu_device *adev)
+{
+	/* force MSI on */
+	if (amdgpu_msi == 1)
+		return true;
+	else if (amdgpu_msi == 0)
+		return false;
+
+	return true;
+}
+
+/**
+ * amdgpu_irq_init - init driver interrupt info
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Sets up the work irq handlers, vblank init, MSIs, etc. (all asics).
+ * Returns 0 for success, error for failure.
+ */
+int amdgpu_irq_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	spin_lock_init(&adev->irq.lock);
+	r = drm_vblank_init(adev->ddev, adev->mode_info.num_crtc);
+	if (r) {
+		return r;
+	}
+	/* enable msi */
+	adev->irq.msi_enabled = false;
+
+	if (amdgpu_msi_ok(adev)) {
+		int ret = pci_enable_msi(adev->pdev);
+		if (!ret) {
+			adev->irq.msi_enabled = true;
+			dev_info(adev->dev, "amdgpu: using MSI.\n");
+		}
+	}
+
+	INIT_WORK(&adev->hotplug_work, amdgpu_hotplug_work_func);
+	INIT_WORK(&adev->reset_work, amdgpu_irq_reset_work_func);
+
+	adev->irq.installed = true;
+	r = drm_irq_install(adev->ddev, adev->ddev->pdev->irq);
+	if (r) {
+		adev->irq.installed = false;
+		flush_work(&adev->hotplug_work);
+		return r;
+	}
+
+	DRM_INFO("amdgpu: irq initialized.\n");
+	return 0;
+}
+
+/**
+ * amdgpu_irq_fini - tear down driver interrupt info
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Tears down the work irq handlers, vblank handlers, MSIs, etc. (all asics).
+ */
+void amdgpu_irq_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	drm_vblank_cleanup(adev->ddev);
+	if (adev->irq.installed) {
+		drm_irq_uninstall(adev->ddev);
+		adev->irq.installed = false;
+		if (adev->irq.msi_enabled)
+			pci_disable_msi(adev->pdev);
+		flush_work(&adev->hotplug_work);
+	}
+
+	for (i = 0; i < AMDGPU_MAX_IRQ_SRC_ID; ++i) {
+		struct amdgpu_irq_src *src = adev->irq.sources[i];
+
+		if (!src)
+			continue;
+
+		kfree(src->enabled_types);
+		src->enabled_types = NULL;
+		if (src->data) {
+			kfree(src->data);
+			kfree(src);
+			adev->irq.sources[i] = NULL;
+		}
+	}
+}
+
+/**
+ * amdgpu_irq_add_id - register irq source
+ *
+ * @adev: amdgpu device pointer
+ * @src_id: source id for this source
+ * @source: irq source
+ *
+ */
+int amdgpu_irq_add_id(struct amdgpu_device *adev, unsigned src_id,
+		      struct amdgpu_irq_src *source)
+{
+	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID)
+		return -EINVAL;
+
+	if (adev->irq.sources[src_id] != NULL)
+		return -EINVAL;
+
+	if (!source->funcs)
+		return -EINVAL;
+
+	if (source->num_types && !source->enabled_types) {
+		atomic_t *types;
+
+		types = kcalloc(source->num_types, sizeof(atomic_t),
+				GFP_KERNEL);
+		if (!types)
+			return -ENOMEM;
+
+		source->enabled_types = types;
+	}
+
+	adev->irq.sources[src_id] = source;
+	return 0;
+}
+
+/**
+ * amdgpu_irq_dispatch - dispatch irq to IP blocks
+ *
+ * @adev: amdgpu device pointer
+ * @entry: interrupt vector
+ *
+ * Dispatches the irq to the different IP blocks
+ */
+void amdgpu_irq_dispatch(struct amdgpu_device *adev,
+			 struct amdgpu_iv_entry *entry)
+{
+	unsigned src_id = entry->src_id;
+	struct amdgpu_irq_src *src;
+	int r;
+
+	if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
+		DRM_DEBUG("Invalid src_id in IV: %d\n", src_id);
+		return;
+	}
+
+	src = adev->irq.sources[src_id];
+	if (!src) {
+		DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id);
+		return;
+	}
+
+	r = src->funcs->process(adev, src, entry);
+	if (r)
+		DRM_ERROR("error processing interrupt (%d)\n", r);
+}
+
+/**
+ * amdgpu_irq_update - update hw interrupt state
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt src you want to enable
+ * @type: type of interrupt you want to update
+ *
+ * Updates the interrupt state for a specific src (all asics).
+ */
+int amdgpu_irq_update(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *src, unsigned type)
+{
+	unsigned long irqflags;
+	enum amdgpu_interrupt_state state;
+	int r;
+
+	spin_lock_irqsave(&adev->irq.lock, irqflags);
+
+	/* we need to determine after taking the lock, otherwise
+	   we might disable just enabled interrupts again */
+	if (amdgpu_irq_enabled(adev, src, type))
+		state = AMDGPU_IRQ_STATE_ENABLE;
+	else
+		state = AMDGPU_IRQ_STATE_DISABLE;
+
+	r = src->funcs->set(adev, src, type, state);
+	spin_unlock_irqrestore(&adev->irq.lock, irqflags);
+	return r;
+}
+
+/**
+ * amdgpu_irq_get - enable interrupt
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt src you want to enable
+ * @type: type of interrupt you want to enable
+ *
+ * Enables the interrupt type for a specific src (all asics).
+ */
+int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type)
+{
+	if (!adev->ddev->irq_enabled)
+		return -ENOENT;
+
+	if (type >= src->num_types)
+		return -EINVAL;
+
+	if (!src->enabled_types || !src->funcs->set)
+		return -EINVAL;
+
+	if (atomic_inc_return(&src->enabled_types[type]) == 1)
+		return amdgpu_irq_update(adev, src, type);
+
+	return 0;
+}
+
+bool amdgpu_irq_get_delayed(struct amdgpu_device *adev,
+			struct amdgpu_irq_src *src,
+			unsigned type)
+{
+	if ((type >= src->num_types) || !src->enabled_types)
+		return false;
+	return atomic_inc_return(&src->enabled_types[type]) == 1;
+}
+
+/**
+ * amdgpu_irq_put - disable interrupt
+ *
+ * @adev: amdgpu device pointer
+ * @src: interrupt src you want to disable
+ * @type: type of interrupt you want to disable
+ *
+ * Disables the interrupt type for a specific src (all asics).
+ */
+int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type)
+{
+	if (!adev->ddev->irq_enabled)
+		return -ENOENT;
+
+	if (type >= src->num_types)
+		return -EINVAL;
+
+	if (!src->enabled_types || !src->funcs->set)
+		return -EINVAL;
+
+	if (atomic_dec_and_test(&src->enabled_types[type]))
+		return amdgpu_irq_update(adev, src, type);
+
+	return 0;
+}
+
+/**
+ * amdgpu_irq_enabled - test if irq is enabled or not
+ *
+ * @adev: amdgpu device pointer
+ * @idx: interrupt src you want to test
+ *
+ * Tests if the given interrupt source is enabled or not
+ */
+bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+			unsigned type)
+{
+	if (!adev->ddev->irq_enabled)
+		return false;
+
+	if (type >= src->num_types)
+		return false;
+
+	if (!src->enabled_types || !src->funcs->set)
+		return false;
+
+	return !!atomic_read(&src->enabled_types[type]);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
new file mode 100644
index 000000000..17b01aef4
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -0,0 +1,93 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_IRQ_H__
+#define __AMDGPU_IRQ_H__
+
+#include "amdgpu_ih.h"
+
+#define AMDGPU_MAX_IRQ_SRC_ID	0x100
+
+struct amdgpu_device;
+struct amdgpu_iv_entry;
+
+enum amdgpu_interrupt_state {
+	AMDGPU_IRQ_STATE_DISABLE,
+	AMDGPU_IRQ_STATE_ENABLE,
+};
+
+struct amdgpu_irq_src {
+	unsigned				num_types;
+	atomic_t				*enabled_types;
+	const struct amdgpu_irq_src_funcs	*funcs;
+	void *data;
+};
+
+/* provided by interrupt generating IP blocks */
+struct amdgpu_irq_src_funcs {
+	int (*set)(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+		   unsigned type, enum amdgpu_interrupt_state state);
+
+	int (*process)(struct amdgpu_device *adev,
+		       struct amdgpu_irq_src *source,
+		       struct amdgpu_iv_entry *entry);
+};
+
+struct amdgpu_irq {
+	bool				installed;
+	spinlock_t			lock;
+	/* interrupt sources */
+	struct amdgpu_irq_src		*sources[AMDGPU_MAX_IRQ_SRC_ID];
+
+	/* status, etc. */
+	bool				msi_enabled; /* msi enabled */
+
+	/* interrupt ring */
+	struct amdgpu_ih_ring		ih;
+	const struct amdgpu_ih_funcs	*ih_funcs;
+};
+
+void amdgpu_irq_preinstall(struct drm_device *dev);
+int amdgpu_irq_postinstall(struct drm_device *dev);
+void amdgpu_irq_uninstall(struct drm_device *dev);
+irqreturn_t amdgpu_irq_handler(int irq, void *arg);
+
+int amdgpu_irq_init(struct amdgpu_device *adev);
+void amdgpu_irq_fini(struct amdgpu_device *adev);
+int amdgpu_irq_add_id(struct amdgpu_device *adev, unsigned src_id,
+		      struct amdgpu_irq_src *source);
+void amdgpu_irq_dispatch(struct amdgpu_device *adev,
+			 struct amdgpu_iv_entry *entry);
+int amdgpu_irq_update(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		      unsigned type);
+int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type);
+bool amdgpu_irq_get_delayed(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *src,
+			    unsigned type);
+int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+		   unsigned type);
+bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
+			unsigned type);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
new file mode 100644
index 000000000..e23843f4d
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -0,0 +1,752 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include <drm/amdgpu_drm.h>
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+
+#include <linux/vga_switcheroo.h>
+#include <linux/slab.h>
+#include <linux/pm_runtime.h>
+#include "amdgpu_amdkfd.h"
+
+#if defined(CONFIG_VGA_SWITCHEROO)
+bool amdgpu_has_atpx(void);
+#else
+static inline bool amdgpu_has_atpx(void) { return false; }
+#endif
+
+/**
+ * amdgpu_driver_unload_kms - Main unload function for KMS.
+ *
+ * @dev: drm dev pointer
+ *
+ * This is the main unload function for KMS (all asics).
+ * Returns 0 on success.
+ */
+int amdgpu_driver_unload_kms(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (adev == NULL)
+		return 0;
+
+	if (adev->rmmio == NULL)
+		goto done_free;
+
+	pm_runtime_get_sync(dev->dev);
+
+	amdgpu_amdkfd_device_fini(adev);
+
+	amdgpu_acpi_fini(adev);
+
+	amdgpu_device_fini(adev);
+
+done_free:
+	kfree(adev);
+	dev->dev_private = NULL;
+	return 0;
+}
+
+/**
+ * amdgpu_driver_load_kms - Main load function for KMS.
+ *
+ * @dev: drm dev pointer
+ * @flags: device flags
+ *
+ * This is the main load function for KMS (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
+{
+	struct amdgpu_device *adev;
+	int r, acpi_status;
+
+	adev = kzalloc(sizeof(struct amdgpu_device), GFP_KERNEL);
+	if (adev == NULL) {
+		return -ENOMEM;
+	}
+	dev->dev_private = (void *)adev;
+
+	if ((amdgpu_runtime_pm != 0) &&
+	    amdgpu_has_atpx() &&
+	    ((flags & AMD_IS_APU) == 0))
+		flags |= AMD_IS_PX;
+
+	/* amdgpu_device_init should report only fatal error
+	 * like memory allocation failure or iomapping failure,
+	 * or memory manager initialization failure, it must
+	 * properly initialize the GPU MC controller and permit
+	 * VRAM allocation
+	 */
+	r = amdgpu_device_init(adev, dev, dev->pdev, flags);
+	if (r) {
+		dev_err(&dev->pdev->dev, "Fatal error during GPU init\n");
+		goto out;
+	}
+
+	/* Call ACPI methods: require modeset init
+	 * but failure is not fatal
+	 */
+	if (!r) {
+		acpi_status = amdgpu_acpi_init(adev);
+		if (acpi_status)
+		dev_dbg(&dev->pdev->dev,
+				"Error during ACPI methods call\n");
+	}
+
+	amdgpu_amdkfd_load_interface(adev);
+	amdgpu_amdkfd_device_probe(adev);
+	amdgpu_amdkfd_device_init(adev);
+
+	if (amdgpu_device_is_px(dev)) {
+		pm_runtime_use_autosuspend(dev->dev);
+		pm_runtime_set_autosuspend_delay(dev->dev, 5000);
+		pm_runtime_set_active(dev->dev);
+		pm_runtime_allow(dev->dev);
+		pm_runtime_mark_last_busy(dev->dev);
+		pm_runtime_put_autosuspend(dev->dev);
+	}
+
+out:
+	if (r)
+		amdgpu_driver_unload_kms(dev);
+
+
+	return r;
+}
+
+/*
+ * Userspace get information ioctl
+ */
+/**
+ * amdgpu_info_ioctl - answer a device specific request.
+ *
+ * @adev: amdgpu device pointer
+ * @data: request object
+ * @filp: drm filp
+ *
+ * This function is used to pass device specific parameters to the userspace
+ * drivers.  Examples include: pci device id, pipeline parms, tiling params,
+ * etc. (all asics).
+ * Returns 0 on success, -EINVAL on failure.
+ */
+static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_info *info = data;
+	struct amdgpu_mode_info *minfo = &adev->mode_info;
+	void __user *out = (void __user *)(long)info->return_pointer;
+	uint32_t size = info->return_size;
+	struct drm_crtc *crtc;
+	uint32_t ui32 = 0;
+	uint64_t ui64 = 0;
+	int i, found;
+
+	if (!info->return_size || !info->return_pointer)
+		return -EINVAL;
+
+	switch (info->query) {
+	case AMDGPU_INFO_ACCEL_WORKING:
+		ui32 = adev->accel_working;
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_CRTC_FROM_ID:
+		for (i = 0, found = 0; i < adev->mode_info.num_crtc; i++) {
+			crtc = (struct drm_crtc *)minfo->crtcs[i];
+			if (crtc && crtc->base.id == info->mode_crtc.id) {
+				struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+				ui32 = amdgpu_crtc->crtc_id;
+				found = 1;
+				break;
+			}
+		}
+		if (!found) {
+			DRM_DEBUG_KMS("unknown crtc id %d\n", info->mode_crtc.id);
+			return -EINVAL;
+		}
+		return copy_to_user(out, &ui32, min(size, 4u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_HW_IP_INFO: {
+		struct drm_amdgpu_info_hw_ip ip = {};
+		enum amd_ip_block_type type;
+		uint32_t ring_mask = 0;
+		uint32_t ib_start_alignment = 0;
+		uint32_t ib_size_alignment = 0;
+
+		if (info->query_hw_ip.ip_instance >= AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
+			return -EINVAL;
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_GFX:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+				ring_mask |= ((adev->gfx.gfx_ring[i].ready ? 1 : 0) << i);
+			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
+			ib_size_alignment = 8;
+			break;
+		case AMDGPU_HW_IP_COMPUTE:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			for (i = 0; i < adev->gfx.num_compute_rings; i++)
+				ring_mask |= ((adev->gfx.compute_ring[i].ready ? 1 : 0) << i);
+			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
+			ib_size_alignment = 8;
+			break;
+		case AMDGPU_HW_IP_DMA:
+			type = AMD_IP_BLOCK_TYPE_SDMA;
+			for (i = 0; i < adev->sdma.num_instances; i++)
+				ring_mask |= ((adev->sdma.instance[i].ring.ready ? 1 : 0) << i);
+			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
+			ib_size_alignment = 1;
+			break;
+		case AMDGPU_HW_IP_UVD:
+			type = AMD_IP_BLOCK_TYPE_UVD;
+			ring_mask = adev->uvd.ring.ready ? 1 : 0;
+			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
+			ib_size_alignment = 8;
+			break;
+		case AMDGPU_HW_IP_VCE:
+			type = AMD_IP_BLOCK_TYPE_VCE;
+			for (i = 0; i < AMDGPU_MAX_VCE_RINGS; i++)
+				ring_mask |= ((adev->vce.ring[i].ready ? 1 : 0) << i);
+			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
+			ib_size_alignment = 8;
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		for (i = 0; i < adev->num_ip_blocks; i++) {
+			if (adev->ip_blocks[i].type == type &&
+			    adev->ip_block_status[i].valid) {
+				ip.hw_ip_version_major = adev->ip_blocks[i].major;
+				ip.hw_ip_version_minor = adev->ip_blocks[i].minor;
+				ip.capabilities_flags = 0;
+				ip.available_rings = ring_mask;
+				ip.ib_start_alignment = ib_start_alignment;
+				ip.ib_size_alignment = ib_size_alignment;
+				break;
+			}
+		}
+		return copy_to_user(out, &ip,
+				    min((size_t)size, sizeof(ip))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_HW_IP_COUNT: {
+		enum amd_ip_block_type type;
+		uint32_t count = 0;
+
+		switch (info->query_hw_ip.type) {
+		case AMDGPU_HW_IP_GFX:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			break;
+		case AMDGPU_HW_IP_COMPUTE:
+			type = AMD_IP_BLOCK_TYPE_GFX;
+			break;
+		case AMDGPU_HW_IP_DMA:
+			type = AMD_IP_BLOCK_TYPE_SDMA;
+			break;
+		case AMDGPU_HW_IP_UVD:
+			type = AMD_IP_BLOCK_TYPE_UVD;
+			break;
+		case AMDGPU_HW_IP_VCE:
+			type = AMD_IP_BLOCK_TYPE_VCE;
+			break;
+		default:
+			return -EINVAL;
+		}
+
+		for (i = 0; i < adev->num_ip_blocks; i++)
+			if (adev->ip_blocks[i].type == type &&
+			    adev->ip_block_status[i].valid &&
+			    count < AMDGPU_HW_IP_INSTANCE_MAX_COUNT)
+				count++;
+
+		return copy_to_user(out, &count, min(size, 4u)) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_TIMESTAMP:
+		ui64 = amdgpu_asic_get_gpu_clock_counter(adev);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_FW_VERSION: {
+		struct drm_amdgpu_info_firmware fw_info;
+
+		/* We only support one instance of each IP block right now. */
+		if (info->query_fw.ip_instance != 0)
+			return -EINVAL;
+
+		switch (info->query_fw.fw_type) {
+		case AMDGPU_INFO_FW_VCE:
+			fw_info.ver = adev->vce.fw_version;
+			fw_info.feature = adev->vce.fb_version;
+			break;
+		case AMDGPU_INFO_FW_UVD:
+			fw_info.ver = 0;
+			fw_info.feature = 0;
+			break;
+		case AMDGPU_INFO_FW_GMC:
+			fw_info.ver = adev->mc.fw_version;
+			fw_info.feature = 0;
+			break;
+		case AMDGPU_INFO_FW_GFX_ME:
+			fw_info.ver = adev->gfx.me_fw_version;
+			fw_info.feature = adev->gfx.me_feature_version;
+			break;
+		case AMDGPU_INFO_FW_GFX_PFP:
+			fw_info.ver = adev->gfx.pfp_fw_version;
+			fw_info.feature = adev->gfx.pfp_feature_version;
+			break;
+		case AMDGPU_INFO_FW_GFX_CE:
+			fw_info.ver = adev->gfx.ce_fw_version;
+			fw_info.feature = adev->gfx.ce_feature_version;
+			break;
+		case AMDGPU_INFO_FW_GFX_RLC:
+			fw_info.ver = adev->gfx.rlc_fw_version;
+			fw_info.feature = adev->gfx.rlc_feature_version;
+			break;
+		case AMDGPU_INFO_FW_GFX_MEC:
+			if (info->query_fw.index == 0) {
+				fw_info.ver = adev->gfx.mec_fw_version;
+				fw_info.feature = adev->gfx.mec_feature_version;
+			} else if (info->query_fw.index == 1) {
+				fw_info.ver = adev->gfx.mec2_fw_version;
+				fw_info.feature = adev->gfx.mec2_feature_version;
+			} else
+				return -EINVAL;
+			break;
+		case AMDGPU_INFO_FW_SMC:
+			fw_info.ver = adev->pm.fw_version;
+			fw_info.feature = 0;
+			break;
+		case AMDGPU_INFO_FW_SDMA:
+			if (info->query_fw.index >= adev->sdma.num_instances)
+				return -EINVAL;
+			fw_info.ver = adev->sdma.instance[info->query_fw.index].fw_version;
+			fw_info.feature = adev->sdma.instance[info->query_fw.index].feature_version;
+			break;
+		default:
+			return -EINVAL;
+		}
+		return copy_to_user(out, &fw_info,
+				    min((size_t)size, sizeof(fw_info))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_NUM_BYTES_MOVED:
+		ui64 = atomic64_read(&adev->num_bytes_moved);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_VRAM_USAGE:
+		ui64 = atomic64_read(&adev->vram_usage);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_VIS_VRAM_USAGE:
+		ui64 = atomic64_read(&adev->vram_vis_usage);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_GTT_USAGE:
+		ui64 = atomic64_read(&adev->gtt_usage);
+		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
+	case AMDGPU_INFO_GDS_CONFIG: {
+		struct drm_amdgpu_info_gds gds_info;
+
+		memset(&gds_info, 0, sizeof(gds_info));
+		gds_info.gds_gfx_partition_size = adev->gds.mem.gfx_partition_size >> AMDGPU_GDS_SHIFT;
+		gds_info.compute_partition_size = adev->gds.mem.cs_partition_size >> AMDGPU_GDS_SHIFT;
+		gds_info.gds_total_size = adev->gds.mem.total_size >> AMDGPU_GDS_SHIFT;
+		gds_info.gws_per_gfx_partition = adev->gds.gws.gfx_partition_size >> AMDGPU_GWS_SHIFT;
+		gds_info.gws_per_compute_partition = adev->gds.gws.cs_partition_size >> AMDGPU_GWS_SHIFT;
+		gds_info.oa_per_gfx_partition = adev->gds.oa.gfx_partition_size >> AMDGPU_OA_SHIFT;
+		gds_info.oa_per_compute_partition = adev->gds.oa.cs_partition_size >> AMDGPU_OA_SHIFT;
+		return copy_to_user(out, &gds_info,
+				    min((size_t)size, sizeof(gds_info))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_VRAM_GTT: {
+		struct drm_amdgpu_info_vram_gtt vram_gtt;
+
+		vram_gtt.vram_size = adev->mc.real_vram_size;
+		vram_gtt.vram_cpu_accessible_size = adev->mc.visible_vram_size;
+		vram_gtt.vram_cpu_accessible_size -= adev->vram_pin_size;
+		vram_gtt.gtt_size  = adev->mc.gtt_size;
+		vram_gtt.gtt_size -= adev->gart_pin_size;
+		return copy_to_user(out, &vram_gtt,
+				    min((size_t)size, sizeof(vram_gtt))) ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_READ_MMR_REG: {
+		unsigned n, alloc_size;
+		uint32_t *regs;
+		unsigned se_num = (info->read_mmr_reg.instance >>
+				   AMDGPU_INFO_MMR_SE_INDEX_SHIFT) &
+				  AMDGPU_INFO_MMR_SE_INDEX_MASK;
+		unsigned sh_num = (info->read_mmr_reg.instance >>
+				   AMDGPU_INFO_MMR_SH_INDEX_SHIFT) &
+				  AMDGPU_INFO_MMR_SH_INDEX_MASK;
+
+		/* set full masks if the userspace set all bits
+		 * in the bitfields */
+		if (se_num == AMDGPU_INFO_MMR_SE_INDEX_MASK)
+			se_num = 0xffffffff;
+		if (sh_num == AMDGPU_INFO_MMR_SH_INDEX_MASK)
+			sh_num = 0xffffffff;
+
+		regs = kmalloc_array(info->read_mmr_reg.count, sizeof(*regs), GFP_KERNEL);
+		if (!regs)
+			return -ENOMEM;
+		alloc_size = info->read_mmr_reg.count * sizeof(*regs);
+
+		for (i = 0; i < info->read_mmr_reg.count; i++)
+			if (amdgpu_asic_read_register(adev, se_num, sh_num,
+						      info->read_mmr_reg.dword_offset + i,
+						      &regs[i])) {
+				DRM_DEBUG_KMS("unallowed offset %#x\n",
+					      info->read_mmr_reg.dword_offset + i);
+				kfree(regs);
+				return -EFAULT;
+			}
+		n = copy_to_user(out, regs, min(size, alloc_size));
+		kfree(regs);
+		return n ? -EFAULT : 0;
+	}
+	case AMDGPU_INFO_DEV_INFO: {
+		struct drm_amdgpu_info_device dev_info = {};
+		struct amdgpu_cu_info cu_info;
+
+		dev_info.device_id = dev->pdev->device;
+		dev_info.chip_rev = adev->rev_id;
+		dev_info.external_rev = adev->external_rev_id;
+		dev_info.pci_rev = dev->pdev->revision;
+		dev_info.family = adev->family;
+		dev_info.num_shader_engines = adev->gfx.config.max_shader_engines;
+		dev_info.num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
+		/* return all clocks in KHz */
+		dev_info.gpu_counter_freq = amdgpu_asic_get_xclk(adev) * 10;
+		if (adev->pm.dpm_enabled) {
+			dev_info.max_engine_clock =
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk * 10;
+			dev_info.max_memory_clock =
+				adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.mclk * 10;
+		} else {
+			dev_info.max_engine_clock = adev->pm.default_sclk * 10;
+			dev_info.max_memory_clock = adev->pm.default_mclk * 10;
+		}
+		dev_info.enabled_rb_pipes_mask = adev->gfx.config.backend_enable_mask;
+		dev_info.num_rb_pipes = adev->gfx.config.max_backends_per_se *
+					adev->gfx.config.max_shader_engines;
+		dev_info.num_hw_gfx_contexts = adev->gfx.config.max_hw_contexts;
+		dev_info._pad = 0;
+		dev_info.ids_flags = 0;
+		if (adev->flags & AMD_IS_APU)
+			dev_info.ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
+		dev_info.virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
+		dev_info.virtual_address_max = (uint64_t)adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
+		dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
+		dev_info.pte_fragment_size = (1 << AMDGPU_LOG2_PAGES_PER_FRAG) *
+					     AMDGPU_GPU_PAGE_SIZE;
+		dev_info.gart_page_size = AMDGPU_GPU_PAGE_SIZE;
+
+		amdgpu_asic_get_cu_info(adev, &cu_info);
+		dev_info.cu_active_number = cu_info.number;
+		dev_info.cu_ao_mask = cu_info.ao_cu_mask;
+		dev_info.ce_ram_size = adev->gfx.ce_ram_size;
+		memcpy(&dev_info.cu_bitmap[0], &cu_info.bitmap[0], sizeof(cu_info.bitmap));
+		dev_info.vram_type = adev->mc.vram_type;
+		dev_info.vram_bit_width = adev->mc.vram_width;
+		dev_info.vce_harvest_config = adev->vce.harvest_config;
+
+		return copy_to_user(out, &dev_info,
+				    min((size_t)size, sizeof(dev_info))) ? -EFAULT : 0;
+	}
+	default:
+		DRM_DEBUG_KMS("Invalid request %d\n", info->query);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+
+/*
+ * Outdated mess for old drm with Xorg being in charge (void function now).
+ */
+/**
+ * amdgpu_driver_lastclose_kms - drm callback for last close
+ *
+ * @dev: drm dev pointer
+ *
+ * Switch vga_switcheroo state after last close (all asics).
+ */
+void amdgpu_driver_lastclose_kms(struct drm_device *dev)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	amdgpu_fbdev_restore_mode(adev);
+	vga_switcheroo_process_delayed_switch();
+}
+
+/**
+ * amdgpu_driver_open_kms - drm callback for open
+ *
+ * @dev: drm dev pointer
+ * @file_priv: drm file
+ *
+ * On device open, init vm on cayman+ (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv;
+	int r;
+
+	file_priv->driver_priv = NULL;
+
+	r = pm_runtime_get_sync(dev->dev);
+	if (r < 0)
+		return r;
+
+	fpriv = kzalloc(sizeof(*fpriv), GFP_KERNEL);
+	if (unlikely(!fpriv))
+		return -ENOMEM;
+
+	r = amdgpu_vm_init(adev, &fpriv->vm);
+	if (r)
+		goto error_free;
+
+	mutex_init(&fpriv->bo_list_lock);
+	idr_init(&fpriv->bo_list_handles);
+
+	amdgpu_ctx_mgr_init(&fpriv->ctx_mgr);
+
+	file_priv->driver_priv = fpriv;
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
+	return 0;
+
+error_free:
+	kfree(fpriv);
+
+	return r;
+}
+
+/**
+ * amdgpu_driver_postclose_kms - drm callback for post close
+ *
+ * @dev: drm dev pointer
+ * @file_priv: drm file
+ *
+ * On device post close, tear down vm on cayman+ (all asics).
+ */
+void amdgpu_driver_postclose_kms(struct drm_device *dev,
+				 struct drm_file *file_priv)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_fpriv *fpriv = file_priv->driver_priv;
+	struct amdgpu_bo_list *list;
+	int handle;
+
+	if (!fpriv)
+		return;
+
+	amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr);
+
+	amdgpu_vm_fini(adev, &fpriv->vm);
+
+	idr_for_each_entry(&fpriv->bo_list_handles, list, handle)
+		amdgpu_bo_list_free(list);
+
+	idr_destroy(&fpriv->bo_list_handles);
+	mutex_destroy(&fpriv->bo_list_lock);
+
+	kfree(fpriv);
+	file_priv->driver_priv = NULL;
+}
+
+/**
+ * amdgpu_driver_preclose_kms - drm callback for pre close
+ *
+ * @dev: drm dev pointer
+ * @file_priv: drm file
+ *
+ * On device pre close, tear down hyperz and cmask filps on r1xx-r5xx
+ * (all asics).
+ */
+void amdgpu_driver_preclose_kms(struct drm_device *dev,
+				struct drm_file *file_priv)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+
+	amdgpu_uvd_free_handles(adev, file_priv);
+	amdgpu_vce_free_handles(adev, file_priv);
+}
+
+/*
+ * VBlank related functions.
+ */
+/**
+ * amdgpu_get_vblank_counter_kms - get frame count
+ *
+ * @dev: drm dev pointer
+ * @pipe: crtc to get the frame count from
+ *
+ * Gets the frame count on the requested crtc (all asics).
+ * Returns frame count on success, -EINVAL on failure.
+ */
+u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	int vpos, hpos, stat;
+	u32 count;
+
+	if (pipe >= adev->mode_info.num_crtc) {
+		DRM_ERROR("Invalid crtc %u\n", pipe);
+		return -EINVAL;
+	}
+
+	/* The hw increments its frame counter at start of vsync, not at start
+	 * of vblank, as is required by DRM core vblank counter handling.
+	 * Cook the hw count here to make it appear to the caller as if it
+	 * incremented at start of vblank. We measure distance to start of
+	 * vblank in vpos. vpos therefore will be >= 0 between start of vblank
+	 * and start of vsync, so vpos >= 0 means to bump the hw frame counter
+	 * result by 1 to give the proper appearance to caller.
+	 */
+	if (adev->mode_info.crtcs[pipe]) {
+		/* Repeat readout if needed to provide stable result if
+		 * we cross start of vsync during the queries.
+		 */
+		do {
+			count = amdgpu_display_vblank_get_counter(adev, pipe);
+			/* Ask amdgpu_get_crtc_scanoutpos to return vpos as
+			 * distance to start of vblank, instead of regular
+			 * vertical scanout pos.
+			 */
+			stat = amdgpu_get_crtc_scanoutpos(
+				dev, pipe, GET_DISTANCE_TO_VBLANKSTART,
+				&vpos, &hpos, NULL, NULL,
+				&adev->mode_info.crtcs[pipe]->base.hwmode);
+		} while (count != amdgpu_display_vblank_get_counter(adev, pipe));
+
+		if (((stat & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE)) !=
+		    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE))) {
+			DRM_DEBUG_VBL("Query failed! stat %d\n", stat);
+		} else {
+			DRM_DEBUG_VBL("crtc %d: dist from vblank start %d\n",
+				      pipe, vpos);
+
+			/* Bump counter if we are at >= leading edge of vblank,
+			 * but before vsync where vpos would turn negative and
+			 * the hw counter really increments.
+			 */
+			if (vpos >= 0)
+				count++;
+		}
+	} else {
+		/* Fallback to use value as is. */
+		count = amdgpu_display_vblank_get_counter(adev, pipe);
+		DRM_DEBUG_VBL("NULL mode info! Returned count may be wrong.\n");
+	}
+
+	return count;
+}
+
+/**
+ * amdgpu_enable_vblank_kms - enable vblank interrupt
+ *
+ * @dev: drm dev pointer
+ * @pipe: crtc to enable vblank interrupt for
+ *
+ * Enable the interrupt on the requested crtc (all asics).
+ * Returns 0 on success, -EINVAL on failure.
+ */
+int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	int idx = amdgpu_crtc_idx_to_irq_type(adev, pipe);
+
+	return amdgpu_irq_get(adev, &adev->crtc_irq, idx);
+}
+
+/**
+ * amdgpu_disable_vblank_kms - disable vblank interrupt
+ *
+ * @dev: drm dev pointer
+ * @pipe: crtc to disable vblank interrupt for
+ *
+ * Disable the interrupt on the requested crtc (all asics).
+ */
+void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe)
+{
+	struct amdgpu_device *adev = dev->dev_private;
+	int idx = amdgpu_crtc_idx_to_irq_type(adev, pipe);
+
+	amdgpu_irq_put(adev, &adev->crtc_irq, idx);
+}
+
+/**
+ * amdgpu_get_vblank_timestamp_kms - get vblank timestamp
+ *
+ * @dev: drm dev pointer
+ * @crtc: crtc to get the timestamp for
+ * @max_error: max error
+ * @vblank_time: time value
+ * @flags: flags passed to the driver
+ *
+ * Gets the timestamp on the requested crtc based on the
+ * scanout position.  (all asics).
+ * Returns postive status flags on success, negative error on failure.
+ */
+int amdgpu_get_vblank_timestamp_kms(struct drm_device *dev, unsigned int pipe,
+				    int *max_error,
+				    struct timeval *vblank_time,
+				    unsigned flags)
+{
+	struct drm_crtc *crtc;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (pipe >= dev->num_crtcs) {
+		DRM_ERROR("Invalid crtc %u\n", pipe);
+		return -EINVAL;
+	}
+
+	/* Get associated drm_crtc: */
+	crtc = &adev->mode_info.crtcs[pipe]->base;
+
+	/* Helper routine in DRM core does all the work: */
+	return drm_calc_vbltimestamp_from_scanoutpos(dev, pipe, max_error,
+						     vblank_time, flags,
+						     &crtc->hwmode);
+}
+
+const struct drm_ioctl_desc amdgpu_ioctls_kms[] = {
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_CREATE, amdgpu_gem_create_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_CTX, amdgpu_ctx_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_BO_LIST, amdgpu_bo_list_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	/* KMS */
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_MMAP, amdgpu_gem_mmap_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_WAIT_IDLE, amdgpu_gem_wait_idle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_CS, amdgpu_cs_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_INFO, amdgpu_info_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_WAIT_CS, amdgpu_cs_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_METADATA, amdgpu_gem_metadata_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_VA, amdgpu_gem_va_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_OP, amdgpu_gem_op_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+};
+int amdgpu_max_kms_ioctl = ARRAY_SIZE(amdgpu_ioctls_kms);
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
new file mode 100644
index 000000000..d4e2780c0
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
@@ -0,0 +1,323 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <linux/mmu_notifier.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+
+struct amdgpu_mn {
+	/* constant after initialisation */
+	struct amdgpu_device	*adev;
+	struct mm_struct	*mm;
+	struct mmu_notifier	mn;
+
+	/* only used on destruction */
+	struct work_struct	work;
+
+	/* protected by adev->mn_lock */
+	struct hlist_node	node;
+
+	/* objects protected by lock */
+	struct mutex		lock;
+	struct rb_root		objects;
+};
+
+struct amdgpu_mn_node {
+	struct interval_tree_node	it;
+	struct list_head		bos;
+};
+
+/**
+ * amdgpu_mn_destroy - destroy the rmn
+ *
+ * @work: previously sheduled work item
+ *
+ * Lazy destroys the notifier from a work item
+ */
+static void amdgpu_mn_destroy(struct work_struct *work)
+{
+	struct amdgpu_mn *rmn = container_of(work, struct amdgpu_mn, work);
+	struct amdgpu_device *adev = rmn->adev;
+	struct amdgpu_mn_node *node, *next_node;
+	struct amdgpu_bo *bo, *next_bo;
+
+	mutex_lock(&adev->mn_lock);
+	mutex_lock(&rmn->lock);
+	hash_del(&rmn->node);
+	rbtree_postorder_for_each_entry_safe(node, next_node, &rmn->objects,
+					     it.rb) {
+
+		interval_tree_remove(&node->it, &rmn->objects);
+		list_for_each_entry_safe(bo, next_bo, &node->bos, mn_list) {
+			bo->mn = NULL;
+			list_del_init(&bo->mn_list);
+		}
+		kfree(node);
+	}
+	mutex_unlock(&rmn->lock);
+	mutex_unlock(&adev->mn_lock);
+	mmu_notifier_unregister(&rmn->mn, rmn->mm);
+	kfree(rmn);
+}
+
+/**
+ * amdgpu_mn_release - callback to notify about mm destruction
+ *
+ * @mn: our notifier
+ * @mn: the mm this callback is about
+ *
+ * Shedule a work item to lazy destroy our notifier.
+ */
+static void amdgpu_mn_release(struct mmu_notifier *mn,
+			      struct mm_struct *mm)
+{
+	struct amdgpu_mn *rmn = container_of(mn, struct amdgpu_mn, mn);
+	INIT_WORK(&rmn->work, amdgpu_mn_destroy);
+	schedule_work(&rmn->work);
+}
+
+/**
+ * amdgpu_mn_invalidate_range_start - callback to notify about mm change
+ *
+ * @mn: our notifier
+ * @mn: the mm this callback is about
+ * @start: start of updated range
+ * @end: end of updated range
+ *
+ * We block for all BOs between start and end to be idle and
+ * unmap them by move them into system domain again.
+ */
+static void amdgpu_mn_invalidate_range_start(struct mmu_notifier *mn,
+					     struct mm_struct *mm,
+					     unsigned long start,
+					     unsigned long end)
+{
+	struct amdgpu_mn *rmn = container_of(mn, struct amdgpu_mn, mn);
+	struct interval_tree_node *it;
+
+	/* notification is exclusive, but interval is inclusive */
+	end -= 1;
+
+	mutex_lock(&rmn->lock);
+
+	it = interval_tree_iter_first(&rmn->objects, start, end);
+	while (it) {
+		struct amdgpu_mn_node *node;
+		struct amdgpu_bo *bo;
+		long r;
+
+		node = container_of(it, struct amdgpu_mn_node, it);
+		it = interval_tree_iter_next(it, start, end);
+
+		list_for_each_entry(bo, &node->bos, mn_list) {
+
+			if (!amdgpu_ttm_tt_affect_userptr(bo->tbo.ttm, start,
+							  end))
+				continue;
+
+			r = amdgpu_bo_reserve(bo, true);
+			if (r) {
+				DRM_ERROR("(%ld) failed to reserve user bo\n", r);
+				continue;
+			}
+
+			r = reservation_object_wait_timeout_rcu(bo->tbo.resv,
+				true, false, MAX_SCHEDULE_TIMEOUT);
+			if (r <= 0)
+				DRM_ERROR("(%ld) failed to wait for user bo\n", r);
+
+			amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
+			r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+			if (r)
+				DRM_ERROR("(%ld) failed to validate user bo\n", r);
+
+			amdgpu_bo_unreserve(bo);
+		}
+	}
+
+	mutex_unlock(&rmn->lock);
+}
+
+static const struct mmu_notifier_ops amdgpu_mn_ops = {
+	.release = amdgpu_mn_release,
+	.invalidate_range_start = amdgpu_mn_invalidate_range_start,
+};
+
+/**
+ * amdgpu_mn_get - create notifier context
+ *
+ * @adev: amdgpu device pointer
+ *
+ * Creates a notifier context for current->mm.
+ */
+static struct amdgpu_mn *amdgpu_mn_get(struct amdgpu_device *adev)
+{
+	struct mm_struct *mm = current->mm;
+	struct amdgpu_mn *rmn;
+	int r;
+
+	down_write(&mm->mmap_sem);
+	mutex_lock(&adev->mn_lock);
+
+	hash_for_each_possible(adev->mn_hash, rmn, node, (unsigned long)mm)
+		if (rmn->mm == mm)
+			goto release_locks;
+
+	rmn = kzalloc(sizeof(*rmn), GFP_KERNEL);
+	if (!rmn) {
+		rmn = ERR_PTR(-ENOMEM);
+		goto release_locks;
+	}
+
+	rmn->adev = adev;
+	rmn->mm = mm;
+	rmn->mn.ops = &amdgpu_mn_ops;
+	mutex_init(&rmn->lock);
+	rmn->objects = RB_ROOT;
+
+	r = __mmu_notifier_register(&rmn->mn, mm);
+	if (r)
+		goto free_rmn;
+
+	hash_add(adev->mn_hash, &rmn->node, (unsigned long)mm);
+
+release_locks:
+	mutex_unlock(&adev->mn_lock);
+	up_write(&mm->mmap_sem);
+
+	return rmn;
+
+free_rmn:
+	mutex_unlock(&adev->mn_lock);
+	up_write(&mm->mmap_sem);
+	kfree(rmn);
+
+	return ERR_PTR(r);
+}
+
+/**
+ * amdgpu_mn_register - register a BO for notifier updates
+ *
+ * @bo: amdgpu buffer object
+ * @addr: userptr addr we should monitor
+ *
+ * Registers an MMU notifier for the given BO at the specified address.
+ * Returns 0 on success, -ERRNO if anything goes wrong.
+ */
+int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr)
+{
+	unsigned long end = addr + amdgpu_bo_size(bo) - 1;
+	struct amdgpu_device *adev = bo->adev;
+	struct amdgpu_mn *rmn;
+	struct amdgpu_mn_node *node = NULL;
+	struct list_head bos;
+	struct interval_tree_node *it;
+
+	rmn = amdgpu_mn_get(adev);
+	if (IS_ERR(rmn))
+		return PTR_ERR(rmn);
+
+	INIT_LIST_HEAD(&bos);
+
+	mutex_lock(&rmn->lock);
+
+	while ((it = interval_tree_iter_first(&rmn->objects, addr, end))) {
+		kfree(node);
+		node = container_of(it, struct amdgpu_mn_node, it);
+		interval_tree_remove(&node->it, &rmn->objects);
+		addr = min(it->start, addr);
+		end = max(it->last, end);
+		list_splice(&node->bos, &bos);
+	}
+
+	if (!node) {
+		node = kmalloc(sizeof(struct amdgpu_mn_node), GFP_KERNEL);
+		if (!node) {
+			mutex_unlock(&rmn->lock);
+			return -ENOMEM;
+		}
+	}
+
+	bo->mn = rmn;
+
+	node->it.start = addr;
+	node->it.last = end;
+	INIT_LIST_HEAD(&node->bos);
+	list_splice(&bos, &node->bos);
+	list_add(&bo->mn_list, &node->bos);
+
+	interval_tree_insert(&node->it, &rmn->objects);
+
+	mutex_unlock(&rmn->lock);
+
+	return 0;
+}
+
+/**
+ * amdgpu_mn_unregister - unregister a BO for notifier updates
+ *
+ * @bo: amdgpu buffer object
+ *
+ * Remove any registration of MMU notifier updates from the buffer object.
+ */
+void amdgpu_mn_unregister(struct amdgpu_bo *bo)
+{
+	struct amdgpu_device *adev = bo->adev;
+	struct amdgpu_mn *rmn;
+	struct list_head *head;
+
+	mutex_lock(&adev->mn_lock);
+	rmn = bo->mn;
+	if (rmn == NULL) {
+		mutex_unlock(&adev->mn_lock);
+		return;
+	}
+
+	mutex_lock(&rmn->lock);
+	/* save the next list entry for later */
+	head = bo->mn_list.next;
+
+	bo->mn = NULL;
+	list_del(&bo->mn_list);
+
+	if (list_empty(head)) {
+		struct amdgpu_mn_node *node;
+		node = container_of(head, struct amdgpu_mn_node, bos);
+		interval_tree_remove(&node->it, &rmn->objects);
+		kfree(node);
+	}
+
+	mutex_unlock(&rmn->lock);
+	mutex_unlock(&adev->mn_lock);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
new file mode 100644
index 000000000..064ebb347
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
@@ -0,0 +1,596 @@
+/*
+ * Copyright 2000 ATI Technologies Inc., Markham, Ontario, and
+ *                VA Linux Systems Inc., Fremont, California.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Original Authors:
+ *   Kevin E. Martin, Rickard E. Faith, Alan Hourihane
+ *
+ * Kernel port Author: Dave Airlie
+ */
+
+#ifndef AMDGPU_MODE_H
+#define AMDGPU_MODE_H
+
+#include <drm/drm_crtc.h>
+#include <drm/drm_edid.h>
+#include <drm/drm_dp_helper.h>
+#include <drm/drm_fixed.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/drm_plane_helper.h>
+#include <linux/i2c.h>
+#include <linux/i2c-algo-bit.h>
+
+struct amdgpu_bo;
+struct amdgpu_device;
+struct amdgpu_encoder;
+struct amdgpu_router;
+struct amdgpu_hpd;
+
+#define to_amdgpu_crtc(x) container_of(x, struct amdgpu_crtc, base)
+#define to_amdgpu_connector(x) container_of(x, struct amdgpu_connector, base)
+#define to_amdgpu_encoder(x) container_of(x, struct amdgpu_encoder, base)
+#define to_amdgpu_framebuffer(x) container_of(x, struct amdgpu_framebuffer, base)
+
+#define AMDGPU_MAX_HPD_PINS 6
+#define AMDGPU_MAX_CRTCS 6
+#define AMDGPU_MAX_AFMT_BLOCKS 7
+
+enum amdgpu_rmx_type {
+	RMX_OFF,
+	RMX_FULL,
+	RMX_CENTER,
+	RMX_ASPECT
+};
+
+enum amdgpu_underscan_type {
+	UNDERSCAN_OFF,
+	UNDERSCAN_ON,
+	UNDERSCAN_AUTO,
+};
+
+#define AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS 50
+#define AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS 10
+
+enum amdgpu_hpd_id {
+	AMDGPU_HPD_1 = 0,
+	AMDGPU_HPD_2,
+	AMDGPU_HPD_3,
+	AMDGPU_HPD_4,
+	AMDGPU_HPD_5,
+	AMDGPU_HPD_6,
+	AMDGPU_HPD_LAST,
+	AMDGPU_HPD_NONE = 0xff,
+};
+
+enum amdgpu_crtc_irq {
+	AMDGPU_CRTC_IRQ_VBLANK1 = 0,
+	AMDGPU_CRTC_IRQ_VBLANK2,
+	AMDGPU_CRTC_IRQ_VBLANK3,
+	AMDGPU_CRTC_IRQ_VBLANK4,
+	AMDGPU_CRTC_IRQ_VBLANK5,
+	AMDGPU_CRTC_IRQ_VBLANK6,
+	AMDGPU_CRTC_IRQ_VLINE1,
+	AMDGPU_CRTC_IRQ_VLINE2,
+	AMDGPU_CRTC_IRQ_VLINE3,
+	AMDGPU_CRTC_IRQ_VLINE4,
+	AMDGPU_CRTC_IRQ_VLINE5,
+	AMDGPU_CRTC_IRQ_VLINE6,
+	AMDGPU_CRTC_IRQ_LAST,
+	AMDGPU_CRTC_IRQ_NONE = 0xff
+};
+
+enum amdgpu_pageflip_irq {
+	AMDGPU_PAGEFLIP_IRQ_D1 = 0,
+	AMDGPU_PAGEFLIP_IRQ_D2,
+	AMDGPU_PAGEFLIP_IRQ_D3,
+	AMDGPU_PAGEFLIP_IRQ_D4,
+	AMDGPU_PAGEFLIP_IRQ_D5,
+	AMDGPU_PAGEFLIP_IRQ_D6,
+	AMDGPU_PAGEFLIP_IRQ_LAST,
+	AMDGPU_PAGEFLIP_IRQ_NONE = 0xff
+};
+
+enum amdgpu_flip_status {
+	AMDGPU_FLIP_NONE,
+	AMDGPU_FLIP_PENDING,
+	AMDGPU_FLIP_SUBMITTED
+};
+
+#define AMDGPU_MAX_I2C_BUS 16
+
+/* amdgpu gpio-based i2c
+ * 1. "mask" reg and bits
+ *    grabs the gpio pins for software use
+ *    0=not held  1=held
+ * 2. "a" reg and bits
+ *    output pin value
+ *    0=low 1=high
+ * 3. "en" reg and bits
+ *    sets the pin direction
+ *    0=input 1=output
+ * 4. "y" reg and bits
+ *    input pin value
+ *    0=low 1=high
+ */
+struct amdgpu_i2c_bus_rec {
+	bool valid;
+	/* id used by atom */
+	uint8_t i2c_id;
+	/* id used by atom */
+	enum amdgpu_hpd_id hpd;
+	/* can be used with hw i2c engine */
+	bool hw_capable;
+	/* uses multi-media i2c engine */
+	bool mm_i2c;
+	/* regs and bits */
+	uint32_t mask_clk_reg;
+	uint32_t mask_data_reg;
+	uint32_t a_clk_reg;
+	uint32_t a_data_reg;
+	uint32_t en_clk_reg;
+	uint32_t en_data_reg;
+	uint32_t y_clk_reg;
+	uint32_t y_data_reg;
+	uint32_t mask_clk_mask;
+	uint32_t mask_data_mask;
+	uint32_t a_clk_mask;
+	uint32_t a_data_mask;
+	uint32_t en_clk_mask;
+	uint32_t en_data_mask;
+	uint32_t y_clk_mask;
+	uint32_t y_data_mask;
+};
+
+#define AMDGPU_MAX_BIOS_CONNECTOR 16
+
+/* pll flags */
+#define AMDGPU_PLL_USE_BIOS_DIVS        (1 << 0)
+#define AMDGPU_PLL_NO_ODD_POST_DIV      (1 << 1)
+#define AMDGPU_PLL_USE_REF_DIV          (1 << 2)
+#define AMDGPU_PLL_LEGACY               (1 << 3)
+#define AMDGPU_PLL_PREFER_LOW_REF_DIV   (1 << 4)
+#define AMDGPU_PLL_PREFER_HIGH_REF_DIV  (1 << 5)
+#define AMDGPU_PLL_PREFER_LOW_FB_DIV    (1 << 6)
+#define AMDGPU_PLL_PREFER_HIGH_FB_DIV   (1 << 7)
+#define AMDGPU_PLL_PREFER_LOW_POST_DIV  (1 << 8)
+#define AMDGPU_PLL_PREFER_HIGH_POST_DIV (1 << 9)
+#define AMDGPU_PLL_USE_FRAC_FB_DIV      (1 << 10)
+#define AMDGPU_PLL_PREFER_CLOSEST_LOWER (1 << 11)
+#define AMDGPU_PLL_USE_POST_DIV         (1 << 12)
+#define AMDGPU_PLL_IS_LCD               (1 << 13)
+#define AMDGPU_PLL_PREFER_MINM_OVER_MAXP (1 << 14)
+
+struct amdgpu_pll {
+	/* reference frequency */
+	uint32_t reference_freq;
+
+	/* fixed dividers */
+	uint32_t reference_div;
+	uint32_t post_div;
+
+	/* pll in/out limits */
+	uint32_t pll_in_min;
+	uint32_t pll_in_max;
+	uint32_t pll_out_min;
+	uint32_t pll_out_max;
+	uint32_t lcd_pll_out_min;
+	uint32_t lcd_pll_out_max;
+	uint32_t best_vco;
+
+	/* divider limits */
+	uint32_t min_ref_div;
+	uint32_t max_ref_div;
+	uint32_t min_post_div;
+	uint32_t max_post_div;
+	uint32_t min_feedback_div;
+	uint32_t max_feedback_div;
+	uint32_t min_frac_feedback_div;
+	uint32_t max_frac_feedback_div;
+
+	/* flags for the current clock */
+	uint32_t flags;
+
+	/* pll id */
+	uint32_t id;
+};
+
+struct amdgpu_i2c_chan {
+	struct i2c_adapter adapter;
+	struct drm_device *dev;
+	struct i2c_algo_bit_data bit;
+	struct amdgpu_i2c_bus_rec rec;
+	struct drm_dp_aux aux;
+	bool has_aux;
+	struct mutex mutex;
+};
+
+struct amdgpu_fbdev;
+
+struct amdgpu_afmt {
+	bool enabled;
+	int offset;
+	bool last_buffer_filled_status;
+	int id;
+	struct amdgpu_audio_pin *pin;
+};
+
+/*
+ * Audio
+ */
+struct amdgpu_audio_pin {
+	int			channels;
+	int			rate;
+	int			bits_per_sample;
+	u8			status_bits;
+	u8			category_code;
+	u32			offset;
+	bool			connected;
+	u32			id;
+};
+
+struct amdgpu_audio {
+	bool enabled;
+	struct amdgpu_audio_pin pin[AMDGPU_MAX_AFMT_BLOCKS];
+	int num_pins;
+};
+
+struct amdgpu_mode_mc_save {
+	u32 vga_render_control;
+	u32 vga_hdp_control;
+	bool crtc_enabled[AMDGPU_MAX_CRTCS];
+};
+
+struct amdgpu_display_funcs {
+	/* vga render */
+	void (*set_vga_render_state)(struct amdgpu_device *adev, bool render);
+	/* display watermarks */
+	void (*bandwidth_update)(struct amdgpu_device *adev);
+	/* get frame count */
+	u32 (*vblank_get_counter)(struct amdgpu_device *adev, int crtc);
+	/* wait for vblank */
+	void (*vblank_wait)(struct amdgpu_device *adev, int crtc);
+	/* is dce hung */
+	bool (*is_display_hung)(struct amdgpu_device *adev);
+	/* set backlight level */
+	void (*backlight_set_level)(struct amdgpu_encoder *amdgpu_encoder,
+				    u8 level);
+	/* get backlight level */
+	u8 (*backlight_get_level)(struct amdgpu_encoder *amdgpu_encoder);
+	/* hotplug detect */
+	bool (*hpd_sense)(struct amdgpu_device *adev, enum amdgpu_hpd_id hpd);
+	void (*hpd_set_polarity)(struct amdgpu_device *adev,
+				 enum amdgpu_hpd_id hpd);
+	u32 (*hpd_get_gpio_reg)(struct amdgpu_device *adev);
+	/* pageflipping */
+	void (*page_flip)(struct amdgpu_device *adev,
+			 int crtc_id, u64 crtc_base);
+	int (*page_flip_get_scanoutpos)(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position);
+	/* display topology setup */
+	void (*add_encoder)(struct amdgpu_device *adev,
+			    uint32_t encoder_enum,
+			    uint32_t supported_device,
+			    u16 caps);
+	void (*add_connector)(struct amdgpu_device *adev,
+			      uint32_t connector_id,
+			      uint32_t supported_device,
+			      int connector_type,
+			      struct amdgpu_i2c_bus_rec *i2c_bus,
+			      uint16_t connector_object_id,
+			      struct amdgpu_hpd *hpd,
+			      struct amdgpu_router *router);
+	void (*stop_mc_access)(struct amdgpu_device *adev,
+			       struct amdgpu_mode_mc_save *save);
+	void (*resume_mc_access)(struct amdgpu_device *adev,
+				 struct amdgpu_mode_mc_save *save);
+};
+
+struct amdgpu_mode_info {
+	struct atom_context *atom_context;
+	struct card_info *atom_card_info;
+	bool mode_config_initialized;
+	struct amdgpu_crtc *crtcs[6];
+	struct amdgpu_afmt *afmt[7];
+	/* DVI-I properties */
+	struct drm_property *coherent_mode_property;
+	/* DAC enable load detect */
+	struct drm_property *load_detect_property;
+	/* underscan */
+	struct drm_property *underscan_property;
+	struct drm_property *underscan_hborder_property;
+	struct drm_property *underscan_vborder_property;
+	/* audio */
+	struct drm_property *audio_property;
+	/* FMT dithering */
+	struct drm_property *dither_property;
+	/* hardcoded DFP edid from BIOS */
+	struct edid *bios_hardcoded_edid;
+	int bios_hardcoded_edid_size;
+
+	/* pointer to fbdev info structure */
+	struct amdgpu_fbdev *rfbdev;
+	/* firmware flags */
+	u16 firmware_flags;
+	/* pointer to backlight encoder */
+	struct amdgpu_encoder *bl_encoder;
+	struct amdgpu_audio	audio; /* audio stuff */
+	int			num_crtc; /* number of crtcs */
+	int			num_hpd; /* number of hpd pins */
+	int			num_dig; /* number of dig blocks */
+	int			disp_priority;
+	const struct amdgpu_display_funcs *funcs;
+};
+
+#define AMDGPU_MAX_BL_LEVEL 0xFF
+
+#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
+
+struct amdgpu_backlight_privdata {
+	struct amdgpu_encoder *encoder;
+	uint8_t negative;
+};
+
+#endif
+
+struct amdgpu_atom_ss {
+	uint16_t percentage;
+	uint16_t percentage_divider;
+	uint8_t type;
+	uint16_t step;
+	uint8_t delay;
+	uint8_t range;
+	uint8_t refdiv;
+	/* asic_ss */
+	uint16_t rate;
+	uint16_t amount;
+};
+
+struct amdgpu_crtc {
+	struct drm_crtc base;
+	int crtc_id;
+	u16 lut_r[256], lut_g[256], lut_b[256];
+	bool enabled;
+	bool can_tile;
+	uint32_t crtc_offset;
+	struct drm_gem_object *cursor_bo;
+	uint64_t cursor_addr;
+	int cursor_x;
+	int cursor_y;
+	int cursor_hot_x;
+	int cursor_hot_y;
+	int cursor_width;
+	int cursor_height;
+	int max_cursor_width;
+	int max_cursor_height;
+	enum amdgpu_rmx_type rmx_type;
+	u8 h_border;
+	u8 v_border;
+	fixed20_12 vsc;
+	fixed20_12 hsc;
+	struct drm_display_mode native_mode;
+	u32 pll_id;
+	/* page flipping */
+	struct workqueue_struct *pflip_queue;
+	struct amdgpu_flip_work *pflip_works;
+	enum amdgpu_flip_status pflip_status;
+	int deferred_flip_completion;
+	/* pll sharing */
+	struct amdgpu_atom_ss ss;
+	bool ss_enabled;
+	u32 adjusted_clock;
+	int bpc;
+	u32 pll_reference_div;
+	u32 pll_post_div;
+	u32 pll_flags;
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+	/* for dpm */
+	u32 line_time;
+	u32 wm_low;
+	u32 wm_high;
+	u32 lb_vblank_lead_lines;
+	struct drm_display_mode hw_mode;
+};
+
+struct amdgpu_encoder_atom_dig {
+	bool linkb;
+	/* atom dig */
+	bool coherent_mode;
+	int dig_encoder; /* -1 disabled, 0 DIGA, 1 DIGB, etc. */
+	/* atom lvds/edp */
+	uint32_t lcd_misc;
+	uint16_t panel_pwr_delay;
+	uint32_t lcd_ss_id;
+	/* panel mode */
+	struct drm_display_mode native_mode;
+	struct backlight_device *bl_dev;
+	int dpms_mode;
+	uint8_t backlight_level;
+	int panel_mode;
+	struct amdgpu_afmt *afmt;
+};
+
+struct amdgpu_encoder {
+	struct drm_encoder base;
+	uint32_t encoder_enum;
+	uint32_t encoder_id;
+	uint32_t devices;
+	uint32_t active_device;
+	uint32_t flags;
+	uint32_t pixel_clock;
+	enum amdgpu_rmx_type rmx_type;
+	enum amdgpu_underscan_type underscan_type;
+	uint32_t underscan_hborder;
+	uint32_t underscan_vborder;
+	struct drm_display_mode native_mode;
+	void *enc_priv;
+	int audio_polling_active;
+	bool is_ext_encoder;
+	u16 caps;
+};
+
+struct amdgpu_connector_atom_dig {
+	/* displayport */
+	u8 dpcd[DP_RECEIVER_CAP_SIZE];
+	u8 dp_sink_type;
+	int dp_clock;
+	int dp_lane_count;
+	bool edp_on;
+};
+
+struct amdgpu_gpio_rec {
+	bool valid;
+	u8 id;
+	u32 reg;
+	u32 mask;
+	u32 shift;
+};
+
+struct amdgpu_hpd {
+	enum amdgpu_hpd_id hpd;
+	u8 plugged_state;
+	struct amdgpu_gpio_rec gpio;
+};
+
+struct amdgpu_router {
+	u32 router_id;
+	struct amdgpu_i2c_bus_rec i2c_info;
+	u8 i2c_addr;
+	/* i2c mux */
+	bool ddc_valid;
+	u8 ddc_mux_type;
+	u8 ddc_mux_control_pin;
+	u8 ddc_mux_state;
+	/* clock/data mux */
+	bool cd_valid;
+	u8 cd_mux_type;
+	u8 cd_mux_control_pin;
+	u8 cd_mux_state;
+};
+
+enum amdgpu_connector_audio {
+	AMDGPU_AUDIO_DISABLE = 0,
+	AMDGPU_AUDIO_ENABLE = 1,
+	AMDGPU_AUDIO_AUTO = 2
+};
+
+enum amdgpu_connector_dither {
+	AMDGPU_FMT_DITHER_DISABLE = 0,
+	AMDGPU_FMT_DITHER_ENABLE = 1,
+};
+
+struct amdgpu_connector {
+	struct drm_connector base;
+	uint32_t connector_id;
+	uint32_t devices;
+	struct amdgpu_i2c_chan *ddc_bus;
+	/* some systems have an hdmi and vga port with a shared ddc line */
+	bool shared_ddc;
+	bool use_digital;
+	/* we need to mind the EDID between detect
+	   and get modes due to analog/digital/tvencoder */
+	struct edid *edid;
+	void *con_priv;
+	bool dac_load_detect;
+	bool detected_by_load; /* if the connection status was determined by load */
+	uint16_t connector_object_id;
+	struct amdgpu_hpd hpd;
+	struct amdgpu_router router;
+	struct amdgpu_i2c_chan *router_bus;
+	enum amdgpu_connector_audio audio;
+	enum amdgpu_connector_dither dither;
+	unsigned pixelclock_for_modeset;
+};
+
+struct amdgpu_framebuffer {
+	struct drm_framebuffer base;
+	struct drm_gem_object *obj;
+};
+
+#define ENCODER_MODE_IS_DP(em) (((em) == ATOM_ENCODER_MODE_DP) || \
+				((em) == ATOM_ENCODER_MODE_DP_MST))
+
+/* Driver internal use only flags of amdgpu_get_crtc_scanoutpos() */
+#define USE_REAL_VBLANKSTART 		(1 << 30)
+#define GET_DISTANCE_TO_VBLANKSTART	(1 << 31)
+
+void amdgpu_link_encoder_connector(struct drm_device *dev);
+
+struct drm_connector *
+amdgpu_get_connector_for_encoder(struct drm_encoder *encoder);
+struct drm_connector *
+amdgpu_get_connector_for_encoder_init(struct drm_encoder *encoder);
+bool amdgpu_dig_monitor_is_duallink(struct drm_encoder *encoder,
+				    u32 pixel_clock);
+
+u16 amdgpu_encoder_get_dp_bridge_encoder_id(struct drm_encoder *encoder);
+struct drm_encoder *amdgpu_get_external_encoder(struct drm_encoder *encoder);
+
+bool amdgpu_ddc_probe(struct amdgpu_connector *amdgpu_connector, bool use_aux);
+
+void amdgpu_encoder_set_active_device(struct drm_encoder *encoder);
+
+int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
+			       unsigned int flags, int *vpos, int *hpos,
+			       ktime_t *stime, ktime_t *etime,
+			       const struct drm_display_mode *mode);
+
+int amdgpu_framebuffer_init(struct drm_device *dev,
+			     struct amdgpu_framebuffer *rfb,
+			     struct drm_mode_fb_cmd2 *mode_cmd,
+			     struct drm_gem_object *obj);
+
+int amdgpufb_remove(struct drm_device *dev, struct drm_framebuffer *fb);
+
+void amdgpu_enc_destroy(struct drm_encoder *encoder);
+void amdgpu_copy_fb(struct drm_device *dev, struct drm_gem_object *dst_obj);
+bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
+					const struct drm_display_mode *mode,
+					struct drm_display_mode *adjusted_mode);
+void amdgpu_panel_mode_fixup(struct drm_encoder *encoder,
+			     struct drm_display_mode *adjusted_mode);
+int amdgpu_crtc_idx_to_irq_type(struct amdgpu_device *adev, int crtc);
+
+/* fbdev layer */
+int amdgpu_fbdev_init(struct amdgpu_device *adev);
+void amdgpu_fbdev_fini(struct amdgpu_device *adev);
+void amdgpu_fbdev_set_suspend(struct amdgpu_device *adev, int state);
+int amdgpu_fbdev_total_size(struct amdgpu_device *adev);
+bool amdgpu_fbdev_robj_is_fb(struct amdgpu_device *adev, struct amdgpu_bo *robj);
+void amdgpu_fbdev_restore_mode(struct amdgpu_device *adev);
+
+void amdgpu_fb_output_poll_changed(struct amdgpu_device *adev);
+
+
+int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int bpp, bool tiled);
+
+/* amdgpu_display.c */
+void amdgpu_print_display_setup(struct drm_device *dev);
+int amdgpu_modeset_create_props(struct amdgpu_device *adev);
+int amdgpu_crtc_set_config(struct drm_mode_set *set);
+int amdgpu_crtc_page_flip(struct drm_crtc *crtc,
+			  struct drm_framebuffer *fb,
+			  struct drm_pending_vblank_event *event,
+			  uint32_t page_flip_flags);
+extern const struct drm_mode_config_funcs amdgpu_mode_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
new file mode 100644
index 000000000..b8fbbd769
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -0,0 +1,667 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ *    Thomas Hellstrom <thomas-at-tungstengraphics-dot-com>
+ *    Dave Airlie
+ */
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_cache.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+
+int amdgpu_ttm_init(struct amdgpu_device *adev);
+void amdgpu_ttm_fini(struct amdgpu_device *adev);
+
+static u64 amdgpu_get_vis_part_size(struct amdgpu_device *adev,
+						struct ttm_mem_reg *mem)
+{
+	u64 ret = 0;
+	if (mem->start << PAGE_SHIFT < adev->mc.visible_vram_size) {
+		ret = (u64)((mem->start << PAGE_SHIFT) + mem->size) >
+			   adev->mc.visible_vram_size ?
+			   adev->mc.visible_vram_size - (mem->start << PAGE_SHIFT) :
+			   mem->size;
+	}
+	return ret;
+}
+
+static void amdgpu_update_memory_usage(struct amdgpu_device *adev,
+		       struct ttm_mem_reg *old_mem,
+		       struct ttm_mem_reg *new_mem)
+{
+	u64 vis_size;
+	if (!adev)
+		return;
+
+	if (new_mem) {
+		switch (new_mem->mem_type) {
+		case TTM_PL_TT:
+			atomic64_add(new_mem->size, &adev->gtt_usage);
+			break;
+		case TTM_PL_VRAM:
+			atomic64_add(new_mem->size, &adev->vram_usage);
+			vis_size = amdgpu_get_vis_part_size(adev, new_mem);
+			atomic64_add(vis_size, &adev->vram_vis_usage);
+			break;
+		}
+	}
+
+	if (old_mem) {
+		switch (old_mem->mem_type) {
+		case TTM_PL_TT:
+			atomic64_sub(old_mem->size, &adev->gtt_usage);
+			break;
+		case TTM_PL_VRAM:
+			atomic64_sub(old_mem->size, &adev->vram_usage);
+			vis_size = amdgpu_get_vis_part_size(adev, old_mem);
+			atomic64_sub(vis_size, &adev->vram_vis_usage);
+			break;
+		}
+	}
+}
+
+static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
+{
+	struct amdgpu_bo *bo;
+
+	bo = container_of(tbo, struct amdgpu_bo, tbo);
+
+	amdgpu_update_memory_usage(bo->adev, &bo->tbo.mem, NULL);
+
+	mutex_lock(&bo->adev->gem.mutex);
+	list_del_init(&bo->list);
+	mutex_unlock(&bo->adev->gem.mutex);
+	drm_gem_object_release(&bo->gem_base);
+	amdgpu_bo_unref(&bo->parent);
+	kfree(bo->metadata);
+	kfree(bo);
+}
+
+bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo)
+{
+	if (bo->destroy == &amdgpu_ttm_bo_destroy)
+		return true;
+	return false;
+}
+
+static void amdgpu_ttm_placement_init(struct amdgpu_device *adev,
+				      struct ttm_placement *placement,
+				      struct ttm_place *placements,
+				      u32 domain, u64 flags)
+{
+	u32 c = 0, i;
+
+	placement->placement = placements;
+	placement->busy_placement = placements;
+
+	if (domain & AMDGPU_GEM_DOMAIN_VRAM) {
+		if (flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS &&
+			adev->mc.visible_vram_size < adev->mc.real_vram_size) {
+			placements[c].fpfn =
+				adev->mc.visible_vram_size >> PAGE_SHIFT;
+			placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
+				TTM_PL_FLAG_VRAM | TTM_PL_FLAG_TOPDOWN;
+		}
+		placements[c].fpfn = 0;
+		placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED |
+			TTM_PL_FLAG_VRAM;
+		if (!(flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED))
+			placements[c - 1].flags |= TTM_PL_FLAG_TOPDOWN;
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_GTT) {
+		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) {
+			placements[c].fpfn = 0;
+			placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_TT |
+				TTM_PL_FLAG_UNCACHED;
+		} else {
+			placements[c].fpfn = 0;
+			placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT;
+		}
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_CPU) {
+		if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) {
+			placements[c].fpfn = 0;
+			placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_SYSTEM |
+				TTM_PL_FLAG_UNCACHED;
+		} else {
+			placements[c].fpfn = 0;
+			placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_SYSTEM;
+		}
+	}
+
+	if (domain & AMDGPU_GEM_DOMAIN_GDS) {
+		placements[c].fpfn = 0;
+		placements[c++].flags = TTM_PL_FLAG_UNCACHED |
+			AMDGPU_PL_FLAG_GDS;
+	}
+	if (domain & AMDGPU_GEM_DOMAIN_GWS) {
+		placements[c].fpfn = 0;
+		placements[c++].flags = TTM_PL_FLAG_UNCACHED |
+			AMDGPU_PL_FLAG_GWS;
+	}
+	if (domain & AMDGPU_GEM_DOMAIN_OA) {
+		placements[c].fpfn = 0;
+		placements[c++].flags = TTM_PL_FLAG_UNCACHED |
+			AMDGPU_PL_FLAG_OA;
+	}
+
+	if (!c) {
+		placements[c].fpfn = 0;
+		placements[c++].flags = TTM_PL_MASK_CACHING |
+			TTM_PL_FLAG_SYSTEM;
+	}
+	placement->num_placement = c;
+	placement->num_busy_placement = c;
+
+	for (i = 0; i < c; i++) {
+		if ((flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) &&
+			(placements[i].flags & TTM_PL_FLAG_VRAM) &&
+			!placements[i].fpfn)
+			placements[i].lpfn =
+				adev->mc.visible_vram_size >> PAGE_SHIFT;
+		else
+			placements[i].lpfn = 0;
+	}
+}
+
+void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *rbo, u32 domain)
+{
+	amdgpu_ttm_placement_init(rbo->adev, &rbo->placement,
+				  rbo->placements, domain, rbo->flags);
+}
+
+static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
+					struct ttm_placement *placement)
+{
+	BUG_ON(placement->num_placement > (AMDGPU_GEM_DOMAIN_MAX + 1));
+
+	memcpy(bo->placements, placement->placement,
+	       placement->num_placement * sizeof(struct ttm_place));
+	bo->placement.num_placement = placement->num_placement;
+	bo->placement.num_busy_placement = placement->num_busy_placement;
+	bo->placement.placement = bo->placements;
+	bo->placement.busy_placement = bo->placements;
+}
+
+int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
+				unsigned long size, int byte_align,
+				bool kernel, u32 domain, u64 flags,
+				struct sg_table *sg,
+				struct ttm_placement *placement,
+				struct reservation_object *resv,
+				struct amdgpu_bo **bo_ptr)
+{
+	struct amdgpu_bo *bo;
+	enum ttm_bo_type type;
+	unsigned long page_align;
+	size_t acc_size;
+	int r;
+
+	page_align = roundup(byte_align, PAGE_SIZE) >> PAGE_SHIFT;
+	size = ALIGN(size, PAGE_SIZE);
+
+	if (kernel) {
+		type = ttm_bo_type_kernel;
+	} else if (sg) {
+		type = ttm_bo_type_sg;
+	} else {
+		type = ttm_bo_type_device;
+	}
+	*bo_ptr = NULL;
+
+	acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
+				       sizeof(struct amdgpu_bo));
+
+	bo = kzalloc(sizeof(struct amdgpu_bo), GFP_KERNEL);
+	if (bo == NULL)
+		return -ENOMEM;
+	r = drm_gem_object_init(adev->ddev, &bo->gem_base, size);
+	if (unlikely(r)) {
+		kfree(bo);
+		return r;
+	}
+	bo->adev = adev;
+	INIT_LIST_HEAD(&bo->list);
+	INIT_LIST_HEAD(&bo->va);
+	bo->initial_domain = domain & (AMDGPU_GEM_DOMAIN_VRAM |
+				       AMDGPU_GEM_DOMAIN_GTT |
+				       AMDGPU_GEM_DOMAIN_CPU |
+				       AMDGPU_GEM_DOMAIN_GDS |
+				       AMDGPU_GEM_DOMAIN_GWS |
+				       AMDGPU_GEM_DOMAIN_OA);
+
+	bo->flags = flags;
+
+	/* For architectures that don't support WC memory,
+	 * mask out the WC flag from the BO
+	 */
+	if (!drm_arch_can_wc_memory())
+		bo->flags &= ~AMDGPU_GEM_CREATE_CPU_GTT_USWC;
+
+	amdgpu_fill_placement_to_bo(bo, placement);
+	/* Kernel allocation are uninterruptible */
+	r = ttm_bo_init(&adev->mman.bdev, &bo->tbo, size, type,
+			&bo->placement, page_align, !kernel, NULL,
+			acc_size, sg, resv, &amdgpu_ttm_bo_destroy);
+	if (unlikely(r != 0)) {
+		return r;
+	}
+	*bo_ptr = bo;
+
+	trace_amdgpu_bo_create(bo);
+
+	return 0;
+}
+
+int amdgpu_bo_create(struct amdgpu_device *adev,
+		     unsigned long size, int byte_align,
+		     bool kernel, u32 domain, u64 flags,
+		     struct sg_table *sg,
+		     struct reservation_object *resv,
+		     struct amdgpu_bo **bo_ptr)
+{
+	struct ttm_placement placement = {0};
+	struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
+
+	memset(&placements, 0,
+	       (AMDGPU_GEM_DOMAIN_MAX + 1) * sizeof(struct ttm_place));
+
+	amdgpu_ttm_placement_init(adev, &placement,
+				  placements, domain, flags);
+
+	return amdgpu_bo_create_restricted(adev, size, byte_align, kernel,
+					   domain, flags, sg, &placement,
+					   resv, bo_ptr);
+}
+
+int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
+{
+	bool is_iomem;
+	int r;
+
+	if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
+		return -EPERM;
+
+	if (bo->kptr) {
+		if (ptr) {
+			*ptr = bo->kptr;
+		}
+		return 0;
+	}
+	r = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages, &bo->kmap);
+	if (r) {
+		return r;
+	}
+	bo->kptr = ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
+	if (ptr) {
+		*ptr = bo->kptr;
+	}
+	return 0;
+}
+
+void amdgpu_bo_kunmap(struct amdgpu_bo *bo)
+{
+	if (bo->kptr == NULL)
+		return;
+	bo->kptr = NULL;
+	ttm_bo_kunmap(&bo->kmap);
+}
+
+struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo)
+{
+	if (bo == NULL)
+		return NULL;
+
+	ttm_bo_reference(&bo->tbo);
+	return bo;
+}
+
+void amdgpu_bo_unref(struct amdgpu_bo **bo)
+{
+	struct ttm_buffer_object *tbo;
+
+	if ((*bo) == NULL)
+		return;
+
+	tbo = &((*bo)->tbo);
+	ttm_bo_unref(&tbo);
+	if (tbo == NULL)
+		*bo = NULL;
+}
+
+int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
+			     u64 min_offset, u64 max_offset,
+			     u64 *gpu_addr)
+{
+	int r, i;
+	unsigned fpfn, lpfn;
+
+	if (amdgpu_ttm_tt_has_userptr(bo->tbo.ttm))
+		return -EPERM;
+
+	if (WARN_ON_ONCE(min_offset > max_offset))
+		return -EINVAL;
+
+	if (bo->pin_count) {
+		bo->pin_count++;
+		if (gpu_addr)
+			*gpu_addr = amdgpu_bo_gpu_offset(bo);
+
+		if (max_offset != 0) {
+			u64 domain_start;
+			if (domain == AMDGPU_GEM_DOMAIN_VRAM)
+				domain_start = bo->adev->mc.vram_start;
+			else
+				domain_start = bo->adev->mc.gtt_start;
+			WARN_ON_ONCE(max_offset <
+				     (amdgpu_bo_gpu_offset(bo) - domain_start));
+		}
+
+		return 0;
+	}
+	amdgpu_ttm_placement_from_domain(bo, domain);
+	for (i = 0; i < bo->placement.num_placement; i++) {
+		/* force to pin into visible video ram */
+		if ((bo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
+		    !(bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) &&
+		    (!max_offset || max_offset > bo->adev->mc.visible_vram_size)) {
+			if (WARN_ON_ONCE(min_offset >
+					 bo->adev->mc.visible_vram_size))
+				return -EINVAL;
+			fpfn = min_offset >> PAGE_SHIFT;
+			lpfn = bo->adev->mc.visible_vram_size >> PAGE_SHIFT;
+		} else {
+			fpfn = min_offset >> PAGE_SHIFT;
+			lpfn = max_offset >> PAGE_SHIFT;
+		}
+		if (fpfn > bo->placements[i].fpfn)
+			bo->placements[i].fpfn = fpfn;
+		if (!bo->placements[i].lpfn ||
+		    (lpfn && lpfn < bo->placements[i].lpfn))
+			bo->placements[i].lpfn = lpfn;
+		bo->placements[i].flags |= TTM_PL_FLAG_NO_EVICT;
+	}
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+	if (likely(r == 0)) {
+		bo->pin_count = 1;
+		if (gpu_addr != NULL)
+			*gpu_addr = amdgpu_bo_gpu_offset(bo);
+		if (domain == AMDGPU_GEM_DOMAIN_VRAM)
+			bo->adev->vram_pin_size += amdgpu_bo_size(bo);
+		else
+			bo->adev->gart_pin_size += amdgpu_bo_size(bo);
+	} else {
+		dev_err(bo->adev->dev, "%p pin failed\n", bo);
+	}
+	return r;
+}
+
+int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain, u64 *gpu_addr)
+{
+	return amdgpu_bo_pin_restricted(bo, domain, 0, 0, gpu_addr);
+}
+
+int amdgpu_bo_unpin(struct amdgpu_bo *bo)
+{
+	int r, i;
+
+	if (!bo->pin_count) {
+		dev_warn(bo->adev->dev, "%p unpin not necessary\n", bo);
+		return 0;
+	}
+	bo->pin_count--;
+	if (bo->pin_count)
+		return 0;
+	for (i = 0; i < bo->placement.num_placement; i++) {
+		bo->placements[i].lpfn = 0;
+		bo->placements[i].flags &= ~TTM_PL_FLAG_NO_EVICT;
+	}
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+	if (likely(r == 0)) {
+		if (bo->tbo.mem.mem_type == TTM_PL_VRAM)
+			bo->adev->vram_pin_size -= amdgpu_bo_size(bo);
+		else
+			bo->adev->gart_pin_size -= amdgpu_bo_size(bo);
+	} else {
+		dev_err(bo->adev->dev, "%p validate failed for unpin\n", bo);
+	}
+	return r;
+}
+
+int amdgpu_bo_evict_vram(struct amdgpu_device *adev)
+{
+	/* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */
+	if (0 && (adev->flags & AMD_IS_APU)) {
+		/* Useless to evict on IGP chips */
+		return 0;
+	}
+	return ttm_bo_evict_mm(&adev->mman.bdev, TTM_PL_VRAM);
+}
+
+void amdgpu_bo_force_delete(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo *bo, *n;
+
+	if (list_empty(&adev->gem.objects)) {
+		return;
+	}
+	dev_err(adev->dev, "Userspace still has active objects !\n");
+	list_for_each_entry_safe(bo, n, &adev->gem.objects, list) {
+		dev_err(adev->dev, "%p %p %lu %lu force free\n",
+			&bo->gem_base, bo, (unsigned long)bo->gem_base.size,
+			*((unsigned long *)&bo->gem_base.refcount));
+		mutex_lock(&bo->adev->gem.mutex);
+		list_del_init(&bo->list);
+		mutex_unlock(&bo->adev->gem.mutex);
+		/* this should unref the ttm bo */
+		drm_gem_object_unreference_unlocked(&bo->gem_base);
+	}
+}
+
+int amdgpu_bo_init(struct amdgpu_device *adev)
+{
+	/* Add an MTRR for the VRAM */
+	adev->mc.vram_mtrr = arch_phys_wc_add(adev->mc.aper_base,
+					      adev->mc.aper_size);
+	DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n",
+		adev->mc.mc_vram_size >> 20,
+		(unsigned long long)adev->mc.aper_size >> 20);
+	DRM_INFO("RAM width %dbits DDR\n",
+			adev->mc.vram_width);
+	return amdgpu_ttm_init(adev);
+}
+
+void amdgpu_bo_fini(struct amdgpu_device *adev)
+{
+	amdgpu_ttm_fini(adev);
+	arch_phys_wc_del(adev->mc.vram_mtrr);
+}
+
+int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
+			     struct vm_area_struct *vma)
+{
+	return ttm_fbdev_mmap(vma, &bo->tbo);
+}
+
+int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags)
+{
+	if (AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT) > 6)
+		return -EINVAL;
+
+	bo->tiling_flags = tiling_flags;
+	return 0;
+}
+
+void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags)
+{
+	lockdep_assert_held(&bo->tbo.resv->lock.base);
+
+	if (tiling_flags)
+		*tiling_flags = bo->tiling_flags;
+}
+
+int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
+			    uint32_t metadata_size, uint64_t flags)
+{
+	void *buffer;
+
+	if (!metadata_size) {
+		if (bo->metadata_size) {
+			kfree(bo->metadata);
+			bo->metadata_size = 0;
+		}
+		return 0;
+	}
+
+	if (metadata == NULL)
+		return -EINVAL;
+
+	buffer = kmemdup(metadata, metadata_size, GFP_KERNEL);
+	if (buffer == NULL)
+		return -ENOMEM;
+
+	kfree(bo->metadata);
+	bo->metadata_flags = flags;
+	bo->metadata = buffer;
+	bo->metadata_size = metadata_size;
+
+	return 0;
+}
+
+int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
+			   size_t buffer_size, uint32_t *metadata_size,
+			   uint64_t *flags)
+{
+	if (!buffer && !metadata_size)
+		return -EINVAL;
+
+	if (buffer) {
+		if (buffer_size < bo->metadata_size)
+			return -EINVAL;
+
+		if (bo->metadata_size)
+			memcpy(buffer, bo->metadata, bo->metadata_size);
+	}
+
+	if (metadata_size)
+		*metadata_size = bo->metadata_size;
+	if (flags)
+		*flags = bo->metadata_flags;
+
+	return 0;
+}
+
+void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
+			   struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_bo *rbo;
+
+	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
+		return;
+
+	rbo = container_of(bo, struct amdgpu_bo, tbo);
+	amdgpu_vm_bo_invalidate(rbo->adev, rbo);
+
+	/* update statistics */
+	if (!new_mem)
+		return;
+
+	/* move_notify is called before move happens */
+	amdgpu_update_memory_usage(rbo->adev, &bo->mem, new_mem);
+}
+
+int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_bo *abo;
+	unsigned long offset, size, lpfn;
+	int i, r;
+
+	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
+		return 0;
+
+	abo = container_of(bo, struct amdgpu_bo, tbo);
+	adev = abo->adev;
+	if (bo->mem.mem_type != TTM_PL_VRAM)
+		return 0;
+
+	size = bo->mem.num_pages << PAGE_SHIFT;
+	offset = bo->mem.start << PAGE_SHIFT;
+	if ((offset + size) <= adev->mc.visible_vram_size)
+		return 0;
+
+	/* hurrah the memory is not visible ! */
+	amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM);
+	lpfn =	adev->mc.visible_vram_size >> PAGE_SHIFT;
+	for (i = 0; i < abo->placement.num_placement; i++) {
+		/* Force into visible VRAM */
+		if ((abo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
+		    (!abo->placements[i].lpfn || abo->placements[i].lpfn > lpfn))
+			abo->placements[i].lpfn = lpfn;
+	}
+	r = ttm_bo_validate(bo, &abo->placement, false, false);
+	if (unlikely(r == -ENOMEM)) {
+		amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
+		return ttm_bo_validate(bo, &abo->placement, false, false);
+	} else if (unlikely(r != 0)) {
+		return r;
+	}
+
+	offset = bo->mem.start << PAGE_SHIFT;
+	/* this should never happen */
+	if ((offset + size) > adev->mc.visible_vram_size)
+		return -EINVAL;
+
+	return 0;
+}
+
+/**
+ * amdgpu_bo_fence - add fence to buffer object
+ *
+ * @bo: buffer object in question
+ * @fence: fence to add
+ * @shared: true if fence should be added shared
+ *
+ */
+void amdgpu_bo_fence(struct amdgpu_bo *bo, struct fence *fence,
+		     bool shared)
+{
+	struct reservation_object *resv = bo->tbo.resv;
+
+	if (shared)
+		reservation_object_add_shared_fence(resv, fence);
+	else
+		reservation_object_add_excl_fence(resv, fence);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
new file mode 100644
index 000000000..ea756e77b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -0,0 +1,204 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#ifndef __AMDGPU_OBJECT_H__
+#define __AMDGPU_OBJECT_H__
+
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+/**
+ * amdgpu_mem_type_to_domain - return domain corresponding to mem_type
+ * @mem_type:	ttm memory type
+ *
+ * Returns corresponding domain of the ttm mem_type
+ */
+static inline unsigned amdgpu_mem_type_to_domain(u32 mem_type)
+{
+	switch (mem_type) {
+	case TTM_PL_VRAM:
+		return AMDGPU_GEM_DOMAIN_VRAM;
+	case TTM_PL_TT:
+		return AMDGPU_GEM_DOMAIN_GTT;
+	case TTM_PL_SYSTEM:
+		return AMDGPU_GEM_DOMAIN_CPU;
+	case AMDGPU_PL_GDS:
+		return AMDGPU_GEM_DOMAIN_GDS;
+	case AMDGPU_PL_GWS:
+		return AMDGPU_GEM_DOMAIN_GWS;
+	case AMDGPU_PL_OA:
+		return AMDGPU_GEM_DOMAIN_OA;
+	default:
+		break;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_bo_reserve - reserve bo
+ * @bo:		bo structure
+ * @no_intr:	don't return -ERESTARTSYS on pending signal
+ *
+ * Returns:
+ * -ERESTARTSYS: A wait for the buffer to become unreserved was interrupted by
+ * a signal. Release all buffer reservations and return to user-space.
+ */
+static inline int amdgpu_bo_reserve(struct amdgpu_bo *bo, bool no_intr)
+{
+	int r;
+
+	r = ttm_bo_reserve(&bo->tbo, !no_intr, false, false, 0);
+	if (unlikely(r != 0)) {
+		if (r != -ERESTARTSYS)
+			dev_err(bo->adev->dev, "%p reserve failed\n", bo);
+		return r;
+	}
+	return 0;
+}
+
+static inline void amdgpu_bo_unreserve(struct amdgpu_bo *bo)
+{
+	ttm_bo_unreserve(&bo->tbo);
+}
+
+/**
+ * amdgpu_bo_gpu_offset - return GPU offset of bo
+ * @bo:	amdgpu object for which we query the offset
+ *
+ * Returns current GPU offset of the object.
+ *
+ * Note: object should either be pinned or reserved when calling this
+ * function, it might be useful to add check for this for debugging.
+ */
+static inline u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
+{
+	return bo->tbo.offset;
+}
+
+static inline unsigned long amdgpu_bo_size(struct amdgpu_bo *bo)
+{
+	return bo->tbo.num_pages << PAGE_SHIFT;
+}
+
+static inline unsigned amdgpu_bo_ngpu_pages(struct amdgpu_bo *bo)
+{
+	return (bo->tbo.num_pages << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
+}
+
+static inline unsigned amdgpu_bo_gpu_page_alignment(struct amdgpu_bo *bo)
+{
+	return (bo->tbo.mem.page_alignment << PAGE_SHIFT) / AMDGPU_GPU_PAGE_SIZE;
+}
+
+/**
+ * amdgpu_bo_mmap_offset - return mmap offset of bo
+ * @bo:	amdgpu object for which we query the offset
+ *
+ * Returns mmap offset of the object.
+ */
+static inline u64 amdgpu_bo_mmap_offset(struct amdgpu_bo *bo)
+{
+	return drm_vma_node_offset_addr(&bo->tbo.vma_node);
+}
+
+int amdgpu_bo_create(struct amdgpu_device *adev,
+			    unsigned long size, int byte_align,
+			    bool kernel, u32 domain, u64 flags,
+			    struct sg_table *sg,
+			    struct reservation_object *resv,
+			    struct amdgpu_bo **bo_ptr);
+int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
+				unsigned long size, int byte_align,
+				bool kernel, u32 domain, u64 flags,
+				struct sg_table *sg,
+				struct ttm_placement *placement,
+			        struct reservation_object *resv,
+				struct amdgpu_bo **bo_ptr);
+int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr);
+void amdgpu_bo_kunmap(struct amdgpu_bo *bo);
+struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo);
+void amdgpu_bo_unref(struct amdgpu_bo **bo);
+int amdgpu_bo_pin(struct amdgpu_bo *bo, u32 domain, u64 *gpu_addr);
+int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
+			     u64 min_offset, u64 max_offset,
+			     u64 *gpu_addr);
+int amdgpu_bo_unpin(struct amdgpu_bo *bo);
+int amdgpu_bo_evict_vram(struct amdgpu_device *adev);
+void amdgpu_bo_force_delete(struct amdgpu_device *adev);
+int amdgpu_bo_init(struct amdgpu_device *adev);
+void amdgpu_bo_fini(struct amdgpu_device *adev);
+int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
+				struct vm_area_struct *vma);
+int amdgpu_bo_set_tiling_flags(struct amdgpu_bo *bo, u64 tiling_flags);
+void amdgpu_bo_get_tiling_flags(struct amdgpu_bo *bo, u64 *tiling_flags);
+int amdgpu_bo_set_metadata (struct amdgpu_bo *bo, void *metadata,
+			    uint32_t metadata_size, uint64_t flags);
+int amdgpu_bo_get_metadata(struct amdgpu_bo *bo, void *buffer,
+			   size_t buffer_size, uint32_t *metadata_size,
+			   uint64_t *flags);
+void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
+				  struct ttm_mem_reg *new_mem);
+int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo);
+void amdgpu_bo_fence(struct amdgpu_bo *bo, struct fence *fence,
+		     bool shared);
+
+/*
+ * sub allocation
+ */
+
+static inline uint64_t amdgpu_sa_bo_gpu_addr(struct amdgpu_sa_bo *sa_bo)
+{
+	return sa_bo->manager->gpu_addr + sa_bo->soffset;
+}
+
+static inline void * amdgpu_sa_bo_cpu_addr(struct amdgpu_sa_bo *sa_bo)
+{
+	return sa_bo->manager->cpu_ptr + sa_bo->soffset;
+}
+
+int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
+				     struct amdgpu_sa_manager *sa_manager,
+				     unsigned size, u32 align, u32 domain);
+void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
+				      struct amdgpu_sa_manager *sa_manager);
+int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
+				      struct amdgpu_sa_manager *sa_manager);
+int amdgpu_sa_bo_manager_suspend(struct amdgpu_device *adev,
+					struct amdgpu_sa_manager *sa_manager);
+int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
+		     struct amdgpu_sa_bo **sa_bo,
+		     unsigned size, unsigned align);
+void amdgpu_sa_bo_free(struct amdgpu_device *adev,
+			      struct amdgpu_sa_bo **sa_bo,
+			      struct fence *fence);
+#if defined(CONFIG_DEBUG_FS)
+void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
+					 struct seq_file *m);
+#endif
+
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
new file mode 100644
index 000000000..d15314957
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
@@ -0,0 +1,350 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+#include <asm/div64.h>
+#include <linux/gcd.h>
+
+/**
+ * amdgpu_pll_reduce_ratio - fractional number reduction
+ *
+ * @nom: nominator
+ * @den: denominator
+ * @nom_min: minimum value for nominator
+ * @den_min: minimum value for denominator
+ *
+ * Find the greatest common divisor and apply it on both nominator and
+ * denominator, but make nominator and denominator are at least as large
+ * as their minimum values.
+ */
+static void amdgpu_pll_reduce_ratio(unsigned *nom, unsigned *den,
+				    unsigned nom_min, unsigned den_min)
+{
+	unsigned tmp;
+
+	/* reduce the numbers to a simpler ratio */
+	tmp = gcd(*nom, *den);
+	*nom /= tmp;
+	*den /= tmp;
+
+	/* make sure nominator is large enough */
+	if (*nom < nom_min) {
+		tmp = DIV_ROUND_UP(nom_min, *nom);
+		*nom *= tmp;
+		*den *= tmp;
+	}
+
+	/* make sure the denominator is large enough */
+	if (*den < den_min) {
+		tmp = DIV_ROUND_UP(den_min, *den);
+		*nom *= tmp;
+		*den *= tmp;
+	}
+}
+
+/**
+ * amdgpu_pll_get_fb_ref_div - feedback and ref divider calculation
+ *
+ * @nom: nominator
+ * @den: denominator
+ * @post_div: post divider
+ * @fb_div_max: feedback divider maximum
+ * @ref_div_max: reference divider maximum
+ * @fb_div: resulting feedback divider
+ * @ref_div: resulting reference divider
+ *
+ * Calculate feedback and reference divider for a given post divider. Makes
+ * sure we stay within the limits.
+ */
+static void amdgpu_pll_get_fb_ref_div(unsigned nom, unsigned den, unsigned post_div,
+				      unsigned fb_div_max, unsigned ref_div_max,
+				      unsigned *fb_div, unsigned *ref_div)
+{
+	/* limit reference * post divider to a maximum */
+	ref_div_max = min(128 / post_div, ref_div_max);
+
+	/* get matching reference and feedback divider */
+	*ref_div = min(max(DIV_ROUND_CLOSEST(den, post_div), 1u), ref_div_max);
+	*fb_div = DIV_ROUND_CLOSEST(nom * *ref_div * post_div, den);
+
+	/* limit fb divider to its maximum */
+	if (*fb_div > fb_div_max) {
+		*ref_div = DIV_ROUND_CLOSEST(*ref_div * fb_div_max, *fb_div);
+		*fb_div = fb_div_max;
+	}
+}
+
+/**
+ * amdgpu_pll_compute - compute PLL paramaters
+ *
+ * @pll: information about the PLL
+ * @dot_clock_p: resulting pixel clock
+ * fb_div_p: resulting feedback divider
+ * frac_fb_div_p: fractional part of the feedback divider
+ * ref_div_p: resulting reference divider
+ * post_div_p: resulting reference divider
+ *
+ * Try to calculate the PLL parameters to generate the given frequency:
+ * dot_clock = (ref_freq * feedback_div) / (ref_div * post_div)
+ */
+void amdgpu_pll_compute(struct amdgpu_pll *pll,
+			u32 freq,
+			u32 *dot_clock_p,
+			u32 *fb_div_p,
+			u32 *frac_fb_div_p,
+			u32 *ref_div_p,
+			u32 *post_div_p)
+{
+	unsigned target_clock = pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV ?
+		freq : freq / 10;
+
+	unsigned fb_div_min, fb_div_max, fb_div;
+	unsigned post_div_min, post_div_max, post_div;
+	unsigned ref_div_min, ref_div_max, ref_div;
+	unsigned post_div_best, diff_best;
+	unsigned nom, den;
+
+	/* determine allowed feedback divider range */
+	fb_div_min = pll->min_feedback_div;
+	fb_div_max = pll->max_feedback_div;
+
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV) {
+		fb_div_min *= 10;
+		fb_div_max *= 10;
+	}
+
+	/* determine allowed ref divider range */
+	if (pll->flags & AMDGPU_PLL_USE_REF_DIV)
+		ref_div_min = pll->reference_div;
+	else
+		ref_div_min = pll->min_ref_div;
+
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV &&
+	    pll->flags & AMDGPU_PLL_USE_REF_DIV)
+		ref_div_max = pll->reference_div;
+	else
+		ref_div_max = pll->max_ref_div;
+
+	/* determine allowed post divider range */
+	if (pll->flags & AMDGPU_PLL_USE_POST_DIV) {
+		post_div_min = pll->post_div;
+		post_div_max = pll->post_div;
+	} else {
+		unsigned vco_min, vco_max;
+
+		if (pll->flags & AMDGPU_PLL_IS_LCD) {
+			vco_min = pll->lcd_pll_out_min;
+			vco_max = pll->lcd_pll_out_max;
+		} else {
+			vco_min = pll->pll_out_min;
+			vco_max = pll->pll_out_max;
+		}
+
+		if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV) {
+			vco_min *= 10;
+			vco_max *= 10;
+		}
+
+		post_div_min = vco_min / target_clock;
+		if ((target_clock * post_div_min) < vco_min)
+			++post_div_min;
+		if (post_div_min < pll->min_post_div)
+			post_div_min = pll->min_post_div;
+
+		post_div_max = vco_max / target_clock;
+		if ((target_clock * post_div_max) > vco_max)
+			--post_div_max;
+		if (post_div_max > pll->max_post_div)
+			post_div_max = pll->max_post_div;
+	}
+
+	/* represent the searched ratio as fractional number */
+	nom = target_clock;
+	den = pll->reference_freq;
+
+	/* reduce the numbers to a simpler ratio */
+	amdgpu_pll_reduce_ratio(&nom, &den, fb_div_min, post_div_min);
+
+	/* now search for a post divider */
+	if (pll->flags & AMDGPU_PLL_PREFER_MINM_OVER_MAXP)
+		post_div_best = post_div_min;
+	else
+		post_div_best = post_div_max;
+	diff_best = ~0;
+
+	for (post_div = post_div_min; post_div <= post_div_max; ++post_div) {
+		unsigned diff;
+		amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max,
+					  ref_div_max, &fb_div, &ref_div);
+		diff = abs(target_clock - (pll->reference_freq * fb_div) /
+			(ref_div * post_div));
+
+		if (diff < diff_best || (diff == diff_best &&
+		    !(pll->flags & AMDGPU_PLL_PREFER_MINM_OVER_MAXP))) {
+
+			post_div_best = post_div;
+			diff_best = diff;
+		}
+	}
+	post_div = post_div_best;
+
+	/* get the feedback and reference divider for the optimal value */
+	amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max, ref_div_max,
+				  &fb_div, &ref_div);
+
+	/* reduce the numbers to a simpler ratio once more */
+	/* this also makes sure that the reference divider is large enough */
+	amdgpu_pll_reduce_ratio(&fb_div, &ref_div, fb_div_min, ref_div_min);
+
+	/* avoid high jitter with small fractional dividers */
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV && (fb_div % 10)) {
+		fb_div_min = max(fb_div_min, (9 - (fb_div % 10)) * 20 + 60);
+		if (fb_div < fb_div_min) {
+			unsigned tmp = DIV_ROUND_UP(fb_div_min, fb_div);
+			fb_div *= tmp;
+			ref_div *= tmp;
+		}
+	}
+
+	/* and finally save the result */
+	if (pll->flags & AMDGPU_PLL_USE_FRAC_FB_DIV) {
+		*fb_div_p = fb_div / 10;
+		*frac_fb_div_p = fb_div % 10;
+	} else {
+		*fb_div_p = fb_div;
+		*frac_fb_div_p = 0;
+	}
+
+	*dot_clock_p = ((pll->reference_freq * *fb_div_p * 10) +
+			(pll->reference_freq * *frac_fb_div_p)) /
+		       (ref_div * post_div * 10);
+	*ref_div_p = ref_div;
+	*post_div_p = post_div;
+
+	DRM_DEBUG_KMS("%d - %d, pll dividers - fb: %d.%d ref: %d, post %d\n",
+		      freq, *dot_clock_p * 10, *fb_div_p, *frac_fb_div_p,
+		      ref_div, post_div);
+}
+
+/**
+ * amdgpu_pll_get_use_mask - look up a mask of which pplls are in use
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the mask of which PPLLs (Pixel PLLs) are in use.
+ */
+u32 amdgpu_pll_get_use_mask(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_crtc *test_crtc;
+	struct amdgpu_crtc *test_amdgpu_crtc;
+	u32 pll_in_use = 0;
+
+	list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) {
+		if (crtc == test_crtc)
+			continue;
+
+		test_amdgpu_crtc = to_amdgpu_crtc(test_crtc);
+		if (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID)
+			pll_in_use |= (1 << test_amdgpu_crtc->pll_id);
+	}
+	return pll_in_use;
+}
+
+/**
+ * amdgpu_pll_get_shared_dp_ppll - return the PPLL used by another crtc for DP
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) used by another crtc/encoder which is
+ * also in DP mode.  For DP, a single PPLL can be used for all DP
+ * crtcs/encoders.
+ */
+int amdgpu_pll_get_shared_dp_ppll(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct drm_crtc *test_crtc;
+	struct amdgpu_crtc *test_amdgpu_crtc;
+
+	list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) {
+		if (crtc == test_crtc)
+			continue;
+		test_amdgpu_crtc = to_amdgpu_crtc(test_crtc);
+		if (test_amdgpu_crtc->encoder &&
+		    ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(test_amdgpu_crtc->encoder))) {
+			/* for DP use the same PLL for all */
+			if (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID)
+				return test_amdgpu_crtc->pll_id;
+		}
+	}
+	return ATOM_PPLL_INVALID;
+}
+
+/**
+ * amdgpu_pll_get_shared_nondp_ppll - return the PPLL used by another non-DP crtc
+ *
+ * @crtc: drm crtc
+ * @encoder: drm encoder
+ *
+ * Returns the PPLL (Pixel PLL) used by another non-DP crtc/encoder which can
+ * be shared (i.e., same clock).
+ */
+int amdgpu_pll_get_shared_nondp_ppll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_crtc *test_crtc;
+	struct amdgpu_crtc *test_amdgpu_crtc;
+	u32 adjusted_clock, test_adjusted_clock;
+
+	adjusted_clock = amdgpu_crtc->adjusted_clock;
+
+	if (adjusted_clock == 0)
+		return ATOM_PPLL_INVALID;
+
+	list_for_each_entry(test_crtc, &dev->mode_config.crtc_list, head) {
+		if (crtc == test_crtc)
+			continue;
+		test_amdgpu_crtc = to_amdgpu_crtc(test_crtc);
+		if (test_amdgpu_crtc->encoder &&
+		    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(test_amdgpu_crtc->encoder))) {
+			/* check if we are already driving this connector with another crtc */
+			if (test_amdgpu_crtc->connector == amdgpu_crtc->connector) {
+				/* if we are, return that pll */
+				if (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID)
+					return test_amdgpu_crtc->pll_id;
+			}
+			/* for non-DP check the clock */
+			test_adjusted_clock = test_amdgpu_crtc->adjusted_clock;
+			if ((crtc->mode.clock == test_crtc->mode.clock) &&
+			    (adjusted_clock == test_adjusted_clock) &&
+			    (amdgpu_crtc->ss_enabled == test_amdgpu_crtc->ss_enabled) &&
+			    (test_amdgpu_crtc->pll_id != ATOM_PPLL_INVALID))
+				return test_amdgpu_crtc->pll_id;
+		}
+	}
+	return ATOM_PPLL_INVALID;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
new file mode 100644
index 000000000..db6136f68
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_PLL_H__
+#define __AMDGPU_PLL_H__
+
+void amdgpu_pll_compute(struct amdgpu_pll *pll,
+			 u32 freq,
+			 u32 *dot_clock_p,
+			 u32 *fb_div_p,
+			 u32 *frac_fb_div_p,
+			 u32 *ref_div_p,
+			 u32 *post_div_p);
+u32 amdgpu_pll_get_use_mask(struct drm_crtc *crtc);
+int amdgpu_pll_get_shared_dp_ppll(struct drm_crtc *crtc);
+int amdgpu_pll_get_shared_nondp_ppll(struct drm_crtc *crtc);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
new file mode 100644
index 000000000..7ae15fad1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
@@ -0,0 +1,815 @@
+/*
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Rafał Miłecki <zajec5@gmail.com>
+ *          Alex Deucher <alexdeucher@gmail.com>
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_drv.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_dpm.h"
+#include "atom.h"
+#include <linux/power_supply.h>
+#include <linux/hwmon.h>
+#include <linux/hwmon-sysfs.h>
+
+static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev);
+
+void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		if (power_supply_is_system_supplied() > 0)
+			adev->pm.dpm.ac_power = true;
+		else
+			adev->pm.dpm.ac_power = false;
+		if (adev->pm.funcs->enable_bapm)
+			amdgpu_dpm_enable_bapm(adev, adev->pm.dpm.ac_power);
+		mutex_unlock(&adev->pm.mutex);
+	}
+}
+
+static ssize_t amdgpu_get_dpm_state(struct device *dev,
+				    struct device_attribute *attr,
+				    char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amdgpu_pm_state_type pm = adev->pm.dpm.user_state;
+
+	return snprintf(buf, PAGE_SIZE, "%s\n",
+			(pm == POWER_STATE_TYPE_BATTERY) ? "battery" :
+			(pm == POWER_STATE_TYPE_BALANCED) ? "balanced" : "performance");
+}
+
+static ssize_t amdgpu_set_dpm_state(struct device *dev,
+				    struct device_attribute *attr,
+				    const char *buf,
+				    size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+
+	mutex_lock(&adev->pm.mutex);
+	if (strncmp("battery", buf, strlen("battery")) == 0)
+		adev->pm.dpm.user_state = POWER_STATE_TYPE_BATTERY;
+	else if (strncmp("balanced", buf, strlen("balanced")) == 0)
+		adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	else if (strncmp("performance", buf, strlen("performance")) == 0)
+		adev->pm.dpm.user_state = POWER_STATE_TYPE_PERFORMANCE;
+	else {
+		mutex_unlock(&adev->pm.mutex);
+		count = -EINVAL;
+		goto fail;
+	}
+	mutex_unlock(&adev->pm.mutex);
+
+	/* Can't set dpm state when the card is off */
+	if (!(adev->flags & AMD_IS_PX) ||
+	    (ddev->switch_power_state == DRM_SWITCH_POWER_ON))
+		amdgpu_pm_compute_clocks(adev);
+fail:
+	return count;
+}
+
+static ssize_t amdgpu_get_dpm_forced_performance_level(struct device *dev,
+						       struct device_attribute *attr,
+						       char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amdgpu_dpm_forced_level level = adev->pm.dpm.forced_level;
+
+	return snprintf(buf, PAGE_SIZE, "%s\n",
+			(level == AMDGPU_DPM_FORCED_LEVEL_AUTO) ? "auto" :
+			(level == AMDGPU_DPM_FORCED_LEVEL_LOW) ? "low" : "high");
+}
+
+static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev,
+						       struct device_attribute *attr,
+						       const char *buf,
+						       size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	enum amdgpu_dpm_forced_level level;
+	int ret = 0;
+
+	mutex_lock(&adev->pm.mutex);
+	if (strncmp("low", buf, strlen("low")) == 0) {
+		level = AMDGPU_DPM_FORCED_LEVEL_LOW;
+	} else if (strncmp("high", buf, strlen("high")) == 0) {
+		level = AMDGPU_DPM_FORCED_LEVEL_HIGH;
+	} else if (strncmp("auto", buf, strlen("auto")) == 0) {
+		level = AMDGPU_DPM_FORCED_LEVEL_AUTO;
+	} else {
+		count = -EINVAL;
+		goto fail;
+	}
+	if (adev->pm.funcs->force_performance_level) {
+		if (adev->pm.dpm.thermal_active) {
+			count = -EINVAL;
+			goto fail;
+		}
+		ret = amdgpu_dpm_force_performance_level(adev, level);
+		if (ret)
+			count = -EINVAL;
+	}
+fail:
+	mutex_unlock(&adev->pm.mutex);
+
+	return count;
+}
+
+static DEVICE_ATTR(power_dpm_state, S_IRUGO | S_IWUSR, amdgpu_get_dpm_state, amdgpu_set_dpm_state);
+static DEVICE_ATTR(power_dpm_force_performance_level, S_IRUGO | S_IWUSR,
+		   amdgpu_get_dpm_forced_performance_level,
+		   amdgpu_set_dpm_forced_performance_level);
+
+static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
+				      struct device_attribute *attr,
+				      char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int temp;
+
+	if (adev->pm.funcs->get_temperature)
+		temp = amdgpu_dpm_get_temperature(adev);
+	else
+		temp = 0;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
+
+static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
+					     struct device_attribute *attr,
+					     char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int hyst = to_sensor_dev_attr(attr)->index;
+	int temp;
+
+	if (hyst)
+		temp = adev->pm.dpm.thermal.min_temp;
+	else
+		temp = adev->pm.dpm.thermal.max_temp;
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
+					    struct device_attribute *attr,
+					    char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	u32 pwm_mode = 0;
+
+	if (adev->pm.funcs->get_fan_control_mode)
+		pwm_mode = amdgpu_dpm_get_fan_control_mode(adev);
+
+	/* never 0 (full-speed), fuse or smc-controlled always */
+	return sprintf(buf, "%i\n", pwm_mode == FDO_PWM_MODE_STATIC ? 1 : 2);
+}
+
+static ssize_t amdgpu_hwmon_set_pwm1_enable(struct device *dev,
+					    struct device_attribute *attr,
+					    const char *buf,
+					    size_t count)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	int value;
+
+	if(!adev->pm.funcs->set_fan_control_mode)
+		return -EINVAL;
+
+	err = kstrtoint(buf, 10, &value);
+	if (err)
+		return err;
+
+	switch (value) {
+	case 1: /* manual, percent-based */
+		amdgpu_dpm_set_fan_control_mode(adev, FDO_PWM_MODE_STATIC);
+		break;
+	default: /* disable */
+		amdgpu_dpm_set_fan_control_mode(adev, 0);
+		break;
+	}
+
+	return count;
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1_min(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	return sprintf(buf, "%i\n", 0);
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1_max(struct device *dev,
+					 struct device_attribute *attr,
+					 char *buf)
+{
+	return sprintf(buf, "%i\n", 255);
+}
+
+static ssize_t amdgpu_hwmon_set_pwm1(struct device *dev,
+				     struct device_attribute *attr,
+				     const char *buf, size_t count)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	u32 value;
+
+	err = kstrtou32(buf, 10, &value);
+	if (err)
+		return err;
+
+	value = (value * 100) / 255;
+
+	err = amdgpu_dpm_set_fan_speed_percent(adev, value);
+	if (err)
+		return err;
+
+	return count;
+}
+
+static ssize_t amdgpu_hwmon_get_pwm1(struct device *dev,
+				     struct device_attribute *attr,
+				     char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int err;
+	u32 speed;
+
+	err = amdgpu_dpm_get_fan_speed_percent(adev, &speed);
+	if (err)
+		return err;
+
+	speed = (speed * 255) / 100;
+
+	return sprintf(buf, "%i\n", speed);
+}
+
+static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(pwm1, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1, amdgpu_hwmon_set_pwm1, 0);
+static SENSOR_DEVICE_ATTR(pwm1_enable, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1_enable, amdgpu_hwmon_set_pwm1_enable, 0);
+static SENSOR_DEVICE_ATTR(pwm1_min, S_IRUGO, amdgpu_hwmon_get_pwm1_min, NULL, 0);
+static SENSOR_DEVICE_ATTR(pwm1_max, S_IRUGO, amdgpu_hwmon_get_pwm1_max, NULL, 0);
+
+static struct attribute *hwmon_attributes[] = {
+	&sensor_dev_attr_temp1_input.dev_attr.attr,
+	&sensor_dev_attr_temp1_crit.dev_attr.attr,
+	&sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+	&sensor_dev_attr_pwm1.dev_attr.attr,
+	&sensor_dev_attr_pwm1_enable.dev_attr.attr,
+	&sensor_dev_attr_pwm1_min.dev_attr.attr,
+	&sensor_dev_attr_pwm1_max.dev_attr.attr,
+	NULL
+};
+
+static umode_t hwmon_attributes_visible(struct kobject *kobj,
+					struct attribute *attr, int index)
+{
+	struct device *dev = container_of(kobj, struct device, kobj);
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	umode_t effective_mode = attr->mode;
+
+	/* Skip attributes if DPM is not enabled */
+	if (!adev->pm.dpm_enabled &&
+	    (attr == &sensor_dev_attr_temp1_crit.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
+		return 0;
+
+	/* Skip fan attributes if fan is not present */
+	if (adev->pm.no_fan &&
+	    (attr == &sensor_dev_attr_pwm1.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
+		return 0;
+
+	/* mask fan attributes if we have no bindings for this asic to expose */
+	if ((!adev->pm.funcs->get_fan_speed_percent &&
+	     attr == &sensor_dev_attr_pwm1.dev_attr.attr) || /* can't query fan */
+	    (!adev->pm.funcs->get_fan_control_mode &&
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't query state */
+		effective_mode &= ~S_IRUGO;
+
+	if ((!adev->pm.funcs->set_fan_speed_percent &&
+	     attr == &sensor_dev_attr_pwm1.dev_attr.attr) || /* can't manage fan */
+	    (!adev->pm.funcs->set_fan_control_mode &&
+	     attr == &sensor_dev_attr_pwm1_enable.dev_attr.attr)) /* can't manage state */
+		effective_mode &= ~S_IWUSR;
+
+	/* hide max/min values if we can't both query and manage the fan */
+	if ((!adev->pm.funcs->set_fan_speed_percent &&
+	     !adev->pm.funcs->get_fan_speed_percent) &&
+	    (attr == &sensor_dev_attr_pwm1_max.dev_attr.attr ||
+	     attr == &sensor_dev_attr_pwm1_min.dev_attr.attr))
+		return 0;
+
+	return effective_mode;
+}
+
+static const struct attribute_group hwmon_attrgroup = {
+	.attrs = hwmon_attributes,
+	.is_visible = hwmon_attributes_visible,
+};
+
+static const struct attribute_group *hwmon_groups[] = {
+	&hwmon_attrgroup,
+	NULL
+};
+
+void amdgpu_dpm_thermal_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device,
+			     pm.dpm.thermal.work);
+	/* switch to the thermal state */
+	enum amdgpu_pm_state_type dpm_state = POWER_STATE_TYPE_INTERNAL_THERMAL;
+
+	if (!adev->pm.dpm_enabled)
+		return;
+
+	if (adev->pm.funcs->get_temperature) {
+		int temp = amdgpu_dpm_get_temperature(adev);
+
+		if (temp < adev->pm.dpm.thermal.min_temp)
+			/* switch back the user state */
+			dpm_state = adev->pm.dpm.user_state;
+	} else {
+		if (adev->pm.dpm.thermal.high_to_low)
+			/* switch back the user state */
+			dpm_state = adev->pm.dpm.user_state;
+	}
+	mutex_lock(&adev->pm.mutex);
+	if (dpm_state == POWER_STATE_TYPE_INTERNAL_THERMAL)
+		adev->pm.dpm.thermal_active = true;
+	else
+		adev->pm.dpm.thermal_active = false;
+	adev->pm.dpm.state = dpm_state;
+	mutex_unlock(&adev->pm.mutex);
+
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static struct amdgpu_ps *amdgpu_dpm_pick_power_state(struct amdgpu_device *adev,
+						     enum amdgpu_pm_state_type dpm_state)
+{
+	int i;
+	struct amdgpu_ps *ps;
+	u32 ui_class;
+	bool single_display = (adev->pm.dpm.new_active_crtc_count < 2) ?
+		true : false;
+
+	/* check if the vblank period is too short to adjust the mclk */
+	if (single_display && adev->pm.funcs->vblank_too_short) {
+		if (amdgpu_dpm_vblank_too_short(adev))
+			single_display = false;
+	}
+
+	/* certain older asics have a separare 3D performance state,
+	 * so try that first if the user selected performance
+	 */
+	if (dpm_state == POWER_STATE_TYPE_PERFORMANCE)
+		dpm_state = POWER_STATE_TYPE_INTERNAL_3DPERF;
+	/* balanced states don't exist at the moment */
+	if (dpm_state == POWER_STATE_TYPE_BALANCED)
+		dpm_state = POWER_STATE_TYPE_PERFORMANCE;
+
+restart_search:
+	/* Pick the best power state based on current conditions */
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		ps = &adev->pm.dpm.ps[i];
+		ui_class = ps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK;
+		switch (dpm_state) {
+		/* user states */
+		case POWER_STATE_TYPE_BATTERY:
+			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY) {
+				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
+					if (single_display)
+						return ps;
+				} else
+					return ps;
+			}
+			break;
+		case POWER_STATE_TYPE_BALANCED:
+			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_BALANCED) {
+				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
+					if (single_display)
+						return ps;
+				} else
+					return ps;
+			}
+			break;
+		case POWER_STATE_TYPE_PERFORMANCE:
+			if (ui_class == ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE) {
+				if (ps->caps & ATOM_PPLIB_SINGLE_DISPLAY_ONLY) {
+					if (single_display)
+						return ps;
+				} else
+					return ps;
+			}
+			break;
+		/* internal states */
+		case POWER_STATE_TYPE_INTERNAL_UVD:
+			if (adev->pm.dpm.uvd_ps)
+				return adev->pm.dpm.uvd_ps;
+			else
+				break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_SD:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_SDSTATE)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_HD:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_HDSTATE)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_HD2:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_HD2STATE)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_UVD_MVC:
+			if (ps->class2 & ATOM_PPLIB_CLASSIFICATION2_MVC)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_BOOT:
+			return adev->pm.dpm.boot_ps;
+		case POWER_STATE_TYPE_INTERNAL_THERMAL:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_THERMAL)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_ACPI:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_ACPI)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_ULV:
+			if (ps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV)
+				return ps;
+			break;
+		case POWER_STATE_TYPE_INTERNAL_3DPERF:
+			if (ps->class & ATOM_PPLIB_CLASSIFICATION_3DPERFORMANCE)
+				return ps;
+			break;
+		default:
+			break;
+		}
+	}
+	/* use a fallback state if we didn't match */
+	switch (dpm_state) {
+	case POWER_STATE_TYPE_INTERNAL_UVD_SD:
+		dpm_state = POWER_STATE_TYPE_INTERNAL_UVD_HD;
+		goto restart_search;
+	case POWER_STATE_TYPE_INTERNAL_UVD_HD:
+	case POWER_STATE_TYPE_INTERNAL_UVD_HD2:
+	case POWER_STATE_TYPE_INTERNAL_UVD_MVC:
+		if (adev->pm.dpm.uvd_ps) {
+			return adev->pm.dpm.uvd_ps;
+		} else {
+			dpm_state = POWER_STATE_TYPE_PERFORMANCE;
+			goto restart_search;
+		}
+	case POWER_STATE_TYPE_INTERNAL_THERMAL:
+		dpm_state = POWER_STATE_TYPE_INTERNAL_ACPI;
+		goto restart_search;
+	case POWER_STATE_TYPE_INTERNAL_ACPI:
+		dpm_state = POWER_STATE_TYPE_BATTERY;
+		goto restart_search;
+	case POWER_STATE_TYPE_BATTERY:
+	case POWER_STATE_TYPE_BALANCED:
+	case POWER_STATE_TYPE_INTERNAL_3DPERF:
+		dpm_state = POWER_STATE_TYPE_PERFORMANCE;
+		goto restart_search;
+	default:
+		break;
+	}
+
+	return NULL;
+}
+
+static void amdgpu_dpm_change_power_state_locked(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_ps *ps;
+	enum amdgpu_pm_state_type dpm_state;
+	int ret;
+
+	/* if dpm init failed */
+	if (!adev->pm.dpm_enabled)
+		return;
+
+	if (adev->pm.dpm.user_state != adev->pm.dpm.state) {
+		/* add other state override checks here */
+		if ((!adev->pm.dpm.thermal_active) &&
+		    (!adev->pm.dpm.uvd_active))
+			adev->pm.dpm.state = adev->pm.dpm.user_state;
+	}
+	dpm_state = adev->pm.dpm.state;
+
+	ps = amdgpu_dpm_pick_power_state(adev, dpm_state);
+	if (ps)
+		adev->pm.dpm.requested_ps = ps;
+	else
+		return;
+
+	/* no need to reprogram if nothing changed unless we are on BTC+ */
+	if (adev->pm.dpm.current_ps == adev->pm.dpm.requested_ps) {
+		/* vce just modifies an existing state so force a change */
+		if (ps->vce_active != adev->pm.dpm.vce_active)
+			goto force;
+		if (adev->flags & AMD_IS_APU) {
+			/* for APUs if the num crtcs changed but state is the same,
+			 * all we need to do is update the display configuration.
+			 */
+			if (adev->pm.dpm.new_active_crtcs != adev->pm.dpm.current_active_crtcs) {
+				/* update display watermarks based on new power state */
+				amdgpu_display_bandwidth_update(adev);
+				/* update displays */
+				amdgpu_dpm_display_configuration_changed(adev);
+				adev->pm.dpm.current_active_crtcs = adev->pm.dpm.new_active_crtcs;
+				adev->pm.dpm.current_active_crtc_count = adev->pm.dpm.new_active_crtc_count;
+			}
+			return;
+		} else {
+			/* for BTC+ if the num crtcs hasn't changed and state is the same,
+			 * nothing to do, if the num crtcs is > 1 and state is the same,
+			 * update display configuration.
+			 */
+			if (adev->pm.dpm.new_active_crtcs ==
+			    adev->pm.dpm.current_active_crtcs) {
+				return;
+			} else if ((adev->pm.dpm.current_active_crtc_count > 1) &&
+				   (adev->pm.dpm.new_active_crtc_count > 1)) {
+				/* update display watermarks based on new power state */
+				amdgpu_display_bandwidth_update(adev);
+				/* update displays */
+				amdgpu_dpm_display_configuration_changed(adev);
+				adev->pm.dpm.current_active_crtcs = adev->pm.dpm.new_active_crtcs;
+				adev->pm.dpm.current_active_crtc_count = adev->pm.dpm.new_active_crtc_count;
+				return;
+			}
+		}
+	}
+
+force:
+	if (amdgpu_dpm == 1) {
+		printk("switching from power state:\n");
+		amdgpu_dpm_print_power_state(adev, adev->pm.dpm.current_ps);
+		printk("switching to power state:\n");
+		amdgpu_dpm_print_power_state(adev, adev->pm.dpm.requested_ps);
+	}
+
+	mutex_lock(&adev->ring_lock);
+
+	/* update whether vce is active */
+	ps->vce_active = adev->pm.dpm.vce_active;
+
+	ret = amdgpu_dpm_pre_set_power_state(adev);
+	if (ret)
+		goto done;
+
+	/* update display watermarks based on new power state */
+	amdgpu_display_bandwidth_update(adev);
+
+	/* wait for the rings to drain */
+	for (i = 0; i < AMDGPU_MAX_RINGS; i++) {
+		struct amdgpu_ring *ring = adev->rings[i];
+		if (ring && ring->ready)
+			amdgpu_fence_wait_empty(ring);
+	}
+
+	/* program the new power state */
+	amdgpu_dpm_set_power_state(adev);
+
+	/* update current power state */
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps;
+
+	amdgpu_dpm_post_set_power_state(adev);
+
+	/* update displays */
+	amdgpu_dpm_display_configuration_changed(adev);
+
+	adev->pm.dpm.current_active_crtcs = adev->pm.dpm.new_active_crtcs;
+	adev->pm.dpm.current_active_crtc_count = adev->pm.dpm.new_active_crtc_count;
+
+	if (adev->pm.funcs->force_performance_level) {
+		if (adev->pm.dpm.thermal_active) {
+			enum amdgpu_dpm_forced_level level = adev->pm.dpm.forced_level;
+			/* force low perf level for thermal */
+			amdgpu_dpm_force_performance_level(adev, AMDGPU_DPM_FORCED_LEVEL_LOW);
+			/* save the user's level */
+			adev->pm.dpm.forced_level = level;
+		} else {
+			/* otherwise, user selected level */
+			amdgpu_dpm_force_performance_level(adev, adev->pm.dpm.forced_level);
+		}
+	}
+
+done:
+	mutex_unlock(&adev->ring_lock);
+}
+
+void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable)
+{
+	if (adev->pm.funcs->powergate_uvd) {
+		mutex_lock(&adev->pm.mutex);
+		/* enable/disable UVD */
+		amdgpu_dpm_powergate_uvd(adev, !enable);
+		mutex_unlock(&adev->pm.mutex);
+	} else {
+		if (enable) {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.uvd_active = true;
+			adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD;
+			mutex_unlock(&adev->pm.mutex);
+		} else {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.uvd_active = false;
+			mutex_unlock(&adev->pm.mutex);
+		}
+
+		amdgpu_pm_compute_clocks(adev);
+	}
+}
+
+void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable)
+{
+	if (adev->pm.funcs->powergate_vce) {
+		mutex_lock(&adev->pm.mutex);
+		/* enable/disable VCE */
+		amdgpu_dpm_powergate_vce(adev, !enable);
+
+		mutex_unlock(&adev->pm.mutex);
+	} else {
+		if (enable) {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.vce_active = true;
+			/* XXX select vce level based on ring/task */
+			adev->pm.dpm.vce_level = AMDGPU_VCE_LEVEL_AC_ALL;
+			mutex_unlock(&adev->pm.mutex);
+		} else {
+			mutex_lock(&adev->pm.mutex);
+			adev->pm.dpm.vce_active = false;
+			mutex_unlock(&adev->pm.mutex);
+		}
+
+		amdgpu_pm_compute_clocks(adev);
+	}
+}
+
+void amdgpu_pm_print_power_states(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		printk("== power state %d ==\n", i);
+		amdgpu_dpm_print_power_state(adev, &adev->pm.dpm.ps[i]);
+	}
+}
+
+int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
+{
+	int ret;
+
+	if (adev->pm.sysfs_initialized)
+		return 0;
+
+	if (adev->pm.funcs->get_temperature == NULL)
+		return 0;
+	adev->pm.int_hwmon_dev = hwmon_device_register_with_groups(adev->dev,
+								   DRIVER_NAME, adev,
+								   hwmon_groups);
+	if (IS_ERR(adev->pm.int_hwmon_dev)) {
+		ret = PTR_ERR(adev->pm.int_hwmon_dev);
+		dev_err(adev->dev,
+			"Unable to register hwmon device: %d\n", ret);
+		return ret;
+	}
+
+	ret = device_create_file(adev->dev, &dev_attr_power_dpm_state);
+	if (ret) {
+		DRM_ERROR("failed to create device file for dpm state\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
+	if (ret) {
+		DRM_ERROR("failed to create device file for dpm state\n");
+		return ret;
+	}
+	ret = amdgpu_debugfs_pm_init(adev);
+	if (ret) {
+		DRM_ERROR("Failed to register debugfs file for dpm!\n");
+		return ret;
+	}
+
+	adev->pm.sysfs_initialized = true;
+
+	return 0;
+}
+
+void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev)
+{
+	if (adev->pm.int_hwmon_dev)
+		hwmon_device_unregister(adev->pm.int_hwmon_dev);
+	device_remove_file(adev->dev, &dev_attr_power_dpm_state);
+	device_remove_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
+}
+
+void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)
+{
+	struct drm_device *ddev = adev->ddev;
+	struct drm_crtc *crtc;
+	struct amdgpu_crtc *amdgpu_crtc;
+
+	if (!adev->pm.dpm_enabled)
+		return;
+
+	mutex_lock(&adev->pm.mutex);
+
+	/* update active crtc counts */
+	adev->pm.dpm.new_active_crtcs = 0;
+	adev->pm.dpm.new_active_crtc_count = 0;
+	if (adev->mode_info.num_crtc && adev->mode_info.mode_config_initialized) {
+		list_for_each_entry(crtc,
+				    &ddev->mode_config.crtc_list, head) {
+			amdgpu_crtc = to_amdgpu_crtc(crtc);
+			if (crtc->enabled) {
+				adev->pm.dpm.new_active_crtcs |= (1 << amdgpu_crtc->crtc_id);
+				adev->pm.dpm.new_active_crtc_count++;
+			}
+		}
+	}
+
+	/* update battery/ac status */
+	if (power_supply_is_system_supplied() > 0)
+		adev->pm.dpm.ac_power = true;
+	else
+		adev->pm.dpm.ac_power = false;
+
+	amdgpu_dpm_change_power_state_locked(adev);
+
+	mutex_unlock(&adev->pm.mutex);
+
+}
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_pm_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		if (adev->pm.funcs->debugfs_print_current_performance_level)
+			amdgpu_dpm_debugfs_print_current_performance_level(adev, m);
+		else
+			seq_printf(m, "Debugfs support not implemented for this asic\n");
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+	return 0;
+}
+
+static struct drm_info_list amdgpu_pm_info_list[] = {
+	{"amdgpu_pm_info", amdgpu_debugfs_pm_info, 0, NULL},
+};
+#endif
+
+static int amdgpu_debugfs_pm_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_pm_info_list, ARRAY_SIZE(amdgpu_pm_info_list));
+#else
+	return 0;
+#endif
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
new file mode 100644
index 000000000..5fd7734f1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_PM_H__
+#define __AMDGPU_PM_H__
+
+int amdgpu_pm_sysfs_init(struct amdgpu_device *adev);
+void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev);
+void amdgpu_pm_print_power_states(struct amdgpu_device *adev);
+void amdgpu_pm_compute_clocks(struct amdgpu_device *adev);
+void amdgpu_dpm_thermal_work_handler(struct work_struct *work);
+void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable);
+void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
new file mode 100644
index 000000000..59f735a93
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
@@ -0,0 +1,128 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * based on nouveau_prime.c
+ *
+ * Authors: Alex Deucher
+ */
+#include <drm/drmP.h>
+
+#include "amdgpu.h"
+#include <drm/amdgpu_drm.h>
+#include <linux/dma-buf.h>
+
+struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int npages = bo->tbo.num_pages;
+
+	return drm_prime_pages_to_sg(bo->tbo.ttm->pages, npages);
+}
+
+void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int ret;
+
+	ret = ttm_bo_kmap(&bo->tbo, 0, bo->tbo.num_pages,
+			  &bo->dma_buf_vmap);
+	if (ret)
+		return ERR_PTR(ret);
+
+	return bo->dma_buf_vmap.virtual;
+}
+
+void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	ttm_bo_kunmap(&bo->dma_buf_vmap);
+}
+
+struct drm_gem_object *amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
+							struct dma_buf_attachment *attach,
+							struct sg_table *sg)
+{
+	struct reservation_object *resv = attach->dmabuf->resv;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_bo *bo;
+	int ret;
+
+	ww_mutex_lock(&resv->lock, NULL);
+	ret = amdgpu_bo_create(adev, attach->dmabuf->size, PAGE_SIZE, false,
+			       AMDGPU_GEM_DOMAIN_GTT, 0, sg, resv, &bo);
+	ww_mutex_unlock(&resv->lock);
+	if (ret)
+		return ERR_PTR(ret);
+
+	mutex_lock(&adev->gem.mutex);
+	list_add_tail(&bo->list, &adev->gem.objects);
+	mutex_unlock(&adev->gem.mutex);
+
+	return &bo->gem_base;
+}
+
+int amdgpu_gem_prime_pin(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int ret = 0;
+
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret != 0))
+		return ret;
+
+	/* pin buffer into GTT */
+	ret = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT, NULL);
+	amdgpu_bo_unreserve(bo);
+	return ret;
+}
+
+void amdgpu_gem_prime_unpin(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+	int ret = 0;
+
+	ret = amdgpu_bo_reserve(bo, false);
+	if (unlikely(ret != 0))
+		return;
+
+	amdgpu_bo_unpin(bo);
+	amdgpu_bo_unreserve(bo);
+}
+
+struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
+
+	return bo->tbo.resv;
+}
+
+struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev,
+					struct drm_gem_object *gobj,
+					int flags)
+{
+	struct amdgpu_bo *bo = gem_to_amdgpu_bo(gobj);
+
+	if (amdgpu_ttm_tt_has_userptr(bo->tbo.ttm))
+		return ERR_PTR(-EPERM);
+
+	return drm_gem_prime_export(dev, gobj, flags);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
new file mode 100644
index 000000000..78e9b0f14
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -0,0 +1,562 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ *          Christian König
+ */
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+
+/*
+ * Rings
+ * Most engines on the GPU are fed via ring buffers.  Ring
+ * buffers are areas of GPU accessible memory that the host
+ * writes commands into and the GPU reads commands out of.
+ * There is a rptr (read pointer) that determines where the
+ * GPU is currently reading, and a wptr (write pointer)
+ * which determines where the host has written.  When the
+ * pointers are equal, the ring is idle.  When the host
+ * writes commands to the ring buffer, it increments the
+ * wptr.  The GPU then starts fetching commands and executes
+ * them until the pointers are equal again.
+ */
+static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring);
+
+/**
+ * amdgpu_ring_free_size - update the free size
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Update the free dw slots in the ring buffer (all asics).
+ */
+void amdgpu_ring_free_size(struct amdgpu_ring *ring)
+{
+	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+
+	/* This works because ring_size is a power of 2 */
+	ring->ring_free_dw = rptr + (ring->ring_size / 4);
+	ring->ring_free_dw -= ring->wptr;
+	ring->ring_free_dw &= ring->ptr_mask;
+	if (!ring->ring_free_dw) {
+		/* this is an empty ring */
+		ring->ring_free_dw = ring->ring_size / 4;
+	}
+}
+
+/**
+ * amdgpu_ring_alloc - allocate space on the ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ * @ndw: number of dwords to allocate in the ring buffer
+ *
+ * Allocate @ndw dwords in the ring buffer (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
+{
+	int r;
+
+	/* make sure we aren't trying to allocate more space than there is on the ring */
+	if (ndw > (ring->ring_size / 4))
+		return -ENOMEM;
+	/* Align requested size with padding so unlock_commit can
+	 * pad safely */
+	amdgpu_ring_free_size(ring);
+	ndw = (ndw + ring->align_mask) & ~ring->align_mask;
+	while (ndw > (ring->ring_free_dw - 1)) {
+		amdgpu_ring_free_size(ring);
+		if (ndw < ring->ring_free_dw) {
+			break;
+		}
+		r = amdgpu_fence_wait_next(ring);
+		if (r)
+			return r;
+	}
+	ring->count_dw = ndw;
+	ring->wptr_old = ring->wptr;
+	return 0;
+}
+
+/**
+ * amdgpu_ring_lock - lock the ring and allocate space on it
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ * @ndw: number of dwords to allocate in the ring buffer
+ *
+ * Lock the ring and allocate @ndw dwords in the ring buffer
+ * (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_lock(struct amdgpu_ring *ring, unsigned ndw)
+{
+	int r;
+
+	mutex_lock(ring->ring_lock);
+	r = amdgpu_ring_alloc(ring, ndw);
+	if (r) {
+		mutex_unlock(ring->ring_lock);
+		return r;
+	}
+	return 0;
+}
+
+/** amdgpu_ring_insert_nop - insert NOP packets
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @count: the number of NOP packets to insert
+ *
+ * This is the generic insert_nop function for rings except SDMA
+ */
+void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	int i;
+
+	for (i = 0; i < count; i++)
+		amdgpu_ring_write(ring, ring->nop);
+}
+
+/**
+ * amdgpu_ring_commit - tell the GPU to execute the new
+ * commands on the ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Update the wptr (write pointer) to tell the GPU to
+ * execute new commands on the ring buffer (all asics).
+ */
+void amdgpu_ring_commit(struct amdgpu_ring *ring)
+{
+	uint32_t count;
+
+	/* We pad to match fetch size */
+	count = ring->align_mask + 1 - (ring->wptr & ring->align_mask);
+	count %= ring->align_mask + 1;
+	ring->funcs->insert_nop(ring, count);
+
+	mb();
+	amdgpu_ring_set_wptr(ring);
+}
+
+/**
+ * amdgpu_ring_unlock_commit - tell the GPU to execute the new
+ * commands on the ring buffer and unlock it
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Call amdgpu_ring_commit() then unlock the ring (all asics).
+ */
+void amdgpu_ring_unlock_commit(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_commit(ring);
+	mutex_unlock(ring->ring_lock);
+}
+
+/**
+ * amdgpu_ring_undo - reset the wptr
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Reset the driver's copy of the wptr (all asics).
+ */
+void amdgpu_ring_undo(struct amdgpu_ring *ring)
+{
+	ring->wptr = ring->wptr_old;
+}
+
+/**
+ * amdgpu_ring_unlock_undo - reset the wptr and unlock the ring
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Call amdgpu_ring_undo() then unlock the ring (all asics).
+ */
+void amdgpu_ring_unlock_undo(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_undo(ring);
+	mutex_unlock(ring->ring_lock);
+}
+
+/**
+ * amdgpu_ring_backup - Back up the content of a ring
+ *
+ * @ring: the ring we want to back up
+ *
+ * Saves all unprocessed commits from a ring, returns the number of dwords saved.
+ */
+unsigned amdgpu_ring_backup(struct amdgpu_ring *ring,
+			    uint32_t **data)
+{
+	unsigned size, ptr, i;
+
+	/* just in case lock the ring */
+	mutex_lock(ring->ring_lock);
+	*data = NULL;
+
+	if (ring->ring_obj == NULL) {
+		mutex_unlock(ring->ring_lock);
+		return 0;
+	}
+
+	/* it doesn't make sense to save anything if all fences are signaled */
+	if (!amdgpu_fence_count_emitted(ring)) {
+		mutex_unlock(ring->ring_lock);
+		return 0;
+	}
+
+	ptr = le32_to_cpu(*ring->next_rptr_cpu_addr);
+
+	size = ring->wptr + (ring->ring_size / 4);
+	size -= ptr;
+	size &= ring->ptr_mask;
+	if (size == 0) {
+		mutex_unlock(ring->ring_lock);
+		return 0;
+	}
+
+	/* and then save the content of the ring */
+	*data = kmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
+	if (!*data) {
+		mutex_unlock(ring->ring_lock);
+		return 0;
+	}
+	for (i = 0; i < size; ++i) {
+		(*data)[i] = ring->ring[ptr++];
+		ptr &= ring->ptr_mask;
+	}
+
+	mutex_unlock(ring->ring_lock);
+	return size;
+}
+
+/**
+ * amdgpu_ring_restore - append saved commands to the ring again
+ *
+ * @ring: ring to append commands to
+ * @size: number of dwords we want to write
+ * @data: saved commands
+ *
+ * Allocates space on the ring and restore the previously saved commands.
+ */
+int amdgpu_ring_restore(struct amdgpu_ring *ring,
+			unsigned size, uint32_t *data)
+{
+	int i, r;
+
+	if (!size || !data)
+		return 0;
+
+	/* restore the saved ring content */
+	r = amdgpu_ring_lock(ring, size);
+	if (r)
+		return r;
+
+	for (i = 0; i < size; ++i) {
+		amdgpu_ring_write(ring, data[i]);
+	}
+
+	amdgpu_ring_unlock_commit(ring);
+	kfree(data);
+	return 0;
+}
+
+/**
+ * amdgpu_ring_init - init driver ring struct.
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ * @ring_size: size of the ring
+ * @nop: nop packet for this ring
+ *
+ * Initialize the driver information for the selected ring (all asics).
+ * Returns 0 on success, error on failure.
+ */
+int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
+		     unsigned ring_size, u32 nop, u32 align_mask,
+		     struct amdgpu_irq_src *irq_src, unsigned irq_type,
+		     enum amdgpu_ring_type ring_type)
+{
+	u32 rb_bufsz;
+	int r;
+
+	if (ring->adev == NULL) {
+		if (adev->num_rings >= AMDGPU_MAX_RINGS)
+			return -EINVAL;
+
+		ring->adev = adev;
+		ring->idx = adev->num_rings++;
+		adev->rings[ring->idx] = ring;
+		r = amdgpu_fence_driver_init_ring(ring);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_wb_get(adev, &ring->rptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_wb_get(adev, &ring->wptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_wb_get(adev, &ring->fence_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_wb_get(adev, &ring->next_rptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring next_rptr wb alloc failed\n", r);
+		return r;
+	}
+	ring->next_rptr_gpu_addr = adev->wb.gpu_addr + (ring->next_rptr_offs * 4);
+	ring->next_rptr_cpu_addr = &adev->wb.wb[ring->next_rptr_offs];
+	spin_lock_init(&ring->fence_lock);
+	r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
+	if (r) {
+		dev_err(adev->dev, "failed initializing fences (%d).\n", r);
+		return r;
+	}
+
+	ring->ring_lock = &adev->ring_lock;
+	/* Align ring size */
+	rb_bufsz = order_base_2(ring_size / 8);
+	ring_size = (1 << (rb_bufsz + 1)) * 4;
+	ring->ring_size = ring_size;
+	ring->align_mask = align_mask;
+	ring->nop = nop;
+	ring->type = ring_type;
+
+	/* Allocate ring buffer */
+	if (ring->ring_obj == NULL) {
+		r = amdgpu_bo_create(adev, ring->ring_size, PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_GTT, 0,
+				     NULL, NULL, &ring->ring_obj);
+		if (r) {
+			dev_err(adev->dev, "(%d) ring create failed\n", r);
+			return r;
+		}
+		r = amdgpu_bo_reserve(ring->ring_obj, false);
+		if (unlikely(r != 0))
+			return r;
+		r = amdgpu_bo_pin(ring->ring_obj, AMDGPU_GEM_DOMAIN_GTT,
+					&ring->gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(ring->ring_obj);
+			dev_err(adev->dev, "(%d) ring pin failed\n", r);
+			return r;
+		}
+		r = amdgpu_bo_kmap(ring->ring_obj,
+				       (void **)&ring->ring);
+		amdgpu_bo_unreserve(ring->ring_obj);
+		if (r) {
+			dev_err(adev->dev, "(%d) ring map failed\n", r);
+			return r;
+		}
+	}
+	ring->ptr_mask = (ring->ring_size / 4) - 1;
+	ring->ring_free_dw = ring->ring_size / 4;
+
+	if (amdgpu_debugfs_ring_init(adev, ring)) {
+		DRM_ERROR("Failed to register debugfs file for rings !\n");
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_ring_fini - tear down the driver ring struct.
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Tear down the driver information for the selected ring (all asics).
+ */
+void amdgpu_ring_fini(struct amdgpu_ring *ring)
+{
+	int r;
+	struct amdgpu_bo *ring_obj;
+
+	if (ring->ring_lock == NULL)
+		return;
+
+	mutex_lock(ring->ring_lock);
+	ring_obj = ring->ring_obj;
+	ring->ready = false;
+	ring->ring = NULL;
+	ring->ring_obj = NULL;
+	mutex_unlock(ring->ring_lock);
+
+	amdgpu_wb_free(ring->adev, ring->fence_offs);
+	amdgpu_wb_free(ring->adev, ring->rptr_offs);
+	amdgpu_wb_free(ring->adev, ring->wptr_offs);
+	amdgpu_wb_free(ring->adev, ring->next_rptr_offs);
+
+	if (ring_obj) {
+		r = amdgpu_bo_reserve(ring_obj, false);
+		if (likely(r == 0)) {
+			amdgpu_bo_kunmap(ring_obj);
+			amdgpu_bo_unpin(ring_obj);
+			amdgpu_bo_unreserve(ring_obj);
+		}
+		amdgpu_bo_unref(&ring_obj);
+	}
+}
+
+/**
+ * amdgpu_ring_from_fence - get ring from fence
+ *
+ * @f: fence structure
+ *
+ * Extract the ring a fence belongs to. Handles both scheduler as
+ * well as hardware fences.
+ */
+struct amdgpu_ring *amdgpu_ring_from_fence(struct fence *f)
+{
+	struct amdgpu_fence *a_fence;
+	struct amd_sched_fence *s_fence;
+
+	s_fence = to_amd_sched_fence(f);
+	if (s_fence)
+		return container_of(s_fence->sched, struct amdgpu_ring, sched);
+
+	a_fence = to_amdgpu_fence(f);
+	if (a_fence)
+		return a_fence->ring;
+
+	return NULL;
+}
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_ring_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int roffset = *(int*)node->info_ent->data;
+	struct amdgpu_ring *ring = (void *)(((uint8_t*)adev) + roffset);
+
+	uint32_t rptr, wptr, rptr_next;
+	unsigned count, i, j;
+
+	amdgpu_ring_free_size(ring);
+	count = (ring->ring_size / 4) - ring->ring_free_dw;
+
+	wptr = amdgpu_ring_get_wptr(ring);
+	seq_printf(m, "wptr: 0x%08x [%5d]\n",
+		   wptr, wptr);
+
+	rptr = amdgpu_ring_get_rptr(ring);
+	seq_printf(m, "rptr: 0x%08x [%5d]\n",
+		   rptr, rptr);
+
+	rptr_next = ~0;
+
+	seq_printf(m, "driver's copy of the wptr: 0x%08x [%5d]\n",
+		   ring->wptr, ring->wptr);
+	seq_printf(m, "last semaphore signal addr : 0x%016llx\n",
+		   ring->last_semaphore_signal_addr);
+	seq_printf(m, "last semaphore wait addr   : 0x%016llx\n",
+		   ring->last_semaphore_wait_addr);
+	seq_printf(m, "%u free dwords in ring\n", ring->ring_free_dw);
+	seq_printf(m, "%u dwords in ring\n", count);
+
+	if (!ring->ready)
+		return 0;
+
+	/* print 8 dw before current rptr as often it's the last executed
+	 * packet that is the root issue
+	 */
+	i = (rptr + ring->ptr_mask + 1 - 32) & ring->ptr_mask;
+	for (j = 0; j <= (count + 32); j++) {
+		seq_printf(m, "r[%5d]=0x%08x", i, ring->ring[i]);
+		if (rptr == i)
+			seq_puts(m, " *");
+		if (rptr_next == i)
+			seq_puts(m, " #");
+		seq_puts(m, "\n");
+		i = (i + 1) & ring->ptr_mask;
+	}
+	return 0;
+}
+
+/* TODO: clean this up !*/
+static int amdgpu_gfx_index = offsetof(struct amdgpu_device, gfx.gfx_ring[0]);
+static int cayman_cp1_index = offsetof(struct amdgpu_device, gfx.compute_ring[0]);
+static int cayman_cp2_index = offsetof(struct amdgpu_device, gfx.compute_ring[1]);
+static int amdgpu_dma1_index = offsetof(struct amdgpu_device, sdma.instance[0].ring);
+static int amdgpu_dma2_index = offsetof(struct amdgpu_device, sdma.instance[1].ring);
+static int r600_uvd_index = offsetof(struct amdgpu_device, uvd.ring);
+static int si_vce1_index = offsetof(struct amdgpu_device, vce.ring[0]);
+static int si_vce2_index = offsetof(struct amdgpu_device, vce.ring[1]);
+
+static struct drm_info_list amdgpu_debugfs_ring_info_list[] = {
+	{"amdgpu_ring_gfx", amdgpu_debugfs_ring_info, 0, &amdgpu_gfx_index},
+	{"amdgpu_ring_cp1", amdgpu_debugfs_ring_info, 0, &cayman_cp1_index},
+	{"amdgpu_ring_cp2", amdgpu_debugfs_ring_info, 0, &cayman_cp2_index},
+	{"amdgpu_ring_dma1", amdgpu_debugfs_ring_info, 0, &amdgpu_dma1_index},
+	{"amdgpu_ring_dma2", amdgpu_debugfs_ring_info, 0, &amdgpu_dma2_index},
+	{"amdgpu_ring_uvd", amdgpu_debugfs_ring_info, 0, &r600_uvd_index},
+	{"amdgpu_ring_vce1", amdgpu_debugfs_ring_info, 0, &si_vce1_index},
+	{"amdgpu_ring_vce2", amdgpu_debugfs_ring_info, 0, &si_vce2_index},
+};
+
+#endif
+
+static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring)
+{
+#if defined(CONFIG_DEBUG_FS)
+	unsigned i;
+	for (i = 0; i < ARRAY_SIZE(amdgpu_debugfs_ring_info_list); ++i) {
+		struct drm_info_list *info = &amdgpu_debugfs_ring_info_list[i];
+		int roffset = *(int*)amdgpu_debugfs_ring_info_list[i].data;
+		struct amdgpu_ring *other = (void *)(((uint8_t*)adev) + roffset);
+		unsigned r;
+
+		if (other != ring)
+			continue;
+
+		r = amdgpu_debugfs_add_files(adev, info, 1);
+		if (r)
+			return r;
+	}
+#endif
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
new file mode 100644
index 000000000..ca72a2e48
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -0,0 +1,451 @@
+/*
+ * Copyright 2011 Red Hat Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ */
+/* Algorithm:
+ *
+ * We store the last allocated bo in "hole", we always try to allocate
+ * after the last allocated bo. Principle is that in a linear GPU ring
+ * progression was is after last is the oldest bo we allocated and thus
+ * the first one that should no longer be in use by the GPU.
+ *
+ * If it's not the case we skip over the bo after last to the closest
+ * done bo if such one exist. If none exist and we are not asked to
+ * block we report failure to allocate.
+ *
+ * If we are asked to block we wait on all the oldest fence of all
+ * rings. We just wait for any of those fence to complete.
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+
+static void amdgpu_sa_bo_remove_locked(struct amdgpu_sa_bo *sa_bo);
+static void amdgpu_sa_bo_try_free(struct amdgpu_sa_manager *sa_manager);
+
+int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
+			      struct amdgpu_sa_manager *sa_manager,
+			      unsigned size, u32 align, u32 domain)
+{
+	int i, r;
+
+	init_waitqueue_head(&sa_manager->wq);
+	sa_manager->bo = NULL;
+	sa_manager->size = size;
+	sa_manager->domain = domain;
+	sa_manager->align = align;
+	sa_manager->hole = &sa_manager->olist;
+	INIT_LIST_HEAD(&sa_manager->olist);
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		INIT_LIST_HEAD(&sa_manager->flist[i]);
+	}
+
+	r = amdgpu_bo_create(adev, size, align, true, domain,
+			     0, NULL, NULL, &sa_manager->bo);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate bo for manager\n", r);
+		return r;
+	}
+
+	return r;
+}
+
+void amdgpu_sa_bo_manager_fini(struct amdgpu_device *adev,
+			       struct amdgpu_sa_manager *sa_manager)
+{
+	struct amdgpu_sa_bo *sa_bo, *tmp;
+
+	if (!list_empty(&sa_manager->olist)) {
+		sa_manager->hole = &sa_manager->olist,
+		amdgpu_sa_bo_try_free(sa_manager);
+		if (!list_empty(&sa_manager->olist)) {
+			dev_err(adev->dev, "sa_manager is not empty, clearing anyway\n");
+		}
+	}
+	list_for_each_entry_safe(sa_bo, tmp, &sa_manager->olist, olist) {
+		amdgpu_sa_bo_remove_locked(sa_bo);
+	}
+	amdgpu_bo_unref(&sa_manager->bo);
+	sa_manager->size = 0;
+}
+
+int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
+			       struct amdgpu_sa_manager *sa_manager)
+{
+	int r;
+
+	if (sa_manager->bo == NULL) {
+		dev_err(adev->dev, "no bo for sa manager\n");
+		return -EINVAL;
+	}
+
+	/* map the buffer */
+	r = amdgpu_bo_reserve(sa_manager->bo, false);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to reserve manager bo\n", r);
+		return r;
+	}
+	r = amdgpu_bo_pin(sa_manager->bo, sa_manager->domain, &sa_manager->gpu_addr);
+	if (r) {
+		amdgpu_bo_unreserve(sa_manager->bo);
+		dev_err(adev->dev, "(%d) failed to pin manager bo\n", r);
+		return r;
+	}
+	r = amdgpu_bo_kmap(sa_manager->bo, &sa_manager->cpu_ptr);
+	amdgpu_bo_unreserve(sa_manager->bo);
+	return r;
+}
+
+int amdgpu_sa_bo_manager_suspend(struct amdgpu_device *adev,
+				 struct amdgpu_sa_manager *sa_manager)
+{
+	int r;
+
+	if (sa_manager->bo == NULL) {
+		dev_err(adev->dev, "no bo for sa manager\n");
+		return -EINVAL;
+	}
+
+	r = amdgpu_bo_reserve(sa_manager->bo, false);
+	if (!r) {
+		amdgpu_bo_kunmap(sa_manager->bo);
+		amdgpu_bo_unpin(sa_manager->bo);
+		amdgpu_bo_unreserve(sa_manager->bo);
+	}
+	return r;
+}
+
+static void amdgpu_sa_bo_remove_locked(struct amdgpu_sa_bo *sa_bo)
+{
+	struct amdgpu_sa_manager *sa_manager = sa_bo->manager;
+	if (sa_manager->hole == &sa_bo->olist) {
+		sa_manager->hole = sa_bo->olist.prev;
+	}
+	list_del_init(&sa_bo->olist);
+	list_del_init(&sa_bo->flist);
+	fence_put(sa_bo->fence);
+	kfree(sa_bo);
+}
+
+static void amdgpu_sa_bo_try_free(struct amdgpu_sa_manager *sa_manager)
+{
+	struct amdgpu_sa_bo *sa_bo, *tmp;
+
+	if (sa_manager->hole->next == &sa_manager->olist)
+		return;
+
+	sa_bo = list_entry(sa_manager->hole->next, struct amdgpu_sa_bo, olist);
+	list_for_each_entry_safe_from(sa_bo, tmp, &sa_manager->olist, olist) {
+		if (sa_bo->fence == NULL ||
+		    !fence_is_signaled(sa_bo->fence)) {
+			return;
+		}
+		amdgpu_sa_bo_remove_locked(sa_bo);
+	}
+}
+
+static inline unsigned amdgpu_sa_bo_hole_soffset(struct amdgpu_sa_manager *sa_manager)
+{
+	struct list_head *hole = sa_manager->hole;
+
+	if (hole != &sa_manager->olist) {
+		return list_entry(hole, struct amdgpu_sa_bo, olist)->eoffset;
+	}
+	return 0;
+}
+
+static inline unsigned amdgpu_sa_bo_hole_eoffset(struct amdgpu_sa_manager *sa_manager)
+{
+	struct list_head *hole = sa_manager->hole;
+
+	if (hole->next != &sa_manager->olist) {
+		return list_entry(hole->next, struct amdgpu_sa_bo, olist)->soffset;
+	}
+	return sa_manager->size;
+}
+
+static bool amdgpu_sa_bo_try_alloc(struct amdgpu_sa_manager *sa_manager,
+				   struct amdgpu_sa_bo *sa_bo,
+				   unsigned size, unsigned align)
+{
+	unsigned soffset, eoffset, wasted;
+
+	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
+	eoffset = amdgpu_sa_bo_hole_eoffset(sa_manager);
+	wasted = (align - (soffset % align)) % align;
+
+	if ((eoffset - soffset) >= (size + wasted)) {
+		soffset += wasted;
+
+		sa_bo->manager = sa_manager;
+		sa_bo->soffset = soffset;
+		sa_bo->eoffset = soffset + size;
+		list_add(&sa_bo->olist, sa_manager->hole);
+		INIT_LIST_HEAD(&sa_bo->flist);
+		sa_manager->hole = &sa_bo->olist;
+		return true;
+	}
+	return false;
+}
+
+/**
+ * amdgpu_sa_event - Check if we can stop waiting
+ *
+ * @sa_manager: pointer to the sa_manager
+ * @size: number of bytes we want to allocate
+ * @align: alignment we need to match
+ *
+ * Check if either there is a fence we can wait for or
+ * enough free memory to satisfy the allocation directly
+ */
+static bool amdgpu_sa_event(struct amdgpu_sa_manager *sa_manager,
+			    unsigned size, unsigned align)
+{
+	unsigned soffset, eoffset, wasted;
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		if (!list_empty(&sa_manager->flist[i])) {
+			return true;
+		}
+	}
+
+	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
+	eoffset = amdgpu_sa_bo_hole_eoffset(sa_manager);
+	wasted = (align - (soffset % align)) % align;
+
+	if ((eoffset - soffset) >= (size + wasted)) {
+		return true;
+	}
+
+	return false;
+}
+
+static bool amdgpu_sa_bo_next_hole(struct amdgpu_sa_manager *sa_manager,
+				   struct fence **fences,
+				   unsigned *tries)
+{
+	struct amdgpu_sa_bo *best_bo = NULL;
+	unsigned i, soffset, best, tmp;
+
+	/* if hole points to the end of the buffer */
+	if (sa_manager->hole->next == &sa_manager->olist) {
+		/* try again with its beginning */
+		sa_manager->hole = &sa_manager->olist;
+		return true;
+	}
+
+	soffset = amdgpu_sa_bo_hole_soffset(sa_manager);
+	/* to handle wrap around we add sa_manager->size */
+	best = sa_manager->size * 2;
+	/* go over all fence list and try to find the closest sa_bo
+	 * of the current last
+	 */
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_sa_bo *sa_bo;
+
+		if (list_empty(&sa_manager->flist[i])) {
+			continue;
+		}
+
+		sa_bo = list_first_entry(&sa_manager->flist[i],
+					 struct amdgpu_sa_bo, flist);
+
+		if (!fence_is_signaled(sa_bo->fence)) {
+			fences[i] = sa_bo->fence;
+			continue;
+		}
+
+		/* limit the number of tries each ring gets */
+		if (tries[i] > 2) {
+			continue;
+		}
+
+		tmp = sa_bo->soffset;
+		if (tmp < soffset) {
+			/* wrap around, pretend it's after */
+			tmp += sa_manager->size;
+		}
+		tmp -= soffset;
+		if (tmp < best) {
+			/* this sa bo is the closest one */
+			best = tmp;
+			best_bo = sa_bo;
+		}
+	}
+
+	if (best_bo) {
+		uint32_t idx = amdgpu_ring_from_fence(best_bo->fence)->idx;
+		++tries[idx];
+		sa_manager->hole = best_bo->olist.prev;
+
+		/* we knew that this one is signaled,
+		   so it's save to remote it */
+		amdgpu_sa_bo_remove_locked(best_bo);
+		return true;
+	}
+	return false;
+}
+
+int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
+		     struct amdgpu_sa_bo **sa_bo,
+		     unsigned size, unsigned align)
+{
+	struct fence *fences[AMDGPU_MAX_RINGS];
+	unsigned tries[AMDGPU_MAX_RINGS];
+	unsigned count;
+	int i, r;
+	signed long t;
+
+	BUG_ON(align > sa_manager->align);
+	BUG_ON(size > sa_manager->size);
+
+	*sa_bo = kmalloc(sizeof(struct amdgpu_sa_bo), GFP_KERNEL);
+	if ((*sa_bo) == NULL) {
+		return -ENOMEM;
+	}
+	(*sa_bo)->manager = sa_manager;
+	(*sa_bo)->fence = NULL;
+	INIT_LIST_HEAD(&(*sa_bo)->olist);
+	INIT_LIST_HEAD(&(*sa_bo)->flist);
+
+	spin_lock(&sa_manager->wq.lock);
+	do {
+		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+			fences[i] = NULL;
+			tries[i] = 0;
+		}
+
+		do {
+			amdgpu_sa_bo_try_free(sa_manager);
+
+			if (amdgpu_sa_bo_try_alloc(sa_manager, *sa_bo,
+						   size, align)) {
+				spin_unlock(&sa_manager->wq.lock);
+				return 0;
+			}
+
+			/* see if we can skip over some allocations */
+		} while (amdgpu_sa_bo_next_hole(sa_manager, fences, tries));
+
+		for (i = 0, count = 0; i < AMDGPU_MAX_RINGS; ++i)
+			if (fences[i])
+				fences[count++] = fence_get(fences[i]);
+
+		if (count) {
+			spin_unlock(&sa_manager->wq.lock);
+			t = fence_wait_any_timeout(fences, count, false,
+						   MAX_SCHEDULE_TIMEOUT);
+			for (i = 0; i < count; ++i)
+				fence_put(fences[i]);
+
+			r = (t > 0) ? 0 : t;
+			spin_lock(&sa_manager->wq.lock);
+		} else {
+			/* if we have nothing to wait for block */
+			r = wait_event_interruptible_locked(
+				sa_manager->wq,
+				amdgpu_sa_event(sa_manager, size, align)
+			);
+		}
+
+	} while (!r);
+
+	spin_unlock(&sa_manager->wq.lock);
+	kfree(*sa_bo);
+	*sa_bo = NULL;
+	return r;
+}
+
+void amdgpu_sa_bo_free(struct amdgpu_device *adev, struct amdgpu_sa_bo **sa_bo,
+		       struct fence *fence)
+{
+	struct amdgpu_sa_manager *sa_manager;
+
+	if (sa_bo == NULL || *sa_bo == NULL) {
+		return;
+	}
+
+	sa_manager = (*sa_bo)->manager;
+	spin_lock(&sa_manager->wq.lock);
+	if (fence && !fence_is_signaled(fence)) {
+		uint32_t idx;
+		(*sa_bo)->fence = fence_get(fence);
+		idx = amdgpu_ring_from_fence(fence)->idx;
+		list_add_tail(&(*sa_bo)->flist, &sa_manager->flist[idx]);
+	} else {
+		amdgpu_sa_bo_remove_locked(*sa_bo);
+	}
+	wake_up_all_locked(&sa_manager->wq);
+	spin_unlock(&sa_manager->wq.lock);
+	*sa_bo = NULL;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static void amdgpu_sa_bo_dump_fence(struct fence *fence, struct seq_file *m)
+{
+	struct amdgpu_fence *a_fence = to_amdgpu_fence(fence);
+	struct amd_sched_fence *s_fence = to_amd_sched_fence(fence);
+
+	if (a_fence)
+		seq_printf(m, " protected by 0x%016llx on ring %d",
+			   a_fence->seq, a_fence->ring->idx);
+
+	if (s_fence) {
+		struct amdgpu_ring *ring;
+
+
+		ring = container_of(s_fence->sched, struct amdgpu_ring, sched);
+		seq_printf(m, " protected by 0x%016x on ring %d",
+			   s_fence->base.seqno, ring->idx);
+	}
+}
+
+void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
+				  struct seq_file *m)
+{
+	struct amdgpu_sa_bo *i;
+
+	spin_lock(&sa_manager->wq.lock);
+	list_for_each_entry(i, &sa_manager->olist, olist) {
+		uint64_t soffset = i->soffset + sa_manager->gpu_addr;
+		uint64_t eoffset = i->eoffset + sa_manager->gpu_addr;
+		if (&i->olist == sa_manager->hole) {
+			seq_printf(m, ">");
+		} else {
+			seq_printf(m, " ");
+		}
+		seq_printf(m, "[0x%010llx 0x%010llx] size %8lld",
+			   soffset, eoffset, eoffset - soffset);
+		if (i->fence)
+			amdgpu_sa_bo_dump_fence(i->fence, m);
+		seq_printf(m, "\n");
+	}
+	spin_unlock(&sa_manager->wq.lock);
+}
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c
new file mode 100644
index 000000000..438c05254
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sched.c
@@ -0,0 +1,108 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ *
+ */
+#include <linux/kthread.h>
+#include <linux/wait.h>
+#include <linux/sched.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+static struct fence *amdgpu_sched_dependency(struct amd_sched_job *sched_job)
+{
+	struct amdgpu_job *job = to_amdgpu_job(sched_job);
+	return amdgpu_sync_get_fence(&job->ibs->sync);
+}
+
+static struct fence *amdgpu_sched_run_job(struct amd_sched_job *sched_job)
+{
+	struct amdgpu_fence *fence = NULL;
+	struct amdgpu_job *job;
+	int r;
+
+	if (!sched_job) {
+		DRM_ERROR("job is null\n");
+		return NULL;
+	}
+	job = to_amdgpu_job(sched_job);
+	trace_amdgpu_sched_run_job(job);
+	r = amdgpu_ib_schedule(job->adev, job->num_ibs, job->ibs, job->owner);
+	if (r) {
+		DRM_ERROR("Error scheduling IBs (%d)\n", r);
+		goto err;
+	}
+
+	fence = job->ibs[job->num_ibs - 1].fence;
+	fence_get(&fence->base);
+
+err:
+	if (job->free_job)
+		job->free_job(job);
+
+	kfree(job);
+	return fence ? &fence->base : NULL;
+}
+
+struct amd_sched_backend_ops amdgpu_sched_ops = {
+	.dependency = amdgpu_sched_dependency,
+	.run_job = amdgpu_sched_run_job,
+};
+
+int amdgpu_sched_ib_submit_kernel_helper(struct amdgpu_device *adev,
+					 struct amdgpu_ring *ring,
+					 struct amdgpu_ib *ibs,
+					 unsigned num_ibs,
+					 int (*free_job)(struct amdgpu_job *),
+					 void *owner,
+					 struct fence **f)
+{
+	int r = 0;
+	if (amdgpu_enable_scheduler) {
+		struct amdgpu_job *job =
+			kzalloc(sizeof(struct amdgpu_job), GFP_KERNEL);
+		if (!job)
+			return -ENOMEM;
+		job->base.sched = &ring->sched;
+		job->base.s_entity = &adev->kernel_ctx.rings[ring->idx].entity;
+		job->base.s_fence = amd_sched_fence_create(job->base.s_entity, owner);
+		if (!job->base.s_fence) {
+			kfree(job);
+			return -ENOMEM;
+		}
+		*f = fence_get(&job->base.s_fence->base);
+
+		job->adev = adev;
+		job->ibs = ibs;
+		job->num_ibs = num_ibs;
+		job->owner = owner;
+		job->free_job = free_job;
+		amd_sched_entity_push_job(&job->base);
+	} else {
+		r = amdgpu_ib_schedule(adev, num_ibs, ibs, owner);
+		if (r)
+			return r;
+		*f = fence_get(&ibs[num_ibs - 1].fence->base);
+	}
+
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_semaphore.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_semaphore.c
new file mode 100644
index 000000000..1caaf201b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_semaphore.c
@@ -0,0 +1,102 @@
+/*
+ * Copyright 2011 Christian König.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+int amdgpu_semaphore_create(struct amdgpu_device *adev,
+			    struct amdgpu_semaphore **semaphore)
+{
+	int r;
+
+	*semaphore = kmalloc(sizeof(struct amdgpu_semaphore), GFP_KERNEL);
+	if (*semaphore == NULL) {
+		return -ENOMEM;
+	}
+	r = amdgpu_sa_bo_new(&adev->ring_tmp_bo,
+			     &(*semaphore)->sa_bo, 8, 8);
+	if (r) {
+		kfree(*semaphore);
+		*semaphore = NULL;
+		return r;
+	}
+	(*semaphore)->waiters = 0;
+	(*semaphore)->gpu_addr = amdgpu_sa_bo_gpu_addr((*semaphore)->sa_bo);
+
+	*((uint64_t *)amdgpu_sa_bo_cpu_addr((*semaphore)->sa_bo)) = 0;
+
+	return 0;
+}
+
+bool amdgpu_semaphore_emit_signal(struct amdgpu_ring *ring,
+				  struct amdgpu_semaphore *semaphore)
+{
+	trace_amdgpu_semaphore_signale(ring->idx, semaphore);
+
+	if (amdgpu_ring_emit_semaphore(ring, semaphore, false)) {
+		--semaphore->waiters;
+
+		/* for debugging lockup only, used by sysfs debug files */
+		ring->last_semaphore_signal_addr = semaphore->gpu_addr;
+		return true;
+	}
+	return false;
+}
+
+bool amdgpu_semaphore_emit_wait(struct amdgpu_ring *ring,
+				struct amdgpu_semaphore *semaphore)
+{
+	trace_amdgpu_semaphore_wait(ring->idx, semaphore);
+
+	if (amdgpu_ring_emit_semaphore(ring, semaphore, true)) {
+		++semaphore->waiters;
+
+		/* for debugging lockup only, used by sysfs debug files */
+		ring->last_semaphore_wait_addr = semaphore->gpu_addr;
+		return true;
+	}
+	return false;
+}
+
+void amdgpu_semaphore_free(struct amdgpu_device *adev,
+			   struct amdgpu_semaphore **semaphore,
+			   struct fence *fence)
+{
+	if (semaphore == NULL || *semaphore == NULL) {
+		return;
+	}
+	if ((*semaphore)->waiters > 0) {
+		dev_err(adev->dev, "semaphore %p has more waiters than signalers,"
+			" hardware lockup imminent!\n", *semaphore);
+	}
+	amdgpu_sa_bo_free(adev, &(*semaphore)->sa_bo, fence);
+	kfree(*semaphore);
+	*semaphore = NULL;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
new file mode 100644
index 000000000..181ce39ef
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
@@ -0,0 +1,388 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <christian.koenig@amd.com>
+ */
+
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+struct amdgpu_sync_entry {
+	struct hlist_node	node;
+	struct fence		*fence;
+};
+
+/**
+ * amdgpu_sync_create - zero init sync object
+ *
+ * @sync: sync object to initialize
+ *
+ * Just clear the sync object for now.
+ */
+void amdgpu_sync_create(struct amdgpu_sync *sync)
+{
+	unsigned i;
+
+	for (i = 0; i < AMDGPU_NUM_SYNCS; ++i)
+		sync->semaphores[i] = NULL;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		sync->sync_to[i] = NULL;
+
+	hash_init(sync->fences);
+	sync->last_vm_update = NULL;
+}
+
+static bool amdgpu_sync_same_dev(struct amdgpu_device *adev, struct fence *f)
+{
+	struct amdgpu_fence *a_fence = to_amdgpu_fence(f);
+	struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+
+	if (a_fence)
+		return a_fence->ring->adev == adev;
+
+	if (s_fence) {
+		struct amdgpu_ring *ring;
+
+		ring = container_of(s_fence->sched, struct amdgpu_ring, sched);
+		return ring->adev == adev;
+	}
+
+	return false;
+}
+
+static bool amdgpu_sync_test_owner(struct fence *f, void *owner)
+{
+	struct amdgpu_fence *a_fence = to_amdgpu_fence(f);
+	struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+	if (s_fence)
+		return s_fence->owner == owner;
+	if (a_fence)
+		return a_fence->owner == owner;
+	return false;
+}
+
+static void amdgpu_sync_keep_later(struct fence **keep, struct fence *fence)
+{
+	if (*keep && fence_is_later(*keep, fence))
+		return;
+
+	fence_put(*keep);
+	*keep = fence_get(fence);
+}
+
+/**
+ * amdgpu_sync_fence - remember to sync to this fence
+ *
+ * @sync: sync object to add fence to
+ * @fence: fence to sync to
+ *
+ */
+int amdgpu_sync_fence(struct amdgpu_device *adev, struct amdgpu_sync *sync,
+		      struct fence *f)
+{
+	struct amdgpu_sync_entry *e;
+	struct amdgpu_fence *fence;
+
+	if (!f)
+		return 0;
+
+	if (amdgpu_sync_same_dev(adev, f) &&
+	    amdgpu_sync_test_owner(f, AMDGPU_FENCE_OWNER_VM))
+		amdgpu_sync_keep_later(&sync->last_vm_update, f);
+
+	fence = to_amdgpu_fence(f);
+	if (!fence || fence->ring->adev != adev) {
+		hash_for_each_possible(sync->fences, e, node, f->context) {
+			if (unlikely(e->fence->context != f->context))
+				continue;
+
+			amdgpu_sync_keep_later(&e->fence, f);
+			return 0;
+		}
+
+		e = kmalloc(sizeof(struct amdgpu_sync_entry), GFP_KERNEL);
+		if (!e)
+			return -ENOMEM;
+
+		hash_add(sync->fences, &e->node, f->context);
+		e->fence = fence_get(f);
+		return 0;
+	}
+
+	amdgpu_sync_keep_later(&sync->sync_to[fence->ring->idx], f);
+
+	return 0;
+}
+
+static void *amdgpu_sync_get_owner(struct fence *f)
+{
+	struct amdgpu_fence *a_fence = to_amdgpu_fence(f);
+	struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+
+	if (s_fence)
+		return s_fence->owner;
+	else if (a_fence)
+		return a_fence->owner;
+	return AMDGPU_FENCE_OWNER_UNDEFINED;
+}
+
+/**
+ * amdgpu_sync_resv - use the semaphores to sync to a reservation object
+ *
+ * @sync: sync object to add fences from reservation object to
+ * @resv: reservation object with embedded fence
+ * @shared: true if we should only sync to the exclusive fence
+ *
+ * Sync to the fence using the semaphore objects
+ */
+int amdgpu_sync_resv(struct amdgpu_device *adev,
+		     struct amdgpu_sync *sync,
+		     struct reservation_object *resv,
+		     void *owner)
+{
+	struct reservation_object_list *flist;
+	struct fence *f;
+	void *fence_owner;
+	unsigned i;
+	int r = 0;
+
+	if (resv == NULL)
+		return -EINVAL;
+
+	/* always sync to the exclusive fence */
+	f = reservation_object_get_excl(resv);
+	r = amdgpu_sync_fence(adev, sync, f);
+
+	flist = reservation_object_get_list(resv);
+	if (!flist || r)
+		return r;
+
+	for (i = 0; i < flist->shared_count; ++i) {
+		f = rcu_dereference_protected(flist->shared[i],
+					      reservation_object_held(resv));
+		if (amdgpu_sync_same_dev(adev, f)) {
+			/* VM updates are only interesting
+			 * for other VM updates and moves.
+			 */
+			fence_owner = amdgpu_sync_get_owner(f);
+			if ((owner != AMDGPU_FENCE_OWNER_UNDEFINED) &&
+			    (fence_owner != AMDGPU_FENCE_OWNER_UNDEFINED) &&
+			    ((owner == AMDGPU_FENCE_OWNER_VM) !=
+			     (fence_owner == AMDGPU_FENCE_OWNER_VM)))
+				continue;
+
+			/* Ignore fence from the same owner as
+			 * long as it isn't undefined.
+			 */
+			if (owner != AMDGPU_FENCE_OWNER_UNDEFINED &&
+			    fence_owner == owner)
+				continue;
+		}
+
+		r = amdgpu_sync_fence(adev, sync, f);
+		if (r)
+			break;
+	}
+	return r;
+}
+
+struct fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	struct fence *f;
+	int i;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+
+		f = e->fence;
+
+		hash_del(&e->node);
+		kfree(e);
+
+		if (!fence_is_signaled(f))
+			return f;
+
+		fence_put(f);
+	}
+	return NULL;
+}
+
+int amdgpu_sync_wait(struct amdgpu_sync *sync)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	int i, r;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+		r = fence_wait(e->fence, false);
+		if (r)
+			return r;
+
+		hash_del(&e->node);
+		fence_put(e->fence);
+		kfree(e);
+	}
+
+	if (amdgpu_enable_semaphores)
+		return 0;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct fence *fence = sync->sync_to[i];
+		if (!fence)
+			continue;
+
+		r = fence_wait(fence, false);
+		if (r)
+			return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_sync_rings - sync ring to all registered fences
+ *
+ * @sync: sync object to use
+ * @ring: ring that needs sync
+ *
+ * Ensure that all registered fences are signaled before letting
+ * the ring continue. The caller must hold the ring lock.
+ */
+int amdgpu_sync_rings(struct amdgpu_sync *sync,
+		      struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned count = 0;
+	int i, r;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *other = adev->rings[i];
+		struct amdgpu_semaphore *semaphore;
+		struct amdgpu_fence *fence;
+
+		if (!sync->sync_to[i])
+			continue;
+
+		fence = to_amdgpu_fence(sync->sync_to[i]);
+
+		/* check if we really need to sync */
+		if (!amdgpu_enable_scheduler &&
+		    !amdgpu_fence_need_sync(fence, ring))
+			continue;
+
+		/* prevent GPU deadlocks */
+		if (!other->ready) {
+			dev_err(adev->dev, "Syncing to a disabled ring!");
+			return -EINVAL;
+		}
+
+		if (amdgpu_enable_scheduler || !amdgpu_enable_semaphores) {
+			r = fence_wait(sync->sync_to[i], true);
+			if (r)
+				return r;
+			continue;
+		}
+
+		if (count >= AMDGPU_NUM_SYNCS) {
+			/* not enough room, wait manually */
+			r = fence_wait(&fence->base, false);
+			if (r)
+				return r;
+			continue;
+		}
+		r = amdgpu_semaphore_create(adev, &semaphore);
+		if (r)
+			return r;
+
+		sync->semaphores[count++] = semaphore;
+
+		/* allocate enough space for sync command */
+		r = amdgpu_ring_alloc(other, 16);
+		if (r)
+			return r;
+
+		/* emit the signal semaphore */
+		if (!amdgpu_semaphore_emit_signal(other, semaphore)) {
+			/* signaling wasn't successful wait manually */
+			amdgpu_ring_undo(other);
+			r = fence_wait(&fence->base, false);
+			if (r)
+				return r;
+			continue;
+		}
+
+		/* we assume caller has already allocated space on waiters ring */
+		if (!amdgpu_semaphore_emit_wait(ring, semaphore)) {
+			/* waiting wasn't successful wait manually */
+			amdgpu_ring_undo(other);
+			r = fence_wait(&fence->base, false);
+			if (r)
+				return r;
+			continue;
+		}
+
+		amdgpu_ring_commit(other);
+		amdgpu_fence_note_sync(fence, ring);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_sync_free - free the sync object
+ *
+ * @adev: amdgpu_device pointer
+ * @sync: sync object to use
+ * @fence: fence to use for the free
+ *
+ * Free the sync object by freeing all semaphores in it.
+ */
+void amdgpu_sync_free(struct amdgpu_device *adev,
+		      struct amdgpu_sync *sync,
+		      struct fence *fence)
+{
+	struct amdgpu_sync_entry *e;
+	struct hlist_node *tmp;
+	unsigned i;
+
+	hash_for_each_safe(sync->fences, i, tmp, e, node) {
+		hash_del(&e->node);
+		fence_put(e->fence);
+		kfree(e);
+	}
+
+	for (i = 0; i < AMDGPU_NUM_SYNCS; ++i)
+		amdgpu_semaphore_free(adev, &sync->semaphores[i], fence);
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		fence_put(sync->sync_to[i]);
+
+	fence_put(sync->last_vm_update);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
new file mode 100644
index 000000000..4865615e9
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
@@ -0,0 +1,555 @@
+/*
+ * Copyright 2009 VMware, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Michel Dänzer
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+
+/* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */
+static void amdgpu_do_test_moves(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
+	struct amdgpu_bo *vram_obj = NULL;
+	struct amdgpu_bo **gtt_obj = NULL;
+	uint64_t gtt_addr, vram_addr;
+	unsigned n, size;
+	int i, r;
+
+	size = 1024 * 1024;
+
+	/* Number of tests =
+	 * (Total GTT - IB pool - writeback page - ring buffers) / test size
+	 */
+	n = adev->mc.gtt_size - AMDGPU_IB_POOL_SIZE*64*1024;
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		if (adev->rings[i])
+			n -= adev->rings[i]->ring_size;
+	if (adev->wb.wb_obj)
+		n -= AMDGPU_GPU_PAGE_SIZE;
+	if (adev->irq.ih.ring_obj)
+		n -= adev->irq.ih.ring_size;
+	n /= size;
+
+	gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL);
+	if (!gtt_obj) {
+		DRM_ERROR("Failed to allocate %d pointers\n", n);
+		r = 1;
+		goto out_cleanup;
+	}
+
+	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
+			     AMDGPU_GEM_DOMAIN_VRAM, 0,
+			     NULL, NULL, &vram_obj);
+	if (r) {
+		DRM_ERROR("Failed to create VRAM object\n");
+		goto out_cleanup;
+	}
+	r = amdgpu_bo_reserve(vram_obj, false);
+	if (unlikely(r != 0))
+		goto out_unref;
+	r = amdgpu_bo_pin(vram_obj, AMDGPU_GEM_DOMAIN_VRAM, &vram_addr);
+	if (r) {
+		DRM_ERROR("Failed to pin VRAM object\n");
+		goto out_unres;
+	}
+	for (i = 0; i < n; i++) {
+		void *gtt_map, *vram_map;
+		void **gtt_start, **gtt_end;
+		void **vram_start, **vram_end;
+		struct fence *fence = NULL;
+
+		r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL,
+				     NULL, gtt_obj + i);
+		if (r) {
+			DRM_ERROR("Failed to create GTT object %d\n", i);
+			goto out_lclean;
+		}
+
+		r = amdgpu_bo_reserve(gtt_obj[i], false);
+		if (unlikely(r != 0))
+			goto out_lclean_unref;
+		r = amdgpu_bo_pin(gtt_obj[i], AMDGPU_GEM_DOMAIN_GTT, &gtt_addr);
+		if (r) {
+			DRM_ERROR("Failed to pin GTT object %d\n", i);
+			goto out_lclean_unres;
+		}
+
+		r = amdgpu_bo_kmap(gtt_obj[i], &gtt_map);
+		if (r) {
+			DRM_ERROR("Failed to map GTT object %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		for (gtt_start = gtt_map, gtt_end = gtt_map + size;
+		     gtt_start < gtt_end;
+		     gtt_start++)
+			*gtt_start = gtt_start;
+
+		amdgpu_bo_kunmap(gtt_obj[i]);
+
+		r = amdgpu_copy_buffer(ring, gtt_addr, vram_addr,
+				       size, NULL, &fence);
+
+		if (r) {
+			DRM_ERROR("Failed GTT->VRAM copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		r = fence_wait(fence, false);
+		if (r) {
+			DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		fence_put(fence);
+
+		r = amdgpu_bo_kmap(vram_obj, &vram_map);
+		if (r) {
+			DRM_ERROR("Failed to map VRAM object after copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		for (gtt_start = gtt_map, gtt_end = gtt_map + size,
+		     vram_start = vram_map, vram_end = vram_map + size;
+		     vram_start < vram_end;
+		     gtt_start++, vram_start++) {
+			if (*vram_start != gtt_start) {
+				DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, "
+					  "expected 0x%p (GTT/VRAM offset "
+					  "0x%16llx/0x%16llx)\n",
+					  i, *vram_start, gtt_start,
+					  (unsigned long long)
+					  (gtt_addr - adev->mc.gtt_start +
+					   (void*)gtt_start - gtt_map),
+					  (unsigned long long)
+					  (vram_addr - adev->mc.vram_start +
+					   (void*)gtt_start - gtt_map));
+				amdgpu_bo_kunmap(vram_obj);
+				goto out_lclean_unpin;
+			}
+			*vram_start = vram_start;
+		}
+
+		amdgpu_bo_kunmap(vram_obj);
+
+		r = amdgpu_copy_buffer(ring, vram_addr, gtt_addr,
+				       size, NULL, &fence);
+
+		if (r) {
+			DRM_ERROR("Failed VRAM->GTT copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		r = fence_wait(fence, false);
+		if (r) {
+			DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		fence_put(fence);
+
+		r = amdgpu_bo_kmap(gtt_obj[i], &gtt_map);
+		if (r) {
+			DRM_ERROR("Failed to map GTT object after copy %d\n", i);
+			goto out_lclean_unpin;
+		}
+
+		for (gtt_start = gtt_map, gtt_end = gtt_map + size,
+		     vram_start = vram_map, vram_end = vram_map + size;
+		     gtt_start < gtt_end;
+		     gtt_start++, vram_start++) {
+			if (*gtt_start != vram_start) {
+				DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, "
+					  "expected 0x%p (VRAM/GTT offset "
+					  "0x%16llx/0x%16llx)\n",
+					  i, *gtt_start, vram_start,
+					  (unsigned long long)
+					  (vram_addr - adev->mc.vram_start +
+					   (void*)vram_start - vram_map),
+					  (unsigned long long)
+					  (gtt_addr - adev->mc.gtt_start +
+					   (void*)vram_start - vram_map));
+				amdgpu_bo_kunmap(gtt_obj[i]);
+				goto out_lclean_unpin;
+			}
+		}
+
+		amdgpu_bo_kunmap(gtt_obj[i]);
+
+		DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n",
+			 gtt_addr - adev->mc.gtt_start);
+		continue;
+
+out_lclean_unpin:
+		amdgpu_bo_unpin(gtt_obj[i]);
+out_lclean_unres:
+		amdgpu_bo_unreserve(gtt_obj[i]);
+out_lclean_unref:
+		amdgpu_bo_unref(&gtt_obj[i]);
+out_lclean:
+		for (--i; i >= 0; --i) {
+			amdgpu_bo_unpin(gtt_obj[i]);
+			amdgpu_bo_unreserve(gtt_obj[i]);
+			amdgpu_bo_unref(&gtt_obj[i]);
+		}
+		if (fence)
+			fence_put(fence);
+		break;
+	}
+
+	amdgpu_bo_unpin(vram_obj);
+out_unres:
+	amdgpu_bo_unreserve(vram_obj);
+out_unref:
+	amdgpu_bo_unref(&vram_obj);
+out_cleanup:
+	kfree(gtt_obj);
+	if (r) {
+		printk(KERN_WARNING "Error while testing BO move.\n");
+	}
+}
+
+void amdgpu_test_moves(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs)
+		amdgpu_do_test_moves(adev);
+}
+
+static int amdgpu_test_create_and_emit_fence(struct amdgpu_device *adev,
+					     struct amdgpu_ring *ring,
+					     struct fence **fence)
+{
+	uint32_t handle = ring->idx ^ 0xdeafbeef;
+	int r;
+
+	if (ring == &adev->uvd.ring) {
+		r = amdgpu_uvd_get_create_msg(ring, handle, NULL);
+		if (r) {
+			DRM_ERROR("Failed to get dummy create msg\n");
+			return r;
+		}
+
+		r = amdgpu_uvd_get_destroy_msg(ring, handle, fence);
+		if (r) {
+			DRM_ERROR("Failed to get dummy destroy msg\n");
+			return r;
+		}
+
+	} else if (ring == &adev->vce.ring[0] ||
+		   ring == &adev->vce.ring[1]) {
+		r = amdgpu_vce_get_create_msg(ring, handle, NULL);
+		if (r) {
+			DRM_ERROR("Failed to get dummy create msg\n");
+			return r;
+		}
+
+		r = amdgpu_vce_get_destroy_msg(ring, handle, fence);
+		if (r) {
+			DRM_ERROR("Failed to get dummy destroy msg\n");
+			return r;
+		}
+	} else {
+		struct amdgpu_fence *a_fence = NULL;
+		r = amdgpu_ring_lock(ring, 64);
+		if (r) {
+			DRM_ERROR("Failed to lock ring A %d\n", ring->idx);
+			return r;
+		}
+		amdgpu_fence_emit(ring, AMDGPU_FENCE_OWNER_UNDEFINED, &a_fence);
+		amdgpu_ring_unlock_commit(ring);
+		*fence = &a_fence->base;
+	}
+	return 0;
+}
+
+void amdgpu_test_ring_sync(struct amdgpu_device *adev,
+			   struct amdgpu_ring *ringA,
+			   struct amdgpu_ring *ringB)
+{
+	struct fence *fence1 = NULL, *fence2 = NULL;
+	struct amdgpu_semaphore *semaphore = NULL;
+	int r;
+
+	r = amdgpu_semaphore_create(adev, &semaphore);
+	if (r) {
+		DRM_ERROR("Failed to create semaphore\n");
+		goto out_cleanup;
+	}
+
+	r = amdgpu_ring_lock(ringA, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring A %d\n", ringA->idx);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_wait(ringA, semaphore);
+	amdgpu_ring_unlock_commit(ringA);
+
+	r = amdgpu_test_create_and_emit_fence(adev, ringA, &fence1);
+	if (r)
+		goto out_cleanup;
+
+	r = amdgpu_ring_lock(ringA, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring A %d\n", ringA->idx);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_wait(ringA, semaphore);
+	amdgpu_ring_unlock_commit(ringA);
+
+	r = amdgpu_test_create_and_emit_fence(adev, ringA, &fence2);
+	if (r)
+		goto out_cleanup;
+
+	mdelay(1000);
+
+	if (fence_is_signaled(fence1)) {
+		DRM_ERROR("Fence 1 signaled without waiting for semaphore.\n");
+		goto out_cleanup;
+	}
+
+	r = amdgpu_ring_lock(ringB, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring B %p\n", ringB);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_signal(ringB, semaphore);
+	amdgpu_ring_unlock_commit(ringB);
+
+	r = fence_wait(fence1, false);
+	if (r) {
+		DRM_ERROR("Failed to wait for sync fence 1\n");
+		goto out_cleanup;
+	}
+
+	mdelay(1000);
+
+	if (fence_is_signaled(fence2)) {
+		DRM_ERROR("Fence 2 signaled without waiting for semaphore.\n");
+		goto out_cleanup;
+	}
+
+	r = amdgpu_ring_lock(ringB, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring B %p\n", ringB);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_signal(ringB, semaphore);
+	amdgpu_ring_unlock_commit(ringB);
+
+	r = fence_wait(fence2, false);
+	if (r) {
+		DRM_ERROR("Failed to wait for sync fence 1\n");
+		goto out_cleanup;
+	}
+
+out_cleanup:
+	amdgpu_semaphore_free(adev, &semaphore, NULL);
+
+	if (fence1)
+		fence_put(fence1);
+
+	if (fence2)
+		fence_put(fence2);
+
+	if (r)
+		printk(KERN_WARNING "Error while testing ring sync (%d).\n", r);
+}
+
+static void amdgpu_test_ring_sync2(struct amdgpu_device *adev,
+			    struct amdgpu_ring *ringA,
+			    struct amdgpu_ring *ringB,
+			    struct amdgpu_ring *ringC)
+{
+	struct fence *fenceA = NULL, *fenceB = NULL;
+	struct amdgpu_semaphore *semaphore = NULL;
+	bool sigA, sigB;
+	int i, r;
+
+	r = amdgpu_semaphore_create(adev, &semaphore);
+	if (r) {
+		DRM_ERROR("Failed to create semaphore\n");
+		goto out_cleanup;
+	}
+
+	r = amdgpu_ring_lock(ringA, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring A %d\n", ringA->idx);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_wait(ringA, semaphore);
+	amdgpu_ring_unlock_commit(ringA);
+
+	r = amdgpu_test_create_and_emit_fence(adev, ringA, &fenceA);
+	if (r)
+		goto out_cleanup;
+
+	r = amdgpu_ring_lock(ringB, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring B %d\n", ringB->idx);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_wait(ringB, semaphore);
+	amdgpu_ring_unlock_commit(ringB);
+	r = amdgpu_test_create_and_emit_fence(adev, ringB, &fenceB);
+	if (r)
+		goto out_cleanup;
+
+	mdelay(1000);
+
+	if (fence_is_signaled(fenceA)) {
+		DRM_ERROR("Fence A signaled without waiting for semaphore.\n");
+		goto out_cleanup;
+	}
+	if (fence_is_signaled(fenceB)) {
+		DRM_ERROR("Fence B signaled without waiting for semaphore.\n");
+		goto out_cleanup;
+	}
+
+	r = amdgpu_ring_lock(ringC, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring B %p\n", ringC);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_signal(ringC, semaphore);
+	amdgpu_ring_unlock_commit(ringC);
+
+	for (i = 0; i < 30; ++i) {
+		mdelay(100);
+		sigA = fence_is_signaled(fenceA);
+		sigB = fence_is_signaled(fenceB);
+		if (sigA || sigB)
+			break;
+	}
+
+	if (!sigA && !sigB) {
+		DRM_ERROR("Neither fence A nor B has been signaled\n");
+		goto out_cleanup;
+	} else if (sigA && sigB) {
+		DRM_ERROR("Both fence A and B has been signaled\n");
+		goto out_cleanup;
+	}
+
+	DRM_INFO("Fence %c was first signaled\n", sigA ? 'A' : 'B');
+
+	r = amdgpu_ring_lock(ringC, 64);
+	if (r) {
+		DRM_ERROR("Failed to lock ring B %p\n", ringC);
+		goto out_cleanup;
+	}
+	amdgpu_semaphore_emit_signal(ringC, semaphore);
+	amdgpu_ring_unlock_commit(ringC);
+
+	mdelay(1000);
+
+	r = fence_wait(fenceA, false);
+	if (r) {
+		DRM_ERROR("Failed to wait for sync fence A\n");
+		goto out_cleanup;
+	}
+	r = fence_wait(fenceB, false);
+	if (r) {
+		DRM_ERROR("Failed to wait for sync fence B\n");
+		goto out_cleanup;
+	}
+
+out_cleanup:
+	amdgpu_semaphore_free(adev, &semaphore, NULL);
+
+	if (fenceA)
+		fence_put(fenceA);
+
+	if (fenceB)
+		fence_put(fenceB);
+
+	if (r)
+		printk(KERN_WARNING "Error while testing ring sync (%d).\n", r);
+}
+
+static bool amdgpu_test_sync_possible(struct amdgpu_ring *ringA,
+				      struct amdgpu_ring *ringB)
+{
+	if (ringA == &ringA->adev->vce.ring[0] &&
+	    ringB == &ringB->adev->vce.ring[1])
+		return false;
+
+	return true;
+}
+
+void amdgpu_test_syncing(struct amdgpu_device *adev)
+{
+	int i, j, k;
+
+	for (i = 1; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ringA = adev->rings[i];
+		if (!ringA || !ringA->ready)
+			continue;
+
+		for (j = 0; j < i; ++j) {
+			struct amdgpu_ring *ringB = adev->rings[j];
+			if (!ringB || !ringB->ready)
+				continue;
+
+			if (!amdgpu_test_sync_possible(ringA, ringB))
+				continue;
+
+			DRM_INFO("Testing syncing between rings %d and %d...\n", i, j);
+			amdgpu_test_ring_sync(adev, ringA, ringB);
+
+			DRM_INFO("Testing syncing between rings %d and %d...\n", j, i);
+			amdgpu_test_ring_sync(adev, ringB, ringA);
+
+			for (k = 0; k < j; ++k) {
+				struct amdgpu_ring *ringC = adev->rings[k];
+				if (!ringC || !ringC->ready)
+					continue;
+
+				if (!amdgpu_test_sync_possible(ringA, ringC))
+					continue;
+
+				if (!amdgpu_test_sync_possible(ringB, ringC))
+					continue;
+
+				DRM_INFO("Testing syncing between rings %d, %d and %d...\n", i, j, k);
+				amdgpu_test_ring_sync2(adev, ringA, ringB, ringC);
+
+				DRM_INFO("Testing syncing between rings %d, %d and %d...\n", i, k, j);
+				amdgpu_test_ring_sync2(adev, ringA, ringC, ringB);
+
+				DRM_INFO("Testing syncing between rings %d, %d and %d...\n", j, i, k);
+				amdgpu_test_ring_sync2(adev, ringB, ringA, ringC);
+
+				DRM_INFO("Testing syncing between rings %d, %d and %d...\n", j, k, i);
+				amdgpu_test_ring_sync2(adev, ringB, ringC, ringA);
+
+				DRM_INFO("Testing syncing between rings %d, %d and %d...\n", k, i, j);
+				amdgpu_test_ring_sync2(adev, ringC, ringA, ringB);
+
+				DRM_INFO("Testing syncing between rings %d, %d and %d...\n", k, j, i);
+				amdgpu_test_ring_sync2(adev, ringC, ringB, ringA);
+			}
+		}
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
new file mode 100644
index 000000000..8f9834ab1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -0,0 +1,291 @@
+#if !defined(_AMDGPU_TRACE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _AMDGPU_TRACE_H_
+
+#include <linux/stringify.h>
+#include <linux/types.h>
+#include <linux/tracepoint.h>
+
+#include <drm/drmP.h>
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM amdgpu
+#define TRACE_INCLUDE_FILE amdgpu_trace
+
+TRACE_EVENT(amdgpu_bo_create,
+	    TP_PROTO(struct amdgpu_bo *bo),
+	    TP_ARGS(bo),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo *, bo)
+			     __field(u32, pages)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo = bo;
+			   __entry->pages = bo->tbo.num_pages;
+			   ),
+	    TP_printk("bo=%p, pages=%u", __entry->bo, __entry->pages)
+);
+
+TRACE_EVENT(amdgpu_cs,
+	    TP_PROTO(struct amdgpu_cs_parser *p, int i),
+	    TP_ARGS(p, i),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo_list *, bo_list)
+			     __field(u32, ring)
+			     __field(u32, dw)
+			     __field(u32, fences)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo_list = p->bo_list;
+			   __entry->ring = p->ibs[i].ring->idx;
+			   __entry->dw = p->ibs[i].length_dw;
+			   __entry->fences = amdgpu_fence_count_emitted(
+				p->ibs[i].ring);
+			   ),
+	    TP_printk("bo_list=%p, ring=%u, dw=%u, fences=%u",
+		      __entry->bo_list, __entry->ring, __entry->dw,
+		      __entry->fences)
+);
+
+TRACE_EVENT(amdgpu_cs_ioctl,
+	    TP_PROTO(struct amdgpu_job *job),
+	    TP_ARGS(job),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_device *, adev)
+			     __field(struct amd_sched_job *, sched_job)
+			     __field(struct amdgpu_ib *, ib)
+			     __field(struct fence *, fence)
+			     __field(char *, ring_name)
+			     __field(u32, num_ibs)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->adev = job->adev;
+			   __entry->sched_job = &job->base;
+			   __entry->ib = job->ibs;
+			   __entry->fence = &job->base.s_fence->base;
+			   __entry->ring_name = job->ibs[0].ring->name;
+			   __entry->num_ibs = job->num_ibs;
+			   ),
+	    TP_printk("adev=%p, sched_job=%p, first ib=%p, sched fence=%p, ring name:%s, num_ibs:%u",
+		      __entry->adev, __entry->sched_job, __entry->ib,
+		      __entry->fence, __entry->ring_name, __entry->num_ibs)
+);
+
+TRACE_EVENT(amdgpu_sched_run_job,
+	    TP_PROTO(struct amdgpu_job *job),
+	    TP_ARGS(job),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_device *, adev)
+			     __field(struct amd_sched_job *, sched_job)
+			     __field(struct amdgpu_ib *, ib)
+			     __field(struct fence *, fence)
+			     __field(char *, ring_name)
+			     __field(u32, num_ibs)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->adev = job->adev;
+			   __entry->sched_job = &job->base;
+			   __entry->ib = job->ibs;
+			   __entry->fence = &job->base.s_fence->base;
+			   __entry->ring_name = job->ibs[0].ring->name;
+			   __entry->num_ibs = job->num_ibs;
+			   ),
+	    TP_printk("adev=%p, sched_job=%p, first ib=%p, sched fence=%p, ring name:%s, num_ibs:%u",
+		      __entry->adev, __entry->sched_job, __entry->ib,
+		      __entry->fence, __entry->ring_name, __entry->num_ibs)
+);
+
+
+TRACE_EVENT(amdgpu_vm_grab_id,
+	    TP_PROTO(unsigned vmid, int ring),
+	    TP_ARGS(vmid, ring),
+	    TP_STRUCT__entry(
+			     __field(u32, vmid)
+			     __field(u32, ring)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->vmid = vmid;
+			   __entry->ring = ring;
+			   ),
+	    TP_printk("vmid=%u, ring=%u", __entry->vmid, __entry->ring)
+);
+
+TRACE_EVENT(amdgpu_vm_bo_map,
+	    TP_PROTO(struct amdgpu_bo_va *bo_va,
+		     struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(bo_va, mapping),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo *, bo)
+			     __field(long, start)
+			     __field(long, last)
+			     __field(u64, offset)
+			     __field(u32, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo = bo_va->bo;
+			   __entry->start = mapping->it.start;
+			   __entry->last = mapping->it.last;
+			   __entry->offset = mapping->offset;
+			   __entry->flags = mapping->flags;
+			   ),
+	    TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%08x",
+		      __entry->bo, __entry->start, __entry->last,
+		      __entry->offset, __entry->flags)
+);
+
+TRACE_EVENT(amdgpu_vm_bo_unmap,
+	    TP_PROTO(struct amdgpu_bo_va *bo_va,
+		     struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(bo_va, mapping),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo *, bo)
+			     __field(long, start)
+			     __field(long, last)
+			     __field(u64, offset)
+			     __field(u32, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->bo = bo_va->bo;
+			   __entry->start = mapping->it.start;
+			   __entry->last = mapping->it.last;
+			   __entry->offset = mapping->offset;
+			   __entry->flags = mapping->flags;
+			   ),
+	    TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%08x",
+		      __entry->bo, __entry->start, __entry->last,
+		      __entry->offset, __entry->flags)
+);
+
+DECLARE_EVENT_CLASS(amdgpu_vm_mapping,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping),
+	    TP_STRUCT__entry(
+			     __field(u64, soffset)
+			     __field(u64, eoffset)
+			     __field(u32, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->soffset = mapping->it.start;
+			   __entry->eoffset = mapping->it.last + 1;
+			   __entry->flags = mapping->flags;
+			   ),
+	    TP_printk("soffs=%010llx, eoffs=%010llx, flags=%08x",
+		      __entry->soffset, __entry->eoffset, __entry->flags)
+);
+
+DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_update,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping)
+);
+
+DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_mapping,
+	    TP_PROTO(struct amdgpu_bo_va_mapping *mapping),
+	    TP_ARGS(mapping)
+);
+
+TRACE_EVENT(amdgpu_vm_set_page,
+	    TP_PROTO(uint64_t pe, uint64_t addr, unsigned count,
+		     uint32_t incr, uint32_t flags),
+	    TP_ARGS(pe, addr, count, incr, flags),
+	    TP_STRUCT__entry(
+			     __field(u64, pe)
+			     __field(u64, addr)
+			     __field(u32, count)
+			     __field(u32, incr)
+			     __field(u32, flags)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->pe = pe;
+			   __entry->addr = addr;
+			   __entry->count = count;
+			   __entry->incr = incr;
+			   __entry->flags = flags;
+			   ),
+	    TP_printk("pe=%010Lx, addr=%010Lx, incr=%u, flags=%08x, count=%u",
+		      __entry->pe, __entry->addr, __entry->incr,
+		      __entry->flags, __entry->count)
+);
+
+TRACE_EVENT(amdgpu_vm_flush,
+	    TP_PROTO(uint64_t pd_addr, unsigned ring, unsigned id),
+	    TP_ARGS(pd_addr, ring, id),
+	    TP_STRUCT__entry(
+			     __field(u64, pd_addr)
+			     __field(u32, ring)
+			     __field(u32, id)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->pd_addr = pd_addr;
+			   __entry->ring = ring;
+			   __entry->id = id;
+			   ),
+	    TP_printk("pd_addr=%010Lx, ring=%u, id=%u",
+		      __entry->pd_addr, __entry->ring, __entry->id)
+);
+
+TRACE_EVENT(amdgpu_bo_list_set,
+	    TP_PROTO(struct amdgpu_bo_list *list, struct amdgpu_bo *bo),
+	    TP_ARGS(list, bo),
+	    TP_STRUCT__entry(
+			     __field(struct amdgpu_bo_list *, list)
+			     __field(struct amdgpu_bo *, bo)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->list = list;
+			   __entry->bo = bo;
+			   ),
+	    TP_printk("list=%p, bo=%p", __entry->list, __entry->bo)
+);
+
+DECLARE_EVENT_CLASS(amdgpu_semaphore_request,
+
+	    TP_PROTO(int ring, struct amdgpu_semaphore *sem),
+
+	    TP_ARGS(ring, sem),
+
+	    TP_STRUCT__entry(
+			     __field(int, ring)
+			     __field(signed, waiters)
+			     __field(uint64_t, gpu_addr)
+			     ),
+
+	    TP_fast_assign(
+			   __entry->ring = ring;
+			   __entry->waiters = sem->waiters;
+			   __entry->gpu_addr = sem->gpu_addr;
+			   ),
+
+	    TP_printk("ring=%u, waiters=%d, addr=%010Lx", __entry->ring,
+		      __entry->waiters, __entry->gpu_addr)
+);
+
+DEFINE_EVENT(amdgpu_semaphore_request, amdgpu_semaphore_signale,
+
+	    TP_PROTO(int ring, struct amdgpu_semaphore *sem),
+
+	    TP_ARGS(ring, sem)
+);
+
+DEFINE_EVENT(amdgpu_semaphore_request, amdgpu_semaphore_wait,
+
+	    TP_PROTO(int ring, struct amdgpu_semaphore *sem),
+
+	    TP_ARGS(ring, sem)
+);
+
+#endif
+
+/* This part must be outside protection */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#include <trace/define_trace.h>
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c
new file mode 100644
index 000000000..385b7e1d7
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_trace_points.c
@@ -0,0 +1,9 @@
+/* Copyright Red Hat Inc 2010.
+ * Author : Dave Airlie <airlied@redhat.com>
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#define CREATE_TRACE_POINTS
+#include "amdgpu_trace.h"
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
new file mode 100644
index 000000000..1cbb16e15
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -0,0 +1,1254 @@
+/*
+ * Copyright 2009 Jerome Glisse.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Jerome Glisse <glisse@freedesktop.org>
+ *    Thomas Hellstrom <thomas-at-tungstengraphics-dot-com>
+ *    Dave Airlie
+ */
+#include <ttm/ttm_bo_api.h>
+#include <ttm/ttm_bo_driver.h>
+#include <ttm/ttm_placement.h>
+#include <ttm/ttm_module.h>
+#include <ttm/ttm_page_alloc.h>
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/swiotlb.h>
+#include <linux/swap.h>
+#include <linux/pagemap.h>
+#include <linux/debugfs.h>
+#include "amdgpu.h"
+#include "bif/bif_4_1_d.h"
+
+#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT)
+
+static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev);
+static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev);
+
+static struct amdgpu_device *amdgpu_get_adev(struct ttm_bo_device *bdev)
+{
+	struct amdgpu_mman *mman;
+	struct amdgpu_device *adev;
+
+	mman = container_of(bdev, struct amdgpu_mman, bdev);
+	adev = container_of(mman, struct amdgpu_device, mman);
+	return adev;
+}
+
+
+/*
+ * Global memory.
+ */
+static int amdgpu_ttm_mem_global_init(struct drm_global_reference *ref)
+{
+	return ttm_mem_global_init(ref->object);
+}
+
+static void amdgpu_ttm_mem_global_release(struct drm_global_reference *ref)
+{
+	ttm_mem_global_release(ref->object);
+}
+
+static int amdgpu_ttm_global_init(struct amdgpu_device *adev)
+{
+	struct drm_global_reference *global_ref;
+	int r;
+
+	adev->mman.mem_global_referenced = false;
+	global_ref = &adev->mman.mem_global_ref;
+	global_ref->global_type = DRM_GLOBAL_TTM_MEM;
+	global_ref->size = sizeof(struct ttm_mem_global);
+	global_ref->init = &amdgpu_ttm_mem_global_init;
+	global_ref->release = &amdgpu_ttm_mem_global_release;
+	r = drm_global_item_ref(global_ref);
+	if (r != 0) {
+		DRM_ERROR("Failed setting up TTM memory accounting "
+			  "subsystem.\n");
+		return r;
+	}
+
+	adev->mman.bo_global_ref.mem_glob =
+		adev->mman.mem_global_ref.object;
+	global_ref = &adev->mman.bo_global_ref.ref;
+	global_ref->global_type = DRM_GLOBAL_TTM_BO;
+	global_ref->size = sizeof(struct ttm_bo_global);
+	global_ref->init = &ttm_bo_global_init;
+	global_ref->release = &ttm_bo_global_release;
+	r = drm_global_item_ref(global_ref);
+	if (r != 0) {
+		DRM_ERROR("Failed setting up TTM BO subsystem.\n");
+		drm_global_item_unref(&adev->mman.mem_global_ref);
+		return r;
+	}
+
+	adev->mman.mem_global_referenced = true;
+	return 0;
+}
+
+static void amdgpu_ttm_global_fini(struct amdgpu_device *adev)
+{
+	if (adev->mman.mem_global_referenced) {
+		drm_global_item_unref(&adev->mman.bo_global_ref.ref);
+		drm_global_item_unref(&adev->mman.mem_global_ref);
+		adev->mman.mem_global_referenced = false;
+	}
+}
+
+static int amdgpu_invalidate_caches(struct ttm_bo_device *bdev, uint32_t flags)
+{
+	return 0;
+}
+
+static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type,
+				struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_device *adev;
+
+	adev = amdgpu_get_adev(bdev);
+
+	switch (type) {
+	case TTM_PL_SYSTEM:
+		/* System memory */
+		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE;
+		man->available_caching = TTM_PL_MASK_CACHING;
+		man->default_caching = TTM_PL_FLAG_CACHED;
+		break;
+	case TTM_PL_TT:
+		man->func = &ttm_bo_manager_func;
+		man->gpu_offset = adev->mc.gtt_start;
+		man->available_caching = TTM_PL_MASK_CACHING;
+		man->default_caching = TTM_PL_FLAG_CACHED;
+		man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | TTM_MEMTYPE_FLAG_CMA;
+		break;
+	case TTM_PL_VRAM:
+		/* "On-card" video ram */
+		man->func = &ttm_bo_manager_func;
+		man->gpu_offset = adev->mc.vram_start;
+		man->flags = TTM_MEMTYPE_FLAG_FIXED |
+			     TTM_MEMTYPE_FLAG_MAPPABLE;
+		man->available_caching = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_WC;
+		man->default_caching = TTM_PL_FLAG_WC;
+		break;
+	case AMDGPU_PL_GDS:
+	case AMDGPU_PL_GWS:
+	case AMDGPU_PL_OA:
+		/* On-chip GDS memory*/
+		man->func = &ttm_bo_manager_func;
+		man->gpu_offset = 0;
+		man->flags = TTM_MEMTYPE_FLAG_FIXED | TTM_MEMTYPE_FLAG_CMA;
+		man->available_caching = TTM_PL_FLAG_UNCACHED;
+		man->default_caching = TTM_PL_FLAG_UNCACHED;
+		break;
+	default:
+		DRM_ERROR("Unsupported memory type %u\n", (unsigned)type);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
+				struct ttm_placement *placement)
+{
+	struct amdgpu_bo *rbo;
+	static struct ttm_place placements = {
+		.fpfn = 0,
+		.lpfn = 0,
+		.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM
+	};
+
+	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo)) {
+		placement->placement = &placements;
+		placement->busy_placement = &placements;
+		placement->num_placement = 1;
+		placement->num_busy_placement = 1;
+		return;
+	}
+	rbo = container_of(bo, struct amdgpu_bo, tbo);
+	switch (bo->mem.mem_type) {
+	case TTM_PL_VRAM:
+		if (rbo->adev->mman.buffer_funcs_ring->ready == false)
+			amdgpu_ttm_placement_from_domain(rbo, AMDGPU_GEM_DOMAIN_CPU);
+		else
+			amdgpu_ttm_placement_from_domain(rbo, AMDGPU_GEM_DOMAIN_GTT);
+		break;
+	case TTM_PL_TT:
+	default:
+		amdgpu_ttm_placement_from_domain(rbo, AMDGPU_GEM_DOMAIN_CPU);
+	}
+	*placement = rbo->placement;
+}
+
+static int amdgpu_verify_access(struct ttm_buffer_object *bo, struct file *filp)
+{
+	struct amdgpu_bo *rbo = container_of(bo, struct amdgpu_bo, tbo);
+
+	return drm_vma_node_verify_access(&rbo->gem_base.vma_node, filp);
+}
+
+static void amdgpu_move_null(struct ttm_buffer_object *bo,
+			     struct ttm_mem_reg *new_mem)
+{
+	struct ttm_mem_reg *old_mem = &bo->mem;
+
+	BUG_ON(old_mem->mm_node != NULL);
+	*old_mem = *new_mem;
+	new_mem->mm_node = NULL;
+}
+
+static int amdgpu_move_blit(struct ttm_buffer_object *bo,
+			bool evict, bool no_wait_gpu,
+			struct ttm_mem_reg *new_mem,
+			struct ttm_mem_reg *old_mem)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_ring *ring;
+	uint64_t old_start, new_start;
+	struct fence *fence;
+	int r;
+
+	adev = amdgpu_get_adev(bo->bdev);
+	ring = adev->mman.buffer_funcs_ring;
+	old_start = old_mem->start << PAGE_SHIFT;
+	new_start = new_mem->start << PAGE_SHIFT;
+
+	switch (old_mem->mem_type) {
+	case TTM_PL_VRAM:
+		old_start += adev->mc.vram_start;
+		break;
+	case TTM_PL_TT:
+		old_start += adev->mc.gtt_start;
+		break;
+	default:
+		DRM_ERROR("Unknown placement %d\n", old_mem->mem_type);
+		return -EINVAL;
+	}
+	switch (new_mem->mem_type) {
+	case TTM_PL_VRAM:
+		new_start += adev->mc.vram_start;
+		break;
+	case TTM_PL_TT:
+		new_start += adev->mc.gtt_start;
+		break;
+	default:
+		DRM_ERROR("Unknown placement %d\n", old_mem->mem_type);
+		return -EINVAL;
+	}
+	if (!ring->ready) {
+		DRM_ERROR("Trying to move memory with ring turned off.\n");
+		return -EINVAL;
+	}
+
+	BUILD_BUG_ON((PAGE_SIZE % AMDGPU_GPU_PAGE_SIZE) != 0);
+
+	r = amdgpu_copy_buffer(ring, old_start, new_start,
+			       new_mem->num_pages * PAGE_SIZE, /* bytes */
+			       bo->resv, &fence);
+	/* FIXME: handle copy error */
+	r = ttm_bo_move_accel_cleanup(bo, fence,
+				      evict, no_wait_gpu, new_mem);
+	fence_put(fence);
+	return r;
+}
+
+static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo,
+				bool evict, bool interruptible,
+				bool no_wait_gpu,
+				struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	struct ttm_mem_reg tmp_mem;
+	struct ttm_place placements;
+	struct ttm_placement placement;
+	int r;
+
+	adev = amdgpu_get_adev(bo->bdev);
+	tmp_mem = *new_mem;
+	tmp_mem.mm_node = NULL;
+	placement.num_placement = 1;
+	placement.placement = &placements;
+	placement.num_busy_placement = 1;
+	placement.busy_placement = &placements;
+	placements.fpfn = 0;
+	placements.lpfn = 0;
+	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
+	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
+			     interruptible, no_wait_gpu);
+	if (unlikely(r)) {
+		return r;
+	}
+
+	r = ttm_tt_set_placement_caching(bo->ttm, tmp_mem.placement);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+
+	r = ttm_tt_bind(bo->ttm, &tmp_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+	r = amdgpu_move_blit(bo, true, no_wait_gpu, &tmp_mem, old_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+	r = ttm_bo_move_ttm(bo, true, no_wait_gpu, new_mem);
+out_cleanup:
+	ttm_bo_mem_put(bo, &tmp_mem);
+	return r;
+}
+
+static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo,
+				bool evict, bool interruptible,
+				bool no_wait_gpu,
+				struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	struct ttm_mem_reg tmp_mem;
+	struct ttm_placement placement;
+	struct ttm_place placements;
+	int r;
+
+	adev = amdgpu_get_adev(bo->bdev);
+	tmp_mem = *new_mem;
+	tmp_mem.mm_node = NULL;
+	placement.num_placement = 1;
+	placement.placement = &placements;
+	placement.num_busy_placement = 1;
+	placement.busy_placement = &placements;
+	placements.fpfn = 0;
+	placements.lpfn = 0;
+	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
+	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
+			     interruptible, no_wait_gpu);
+	if (unlikely(r)) {
+		return r;
+	}
+	r = ttm_bo_move_ttm(bo, true, no_wait_gpu, &tmp_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+	r = amdgpu_move_blit(bo, true, no_wait_gpu, new_mem, old_mem);
+	if (unlikely(r)) {
+		goto out_cleanup;
+	}
+out_cleanup:
+	ttm_bo_mem_put(bo, &tmp_mem);
+	return r;
+}
+
+static int amdgpu_bo_move(struct ttm_buffer_object *bo,
+			bool evict, bool interruptible,
+			bool no_wait_gpu,
+			struct ttm_mem_reg *new_mem)
+{
+	struct amdgpu_device *adev;
+	struct ttm_mem_reg *old_mem = &bo->mem;
+	int r;
+
+	adev = amdgpu_get_adev(bo->bdev);
+	if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) {
+		amdgpu_move_null(bo, new_mem);
+		return 0;
+	}
+	if ((old_mem->mem_type == TTM_PL_TT &&
+	     new_mem->mem_type == TTM_PL_SYSTEM) ||
+	    (old_mem->mem_type == TTM_PL_SYSTEM &&
+	     new_mem->mem_type == TTM_PL_TT)) {
+		/* bind is enough */
+		amdgpu_move_null(bo, new_mem);
+		return 0;
+	}
+	if (adev->mman.buffer_funcs == NULL ||
+	    adev->mman.buffer_funcs_ring == NULL ||
+	    !adev->mman.buffer_funcs_ring->ready) {
+		/* use memcpy */
+		goto memcpy;
+	}
+
+	if (old_mem->mem_type == TTM_PL_VRAM &&
+	    new_mem->mem_type == TTM_PL_SYSTEM) {
+		r = amdgpu_move_vram_ram(bo, evict, interruptible,
+					no_wait_gpu, new_mem);
+	} else if (old_mem->mem_type == TTM_PL_SYSTEM &&
+		   new_mem->mem_type == TTM_PL_VRAM) {
+		r = amdgpu_move_ram_vram(bo, evict, interruptible,
+					    no_wait_gpu, new_mem);
+	} else {
+		r = amdgpu_move_blit(bo, evict, no_wait_gpu, new_mem, old_mem);
+	}
+
+	if (r) {
+memcpy:
+		r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem);
+		if (r) {
+			return r;
+		}
+	}
+
+	/* update statistics */
+	atomic64_add((u64)bo->num_pages << PAGE_SHIFT, &adev->num_bytes_moved);
+	return 0;
+}
+
+static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
+{
+	struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type];
+	struct amdgpu_device *adev = amdgpu_get_adev(bdev);
+
+	mem->bus.addr = NULL;
+	mem->bus.offset = 0;
+	mem->bus.size = mem->num_pages << PAGE_SHIFT;
+	mem->bus.base = 0;
+	mem->bus.is_iomem = false;
+	if (!(man->flags & TTM_MEMTYPE_FLAG_MAPPABLE))
+		return -EINVAL;
+	switch (mem->mem_type) {
+	case TTM_PL_SYSTEM:
+		/* system memory */
+		return 0;
+	case TTM_PL_TT:
+		break;
+	case TTM_PL_VRAM:
+		mem->bus.offset = mem->start << PAGE_SHIFT;
+		/* check if it's visible */
+		if ((mem->bus.offset + mem->bus.size) > adev->mc.visible_vram_size)
+			return -EINVAL;
+		mem->bus.base = adev->mc.aper_base;
+		mem->bus.is_iomem = true;
+#ifdef __alpha__
+		/*
+		 * Alpha: use bus.addr to hold the ioremap() return,
+		 * so we can modify bus.base below.
+		 */
+		if (mem->placement & TTM_PL_FLAG_WC)
+			mem->bus.addr =
+				ioremap_wc(mem->bus.base + mem->bus.offset,
+					   mem->bus.size);
+		else
+			mem->bus.addr =
+				ioremap_nocache(mem->bus.base + mem->bus.offset,
+						mem->bus.size);
+
+		/*
+		 * Alpha: Use just the bus offset plus
+		 * the hose/domain memory base for bus.base.
+		 * It then can be used to build PTEs for VRAM
+		 * access, as done in ttm_bo_vm_fault().
+		 */
+		mem->bus.base = (mem->bus.base & 0x0ffffffffUL) +
+			adev->ddev->hose->dense_mem_base;
+#endif
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static void amdgpu_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem)
+{
+}
+
+/*
+ * TTM backend functions.
+ */
+struct amdgpu_ttm_tt {
+	struct ttm_dma_tt		ttm;
+	struct amdgpu_device		*adev;
+	u64				offset;
+	uint64_t			userptr;
+	struct mm_struct		*usermm;
+	uint32_t			userflags;
+};
+
+/* prepare the sg table with the user pages */
+static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_get_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned pinned = 0, nents;
+	int r;
+
+	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+	enum dma_data_direction direction = write ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+
+	if (current->mm != gtt->usermm)
+		return -EPERM;
+
+	if (gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) {
+		/* check that we only pin down anonymous memory
+		   to prevent problems with writeback */
+		unsigned long end = gtt->userptr + ttm->num_pages * PAGE_SIZE;
+		struct vm_area_struct *vma;
+
+		vma = find_vma(gtt->usermm, gtt->userptr);
+		if (!vma || vma->vm_file || vma->vm_end < end)
+			return -EPERM;
+	}
+
+	do {
+		unsigned num_pages = ttm->num_pages - pinned;
+		uint64_t userptr = gtt->userptr + pinned * PAGE_SIZE;
+		struct page **pages = ttm->pages + pinned;
+
+		r = get_user_pages(current, current->mm, userptr, num_pages,
+				   write, 0, pages, NULL);
+		if (r < 0)
+			goto release_pages;
+
+		pinned += r;
+
+	} while (pinned < ttm->num_pages);
+
+	r = sg_alloc_table_from_pages(ttm->sg, ttm->pages, ttm->num_pages, 0,
+				      ttm->num_pages << PAGE_SHIFT,
+				      GFP_KERNEL);
+	if (r)
+		goto release_sg;
+
+	r = -ENOMEM;
+	nents = dma_map_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
+	if (nents != ttm->sg->nents)
+		goto release_sg;
+
+	drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
+					 gtt->ttm.dma_address, ttm->num_pages);
+
+	return 0;
+
+release_sg:
+	kfree(ttm->sg);
+
+release_pages:
+	release_pages(ttm->pages, pinned, 0);
+	return r;
+}
+
+static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_get_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	struct sg_page_iter sg_iter;
+
+	int write = !(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+	enum dma_data_direction direction = write ?
+		DMA_BIDIRECTIONAL : DMA_TO_DEVICE;
+
+	/* double check that we don't free the table twice */
+	if (!ttm->sg->sgl)
+		return;
+
+	/* free the sg table and pages again */
+	dma_unmap_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
+
+	for_each_sg_page(ttm->sg->sgl, &sg_iter, ttm->sg->nents, 0) {
+		struct page *page = sg_page_iter_page(&sg_iter);
+		if (!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY))
+			set_page_dirty(page);
+
+		mark_page_accessed(page);
+		page_cache_release(page);
+	}
+
+	sg_free_table(ttm->sg);
+}
+
+static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
+				   struct ttm_mem_reg *bo_mem)
+{
+	struct amdgpu_ttm_tt *gtt = (void*)ttm;
+	uint32_t flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
+	int r;
+
+	if (gtt->userptr) {
+		r = amdgpu_ttm_tt_pin_userptr(ttm);
+		if (r) {
+			DRM_ERROR("failed to pin userptr\n");
+			return r;
+		}
+	}
+	gtt->offset = (unsigned long)(bo_mem->start << PAGE_SHIFT);
+	if (!ttm->num_pages) {
+		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
+		     ttm->num_pages, bo_mem, ttm);
+	}
+
+	if (bo_mem->mem_type == AMDGPU_PL_GDS ||
+	    bo_mem->mem_type == AMDGPU_PL_GWS ||
+	    bo_mem->mem_type == AMDGPU_PL_OA)
+		return -EINVAL;
+
+	r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
+		ttm->pages, gtt->ttm.dma_address, flags);
+
+	if (r) {
+		DRM_ERROR("failed to bind %lu pages at 0x%08X\n",
+			  ttm->num_pages, (unsigned)gtt->offset);
+		return r;
+	}
+	return 0;
+}
+
+static int amdgpu_ttm_backend_unbind(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	/* unbind shouldn't be done for GDS/GWS/OA in ttm_bo_clean_mm */
+	if (gtt->adev->gart.ready)
+		amdgpu_gart_unbind(gtt->adev, gtt->offset, ttm->num_pages);
+
+	if (gtt->userptr)
+		amdgpu_ttm_tt_unpin_userptr(ttm);
+
+	return 0;
+}
+
+static void amdgpu_ttm_backend_destroy(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	ttm_dma_tt_fini(&gtt->ttm);
+	kfree(gtt);
+}
+
+static struct ttm_backend_func amdgpu_backend_func = {
+	.bind = &amdgpu_ttm_backend_bind,
+	.unbind = &amdgpu_ttm_backend_unbind,
+	.destroy = &amdgpu_ttm_backend_destroy,
+};
+
+static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_bo_device *bdev,
+				    unsigned long size, uint32_t page_flags,
+				    struct page *dummy_read_page)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_ttm_tt *gtt;
+
+	adev = amdgpu_get_adev(bdev);
+
+	gtt = kzalloc(sizeof(struct amdgpu_ttm_tt), GFP_KERNEL);
+	if (gtt == NULL) {
+		return NULL;
+	}
+	gtt->ttm.ttm.func = &amdgpu_backend_func;
+	gtt->adev = adev;
+	if (ttm_dma_tt_init(&gtt->ttm, bdev, size, page_flags, dummy_read_page)) {
+		kfree(gtt);
+		return NULL;
+	}
+	return &gtt->ttm.ttm;
+}
+
+static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned i;
+	int r;
+	bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG);
+
+	if (ttm->state != tt_unpopulated)
+		return 0;
+
+	if (gtt && gtt->userptr) {
+		ttm->sg = kzalloc(sizeof(struct sg_table), GFP_KERNEL);
+		if (!ttm->sg)
+			return -ENOMEM;
+
+		ttm->page_flags |= TTM_PAGE_FLAG_SG;
+		ttm->state = tt_unbound;
+		return 0;
+	}
+
+	if (slave && ttm->sg) {
+		drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
+						 gtt->ttm.dma_address, ttm->num_pages);
+		ttm->state = tt_unbound;
+		return 0;
+	}
+
+	adev = amdgpu_get_adev(ttm->bdev);
+
+#ifdef CONFIG_SWIOTLB
+	if (swiotlb_nr_tbl()) {
+		return ttm_dma_populate(&gtt->ttm, adev->dev);
+	}
+#endif
+
+	r = ttm_pool_populate(ttm);
+	if (r) {
+		return r;
+	}
+
+	for (i = 0; i < ttm->num_pages; i++) {
+		gtt->ttm.dma_address[i] = pci_map_page(adev->pdev, ttm->pages[i],
+						       0, PAGE_SIZE,
+						       PCI_DMA_BIDIRECTIONAL);
+		if (pci_dma_mapping_error(adev->pdev, gtt->ttm.dma_address[i])) {
+			while (i--) {
+				pci_unmap_page(adev->pdev, gtt->ttm.dma_address[i],
+					       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+				gtt->ttm.dma_address[i] = 0;
+			}
+			ttm_pool_unpopulate(ttm);
+			return -EFAULT;
+		}
+	}
+	return 0;
+}
+
+static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev;
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned i;
+	bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG);
+
+	if (gtt && gtt->userptr) {
+		kfree(ttm->sg);
+		ttm->page_flags &= ~TTM_PAGE_FLAG_SG;
+		return;
+	}
+
+	if (slave)
+		return;
+
+	adev = amdgpu_get_adev(ttm->bdev);
+
+#ifdef CONFIG_SWIOTLB
+	if (swiotlb_nr_tbl()) {
+		ttm_dma_unpopulate(&gtt->ttm, adev->dev);
+		return;
+	}
+#endif
+
+	for (i = 0; i < ttm->num_pages; i++) {
+		if (gtt->ttm.dma_address[i]) {
+			pci_unmap_page(adev->pdev, gtt->ttm.dma_address[i],
+				       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
+		}
+	}
+
+	ttm_pool_unpopulate(ttm);
+}
+
+int amdgpu_ttm_tt_set_userptr(struct ttm_tt *ttm, uint64_t addr,
+			      uint32_t flags)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL)
+		return -EINVAL;
+
+	gtt->userptr = addr;
+	gtt->usermm = current->mm;
+	gtt->userflags = flags;
+	return 0;
+}
+
+bool amdgpu_ttm_tt_has_userptr(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL)
+		return false;
+
+	return !!gtt->userptr;
+}
+
+bool amdgpu_ttm_tt_affect_userptr(struct ttm_tt *ttm, unsigned long start,
+				  unsigned long end)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned long size;
+
+	if (gtt == NULL)
+		return false;
+
+	if (gtt->ttm.ttm.state != tt_bound || !gtt->userptr)
+		return false;
+
+	size = (unsigned long)gtt->ttm.ttm.num_pages * PAGE_SIZE;
+	if (gtt->userptr > end || gtt->userptr + size <= start)
+		return false;
+
+	return true;
+}
+
+bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	if (gtt == NULL)
+		return false;
+
+	return !!(gtt->userflags & AMDGPU_GEM_USERPTR_READONLY);
+}
+
+uint32_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
+				 struct ttm_mem_reg *mem)
+{
+	uint32_t flags = 0;
+
+	if (mem && mem->mem_type != TTM_PL_SYSTEM)
+		flags |= AMDGPU_PTE_VALID;
+
+	if (mem && mem->mem_type == TTM_PL_TT) {
+		flags |= AMDGPU_PTE_SYSTEM;
+
+		if (ttm->caching_state == tt_cached)
+			flags |= AMDGPU_PTE_SNOOPED;
+	}
+
+	if (adev->asic_type >= CHIP_TONGA)
+		flags |= AMDGPU_PTE_EXECUTABLE;
+
+	flags |= AMDGPU_PTE_READABLE;
+
+	if (!amdgpu_ttm_tt_is_readonly(ttm))
+		flags |= AMDGPU_PTE_WRITEABLE;
+
+	return flags;
+}
+
+static struct ttm_bo_driver amdgpu_bo_driver = {
+	.ttm_tt_create = &amdgpu_ttm_tt_create,
+	.ttm_tt_populate = &amdgpu_ttm_tt_populate,
+	.ttm_tt_unpopulate = &amdgpu_ttm_tt_unpopulate,
+	.invalidate_caches = &amdgpu_invalidate_caches,
+	.init_mem_type = &amdgpu_init_mem_type,
+	.evict_flags = &amdgpu_evict_flags,
+	.move = &amdgpu_bo_move,
+	.verify_access = &amdgpu_verify_access,
+	.move_notify = &amdgpu_bo_move_notify,
+	.fault_reserve_notify = &amdgpu_bo_fault_reserve_notify,
+	.io_mem_reserve = &amdgpu_ttm_io_mem_reserve,
+	.io_mem_free = &amdgpu_ttm_io_mem_free,
+};
+
+int amdgpu_ttm_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = amdgpu_ttm_global_init(adev);
+	if (r) {
+		return r;
+	}
+	/* No others user of address space so set it to 0 */
+	r = ttm_bo_device_init(&adev->mman.bdev,
+			       adev->mman.bo_global_ref.ref.object,
+			       &amdgpu_bo_driver,
+			       adev->ddev->anon_inode->i_mapping,
+			       DRM_FILE_PAGE_OFFSET,
+			       adev->need_dma32);
+	if (r) {
+		DRM_ERROR("failed initializing buffer object driver(%d).\n", r);
+		return r;
+	}
+	adev->mman.initialized = true;
+	r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_VRAM,
+				adev->mc.real_vram_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing VRAM heap.\n");
+		return r;
+	}
+	/* Change the size here instead of the init above so only lpfn is affected */
+	amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
+
+	r = amdgpu_bo_create(adev, 256 * 1024, PAGE_SIZE, true,
+			     AMDGPU_GEM_DOMAIN_VRAM,
+			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			     NULL, NULL, &adev->stollen_vga_memory);
+	if (r) {
+		return r;
+	}
+	r = amdgpu_bo_reserve(adev->stollen_vga_memory, false);
+	if (r)
+		return r;
+	r = amdgpu_bo_pin(adev->stollen_vga_memory, AMDGPU_GEM_DOMAIN_VRAM, NULL);
+	amdgpu_bo_unreserve(adev->stollen_vga_memory);
+	if (r) {
+		amdgpu_bo_unref(&adev->stollen_vga_memory);
+		return r;
+	}
+	DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
+		 (unsigned) (adev->mc.real_vram_size / (1024 * 1024)));
+	r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_TT,
+				adev->mc.gtt_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing GTT heap.\n");
+		return r;
+	}
+	DRM_INFO("amdgpu: %uM of GTT memory ready.\n",
+		 (unsigned)(adev->mc.gtt_size / (1024 * 1024)));
+
+	adev->gds.mem.total_size = adev->gds.mem.total_size << AMDGPU_GDS_SHIFT;
+	adev->gds.mem.gfx_partition_size = adev->gds.mem.gfx_partition_size << AMDGPU_GDS_SHIFT;
+	adev->gds.mem.cs_partition_size = adev->gds.mem.cs_partition_size << AMDGPU_GDS_SHIFT;
+	adev->gds.gws.total_size = adev->gds.gws.total_size << AMDGPU_GWS_SHIFT;
+	adev->gds.gws.gfx_partition_size = adev->gds.gws.gfx_partition_size << AMDGPU_GWS_SHIFT;
+	adev->gds.gws.cs_partition_size = adev->gds.gws.cs_partition_size << AMDGPU_GWS_SHIFT;
+	adev->gds.oa.total_size = adev->gds.oa.total_size << AMDGPU_OA_SHIFT;
+	adev->gds.oa.gfx_partition_size = adev->gds.oa.gfx_partition_size << AMDGPU_OA_SHIFT;
+	adev->gds.oa.cs_partition_size = adev->gds.oa.cs_partition_size << AMDGPU_OA_SHIFT;
+	/* GDS Memory */
+	r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_GDS,
+				adev->gds.mem.total_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing GDS heap.\n");
+		return r;
+	}
+
+	/* GWS */
+	r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_GWS,
+				adev->gds.gws.total_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing gws heap.\n");
+		return r;
+	}
+
+	/* OA */
+	r = ttm_bo_init_mm(&adev->mman.bdev, AMDGPU_PL_OA,
+				adev->gds.oa.total_size >> PAGE_SHIFT);
+	if (r) {
+		DRM_ERROR("Failed initializing oa heap.\n");
+		return r;
+	}
+
+	r = amdgpu_ttm_debugfs_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init debugfs\n");
+		return r;
+	}
+	return 0;
+}
+
+void amdgpu_ttm_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!adev->mman.initialized)
+		return;
+	amdgpu_ttm_debugfs_fini(adev);
+	if (adev->stollen_vga_memory) {
+		r = amdgpu_bo_reserve(adev->stollen_vga_memory, false);
+		if (r == 0) {
+			amdgpu_bo_unpin(adev->stollen_vga_memory);
+			amdgpu_bo_unreserve(adev->stollen_vga_memory);
+		}
+		amdgpu_bo_unref(&adev->stollen_vga_memory);
+	}
+	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_VRAM);
+	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_TT);
+	ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GDS);
+	ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_GWS);
+	ttm_bo_clean_mm(&adev->mman.bdev, AMDGPU_PL_OA);
+	ttm_bo_device_release(&adev->mman.bdev);
+	amdgpu_gart_fini(adev);
+	amdgpu_ttm_global_fini(adev);
+	adev->mman.initialized = false;
+	DRM_INFO("amdgpu: ttm finalized\n");
+}
+
+/* this should only be called at bootup or when userspace
+ * isn't running */
+void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size)
+{
+	struct ttm_mem_type_manager *man;
+
+	if (!adev->mman.initialized)
+		return;
+
+	man = &adev->mman.bdev.man[TTM_PL_VRAM];
+	/* this just adjusts TTM size idea, which sets lpfn to the correct value */
+	man->size = size >> PAGE_SHIFT;
+}
+
+int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct drm_file *file_priv;
+	struct amdgpu_device *adev;
+
+	if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET))
+		return -EINVAL;
+
+	file_priv = filp->private_data;
+	adev = file_priv->minor->dev->dev_private;
+	if (adev == NULL)
+		return -EINVAL;
+
+	return ttm_bo_mmap(filp, vma, &adev->mman.bdev);
+}
+
+int amdgpu_copy_buffer(struct amdgpu_ring *ring,
+		       uint64_t src_offset,
+		       uint64_t dst_offset,
+		       uint32_t byte_count,
+		       struct reservation_object *resv,
+		       struct fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t max_bytes;
+	unsigned num_loops, num_dw;
+	struct amdgpu_ib *ib;
+	unsigned i;
+	int r;
+
+	max_bytes = adev->mman.buffer_funcs->copy_max_bytes;
+	num_loops = DIV_ROUND_UP(byte_count, max_bytes);
+	num_dw = num_loops * adev->mman.buffer_funcs->copy_num_dw;
+
+	/* for IB padding */
+	while (num_dw & 0x7)
+		num_dw++;
+
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib)
+		return -ENOMEM;
+
+	r = amdgpu_ib_get(ring, NULL, num_dw * 4, ib);
+	if (r) {
+		kfree(ib);
+		return r;
+	}
+
+	ib->length_dw = 0;
+
+	if (resv) {
+		r = amdgpu_sync_resv(adev, &ib->sync, resv,
+				     AMDGPU_FENCE_OWNER_UNDEFINED);
+		if (r) {
+			DRM_ERROR("sync failed (%d).\n", r);
+			goto error_free;
+		}
+	}
+
+	for (i = 0; i < num_loops; i++) {
+		uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
+
+		amdgpu_emit_copy_buffer(adev, ib, src_offset, dst_offset,
+					cur_size_in_bytes);
+
+		src_offset += cur_size_in_bytes;
+		dst_offset += cur_size_in_bytes;
+		byte_count -= cur_size_in_bytes;
+	}
+
+	amdgpu_vm_pad_ib(adev, ib);
+	WARN_ON(ib->length_dw > num_dw);
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+						 &amdgpu_vm_free_job,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 fence);
+	if (r)
+		goto error_free;
+
+	if (!amdgpu_enable_scheduler) {
+		amdgpu_ib_free(adev, ib);
+		kfree(ib);
+	}
+	return 0;
+error_free:
+	amdgpu_ib_free(adev, ib);
+	kfree(ib);
+	return r;
+}
+
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_mm_dump_table(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *)m->private;
+	unsigned ttm_pl = *(int *)node->info_ent->data;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_mm *mm = (struct drm_mm *)adev->mman.bdev.man[ttm_pl].priv;
+	int ret;
+	struct ttm_bo_global *glob = adev->mman.bdev.glob;
+
+	spin_lock(&glob->lru_lock);
+	ret = drm_mm_dump_table(m, mm);
+	spin_unlock(&glob->lru_lock);
+	if (ttm_pl == TTM_PL_VRAM)
+		seq_printf(m, "man size:%llu pages, ram usage:%lluMB, vis usage:%lluMB\n",
+			   adev->mman.bdev.man[ttm_pl].size,
+			   (u64)atomic64_read(&adev->vram_usage) >> 20,
+			   (u64)atomic64_read(&adev->vram_vis_usage) >> 20);
+	return ret;
+}
+
+static int ttm_pl_vram = TTM_PL_VRAM;
+static int ttm_pl_tt = TTM_PL_TT;
+
+static struct drm_info_list amdgpu_ttm_debugfs_list[] = {
+	{"amdgpu_vram_mm", amdgpu_mm_dump_table, 0, &ttm_pl_vram},
+	{"amdgpu_gtt_mm", amdgpu_mm_dump_table, 0, &ttm_pl_tt},
+	{"ttm_page_pool", ttm_page_alloc_debugfs, 0, NULL},
+#ifdef CONFIG_SWIOTLB
+	{"ttm_dma_page_pool", ttm_dma_page_alloc_debugfs, 0, NULL}
+#endif
+};
+
+static ssize_t amdgpu_ttm_vram_read(struct file *f, char __user *buf,
+				    size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	ssize_t result = 0;
+	int r;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	while (size) {
+		unsigned long flags;
+		uint32_t value;
+
+		if (*pos >= adev->mc.mc_vram_size)
+			return result;
+
+		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
+		WREG32(mmMM_INDEX, ((uint32_t)*pos) | 0x80000000);
+		WREG32(mmMM_INDEX_HI, *pos >> 31);
+		value = RREG32(mmMM_DATA);
+		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
+
+		r = put_user(value, (uint32_t *)buf);
+		if (r)
+			return r;
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_ttm_vram_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ttm_vram_read,
+	.llseek = default_llseek
+};
+
+static ssize_t amdgpu_ttm_gtt_read(struct file *f, char __user *buf,
+				   size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	ssize_t result = 0;
+	int r;
+
+	while (size) {
+		loff_t p = *pos / PAGE_SIZE;
+		unsigned off = *pos & ~PAGE_MASK;
+		size_t cur_size = min_t(size_t, size, PAGE_SIZE - off);
+		struct page *page;
+		void *ptr;
+
+		if (p >= adev->gart.num_cpu_pages)
+			return result;
+
+		page = adev->gart.pages[p];
+		if (page) {
+			ptr = kmap(page);
+			ptr += off;
+
+			r = copy_to_user(buf, ptr, cur_size);
+			kunmap(adev->gart.pages[p]);
+		} else
+			r = clear_user(buf, cur_size);
+
+		if (r)
+			return -EFAULT;
+
+		result += cur_size;
+		buf += cur_size;
+		*pos += cur_size;
+		size -= cur_size;
+	}
+
+	return result;
+}
+
+static const struct file_operations amdgpu_ttm_gtt_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_ttm_gtt_read,
+	.llseek = default_llseek
+};
+
+#endif
+
+static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	unsigned count;
+
+	struct drm_minor *minor = adev->ddev->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+
+	ent = debugfs_create_file("amdgpu_vram", S_IFREG | S_IRUGO, root,
+				  adev, &amdgpu_ttm_vram_fops);
+	if (IS_ERR(ent))
+		return PTR_ERR(ent);
+	i_size_write(ent->d_inode, adev->mc.mc_vram_size);
+	adev->mman.vram = ent;
+
+	ent = debugfs_create_file("amdgpu_gtt", S_IFREG | S_IRUGO, root,
+				  adev, &amdgpu_ttm_gtt_fops);
+	if (IS_ERR(ent))
+		return PTR_ERR(ent);
+	i_size_write(ent->d_inode, adev->mc.gtt_size);
+	adev->mman.gtt = ent;
+
+	count = ARRAY_SIZE(amdgpu_ttm_debugfs_list);
+
+#ifdef CONFIG_SWIOTLB
+	if (!swiotlb_nr_tbl())
+		--count;
+#endif
+
+	return amdgpu_debugfs_add_files(adev, amdgpu_ttm_debugfs_list, count);
+#else
+
+	return 0;
+#endif
+}
+
+static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+
+	debugfs_remove(adev->mman.vram);
+	adev->mman.vram = NULL;
+
+	debugfs_remove(adev->mman.gtt);
+	adev->mman.gtt = NULL;
+#endif
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
new file mode 100644
index 000000000..5cc95f1a7
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
@@ -0,0 +1,317 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+
+static void amdgpu_ucode_print_common_hdr(const struct common_firmware_header *hdr)
+{
+	DRM_DEBUG("size_bytes: %u\n", le32_to_cpu(hdr->size_bytes));
+	DRM_DEBUG("header_size_bytes: %u\n", le32_to_cpu(hdr->header_size_bytes));
+	DRM_DEBUG("header_version_major: %u\n", le16_to_cpu(hdr->header_version_major));
+	DRM_DEBUG("header_version_minor: %u\n", le16_to_cpu(hdr->header_version_minor));
+	DRM_DEBUG("ip_version_major: %u\n", le16_to_cpu(hdr->ip_version_major));
+	DRM_DEBUG("ip_version_minor: %u\n", le16_to_cpu(hdr->ip_version_minor));
+	DRM_DEBUG("ucode_version: 0x%08x\n", le32_to_cpu(hdr->ucode_version));
+	DRM_DEBUG("ucode_size_bytes: %u\n", le32_to_cpu(hdr->ucode_size_bytes));
+	DRM_DEBUG("ucode_array_offset_bytes: %u\n",
+		  le32_to_cpu(hdr->ucode_array_offset_bytes));
+	DRM_DEBUG("crc32: 0x%08x\n", le32_to_cpu(hdr->crc32));
+}
+
+void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("MC\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct mc_firmware_header_v1_0 *mc_hdr =
+			container_of(hdr, struct mc_firmware_header_v1_0, header);
+
+		DRM_DEBUG("io_debug_size_bytes: %u\n",
+			  le32_to_cpu(mc_hdr->io_debug_size_bytes));
+		DRM_DEBUG("io_debug_array_offset_bytes: %u\n",
+			  le32_to_cpu(mc_hdr->io_debug_array_offset_bytes));
+	} else {
+		DRM_ERROR("Unknown MC ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("SMC\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct smc_firmware_header_v1_0 *smc_hdr =
+			container_of(hdr, struct smc_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_start_addr: %u\n", le32_to_cpu(smc_hdr->ucode_start_addr));
+	} else {
+		DRM_ERROR("Unknown SMC ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("GFX\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct gfx_firmware_header_v1_0 *gfx_hdr =
+			container_of(hdr, struct gfx_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(gfx_hdr->ucode_feature_version));
+		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(gfx_hdr->jt_offset));
+		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(gfx_hdr->jt_size));
+	} else {
+		DRM_ERROR("Unknown GFX ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("RLC\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct rlc_firmware_header_v1_0 *rlc_hdr =
+			container_of(hdr, struct rlc_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(rlc_hdr->ucode_feature_version));
+		DRM_DEBUG("save_and_restore_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->save_and_restore_offset));
+		DRM_DEBUG("clear_state_descriptor_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
+		DRM_DEBUG("avail_scratch_ram_locations: %u\n",
+			  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
+		DRM_DEBUG("master_pkt_description_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->master_pkt_description_offset));
+	} else if (version_major == 2) {
+		const struct rlc_firmware_header_v2_0 *rlc_hdr =
+			container_of(hdr, struct rlc_firmware_header_v2_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(rlc_hdr->ucode_feature_version));
+		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(rlc_hdr->jt_offset));
+		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(rlc_hdr->jt_size));
+		DRM_DEBUG("save_and_restore_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->save_and_restore_offset));
+		DRM_DEBUG("clear_state_descriptor_offset: %u\n",
+			  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
+		DRM_DEBUG("avail_scratch_ram_locations: %u\n",
+			  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
+		DRM_DEBUG("reg_restore_list_size: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_restore_list_size));
+		DRM_DEBUG("reg_list_format_start: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_start));
+		DRM_DEBUG("reg_list_format_separate_start: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_separate_start));
+		DRM_DEBUG("starting_offsets_start: %u\n",
+			  le32_to_cpu(rlc_hdr->starting_offsets_start));
+		DRM_DEBUG("reg_list_format_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_size_bytes));
+		DRM_DEBUG("reg_list_format_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+		DRM_DEBUG("reg_list_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_size_bytes));
+		DRM_DEBUG("reg_list_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+		DRM_DEBUG("reg_list_format_separate_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_separate_size_bytes));
+		DRM_DEBUG("reg_list_format_separate_array_offset_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_format_separate_array_offset_bytes));
+		DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
+		DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
+			  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
+	} else {
+		DRM_ERROR("Unknown RLC ucode version: %u.%u\n", version_major, version_minor);
+	}
+}
+
+void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr)
+{
+	uint16_t version_major = le16_to_cpu(hdr->header_version_major);
+	uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
+
+	DRM_DEBUG("SDMA\n");
+	amdgpu_ucode_print_common_hdr(hdr);
+
+	if (version_major == 1) {
+		const struct sdma_firmware_header_v1_0 *sdma_hdr =
+			container_of(hdr, struct sdma_firmware_header_v1_0, header);
+
+		DRM_DEBUG("ucode_feature_version: %u\n",
+			  le32_to_cpu(sdma_hdr->ucode_feature_version));
+		DRM_DEBUG("ucode_change_version: %u\n",
+			  le32_to_cpu(sdma_hdr->ucode_change_version));
+		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(sdma_hdr->jt_offset));
+		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(sdma_hdr->jt_size));
+		if (version_minor >= 1) {
+			const struct sdma_firmware_header_v1_1 *sdma_v1_1_hdr =
+				container_of(sdma_hdr, struct sdma_firmware_header_v1_1, v1_0);
+			DRM_DEBUG("digest_size: %u\n", le32_to_cpu(sdma_v1_1_hdr->digest_size));
+		}
+	} else {
+		DRM_ERROR("Unknown SDMA ucode version: %u.%u\n",
+			  version_major, version_minor);
+	}
+}
+
+int amdgpu_ucode_validate(const struct firmware *fw)
+{
+	const struct common_firmware_header *hdr =
+		(const struct common_firmware_header *)fw->data;
+
+	if (fw->size == le32_to_cpu(hdr->size_bytes))
+		return 0;
+
+	return -EINVAL;
+}
+
+bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,
+				uint16_t hdr_major, uint16_t hdr_minor)
+{
+	if ((hdr->common.header_version_major == hdr_major) &&
+		(hdr->common.header_version_minor == hdr_minor))
+		return false;
+	return true;
+}
+
+static int amdgpu_ucode_init_single_fw(struct amdgpu_firmware_info *ucode,
+				uint64_t mc_addr, void *kptr)
+{
+	const struct common_firmware_header *header = NULL;
+
+	if (NULL == ucode->fw)
+		return 0;
+
+	ucode->mc_addr = mc_addr;
+	ucode->kaddr = kptr;
+
+	header = (const struct common_firmware_header *)ucode->fw->data;
+	memcpy(ucode->kaddr, (void *)((uint8_t *)ucode->fw->data +
+		le32_to_cpu(header->ucode_array_offset_bytes)),
+		le32_to_cpu(header->ucode_size_bytes));
+
+	return 0;
+}
+
+int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
+{
+	struct amdgpu_bo **bo = &adev->firmware.fw_buf;
+	uint64_t fw_mc_addr;
+	void *fw_buf_ptr = NULL;
+	uint64_t fw_offset = 0;
+	int i, err;
+	struct amdgpu_firmware_info *ucode = NULL;
+	const struct common_firmware_header *header = NULL;
+
+	err = amdgpu_bo_create(adev, adev->firmware.fw_size, PAGE_SIZE, true,
+			AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL, bo);
+	if (err) {
+		dev_err(adev->dev, "(%d) Firmware buffer allocate failed\n", err);
+		err = -ENOMEM;
+		goto failed;
+	}
+
+	err = amdgpu_bo_reserve(*bo, false);
+	if (err) {
+		amdgpu_bo_unref(bo);
+		dev_err(adev->dev, "(%d) Firmware buffer reserve failed\n", err);
+		goto failed;
+	}
+
+	err = amdgpu_bo_pin(*bo, AMDGPU_GEM_DOMAIN_GTT, &fw_mc_addr);
+	if (err) {
+		amdgpu_bo_unreserve(*bo);
+		amdgpu_bo_unref(bo);
+		dev_err(adev->dev, "(%d) Firmware buffer pin failed\n", err);
+		goto failed;
+	}
+
+	err = amdgpu_bo_kmap(*bo, &fw_buf_ptr);
+	if (err) {
+		dev_err(adev->dev, "(%d) Firmware buffer kmap failed\n", err);
+		amdgpu_bo_unpin(*bo);
+		amdgpu_bo_unreserve(*bo);
+		amdgpu_bo_unref(bo);
+		goto failed;
+	}
+
+	amdgpu_bo_unreserve(*bo);
+
+	fw_offset = 0;
+	for (i = 0; i < AMDGPU_UCODE_ID_MAXIMUM; i++) {
+		ucode = &adev->firmware.ucode[i];
+		if (ucode->fw) {
+			header = (const struct common_firmware_header *)ucode->fw->data;
+			amdgpu_ucode_init_single_fw(ucode, fw_mc_addr + fw_offset,
+						    fw_buf_ptr + fw_offset);
+			fw_offset += ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+
+failed:
+	if (err)
+		adev->firmware.smu_load = false;
+
+	return err;
+}
+
+int amdgpu_ucode_fini_bo(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_firmware_info *ucode = NULL;
+
+	for (i = 0; i < AMDGPU_UCODE_ID_MAXIMUM; i++) {
+		ucode = &adev->firmware.ucode[i];
+		if (ucode->fw) {
+			ucode->mc_addr = 0;
+			ucode->kaddr = NULL;
+		}
+	}
+	amdgpu_bo_unref(&adev->firmware.fw_buf);
+	adev->firmware.fw_buf = NULL;
+
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
new file mode 100644
index 000000000..e468be4e2
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
@@ -0,0 +1,176 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __AMDGPU_UCODE_H__
+#define __AMDGPU_UCODE_H__
+
+struct common_firmware_header {
+	uint32_t size_bytes; /* size of the entire header+image(s) in bytes */
+	uint32_t header_size_bytes; /* size of just the header in bytes */
+	uint16_t header_version_major; /* header version */
+	uint16_t header_version_minor; /* header version */
+	uint16_t ip_version_major; /* IP version */
+	uint16_t ip_version_minor; /* IP version */
+	uint32_t ucode_version;
+	uint32_t ucode_size_bytes; /* size of ucode in bytes */
+	uint32_t ucode_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t crc32;  /* crc32 checksum of the payload */
+};
+
+/* version_major=1, version_minor=0 */
+struct mc_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t io_debug_size_bytes; /* size of debug array in dwords */
+	uint32_t io_debug_array_offset_bytes; /* payload offset from the start of the header */
+};
+
+/* version_major=1, version_minor=0 */
+struct smc_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_start_addr;
+};
+
+/* version_major=1, version_minor=0 */
+struct gfx_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t jt_offset; /* jt location */
+	uint32_t jt_size;  /* size of jt */
+};
+
+/* version_major=1, version_minor=0 */
+struct rlc_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t save_and_restore_offset;
+	uint32_t clear_state_descriptor_offset;
+	uint32_t avail_scratch_ram_locations;
+	uint32_t master_pkt_description_offset;
+};
+
+/* version_major=2, version_minor=0 */
+struct rlc_firmware_header_v2_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t jt_offset; /* jt location */
+	uint32_t jt_size;  /* size of jt */
+	uint32_t save_and_restore_offset;
+	uint32_t clear_state_descriptor_offset;
+	uint32_t avail_scratch_ram_locations;
+	uint32_t reg_restore_list_size;
+	uint32_t reg_list_format_start;
+	uint32_t reg_list_format_separate_start;
+	uint32_t starting_offsets_start;
+	uint32_t reg_list_format_size_bytes; /* size of reg list format array in bytes */
+	uint32_t reg_list_format_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t reg_list_size_bytes; /* size of reg list array in bytes */
+	uint32_t reg_list_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t reg_list_format_separate_size_bytes; /* size of reg list format array in bytes */
+	uint32_t reg_list_format_separate_array_offset_bytes; /* payload offset from the start of the header */
+	uint32_t reg_list_separate_size_bytes; /* size of reg list array in bytes */
+	uint32_t reg_list_separate_array_offset_bytes; /* payload offset from the start of the header */
+};
+
+/* version_major=1, version_minor=0 */
+struct sdma_firmware_header_v1_0 {
+	struct common_firmware_header header;
+	uint32_t ucode_feature_version;
+	uint32_t ucode_change_version;
+	uint32_t jt_offset; /* jt location */
+	uint32_t jt_size; /* size of jt */
+};
+
+/* version_major=1, version_minor=1 */
+struct sdma_firmware_header_v1_1 {
+	struct sdma_firmware_header_v1_0 v1_0;
+	uint32_t digest_size;
+};
+
+/* header is fixed size */
+union amdgpu_firmware_header {
+	struct common_firmware_header common;
+	struct mc_firmware_header_v1_0 mc;
+	struct smc_firmware_header_v1_0 smc;
+	struct gfx_firmware_header_v1_0 gfx;
+	struct rlc_firmware_header_v1_0 rlc;
+	struct rlc_firmware_header_v2_0 rlc_v2_0;
+	struct sdma_firmware_header_v1_0 sdma;
+	struct sdma_firmware_header_v1_1 sdma_v1_1;
+	uint8_t raw[0x100];
+};
+
+/*
+ * fw loading support
+ */
+enum AMDGPU_UCODE_ID {
+	AMDGPU_UCODE_ID_SDMA0 = 0,
+	AMDGPU_UCODE_ID_SDMA1,
+	AMDGPU_UCODE_ID_CP_CE,
+	AMDGPU_UCODE_ID_CP_PFP,
+	AMDGPU_UCODE_ID_CP_ME,
+	AMDGPU_UCODE_ID_CP_MEC1,
+	AMDGPU_UCODE_ID_CP_MEC2,
+	AMDGPU_UCODE_ID_RLC_G,
+	AMDGPU_UCODE_ID_MAXIMUM,
+};
+
+/* engine firmware status */
+enum AMDGPU_UCODE_STATUS {
+	AMDGPU_UCODE_STATUS_INVALID,
+	AMDGPU_UCODE_STATUS_NOT_LOADED,
+	AMDGPU_UCODE_STATUS_LOADED,
+};
+
+/* conform to smu_ucode_xfer_cz.h */
+#define AMDGPU_SDMA0_UCODE_LOADED	0x00000001
+#define AMDGPU_SDMA1_UCODE_LOADED	0x00000002
+#define AMDGPU_CPCE_UCODE_LOADED	0x00000004
+#define AMDGPU_CPPFP_UCODE_LOADED	0x00000008
+#define AMDGPU_CPME_UCODE_LOADED	0x00000010
+#define AMDGPU_CPMEC1_UCODE_LOADED	0x00000020
+#define AMDGPU_CPMEC2_UCODE_LOADED	0x00000040
+#define AMDGPU_CPRLC_UCODE_LOADED	0x00000100
+
+/* amdgpu firmware info */
+struct amdgpu_firmware_info {
+	/* ucode ID */
+	enum AMDGPU_UCODE_ID ucode_id;
+	/* request_firmware */
+	const struct firmware *fw;
+	/* starting mc address */
+	uint64_t mc_addr;
+	/* kernel linear address */
+	void *kaddr;
+};
+
+void amdgpu_ucode_print_mc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr);
+void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr);
+int amdgpu_ucode_validate(const struct firmware *fw);
+bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,
+				uint16_t hdr_major, uint16_t hdr_minor);
+int amdgpu_ucode_init_bo(struct amdgpu_device *adev);
+int amdgpu_ucode_fini_bo(struct amdgpu_device *adev);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
new file mode 100644
index 000000000..53f987aee
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -0,0 +1,1044 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_uvd.h"
+#include "cikd.h"
+#include "uvd/uvd_4_2_d.h"
+
+/* 1 second timeout */
+#define UVD_IDLE_TIMEOUT_MS	1000
+
+/* Firmware Names */
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#define FIRMWARE_BONAIRE	"radeon/bonaire_uvd.bin"
+#define FIRMWARE_KABINI 	"radeon/kabini_uvd.bin"
+#define FIRMWARE_KAVERI 	"radeon/kaveri_uvd.bin"
+#define FIRMWARE_HAWAII 	"radeon/hawaii_uvd.bin"
+#define FIRMWARE_MULLINS	"radeon/mullins_uvd.bin"
+#endif
+#define FIRMWARE_TONGA		"amdgpu/tonga_uvd.bin"
+#define FIRMWARE_CARRIZO	"amdgpu/carrizo_uvd.bin"
+#define FIRMWARE_FIJI		"amdgpu/fiji_uvd.bin"
+#define FIRMWARE_STONEY		"amdgpu/stoney_uvd.bin"
+
+/**
+ * amdgpu_uvd_cs_ctx - Command submission parser context
+ *
+ * Used for emulating virtual memory support on UVD 4.2.
+ */
+struct amdgpu_uvd_cs_ctx {
+	struct amdgpu_cs_parser *parser;
+	unsigned reg, count;
+	unsigned data0, data1;
+	unsigned idx;
+	unsigned ib_idx;
+
+	/* does the IB has a msg command */
+	bool has_msg_cmd;
+
+	/* minimum buffer sizes */
+	unsigned *buf_sizes;
+};
+
+#ifdef CONFIG_DRM_AMDGPU_CIK
+MODULE_FIRMWARE(FIRMWARE_BONAIRE);
+MODULE_FIRMWARE(FIRMWARE_KABINI);
+MODULE_FIRMWARE(FIRMWARE_KAVERI);
+MODULE_FIRMWARE(FIRMWARE_HAWAII);
+MODULE_FIRMWARE(FIRMWARE_MULLINS);
+#endif
+MODULE_FIRMWARE(FIRMWARE_TONGA);
+MODULE_FIRMWARE(FIRMWARE_CARRIZO);
+MODULE_FIRMWARE(FIRMWARE_FIJI);
+MODULE_FIRMWARE(FIRMWARE_STONEY);
+
+static void amdgpu_uvd_note_usage(struct amdgpu_device *adev);
+static void amdgpu_uvd_idle_work_handler(struct work_struct *work);
+
+int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
+{
+	unsigned long bo_size;
+	const char *fw_name;
+	const struct common_firmware_header *hdr;
+	unsigned version_major, version_minor, family_id;
+	int i, r;
+
+	INIT_DELAYED_WORK(&adev->uvd.idle_work, amdgpu_uvd_idle_work_handler);
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+		fw_name = FIRMWARE_BONAIRE;
+		break;
+	case CHIP_KABINI:
+		fw_name = FIRMWARE_KABINI;
+		break;
+	case CHIP_KAVERI:
+		fw_name = FIRMWARE_KAVERI;
+		break;
+	case CHIP_HAWAII:
+		fw_name = FIRMWARE_HAWAII;
+		break;
+	case CHIP_MULLINS:
+		fw_name = FIRMWARE_MULLINS;
+		break;
+#endif
+	case CHIP_TONGA:
+		fw_name = FIRMWARE_TONGA;
+		break;
+	case CHIP_FIJI:
+		fw_name = FIRMWARE_FIJI;
+		break;
+	case CHIP_CARRIZO:
+		fw_name = FIRMWARE_CARRIZO;
+		break;
+	case CHIP_STONEY:
+		fw_name = FIRMWARE_STONEY;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&adev->uvd.fw, fw_name, adev->dev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_uvd: Can't load firmware \"%s\"\n",
+			fw_name);
+		return r;
+	}
+
+	r = amdgpu_ucode_validate(adev->uvd.fw);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_uvd: Can't validate firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->uvd.fw);
+		adev->uvd.fw = NULL;
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
+	family_id = le32_to_cpu(hdr->ucode_version) & 0xff;
+	version_major = (le32_to_cpu(hdr->ucode_version) >> 24) & 0xff;
+	version_minor = (le32_to_cpu(hdr->ucode_version) >> 8) & 0xff;
+	DRM_INFO("Found UVD firmware Version: %hu.%hu Family ID: %hu\n",
+		version_major, version_minor, family_id);
+
+	bo_size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8)
+		 +  AMDGPU_UVD_STACK_SIZE + AMDGPU_UVD_HEAP_SIZE;
+	r = amdgpu_bo_create(adev, bo_size, PAGE_SIZE, true,
+			     AMDGPU_GEM_DOMAIN_VRAM,
+			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			     NULL, NULL, &adev->uvd.vcpu_bo);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_reserve(adev->uvd.vcpu_bo, false);
+	if (r) {
+		amdgpu_bo_unref(&adev->uvd.vcpu_bo);
+		dev_err(adev->dev, "(%d) failed to reserve UVD bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_pin(adev->uvd.vcpu_bo, AMDGPU_GEM_DOMAIN_VRAM,
+			  &adev->uvd.gpu_addr);
+	if (r) {
+		amdgpu_bo_unreserve(adev->uvd.vcpu_bo);
+		amdgpu_bo_unref(&adev->uvd.vcpu_bo);
+		dev_err(adev->dev, "(%d) UVD bo pin failed\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(adev->uvd.vcpu_bo, &adev->uvd.cpu_addr);
+	if (r) {
+		dev_err(adev->dev, "(%d) UVD map failed\n", r);
+		return r;
+	}
+
+	amdgpu_bo_unreserve(adev->uvd.vcpu_bo);
+
+	for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i) {
+		atomic_set(&adev->uvd.handles[i], 0);
+		adev->uvd.filp[i] = NULL;
+	}
+
+	/* from uvd v5.0 HW addressing capacity increased to 64 bits */
+	if (!amdgpu_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0))
+		adev->uvd.address_64_bit = true;
+
+	return 0;
+}
+
+int amdgpu_uvd_sw_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->uvd.vcpu_bo == NULL)
+		return 0;
+
+	r = amdgpu_bo_reserve(adev->uvd.vcpu_bo, false);
+	if (!r) {
+		amdgpu_bo_kunmap(adev->uvd.vcpu_bo);
+		amdgpu_bo_unpin(adev->uvd.vcpu_bo);
+		amdgpu_bo_unreserve(adev->uvd.vcpu_bo);
+	}
+
+	amdgpu_bo_unref(&adev->uvd.vcpu_bo);
+
+	amdgpu_ring_fini(&adev->uvd.ring);
+
+	release_firmware(adev->uvd.fw);
+
+	return 0;
+}
+
+int amdgpu_uvd_suspend(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	int i, r;
+
+	if (adev->uvd.vcpu_bo == NULL)
+		return 0;
+
+	for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i) {
+		uint32_t handle = atomic_read(&adev->uvd.handles[i]);
+		if (handle != 0) {
+			struct fence *fence;
+
+			amdgpu_uvd_note_usage(adev);
+
+			r = amdgpu_uvd_get_destroy_msg(ring, handle, &fence);
+			if (r) {
+				DRM_ERROR("Error destroying UVD (%d)!\n", r);
+				continue;
+			}
+
+			fence_wait(fence, false);
+			fence_put(fence);
+
+			adev->uvd.filp[i] = NULL;
+			atomic_set(&adev->uvd.handles[i], 0);
+		}
+	}
+
+	return 0;
+}
+
+int amdgpu_uvd_resume(struct amdgpu_device *adev)
+{
+	unsigned size;
+	void *ptr;
+	const struct common_firmware_header *hdr;
+	unsigned offset;
+
+	if (adev->uvd.vcpu_bo == NULL)
+		return -EINVAL;
+
+	hdr = (const struct common_firmware_header *)adev->uvd.fw->data;
+	offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+	memcpy(adev->uvd.cpu_addr, (adev->uvd.fw->data) + offset,
+		(adev->uvd.fw->size) - offset);
+
+	size = amdgpu_bo_size(adev->uvd.vcpu_bo);
+	size -= le32_to_cpu(hdr->ucode_size_bytes);
+	ptr = adev->uvd.cpu_addr;
+	ptr += le32_to_cpu(hdr->ucode_size_bytes);
+
+	memset(ptr, 0, size);
+
+	return 0;
+}
+
+void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
+{
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	int i, r;
+
+	for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i) {
+		uint32_t handle = atomic_read(&adev->uvd.handles[i]);
+		if (handle != 0 && adev->uvd.filp[i] == filp) {
+			struct fence *fence;
+
+			amdgpu_uvd_note_usage(adev);
+
+			r = amdgpu_uvd_get_destroy_msg(ring, handle, &fence);
+			if (r) {
+				DRM_ERROR("Error destroying UVD (%d)!\n", r);
+				continue;
+			}
+
+			fence_wait(fence, false);
+			fence_put(fence);
+
+			adev->uvd.filp[i] = NULL;
+			atomic_set(&adev->uvd.handles[i], 0);
+		}
+	}
+}
+
+static void amdgpu_uvd_force_into_uvd_segment(struct amdgpu_bo *rbo)
+{
+	int i;
+	for (i = 0; i < rbo->placement.num_placement; ++i) {
+		rbo->placements[i].fpfn = 0 >> PAGE_SHIFT;
+		rbo->placements[i].lpfn = (256 * 1024 * 1024) >> PAGE_SHIFT;
+	}
+}
+
+/**
+ * amdgpu_uvd_cs_pass1 - first parsing round
+ *
+ * @ctx: UVD parser context
+ *
+ * Make sure UVD message and feedback buffers are in VRAM and
+ * nobody is violating an 256MB boundary.
+ */
+static int amdgpu_uvd_cs_pass1(struct amdgpu_uvd_cs_ctx *ctx)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	uint32_t cmd, lo, hi;
+	uint64_t addr;
+	int r = 0;
+
+	lo = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data0);
+	hi = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data1);
+	addr = ((uint64_t)lo) | (((uint64_t)hi) << 32);
+
+	mapping = amdgpu_cs_find_mapping(ctx->parser, addr, &bo);
+	if (mapping == NULL) {
+		DRM_ERROR("Can't find BO for addr 0x%08Lx\n", addr);
+		return -EINVAL;
+	}
+
+	if (!ctx->parser->adev->uvd.address_64_bit) {
+		/* check if it's a message or feedback command */
+		cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1;
+		if (cmd == 0x0 || cmd == 0x3) {
+			/* yes, force it into VRAM */
+			uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM;
+			amdgpu_ttm_placement_from_domain(bo, domain);
+		}
+		amdgpu_uvd_force_into_uvd_segment(bo);
+
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false);
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_uvd_cs_msg_decode - handle UVD decode message
+ *
+ * @msg: pointer to message structure
+ * @buf_sizes: returned buffer sizes
+ *
+ * Peek into the decode message and calculate the necessary buffer sizes.
+ */
+static int amdgpu_uvd_cs_msg_decode(uint32_t *msg, unsigned buf_sizes[])
+{
+	unsigned stream_type = msg[4];
+	unsigned width = msg[6];
+	unsigned height = msg[7];
+	unsigned dpb_size = msg[9];
+	unsigned pitch = msg[28];
+	unsigned level = msg[57];
+
+	unsigned width_in_mb = width / 16;
+	unsigned height_in_mb = ALIGN(height / 16, 2);
+	unsigned fs_in_mb = width_in_mb * height_in_mb;
+
+	unsigned image_size, tmp, min_dpb_size, num_dpb_buffer;
+	unsigned min_ctx_size = 0;
+
+	image_size = width * height;
+	image_size += image_size / 2;
+	image_size = ALIGN(image_size, 1024);
+
+	switch (stream_type) {
+	case 0: /* H264 */
+	case 7: /* H264 Perf */
+		switch(level) {
+		case 30:
+			num_dpb_buffer = 8100 / fs_in_mb;
+			break;
+		case 31:
+			num_dpb_buffer = 18000 / fs_in_mb;
+			break;
+		case 32:
+			num_dpb_buffer = 20480 / fs_in_mb;
+			break;
+		case 41:
+			num_dpb_buffer = 32768 / fs_in_mb;
+			break;
+		case 42:
+			num_dpb_buffer = 34816 / fs_in_mb;
+			break;
+		case 50:
+			num_dpb_buffer = 110400 / fs_in_mb;
+			break;
+		case 51:
+			num_dpb_buffer = 184320 / fs_in_mb;
+			break;
+		default:
+			num_dpb_buffer = 184320 / fs_in_mb;
+			break;
+		}
+		num_dpb_buffer++;
+		if (num_dpb_buffer > 17)
+			num_dpb_buffer = 17;
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * num_dpb_buffer;
+
+		/* macroblock context buffer */
+		min_dpb_size += width_in_mb * height_in_mb * num_dpb_buffer * 192;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * height_in_mb * 32;
+		break;
+
+	case 1: /* VC1 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CONTEXT_BUFFER */
+		min_dpb_size += width_in_mb * height_in_mb * 128;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * 64;
+
+		/* DB surface buffer */
+		min_dpb_size += width_in_mb * 128;
+
+		/* BP */
+		tmp = max(width_in_mb, height_in_mb);
+		min_dpb_size += ALIGN(tmp * 7 * 16, 64);
+		break;
+
+	case 3: /* MPEG2 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+		break;
+
+	case 4: /* MPEG4 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CM */
+		min_dpb_size += width_in_mb * height_in_mb * 64;
+
+		/* IT surface buffer */
+		min_dpb_size += ALIGN(width_in_mb * height_in_mb * 32, 64);
+		break;
+
+	case 16: /* H265 */
+		image_size = (ALIGN(width, 16) * ALIGN(height, 16) * 3) / 2;
+		image_size = ALIGN(image_size, 256);
+
+		num_dpb_buffer = (le32_to_cpu(msg[59]) & 0xff) + 2;
+		min_dpb_size = image_size * num_dpb_buffer;
+		min_ctx_size = ((width + 255) / 16) * ((height + 255) / 16)
+					   * 16 * num_dpb_buffer + 52 * 1024;
+		break;
+
+	default:
+		DRM_ERROR("UVD codec not handled %d!\n", stream_type);
+		return -EINVAL;
+	}
+
+	if (width > pitch) {
+		DRM_ERROR("Invalid UVD decoding target pitch!\n");
+		return -EINVAL;
+	}
+
+	if (dpb_size < min_dpb_size) {
+		DRM_ERROR("Invalid dpb_size in UVD message (%d / %d)!\n",
+			  dpb_size, min_dpb_size);
+		return -EINVAL;
+	}
+
+	buf_sizes[0x1] = dpb_size;
+	buf_sizes[0x2] = image_size;
+	buf_sizes[0x4] = min_ctx_size;
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_cs_msg - handle UVD message
+ *
+ * @ctx: UVD parser context
+ * @bo: buffer object containing the message
+ * @offset: offset into the buffer object
+ *
+ * Peek into the UVD message and extract the session id.
+ * Make sure that we don't open up to many sessions.
+ */
+static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
+			     struct amdgpu_bo *bo, unsigned offset)
+{
+	struct amdgpu_device *adev = ctx->parser->adev;
+	int32_t *msg, msg_type, handle;
+	void *ptr;
+	long r;
+	int i;
+
+	if (offset & 0x3F) {
+		DRM_ERROR("UVD messages must be 64 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	r = reservation_object_wait_timeout_rcu(bo->tbo.resv, true, false,
+						MAX_SCHEDULE_TIMEOUT);
+	if (r < 0) {
+		DRM_ERROR("Failed waiting for UVD message (%ld)!\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, &ptr);
+	if (r) {
+		DRM_ERROR("Failed mapping the UVD message (%ld)!\n", r);
+		return r;
+	}
+
+	msg = ptr + offset;
+
+	msg_type = msg[1];
+	handle = msg[2];
+
+	if (handle == 0) {
+		DRM_ERROR("Invalid UVD handle!\n");
+		return -EINVAL;
+	}
+
+	switch (msg_type) {
+	case 0:
+		/* it's a create msg, calc image size (width * height) */
+		amdgpu_bo_kunmap(bo);
+
+		/* try to alloc a new handle */
+		for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i) {
+			if (atomic_read(&adev->uvd.handles[i]) == handle) {
+				DRM_ERROR("Handle 0x%x already in use!\n", handle);
+				return -EINVAL;
+			}
+
+			if (!atomic_cmpxchg(&adev->uvd.handles[i], 0, handle)) {
+				adev->uvd.filp[i] = ctx->parser->filp;
+				return 0;
+			}
+		}
+
+		DRM_ERROR("No more free UVD handles!\n");
+		return -EINVAL;
+
+	case 1:
+		/* it's a decode msg, calc buffer sizes */
+		r = amdgpu_uvd_cs_msg_decode(msg, ctx->buf_sizes);
+		amdgpu_bo_kunmap(bo);
+		if (r)
+			return r;
+
+		/* validate the handle */
+		for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i) {
+			if (atomic_read(&adev->uvd.handles[i]) == handle) {
+				if (adev->uvd.filp[i] != ctx->parser->filp) {
+					DRM_ERROR("UVD handle collision detected!\n");
+					return -EINVAL;
+				}
+				return 0;
+			}
+		}
+
+		DRM_ERROR("Invalid UVD handle 0x%x!\n", handle);
+		return -ENOENT;
+
+	case 2:
+		/* it's a destroy msg, free the handle */
+		for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i)
+			atomic_cmpxchg(&adev->uvd.handles[i], handle, 0);
+		amdgpu_bo_kunmap(bo);
+		return 0;
+
+	default:
+		DRM_ERROR("Illegal UVD message type (%d)!\n", msg_type);
+		return -EINVAL;
+	}
+	BUG();
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_uvd_cs_pass2 - second parsing round
+ *
+ * @ctx: UVD parser context
+ *
+ * Patch buffer addresses, make sure buffer sizes are correct.
+ */
+static int amdgpu_uvd_cs_pass2(struct amdgpu_uvd_cs_ctx *ctx)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_bo *bo;
+	struct amdgpu_ib *ib;
+	uint32_t cmd, lo, hi;
+	uint64_t start, end;
+	uint64_t addr;
+	int r;
+
+	lo = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data0);
+	hi = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->data1);
+	addr = ((uint64_t)lo) | (((uint64_t)hi) << 32);
+
+	mapping = amdgpu_cs_find_mapping(ctx->parser, addr, &bo);
+	if (mapping == NULL)
+		return -EINVAL;
+
+	start = amdgpu_bo_gpu_offset(bo);
+
+	end = (mapping->it.last + 1 - mapping->it.start);
+	end = end * AMDGPU_GPU_PAGE_SIZE + start;
+
+	addr -= ((uint64_t)mapping->it.start) * AMDGPU_GPU_PAGE_SIZE;
+	start += addr;
+
+	ib = &ctx->parser->ibs[ctx->ib_idx];
+	ib->ptr[ctx->data0] = start & 0xFFFFFFFF;
+	ib->ptr[ctx->data1] = start >> 32;
+
+	cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx) >> 1;
+	if (cmd < 0x4) {
+		if ((end - start) < ctx->buf_sizes[cmd]) {
+			DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd,
+				  (unsigned)(end - start),
+				  ctx->buf_sizes[cmd]);
+			return -EINVAL;
+		}
+
+	} else if (cmd == 0x206) {
+		if ((end - start) < ctx->buf_sizes[4]) {
+			DRM_ERROR("buffer (%d) to small (%d / %d)!\n", cmd,
+					  (unsigned)(end - start),
+					  ctx->buf_sizes[4]);
+			return -EINVAL;
+		}
+	} else if ((cmd != 0x100) && (cmd != 0x204)) {
+		DRM_ERROR("invalid UVD command %X!\n", cmd);
+		return -EINVAL;
+	}
+
+	if (!ctx->parser->adev->uvd.address_64_bit) {
+		if ((start >> 28) != ((end - 1) >> 28)) {
+			DRM_ERROR("reloc %LX-%LX crossing 256MB boundary!\n",
+				  start, end);
+			return -EINVAL;
+		}
+
+		if ((cmd == 0 || cmd == 0x3) &&
+		    (start >> 28) != (ctx->parser->adev->uvd.gpu_addr >> 28)) {
+			DRM_ERROR("msg/fb buffer %LX-%LX out of 256MB segment!\n",
+				  start, end);
+			return -EINVAL;
+		}
+	}
+
+	if (cmd == 0) {
+		ctx->has_msg_cmd = true;
+		r = amdgpu_uvd_cs_msg(ctx, bo, addr);
+		if (r)
+			return r;
+	} else if (!ctx->has_msg_cmd) {
+		DRM_ERROR("Message needed before other commands are send!\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_cs_reg - parse register writes
+ *
+ * @ctx: UVD parser context
+ * @cb: callback function
+ *
+ * Parse the register writes, call cb on each complete command.
+ */
+static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx,
+			     int (*cb)(struct amdgpu_uvd_cs_ctx *ctx))
+{
+	struct amdgpu_ib *ib = &ctx->parser->ibs[ctx->ib_idx];
+	int i, r;
+
+	ctx->idx++;
+	for (i = 0; i <= ctx->count; ++i) {
+		unsigned reg = ctx->reg + i;
+
+		if (ctx->idx >= ib->length_dw) {
+			DRM_ERROR("Register command after end of CS!\n");
+			return -EINVAL;
+		}
+
+		switch (reg) {
+		case mmUVD_GPCOM_VCPU_DATA0:
+			ctx->data0 = ctx->idx;
+			break;
+		case mmUVD_GPCOM_VCPU_DATA1:
+			ctx->data1 = ctx->idx;
+			break;
+		case mmUVD_GPCOM_VCPU_CMD:
+			r = cb(ctx);
+			if (r)
+				return r;
+			break;
+		case mmUVD_ENGINE_CNTL:
+			break;
+		default:
+			DRM_ERROR("Invalid reg 0x%X!\n", reg);
+			return -EINVAL;
+		}
+		ctx->idx++;
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_cs_packets - parse UVD packets
+ *
+ * @ctx: UVD parser context
+ * @cb: callback function
+ *
+ * Parse the command stream packets.
+ */
+static int amdgpu_uvd_cs_packets(struct amdgpu_uvd_cs_ctx *ctx,
+				 int (*cb)(struct amdgpu_uvd_cs_ctx *ctx))
+{
+	struct amdgpu_ib *ib = &ctx->parser->ibs[ctx->ib_idx];
+	int r;
+
+	for (ctx->idx = 0 ; ctx->idx < ib->length_dw; ) {
+		uint32_t cmd = amdgpu_get_ib_value(ctx->parser, ctx->ib_idx, ctx->idx);
+		unsigned type = CP_PACKET_GET_TYPE(cmd);
+		switch (type) {
+		case PACKET_TYPE0:
+			ctx->reg = CP_PACKET0_GET_REG(cmd);
+			ctx->count = CP_PACKET_GET_COUNT(cmd);
+			r = amdgpu_uvd_cs_reg(ctx, cb);
+			if (r)
+				return r;
+			break;
+		case PACKET_TYPE2:
+			++ctx->idx;
+			break;
+		default:
+			DRM_ERROR("Unknown packet type %d !\n", type);
+			return -EINVAL;
+		}
+	}
+	return 0;
+}
+
+/**
+ * amdgpu_uvd_ring_parse_cs - UVD command submission parser
+ *
+ * @parser: Command submission parser context
+ *
+ * Parse the command stream, patch in addresses as necessary.
+ */
+int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
+{
+	struct amdgpu_uvd_cs_ctx ctx = {};
+	unsigned buf_sizes[] = {
+		[0x00000000]	=	2048,
+		[0x00000001]	=	0xFFFFFFFF,
+		[0x00000002]	=	0xFFFFFFFF,
+		[0x00000003]	=	2048,
+		[0x00000004]	=	0xFFFFFFFF,
+	};
+	struct amdgpu_ib *ib = &parser->ibs[ib_idx];
+	int r;
+
+	if (ib->length_dw % 16) {
+		DRM_ERROR("UVD IB length (%d) not 16 dwords aligned!\n",
+			  ib->length_dw);
+		return -EINVAL;
+	}
+
+	ctx.parser = parser;
+	ctx.buf_sizes = buf_sizes;
+	ctx.ib_idx = ib_idx;
+
+	/* first round, make sure the buffers are actually in the UVD segment */
+	r = amdgpu_uvd_cs_packets(&ctx, amdgpu_uvd_cs_pass1);
+	if (r)
+		return r;
+
+	/* second round, patch buffer addresses into the command stream */
+	r = amdgpu_uvd_cs_packets(&ctx, amdgpu_uvd_cs_pass2);
+	if (r)
+		return r;
+
+	if (!ctx.has_msg_cmd) {
+		DRM_ERROR("UVD-IBs need a msg command!\n");
+		return -EINVAL;
+	}
+
+	amdgpu_uvd_note_usage(ctx.parser->adev);
+
+	return 0;
+}
+
+static int amdgpu_uvd_free_job(
+	struct amdgpu_job *job)
+{
+	amdgpu_ib_free(job->adev, job->ibs);
+	kfree(job->ibs);
+	return 0;
+}
+
+static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring,
+			       struct amdgpu_bo *bo,
+			       struct fence **fence)
+{
+	struct ttm_validate_buffer tv;
+	struct ww_acquire_ctx ticket;
+	struct list_head head;
+	struct amdgpu_ib *ib = NULL;
+	struct fence *f = NULL;
+	struct amdgpu_device *adev = ring->adev;
+	uint64_t addr;
+	int i, r;
+
+	memset(&tv, 0, sizeof(tv));
+	tv.bo = &bo->tbo;
+
+	INIT_LIST_HEAD(&head);
+	list_add(&tv.head, &head);
+
+	r = ttm_eu_reserve_buffers(&ticket, &head, true, NULL);
+	if (r)
+		return r;
+
+	if (!bo->adev->uvd.address_64_bit) {
+		amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM);
+		amdgpu_uvd_force_into_uvd_segment(bo);
+	}
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+	if (r)
+		goto err;
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib) {
+		r = -ENOMEM;
+		goto err;
+	}
+	r = amdgpu_ib_get(ring, NULL, 64, ib);
+	if (r)
+		goto err1;
+
+	addr = amdgpu_bo_gpu_offset(bo);
+	ib->ptr[0] = PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0);
+	ib->ptr[1] = addr;
+	ib->ptr[2] = PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0);
+	ib->ptr[3] = addr >> 32;
+	ib->ptr[4] = PACKET0(mmUVD_GPCOM_VCPU_CMD, 0);
+	ib->ptr[5] = 0;
+	for (i = 6; i < 16; ++i)
+		ib->ptr[i] = PACKET2(0);
+	ib->length_dw = 16;
+
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+						 &amdgpu_uvd_free_job,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err2;
+
+	ttm_eu_fence_buffer_objects(&ticket, &head, f);
+
+	if (fence)
+		*fence = fence_get(f);
+	amdgpu_bo_unref(&bo);
+	fence_put(f);
+	if (amdgpu_enable_scheduler)
+		return 0;
+
+	amdgpu_ib_free(ring->adev, ib);
+	kfree(ib);
+	return 0;
+err2:
+	amdgpu_ib_free(ring->adev, ib);
+err1:
+	kfree(ib);
+err:
+	ttm_eu_backoff_reservation(&ticket, &head);
+	return r;
+}
+
+/* multiple fence commands without any stream commands in between can
+   crash the vcpu so just try to emmit a dummy create/destroy msg to
+   avoid this */
+int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_bo *bo;
+	uint32_t *msg;
+	int r, i;
+
+	r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
+			     AMDGPU_GEM_DOMAIN_VRAM,
+			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			     NULL, NULL, &bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_reserve(bo, false);
+	if (r) {
+		amdgpu_bo_unref(&bo);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, (void **)&msg);
+	if (r) {
+		amdgpu_bo_unreserve(bo);
+		amdgpu_bo_unref(&bo);
+		return r;
+	}
+
+	/* stitch together an UVD create msg */
+	msg[0] = cpu_to_le32(0x00000de4);
+	msg[1] = cpu_to_le32(0x00000000);
+	msg[2] = cpu_to_le32(handle);
+	msg[3] = cpu_to_le32(0x00000000);
+	msg[4] = cpu_to_le32(0x00000000);
+	msg[5] = cpu_to_le32(0x00000000);
+	msg[6] = cpu_to_le32(0x00000000);
+	msg[7] = cpu_to_le32(0x00000780);
+	msg[8] = cpu_to_le32(0x00000440);
+	msg[9] = cpu_to_le32(0x00000000);
+	msg[10] = cpu_to_le32(0x01b37000);
+	for (i = 11; i < 1024; ++i)
+		msg[i] = cpu_to_le32(0x0);
+
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unreserve(bo);
+
+	return amdgpu_uvd_send_msg(ring, bo, fence);
+}
+
+int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       struct fence **fence)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_bo *bo;
+	uint32_t *msg;
+	int r, i;
+
+	r = amdgpu_bo_create(adev, 1024, PAGE_SIZE, true,
+			     AMDGPU_GEM_DOMAIN_VRAM,
+			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			     NULL, NULL, &bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_reserve(bo, false);
+	if (r) {
+		amdgpu_bo_unref(&bo);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(bo, (void **)&msg);
+	if (r) {
+		amdgpu_bo_unreserve(bo);
+		amdgpu_bo_unref(&bo);
+		return r;
+	}
+
+	/* stitch together an UVD destroy msg */
+	msg[0] = cpu_to_le32(0x00000de4);
+	msg[1] = cpu_to_le32(0x00000002);
+	msg[2] = cpu_to_le32(handle);
+	msg[3] = cpu_to_le32(0x00000000);
+	for (i = 4; i < 1024; ++i)
+		msg[i] = cpu_to_le32(0x0);
+
+	amdgpu_bo_kunmap(bo);
+	amdgpu_bo_unreserve(bo);
+
+	return amdgpu_uvd_send_msg(ring, bo, fence);
+}
+
+static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, uvd.idle_work.work);
+	unsigned i, fences, handles = 0;
+
+	fences = amdgpu_fence_count_emitted(&adev->uvd.ring);
+
+	for (i = 0; i < AMDGPU_MAX_UVD_HANDLES; ++i)
+		if (atomic_read(&adev->uvd.handles[i]))
+			++handles;
+
+	if (fences == 0 && handles == 0) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_uvd(adev, false);
+		} else {
+			amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+		}
+	} else {
+		schedule_delayed_work(&adev->uvd.idle_work,
+				      msecs_to_jiffies(UVD_IDLE_TIMEOUT_MS));
+	}
+}
+
+static void amdgpu_uvd_note_usage(struct amdgpu_device *adev)
+{
+	bool set_clocks = !cancel_delayed_work_sync(&adev->uvd.idle_work);
+	set_clocks &= schedule_delayed_work(&adev->uvd.idle_work,
+					    msecs_to_jiffies(UVD_IDLE_TIMEOUT_MS));
+
+	if (set_clocks) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_uvd(adev, true);
+		} else {
+			amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+		}
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
new file mode 100644
index 000000000..1724c2c86
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_UVD_H__
+#define __AMDGPU_UVD_H__
+
+int amdgpu_uvd_sw_init(struct amdgpu_device *adev);
+int amdgpu_uvd_sw_fini(struct amdgpu_device *adev);
+int amdgpu_uvd_suspend(struct amdgpu_device *adev);
+int amdgpu_uvd_resume(struct amdgpu_device *adev);
+int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct fence **fence);
+int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       struct fence **fence);
+void amdgpu_uvd_free_handles(struct amdgpu_device *adev,
+			     struct drm_file *filp);
+int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
new file mode 100644
index 000000000..a745eeeb5
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
@@ -0,0 +1,880 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_vce.h"
+#include "cikd.h"
+
+/* 1 second timeout */
+#define VCE_IDLE_TIMEOUT_MS	1000
+
+/* Firmware Names */
+#ifdef CONFIG_DRM_AMDGPU_CIK
+#define FIRMWARE_BONAIRE	"radeon/bonaire_vce.bin"
+#define FIRMWARE_KABINI 	"radeon/kabini_vce.bin"
+#define FIRMWARE_KAVERI 	"radeon/kaveri_vce.bin"
+#define FIRMWARE_HAWAII 	"radeon/hawaii_vce.bin"
+#define FIRMWARE_MULLINS	"radeon/mullins_vce.bin"
+#endif
+#define FIRMWARE_TONGA		"amdgpu/tonga_vce.bin"
+#define FIRMWARE_CARRIZO	"amdgpu/carrizo_vce.bin"
+#define FIRMWARE_FIJI		"amdgpu/fiji_vce.bin"
+#define FIRMWARE_STONEY		"amdgpu/stoney_vce.bin"
+
+#ifdef CONFIG_DRM_AMDGPU_CIK
+MODULE_FIRMWARE(FIRMWARE_BONAIRE);
+MODULE_FIRMWARE(FIRMWARE_KABINI);
+MODULE_FIRMWARE(FIRMWARE_KAVERI);
+MODULE_FIRMWARE(FIRMWARE_HAWAII);
+MODULE_FIRMWARE(FIRMWARE_MULLINS);
+#endif
+MODULE_FIRMWARE(FIRMWARE_TONGA);
+MODULE_FIRMWARE(FIRMWARE_CARRIZO);
+MODULE_FIRMWARE(FIRMWARE_FIJI);
+MODULE_FIRMWARE(FIRMWARE_STONEY);
+
+static void amdgpu_vce_idle_work_handler(struct work_struct *work);
+
+/**
+ * amdgpu_vce_init - allocate memory, load vce firmware
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * First step to get VCE online, allocate memory and load the firmware
+ */
+int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
+{
+	const char *fw_name;
+	const struct common_firmware_header *hdr;
+	unsigned ucode_version, version_major, version_minor, binary_id;
+	int i, r;
+
+	INIT_DELAYED_WORK(&adev->vce.idle_work, amdgpu_vce_idle_work_handler);
+
+	switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_CIK
+	case CHIP_BONAIRE:
+		fw_name = FIRMWARE_BONAIRE;
+		break;
+	case CHIP_KAVERI:
+		fw_name = FIRMWARE_KAVERI;
+		break;
+	case CHIP_KABINI:
+		fw_name = FIRMWARE_KABINI;
+		break;
+	case CHIP_HAWAII:
+		fw_name = FIRMWARE_HAWAII;
+		break;
+	case CHIP_MULLINS:
+		fw_name = FIRMWARE_MULLINS;
+		break;
+#endif
+	case CHIP_TONGA:
+		fw_name = FIRMWARE_TONGA;
+		break;
+	case CHIP_CARRIZO:
+		fw_name = FIRMWARE_CARRIZO;
+		break;
+	case CHIP_FIJI:
+		fw_name = FIRMWARE_FIJI;
+		break;
+	case CHIP_STONEY:
+		fw_name = FIRMWARE_STONEY;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&adev->vce.fw, fw_name, adev->dev);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_vce: Can't load firmware \"%s\"\n",
+			fw_name);
+		return r;
+	}
+
+	r = amdgpu_ucode_validate(adev->vce.fw);
+	if (r) {
+		dev_err(adev->dev, "amdgpu_vce: Can't validate firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->vce.fw);
+		adev->vce.fw = NULL;
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
+
+	ucode_version = le32_to_cpu(hdr->ucode_version);
+	version_major = (ucode_version >> 20) & 0xfff;
+	version_minor = (ucode_version >> 8) & 0xfff;
+	binary_id = ucode_version & 0xff;
+	DRM_INFO("Found VCE firmware Version: %hhd.%hhd Binary ID: %hhd\n",
+		version_major, version_minor, binary_id);
+	adev->vce.fw_version = ((version_major << 24) | (version_minor << 16) |
+				(binary_id << 8));
+
+	/* allocate firmware, stack and heap BO */
+
+	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
+			     AMDGPU_GEM_DOMAIN_VRAM,
+			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			     NULL, NULL, &adev->vce.vcpu_bo);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate VCE bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_reserve(adev->vce.vcpu_bo, false);
+	if (r) {
+		amdgpu_bo_unref(&adev->vce.vcpu_bo);
+		dev_err(adev->dev, "(%d) failed to reserve VCE bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_pin(adev->vce.vcpu_bo, AMDGPU_GEM_DOMAIN_VRAM,
+			  &adev->vce.gpu_addr);
+	amdgpu_bo_unreserve(adev->vce.vcpu_bo);
+	if (r) {
+		amdgpu_bo_unref(&adev->vce.vcpu_bo);
+		dev_err(adev->dev, "(%d) VCE bo pin failed\n", r);
+		return r;
+	}
+
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		atomic_set(&adev->vce.handles[i], 0);
+		adev->vce.filp[i] = NULL;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_fini - free memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Last step on VCE teardown, free firmware memory
+ */
+int amdgpu_vce_sw_fini(struct amdgpu_device *adev)
+{
+	if (adev->vce.vcpu_bo == NULL)
+		return 0;
+
+	amdgpu_bo_unref(&adev->vce.vcpu_bo);
+
+	amdgpu_ring_fini(&adev->vce.ring[0]);
+	amdgpu_ring_fini(&adev->vce.ring[1]);
+
+	release_firmware(adev->vce.fw);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_suspend - unpin VCE fw memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+int amdgpu_vce_suspend(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (adev->vce.vcpu_bo == NULL)
+		return 0;
+
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
+		if (atomic_read(&adev->vce.handles[i]))
+			break;
+
+	if (i == AMDGPU_MAX_VCE_HANDLES)
+		return 0;
+
+	/* TODO: suspending running encoding sessions isn't supported */
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_vce_resume - pin VCE fw memory
+ *
+ * @adev: amdgpu_device pointer
+ *
+ */
+int amdgpu_vce_resume(struct amdgpu_device *adev)
+{
+	void *cpu_addr;
+	const struct common_firmware_header *hdr;
+	unsigned offset;
+	int r;
+
+	if (adev->vce.vcpu_bo == NULL)
+		return -EINVAL;
+
+	r = amdgpu_bo_reserve(adev->vce.vcpu_bo, false);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to reserve VCE bo\n", r);
+		return r;
+	}
+
+	r = amdgpu_bo_kmap(adev->vce.vcpu_bo, &cpu_addr);
+	if (r) {
+		amdgpu_bo_unreserve(adev->vce.vcpu_bo);
+		dev_err(adev->dev, "(%d) VCE map failed\n", r);
+		return r;
+	}
+
+	hdr = (const struct common_firmware_header *)adev->vce.fw->data;
+	offset = le32_to_cpu(hdr->ucode_array_offset_bytes);
+	memcpy(cpu_addr, (adev->vce.fw->data) + offset,
+		(adev->vce.fw->size) - offset);
+
+	amdgpu_bo_kunmap(adev->vce.vcpu_bo);
+
+	amdgpu_bo_unreserve(adev->vce.vcpu_bo);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_idle_work_handler - power off VCE
+ *
+ * @work: pointer to work structure
+ *
+ * power of VCE when it's not used any more
+ */
+static void amdgpu_vce_idle_work_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, vce.idle_work.work);
+
+	if ((amdgpu_fence_count_emitted(&adev->vce.ring[0]) == 0) &&
+	    (amdgpu_fence_count_emitted(&adev->vce.ring[1]) == 0)) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_vce(adev, false);
+		} else {
+			amdgpu_asic_set_vce_clocks(adev, 0, 0);
+		}
+	} else {
+		schedule_delayed_work(&adev->vce.idle_work,
+				      msecs_to_jiffies(VCE_IDLE_TIMEOUT_MS));
+	}
+}
+
+/**
+ * amdgpu_vce_note_usage - power up VCE
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Make sure VCE is powerd up when we want to use it
+ */
+static void amdgpu_vce_note_usage(struct amdgpu_device *adev)
+{
+	bool streams_changed = false;
+	bool set_clocks = !cancel_delayed_work_sync(&adev->vce.idle_work);
+	set_clocks &= schedule_delayed_work(&adev->vce.idle_work,
+					    msecs_to_jiffies(VCE_IDLE_TIMEOUT_MS));
+
+	if (adev->pm.dpm_enabled) {
+		/* XXX figure out if the streams changed */
+		streams_changed = false;
+	}
+
+	if (set_clocks || streams_changed) {
+		if (adev->pm.dpm_enabled) {
+			amdgpu_dpm_enable_vce(adev, true);
+		} else {
+			amdgpu_asic_set_vce_clocks(adev, 53300, 40000);
+		}
+	}
+}
+
+/**
+ * amdgpu_vce_free_handles - free still open VCE handles
+ *
+ * @adev: amdgpu_device pointer
+ * @filp: drm file pointer
+ *
+ * Close all VCE handles still open by this file pointer
+ */
+void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
+{
+	struct amdgpu_ring *ring = &adev->vce.ring[0];
+	int i, r;
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		uint32_t handle = atomic_read(&adev->vce.handles[i]);
+		if (!handle || adev->vce.filp[i] != filp)
+			continue;
+
+		amdgpu_vce_note_usage(adev);
+
+		r = amdgpu_vce_get_destroy_msg(ring, handle, NULL);
+		if (r)
+			DRM_ERROR("Error destroying VCE handle (%d)!\n", r);
+
+		adev->vce.filp[i] = NULL;
+		atomic_set(&adev->vce.handles[i], 0);
+	}
+}
+
+static int amdgpu_vce_free_job(
+	struct amdgpu_job *job)
+{
+	amdgpu_ib_free(job->adev, job->ibs);
+	kfree(job->ibs);
+	return 0;
+}
+
+/**
+ * amdgpu_vce_get_create_msg - generate a VCE create msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: VCE session handle to use
+ * @fence: optional fence to return
+ *
+ * Open up a stream for HW test
+ */
+int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct fence **fence)
+{
+	const unsigned ib_size_dw = 1024;
+	struct amdgpu_ib *ib = NULL;
+	struct fence *f = NULL;
+	struct amdgpu_device *adev = ring->adev;
+	uint64_t dummy;
+	int i, r;
+
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib)
+		return -ENOMEM;
+	r = amdgpu_ib_get(ring, NULL, ib_size_dw * 4, ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		kfree(ib);
+		return r;
+	}
+
+	dummy = ib->gpu_addr + 1024;
+
+	/* stitch together an VCE create msg */
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x0000000c; /* len */
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session cmd */
+	ib->ptr[ib->length_dw++] = handle;
+
+	if ((ring->adev->vce.fw_version >> 24) >= 52)
+		ib->ptr[ib->length_dw++] = 0x00000040; /* len */
+	else
+		ib->ptr[ib->length_dw++] = 0x00000030; /* len */
+	ib->ptr[ib->length_dw++] = 0x01000001; /* create cmd */
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000042;
+	ib->ptr[ib->length_dw++] = 0x0000000a;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000080;
+	ib->ptr[ib->length_dw++] = 0x00000060;
+	ib->ptr[ib->length_dw++] = 0x00000100;
+	ib->ptr[ib->length_dw++] = 0x00000100;
+	ib->ptr[ib->length_dw++] = 0x0000000c;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	if ((ring->adev->vce.fw_version >> 24) >= 52) {
+		ib->ptr[ib->length_dw++] = 0x00000000;
+		ib->ptr[ib->length_dw++] = 0x00000000;
+		ib->ptr[ib->length_dw++] = 0x00000000;
+		ib->ptr[ib->length_dw++] = 0x00000000;
+	}
+
+	ib->ptr[ib->length_dw++] = 0x00000014; /* len */
+	ib->ptr[ib->length_dw++] = 0x05000005; /* feedback buffer */
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+						 &amdgpu_vce_free_job,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err;
+	if (fence)
+		*fence = fence_get(f);
+	fence_put(f);
+	if (amdgpu_enable_scheduler)
+		return 0;
+err:
+	amdgpu_ib_free(adev, ib);
+	kfree(ib);
+	return r;
+}
+
+/**
+ * amdgpu_vce_get_destroy_msg - generate a VCE destroy msg
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: ring we should submit the msg to
+ * @handle: VCE session handle to use
+ * @fence: optional fence to return
+ *
+ * Close up a stream for HW test or if userspace failed to do so
+ */
+int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       struct fence **fence)
+{
+	const unsigned ib_size_dw = 1024;
+	struct amdgpu_ib *ib = NULL;
+	struct fence *f = NULL;
+	struct amdgpu_device *adev = ring->adev;
+	uint64_t dummy;
+	int i, r;
+
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib)
+		return -ENOMEM;
+
+	r = amdgpu_ib_get(ring, NULL, ib_size_dw * 4, ib);
+	if (r) {
+		kfree(ib);
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		return r;
+	}
+
+	dummy = ib->gpu_addr + 1024;
+
+	/* stitch together an VCE destroy msg */
+	ib->length_dw = 0;
+	ib->ptr[ib->length_dw++] = 0x0000000c; /* len */
+	ib->ptr[ib->length_dw++] = 0x00000001; /* session cmd */
+	ib->ptr[ib->length_dw++] = handle;
+
+	ib->ptr[ib->length_dw++] = 0x00000014; /* len */
+	ib->ptr[ib->length_dw++] = 0x05000005; /* feedback buffer */
+	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
+	ib->ptr[ib->length_dw++] = dummy;
+	ib->ptr[ib->length_dw++] = 0x00000001;
+
+	ib->ptr[ib->length_dw++] = 0x00000008; /* len */
+	ib->ptr[ib->length_dw++] = 0x02000001; /* destroy cmd */
+
+	for (i = ib->length_dw; i < ib_size_dw; ++i)
+		ib->ptr[i] = 0x0;
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+						 &amdgpu_vce_free_job,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err;
+	if (fence)
+		*fence = fence_get(f);
+	fence_put(f);
+	if (amdgpu_enable_scheduler)
+		return 0;
+err:
+	amdgpu_ib_free(adev, ib);
+	kfree(ib);
+	return r;
+}
+
+/**
+ * amdgpu_vce_cs_reloc - command submission relocation
+ *
+ * @p: parser context
+ * @lo: address of lower dword
+ * @hi: address of higher dword
+ * @size: minimum size
+ *
+ * Patch relocation inside command stream with real buffer address
+ */
+static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx,
+			       int lo, int hi, unsigned size, uint32_t index)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_ib *ib = &p->ibs[ib_idx];
+	struct amdgpu_bo *bo;
+	uint64_t addr;
+
+	if (index == 0xffffffff)
+		index = 0;
+
+	addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) |
+	       ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32;
+	addr += ((uint64_t)size) * ((uint64_t)index);
+
+	mapping = amdgpu_cs_find_mapping(p, addr, &bo);
+	if (mapping == NULL) {
+		DRM_ERROR("Can't find BO for addr 0x%010Lx %d %d %d %d\n",
+			  addr, lo, hi, size, index);
+		return -EINVAL;
+	}
+
+	if ((addr + (uint64_t)size) >
+	    ((uint64_t)mapping->it.last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+		DRM_ERROR("BO to small for addr 0x%010Lx %d %d\n",
+			  addr, lo, hi);
+		return -EINVAL;
+	}
+
+	addr -= ((uint64_t)mapping->it.start) * AMDGPU_GPU_PAGE_SIZE;
+	addr += amdgpu_bo_gpu_offset(bo);
+	addr -= ((uint64_t)size) * ((uint64_t)index);
+
+	ib->ptr[lo] = addr & 0xFFFFFFFF;
+	ib->ptr[hi] = addr >> 32;
+
+	return 0;
+}
+
+/**
+ * amdgpu_vce_validate_handle - validate stream handle
+ *
+ * @p: parser context
+ * @handle: handle to validate
+ * @allocated: allocated a new handle?
+ *
+ * Validates the handle and return the found session index or -EINVAL
+ * we we don't have another free session index.
+ */
+static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p,
+				      uint32_t handle, bool *allocated)
+{
+	unsigned i;
+
+	*allocated = false;
+
+	/* validate the handle */
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		if (atomic_read(&p->adev->vce.handles[i]) == handle) {
+			if (p->adev->vce.filp[i] != p->filp) {
+				DRM_ERROR("VCE handle collision detected!\n");
+				return -EINVAL;
+			}
+			return i;
+		}
+	}
+
+	/* handle not found try to alloc a new one */
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
+		if (!atomic_cmpxchg(&p->adev->vce.handles[i], 0, handle)) {
+			p->adev->vce.filp[i] = p->filp;
+			p->adev->vce.img_size[i] = 0;
+			*allocated = true;
+			return i;
+		}
+	}
+
+	DRM_ERROR("No more free VCE handles!\n");
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_vce_cs_parse - parse and validate the command stream
+ *
+ * @p: parser context
+ *
+ */
+int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
+{
+	struct amdgpu_ib *ib = &p->ibs[ib_idx];
+	unsigned fb_idx = 0, bs_idx = 0;
+	int session_idx = -1;
+	bool destroyed = false;
+	bool created = false;
+	bool allocated = false;
+	uint32_t tmp, handle = 0;
+	uint32_t *size = &tmp;
+	int i, r = 0, idx = 0;
+
+	amdgpu_vce_note_usage(p->adev);
+
+	while (idx < ib->length_dw) {
+		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
+		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
+
+		if ((len < 8) || (len & 3)) {
+			DRM_ERROR("invalid VCE command length (%d)!\n", len);
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (destroyed) {
+			DRM_ERROR("No other command allowed after destroy!\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		switch (cmd) {
+		case 0x00000001: // session
+			handle = amdgpu_get_ib_value(p, ib_idx, idx + 2);
+			session_idx = amdgpu_vce_validate_handle(p, handle,
+								 &allocated);
+			if (session_idx < 0)
+				return session_idx;
+			size = &p->adev->vce.img_size[session_idx];
+			break;
+
+		case 0x00000002: // task info
+			fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6);
+			bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7);
+			break;
+
+		case 0x01000001: // create
+			created = true;
+			if (!allocated) {
+				DRM_ERROR("Handle already in use!\n");
+				r = -EINVAL;
+				goto out;
+			}
+
+			*size = amdgpu_get_ib_value(p, ib_idx, idx + 8) *
+				amdgpu_get_ib_value(p, ib_idx, idx + 10) *
+				8 * 3 / 2;
+			break;
+
+		case 0x04000001: // config extension
+		case 0x04000002: // pic control
+		case 0x04000005: // rate control
+		case 0x04000007: // motion estimation
+		case 0x04000008: // rdo
+		case 0x04000009: // vui
+		case 0x05000002: // auxiliary buffer
+			break;
+
+		case 0x03000001: // encode
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9,
+						*size, 0);
+			if (r)
+				goto out;
+
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11,
+						*size / 3, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x02000001: // destroy
+			destroyed = true;
+			break;
+
+		case 0x05000001: // context buffer
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+						*size * 2, 0);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000004: // video bitstream buffer
+			tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4);
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+						tmp, bs_idx);
+			if (r)
+				goto out;
+			break;
+
+		case 0x05000005: // feedback buffer
+			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
+						4096, fb_idx);
+			if (r)
+				goto out;
+			break;
+
+		default:
+			DRM_ERROR("invalid VCE command (0x%x)!\n", cmd);
+			r = -EINVAL;
+			goto out;
+		}
+
+		if (session_idx == -1) {
+			DRM_ERROR("no session command at start of IB\n");
+			r = -EINVAL;
+			goto out;
+		}
+
+		idx += len / 4;
+	}
+
+	if (allocated && !created) {
+		DRM_ERROR("New session without create command!\n");
+		r = -ENOENT;
+	}
+
+out:
+	if ((!r && destroyed) || (r && allocated)) {
+		/*
+		 * IB contains a destroy msg or we have allocated an
+		 * handle and got an error, anyway free the handle
+		 */
+		for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
+			atomic_cmpxchg(&p->adev->vce.handles[i], handle, 0);
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_vce_ring_emit_semaphore - emit a semaphore command
+ *
+ * @ring: engine to use
+ * @semaphore: address of semaphore
+ * @emit_wait: true=emit wait, false=emit signal
+ *
+ */
+bool amdgpu_vce_ring_emit_semaphore(struct amdgpu_ring *ring,
+				    struct amdgpu_semaphore *semaphore,
+				    bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	amdgpu_ring_write(ring, VCE_CMD_SEMAPHORE);
+	amdgpu_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+	amdgpu_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+	amdgpu_ring_write(ring, 0x01003000 | (emit_wait ? 1 : 0));
+	if (!emit_wait)
+		amdgpu_ring_write(ring, VCE_CMD_END);
+
+	return true;
+}
+
+/**
+ * amdgpu_vce_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: engine to use
+ * @ib: the IB to execute
+ *
+ */
+void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib)
+{
+	amdgpu_ring_write(ring, VCE_CMD_IB);
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * amdgpu_vce_ring_emit_fence - add a fence command to the ring
+ *
+ * @ring: engine to use
+ * @fence: the fence
+ *
+ */
+void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, VCE_CMD_FENCE);
+	amdgpu_ring_write(ring, addr);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, VCE_CMD_TRAP);
+	amdgpu_ring_write(ring, VCE_CMD_END);
+}
+
+/**
+ * amdgpu_vce_ring_test_ring - test if VCE ring is working
+ *
+ * @ring: the engine to test on
+ *
+ */
+int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t rptr = amdgpu_ring_get_rptr(ring);
+	unsigned i;
+	int r;
+
+	r = amdgpu_ring_lock(ring, 16);
+	if (r) {
+		DRM_ERROR("amdgpu: vce failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, VCE_CMD_END);
+	amdgpu_ring_unlock_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (amdgpu_ring_get_rptr(ring) != rptr)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed\n",
+			  ring->idx);
+		r = -ETIMEDOUT;
+	}
+
+	return r;
+}
+
+/**
+ * amdgpu_vce_ring_test_ib - test if VCE IBs are working
+ *
+ * @ring: the engine to test on
+ *
+ */
+int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct fence *fence = NULL;
+	int r;
+
+	/* skip vce ring1 ib test for now, since it's not reliable */
+	if (ring == &ring->adev->vce.ring[1])
+		return 0;
+
+	r = amdgpu_vce_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_vce_get_destroy_msg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
+		goto error;
+	}
+
+	r = fence_wait(fence, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	} else {
+		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+	}
+error:
+	fence_put(fence);
+	return r;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
new file mode 100644
index 000000000..ba2da8ee5
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_VCE_H__
+#define __AMDGPU_VCE_H__
+
+int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size);
+int amdgpu_vce_sw_fini(struct amdgpu_device *adev);
+int amdgpu_vce_suspend(struct amdgpu_device *adev);
+int amdgpu_vce_resume(struct amdgpu_device *adev);
+int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+			      struct fence **fence);
+int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+			       struct fence **fence);
+void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp);
+int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx);
+bool amdgpu_vce_ring_emit_semaphore(struct amdgpu_ring *ring,
+				    struct amdgpu_semaphore *semaphore,
+				    bool emit_wait);
+void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
+void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				unsigned flags);
+int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);
+int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
new file mode 100644
index 000000000..8c5ec151d
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -0,0 +1,1356 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ * Copyright 2009 Jerome Glisse.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_trace.h"
+
+/*
+ * GPUVM
+ * GPUVM is similar to the legacy gart on older asics, however
+ * rather than there being a single global gart table
+ * for the entire GPU, there are multiple VM page tables active
+ * at any given time.  The VM page tables can contain a mix
+ * vram pages and system memory pages and system memory pages
+ * can be mapped as snooped (cached system pages) or unsnooped
+ * (uncached system pages).
+ * Each VM has an ID associated with it and there is a page table
+ * associated with each VMID.  When execting a command buffer,
+ * the kernel tells the the ring what VMID to use for that command
+ * buffer.  VMIDs are allocated dynamically as commands are submitted.
+ * The userspace drivers maintain their own address space and the kernel
+ * sets up their pages tables accordingly when they submit their
+ * command buffers and a VMID is assigned.
+ * Cayman/Trinity support up to 8 active VMs at any given time;
+ * SI supports 16.
+ */
+
+/**
+ * amdgpu_vm_num_pde - return the number of page directory entries
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate the number of page directory entries (cayman+).
+ */
+static unsigned amdgpu_vm_num_pdes(struct amdgpu_device *adev)
+{
+	return adev->vm_manager.max_pfn >> amdgpu_vm_block_size;
+}
+
+/**
+ * amdgpu_vm_directory_size - returns the size of the page directory in bytes
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate the size of the page directory in bytes (cayman+).
+ */
+static unsigned amdgpu_vm_directory_size(struct amdgpu_device *adev)
+{
+	return AMDGPU_GPU_PAGE_ALIGN(amdgpu_vm_num_pdes(adev) * 8);
+}
+
+/**
+ * amdgpu_vm_get_bos - add the vm BOs to a validation list
+ *
+ * @vm: vm providing the BOs
+ * @head: head of validation list
+ *
+ * Add the page directory to the list of BOs to
+ * validate for command submission (cayman+).
+ */
+struct amdgpu_bo_list_entry *amdgpu_vm_get_bos(struct amdgpu_device *adev,
+					  struct amdgpu_vm *vm,
+					  struct list_head *head)
+{
+	struct amdgpu_bo_list_entry *list;
+	unsigned i, idx;
+
+	list = drm_malloc_ab(vm->max_pde_used + 2,
+			     sizeof(struct amdgpu_bo_list_entry));
+	if (!list) {
+		return NULL;
+	}
+
+	/* add the vm page table to the list */
+	list[0].robj = vm->page_directory;
+	list[0].prefered_domains = AMDGPU_GEM_DOMAIN_VRAM;
+	list[0].allowed_domains = AMDGPU_GEM_DOMAIN_VRAM;
+	list[0].priority = 0;
+	list[0].tv.bo = &vm->page_directory->tbo;
+	list[0].tv.shared = true;
+	list_add(&list[0].tv.head, head);
+
+	for (i = 0, idx = 1; i <= vm->max_pde_used; i++) {
+		if (!vm->page_tables[i].bo)
+			continue;
+
+		list[idx].robj = vm->page_tables[i].bo;
+		list[idx].prefered_domains = AMDGPU_GEM_DOMAIN_VRAM;
+		list[idx].allowed_domains = AMDGPU_GEM_DOMAIN_VRAM;
+		list[idx].priority = 0;
+		list[idx].tv.bo = &list[idx].robj->tbo;
+		list[idx].tv.shared = true;
+		list_add(&list[idx++].tv.head, head);
+	}
+
+	return list;
+}
+
+/**
+ * amdgpu_vm_grab_id - allocate the next free VMID
+ *
+ * @vm: vm to allocate id for
+ * @ring: ring we want to submit job to
+ * @sync: sync object where we add dependencies
+ *
+ * Allocate an id for the vm, adding fences to the sync obj as necessary.
+ *
+ * Global mutex must be locked!
+ */
+int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
+		      struct amdgpu_sync *sync)
+{
+	struct fence *best[AMDGPU_MAX_RINGS] = {};
+	struct amdgpu_vm_id *vm_id = &vm->ids[ring->idx];
+	struct amdgpu_device *adev = ring->adev;
+
+	unsigned choices[2] = {};
+	unsigned i;
+
+	/* check if the id is still valid */
+	if (vm_id->id) {
+		unsigned id = vm_id->id;
+		long owner;
+
+		owner = atomic_long_read(&adev->vm_manager.ids[id].owner);
+		if (owner == (long)vm) {
+			trace_amdgpu_vm_grab_id(vm_id->id, ring->idx);
+			return 0;
+		}
+	}
+
+	/* we definately need to flush */
+	vm_id->pd_gpu_addr = ~0ll;
+
+	/* skip over VMID 0, since it is the system VM */
+	for (i = 1; i < adev->vm_manager.nvm; ++i) {
+		struct fence *fence = adev->vm_manager.ids[i].active;
+		struct amdgpu_ring *fring;
+
+		if (fence == NULL) {
+			/* found a free one */
+			vm_id->id = i;
+			trace_amdgpu_vm_grab_id(i, ring->idx);
+			return 0;
+		}
+
+		fring = amdgpu_ring_from_fence(fence);
+		if (best[fring->idx] == NULL ||
+		    fence_is_later(best[fring->idx], fence)) {
+			best[fring->idx] = fence;
+			choices[fring == ring ? 0 : 1] = i;
+		}
+	}
+
+	for (i = 0; i < 2; ++i) {
+		if (choices[i]) {
+			struct fence *fence;
+
+			fence  = adev->vm_manager.ids[choices[i]].active;
+			vm_id->id = choices[i];
+
+			trace_amdgpu_vm_grab_id(choices[i], ring->idx);
+			return amdgpu_sync_fence(ring->adev, sync, fence);
+		}
+	}
+
+	/* should never happen */
+	BUG();
+	return -EINVAL;
+}
+
+/**
+ * amdgpu_vm_flush - hardware flush the vm
+ *
+ * @ring: ring to use for flush
+ * @vm: vm we want to flush
+ * @updates: last vm update that we waited for
+ *
+ * Flush the vm (cayman+).
+ *
+ * Global and local mutex must be locked!
+ */
+void amdgpu_vm_flush(struct amdgpu_ring *ring,
+		     struct amdgpu_vm *vm,
+		     struct fence *updates)
+{
+	uint64_t pd_addr = amdgpu_bo_gpu_offset(vm->page_directory);
+	struct amdgpu_vm_id *vm_id = &vm->ids[ring->idx];
+	struct fence *flushed_updates = vm_id->flushed_updates;
+	bool is_later;
+
+	if (!flushed_updates)
+		is_later = true;
+	else if (!updates)
+		is_later = false;
+	else
+		is_later = fence_is_later(updates, flushed_updates);
+
+	if (pd_addr != vm_id->pd_gpu_addr || is_later) {
+		trace_amdgpu_vm_flush(pd_addr, ring->idx, vm_id->id);
+		if (is_later) {
+			vm_id->flushed_updates = fence_get(updates);
+			fence_put(flushed_updates);
+		}
+		vm_id->pd_gpu_addr = pd_addr;
+		amdgpu_ring_emit_vm_flush(ring, vm_id->id, vm_id->pd_gpu_addr);
+	}
+}
+
+/**
+ * amdgpu_vm_fence - remember fence for vm
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: vm we want to fence
+ * @fence: fence to remember
+ *
+ * Fence the vm (cayman+).
+ * Set the fence used to protect page table and id.
+ *
+ * Global and local mutex must be locked!
+ */
+void amdgpu_vm_fence(struct amdgpu_device *adev,
+		     struct amdgpu_vm *vm,
+		     struct fence *fence)
+{
+	struct amdgpu_ring *ring = amdgpu_ring_from_fence(fence);
+	unsigned vm_id = vm->ids[ring->idx].id;
+
+	fence_put(adev->vm_manager.ids[vm_id].active);
+	adev->vm_manager.ids[vm_id].active = fence_get(fence);
+	atomic_long_set(&adev->vm_manager.ids[vm_id].owner, (long)vm);
+}
+
+/**
+ * amdgpu_vm_bo_find - find the bo_va for a specific vm & bo
+ *
+ * @vm: requested vm
+ * @bo: requested buffer object
+ *
+ * Find @bo inside the requested vm (cayman+).
+ * Search inside the @bos vm list for the requested vm
+ * Returns the found bo_va or NULL if none is found
+ *
+ * Object has to be reserved!
+ */
+struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
+				       struct amdgpu_bo *bo)
+{
+	struct amdgpu_bo_va *bo_va;
+
+	list_for_each_entry(bo_va, &bo->va, bo_list) {
+		if (bo_va->vm == vm) {
+			return bo_va;
+		}
+	}
+	return NULL;
+}
+
+/**
+ * amdgpu_vm_update_pages - helper to call the right asic function
+ *
+ * @adev: amdgpu_device pointer
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: hw access flags
+ * @gtt_flags: GTT hw access flags
+ *
+ * Traces the parameters and calls the right asic functions
+ * to setup the page table using the DMA.
+ */
+static void amdgpu_vm_update_pages(struct amdgpu_device *adev,
+				   struct amdgpu_ib *ib,
+				   uint64_t pe, uint64_t addr,
+				   unsigned count, uint32_t incr,
+				   uint32_t flags, uint32_t gtt_flags)
+{
+	trace_amdgpu_vm_set_page(pe, addr, count, incr, flags);
+
+	if ((flags & AMDGPU_PTE_SYSTEM) && (flags == gtt_flags)) {
+		uint64_t src = adev->gart.table_addr + (addr >> 12) * 8;
+		amdgpu_vm_copy_pte(adev, ib, pe, src, count);
+
+	} else if ((flags & AMDGPU_PTE_SYSTEM) || (count < 3)) {
+		amdgpu_vm_write_pte(adev, ib, pe, addr,
+				      count, incr, flags);
+
+	} else {
+		amdgpu_vm_set_pte_pde(adev, ib, pe, addr,
+				      count, incr, flags);
+	}
+}
+
+int amdgpu_vm_free_job(struct amdgpu_job *job)
+{
+	int i;
+	for (i = 0; i < job->num_ibs; i++)
+		amdgpu_ib_free(job->adev, &job->ibs[i]);
+	kfree(job->ibs);
+	return 0;
+}
+
+/**
+ * amdgpu_vm_clear_bo - initially clear the page dir/table
+ *
+ * @adev: amdgpu_device pointer
+ * @bo: bo to clear
+ *
+ * need to reserve bo first before calling it.
+ */
+static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
+			      struct amdgpu_bo *bo)
+{
+	struct amdgpu_ring *ring = adev->vm_manager.vm_pte_funcs_ring;
+	struct fence *fence = NULL;
+	struct amdgpu_ib *ib;
+	unsigned entries;
+	uint64_t addr;
+	int r;
+
+	r = reservation_object_reserve_shared(bo->tbo.resv);
+	if (r)
+		return r;
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+	if (r)
+		goto error;
+
+	addr = amdgpu_bo_gpu_offset(bo);
+	entries = amdgpu_bo_size(bo) / 8;
+
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib)
+		goto error;
+
+	r = amdgpu_ib_get(ring, NULL, entries * 2 + 64, ib);
+	if (r)
+		goto error_free;
+
+	ib->length_dw = 0;
+
+	amdgpu_vm_update_pages(adev, ib, addr, 0, entries, 0, 0, 0);
+	amdgpu_vm_pad_ib(adev, ib);
+	WARN_ON(ib->length_dw > 64);
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+						 &amdgpu_vm_free_job,
+						 AMDGPU_FENCE_OWNER_VM,
+						 &fence);
+	if (!r)
+		amdgpu_bo_fence(bo, fence, true);
+	fence_put(fence);
+	if (amdgpu_enable_scheduler)
+		return 0;
+
+error_free:
+	amdgpu_ib_free(adev, ib);
+	kfree(ib);
+
+error:
+	return r;
+}
+
+/**
+ * amdgpu_vm_map_gart - get the physical address of a gart page
+ *
+ * @adev: amdgpu_device pointer
+ * @addr: the unmapped addr
+ *
+ * Look up the physical address of the page that the pte resolves
+ * to (cayman+).
+ * Returns the physical address of the page.
+ */
+uint64_t amdgpu_vm_map_gart(struct amdgpu_device *adev, uint64_t addr)
+{
+	uint64_t result;
+
+	/* page table offset */
+	result = adev->gart.pages_addr[addr >> PAGE_SHIFT];
+
+	/* in case cpu page size != gpu page size*/
+	result |= addr & (~PAGE_MASK);
+
+	return result;
+}
+
+/**
+ * amdgpu_vm_update_pdes - make sure that page directory is valid
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @start: start of GPU address range
+ * @end: end of GPU address range
+ *
+ * Allocates new page tables if necessary
+ * and updates the page directory (cayman+).
+ * Returns 0 for success, error for failure.
+ *
+ * Global and local mutex must be locked!
+ */
+int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
+				    struct amdgpu_vm *vm)
+{
+	struct amdgpu_ring *ring = adev->vm_manager.vm_pte_funcs_ring;
+	struct amdgpu_bo *pd = vm->page_directory;
+	uint64_t pd_addr = amdgpu_bo_gpu_offset(pd);
+	uint32_t incr = AMDGPU_VM_PTE_COUNT * 8;
+	uint64_t last_pde = ~0, last_pt = ~0;
+	unsigned count = 0, pt_idx, ndw;
+	struct amdgpu_ib *ib;
+	struct fence *fence = NULL;
+
+	int r;
+
+	/* padding, etc. */
+	ndw = 64;
+
+	/* assume the worst case */
+	ndw += vm->max_pde_used * 6;
+
+	/* update too big for an IB */
+	if (ndw > 0xfffff)
+		return -ENOMEM;
+
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib)
+		return -ENOMEM;
+
+	r = amdgpu_ib_get(ring, NULL, ndw * 4, ib);
+	if (r) {
+		kfree(ib);
+		return r;
+	}
+	ib->length_dw = 0;
+
+	/* walk over the address space and update the page directory */
+	for (pt_idx = 0; pt_idx <= vm->max_pde_used; ++pt_idx) {
+		struct amdgpu_bo *bo = vm->page_tables[pt_idx].bo;
+		uint64_t pde, pt;
+
+		if (bo == NULL)
+			continue;
+
+		pt = amdgpu_bo_gpu_offset(bo);
+		if (vm->page_tables[pt_idx].addr == pt)
+			continue;
+		vm->page_tables[pt_idx].addr = pt;
+
+		pde = pd_addr + pt_idx * 8;
+		if (((last_pde + 8 * count) != pde) ||
+		    ((last_pt + incr * count) != pt)) {
+
+			if (count) {
+				amdgpu_vm_update_pages(adev, ib, last_pde,
+						       last_pt, count, incr,
+						       AMDGPU_PTE_VALID, 0);
+			}
+
+			count = 1;
+			last_pde = pde;
+			last_pt = pt;
+		} else {
+			++count;
+		}
+	}
+
+	if (count)
+		amdgpu_vm_update_pages(adev, ib, last_pde, last_pt, count,
+				       incr, AMDGPU_PTE_VALID, 0);
+
+	if (ib->length_dw != 0) {
+		amdgpu_vm_pad_ib(adev, ib);
+		amdgpu_sync_resv(adev, &ib->sync, pd->tbo.resv, AMDGPU_FENCE_OWNER_VM);
+		WARN_ON(ib->length_dw > ndw);
+		r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+							 &amdgpu_vm_free_job,
+							 AMDGPU_FENCE_OWNER_VM,
+							 &fence);
+		if (r)
+			goto error_free;
+
+		amdgpu_bo_fence(pd, fence, true);
+		fence_put(vm->page_directory_fence);
+		vm->page_directory_fence = fence_get(fence);
+		fence_put(fence);
+	}
+
+	if (!amdgpu_enable_scheduler || ib->length_dw == 0) {
+		amdgpu_ib_free(adev, ib);
+		kfree(ib);
+	}
+
+	return 0;
+
+error_free:
+	amdgpu_ib_free(adev, ib);
+	kfree(ib);
+	return r;
+}
+
+/**
+ * amdgpu_vm_frag_ptes - add fragment information to PTEs
+ *
+ * @adev: amdgpu_device pointer
+ * @ib: IB for the update
+ * @pe_start: first PTE to handle
+ * @pe_end: last PTE to handle
+ * @addr: addr those PTEs should point to
+ * @flags: hw mapping flags
+ * @gtt_flags: GTT hw mapping flags
+ *
+ * Global and local mutex must be locked!
+ */
+static void amdgpu_vm_frag_ptes(struct amdgpu_device *adev,
+				struct amdgpu_ib *ib,
+				uint64_t pe_start, uint64_t pe_end,
+				uint64_t addr, uint32_t flags,
+				uint32_t gtt_flags)
+{
+	/**
+	 * The MC L1 TLB supports variable sized pages, based on a fragment
+	 * field in the PTE. When this field is set to a non-zero value, page
+	 * granularity is increased from 4KB to (1 << (12 + frag)). The PTE
+	 * flags are considered valid for all PTEs within the fragment range
+	 * and corresponding mappings are assumed to be physically contiguous.
+	 *
+	 * The L1 TLB can store a single PTE for the whole fragment,
+	 * significantly increasing the space available for translation
+	 * caching. This leads to large improvements in throughput when the
+	 * TLB is under pressure.
+	 *
+	 * The L2 TLB distributes small and large fragments into two
+	 * asymmetric partitions. The large fragment cache is significantly
+	 * larger. Thus, we try to use large fragments wherever possible.
+	 * Userspace can support this by aligning virtual base address and
+	 * allocation size to the fragment size.
+	 */
+
+	/* SI and newer are optimized for 64KB */
+	uint64_t frag_flags = AMDGPU_PTE_FRAG_64KB;
+	uint64_t frag_align = 0x80;
+
+	uint64_t frag_start = ALIGN(pe_start, frag_align);
+	uint64_t frag_end = pe_end & ~(frag_align - 1);
+
+	unsigned count;
+
+	/* system pages are non continuously */
+	if ((flags & AMDGPU_PTE_SYSTEM) || !(flags & AMDGPU_PTE_VALID) ||
+	    (frag_start >= frag_end)) {
+
+		count = (pe_end - pe_start) / 8;
+		amdgpu_vm_update_pages(adev, ib, pe_start, addr, count,
+				       AMDGPU_GPU_PAGE_SIZE, flags, gtt_flags);
+		return;
+	}
+
+	/* handle the 4K area at the beginning */
+	if (pe_start != frag_start) {
+		count = (frag_start - pe_start) / 8;
+		amdgpu_vm_update_pages(adev, ib, pe_start, addr, count,
+				       AMDGPU_GPU_PAGE_SIZE, flags, gtt_flags);
+		addr += AMDGPU_GPU_PAGE_SIZE * count;
+	}
+
+	/* handle the area in the middle */
+	count = (frag_end - frag_start) / 8;
+	amdgpu_vm_update_pages(adev, ib, frag_start, addr, count,
+			       AMDGPU_GPU_PAGE_SIZE, flags | frag_flags,
+			       gtt_flags);
+
+	/* handle the 4K area at the end */
+	if (frag_end != pe_end) {
+		addr += AMDGPU_GPU_PAGE_SIZE * count;
+		count = (pe_end - frag_end) / 8;
+		amdgpu_vm_update_pages(adev, ib, frag_end, addr, count,
+				       AMDGPU_GPU_PAGE_SIZE, flags, gtt_flags);
+	}
+}
+
+/**
+ * amdgpu_vm_update_ptes - make sure that page tables are valid
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @start: start of GPU address range
+ * @end: end of GPU address range
+ * @dst: destination address to map to
+ * @flags: mapping flags
+ *
+ * Update the page tables in the range @start - @end (cayman+).
+ *
+ * Global and local mutex must be locked!
+ */
+static int amdgpu_vm_update_ptes(struct amdgpu_device *adev,
+				 struct amdgpu_vm *vm,
+				 struct amdgpu_ib *ib,
+				 uint64_t start, uint64_t end,
+				 uint64_t dst, uint32_t flags,
+				 uint32_t gtt_flags)
+{
+	uint64_t mask = AMDGPU_VM_PTE_COUNT - 1;
+	uint64_t last_pte = ~0, last_dst = ~0;
+	void *owner = AMDGPU_FENCE_OWNER_VM;
+	unsigned count = 0;
+	uint64_t addr;
+
+	/* sync to everything on unmapping */
+	if (!(flags & AMDGPU_PTE_VALID))
+		owner = AMDGPU_FENCE_OWNER_UNDEFINED;
+
+	/* walk over the address space and update the page tables */
+	for (addr = start; addr < end; ) {
+		uint64_t pt_idx = addr >> amdgpu_vm_block_size;
+		struct amdgpu_bo *pt = vm->page_tables[pt_idx].bo;
+		unsigned nptes;
+		uint64_t pte;
+		int r;
+
+		amdgpu_sync_resv(adev, &ib->sync, pt->tbo.resv, owner);
+		r = reservation_object_reserve_shared(pt->tbo.resv);
+		if (r)
+			return r;
+
+		if ((addr & ~mask) == (end & ~mask))
+			nptes = end - addr;
+		else
+			nptes = AMDGPU_VM_PTE_COUNT - (addr & mask);
+
+		pte = amdgpu_bo_gpu_offset(pt);
+		pte += (addr & mask) * 8;
+
+		if ((last_pte + 8 * count) != pte) {
+
+			if (count) {
+				amdgpu_vm_frag_ptes(adev, ib, last_pte,
+						    last_pte + 8 * count,
+						    last_dst, flags,
+						    gtt_flags);
+			}
+
+			count = nptes;
+			last_pte = pte;
+			last_dst = dst;
+		} else {
+			count += nptes;
+		}
+
+		addr += nptes;
+		dst += nptes * AMDGPU_GPU_PAGE_SIZE;
+	}
+
+	if (count) {
+		amdgpu_vm_frag_ptes(adev, ib, last_pte,
+				    last_pte + 8 * count,
+				    last_dst, flags, gtt_flags);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_update_mapping - update a mapping in the vm page table
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @mapping: mapped range and flags to use for the update
+ * @addr: addr to set the area to
+ * @gtt_flags: flags as they are used for GTT
+ * @fence: optional resulting fence
+ *
+ * Fill in the page table entries for @mapping.
+ * Returns 0 for success, -EINVAL for failure.
+ *
+ * Object have to be reserved and mutex must be locked!
+ */
+static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
+				       struct amdgpu_vm *vm,
+				       struct amdgpu_bo_va_mapping *mapping,
+				       uint64_t addr, uint32_t gtt_flags,
+				       struct fence **fence)
+{
+	struct amdgpu_ring *ring = adev->vm_manager.vm_pte_funcs_ring;
+	unsigned nptes, ncmds, ndw;
+	uint32_t flags = gtt_flags;
+	struct amdgpu_ib *ib;
+	struct fence *f = NULL;
+	int r;
+
+	/* normally,bo_va->flags only contians READABLE and WIRTEABLE bit go here
+	 * but in case of something, we filter the flags in first place
+	 */
+	if (!(mapping->flags & AMDGPU_PTE_READABLE))
+		flags &= ~AMDGPU_PTE_READABLE;
+	if (!(mapping->flags & AMDGPU_PTE_WRITEABLE))
+		flags &= ~AMDGPU_PTE_WRITEABLE;
+
+	trace_amdgpu_vm_bo_update(mapping);
+
+	nptes = mapping->it.last - mapping->it.start + 1;
+
+	/*
+	 * reserve space for one command every (1 << BLOCK_SIZE)
+	 *  entries or 2k dwords (whatever is smaller)
+	 */
+	ncmds = (nptes >> min(amdgpu_vm_block_size, 11)) + 1;
+
+	/* padding, etc. */
+	ndw = 64;
+
+	if ((flags & AMDGPU_PTE_SYSTEM) && (flags == gtt_flags)) {
+		/* only copy commands needed */
+		ndw += ncmds * 7;
+
+	} else if (flags & AMDGPU_PTE_SYSTEM) {
+		/* header for write data commands */
+		ndw += ncmds * 4;
+
+		/* body of write data command */
+		ndw += nptes * 2;
+
+	} else {
+		/* set page commands needed */
+		ndw += ncmds * 10;
+
+		/* two extra commands for begin/end of fragment */
+		ndw += 2 * 10;
+	}
+
+	/* update too big for an IB */
+	if (ndw > 0xfffff)
+		return -ENOMEM;
+
+	ib = kzalloc(sizeof(struct amdgpu_ib), GFP_KERNEL);
+	if (!ib)
+		return -ENOMEM;
+
+	r = amdgpu_ib_get(ring, NULL, ndw * 4, ib);
+	if (r) {
+		kfree(ib);
+		return r;
+	}
+
+	ib->length_dw = 0;
+
+	r = amdgpu_vm_update_ptes(adev, vm, ib, mapping->it.start,
+				  mapping->it.last + 1, addr + mapping->offset,
+				  flags, gtt_flags);
+
+	if (r) {
+		amdgpu_ib_free(adev, ib);
+		kfree(ib);
+		return r;
+	}
+
+	amdgpu_vm_pad_ib(adev, ib);
+	WARN_ON(ib->length_dw > ndw);
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, ib, 1,
+						 &amdgpu_vm_free_job,
+						 AMDGPU_FENCE_OWNER_VM,
+						 &f);
+	if (r)
+		goto error_free;
+
+	amdgpu_bo_fence(vm->page_directory, f, true);
+	if (fence) {
+		fence_put(*fence);
+		*fence = fence_get(f);
+	}
+	fence_put(f);
+	if (!amdgpu_enable_scheduler) {
+		amdgpu_ib_free(adev, ib);
+		kfree(ib);
+	}
+	return 0;
+
+error_free:
+	amdgpu_ib_free(adev, ib);
+	kfree(ib);
+	return r;
+}
+
+/**
+ * amdgpu_vm_bo_update - update all BO mappings in the vm page table
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: requested BO and VM object
+ * @mem: ttm mem
+ *
+ * Fill in the page table entries for @bo_va.
+ * Returns 0 for success, -EINVAL for failure.
+ *
+ * Object have to be reserved and mutex must be locked!
+ */
+int amdgpu_vm_bo_update(struct amdgpu_device *adev,
+			struct amdgpu_bo_va *bo_va,
+			struct ttm_mem_reg *mem)
+{
+	struct amdgpu_vm *vm = bo_va->vm;
+	struct amdgpu_bo_va_mapping *mapping;
+	uint32_t flags;
+	uint64_t addr;
+	int r;
+
+	if (mem) {
+		addr = (u64)mem->start << PAGE_SHIFT;
+		if (mem->mem_type != TTM_PL_TT)
+			addr += adev->vm_manager.vram_base_offset;
+	} else {
+		addr = 0;
+	}
+
+	flags = amdgpu_ttm_tt_pte_flags(adev, bo_va->bo->tbo.ttm, mem);
+
+	spin_lock(&vm->status_lock);
+	if (!list_empty(&bo_va->vm_status))
+		list_splice_init(&bo_va->valids, &bo_va->invalids);
+	spin_unlock(&vm->status_lock);
+
+	list_for_each_entry(mapping, &bo_va->invalids, list) {
+		r = amdgpu_vm_bo_update_mapping(adev, vm, mapping, addr,
+						flags, &bo_va->last_pt_update);
+		if (r)
+			return r;
+	}
+
+	if (trace_amdgpu_vm_bo_mapping_enabled()) {
+		list_for_each_entry(mapping, &bo_va->valids, list)
+			trace_amdgpu_vm_bo_mapping(mapping);
+
+		list_for_each_entry(mapping, &bo_va->invalids, list)
+			trace_amdgpu_vm_bo_mapping(mapping);
+	}
+
+	spin_lock(&vm->status_lock);
+	list_splice_init(&bo_va->invalids, &bo_va->valids);
+	list_del_init(&bo_va->vm_status);
+	if (!mem)
+		list_add(&bo_va->vm_status, &vm->cleared);
+	spin_unlock(&vm->status_lock);
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_clear_freed - clear freed BOs in the PT
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Make sure all freed BOs are cleared in the PT.
+ * Returns 0 for success.
+ *
+ * PTs have to be reserved and mutex must be locked!
+ */
+int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
+			  struct amdgpu_vm *vm)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	int r;
+
+	spin_lock(&vm->freed_lock);
+	while (!list_empty(&vm->freed)) {
+		mapping = list_first_entry(&vm->freed,
+			struct amdgpu_bo_va_mapping, list);
+		list_del(&mapping->list);
+		spin_unlock(&vm->freed_lock);
+		r = amdgpu_vm_bo_update_mapping(adev, vm, mapping, 0, 0, NULL);
+		kfree(mapping);
+		if (r)
+			return r;
+
+		spin_lock(&vm->freed_lock);
+	}
+	spin_unlock(&vm->freed_lock);
+
+	return 0;
+
+}
+
+/**
+ * amdgpu_vm_clear_invalids - clear invalidated BOs in the PT
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Make sure all invalidated BOs are cleared in the PT.
+ * Returns 0 for success.
+ *
+ * PTs have to be reserved and mutex must be locked!
+ */
+int amdgpu_vm_clear_invalids(struct amdgpu_device *adev,
+			     struct amdgpu_vm *vm, struct amdgpu_sync *sync)
+{
+	struct amdgpu_bo_va *bo_va = NULL;
+	int r = 0;
+
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->invalidated)) {
+		bo_va = list_first_entry(&vm->invalidated,
+			struct amdgpu_bo_va, vm_status);
+		spin_unlock(&vm->status_lock);
+		mutex_lock(&bo_va->mutex);
+		r = amdgpu_vm_bo_update(adev, bo_va, NULL);
+		mutex_unlock(&bo_va->mutex);
+		if (r)
+			return r;
+
+		spin_lock(&vm->status_lock);
+	}
+	spin_unlock(&vm->status_lock);
+
+	if (bo_va)
+		r = amdgpu_sync_fence(adev, sync, bo_va->last_pt_update);
+
+	return r;
+}
+
+/**
+ * amdgpu_vm_bo_add - add a bo to a specific vm
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @bo: amdgpu buffer object
+ *
+ * Add @bo into the requested vm (cayman+).
+ * Add @bo to the list of bos associated with the vm
+ * Returns newly added bo_va or NULL for failure
+ *
+ * Object has to be reserved!
+ */
+struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
+				      struct amdgpu_vm *vm,
+				      struct amdgpu_bo *bo)
+{
+	struct amdgpu_bo_va *bo_va;
+
+	bo_va = kzalloc(sizeof(struct amdgpu_bo_va), GFP_KERNEL);
+	if (bo_va == NULL) {
+		return NULL;
+	}
+	bo_va->vm = vm;
+	bo_va->bo = bo;
+	bo_va->ref_count = 1;
+	INIT_LIST_HEAD(&bo_va->bo_list);
+	INIT_LIST_HEAD(&bo_va->valids);
+	INIT_LIST_HEAD(&bo_va->invalids);
+	INIT_LIST_HEAD(&bo_va->vm_status);
+	mutex_init(&bo_va->mutex);
+	list_add_tail(&bo_va->bo_list, &bo->va);
+
+	return bo_va;
+}
+
+/**
+ * amdgpu_vm_bo_map - map bo inside a vm
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to store the address
+ * @saddr: where to map the BO
+ * @offset: requested offset in the BO
+ * @flags: attributes of pages (read/write/valid/etc.)
+ *
+ * Add a mapping of the BO at the specefied addr into the VM.
+ * Returns 0 for success, error for failure.
+ *
+ * Object has to be reserved and unreserved outside!
+ */
+int amdgpu_vm_bo_map(struct amdgpu_device *adev,
+		     struct amdgpu_bo_va *bo_va,
+		     uint64_t saddr, uint64_t offset,
+		     uint64_t size, uint32_t flags)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_vm *vm = bo_va->vm;
+	struct interval_tree_node *it;
+	unsigned last_pfn, pt_idx;
+	uint64_t eaddr;
+	int r;
+
+	/* validate the parameters */
+	if (saddr & AMDGPU_GPU_PAGE_MASK || offset & AMDGPU_GPU_PAGE_MASK ||
+	    size == 0 || size & AMDGPU_GPU_PAGE_MASK)
+		return -EINVAL;
+
+	/* make sure object fit at this offset */
+	eaddr = saddr + size - 1;
+	if ((saddr >= eaddr) || (offset + size > amdgpu_bo_size(bo_va->bo)))
+		return -EINVAL;
+
+	last_pfn = eaddr / AMDGPU_GPU_PAGE_SIZE;
+	if (last_pfn >= adev->vm_manager.max_pfn) {
+		dev_err(adev->dev, "va above limit (0x%08X >= 0x%08X)\n",
+			last_pfn, adev->vm_manager.max_pfn);
+		return -EINVAL;
+	}
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	eaddr /= AMDGPU_GPU_PAGE_SIZE;
+
+	spin_lock(&vm->it_lock);
+	it = interval_tree_iter_first(&vm->va, saddr, eaddr);
+	spin_unlock(&vm->it_lock);
+	if (it) {
+		struct amdgpu_bo_va_mapping *tmp;
+		tmp = container_of(it, struct amdgpu_bo_va_mapping, it);
+		/* bo and tmp overlap, invalid addr */
+		dev_err(adev->dev, "bo %p va 0x%010Lx-0x%010Lx conflict with "
+			"0x%010lx-0x%010lx\n", bo_va->bo, saddr, eaddr,
+			tmp->it.start, tmp->it.last + 1);
+		r = -EINVAL;
+		goto error;
+	}
+
+	mapping = kmalloc(sizeof(*mapping), GFP_KERNEL);
+	if (!mapping) {
+		r = -ENOMEM;
+		goto error;
+	}
+
+	INIT_LIST_HEAD(&mapping->list);
+	mapping->it.start = saddr;
+	mapping->it.last = eaddr;
+	mapping->offset = offset;
+	mapping->flags = flags;
+
+	mutex_lock(&bo_va->mutex);
+	list_add(&mapping->list, &bo_va->invalids);
+	mutex_unlock(&bo_va->mutex);
+	spin_lock(&vm->it_lock);
+	interval_tree_insert(&mapping->it, &vm->va);
+	spin_unlock(&vm->it_lock);
+	trace_amdgpu_vm_bo_map(bo_va, mapping);
+
+	/* Make sure the page tables are allocated */
+	saddr >>= amdgpu_vm_block_size;
+	eaddr >>= amdgpu_vm_block_size;
+
+	BUG_ON(eaddr >= amdgpu_vm_num_pdes(adev));
+
+	if (eaddr > vm->max_pde_used)
+		vm->max_pde_used = eaddr;
+
+	/* walk over the address space and allocate the page tables */
+	for (pt_idx = saddr; pt_idx <= eaddr; ++pt_idx) {
+		struct reservation_object *resv = vm->page_directory->tbo.resv;
+		struct amdgpu_bo *pt;
+
+		if (vm->page_tables[pt_idx].bo)
+			continue;
+
+		r = amdgpu_bo_create(adev, AMDGPU_VM_PTE_COUNT * 8,
+				     AMDGPU_GPU_PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_VRAM,
+				     AMDGPU_GEM_CREATE_NO_CPU_ACCESS,
+				     NULL, resv, &pt);
+		if (r)
+			goto error_free;
+
+		/* Keep a reference to the page table to avoid freeing
+		 * them up in the wrong order.
+		 */
+		pt->parent = amdgpu_bo_ref(vm->page_directory);
+
+		r = amdgpu_vm_clear_bo(adev, pt);
+		if (r) {
+			amdgpu_bo_unref(&pt);
+			goto error_free;
+		}
+
+		vm->page_tables[pt_idx].addr = 0;
+		vm->page_tables[pt_idx].bo = pt;
+	}
+
+	return 0;
+
+error_free:
+	list_del(&mapping->list);
+	spin_lock(&vm->it_lock);
+	interval_tree_remove(&mapping->it, &vm->va);
+	spin_unlock(&vm->it_lock);
+	trace_amdgpu_vm_bo_unmap(bo_va, mapping);
+	kfree(mapping);
+
+error:
+	return r;
+}
+
+/**
+ * amdgpu_vm_bo_unmap - remove bo mapping from vm
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: bo_va to remove the address from
+ * @saddr: where to the BO is mapped
+ *
+ * Remove a mapping of the BO at the specefied addr from the VM.
+ * Returns 0 for success, error for failure.
+ *
+ * Object has to be reserved and unreserved outside!
+ */
+int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
+		       struct amdgpu_bo_va *bo_va,
+		       uint64_t saddr)
+{
+	struct amdgpu_bo_va_mapping *mapping;
+	struct amdgpu_vm *vm = bo_va->vm;
+	bool valid = true;
+
+	saddr /= AMDGPU_GPU_PAGE_SIZE;
+	mutex_lock(&bo_va->mutex);
+	list_for_each_entry(mapping, &bo_va->valids, list) {
+		if (mapping->it.start == saddr)
+			break;
+	}
+
+	if (&mapping->list == &bo_va->valids) {
+		valid = false;
+
+		list_for_each_entry(mapping, &bo_va->invalids, list) {
+			if (mapping->it.start == saddr)
+				break;
+		}
+
+		if (&mapping->list == &bo_va->invalids) {
+			mutex_unlock(&bo_va->mutex);
+			return -ENOENT;
+		}
+	}
+	mutex_unlock(&bo_va->mutex);
+	list_del(&mapping->list);
+	spin_lock(&vm->it_lock);
+	interval_tree_remove(&mapping->it, &vm->va);
+	spin_unlock(&vm->it_lock);
+	trace_amdgpu_vm_bo_unmap(bo_va, mapping);
+
+	if (valid) {
+		spin_lock(&vm->freed_lock);
+		list_add(&mapping->list, &vm->freed);
+		spin_unlock(&vm->freed_lock);
+	} else {
+		kfree(mapping);
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_bo_rmv - remove a bo to a specific vm
+ *
+ * @adev: amdgpu_device pointer
+ * @bo_va: requested bo_va
+ *
+ * Remove @bo_va->bo from the requested vm (cayman+).
+ *
+ * Object have to be reserved!
+ */
+void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
+		      struct amdgpu_bo_va *bo_va)
+{
+	struct amdgpu_bo_va_mapping *mapping, *next;
+	struct amdgpu_vm *vm = bo_va->vm;
+
+	list_del(&bo_va->bo_list);
+
+	spin_lock(&vm->status_lock);
+	list_del(&bo_va->vm_status);
+	spin_unlock(&vm->status_lock);
+
+	list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
+		list_del(&mapping->list);
+		spin_lock(&vm->it_lock);
+		interval_tree_remove(&mapping->it, &vm->va);
+		spin_unlock(&vm->it_lock);
+		trace_amdgpu_vm_bo_unmap(bo_va, mapping);
+		spin_lock(&vm->freed_lock);
+		list_add(&mapping->list, &vm->freed);
+		spin_unlock(&vm->freed_lock);
+	}
+	list_for_each_entry_safe(mapping, next, &bo_va->invalids, list) {
+		list_del(&mapping->list);
+		spin_lock(&vm->it_lock);
+		interval_tree_remove(&mapping->it, &vm->va);
+		spin_unlock(&vm->it_lock);
+		kfree(mapping);
+	}
+	fence_put(bo_va->last_pt_update);
+	mutex_destroy(&bo_va->mutex);
+	kfree(bo_va);
+}
+
+/**
+ * amdgpu_vm_bo_invalidate - mark the bo as invalid
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ * @bo: amdgpu buffer object
+ *
+ * Mark @bo as invalid (cayman+).
+ */
+void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
+			     struct amdgpu_bo *bo)
+{
+	struct amdgpu_bo_va *bo_va;
+
+	list_for_each_entry(bo_va, &bo->va, bo_list) {
+		spin_lock(&bo_va->vm->status_lock);
+		if (list_empty(&bo_va->vm_status))
+			list_add(&bo_va->vm_status, &bo_va->vm->invalidated);
+		spin_unlock(&bo_va->vm->status_lock);
+	}
+}
+
+/**
+ * amdgpu_vm_init - initialize a vm instance
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Init @vm fields (cayman+).
+ */
+int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	const unsigned align = min(AMDGPU_VM_PTB_ALIGN_SIZE,
+		AMDGPU_VM_PTE_COUNT * 8);
+	unsigned pd_size, pd_entries;
+	int i, r;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		vm->ids[i].id = 0;
+		vm->ids[i].flushed_updates = NULL;
+	}
+	vm->va = RB_ROOT;
+	spin_lock_init(&vm->status_lock);
+	INIT_LIST_HEAD(&vm->invalidated);
+	INIT_LIST_HEAD(&vm->cleared);
+	INIT_LIST_HEAD(&vm->freed);
+	spin_lock_init(&vm->it_lock);
+	spin_lock_init(&vm->freed_lock);
+	pd_size = amdgpu_vm_directory_size(adev);
+	pd_entries = amdgpu_vm_num_pdes(adev);
+
+	/* allocate page table array */
+	vm->page_tables = drm_calloc_large(pd_entries, sizeof(struct amdgpu_vm_pt));
+	if (vm->page_tables == NULL) {
+		DRM_ERROR("Cannot allocate memory for page table array\n");
+		return -ENOMEM;
+	}
+
+	vm->page_directory_fence = NULL;
+
+	r = amdgpu_bo_create(adev, pd_size, align, true,
+			     AMDGPU_GEM_DOMAIN_VRAM,
+			     AMDGPU_GEM_CREATE_NO_CPU_ACCESS,
+			     NULL, NULL, &vm->page_directory);
+	if (r)
+		return r;
+	r = amdgpu_bo_reserve(vm->page_directory, false);
+	if (r) {
+		amdgpu_bo_unref(&vm->page_directory);
+		vm->page_directory = NULL;
+		return r;
+	}
+	r = amdgpu_vm_clear_bo(adev, vm->page_directory);
+	amdgpu_bo_unreserve(vm->page_directory);
+	if (r) {
+		amdgpu_bo_unref(&vm->page_directory);
+		vm->page_directory = NULL;
+		return r;
+	}
+
+	return 0;
+}
+
+/**
+ * amdgpu_vm_fini - tear down a vm instance
+ *
+ * @adev: amdgpu_device pointer
+ * @vm: requested vm
+ *
+ * Tear down @vm (cayman+).
+ * Unbind the VM and remove all bos from the vm bo list
+ */
+void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	struct amdgpu_bo_va_mapping *mapping, *tmp;
+	int i;
+
+	if (!RB_EMPTY_ROOT(&vm->va)) {
+		dev_err(adev->dev, "still active bo inside vm\n");
+	}
+	rbtree_postorder_for_each_entry_safe(mapping, tmp, &vm->va, it.rb) {
+		list_del(&mapping->list);
+		interval_tree_remove(&mapping->it, &vm->va);
+		kfree(mapping);
+	}
+	list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
+		list_del(&mapping->list);
+		kfree(mapping);
+	}
+
+	for (i = 0; i < amdgpu_vm_num_pdes(adev); i++)
+		amdgpu_bo_unref(&vm->page_tables[i].bo);
+	drm_free_large(vm->page_tables);
+
+	amdgpu_bo_unref(&vm->page_directory);
+	fence_put(vm->page_directory_fence);
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		unsigned id = vm->ids[i].id;
+
+		atomic_long_cmpxchg(&adev->vm_manager.ids[id].owner,
+				    (long)vm, 0);
+		fence_put(vm->ids[i].flushed_updates);
+	}
+
+}
+
+/**
+ * amdgpu_vm_manager_fini - cleanup VM manager
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Cleanup the VM manager and free resources.
+ */
+void amdgpu_vm_manager_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	for (i = 0; i < AMDGPU_NUM_VM; ++i)
+		fence_put(adev->vm_manager.ids[i].active);
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atom.c b/kernel/drivers/gpu/drm/amd/amdgpu/atom.c
new file mode 100644
index 000000000..1b50e6c13
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atom.c
@@ -0,0 +1,1457 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Stanislaw Skowronek
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <asm/unaligned.h>
+
+#define ATOM_DEBUG
+
+#include "atom.h"
+#include "atom-names.h"
+#include "atom-bits.h"
+#include "amdgpu.h"
+
+#define ATOM_COND_ABOVE		0
+#define ATOM_COND_ABOVEOREQUAL	1
+#define ATOM_COND_ALWAYS	2
+#define ATOM_COND_BELOW		3
+#define ATOM_COND_BELOWOREQUAL	4
+#define ATOM_COND_EQUAL		5
+#define ATOM_COND_NOTEQUAL	6
+
+#define ATOM_PORT_ATI	0
+#define ATOM_PORT_PCI	1
+#define ATOM_PORT_SYSIO	2
+
+#define ATOM_UNIT_MICROSEC	0
+#define ATOM_UNIT_MILLISEC	1
+
+#define PLL_INDEX	2
+#define PLL_DATA	3
+
+typedef struct {
+	struct atom_context *ctx;
+	uint32_t *ps, *ws;
+	int ps_shift;
+	uint16_t start;
+	unsigned last_jump;
+	unsigned long last_jump_jiffies;
+	bool abort;
+} atom_exec_context;
+
+int amdgpu_atom_debug = 0;
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params);
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params);
+
+static uint32_t atom_arg_mask[8] =
+    { 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000,
+0xFF000000 };
+static int atom_arg_shift[8] = { 0, 0, 8, 16, 0, 8, 16, 24 };
+
+static int atom_dst_to_src[8][4] = {
+	/* translate destination alignment field to the source alignment encoding */
+	{0, 0, 0, 0},
+	{1, 2, 3, 0},
+	{1, 2, 3, 0},
+	{1, 2, 3, 0},
+	{4, 5, 6, 7},
+	{4, 5, 6, 7},
+	{4, 5, 6, 7},
+	{4, 5, 6, 7},
+};
+static int atom_def_dst[8] = { 0, 0, 1, 2, 0, 1, 2, 3 };
+
+static int debug_depth = 0;
+#ifdef ATOM_DEBUG
+static void debug_print_spaces(int n)
+{
+	while (n--)
+		printk("   ");
+}
+
+#define DEBUG(...) do if (amdgpu_atom_debug) { printk(KERN_DEBUG __VA_ARGS__); } while (0)
+#define SDEBUG(...) do if (amdgpu_atom_debug) { printk(KERN_DEBUG); debug_print_spaces(debug_depth); printk(__VA_ARGS__); } while (0)
+#else
+#define DEBUG(...) do { } while (0)
+#define SDEBUG(...) do { } while (0)
+#endif
+
+static uint32_t atom_iio_execute(struct atom_context *ctx, int base,
+				 uint32_t index, uint32_t data)
+{
+	uint32_t temp = 0xCDCDCDCD;
+
+	while (1)
+		switch (CU8(base)) {
+		case ATOM_IIO_NOP:
+			base++;
+			break;
+		case ATOM_IIO_READ:
+			temp = ctx->card->ioreg_read(ctx->card, CU16(base + 1));
+			base += 3;
+			break;
+		case ATOM_IIO_WRITE:
+			ctx->card->ioreg_write(ctx->card, CU16(base + 1), temp);
+			base += 3;
+			break;
+		case ATOM_IIO_CLEAR:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 2));
+			base += 3;
+			break;
+		case ATOM_IIO_SET:
+			temp |=
+			    (0xFFFFFFFF >> (32 - CU8(base + 1))) << CU8(base +
+									2);
+			base += 3;
+			break;
+		case ATOM_IIO_MOVE_INDEX:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 3));
+			temp |=
+			    ((index >> CU8(base + 2)) &
+			     (0xFFFFFFFF >> (32 - CU8(base + 1)))) << CU8(base +
+									  3);
+			base += 4;
+			break;
+		case ATOM_IIO_MOVE_DATA:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 3));
+			temp |=
+			    ((data >> CU8(base + 2)) &
+			     (0xFFFFFFFF >> (32 - CU8(base + 1)))) << CU8(base +
+									  3);
+			base += 4;
+			break;
+		case ATOM_IIO_MOVE_ATTR:
+			temp &=
+			    ~((0xFFFFFFFF >> (32 - CU8(base + 1))) <<
+			      CU8(base + 3));
+			temp |=
+			    ((ctx->
+			      io_attr >> CU8(base + 2)) & (0xFFFFFFFF >> (32 -
+									  CU8
+									  (base
+									   +
+									   1))))
+			    << CU8(base + 3);
+			base += 4;
+			break;
+		case ATOM_IIO_END:
+			return temp;
+		default:
+			printk(KERN_INFO "Unknown IIO opcode.\n");
+			return 0;
+		}
+}
+
+static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
+				 int *ptr, uint32_t *saved, int print)
+{
+	uint32_t idx, val = 0xCDCDCDCD, align, arg;
+	struct atom_context *gctx = ctx->ctx;
+	arg = attr & 7;
+	align = (attr >> 3) & 7;
+	switch (arg) {
+	case ATOM_ARG_REG:
+		idx = U16(*ptr);
+		(*ptr) += 2;
+		if (print)
+			DEBUG("REG[0x%04X]", idx);
+		idx += gctx->reg_block;
+		switch (gctx->io_mode) {
+		case ATOM_IO_MM:
+			val = gctx->card->reg_read(gctx->card, idx);
+			break;
+		case ATOM_IO_PCI:
+			printk(KERN_INFO
+			       "PCI registers are not implemented.\n");
+			return 0;
+		case ATOM_IO_SYSIO:
+			printk(KERN_INFO
+			       "SYSIO registers are not implemented.\n");
+			return 0;
+		default:
+			if (!(gctx->io_mode & 0x80)) {
+				printk(KERN_INFO "Bad IO mode.\n");
+				return 0;
+			}
+			if (!gctx->iio[gctx->io_mode & 0x7F]) {
+				printk(KERN_INFO
+				       "Undefined indirect IO read method %d.\n",
+				       gctx->io_mode & 0x7F);
+				return 0;
+			}
+			val =
+			    atom_iio_execute(gctx,
+					     gctx->iio[gctx->io_mode & 0x7F],
+					     idx, 0);
+		}
+		break;
+	case ATOM_ARG_PS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		/* get_unaligned_le32 avoids unaligned accesses from atombios
+		 * tables, noticed on a DEC Alpha. */
+		val = get_unaligned_le32((u32 *)&ctx->ps[idx]);
+		if (print)
+			DEBUG("PS[0x%02X,0x%04X]", idx, val);
+		break;
+	case ATOM_ARG_WS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if (print)
+			DEBUG("WS[0x%02X]", idx);
+		switch (idx) {
+		case ATOM_WS_QUOTIENT:
+			val = gctx->divmul[0];
+			break;
+		case ATOM_WS_REMAINDER:
+			val = gctx->divmul[1];
+			break;
+		case ATOM_WS_DATAPTR:
+			val = gctx->data_block;
+			break;
+		case ATOM_WS_SHIFT:
+			val = gctx->shift;
+			break;
+		case ATOM_WS_OR_MASK:
+			val = 1 << gctx->shift;
+			break;
+		case ATOM_WS_AND_MASK:
+			val = ~(1 << gctx->shift);
+			break;
+		case ATOM_WS_FB_WINDOW:
+			val = gctx->fb_base;
+			break;
+		case ATOM_WS_ATTRIBUTES:
+			val = gctx->io_attr;
+			break;
+		case ATOM_WS_REGPTR:
+			val = gctx->reg_block;
+			break;
+		default:
+			val = ctx->ws[idx];
+		}
+		break;
+	case ATOM_ARG_ID:
+		idx = U16(*ptr);
+		(*ptr) += 2;
+		if (print) {
+			if (gctx->data_block)
+				DEBUG("ID[0x%04X+%04X]", idx, gctx->data_block);
+			else
+				DEBUG("ID[0x%04X]", idx);
+		}
+		val = U32(idx + gctx->data_block);
+		break;
+	case ATOM_ARG_FB:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if ((gctx->fb_base + (idx * 4)) > gctx->scratch_size_bytes) {
+			DRM_ERROR("ATOM: fb read beyond scratch region: %d vs. %d\n",
+				  gctx->fb_base + (idx * 4), gctx->scratch_size_bytes);
+			val = 0;
+		} else
+			val = gctx->scratch[(gctx->fb_base / 4) + idx];
+		if (print)
+			DEBUG("FB[0x%02X]", idx);
+		break;
+	case ATOM_ARG_IMM:
+		switch (align) {
+		case ATOM_SRC_DWORD:
+			val = U32(*ptr);
+			(*ptr) += 4;
+			if (print)
+				DEBUG("IMM 0x%08X\n", val);
+			return val;
+		case ATOM_SRC_WORD0:
+		case ATOM_SRC_WORD8:
+		case ATOM_SRC_WORD16:
+			val = U16(*ptr);
+			(*ptr) += 2;
+			if (print)
+				DEBUG("IMM 0x%04X\n", val);
+			return val;
+		case ATOM_SRC_BYTE0:
+		case ATOM_SRC_BYTE8:
+		case ATOM_SRC_BYTE16:
+		case ATOM_SRC_BYTE24:
+			val = U8(*ptr);
+			(*ptr)++;
+			if (print)
+				DEBUG("IMM 0x%02X\n", val);
+			return val;
+		}
+		return 0;
+	case ATOM_ARG_PLL:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if (print)
+			DEBUG("PLL[0x%02X]", idx);
+		val = gctx->card->pll_read(gctx->card, idx);
+		break;
+	case ATOM_ARG_MC:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if (print)
+			DEBUG("MC[0x%02X]", idx);
+		val = gctx->card->mc_read(gctx->card, idx);
+		break;
+	}
+	if (saved)
+		*saved = val;
+	val &= atom_arg_mask[align];
+	val >>= atom_arg_shift[align];
+	if (print)
+		switch (align) {
+		case ATOM_SRC_DWORD:
+			DEBUG(".[31:0] -> 0x%08X\n", val);
+			break;
+		case ATOM_SRC_WORD0:
+			DEBUG(".[15:0] -> 0x%04X\n", val);
+			break;
+		case ATOM_SRC_WORD8:
+			DEBUG(".[23:8] -> 0x%04X\n", val);
+			break;
+		case ATOM_SRC_WORD16:
+			DEBUG(".[31:16] -> 0x%04X\n", val);
+			break;
+		case ATOM_SRC_BYTE0:
+			DEBUG(".[7:0] -> 0x%02X\n", val);
+			break;
+		case ATOM_SRC_BYTE8:
+			DEBUG(".[15:8] -> 0x%02X\n", val);
+			break;
+		case ATOM_SRC_BYTE16:
+			DEBUG(".[23:16] -> 0x%02X\n", val);
+			break;
+		case ATOM_SRC_BYTE24:
+			DEBUG(".[31:24] -> 0x%02X\n", val);
+			break;
+		}
+	return val;
+}
+
+static void atom_skip_src_int(atom_exec_context *ctx, uint8_t attr, int *ptr)
+{
+	uint32_t align = (attr >> 3) & 7, arg = attr & 7;
+	switch (arg) {
+	case ATOM_ARG_REG:
+	case ATOM_ARG_ID:
+		(*ptr) += 2;
+		break;
+	case ATOM_ARG_PLL:
+	case ATOM_ARG_MC:
+	case ATOM_ARG_PS:
+	case ATOM_ARG_WS:
+	case ATOM_ARG_FB:
+		(*ptr)++;
+		break;
+	case ATOM_ARG_IMM:
+		switch (align) {
+		case ATOM_SRC_DWORD:
+			(*ptr) += 4;
+			return;
+		case ATOM_SRC_WORD0:
+		case ATOM_SRC_WORD8:
+		case ATOM_SRC_WORD16:
+			(*ptr) += 2;
+			return;
+		case ATOM_SRC_BYTE0:
+		case ATOM_SRC_BYTE8:
+		case ATOM_SRC_BYTE16:
+		case ATOM_SRC_BYTE24:
+			(*ptr)++;
+			return;
+		}
+		return;
+	}
+}
+
+static uint32_t atom_get_src(atom_exec_context *ctx, uint8_t attr, int *ptr)
+{
+	return atom_get_src_int(ctx, attr, ptr, NULL, 1);
+}
+
+static uint32_t atom_get_src_direct(atom_exec_context *ctx, uint8_t align, int *ptr)
+{
+	uint32_t val = 0xCDCDCDCD;
+
+	switch (align) {
+	case ATOM_SRC_DWORD:
+		val = U32(*ptr);
+		(*ptr) += 4;
+		break;
+	case ATOM_SRC_WORD0:
+	case ATOM_SRC_WORD8:
+	case ATOM_SRC_WORD16:
+		val = U16(*ptr);
+		(*ptr) += 2;
+		break;
+	case ATOM_SRC_BYTE0:
+	case ATOM_SRC_BYTE8:
+	case ATOM_SRC_BYTE16:
+	case ATOM_SRC_BYTE24:
+		val = U8(*ptr);
+		(*ptr)++;
+		break;
+	}
+	return val;
+}
+
+static uint32_t atom_get_dst(atom_exec_context *ctx, int arg, uint8_t attr,
+			     int *ptr, uint32_t *saved, int print)
+{
+	return atom_get_src_int(ctx,
+				arg | atom_dst_to_src[(attr >> 3) &
+						      7][(attr >> 6) & 3] << 3,
+				ptr, saved, print);
+}
+
+static void atom_skip_dst(atom_exec_context *ctx, int arg, uint8_t attr, int *ptr)
+{
+	atom_skip_src_int(ctx,
+			  arg | atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) &
+								 3] << 3, ptr);
+}
+
+static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
+			 int *ptr, uint32_t val, uint32_t saved)
+{
+	uint32_t align =
+	    atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3], old_val =
+	    val, idx;
+	struct atom_context *gctx = ctx->ctx;
+	old_val &= atom_arg_mask[align] >> atom_arg_shift[align];
+	val <<= atom_arg_shift[align];
+	val &= atom_arg_mask[align];
+	saved &= ~atom_arg_mask[align];
+	val |= saved;
+	switch (arg) {
+	case ATOM_ARG_REG:
+		idx = U16(*ptr);
+		(*ptr) += 2;
+		DEBUG("REG[0x%04X]", idx);
+		idx += gctx->reg_block;
+		switch (gctx->io_mode) {
+		case ATOM_IO_MM:
+			if (idx == 0)
+				gctx->card->reg_write(gctx->card, idx,
+						      val << 2);
+			else
+				gctx->card->reg_write(gctx->card, idx, val);
+			break;
+		case ATOM_IO_PCI:
+			printk(KERN_INFO
+			       "PCI registers are not implemented.\n");
+			return;
+		case ATOM_IO_SYSIO:
+			printk(KERN_INFO
+			       "SYSIO registers are not implemented.\n");
+			return;
+		default:
+			if (!(gctx->io_mode & 0x80)) {
+				printk(KERN_INFO "Bad IO mode.\n");
+				return;
+			}
+			if (!gctx->iio[gctx->io_mode & 0xFF]) {
+				printk(KERN_INFO
+				       "Undefined indirect IO write method %d.\n",
+				       gctx->io_mode & 0x7F);
+				return;
+			}
+			atom_iio_execute(gctx, gctx->iio[gctx->io_mode & 0xFF],
+					 idx, val);
+		}
+		break;
+	case ATOM_ARG_PS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("PS[0x%02X]", idx);
+		ctx->ps[idx] = cpu_to_le32(val);
+		break;
+	case ATOM_ARG_WS:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("WS[0x%02X]", idx);
+		switch (idx) {
+		case ATOM_WS_QUOTIENT:
+			gctx->divmul[0] = val;
+			break;
+		case ATOM_WS_REMAINDER:
+			gctx->divmul[1] = val;
+			break;
+		case ATOM_WS_DATAPTR:
+			gctx->data_block = val;
+			break;
+		case ATOM_WS_SHIFT:
+			gctx->shift = val;
+			break;
+		case ATOM_WS_OR_MASK:
+		case ATOM_WS_AND_MASK:
+			break;
+		case ATOM_WS_FB_WINDOW:
+			gctx->fb_base = val;
+			break;
+		case ATOM_WS_ATTRIBUTES:
+			gctx->io_attr = val;
+			break;
+		case ATOM_WS_REGPTR:
+			gctx->reg_block = val;
+			break;
+		default:
+			ctx->ws[idx] = val;
+		}
+		break;
+	case ATOM_ARG_FB:
+		idx = U8(*ptr);
+		(*ptr)++;
+		if ((gctx->fb_base + (idx * 4)) > gctx->scratch_size_bytes) {
+			DRM_ERROR("ATOM: fb write beyond scratch region: %d vs. %d\n",
+				  gctx->fb_base + (idx * 4), gctx->scratch_size_bytes);
+		} else
+			gctx->scratch[(gctx->fb_base / 4) + idx] = val;
+		DEBUG("FB[0x%02X]", idx);
+		break;
+	case ATOM_ARG_PLL:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("PLL[0x%02X]", idx);
+		gctx->card->pll_write(gctx->card, idx, val);
+		break;
+	case ATOM_ARG_MC:
+		idx = U8(*ptr);
+		(*ptr)++;
+		DEBUG("MC[0x%02X]", idx);
+		gctx->card->mc_write(gctx->card, idx, val);
+		return;
+	}
+	switch (align) {
+	case ATOM_SRC_DWORD:
+		DEBUG(".[31:0] <- 0x%08X\n", old_val);
+		break;
+	case ATOM_SRC_WORD0:
+		DEBUG(".[15:0] <- 0x%04X\n", old_val);
+		break;
+	case ATOM_SRC_WORD8:
+		DEBUG(".[23:8] <- 0x%04X\n", old_val);
+		break;
+	case ATOM_SRC_WORD16:
+		DEBUG(".[31:16] <- 0x%04X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE0:
+		DEBUG(".[7:0] <- 0x%02X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE8:
+		DEBUG(".[15:8] <- 0x%02X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE16:
+		DEBUG(".[23:16] <- 0x%02X\n", old_val);
+		break;
+	case ATOM_SRC_BYTE24:
+		DEBUG(".[31:24] <- 0x%02X\n", old_val);
+		break;
+	}
+}
+
+static void atom_op_add(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst += src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_and(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst &= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_beep(atom_exec_context *ctx, int *ptr, int arg)
+{
+	printk("ATOM BIOS beeped!\n");
+}
+
+static void atom_op_calltable(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int idx = U8((*ptr)++);
+	int r = 0;
+
+	if (idx < ATOM_TABLE_NAMES_CNT)
+		SDEBUG("   table: %d (%s)\n", idx, atom_table_names[idx]);
+	else
+		SDEBUG("   table: %d\n", idx);
+	if (U16(ctx->ctx->cmd_table + 4 + 2 * idx))
+		r = amdgpu_atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift);
+	if (r) {
+		ctx->abort = true;
+	}
+}
+
+static void atom_op_clear(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t saved;
+	int dptr = *ptr;
+	attr &= 0x38;
+	attr |= atom_def_dst[attr >> 3] << 6;
+	atom_get_dst(ctx, arg, attr, ptr, &saved, 0);
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, 0, saved);
+}
+
+static void atom_op_compare(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	ctx->ctx->cs_equal = (dst == src);
+	ctx->ctx->cs_above = (dst > src);
+	SDEBUG("   result: %s %s\n", ctx->ctx->cs_equal ? "EQ" : "NE",
+	       ctx->ctx->cs_above ? "GT" : "LE");
+}
+
+static void atom_op_delay(atom_exec_context *ctx, int *ptr, int arg)
+{
+	unsigned count = U8((*ptr)++);
+	SDEBUG("   count: %d\n", count);
+	if (arg == ATOM_UNIT_MICROSEC)
+		udelay(count);
+	else if (!drm_can_sleep())
+		mdelay(count);
+	else
+		msleep(count);
+}
+
+static void atom_op_div(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	if (src != 0) {
+		ctx->ctx->divmul[0] = dst / src;
+		ctx->ctx->divmul[1] = dst % src;
+	} else {
+		ctx->ctx->divmul[0] = 0;
+		ctx->ctx->divmul[1] = 0;
+	}
+}
+
+static void atom_op_div32(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint64_t val64;
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	if (src != 0) {
+		val64 = dst;
+		val64 |= ((uint64_t)ctx->ctx->divmul[1]) << 32;
+		do_div(val64, src);
+		ctx->ctx->divmul[0] = lower_32_bits(val64);
+		ctx->ctx->divmul[1] = upper_32_bits(val64);
+	} else {
+		ctx->ctx->divmul[0] = 0;
+		ctx->ctx->divmul[1] = 0;
+	}
+}
+
+static void atom_op_eot(atom_exec_context *ctx, int *ptr, int arg)
+{
+	/* functionally, a nop */
+}
+
+static void atom_op_jump(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int execute = 0, target = U16(*ptr);
+	unsigned long cjiffies;
+
+	(*ptr) += 2;
+	switch (arg) {
+	case ATOM_COND_ABOVE:
+		execute = ctx->ctx->cs_above;
+		break;
+	case ATOM_COND_ABOVEOREQUAL:
+		execute = ctx->ctx->cs_above || ctx->ctx->cs_equal;
+		break;
+	case ATOM_COND_ALWAYS:
+		execute = 1;
+		break;
+	case ATOM_COND_BELOW:
+		execute = !(ctx->ctx->cs_above || ctx->ctx->cs_equal);
+		break;
+	case ATOM_COND_BELOWOREQUAL:
+		execute = !ctx->ctx->cs_above;
+		break;
+	case ATOM_COND_EQUAL:
+		execute = ctx->ctx->cs_equal;
+		break;
+	case ATOM_COND_NOTEQUAL:
+		execute = !ctx->ctx->cs_equal;
+		break;
+	}
+	if (arg != ATOM_COND_ALWAYS)
+		SDEBUG("   taken: %s\n", execute ? "yes" : "no");
+	SDEBUG("   target: 0x%04X\n", target);
+	if (execute) {
+		if (ctx->last_jump == (ctx->start + target)) {
+			cjiffies = jiffies;
+			if (time_after(cjiffies, ctx->last_jump_jiffies)) {
+				cjiffies -= ctx->last_jump_jiffies;
+				if ((jiffies_to_msecs(cjiffies) > 5000)) {
+					DRM_ERROR("atombios stuck in loop for more than 5secs aborting\n");
+					ctx->abort = true;
+				}
+			} else {
+				/* jiffies wrap around we will just wait a little longer */
+				ctx->last_jump_jiffies = jiffies;
+			}
+		} else {
+			ctx->last_jump = ctx->start + target;
+			ctx->last_jump_jiffies = jiffies;
+		}
+		*ptr = ctx->start + target;
+	}
+}
+
+static void atom_op_mask(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, mask, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	mask = atom_get_src_direct(ctx, ((attr >> 3) & 7), ptr);
+	SDEBUG("   mask: 0x%08x", mask);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst &= mask;
+	dst |= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_move(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t src, saved;
+	int dptr = *ptr;
+	if (((attr >> 3) & 7) != ATOM_SRC_DWORD)
+		atom_get_dst(ctx, arg, attr, ptr, &saved, 0);
+	else {
+		atom_skip_dst(ctx, arg, attr, ptr);
+		saved = 0xCDCDCDCD;
+	}
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, src, saved);
+}
+
+static void atom_op_mul(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	ctx->ctx->divmul[0] = dst * src;
+}
+
+static void atom_op_mul32(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint64_t val64;
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	val64 = (uint64_t)dst * (uint64_t)src;
+	ctx->ctx->divmul[0] = lower_32_bits(val64);
+	ctx->ctx->divmul[1] = upper_32_bits(val64);
+}
+
+static void atom_op_nop(atom_exec_context *ctx, int *ptr, int arg)
+{
+	/* nothing */
+}
+
+static void atom_op_or(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst |= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_postcard(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t val = U8((*ptr)++);
+	SDEBUG("POST card output: 0x%02X\n", val);
+}
+
+static void atom_op_repeat(atom_exec_context *ctx, int *ptr, int arg)
+{
+	printk(KERN_INFO "unimplemented!\n");
+}
+
+static void atom_op_restorereg(atom_exec_context *ctx, int *ptr, int arg)
+{
+	printk(KERN_INFO "unimplemented!\n");
+}
+
+static void atom_op_savereg(atom_exec_context *ctx, int *ptr, int arg)
+{
+	printk(KERN_INFO "unimplemented!\n");
+}
+
+static void atom_op_setdatablock(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int idx = U8(*ptr);
+	(*ptr)++;
+	SDEBUG("   block: %d\n", idx);
+	if (!idx)
+		ctx->ctx->data_block = 0;
+	else if (idx == 255)
+		ctx->ctx->data_block = ctx->start;
+	else
+		ctx->ctx->data_block = U16(ctx->ctx->data_table + 4 + 2 * idx);
+	SDEBUG("   base: 0x%04X\n", ctx->ctx->data_block);
+}
+
+static void atom_op_setfbbase(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	SDEBUG("   fb_base: ");
+	ctx->ctx->fb_base = atom_get_src(ctx, attr, ptr);
+}
+
+static void atom_op_setport(atom_exec_context *ctx, int *ptr, int arg)
+{
+	int port;
+	switch (arg) {
+	case ATOM_PORT_ATI:
+		port = U16(*ptr);
+		if (port < ATOM_IO_NAMES_CNT)
+			SDEBUG("   port: %d (%s)\n", port, atom_io_names[port]);
+		else
+			SDEBUG("   port: %d\n", port);
+		if (!port)
+			ctx->ctx->io_mode = ATOM_IO_MM;
+		else
+			ctx->ctx->io_mode = ATOM_IO_IIO | port;
+		(*ptr) += 2;
+		break;
+	case ATOM_PORT_PCI:
+		ctx->ctx->io_mode = ATOM_IO_PCI;
+		(*ptr)++;
+		break;
+	case ATOM_PORT_SYSIO:
+		ctx->ctx->io_mode = ATOM_IO_SYSIO;
+		(*ptr)++;
+		break;
+	}
+}
+
+static void atom_op_setregblock(atom_exec_context *ctx, int *ptr, int arg)
+{
+	ctx->ctx->reg_block = U16(*ptr);
+	(*ptr) += 2;
+	SDEBUG("   base: 0x%04X\n", ctx->ctx->reg_block);
+}
+
+static void atom_op_shift_left(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	attr &= 0x38;
+	attr |= atom_def_dst[attr >> 3] << 6;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst <<= shift;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shift_right(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	attr &= 0x38;
+	attr |= atom_def_dst[attr >> 3] << 6;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst >>= shift;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shl(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	uint32_t dst_align = atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3];
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	/* op needs to full dst value */
+	dst = saved;
+	shift = atom_get_src(ctx, attr, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst <<= shift;
+	dst &= atom_arg_mask[dst_align];
+	dst >>= atom_arg_shift[dst_align];
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shr(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++), shift;
+	uint32_t saved, dst;
+	int dptr = *ptr;
+	uint32_t dst_align = atom_dst_to_src[(attr >> 3) & 7][(attr >> 6) & 3];
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	/* op needs to full dst value */
+	dst = saved;
+	shift = atom_get_src(ctx, attr, ptr);
+	SDEBUG("   shift: %d\n", shift);
+	dst >>= shift;
+	dst &= atom_arg_mask[dst_align];
+	dst >>= atom_arg_shift[dst_align];
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_sub(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst -= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_switch(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t src, val, target;
+	SDEBUG("   switch: ");
+	src = atom_get_src(ctx, attr, ptr);
+	while (U16(*ptr) != ATOM_CASE_END)
+		if (U8(*ptr) == ATOM_CASE_MAGIC) {
+			(*ptr)++;
+			SDEBUG("   case: ");
+			val =
+			    atom_get_src(ctx, (attr & 0x38) | ATOM_ARG_IMM,
+					 ptr);
+			target = U16(*ptr);
+			if (val == src) {
+				SDEBUG("   target: %04X\n", target);
+				*ptr = ctx->start + target;
+				return;
+			}
+			(*ptr) += 2;
+		} else {
+			printk(KERN_INFO "Bad case.\n");
+			return;
+		}
+	(*ptr) += 2;
+}
+
+static void atom_op_test(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src;
+	SDEBUG("   src1: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, NULL, 1);
+	SDEBUG("   src2: ");
+	src = atom_get_src(ctx, attr, ptr);
+	ctx->ctx->cs_equal = ((dst & src) == 0);
+	SDEBUG("   result: %s\n", ctx->ctx->cs_equal ? "EQ" : "NE");
+}
+
+static void atom_op_xor(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t attr = U8((*ptr)++);
+	uint32_t dst, src, saved;
+	int dptr = *ptr;
+	SDEBUG("   dst: ");
+	dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+	SDEBUG("   src: ");
+	src = atom_get_src(ctx, attr, ptr);
+	dst ^= src;
+	SDEBUG("   dst: ");
+	atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_debug(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint8_t val = U8((*ptr)++);
+	SDEBUG("DEBUG output: 0x%02X\n", val);
+}
+
+static void atom_op_processds(atom_exec_context *ctx, int *ptr, int arg)
+{
+	uint16_t val = U16(*ptr);
+	(*ptr) += val + 2;
+	SDEBUG("PROCESSDS output: 0x%02X\n", val);
+}
+
+static struct {
+	void (*func) (atom_exec_context *, int *, int);
+	int arg;
+} opcode_table[ATOM_OP_CNT] = {
+	{
+	NULL, 0}, {
+	atom_op_move, ATOM_ARG_REG}, {
+	atom_op_move, ATOM_ARG_PS}, {
+	atom_op_move, ATOM_ARG_WS}, {
+	atom_op_move, ATOM_ARG_FB}, {
+	atom_op_move, ATOM_ARG_PLL}, {
+	atom_op_move, ATOM_ARG_MC}, {
+	atom_op_and, ATOM_ARG_REG}, {
+	atom_op_and, ATOM_ARG_PS}, {
+	atom_op_and, ATOM_ARG_WS}, {
+	atom_op_and, ATOM_ARG_FB}, {
+	atom_op_and, ATOM_ARG_PLL}, {
+	atom_op_and, ATOM_ARG_MC}, {
+	atom_op_or, ATOM_ARG_REG}, {
+	atom_op_or, ATOM_ARG_PS}, {
+	atom_op_or, ATOM_ARG_WS}, {
+	atom_op_or, ATOM_ARG_FB}, {
+	atom_op_or, ATOM_ARG_PLL}, {
+	atom_op_or, ATOM_ARG_MC}, {
+	atom_op_shift_left, ATOM_ARG_REG}, {
+	atom_op_shift_left, ATOM_ARG_PS}, {
+	atom_op_shift_left, ATOM_ARG_WS}, {
+	atom_op_shift_left, ATOM_ARG_FB}, {
+	atom_op_shift_left, ATOM_ARG_PLL}, {
+	atom_op_shift_left, ATOM_ARG_MC}, {
+	atom_op_shift_right, ATOM_ARG_REG}, {
+	atom_op_shift_right, ATOM_ARG_PS}, {
+	atom_op_shift_right, ATOM_ARG_WS}, {
+	atom_op_shift_right, ATOM_ARG_FB}, {
+	atom_op_shift_right, ATOM_ARG_PLL}, {
+	atom_op_shift_right, ATOM_ARG_MC}, {
+	atom_op_mul, ATOM_ARG_REG}, {
+	atom_op_mul, ATOM_ARG_PS}, {
+	atom_op_mul, ATOM_ARG_WS}, {
+	atom_op_mul, ATOM_ARG_FB}, {
+	atom_op_mul, ATOM_ARG_PLL}, {
+	atom_op_mul, ATOM_ARG_MC}, {
+	atom_op_div, ATOM_ARG_REG}, {
+	atom_op_div, ATOM_ARG_PS}, {
+	atom_op_div, ATOM_ARG_WS}, {
+	atom_op_div, ATOM_ARG_FB}, {
+	atom_op_div, ATOM_ARG_PLL}, {
+	atom_op_div, ATOM_ARG_MC}, {
+	atom_op_add, ATOM_ARG_REG}, {
+	atom_op_add, ATOM_ARG_PS}, {
+	atom_op_add, ATOM_ARG_WS}, {
+	atom_op_add, ATOM_ARG_FB}, {
+	atom_op_add, ATOM_ARG_PLL}, {
+	atom_op_add, ATOM_ARG_MC}, {
+	atom_op_sub, ATOM_ARG_REG}, {
+	atom_op_sub, ATOM_ARG_PS}, {
+	atom_op_sub, ATOM_ARG_WS}, {
+	atom_op_sub, ATOM_ARG_FB}, {
+	atom_op_sub, ATOM_ARG_PLL}, {
+	atom_op_sub, ATOM_ARG_MC}, {
+	atom_op_setport, ATOM_PORT_ATI}, {
+	atom_op_setport, ATOM_PORT_PCI}, {
+	atom_op_setport, ATOM_PORT_SYSIO}, {
+	atom_op_setregblock, 0}, {
+	atom_op_setfbbase, 0}, {
+	atom_op_compare, ATOM_ARG_REG}, {
+	atom_op_compare, ATOM_ARG_PS}, {
+	atom_op_compare, ATOM_ARG_WS}, {
+	atom_op_compare, ATOM_ARG_FB}, {
+	atom_op_compare, ATOM_ARG_PLL}, {
+	atom_op_compare, ATOM_ARG_MC}, {
+	atom_op_switch, 0}, {
+	atom_op_jump, ATOM_COND_ALWAYS}, {
+	atom_op_jump, ATOM_COND_EQUAL}, {
+	atom_op_jump, ATOM_COND_BELOW}, {
+	atom_op_jump, ATOM_COND_ABOVE}, {
+	atom_op_jump, ATOM_COND_BELOWOREQUAL}, {
+	atom_op_jump, ATOM_COND_ABOVEOREQUAL}, {
+	atom_op_jump, ATOM_COND_NOTEQUAL}, {
+	atom_op_test, ATOM_ARG_REG}, {
+	atom_op_test, ATOM_ARG_PS}, {
+	atom_op_test, ATOM_ARG_WS}, {
+	atom_op_test, ATOM_ARG_FB}, {
+	atom_op_test, ATOM_ARG_PLL}, {
+	atom_op_test, ATOM_ARG_MC}, {
+	atom_op_delay, ATOM_UNIT_MILLISEC}, {
+	atom_op_delay, ATOM_UNIT_MICROSEC}, {
+	atom_op_calltable, 0}, {
+	atom_op_repeat, 0}, {
+	atom_op_clear, ATOM_ARG_REG}, {
+	atom_op_clear, ATOM_ARG_PS}, {
+	atom_op_clear, ATOM_ARG_WS}, {
+	atom_op_clear, ATOM_ARG_FB}, {
+	atom_op_clear, ATOM_ARG_PLL}, {
+	atom_op_clear, ATOM_ARG_MC}, {
+	atom_op_nop, 0}, {
+	atom_op_eot, 0}, {
+	atom_op_mask, ATOM_ARG_REG}, {
+	atom_op_mask, ATOM_ARG_PS}, {
+	atom_op_mask, ATOM_ARG_WS}, {
+	atom_op_mask, ATOM_ARG_FB}, {
+	atom_op_mask, ATOM_ARG_PLL}, {
+	atom_op_mask, ATOM_ARG_MC}, {
+	atom_op_postcard, 0}, {
+	atom_op_beep, 0}, {
+	atom_op_savereg, 0}, {
+	atom_op_restorereg, 0}, {
+	atom_op_setdatablock, 0}, {
+	atom_op_xor, ATOM_ARG_REG}, {
+	atom_op_xor, ATOM_ARG_PS}, {
+	atom_op_xor, ATOM_ARG_WS}, {
+	atom_op_xor, ATOM_ARG_FB}, {
+	atom_op_xor, ATOM_ARG_PLL}, {
+	atom_op_xor, ATOM_ARG_MC}, {
+	atom_op_shl, ATOM_ARG_REG}, {
+	atom_op_shl, ATOM_ARG_PS}, {
+	atom_op_shl, ATOM_ARG_WS}, {
+	atom_op_shl, ATOM_ARG_FB}, {
+	atom_op_shl, ATOM_ARG_PLL}, {
+	atom_op_shl, ATOM_ARG_MC}, {
+	atom_op_shr, ATOM_ARG_REG}, {
+	atom_op_shr, ATOM_ARG_PS}, {
+	atom_op_shr, ATOM_ARG_WS}, {
+	atom_op_shr, ATOM_ARG_FB}, {
+	atom_op_shr, ATOM_ARG_PLL}, {
+	atom_op_shr, ATOM_ARG_MC}, {
+	atom_op_debug, 0}, {
+	atom_op_processds, 0}, {
+	atom_op_mul32, ATOM_ARG_PS}, {
+	atom_op_mul32, ATOM_ARG_WS}, {
+	atom_op_div32, ATOM_ARG_PS}, {
+	atom_op_div32, ATOM_ARG_WS},
+};
+
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t * params)
+{
+	int base = CU16(ctx->cmd_table + 4 + 2 * index);
+	int len, ws, ps, ptr;
+	unsigned char op;
+	atom_exec_context ectx;
+	int ret = 0;
+
+	if (!base)
+		return -EINVAL;
+
+	len = CU16(base + ATOM_CT_SIZE_PTR);
+	ws = CU8(base + ATOM_CT_WS_PTR);
+	ps = CU8(base + ATOM_CT_PS_PTR) & ATOM_CT_PS_MASK;
+	ptr = base + ATOM_CT_CODE_PTR;
+
+	SDEBUG(">> execute %04X (len %d, WS %d, PS %d)\n", base, len, ws, ps);
+
+	ectx.ctx = ctx;
+	ectx.ps_shift = ps / 4;
+	ectx.start = base;
+	ectx.ps = params;
+	ectx.abort = false;
+	ectx.last_jump = 0;
+	if (ws)
+		ectx.ws = kzalloc(4 * ws, GFP_KERNEL);
+	else
+		ectx.ws = NULL;
+
+	debug_depth++;
+	while (1) {
+		op = CU8(ptr++);
+		if (op < ATOM_OP_NAMES_CNT)
+			SDEBUG("%s @ 0x%04X\n", atom_op_names[op], ptr - 1);
+		else
+			SDEBUG("[%d] @ 0x%04X\n", op, ptr - 1);
+		if (ectx.abort) {
+			DRM_ERROR("atombios stuck executing %04X (len %d, WS %d, PS %d) @ 0x%04X\n",
+				base, len, ws, ps, ptr - 1);
+			ret = -EINVAL;
+			goto free;
+		}
+
+		if (op < ATOM_OP_CNT && op > 0)
+			opcode_table[op].func(&ectx, &ptr,
+					      opcode_table[op].arg);
+		else
+			break;
+
+		if (op == ATOM_OP_EOT)
+			break;
+	}
+	debug_depth--;
+	SDEBUG("<<\n");
+
+free:
+	if (ws)
+		kfree(ectx.ws);
+	return ret;
+}
+
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t * params)
+{
+	int r;
+
+	mutex_lock(&ctx->mutex);
+	/* reset data block */
+	ctx->data_block = 0;
+	/* reset reg block */
+	ctx->reg_block = 0;
+	/* reset fb window */
+	ctx->fb_base = 0;
+	/* reset io mode */
+	ctx->io_mode = ATOM_IO_MM;
+	/* reset divmul */
+	ctx->divmul[0] = 0;
+	ctx->divmul[1] = 0;
+	r = amdgpu_atom_execute_table_locked(ctx, index, params);
+	mutex_unlock(&ctx->mutex);
+	return r;
+}
+
+static int atom_iio_len[] = { 1, 2, 3, 3, 3, 3, 4, 4, 4, 3 };
+
+static void atom_index_iio(struct atom_context *ctx, int base)
+{
+	ctx->iio = kzalloc(2 * 256, GFP_KERNEL);
+	if (!ctx->iio)
+		return;
+	while (CU8(base) == ATOM_IIO_START) {
+		ctx->iio[CU8(base + 1)] = base + 2;
+		base += 2;
+		while (CU8(base) != ATOM_IIO_END)
+			base += atom_iio_len[CU8(base)];
+		base += 3;
+	}
+}
+
+struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios)
+{
+	int base;
+	struct atom_context *ctx =
+	    kzalloc(sizeof(struct atom_context), GFP_KERNEL);
+	char *str;
+	char name[512];
+	int i;
+
+	if (!ctx)
+		return NULL;
+
+	ctx->card = card;
+	ctx->bios = bios;
+
+	if (CU16(0) != ATOM_BIOS_MAGIC) {
+		printk(KERN_INFO "Invalid BIOS magic.\n");
+		kfree(ctx);
+		return NULL;
+	}
+	if (strncmp
+	    (CSTR(ATOM_ATI_MAGIC_PTR), ATOM_ATI_MAGIC,
+	     strlen(ATOM_ATI_MAGIC))) {
+		printk(KERN_INFO "Invalid ATI magic.\n");
+		kfree(ctx);
+		return NULL;
+	}
+
+	base = CU16(ATOM_ROM_TABLE_PTR);
+	if (strncmp
+	    (CSTR(base + ATOM_ROM_MAGIC_PTR), ATOM_ROM_MAGIC,
+	     strlen(ATOM_ROM_MAGIC))) {
+		printk(KERN_INFO "Invalid ATOM magic.\n");
+		kfree(ctx);
+		return NULL;
+	}
+
+	ctx->cmd_table = CU16(base + ATOM_ROM_CMD_PTR);
+	ctx->data_table = CU16(base + ATOM_ROM_DATA_PTR);
+	atom_index_iio(ctx, CU16(ctx->data_table + ATOM_DATA_IIO_PTR) + 4);
+	if (!ctx->iio) {
+		amdgpu_atom_destroy(ctx);
+		return NULL;
+	}
+
+	str = CSTR(CU16(base + ATOM_ROM_MSG_PTR));
+	while (*str && ((*str == '\n') || (*str == '\r')))
+		str++;
+	/* name string isn't always 0 terminated */
+	for (i = 0; i < 511; i++) {
+		name[i] = str[i];
+		if (name[i] < '.' || name[i] > 'z') {
+			name[i] = 0;
+			break;
+		}
+	}
+	printk(KERN_INFO "ATOM BIOS: %s\n", name);
+
+	return ctx;
+}
+
+int amdgpu_atom_asic_init(struct atom_context *ctx)
+{
+	int hwi = CU16(ctx->data_table + ATOM_DATA_FWI_PTR);
+	uint32_t ps[16];
+	int ret;
+
+	memset(ps, 0, 64);
+
+	ps[0] = cpu_to_le32(CU32(hwi + ATOM_FWI_DEFSCLK_PTR));
+	ps[1] = cpu_to_le32(CU32(hwi + ATOM_FWI_DEFMCLK_PTR));
+	if (!ps[0] || !ps[1])
+		return 1;
+
+	if (!CU16(ctx->cmd_table + 4 + 2 * ATOM_CMD_INIT))
+		return 1;
+	ret = amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, ps);
+	if (ret)
+		return ret;
+
+	memset(ps, 0, 64);
+
+	return ret;
+}
+
+void amdgpu_atom_destroy(struct atom_context *ctx)
+{
+	kfree(ctx->iio);
+	kfree(ctx);
+}
+
+bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index,
+			    uint16_t * size, uint8_t * frev, uint8_t * crev,
+			    uint16_t * data_start)
+{
+	int offset = index * 2 + 4;
+	int idx = CU16(ctx->data_table + offset);
+	u16 *mdt = (u16 *)(ctx->bios + ctx->data_table + 4);
+
+	if (!mdt[index])
+		return false;
+
+	if (size)
+		*size = CU16(idx);
+	if (frev)
+		*frev = CU8(idx + 2);
+	if (crev)
+		*crev = CU8(idx + 3);
+	*data_start = idx;
+	return true;
+}
+
+bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index, uint8_t * frev,
+			   uint8_t * crev)
+{
+	int offset = index * 2 + 4;
+	int idx = CU16(ctx->cmd_table + offset);
+	u16 *mct = (u16 *)(ctx->bios + ctx->cmd_table + 4);
+
+	if (!mct[index])
+		return false;
+
+	if (frev)
+		*frev = CU8(idx + 2);
+	if (crev)
+		*crev = CU8(idx + 3);
+	return true;
+}
+
+int amdgpu_atom_allocate_fb_scratch(struct atom_context *ctx)
+{
+	int index = GetIndexIntoMasterTable(DATA, VRAM_UsageByFirmware);
+	uint16_t data_offset;
+	int usage_bytes = 0;
+	struct _ATOM_VRAM_USAGE_BY_FIRMWARE *firmware_usage;
+
+	if (amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) {
+		firmware_usage = (struct _ATOM_VRAM_USAGE_BY_FIRMWARE *)(ctx->bios + data_offset);
+
+		DRM_DEBUG("atom firmware requested %08x %dkb\n",
+			  le32_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].ulStartAddrUsedByFirmware),
+			  le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb));
+
+		usage_bytes = le16_to_cpu(firmware_usage->asFirmwareVramReserveInfo[0].usFirmwareUseInKb) * 1024;
+	}
+	ctx->scratch_size_bytes = 0;
+	if (usage_bytes == 0)
+		usage_bytes = 20 * 1024;
+	/* allocate some scratch memory */
+	ctx->scratch = kzalloc(usage_bytes, GFP_KERNEL);
+	if (!ctx->scratch)
+		return -ENOMEM;
+	ctx->scratch_size_bytes = usage_bytes;
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atom.h b/kernel/drivers/gpu/drm/amd/amdgpu/atom.h
new file mode 100644
index 000000000..fece8f45d
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atom.h
@@ -0,0 +1,159 @@
+/*
+ * Copyright 2008 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Author: Stanislaw Skowronek
+ */
+
+#ifndef ATOM_H
+#define ATOM_H
+
+#include <linux/types.h>
+#include <drm/drmP.h>
+
+#define ATOM_BIOS_MAGIC		0xAA55
+#define ATOM_ATI_MAGIC_PTR	0x30
+#define ATOM_ATI_MAGIC		" 761295520"
+#define ATOM_ROM_TABLE_PTR	0x48
+
+#define ATOM_ROM_MAGIC		"ATOM"
+#define ATOM_ROM_MAGIC_PTR	4
+
+#define ATOM_ROM_MSG_PTR	0x10
+#define ATOM_ROM_CMD_PTR	0x1E
+#define ATOM_ROM_DATA_PTR	0x20
+
+#define ATOM_CMD_INIT		0
+#define ATOM_CMD_SETSCLK	0x0A
+#define ATOM_CMD_SETMCLK	0x0B
+#define ATOM_CMD_SETPCLK	0x0C
+#define ATOM_CMD_SPDFANCNTL	0x39
+
+#define ATOM_DATA_FWI_PTR	0xC
+#define ATOM_DATA_IIO_PTR	0x32
+
+#define ATOM_FWI_DEFSCLK_PTR	8
+#define ATOM_FWI_DEFMCLK_PTR	0xC
+#define ATOM_FWI_MAXSCLK_PTR	0x24
+#define ATOM_FWI_MAXMCLK_PTR	0x28
+
+#define ATOM_CT_SIZE_PTR	0
+#define ATOM_CT_WS_PTR		4
+#define ATOM_CT_PS_PTR		5
+#define ATOM_CT_PS_MASK		0x7F
+#define ATOM_CT_CODE_PTR	6
+
+#define ATOM_OP_CNT		127
+#define ATOM_OP_EOT		91
+
+#define ATOM_CASE_MAGIC		0x63
+#define ATOM_CASE_END		0x5A5A
+
+#define ATOM_ARG_REG		0
+#define ATOM_ARG_PS		1
+#define ATOM_ARG_WS		2
+#define ATOM_ARG_FB		3
+#define ATOM_ARG_ID		4
+#define ATOM_ARG_IMM		5
+#define ATOM_ARG_PLL		6
+#define ATOM_ARG_MC		7
+
+#define ATOM_SRC_DWORD		0
+#define ATOM_SRC_WORD0		1
+#define ATOM_SRC_WORD8		2
+#define ATOM_SRC_WORD16		3
+#define ATOM_SRC_BYTE0		4
+#define ATOM_SRC_BYTE8		5
+#define ATOM_SRC_BYTE16		6
+#define ATOM_SRC_BYTE24		7
+
+#define ATOM_WS_QUOTIENT	0x40
+#define ATOM_WS_REMAINDER	0x41
+#define ATOM_WS_DATAPTR		0x42
+#define ATOM_WS_SHIFT		0x43
+#define ATOM_WS_OR_MASK		0x44
+#define ATOM_WS_AND_MASK	0x45
+#define ATOM_WS_FB_WINDOW	0x46
+#define ATOM_WS_ATTRIBUTES	0x47
+#define ATOM_WS_REGPTR  	0x48
+
+#define ATOM_IIO_NOP		0
+#define ATOM_IIO_START		1
+#define ATOM_IIO_READ		2
+#define ATOM_IIO_WRITE		3
+#define ATOM_IIO_CLEAR		4
+#define ATOM_IIO_SET		5
+#define ATOM_IIO_MOVE_INDEX	6
+#define ATOM_IIO_MOVE_ATTR	7
+#define ATOM_IIO_MOVE_DATA	8
+#define ATOM_IIO_END		9
+
+#define ATOM_IO_MM		0
+#define ATOM_IO_PCI		1
+#define ATOM_IO_SYSIO		2
+#define ATOM_IO_IIO		0x80
+
+struct card_info {
+	struct drm_device *dev;
+	void (* reg_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* reg_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (* ioreg_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* ioreg_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (* mc_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* mc_read)(struct card_info *, uint32_t);          /*  filled by driver */
+	void (* pll_write)(struct card_info *, uint32_t, uint32_t);   /*  filled by driver */
+	uint32_t (* pll_read)(struct card_info *, uint32_t);          /*  filled by driver */
+};
+
+struct atom_context {
+	struct card_info *card;
+	struct mutex mutex;
+	void *bios;
+	uint32_t cmd_table, data_table;
+	uint16_t *iio;
+
+	uint16_t data_block;
+	uint32_t fb_base;
+	uint32_t divmul[2];
+	uint16_t io_attr;
+	uint16_t reg_block;
+	uint8_t shift;
+	int cs_equal, cs_above;
+	int io_mode;
+	uint32_t *scratch;
+	int scratch_size_bytes;
+};
+
+extern int amdgpu_atom_debug;
+
+struct atom_context *amdgpu_atom_parse(struct card_info *, void *);
+int amdgpu_atom_execute_table(struct atom_context *, int, uint32_t *);
+int amdgpu_atom_asic_init(struct atom_context *);
+void amdgpu_atom_destroy(struct atom_context *);
+bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, uint16_t *size,
+			    uint8_t *frev, uint8_t *crev, uint16_t *data_start);
+bool amdgpu_atom_parse_cmd_header(struct atom_context *ctx, int index,
+			   uint8_t *frev, uint8_t *crev);
+int amdgpu_atom_allocate_fb_scratch(struct atom_context *ctx);
+#include "atom-types.h"
+#include "atombios.h"
+#include "ObjectID.h"
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
new file mode 100644
index 000000000..49aa35016
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
@@ -0,0 +1,807 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include <drm/drm_fixed.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "atom-bits.h"
+#include "atombios_encoders.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+
+void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	SET_CRTC_OVERSCAN_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_OverScan);
+	int a1, a2;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+
+	switch (amdgpu_crtc->rmx_type) {
+	case RMX_CENTER:
+		args.usOverscanTop = cpu_to_le16((adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2);
+		args.usOverscanBottom = cpu_to_le16((adjusted_mode->crtc_vdisplay - mode->crtc_vdisplay) / 2);
+		args.usOverscanLeft = cpu_to_le16((adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2);
+		args.usOverscanRight = cpu_to_le16((adjusted_mode->crtc_hdisplay - mode->crtc_hdisplay) / 2);
+		break;
+	case RMX_ASPECT:
+		a1 = mode->crtc_vdisplay * adjusted_mode->crtc_hdisplay;
+		a2 = adjusted_mode->crtc_vdisplay * mode->crtc_hdisplay;
+
+		if (a1 > a2) {
+			args.usOverscanLeft = cpu_to_le16((adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2);
+			args.usOverscanRight = cpu_to_le16((adjusted_mode->crtc_hdisplay - (a2 / mode->crtc_vdisplay)) / 2);
+		} else if (a2 > a1) {
+			args.usOverscanTop = cpu_to_le16((adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2);
+			args.usOverscanBottom = cpu_to_le16((adjusted_mode->crtc_vdisplay - (a1 / mode->crtc_hdisplay)) / 2);
+		}
+		break;
+	case RMX_FULL:
+	default:
+		args.usOverscanRight = cpu_to_le16(amdgpu_crtc->h_border);
+		args.usOverscanLeft = cpu_to_le16(amdgpu_crtc->h_border);
+		args.usOverscanBottom = cpu_to_le16(amdgpu_crtc->v_border);
+		args.usOverscanTop = cpu_to_le16(amdgpu_crtc->v_border);
+		break;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	ENABLE_SCALER_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableScaler);
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucScaler = amdgpu_crtc->crtc_id;
+
+	switch (amdgpu_crtc->rmx_type) {
+	case RMX_FULL:
+		args.ucEnable = ATOM_SCALER_EXPANSION;
+		break;
+	case RMX_CENTER:
+		args.ucEnable = ATOM_SCALER_CENTER;
+		break;
+	case RMX_ASPECT:
+		args.ucEnable = ATOM_SCALER_EXPANSION;
+		break;
+	default:
+		args.ucEnable = ATOM_SCALER_DISABLE;
+		break;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index =
+	    GetIndexIntoMasterTable(COMMAND, UpdateCRTC_DoubleBufferRegisters);
+	ENABLE_CRTC_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+	args.ucEnable = lock;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableCRTC);
+	ENABLE_CRTC_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+	args.ucEnable = state;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index = GetIndexIntoMasterTable(COMMAND, BlankCRTC);
+	BLANK_CRTC_PS_ALLOCATION args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+	args.ucBlanking = state;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableDispPowerGating);
+	ENABLE_DISP_POWER_GATING_PARAMETERS_V2_1 args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucDispPipeId = amdgpu_crtc->crtc_id;
+	args.ucEnable = state;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev)
+{
+	int index = GetIndexIntoMasterTable(COMMAND, EnableDispPowerGating);
+	ENABLE_DISP_POWER_GATING_PARAMETERS_V2_1 args;
+
+	memset(&args, 0, sizeof(args));
+
+	args.ucEnable = ATOM_INIT;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	SET_CRTC_USING_DTD_TIMING_PARAMETERS args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetCRTC_UsingDTDTiming);
+	u16 misc = 0;
+
+	memset(&args, 0, sizeof(args));
+	args.usH_Size = cpu_to_le16(mode->crtc_hdisplay - (amdgpu_crtc->h_border * 2));
+	args.usH_Blanking_Time =
+		cpu_to_le16(mode->crtc_hblank_end - mode->crtc_hdisplay + (amdgpu_crtc->h_border * 2));
+	args.usV_Size = cpu_to_le16(mode->crtc_vdisplay - (amdgpu_crtc->v_border * 2));
+	args.usV_Blanking_Time =
+		cpu_to_le16(mode->crtc_vblank_end - mode->crtc_vdisplay + (amdgpu_crtc->v_border * 2));
+	args.usH_SyncOffset =
+		cpu_to_le16(mode->crtc_hsync_start - mode->crtc_hdisplay + amdgpu_crtc->h_border);
+	args.usH_SyncWidth =
+		cpu_to_le16(mode->crtc_hsync_end - mode->crtc_hsync_start);
+	args.usV_SyncOffset =
+		cpu_to_le16(mode->crtc_vsync_start - mode->crtc_vdisplay + amdgpu_crtc->v_border);
+	args.usV_SyncWidth =
+		cpu_to_le16(mode->crtc_vsync_end - mode->crtc_vsync_start);
+	args.ucH_Border = amdgpu_crtc->h_border;
+	args.ucV_Border = amdgpu_crtc->v_border;
+
+	if (mode->flags & DRM_MODE_FLAG_NVSYNC)
+		misc |= ATOM_VSYNC_POLARITY;
+	if (mode->flags & DRM_MODE_FLAG_NHSYNC)
+		misc |= ATOM_HSYNC_POLARITY;
+	if (mode->flags & DRM_MODE_FLAG_CSYNC)
+		misc |= ATOM_COMPOSITESYNC;
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		misc |= ATOM_INTERLACE;
+	if (mode->flags & DRM_MODE_FLAG_DBLSCAN)
+		misc |= ATOM_DOUBLE_CLOCK_MODE;
+
+	args.susModeMiscInfo.usAccess = cpu_to_le16(misc);
+	args.ucCRTC = amdgpu_crtc->crtc_id;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union atom_enable_ss {
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_PS_ALLOCATION v1;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_V2 v2;
+	ENABLE_SPREAD_SPECTRUM_ON_PPLL_V3 v3;
+};
+
+static void amdgpu_atombios_crtc_program_ss(struct amdgpu_device *adev,
+				     int enable,
+				     int pll_id,
+				     int crtc_id,
+				     struct amdgpu_atom_ss *ss)
+{
+	unsigned i;
+	int index = GetIndexIntoMasterTable(COMMAND, EnableSpreadSpectrumOnPPLL);
+	union atom_enable_ss args;
+
+	if (enable) {
+		/* Don't mess with SS if percentage is 0 or external ss.
+		 * SS is already disabled previously, and disabling it
+		 * again can cause display problems if the pll is already
+		 * programmed.
+		 */
+		if (ss->percentage == 0)
+			return;
+		if (ss->type & ATOM_EXTERNAL_SS_MASK)
+			return;
+	} else {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (adev->mode_info.crtcs[i] &&
+			    adev->mode_info.crtcs[i]->enabled &&
+			    i != crtc_id &&
+			    pll_id == adev->mode_info.crtcs[i]->pll_id) {
+				/* one other crtc is using this pll don't turn
+				 * off spread spectrum as it might turn off
+				 * display on active crtc
+				 */
+				return;
+			}
+		}
+	}
+
+	memset(&args, 0, sizeof(args));
+
+	args.v3.usSpreadSpectrumAmountFrac = cpu_to_le16(0);
+	args.v3.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK;
+	switch (pll_id) {
+	case ATOM_PPLL1:
+		args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P1PLL;
+		break;
+	case ATOM_PPLL2:
+		args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P2PLL;
+		break;
+	case ATOM_DCPLL:
+		args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_DCPLL;
+		break;
+	case ATOM_PPLL_INVALID:
+		return;
+	}
+	args.v3.usSpreadSpectrumAmount = cpu_to_le16(ss->amount);
+	args.v3.usSpreadSpectrumStep = cpu_to_le16(ss->step);
+	args.v3.ucEnable = enable;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union adjust_pixel_clock {
+	ADJUST_DISPLAY_PLL_PS_ALLOCATION v1;
+	ADJUST_DISPLAY_PLL_PS_ALLOCATION_V3 v3;
+};
+
+static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
+				    struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_encoder *encoder = amdgpu_crtc->encoder;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u32 adjusted_clock = mode->clock;
+	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+	u32 dp_clock = mode->clock;
+	u32 clock = mode->clock;
+	int bpc = amdgpu_crtc->bpc;
+	bool is_duallink = amdgpu_dig_monitor_is_duallink(encoder, mode->clock);
+	union adjust_pixel_clock args;
+	u8 frev, crev;
+	int index;
+
+	amdgpu_crtc->pll_flags = AMDGPU_PLL_USE_FRAC_FB_DIV;
+
+	if ((amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE)) {
+		if (connector) {
+			struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+			struct amdgpu_connector_atom_dig *dig_connector =
+				amdgpu_connector->con_priv;
+
+			dp_clock = dig_connector->dp_clock;
+		}
+	}
+
+	/* use recommended ref_div for ss */
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+		if (amdgpu_crtc->ss_enabled) {
+			if (amdgpu_crtc->ss.refdiv) {
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_REF_DIV;
+				amdgpu_crtc->pll_reference_div = amdgpu_crtc->ss.refdiv;
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
+			}
+		}
+	}
+
+	/* DVO wants 2x pixel clock if the DVO chip is in 12 bit mode */
+	if (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1)
+		adjusted_clock = mode->clock * 2;
+	if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+		amdgpu_crtc->pll_flags |= AMDGPU_PLL_PREFER_CLOSEST_LOWER;
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_crtc->pll_flags |= AMDGPU_PLL_IS_LCD;
+
+
+	/* adjust pll for deep color modes */
+	if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+		switch (bpc) {
+		case 8:
+		default:
+			break;
+		case 10:
+			clock = (clock * 5) / 4;
+			break;
+		case 12:
+			clock = (clock * 3) / 2;
+			break;
+		case 16:
+			clock = clock * 2;
+			break;
+		}
+	}
+
+	/* DCE3+ has an AdjustDisplayPll that will adjust the pixel clock
+	 * accordingly based on the encoder/transmitter to work around
+	 * special hw requirements.
+	 */
+	index = GetIndexIntoMasterTable(COMMAND, AdjustDisplayPll);
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return adjusted_clock;
+
+	memset(&args, 0, sizeof(args));
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+		case 2:
+			args.v1.usPixelClock = cpu_to_le16(clock / 10);
+			args.v1.ucTransmitterID = amdgpu_encoder->encoder_id;
+			args.v1.ucEncodeMode = encoder_mode;
+			if (amdgpu_crtc->ss_enabled && amdgpu_crtc->ss.percentage)
+				args.v1.ucConfig |=
+					ADJUST_DISPLAY_CONFIG_SS_ENABLE;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context,
+					   index, (uint32_t *)&args);
+			adjusted_clock = le16_to_cpu(args.v1.usPixelClock) * 10;
+			break;
+		case 3:
+			args.v3.sInput.usPixelClock = cpu_to_le16(clock / 10);
+			args.v3.sInput.ucTransmitterID = amdgpu_encoder->encoder_id;
+			args.v3.sInput.ucEncodeMode = encoder_mode;
+			args.v3.sInput.ucDispPllConfig = 0;
+			if (amdgpu_crtc->ss_enabled && amdgpu_crtc->ss.percentage)
+				args.v3.sInput.ucDispPllConfig |=
+					DISPPLL_CONFIG_SS_ENABLE;
+			if (ENCODER_MODE_IS_DP(encoder_mode)) {
+				args.v3.sInput.ucDispPllConfig |=
+					DISPPLL_CONFIG_COHERENT_MODE;
+				/* 16200 or 27000 */
+				args.v3.sInput.usPixelClock = cpu_to_le16(dp_clock / 10);
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+				if (dig->coherent_mode)
+					args.v3.sInput.ucDispPllConfig |=
+						DISPPLL_CONFIG_COHERENT_MODE;
+				if (is_duallink)
+					args.v3.sInput.ucDispPllConfig |=
+						DISPPLL_CONFIG_DUAL_LINK;
+			}
+			if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+			    ENCODER_OBJECT_ID_NONE)
+				args.v3.sInput.ucExtTransmitterID =
+					amdgpu_encoder_get_dp_bridge_encoder_id(encoder);
+			else
+				args.v3.sInput.ucExtTransmitterID = 0;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context,
+					   index, (uint32_t *)&args);
+			adjusted_clock = le32_to_cpu(args.v3.sOutput.ulDispPllFreq) * 10;
+			if (args.v3.sOutput.ucRefDiv) {
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_REF_DIV;
+				amdgpu_crtc->pll_reference_div = args.v3.sOutput.ucRefDiv;
+			}
+			if (args.v3.sOutput.ucPostDiv) {
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
+				amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_POST_DIV;
+				amdgpu_crtc->pll_post_div = args.v3.sOutput.ucPostDiv;
+			}
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return adjusted_clock;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return adjusted_clock;
+	}
+
+	return adjusted_clock;
+}
+
+union set_pixel_clock {
+	SET_PIXEL_CLOCK_PS_ALLOCATION base;
+	PIXEL_CLOCK_PARAMETERS v1;
+	PIXEL_CLOCK_PARAMETERS_V2 v2;
+	PIXEL_CLOCK_PARAMETERS_V3 v3;
+	PIXEL_CLOCK_PARAMETERS_V5 v5;
+	PIXEL_CLOCK_PARAMETERS_V6 v6;
+};
+
+/* on DCE5, make sure the voltage is high enough to support the
+ * required disp clk.
+ */
+void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
+				    u32 dispclk)
+{
+	u8 frev, crev;
+	int index;
+	union set_pixel_clock args;
+
+	memset(&args, 0, sizeof(args));
+
+	index = GetIndexIntoMasterTable(COMMAND, SetPixelClock);
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 5:
+			/* if the default dcpll clock is specified,
+			 * SetPixelClock provides the dividers
+			 */
+			args.v5.ucCRTC = ATOM_CRTC_INVALID;
+			args.v5.usPixelClock = cpu_to_le16(dispclk);
+			args.v5.ucPpll = ATOM_DCPLL;
+			break;
+		case 6:
+			/* if the default dcpll clock is specified,
+			 * SetPixelClock provides the dividers
+			 */
+			args.v6.ulDispEngClkFreq = cpu_to_le32(dispclk);
+			args.v6.ucPpll = ATOM_EXT_PLL1;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+static bool is_pixel_clock_source_from_pll(u32 encoder_mode, int pll_id)
+{
+	if (ENCODER_MODE_IS_DP(encoder_mode)) {
+		if (pll_id < ATOM_EXT_PLL1)
+			return true;
+		else
+			return false;
+	} else {
+		return true;
+	}
+}
+
+void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
+			       u32 crtc_id,
+			       int pll_id,
+			       u32 encoder_mode,
+			       u32 encoder_id,
+			       u32 clock,
+			       u32 ref_div,
+			       u32 fb_div,
+			       u32 frac_fb_div,
+			       u32 post_div,
+			       int bpc,
+			       bool ss_enabled,
+			       struct amdgpu_atom_ss *ss)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u8 frev, crev;
+	int index = GetIndexIntoMasterTable(COMMAND, SetPixelClock);
+	union set_pixel_clock args;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev,
+				   &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			if (clock == ATOM_DISABLE)
+				return;
+			args.v1.usPixelClock = cpu_to_le16(clock / 10);
+			args.v1.usRefDiv = cpu_to_le16(ref_div);
+			args.v1.usFbDiv = cpu_to_le16(fb_div);
+			args.v1.ucFracFbDiv = frac_fb_div;
+			args.v1.ucPostDiv = post_div;
+			args.v1.ucPpll = pll_id;
+			args.v1.ucCRTC = crtc_id;
+			args.v1.ucRefDivSrc = 1;
+			break;
+		case 2:
+			args.v2.usPixelClock = cpu_to_le16(clock / 10);
+			args.v2.usRefDiv = cpu_to_le16(ref_div);
+			args.v2.usFbDiv = cpu_to_le16(fb_div);
+			args.v2.ucFracFbDiv = frac_fb_div;
+			args.v2.ucPostDiv = post_div;
+			args.v2.ucPpll = pll_id;
+			args.v2.ucCRTC = crtc_id;
+			args.v2.ucRefDivSrc = 1;
+			break;
+		case 3:
+			args.v3.usPixelClock = cpu_to_le16(clock / 10);
+			args.v3.usRefDiv = cpu_to_le16(ref_div);
+			args.v3.usFbDiv = cpu_to_le16(fb_div);
+			args.v3.ucFracFbDiv = frac_fb_div;
+			args.v3.ucPostDiv = post_div;
+			args.v3.ucPpll = pll_id;
+			if (crtc_id == ATOM_CRTC2)
+				args.v3.ucMiscInfo = PIXEL_CLOCK_MISC_CRTC_SEL_CRTC2;
+			else
+				args.v3.ucMiscInfo = PIXEL_CLOCK_MISC_CRTC_SEL_CRTC1;
+			if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK))
+				args.v3.ucMiscInfo |= PIXEL_CLOCK_MISC_REF_DIV_SRC;
+			args.v3.ucTransmitterId = encoder_id;
+			args.v3.ucEncoderMode = encoder_mode;
+			break;
+		case 5:
+			args.v5.ucCRTC = crtc_id;
+			args.v5.usPixelClock = cpu_to_le16(clock / 10);
+			args.v5.ucRefDiv = ref_div;
+			args.v5.usFbDiv = cpu_to_le16(fb_div);
+			args.v5.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000);
+			args.v5.ucPostDiv = post_div;
+			args.v5.ucMiscInfo = 0; /* HDMI depth, etc. */
+			if ((ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) &&
+			    (pll_id < ATOM_EXT_PLL1))
+				args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_REF_DIV_SRC;
+			if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+				switch (bpc) {
+				case 8:
+				default:
+					args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_24BPP;
+					break;
+				case 10:
+					/* yes this is correct, the atom define is wrong */
+					args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_32BPP;
+					break;
+				case 12:
+					/* yes this is correct, the atom define is wrong */
+					args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_30BPP;
+					break;
+				}
+			}
+			args.v5.ucTransmitterID = encoder_id;
+			args.v5.ucEncoderMode = encoder_mode;
+			args.v5.ucPpll = pll_id;
+			break;
+		case 6:
+			args.v6.ulDispEngClkFreq = cpu_to_le32(crtc_id << 24 | clock / 10);
+			args.v6.ucRefDiv = ref_div;
+			args.v6.usFbDiv = cpu_to_le16(fb_div);
+			args.v6.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000);
+			args.v6.ucPostDiv = post_div;
+			args.v6.ucMiscInfo = 0; /* HDMI depth, etc. */
+			if ((ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) &&
+			    (pll_id < ATOM_EXT_PLL1) &&
+			    !is_pixel_clock_source_from_pll(encoder_mode, pll_id))
+				args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_REF_DIV_SRC;
+			if (encoder_mode == ATOM_ENCODER_MODE_HDMI) {
+				switch (bpc) {
+				case 8:
+				default:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_24BPP;
+					break;
+				case 10:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_30BPP_V6;
+					break;
+				case 12:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_36BPP_V6;
+					break;
+				case 16:
+					args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_48BPP;
+					break;
+				}
+			}
+			args.v6.ucTransmitterID = encoder_id;
+			args.v6.ucEncoderMode = encoder_mode;
+			args.v6.ucPpll = pll_id;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+		return;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
+			      struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder =
+		to_amdgpu_encoder(amdgpu_crtc->encoder);
+	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
+
+	amdgpu_crtc->bpc = 8;
+	amdgpu_crtc->ss_enabled = false;
+
+	if ((amdgpu_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(amdgpu_crtc->encoder) != ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		struct drm_connector *connector =
+			amdgpu_get_connector_for_encoder(amdgpu_crtc->encoder);
+		struct amdgpu_connector *amdgpu_connector =
+			to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+		int dp_clock;
+
+		/* Assign mode clock for hdmi deep color max clock limit check */
+		amdgpu_connector->pixelclock_for_modeset = mode->clock;
+		amdgpu_crtc->bpc = amdgpu_connector_get_monitor_bpc(connector);
+
+		switch (encoder_mode) {
+		case ATOM_ENCODER_MODE_DP_MST:
+		case ATOM_ENCODER_MODE_DP:
+			/* DP/eDP */
+			dp_clock = dig_connector->dp_clock / 10;
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev, &amdgpu_crtc->ss,
+								 ASIC_INTERNAL_SS_ON_DP,
+								 dp_clock);
+			break;
+		case ATOM_ENCODER_MODE_LVDS:
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev,
+								 &amdgpu_crtc->ss,
+								 dig->lcd_ss_id,
+								 mode->clock / 10);
+			break;
+		case ATOM_ENCODER_MODE_DVI:
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev,
+								 &amdgpu_crtc->ss,
+								 ASIC_INTERNAL_SS_ON_TMDS,
+								 mode->clock / 10);
+			break;
+		case ATOM_ENCODER_MODE_HDMI:
+			amdgpu_crtc->ss_enabled =
+				amdgpu_atombios_get_asic_ss_info(adev,
+								 &amdgpu_crtc->ss,
+								 ASIC_INTERNAL_SS_ON_HDMI,
+								 mode->clock / 10);
+			break;
+		default:
+			break;
+		}
+	}
+
+	/* adjust pixel clock as needed */
+	amdgpu_crtc->adjusted_clock = amdgpu_atombios_crtc_adjust_pll(crtc, mode);
+
+	return 0;
+}
+
+void amdgpu_atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder =
+		to_amdgpu_encoder(amdgpu_crtc->encoder);
+	u32 pll_clock = mode->clock;
+	u32 clock = mode->clock;
+	u32 ref_div = 0, fb_div = 0, frac_fb_div = 0, post_div = 0;
+	struct amdgpu_pll *pll;
+	int encoder_mode = amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder);
+
+	/* pass the actual clock to amdgpu_atombios_crtc_program_pll for HDMI */
+	if ((encoder_mode == ATOM_ENCODER_MODE_HDMI) &&
+	    (amdgpu_crtc->bpc > 8))
+		clock = amdgpu_crtc->adjusted_clock;
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL1:
+		pll = &adev->clock.ppll[0];
+		break;
+	case ATOM_PPLL2:
+		pll = &adev->clock.ppll[1];
+		break;
+	case ATOM_PPLL0:
+	case ATOM_PPLL_INVALID:
+	default:
+		pll = &adev->clock.ppll[2];
+		break;
+	}
+
+	/* update pll params */
+	pll->flags = amdgpu_crtc->pll_flags;
+	pll->reference_div = amdgpu_crtc->pll_reference_div;
+	pll->post_div = amdgpu_crtc->pll_post_div;
+
+	amdgpu_pll_compute(pll, amdgpu_crtc->adjusted_clock, &pll_clock,
+			    &fb_div, &frac_fb_div, &ref_div, &post_div);
+
+	amdgpu_atombios_crtc_program_ss(adev, ATOM_DISABLE, amdgpu_crtc->pll_id,
+				 amdgpu_crtc->crtc_id, &amdgpu_crtc->ss);
+
+	amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+				  encoder_mode, amdgpu_encoder->encoder_id, clock,
+				  ref_div, fb_div, frac_fb_div, post_div,
+				  amdgpu_crtc->bpc, amdgpu_crtc->ss_enabled, &amdgpu_crtc->ss);
+
+	if (amdgpu_crtc->ss_enabled) {
+		/* calculate ss amount and step size */
+		u32 step_size;
+		u32 amount = (((fb_div * 10) + frac_fb_div) *
+			      (u32)amdgpu_crtc->ss.percentage) /
+			(100 * (u32)amdgpu_crtc->ss.percentage_divider);
+		amdgpu_crtc->ss.amount = (amount / 10) & ATOM_PPLL_SS_AMOUNT_V2_FBDIV_MASK;
+		amdgpu_crtc->ss.amount |= ((amount - (amount / 10)) << ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT) &
+			ATOM_PPLL_SS_AMOUNT_V2_NFRAC_MASK;
+		if (amdgpu_crtc->ss.type & ATOM_PPLL_SS_TYPE_V2_CENTRE_SPREAD)
+			step_size = (4 * amount * ref_div * ((u32)amdgpu_crtc->ss.rate * 2048)) /
+				(125 * 25 * pll->reference_freq / 100);
+		else
+			step_size = (2 * amount * ref_div * ((u32)amdgpu_crtc->ss.rate * 2048)) /
+				(125 * 25 * pll->reference_freq / 100);
+		amdgpu_crtc->ss.step = step_size;
+
+		amdgpu_atombios_crtc_program_ss(adev, ATOM_ENABLE, amdgpu_crtc->pll_id,
+					 amdgpu_crtc->crtc_id, &amdgpu_crtc->ss);
+	}
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.h b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.h
new file mode 100644
index 000000000..c67083335
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_crtc.h
@@ -0,0 +1,58 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_CRTC_H__
+#define __ATOMBIOS_CRTC_H__
+
+void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode);
+void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc);
+void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock);
+void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state);
+void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state);
+void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state);
+void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev);
+void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode);
+void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
+				    u32 dispclk);
+void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
+			       u32 crtc_id,
+			       int pll_id,
+			       u32 encoder_mode,
+			       u32 encoder_id,
+			       u32 clock,
+			       u32 ref_div,
+			       u32 fb_div,
+			       u32 frac_fb_div,
+			       u32 post_div,
+			       int bpc,
+			       bool ss_enabled,
+			       struct amdgpu_atom_ss *ss);
+int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
+			      struct drm_display_mode *mode);
+void amdgpu_atombios_crtc_set_pll(struct drm_crtc *crtc,
+			   struct drm_display_mode *mode);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.c b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
new file mode 100644
index 000000000..92b6acadf
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
@@ -0,0 +1,776 @@
+/*
+ * Copyright 2007-8 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ *          Jerome Glisse
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+
+#include "atom.h"
+#include "atom-bits.h"
+#include "atombios_encoders.h"
+#include "atombios_dp.h"
+#include "amdgpu_connectors.h"
+#include "amdgpu_atombios.h"
+#include <drm/drm_dp_helper.h>
+
+/* move these to drm_dp_helper.c/h */
+#define DP_LINK_CONFIGURATION_SIZE 9
+#define DP_DPCD_SIZE DP_RECEIVER_CAP_SIZE
+
+static char *voltage_names[] = {
+	"0.4V", "0.6V", "0.8V", "1.2V"
+};
+static char *pre_emph_names[] = {
+	"0dB", "3.5dB", "6dB", "9.5dB"
+};
+
+/***** amdgpu AUX functions *****/
+
+union aux_channel_transaction {
+	PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION v1;
+	PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 v2;
+};
+
+static int amdgpu_atombios_dp_process_aux_ch(struct amdgpu_i2c_chan *chan,
+				      u8 *send, int send_bytes,
+				      u8 *recv, int recv_size,
+				      u8 delay, u8 *ack)
+{
+	struct drm_device *dev = chan->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	union aux_channel_transaction args;
+	int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction);
+	unsigned char *base;
+	int recv_bytes;
+	int r = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	mutex_lock(&chan->mutex);
+
+	base = (unsigned char *)(adev->mode_info.atom_context->scratch + 1);
+
+	amdgpu_atombios_copy_swap(base, send, send_bytes, true);
+
+	args.v2.lpAuxRequest = cpu_to_le16((u16)(0 + 4));
+	args.v2.lpDataOut = cpu_to_le16((u16)(16 + 4));
+	args.v2.ucDataOutLen = 0;
+	args.v2.ucChannelID = chan->rec.i2c_id;
+	args.v2.ucDelay = delay / 10;
+	args.v2.ucHPD_ID = chan->rec.hpd;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	*ack = args.v2.ucReplyStatus;
+
+	/* timeout */
+	if (args.v2.ucReplyStatus == 1) {
+		DRM_DEBUG_KMS("dp_aux_ch timeout\n");
+		r = -ETIMEDOUT;
+		goto done;
+	}
+
+	/* flags not zero */
+	if (args.v2.ucReplyStatus == 2) {
+		DRM_DEBUG_KMS("dp_aux_ch flags not zero\n");
+		r = -EIO;
+		goto done;
+	}
+
+	/* error */
+	if (args.v2.ucReplyStatus == 3) {
+		DRM_DEBUG_KMS("dp_aux_ch error\n");
+		r = -EIO;
+		goto done;
+	}
+
+	recv_bytes = args.v1.ucDataOutLen;
+	if (recv_bytes > recv_size)
+		recv_bytes = recv_size;
+
+	if (recv && recv_size)
+		amdgpu_atombios_copy_swap(recv, base + 16, recv_bytes, false);
+
+	r = recv_bytes;
+done:
+	mutex_unlock(&chan->mutex);
+
+	return r;
+}
+
+#define BARE_ADDRESS_SIZE 3
+#define HEADER_SIZE (BARE_ADDRESS_SIZE + 1)
+
+static ssize_t
+amdgpu_atombios_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg)
+{
+	struct amdgpu_i2c_chan *chan =
+		container_of(aux, struct amdgpu_i2c_chan, aux);
+	int ret;
+	u8 tx_buf[20];
+	size_t tx_size;
+	u8 ack, delay = 0;
+
+	if (WARN_ON(msg->size > 16))
+		return -E2BIG;
+
+	tx_buf[0] = msg->address & 0xff;
+	tx_buf[1] = msg->address >> 8;
+	tx_buf[2] = (msg->request << 4) |
+		((msg->address >> 16) & 0xf);
+	tx_buf[3] = msg->size ? (msg->size - 1) : 0;
+
+	switch (msg->request & ~DP_AUX_I2C_MOT) {
+	case DP_AUX_NATIVE_WRITE:
+	case DP_AUX_I2C_WRITE:
+		/* tx_size needs to be 4 even for bare address packets since the atom
+		 * table needs the info in tx_buf[3].
+		 */
+		tx_size = HEADER_SIZE + msg->size;
+		if (msg->size == 0)
+			tx_buf[3] |= BARE_ADDRESS_SIZE << 4;
+		else
+			tx_buf[3] |= tx_size << 4;
+		memcpy(tx_buf + HEADER_SIZE, msg->buffer, msg->size);
+		ret = amdgpu_atombios_dp_process_aux_ch(chan,
+						 tx_buf, tx_size, NULL, 0, delay, &ack);
+		if (ret >= 0)
+			/* Return payload size. */
+			ret = msg->size;
+		break;
+	case DP_AUX_NATIVE_READ:
+	case DP_AUX_I2C_READ:
+		/* tx_size needs to be 4 even for bare address packets since the atom
+		 * table needs the info in tx_buf[3].
+		 */
+		tx_size = HEADER_SIZE;
+		if (msg->size == 0)
+			tx_buf[3] |= BARE_ADDRESS_SIZE << 4;
+		else
+			tx_buf[3] |= tx_size << 4;
+		ret = amdgpu_atombios_dp_process_aux_ch(chan,
+						 tx_buf, tx_size, msg->buffer, msg->size, delay, &ack);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
+	if (ret >= 0)
+		msg->reply = ack >> 4;
+
+	return ret;
+}
+
+void amdgpu_atombios_dp_aux_init(struct amdgpu_connector *amdgpu_connector)
+{
+	int ret;
+
+	amdgpu_connector->ddc_bus->rec.hpd = amdgpu_connector->hpd.hpd;
+	amdgpu_connector->ddc_bus->aux.dev = amdgpu_connector->base.kdev;
+	amdgpu_connector->ddc_bus->aux.transfer = amdgpu_atombios_dp_aux_transfer;
+	ret = drm_dp_aux_register(&amdgpu_connector->ddc_bus->aux);
+	if (!ret)
+		amdgpu_connector->ddc_bus->has_aux = true;
+
+	WARN(ret, "drm_dp_aux_register_i2c_bus() failed with error %d\n", ret);
+}
+
+/***** general DP utility functions *****/
+
+#define DP_VOLTAGE_MAX         DP_TRAIN_VOLTAGE_SWING_LEVEL_3
+#define DP_PRE_EMPHASIS_MAX    DP_TRAIN_PRE_EMPH_LEVEL_3
+
+static void amdgpu_atombios_dp_get_adjust_train(const u8 link_status[DP_LINK_STATUS_SIZE],
+						int lane_count,
+						u8 train_set[4])
+{
+	u8 v = 0;
+	u8 p = 0;
+	int lane;
+
+	for (lane = 0; lane < lane_count; lane++) {
+		u8 this_v = drm_dp_get_adjust_request_voltage(link_status, lane);
+		u8 this_p = drm_dp_get_adjust_request_pre_emphasis(link_status, lane);
+
+		DRM_DEBUG_KMS("requested signal parameters: lane %d voltage %s pre_emph %s\n",
+			  lane,
+			  voltage_names[this_v >> DP_TRAIN_VOLTAGE_SWING_SHIFT],
+			  pre_emph_names[this_p >> DP_TRAIN_PRE_EMPHASIS_SHIFT]);
+
+		if (this_v > v)
+			v = this_v;
+		if (this_p > p)
+			p = this_p;
+	}
+
+	if (v >= DP_VOLTAGE_MAX)
+		v |= DP_TRAIN_MAX_SWING_REACHED;
+
+	if (p >= DP_PRE_EMPHASIS_MAX)
+		p |= DP_TRAIN_MAX_PRE_EMPHASIS_REACHED;
+
+	DRM_DEBUG_KMS("using signal parameters: voltage %s pre_emph %s\n",
+		  voltage_names[(v & DP_TRAIN_VOLTAGE_SWING_MASK) >> DP_TRAIN_VOLTAGE_SWING_SHIFT],
+		  pre_emph_names[(p & DP_TRAIN_PRE_EMPHASIS_MASK) >> DP_TRAIN_PRE_EMPHASIS_SHIFT]);
+
+	for (lane = 0; lane < 4; lane++)
+		train_set[lane] = v | p;
+}
+
+/* convert bits per color to bits per pixel */
+/* get bpc from the EDID */
+static int amdgpu_atombios_dp_convert_bpc_to_bpp(int bpc)
+{
+	if (bpc == 0)
+		return 24;
+	else
+		return bpc * 3;
+}
+
+/* get the max pix clock supported by the link rate and lane num */
+static int amdgpu_atombios_dp_get_max_dp_pix_clock(int link_rate,
+					    int lane_num,
+					    int bpp)
+{
+	return (link_rate * lane_num * 8) / bpp;
+}
+
+/***** amdgpu specific DP functions *****/
+
+/* First get the min lane# when low rate is used according to pixel clock
+ * (prefer low rate), second check max lane# supported by DP panel,
+ * if the max lane# < low rate lane# then use max lane# instead.
+ */
+static int amdgpu_atombios_dp_get_dp_lane_number(struct drm_connector *connector,
+						 const u8 dpcd[DP_DPCD_SIZE],
+						 int pix_clock)
+{
+	int bpp = amdgpu_atombios_dp_convert_bpc_to_bpp(amdgpu_connector_get_monitor_bpc(connector));
+	int max_link_rate = drm_dp_max_link_rate(dpcd);
+	int max_lane_num = drm_dp_max_lane_count(dpcd);
+	int lane_num;
+	int max_dp_pix_clock;
+
+	for (lane_num = 1; lane_num < max_lane_num; lane_num <<= 1) {
+		max_dp_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(max_link_rate, lane_num, bpp);
+		if (pix_clock <= max_dp_pix_clock)
+			break;
+	}
+
+	return lane_num;
+}
+
+static int amdgpu_atombios_dp_get_dp_link_clock(struct drm_connector *connector,
+						const u8 dpcd[DP_DPCD_SIZE],
+						int pix_clock)
+{
+	int bpp = amdgpu_atombios_dp_convert_bpc_to_bpp(amdgpu_connector_get_monitor_bpc(connector));
+	int lane_num, max_pix_clock;
+
+	if (amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector) ==
+	    ENCODER_OBJECT_ID_NUTMEG)
+		return 270000;
+
+	lane_num = amdgpu_atombios_dp_get_dp_lane_number(connector, dpcd, pix_clock);
+	max_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(162000, lane_num, bpp);
+	if (pix_clock <= max_pix_clock)
+		return 162000;
+	max_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(270000, lane_num, bpp);
+	if (pix_clock <= max_pix_clock)
+		return 270000;
+	if (amdgpu_connector_is_dp12_capable(connector)) {
+		max_pix_clock = amdgpu_atombios_dp_get_max_dp_pix_clock(540000, lane_num, bpp);
+		if (pix_clock <= max_pix_clock)
+			return 540000;
+	}
+
+	return drm_dp_max_link_rate(dpcd);
+}
+
+static u8 amdgpu_atombios_dp_encoder_service(struct amdgpu_device *adev,
+				      int action, int dp_clock,
+				      u8 ucconfig, u8 lane_num)
+{
+	DP_ENCODER_SERVICE_PARAMETERS args;
+	int index = GetIndexIntoMasterTable(COMMAND, DPEncoderService);
+
+	memset(&args, 0, sizeof(args));
+	args.ucLinkClock = dp_clock / 10;
+	args.ucConfig = ucconfig;
+	args.ucAction = action;
+	args.ucLaneNum = lane_num;
+	args.ucStatus = 0;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+	return args.ucStatus;
+}
+
+u8 amdgpu_atombios_dp_get_sinktype(struct amdgpu_connector *amdgpu_connector)
+{
+	struct drm_device *dev = amdgpu_connector->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	return amdgpu_atombios_dp_encoder_service(adev, ATOM_DP_ACTION_GET_SINK_TYPE, 0,
+					   amdgpu_connector->ddc_bus->rec.i2c_id, 0);
+}
+
+static void amdgpu_atombios_dp_probe_oui(struct amdgpu_connector *amdgpu_connector)
+{
+	struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv;
+	u8 buf[3];
+
+	if (!(dig_connector->dpcd[DP_DOWN_STREAM_PORT_COUNT] & DP_OUI_SUPPORT))
+		return;
+
+	if (drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_SINK_OUI, buf, 3) == 3)
+		DRM_DEBUG_KMS("Sink OUI: %02hx%02hx%02hx\n",
+			      buf[0], buf[1], buf[2]);
+
+	if (drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_BRANCH_OUI, buf, 3) == 3)
+		DRM_DEBUG_KMS("Branch OUI: %02hx%02hx%02hx\n",
+			      buf[0], buf[1], buf[2]);
+}
+
+int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector)
+{
+	struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv;
+	u8 msg[DP_DPCD_SIZE];
+	int ret, i;
+
+	for (i = 0; i < 7; i++) {
+		ret = drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_DPCD_REV, msg,
+				       DP_DPCD_SIZE);
+		if (ret == DP_DPCD_SIZE) {
+			memcpy(dig_connector->dpcd, msg, DP_DPCD_SIZE);
+
+			DRM_DEBUG_KMS("DPCD: %*ph\n", (int)sizeof(dig_connector->dpcd),
+				      dig_connector->dpcd);
+
+			amdgpu_atombios_dp_probe_oui(amdgpu_connector);
+
+			return 0;
+		}
+	}
+	dig_connector->dpcd[0] = 0;
+	return -EINVAL;
+}
+
+int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
+			       struct drm_connector *connector)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	int panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
+	u16 dp_bridge = amdgpu_connector_encoder_get_dp_bridge_encoder_id(connector);
+	u8 tmp;
+
+	if (!amdgpu_connector->con_priv)
+		return panel_mode;
+
+	dig_connector = amdgpu_connector->con_priv;
+
+	if (dp_bridge != ENCODER_OBJECT_ID_NONE) {
+		/* DP bridge chips */
+		if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux,
+				      DP_EDP_CONFIGURATION_CAP, &tmp) == 1) {
+			if (tmp & 1)
+				panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE;
+			else if ((dp_bridge == ENCODER_OBJECT_ID_NUTMEG) ||
+				 (dp_bridge == ENCODER_OBJECT_ID_TRAVIS))
+				panel_mode = DP_PANEL_MODE_INTERNAL_DP1_MODE;
+			else
+				panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
+		}
+	} else if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+		/* eDP */
+		if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux,
+				      DP_EDP_CONFIGURATION_CAP, &tmp) == 1) {
+			if (tmp & 1)
+				panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE;
+		}
+	}
+
+	return panel_mode;
+}
+
+void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
+				 const struct drm_display_mode *mode)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	if (!amdgpu_connector->con_priv)
+		return;
+	dig_connector = amdgpu_connector->con_priv;
+
+	if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+	    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
+		dig_connector->dp_clock =
+			amdgpu_atombios_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock);
+		dig_connector->dp_lane_count =
+			amdgpu_atombios_dp_get_dp_lane_number(connector, dig_connector->dpcd, mode->clock);
+	}
+}
+
+int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
+				  struct drm_display_mode *mode)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+	int dp_clock;
+
+	if (!amdgpu_connector->con_priv)
+		return MODE_CLOCK_HIGH;
+	dig_connector = amdgpu_connector->con_priv;
+
+	dp_clock =
+		amdgpu_atombios_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock);
+
+	if ((dp_clock == 540000) &&
+	    (!amdgpu_connector_is_dp12_capable(connector)))
+		return MODE_CLOCK_HIGH;
+
+	return MODE_OK;
+}
+
+bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connector)
+{
+	u8 link_status[DP_LINK_STATUS_SIZE];
+	struct amdgpu_connector_atom_dig *dig = amdgpu_connector->con_priv;
+
+	if (drm_dp_dpcd_read_link_status(&amdgpu_connector->ddc_bus->aux, link_status)
+	    <= 0)
+		return false;
+	if (drm_dp_channel_eq_ok(link_status, dig->dp_lane_count))
+		return false;
+	return true;
+}
+
+void amdgpu_atombios_dp_set_rx_power_state(struct drm_connector *connector,
+				    u8 power_state)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	if (!amdgpu_connector->con_priv)
+		return;
+
+	dig_connector = amdgpu_connector->con_priv;
+
+	/* power up/down the sink */
+	if (dig_connector->dpcd[0] >= 0x11) {
+		drm_dp_dpcd_writeb(&amdgpu_connector->ddc_bus->aux,
+				   DP_SET_POWER, power_state);
+		usleep_range(1000, 2000);
+	}
+}
+
+struct amdgpu_atombios_dp_link_train_info {
+	struct amdgpu_device *adev;
+	struct drm_encoder *encoder;
+	struct drm_connector *connector;
+	int dp_clock;
+	int dp_lane_count;
+	bool tp3_supported;
+	u8 dpcd[DP_RECEIVER_CAP_SIZE];
+	u8 train_set[4];
+	u8 link_status[DP_LINK_STATUS_SIZE];
+	u8 tries;
+	struct drm_dp_aux *aux;
+};
+
+static void
+amdgpu_atombios_dp_update_vs_emph(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	/* set the initial vs/emph on the source */
+	amdgpu_atombios_encoder_setup_dig_transmitter(dp_info->encoder,
+					       ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH,
+					       0, dp_info->train_set[0]); /* sets all lanes at once */
+
+	/* set the vs/emph on the sink */
+	drm_dp_dpcd_write(dp_info->aux, DP_TRAINING_LANE0_SET,
+			  dp_info->train_set, dp_info->dp_lane_count);
+}
+
+static void
+amdgpu_atombios_dp_set_tp(struct amdgpu_atombios_dp_link_train_info *dp_info, int tp)
+{
+	int rtp = 0;
+
+	/* set training pattern on the source */
+	switch (tp) {
+	case DP_TRAINING_PATTERN_1:
+		rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1;
+		break;
+	case DP_TRAINING_PATTERN_2:
+		rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2;
+		break;
+	case DP_TRAINING_PATTERN_3:
+		rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN3;
+			break;
+	}
+	amdgpu_atombios_encoder_setup_dig_encoder(dp_info->encoder, rtp, 0);
+
+	/* enable training pattern on the sink */
+	drm_dp_dpcd_writeb(dp_info->aux, DP_TRAINING_PATTERN_SET, tp);
+}
+
+static int
+amdgpu_atombios_dp_link_train_init(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(dp_info->encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u8 tmp;
+
+	/* power up the sink */
+	amdgpu_atombios_dp_set_rx_power_state(dp_info->connector, DP_SET_POWER_D0);
+
+	/* possibly enable downspread on the sink */
+	if (dp_info->dpcd[3] & 0x1)
+		drm_dp_dpcd_writeb(dp_info->aux,
+				   DP_DOWNSPREAD_CTRL, DP_SPREAD_AMP_0_5);
+	else
+		drm_dp_dpcd_writeb(dp_info->aux,
+				   DP_DOWNSPREAD_CTRL, 0);
+
+	if (dig->panel_mode == DP_PANEL_MODE_INTERNAL_DP2_MODE)
+		drm_dp_dpcd_writeb(dp_info->aux, DP_EDP_CONFIGURATION_SET, 1);
+
+	/* set the lane count on the sink */
+	tmp = dp_info->dp_lane_count;
+	if (drm_dp_enhanced_frame_cap(dp_info->dpcd))
+		tmp |= DP_LANE_COUNT_ENHANCED_FRAME_EN;
+	drm_dp_dpcd_writeb(dp_info->aux, DP_LANE_COUNT_SET, tmp);
+
+	/* set the link rate on the sink */
+	tmp = drm_dp_link_rate_to_bw_code(dp_info->dp_clock);
+	drm_dp_dpcd_writeb(dp_info->aux, DP_LINK_BW_SET, tmp);
+
+	/* start training on the source */
+	amdgpu_atombios_encoder_setup_dig_encoder(dp_info->encoder,
+					   ATOM_ENCODER_CMD_DP_LINK_TRAINING_START, 0);
+
+	/* disable the training pattern on the sink */
+	drm_dp_dpcd_writeb(dp_info->aux,
+			   DP_TRAINING_PATTERN_SET,
+			   DP_TRAINING_PATTERN_DISABLE);
+
+	return 0;
+}
+
+static int
+amdgpu_atombios_dp_link_train_finish(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	udelay(400);
+
+	/* disable the training pattern on the sink */
+	drm_dp_dpcd_writeb(dp_info->aux,
+			   DP_TRAINING_PATTERN_SET,
+			   DP_TRAINING_PATTERN_DISABLE);
+
+	/* disable the training pattern on the source */
+	amdgpu_atombios_encoder_setup_dig_encoder(dp_info->encoder,
+					   ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE, 0);
+
+	return 0;
+}
+
+static int
+amdgpu_atombios_dp_link_train_cr(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	bool clock_recovery;
+	u8 voltage;
+	int i;
+
+	amdgpu_atombios_dp_set_tp(dp_info, DP_TRAINING_PATTERN_1);
+	memset(dp_info->train_set, 0, 4);
+	amdgpu_atombios_dp_update_vs_emph(dp_info);
+
+	udelay(400);
+
+	/* clock recovery loop */
+	clock_recovery = false;
+	dp_info->tries = 0;
+	voltage = 0xff;
+	while (1) {
+		drm_dp_link_train_clock_recovery_delay(dp_info->dpcd);
+
+		if (drm_dp_dpcd_read_link_status(dp_info->aux,
+						 dp_info->link_status) <= 0) {
+			DRM_ERROR("displayport link status failed\n");
+			break;
+		}
+
+		if (drm_dp_clock_recovery_ok(dp_info->link_status, dp_info->dp_lane_count)) {
+			clock_recovery = true;
+			break;
+		}
+
+		for (i = 0; i < dp_info->dp_lane_count; i++) {
+			if ((dp_info->train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0)
+				break;
+		}
+		if (i == dp_info->dp_lane_count) {
+			DRM_ERROR("clock recovery reached max voltage\n");
+			break;
+		}
+
+		if ((dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK) == voltage) {
+			++dp_info->tries;
+			if (dp_info->tries == 5) {
+				DRM_ERROR("clock recovery tried 5 times\n");
+				break;
+			}
+		} else
+			dp_info->tries = 0;
+
+		voltage = dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK;
+
+		/* Compute new train_set as requested by sink */
+		amdgpu_atombios_dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count,
+					     dp_info->train_set);
+
+		amdgpu_atombios_dp_update_vs_emph(dp_info);
+	}
+	if (!clock_recovery) {
+		DRM_ERROR("clock recovery failed\n");
+		return -1;
+	} else {
+		DRM_DEBUG_KMS("clock recovery at voltage %d pre-emphasis %d\n",
+			  dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK,
+			  (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) >>
+			  DP_TRAIN_PRE_EMPHASIS_SHIFT);
+		return 0;
+	}
+}
+
+static int
+amdgpu_atombios_dp_link_train_ce(struct amdgpu_atombios_dp_link_train_info *dp_info)
+{
+	bool channel_eq;
+
+	if (dp_info->tp3_supported)
+		amdgpu_atombios_dp_set_tp(dp_info, DP_TRAINING_PATTERN_3);
+	else
+		amdgpu_atombios_dp_set_tp(dp_info, DP_TRAINING_PATTERN_2);
+
+	/* channel equalization loop */
+	dp_info->tries = 0;
+	channel_eq = false;
+	while (1) {
+		drm_dp_link_train_channel_eq_delay(dp_info->dpcd);
+
+		if (drm_dp_dpcd_read_link_status(dp_info->aux,
+						 dp_info->link_status) <= 0) {
+			DRM_ERROR("displayport link status failed\n");
+			break;
+		}
+
+		if (drm_dp_channel_eq_ok(dp_info->link_status, dp_info->dp_lane_count)) {
+			channel_eq = true;
+			break;
+		}
+
+		/* Try 5 times */
+		if (dp_info->tries > 5) {
+			DRM_ERROR("channel eq failed: 5 tries\n");
+			break;
+		}
+
+		/* Compute new train_set as requested by sink */
+		amdgpu_atombios_dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count,
+					     dp_info->train_set);
+
+		amdgpu_atombios_dp_update_vs_emph(dp_info);
+		dp_info->tries++;
+	}
+
+	if (!channel_eq) {
+		DRM_ERROR("channel eq failed\n");
+		return -1;
+	} else {
+		DRM_DEBUG_KMS("channel eq at voltage %d pre-emphasis %d\n",
+			  dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK,
+			  (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK)
+			  >> DP_TRAIN_PRE_EMPHASIS_SHIFT);
+		return 0;
+	}
+}
+
+void amdgpu_atombios_dp_link_train(struct drm_encoder *encoder,
+			    struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *dig_connector;
+	struct amdgpu_atombios_dp_link_train_info dp_info;
+	u8 tmp;
+
+	if (!amdgpu_encoder->enc_priv)
+		return;
+	dig = amdgpu_encoder->enc_priv;
+
+	amdgpu_connector = to_amdgpu_connector(connector);
+	if (!amdgpu_connector->con_priv)
+		return;
+	dig_connector = amdgpu_connector->con_priv;
+
+	if ((dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_DISPLAYPORT) &&
+	    (dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_eDP))
+		return;
+
+	if (drm_dp_dpcd_readb(&amdgpu_connector->ddc_bus->aux, DP_MAX_LANE_COUNT, &tmp)
+	    == 1) {
+		if (tmp & DP_TPS3_SUPPORTED)
+			dp_info.tp3_supported = true;
+		else
+			dp_info.tp3_supported = false;
+	} else {
+		dp_info.tp3_supported = false;
+	}
+
+	memcpy(dp_info.dpcd, dig_connector->dpcd, DP_RECEIVER_CAP_SIZE);
+	dp_info.adev = adev;
+	dp_info.encoder = encoder;
+	dp_info.connector = connector;
+	dp_info.dp_lane_count = dig_connector->dp_lane_count;
+	dp_info.dp_clock = dig_connector->dp_clock;
+	dp_info.aux = &amdgpu_connector->ddc_bus->aux;
+
+	if (amdgpu_atombios_dp_link_train_init(&dp_info))
+		goto done;
+	if (amdgpu_atombios_dp_link_train_cr(&dp_info))
+		goto done;
+	if (amdgpu_atombios_dp_link_train_ce(&dp_info))
+		goto done;
+done:
+	if (amdgpu_atombios_dp_link_train_finish(&dp_info))
+		return;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.h b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
new file mode 100644
index 000000000..f59d85ead
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_dp.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_DP_H__
+#define __ATOMBIOS_DP_H__
+
+void amdgpu_atombios_dp_aux_init(struct amdgpu_connector *amdgpu_connector);
+u8 amdgpu_atombios_dp_get_sinktype(struct amdgpu_connector *amdgpu_connector);
+int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector);
+int amdgpu_atombios_dp_get_panel_mode(struct drm_encoder *encoder,
+			       struct drm_connector *connector);
+void amdgpu_atombios_dp_set_link_config(struct drm_connector *connector,
+				 const struct drm_display_mode *mode);
+int amdgpu_atombios_dp_mode_valid_helper(struct drm_connector *connector,
+				  struct drm_display_mode *mode);
+bool amdgpu_atombios_dp_needs_link_train(struct amdgpu_connector *amdgpu_connector);
+void amdgpu_atombios_dp_set_rx_power_state(struct drm_connector *connector,
+				    u8 power_state);
+void amdgpu_atombios_dp_link_train(struct drm_encoder *encoder,
+			    struct drm_connector *connector);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
new file mode 100644
index 000000000..1e0bba29e
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
@@ -0,0 +1,2065 @@
+/*
+ * Copyright 2007-11 Advanced Micro Devices, Inc.
+ * Copyright 2008 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Dave Airlie
+ *          Alex Deucher
+ */
+#include <drm/drmP.h>
+#include <drm/drm_crtc_helper.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "amdgpu_connectors.h"
+#include "atom.h"
+#include "atombios_encoders.h"
+#include "atombios_dp.h"
+#include <linux/backlight.h>
+#include "bif/bif_4_1_d.h"
+
+static u8
+amdgpu_atombios_encoder_get_backlight_level_from_reg(struct amdgpu_device *adev)
+{
+	u8 backlight_level;
+	u32 bios_2_scratch;
+
+	bios_2_scratch = RREG32(mmBIOS_SCRATCH_2);
+
+	backlight_level = ((bios_2_scratch & ATOM_S2_CURRENT_BL_LEVEL_MASK) >>
+			   ATOM_S2_CURRENT_BL_LEVEL_SHIFT);
+
+	return backlight_level;
+}
+
+static void
+amdgpu_atombios_encoder_set_backlight_level_to_reg(struct amdgpu_device *adev,
+					    u8 backlight_level)
+{
+	u32 bios_2_scratch;
+
+	bios_2_scratch = RREG32(mmBIOS_SCRATCH_2);
+
+	bios_2_scratch &= ~ATOM_S2_CURRENT_BL_LEVEL_MASK;
+	bios_2_scratch |= ((backlight_level << ATOM_S2_CURRENT_BL_LEVEL_SHIFT) &
+			   ATOM_S2_CURRENT_BL_LEVEL_MASK);
+
+	WREG32(mmBIOS_SCRATCH_2, bios_2_scratch);
+}
+
+u8
+amdgpu_atombios_encoder_get_backlight_level(struct amdgpu_encoder *amdgpu_encoder)
+{
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return 0;
+
+	return amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+}
+
+void
+amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encoder,
+				     u8 level)
+{
+	struct drm_encoder *encoder = &amdgpu_encoder->base;
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder_atom_dig *dig;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return;
+
+	if ((amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) &&
+	    amdgpu_encoder->enc_priv) {
+		dig = amdgpu_encoder->enc_priv;
+		dig->backlight_level = level;
+		amdgpu_atombios_encoder_set_backlight_level_to_reg(adev, dig->backlight_level);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+			if (dig->backlight_level == 0)
+				amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+								       ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0);
+			else {
+				amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+								       ATOM_TRANSMITTER_ACTION_BL_BRIGHTNESS_CONTROL, 0, 0);
+				amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+								       ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0);
+			}
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+#if defined(CONFIG_BACKLIGHT_CLASS_DEVICE) || defined(CONFIG_BACKLIGHT_CLASS_DEVICE_MODULE)
+
+static u8 amdgpu_atombios_encoder_backlight_level(struct backlight_device *bd)
+{
+	u8 level;
+
+	/* Convert brightness to hardware level */
+	if (bd->props.brightness < 0)
+		level = 0;
+	else if (bd->props.brightness > AMDGPU_MAX_BL_LEVEL)
+		level = AMDGPU_MAX_BL_LEVEL;
+	else
+		level = bd->props.brightness;
+
+	return level;
+}
+
+static int amdgpu_atombios_encoder_update_backlight_status(struct backlight_device *bd)
+{
+	struct amdgpu_backlight_privdata *pdata = bl_get_data(bd);
+	struct amdgpu_encoder *amdgpu_encoder = pdata->encoder;
+
+	amdgpu_atombios_encoder_set_backlight_level(amdgpu_encoder,
+					     amdgpu_atombios_encoder_backlight_level(bd));
+
+	return 0;
+}
+
+static int
+amdgpu_atombios_encoder_get_backlight_brightness(struct backlight_device *bd)
+{
+	struct amdgpu_backlight_privdata *pdata = bl_get_data(bd);
+	struct amdgpu_encoder *amdgpu_encoder = pdata->encoder;
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	return amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+}
+
+static const struct backlight_ops amdgpu_atombios_encoder_backlight_ops = {
+	.get_brightness = amdgpu_atombios_encoder_get_backlight_brightness,
+	.update_status	= amdgpu_atombios_encoder_update_backlight_status,
+};
+
+void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encoder,
+				     struct drm_connector *drm_connector)
+{
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct backlight_device *bd;
+	struct backlight_properties props;
+	struct amdgpu_backlight_privdata *pdata;
+	struct amdgpu_encoder_atom_dig *dig;
+	u8 backlight_level;
+	char bl_name[16];
+
+	/* Mac laptops with multiple GPUs use the gmux driver for backlight
+	 * so don't register a backlight device
+	 */
+	if ((adev->pdev->subsystem_vendor == PCI_VENDOR_ID_APPLE) &&
+	    (adev->pdev->device == 0x6741))
+		return;
+
+	if (!amdgpu_encoder->enc_priv)
+		return;
+
+	if (!adev->is_atom_bios)
+		return;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return;
+
+	pdata = kmalloc(sizeof(struct amdgpu_backlight_privdata), GFP_KERNEL);
+	if (!pdata) {
+		DRM_ERROR("Memory allocation failed\n");
+		goto error;
+	}
+
+	memset(&props, 0, sizeof(props));
+	props.max_brightness = AMDGPU_MAX_BL_LEVEL;
+	props.type = BACKLIGHT_RAW;
+	snprintf(bl_name, sizeof(bl_name),
+		 "amdgpu_bl%d", dev->primary->index);
+	bd = backlight_device_register(bl_name, drm_connector->kdev,
+				       pdata, &amdgpu_atombios_encoder_backlight_ops, &props);
+	if (IS_ERR(bd)) {
+		DRM_ERROR("Backlight registration failed\n");
+		goto error;
+	}
+
+	pdata->encoder = amdgpu_encoder;
+
+	backlight_level = amdgpu_atombios_encoder_get_backlight_level_from_reg(adev);
+
+	dig = amdgpu_encoder->enc_priv;
+	dig->bl_dev = bd;
+
+	bd->props.brightness = amdgpu_atombios_encoder_get_backlight_brightness(bd);
+	bd->props.power = FB_BLANK_UNBLANK;
+	backlight_update_status(bd);
+
+	DRM_INFO("amdgpu atom DIG backlight initialized\n");
+
+	return;
+
+error:
+	kfree(pdata);
+	return;
+}
+
+void
+amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *amdgpu_encoder)
+{
+	struct drm_device *dev = amdgpu_encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct backlight_device *bd = NULL;
+	struct amdgpu_encoder_atom_dig *dig;
+
+	if (!amdgpu_encoder->enc_priv)
+		return;
+
+	if (!adev->is_atom_bios)
+		return;
+
+	if (!(adev->mode_info.firmware_flags & ATOM_BIOS_INFO_BL_CONTROLLED_BY_GPU))
+		return;
+
+	dig = amdgpu_encoder->enc_priv;
+	bd = dig->bl_dev;
+	dig->bl_dev = NULL;
+
+	if (bd) {
+		struct amdgpu_legacy_backlight_privdata *pdata;
+
+		pdata = bl_get_data(bd);
+		backlight_device_unregister(bd);
+		kfree(pdata);
+
+		DRM_INFO("amdgpu atom LVDS backlight unloaded\n");
+	}
+}
+
+#else /* !CONFIG_BACKLIGHT_CLASS_DEVICE */
+
+void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *encoder)
+{
+}
+
+void amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *encoder)
+{
+}
+
+#endif
+
+bool amdgpu_atombios_encoder_is_digital(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return true;
+	default:
+		return false;
+	}
+}
+
+bool amdgpu_atombios_encoder_mode_fixup(struct drm_encoder *encoder,
+				 const struct drm_display_mode *mode,
+				 struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	/* set the active encoder to connector routing */
+	amdgpu_encoder_set_active_device(encoder);
+	drm_mode_set_crtcinfo(adjusted_mode, 0);
+
+	/* hw bug */
+	if ((mode->flags & DRM_MODE_FLAG_INTERLACE)
+	    && (mode->crtc_vsync_start < (mode->crtc_vdisplay + 2)))
+		adjusted_mode->crtc_vsync_start = adjusted_mode->crtc_vdisplay + 2;
+
+	/* get the native mode for scaling */
+	if (amdgpu_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_panel_mode_fixup(encoder, adjusted_mode);
+	else if (amdgpu_encoder->rmx_type != RMX_OFF)
+		amdgpu_panel_mode_fixup(encoder, adjusted_mode);
+
+	if ((amdgpu_encoder->active_device & (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE)) {
+		struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+		amdgpu_atombios_dp_set_link_config(connector, adjusted_mode);
+	}
+
+	return true;
+}
+
+static void
+amdgpu_atombios_encoder_setup_dac(struct drm_encoder *encoder, int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	DAC_ENCODER_CONTROL_PS_ALLOCATION args;
+	int index = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+		index = GetIndexIntoMasterTable(COMMAND, DAC1EncoderControl);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_DAC2:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		index = GetIndexIntoMasterTable(COMMAND, DAC2EncoderControl);
+		break;
+	}
+
+	args.ucAction = action;
+	args.ucDacStandard = ATOM_DAC1_PS2;
+	args.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+}
+
+static u8 amdgpu_atombios_encoder_get_bpc(struct drm_encoder *encoder)
+{
+	int bpc = 8;
+
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	switch (bpc) {
+	case 0:
+		return PANEL_BPC_UNDEFINE;
+	case 6:
+		return PANEL_6BIT_PER_COLOR;
+	case 8:
+	default:
+		return PANEL_8BIT_PER_COLOR;
+	case 10:
+		return PANEL_10BIT_PER_COLOR;
+	case 12:
+		return PANEL_12BIT_PER_COLOR;
+	case 16:
+		return PANEL_16BIT_PER_COLOR;
+	}
+}
+
+union dvo_encoder_control {
+	ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION ext_tmds;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION dvo;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION_V3 dvo_v3;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION_V1_4 dvo_v4;
+};
+
+static void
+amdgpu_atombios_encoder_setup_dvo(struct drm_encoder *encoder, int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	union dvo_encoder_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl);
+	uint8_t frev, crev;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			/* R4xx, R5xx */
+			args.ext_tmds.sXTmdsEncoder.ucEnable = action;
+
+			if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.ext_tmds.sXTmdsEncoder.ucMisc |= PANEL_ENCODER_MISC_DUAL;
+
+			args.ext_tmds.sXTmdsEncoder.ucMisc |= ATOM_PANEL_MISC_888RGB;
+			break;
+		case 2:
+			/* RS600/690/740 */
+			args.dvo.sDVOEncoder.ucAction = action;
+			args.dvo.sDVOEncoder.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			/* DFP1, CRT1, TV1 depending on the type of port */
+			args.dvo.sDVOEncoder.ucDeviceType = ATOM_DEVICE_DFP1_INDEX;
+
+			if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.dvo.sDVOEncoder.usDevAttr.sDigAttrib.ucAttribute |= PANEL_ENCODER_MISC_DUAL;
+			break;
+		case 3:
+			/* R6xx */
+			args.dvo_v3.ucAction = action;
+			args.dvo_v3.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.dvo_v3.ucDVOConfig = 0; /* XXX */
+			break;
+		case 4:
+			/* DCE8 */
+			args.dvo_v4.ucAction = action;
+			args.dvo_v4.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.dvo_v4.ucDVOConfig = 0; /* XXX */
+			args.dvo_v4.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		break;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector;
+	struct amdgpu_connector_atom_dig *dig_connector;
+
+	/* dp bridges are always DP */
+	if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE)
+		return ATOM_ENCODER_MODE_DP;
+
+	/* DVO is always DVO */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DVO1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1))
+		return ATOM_ENCODER_MODE_DVO;
+
+	connector = amdgpu_get_connector_for_encoder(encoder);
+	/* if we don't have an active device yet, just use one of
+	 * the connectors tied to the encoder.
+	 */
+	if (!connector)
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+	amdgpu_connector = to_amdgpu_connector(connector);
+
+	switch (connector->connector_type) {
+	case DRM_MODE_CONNECTOR_DVII:
+	case DRM_MODE_CONNECTOR_HDMIB: /* HDMI-B is basically DL-DVI; analog works fine */
+		if (amdgpu_audio != 0) {
+			if (amdgpu_connector->use_digital &&
+			    (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE))
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (amdgpu_connector->use_digital)
+				return ATOM_ENCODER_MODE_DVI;
+			else
+				return ATOM_ENCODER_MODE_CRT;
+		} else if (amdgpu_connector->use_digital) {
+			return ATOM_ENCODER_MODE_DVI;
+		} else {
+			return ATOM_ENCODER_MODE_CRT;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_DVID:
+	case DRM_MODE_CONNECTOR_HDMIA:
+	default:
+		if (amdgpu_audio != 0) {
+			if (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE)
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
+				return ATOM_ENCODER_MODE_HDMI;
+			else
+				return ATOM_ENCODER_MODE_DVI;
+		} else {
+			return ATOM_ENCODER_MODE_DVI;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_LVDS:
+		return ATOM_ENCODER_MODE_LVDS;
+		break;
+	case DRM_MODE_CONNECTOR_DisplayPort:
+		dig_connector = amdgpu_connector->con_priv;
+		if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) ||
+		    (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) {
+			return ATOM_ENCODER_MODE_DP;
+		} else if (amdgpu_audio != 0) {
+			if (amdgpu_connector->audio == AMDGPU_AUDIO_ENABLE)
+				return ATOM_ENCODER_MODE_HDMI;
+			else if (drm_detect_hdmi_monitor(amdgpu_connector_edid(connector)) &&
+				 (amdgpu_connector->audio == AMDGPU_AUDIO_AUTO))
+				return ATOM_ENCODER_MODE_HDMI;
+			else
+				return ATOM_ENCODER_MODE_DVI;
+		} else {
+			return ATOM_ENCODER_MODE_DVI;
+		}
+		break;
+	case DRM_MODE_CONNECTOR_eDP:
+		return ATOM_ENCODER_MODE_DP;
+	case DRM_MODE_CONNECTOR_DVIA:
+	case DRM_MODE_CONNECTOR_VGA:
+		return ATOM_ENCODER_MODE_CRT;
+		break;
+	case DRM_MODE_CONNECTOR_Composite:
+	case DRM_MODE_CONNECTOR_SVIDEO:
+	case DRM_MODE_CONNECTOR_9PinDIN:
+		/* fix me */
+		return ATOM_ENCODER_MODE_TV;
+		/*return ATOM_ENCODER_MODE_CV;*/
+		break;
+	}
+}
+
+/*
+ * DIG Encoder/Transmitter Setup
+ *
+ * DCE 6.0
+ * - 3 DIG transmitter blocks UNIPHY0/1/2 (links A and B).
+ * Supports up to 6 digital outputs
+ * - 6 DIG encoder blocks.
+ * - DIG to PHY mapping is hardcoded
+ * DIG1 drives UNIPHY0 link A, A+B
+ * DIG2 drives UNIPHY0 link B
+ * DIG3 drives UNIPHY1 link A, A+B
+ * DIG4 drives UNIPHY1 link B
+ * DIG5 drives UNIPHY2 link A, A+B
+ * DIG6 drives UNIPHY2 link B
+ *
+ * Routing
+ * crtc -> dig encoder -> UNIPHY/LVTMA (1 or 2 links)
+ * Examples:
+ * crtc0 -> dig2 -> LVTMA   links A+B -> TMDS/HDMI
+ * crtc1 -> dig1 -> UNIPHY0 link  B   -> DP
+ * crtc0 -> dig1 -> UNIPHY2 link  A   -> LVDS
+ * crtc1 -> dig2 -> UNIPHY1 link  B+A -> TMDS/HDMI
+ */
+
+union dig_encoder_control {
+	DIG_ENCODER_CONTROL_PS_ALLOCATION v1;
+	DIG_ENCODER_CONTROL_PARAMETERS_V2 v2;
+	DIG_ENCODER_CONTROL_PARAMETERS_V3 v3;
+	DIG_ENCODER_CONTROL_PARAMETERS_V4 v4;
+};
+
+void
+amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
+				   int action, int panel_mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	union dig_encoder_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, DIGxEncoderControl);
+	uint8_t frev, crev;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int hpd_id = AMDGPU_HPD_NONE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		hpd_id = amdgpu_connector->hpd.hpd;
+	}
+
+	/* no dig encoder assigned */
+	if (dig->dig_encoder == -1)
+		return;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			args.v1.ucAction = action;
+			args.v1.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE)
+				args.v3.ucPanelMode = panel_mode;
+			else
+				args.v1.ucEncoderMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v1.ucEncoderMode))
+				args.v1.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v1.ucLaneNum = 8;
+			else
+				args.v1.ucLaneNum = 4;
+
+			if (ENCODER_MODE_IS_DP(args.v1.ucEncoderMode) && (dp_clock == 270000))
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ;
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+				args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER2;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v1.ucConfig = ATOM_ENCODER_CONFIG_V2_TRANSMITTER3;
+				break;
+			}
+			if (dig->linkb)
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_LINKB;
+			else
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_LINKA;
+			break;
+		case 2:
+		case 3:
+			args.v3.ucAction = action;
+			args.v3.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE)
+				args.v3.ucPanelMode = panel_mode;
+			else
+				args.v3.ucEncoderMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v3.ucEncoderMode))
+				args.v3.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v3.ucLaneNum = 8;
+			else
+				args.v3.ucLaneNum = 4;
+
+			if (ENCODER_MODE_IS_DP(args.v3.ucEncoderMode) && (dp_clock == 270000))
+				args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ;
+			args.v3.acConfig.ucDigSel = dig->dig_encoder;
+			args.v3.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			break;
+		case 4:
+			args.v4.ucAction = action;
+			args.v4.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE)
+				args.v4.ucPanelMode = panel_mode;
+			else
+				args.v4.ucEncoderMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v4.ucEncoderMode))
+				args.v4.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v4.ucLaneNum = 8;
+			else
+				args.v4.ucLaneNum = 4;
+
+			if (ENCODER_MODE_IS_DP(args.v4.ucEncoderMode)) {
+				if (dp_clock == 540000)
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_5_40GHZ;
+				else if (dp_clock == 324000)
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_3_24GHZ;
+				else if (dp_clock == 270000)
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_2_70GHZ;
+				else
+					args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_1_62GHZ;
+			}
+			args.v4.acConfig.ucDigSel = dig->dig_encoder;
+			args.v4.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			if (hpd_id == AMDGPU_HPD_NONE)
+				args.v4.ucHPD_ID = 0;
+			else
+				args.v4.ucHPD_ID = hpd_id + 1;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		break;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+}
+
+union dig_transmitter_control {
+	DIG_TRANSMITTER_CONTROL_PS_ALLOCATION v1;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V2 v2;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V3 v3;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V4 v4;
+	DIG_TRANSMITTER_CONTROL_PARAMETERS_V1_5 v5;
+};
+
+void
+amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int action,
+				       uint8_t lane_num, uint8_t lane_set)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	union dig_transmitter_control args;
+	int index = 0;
+	uint8_t frev, crev;
+	bool is_dp = false;
+	int pll_id = 0;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int connector_object_id = 0;
+	int igp_lane_info = 0;
+	int dig_encoder = dig->dig_encoder;
+	int hpd_id = AMDGPU_HPD_NONE;
+
+	if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+		/* just needed to avoid bailing in the encoder check.  the encoder
+		 * isn't used for init
+		 */
+		dig_encoder = 0;
+	} else
+		connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		hpd_id = amdgpu_connector->hpd.hpd;
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		connector_object_id =
+			(amdgpu_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	}
+
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		pll_id = amdgpu_crtc->pll_id;
+	}
+
+	/* no dig encoder assigned */
+	if (dig_encoder == -1)
+		return;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)))
+		is_dp = true;
+
+	memset(&args, 0, sizeof(args));
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+		index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+		index = GetIndexIntoMasterTable(COMMAND, LVTMATransmitterControl);
+		break;
+	}
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+			args.v1.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v1.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v1.asMode.ucLaneSel = lane_num;
+				args.v1.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v1.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v1.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v1.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			args.v1.ucConfig = ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL;
+
+			if (dig_encoder)
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER;
+			else
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER;
+
+			if ((adev->flags & AMD_IS_APU) &&
+			    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_UNIPHY)) {
+				if (is_dp ||
+				    !amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock)) {
+					if (igp_lane_info & 0x1)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_3;
+					else if (igp_lane_info & 0x2)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_4_7;
+					else if (igp_lane_info & 0x4)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_11;
+					else if (igp_lane_info & 0x8)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_12_15;
+				} else {
+					if (igp_lane_info & 0x3)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_0_7;
+					else if (igp_lane_info & 0xc)
+						args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LANE_8_15;
+				}
+			}
+
+			if (dig->linkb)
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKB;
+			else
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_LINKA;
+
+			if (is_dp)
+				args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_COHERENT;
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_COHERENT;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_8LANE_LINK;
+			}
+			break;
+		case 2:
+			args.v2.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v2.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v2.asMode.ucLaneSel = lane_num;
+				args.v2.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v2.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v2.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v2.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			args.v2.acConfig.ucEncoderSel = dig_encoder;
+			if (dig->linkb)
+				args.v2.acConfig.ucLinkSel = 1;
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v2.acConfig.ucTransmitterSel = 0;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				args.v2.acConfig.ucTransmitterSel = 1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v2.acConfig.ucTransmitterSel = 2;
+				break;
+			}
+
+			if (is_dp) {
+				args.v2.acConfig.fCoherentMode = 1;
+				args.v2.acConfig.fDPConnector = 1;
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v2.acConfig.fCoherentMode = 1;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v2.acConfig.fDualLinkConnector = 1;
+			}
+			break;
+		case 3:
+			args.v3.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v3.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v3.asMode.ucLaneSel = lane_num;
+				args.v3.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v3.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v3.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v3.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			if (is_dp)
+				args.v3.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v3.ucLaneNum = 8;
+			else
+				args.v3.ucLaneNum = 4;
+
+			if (dig->linkb)
+				args.v3.acConfig.ucLinkSel = 1;
+			if (dig_encoder & 1)
+				args.v3.acConfig.ucEncoderSel = 1;
+
+			/* Select the PLL for the PHY
+			 * DP PHY should be clocked from external src if there is
+			 * one.
+			 */
+			/* On DCE4, if there is an external clock, it generates the DP ref clock */
+			if (is_dp && adev->clock.dp_extclk)
+				args.v3.acConfig.ucRefClkSource = 2; /* external src */
+			else
+				args.v3.acConfig.ucRefClkSource = pll_id;
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v3.acConfig.ucTransmitterSel = 0;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				args.v3.acConfig.ucTransmitterSel = 1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v3.acConfig.ucTransmitterSel = 2;
+				break;
+			}
+
+			if (is_dp)
+				args.v3.acConfig.fCoherentMode = 1; /* DP requires coherent */
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v3.acConfig.fCoherentMode = 1;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v3.acConfig.fDualLinkConnector = 1;
+			}
+			break;
+		case 4:
+			args.v4.ucAction = action;
+			if (action == ATOM_TRANSMITTER_ACTION_INIT) {
+				args.v4.usInitInfo = cpu_to_le16(connector_object_id);
+			} else if (action == ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH) {
+				args.v4.asMode.ucLaneSel = lane_num;
+				args.v4.asMode.ucLaneSet = lane_set;
+			} else {
+				if (is_dp)
+					args.v4.usPixelClock = cpu_to_le16(dp_clock / 10);
+				else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v4.usPixelClock = cpu_to_le16((amdgpu_encoder->pixel_clock / 2) / 10);
+				else
+					args.v4.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			}
+
+			if (is_dp)
+				args.v4.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v4.ucLaneNum = 8;
+			else
+				args.v4.ucLaneNum = 4;
+
+			if (dig->linkb)
+				args.v4.acConfig.ucLinkSel = 1;
+			if (dig_encoder & 1)
+				args.v4.acConfig.ucEncoderSel = 1;
+
+			/* Select the PLL for the PHY
+			 * DP PHY should be clocked from external src if there is
+			 * one.
+			 */
+			/* On DCE5 DCPLL usually generates the DP ref clock */
+			if (is_dp) {
+				if (adev->clock.dp_extclk)
+					args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_EXTCLK;
+				else
+					args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_DCPLL;
+			} else
+				args.v4.acConfig.ucRefClkSource = pll_id;
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				args.v4.acConfig.ucTransmitterSel = 0;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				args.v4.acConfig.ucTransmitterSel = 1;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				args.v4.acConfig.ucTransmitterSel = 2;
+				break;
+			}
+
+			if (is_dp)
+				args.v4.acConfig.fCoherentMode = 1; /* DP requires coherent */
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v4.acConfig.fCoherentMode = 1;
+				if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+					args.v4.acConfig.fDualLinkConnector = 1;
+			}
+			break;
+		case 5:
+			args.v5.ucAction = action;
+			if (is_dp)
+				args.v5.usSymClock = cpu_to_le16(dp_clock / 10);
+			else
+				args.v5.usSymClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+				if (dig->linkb)
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYB;
+				else
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYA;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+				if (dig->linkb)
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYD;
+				else
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYC;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+				if (dig->linkb)
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYF;
+				else
+					args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYE;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+				args.v5.ucPhyId = ATOM_PHY_ID_UNIPHYG;
+				break;
+			}
+			if (is_dp)
+				args.v5.ucLaneNum = dp_lane_count;
+			else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v5.ucLaneNum = 8;
+			else
+				args.v5.ucLaneNum = 4;
+			args.v5.ucConnObjId = connector_object_id;
+			args.v5.ucDigMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (is_dp && adev->clock.dp_extclk)
+				args.v5.asConfig.ucPhyClkSrcId = ENCODER_REFCLK_SRC_EXTCLK;
+			else
+				args.v5.asConfig.ucPhyClkSrcId = pll_id;
+
+			if (is_dp)
+				args.v5.asConfig.ucCoherentMode = 1; /* DP requires coherent */
+			else if (amdgpu_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) {
+				if (dig->coherent_mode)
+					args.v5.asConfig.ucCoherentMode = 1;
+			}
+			if (hpd_id == AMDGPU_HPD_NONE)
+				args.v5.asConfig.ucHPDSel = 0;
+			else
+				args.v5.asConfig.ucHPDSel = hpd_id + 1;
+			args.v5.ucDigEncoderSel = 1 << dig_encoder;
+			args.v5.ucDPLaneSet = lane_set;
+			break;
+		default:
+			DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		break;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+bool
+amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
+				     int action)
+{
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_device *dev = amdgpu_connector->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	union dig_transmitter_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl);
+	uint8_t frev, crev;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_eDP)
+		goto done;
+
+	if ((action != ATOM_TRANSMITTER_ACTION_POWER_ON) &&
+	    (action != ATOM_TRANSMITTER_ACTION_POWER_OFF))
+		goto done;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		goto done;
+
+	memset(&args, 0, sizeof(args));
+
+	args.v1.ucAction = action;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	/* wait for the panel to power up */
+	if (action == ATOM_TRANSMITTER_ACTION_POWER_ON) {
+		int i;
+
+		for (i = 0; i < 300; i++) {
+			if (amdgpu_display_hpd_sense(adev, amdgpu_connector->hpd.hpd))
+				return true;
+			mdelay(1);
+		}
+		return false;
+	}
+done:
+	return true;
+}
+
+union external_encoder_control {
+	EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION v1;
+	EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION_V3 v3;
+};
+
+static void
+amdgpu_atombios_encoder_setup_external_encoder(struct drm_encoder *encoder,
+					struct drm_encoder *ext_encoder,
+					int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder *ext_amdgpu_encoder = to_amdgpu_encoder(ext_encoder);
+	union external_encoder_control args;
+	struct drm_connector *connector;
+	int index = GetIndexIntoMasterTable(COMMAND, ExternalEncoderControl);
+	u8 frev, crev;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int connector_object_id = 0;
+	u32 ext_enum = (ext_amdgpu_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+
+	if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT)
+		connector = amdgpu_get_connector_for_encoder_init(encoder);
+	else
+		connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		struct amdgpu_connector_atom_dig *dig_connector =
+			amdgpu_connector->con_priv;
+
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		connector_object_id =
+			(amdgpu_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	}
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		/* no params on frev 1 */
+		break;
+	case 2:
+		switch (crev) {
+		case 1:
+		case 2:
+			args.v1.sDigEncoder.ucAction = action;
+			args.v1.sDigEncoder.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.v1.sDigEncoder.ucEncoderMode =
+				amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v1.sDigEncoder.ucEncoderMode)) {
+				if (dp_clock == 270000)
+					args.v1.sDigEncoder.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ;
+				args.v1.sDigEncoder.ucLaneNum = dp_lane_count;
+			} else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v1.sDigEncoder.ucLaneNum = 8;
+			else
+				args.v1.sDigEncoder.ucLaneNum = 4;
+			break;
+		case 3:
+			args.v3.sExtEncoder.ucAction = action;
+			if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT)
+				args.v3.sExtEncoder.usConnectorId = cpu_to_le16(connector_object_id);
+			else
+				args.v3.sExtEncoder.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
+			args.v3.sExtEncoder.ucEncoderMode =
+				amdgpu_atombios_encoder_get_encoder_mode(encoder);
+
+			if (ENCODER_MODE_IS_DP(args.v3.sExtEncoder.ucEncoderMode)) {
+				if (dp_clock == 270000)
+					args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ;
+				else if (dp_clock == 540000)
+					args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_5_40GHZ;
+				args.v3.sExtEncoder.ucLaneNum = dp_lane_count;
+			} else if (amdgpu_dig_monitor_is_duallink(encoder, amdgpu_encoder->pixel_clock))
+				args.v3.sExtEncoder.ucLaneNum = 8;
+			else
+				args.v3.sExtEncoder.ucLaneNum = 4;
+			switch (ext_enum) {
+			case GRAPH_OBJECT_ENUM_ID1:
+				args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER1;
+				break;
+			case GRAPH_OBJECT_ENUM_ID2:
+				args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER2;
+				break;
+			case GRAPH_OBJECT_ENUM_ID3:
+				args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER3;
+				break;
+			}
+			args.v3.sExtEncoder.ucBitPerColor = amdgpu_atombios_encoder_get_bpc(encoder);
+			break;
+		default:
+			DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+		return;
+	}
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+static void
+amdgpu_atombios_encoder_setup_dig(struct drm_encoder *encoder, int action)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct amdgpu_connector_atom_dig *amdgpu_dig_connector = NULL;
+
+	if (connector) {
+		amdgpu_connector = to_amdgpu_connector(connector);
+		amdgpu_dig_connector = amdgpu_connector->con_priv;
+	}
+
+	if (action == ATOM_ENABLE) {
+		if (!connector)
+			dig->panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE;
+		else
+			dig->panel_mode = amdgpu_atombios_dp_get_panel_mode(encoder, connector);
+
+		/* setup and enable the encoder */
+		amdgpu_atombios_encoder_setup_dig_encoder(encoder, ATOM_ENCODER_CMD_SETUP, 0);
+		amdgpu_atombios_encoder_setup_dig_encoder(encoder,
+						   ATOM_ENCODER_CMD_SETUP_PANEL_MODE,
+						   dig->panel_mode);
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+								EXTERNAL_ENCODER_ACTION_V3_ENCODER_SETUP);
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector) {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_ON);
+				amdgpu_dig_connector->edp_on = true;
+			}
+		}
+		/* enable the transmitter */
+		amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+						       ATOM_TRANSMITTER_ACTION_ENABLE,
+						       0, 0);
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector) {
+			/* DP_SET_POWER_D0 is set in amdgpu_atombios_dp_link_train */
+			amdgpu_atombios_dp_link_train(encoder, connector);
+			amdgpu_atombios_encoder_setup_dig_encoder(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON, 0);
+		}
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			amdgpu_atombios_encoder_set_backlight_level(amdgpu_encoder, dig->backlight_level);
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder, ATOM_ENABLE);
+	} else {
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector)
+			amdgpu_atombios_encoder_setup_dig_encoder(encoder,
+							   ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0);
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder, ATOM_DISABLE);
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+							       ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0);
+
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector)
+			amdgpu_atombios_dp_set_rx_power_state(connector, DP_SET_POWER_D3);
+		/* disable the transmitter */
+		amdgpu_atombios_encoder_setup_dig_transmitter(encoder,
+						       ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0);
+		if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(encoder)) &&
+		    connector) {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+				amdgpu_atombios_encoder_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+				amdgpu_dig_connector->edp_on = false;
+			}
+		}
+	}
+}
+
+void
+amdgpu_atombios_encoder_dpms(struct drm_encoder *encoder, int mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	DRM_DEBUG_KMS("encoder dpms %d to mode %d, devices %08x, active_devices %08x\n",
+		  amdgpu_encoder->encoder_id, mode, amdgpu_encoder->devices,
+		  amdgpu_encoder->active_device);
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+			amdgpu_atombios_encoder_setup_dig(encoder, ATOM_ENABLE);
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			amdgpu_atombios_encoder_setup_dig(encoder, ATOM_DISABLE);
+			break;
+		}
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+			amdgpu_atombios_encoder_setup_dvo(encoder, ATOM_ENABLE);
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			amdgpu_atombios_encoder_setup_dvo(encoder, ATOM_DISABLE);
+			break;
+		}
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+			amdgpu_atombios_encoder_setup_dac(encoder, ATOM_ENABLE);
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			amdgpu_atombios_encoder_setup_dac(encoder, ATOM_DISABLE);
+			break;
+		}
+		break;
+	default:
+		return;
+	}
+}
+
+union crtc_source_param {
+	SELECT_CRTC_SOURCE_PS_ALLOCATION v1;
+	SELECT_CRTC_SOURCE_PARAMETERS_V2 v2;
+	SELECT_CRTC_SOURCE_PARAMETERS_V3 v3;
+};
+
+void
+amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	union crtc_source_param args;
+	int index = GetIndexIntoMasterTable(COMMAND, SelectCRTC_Source);
+	uint8_t frev, crev;
+	struct amdgpu_encoder_atom_dig *dig;
+
+	memset(&args, 0, sizeof(args));
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		switch (crev) {
+		case 1:
+		default:
+			args.v1.ucCRTC = amdgpu_crtc->crtc_id;
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_TMDS1:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1:
+				args.v1.ucDevice = ATOM_DEVICE_DFP1_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_LVDS:
+			case ENCODER_OBJECT_ID_INTERNAL_LVTM1:
+				if (amdgpu_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT)
+					args.v1.ucDevice = ATOM_DEVICE_LCD1_INDEX;
+				else
+					args.v1.ucDevice = ATOM_DEVICE_DFP3_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_DVO1:
+			case ENCODER_OBJECT_ID_INTERNAL_DDI:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+				args.v1.ucDevice = ATOM_DEVICE_DFP2_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_DAC1:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_TV1_INDEX;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_CV_INDEX;
+				else
+					args.v1.ucDevice = ATOM_DEVICE_CRT1_INDEX;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_DAC2:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_TV1_INDEX;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v1.ucDevice = ATOM_DEVICE_CV_INDEX;
+				else
+					args.v1.ucDevice = ATOM_DEVICE_CRT2_INDEX;
+				break;
+			}
+			break;
+		case 2:
+			args.v2.ucCRTC = amdgpu_crtc->crtc_id;
+			if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE) {
+				struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+				if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+				else if (connector->connector_type == DRM_MODE_CONNECTOR_VGA)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_CRT;
+				else
+					args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+				args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+			} else {
+				args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			}
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+				dig = amdgpu_encoder->enc_priv;
+				switch (dig->dig_encoder) {
+				case 0:
+					args.v2.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
+					break;
+				case 1:
+					args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
+					break;
+				case 2:
+					args.v2.ucEncoderID = ASIC_INT_DIG3_ENCODER_ID;
+					break;
+				case 3:
+					args.v2.ucEncoderID = ASIC_INT_DIG4_ENCODER_ID;
+					break;
+				case 4:
+					args.v2.ucEncoderID = ASIC_INT_DIG5_ENCODER_ID;
+					break;
+				case 5:
+					args.v2.ucEncoderID = ASIC_INT_DIG6_ENCODER_ID;
+					break;
+				case 6:
+					args.v2.ucEncoderID = ASIC_INT_DIG7_ENCODER_ID;
+					break;
+				}
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+				args.v2.ucEncoderID = ASIC_INT_DVO_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v2.ucEncoderID = ASIC_INT_DAC1_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v2.ucEncoderID = ASIC_INT_DAC2_ENCODER_ID;
+				break;
+			}
+			break;
+		case 3:
+			args.v3.ucCRTC = amdgpu_crtc->crtc_id;
+			if (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) != ENCODER_OBJECT_ID_NONE) {
+				struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+				if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+				else if (connector->connector_type == DRM_MODE_CONNECTOR_VGA)
+					args.v2.ucEncodeMode = ATOM_ENCODER_MODE_CRT;
+				else
+					args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			} else if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+				args.v2.ucEncodeMode = ATOM_ENCODER_MODE_LVDS;
+			} else {
+				args.v2.ucEncodeMode = amdgpu_atombios_encoder_get_encoder_mode(encoder);
+			}
+			args.v3.ucDstBpc = amdgpu_atombios_encoder_get_bpc(encoder);
+			switch (amdgpu_encoder->encoder_id) {
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+			case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+				dig = amdgpu_encoder->enc_priv;
+				switch (dig->dig_encoder) {
+				case 0:
+					args.v3.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
+					break;
+				case 1:
+					args.v3.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
+					break;
+				case 2:
+					args.v3.ucEncoderID = ASIC_INT_DIG3_ENCODER_ID;
+					break;
+				case 3:
+					args.v3.ucEncoderID = ASIC_INT_DIG4_ENCODER_ID;
+					break;
+				case 4:
+					args.v3.ucEncoderID = ASIC_INT_DIG5_ENCODER_ID;
+					break;
+				case 5:
+					args.v3.ucEncoderID = ASIC_INT_DIG6_ENCODER_ID;
+					break;
+				case 6:
+					args.v3.ucEncoderID = ASIC_INT_DIG7_ENCODER_ID;
+					break;
+				}
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+				args.v3.ucEncoderID = ASIC_INT_DVO_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v3.ucEncoderID = ASIC_INT_DAC1_ENCODER_ID;
+				break;
+			case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+				if (amdgpu_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else if (amdgpu_encoder->active_device & (ATOM_DEVICE_CV_SUPPORT))
+					args.v3.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
+				else
+					args.v3.ucEncoderID = ASIC_INT_DAC2_ENCODER_ID;
+				break;
+			}
+			break;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+		return;
+	}
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+/* This only needs to be called once at startup */
+void
+amdgpu_atombios_encoder_init_dig(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+		struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+
+		switch (amdgpu_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+			amdgpu_atombios_encoder_setup_dig_transmitter(encoder, ATOM_TRANSMITTER_ACTION_INIT,
+							       0, 0);
+			break;
+		}
+
+		if (ext_encoder)
+			amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+								EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT);
+	}
+}
+
+static bool
+amdgpu_atombios_encoder_dac_load_detect(struct drm_encoder *encoder,
+				 struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_TV_SUPPORT |
+				       ATOM_DEVICE_CV_SUPPORT |
+				       ATOM_DEVICE_CRT_SUPPORT)) {
+		DAC_LOAD_DETECTION_PS_ALLOCATION args;
+		int index = GetIndexIntoMasterTable(COMMAND, DAC_LoadDetection);
+		uint8_t frev, crev;
+
+		memset(&args, 0, sizeof(args));
+
+		if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+			return false;
+
+		args.sDacload.ucMisc = 0;
+
+		if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_DAC1) ||
+		    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1))
+			args.sDacload.ucDacType = ATOM_DAC_A;
+		else
+			args.sDacload.ucDacType = ATOM_DAC_B;
+
+		if (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT)
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CRT1_SUPPORT);
+		else if (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT)
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CRT2_SUPPORT);
+		else if (amdgpu_connector->devices & ATOM_DEVICE_CV_SUPPORT) {
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_CV_SUPPORT);
+			if (crev >= 3)
+				args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb;
+		} else if (amdgpu_connector->devices & ATOM_DEVICE_TV1_SUPPORT) {
+			args.sDacload.usDeviceID = cpu_to_le16(ATOM_DEVICE_TV1_SUPPORT);
+			if (crev >= 3)
+				args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb;
+		}
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		return true;
+	} else
+		return false;
+}
+
+enum drm_connector_status
+amdgpu_atombios_encoder_dac_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	uint32_t bios_0_scratch;
+
+	if (!amdgpu_atombios_encoder_dac_load_detect(encoder, connector)) {
+		DRM_DEBUG_KMS("detect returned false \n");
+		return connector_status_unknown;
+	}
+
+	bios_0_scratch = RREG32(mmBIOS_SCRATCH_0);
+
+	DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, amdgpu_encoder->devices);
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT1_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT2_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CV_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A))
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_TV1_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A))
+			return connector_status_connected; /* CTV */
+		else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A))
+			return connector_status_connected; /* STV */
+	}
+	return connector_status_disconnected;
+}
+
+enum drm_connector_status
+amdgpu_atombios_encoder_dig_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+	u32 bios_0_scratch;
+
+	if (!ext_encoder)
+		return connector_status_unknown;
+
+	if ((amdgpu_connector->devices & ATOM_DEVICE_CRT_SUPPORT) == 0)
+		return connector_status_unknown;
+
+	/* load detect on the dp bridge */
+	amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+						EXTERNAL_ENCODER_ACTION_V3_DACLOAD_DETECTION);
+
+	bios_0_scratch = RREG32(mmBIOS_SCRATCH_0);
+
+	DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, amdgpu_encoder->devices);
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT1_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) {
+		if (bios_0_scratch & ATOM_S0_CRT2_MASK)
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_CV_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A))
+			return connector_status_connected;
+	}
+	if (amdgpu_connector->devices & ATOM_DEVICE_TV1_SUPPORT) {
+		if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A))
+			return connector_status_connected; /* CTV */
+		else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A))
+			return connector_status_connected; /* STV */
+	}
+	return connector_status_disconnected;
+}
+
+void
+amdgpu_atombios_encoder_setup_ext_encoder_ddc(struct drm_encoder *encoder)
+{
+	struct drm_encoder *ext_encoder = amdgpu_get_external_encoder(encoder);
+
+	if (ext_encoder)
+		/* ddc_setup on the dp bridge */
+		amdgpu_atombios_encoder_setup_external_encoder(encoder, ext_encoder,
+							EXTERNAL_ENCODER_ACTION_V3_DDC_SETUP);
+
+}
+
+void
+amdgpu_atombios_encoder_set_bios_scratch_regs(struct drm_connector *connector,
+				       struct drm_encoder *encoder,
+				       bool connected)
+{
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_connector *amdgpu_connector =
+	    to_amdgpu_connector(connector);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	uint32_t bios_0_scratch, bios_3_scratch, bios_6_scratch;
+
+	bios_0_scratch = RREG32(mmBIOS_SCRATCH_0);
+	bios_3_scratch = RREG32(mmBIOS_SCRATCH_3);
+	bios_6_scratch = RREG32(mmBIOS_SCRATCH_6);
+
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_LCD1_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_LCD1_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("LCD1 connected\n");
+			bios_0_scratch |= ATOM_S0_LCD1;
+			bios_3_scratch |= ATOM_S3_LCD1_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_LCD1;
+		} else {
+			DRM_DEBUG_KMS("LCD1 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_LCD1;
+			bios_3_scratch &= ~ATOM_S3_LCD1_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_LCD1;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_CRT1_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_CRT1_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("CRT1 connected\n");
+			bios_0_scratch |= ATOM_S0_CRT1_COLOR;
+			bios_3_scratch |= ATOM_S3_CRT1_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_CRT1;
+		} else {
+			DRM_DEBUG_KMS("CRT1 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_CRT1_MASK;
+			bios_3_scratch &= ~ATOM_S3_CRT1_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_CRT1;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_CRT2_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_CRT2_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("CRT2 connected\n");
+			bios_0_scratch |= ATOM_S0_CRT2_COLOR;
+			bios_3_scratch |= ATOM_S3_CRT2_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_CRT2;
+		} else {
+			DRM_DEBUG_KMS("CRT2 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_CRT2_MASK;
+			bios_3_scratch &= ~ATOM_S3_CRT2_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_CRT2;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP1_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP1_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP1 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP1;
+			bios_3_scratch |= ATOM_S3_DFP1_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP1;
+		} else {
+			DRM_DEBUG_KMS("DFP1 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP1;
+			bios_3_scratch &= ~ATOM_S3_DFP1_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP1;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP2_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP2_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP2 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP2;
+			bios_3_scratch |= ATOM_S3_DFP2_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP2;
+		} else {
+			DRM_DEBUG_KMS("DFP2 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP2;
+			bios_3_scratch &= ~ATOM_S3_DFP2_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP2;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP3_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP3_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP3 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP3;
+			bios_3_scratch |= ATOM_S3_DFP3_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP3;
+		} else {
+			DRM_DEBUG_KMS("DFP3 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP3;
+			bios_3_scratch &= ~ATOM_S3_DFP3_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP3;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP4_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP4_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP4 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP4;
+			bios_3_scratch |= ATOM_S3_DFP4_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP4;
+		} else {
+			DRM_DEBUG_KMS("DFP4 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP4;
+			bios_3_scratch &= ~ATOM_S3_DFP4_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP4;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP5_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP5_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP5 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP5;
+			bios_3_scratch |= ATOM_S3_DFP5_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP5;
+		} else {
+			DRM_DEBUG_KMS("DFP5 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP5;
+			bios_3_scratch &= ~ATOM_S3_DFP5_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP5;
+		}
+	}
+	if ((amdgpu_encoder->devices & ATOM_DEVICE_DFP6_SUPPORT) &&
+	    (amdgpu_connector->devices & ATOM_DEVICE_DFP6_SUPPORT)) {
+		if (connected) {
+			DRM_DEBUG_KMS("DFP6 connected\n");
+			bios_0_scratch |= ATOM_S0_DFP6;
+			bios_3_scratch |= ATOM_S3_DFP6_ACTIVE;
+			bios_6_scratch |= ATOM_S6_ACC_REQ_DFP6;
+		} else {
+			DRM_DEBUG_KMS("DFP6 disconnected\n");
+			bios_0_scratch &= ~ATOM_S0_DFP6;
+			bios_3_scratch &= ~ATOM_S3_DFP6_ACTIVE;
+			bios_6_scratch &= ~ATOM_S6_ACC_REQ_DFP6;
+		}
+	}
+
+	WREG32(mmBIOS_SCRATCH_0, bios_0_scratch);
+	WREG32(mmBIOS_SCRATCH_3, bios_3_scratch);
+	WREG32(mmBIOS_SCRATCH_6, bios_6_scratch);
+}
+
+union lvds_info {
+	struct _ATOM_LVDS_INFO info;
+	struct _ATOM_LVDS_INFO_V12 info_12;
+};
+
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder)
+{
+	struct drm_device *dev = encoder->base.dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, LVDS_Info);
+	uint16_t data_offset, misc;
+	union lvds_info *lvds_info;
+	uint8_t frev, crev;
+	struct amdgpu_encoder_atom_dig *lvds = NULL;
+	int encoder_enum = (encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		lvds_info =
+			(union lvds_info *)(mode_info->atom_context->bios + data_offset);
+		lvds =
+		    kzalloc(sizeof(struct amdgpu_encoder_atom_dig), GFP_KERNEL);
+
+		if (!lvds)
+			return NULL;
+
+		lvds->native_mode.clock =
+		    le16_to_cpu(lvds_info->info.sLCDTiming.usPixClk) * 10;
+		lvds->native_mode.hdisplay =
+		    le16_to_cpu(lvds_info->info.sLCDTiming.usHActive);
+		lvds->native_mode.vdisplay =
+		    le16_to_cpu(lvds_info->info.sLCDTiming.usVActive);
+		lvds->native_mode.htotal = lvds->native_mode.hdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usHBlanking_Time);
+		lvds->native_mode.hsync_start = lvds->native_mode.hdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncOffset);
+		lvds->native_mode.hsync_end = lvds->native_mode.hsync_start +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usHSyncWidth);
+		lvds->native_mode.vtotal = lvds->native_mode.vdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usVBlanking_Time);
+		lvds->native_mode.vsync_start = lvds->native_mode.vdisplay +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncOffset);
+		lvds->native_mode.vsync_end = lvds->native_mode.vsync_start +
+			le16_to_cpu(lvds_info->info.sLCDTiming.usVSyncWidth);
+		lvds->panel_pwr_delay =
+		    le16_to_cpu(lvds_info->info.usOffDelayInMs);
+		lvds->lcd_misc = lvds_info->info.ucLVDS_Misc;
+
+		misc = le16_to_cpu(lvds_info->info.sLCDTiming.susModeMiscInfo.usAccess);
+		if (misc & ATOM_VSYNC_POLARITY)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_NVSYNC;
+		if (misc & ATOM_HSYNC_POLARITY)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_NHSYNC;
+		if (misc & ATOM_COMPOSITESYNC)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_CSYNC;
+		if (misc & ATOM_INTERLACE)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_INTERLACE;
+		if (misc & ATOM_DOUBLE_CLOCK_MODE)
+			lvds->native_mode.flags |= DRM_MODE_FLAG_DBLSCAN;
+
+		lvds->native_mode.width_mm = le16_to_cpu(lvds_info->info.sLCDTiming.usImageHSize);
+		lvds->native_mode.height_mm = le16_to_cpu(lvds_info->info.sLCDTiming.usImageVSize);
+
+		/* set crtc values */
+		drm_mode_set_crtcinfo(&lvds->native_mode, CRTC_INTERLACE_HALVE_V);
+
+		lvds->lcd_ss_id = lvds_info->info.ucSS_Id;
+
+		encoder->native_mode = lvds->native_mode;
+
+		if (encoder_enum == 2)
+			lvds->linkb = true;
+		else
+			lvds->linkb = false;
+
+		/* parse the lcd record table */
+		if (le16_to_cpu(lvds_info->info.usModePatchTableOffset)) {
+			ATOM_FAKE_EDID_PATCH_RECORD *fake_edid_record;
+			ATOM_PANEL_RESOLUTION_PATCH_RECORD *panel_res_record;
+			bool bad_record = false;
+			u8 *record;
+
+			if ((frev == 1) && (crev < 2))
+				/* absolute */
+				record = (u8 *)(mode_info->atom_context->bios +
+						le16_to_cpu(lvds_info->info.usModePatchTableOffset));
+			else
+				/* relative */
+				record = (u8 *)(mode_info->atom_context->bios +
+						data_offset +
+						le16_to_cpu(lvds_info->info.usModePatchTableOffset));
+			while (*record != ATOM_RECORD_END_TYPE) {
+				switch (*record) {
+				case LCD_MODE_PATCH_RECORD_MODE_TYPE:
+					record += sizeof(ATOM_PATCH_RECORD_MODE);
+					break;
+				case LCD_RTS_RECORD_TYPE:
+					record += sizeof(ATOM_LCD_RTS_RECORD);
+					break;
+				case LCD_CAP_RECORD_TYPE:
+					record += sizeof(ATOM_LCD_MODE_CONTROL_CAP);
+					break;
+				case LCD_FAKE_EDID_PATCH_RECORD_TYPE:
+					fake_edid_record = (ATOM_FAKE_EDID_PATCH_RECORD *)record;
+					if (fake_edid_record->ucFakeEDIDLength) {
+						struct edid *edid;
+						int edid_size =
+							max((int)EDID_LENGTH, (int)fake_edid_record->ucFakeEDIDLength);
+						edid = kmalloc(edid_size, GFP_KERNEL);
+						if (edid) {
+							memcpy((u8 *)edid, (u8 *)&fake_edid_record->ucFakeEDIDString[0],
+							       fake_edid_record->ucFakeEDIDLength);
+
+							if (drm_edid_is_valid(edid)) {
+								adev->mode_info.bios_hardcoded_edid = edid;
+								adev->mode_info.bios_hardcoded_edid_size = edid_size;
+							} else
+								kfree(edid);
+						}
+					}
+					record += fake_edid_record->ucFakeEDIDLength ?
+						fake_edid_record->ucFakeEDIDLength + 2 :
+						sizeof(ATOM_FAKE_EDID_PATCH_RECORD);
+					break;
+				case LCD_PANEL_RESOLUTION_RECORD_TYPE:
+					panel_res_record = (ATOM_PANEL_RESOLUTION_PATCH_RECORD *)record;
+					lvds->native_mode.width_mm = panel_res_record->usHSize;
+					lvds->native_mode.height_mm = panel_res_record->usVSize;
+					record += sizeof(ATOM_PANEL_RESOLUTION_PATCH_RECORD);
+					break;
+				default:
+					DRM_ERROR("Bad LCD record %d\n", *record);
+					bad_record = true;
+					break;
+				}
+				if (bad_record)
+					break;
+			}
+		}
+	}
+	return lvds;
+}
+
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_dig_info(struct amdgpu_encoder *amdgpu_encoder)
+{
+	int encoder_enum = (amdgpu_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT;
+	struct amdgpu_encoder_atom_dig *dig = kzalloc(sizeof(struct amdgpu_encoder_atom_dig), GFP_KERNEL);
+
+	if (!dig)
+		return NULL;
+
+	/* coherent mode by default */
+	dig->coherent_mode = true;
+	dig->dig_encoder = -1;
+
+	if (encoder_enum == 2)
+		dig->linkb = true;
+	else
+		dig->linkb = false;
+
+	return dig;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h
new file mode 100644
index 000000000..2bdec4051
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_encoders.h
@@ -0,0 +1,73 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_ENCODER_H__
+#define __ATOMBIOS_ENCODER_H__
+
+u8
+amdgpu_atombios_encoder_get_backlight_level(struct amdgpu_encoder *amdgpu_encoder);
+void
+amdgpu_atombios_encoder_set_backlight_level(struct amdgpu_encoder *amdgpu_encoder,
+				     u8 level);
+void amdgpu_atombios_encoder_init_backlight(struct amdgpu_encoder *amdgpu_encoder,
+				     struct drm_connector *drm_connector);
+void
+amdgpu_atombios_encoder_fini_backlight(struct amdgpu_encoder *amdgpu_encoder);
+bool amdgpu_atombios_encoder_is_digital(struct drm_encoder *encoder);
+bool amdgpu_atombios_encoder_mode_fixup(struct drm_encoder *encoder,
+				 const struct drm_display_mode *mode,
+				 struct drm_display_mode *adjusted_mode);
+int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder);
+void
+amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
+				   int action, int panel_mode);
+void
+amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int action,
+				       uint8_t lane_num, uint8_t lane_set);
+bool
+amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
+				     int action);
+void
+amdgpu_atombios_encoder_dpms(struct drm_encoder *encoder, int mode);
+void
+amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder);
+void
+amdgpu_atombios_encoder_init_dig(struct amdgpu_device *adev);
+enum drm_connector_status
+amdgpu_atombios_encoder_dac_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector);
+enum drm_connector_status
+amdgpu_atombios_encoder_dig_detect(struct drm_encoder *encoder,
+			    struct drm_connector *connector);
+void
+amdgpu_atombios_encoder_setup_ext_encoder_ddc(struct drm_encoder *encoder);
+void
+amdgpu_atombios_encoder_set_bios_scratch_regs(struct drm_connector *connector,
+				       struct drm_encoder *encoder,
+				       bool connected);
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder);
+struct amdgpu_encoder_atom_dig *
+amdgpu_atombios_encoder_get_dig_info(struct amdgpu_encoder *amdgpu_encoder);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
new file mode 100644
index 000000000..13cdb01e9
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
@@ -0,0 +1,158 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ *
+ */
+#include <drm/drmP.h>
+#include <drm/amdgpu_drm.h>
+#include "amdgpu.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+
+#define TARGET_HW_I2C_CLOCK 50
+
+/* these are a limitation of ProcessI2cChannelTransaction not the hw */
+#define ATOM_MAX_HW_I2C_WRITE 3
+#define ATOM_MAX_HW_I2C_READ  255
+
+static int amdgpu_atombios_i2c_process_i2c_ch(struct amdgpu_i2c_chan *chan,
+				       u8 slave_addr, u8 flags,
+				       u8 *buf, u8 num)
+{
+	struct drm_device *dev = chan->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	PROCESS_I2C_CHANNEL_TRANSACTION_PS_ALLOCATION args;
+	int index = GetIndexIntoMasterTable(COMMAND, ProcessI2cChannelTransaction);
+	unsigned char *base;
+	u16 out = cpu_to_le16(0);
+	int r = 0;
+
+	memset(&args, 0, sizeof(args));
+
+	mutex_lock(&chan->mutex);
+
+	base = (unsigned char *)adev->mode_info.atom_context->scratch;
+
+	if (flags & HW_I2C_WRITE) {
+		if (num > ATOM_MAX_HW_I2C_WRITE) {
+			DRM_ERROR("hw i2c: tried to write too many bytes (%d vs 3)\n", num);
+			r = -EINVAL;
+			goto done;
+		}
+		if (buf == NULL)
+			args.ucRegIndex = 0;
+		else
+			args.ucRegIndex = buf[0];
+		if (num)
+			num--;
+		if (num)
+			memcpy(&out, &buf[1], num);
+		args.lpI2CDataOut = cpu_to_le16(out);
+	} else {
+		if (num > ATOM_MAX_HW_I2C_READ) {
+			DRM_ERROR("hw i2c: tried to read too many bytes (%d vs 255)\n", num);
+			r = -EINVAL;
+			goto done;
+		}
+		args.ucRegIndex = 0;
+		args.lpI2CDataOut = 0;
+	}
+
+	args.ucFlag = flags;
+	args.ucI2CSpeed = TARGET_HW_I2C_CLOCK;
+	args.ucTransBytes = num;
+	args.ucSlaveAddr = slave_addr << 1;
+	args.ucLineNumber = chan->rec.i2c_id;
+
+	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+	/* error */
+	if (args.ucStatus != HW_ASSISTED_I2C_STATUS_SUCCESS) {
+		DRM_DEBUG_KMS("hw_i2c error\n");
+		r = -EIO;
+		goto done;
+	}
+
+	if (!(flags & HW_I2C_WRITE))
+		amdgpu_atombios_copy_swap(buf, base, num, false);
+
+done:
+	mutex_unlock(&chan->mutex);
+
+	return r;
+}
+
+int amdgpu_atombios_i2c_xfer(struct i2c_adapter *i2c_adap,
+		      struct i2c_msg *msgs, int num)
+{
+	struct amdgpu_i2c_chan *i2c = i2c_get_adapdata(i2c_adap);
+	struct i2c_msg *p;
+	int i, remaining, current_count, buffer_offset, max_bytes, ret;
+	u8 flags;
+
+	/* check for bus probe */
+	p = &msgs[0];
+	if ((num == 1) && (p->len == 0)) {
+		ret = amdgpu_atombios_i2c_process_i2c_ch(i2c,
+						  p->addr, HW_I2C_WRITE,
+						  NULL, 0);
+		if (ret)
+			return ret;
+		else
+			return num;
+	}
+
+	for (i = 0; i < num; i++) {
+		p = &msgs[i];
+		remaining = p->len;
+		buffer_offset = 0;
+		/* max_bytes are a limitation of ProcessI2cChannelTransaction not the hw */
+		if (p->flags & I2C_M_RD) {
+			max_bytes = ATOM_MAX_HW_I2C_READ;
+			flags = HW_I2C_READ;
+		} else {
+			max_bytes = ATOM_MAX_HW_I2C_WRITE;
+			flags = HW_I2C_WRITE;
+		}
+		while (remaining) {
+			if (remaining > max_bytes)
+				current_count = max_bytes;
+			else
+				current_count = remaining;
+			ret = amdgpu_atombios_i2c_process_i2c_ch(i2c,
+							  p->addr, flags,
+							  &p->buf[buffer_offset], current_count);
+			if (ret)
+				return ret;
+			remaining -= current_count;
+			buffer_offset += current_count;
+		}
+	}
+
+	return num;
+}
+
+u32 amdgpu_atombios_i2c_func(struct i2c_adapter *adap)
+{
+	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.h b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.h
new file mode 100644
index 000000000..d6128d9de
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/atombios_i2c.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ATOMBIOS_I2C_H__
+#define __ATOMBIOS_I2C_H__
+
+int amdgpu_atombios_i2c_xfer(struct i2c_adapter *i2c_adap,
+		      struct i2c_msg *msgs, int num);
+u32 amdgpu_atombios_i2c_func(struct i2c_adapter *adap);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
new file mode 100644
index 000000000..57a2e347f
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
@@ -0,0 +1,6701 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_ucode.h"
+#include "cikd.h"
+#include "amdgpu_dpm.h"
+#include "ci_dpm.h"
+#include "gfx_v7_0.h"
+#include "atom.h"
+#include <linux/seq_file.h>
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+MODULE_FIRMWARE("radeon/bonaire_smc.bin");
+MODULE_FIRMWARE("radeon/hawaii_smc.bin");
+
+#define MC_CG_ARB_FREQ_F0           0x0a
+#define MC_CG_ARB_FREQ_F1           0x0b
+#define MC_CG_ARB_FREQ_F2           0x0c
+#define MC_CG_ARB_FREQ_F3           0x0d
+
+#define SMC_RAM_END 0x40000
+
+#define VOLTAGE_SCALE               4
+#define VOLTAGE_VID_OFFSET_SCALE1    625
+#define VOLTAGE_VID_OFFSET_SCALE2    100
+
+static const struct ci_pt_defaults defaults_hawaii_xt =
+{
+	1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0xB0000,
+	{ 0x2E,  0x00,  0x00,  0x88,  0x00,  0x00,  0x72,  0x60,  0x51,  0xA7,  0x79,  0x6B,  0x90,  0xBD,  0x79  },
+	{ 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 }
+};
+
+static const struct ci_pt_defaults defaults_hawaii_pro =
+{
+	1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0x65062,
+	{ 0x2E,  0x00,  0x00,  0x88,  0x00,  0x00,  0x72,  0x60,  0x51,  0xA7,  0x79,  0x6B,  0x90,  0xBD,  0x79  },
+	{ 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 }
+};
+
+static const struct ci_pt_defaults defaults_bonaire_xt =
+{
+	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0xB0000,
+	{ 0x79,  0x253, 0x25D, 0xAE,  0x72,  0x80,  0x83,  0x86,  0x6F,  0xC8,  0xC9,  0xC9,  0x2F,  0x4D,  0x61  },
+	{ 0x17C, 0x172, 0x180, 0x1BC, 0x1B3, 0x1BD, 0x206, 0x200, 0x203, 0x25D, 0x25A, 0x255, 0x2C3, 0x2C5, 0x2B4 }
+};
+
+static const struct ci_pt_defaults defaults_bonaire_pro =
+{
+	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0x65062,
+	{ 0x8C,  0x23F, 0x244, 0xA6,  0x83,  0x85,  0x86,  0x86,  0x83,  0xDB,  0xDB,  0xDA,  0x67,  0x60,  0x5F  },
+	{ 0x187, 0x193, 0x193, 0x1C7, 0x1D1, 0x1D1, 0x210, 0x219, 0x219, 0x266, 0x26C, 0x26C, 0x2C9, 0x2CB, 0x2CB }
+};
+
+static const struct ci_pt_defaults defaults_saturn_xt =
+{
+	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x70000,
+	{ 0x8C,  0x247, 0x249, 0xA6,  0x80,  0x81,  0x8B,  0x89,  0x86,  0xC9,  0xCA,  0xC9,  0x4D,  0x4D,  0x4D  },
+	{ 0x187, 0x187, 0x187, 0x1C7, 0x1C7, 0x1C7, 0x210, 0x210, 0x210, 0x266, 0x266, 0x266, 0x2C9, 0x2C9, 0x2C9 }
+};
+
+static const struct ci_pt_defaults defaults_saturn_pro =
+{
+	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x30000,
+	{ 0x96,  0x21D, 0x23B, 0xA1,  0x85,  0x87,  0x83,  0x84,  0x81,  0xE6,  0xE6,  0xE6,  0x71,  0x6A,  0x6A  },
+	{ 0x193, 0x19E, 0x19E, 0x1D2, 0x1DC, 0x1DC, 0x21A, 0x223, 0x223, 0x26E, 0x27E, 0x274, 0x2CF, 0x2D2, 0x2D2 }
+};
+
+static const struct ci_pt_config_reg didt_config_ci[] =
+{
+	{ 0x10, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x10, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x11, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x12, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x1, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x1, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x0, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x30, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x31, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x32, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x21, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x21, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x20, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x50, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x51, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x52, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x41, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x41, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x40, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x70, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x71, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x000000ff, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x0000ff00, 8, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x00ff0000, 16, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x72, 0xff000000, 24, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x00003fff, 0, 0x4, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x03ff0000, 16, 0x80, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x78000000, 27, 0x3, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x61, 0x0000ffff, 0, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x61, 0xffff0000, 16, 0x3FFF, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0x60, 0x00000001, 0, 0x0, CISLANDS_CONFIGREG_DIDT_IND },
+	{ 0xFFFFFFFF }
+};
+
+static u8 ci_get_memory_module_index(struct amdgpu_device *adev)
+{
+	return (u8) ((RREG32(mmBIOS_SCRATCH_4) >> 16) & 0xff);
+}
+
+#define MC_CG_ARB_FREQ_F0           0x0a
+#define MC_CG_ARB_FREQ_F1           0x0b
+#define MC_CG_ARB_FREQ_F2           0x0c
+#define MC_CG_ARB_FREQ_F3           0x0d
+
+static int ci_copy_and_switch_arb_sets(struct amdgpu_device *adev,
+				       u32 arb_freq_src, u32 arb_freq_dest)
+{
+	u32 mc_arb_dram_timing;
+	u32 mc_arb_dram_timing2;
+	u32 burst_time;
+	u32 mc_cg_config;
+
+	switch (arb_freq_src) {
+	case MC_CG_ARB_FREQ_F0:
+		mc_arb_dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING);
+		mc_arb_dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2);
+		burst_time = (RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE0_MASK) >>
+			 MC_ARB_BURST_TIME__STATE0__SHIFT;
+		break;
+	case MC_CG_ARB_FREQ_F1:
+		mc_arb_dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING_1);
+		mc_arb_dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2_1);
+		burst_time = (RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE1_MASK) >>
+			 MC_ARB_BURST_TIME__STATE1__SHIFT;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	switch (arb_freq_dest) {
+	case MC_CG_ARB_FREQ_F0:
+		WREG32(mmMC_ARB_DRAM_TIMING, mc_arb_dram_timing);
+		WREG32(mmMC_ARB_DRAM_TIMING2, mc_arb_dram_timing2);
+		WREG32_P(mmMC_ARB_BURST_TIME, (burst_time << MC_ARB_BURST_TIME__STATE0__SHIFT),
+			~MC_ARB_BURST_TIME__STATE0_MASK);
+		break;
+	case MC_CG_ARB_FREQ_F1:
+		WREG32(mmMC_ARB_DRAM_TIMING_1, mc_arb_dram_timing);
+		WREG32(mmMC_ARB_DRAM_TIMING2_1, mc_arb_dram_timing2);
+		WREG32_P(mmMC_ARB_BURST_TIME, (burst_time << MC_ARB_BURST_TIME__STATE1__SHIFT),
+			~MC_ARB_BURST_TIME__STATE1_MASK);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	mc_cg_config = RREG32(mmMC_CG_CONFIG) | 0x0000000F;
+	WREG32(mmMC_CG_CONFIG, mc_cg_config);
+	WREG32_P(mmMC_ARB_CG, (arb_freq_dest) << MC_ARB_CG__CG_ARB_REQ__SHIFT,
+		~MC_ARB_CG__CG_ARB_REQ_MASK);
+
+	return 0;
+}
+
+static u8 ci_get_ddr3_mclk_frequency_ratio(u32 memory_clock)
+{
+	u8 mc_para_index;
+
+	if (memory_clock < 10000)
+		mc_para_index = 0;
+	else if (memory_clock >= 80000)
+		mc_para_index = 0x0f;
+	else
+		mc_para_index = (u8)((memory_clock - 10000) / 5000 + 1);
+	return mc_para_index;
+}
+
+static u8 ci_get_mclk_frequency_ratio(u32 memory_clock, bool strobe_mode)
+{
+	u8 mc_para_index;
+
+	if (strobe_mode) {
+		if (memory_clock < 12500)
+			mc_para_index = 0x00;
+		else if (memory_clock > 47500)
+			mc_para_index = 0x0f;
+		else
+			mc_para_index = (u8)((memory_clock - 10000) / 2500);
+	} else {
+		if (memory_clock < 65000)
+			mc_para_index = 0x00;
+		else if (memory_clock > 135000)
+			mc_para_index = 0x0f;
+		else
+			mc_para_index = (u8)((memory_clock - 60000) / 5000);
+	}
+	return mc_para_index;
+}
+
+static void ci_trim_voltage_table_to_fit_state_table(struct amdgpu_device *adev,
+						     u32 max_voltage_steps,
+						     struct atom_voltage_table *voltage_table)
+{
+	unsigned int i, diff;
+
+	if (voltage_table->count <= max_voltage_steps)
+		return;
+
+	diff = voltage_table->count - max_voltage_steps;
+
+	for (i = 0; i < max_voltage_steps; i++)
+		voltage_table->entries[i] = voltage_table->entries[i + diff];
+
+	voltage_table->count = max_voltage_steps;
+}
+
+static int ci_get_std_voltage_value_sidd(struct amdgpu_device *adev,
+					 struct atom_voltage_table_entry *voltage_table,
+					 u16 *std_voltage_hi_sidd, u16 *std_voltage_lo_sidd);
+static int ci_set_power_limit(struct amdgpu_device *adev, u32 n);
+static int ci_set_overdrive_target_tdp(struct amdgpu_device *adev,
+				       u32 target_tdp);
+static int ci_update_uvd_dpm(struct amdgpu_device *adev, bool gate);
+static void ci_dpm_set_dpm_funcs(struct amdgpu_device *adev);
+static void ci_dpm_set_irq_funcs(struct amdgpu_device *adev);
+
+static PPSMC_Result amdgpu_ci_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+							     PPSMC_Msg msg, u32 parameter);
+static void ci_thermal_start_smc_fan_control(struct amdgpu_device *adev);
+static void ci_fan_ctrl_set_default_mode(struct amdgpu_device *adev);
+
+static struct ci_power_info *ci_get_pi(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+static struct ci_ps *ci_get_ps(struct amdgpu_ps *rps)
+{
+	struct ci_ps *ps = rps->ps_priv;
+
+	return ps;
+}
+
+static void ci_initialize_powertune_defaults(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	switch (adev->pdev->device) {
+	case 0x6649:
+	case 0x6650:
+	case 0x6651:
+	case 0x6658:
+	case 0x665C:
+	case 0x665D:
+	default:
+		pi->powertune_defaults = &defaults_bonaire_xt;
+		break;
+	case 0x6640:
+	case 0x6641:
+	case 0x6646:
+	case 0x6647:
+		pi->powertune_defaults = &defaults_saturn_xt;
+		break;
+	case 0x67B8:
+	case 0x67B0:
+		pi->powertune_defaults = &defaults_hawaii_xt;
+		break;
+	case 0x67BA:
+	case 0x67B1:
+		pi->powertune_defaults = &defaults_hawaii_pro;
+		break;
+	case 0x67A0:
+	case 0x67A1:
+	case 0x67A2:
+	case 0x67A8:
+	case 0x67A9:
+	case 0x67AA:
+	case 0x67B9:
+	case 0x67BE:
+		pi->powertune_defaults = &defaults_bonaire_xt;
+		break;
+	}
+
+	pi->dte_tj_offset = 0;
+
+	pi->caps_power_containment = true;
+	pi->caps_cac = false;
+	pi->caps_sq_ramping = false;
+	pi->caps_db_ramping = false;
+	pi->caps_td_ramping = false;
+	pi->caps_tcp_ramping = false;
+
+	if (pi->caps_power_containment) {
+		pi->caps_cac = true;
+		if (adev->asic_type == CHIP_HAWAII)
+			pi->enable_bapm_feature = false;
+		else
+			pi->enable_bapm_feature = true;
+		pi->enable_tdc_limit_feature = true;
+		pi->enable_pkg_pwr_tracking_feature = true;
+	}
+}
+
+static u8 ci_convert_to_vid(u16 vddc)
+{
+	return (6200 - (vddc * VOLTAGE_SCALE)) / 25;
+}
+
+static int ci_populate_bapm_vddc_vid_sidd(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u8 *hi_vid = pi->smc_powertune_table.BapmVddCVidHiSidd;
+	u8 *lo_vid = pi->smc_powertune_table.BapmVddCVidLoSidd;
+	u8 *hi2_vid = pi->smc_powertune_table.BapmVddCVidHiSidd2;
+	u32 i;
+
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries == NULL)
+		return -EINVAL;
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.count > 8)
+		return -EINVAL;
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.count !=
+	    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count)
+		return -EINVAL;
+
+	for (i = 0; i < adev->pm.dpm.dyn_state.cac_leakage_table.count; i++) {
+		if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
+			lo_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc1);
+			hi_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc2);
+			hi2_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc3);
+		} else {
+			lo_vid[i] = ci_convert_to_vid(adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].vddc);
+			hi_vid[i] = ci_convert_to_vid((u16)adev->pm.dpm.dyn_state.cac_leakage_table.entries[i].leakage);
+		}
+	}
+	return 0;
+}
+
+static int ci_populate_vddc_vid(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u8 *vid = pi->smc_powertune_table.VddCVid;
+	u32 i;
+
+	if (pi->vddc_voltage_table.count > 8)
+		return -EINVAL;
+
+	for (i = 0; i < pi->vddc_voltage_table.count; i++)
+		vid[i] = ci_convert_to_vid(pi->vddc_voltage_table.entries[i].value);
+
+	return 0;
+}
+
+static int ci_populate_svi_load_line(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+
+	pi->smc_powertune_table.SviLoadLineEn = pt_defaults->svi_load_line_en;
+	pi->smc_powertune_table.SviLoadLineVddC = pt_defaults->svi_load_line_vddc;
+	pi->smc_powertune_table.SviLoadLineTrimVddC = 3;
+	pi->smc_powertune_table.SviLoadLineOffsetVddC = 0;
+
+	return 0;
+}
+
+static int ci_populate_tdc_limit(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+	u16 tdc_limit;
+
+	tdc_limit = adev->pm.dpm.dyn_state.cac_tdp_table->tdc * 256;
+	pi->smc_powertune_table.TDC_VDDC_PkgLimit = cpu_to_be16(tdc_limit);
+	pi->smc_powertune_table.TDC_VDDC_ThrottleReleaseLimitPerc =
+		pt_defaults->tdc_vddc_throttle_release_limit_perc;
+	pi->smc_powertune_table.TDC_MAWt = pt_defaults->tdc_mawt;
+
+	return 0;
+}
+
+static int ci_populate_dw8(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+	int ret;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, PmFuseTable) +
+				     offsetof(SMU7_Discrete_PmFuses, TdcWaterfallCtl),
+				     (u32 *)&pi->smc_powertune_table.TdcWaterfallCtl,
+				     pi->sram_end);
+	if (ret)
+		return -EINVAL;
+	else
+		pi->smc_powertune_table.TdcWaterfallCtl = pt_defaults->tdc_waterfall_ctl;
+
+	return 0;
+}
+
+static int ci_populate_fuzzy_fan(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if ((adev->pm.dpm.fan.fan_output_sensitivity & (1 << 15)) ||
+	    (adev->pm.dpm.fan.fan_output_sensitivity == 0))
+		adev->pm.dpm.fan.fan_output_sensitivity =
+			adev->pm.dpm.fan.default_fan_output_sensitivity;
+
+	pi->smc_powertune_table.FuzzyFan_PwmSetDelta =
+		cpu_to_be16(adev->pm.dpm.fan.fan_output_sensitivity);
+
+	return 0;
+}
+
+static int ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u8 *hi_vid = pi->smc_powertune_table.BapmVddCVidHiSidd;
+	u8 *lo_vid = pi->smc_powertune_table.BapmVddCVidLoSidd;
+	int i, min, max;
+
+	min = max = hi_vid[0];
+	for (i = 0; i < 8; i++) {
+		if (0 != hi_vid[i]) {
+			if (min > hi_vid[i])
+				min = hi_vid[i];
+			if (max < hi_vid[i])
+				max = hi_vid[i];
+		}
+
+		if (0 != lo_vid[i]) {
+			if (min > lo_vid[i])
+				min = lo_vid[i];
+			if (max < lo_vid[i])
+				max = lo_vid[i];
+		}
+	}
+
+	if ((min == 0) || (max == 0))
+		return -EINVAL;
+	pi->smc_powertune_table.GnbLPMLMaxVid = (u8)max;
+	pi->smc_powertune_table.GnbLPMLMinVid = (u8)min;
+
+	return 0;
+}
+
+static int ci_populate_bapm_vddc_base_leakage_sidd(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u16 hi_sidd = pi->smc_powertune_table.BapmVddCBaseLeakageHiSidd;
+	u16 lo_sidd = pi->smc_powertune_table.BapmVddCBaseLeakageLoSidd;
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+
+	hi_sidd = cac_tdp_table->high_cac_leakage / 100 * 256;
+	lo_sidd = cac_tdp_table->low_cac_leakage / 100 * 256;
+
+	pi->smc_powertune_table.BapmVddCBaseLeakageHiSidd = cpu_to_be16(hi_sidd);
+	pi->smc_powertune_table.BapmVddCBaseLeakageLoSidd = cpu_to_be16(lo_sidd);
+
+	return 0;
+}
+
+static int ci_populate_bapm_parameters_in_dpm_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct ci_pt_defaults *pt_defaults = pi->powertune_defaults;
+	SMU7_Discrete_DpmTable  *dpm_table = &pi->smc_state_table;
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+	struct amdgpu_ppm_table *ppm = adev->pm.dpm.dyn_state.ppm_table;
+	int i, j, k;
+	const u16 *def1;
+	const u16 *def2;
+
+	dpm_table->DefaultTdp = cac_tdp_table->tdp * 256;
+	dpm_table->TargetTdp = cac_tdp_table->configurable_tdp * 256;
+
+	dpm_table->DTETjOffset = (u8)pi->dte_tj_offset;
+	dpm_table->GpuTjMax =
+		(u8)(pi->thermal_temp_setting.temperature_high / 1000);
+	dpm_table->GpuTjHyst = 8;
+
+	dpm_table->DTEAmbientTempBase = pt_defaults->dte_ambient_temp_base;
+
+	if (ppm) {
+		dpm_table->PPM_PkgPwrLimit = cpu_to_be16((u16)ppm->dgpu_tdp * 256 / 1000);
+		dpm_table->PPM_TemperatureLimit = cpu_to_be16((u16)ppm->tj_max * 256);
+	} else {
+		dpm_table->PPM_PkgPwrLimit = cpu_to_be16(0);
+		dpm_table->PPM_TemperatureLimit = cpu_to_be16(0);
+	}
+
+	dpm_table->BAPM_TEMP_GRADIENT = cpu_to_be32(pt_defaults->bapm_temp_gradient);
+	def1 = pt_defaults->bapmti_r;
+	def2 = pt_defaults->bapmti_rc;
+
+	for (i = 0; i < SMU7_DTE_ITERATIONS; i++) {
+		for (j = 0; j < SMU7_DTE_SOURCES; j++) {
+			for (k = 0; k < SMU7_DTE_SINKS; k++) {
+				dpm_table->BAPMTI_R[i][j][k] = cpu_to_be16(*def1);
+				dpm_table->BAPMTI_RC[i][j][k] = cpu_to_be16(*def2);
+				def1++;
+				def2++;
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int ci_populate_pm_base(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 pm_fuse_table_offset;
+	int ret;
+
+	if (pi->caps_power_containment) {
+		ret = amdgpu_ci_read_smc_sram_dword(adev,
+					     SMU7_FIRMWARE_HEADER_LOCATION +
+					     offsetof(SMU7_Firmware_Header, PmFuseTable),
+					     &pm_fuse_table_offset, pi->sram_end);
+		if (ret)
+			return ret;
+		ret = ci_populate_bapm_vddc_vid_sidd(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_vddc_vid(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_svi_load_line(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_tdc_limit(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_dw8(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_fuzzy_fan(adev);
+		if (ret)
+			return ret;
+		ret = ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(adev);
+		if (ret)
+			return ret;
+		ret = ci_populate_bapm_vddc_base_leakage_sidd(adev);
+		if (ret)
+			return ret;
+		ret = amdgpu_ci_copy_bytes_to_smc(adev, pm_fuse_table_offset,
+					   (u8 *)&pi->smc_powertune_table,
+					   sizeof(SMU7_Discrete_PmFuses), pi->sram_end);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static void ci_do_enable_didt(struct amdgpu_device *adev, const bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 data;
+
+	if (pi->caps_sq_ramping) {
+		data = RREG32_DIDT(ixDIDT_SQ_CTRL0);
+		if (enable)
+			data |= DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_SQ_CTRL0, data);
+	}
+
+	if (pi->caps_db_ramping) {
+		data = RREG32_DIDT(ixDIDT_DB_CTRL0);
+		if (enable)
+			data |= DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_DB_CTRL0, data);
+	}
+
+	if (pi->caps_td_ramping) {
+		data = RREG32_DIDT(ixDIDT_TD_CTRL0);
+		if (enable)
+			data |= DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TD_CTRL0, data);
+	}
+
+	if (pi->caps_tcp_ramping) {
+		data = RREG32_DIDT(ixDIDT_TCP_CTRL0);
+		if (enable)
+			data |= DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TCP_CTRL0, data);
+	}
+}
+
+static int ci_program_pt_config_registers(struct amdgpu_device *adev,
+					  const struct ci_pt_config_reg *cac_config_regs)
+{
+	const struct ci_pt_config_reg *config_regs = cac_config_regs;
+	u32 data;
+	u32 cache = 0;
+
+	if (config_regs == NULL)
+		return -EINVAL;
+
+	while (config_regs->offset != 0xFFFFFFFF) {
+		if (config_regs->type == CISLANDS_CONFIGREG_CACHE) {
+			cache |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+		} else {
+			switch (config_regs->type) {
+			case CISLANDS_CONFIGREG_SMC_IND:
+				data = RREG32_SMC(config_regs->offset);
+				break;
+			case CISLANDS_CONFIGREG_DIDT_IND:
+				data = RREG32_DIDT(config_regs->offset);
+				break;
+			default:
+				data = RREG32(config_regs->offset);
+				break;
+			}
+
+			data &= ~config_regs->mask;
+			data |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+			data |= cache;
+
+			switch (config_regs->type) {
+			case CISLANDS_CONFIGREG_SMC_IND:
+				WREG32_SMC(config_regs->offset, data);
+				break;
+			case CISLANDS_CONFIGREG_DIDT_IND:
+				WREG32_DIDT(config_regs->offset, data);
+				break;
+			default:
+				WREG32(config_regs->offset, data);
+				break;
+			}
+			cache = 0;
+		}
+		config_regs++;
+	}
+	return 0;
+}
+
+static int ci_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	if (pi->caps_sq_ramping || pi->caps_db_ramping ||
+	    pi->caps_td_ramping || pi->caps_tcp_ramping) {
+		gfx_v7_0_enter_rlc_safe_mode(adev);
+
+		if (enable) {
+			ret = ci_program_pt_config_registers(adev, didt_config_ci);
+			if (ret) {
+				gfx_v7_0_exit_rlc_safe_mode(adev);
+				return ret;
+			}
+		}
+
+		ci_do_enable_didt(adev, enable);
+
+		gfx_v7_0_exit_rlc_safe_mode(adev);
+	}
+
+	return 0;
+}
+
+static int ci_enable_power_containment(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret = 0;
+
+	if (enable) {
+		pi->power_containment_features = 0;
+		if (pi->caps_power_containment) {
+			if (pi->enable_bapm_feature) {
+				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableDTE);
+				if (smc_result != PPSMC_Result_OK)
+					ret = -EINVAL;
+				else
+					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_BAPM;
+			}
+
+			if (pi->enable_tdc_limit_feature) {
+				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_TDCLimitEnable);
+				if (smc_result != PPSMC_Result_OK)
+					ret = -EINVAL;
+				else
+					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_TDCLimit;
+			}
+
+			if (pi->enable_pkg_pwr_tracking_feature) {
+				smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PkgPwrLimitEnable);
+				if (smc_result != PPSMC_Result_OK) {
+					ret = -EINVAL;
+				} else {
+					struct amdgpu_cac_tdp_table *cac_tdp_table =
+						adev->pm.dpm.dyn_state.cac_tdp_table;
+					u32 default_pwr_limit =
+						(u32)(cac_tdp_table->maximum_power_delivery_limit * 256);
+
+					pi->power_containment_features |= POWERCONTAINMENT_FEATURE_PkgPwrLimit;
+
+					ci_set_power_limit(adev, default_pwr_limit);
+				}
+			}
+		}
+	} else {
+		if (pi->caps_power_containment && pi->power_containment_features) {
+			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_TDCLimit)
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_TDCLimitDisable);
+
+			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_BAPM)
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableDTE);
+
+			if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_PkgPwrLimit)
+				amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PkgPwrLimitDisable);
+			pi->power_containment_features = 0;
+		}
+	}
+
+	return ret;
+}
+
+static int ci_enable_smc_cac(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret = 0;
+
+	if (pi->caps_cac) {
+		if (enable) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableCac);
+			if (smc_result != PPSMC_Result_OK) {
+				ret = -EINVAL;
+				pi->cac_enabled = false;
+			} else {
+				pi->cac_enabled = true;
+			}
+		} else if (pi->cac_enabled) {
+			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableCac);
+			pi->cac_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int ci_enable_thermal_based_sclk_dpm(struct amdgpu_device *adev,
+					    bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result = PPSMC_Result_OK;
+
+	if (pi->thermal_sclk_dpm_enabled) {
+		if (enable)
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_ENABLE_THERMAL_DPM);
+		else
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DISABLE_THERMAL_DPM);
+	}
+
+	if (smc_result == PPSMC_Result_OK)
+		return 0;
+	else
+		return -EINVAL;
+}
+
+static int ci_power_control_set_level(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+	s32 adjust_percent;
+	s32 target_tdp;
+	int ret = 0;
+	bool adjust_polarity = false; /* ??? */
+
+	if (pi->caps_power_containment) {
+		adjust_percent = adjust_polarity ?
+			adev->pm.dpm.tdp_adjustment : (-1 * adev->pm.dpm.tdp_adjustment);
+		target_tdp = ((100 + adjust_percent) *
+			      (s32)cac_tdp_table->configurable_tdp) / 100;
+
+		ret = ci_set_overdrive_target_tdp(adev, (u32)target_tdp);
+	}
+
+	return ret;
+}
+
+static void ci_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (pi->uvd_power_gated == gate)
+		return;
+
+	pi->uvd_power_gated = gate;
+
+	ci_update_uvd_dpm(adev, gate);
+}
+
+static bool ci_dpm_vblank_too_short(struct amdgpu_device *adev)
+{
+	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
+	u32 switch_limit = adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5 ? 450 : 300;
+
+	if (vblank_time < switch_limit)
+		return true;
+	else
+		return false;
+
+}
+
+static void ci_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps)
+{
+	struct ci_ps *ps = ci_get_ps(rps);
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_clock_and_voltage_limits *max_limits;
+	bool disable_mclk_switching;
+	u32 sclk, mclk;
+	int i;
+
+	if (rps->vce_active) {
+		rps->evclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].evclk;
+		rps->ecclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].ecclk;
+	} else {
+		rps->evclk = 0;
+		rps->ecclk = 0;
+	}
+
+	if ((adev->pm.dpm.new_active_crtc_count > 1) ||
+	    ci_dpm_vblank_too_short(adev))
+		disable_mclk_switching = true;
+	else
+		disable_mclk_switching = false;
+
+	if ((rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
+		pi->battery_state = true;
+	else
+		pi->battery_state = false;
+
+	if (adev->pm.dpm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (adev->pm.dpm.ac_power == false) {
+		for (i = 0; i < ps->performance_level_count; i++) {
+			if (ps->performance_levels[i].mclk > max_limits->mclk)
+				ps->performance_levels[i].mclk = max_limits->mclk;
+			if (ps->performance_levels[i].sclk > max_limits->sclk)
+				ps->performance_levels[i].sclk = max_limits->sclk;
+		}
+	}
+
+	/* XXX validate the min clocks required for display */
+
+	if (disable_mclk_switching) {
+		mclk  = ps->performance_levels[ps->performance_level_count - 1].mclk;
+		sclk = ps->performance_levels[0].sclk;
+	} else {
+		mclk = ps->performance_levels[0].mclk;
+		sclk = ps->performance_levels[0].sclk;
+	}
+
+	if (rps->vce_active) {
+		if (sclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk)
+			sclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk;
+		if (mclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk)
+			mclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].mclk;
+	}
+
+	ps->performance_levels[0].sclk = sclk;
+	ps->performance_levels[0].mclk = mclk;
+
+	if (ps->performance_levels[1].sclk < ps->performance_levels[0].sclk)
+		ps->performance_levels[1].sclk = ps->performance_levels[0].sclk;
+
+	if (disable_mclk_switching) {
+		if (ps->performance_levels[0].mclk < ps->performance_levels[1].mclk)
+			ps->performance_levels[0].mclk = ps->performance_levels[1].mclk;
+	} else {
+		if (ps->performance_levels[1].mclk < ps->performance_levels[0].mclk)
+			ps->performance_levels[1].mclk = ps->performance_levels[0].mclk;
+	}
+}
+
+static int ci_thermal_set_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp)
+{
+	int low_temp = 0 * 1000;
+	int high_temp = 255 * 1000;
+	u32 tmp;
+
+	if (low_temp < min_temp)
+		low_temp = min_temp;
+	if (high_temp > max_temp)
+		high_temp = max_temp;
+	if (high_temp < low_temp) {
+		DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp);
+		return -EINVAL;
+	}
+
+	tmp = RREG32_SMC(ixCG_THERMAL_INT);
+	tmp &= ~(CG_THERMAL_INT__DIG_THERM_INTH_MASK | CG_THERMAL_INT__DIG_THERM_INTL_MASK);
+	tmp |= ((high_temp / 1000) << CG_THERMAL_INT__DIG_THERM_INTH__SHIFT) |
+		((low_temp / 1000)) << CG_THERMAL_INT__DIG_THERM_INTL__SHIFT;
+	WREG32_SMC(ixCG_THERMAL_INT, tmp);
+
+#if 0
+	/* XXX: need to figure out how to handle this properly */
+	tmp = RREG32_SMC(ixCG_THERMAL_CTRL);
+	tmp &= DIG_THERM_DPM_MASK;
+	tmp |= DIG_THERM_DPM(high_temp / 1000);
+	WREG32_SMC(ixCG_THERMAL_CTRL, tmp);
+#endif
+
+	adev->pm.dpm.thermal.min_temp = low_temp;
+	adev->pm.dpm.thermal.max_temp = high_temp;
+	return 0;
+}
+
+static int ci_thermal_enable_alert(struct amdgpu_device *adev,
+				   bool enable)
+{
+	u32 thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+	PPSMC_Result result;
+
+	if (enable) {
+		thermal_int &= ~(CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK |
+				 CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK);
+		WREG32_SMC(ixCG_THERMAL_INT, thermal_int);
+		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Thermal_Cntl_Enable);
+		if (result != PPSMC_Result_OK) {
+			DRM_DEBUG_KMS("Could not enable thermal interrupts.\n");
+			return -EINVAL;
+		}
+	} else {
+		thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK |
+			CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+		WREG32_SMC(ixCG_THERMAL_INT, thermal_int);
+		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Thermal_Cntl_Disable);
+		if (result != PPSMC_Result_OK) {
+			DRM_DEBUG_KMS("Could not disable thermal interrupts.\n");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void ci_fan_ctrl_set_static_mode(struct amdgpu_device *adev, u32 mode)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (pi->fan_ctrl_is_in_default_mode) {
+		tmp = (RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__FDO_PWM_MODE_MASK)
+			>> CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
+		pi->fan_ctrl_default_mode = tmp;
+		tmp = (RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__TMIN_MASK)
+			>> CG_FDO_CTRL2__TMIN__SHIFT;
+		pi->t_min = tmp;
+		pi->fan_ctrl_is_in_default_mode = false;
+	}
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TMIN_MASK;
+	tmp |= 0 << CG_FDO_CTRL2__TMIN__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
+	tmp |= mode << CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+}
+
+static int ci_thermal_setup_fan_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	SMU7_Discrete_FanTable fan_table = { FDO_MODE_HARDWARE };
+	u32 duty100;
+	u32 t_diff1, t_diff2, pwm_diff1, pwm_diff2;
+	u16 fdo_min, slope1, slope2;
+	u32 reference_clock, tmp;
+	int ret;
+	u64 tmp64;
+
+	if (!pi->fan_table_start) {
+		adev->pm.dpm.fan.ucode_fan_control = false;
+		return 0;
+	}
+
+	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
+		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
+
+	if (duty100 == 0) {
+		adev->pm.dpm.fan.ucode_fan_control = false;
+		return 0;
+	}
+
+	tmp64 = (u64)adev->pm.dpm.fan.pwm_min * duty100;
+	do_div(tmp64, 10000);
+	fdo_min = (u16)tmp64;
+
+	t_diff1 = adev->pm.dpm.fan.t_med - adev->pm.dpm.fan.t_min;
+	t_diff2 = adev->pm.dpm.fan.t_high - adev->pm.dpm.fan.t_med;
+
+	pwm_diff1 = adev->pm.dpm.fan.pwm_med - adev->pm.dpm.fan.pwm_min;
+	pwm_diff2 = adev->pm.dpm.fan.pwm_high - adev->pm.dpm.fan.pwm_med;
+
+	slope1 = (u16)((50 + ((16 * duty100 * pwm_diff1) / t_diff1)) / 100);
+	slope2 = (u16)((50 + ((16 * duty100 * pwm_diff2) / t_diff2)) / 100);
+
+	fan_table.TempMin = cpu_to_be16((50 + adev->pm.dpm.fan.t_min) / 100);
+	fan_table.TempMed = cpu_to_be16((50 + adev->pm.dpm.fan.t_med) / 100);
+	fan_table.TempMax = cpu_to_be16((50 + adev->pm.dpm.fan.t_max) / 100);
+
+	fan_table.Slope1 = cpu_to_be16(slope1);
+	fan_table.Slope2 = cpu_to_be16(slope2);
+
+	fan_table.FdoMin = cpu_to_be16(fdo_min);
+
+	fan_table.HystDown = cpu_to_be16(adev->pm.dpm.fan.t_hyst);
+
+	fan_table.HystUp = cpu_to_be16(1);
+
+	fan_table.HystSlope = cpu_to_be16(1);
+
+	fan_table.TempRespLim = cpu_to_be16(5);
+
+	reference_clock = amdgpu_asic_get_xclk(adev);
+
+	fan_table.RefreshPeriod = cpu_to_be32((adev->pm.dpm.fan.cycle_delay *
+					       reference_clock) / 1600);
+
+	fan_table.FdoMax = cpu_to_be16((u16)duty100);
+
+	tmp = (RREG32_SMC(ixCG_MULT_THERMAL_CTRL) & CG_MULT_THERMAL_CTRL__TEMP_SEL_MASK)
+		>> CG_MULT_THERMAL_CTRL__TEMP_SEL__SHIFT;
+	fan_table.TempSrc = (uint8_t)tmp;
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev,
+					  pi->fan_table_start,
+					  (u8 *)(&fan_table),
+					  sizeof(fan_table),
+					  pi->sram_end);
+
+	if (ret) {
+		DRM_ERROR("Failed to load fan table to the SMC.");
+		adev->pm.dpm.fan.ucode_fan_control = false;
+	}
+
+	return 0;
+}
+
+static int ci_fan_ctrl_start_smc_fan_control(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result ret;
+
+	if (pi->caps_od_fuzzy_fan_control_support) {
+		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							       PPSMC_StartFanControl,
+							       FAN_CONTROL_FUZZY);
+		if (ret != PPSMC_Result_OK)
+			return -EINVAL;
+		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							       PPSMC_MSG_SetFanPwmMax,
+							       adev->pm.dpm.fan.default_max_fan_pwm);
+		if (ret != PPSMC_Result_OK)
+			return -EINVAL;
+	} else {
+		ret = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							       PPSMC_StartFanControl,
+							       FAN_CONTROL_TABLE);
+		if (ret != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	pi->fan_is_controlled_by_smc = true;
+	return 0;
+}
+
+
+static int ci_fan_ctrl_stop_smc_fan_control(struct amdgpu_device *adev)
+{
+	PPSMC_Result ret;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	ret = amdgpu_ci_send_msg_to_smc(adev, PPSMC_StopFanControl);
+	if (ret == PPSMC_Result_OK) {
+		pi->fan_is_controlled_by_smc = false;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+static int ci_dpm_get_fan_speed_percent(struct amdgpu_device *adev,
+					u32 *speed)
+{
+	u32 duty, duty100;
+	u64 tmp64;
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
+		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
+	duty = (RREG32_SMC(ixCG_THERMAL_STATUS) & CG_THERMAL_STATUS__FDO_PWM_DUTY_MASK)
+		>> CG_THERMAL_STATUS__FDO_PWM_DUTY__SHIFT;
+
+	if (duty100 == 0)
+		return -EINVAL;
+
+	tmp64 = (u64)duty * 100;
+	do_div(tmp64, duty100);
+	*speed = (u32)tmp64;
+
+	if (*speed > 100)
+		*speed = 100;
+
+	return 0;
+}
+
+static int ci_dpm_set_fan_speed_percent(struct amdgpu_device *adev,
+					u32 speed)
+{
+	u32 tmp;
+	u32 duty, duty100;
+	u64 tmp64;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (pi->fan_is_controlled_by_smc)
+		return -EINVAL;
+
+	if (speed > 100)
+		return -EINVAL;
+
+	duty100 = (RREG32_SMC(ixCG_FDO_CTRL1) & CG_FDO_CTRL1__FMAX_DUTY100_MASK)
+		>> CG_FDO_CTRL1__FMAX_DUTY100__SHIFT;
+
+	if (duty100 == 0)
+		return -EINVAL;
+
+	tmp64 = (u64)speed * duty100;
+	do_div(tmp64, 100);
+	duty = (u32)tmp64;
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL0) & ~CG_FDO_CTRL0__FDO_STATIC_DUTY_MASK;
+	tmp |= duty << CG_FDO_CTRL0__FDO_STATIC_DUTY__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL0, tmp);
+
+	return 0;
+}
+
+static void ci_dpm_set_fan_control_mode(struct amdgpu_device *adev, u32 mode)
+{
+	if (mode) {
+		/* stop auto-manage */
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			ci_fan_ctrl_stop_smc_fan_control(adev);
+		ci_fan_ctrl_set_static_mode(adev, mode);
+	} else {
+		/* restart auto-manage */
+		if (adev->pm.dpm.fan.ucode_fan_control)
+			ci_thermal_start_smc_fan_control(adev);
+		else
+			ci_fan_ctrl_set_default_mode(adev);
+	}
+}
+
+static u32 ci_dpm_get_fan_control_mode(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (pi->fan_is_controlled_by_smc)
+		return 0;
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
+	return (tmp >> CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT);
+}
+
+#if 0
+static int ci_fan_ctrl_get_fan_speed_rpm(struct amdgpu_device *adev,
+					 u32 *speed)
+{
+	u32 tach_period;
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (adev->pm.fan_pulses_per_revolution == 0)
+		return -ENOENT;
+
+	tach_period = (RREG32_SMC(ixCG_TACH_STATUS) & CG_TACH_STATUS__TACH_PERIOD_MASK)
+		>> CG_TACH_STATUS__TACH_PERIOD__SHIFT;
+	if (tach_period == 0)
+		return -ENOENT;
+
+	*speed = 60 * xclk * 10000 / tach_period;
+
+	return 0;
+}
+
+static int ci_fan_ctrl_set_fan_speed_rpm(struct amdgpu_device *adev,
+					 u32 speed)
+{
+	u32 tach_period, tmp;
+	u32 xclk = amdgpu_asic_get_xclk(adev);
+
+	if (adev->pm.no_fan)
+		return -ENOENT;
+
+	if (adev->pm.fan_pulses_per_revolution == 0)
+		return -ENOENT;
+
+	if ((speed < adev->pm.fan_min_rpm) ||
+	    (speed > adev->pm.fan_max_rpm))
+		return -EINVAL;
+
+	if (adev->pm.dpm.fan.ucode_fan_control)
+		ci_fan_ctrl_stop_smc_fan_control(adev);
+
+	tach_period = 60 * xclk * 10000 / (8 * speed);
+	tmp = RREG32_SMC(ixCG_TACH_CTRL) & ~CG_TACH_CTRL__TARGET_PERIOD_MASK;
+	tmp |= tach_period << CG_TACH_CTRL__TARGET_PERIOD__SHIFT;
+	WREG32_SMC(CG_TACH_CTRL, tmp);
+
+	ci_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC_RPM);
+
+	return 0;
+}
+#endif
+
+static void ci_fan_ctrl_set_default_mode(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (!pi->fan_ctrl_is_in_default_mode) {
+		tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__FDO_PWM_MODE_MASK;
+		tmp |= pi->fan_ctrl_default_mode << CG_FDO_CTRL2__FDO_PWM_MODE__SHIFT;
+		WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+
+		tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TMIN_MASK;
+		tmp |= pi->t_min << CG_FDO_CTRL2__TMIN__SHIFT;
+		WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+		pi->fan_ctrl_is_in_default_mode = true;
+	}
+}
+
+static void ci_thermal_start_smc_fan_control(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm.fan.ucode_fan_control) {
+		ci_fan_ctrl_start_smc_fan_control(adev);
+		ci_fan_ctrl_set_static_mode(adev, FDO_PWM_MODE_STATIC);
+	}
+}
+
+static void ci_thermal_initialize(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	if (adev->pm.fan_pulses_per_revolution) {
+		tmp = RREG32_SMC(ixCG_TACH_CTRL) & ~CG_TACH_CTRL__EDGE_PER_REV_MASK;
+		tmp |= (adev->pm.fan_pulses_per_revolution - 1)
+			<< CG_TACH_CTRL__EDGE_PER_REV__SHIFT;
+		WREG32_SMC(ixCG_TACH_CTRL, tmp);
+	}
+
+	tmp = RREG32_SMC(ixCG_FDO_CTRL2) & ~CG_FDO_CTRL2__TACH_PWM_RESP_RATE_MASK;
+	tmp |= 0x28 << CG_FDO_CTRL2__TACH_PWM_RESP_RATE__SHIFT;
+	WREG32_SMC(ixCG_FDO_CTRL2, tmp);
+}
+
+static int ci_thermal_start_thermal_controller(struct amdgpu_device *adev)
+{
+	int ret;
+
+	ci_thermal_initialize(adev);
+	ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN, CISLANDS_TEMP_RANGE_MAX);
+	if (ret)
+		return ret;
+	ret = ci_thermal_enable_alert(adev, true);
+	if (ret)
+		return ret;
+	if (adev->pm.dpm.fan.ucode_fan_control) {
+		ret = ci_thermal_setup_fan_table(adev);
+		if (ret)
+			return ret;
+		ci_thermal_start_smc_fan_control(adev);
+	}
+
+	return 0;
+}
+
+static void ci_thermal_stop_thermal_controller(struct amdgpu_device *adev)
+{
+	if (!adev->pm.no_fan)
+		ci_fan_ctrl_set_default_mode(adev);
+}
+
+#if 0
+static int ci_read_smc_soft_register(struct amdgpu_device *adev,
+				     u16 reg_offset, u32 *value)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	return amdgpu_ci_read_smc_sram_dword(adev,
+				      pi->soft_regs_start + reg_offset,
+				      value, pi->sram_end);
+}
+#endif
+
+static int ci_write_smc_soft_register(struct amdgpu_device *adev,
+				      u16 reg_offset, u32 value)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	return amdgpu_ci_write_smc_sram_dword(adev,
+				       pi->soft_regs_start + reg_offset,
+				       value, pi->sram_end);
+}
+
+static void ci_init_fps_limits(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	SMU7_Discrete_DpmTable *table = &pi->smc_state_table;
+
+	if (pi->caps_fps) {
+		u16 tmp;
+
+		tmp = 45;
+		table->FpsHighT = cpu_to_be16(tmp);
+
+		tmp = 30;
+		table->FpsLowT = cpu_to_be16(tmp);
+	}
+}
+
+static int ci_update_sclk_t(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret = 0;
+	u32 low_sclk_interrupt_t = 0;
+
+	if (pi->caps_sclk_throttle_low_notification) {
+		low_sclk_interrupt_t = cpu_to_be32(pi->low_sclk_interrupt_t);
+
+		ret = amdgpu_ci_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Discrete_DpmTable, LowSclkInterruptT),
+					   (u8 *)&low_sclk_interrupt_t,
+					   sizeof(u32), pi->sram_end);
+
+	}
+
+	return ret;
+}
+
+static void ci_get_leakage_voltages(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u16 leakage_id, virtual_voltage_id;
+	u16 vddc, vddci;
+	int i;
+
+	pi->vddc_leakage.count = 0;
+	pi->vddci_leakage.count = 0;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_EVV) {
+		for (i = 0; i < CISLANDS_MAX_LEAKAGE_COUNT; i++) {
+			virtual_voltage_id = ATOM_VIRTUAL_VOLTAGE_ID0 + i;
+			if (amdgpu_atombios_get_voltage_evv(adev, virtual_voltage_id, &vddc) != 0)
+				continue;
+			if (vddc != 0 && vddc != virtual_voltage_id) {
+				pi->vddc_leakage.actual_voltage[pi->vddc_leakage.count] = vddc;
+				pi->vddc_leakage.leakage_id[pi->vddc_leakage.count] = virtual_voltage_id;
+				pi->vddc_leakage.count++;
+			}
+		}
+	} else if (amdgpu_atombios_get_leakage_id_from_vbios(adev, &leakage_id) == 0) {
+		for (i = 0; i < CISLANDS_MAX_LEAKAGE_COUNT; i++) {
+			virtual_voltage_id = ATOM_VIRTUAL_VOLTAGE_ID0 + i;
+			if (amdgpu_atombios_get_leakage_vddc_based_on_leakage_params(adev, &vddc, &vddci,
+										     virtual_voltage_id,
+										     leakage_id) == 0) {
+				if (vddc != 0 && vddc != virtual_voltage_id) {
+					pi->vddc_leakage.actual_voltage[pi->vddc_leakage.count] = vddc;
+					pi->vddc_leakage.leakage_id[pi->vddc_leakage.count] = virtual_voltage_id;
+					pi->vddc_leakage.count++;
+				}
+				if (vddci != 0 && vddci != virtual_voltage_id) {
+					pi->vddci_leakage.actual_voltage[pi->vddci_leakage.count] = vddci;
+					pi->vddci_leakage.leakage_id[pi->vddci_leakage.count] = virtual_voltage_id;
+					pi->vddci_leakage.count++;
+				}
+			}
+		}
+	}
+}
+
+static void ci_set_dpm_event_sources(struct amdgpu_device *adev, u32 sources)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	bool want_thermal_protection;
+	enum amdgpu_dpm_event_src dpm_event_src;
+	u32 tmp;
+
+	switch (sources) {
+	case 0:
+	default:
+		want_thermal_protection = false;
+		break;
+	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGITAL;
+		break;
+	case (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_EXTERNAL;
+		break;
+	case ((1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_EXTERNAL) |
+	      (1 << AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL)):
+		want_thermal_protection = true;
+		dpm_event_src = AMDGPU_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL;
+		break;
+	}
+
+	if (want_thermal_protection) {
+#if 0
+		/* XXX: need to figure out how to handle this properly */
+		tmp = RREG32_SMC(ixCG_THERMAL_CTRL);
+		tmp &= DPM_EVENT_SRC_MASK;
+		tmp |= DPM_EVENT_SRC(dpm_event_src);
+		WREG32_SMC(ixCG_THERMAL_CTRL, tmp);
+#endif
+
+		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+		if (pi->thermal_protection)
+			tmp &= ~GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+		else
+			tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+	} else {
+		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+		tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+	}
+}
+
+static void ci_enable_auto_throttle_source(struct amdgpu_device *adev,
+					   enum amdgpu_dpm_auto_throttle_src source,
+					   bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (enable) {
+		if (!(pi->active_auto_throttle_sources & (1 << source))) {
+			pi->active_auto_throttle_sources |= 1 << source;
+			ci_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
+		}
+	} else {
+		if (pi->active_auto_throttle_sources & (1 << source)) {
+			pi->active_auto_throttle_sources &= ~(1 << source);
+			ci_set_dpm_event_sources(adev, pi->active_auto_throttle_sources);
+		}
+	}
+}
+
+static void ci_enable_vr_hot_gpio_interrupt(struct amdgpu_device *adev)
+{
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_REGULATOR_HOT)
+		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableVRHotGPIOInterrupt);
+}
+
+static int ci_unfreeze_sclk_mclk_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+
+	if (!pi->need_update_smu7_dpm_table)
+		return 0;
+
+	if ((!pi->sclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK))) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_SCLKDPM_UnfreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	if ((!pi->mclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_UnfreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	pi->need_update_smu7_dpm_table = 0;
+	return 0;
+}
+
+static int ci_enable_sclk_mclk_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+
+	if (enable) {
+		if (!pi->sclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DPM_Enable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+
+		if (!pi->mclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_Enable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+
+			WREG32_P(mmMC_SEQ_CNTL_3, MC_SEQ_CNTL_3__CAC_EN_MASK,
+					~MC_SEQ_CNTL_3__CAC_EN_MASK);
+
+			WREG32_SMC(ixLCAC_MC0_CNTL, 0x05);
+			WREG32_SMC(ixLCAC_MC1_CNTL, 0x05);
+			WREG32_SMC(ixLCAC_CPL_CNTL, 0x100005);
+
+			udelay(10);
+
+			WREG32_SMC(ixLCAC_MC0_CNTL, 0x400005);
+			WREG32_SMC(ixLCAC_MC1_CNTL, 0x400005);
+			WREG32_SMC(ixLCAC_CPL_CNTL, 0x500005);
+		}
+	} else {
+		if (!pi->sclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DPM_Disable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+
+		if (!pi->mclk_dpm_key_disabled) {
+			smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_Disable);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static int ci_start_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret;
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+	tmp |= GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp |= SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, VoltageChangeTimeout), 0x1000);
+
+	WREG32_P(mmBIF_LNCNT_RESET, 0, ~BIF_LNCNT_RESET__RESET_LNCNT_EN_MASK);
+
+	smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Voltage_Cntl_Enable);
+	if (smc_result != PPSMC_Result_OK)
+		return -EINVAL;
+
+	ret = ci_enable_sclk_mclk_dpm(adev, true);
+	if (ret)
+		return ret;
+
+	if (!pi->pcie_dpm_key_disabled) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_Enable);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_freeze_sclk_mclk_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+
+	if (!pi->need_update_smu7_dpm_table)
+		return 0;
+
+	if ((!pi->sclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK))) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_SCLKDPM_FreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	if ((!pi->mclk_dpm_key_disabled) &&
+	    (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MCLKDPM_FreezeLevel);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_stop_dpm(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result smc_result;
+	int ret;
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+	tmp &= ~GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp &= ~SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	if (!pi->pcie_dpm_key_disabled) {
+		smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_PCIeDPM_Disable);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	ret = ci_enable_sclk_mclk_dpm(adev, false);
+	if (ret)
+		return ret;
+
+	smc_result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Voltage_Cntl_Disable);
+	if (smc_result != PPSMC_Result_OK)
+		return -EINVAL;
+
+	return 0;
+}
+
+static void ci_enable_sclk_control(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+
+	if (enable)
+		tmp &= ~SCLK_PWRMGT_CNTL__SCLK_PWRMGT_OFF_MASK;
+	else
+		tmp |= SCLK_PWRMGT_CNTL__SCLK_PWRMGT_OFF_MASK;
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+}
+
+#if 0
+static int ci_notify_hw_of_power_source(struct amdgpu_device *adev,
+					bool ac_power)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_cac_tdp_table *cac_tdp_table =
+		adev->pm.dpm.dyn_state.cac_tdp_table;
+	u32 power_limit;
+
+	if (ac_power)
+		power_limit = (u32)(cac_tdp_table->maximum_power_delivery_limit * 256);
+	else
+		power_limit = (u32)(cac_tdp_table->battery_power_limit * 256);
+
+	ci_set_power_limit(adev, power_limit);
+
+	if (pi->caps_automatic_dc_transition) {
+		if (ac_power)
+			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_RunningOnAC);
+		else
+			amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_Remove_DC_Clamp);
+	}
+
+	return 0;
+}
+#endif
+
+static PPSMC_Result amdgpu_ci_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						      PPSMC_Msg msg, u32 parameter)
+{
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+	return amdgpu_ci_send_msg_to_smc(adev, msg);
+}
+
+static PPSMC_Result amdgpu_ci_send_msg_to_smc_return_parameter(struct amdgpu_device *adev,
+							PPSMC_Msg msg, u32 *parameter)
+{
+	PPSMC_Result smc_result;
+
+	smc_result = amdgpu_ci_send_msg_to_smc(adev, msg);
+
+	if ((smc_result == PPSMC_Result_OK) && parameter)
+		*parameter = RREG32(mmSMC_MSG_ARG_0);
+
+	return smc_result;
+}
+
+static int ci_dpm_force_state_sclk(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->sclk_dpm_key_disabled) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SCLKDPM_SetEnabledMask, 1 << n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_dpm_force_state_mclk(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->mclk_dpm_key_disabled) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_MCLKDPM_SetEnabledMask, 1 << n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_dpm_force_state_pcie(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->pcie_dpm_key_disabled) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_PCIeDPM_ForceLevel, n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_set_power_limit(struct amdgpu_device *adev, u32 n)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (pi->power_containment_features & POWERCONTAINMENT_FEATURE_PkgPwrLimit) {
+		PPSMC_Result smc_result =
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_PkgPwrSetLimit, n);
+		if (smc_result != PPSMC_Result_OK)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_set_overdrive_target_tdp(struct amdgpu_device *adev,
+				       u32 target_tdp)
+{
+	PPSMC_Result smc_result =
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_OverDriveSetTargetTdp, target_tdp);
+	if (smc_result != PPSMC_Result_OK)
+		return -EINVAL;
+	return 0;
+}
+
+#if 0
+static int ci_set_boot_state(struct amdgpu_device *adev)
+{
+	return ci_enable_sclk_mclk_dpm(adev, false);
+}
+#endif
+
+static u32 ci_get_average_sclk_freq(struct amdgpu_device *adev)
+{
+	u32 sclk_freq;
+	PPSMC_Result smc_result =
+		amdgpu_ci_send_msg_to_smc_return_parameter(adev,
+						    PPSMC_MSG_API_GetSclkFrequency,
+						    &sclk_freq);
+	if (smc_result != PPSMC_Result_OK)
+		sclk_freq = 0;
+
+	return sclk_freq;
+}
+
+static u32 ci_get_average_mclk_freq(struct amdgpu_device *adev)
+{
+	u32 mclk_freq;
+	PPSMC_Result smc_result =
+		amdgpu_ci_send_msg_to_smc_return_parameter(adev,
+						    PPSMC_MSG_API_GetMclkFrequency,
+						    &mclk_freq);
+	if (smc_result != PPSMC_Result_OK)
+		mclk_freq = 0;
+
+	return mclk_freq;
+}
+
+static void ci_dpm_start_smc(struct amdgpu_device *adev)
+{
+	int i;
+
+	amdgpu_ci_program_jump_on_start(adev);
+	amdgpu_ci_start_smc_clock(adev);
+	amdgpu_ci_start_smc(adev);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SMC(ixFIRMWARE_FLAGS) & FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK)
+			break;
+	}
+}
+
+static void ci_dpm_stop_smc(struct amdgpu_device *adev)
+{
+	amdgpu_ci_reset_smc(adev);
+	amdgpu_ci_stop_smc_clock(adev);
+}
+
+static int ci_process_firmware_header(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, DpmTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->dpm_table_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, SoftRegisters),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->soft_regs_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, mcRegisterTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->mc_reg_table_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, FanTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->fan_table_start = tmp;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev,
+				     SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, mcArbDramTimingTable),
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->arb_table_start = tmp;
+
+	return 0;
+}
+
+static void ci_read_clock_registers(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->clock_registers.cg_spll_func_cntl =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL);
+	pi->clock_registers.cg_spll_func_cntl_2 =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_2);
+	pi->clock_registers.cg_spll_func_cntl_3 =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_3);
+	pi->clock_registers.cg_spll_func_cntl_4 =
+		RREG32_SMC(ixCG_SPLL_FUNC_CNTL_4);
+	pi->clock_registers.cg_spll_spread_spectrum =
+		RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM);
+	pi->clock_registers.cg_spll_spread_spectrum_2 =
+		RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM_2);
+	pi->clock_registers.dll_cntl = RREG32(mmDLL_CNTL);
+	pi->clock_registers.mclk_pwrmgt_cntl = RREG32(mmMCLK_PWRMGT_CNTL);
+	pi->clock_registers.mpll_ad_func_cntl = RREG32(mmMPLL_AD_FUNC_CNTL);
+	pi->clock_registers.mpll_dq_func_cntl = RREG32(mmMPLL_DQ_FUNC_CNTL);
+	pi->clock_registers.mpll_func_cntl = RREG32(mmMPLL_FUNC_CNTL);
+	pi->clock_registers.mpll_func_cntl_1 = RREG32(mmMPLL_FUNC_CNTL_1);
+	pi->clock_registers.mpll_func_cntl_2 = RREG32(mmMPLL_FUNC_CNTL_2);
+	pi->clock_registers.mpll_ss1 = RREG32(mmMPLL_SS1);
+	pi->clock_registers.mpll_ss2 = RREG32(mmMPLL_SS2);
+}
+
+static void ci_init_sclk_t(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->low_sclk_interrupt_t = 0;
+}
+
+static void ci_enable_thermal_protection(struct amdgpu_device *adev,
+					 bool enable)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	if (enable)
+		tmp &= ~GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+	else
+		tmp |= GENERAL_PWRMGT__THERMAL_PROTECTION_DIS_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+}
+
+static void ci_enable_acpi_power_management(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	tmp |= GENERAL_PWRMGT__STATIC_PM_EN_MASK;
+
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+}
+
+#if 0
+static int ci_enter_ulp_state(struct amdgpu_device *adev)
+{
+
+	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_SwitchToMinimumPower);
+
+	udelay(25000);
+
+	return 0;
+}
+
+static int ci_exit_ulp_state(struct amdgpu_device *adev)
+{
+	int i;
+
+	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_ResumeFromMinimumPower);
+
+	udelay(7000);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmSMC_RESP_0) == 1)
+			break;
+		udelay(1000);
+	}
+
+	return 0;
+}
+#endif
+
+static int ci_notify_smc_display_change(struct amdgpu_device *adev,
+					bool has_display)
+{
+	PPSMC_Msg msg = has_display ? PPSMC_MSG_HasDisplay : PPSMC_MSG_NoDisplay;
+
+	return (amdgpu_ci_send_msg_to_smc(adev, msg) == PPSMC_Result_OK) ?  0 : -EINVAL;
+}
+
+static int ci_enable_ds_master_switch(struct amdgpu_device *adev,
+				      bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (enable) {
+		if (pi->caps_sclk_ds) {
+			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_ON) != PPSMC_Result_OK)
+				return -EINVAL;
+		} else {
+			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_OFF) != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	} else {
+		if (pi->caps_sclk_ds) {
+			if (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_MASTER_DeepSleep_OFF) != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static void ci_program_display_gap(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixCG_DISPLAY_GAP_CNTL);
+	u32 pre_vbi_time_in_us;
+	u32 frame_time_in_us;
+	u32 ref_clock = adev->clock.spll.reference_freq;
+	u32 refresh_rate = amdgpu_dpm_get_vrefresh(adev);
+	u32 vblank_time = amdgpu_dpm_get_vblank_time(adev);
+
+	tmp &= ~CG_DISPLAY_GAP_CNTL__DISP_GAP_MASK;
+	if (adev->pm.dpm.new_active_crtc_count > 0)
+		tmp |= (AMDGPU_PM_DISPLAY_GAP_VBLANK_OR_WM << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT);
+	else
+		tmp |= (AMDGPU_PM_DISPLAY_GAP_IGNORE << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT);
+	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL, tmp);
+
+	if (refresh_rate == 0)
+		refresh_rate = 60;
+	if (vblank_time == 0xffffffff)
+		vblank_time = 500;
+	frame_time_in_us = 1000000 / refresh_rate;
+	pre_vbi_time_in_us =
+		frame_time_in_us - 200 - vblank_time;
+	tmp = pre_vbi_time_in_us * (ref_clock / 100);
+
+	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL2, tmp);
+	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, PreVBlankGap), 0x64);
+	ci_write_smc_soft_register(adev, offsetof(SMU7_SoftRegisters, VBlankTimeout), (frame_time_in_us - pre_vbi_time_in_us));
+
+
+	ci_notify_smc_display_change(adev, (adev->pm.dpm.new_active_crtc_count == 1));
+
+}
+
+static void ci_enable_spread_spectrum(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (enable) {
+		if (pi->caps_sclk_ss_support) {
+			tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+			tmp |= GENERAL_PWRMGT__DYN_SPREAD_SPECTRUM_EN_MASK;
+			WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+		}
+	} else {
+		tmp = RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM);
+		tmp &= ~CG_SPLL_SPREAD_SPECTRUM__SSEN_MASK;
+		WREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM, tmp);
+
+		tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+		tmp &= ~GENERAL_PWRMGT__DYN_SPREAD_SPECTRUM_EN_MASK;
+		WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+	}
+}
+
+static void ci_program_sstp(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_STATIC_SCREEN_PARAMETER,
+	((CISLANDS_SSTU_DFLT << CG_STATIC_SCREEN_PARAMETER__STATIC_SCREEN_THRESHOLD_UNIT__SHIFT) |
+	 (CISLANDS_SST_DFLT << CG_STATIC_SCREEN_PARAMETER__STATIC_SCREEN_THRESHOLD__SHIFT)));
+}
+
+static void ci_enable_display_gap(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixCG_DISPLAY_GAP_CNTL);
+
+	tmp &= ~(CG_DISPLAY_GAP_CNTL__DISP_GAP_MASK |
+			CG_DISPLAY_GAP_CNTL__DISP_GAP_MCHG_MASK);
+	tmp |= ((AMDGPU_PM_DISPLAY_GAP_IGNORE << CG_DISPLAY_GAP_CNTL__DISP_GAP__SHIFT) |
+		(AMDGPU_PM_DISPLAY_GAP_VBLANK << CG_DISPLAY_GAP_CNTL__DISP_GAP_MCHG__SHIFT));
+
+	WREG32_SMC(ixCG_DISPLAY_GAP_CNTL, tmp);
+}
+
+static void ci_program_vc(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp &= ~(SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK | SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, CISLANDS_VRC_DFLT0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_1, CISLANDS_VRC_DFLT1);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_2, CISLANDS_VRC_DFLT2);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_3, CISLANDS_VRC_DFLT3);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_4, CISLANDS_VRC_DFLT4);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_5, CISLANDS_VRC_DFLT5);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_6, CISLANDS_VRC_DFLT6);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_7, CISLANDS_VRC_DFLT7);
+}
+
+static void ci_clear_vc(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+	tmp |= (SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK | SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, tmp);
+
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_1, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_2, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_3, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_4, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_5, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_6, 0);
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_7, 0);
+}
+
+static int ci_upload_firmware(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int i, ret;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32_SMC(ixRCU_UC_EVENTS) & RCU_UC_EVENTS__boot_seq_done_MASK)
+			break;
+	}
+	WREG32_SMC(ixSMC_SYSCON_MISC_CNTL, 1);
+
+	amdgpu_ci_stop_smc_clock(adev);
+	amdgpu_ci_reset_smc(adev);
+
+	ret = amdgpu_ci_load_smc_ucode(adev, pi->sram_end);
+
+	return ret;
+
+}
+
+static int ci_get_svi2_voltage_table(struct amdgpu_device *adev,
+				     struct amdgpu_clock_voltage_dependency_table *voltage_dependency_table,
+				     struct atom_voltage_table *voltage_table)
+{
+	u32 i;
+
+	if (voltage_dependency_table == NULL)
+		return -EINVAL;
+
+	voltage_table->mask_low = 0;
+	voltage_table->phase_delay = 0;
+
+	voltage_table->count = voltage_dependency_table->count;
+	for (i = 0; i < voltage_table->count; i++) {
+		voltage_table->entries[i].value = voltage_dependency_table->entries[i].v;
+		voltage_table->entries[i].smio_low = 0;
+	}
+
+	return 0;
+}
+
+static int ci_construct_voltage_tables(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDC,
+							VOLTAGE_OBJ_GPIO_LUT,
+							&pi->vddc_voltage_table);
+		if (ret)
+			return ret;
+	} else if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		ret = ci_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+						&pi->vddc_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->vddc_voltage_table.count > SMU7_MAX_LEVELS_VDDC)
+		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_VDDC,
+							 &pi->vddc_voltage_table);
+
+	if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_VDDCI,
+							VOLTAGE_OBJ_GPIO_LUT,
+							&pi->vddci_voltage_table);
+		if (ret)
+			return ret;
+	} else if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		ret = ci_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+						&pi->vddci_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->vddci_voltage_table.count > SMU7_MAX_LEVELS_VDDCI)
+		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_VDDCI,
+							 &pi->vddci_voltage_table);
+
+	if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO) {
+		ret = amdgpu_atombios_get_voltage_table(adev, VOLTAGE_TYPE_MVDDC,
+							VOLTAGE_OBJ_GPIO_LUT,
+							&pi->mvdd_voltage_table);
+		if (ret)
+			return ret;
+	} else if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		ret = ci_get_svi2_voltage_table(adev,
+						&adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
+						&pi->mvdd_voltage_table);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->mvdd_voltage_table.count > SMU7_MAX_LEVELS_MVDD)
+		ci_trim_voltage_table_to_fit_state_table(adev, SMU7_MAX_LEVELS_MVDD,
+							 &pi->mvdd_voltage_table);
+
+	return 0;
+}
+
+static void ci_populate_smc_voltage_table(struct amdgpu_device *adev,
+					  struct atom_voltage_table_entry *voltage_table,
+					  SMU7_Discrete_VoltageLevel *smc_voltage_table)
+{
+	int ret;
+
+	ret = ci_get_std_voltage_value_sidd(adev, voltage_table,
+					    &smc_voltage_table->StdVoltageHiSidd,
+					    &smc_voltage_table->StdVoltageLoSidd);
+
+	if (ret) {
+		smc_voltage_table->StdVoltageHiSidd = voltage_table->value * VOLTAGE_SCALE;
+		smc_voltage_table->StdVoltageLoSidd = voltage_table->value * VOLTAGE_SCALE;
+	}
+
+	smc_voltage_table->Voltage = cpu_to_be16(voltage_table->value * VOLTAGE_SCALE);
+	smc_voltage_table->StdVoltageHiSidd =
+		cpu_to_be16(smc_voltage_table->StdVoltageHiSidd);
+	smc_voltage_table->StdVoltageLoSidd =
+		cpu_to_be16(smc_voltage_table->StdVoltageLoSidd);
+}
+
+static int ci_populate_smc_vddc_table(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	unsigned int count;
+
+	table->VddcLevelCount = pi->vddc_voltage_table.count;
+	for (count = 0; count < table->VddcLevelCount; count++) {
+		ci_populate_smc_voltage_table(adev,
+					      &pi->vddc_voltage_table.entries[count],
+					      &table->VddcLevel[count]);
+
+		if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
+			table->VddcLevel[count].Smio |=
+				pi->vddc_voltage_table.entries[count].smio_low;
+		else
+			table->VddcLevel[count].Smio = 0;
+	}
+	table->VddcLevelCount = cpu_to_be32(table->VddcLevelCount);
+
+	return 0;
+}
+
+static int ci_populate_smc_vddci_table(struct amdgpu_device *adev,
+				       SMU7_Discrete_DpmTable *table)
+{
+	unsigned int count;
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	table->VddciLevelCount = pi->vddci_voltage_table.count;
+	for (count = 0; count < table->VddciLevelCount; count++) {
+		ci_populate_smc_voltage_table(adev,
+					      &pi->vddci_voltage_table.entries[count],
+					      &table->VddciLevel[count]);
+
+		if (pi->vddci_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
+			table->VddciLevel[count].Smio |=
+				pi->vddci_voltage_table.entries[count].smio_low;
+		else
+			table->VddciLevel[count].Smio = 0;
+	}
+	table->VddciLevelCount = cpu_to_be32(table->VddciLevelCount);
+
+	return 0;
+}
+
+static int ci_populate_smc_mvdd_table(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	unsigned int count;
+
+	table->MvddLevelCount = pi->mvdd_voltage_table.count;
+	for (count = 0; count < table->MvddLevelCount; count++) {
+		ci_populate_smc_voltage_table(adev,
+					      &pi->mvdd_voltage_table.entries[count],
+					      &table->MvddLevel[count]);
+
+		if (pi->mvdd_control == CISLANDS_VOLTAGE_CONTROL_BY_GPIO)
+			table->MvddLevel[count].Smio |=
+				pi->mvdd_voltage_table.entries[count].smio_low;
+		else
+			table->MvddLevel[count].Smio = 0;
+	}
+	table->MvddLevelCount = cpu_to_be32(table->MvddLevelCount);
+
+	return 0;
+}
+
+static int ci_populate_smc_voltage_tables(struct amdgpu_device *adev,
+					  SMU7_Discrete_DpmTable *table)
+{
+	int ret;
+
+	ret = ci_populate_smc_vddc_table(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_vddci_table(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_mvdd_table(adev, table);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int ci_populate_mvdd_value(struct amdgpu_device *adev, u32 mclk,
+				  SMU7_Discrete_VoltageLevel *voltage)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i = 0;
+
+	if (pi->mvdd_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
+		for (i = 0; i < adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.count; i++) {
+			if (mclk <= adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.entries[i].clk) {
+				voltage->Voltage = pi->mvdd_voltage_table.entries[i].value;
+				break;
+			}
+		}
+
+		if (i >= adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.count)
+			return -EINVAL;
+	}
+
+	return -EINVAL;
+}
+
+static int ci_get_std_voltage_value_sidd(struct amdgpu_device *adev,
+					 struct atom_voltage_table_entry *voltage_table,
+					 u16 *std_voltage_hi_sidd, u16 *std_voltage_lo_sidd)
+{
+	u16 v_index, idx;
+	bool voltage_found = false;
+	*std_voltage_hi_sidd = voltage_table->value * VOLTAGE_SCALE;
+	*std_voltage_lo_sidd = voltage_table->value * VOLTAGE_SCALE;
+
+	if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries == NULL)
+		return -EINVAL;
+
+	if (adev->pm.dpm.dyn_state.cac_leakage_table.entries) {
+		for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
+			if (voltage_table->value ==
+			    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
+				voltage_found = true;
+				if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+					idx = v_index;
+				else
+					idx = adev->pm.dpm.dyn_state.cac_leakage_table.count - 1;
+				*std_voltage_lo_sidd =
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].vddc * VOLTAGE_SCALE;
+				*std_voltage_hi_sidd =
+					adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].leakage * VOLTAGE_SCALE;
+				break;
+			}
+		}
+
+		if (!voltage_found) {
+			for (v_index = 0; (u32)v_index < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; v_index++) {
+				if (voltage_table->value <=
+				    adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[v_index].v) {
+					voltage_found = true;
+					if ((u32)v_index < adev->pm.dpm.dyn_state.cac_leakage_table.count)
+						idx = v_index;
+					else
+						idx = adev->pm.dpm.dyn_state.cac_leakage_table.count - 1;
+					*std_voltage_lo_sidd =
+						adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].vddc * VOLTAGE_SCALE;
+					*std_voltage_hi_sidd =
+						adev->pm.dpm.dyn_state.cac_leakage_table.entries[idx].leakage * VOLTAGE_SCALE;
+					break;
+				}
+			}
+		}
+	}
+
+	return 0;
+}
+
+static void ci_populate_phase_value_based_on_sclk(struct amdgpu_device *adev,
+						  const struct amdgpu_phase_shedding_limits_table *limits,
+						  u32 sclk,
+						  u32 *phase_shedding)
+{
+	unsigned int i;
+
+	*phase_shedding = 1;
+
+	for (i = 0; i < limits->count; i++) {
+		if (sclk < limits->entries[i].sclk) {
+			*phase_shedding = i;
+			break;
+		}
+	}
+}
+
+static void ci_populate_phase_value_based_on_mclk(struct amdgpu_device *adev,
+						  const struct amdgpu_phase_shedding_limits_table *limits,
+						  u32 mclk,
+						  u32 *phase_shedding)
+{
+	unsigned int i;
+
+	*phase_shedding = 1;
+
+	for (i = 0; i < limits->count; i++) {
+		if (mclk < limits->entries[i].mclk) {
+			*phase_shedding = i;
+			break;
+		}
+	}
+}
+
+static int ci_init_arb_table_index(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_ci_read_smc_sram_dword(adev, pi->arb_table_start,
+				     &tmp, pi->sram_end);
+	if (ret)
+		return ret;
+
+	tmp &= 0x00FFFFFF;
+	tmp |= MC_CG_ARB_FREQ_F1 << 24;
+
+	return amdgpu_ci_write_smc_sram_dword(adev, pi->arb_table_start,
+				       tmp, pi->sram_end);
+}
+
+static int ci_get_dependency_volt_by_clk(struct amdgpu_device *adev,
+					 struct amdgpu_clock_voltage_dependency_table *allowed_clock_voltage_table,
+					 u32 clock, u32 *voltage)
+{
+	u32 i = 0;
+
+	if (allowed_clock_voltage_table->count == 0)
+		return -EINVAL;
+
+	for (i = 0; i < allowed_clock_voltage_table->count; i++) {
+		if (allowed_clock_voltage_table->entries[i].clk >= clock) {
+			*voltage = allowed_clock_voltage_table->entries[i].v;
+			return 0;
+		}
+	}
+
+	*voltage = allowed_clock_voltage_table->entries[i-1].v;
+
+	return 0;
+}
+
+static u8 ci_get_sleep_divider_id_from_clock(struct amdgpu_device *adev,
+					     u32 sclk, u32 min_sclk_in_sr)
+{
+	u32 i;
+	u32 tmp;
+	u32 min = (min_sclk_in_sr > CISLAND_MINIMUM_ENGINE_CLOCK) ?
+		min_sclk_in_sr : CISLAND_MINIMUM_ENGINE_CLOCK;
+
+	if (sclk < min)
+		return 0;
+
+	for (i = CISLAND_MAX_DEEPSLEEP_DIVIDER_ID;  ; i--) {
+		tmp = sclk / (1 << i);
+		if (tmp >= min || i == 0)
+			break;
+	}
+
+	return (u8)i;
+}
+
+static int ci_initial_switch_from_arb_f0_to_f1(struct amdgpu_device *adev)
+{
+	return ci_copy_and_switch_arb_sets(adev, MC_CG_ARB_FREQ_F0, MC_CG_ARB_FREQ_F1);
+}
+
+static int ci_reset_to_default(struct amdgpu_device *adev)
+{
+	return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_ResetToDefaults) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int ci_force_switch_to_arb_f0(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = (RREG32_SMC(ixSMC_SCRATCH9) & 0x0000ff00) >> 8;
+
+	if (tmp == MC_CG_ARB_FREQ_F0)
+		return 0;
+
+	return ci_copy_and_switch_arb_sets(adev, tmp, MC_CG_ARB_FREQ_F0);
+}
+
+static void ci_register_patching_mc_arb(struct amdgpu_device *adev,
+					const u32 engine_clock,
+					const u32 memory_clock,
+					u32 *dram_timimg2)
+{
+	bool patch;
+	u32 tmp, tmp2;
+
+	tmp = RREG32(mmMC_SEQ_MISC0);
+	patch = ((tmp & 0x0000f00) == 0x300) ? true : false;
+
+	if (patch &&
+	    ((adev->pdev->device == 0x67B0) ||
+	     (adev->pdev->device == 0x67B1))) {
+		if ((memory_clock > 100000) && (memory_clock <= 125000)) {
+			tmp2 = (((0x31 * engine_clock) / 125000) - 1) & 0xff;
+			*dram_timimg2 &= ~0x00ff0000;
+			*dram_timimg2 |= tmp2 << 16;
+		} else if ((memory_clock > 125000) && (memory_clock <= 137500)) {
+			tmp2 = (((0x36 * engine_clock) / 137500) - 1) & 0xff;
+			*dram_timimg2 &= ~0x00ff0000;
+			*dram_timimg2 |= tmp2 << 16;
+		}
+	}
+}
+
+static int ci_populate_memory_timing_parameters(struct amdgpu_device *adev,
+						u32 sclk,
+						u32 mclk,
+						SMU7_Discrete_MCArbDramTimingTableEntry *arb_regs)
+{
+	u32 dram_timing;
+	u32 dram_timing2;
+	u32 burst_time;
+
+	amdgpu_atombios_set_engine_dram_timings(adev, sclk, mclk);
+
+	dram_timing  = RREG32(mmMC_ARB_DRAM_TIMING);
+	dram_timing2 = RREG32(mmMC_ARB_DRAM_TIMING2);
+	burst_time = RREG32(mmMC_ARB_BURST_TIME) & MC_ARB_BURST_TIME__STATE0_MASK;
+
+	ci_register_patching_mc_arb(adev, sclk, mclk, &dram_timing2);
+
+	arb_regs->McArbDramTiming  = cpu_to_be32(dram_timing);
+	arb_regs->McArbDramTiming2 = cpu_to_be32(dram_timing2);
+	arb_regs->McArbBurstTime = (u8)burst_time;
+
+	return 0;
+}
+
+static int ci_do_program_memory_timing_parameters(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	SMU7_Discrete_MCArbDramTimingTable arb_regs;
+	u32 i, j;
+	int ret =  0;
+
+	memset(&arb_regs, 0, sizeof(SMU7_Discrete_MCArbDramTimingTable));
+
+	for (i = 0; i < pi->dpm_table.sclk_table.count; i++) {
+		for (j = 0; j < pi->dpm_table.mclk_table.count; j++) {
+			ret = ci_populate_memory_timing_parameters(adev,
+								   pi->dpm_table.sclk_table.dpm_levels[i].value,
+								   pi->dpm_table.mclk_table.dpm_levels[j].value,
+								   &arb_regs.entries[i][j]);
+			if (ret)
+				break;
+		}
+	}
+
+	if (ret == 0)
+		ret = amdgpu_ci_copy_bytes_to_smc(adev,
+					   pi->arb_table_start,
+					   (u8 *)&arb_regs,
+					   sizeof(SMU7_Discrete_MCArbDramTimingTable),
+					   pi->sram_end);
+
+	return ret;
+}
+
+static int ci_program_memory_timing_parameters(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (pi->need_update_smu7_dpm_table == 0)
+		return 0;
+
+	return ci_do_program_memory_timing_parameters(adev);
+}
+
+static void ci_populate_smc_initial_state(struct amdgpu_device *adev,
+					  struct amdgpu_ps *amdgpu_boot_state)
+{
+	struct ci_ps *boot_state = ci_get_ps(amdgpu_boot_state);
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 level = 0;
+
+	for (level = 0; level < adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.count; level++) {
+		if (adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[level].clk >=
+		    boot_state->performance_levels[0].sclk) {
+			pi->smc_state_table.GraphicsBootLevel = level;
+			break;
+		}
+	}
+
+	for (level = 0; level < adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.count; level++) {
+		if (adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.entries[level].clk >=
+		    boot_state->performance_levels[0].mclk) {
+			pi->smc_state_table.MemoryBootLevel = level;
+			break;
+		}
+	}
+}
+
+static u32 ci_get_dpm_level_enable_mask_value(struct ci_single_dpm_table *dpm_table)
+{
+	u32 i;
+	u32 mask_value = 0;
+
+	for (i = dpm_table->count; i > 0; i--) {
+		mask_value = mask_value << 1;
+		if (dpm_table->dpm_levels[i-1].enabled)
+			mask_value |= 0x1;
+		else
+			mask_value &= 0xFFFFFFFE;
+	}
+
+	return mask_value;
+}
+
+static void ci_populate_smc_link_level(struct amdgpu_device *adev,
+				       SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	u32 i;
+
+	for (i = 0; i < dpm_table->pcie_speed_table.count; i++) {
+		table->LinkLevel[i].PcieGenSpeed =
+			(u8)dpm_table->pcie_speed_table.dpm_levels[i].value;
+		table->LinkLevel[i].PcieLaneCount =
+			amdgpu_encode_pci_lane_width(dpm_table->pcie_speed_table.dpm_levels[i].param1);
+		table->LinkLevel[i].EnabledForActivity = 1;
+		table->LinkLevel[i].DownT = cpu_to_be32(5);
+		table->LinkLevel[i].UpT = cpu_to_be32(30);
+	}
+
+	pi->smc_state_table.LinkLevelCount = (u8)dpm_table->pcie_speed_table.count;
+	pi->dpm_level_enable_mask.pcie_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&dpm_table->pcie_speed_table);
+}
+
+static int ci_populate_smc_uvd_level(struct amdgpu_device *adev,
+				     SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->UvdLevelCount =
+		adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count;
+
+	for (count = 0; count < table->UvdLevelCount; count++) {
+		table->UvdLevel[count].VclkFrequency =
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].vclk;
+		table->UvdLevel[count].DclkFrequency =
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].dclk;
+		table->UvdLevel[count].MinVddc =
+			adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
+		table->UvdLevel[count].MinVddcPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->UvdLevel[count].VclkFrequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->UvdLevel[count].VclkDivider = (u8)dividers.post_divider;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->UvdLevel[count].DclkFrequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->UvdLevel[count].DclkDivider = (u8)dividers.post_divider;
+
+		table->UvdLevel[count].VclkFrequency = cpu_to_be32(table->UvdLevel[count].VclkFrequency);
+		table->UvdLevel[count].DclkFrequency = cpu_to_be32(table->UvdLevel[count].DclkFrequency);
+		table->UvdLevel[count].MinVddc = cpu_to_be16(table->UvdLevel[count].MinVddc);
+	}
+
+	return ret;
+}
+
+static int ci_populate_smc_vce_level(struct amdgpu_device *adev,
+				     SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->VceLevelCount =
+		adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count;
+
+	for (count = 0; count < table->VceLevelCount; count++) {
+		table->VceLevel[count].Frequency =
+			adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[count].evclk;
+		table->VceLevel[count].MinVoltage =
+			(u16)adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
+		table->VceLevel[count].MinPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->VceLevel[count].Frequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->VceLevel[count].Divider = (u8)dividers.post_divider;
+
+		table->VceLevel[count].Frequency = cpu_to_be32(table->VceLevel[count].Frequency);
+		table->VceLevel[count].MinVoltage = cpu_to_be16(table->VceLevel[count].MinVoltage);
+	}
+
+	return ret;
+
+}
+
+static int ci_populate_smc_acp_level(struct amdgpu_device *adev,
+				     SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->AcpLevelCount = (u8)
+		(adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count);
+
+	for (count = 0; count < table->AcpLevelCount; count++) {
+		table->AcpLevel[count].Frequency =
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[count].clk;
+		table->AcpLevel[count].MinVoltage =
+			adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[count].v;
+		table->AcpLevel[count].MinPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->AcpLevel[count].Frequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->AcpLevel[count].Divider = (u8)dividers.post_divider;
+
+		table->AcpLevel[count].Frequency = cpu_to_be32(table->AcpLevel[count].Frequency);
+		table->AcpLevel[count].MinVoltage = cpu_to_be16(table->AcpLevel[count].MinVoltage);
+	}
+
+	return ret;
+}
+
+static int ci_populate_smc_samu_level(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	u32 count;
+	struct atom_clock_dividers dividers;
+	int ret = -EINVAL;
+
+	table->SamuLevelCount =
+		adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count;
+
+	for (count = 0; count < table->SamuLevelCount; count++) {
+		table->SamuLevel[count].Frequency =
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[count].clk;
+		table->SamuLevel[count].MinVoltage =
+			adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[count].v * VOLTAGE_SCALE;
+		table->SamuLevel[count].MinPhases = 1;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev,
+							 COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+							 table->SamuLevel[count].Frequency, false, &dividers);
+		if (ret)
+			return ret;
+
+		table->SamuLevel[count].Divider = (u8)dividers.post_divider;
+
+		table->SamuLevel[count].Frequency = cpu_to_be32(table->SamuLevel[count].Frequency);
+		table->SamuLevel[count].MinVoltage = cpu_to_be16(table->SamuLevel[count].MinVoltage);
+	}
+
+	return ret;
+}
+
+static int ci_calculate_mclk_params(struct amdgpu_device *adev,
+				    u32 memory_clock,
+				    SMU7_Discrete_MemoryLevel *mclk,
+				    bool strobe_mode,
+				    bool dll_state_on)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32  dll_cntl = pi->clock_registers.dll_cntl;
+	u32  mclk_pwrmgt_cntl = pi->clock_registers.mclk_pwrmgt_cntl;
+	u32  mpll_ad_func_cntl = pi->clock_registers.mpll_ad_func_cntl;
+	u32  mpll_dq_func_cntl = pi->clock_registers.mpll_dq_func_cntl;
+	u32  mpll_func_cntl = pi->clock_registers.mpll_func_cntl;
+	u32  mpll_func_cntl_1 = pi->clock_registers.mpll_func_cntl_1;
+	u32  mpll_func_cntl_2 = pi->clock_registers.mpll_func_cntl_2;
+	u32  mpll_ss1 = pi->clock_registers.mpll_ss1;
+	u32  mpll_ss2 = pi->clock_registers.mpll_ss2;
+	struct atom_mpll_param mpll_param;
+	int ret;
+
+	ret = amdgpu_atombios_get_memory_pll_dividers(adev, memory_clock, strobe_mode, &mpll_param);
+	if (ret)
+		return ret;
+
+	mpll_func_cntl &= ~MPLL_FUNC_CNTL__BWCTRL_MASK;
+	mpll_func_cntl |= (mpll_param.bwcntl << MPLL_FUNC_CNTL__BWCTRL__SHIFT);
+
+	mpll_func_cntl_1 &= ~(MPLL_FUNC_CNTL_1__CLKF_MASK | MPLL_FUNC_CNTL_1__CLKFRAC_MASK |
+			MPLL_FUNC_CNTL_1__VCO_MODE_MASK);
+	mpll_func_cntl_1 |= (mpll_param.clkf) << MPLL_FUNC_CNTL_1__CLKF__SHIFT |
+		(mpll_param.clkfrac << MPLL_FUNC_CNTL_1__CLKFRAC__SHIFT) |
+		(mpll_param.vco_mode << MPLL_FUNC_CNTL_1__VCO_MODE__SHIFT);
+
+	mpll_ad_func_cntl &= ~MPLL_AD_FUNC_CNTL__YCLK_POST_DIV_MASK;
+	mpll_ad_func_cntl |= (mpll_param.post_div << MPLL_AD_FUNC_CNTL__YCLK_POST_DIV__SHIFT);
+
+	if (adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		mpll_dq_func_cntl &= ~(MPLL_DQ_FUNC_CNTL__YCLK_SEL_MASK |
+				MPLL_AD_FUNC_CNTL__YCLK_POST_DIV_MASK);
+		mpll_dq_func_cntl |= (mpll_param.yclk_sel << MPLL_DQ_FUNC_CNTL__YCLK_SEL__SHIFT) |
+				(mpll_param.post_div << MPLL_AD_FUNC_CNTL__YCLK_POST_DIV__SHIFT);
+	}
+
+	if (pi->caps_mclk_ss_support) {
+		struct amdgpu_atom_ss ss;
+		u32 freq_nom;
+		u32 tmp;
+		u32 reference_clock = adev->clock.mpll.reference_freq;
+
+		if (mpll_param.qdr == 1)
+			freq_nom = memory_clock * 4 * (1 << mpll_param.post_div);
+		else
+			freq_nom = memory_clock * 2 * (1 << mpll_param.post_div);
+
+		tmp = (freq_nom / reference_clock);
+		tmp = tmp * tmp;
+		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						     ASIC_INTERNAL_MEMORY_SS, freq_nom)) {
+			u32 clks = reference_clock * 5 / ss.rate;
+			u32 clkv = (u32)((((131 * ss.percentage * ss.rate) / 100) * tmp) / freq_nom);
+
+			mpll_ss1 &= ~MPLL_SS1__CLKV_MASK;
+			mpll_ss1 |= (clkv << MPLL_SS1__CLKV__SHIFT);
+
+			mpll_ss2 &= ~MPLL_SS2__CLKS_MASK;
+			mpll_ss2 |= (clks << MPLL_SS2__CLKS__SHIFT);
+		}
+	}
+
+	mclk_pwrmgt_cntl &= ~MCLK_PWRMGT_CNTL__DLL_SPEED_MASK;
+	mclk_pwrmgt_cntl |= (mpll_param.dll_speed << MCLK_PWRMGT_CNTL__DLL_SPEED__SHIFT);
+
+	if (dll_state_on)
+		mclk_pwrmgt_cntl |= MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
+			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK;
+	else
+		mclk_pwrmgt_cntl &= ~(MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
+			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK);
+
+	mclk->MclkFrequency = memory_clock;
+	mclk->MpllFuncCntl = mpll_func_cntl;
+	mclk->MpllFuncCntl_1 = mpll_func_cntl_1;
+	mclk->MpllFuncCntl_2 = mpll_func_cntl_2;
+	mclk->MpllAdFuncCntl = mpll_ad_func_cntl;
+	mclk->MpllDqFuncCntl = mpll_dq_func_cntl;
+	mclk->MclkPwrmgtCntl = mclk_pwrmgt_cntl;
+	mclk->DllCntl = dll_cntl;
+	mclk->MpllSs1 = mpll_ss1;
+	mclk->MpllSs2 = mpll_ss2;
+
+	return 0;
+}
+
+static int ci_populate_single_memory_level(struct amdgpu_device *adev,
+					   u32 memory_clock,
+					   SMU7_Discrete_MemoryLevel *memory_level)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+	bool dll_state_on;
+
+	if (adev->pm.dpm.dyn_state.vddc_dependency_on_mclk.entries) {
+		ret = ci_get_dependency_volt_by_clk(adev,
+						    &adev->pm.dpm.dyn_state.vddc_dependency_on_mclk,
+						    memory_clock, &memory_level->MinVddc);
+		if (ret)
+			return ret;
+	}
+
+	if (adev->pm.dpm.dyn_state.vddci_dependency_on_mclk.entries) {
+		ret = ci_get_dependency_volt_by_clk(adev,
+						    &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk,
+						    memory_clock, &memory_level->MinVddci);
+		if (ret)
+			return ret;
+	}
+
+	if (adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk.entries) {
+		ret = ci_get_dependency_volt_by_clk(adev,
+						    &adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk,
+						    memory_clock, &memory_level->MinMvdd);
+		if (ret)
+			return ret;
+	}
+
+	memory_level->MinVddcPhases = 1;
+
+	if (pi->vddc_phase_shed_control)
+		ci_populate_phase_value_based_on_mclk(adev,
+						      &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+						      memory_clock,
+						      &memory_level->MinVddcPhases);
+
+	memory_level->EnabledForThrottle = 1;
+	memory_level->EnabledForActivity = 1;
+	memory_level->UpH = 0;
+	memory_level->DownH = 100;
+	memory_level->VoltageDownH = 0;
+	memory_level->ActivityLevel = (u16)pi->mclk_activity_target;
+
+	memory_level->StutterEnable = false;
+	memory_level->StrobeEnable = false;
+	memory_level->EdcReadEnable = false;
+	memory_level->EdcWriteEnable = false;
+	memory_level->RttEnable = false;
+
+	memory_level->DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
+
+	if (pi->mclk_stutter_mode_threshold &&
+	    (memory_clock <= pi->mclk_stutter_mode_threshold) &&
+	    (pi->uvd_enabled == false) &&
+	    (RREG32(mmDPG_PIPE_STUTTER_CONTROL) & DPG_PIPE_STUTTER_CONTROL__STUTTER_ENABLE_MASK) &&
+	    (adev->pm.dpm.new_active_crtc_count <= 2))
+		memory_level->StutterEnable = true;
+
+	if (pi->mclk_strobe_mode_threshold &&
+	    (memory_clock <= pi->mclk_strobe_mode_threshold))
+		memory_level->StrobeEnable = 1;
+
+	if (adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5) {
+		memory_level->StrobeRatio =
+			ci_get_mclk_frequency_ratio(memory_clock, memory_level->StrobeEnable);
+		if (pi->mclk_edc_enable_threshold &&
+		    (memory_clock > pi->mclk_edc_enable_threshold))
+			memory_level->EdcReadEnable = true;
+
+		if (pi->mclk_edc_wr_enable_threshold &&
+		    (memory_clock > pi->mclk_edc_wr_enable_threshold))
+			memory_level->EdcWriteEnable = true;
+
+		if (memory_level->StrobeEnable) {
+			if (ci_get_mclk_frequency_ratio(memory_clock, true) >=
+			    ((RREG32(mmMC_SEQ_MISC7) >> 16) & 0xf))
+				dll_state_on = ((RREG32(mmMC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
+			else
+				dll_state_on = ((RREG32(mmMC_SEQ_MISC6) >> 1) & 0x1) ? true : false;
+		} else {
+			dll_state_on = pi->dll_default_on;
+		}
+	} else {
+		memory_level->StrobeRatio = ci_get_ddr3_mclk_frequency_ratio(memory_clock);
+		dll_state_on = ((RREG32(mmMC_SEQ_MISC5) >> 1) & 0x1) ? true : false;
+	}
+
+	ret = ci_calculate_mclk_params(adev, memory_clock, memory_level, memory_level->StrobeEnable, dll_state_on);
+	if (ret)
+		return ret;
+
+	memory_level->MinVddc = cpu_to_be32(memory_level->MinVddc * VOLTAGE_SCALE);
+	memory_level->MinVddcPhases = cpu_to_be32(memory_level->MinVddcPhases);
+	memory_level->MinVddci = cpu_to_be32(memory_level->MinVddci * VOLTAGE_SCALE);
+	memory_level->MinMvdd = cpu_to_be32(memory_level->MinMvdd * VOLTAGE_SCALE);
+
+	memory_level->MclkFrequency = cpu_to_be32(memory_level->MclkFrequency);
+	memory_level->ActivityLevel = cpu_to_be16(memory_level->ActivityLevel);
+	memory_level->MpllFuncCntl = cpu_to_be32(memory_level->MpllFuncCntl);
+	memory_level->MpllFuncCntl_1 = cpu_to_be32(memory_level->MpllFuncCntl_1);
+	memory_level->MpllFuncCntl_2 = cpu_to_be32(memory_level->MpllFuncCntl_2);
+	memory_level->MpllAdFuncCntl = cpu_to_be32(memory_level->MpllAdFuncCntl);
+	memory_level->MpllDqFuncCntl = cpu_to_be32(memory_level->MpllDqFuncCntl);
+	memory_level->MclkPwrmgtCntl = cpu_to_be32(memory_level->MclkPwrmgtCntl);
+	memory_level->DllCntl = cpu_to_be32(memory_level->DllCntl);
+	memory_level->MpllSs1 = cpu_to_be32(memory_level->MpllSs1);
+	memory_level->MpllSs2 = cpu_to_be32(memory_level->MpllSs2);
+
+	return 0;
+}
+
+static int ci_populate_smc_acpi_level(struct amdgpu_device *adev,
+				      SMU7_Discrete_DpmTable *table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	SMU7_Discrete_VoltageLevel voltage_level;
+	u32 spll_func_cntl = pi->clock_registers.cg_spll_func_cntl;
+	u32 spll_func_cntl_2 = pi->clock_registers.cg_spll_func_cntl_2;
+	u32 dll_cntl = pi->clock_registers.dll_cntl;
+	u32 mclk_pwrmgt_cntl = pi->clock_registers.mclk_pwrmgt_cntl;
+	int ret;
+
+	table->ACPILevel.Flags &= ~PPSMC_SWSTATE_FLAG_DC;
+
+	if (pi->acpi_vddc)
+		table->ACPILevel.MinVddc = cpu_to_be32(pi->acpi_vddc * VOLTAGE_SCALE);
+	else
+		table->ACPILevel.MinVddc = cpu_to_be32(pi->min_vddc_in_pp_table * VOLTAGE_SCALE);
+
+	table->ACPILevel.MinVddcPhases = pi->vddc_phase_shed_control ? 0 : 1;
+
+	table->ACPILevel.SclkFrequency = adev->clock.spll.reference_freq;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev,
+						 COMPUTE_GPUCLK_INPUT_FLAG_SCLK,
+						 table->ACPILevel.SclkFrequency, false, &dividers);
+	if (ret)
+		return ret;
+
+	table->ACPILevel.SclkDid = (u8)dividers.post_divider;
+	table->ACPILevel.DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
+	table->ACPILevel.DeepSleepDivId = 0;
+
+	spll_func_cntl &= ~CG_SPLL_FUNC_CNTL__SPLL_PWRON_MASK;
+	spll_func_cntl |= CG_SPLL_FUNC_CNTL__SPLL_RESET_MASK;
+
+	spll_func_cntl_2 &= ~CG_SPLL_FUNC_CNTL_2__SCLK_MUX_SEL_MASK;
+	spll_func_cntl_2 |= (4 << CG_SPLL_FUNC_CNTL_2__SCLK_MUX_SEL__SHIFT);
+
+	table->ACPILevel.CgSpllFuncCntl = spll_func_cntl;
+	table->ACPILevel.CgSpllFuncCntl2 = spll_func_cntl_2;
+	table->ACPILevel.CgSpllFuncCntl3 = pi->clock_registers.cg_spll_func_cntl_3;
+	table->ACPILevel.CgSpllFuncCntl4 = pi->clock_registers.cg_spll_func_cntl_4;
+	table->ACPILevel.SpllSpreadSpectrum = pi->clock_registers.cg_spll_spread_spectrum;
+	table->ACPILevel.SpllSpreadSpectrum2 = pi->clock_registers.cg_spll_spread_spectrum_2;
+	table->ACPILevel.CcPwrDynRm = 0;
+	table->ACPILevel.CcPwrDynRm1 = 0;
+
+	table->ACPILevel.Flags = cpu_to_be32(table->ACPILevel.Flags);
+	table->ACPILevel.MinVddcPhases = cpu_to_be32(table->ACPILevel.MinVddcPhases);
+	table->ACPILevel.SclkFrequency = cpu_to_be32(table->ACPILevel.SclkFrequency);
+	table->ACPILevel.CgSpllFuncCntl = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl);
+	table->ACPILevel.CgSpllFuncCntl2 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl2);
+	table->ACPILevel.CgSpllFuncCntl3 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl3);
+	table->ACPILevel.CgSpllFuncCntl4 = cpu_to_be32(table->ACPILevel.CgSpllFuncCntl4);
+	table->ACPILevel.SpllSpreadSpectrum = cpu_to_be32(table->ACPILevel.SpllSpreadSpectrum);
+	table->ACPILevel.SpllSpreadSpectrum2 = cpu_to_be32(table->ACPILevel.SpllSpreadSpectrum2);
+	table->ACPILevel.CcPwrDynRm = cpu_to_be32(table->ACPILevel.CcPwrDynRm);
+	table->ACPILevel.CcPwrDynRm1 = cpu_to_be32(table->ACPILevel.CcPwrDynRm1);
+
+	table->MemoryACPILevel.MinVddc = table->ACPILevel.MinVddc;
+	table->MemoryACPILevel.MinVddcPhases = table->ACPILevel.MinVddcPhases;
+
+	if (pi->vddci_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
+		if (pi->acpi_vddci)
+			table->MemoryACPILevel.MinVddci =
+				cpu_to_be32(pi->acpi_vddci * VOLTAGE_SCALE);
+		else
+			table->MemoryACPILevel.MinVddci =
+				cpu_to_be32(pi->min_vddci_in_pp_table * VOLTAGE_SCALE);
+	}
+
+	if (ci_populate_mvdd_value(adev, 0, &voltage_level))
+		table->MemoryACPILevel.MinMvdd = 0;
+	else
+		table->MemoryACPILevel.MinMvdd =
+			cpu_to_be32(voltage_level.Voltage * VOLTAGE_SCALE);
+
+	mclk_pwrmgt_cntl |= MCLK_PWRMGT_CNTL__MRDCK0_RESET_MASK |
+		MCLK_PWRMGT_CNTL__MRDCK1_RESET_MASK;
+	mclk_pwrmgt_cntl &= ~(MCLK_PWRMGT_CNTL__MRDCK0_PDNB_MASK |
+			MCLK_PWRMGT_CNTL__MRDCK1_PDNB_MASK);
+
+	dll_cntl &= ~(DLL_CNTL__MRDCK0_BYPASS_MASK | DLL_CNTL__MRDCK1_BYPASS_MASK);
+
+	table->MemoryACPILevel.DllCntl = cpu_to_be32(dll_cntl);
+	table->MemoryACPILevel.MclkPwrmgtCntl = cpu_to_be32(mclk_pwrmgt_cntl);
+	table->MemoryACPILevel.MpllAdFuncCntl =
+		cpu_to_be32(pi->clock_registers.mpll_ad_func_cntl);
+	table->MemoryACPILevel.MpllDqFuncCntl =
+		cpu_to_be32(pi->clock_registers.mpll_dq_func_cntl);
+	table->MemoryACPILevel.MpllFuncCntl =
+		cpu_to_be32(pi->clock_registers.mpll_func_cntl);
+	table->MemoryACPILevel.MpllFuncCntl_1 =
+		cpu_to_be32(pi->clock_registers.mpll_func_cntl_1);
+	table->MemoryACPILevel.MpllFuncCntl_2 =
+		cpu_to_be32(pi->clock_registers.mpll_func_cntl_2);
+	table->MemoryACPILevel.MpllSs1 = cpu_to_be32(pi->clock_registers.mpll_ss1);
+	table->MemoryACPILevel.MpllSs2 = cpu_to_be32(pi->clock_registers.mpll_ss2);
+
+	table->MemoryACPILevel.EnabledForThrottle = 0;
+	table->MemoryACPILevel.EnabledForActivity = 0;
+	table->MemoryACPILevel.UpH = 0;
+	table->MemoryACPILevel.DownH = 100;
+	table->MemoryACPILevel.VoltageDownH = 0;
+	table->MemoryACPILevel.ActivityLevel =
+		cpu_to_be16((u16)pi->mclk_activity_target);
+
+	table->MemoryACPILevel.StutterEnable = false;
+	table->MemoryACPILevel.StrobeEnable = false;
+	table->MemoryACPILevel.EdcReadEnable = false;
+	table->MemoryACPILevel.EdcWriteEnable = false;
+	table->MemoryACPILevel.RttEnable = false;
+
+	return 0;
+}
+
+
+static int ci_enable_ulv(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ulv_parm *ulv = &pi->ulv;
+
+	if (ulv->supported) {
+		if (enable)
+			return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableULV) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+		else
+			return (amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_DisableULV) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+	}
+
+	return 0;
+}
+
+static int ci_populate_ulv_level(struct amdgpu_device *adev,
+				 SMU7_Discrete_Ulv *state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u16 ulv_voltage = adev->pm.dpm.backbias_response_time;
+
+	state->CcPwrDynRm = 0;
+	state->CcPwrDynRm1 = 0;
+
+	if (ulv_voltage == 0) {
+		pi->ulv.supported = false;
+		return 0;
+	}
+
+	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_BY_SVID2) {
+		if (ulv_voltage > adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v)
+			state->VddcOffset = 0;
+		else
+			state->VddcOffset =
+				adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v - ulv_voltage;
+	} else {
+		if (ulv_voltage > adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v)
+			state->VddcOffsetVid = 0;
+		else
+			state->VddcOffsetVid = (u8)
+				((adev->pm.dpm.dyn_state.vddc_dependency_on_sclk.entries[0].v - ulv_voltage) *
+				 VOLTAGE_VID_OFFSET_SCALE2 / VOLTAGE_VID_OFFSET_SCALE1);
+	}
+	state->VddcPhase = pi->vddc_phase_shed_control ? 0 : 1;
+
+	state->CcPwrDynRm = cpu_to_be32(state->CcPwrDynRm);
+	state->CcPwrDynRm1 = cpu_to_be32(state->CcPwrDynRm1);
+	state->VddcOffset = cpu_to_be16(state->VddcOffset);
+
+	return 0;
+}
+
+static int ci_calculate_sclk_params(struct amdgpu_device *adev,
+				    u32 engine_clock,
+				    SMU7_Discrete_GraphicsLevel *sclk)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	u32 spll_func_cntl_3 = pi->clock_registers.cg_spll_func_cntl_3;
+	u32 spll_func_cntl_4 = pi->clock_registers.cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum = pi->clock_registers.cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2 = pi->clock_registers.cg_spll_spread_spectrum_2;
+	u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 reference_divider;
+	u32 fbdiv;
+	int ret;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev,
+						 COMPUTE_GPUCLK_INPUT_FLAG_SCLK,
+						 engine_clock, false, &dividers);
+	if (ret)
+		return ret;
+
+	reference_divider = 1 + dividers.ref_div;
+	fbdiv = dividers.fb_div & 0x3FFFFFF;
+
+	spll_func_cntl_3 &= ~CG_SPLL_FUNC_CNTL_3__SPLL_FB_DIV_MASK;
+	spll_func_cntl_3 |= (fbdiv << CG_SPLL_FUNC_CNTL_3__SPLL_FB_DIV__SHIFT);
+	spll_func_cntl_3 |= CG_SPLL_FUNC_CNTL_3__SPLL_DITHEN_MASK;
+
+	if (pi->caps_sclk_ss_support) {
+		struct amdgpu_atom_ss ss;
+		u32 vco_freq = engine_clock * dividers.post_div;
+
+		if (amdgpu_atombios_get_asic_ss_info(adev, &ss,
+						     ASIC_INTERNAL_ENGINE_SS, vco_freq)) {
+			u32 clk_s = reference_clock * 5 / (reference_divider * ss.rate);
+			u32 clk_v = 4 * ss.percentage * fbdiv / (clk_s * 10000);
+
+			cg_spll_spread_spectrum &= ~(CG_SPLL_SPREAD_SPECTRUM__CLKS_MASK | CG_SPLL_SPREAD_SPECTRUM__SSEN_MASK);
+			cg_spll_spread_spectrum |= (clk_s << CG_SPLL_SPREAD_SPECTRUM__CLKS__SHIFT);
+			cg_spll_spread_spectrum |= (1 << CG_SPLL_SPREAD_SPECTRUM__SSEN__SHIFT);
+
+			cg_spll_spread_spectrum_2 &= ~CG_SPLL_SPREAD_SPECTRUM_2__CLKV_MASK;
+			cg_spll_spread_spectrum_2 |= (clk_v << CG_SPLL_SPREAD_SPECTRUM_2__CLKV__SHIFT);
+		}
+	}
+
+	sclk->SclkFrequency = engine_clock;
+	sclk->CgSpllFuncCntl3 = spll_func_cntl_3;
+	sclk->CgSpllFuncCntl4 = spll_func_cntl_4;
+	sclk->SpllSpreadSpectrum = cg_spll_spread_spectrum;
+	sclk->SpllSpreadSpectrum2  = cg_spll_spread_spectrum_2;
+	sclk->SclkDid = (u8)dividers.post_divider;
+
+	return 0;
+}
+
+static int ci_populate_single_graphic_level(struct amdgpu_device *adev,
+					    u32 engine_clock,
+					    u16 sclk_activity_level_t,
+					    SMU7_Discrete_GraphicsLevel *graphic_level)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	ret = ci_calculate_sclk_params(adev, engine_clock, graphic_level);
+	if (ret)
+		return ret;
+
+	ret = ci_get_dependency_volt_by_clk(adev,
+					    &adev->pm.dpm.dyn_state.vddc_dependency_on_sclk,
+					    engine_clock, &graphic_level->MinVddc);
+	if (ret)
+		return ret;
+
+	graphic_level->SclkFrequency = engine_clock;
+
+	graphic_level->Flags =  0;
+	graphic_level->MinVddcPhases = 1;
+
+	if (pi->vddc_phase_shed_control)
+		ci_populate_phase_value_based_on_sclk(adev,
+						      &adev->pm.dpm.dyn_state.phase_shedding_limits_table,
+						      engine_clock,
+						      &graphic_level->MinVddcPhases);
+
+	graphic_level->ActivityLevel = sclk_activity_level_t;
+
+	graphic_level->CcPwrDynRm = 0;
+	graphic_level->CcPwrDynRm1 = 0;
+	graphic_level->EnabledForThrottle = 1;
+	graphic_level->UpH = 0;
+	graphic_level->DownH = 0;
+	graphic_level->VoltageDownH = 0;
+	graphic_level->PowerThrottle = 0;
+
+	if (pi->caps_sclk_ds)
+		graphic_level->DeepSleepDivId = ci_get_sleep_divider_id_from_clock(adev,
+										   engine_clock,
+										   CISLAND_MINIMUM_ENGINE_CLOCK);
+
+	graphic_level->DisplayWatermark = PPSMC_DISPLAY_WATERMARK_LOW;
+
+	graphic_level->Flags = cpu_to_be32(graphic_level->Flags);
+	graphic_level->MinVddc = cpu_to_be32(graphic_level->MinVddc * VOLTAGE_SCALE);
+	graphic_level->MinVddcPhases = cpu_to_be32(graphic_level->MinVddcPhases);
+	graphic_level->SclkFrequency = cpu_to_be32(graphic_level->SclkFrequency);
+	graphic_level->ActivityLevel = cpu_to_be16(graphic_level->ActivityLevel);
+	graphic_level->CgSpllFuncCntl3 = cpu_to_be32(graphic_level->CgSpllFuncCntl3);
+	graphic_level->CgSpllFuncCntl4 = cpu_to_be32(graphic_level->CgSpllFuncCntl4);
+	graphic_level->SpllSpreadSpectrum = cpu_to_be32(graphic_level->SpllSpreadSpectrum);
+	graphic_level->SpllSpreadSpectrum2 = cpu_to_be32(graphic_level->SpllSpreadSpectrum2);
+	graphic_level->CcPwrDynRm = cpu_to_be32(graphic_level->CcPwrDynRm);
+	graphic_level->CcPwrDynRm1 = cpu_to_be32(graphic_level->CcPwrDynRm1);
+	graphic_level->EnabledForActivity = 1;
+
+	return 0;
+}
+
+static int ci_populate_all_graphic_levels(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	u32 level_array_address = pi->dpm_table_start +
+		offsetof(SMU7_Discrete_DpmTable, GraphicsLevel);
+	u32 level_array_size = sizeof(SMU7_Discrete_GraphicsLevel) *
+		SMU7_MAX_LEVELS_GRAPHICS;
+	SMU7_Discrete_GraphicsLevel *levels = pi->smc_state_table.GraphicsLevel;
+	u32 i, ret;
+
+	memset(levels, 0, level_array_size);
+
+	for (i = 0; i < dpm_table->sclk_table.count; i++) {
+		ret = ci_populate_single_graphic_level(adev,
+						       dpm_table->sclk_table.dpm_levels[i].value,
+						       (u16)pi->activity_target[i],
+						       &pi->smc_state_table.GraphicsLevel[i]);
+		if (ret)
+			return ret;
+		if (i > 1)
+			pi->smc_state_table.GraphicsLevel[i].DeepSleepDivId = 0;
+		if (i == (dpm_table->sclk_table.count - 1))
+			pi->smc_state_table.GraphicsLevel[i].DisplayWatermark =
+				PPSMC_DISPLAY_WATERMARK_HIGH;
+	}
+
+	pi->smc_state_table.GraphicsDpmLevelCount = (u8)dpm_table->sclk_table.count;
+	pi->dpm_level_enable_mask.sclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&dpm_table->sclk_table);
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev, level_array_address,
+				   (u8 *)levels, level_array_size,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int ci_populate_ulv_state(struct amdgpu_device *adev,
+				 SMU7_Discrete_Ulv *ulv_level)
+{
+	return ci_populate_ulv_level(adev, ulv_level);
+}
+
+static int ci_populate_all_memory_levels(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	u32 level_array_address = pi->dpm_table_start +
+		offsetof(SMU7_Discrete_DpmTable, MemoryLevel);
+	u32 level_array_size = sizeof(SMU7_Discrete_MemoryLevel) *
+		SMU7_MAX_LEVELS_MEMORY;
+	SMU7_Discrete_MemoryLevel *levels = pi->smc_state_table.MemoryLevel;
+	u32 i, ret;
+
+	memset(levels, 0, level_array_size);
+
+	for (i = 0; i < dpm_table->mclk_table.count; i++) {
+		if (dpm_table->mclk_table.dpm_levels[i].value == 0)
+			return -EINVAL;
+		ret = ci_populate_single_memory_level(adev,
+						      dpm_table->mclk_table.dpm_levels[i].value,
+						      &pi->smc_state_table.MemoryLevel[i]);
+		if (ret)
+			return ret;
+	}
+
+	if ((dpm_table->mclk_table.count >= 2) &&
+	    ((adev->pdev->device == 0x67B0) || (adev->pdev->device == 0x67B1))) {
+		pi->smc_state_table.MemoryLevel[1].MinVddc =
+			pi->smc_state_table.MemoryLevel[0].MinVddc;
+		pi->smc_state_table.MemoryLevel[1].MinVddcPhases =
+			pi->smc_state_table.MemoryLevel[0].MinVddcPhases;
+	}
+
+	pi->smc_state_table.MemoryLevel[0].ActivityLevel = cpu_to_be16(0x1F);
+
+	pi->smc_state_table.MemoryDpmLevelCount = (u8)dpm_table->mclk_table.count;
+	pi->dpm_level_enable_mask.mclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&dpm_table->mclk_table);
+
+	pi->smc_state_table.MemoryLevel[dpm_table->mclk_table.count - 1].DisplayWatermark =
+		PPSMC_DISPLAY_WATERMARK_HIGH;
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev, level_array_address,
+				   (u8 *)levels, level_array_size,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void ci_reset_single_dpm_table(struct amdgpu_device *adev,
+				      struct ci_single_dpm_table* dpm_table,
+				      u32 count)
+{
+	u32 i;
+
+	dpm_table->count = count;
+	for (i = 0; i < MAX_REGULAR_DPM_NUMBER; i++)
+		dpm_table->dpm_levels[i].enabled = false;
+}
+
+static void ci_setup_pcie_table_entry(struct ci_single_dpm_table* dpm_table,
+				      u32 index, u32 pcie_gen, u32 pcie_lanes)
+{
+	dpm_table->dpm_levels[index].value = pcie_gen;
+	dpm_table->dpm_levels[index].param1 = pcie_lanes;
+	dpm_table->dpm_levels[index].enabled = true;
+}
+
+static int ci_setup_default_pcie_tables(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!pi->use_pcie_performance_levels && !pi->use_pcie_powersaving_levels)
+		return -EINVAL;
+
+	if (pi->use_pcie_performance_levels && !pi->use_pcie_powersaving_levels) {
+		pi->pcie_gen_powersaving = pi->pcie_gen_performance;
+		pi->pcie_lane_powersaving = pi->pcie_lane_performance;
+	} else if (!pi->use_pcie_performance_levels && pi->use_pcie_powersaving_levels) {
+		pi->pcie_gen_performance = pi->pcie_gen_powersaving;
+		pi->pcie_lane_performance = pi->pcie_lane_powersaving;
+	}
+
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.pcie_speed_table,
+				  SMU7_MAX_LEVELS_LINK);
+
+	if (adev->asic_type == CHIP_BONAIRE)
+		ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0,
+					  pi->pcie_gen_powersaving.min,
+					  pi->pcie_lane_powersaving.max);
+	else
+		ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0,
+					  pi->pcie_gen_powersaving.min,
+					  pi->pcie_lane_powersaving.min);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 1,
+				  pi->pcie_gen_performance.min,
+				  pi->pcie_lane_performance.min);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 2,
+				  pi->pcie_gen_powersaving.min,
+				  pi->pcie_lane_powersaving.max);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 3,
+				  pi->pcie_gen_performance.min,
+				  pi->pcie_lane_performance.max);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 4,
+				  pi->pcie_gen_powersaving.max,
+				  pi->pcie_lane_powersaving.max);
+	ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 5,
+				  pi->pcie_gen_performance.max,
+				  pi->pcie_lane_performance.max);
+
+	pi->dpm_table.pcie_speed_table.count = 6;
+
+	return 0;
+}
+
+static int ci_setup_default_dpm_tables(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *allowed_sclk_vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk;
+	struct amdgpu_cac_leakage_table *std_voltage_table =
+		&adev->pm.dpm.dyn_state.cac_leakage_table;
+	u32 i;
+
+	if (allowed_sclk_vddc_table == NULL)
+		return -EINVAL;
+	if (allowed_sclk_vddc_table->count < 1)
+		return -EINVAL;
+	if (allowed_mclk_table == NULL)
+		return -EINVAL;
+	if (allowed_mclk_table->count < 1)
+		return -EINVAL;
+
+	memset(&pi->dpm_table, 0, sizeof(struct ci_dpm_table));
+
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.sclk_table,
+				  SMU7_MAX_LEVELS_GRAPHICS);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.mclk_table,
+				  SMU7_MAX_LEVELS_MEMORY);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.vddc_table,
+				  SMU7_MAX_LEVELS_VDDC);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.vddci_table,
+				  SMU7_MAX_LEVELS_VDDCI);
+	ci_reset_single_dpm_table(adev,
+				  &pi->dpm_table.mvdd_table,
+				  SMU7_MAX_LEVELS_MVDD);
+
+	pi->dpm_table.sclk_table.count = 0;
+	for (i = 0; i < allowed_sclk_vddc_table->count; i++) {
+		if ((i == 0) ||
+		    (pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count-1].value !=
+		     allowed_sclk_vddc_table->entries[i].clk)) {
+			pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].value =
+				allowed_sclk_vddc_table->entries[i].clk;
+			pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].enabled =
+				(i == 0) ? true : false;
+			pi->dpm_table.sclk_table.count++;
+		}
+	}
+
+	pi->dpm_table.mclk_table.count = 0;
+	for (i = 0; i < allowed_mclk_table->count; i++) {
+		if ((i == 0) ||
+		    (pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count-1].value !=
+		     allowed_mclk_table->entries[i].clk)) {
+			pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].value =
+				allowed_mclk_table->entries[i].clk;
+			pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].enabled =
+				(i == 0) ? true : false;
+			pi->dpm_table.mclk_table.count++;
+		}
+	}
+
+	for (i = 0; i < allowed_sclk_vddc_table->count; i++) {
+		pi->dpm_table.vddc_table.dpm_levels[i].value =
+			allowed_sclk_vddc_table->entries[i].v;
+		pi->dpm_table.vddc_table.dpm_levels[i].param1 =
+			std_voltage_table->entries[i].leakage;
+		pi->dpm_table.vddc_table.dpm_levels[i].enabled = true;
+	}
+	pi->dpm_table.vddc_table.count = allowed_sclk_vddc_table->count;
+
+	allowed_mclk_table = &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk;
+	if (allowed_mclk_table) {
+		for (i = 0; i < allowed_mclk_table->count; i++) {
+			pi->dpm_table.vddci_table.dpm_levels[i].value =
+				allowed_mclk_table->entries[i].v;
+			pi->dpm_table.vddci_table.dpm_levels[i].enabled = true;
+		}
+		pi->dpm_table.vddci_table.count = allowed_mclk_table->count;
+	}
+
+	allowed_mclk_table = &adev->pm.dpm.dyn_state.mvdd_dependency_on_mclk;
+	if (allowed_mclk_table) {
+		for (i = 0; i < allowed_mclk_table->count; i++) {
+			pi->dpm_table.mvdd_table.dpm_levels[i].value =
+				allowed_mclk_table->entries[i].v;
+			pi->dpm_table.mvdd_table.dpm_levels[i].enabled = true;
+		}
+		pi->dpm_table.mvdd_table.count = allowed_mclk_table->count;
+	}
+
+	ci_setup_default_pcie_tables(adev);
+
+	return 0;
+}
+
+static int ci_find_boot_level(struct ci_single_dpm_table *table,
+			      u32 value, u32 *boot_level)
+{
+	u32 i;
+	int ret = -EINVAL;
+
+	for(i = 0; i < table->count; i++) {
+		if (value == table->dpm_levels[i].value) {
+			*boot_level = i;
+			ret = 0;
+		}
+	}
+
+	return ret;
+}
+
+static int ci_init_smc_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ulv_parm *ulv = &pi->ulv;
+	struct amdgpu_ps *amdgpu_boot_state = adev->pm.dpm.boot_ps;
+	SMU7_Discrete_DpmTable *table = &pi->smc_state_table;
+	int ret;
+
+	ret = ci_setup_default_dpm_tables(adev);
+	if (ret)
+		return ret;
+
+	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_NONE)
+		ci_populate_smc_voltage_tables(adev, table);
+
+	ci_init_fps_limits(adev);
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_HARDWAREDC)
+		table->SystemFlags |= PPSMC_SYSTEMFLAG_GPIO_DC;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_STEPVDDC)
+		table->SystemFlags |= PPSMC_SYSTEMFLAG_STEPVDDC;
+
+	if (adev->mc.vram_type == AMDGPU_VRAM_TYPE_GDDR5)
+		table->SystemFlags |= PPSMC_SYSTEMFLAG_GDDR5;
+
+	if (ulv->supported) {
+		ret = ci_populate_ulv_state(adev, &pi->smc_state_table.Ulv);
+		if (ret)
+			return ret;
+		WREG32_SMC(ixCG_ULV_PARAMETER, ulv->cg_ulv_parameter);
+	}
+
+	ret = ci_populate_all_graphic_levels(adev);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_all_memory_levels(adev);
+	if (ret)
+		return ret;
+
+	ci_populate_smc_link_level(adev, table);
+
+	ret = ci_populate_smc_acpi_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_vce_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_acp_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_samu_level(adev, table);
+	if (ret)
+		return ret;
+
+	ret = ci_do_program_memory_timing_parameters(adev);
+	if (ret)
+		return ret;
+
+	ret = ci_populate_smc_uvd_level(adev, table);
+	if (ret)
+		return ret;
+
+	table->UvdBootLevel  = 0;
+	table->VceBootLevel  = 0;
+	table->AcpBootLevel  = 0;
+	table->SamuBootLevel  = 0;
+	table->GraphicsBootLevel  = 0;
+	table->MemoryBootLevel  = 0;
+
+	ret = ci_find_boot_level(&pi->dpm_table.sclk_table,
+				 pi->vbios_boot_state.sclk_bootup_value,
+				 (u32 *)&pi->smc_state_table.GraphicsBootLevel);
+
+	ret = ci_find_boot_level(&pi->dpm_table.mclk_table,
+				 pi->vbios_boot_state.mclk_bootup_value,
+				 (u32 *)&pi->smc_state_table.MemoryBootLevel);
+
+	table->BootVddc = pi->vbios_boot_state.vddc_bootup_value;
+	table->BootVddci = pi->vbios_boot_state.vddci_bootup_value;
+	table->BootMVdd = pi->vbios_boot_state.mvdd_bootup_value;
+
+	ci_populate_smc_initial_state(adev, amdgpu_boot_state);
+
+	ret = ci_populate_bapm_parameters_in_dpm_table(adev);
+	if (ret)
+		return ret;
+
+	table->UVDInterval = 1;
+	table->VCEInterval = 1;
+	table->ACPInterval = 1;
+	table->SAMUInterval = 1;
+	table->GraphicsVoltageChangeEnable = 1;
+	table->GraphicsThermThrottleEnable = 1;
+	table->GraphicsInterval = 1;
+	table->VoltageInterval = 1;
+	table->ThermalInterval = 1;
+	table->TemperatureLimitHigh = (u16)((pi->thermal_temp_setting.temperature_high *
+					     CISLANDS_Q88_FORMAT_CONVERSION_UNIT) / 1000);
+	table->TemperatureLimitLow = (u16)((pi->thermal_temp_setting.temperature_low *
+					    CISLANDS_Q88_FORMAT_CONVERSION_UNIT) / 1000);
+	table->MemoryVoltageChangeEnable = 1;
+	table->MemoryInterval = 1;
+	table->VoltageResponseTime = 0;
+	table->VddcVddciDelta = 4000;
+	table->PhaseResponseTime = 0;
+	table->MemoryThermThrottleEnable = 1;
+	table->PCIeBootLinkLevel = pi->dpm_table.pcie_speed_table.count - 1;
+	table->PCIeGenInterval = 1;
+	if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2)
+		table->SVI2Enable  = 1;
+	else
+		table->SVI2Enable  = 0;
+
+	table->ThermGpio = 17;
+	table->SclkStepSize = 0x4000;
+
+	table->SystemFlags = cpu_to_be32(table->SystemFlags);
+	table->SmioMaskVddcVid = cpu_to_be32(table->SmioMaskVddcVid);
+	table->SmioMaskVddcPhase = cpu_to_be32(table->SmioMaskVddcPhase);
+	table->SmioMaskVddciVid = cpu_to_be32(table->SmioMaskVddciVid);
+	table->SmioMaskMvddVid = cpu_to_be32(table->SmioMaskMvddVid);
+	table->SclkStepSize = cpu_to_be32(table->SclkStepSize);
+	table->TemperatureLimitHigh = cpu_to_be16(table->TemperatureLimitHigh);
+	table->TemperatureLimitLow = cpu_to_be16(table->TemperatureLimitLow);
+	table->VddcVddciDelta = cpu_to_be16(table->VddcVddciDelta);
+	table->VoltageResponseTime = cpu_to_be16(table->VoltageResponseTime);
+	table->PhaseResponseTime = cpu_to_be16(table->PhaseResponseTime);
+	table->BootVddc = cpu_to_be16(table->BootVddc * VOLTAGE_SCALE);
+	table->BootVddci = cpu_to_be16(table->BootVddci * VOLTAGE_SCALE);
+	table->BootMVdd = cpu_to_be16(table->BootMVdd * VOLTAGE_SCALE);
+
+	ret = amdgpu_ci_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Discrete_DpmTable, SystemFlags),
+				   (u8 *)&table->SystemFlags,
+				   sizeof(SMU7_Discrete_DpmTable) - 3 * sizeof(SMU7_PIDController),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static void ci_trim_single_dpm_states(struct amdgpu_device *adev,
+				      struct ci_single_dpm_table *dpm_table,
+				      u32 low_limit, u32 high_limit)
+{
+	u32 i;
+
+	for (i = 0; i < dpm_table->count; i++) {
+		if ((dpm_table->dpm_levels[i].value < low_limit) ||
+		    (dpm_table->dpm_levels[i].value > high_limit))
+			dpm_table->dpm_levels[i].enabled = false;
+		else
+			dpm_table->dpm_levels[i].enabled = true;
+	}
+}
+
+static void ci_trim_pcie_dpm_states(struct amdgpu_device *adev,
+				    u32 speed_low, u32 lanes_low,
+				    u32 speed_high, u32 lanes_high)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *pcie_table = &pi->dpm_table.pcie_speed_table;
+	u32 i, j;
+
+	for (i = 0; i < pcie_table->count; i++) {
+		if ((pcie_table->dpm_levels[i].value < speed_low) ||
+		    (pcie_table->dpm_levels[i].param1 < lanes_low) ||
+		    (pcie_table->dpm_levels[i].value > speed_high) ||
+		    (pcie_table->dpm_levels[i].param1 > lanes_high))
+			pcie_table->dpm_levels[i].enabled = false;
+		else
+			pcie_table->dpm_levels[i].enabled = true;
+	}
+
+	for (i = 0; i < pcie_table->count; i++) {
+		if (pcie_table->dpm_levels[i].enabled) {
+			for (j = i + 1; j < pcie_table->count; j++) {
+				if (pcie_table->dpm_levels[j].enabled) {
+					if ((pcie_table->dpm_levels[i].value == pcie_table->dpm_levels[j].value) &&
+					    (pcie_table->dpm_levels[i].param1 == pcie_table->dpm_levels[j].param1))
+						pcie_table->dpm_levels[j].enabled = false;
+				}
+			}
+		}
+	}
+}
+
+static int ci_trim_dpm_states(struct amdgpu_device *adev,
+			      struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 high_limit_count;
+
+	if (state->performance_level_count < 1)
+		return -EINVAL;
+
+	if (state->performance_level_count == 1)
+		high_limit_count = 0;
+	else
+		high_limit_count = 1;
+
+	ci_trim_single_dpm_states(adev,
+				  &pi->dpm_table.sclk_table,
+				  state->performance_levels[0].sclk,
+				  state->performance_levels[high_limit_count].sclk);
+
+	ci_trim_single_dpm_states(adev,
+				  &pi->dpm_table.mclk_table,
+				  state->performance_levels[0].mclk,
+				  state->performance_levels[high_limit_count].mclk);
+
+	ci_trim_pcie_dpm_states(adev,
+				state->performance_levels[0].pcie_gen,
+				state->performance_levels[0].pcie_lane,
+				state->performance_levels[high_limit_count].pcie_gen,
+				state->performance_levels[high_limit_count].pcie_lane);
+
+	return 0;
+}
+
+static int ci_apply_disp_minimum_voltage_request(struct amdgpu_device *adev)
+{
+	struct amdgpu_clock_voltage_dependency_table *disp_voltage_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk;
+	struct amdgpu_clock_voltage_dependency_table *vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 requested_voltage = 0;
+	u32 i;
+
+	if (disp_voltage_table == NULL)
+		return -EINVAL;
+	if (!disp_voltage_table->count)
+		return -EINVAL;
+
+	for (i = 0; i < disp_voltage_table->count; i++) {
+		if (adev->clock.current_dispclk == disp_voltage_table->entries[i].clk)
+			requested_voltage = disp_voltage_table->entries[i].v;
+	}
+
+	for (i = 0; i < vddc_table->count; i++) {
+		if (requested_voltage <= vddc_table->entries[i].v) {
+			requested_voltage = vddc_table->entries[i].v;
+			return (amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								  PPSMC_MSG_VddC_Request,
+								  requested_voltage * VOLTAGE_SCALE) == PPSMC_Result_OK) ?
+				0 : -EINVAL;
+		}
+	}
+
+	return -EINVAL;
+}
+
+static int ci_upload_dpm_level_enable_mask(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	PPSMC_Result result;
+
+	ci_apply_disp_minimum_voltage_request(adev);
+
+	if (!pi->sclk_dpm_key_disabled) {
+		if (pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
+			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								   PPSMC_MSG_SCLKDPM_SetEnabledMask,
+								   pi->dpm_level_enable_mask.sclk_dpm_enable_mask);
+			if (result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+	if (!pi->mclk_dpm_key_disabled) {
+		if (pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
+			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								   PPSMC_MSG_MCLKDPM_SetEnabledMask,
+								   pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+			if (result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+
+#if 0
+	if (!pi->pcie_dpm_key_disabled) {
+		if (pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			result = amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+								   PPSMC_MSG_PCIeDPM_SetEnabledMask,
+								   pi->dpm_level_enable_mask.pcie_dpm_enable_mask);
+			if (result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+	}
+#endif
+
+	return 0;
+}
+
+static void ci_find_dpm_states_clocks_in_dpm_table(struct amdgpu_device *adev,
+						   struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	struct ci_single_dpm_table *sclk_table = &pi->dpm_table.sclk_table;
+	u32 sclk = state->performance_levels[state->performance_level_count-1].sclk;
+	struct ci_single_dpm_table *mclk_table = &pi->dpm_table.mclk_table;
+	u32 mclk = state->performance_levels[state->performance_level_count-1].mclk;
+	u32 i;
+
+	pi->need_update_smu7_dpm_table = 0;
+
+	for (i = 0; i < sclk_table->count; i++) {
+		if (sclk == sclk_table->dpm_levels[i].value)
+			break;
+	}
+
+	if (i >= sclk_table->count) {
+		pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_SCLK;
+	} else {
+		/* XXX check display min clock requirements */
+		if (CISLAND_MINIMUM_ENGINE_CLOCK != CISLAND_MINIMUM_ENGINE_CLOCK)
+			pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_SCLK;
+	}
+
+	for (i = 0; i < mclk_table->count; i++) {
+		if (mclk == mclk_table->dpm_levels[i].value)
+			break;
+	}
+
+	if (i >= mclk_table->count)
+		pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_MCLK;
+
+	if (adev->pm.dpm.current_active_crtc_count !=
+	    adev->pm.dpm.new_active_crtc_count)
+		pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_MCLK;
+}
+
+static int ci_populate_and_upload_sclk_mclk_dpm_levels(struct amdgpu_device *adev,
+						       struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	u32 sclk = state->performance_levels[state->performance_level_count-1].sclk;
+	u32 mclk = state->performance_levels[state->performance_level_count-1].mclk;
+	struct ci_dpm_table *dpm_table = &pi->dpm_table;
+	int ret;
+
+	if (!pi->need_update_smu7_dpm_table)
+		return 0;
+
+	if (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_SCLK)
+		dpm_table->sclk_table.dpm_levels[dpm_table->sclk_table.count-1].value = sclk;
+
+	if (pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK)
+		dpm_table->mclk_table.dpm_levels[dpm_table->mclk_table.count-1].value = mclk;
+
+	if (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_SCLK | DPMTABLE_UPDATE_SCLK)) {
+		ret = ci_populate_all_graphic_levels(adev);
+		if (ret)
+			return ret;
+	}
+
+	if (pi->need_update_smu7_dpm_table & (DPMTABLE_OD_UPDATE_MCLK | DPMTABLE_UPDATE_MCLK)) {
+		ret = ci_populate_all_memory_levels(adev);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int ci_enable_uvd_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.dpm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.uvd_dpm_enable_mask = 0;
+
+		for (i = adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.uvd_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_uvd_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_UVDDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.uvd_dpm_enable_mask);
+
+		if (pi->last_mclk_dpm_enable_mask & 0x1) {
+			pi->uvd_enabled = true;
+			pi->dpm_level_enable_mask.mclk_dpm_enable_mask &= 0xFFFFFFFE;
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_MCLKDPM_SetEnabledMask,
+							  pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+		}
+	} else {
+		if (pi->last_mclk_dpm_enable_mask & 0x1) {
+			pi->uvd_enabled = false;
+			pi->dpm_level_enable_mask.mclk_dpm_enable_mask |= 1;
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_MCLKDPM_SetEnabledMask,
+							  pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+		}
+	}
+
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_UVDDPM_Enable : PPSMC_MSG_UVDDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int ci_enable_vce_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.dpm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.vce_dpm_enable_mask = 0;
+		for (i = adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.vce_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_vce_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_VCEDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.vce_dpm_enable_mask);
+	}
+
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_VCEDPM_Enable : PPSMC_MSG_VCEDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+#if 0
+static int ci_enable_samu_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.dpm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.samu_dpm_enable_mask = 0;
+		for (i = adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.samu_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_samu_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_SAMUDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.samu_dpm_enable_mask);
+	}
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_SAMUDPM_Enable : PPSMC_MSG_SAMUDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+
+static int ci_enable_acp_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	const struct amdgpu_clock_and_voltage_limits *max_limits;
+	int i;
+
+	if (adev->pm.dpm.ac_power)
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	else
+		max_limits = &adev->pm.dpm.dyn_state.max_clock_voltage_on_dc;
+
+	if (enable) {
+		pi->dpm_level_enable_mask.acp_dpm_enable_mask = 0;
+		for (i = adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.count - 1; i >= 0; i--) {
+			if (adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table.entries[i].v <= max_limits->vddc) {
+				pi->dpm_level_enable_mask.acp_dpm_enable_mask |= 1 << i;
+
+				if (!pi->caps_acp_dpm)
+					break;
+			}
+		}
+
+		amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_ACPDPM_SetEnabledMask,
+						  pi->dpm_level_enable_mask.acp_dpm_enable_mask);
+	}
+
+	return (amdgpu_ci_send_msg_to_smc(adev, enable ?
+				   PPSMC_MSG_ACPDPM_Enable : PPSMC_MSG_ACPDPM_Disable) == PPSMC_Result_OK) ?
+		0 : -EINVAL;
+}
+#endif
+
+static int ci_update_uvd_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (!gate) {
+		if (pi->caps_uvd_dpm ||
+		    (adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count <= 0))
+			pi->smc_state_table.UvdBootLevel = 0;
+		else
+			pi->smc_state_table.UvdBootLevel =
+				adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table.count - 1;
+
+		tmp = RREG32_SMC(ixDPM_TABLE_475);
+		tmp &= ~DPM_TABLE_475__UvdBootLevel_MASK;
+		tmp |= (pi->smc_state_table.UvdBootLevel << DPM_TABLE_475__UvdBootLevel__SHIFT);
+		WREG32_SMC(ixDPM_TABLE_475, tmp);
+	}
+
+	return ci_enable_uvd_dpm(adev, !gate);
+}
+
+static u8 ci_get_vce_boot_level(struct amdgpu_device *adev)
+{
+	u8 i;
+	u32 min_evclk = 30000; /* ??? */
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].evclk >= min_evclk)
+			return i;
+	}
+
+	return table->count - 1;
+}
+
+static int ci_update_vce_dpm(struct amdgpu_device *adev,
+			     struct amdgpu_ps *amdgpu_new_state,
+			     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret = 0;
+	u32 tmp;
+
+	if (amdgpu_current_state->evclk != amdgpu_new_state->evclk) {
+		if (amdgpu_new_state->evclk) {
+			/* turn the clocks on when encoding */
+			ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_CG_STATE_UNGATE);
+			if (ret)
+				return ret;
+
+			pi->smc_state_table.VceBootLevel = ci_get_vce_boot_level(adev);
+			tmp = RREG32_SMC(ixDPM_TABLE_475);
+			tmp &= ~DPM_TABLE_475__VceBootLevel_MASK;
+			tmp |= (pi->smc_state_table.VceBootLevel << DPM_TABLE_475__VceBootLevel__SHIFT);
+			WREG32_SMC(ixDPM_TABLE_475, tmp);
+
+			ret = ci_enable_vce_dpm(adev, true);
+		} else {
+			/* turn the clocks off when not encoding */
+			ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_CG_STATE_GATE);
+			if (ret)
+				return ret;
+
+			ret = ci_enable_vce_dpm(adev, false);
+		}
+	}
+	return ret;
+}
+
+#if 0
+static int ci_update_samu_dpm(struct amdgpu_device *adev, bool gate)
+{
+	return ci_enable_samu_dpm(adev, gate);
+}
+
+static int ci_update_acp_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp;
+
+	if (!gate) {
+		pi->smc_state_table.AcpBootLevel = 0;
+
+		tmp = RREG32_SMC(ixDPM_TABLE_475);
+		tmp &= ~AcpBootLevel_MASK;
+		tmp |= AcpBootLevel(pi->smc_state_table.AcpBootLevel);
+		WREG32_SMC(ixDPM_TABLE_475, tmp);
+	}
+
+	return ci_enable_acp_dpm(adev, !gate);
+}
+#endif
+
+static int ci_generate_dpm_level_enable_mask(struct amdgpu_device *adev,
+					     struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	ret = ci_trim_dpm_states(adev, amdgpu_state);
+	if (ret)
+		return ret;
+
+	pi->dpm_level_enable_mask.sclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.sclk_table);
+	pi->dpm_level_enable_mask.mclk_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.mclk_table);
+	pi->last_mclk_dpm_enable_mask =
+		pi->dpm_level_enable_mask.mclk_dpm_enable_mask;
+	if (pi->uvd_enabled) {
+		if (pi->dpm_level_enable_mask.mclk_dpm_enable_mask & 1)
+			pi->dpm_level_enable_mask.mclk_dpm_enable_mask &= 0xFFFFFFFE;
+	}
+	pi->dpm_level_enable_mask.pcie_dpm_enable_mask =
+		ci_get_dpm_level_enable_mask_value(&pi->dpm_table.pcie_speed_table);
+
+	return 0;
+}
+
+static u32 ci_get_lowest_enabled_level(struct amdgpu_device *adev,
+				       u32 level_mask)
+{
+	u32 level = 0;
+
+	while ((level_mask & (1 << level)) == 0)
+		level++;
+
+	return level;
+}
+
+
+static int ci_dpm_force_performance_level(struct amdgpu_device *adev,
+					  enum amdgpu_dpm_forced_level level)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 tmp, levels, i;
+	int ret;
+
+	if (level == AMDGPU_DPM_FORCED_LEVEL_HIGH) {
+		if ((!pi->pcie_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_pcie(adev, level);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
+					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+		if ((!pi->sclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.sclk_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_sclk(adev, levels);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+		if ((!pi->mclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.mclk_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_mclk(adev, levels);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+		if ((!pi->pcie_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			levels = 0;
+			tmp = pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
+			while (tmp >>= 1)
+				levels++;
+			if (levels) {
+				ret = ci_dpm_force_state_pcie(adev, level);
+				if (ret)
+					return ret;
+				for (i = 0; i < adev->usec_timeout; i++) {
+					tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
+					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
+					TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
+					if (tmp == levels)
+						break;
+					udelay(1);
+				}
+			}
+		}
+	} else if (level == AMDGPU_DPM_FORCED_LEVEL_LOW) {
+		if ((!pi->sclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.sclk_dpm_enable_mask) {
+			levels = ci_get_lowest_enabled_level(adev,
+							     pi->dpm_level_enable_mask.sclk_dpm_enable_mask);
+			ret = ci_dpm_force_state_sclk(adev, levels);
+			if (ret)
+				return ret;
+			for (i = 0; i < adev->usec_timeout; i++) {
+				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+				if (tmp == levels)
+					break;
+				udelay(1);
+			}
+		}
+		if ((!pi->mclk_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.mclk_dpm_enable_mask) {
+			levels = ci_get_lowest_enabled_level(adev,
+							     pi->dpm_level_enable_mask.mclk_dpm_enable_mask);
+			ret = ci_dpm_force_state_mclk(adev, levels);
+			if (ret)
+				return ret;
+			for (i = 0; i < adev->usec_timeout; i++) {
+				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX_MASK) >>
+				TARGET_AND_CURRENT_PROFILE_INDEX__CURR_MCLK_INDEX__SHIFT;
+				if (tmp == levels)
+					break;
+				udelay(1);
+			}
+		}
+		if ((!pi->pcie_dpm_key_disabled) &&
+		    pi->dpm_level_enable_mask.pcie_dpm_enable_mask) {
+			levels = ci_get_lowest_enabled_level(adev,
+							     pi->dpm_level_enable_mask.pcie_dpm_enable_mask);
+			ret = ci_dpm_force_state_pcie(adev, levels);
+			if (ret)
+				return ret;
+			for (i = 0; i < adev->usec_timeout; i++) {
+				tmp = (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_1) &
+				TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX_MASK) >>
+				TARGET_AND_CURRENT_PROFILE_INDEX_1__CURR_PCIE_INDEX__SHIFT;
+				if (tmp == levels)
+					break;
+				udelay(1);
+			}
+		}
+	} else if (level == AMDGPU_DPM_FORCED_LEVEL_AUTO) {
+		if (!pi->pcie_dpm_key_disabled) {
+			PPSMC_Result smc_result;
+
+			smc_result = amdgpu_ci_send_msg_to_smc(adev,
+							       PPSMC_MSG_PCIeDPM_UnForceLevel);
+			if (smc_result != PPSMC_Result_OK)
+				return -EINVAL;
+		}
+		ret = ci_upload_dpm_level_enable_mask(adev);
+		if (ret)
+			return ret;
+	}
+
+	adev->pm.dpm.forced_level = level;
+
+	return 0;
+}
+
+static int ci_set_mc_special_registers(struct amdgpu_device *adev,
+				       struct ci_mc_reg_table *table)
+{
+	u8 i, j, k;
+	u32 temp_reg;
+
+	for (i = 0, j = table->last; i < table->last; i++) {
+		if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+			return -EINVAL;
+		switch(table->mc_reg_address[i].s1) {
+		case mmMC_SEQ_MISC1:
+			temp_reg = RREG32(mmMC_PMG_CMD_EMRS);
+			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_EMRS;
+			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_EMRS_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					((temp_reg & 0xffff0000)) | ((table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16);
+			}
+			j++;
+			if (j >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+
+			temp_reg = RREG32(mmMC_PMG_CMD_MRS);
+			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_MRS;
+			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_MRS_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					(temp_reg & 0xffff0000) | (table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
+				if (adev->mc.vram_type != AMDGPU_VRAM_TYPE_GDDR5)
+					table->mc_reg_table_entry[k].mc_data[j] |= 0x100;
+			}
+			j++;
+			if (j > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+
+			if (adev->mc.vram_type != AMDGPU_VRAM_TYPE_GDDR5) {
+				table->mc_reg_address[j].s1 = mmMC_PMG_AUTO_CMD;
+				table->mc_reg_address[j].s0 = mmMC_PMG_AUTO_CMD;
+				for (k = 0; k < table->num_entries; k++) {
+					table->mc_reg_table_entry[k].mc_data[j] =
+						(table->mc_reg_table_entry[k].mc_data[i] & 0xffff0000) >> 16;
+				}
+				j++;
+				if (j > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+					return -EINVAL;
+			}
+			break;
+		case mmMC_SEQ_RESERVE_M:
+			temp_reg = RREG32(mmMC_PMG_CMD_MRS1);
+			table->mc_reg_address[j].s1 = mmMC_PMG_CMD_MRS1;
+			table->mc_reg_address[j].s0 = mmMC_SEQ_PMG_CMD_MRS1_LP;
+			for (k = 0; k < table->num_entries; k++) {
+				table->mc_reg_table_entry[k].mc_data[j] =
+					(temp_reg & 0xffff0000) | (table->mc_reg_table_entry[k].mc_data[i] & 0x0000ffff);
+			}
+			j++;
+			if (j > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			break;
+		default:
+			break;
+		}
+
+	}
+
+	table->last = j;
+
+	return 0;
+}
+
+static bool ci_check_s0_mc_reg_index(u16 in_reg, u16 *out_reg)
+{
+	bool result = true;
+
+	switch(in_reg) {
+	case mmMC_SEQ_RAS_TIMING:
+		*out_reg = mmMC_SEQ_RAS_TIMING_LP;
+		break;
+	case mmMC_SEQ_DLL_STBY:
+		*out_reg = mmMC_SEQ_DLL_STBY_LP;
+		break;
+	case mmMC_SEQ_G5PDX_CMD0:
+		*out_reg = mmMC_SEQ_G5PDX_CMD0_LP;
+		break;
+	case mmMC_SEQ_G5PDX_CMD1:
+		*out_reg = mmMC_SEQ_G5PDX_CMD1_LP;
+		break;
+	case mmMC_SEQ_G5PDX_CTRL:
+		*out_reg = mmMC_SEQ_G5PDX_CTRL_LP;
+		break;
+	case mmMC_SEQ_CAS_TIMING:
+		*out_reg = mmMC_SEQ_CAS_TIMING_LP;
+	    break;
+	case mmMC_SEQ_MISC_TIMING:
+		*out_reg = mmMC_SEQ_MISC_TIMING_LP;
+		break;
+	case mmMC_SEQ_MISC_TIMING2:
+		*out_reg = mmMC_SEQ_MISC_TIMING2_LP;
+		break;
+	case mmMC_SEQ_PMG_DVS_CMD:
+		*out_reg = mmMC_SEQ_PMG_DVS_CMD_LP;
+		break;
+	case mmMC_SEQ_PMG_DVS_CTL:
+		*out_reg = mmMC_SEQ_PMG_DVS_CTL_LP;
+		break;
+	case mmMC_SEQ_RD_CTL_D0:
+		*out_reg = mmMC_SEQ_RD_CTL_D0_LP;
+		break;
+	case mmMC_SEQ_RD_CTL_D1:
+		*out_reg = mmMC_SEQ_RD_CTL_D1_LP;
+		break;
+	case mmMC_SEQ_WR_CTL_D0:
+		*out_reg = mmMC_SEQ_WR_CTL_D0_LP;
+		break;
+	case mmMC_SEQ_WR_CTL_D1:
+		*out_reg = mmMC_SEQ_WR_CTL_D1_LP;
+		break;
+	case mmMC_PMG_CMD_EMRS:
+		*out_reg = mmMC_SEQ_PMG_CMD_EMRS_LP;
+		break;
+	case mmMC_PMG_CMD_MRS:
+		*out_reg = mmMC_SEQ_PMG_CMD_MRS_LP;
+		break;
+	case mmMC_PMG_CMD_MRS1:
+		*out_reg = mmMC_SEQ_PMG_CMD_MRS1_LP;
+		break;
+	case mmMC_SEQ_PMG_TIMING:
+		*out_reg = mmMC_SEQ_PMG_TIMING_LP;
+		break;
+	case mmMC_PMG_CMD_MRS2:
+		*out_reg = mmMC_SEQ_PMG_CMD_MRS2_LP;
+		break;
+	case mmMC_SEQ_WR_CTL_2:
+		*out_reg = mmMC_SEQ_WR_CTL_2_LP;
+		break;
+	default:
+		result = false;
+		break;
+	}
+
+	return result;
+}
+
+static void ci_set_valid_flag(struct ci_mc_reg_table *table)
+{
+	u8 i, j;
+
+	for (i = 0; i < table->last; i++) {
+		for (j = 1; j < table->num_entries; j++) {
+			if (table->mc_reg_table_entry[j-1].mc_data[i] !=
+			    table->mc_reg_table_entry[j].mc_data[i]) {
+				table->valid_flag |= 1 << i;
+				break;
+			}
+		}
+	}
+}
+
+static void ci_set_s0_mc_reg_index(struct ci_mc_reg_table *table)
+{
+	u32 i;
+	u16 address;
+
+	for (i = 0; i < table->last; i++) {
+		table->mc_reg_address[i].s0 =
+			ci_check_s0_mc_reg_index(table->mc_reg_address[i].s1, &address) ?
+			address : table->mc_reg_address[i].s1;
+	}
+}
+
+static int ci_copy_vbios_mc_reg_table(const struct atom_mc_reg_table *table,
+				      struct ci_mc_reg_table *ci_table)
+{
+	u8 i, j;
+
+	if (table->last > SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+		return -EINVAL;
+	if (table->num_entries > MAX_AC_TIMING_ENTRIES)
+		return -EINVAL;
+
+	for (i = 0; i < table->last; i++)
+		ci_table->mc_reg_address[i].s1 = table->mc_reg_address[i].s1;
+
+	ci_table->last = table->last;
+
+	for (i = 0; i < table->num_entries; i++) {
+		ci_table->mc_reg_table_entry[i].mclk_max =
+			table->mc_reg_table_entry[i].mclk_max;
+		for (j = 0; j < table->last; j++)
+			ci_table->mc_reg_table_entry[i].mc_data[j] =
+				table->mc_reg_table_entry[i].mc_data[j];
+	}
+	ci_table->num_entries = table->num_entries;
+
+	return 0;
+}
+
+static int ci_register_patching_mc_seq(struct amdgpu_device *adev,
+				       struct ci_mc_reg_table *table)
+{
+	u8 i, k;
+	u32 tmp;
+	bool patch;
+
+	tmp = RREG32(mmMC_SEQ_MISC0);
+	patch = ((tmp & 0x0000f00) == 0x300) ? true : false;
+
+	if (patch &&
+	    ((adev->pdev->device == 0x67B0) ||
+	     (adev->pdev->device == 0x67B1))) {
+		for (i = 0; i < table->last; i++) {
+			if (table->last >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			switch (table->mc_reg_address[i].s1) {
+			case mmMC_SEQ_MISC1:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFF8) |
+							0x00000007;
+				}
+				break;
+			case mmMC_SEQ_WR_CTL_D0:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) |
+							0x0000D0DD;
+				}
+				break;
+			case mmMC_SEQ_WR_CTL_D1:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) |
+							0x0000D0DD;
+				}
+				break;
+			case mmMC_SEQ_WR_CTL_2:
+				for (k = 0; k < table->num_entries; k++) {
+					if ((table->mc_reg_table_entry[k].mclk_max == 125000) ||
+					    (table->mc_reg_table_entry[k].mclk_max == 137500))
+						table->mc_reg_table_entry[k].mc_data[i] = 0;
+				}
+				break;
+			case mmMC_SEQ_CAS_TIMING:
+				for (k = 0; k < table->num_entries; k++) {
+					if (table->mc_reg_table_entry[k].mclk_max == 125000)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) |
+							0x000C0140;
+					else if (table->mc_reg_table_entry[k].mclk_max == 137500)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) |
+							0x000C0150;
+				}
+				break;
+			case mmMC_SEQ_MISC_TIMING:
+				for (k = 0; k < table->num_entries; k++) {
+					if (table->mc_reg_table_entry[k].mclk_max == 125000)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) |
+							0x00000030;
+					else if (table->mc_reg_table_entry[k].mclk_max == 137500)
+						table->mc_reg_table_entry[k].mc_data[i] =
+							(table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) |
+							0x00000035;
+				}
+				break;
+			default:
+				break;
+			}
+		}
+
+		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 3);
+		tmp = RREG32(mmMC_SEQ_IO_DEBUG_DATA);
+		tmp = (tmp & 0xFFF8FFFF) | (1 << 16);
+		WREG32(mmMC_SEQ_IO_DEBUG_INDEX, 3);
+		WREG32(mmMC_SEQ_IO_DEBUG_DATA, tmp);
+	}
+
+	return 0;
+}
+
+static int ci_initialize_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct atom_mc_reg_table *table;
+	struct ci_mc_reg_table *ci_table = &pi->mc_reg_table;
+	u8 module_index = ci_get_memory_module_index(adev);
+	int ret;
+
+	table = kzalloc(sizeof(struct atom_mc_reg_table), GFP_KERNEL);
+	if (!table)
+		return -ENOMEM;
+
+	WREG32(mmMC_SEQ_RAS_TIMING_LP, RREG32(mmMC_SEQ_RAS_TIMING));
+	WREG32(mmMC_SEQ_CAS_TIMING_LP, RREG32(mmMC_SEQ_CAS_TIMING));
+	WREG32(mmMC_SEQ_DLL_STBY_LP, RREG32(mmMC_SEQ_DLL_STBY));
+	WREG32(mmMC_SEQ_G5PDX_CMD0_LP, RREG32(mmMC_SEQ_G5PDX_CMD0));
+	WREG32(mmMC_SEQ_G5PDX_CMD1_LP, RREG32(mmMC_SEQ_G5PDX_CMD1));
+	WREG32(mmMC_SEQ_G5PDX_CTRL_LP, RREG32(mmMC_SEQ_G5PDX_CTRL));
+	WREG32(mmMC_SEQ_PMG_DVS_CMD_LP, RREG32(mmMC_SEQ_PMG_DVS_CMD));
+	WREG32(mmMC_SEQ_PMG_DVS_CTL_LP, RREG32(mmMC_SEQ_PMG_DVS_CTL));
+	WREG32(mmMC_SEQ_MISC_TIMING_LP, RREG32(mmMC_SEQ_MISC_TIMING));
+	WREG32(mmMC_SEQ_MISC_TIMING2_LP, RREG32(mmMC_SEQ_MISC_TIMING2));
+	WREG32(mmMC_SEQ_PMG_CMD_EMRS_LP, RREG32(mmMC_PMG_CMD_EMRS));
+	WREG32(mmMC_SEQ_PMG_CMD_MRS_LP, RREG32(mmMC_PMG_CMD_MRS));
+	WREG32(mmMC_SEQ_PMG_CMD_MRS1_LP, RREG32(mmMC_PMG_CMD_MRS1));
+	WREG32(mmMC_SEQ_WR_CTL_D0_LP, RREG32(mmMC_SEQ_WR_CTL_D0));
+	WREG32(mmMC_SEQ_WR_CTL_D1_LP, RREG32(mmMC_SEQ_WR_CTL_D1));
+	WREG32(mmMC_SEQ_RD_CTL_D0_LP, RREG32(mmMC_SEQ_RD_CTL_D0));
+	WREG32(mmMC_SEQ_RD_CTL_D1_LP, RREG32(mmMC_SEQ_RD_CTL_D1));
+	WREG32(mmMC_SEQ_PMG_TIMING_LP, RREG32(mmMC_SEQ_PMG_TIMING));
+	WREG32(mmMC_SEQ_PMG_CMD_MRS2_LP, RREG32(mmMC_PMG_CMD_MRS2));
+	WREG32(mmMC_SEQ_WR_CTL_2_LP, RREG32(mmMC_SEQ_WR_CTL_2));
+
+	ret = amdgpu_atombios_init_mc_reg_table(adev, module_index, table);
+	if (ret)
+		goto init_mc_done;
+
+	ret = ci_copy_vbios_mc_reg_table(table, ci_table);
+	if (ret)
+		goto init_mc_done;
+
+	ci_set_s0_mc_reg_index(ci_table);
+
+	ret = ci_register_patching_mc_seq(adev, ci_table);
+	if (ret)
+		goto init_mc_done;
+
+	ret = ci_set_mc_special_registers(adev, ci_table);
+	if (ret)
+		goto init_mc_done;
+
+	ci_set_valid_flag(ci_table);
+
+init_mc_done:
+	kfree(table);
+
+	return ret;
+}
+
+static int ci_populate_mc_reg_addresses(struct amdgpu_device *adev,
+					SMU7_Discrete_MCRegisters *mc_reg_table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i, j;
+
+	for (i = 0, j = 0; j < pi->mc_reg_table.last; j++) {
+		if (pi->mc_reg_table.valid_flag & (1 << j)) {
+			if (i >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE)
+				return -EINVAL;
+			mc_reg_table->address[i].s0 = cpu_to_be16(pi->mc_reg_table.mc_reg_address[j].s0);
+			mc_reg_table->address[i].s1 = cpu_to_be16(pi->mc_reg_table.mc_reg_address[j].s1);
+			i++;
+		}
+	}
+
+	mc_reg_table->last = (u8)i;
+
+	return 0;
+}
+
+static void ci_convert_mc_registers(const struct ci_mc_reg_entry *entry,
+				    SMU7_Discrete_MCRegisterSet *data,
+				    u32 num_entries, u32 valid_flag)
+{
+	u32 i, j;
+
+	for (i = 0, j = 0; j < num_entries; j++) {
+		if (valid_flag & (1 << j)) {
+			data->value[i] = cpu_to_be32(entry->mc_data[j]);
+			i++;
+		}
+	}
+}
+
+static void ci_convert_mc_reg_table_entry_to_smc(struct amdgpu_device *adev,
+						 const u32 memory_clock,
+						 SMU7_Discrete_MCRegisterSet *mc_reg_table_data)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i = 0;
+
+	for(i = 0; i < pi->mc_reg_table.num_entries; i++) {
+		if (memory_clock <= pi->mc_reg_table.mc_reg_table_entry[i].mclk_max)
+			break;
+	}
+
+	if ((i == pi->mc_reg_table.num_entries) && (i > 0))
+		--i;
+
+	ci_convert_mc_registers(&pi->mc_reg_table.mc_reg_table_entry[i],
+				mc_reg_table_data, pi->mc_reg_table.last,
+				pi->mc_reg_table.valid_flag);
+}
+
+static void ci_convert_mc_reg_table_to_smc(struct amdgpu_device *adev,
+					   SMU7_Discrete_MCRegisters *mc_reg_table)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	u32 i;
+
+	for (i = 0; i < pi->dpm_table.mclk_table.count; i++)
+		ci_convert_mc_reg_table_entry_to_smc(adev,
+						     pi->dpm_table.mclk_table.dpm_levels[i].value,
+						     &mc_reg_table->data[i]);
+}
+
+static int ci_populate_initial_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	int ret;
+
+	memset(&pi->smc_mc_reg_table, 0, sizeof(SMU7_Discrete_MCRegisters));
+
+	ret = ci_populate_mc_reg_addresses(adev, &pi->smc_mc_reg_table);
+	if (ret)
+		return ret;
+	ci_convert_mc_reg_table_to_smc(adev, &pi->smc_mc_reg_table);
+
+	return amdgpu_ci_copy_bytes_to_smc(adev,
+				    pi->mc_reg_table_start,
+				    (u8 *)&pi->smc_mc_reg_table,
+				    sizeof(SMU7_Discrete_MCRegisters),
+				    pi->sram_end);
+}
+
+static int ci_update_and_upload_mc_reg_table(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (!(pi->need_update_smu7_dpm_table & DPMTABLE_OD_UPDATE_MCLK))
+		return 0;
+
+	memset(&pi->smc_mc_reg_table, 0, sizeof(SMU7_Discrete_MCRegisters));
+
+	ci_convert_mc_reg_table_to_smc(adev, &pi->smc_mc_reg_table);
+
+	return amdgpu_ci_copy_bytes_to_smc(adev,
+				    pi->mc_reg_table_start +
+				    offsetof(SMU7_Discrete_MCRegisters, data[0]),
+				    (u8 *)&pi->smc_mc_reg_table.data[0],
+				    sizeof(SMU7_Discrete_MCRegisterSet) *
+				    pi->dpm_table.mclk_table.count,
+				    pi->sram_end);
+}
+
+static void ci_enable_voltage_control(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	tmp |= GENERAL_PWRMGT__VOLT_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+}
+
+static enum amdgpu_pcie_gen ci_get_maximum_link_speed(struct amdgpu_device *adev,
+						      struct amdgpu_ps *amdgpu_state)
+{
+	struct ci_ps *state = ci_get_ps(amdgpu_state);
+	int i;
+	u16 pcie_speed, max_speed = 0;
+
+	for (i = 0; i < state->performance_level_count; i++) {
+		pcie_speed = state->performance_levels[i].pcie_gen;
+		if (max_speed < pcie_speed)
+			max_speed = pcie_speed;
+	}
+
+	return max_speed;
+}
+
+static u16 ci_get_current_pcie_speed(struct amdgpu_device *adev)
+{
+	u32 speed_cntl = 0;
+
+	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL) &
+		PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK;
+	speed_cntl >>= PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT;
+
+	return (u16)speed_cntl;
+}
+
+static int ci_get_current_pcie_lane_number(struct amdgpu_device *adev)
+{
+	u32 link_width = 0;
+
+	link_width = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL) &
+		PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD_MASK;
+	link_width >>= PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_RD__SHIFT;
+
+	switch (link_width) {
+	case 1:
+		return 1;
+	case 2:
+		return 2;
+	case 3:
+		return 4;
+	case 4:
+		return 8;
+	case 0:
+	case 6:
+	default:
+		return 16;
+	}
+}
+
+static void ci_request_link_speed_change_before_state_change(struct amdgpu_device *adev,
+							     struct amdgpu_ps *amdgpu_new_state,
+							     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	enum amdgpu_pcie_gen target_link_speed =
+		ci_get_maximum_link_speed(adev, amdgpu_new_state);
+	enum amdgpu_pcie_gen current_link_speed;
+
+	if (pi->force_pcie_gen == AMDGPU_PCIE_GEN_INVALID)
+		current_link_speed = ci_get_maximum_link_speed(adev, amdgpu_current_state);
+	else
+		current_link_speed = pi->force_pcie_gen;
+
+	pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
+	pi->pspp_notify_required = false;
+	if (target_link_speed > current_link_speed) {
+		switch (target_link_speed) {
+#ifdef CONFIG_ACPI
+		case AMDGPU_PCIE_GEN3:
+			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN3, false) == 0)
+				break;
+			pi->force_pcie_gen = AMDGPU_PCIE_GEN2;
+			if (current_link_speed == AMDGPU_PCIE_GEN2)
+				break;
+		case AMDGPU_PCIE_GEN2:
+			if (amdgpu_acpi_pcie_performance_request(adev, PCIE_PERF_REQ_PECI_GEN2, false) == 0)
+				break;
+#endif
+		default:
+			pi->force_pcie_gen = ci_get_current_pcie_speed(adev);
+			break;
+		}
+	} else {
+		if (target_link_speed < current_link_speed)
+			pi->pspp_notify_required = true;
+	}
+}
+
+static void ci_notify_link_speed_change_after_state_change(struct amdgpu_device *adev,
+							   struct amdgpu_ps *amdgpu_new_state,
+							   struct amdgpu_ps *amdgpu_current_state)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	enum amdgpu_pcie_gen target_link_speed =
+		ci_get_maximum_link_speed(adev, amdgpu_new_state);
+	u8 request;
+
+	if (pi->pspp_notify_required) {
+		if (target_link_speed == AMDGPU_PCIE_GEN3)
+			request = PCIE_PERF_REQ_PECI_GEN3;
+		else if (target_link_speed == AMDGPU_PCIE_GEN2)
+			request = PCIE_PERF_REQ_PECI_GEN2;
+		else
+			request = PCIE_PERF_REQ_PECI_GEN1;
+
+		if ((request == PCIE_PERF_REQ_PECI_GEN1) &&
+		    (ci_get_current_pcie_speed(adev) > 0))
+			return;
+
+#ifdef CONFIG_ACPI
+		amdgpu_acpi_pcie_performance_request(adev, request, false);
+#endif
+	}
+}
+
+static int ci_set_private_data_variables_based_on_pptable(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *allowed_sclk_vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_mclk;
+	struct amdgpu_clock_voltage_dependency_table *allowed_mclk_vddci_table =
+		&adev->pm.dpm.dyn_state.vddci_dependency_on_mclk;
+
+	if (allowed_sclk_vddc_table == NULL)
+		return -EINVAL;
+	if (allowed_sclk_vddc_table->count < 1)
+		return -EINVAL;
+	if (allowed_mclk_vddc_table == NULL)
+		return -EINVAL;
+	if (allowed_mclk_vddc_table->count < 1)
+		return -EINVAL;
+	if (allowed_mclk_vddci_table == NULL)
+		return -EINVAL;
+	if (allowed_mclk_vddci_table->count < 1)
+		return -EINVAL;
+
+	pi->min_vddc_in_pp_table = allowed_sclk_vddc_table->entries[0].v;
+	pi->max_vddc_in_pp_table =
+		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].v;
+
+	pi->min_vddci_in_pp_table = allowed_mclk_vddci_table->entries[0].v;
+	pi->max_vddci_in_pp_table =
+		allowed_mclk_vddci_table->entries[allowed_mclk_vddci_table->count - 1].v;
+
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk =
+		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].clk;
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.mclk =
+		allowed_mclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].clk;
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddc =
+		allowed_sclk_vddc_table->entries[allowed_sclk_vddc_table->count - 1].v;
+	adev->pm.dpm.dyn_state.max_clock_voltage_on_ac.vddci =
+		allowed_mclk_vddci_table->entries[allowed_mclk_vddci_table->count - 1].v;
+
+	return 0;
+}
+
+static void ci_patch_with_vddc_leakage(struct amdgpu_device *adev, u16 *vddc)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_leakage_voltage *leakage_table = &pi->vddc_leakage;
+	u32 leakage_index;
+
+	for (leakage_index = 0; leakage_index < leakage_table->count; leakage_index++) {
+		if (leakage_table->leakage_id[leakage_index] == *vddc) {
+			*vddc = leakage_table->actual_voltage[leakage_index];
+			break;
+		}
+	}
+}
+
+static void ci_patch_with_vddci_leakage(struct amdgpu_device *adev, u16 *vddci)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_leakage_voltage *leakage_table = &pi->vddci_leakage;
+	u32 leakage_index;
+
+	for (leakage_index = 0; leakage_index < leakage_table->count; leakage_index++) {
+		if (leakage_table->leakage_id[leakage_index] == *vddci) {
+			*vddci = leakage_table->actual_voltage[leakage_index];
+			break;
+		}
+	}
+}
+
+static void ci_patch_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
+								      struct amdgpu_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_clock_voltage_dependency_table_with_vddci_leakage(struct amdgpu_device *adev,
+								       struct amdgpu_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddci_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_vce_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
+									  struct amdgpu_vce_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_uvd_clock_voltage_dependency_table_with_vddc_leakage(struct amdgpu_device *adev,
+									  struct amdgpu_uvd_clock_voltage_dependency_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].v);
+	}
+}
+
+static void ci_patch_vddc_phase_shed_limit_table_with_vddc_leakage(struct amdgpu_device *adev,
+								   struct amdgpu_phase_shedding_limits_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].voltage);
+	}
+}
+
+static void ci_patch_clock_voltage_limits_with_vddc_leakage(struct amdgpu_device *adev,
+							    struct amdgpu_clock_and_voltage_limits *table)
+{
+	if (table) {
+		ci_patch_with_vddc_leakage(adev, (u16 *)&table->vddc);
+		ci_patch_with_vddci_leakage(adev, (u16 *)&table->vddci);
+	}
+}
+
+static void ci_patch_cac_leakage_table_with_vddc_leakage(struct amdgpu_device *adev,
+							 struct amdgpu_cac_leakage_table *table)
+{
+	u32 i;
+
+	if (table) {
+		for (i = 0; i < table->count; i++)
+			ci_patch_with_vddc_leakage(adev, &table->entries[i].vddc);
+	}
+}
+
+static void ci_patch_dependency_tables_with_leakage(struct amdgpu_device *adev)
+{
+
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.vddc_dependency_on_sclk);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.vddc_dependency_on_mclk);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk);
+	ci_patch_clock_voltage_dependency_table_with_vddci_leakage(adev,
+								   &adev->pm.dpm.dyn_state.vddci_dependency_on_mclk);
+	ci_patch_vce_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								      &adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table);
+	ci_patch_uvd_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								      &adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table);
+	ci_patch_clock_voltage_dependency_table_with_vddc_leakage(adev,
+								  &adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table);
+	ci_patch_vddc_phase_shed_limit_table_with_vddc_leakage(adev,
+							       &adev->pm.dpm.dyn_state.phase_shedding_limits_table);
+	ci_patch_clock_voltage_limits_with_vddc_leakage(adev,
+							&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac);
+	ci_patch_clock_voltage_limits_with_vddc_leakage(adev,
+							&adev->pm.dpm.dyn_state.max_clock_voltage_on_dc);
+	ci_patch_cac_leakage_table_with_vddc_leakage(adev,
+						     &adev->pm.dpm.dyn_state.cac_leakage_table);
+
+}
+
+static void ci_update_current_ps(struct amdgpu_device *adev,
+				 struct amdgpu_ps *rps)
+{
+	struct ci_ps *new_ps = ci_get_ps(rps);
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->current_rps = *rps;
+	pi->current_ps = *new_ps;
+	pi->current_rps.ps_priv = &pi->current_ps;
+}
+
+static void ci_update_requested_ps(struct amdgpu_device *adev,
+				   struct amdgpu_ps *rps)
+{
+	struct ci_ps *new_ps = ci_get_ps(rps);
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	pi->requested_rps = *rps;
+	pi->requested_ps = *new_ps;
+	pi->requested_rps.ps_priv = &pi->requested_ps;
+}
+
+static int ci_dpm_pre_set_power_state(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
+	struct amdgpu_ps *new_ps = &requested_ps;
+
+	ci_update_requested_ps(adev, new_ps);
+
+	ci_apply_state_adjust_rules(adev, &pi->requested_rps);
+
+	return 0;
+}
+
+static void ci_dpm_post_set_power_state(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+
+	ci_update_current_ps(adev, new_ps);
+}
+
+
+static void ci_dpm_setup_asic(struct amdgpu_device *adev)
+{
+	ci_read_clock_registers(adev);
+	ci_enable_acpi_power_management(adev);
+	ci_init_sclk_t(adev);
+}
+
+static int ci_dpm_enable(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
+	int ret;
+
+	if (amdgpu_ci_is_smc_running(adev))
+		return -EINVAL;
+	if (pi->voltage_control != CISLANDS_VOLTAGE_CONTROL_NONE) {
+		ci_enable_voltage_control(adev);
+		ret = ci_construct_voltage_tables(adev);
+		if (ret) {
+			DRM_ERROR("ci_construct_voltage_tables failed\n");
+			return ret;
+		}
+	}
+	if (pi->caps_dynamic_ac_timing) {
+		ret = ci_initialize_mc_reg_table(adev);
+		if (ret)
+			pi->caps_dynamic_ac_timing = false;
+	}
+	if (pi->dynamic_ss)
+		ci_enable_spread_spectrum(adev, true);
+	if (pi->thermal_protection)
+		ci_enable_thermal_protection(adev, true);
+	ci_program_sstp(adev);
+	ci_enable_display_gap(adev);
+	ci_program_vc(adev);
+	ret = ci_upload_firmware(adev);
+	if (ret) {
+		DRM_ERROR("ci_upload_firmware failed\n");
+		return ret;
+	}
+	ret = ci_process_firmware_header(adev);
+	if (ret) {
+		DRM_ERROR("ci_process_firmware_header failed\n");
+		return ret;
+	}
+	ret = ci_initial_switch_from_arb_f0_to_f1(adev);
+	if (ret) {
+		DRM_ERROR("ci_initial_switch_from_arb_f0_to_f1 failed\n");
+		return ret;
+	}
+	ret = ci_init_smc_table(adev);
+	if (ret) {
+		DRM_ERROR("ci_init_smc_table failed\n");
+		return ret;
+	}
+	ret = ci_init_arb_table_index(adev);
+	if (ret) {
+		DRM_ERROR("ci_init_arb_table_index failed\n");
+		return ret;
+	}
+	if (pi->caps_dynamic_ac_timing) {
+		ret = ci_populate_initial_mc_reg_table(adev);
+		if (ret) {
+			DRM_ERROR("ci_populate_initial_mc_reg_table failed\n");
+			return ret;
+		}
+	}
+	ret = ci_populate_pm_base(adev);
+	if (ret) {
+		DRM_ERROR("ci_populate_pm_base failed\n");
+		return ret;
+	}
+	ci_dpm_start_smc(adev);
+	ci_enable_vr_hot_gpio_interrupt(adev);
+	ret = ci_notify_smc_display_change(adev, false);
+	if (ret) {
+		DRM_ERROR("ci_notify_smc_display_change failed\n");
+		return ret;
+	}
+	ci_enable_sclk_control(adev, true);
+	ret = ci_enable_ulv(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_ulv failed\n");
+		return ret;
+	}
+	ret = ci_enable_ds_master_switch(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_ds_master_switch failed\n");
+		return ret;
+	}
+	ret = ci_start_dpm(adev);
+	if (ret) {
+		DRM_ERROR("ci_start_dpm failed\n");
+		return ret;
+	}
+	ret = ci_enable_didt(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_didt failed\n");
+		return ret;
+	}
+	ret = ci_enable_smc_cac(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_smc_cac failed\n");
+		return ret;
+	}
+	ret = ci_enable_power_containment(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_power_containment failed\n");
+		return ret;
+	}
+
+	ret = ci_power_control_set_level(adev);
+	if (ret) {
+		DRM_ERROR("ci_power_control_set_level failed\n");
+		return ret;
+	}
+
+	ci_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, true);
+
+	ret = ci_enable_thermal_based_sclk_dpm(adev, true);
+	if (ret) {
+		DRM_ERROR("ci_enable_thermal_based_sclk_dpm failed\n");
+		return ret;
+	}
+
+	ci_thermal_start_thermal_controller(adev);
+
+	ci_update_current_ps(adev, boot_ps);
+
+	if (adev->irq.installed &&
+	    amdgpu_is_internal_thermal_sensor(adev->pm.int_thermal_type)) {
+#if 0
+		PPSMC_Result result;
+#endif
+		ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN,
+						       CISLANDS_TEMP_RANGE_MAX);
+		if (ret) {
+			DRM_ERROR("ci_thermal_set_temperature_range failed\n");
+			return ret;
+		}
+		amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+		amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+
+#if 0
+		result = amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_EnableThermalInterrupt);
+
+		if (result != PPSMC_Result_OK)
+			DRM_DEBUG_KMS("Could not enable thermal interrupts.\n");
+#endif
+	}
+
+	return 0;
+}
+
+static void ci_dpm_disable(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *boot_ps = adev->pm.dpm.boot_ps;
+
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+
+	ci_dpm_powergate_uvd(adev, false);
+
+	if (!amdgpu_ci_is_smc_running(adev))
+		return;
+
+	ci_thermal_stop_thermal_controller(adev);
+
+	if (pi->thermal_protection)
+		ci_enable_thermal_protection(adev, false);
+	ci_enable_power_containment(adev, false);
+	ci_enable_smc_cac(adev, false);
+	ci_enable_didt(adev, false);
+	ci_enable_spread_spectrum(adev, false);
+	ci_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, false);
+	ci_stop_dpm(adev);
+	ci_enable_ds_master_switch(adev, false);
+	ci_enable_ulv(adev, false);
+	ci_clear_vc(adev);
+	ci_reset_to_default(adev);
+	ci_dpm_stop_smc(adev);
+	ci_force_switch_to_arb_f0(adev);
+	ci_enable_thermal_based_sclk_dpm(adev, false);
+
+	ci_update_current_ps(adev, boot_ps);
+}
+
+static int ci_dpm_set_power_state(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+	struct amdgpu_ps *old_ps = &pi->current_rps;
+	int ret;
+
+	ci_find_dpm_states_clocks_in_dpm_table(adev, new_ps);
+	if (pi->pcie_performance_request)
+		ci_request_link_speed_change_before_state_change(adev, new_ps, old_ps);
+	ret = ci_freeze_sclk_mclk_dpm(adev);
+	if (ret) {
+		DRM_ERROR("ci_freeze_sclk_mclk_dpm failed\n");
+		return ret;
+	}
+	ret = ci_populate_and_upload_sclk_mclk_dpm_levels(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("ci_populate_and_upload_sclk_mclk_dpm_levels failed\n");
+		return ret;
+	}
+	ret = ci_generate_dpm_level_enable_mask(adev, new_ps);
+	if (ret) {
+		DRM_ERROR("ci_generate_dpm_level_enable_mask failed\n");
+		return ret;
+	}
+
+	ret = ci_update_vce_dpm(adev, new_ps, old_ps);
+	if (ret) {
+		DRM_ERROR("ci_update_vce_dpm failed\n");
+		return ret;
+	}
+
+	ret = ci_update_sclk_t(adev);
+	if (ret) {
+		DRM_ERROR("ci_update_sclk_t failed\n");
+		return ret;
+	}
+	if (pi->caps_dynamic_ac_timing) {
+		ret = ci_update_and_upload_mc_reg_table(adev);
+		if (ret) {
+			DRM_ERROR("ci_update_and_upload_mc_reg_table failed\n");
+			return ret;
+		}
+	}
+	ret = ci_program_memory_timing_parameters(adev);
+	if (ret) {
+		DRM_ERROR("ci_program_memory_timing_parameters failed\n");
+		return ret;
+	}
+	ret = ci_unfreeze_sclk_mclk_dpm(adev);
+	if (ret) {
+		DRM_ERROR("ci_unfreeze_sclk_mclk_dpm failed\n");
+		return ret;
+	}
+	ret = ci_upload_dpm_level_enable_mask(adev);
+	if (ret) {
+		DRM_ERROR("ci_upload_dpm_level_enable_mask failed\n");
+		return ret;
+	}
+	if (pi->pcie_performance_request)
+		ci_notify_link_speed_change_after_state_change(adev, new_ps, old_ps);
+
+	return 0;
+}
+
+#if 0
+static void ci_dpm_reset_asic(struct amdgpu_device *adev)
+{
+	ci_set_boot_state(adev);
+}
+#endif
+
+static void ci_dpm_display_configuration_changed(struct amdgpu_device *adev)
+{
+	ci_program_display_gap(adev);
+}
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+};
+
+union pplib_clock_info {
+	struct _ATOM_PPLIB_R600_CLOCK_INFO r600;
+	struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780;
+	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
+	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
+	struct _ATOM_PPLIB_SI_CLOCK_INFO si;
+	struct _ATOM_PPLIB_CI_CLOCK_INFO ci;
+};
+
+union pplib_power_state {
+	struct _ATOM_PPLIB_STATE v1;
+	struct _ATOM_PPLIB_STATE_V2 v2;
+};
+
+static void ci_parse_pplib_non_clock_info(struct amdgpu_device *adev,
+					  struct amdgpu_ps *rps,
+					  struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
+					  u8 table_rev)
+{
+	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
+	rps->class = le16_to_cpu(non_clock_info->usClassification);
+	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
+
+	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
+		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
+		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
+	} else {
+		rps->vclk = 0;
+		rps->dclk = 0;
+	}
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT)
+		adev->pm.dpm.boot_ps = rps;
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		adev->pm.dpm.uvd_ps = rps;
+}
+
+static void ci_parse_pplib_clock_info(struct amdgpu_device *adev,
+				      struct amdgpu_ps *rps, int index,
+				      union pplib_clock_info *clock_info)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *ps = ci_get_ps(rps);
+	struct ci_pl *pl = &ps->performance_levels[index];
+
+	ps->performance_level_count = index + 1;
+
+	pl->sclk = le16_to_cpu(clock_info->ci.usEngineClockLow);
+	pl->sclk |= clock_info->ci.ucEngineClockHigh << 16;
+	pl->mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow);
+	pl->mclk |= clock_info->ci.ucMemoryClockHigh << 16;
+
+	pl->pcie_gen = amdgpu_get_pcie_gen_support(adev,
+						   pi->sys_pcie_mask,
+						   pi->vbios_boot_state.pcie_gen_bootup_value,
+						   clock_info->ci.ucPCIEGen);
+	pl->pcie_lane = amdgpu_get_pcie_lane_support(adev,
+						     pi->vbios_boot_state.pcie_lane_bootup_value,
+						     le16_to_cpu(clock_info->ci.usPCIELane));
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_ACPI) {
+		pi->acpi_pcie_gen = pl->pcie_gen;
+	}
+
+	if (rps->class2 & ATOM_PPLIB_CLASSIFICATION2_ULV) {
+		pi->ulv.supported = true;
+		pi->ulv.pl = *pl;
+		pi->ulv.cg_ulv_parameter = CISLANDS_CGULVPARAMETER_DFLT;
+	}
+
+	/* patch up boot state */
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
+		pl->mclk = pi->vbios_boot_state.mclk_bootup_value;
+		pl->sclk = pi->vbios_boot_state.sclk_bootup_value;
+		pl->pcie_gen = pi->vbios_boot_state.pcie_gen_bootup_value;
+		pl->pcie_lane = pi->vbios_boot_state.pcie_lane_bootup_value;
+	}
+
+	switch (rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) {
+	case ATOM_PPLIB_CLASSIFICATION_UI_BATTERY:
+		pi->use_pcie_powersaving_levels = true;
+		if (pi->pcie_gen_powersaving.max < pl->pcie_gen)
+			pi->pcie_gen_powersaving.max = pl->pcie_gen;
+		if (pi->pcie_gen_powersaving.min > pl->pcie_gen)
+			pi->pcie_gen_powersaving.min = pl->pcie_gen;
+		if (pi->pcie_lane_powersaving.max < pl->pcie_lane)
+			pi->pcie_lane_powersaving.max = pl->pcie_lane;
+		if (pi->pcie_lane_powersaving.min > pl->pcie_lane)
+			pi->pcie_lane_powersaving.min = pl->pcie_lane;
+		break;
+	case ATOM_PPLIB_CLASSIFICATION_UI_PERFORMANCE:
+		pi->use_pcie_performance_levels = true;
+		if (pi->pcie_gen_performance.max < pl->pcie_gen)
+			pi->pcie_gen_performance.max = pl->pcie_gen;
+		if (pi->pcie_gen_performance.min > pl->pcie_gen)
+			pi->pcie_gen_performance.min = pl->pcie_gen;
+		if (pi->pcie_lane_performance.max < pl->pcie_lane)
+			pi->pcie_lane_performance.max = pl->pcie_lane;
+		if (pi->pcie_lane_performance.min > pl->pcie_lane)
+			pi->pcie_lane_performance.min = pl->pcie_lane;
+		break;
+	default:
+		break;
+	}
+}
+
+static int ci_parse_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
+	union pplib_power_state *power_state;
+	int i, j, k, non_clock_array_index, clock_array_index;
+	union pplib_clock_info *clock_info;
+	struct _StateArray *state_array;
+	struct _ClockInfoArray *clock_info_array;
+	struct _NonClockInfoArray *non_clock_info_array;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	u8 *power_state_offset;
+	struct ci_ps *ps;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	amdgpu_add_thermal_controller(adev);
+
+	state_array = (struct _StateArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usStateArrayOffset));
+	clock_info_array = (struct _ClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
+	non_clock_info_array = (struct _NonClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
+
+	adev->pm.dpm.ps = kzalloc(sizeof(struct amdgpu_ps) *
+				  state_array->ucNumEntries, GFP_KERNEL);
+	if (!adev->pm.dpm.ps)
+		return -ENOMEM;
+	power_state_offset = (u8 *)state_array->states;
+	for (i = 0; i < state_array->ucNumEntries; i++) {
+		u8 *idx;
+		power_state = (union pplib_power_state *)power_state_offset;
+		non_clock_array_index = power_state->v2.nonClockInfoIndex;
+		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
+			&non_clock_info_array->nonClockInfo[non_clock_array_index];
+		ps = kzalloc(sizeof(struct ci_ps), GFP_KERNEL);
+		if (ps == NULL) {
+			kfree(adev->pm.dpm.ps);
+			return -ENOMEM;
+		}
+		adev->pm.dpm.ps[i].ps_priv = ps;
+		ci_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
+					      non_clock_info,
+					      non_clock_info_array->ucEntrySize);
+		k = 0;
+		idx = (u8 *)&power_state->v2.clockInfoIndex[0];
+		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
+			clock_array_index = idx[j];
+			if (clock_array_index >= clock_info_array->ucNumEntries)
+				continue;
+			if (k >= CISLANDS_MAX_HARDWARE_POWERLEVELS)
+				break;
+			clock_info = (union pplib_clock_info *)
+				((u8 *)&clock_info_array->clockInfo[0] +
+				 (clock_array_index * clock_info_array->ucEntrySize));
+			ci_parse_pplib_clock_info(adev,
+						  &adev->pm.dpm.ps[i], k,
+						  clock_info);
+			k++;
+		}
+		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
+	}
+	adev->pm.dpm.num_ps = state_array->ucNumEntries;
+
+	/* fill in the vce power states */
+	for (i = 0; i < AMDGPU_MAX_VCE_LEVELS; i++) {
+		u32 sclk, mclk;
+		clock_array_index = adev->pm.dpm.vce_states[i].clk_idx;
+		clock_info = (union pplib_clock_info *)
+			&clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize];
+		sclk = le16_to_cpu(clock_info->ci.usEngineClockLow);
+		sclk |= clock_info->ci.ucEngineClockHigh << 16;
+		mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow);
+		mclk |= clock_info->ci.ucMemoryClockHigh << 16;
+		adev->pm.dpm.vce_states[i].sclk = sclk;
+		adev->pm.dpm.vce_states[i].mclk = mclk;
+	}
+
+	return 0;
+}
+
+static int ci_get_vbios_boot_values(struct amdgpu_device *adev,
+				    struct ci_vbios_boot_state *boot_state)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
+	ATOM_FIRMWARE_INFO_V2_2 *firmware_info;
+	u8 frev, crev;
+	u16 data_offset;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		firmware_info =
+			(ATOM_FIRMWARE_INFO_V2_2 *)(mode_info->atom_context->bios +
+						    data_offset);
+		boot_state->mvdd_bootup_value = le16_to_cpu(firmware_info->usBootUpMVDDCVoltage);
+		boot_state->vddc_bootup_value = le16_to_cpu(firmware_info->usBootUpVDDCVoltage);
+		boot_state->vddci_bootup_value = le16_to_cpu(firmware_info->usBootUpVDDCIVoltage);
+		boot_state->pcie_gen_bootup_value = ci_get_current_pcie_speed(adev);
+		boot_state->pcie_lane_bootup_value = ci_get_current_pcie_lane_number(adev);
+		boot_state->sclk_bootup_value = le32_to_cpu(firmware_info->ulDefaultEngineClock);
+		boot_state->mclk_bootup_value = le32_to_cpu(firmware_info->ulDefaultMemoryClock);
+
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static void ci_dpm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		kfree(adev->pm.dpm.ps[i].ps_priv);
+	}
+	kfree(adev->pm.dpm.ps);
+	kfree(adev->pm.dpm.priv);
+	kfree(adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries);
+	amdgpu_free_extended_power_table(adev);
+}
+
+/**
+ * ci_dpm_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int ci_dpm_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "radeon/%s_smc.bin", chip_name);
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "cik_smc: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+}
+
+static int ci_dpm_init(struct amdgpu_device *adev)
+{
+	int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info);
+	SMU7_Discrete_DpmTable *dpm_table;
+	struct amdgpu_gpio_rec gpio;
+	u16 data_offset, size;
+	u8 frev, crev;
+	struct ci_power_info *pi;
+	int ret;
+	u32 mask;
+
+	pi = kzalloc(sizeof(struct ci_power_info), GFP_KERNEL);
+	if (pi == NULL)
+		return -ENOMEM;
+	adev->pm.dpm.priv = pi;
+
+	ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask);
+	if (ret)
+		pi->sys_pcie_mask = 0;
+	else
+		pi->sys_pcie_mask = mask;
+	pi->force_pcie_gen = AMDGPU_PCIE_GEN_INVALID;
+
+	pi->pcie_gen_performance.max = AMDGPU_PCIE_GEN1;
+	pi->pcie_gen_performance.min = AMDGPU_PCIE_GEN3;
+	pi->pcie_gen_powersaving.max = AMDGPU_PCIE_GEN1;
+	pi->pcie_gen_powersaving.min = AMDGPU_PCIE_GEN3;
+
+	pi->pcie_lane_performance.max = 0;
+	pi->pcie_lane_performance.min = 16;
+	pi->pcie_lane_powersaving.max = 0;
+	pi->pcie_lane_powersaving.min = 16;
+
+	ret = ci_get_vbios_boot_values(adev, &pi->vbios_boot_state);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	ret = amdgpu_get_platform_caps(adev);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	ret = amdgpu_parse_extended_power_table(adev);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	ret = ci_parse_power_table(adev);
+	if (ret) {
+		ci_dpm_fini(adev);
+		return ret;
+	}
+
+	pi->dll_default_on = false;
+	pi->sram_end = SMC_RAM_END;
+
+	pi->activity_target[0] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[1] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[2] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[3] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[4] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[5] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[6] = CISLAND_TARGETACTIVITY_DFLT;
+	pi->activity_target[7] = CISLAND_TARGETACTIVITY_DFLT;
+
+	pi->mclk_activity_target = CISLAND_MCLK_TARGETACTIVITY_DFLT;
+
+	pi->sclk_dpm_key_disabled = 0;
+	pi->mclk_dpm_key_disabled = 0;
+	pi->pcie_dpm_key_disabled = 0;
+	pi->thermal_sclk_dpm_enabled = 0;
+
+	pi->caps_sclk_ds = true;
+
+	pi->mclk_strobe_mode_threshold = 40000;
+	pi->mclk_stutter_mode_threshold = 40000;
+	pi->mclk_edc_enable_threshold = 40000;
+	pi->mclk_edc_wr_enable_threshold = 40000;
+
+	ci_initialize_powertune_defaults(adev);
+
+	pi->caps_fps = false;
+
+	pi->caps_sclk_throttle_low_notification = false;
+
+	pi->caps_uvd_dpm = true;
+	pi->caps_vce_dpm = true;
+
+	ci_get_leakage_voltages(adev);
+	ci_patch_dependency_tables_with_leakage(adev);
+	ci_set_private_data_variables_based_on_pptable(adev);
+
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries =
+		kzalloc(4 * sizeof(struct amdgpu_clock_voltage_dependency_entry), GFP_KERNEL);
+	if (!adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries) {
+		ci_dpm_fini(adev);
+		return -ENOMEM;
+	}
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.count = 4;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].clk = 0;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[0].v = 0;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].clk = 36000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[1].v = 720;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].clk = 54000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[2].v = 810;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].clk = 72000;
+	adev->pm.dpm.dyn_state.vddc_dependency_on_dispclk.entries[3].v = 900;
+
+	adev->pm.dpm.dyn_state.mclk_sclk_ratio = 4;
+	adev->pm.dpm.dyn_state.sclk_mclk_delta = 15000;
+	adev->pm.dpm.dyn_state.vddc_vddci_delta = 200;
+
+	adev->pm.dpm.dyn_state.valid_sclk_values.count = 0;
+	adev->pm.dpm.dyn_state.valid_sclk_values.values = NULL;
+	adev->pm.dpm.dyn_state.valid_mclk_values.count = 0;
+	adev->pm.dpm.dyn_state.valid_mclk_values.values = NULL;
+
+	if (adev->asic_type == CHIP_HAWAII) {
+		pi->thermal_temp_setting.temperature_low = 94500;
+		pi->thermal_temp_setting.temperature_high = 95000;
+		pi->thermal_temp_setting.temperature_shutdown = 104000;
+	} else {
+		pi->thermal_temp_setting.temperature_low = 99500;
+		pi->thermal_temp_setting.temperature_high = 100000;
+		pi->thermal_temp_setting.temperature_shutdown = 104000;
+	}
+
+	pi->uvd_enabled = false;
+
+	dpm_table = &pi->smc_state_table;
+
+	gpio = amdgpu_atombios_lookup_gpio(adev, VDDC_VRHOT_GPIO_PINID);
+	if (gpio.valid) {
+		dpm_table->VRHotGpio = gpio.shift;
+		adev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_REGULATOR_HOT;
+	} else {
+		dpm_table->VRHotGpio = CISLANDS_UNUSED_GPIO_PIN;
+		adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_REGULATOR_HOT;
+	}
+
+	gpio = amdgpu_atombios_lookup_gpio(adev, PP_AC_DC_SWITCH_GPIO_PINID);
+	if (gpio.valid) {
+		dpm_table->AcDcGpio = gpio.shift;
+		adev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_HARDWAREDC;
+	} else {
+		dpm_table->AcDcGpio = CISLANDS_UNUSED_GPIO_PIN;
+		adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_HARDWAREDC;
+	}
+
+	gpio = amdgpu_atombios_lookup_gpio(adev, VDDC_PCC_GPIO_PINID);
+	if (gpio.valid) {
+		u32 tmp = RREG32_SMC(ixCNB_PWRMGT_CNTL);
+
+		switch (gpio.shift) {
+		case 0:
+			tmp &= ~CNB_PWRMGT_CNTL__GNB_SLOW_MODE_MASK;
+			tmp |= 1 << CNB_PWRMGT_CNTL__GNB_SLOW_MODE__SHIFT;
+			break;
+		case 1:
+			tmp &= ~CNB_PWRMGT_CNTL__GNB_SLOW_MODE_MASK;
+			tmp |= 2 << CNB_PWRMGT_CNTL__GNB_SLOW_MODE__SHIFT;
+			break;
+		case 2:
+			tmp |= CNB_PWRMGT_CNTL__GNB_SLOW_MASK;
+			break;
+		case 3:
+			tmp |= CNB_PWRMGT_CNTL__FORCE_NB_PS1_MASK;
+			break;
+		case 4:
+			tmp |= CNB_PWRMGT_CNTL__DPM_ENABLED_MASK;
+			break;
+		default:
+			DRM_ERROR("Invalid PCC GPIO: %u!\n", gpio.shift);
+			break;
+		}
+		WREG32_SMC(ixCNB_PWRMGT_CNTL, tmp);
+	}
+
+	pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_NONE;
+	pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_NONE;
+	pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_NONE;
+	if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDC, VOLTAGE_OBJ_GPIO_LUT))
+		pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
+	else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDC, VOLTAGE_OBJ_SVID2))
+		pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL) {
+		if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDCI, VOLTAGE_OBJ_GPIO_LUT))
+			pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
+		else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_VDDCI, VOLTAGE_OBJ_SVID2))
+			pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
+		else
+			adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_VDDCI_CONTROL;
+	}
+
+	if (adev->pm.dpm.platform_caps & ATOM_PP_PLATFORM_CAP_MVDDCONTROL) {
+		if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_MVDDC, VOLTAGE_OBJ_GPIO_LUT))
+			pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_BY_GPIO;
+		else if (amdgpu_atombios_is_voltage_gpio(adev, VOLTAGE_TYPE_MVDDC, VOLTAGE_OBJ_SVID2))
+			pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_BY_SVID2;
+		else
+			adev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_MVDDCONTROL;
+	}
+
+	pi->vddc_phase_shed_control = true;
+
+#if defined(CONFIG_ACPI)
+	pi->pcie_performance_request =
+		amdgpu_acpi_is_pcie_performance_request_supported(adev);
+#else
+	pi->pcie_performance_request = false;
+#endif
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		pi->caps_sclk_ss_support = true;
+		pi->caps_mclk_ss_support = true;
+		pi->dynamic_ss = true;
+	} else {
+		pi->caps_sclk_ss_support = false;
+		pi->caps_mclk_ss_support = false;
+		pi->dynamic_ss = true;
+	}
+
+	if (adev->pm.int_thermal_type != THERMAL_TYPE_NONE)
+		pi->thermal_protection = true;
+	else
+		pi->thermal_protection = false;
+
+	pi->caps_dynamic_ac_timing = true;
+
+	pi->uvd_power_gated = false;
+
+	/* make sure dc limits are valid */
+	if ((adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk == 0) ||
+	    (adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.mclk == 0))
+		adev->pm.dpm.dyn_state.max_clock_voltage_on_dc =
+			adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+
+	pi->fan_ctrl_is_in_default_mode = true;
+
+	return 0;
+}
+
+static void
+ci_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev,
+					       struct seq_file *m)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct amdgpu_ps *rps = &pi->current_rps;
+	u32 sclk = ci_get_average_sclk_freq(adev);
+	u32 mclk = ci_get_average_mclk_freq(adev);
+
+	seq_printf(m, "uvd %sabled\n", pi->uvd_enabled ? "en" : "dis");
+	seq_printf(m, "vce %sabled\n", rps->vce_active ? "en" : "dis");
+	seq_printf(m, "power level avg    sclk: %u mclk: %u\n",
+		   sclk, mclk);
+}
+
+static void ci_dpm_print_power_state(struct amdgpu_device *adev,
+				     struct amdgpu_ps *rps)
+{
+	struct ci_ps *ps = ci_get_ps(rps);
+	struct ci_pl *pl;
+	int i;
+
+	amdgpu_dpm_print_class_info(rps->class, rps->class2);
+	amdgpu_dpm_print_cap_info(rps->caps);
+	printk("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+	for (i = 0; i < ps->performance_level_count; i++) {
+		pl = &ps->performance_levels[i];
+		printk("\t\tpower level %d    sclk: %u mclk: %u pcie gen: %u pcie lanes: %u\n",
+		       i, pl->sclk, pl->mclk, pl->pcie_gen + 1, pl->pcie_lane);
+	}
+	amdgpu_dpm_print_ps_status(adev, rps);
+}
+
+static u32 ci_dpm_get_sclk(struct amdgpu_device *adev, bool low)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *requested_state = ci_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->performance_levels[0].sclk;
+	else
+		return requested_state->performance_levels[requested_state->performance_level_count - 1].sclk;
+}
+
+static u32 ci_dpm_get_mclk(struct amdgpu_device *adev, bool low)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *requested_state = ci_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->performance_levels[0].mclk;
+	else
+		return requested_state->performance_levels[requested_state->performance_level_count - 1].mclk;
+}
+
+/* get temperature in millidegrees */
+static int ci_dpm_get_temp(struct amdgpu_device *adev)
+{
+	u32 temp;
+	int actual_temp = 0;
+
+	temp = (RREG32_SMC(ixCG_MULT_THERMAL_STATUS) & CG_MULT_THERMAL_STATUS__CTF_TEMP_MASK) >>
+		CG_MULT_THERMAL_STATUS__CTF_TEMP__SHIFT;
+
+	if (temp & 0x200)
+		actual_temp = 255;
+	else
+		actual_temp = temp & 0x1ff;
+
+	actual_temp = actual_temp * 1000;
+
+	return actual_temp;
+}
+
+static int ci_set_temperature_range(struct amdgpu_device *adev)
+{
+	int ret;
+
+	ret = ci_thermal_enable_alert(adev, false);
+	if (ret)
+		return ret;
+	ret = ci_thermal_set_temperature_range(adev, CISLANDS_TEMP_RANGE_MIN,
+					       CISLANDS_TEMP_RANGE_MAX);
+	if (ret)
+		return ret;
+	ret = ci_thermal_enable_alert(adev, true);
+	if (ret)
+		return ret;
+	return ret;
+}
+
+static int ci_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ci_dpm_set_dpm_funcs(adev);
+	ci_dpm_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int ci_dpm_late_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!amdgpu_dpm)
+		return 0;
+
+	/* init the sysfs and debugfs files late */
+	ret = amdgpu_pm_sysfs_init(adev);
+	if (ret)
+		return ret;
+
+	ret = ci_set_temperature_range(adev);
+	if (ret)
+		return ret;
+
+	ci_dpm_powergate_uvd(adev, true);
+
+	return 0;
+}
+
+static int ci_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = amdgpu_irq_add_id(adev, 230, &adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_irq_add_id(adev, 231, &adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	/* default to balanced state */
+	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.forced_level = AMDGPU_DPM_FORCED_LEVEL_AUTO;
+	adev->pm.default_sclk = adev->clock.default_sclk;
+	adev->pm.default_mclk = adev->clock.default_mclk;
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
+
+	if (amdgpu_dpm == 0)
+		return 0;
+
+	ret = ci_dpm_init_microcode(adev);
+	if (ret)
+		return ret;
+
+	INIT_WORK(&adev->pm.dpm.thermal.work, amdgpu_dpm_thermal_work_handler);
+	mutex_lock(&adev->pm.mutex);
+	ret = ci_dpm_init(adev);
+	if (ret)
+		goto dpm_failed;
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+	if (amdgpu_dpm == 1)
+		amdgpu_pm_print_power_states(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_INFO("amdgpu: dpm initialized\n");
+
+	return 0;
+
+dpm_failed:
+	ci_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_ERROR("amdgpu: dpm initialization failed\n");
+	return ret;
+}
+
+static int ci_dpm_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+	amdgpu_pm_sysfs_fini(adev);
+	ci_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+
+	return 0;
+}
+
+static int ci_dpm_hw_init(void *handle)
+{
+	int ret;
+
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (!amdgpu_dpm)
+		return 0;
+
+	mutex_lock(&adev->pm.mutex);
+	ci_dpm_setup_asic(adev);
+	ret = ci_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+	mutex_unlock(&adev->pm.mutex);
+
+	return ret;
+}
+
+static int ci_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		ci_dpm_disable(adev);
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+	return 0;
+}
+
+static int ci_dpm_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		/* disable dpm */
+		ci_dpm_disable(adev);
+		/* reset the power state */
+		adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+		mutex_unlock(&adev->pm.mutex);
+	}
+	return 0;
+}
+
+static int ci_dpm_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		/* asic init will reset to the boot state */
+		mutex_lock(&adev->pm.mutex);
+		ci_dpm_setup_asic(adev);
+		ret = ci_dpm_enable(adev);
+		if (ret)
+			adev->pm.dpm_enabled = false;
+		else
+			adev->pm.dpm_enabled = true;
+		mutex_unlock(&adev->pm.mutex);
+		if (adev->pm.dpm_enabled)
+			amdgpu_pm_compute_clocks(adev);
+	}
+	return 0;
+}
+
+static bool ci_dpm_is_idle(void *handle)
+{
+	/* XXX */
+	return true;
+}
+
+static int ci_dpm_wait_for_idle(void *handle)
+{
+	/* XXX */
+	return 0;
+}
+
+static void ci_dpm_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "CIK DPM registers\n");
+	dev_info(adev->dev, "  BIOS_SCRATCH_4=0x%08X\n",
+		 RREG32(mmBIOS_SCRATCH_4));
+	dev_info(adev->dev, "  MC_ARB_DRAM_TIMING=0x%08X\n",
+		 RREG32(mmMC_ARB_DRAM_TIMING));
+	dev_info(adev->dev, "  MC_ARB_DRAM_TIMING2=0x%08X\n",
+		 RREG32(mmMC_ARB_DRAM_TIMING2));
+	dev_info(adev->dev, "  MC_ARB_BURST_TIME=0x%08X\n",
+		 RREG32(mmMC_ARB_BURST_TIME));
+	dev_info(adev->dev, "  MC_ARB_DRAM_TIMING_1=0x%08X\n",
+		 RREG32(mmMC_ARB_DRAM_TIMING_1));
+	dev_info(adev->dev, "  MC_ARB_DRAM_TIMING2_1=0x%08X\n",
+		 RREG32(mmMC_ARB_DRAM_TIMING2_1));
+	dev_info(adev->dev, "  MC_CG_CONFIG=0x%08X\n",
+		 RREG32(mmMC_CG_CONFIG));
+	dev_info(adev->dev, "  MC_ARB_CG=0x%08X\n",
+		 RREG32(mmMC_ARB_CG));
+	dev_info(adev->dev, "  DIDT_SQ_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_SQ_CTRL0));
+	dev_info(adev->dev, "  DIDT_DB_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_DB_CTRL0));
+	dev_info(adev->dev, "  DIDT_TD_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_TD_CTRL0));
+	dev_info(adev->dev, "  DIDT_TCP_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_TCP_CTRL0));
+	dev_info(adev->dev, "  CG_THERMAL_INT=0x%08X\n",
+		 RREG32_SMC(ixCG_THERMAL_INT));
+	dev_info(adev->dev, "  CG_THERMAL_CTRL=0x%08X\n",
+		 RREG32_SMC(ixCG_THERMAL_CTRL));
+	dev_info(adev->dev, "  GENERAL_PWRMGT=0x%08X\n",
+		 RREG32_SMC(ixGENERAL_PWRMGT));
+	dev_info(adev->dev, "  MC_SEQ_CNTL_3=0x%08X\n",
+		 RREG32(mmMC_SEQ_CNTL_3));
+	dev_info(adev->dev, "  LCAC_MC0_CNTL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC0_CNTL));
+	dev_info(adev->dev, "  LCAC_MC1_CNTL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC1_CNTL));
+	dev_info(adev->dev, "  LCAC_CPL_CNTL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_CPL_CNTL));
+	dev_info(adev->dev, "  SCLK_PWRMGT_CNTL=0x%08X\n",
+		 RREG32_SMC(ixSCLK_PWRMGT_CNTL));
+	dev_info(adev->dev, "  BIF_LNCNT_RESET=0x%08X\n",
+		 RREG32(mmBIF_LNCNT_RESET));
+	dev_info(adev->dev, "  FIRMWARE_FLAGS=0x%08X\n",
+		 RREG32_SMC(ixFIRMWARE_FLAGS));
+	dev_info(adev->dev, "  CG_SPLL_FUNC_CNTL=0x%08X\n",
+		 RREG32_SMC(ixCG_SPLL_FUNC_CNTL));
+	dev_info(adev->dev, "  CG_SPLL_FUNC_CNTL_2=0x%08X\n",
+		 RREG32_SMC(ixCG_SPLL_FUNC_CNTL_2));
+	dev_info(adev->dev, "  CG_SPLL_FUNC_CNTL_3=0x%08X\n",
+		 RREG32_SMC(ixCG_SPLL_FUNC_CNTL_3));
+	dev_info(adev->dev, "  CG_SPLL_FUNC_CNTL_4=0x%08X\n",
+		 RREG32_SMC(ixCG_SPLL_FUNC_CNTL_4));
+	dev_info(adev->dev, "  CG_SPLL_SPREAD_SPECTRUM=0x%08X\n",
+		 RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM));
+	dev_info(adev->dev, "  CG_SPLL_SPREAD_SPECTRUM_2=0x%08X\n",
+		 RREG32_SMC(ixCG_SPLL_SPREAD_SPECTRUM_2));
+	dev_info(adev->dev, "  DLL_CNTL=0x%08X\n",
+		 RREG32(mmDLL_CNTL));
+	dev_info(adev->dev, "  MCLK_PWRMGT_CNTL=0x%08X\n",
+		 RREG32(mmMCLK_PWRMGT_CNTL));
+	dev_info(adev->dev, "  MPLL_AD_FUNC_CNTL=0x%08X\n",
+		 RREG32(mmMPLL_AD_FUNC_CNTL));
+	dev_info(adev->dev, "  MPLL_DQ_FUNC_CNTL=0x%08X\n",
+		 RREG32(mmMPLL_DQ_FUNC_CNTL));
+	dev_info(adev->dev, "  MPLL_FUNC_CNTL=0x%08X\n",
+		 RREG32(mmMPLL_FUNC_CNTL));
+	dev_info(adev->dev, "  MPLL_FUNC_CNTL_1=0x%08X\n",
+		 RREG32(mmMPLL_FUNC_CNTL_1));
+	dev_info(adev->dev, "  MPLL_FUNC_CNTL_2=0x%08X\n",
+		 RREG32(mmMPLL_FUNC_CNTL_2));
+	dev_info(adev->dev, "  MPLL_SS1=0x%08X\n",
+		 RREG32(mmMPLL_SS1));
+	dev_info(adev->dev, "  MPLL_SS2=0x%08X\n",
+		 RREG32(mmMPLL_SS2));
+	dev_info(adev->dev, "  CG_DISPLAY_GAP_CNTL=0x%08X\n",
+		 RREG32_SMC(ixCG_DISPLAY_GAP_CNTL));
+	dev_info(adev->dev, "  CG_DISPLAY_GAP_CNTL2=0x%08X\n",
+		 RREG32_SMC(ixCG_DISPLAY_GAP_CNTL2));
+	dev_info(adev->dev, "  CG_STATIC_SCREEN_PARAMETER=0x%08X\n",
+		 RREG32_SMC(ixCG_STATIC_SCREEN_PARAMETER));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_0=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_0));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_1=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_1));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_2=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_2));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_3=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_3));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_4=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_4));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_5=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_5));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_6=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_6));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_7=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_7));
+	dev_info(adev->dev, "  RCU_UC_EVENTS=0x%08X\n",
+		 RREG32_SMC(ixRCU_UC_EVENTS));
+	dev_info(adev->dev, "  DPM_TABLE_475=0x%08X\n",
+		 RREG32_SMC(ixDPM_TABLE_475));
+	dev_info(adev->dev, "  MC_SEQ_RAS_TIMING_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_RAS_TIMING_LP));
+	dev_info(adev->dev, "  MC_SEQ_RAS_TIMING=0x%08X\n",
+		 RREG32(mmMC_SEQ_RAS_TIMING));
+	dev_info(adev->dev, "  MC_SEQ_CAS_TIMING_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_CAS_TIMING_LP));
+	dev_info(adev->dev, "  MC_SEQ_CAS_TIMING=0x%08X\n",
+		 RREG32(mmMC_SEQ_CAS_TIMING));
+	dev_info(adev->dev, "  MC_SEQ_DLL_STBY_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_DLL_STBY_LP));
+	dev_info(adev->dev, "  MC_SEQ_DLL_STBY=0x%08X\n",
+		 RREG32(mmMC_SEQ_DLL_STBY));
+	dev_info(adev->dev, "  MC_SEQ_G5PDX_CMD0_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_G5PDX_CMD0_LP));
+	dev_info(adev->dev, "  MC_SEQ_G5PDX_CMD0=0x%08X\n",
+		 RREG32(mmMC_SEQ_G5PDX_CMD0));
+	dev_info(adev->dev, "  MC_SEQ_G5PDX_CMD1_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_G5PDX_CMD1_LP));
+	dev_info(adev->dev, "  MC_SEQ_G5PDX_CMD1=0x%08X\n",
+		 RREG32(mmMC_SEQ_G5PDX_CMD1));
+	dev_info(adev->dev, "  MC_SEQ_G5PDX_CTRL_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_G5PDX_CTRL_LP));
+	dev_info(adev->dev, "  MC_SEQ_G5PDX_CTRL=0x%08X\n",
+		 RREG32(mmMC_SEQ_G5PDX_CTRL));
+	dev_info(adev->dev, "  MC_SEQ_PMG_DVS_CMD_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_DVS_CMD_LP));
+	dev_info(adev->dev, "  MC_SEQ_PMG_DVS_CMD=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_DVS_CMD));
+	dev_info(adev->dev, "  MC_SEQ_PMG_DVS_CTL_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_DVS_CTL_LP));
+	dev_info(adev->dev, "  MC_SEQ_PMG_DVS_CTL=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_DVS_CTL));
+	dev_info(adev->dev, "  MC_SEQ_MISC_TIMING_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_MISC_TIMING_LP));
+	dev_info(adev->dev, "  MC_SEQ_MISC_TIMING=0x%08X\n",
+		 RREG32(mmMC_SEQ_MISC_TIMING));
+	dev_info(adev->dev, "  MC_SEQ_MISC_TIMING2_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_MISC_TIMING2_LP));
+	dev_info(adev->dev, "  MC_SEQ_MISC_TIMING2=0x%08X\n",
+		 RREG32(mmMC_SEQ_MISC_TIMING2));
+	dev_info(adev->dev, "  MC_SEQ_PMG_CMD_EMRS_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_CMD_EMRS_LP));
+	dev_info(adev->dev, "  MC_PMG_CMD_EMRS=0x%08X\n",
+		 RREG32(mmMC_PMG_CMD_EMRS));
+	dev_info(adev->dev, "  MC_SEQ_PMG_CMD_MRS_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_CMD_MRS_LP));
+	dev_info(adev->dev, "  MC_PMG_CMD_MRS=0x%08X\n",
+		 RREG32(mmMC_PMG_CMD_MRS));
+	dev_info(adev->dev, "  MC_SEQ_PMG_CMD_MRS1_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_CMD_MRS1_LP));
+	dev_info(adev->dev, "  MC_PMG_CMD_MRS1=0x%08X\n",
+		 RREG32(mmMC_PMG_CMD_MRS1));
+	dev_info(adev->dev, "  MC_SEQ_WR_CTL_D0_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_WR_CTL_D0_LP));
+	dev_info(adev->dev, "  MC_SEQ_WR_CTL_D0=0x%08X\n",
+		 RREG32(mmMC_SEQ_WR_CTL_D0));
+	dev_info(adev->dev, "  MC_SEQ_WR_CTL_D1_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_WR_CTL_D1_LP));
+	dev_info(adev->dev, "  MC_SEQ_WR_CTL_D1=0x%08X\n",
+		 RREG32(mmMC_SEQ_WR_CTL_D1));
+	dev_info(adev->dev, "  MC_SEQ_RD_CTL_D0_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_RD_CTL_D0_LP));
+	dev_info(adev->dev, "  MC_SEQ_RD_CTL_D0=0x%08X\n",
+		 RREG32(mmMC_SEQ_RD_CTL_D0));
+	dev_info(adev->dev, "  MC_SEQ_RD_CTL_D1_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_RD_CTL_D1_LP));
+	dev_info(adev->dev, "  MC_SEQ_RD_CTL_D1=0x%08X\n",
+		 RREG32(mmMC_SEQ_RD_CTL_D1));
+	dev_info(adev->dev, "  MC_SEQ_PMG_TIMING_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_TIMING_LP));
+	dev_info(adev->dev, "  MC_SEQ_PMG_TIMING=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_TIMING));
+	dev_info(adev->dev, "  MC_SEQ_PMG_CMD_MRS2_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_PMG_CMD_MRS2_LP));
+	dev_info(adev->dev, "  MC_PMG_CMD_MRS2=0x%08X\n",
+		 RREG32(mmMC_PMG_CMD_MRS2));
+	dev_info(adev->dev, "  MC_SEQ_WR_CTL_2_LP=0x%08X\n",
+		 RREG32(mmMC_SEQ_WR_CTL_2_LP));
+	dev_info(adev->dev, "  MC_SEQ_WR_CTL_2=0x%08X\n",
+		 RREG32(mmMC_SEQ_WR_CTL_2));
+	dev_info(adev->dev, "  PCIE_LC_SPEED_CNTL=0x%08X\n",
+		 RREG32_PCIE(ixPCIE_LC_SPEED_CNTL));
+	dev_info(adev->dev, "  PCIE_LC_LINK_WIDTH_CNTL=0x%08X\n",
+		 RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL));
+	dev_info(adev->dev, "  SMC_IND_INDEX_0=0x%08X\n",
+		 RREG32(mmSMC_IND_INDEX_0));
+	dev_info(adev->dev, "  SMC_IND_DATA_0=0x%08X\n",
+		 RREG32(mmSMC_IND_DATA_0));
+	dev_info(adev->dev, "  SMC_IND_ACCESS_CNTL=0x%08X\n",
+		 RREG32(mmSMC_IND_ACCESS_CNTL));
+	dev_info(adev->dev, "  SMC_RESP_0=0x%08X\n",
+		 RREG32(mmSMC_RESP_0));
+	dev_info(adev->dev, "  SMC_MESSAGE_0=0x%08X\n",
+		 RREG32(mmSMC_MESSAGE_0));
+	dev_info(adev->dev, "  SMC_SYSCON_RESET_CNTL=0x%08X\n",
+		 RREG32_SMC(ixSMC_SYSCON_RESET_CNTL));
+	dev_info(adev->dev, "  SMC_SYSCON_CLOCK_CNTL_0=0x%08X\n",
+		 RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0));
+	dev_info(adev->dev, "  SMC_SYSCON_MISC_CNTL=0x%08X\n",
+		 RREG32_SMC(ixSMC_SYSCON_MISC_CNTL));
+	dev_info(adev->dev, "  SMC_PC_C=0x%08X\n",
+		 RREG32_SMC(ixSMC_PC_C));
+}
+
+static int ci_dpm_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int ci_dpm_set_interrupt_state(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      unsigned type,
+				      enum amdgpu_interrupt_state state)
+{
+	u32 cg_thermal_int;
+
+	switch (type) {
+	case AMDGPU_THERMAL_IRQ_LOW_TO_HIGH:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case AMDGPU_THERMAL_IRQ_HIGH_TO_LOW:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int ci_dpm_process_interrupt(struct amdgpu_device *adev,
+				    struct amdgpu_irq_src *source, 
+				    struct amdgpu_iv_entry *entry)
+{
+	bool queue_thermal = false;
+
+	if (entry == NULL)
+		return -EINVAL;
+
+	switch (entry->src_id) {
+	case 230: /* thermal low to high */
+		DRM_DEBUG("IH: thermal low to high\n");
+		adev->pm.dpm.thermal.high_to_low = false;
+		queue_thermal = true;
+		break;
+	case 231: /* thermal high to low */
+		DRM_DEBUG("IH: thermal high to low\n");
+		adev->pm.dpm.thermal.high_to_low = true;
+		queue_thermal = true;
+		break;
+	default:
+		break;
+	}
+
+	if (queue_thermal)
+		schedule_work(&adev->pm.dpm.thermal.work);
+
+	return 0;
+}
+
+static int ci_dpm_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int ci_dpm_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs ci_dpm_ip_funcs = {
+	.early_init = ci_dpm_early_init,
+	.late_init = ci_dpm_late_init,
+	.sw_init = ci_dpm_sw_init,
+	.sw_fini = ci_dpm_sw_fini,
+	.hw_init = ci_dpm_hw_init,
+	.hw_fini = ci_dpm_hw_fini,
+	.suspend = ci_dpm_suspend,
+	.resume = ci_dpm_resume,
+	.is_idle = ci_dpm_is_idle,
+	.wait_for_idle = ci_dpm_wait_for_idle,
+	.soft_reset = ci_dpm_soft_reset,
+	.print_status = ci_dpm_print_status,
+	.set_clockgating_state = ci_dpm_set_clockgating_state,
+	.set_powergating_state = ci_dpm_set_powergating_state,
+};
+
+static const struct amdgpu_dpm_funcs ci_dpm_funcs = {
+	.get_temperature = &ci_dpm_get_temp,
+	.pre_set_power_state = &ci_dpm_pre_set_power_state,
+	.set_power_state = &ci_dpm_set_power_state,
+	.post_set_power_state = &ci_dpm_post_set_power_state,
+	.display_configuration_changed = &ci_dpm_display_configuration_changed,
+	.get_sclk = &ci_dpm_get_sclk,
+	.get_mclk = &ci_dpm_get_mclk,
+	.print_power_state = &ci_dpm_print_power_state,
+	.debugfs_print_current_performance_level = &ci_dpm_debugfs_print_current_performance_level,
+	.force_performance_level = &ci_dpm_force_performance_level,
+	.vblank_too_short = &ci_dpm_vblank_too_short,
+	.powergate_uvd = &ci_dpm_powergate_uvd,
+	.set_fan_control_mode = &ci_dpm_set_fan_control_mode,
+	.get_fan_control_mode = &ci_dpm_get_fan_control_mode,
+	.set_fan_speed_percent = &ci_dpm_set_fan_speed_percent,
+	.get_fan_speed_percent = &ci_dpm_get_fan_speed_percent,
+};
+
+static void ci_dpm_set_dpm_funcs(struct amdgpu_device *adev)
+{
+	if (adev->pm.funcs == NULL)
+		adev->pm.funcs = &ci_dpm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs ci_dpm_irq_funcs = {
+	.set = ci_dpm_set_interrupt_state,
+	.process = ci_dpm_process_interrupt,
+};
+
+static void ci_dpm_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->pm.dpm.thermal.irq.num_types = AMDGPU_THERMAL_IRQ_LAST;
+	adev->pm.dpm.thermal.irq.funcs = &ci_dpm_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.h b/kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
new file mode 100644
index 000000000..faccc30c9
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
@@ -0,0 +1,348 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __CI_DPM_H__
+#define __CI_DPM_H__
+
+#include "amdgpu_atombios.h"
+#include "ppsmc.h"
+
+#define SMU__NUM_SCLK_DPM_STATE  8
+#define SMU__NUM_MCLK_DPM_LEVELS 6
+#define SMU__NUM_LCLK_DPM_LEVELS 8
+#define SMU__NUM_PCIE_DPM_LEVELS 8
+#include "smu7_discrete.h"
+
+#define CISLANDS_MAX_HARDWARE_POWERLEVELS 2
+
+#define CISLANDS_UNUSED_GPIO_PIN 0x7F
+
+struct ci_pl {
+	u32 mclk;
+	u32 sclk;
+	enum amdgpu_pcie_gen pcie_gen;
+	u16 pcie_lane;
+};
+
+struct ci_ps {
+	u16 performance_level_count;
+	bool dc_compatible;
+	u32 sclk_t;
+	struct ci_pl performance_levels[CISLANDS_MAX_HARDWARE_POWERLEVELS];
+};
+
+struct ci_dpm_level {
+	bool enabled;
+	u32 value;
+	u32 param1;
+};
+
+#define CISLAND_MAX_DEEPSLEEP_DIVIDER_ID 5
+#define MAX_REGULAR_DPM_NUMBER 8
+#define CISLAND_MINIMUM_ENGINE_CLOCK 800
+
+struct ci_single_dpm_table {
+	u32 count;
+	struct ci_dpm_level dpm_levels[MAX_REGULAR_DPM_NUMBER];
+};
+
+struct ci_dpm_table {
+	struct ci_single_dpm_table sclk_table;
+	struct ci_single_dpm_table mclk_table;
+	struct ci_single_dpm_table pcie_speed_table;
+	struct ci_single_dpm_table vddc_table;
+	struct ci_single_dpm_table vddci_table;
+	struct ci_single_dpm_table mvdd_table;
+};
+
+struct ci_mc_reg_entry {
+	u32 mclk_max;
+	u32 mc_data[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct ci_mc_reg_table {
+	u8 last;
+	u8 num_entries;
+	u16 valid_flag;
+	struct ci_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
+	SMU7_Discrete_MCRegisterAddress mc_reg_address[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
+};
+
+struct ci_ulv_parm
+{
+	bool supported;
+	u32 cg_ulv_parameter;
+	u32 volt_change_delay;
+	struct ci_pl pl;
+};
+
+#define CISLANDS_MAX_LEAKAGE_COUNT  8
+
+struct ci_leakage_voltage {
+	u16 count;
+	u16 leakage_id[CISLANDS_MAX_LEAKAGE_COUNT];
+	u16 actual_voltage[CISLANDS_MAX_LEAKAGE_COUNT];
+};
+
+struct ci_dpm_level_enable_mask {
+	u32 uvd_dpm_enable_mask;
+	u32 vce_dpm_enable_mask;
+	u32 acp_dpm_enable_mask;
+	u32 samu_dpm_enable_mask;
+	u32 sclk_dpm_enable_mask;
+	u32 mclk_dpm_enable_mask;
+	u32 pcie_dpm_enable_mask;
+};
+
+struct ci_vbios_boot_state
+{
+	u16 mvdd_bootup_value;
+	u16 vddc_bootup_value;
+	u16 vddci_bootup_value;
+	u32 sclk_bootup_value;
+	u32 mclk_bootup_value;
+	u16 pcie_gen_bootup_value;
+	u16 pcie_lane_bootup_value;
+};
+
+struct ci_clock_registers {
+	u32 cg_spll_func_cntl;
+	u32 cg_spll_func_cntl_2;
+	u32 cg_spll_func_cntl_3;
+	u32 cg_spll_func_cntl_4;
+	u32 cg_spll_spread_spectrum;
+	u32 cg_spll_spread_spectrum_2;
+	u32 dll_cntl;
+	u32 mclk_pwrmgt_cntl;
+	u32 mpll_ad_func_cntl;
+	u32 mpll_dq_func_cntl;
+	u32 mpll_func_cntl;
+	u32 mpll_func_cntl_1;
+	u32 mpll_func_cntl_2;
+	u32 mpll_ss1;
+	u32 mpll_ss2;
+};
+
+struct ci_thermal_temperature_setting {
+	s32 temperature_low;
+	s32 temperature_high;
+	s32 temperature_shutdown;
+};
+
+struct ci_pcie_perf_range {
+	u16 max;
+	u16 min;
+};
+
+enum ci_pt_config_reg_type {
+	CISLANDS_CONFIGREG_MMR = 0,
+	CISLANDS_CONFIGREG_SMC_IND,
+	CISLANDS_CONFIGREG_DIDT_IND,
+	CISLANDS_CONFIGREG_CACHE,
+	CISLANDS_CONFIGREG_MAX
+};
+
+#define POWERCONTAINMENT_FEATURE_BAPM            0x00000001
+#define POWERCONTAINMENT_FEATURE_TDCLimit        0x00000002
+#define POWERCONTAINMENT_FEATURE_PkgPwrLimit     0x00000004
+
+struct ci_pt_config_reg {
+	u32 offset;
+	u32 mask;
+	u32 shift;
+	u32 value;
+	enum ci_pt_config_reg_type type;
+};
+
+struct ci_pt_defaults {
+	u8 svi_load_line_en;
+	u8 svi_load_line_vddc;
+	u8 tdc_vddc_throttle_release_limit_perc;
+	u8 tdc_mawt;
+	u8 tdc_waterfall_ctl;
+	u8 dte_ambient_temp_base;
+	u32 display_cac;
+	u32 bapm_temp_gradient;
+	u16 bapmti_r[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
+	u16 bapmti_rc[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
+};
+
+#define DPMTABLE_OD_UPDATE_SCLK     0x00000001
+#define DPMTABLE_OD_UPDATE_MCLK     0x00000002
+#define DPMTABLE_UPDATE_SCLK        0x00000004
+#define DPMTABLE_UPDATE_MCLK        0x00000008
+
+struct ci_power_info {
+	struct ci_dpm_table dpm_table;
+	u32 voltage_control;
+	u32 mvdd_control;
+	u32 vddci_control;
+	u32 active_auto_throttle_sources;
+	struct ci_clock_registers clock_registers;
+	u16 acpi_vddc;
+	u16 acpi_vddci;
+	enum amdgpu_pcie_gen force_pcie_gen;
+	enum amdgpu_pcie_gen acpi_pcie_gen;
+	struct ci_leakage_voltage vddc_leakage;
+	struct ci_leakage_voltage vddci_leakage;
+	u16 max_vddc_in_pp_table;
+	u16 min_vddc_in_pp_table;
+	u16 max_vddci_in_pp_table;
+	u16 min_vddci_in_pp_table;
+	u32 mclk_strobe_mode_threshold;
+	u32 mclk_stutter_mode_threshold;
+	u32 mclk_edc_enable_threshold;
+	u32 mclk_edc_wr_enable_threshold;
+	struct ci_vbios_boot_state vbios_boot_state;
+	/* smc offsets */
+	u32 sram_end;
+	u32 dpm_table_start;
+	u32 soft_regs_start;
+	u32 mc_reg_table_start;
+	u32 fan_table_start;
+	u32 arb_table_start;
+	/* smc tables */
+	SMU7_Discrete_DpmTable smc_state_table;
+	SMU7_Discrete_MCRegisters smc_mc_reg_table;
+	SMU7_Discrete_PmFuses smc_powertune_table;
+	/* other stuff */
+	struct ci_mc_reg_table mc_reg_table;
+	struct atom_voltage_table vddc_voltage_table;
+	struct atom_voltage_table vddci_voltage_table;
+	struct atom_voltage_table mvdd_voltage_table;
+	struct ci_ulv_parm ulv;
+	u32 power_containment_features;
+	const struct ci_pt_defaults *powertune_defaults;
+	u32 dte_tj_offset;
+	bool vddc_phase_shed_control;
+	struct ci_thermal_temperature_setting thermal_temp_setting;
+	struct ci_dpm_level_enable_mask dpm_level_enable_mask;
+	u32 need_update_smu7_dpm_table;
+	u32 sclk_dpm_key_disabled;
+	u32 mclk_dpm_key_disabled;
+	u32 pcie_dpm_key_disabled;
+	u32 thermal_sclk_dpm_enabled;
+	struct ci_pcie_perf_range pcie_gen_performance;
+	struct ci_pcie_perf_range pcie_lane_performance;
+	struct ci_pcie_perf_range pcie_gen_powersaving;
+	struct ci_pcie_perf_range pcie_lane_powersaving;
+	u32 activity_target[SMU7_MAX_LEVELS_GRAPHICS];
+	u32 mclk_activity_target;
+	u32 low_sclk_interrupt_t;
+	u32 last_mclk_dpm_enable_mask;
+	u32 sys_pcie_mask;
+	/* caps */
+	bool caps_power_containment;
+	bool caps_cac;
+	bool caps_sq_ramping;
+	bool caps_db_ramping;
+	bool caps_td_ramping;
+	bool caps_tcp_ramping;
+	bool caps_fps;
+	bool caps_sclk_ds;
+	bool caps_sclk_ss_support;
+	bool caps_mclk_ss_support;
+	bool caps_uvd_dpm;
+	bool caps_vce_dpm;
+	bool caps_samu_dpm;
+	bool caps_acp_dpm;
+	bool caps_automatic_dc_transition;
+	bool caps_sclk_throttle_low_notification;
+	bool caps_dynamic_ac_timing;
+	bool caps_od_fuzzy_fan_control_support;
+	/* flags */
+	bool thermal_protection;
+	bool pcie_performance_request;
+	bool dynamic_ss;
+	bool dll_default_on;
+	bool cac_enabled;
+	bool uvd_enabled;
+	bool battery_state;
+	bool pspp_notify_required;
+	bool enable_bapm_feature;
+	bool enable_tdc_limit_feature;
+	bool enable_pkg_pwr_tracking_feature;
+	bool use_pcie_performance_levels;
+	bool use_pcie_powersaving_levels;
+	bool uvd_power_gated;
+	/* driver states */
+	struct amdgpu_ps current_rps;
+	struct ci_ps current_ps;
+	struct amdgpu_ps requested_rps;
+	struct ci_ps requested_ps;
+	/* fan control */
+	bool fan_ctrl_is_in_default_mode;
+	bool fan_is_controlled_by_smc;
+	u32 t_min;
+	u32 fan_ctrl_default_mode;
+};
+
+#define CISLANDS_VOLTAGE_CONTROL_NONE                   0x0
+#define CISLANDS_VOLTAGE_CONTROL_BY_GPIO                0x1
+#define CISLANDS_VOLTAGE_CONTROL_BY_SVID2               0x2
+
+#define CISLANDS_Q88_FORMAT_CONVERSION_UNIT             256
+
+#define CISLANDS_VRC_DFLT0                              0x3FFFC000
+#define CISLANDS_VRC_DFLT1                              0x000400
+#define CISLANDS_VRC_DFLT2                              0xC00080
+#define CISLANDS_VRC_DFLT3                              0xC00200
+#define CISLANDS_VRC_DFLT4                              0xC01680
+#define CISLANDS_VRC_DFLT5                              0xC00033
+#define CISLANDS_VRC_DFLT6                              0xC00033
+#define CISLANDS_VRC_DFLT7                              0x3FFFC000
+
+#define CISLANDS_CGULVPARAMETER_DFLT                    0x00040035
+#define CISLAND_TARGETACTIVITY_DFLT                     30
+#define CISLAND_MCLK_TARGETACTIVITY_DFLT                10
+
+#define PCIE_PERF_REQ_REMOVE_REGISTRY   0
+#define PCIE_PERF_REQ_FORCE_LOWPOWER    1
+#define PCIE_PERF_REQ_PECI_GEN1         2
+#define PCIE_PERF_REQ_PECI_GEN2         3
+#define PCIE_PERF_REQ_PECI_GEN3         4
+
+#define CISLANDS_SSTU_DFLT                               0
+#define CISLANDS_SST_DFLT                                0x00C8
+
+/* XXX are these ok? */
+#define CISLANDS_TEMP_RANGE_MIN (90 * 1000)
+#define CISLANDS_TEMP_RANGE_MAX (120 * 1000)
+
+int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit);
+void amdgpu_ci_start_smc(struct amdgpu_device *adev);
+void amdgpu_ci_reset_smc(struct amdgpu_device *adev);
+int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev);
+void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev);
+void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev);
+bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev);
+PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg);
+PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev);
+int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit);
+int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
+			   u32 smc_address, u32 *value, u32 limit);
+int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
+			    u32 smc_address, u32 value, u32 limit);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/ci_smc.c b/kernel/drivers/gpu/drm/amd/amdgpu/ci_smc.c
new file mode 100644
index 000000000..7eb9069db
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/ci_smc.c
@@ -0,0 +1,279 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "cikd.h"
+#include "ppsmc.h"
+#include "amdgpu_ucode.h"
+#include "ci_dpm.h"
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+static int ci_set_smc_sram_address(struct amdgpu_device *adev,
+				   u32 smc_address, u32 limit)
+{
+	if (smc_address & 3)
+		return -EINVAL;
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+
+	return 0;
+}
+
+int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit)
+{
+	unsigned long flags;
+	u32 data, original_data;
+	u32 addr;
+	u32 extra_shift;
+	int ret = 0;
+
+	if (smc_start_address & 3)
+		return -EINVAL;
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	while (byte_count >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
+
+		ret = ci_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	/* RMW for the final bytes */
+	if (byte_count > 0) {
+		data = 0;
+
+		ret = ci_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		original_data = RREG32(mmSMC_IND_DATA_0);
+
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+
+		data |= (original_data & ~((~0UL) << extra_shift));
+
+		ret = ci_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			goto done;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+
+done:
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
+
+void amdgpu_ci_start_smc(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+
+	tmp &= ~SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
+}
+
+void amdgpu_ci_reset_smc(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+
+	tmp |= SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
+}
+
+int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev)
+{
+	static u8 data[] = { 0xE0, 0x00, 0x80, 0x40 };
+
+	return amdgpu_ci_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
+}
+
+void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+
+	tmp |= SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
+
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
+}
+
+void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+
+	tmp &= ~SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
+
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
+}
+
+bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev)
+{
+	u32 clk = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	u32 pc_c = RREG32_SMC(ixSMC_PC_C);
+
+	if (!(clk & SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK) && (0x20100 <= pc_c))
+		return true;
+
+	return false;
+}
+
+PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_ci_is_smc_running(adev))
+		return PPSMC_Result_Failed;
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSMC_RESP_0);
+		if (tmp != 0)
+			break;
+		udelay(1);
+	}
+	tmp = RREG32(mmSMC_RESP_0);
+
+	return (PPSMC_Result)tmp;
+}
+
+PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	if (!amdgpu_ci_is_smc_running(adev))
+		return PPSMC_Result_OK;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+		if ((tmp & SMC_SYSCON_CLOCK_CNTL_0__cken_MASK) == 0)
+			break;
+		udelay(1);
+	}
+
+	return PPSMC_Result_OK;
+}
+
+int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit)
+{
+	const struct smc_firmware_header_v1_0 *hdr;
+	unsigned long flags;
+	u32 ucode_start_address;
+	u32 ucode_size;
+	const u8 *src;
+	u32 data;
+
+	if (!adev->pm.fw)
+		return -EINVAL;
+
+	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	src = (const u8 *)
+		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	if (ucode_size & 3)
+		return -EINVAL;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK,
+		~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+	while (ucode_size >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		ucode_size -= 4;
+	}
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return 0;
+}
+
+int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
+			   u32 smc_address, u32 *value, u32 limit)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	ret = ci_set_smc_sram_address(adev, smc_address, limit);
+	if (ret == 0)
+		*value = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
+
+int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
+			    u32 smc_address, u32 value, u32 limit)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	ret = ci_set_smc_sram_address(adev, smc_address, limit);
+	if (ret == 0)
+		WREG32(mmSMC_IND_DATA_0, value);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return ret;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik.c b/kernel/drivers/gpu/drm/amd/amdgpu/cik.c
new file mode 100644
index 000000000..484710cfd
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -0,0 +1,2521 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "cikd.h"
+#include "atom.h"
+
+#include "cik.h"
+#include "gmc_v7_0.h"
+#include "cik_ih.h"
+#include "dce_v8_0.h"
+#include "gfx_v7_0.h"
+#include "cik_sdma.h"
+#include "uvd_v4_2.h"
+#include "vce_v2_0.h"
+#include "cik_dpm.h"
+
+#include "uvd/uvd_4_2_d.h"
+
+#include "smu/smu_7_0_1_d.h"
+#include "smu/smu_7_0_1_sh_mask.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#include "amdgpu_amdkfd.h"
+
+/*
+ * Indirect registers accessor
+ */
+static u32 cik_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	r = RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void cik_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	WREG32(mmPCIE_DATA, v);
+	(void)RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 cik_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	r = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void cik_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	WREG32(mmSMC_IND_DATA_0, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+static u32 cik_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	r = RREG32(mmUVD_CTX_DATA);
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+	return r;
+}
+
+static void cik_uvd_ctx_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	WREG32(mmUVD_CTX_DATA, (v));
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+}
+
+static u32 cik_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	r = RREG32(mmDIDT_IND_DATA);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void cik_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	WREG32(mmDIDT_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static const u32 bonaire_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 bonaire_golden_common_registers[] =
+{
+	0x31dc, 0xffffffff, 0x00000800,
+	0x31dd, 0xffffffff, 0x00000800,
+	0x31e6, 0xffffffff, 0x00007fbf,
+	0x31e7, 0xffffffff, 0x00007faf
+};
+
+static const u32 bonaire_golden_registers[] =
+{
+	0xcd5, 0x00000333, 0x00000333,
+	0xcd4, 0x000c0fc0, 0x00040200,
+	0x2684, 0x00010000, 0x00058208,
+	0xf000, 0xffff1fff, 0x00140000,
+	0xf080, 0xfdfc0fff, 0x00000100,
+	0xf08d, 0x40000000, 0x40000200,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x260e, 0x0002021c, 0x00020200,
+	0x31e, 0x00000080, 0x00000000,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0xd43, 0x00810000, 0x408af000,
+	0x1c0c, 0x31000111, 0x00000011,
+	0xbd2, 0x73773777, 0x12010001,
+	0x883, 0x00007fb6, 0x0021a1b1,
+	0x884, 0x00007fb6, 0x002021b1,
+	0x860, 0x00007fb6, 0x00002191,
+	0x886, 0x00007fb6, 0x002121b1,
+	0x887, 0x00007fb6, 0x002021b1,
+	0x877, 0x00007fb6, 0x00002191,
+	0x878, 0x00007fb6, 0x00002191,
+	0xd8a, 0x0000003f, 0x0000000a,
+	0xd8b, 0x0000003f, 0x0000000a,
+	0xab9, 0x00073ffe, 0x000022a2,
+	0x903, 0x000007ff, 0x00000000,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22fc, 0x00002001, 0x00000001,
+	0x22c9, 0xffffffff, 0x00ffffff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0x136, 0x00000fff, 0x00000100,
+	0xf9e, 0x00000001, 0x00000002,
+	0x2440, 0x03000000, 0x0362c688,
+	0x2300, 0x000000ff, 0x00000001,
+	0x390, 0x00001fff, 0x00001fff,
+	0x2418, 0x0000007f, 0x00000020,
+	0x2542, 0x00010000, 0x00010000,
+	0x2b05, 0x000003ff, 0x000000f3,
+	0x2b03, 0xffffffff, 0x00001032
+};
+
+static const u32 bonaire_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffc,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0xc0000100,
+	0xf0b2, 0xffffffff, 0xc0000100,
+	0xf0b1, 0xffffffff, 0xc0000100,
+	0x1579, 0xffffffff, 0x00600100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf012, 0xffffffff, 0x00010000,
+	0xf013, 0xffffffff, 0x00030002,
+	0xf014, 0xffffffff, 0x00040007,
+	0xf015, 0xffffffff, 0x00060005,
+	0xf016, 0xffffffff, 0x00090008,
+	0xf017, 0xffffffff, 0x00010000,
+	0xf018, 0xffffffff, 0x00030002,
+	0xf019, 0xffffffff, 0x00040007,
+	0xf01a, 0xffffffff, 0x00060005,
+	0xf01b, 0xffffffff, 0x00090008,
+	0xf01c, 0xffffffff, 0x00010000,
+	0xf01d, 0xffffffff, 0x00030002,
+	0xf01e, 0xffffffff, 0x00040007,
+	0xf01f, 0xffffffff, 0x00060005,
+	0xf020, 0xffffffff, 0x00090008,
+	0xf021, 0xffffffff, 0x00010000,
+	0xf022, 0xffffffff, 0x00030002,
+	0xf023, 0xffffffff, 0x00040007,
+	0xf024, 0xffffffff, 0x00060005,
+	0xf025, 0xffffffff, 0x00090008,
+	0xf026, 0xffffffff, 0x00010000,
+	0xf027, 0xffffffff, 0x00030002,
+	0xf028, 0xffffffff, 0x00040007,
+	0xf029, 0xffffffff, 0x00060005,
+	0xf02a, 0xffffffff, 0x00090008,
+	0xf000, 0xffffffff, 0x96e00200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x3e4, 0xffffffff, 0x00000100,
+	0x3e6, 0x00000101, 0x00000000,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 spectre_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 spectre_golden_common_registers[] =
+{
+	0x31dc, 0xffffffff, 0x00000800,
+	0x31dd, 0xffffffff, 0x00000800,
+	0x31e6, 0xffffffff, 0x00007fbf,
+	0x31e7, 0xffffffff, 0x00007faf
+};
+
+static const u32 spectre_golden_registers[] =
+{
+	0xf000, 0xffff1fff, 0x96940200,
+	0xf003, 0xffff0001, 0xff000000,
+	0xf080, 0xfffc0fff, 0x00000100,
+	0x1bb6, 0x00010101, 0x00010000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x260e, 0xfffffffc, 0x00020200,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0x26df, 0x00ff0000, 0x00fc0000,
+	0xbd2, 0x73773777, 0x12010001,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22c9, 0xffffffff, 0x00ffffff,
+	0xa0d4, 0x3f3f3fff, 0x00000082,
+	0xa0d5, 0x0000003f, 0x00000000,
+	0xf9e, 0x00000001, 0x00000002,
+	0x244f, 0xffff03df, 0x00000004,
+	0x31da, 0x00000008, 0x00000008,
+	0x2300, 0x000008ff, 0x00000800,
+	0x2542, 0x00010000, 0x00010000,
+	0x2b03, 0xffffffff, 0x54763210,
+	0x853e, 0x01ff01ff, 0x00000002,
+	0x8526, 0x007ff800, 0x00200000,
+	0x8057, 0xffffffff, 0x00000f40,
+	0xc24d, 0xffffffff, 0x00000001
+};
+
+static const u32 spectre_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffc,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0x00000100,
+	0xf0b2, 0xffffffff, 0x00000100,
+	0xf0b1, 0xffffffff, 0x00000100,
+	0x1579, 0xffffffff, 0x00600100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf012, 0xffffffff, 0x00010000,
+	0xf013, 0xffffffff, 0x00030002,
+	0xf014, 0xffffffff, 0x00040007,
+	0xf015, 0xffffffff, 0x00060005,
+	0xf016, 0xffffffff, 0x00090008,
+	0xf017, 0xffffffff, 0x00010000,
+	0xf018, 0xffffffff, 0x00030002,
+	0xf019, 0xffffffff, 0x00040007,
+	0xf01a, 0xffffffff, 0x00060005,
+	0xf01b, 0xffffffff, 0x00090008,
+	0xf01c, 0xffffffff, 0x00010000,
+	0xf01d, 0xffffffff, 0x00030002,
+	0xf01e, 0xffffffff, 0x00040007,
+	0xf01f, 0xffffffff, 0x00060005,
+	0xf020, 0xffffffff, 0x00090008,
+	0xf021, 0xffffffff, 0x00010000,
+	0xf022, 0xffffffff, 0x00030002,
+	0xf023, 0xffffffff, 0x00040007,
+	0xf024, 0xffffffff, 0x00060005,
+	0xf025, 0xffffffff, 0x00090008,
+	0xf026, 0xffffffff, 0x00010000,
+	0xf027, 0xffffffff, 0x00030002,
+	0xf028, 0xffffffff, 0x00040007,
+	0xf029, 0xffffffff, 0x00060005,
+	0xf02a, 0xffffffff, 0x00090008,
+	0xf02b, 0xffffffff, 0x00010000,
+	0xf02c, 0xffffffff, 0x00030002,
+	0xf02d, 0xffffffff, 0x00040007,
+	0xf02e, 0xffffffff, 0x00060005,
+	0xf02f, 0xffffffff, 0x00090008,
+	0xf000, 0xffffffff, 0x96e00200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x3e4, 0xffffffff, 0x00000100,
+	0x3e6, 0x00000101, 0x00000000,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 kalindi_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 kalindi_golden_common_registers[] =
+{
+	0x31dc, 0xffffffff, 0x00000800,
+	0x31dd, 0xffffffff, 0x00000800,
+	0x31e6, 0xffffffff, 0x00007fbf,
+	0x31e7, 0xffffffff, 0x00007faf
+};
+
+static const u32 kalindi_golden_registers[] =
+{
+	0xf000, 0xffffdfff, 0x6e944040,
+	0x1579, 0xff607fff, 0xfc000100,
+	0xf088, 0xff000fff, 0x00000100,
+	0xf089, 0xff000fff, 0x00000100,
+	0xf080, 0xfffc0fff, 0x00000100,
+	0x1bb6, 0x00010101, 0x00010000,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0x263f, 0xffffffff, 0x00000010,
+	0x26df, 0x00ff0000, 0x00fc0000,
+	0x200c, 0x00001f0f, 0x0000100a,
+	0xbd2, 0x73773777, 0x12010001,
+	0x902, 0x000fffff, 0x000c007f,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22c9, 0x3fff3fff, 0x00ffcfff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0x136, 0x00000fff, 0x00000100,
+	0xf9e, 0x00000001, 0x00000002,
+	0x31da, 0x00000008, 0x00000008,
+	0x2300, 0x000000ff, 0x00000003,
+	0x853e, 0x01ff01ff, 0x00000002,
+	0x8526, 0x007ff800, 0x00200000,
+	0x8057, 0xffffffff, 0x00000f40,
+	0x2231, 0x001f3ae3, 0x00000082,
+	0x2235, 0x0000001f, 0x00000010,
+	0xc24d, 0xffffffff, 0x00000000
+};
+
+static const u32 kalindi_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffc,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0x00000100,
+	0xf0b2, 0xffffffff, 0x00000100,
+	0xf0b1, 0xffffffff, 0x00000100,
+	0x1579, 0xffffffff, 0x00600100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf000, 0xffffffff, 0x96e00200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 hawaii_golden_spm_registers[] =
+{
+	0xc200, 0xe0ffffff, 0xe0000000
+};
+
+static const u32 hawaii_golden_common_registers[] =
+{
+	0xc200, 0xffffffff, 0xe0000000,
+	0xa0d4, 0xffffffff, 0x3a00161a,
+	0xa0d5, 0xffffffff, 0x0000002e,
+	0x2684, 0xffffffff, 0x00018208,
+	0x263e, 0xffffffff, 0x12011003
+};
+
+static const u32 hawaii_golden_registers[] =
+{
+	0xcd5, 0x00000333, 0x00000333,
+	0x2684, 0x00010000, 0x00058208,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260d, 0xf00fffff, 0x00000400,
+	0x260e, 0x0002021c, 0x00020200,
+	0x31e, 0x00000080, 0x00000000,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0xd43, 0x00810000, 0x408af000,
+	0x1c0c, 0x31000111, 0x00000011,
+	0xbd2, 0x73773777, 0x12010001,
+	0x848, 0x0000007f, 0x0000001b,
+	0x877, 0x00007fb6, 0x00002191,
+	0xd8a, 0x0000003f, 0x0000000a,
+	0xd8b, 0x0000003f, 0x0000000a,
+	0xab9, 0x00073ffe, 0x000022a2,
+	0x903, 0x000007ff, 0x00000000,
+	0x22fc, 0x00002001, 0x00000001,
+	0x22c9, 0xffffffff, 0x00ffffff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0xf9e, 0x00000001, 0x00000002,
+	0x31da, 0x00000008, 0x00000008,
+	0x31dc, 0x00000f00, 0x00000800,
+	0x31dd, 0x00000f00, 0x00000800,
+	0x31e6, 0x00ffffff, 0x00ff7fbf,
+	0x31e7, 0x00ffffff, 0x00ff7faf,
+	0x2300, 0x000000ff, 0x00000800,
+	0x390, 0x00001fff, 0x00001fff,
+	0x2418, 0x0000007f, 0x00000020,
+	0x2542, 0x00010000, 0x00010000,
+	0x2b80, 0x00100000, 0x000ff07c,
+	0x2b05, 0x000003ff, 0x0000000f,
+	0x2b04, 0xffffffff, 0x7564fdec,
+	0x2b03, 0xffffffff, 0x3120b9a8,
+	0x2b02, 0x20000000, 0x0f9c0000
+};
+
+static const u32 hawaii_mgcg_cgcg_init[] =
+{
+	0x3108, 0xffffffff, 0xfffffffd,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf0a8, 0xffffffff, 0x00000100,
+	0xf082, 0xffffffff, 0x00000100,
+	0xf0b0, 0xffffffff, 0x00000100,
+	0xf0b2, 0xffffffff, 0x00000100,
+	0xf0b1, 0xffffffff, 0x00000100,
+	0x1579, 0xffffffff, 0x00200100,
+	0xf0a0, 0xffffffff, 0x00000100,
+	0xf085, 0xffffffff, 0x06000100,
+	0xf088, 0xffffffff, 0x00000100,
+	0xf086, 0xffffffff, 0x06000100,
+	0xf081, 0xffffffff, 0x00000100,
+	0xf0b8, 0xffffffff, 0x00000100,
+	0xf089, 0xffffffff, 0x00000100,
+	0xf080, 0xffffffff, 0x00000100,
+	0xf08c, 0xffffffff, 0x00000100,
+	0xf08d, 0xffffffff, 0x00000100,
+	0xf094, 0xffffffff, 0x00000100,
+	0xf095, 0xffffffff, 0x00000100,
+	0xf096, 0xffffffff, 0x00000100,
+	0xf097, 0xffffffff, 0x00000100,
+	0xf098, 0xffffffff, 0x00000100,
+	0xf09f, 0xffffffff, 0x00000100,
+	0xf09e, 0xffffffff, 0x00000100,
+	0xf084, 0xffffffff, 0x06000100,
+	0xf0a4, 0xffffffff, 0x00000100,
+	0xf09d, 0xffffffff, 0x00000100,
+	0xf0ad, 0xffffffff, 0x00000100,
+	0xf0ac, 0xffffffff, 0x00000100,
+	0xf09c, 0xffffffff, 0x00000100,
+	0xc200, 0xffffffff, 0xe0000000,
+	0xf008, 0xffffffff, 0x00010000,
+	0xf009, 0xffffffff, 0x00030002,
+	0xf00a, 0xffffffff, 0x00040007,
+	0xf00b, 0xffffffff, 0x00060005,
+	0xf00c, 0xffffffff, 0x00090008,
+	0xf00d, 0xffffffff, 0x00010000,
+	0xf00e, 0xffffffff, 0x00030002,
+	0xf00f, 0xffffffff, 0x00040007,
+	0xf010, 0xffffffff, 0x00060005,
+	0xf011, 0xffffffff, 0x00090008,
+	0xf012, 0xffffffff, 0x00010000,
+	0xf013, 0xffffffff, 0x00030002,
+	0xf014, 0xffffffff, 0x00040007,
+	0xf015, 0xffffffff, 0x00060005,
+	0xf016, 0xffffffff, 0x00090008,
+	0xf017, 0xffffffff, 0x00010000,
+	0xf018, 0xffffffff, 0x00030002,
+	0xf019, 0xffffffff, 0x00040007,
+	0xf01a, 0xffffffff, 0x00060005,
+	0xf01b, 0xffffffff, 0x00090008,
+	0xf01c, 0xffffffff, 0x00010000,
+	0xf01d, 0xffffffff, 0x00030002,
+	0xf01e, 0xffffffff, 0x00040007,
+	0xf01f, 0xffffffff, 0x00060005,
+	0xf020, 0xffffffff, 0x00090008,
+	0xf021, 0xffffffff, 0x00010000,
+	0xf022, 0xffffffff, 0x00030002,
+	0xf023, 0xffffffff, 0x00040007,
+	0xf024, 0xffffffff, 0x00060005,
+	0xf025, 0xffffffff, 0x00090008,
+	0xf026, 0xffffffff, 0x00010000,
+	0xf027, 0xffffffff, 0x00030002,
+	0xf028, 0xffffffff, 0x00040007,
+	0xf029, 0xffffffff, 0x00060005,
+	0xf02a, 0xffffffff, 0x00090008,
+	0xf02b, 0xffffffff, 0x00010000,
+	0xf02c, 0xffffffff, 0x00030002,
+	0xf02d, 0xffffffff, 0x00040007,
+	0xf02e, 0xffffffff, 0x00060005,
+	0xf02f, 0xffffffff, 0x00090008,
+	0xf030, 0xffffffff, 0x00010000,
+	0xf031, 0xffffffff, 0x00030002,
+	0xf032, 0xffffffff, 0x00040007,
+	0xf033, 0xffffffff, 0x00060005,
+	0xf034, 0xffffffff, 0x00090008,
+	0xf035, 0xffffffff, 0x00010000,
+	0xf036, 0xffffffff, 0x00030002,
+	0xf037, 0xffffffff, 0x00040007,
+	0xf038, 0xffffffff, 0x00060005,
+	0xf039, 0xffffffff, 0x00090008,
+	0xf03a, 0xffffffff, 0x00010000,
+	0xf03b, 0xffffffff, 0x00030002,
+	0xf03c, 0xffffffff, 0x00040007,
+	0xf03d, 0xffffffff, 0x00060005,
+	0xf03e, 0xffffffff, 0x00090008,
+	0x30c6, 0xffffffff, 0x00020200,
+	0xcd4, 0xffffffff, 0x00000200,
+	0x570, 0xffffffff, 0x00000400,
+	0x157a, 0xffffffff, 0x00000000,
+	0xbd4, 0xffffffff, 0x00000902,
+	0xf000, 0xffffffff, 0x96940200,
+	0x21c2, 0xffffffff, 0x00900100,
+	0x3109, 0xffffffff, 0x0020003f,
+	0xe, 0xffffffff, 0x0140001c,
+	0xf, 0x000f0000, 0x000f0000,
+	0x88, 0xffffffff, 0xc060000c,
+	0x89, 0xc0000fff, 0x00000100,
+	0x3e4, 0xffffffff, 0x00000100,
+	0x3e6, 0x00000101, 0x00000000,
+	0x82a, 0xffffffff, 0x00000104,
+	0x1579, 0xff000fff, 0x00000100,
+	0xc33, 0xc0000fff, 0x00000104,
+	0x3079, 0x00000001, 0x00000001,
+	0x3403, 0xff000ff0, 0x00000100,
+	0x3603, 0xff000ff0, 0x00000100
+};
+
+static const u32 godavari_golden_registers[] =
+{
+	0x1579, 0xff607fff, 0xfc000100,
+	0x1bb6, 0x00010101, 0x00010000,
+	0x260c, 0xffffffff, 0x00000000,
+	0x260c0, 0xf00fffff, 0x00000400,
+	0x184c, 0xffffffff, 0x00010000,
+	0x16ec, 0x000000f0, 0x00000070,
+	0x16f0, 0xf0311fff, 0x80300000,
+	0x263e, 0x73773777, 0x12010001,
+	0x263f, 0xffffffff, 0x00000010,
+	0x200c, 0x00001f0f, 0x0000100a,
+	0xbd2, 0x73773777, 0x12010001,
+	0x902, 0x000fffff, 0x000c007f,
+	0x2285, 0xf000003f, 0x00000007,
+	0x22c9, 0xffffffff, 0x00ff0fff,
+	0xc281, 0x0000ff0f, 0x00000000,
+	0xa293, 0x07ffffff, 0x06000000,
+	0x136, 0x00000fff, 0x00000100,
+	0x3405, 0x00010000, 0x00810001,
+	0x3605, 0x00010000, 0x00810001,
+	0xf9e, 0x00000001, 0x00000002,
+	0x31da, 0x00000008, 0x00000008,
+	0x31dc, 0x00000f00, 0x00000800,
+	0x31dd, 0x00000f00, 0x00000800,
+	0x31e6, 0x00ffffff, 0x00ff7fbf,
+	0x31e7, 0x00ffffff, 0x00ff7faf,
+	0x2300, 0x000000ff, 0x00000001,
+	0x853e, 0x01ff01ff, 0x00000002,
+	0x8526, 0x007ff800, 0x00200000,
+	0x8057, 0xffffffff, 0x00000f40,
+	0x2231, 0x001f3ae3, 0x00000082,
+	0x2235, 0x0000001f, 0x00000010,
+	0xc24d, 0xffffffff, 0x00000000
+};
+
+static void cik_init_golden_registers(struct amdgpu_device *adev)
+{
+	/* Some of the registers might be dependent on GRBM_GFX_INDEX */
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		amdgpu_program_register_sequence(adev,
+						 bonaire_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(bonaire_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 bonaire_golden_registers,
+						 (const u32)ARRAY_SIZE(bonaire_golden_registers));
+		amdgpu_program_register_sequence(adev,
+						 bonaire_golden_common_registers,
+						 (const u32)ARRAY_SIZE(bonaire_golden_common_registers));
+		amdgpu_program_register_sequence(adev,
+						 bonaire_golden_spm_registers,
+						 (const u32)ARRAY_SIZE(bonaire_golden_spm_registers));
+		break;
+	case CHIP_KABINI:
+		amdgpu_program_register_sequence(adev,
+						 kalindi_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(kalindi_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 kalindi_golden_registers,
+						 (const u32)ARRAY_SIZE(kalindi_golden_registers));
+		amdgpu_program_register_sequence(adev,
+						 kalindi_golden_common_registers,
+						 (const u32)ARRAY_SIZE(kalindi_golden_common_registers));
+		amdgpu_program_register_sequence(adev,
+						 kalindi_golden_spm_registers,
+						 (const u32)ARRAY_SIZE(kalindi_golden_spm_registers));
+		break;
+	case CHIP_MULLINS:
+		amdgpu_program_register_sequence(adev,
+						 kalindi_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(kalindi_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 godavari_golden_registers,
+						 (const u32)ARRAY_SIZE(godavari_golden_registers));
+		amdgpu_program_register_sequence(adev,
+						 kalindi_golden_common_registers,
+						 (const u32)ARRAY_SIZE(kalindi_golden_common_registers));
+		amdgpu_program_register_sequence(adev,
+						 kalindi_golden_spm_registers,
+						 (const u32)ARRAY_SIZE(kalindi_golden_spm_registers));
+		break;
+	case CHIP_KAVERI:
+		amdgpu_program_register_sequence(adev,
+						 spectre_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(spectre_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 spectre_golden_registers,
+						 (const u32)ARRAY_SIZE(spectre_golden_registers));
+		amdgpu_program_register_sequence(adev,
+						 spectre_golden_common_registers,
+						 (const u32)ARRAY_SIZE(spectre_golden_common_registers));
+		amdgpu_program_register_sequence(adev,
+						 spectre_golden_spm_registers,
+						 (const u32)ARRAY_SIZE(spectre_golden_spm_registers));
+		break;
+	case CHIP_HAWAII:
+		amdgpu_program_register_sequence(adev,
+						 hawaii_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(hawaii_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 hawaii_golden_registers,
+						 (const u32)ARRAY_SIZE(hawaii_golden_registers));
+		amdgpu_program_register_sequence(adev,
+						 hawaii_golden_common_registers,
+						 (const u32)ARRAY_SIZE(hawaii_golden_common_registers));
+		amdgpu_program_register_sequence(adev,
+						 hawaii_golden_spm_registers,
+						 (const u32)ARRAY_SIZE(hawaii_golden_spm_registers));
+		break;
+	default:
+		break;
+	}
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * cik_get_xclk - get the xclk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns the reference clock used by the gfx engine
+ * (CIK).
+ */
+static u32 cik_get_xclk(struct amdgpu_device *adev)
+{
+	u32 reference_clock = adev->clock.spll.reference_freq;
+
+	if (adev->flags & AMD_IS_APU) {
+		if (RREG32_SMC(ixGENERAL_PWRMGT) & GENERAL_PWRMGT__GPU_COUNTER_CLK_MASK)
+			return reference_clock / 2;
+	} else {
+		if (RREG32_SMC(ixCG_CLKPIN_CNTL) & CG_CLKPIN_CNTL__XTALIN_DIVIDE_MASK)
+			return reference_clock / 4;
+	}
+	return reference_clock;
+}
+
+/**
+ * cik_srbm_select - select specific register instances
+ *
+ * @adev: amdgpu_device pointer
+ * @me: selected ME (micro engine)
+ * @pipe: pipe
+ * @queue: queue
+ * @vmid: VMID
+ *
+ * Switches the currently active registers instances.  Some
+ * registers are instanced per VMID, others are instanced per
+ * me/pipe/queue combination.
+ */
+void cik_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 srbm_gfx_cntl =
+		(((pipe << SRBM_GFX_CNTL__PIPEID__SHIFT) & SRBM_GFX_CNTL__PIPEID_MASK)|
+		((me << SRBM_GFX_CNTL__MEID__SHIFT) & SRBM_GFX_CNTL__MEID_MASK)|
+		((vmid << SRBM_GFX_CNTL__VMID__SHIFT) & SRBM_GFX_CNTL__VMID_MASK)|
+		((queue << SRBM_GFX_CNTL__QUEUEID__SHIFT) & SRBM_GFX_CNTL__QUEUEID_MASK));
+	WREG32(mmSRBM_GFX_CNTL, srbm_gfx_cntl);
+}
+
+static void cik_vga_set_state(struct amdgpu_device *adev, bool state)
+{
+	uint32_t tmp;
+
+	tmp = RREG32(mmCONFIG_CNTL);
+	if (state == false)
+		tmp |= CONFIG_CNTL__VGA_DIS_MASK;
+	else
+		tmp &= ~CONFIG_CNTL__VGA_DIS_MASK;
+	WREG32(mmCONFIG_CNTL, tmp);
+}
+
+static bool cik_read_disabled_bios(struct amdgpu_device *adev)
+{
+	u32 bus_cntl;
+	u32 d1vga_control = 0;
+	u32 d2vga_control = 0;
+	u32 vga_render_control = 0;
+	u32 rom_cntl;
+	bool r;
+
+	bus_cntl = RREG32(mmBUS_CNTL);
+	if (adev->mode_info.num_crtc) {
+		d1vga_control = RREG32(mmD1VGA_CONTROL);
+		d2vga_control = RREG32(mmD2VGA_CONTROL);
+		vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	}
+	rom_cntl = RREG32_SMC(ixROM_CNTL);
+
+	/* enable the rom */
+	WREG32(mmBUS_CNTL, (bus_cntl & ~BUS_CNTL__BIOS_ROM_DIS_MASK));
+	if (adev->mode_info.num_crtc) {
+		/* Disable VGA mode */
+		WREG32(mmD1VGA_CONTROL,
+		       (d1vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmD2VGA_CONTROL,
+		       (d2vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmVGA_RENDER_CONTROL,
+		       (vga_render_control & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK));
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl | ROM_CNTL__SCK_OVERWRITE_MASK);
+
+	r = amdgpu_read_bios(adev);
+
+	/* restore regs */
+	WREG32(mmBUS_CNTL, bus_cntl);
+	if (adev->mode_info.num_crtc) {
+		WREG32(mmD1VGA_CONTROL, d1vga_control);
+		WREG32(mmD2VGA_CONTROL, d2vga_control);
+		WREG32(mmVGA_RENDER_CONTROL, vga_render_control);
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl);
+	return r;
+}
+
+static struct amdgpu_allowed_register_entry cik_allowed_read_registers[] = {
+	{mmGRBM_STATUS, false},
+	{mmGB_ADDR_CONFIG, false},
+	{mmMC_ARB_RAMCFG, false},
+	{mmGB_TILE_MODE0, false},
+	{mmGB_TILE_MODE1, false},
+	{mmGB_TILE_MODE2, false},
+	{mmGB_TILE_MODE3, false},
+	{mmGB_TILE_MODE4, false},
+	{mmGB_TILE_MODE5, false},
+	{mmGB_TILE_MODE6, false},
+	{mmGB_TILE_MODE7, false},
+	{mmGB_TILE_MODE8, false},
+	{mmGB_TILE_MODE9, false},
+	{mmGB_TILE_MODE10, false},
+	{mmGB_TILE_MODE11, false},
+	{mmGB_TILE_MODE12, false},
+	{mmGB_TILE_MODE13, false},
+	{mmGB_TILE_MODE14, false},
+	{mmGB_TILE_MODE15, false},
+	{mmGB_TILE_MODE16, false},
+	{mmGB_TILE_MODE17, false},
+	{mmGB_TILE_MODE18, false},
+	{mmGB_TILE_MODE19, false},
+	{mmGB_TILE_MODE20, false},
+	{mmGB_TILE_MODE21, false},
+	{mmGB_TILE_MODE22, false},
+	{mmGB_TILE_MODE23, false},
+	{mmGB_TILE_MODE24, false},
+	{mmGB_TILE_MODE25, false},
+	{mmGB_TILE_MODE26, false},
+	{mmGB_TILE_MODE27, false},
+	{mmGB_TILE_MODE28, false},
+	{mmGB_TILE_MODE29, false},
+	{mmGB_TILE_MODE30, false},
+	{mmGB_TILE_MODE31, false},
+	{mmGB_MACROTILE_MODE0, false},
+	{mmGB_MACROTILE_MODE1, false},
+	{mmGB_MACROTILE_MODE2, false},
+	{mmGB_MACROTILE_MODE3, false},
+	{mmGB_MACROTILE_MODE4, false},
+	{mmGB_MACROTILE_MODE5, false},
+	{mmGB_MACROTILE_MODE6, false},
+	{mmGB_MACROTILE_MODE7, false},
+	{mmGB_MACROTILE_MODE8, false},
+	{mmGB_MACROTILE_MODE9, false},
+	{mmGB_MACROTILE_MODE10, false},
+	{mmGB_MACROTILE_MODE11, false},
+	{mmGB_MACROTILE_MODE12, false},
+	{mmGB_MACROTILE_MODE13, false},
+	{mmGB_MACROTILE_MODE14, false},
+	{mmGB_MACROTILE_MODE15, false},
+	{mmCC_RB_BACKEND_DISABLE, false, true},
+	{mmGC_USER_RB_BACKEND_DISABLE, false, true},
+	{mmGB_BACKEND_MAP, false, false},
+	{mmPA_SC_RASTER_CONFIG, false, true},
+	{mmPA_SC_RASTER_CONFIG_1, false, true},
+};
+
+static uint32_t cik_read_indexed_register(struct amdgpu_device *adev,
+					  u32 se_num, u32 sh_num,
+					  u32 reg_offset)
+{
+	uint32_t val;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		gfx_v7_0_select_se_sh(adev, se_num, sh_num);
+
+	val = RREG32(reg_offset);
+
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return val;
+}
+
+static int cik_read_register(struct amdgpu_device *adev, u32 se_num,
+			     u32 sh_num, u32 reg_offset, u32 *value)
+{
+	uint32_t i;
+
+	*value = 0;
+	for (i = 0; i < ARRAY_SIZE(cik_allowed_read_registers); i++) {
+		if (reg_offset != cik_allowed_read_registers[i].reg_offset)
+			continue;
+
+		if (!cik_allowed_read_registers[i].untouched)
+			*value = cik_allowed_read_registers[i].grbm_indexed ?
+				 cik_read_indexed_register(adev, se_num,
+							   sh_num, reg_offset) :
+				 RREG32(reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static void cik_print_gpu_status_regs(struct amdgpu_device *adev)
+{
+	dev_info(adev->dev, "  GRBM_STATUS=0x%08X\n",
+		RREG32(mmGRBM_STATUS));
+	dev_info(adev->dev, "  GRBM_STATUS2=0x%08X\n",
+		RREG32(mmGRBM_STATUS2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE0=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE0));
+	dev_info(adev->dev, "  GRBM_STATUS_SE1=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE1));
+	dev_info(adev->dev, "  GRBM_STATUS_SE2=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE3=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE3));
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+	dev_info(adev->dev, "  SDMA0_STATUS_REG   = 0x%08X\n",
+		RREG32(mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET));
+	dev_info(adev->dev, "  SDMA1_STATUS_REG   = 0x%08X\n",
+		 RREG32(mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET));
+	dev_info(adev->dev, "  CP_STAT = 0x%08x\n", RREG32(mmCP_STAT));
+	dev_info(adev->dev, "  CP_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_STALLED_STAT2 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT2));
+	dev_info(adev->dev, "  CP_STALLED_STAT3 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT3));
+	dev_info(adev->dev, "  CP_CPF_BUSY_STAT = 0x%08x\n",
+		 RREG32(mmCP_CPF_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPF_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPF_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPF_STATUS = 0x%08x\n", RREG32(mmCP_CPF_STATUS));
+	dev_info(adev->dev, "  CP_CPC_BUSY_STAT = 0x%08x\n", RREG32(mmCP_CPC_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPC_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPC_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPC_STATUS = 0x%08x\n", RREG32(mmCP_CPC_STATUS));
+}
+
+/**
+ * cik_gpu_check_soft_reset - check which blocks are busy
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check which blocks are busy and return the relevant reset
+ * mask to be used by cik_gpu_soft_reset().
+ * Returns a mask of the blocks to be reset.
+ */
+u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev)
+{
+	u32 reset_mask = 0;
+	u32 tmp;
+
+	/* GRBM_STATUS */
+	tmp = RREG32(mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK))
+		reset_mask |= AMDGPU_RESET_GFX;
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK))
+		reset_mask |= AMDGPU_RESET_CP;
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32(mmGRBM_STATUS2);
+	if (tmp & GRBM_STATUS2__RLC_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_RLC;
+
+	/* SDMA0_STATUS_REG */
+	tmp = RREG32(mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET);
+	if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+		reset_mask |= AMDGPU_RESET_DMA;
+
+	/* SDMA1_STATUS_REG */
+	tmp = RREG32(mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET);
+	if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+		reset_mask |= AMDGPU_RESET_DMA1;
+
+	/* SRBM_STATUS2 */
+	tmp = RREG32(mmSRBM_STATUS2);
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_DMA;
+
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_DMA1;
+
+	/* SRBM_STATUS */
+	tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_IH;
+
+	if (tmp & SRBM_STATUS__SEM_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_SEM;
+
+	if (tmp & SRBM_STATUS__GRBM_RQ_PENDING_MASK)
+		reset_mask |= AMDGPU_RESET_GRBM;
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_VMC;
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK))
+		reset_mask |= AMDGPU_RESET_MC;
+
+	if (amdgpu_display_is_display_hung(adev))
+		reset_mask |= AMDGPU_RESET_DISPLAY;
+
+	/* Skip MC reset as it's mostly likely not hung, just busy */
+	if (reset_mask & AMDGPU_RESET_MC) {
+		DRM_DEBUG("MC busy: 0x%08X, clearing.\n", reset_mask);
+		reset_mask &= ~AMDGPU_RESET_MC;
+	}
+
+	return reset_mask;
+}
+
+/**
+ * cik_gpu_soft_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ * @reset_mask: mask of which blocks to reset
+ *
+ * Soft reset the blocks specified in @reset_mask.
+ */
+static void cik_gpu_soft_reset(struct amdgpu_device *adev, u32 reset_mask)
+{
+	struct amdgpu_mode_mc_save save;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+
+	if (reset_mask == 0)
+		return;
+
+	dev_info(adev->dev, "GPU softreset: 0x%08X\n", reset_mask);
+
+	cik_print_gpu_status_regs(adev);
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS));
+
+	/* disable CG/PG */
+
+	/* stop the rlc */
+	gfx_v7_0_rlc_stop(adev);
+
+	/* Disable GFX parsing/prefetching */
+	WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK);
+
+	/* Disable MEC parsing/prefetching */
+	WREG32(mmCP_MEC_CNTL, CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK);
+
+	if (reset_mask & AMDGPU_RESET_DMA) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+	}
+	if (reset_mask & AMDGPU_RESET_DMA1) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+	}
+
+	gmc_v7_0_mc_stop(adev, &save);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+
+	if (reset_mask & (AMDGPU_RESET_GFX | AMDGPU_RESET_COMPUTE | AMDGPU_RESET_CP))
+		grbm_soft_reset = GRBM_SOFT_RESET__SOFT_RESET_CP_MASK |
+			GRBM_SOFT_RESET__SOFT_RESET_GFX_MASK;
+
+	if (reset_mask & AMDGPU_RESET_CP) {
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_CP_MASK;
+
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_GRBM_MASK;
+	}
+
+	if (reset_mask & AMDGPU_RESET_DMA)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+
+	if (reset_mask & AMDGPU_RESET_DMA1)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+
+	if (reset_mask & AMDGPU_RESET_DISPLAY)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (reset_mask & AMDGPU_RESET_RLC)
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+
+	if (reset_mask & AMDGPU_RESET_SEM)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SEM_MASK;
+
+	if (reset_mask & AMDGPU_RESET_IH)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_IH_MASK;
+
+	if (reset_mask & AMDGPU_RESET_GRBM)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_GRBM_MASK;
+
+	if (reset_mask & AMDGPU_RESET_VMC)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_VMC_MASK;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		if (reset_mask & AMDGPU_RESET_MC)
+			srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_MC_MASK;
+	}
+
+	if (grbm_soft_reset) {
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~grbm_soft_reset;
+		WREG32(mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+	}
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+	}
+
+	/* Wait a little for things to settle down */
+	udelay(50);
+
+	gmc_v7_0_mc_resume(adev, &save);
+	udelay(50);
+
+	cik_print_gpu_status_regs(adev);
+}
+
+struct kv_reset_save_regs {
+	u32 gmcon_reng_execute;
+	u32 gmcon_misc;
+	u32 gmcon_misc3;
+};
+
+static void kv_save_regs_for_reset(struct amdgpu_device *adev,
+				   struct kv_reset_save_regs *save)
+{
+	save->gmcon_reng_execute = RREG32(mmGMCON_RENG_EXECUTE);
+	save->gmcon_misc = RREG32(mmGMCON_MISC);
+	save->gmcon_misc3 = RREG32(mmGMCON_MISC3);
+
+	WREG32(mmGMCON_RENG_EXECUTE, save->gmcon_reng_execute &
+		~GMCON_RENG_EXECUTE__RENG_EXECUTE_ON_PWR_UP_MASK);
+	WREG32(mmGMCON_MISC, save->gmcon_misc &
+		~(GMCON_MISC__RENG_EXECUTE_ON_REG_UPDATE_MASK |
+			GMCON_MISC__STCTRL_STUTTER_EN_MASK));
+}
+
+static void kv_restore_regs_for_reset(struct amdgpu_device *adev,
+				      struct kv_reset_save_regs *save)
+{
+	int i;
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x200010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x300010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x210000);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xa00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x21003);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xb00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x2b00);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xc00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xd00010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x420000);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x100010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x120202);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x500010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x3e3e36);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x600010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x373f3e);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0x700010ff);
+
+	for (i = 0; i < 5; i++)
+		WREG32(mmGMCON_PGFSM_WRITE, 0);
+
+	WREG32(mmGMCON_PGFSM_WRITE, 0x3e1332);
+	WREG32(mmGMCON_PGFSM_CONFIG, 0xe00010ff);
+
+	WREG32(mmGMCON_MISC3, save->gmcon_misc3);
+	WREG32(mmGMCON_MISC, save->gmcon_misc);
+	WREG32(mmGMCON_RENG_EXECUTE, save->gmcon_reng_execute);
+}
+
+static void cik_gpu_pci_config_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_mc_save save;
+	struct kv_reset_save_regs kv_save = { 0 };
+	u32 tmp, i;
+
+	dev_info(adev->dev, "GPU pci config reset\n");
+
+	/* disable dpm? */
+
+	/* disable cg/pg */
+
+	/* Disable GFX parsing/prefetching */
+	WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK |
+		CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK);
+
+	/* Disable MEC parsing/prefetching */
+	WREG32(mmCP_MEC_CNTL,
+			CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK);
+
+	/* sdma0 */
+	tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+	tmp |= SDMA0_F32_CNTL__HALT_MASK;
+	WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+	/* sdma1 */
+	tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+	tmp |= SDMA0_F32_CNTL__HALT_MASK;
+	WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+	/* XXX other engines? */
+
+	/* halt the rlc, disable cp internal ints */
+	gfx_v7_0_rlc_stop(adev);
+
+	udelay(50);
+
+	/* disable mem access */
+	gmc_v7_0_mc_stop(adev, &save);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timed out !\n");
+	}
+
+	if (adev->flags & AMD_IS_APU)
+		kv_save_regs_for_reset(adev, &kv_save);
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	/* reset */
+	amdgpu_pci_config_reset(adev);
+
+	udelay(100);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff)
+			break;
+		udelay(1);
+	}
+
+	/* does asic init need to be run first??? */
+	if (adev->flags & AMD_IS_APU)
+		kv_restore_regs_for_reset(adev, &kv_save);
+}
+
+static void cik_set_bios_scratch_engine_hung(struct amdgpu_device *adev, bool hung)
+{
+	u32 tmp = RREG32(mmBIOS_SCRATCH_3);
+
+	if (hung)
+		tmp |= ATOM_S3_ASIC_GUI_ENGINE_HUNG;
+	else
+		tmp &= ~ATOM_S3_ASIC_GUI_ENGINE_HUNG;
+
+	WREG32(mmBIOS_SCRATCH_3, tmp);
+}
+
+/**
+ * cik_asic_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up which blocks are hung and attempt
+ * to reset them.
+ * Returns 0 for success.
+ */
+static int cik_asic_reset(struct amdgpu_device *adev)
+{
+	u32 reset_mask;
+
+	reset_mask = amdgpu_cik_gpu_check_soft_reset(adev);
+
+	if (reset_mask)
+		cik_set_bios_scratch_engine_hung(adev, true);
+
+	/* try soft reset */
+	cik_gpu_soft_reset(adev, reset_mask);
+
+	reset_mask = amdgpu_cik_gpu_check_soft_reset(adev);
+
+	/* try pci config reset */
+	if (reset_mask && amdgpu_hard_reset)
+		cik_gpu_pci_config_reset(adev);
+
+	reset_mask = amdgpu_cik_gpu_check_soft_reset(adev);
+
+	if (!reset_mask)
+		cik_set_bios_scratch_engine_hung(adev, false);
+
+	return 0;
+}
+
+static int cik_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
+			      u32 cntl_reg, u32 status_reg)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	uint32_t tmp;
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       clock, false, &dividers);
+	if (r)
+		return r;
+
+	tmp = RREG32_SMC(cntl_reg);
+	tmp &= ~(CG_DCLK_CNTL__DCLK_DIR_CNTL_EN_MASK |
+		CG_DCLK_CNTL__DCLK_DIVIDER_MASK);
+	tmp |= dividers.post_divider;
+	WREG32_SMC(cntl_reg, tmp);
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(status_reg) & CG_DCLK_STATUS__DCLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static int cik_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	int r = 0;
+
+	r = cik_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
+	if (r)
+		return r;
+
+	r = cik_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
+	return r;
+}
+
+static int cik_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	u32 tmp;
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       ecclk, false, &dividers);
+	if (r)
+		return r;
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(ixCG_ECLK_STATUS) & CG_ECLK_STATUS__ECLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	tmp = RREG32_SMC(ixCG_ECLK_CNTL);
+	tmp &= ~(CG_ECLK_CNTL__ECLK_DIR_CNTL_EN_MASK |
+		CG_ECLK_CNTL__ECLK_DIVIDER_MASK);
+	tmp |= dividers.post_divider;
+	WREG32_SMC(ixCG_ECLK_CNTL, tmp);
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(ixCG_ECLK_STATUS) & CG_ECLK_STATUS__ECLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static void cik_pcie_gen3_enable(struct amdgpu_device *adev)
+{
+	struct pci_dev *root = adev->pdev->bus->self;
+	int bridge_pos, gpu_pos;
+	u32 speed_cntl, mask, current_data_rate;
+	int ret, i;
+	u16 tmp16;
+
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	if (amdgpu_pcie_gen2 == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask);
+	if (ret != 0)
+		return;
+
+	if (!(mask & (DRM_PCIE_SPEED_50 | DRM_PCIE_SPEED_80)))
+		return;
+
+	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
+	current_data_rate = (speed_cntl & PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE_MASK) >>
+		PCIE_LC_SPEED_CNTL__LC_CURRENT_DATA_RATE__SHIFT;
+	if (mask & DRM_PCIE_SPEED_80) {
+		if (current_data_rate == 2) {
+			DRM_INFO("PCIE gen 3 link speeds already enabled\n");
+			return;
+		}
+		DRM_INFO("enabling PCIE gen 3 link speeds, disable with amdgpu.pcie_gen2=0\n");
+	} else if (mask & DRM_PCIE_SPEED_50) {
+		if (current_data_rate == 1) {
+			DRM_INFO("PCIE gen 2 link speeds already enabled\n");
+			return;
+		}
+		DRM_INFO("enabling PCIE gen 2 link speeds, disable with amdgpu.pcie_gen2=0\n");
+	}
+
+	bridge_pos = pci_pcie_cap(root);
+	if (!bridge_pos)
+		return;
+
+	gpu_pos = pci_pcie_cap(adev->pdev);
+	if (!gpu_pos)
+		return;
+
+	if (mask & DRM_PCIE_SPEED_80) {
+		/* re-try equalization if gen3 is not already enabled */
+		if (current_data_rate != 2) {
+			u16 bridge_cfg, gpu_cfg;
+			u16 bridge_cfg2, gpu_cfg2;
+			u32 max_lw, current_lw, tmp;
+
+			pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
+			pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+
+			tmp16 = bridge_cfg | PCI_EXP_LNKCTL_HAWD;
+			pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
+
+			tmp16 = gpu_cfg | PCI_EXP_LNKCTL_HAWD;
+			pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+
+			tmp = RREG32_PCIE(ixPCIE_LC_STATUS1);
+			max_lw = (tmp & PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH_MASK) >>
+				PCIE_LC_STATUS1__LC_DETECTED_LINK_WIDTH__SHIFT;
+			current_lw = (tmp & PCIE_LC_STATUS1__LC_OPERATING_LINK_WIDTH_MASK)
+				>> PCIE_LC_STATUS1__LC_OPERATING_LINK_WIDTH__SHIFT;
+
+			if (current_lw < max_lw) {
+				tmp = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL);
+				if (tmp & PCIE_LC_LINK_WIDTH_CNTL__LC_RENEGOTIATION_SUPPORT_MASK) {
+					tmp &= ~(PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH_MASK |
+						PCIE_LC_LINK_WIDTH_CNTL__LC_UPCONFIGURE_DIS_MASK);
+					tmp |= (max_lw <<
+						PCIE_LC_LINK_WIDTH_CNTL__LC_LINK_WIDTH__SHIFT);
+					tmp |= PCIE_LC_LINK_WIDTH_CNTL__LC_UPCONFIGURE_SUPPORT_MASK |
+					PCIE_LC_LINK_WIDTH_CNTL__LC_RENEGOTIATE_EN_MASK |
+					PCIE_LC_LINK_WIDTH_CNTL__LC_RECONFIG_NOW_MASK;
+					WREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL, tmp);
+				}
+			}
+
+			for (i = 0; i < 10; i++) {
+				/* check status */
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_DEVSTA, &tmp16);
+				if (tmp16 & PCI_EXP_DEVSTA_TRPND)
+					break;
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &bridge_cfg);
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &gpu_cfg);
+
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &bridge_cfg2);
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &gpu_cfg2);
+
+				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
+				tmp |= PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
+				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
+
+				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
+				tmp |= PCIE_LC_CNTL4__LC_REDO_EQ_MASK;
+				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
+
+				mdelay(100);
+
+				/* linkctl */
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL, &tmp16);
+				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
+				tmp16 |= (bridge_cfg & PCI_EXP_LNKCTL_HAWD);
+				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL, tmp16);
+
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, &tmp16);
+				tmp16 &= ~PCI_EXP_LNKCTL_HAWD;
+				tmp16 |= (gpu_cfg & PCI_EXP_LNKCTL_HAWD);
+				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL, tmp16);
+
+				/* linkctl2 */
+				pci_read_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, &tmp16);
+				tmp16 &= ~((1 << 4) | (7 << 9));
+				tmp16 |= (bridge_cfg2 & ((1 << 4) | (7 << 9)));
+				pci_write_config_word(root, bridge_pos + PCI_EXP_LNKCTL2, tmp16);
+
+				pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
+				tmp16 &= ~((1 << 4) | (7 << 9));
+				tmp16 |= (gpu_cfg2 & ((1 << 4) | (7 << 9)));
+				pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+
+				tmp = RREG32_PCIE(ixPCIE_LC_CNTL4);
+				tmp &= ~PCIE_LC_CNTL4__LC_SET_QUIESCE_MASK;
+				WREG32_PCIE(ixPCIE_LC_CNTL4, tmp);
+			}
+		}
+	}
+
+	/* set the link speed */
+	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_FORCE_EN_SW_SPEED_CHANGE_MASK |
+		PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_HW_SPEED_CHANGE_MASK;
+	speed_cntl &= ~PCIE_LC_SPEED_CNTL__LC_FORCE_DIS_SW_SPEED_CHANGE_MASK;
+	WREG32_PCIE(ixPCIE_LC_SPEED_CNTL, speed_cntl);
+
+	pci_read_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, &tmp16);
+	tmp16 &= ~0xf;
+	if (mask & DRM_PCIE_SPEED_80)
+		tmp16 |= 3; /* gen3 */
+	else if (mask & DRM_PCIE_SPEED_50)
+		tmp16 |= 2; /* gen2 */
+	else
+		tmp16 |= 1; /* gen1 */
+	pci_write_config_word(adev->pdev, gpu_pos + PCI_EXP_LNKCTL2, tmp16);
+
+	speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
+	speed_cntl |= PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK;
+	WREG32_PCIE(ixPCIE_LC_SPEED_CNTL, speed_cntl);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		speed_cntl = RREG32_PCIE(ixPCIE_LC_SPEED_CNTL);
+		if ((speed_cntl & PCIE_LC_SPEED_CNTL__LC_INITIATE_LINK_SPEED_CHANGE_MASK) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void cik_program_aspm(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+	bool disable_l0s = false, disable_l1 = false, disable_plloff_in_l1 = false;
+	bool disable_clkreq = false;
+
+	if (amdgpu_aspm == 0)
+		return;
+
+	/* XXX double check APUs */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
+	data &= ~PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_MASK;
+	data |= (0x24 << PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS__SHIFT) |
+		PCIE_LC_N_FTS_CNTL__LC_XMIT_N_FTS_OVERRIDE_EN_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_N_FTS_CNTL, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL3);
+	data |= PCIE_LC_CNTL3__LC_GO_TO_RECOVERY_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_LC_CNTL3, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_P_CNTL);
+	data |= PCIE_P_CNTL__P_IGNORE_EDB_ERR_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_P_CNTL, data);
+
+	orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+	data &= ~(PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK |
+		PCIE_LC_CNTL__LC_L1_INACTIVITY_MASK);
+	data |= PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+	if (!disable_l0s)
+		data |= (7 << PCIE_LC_CNTL__LC_L0S_INACTIVITY__SHIFT);
+
+	if (!disable_l1) {
+		data |= (7 << PCIE_LC_CNTL__LC_L1_INACTIVITY__SHIFT);
+		data &= ~PCIE_LC_CNTL__LC_PMI_TO_L1_DIS_MASK;
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL, data);
+
+		if (!disable_plloff_in_l1) {
+			bool clk_req_support;
+
+			orig = data = RREG32_PCIE(ixPB0_PIF_PWRDOWN_0);
+			data &= ~(PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0_MASK |
+				PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= (7 << PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0__SHIFT) |
+				(7 << PB0_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB0_PIF_PWRDOWN_0, data);
+
+			orig = data = RREG32_PCIE(ixPB0_PIF_PWRDOWN_1);
+			data &= ~(PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1_MASK |
+				PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= (7 << PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1__SHIFT) |
+				(7 << PB0_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB0_PIF_PWRDOWN_1, data);
+
+			orig = data = RREG32_PCIE(ixPB1_PIF_PWRDOWN_0);
+			data &= ~(PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0_MASK |
+				PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0_MASK);
+			data |= (7 << PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_OFF_0__SHIFT) |
+				(7 << PB1_PIF_PWRDOWN_0__PLL_POWER_STATE_IN_TXS2_0__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB1_PIF_PWRDOWN_0, data);
+
+			orig = data = RREG32_PCIE(ixPB1_PIF_PWRDOWN_1);
+			data &= ~(PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1_MASK |
+				PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1_MASK);
+			data |= (7 << PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_OFF_1__SHIFT) |
+				(7 << PB1_PIF_PWRDOWN_1__PLL_POWER_STATE_IN_TXS2_1__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPB1_PIF_PWRDOWN_1, data);
+
+			orig = data = RREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL);
+			data &= ~PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE_MASK;
+			data |= ~(3 << PCIE_LC_LINK_WIDTH_CNTL__LC_DYN_LANES_PWR_STATE__SHIFT);
+			if (orig != data)
+				WREG32_PCIE(ixPCIE_LC_LINK_WIDTH_CNTL, data);
+
+			if (!disable_clkreq) {
+				struct pci_dev *root = adev->pdev->bus->self;
+				u32 lnkcap;
+
+				clk_req_support = false;
+				pcie_capability_read_dword(root, PCI_EXP_LNKCAP, &lnkcap);
+				if (lnkcap & PCI_EXP_LNKCAP_CLKPM)
+					clk_req_support = true;
+			} else {
+				clk_req_support = false;
+			}
+
+			if (clk_req_support) {
+				orig = data = RREG32_PCIE(ixPCIE_LC_CNTL2);
+				data |= PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L1_MASK |
+					PCIE_LC_CNTL2__LC_ALLOW_PDWN_IN_L23_MASK;
+				if (orig != data)
+					WREG32_PCIE(ixPCIE_LC_CNTL2, data);
+
+				orig = data = RREG32_SMC(ixTHM_CLK_CNTL);
+				data &= ~(THM_CLK_CNTL__CMON_CLK_SEL_MASK |
+					THM_CLK_CNTL__TMON_CLK_SEL_MASK);
+				data |= (1 << THM_CLK_CNTL__CMON_CLK_SEL__SHIFT) |
+					(1 << THM_CLK_CNTL__TMON_CLK_SEL__SHIFT);
+				if (orig != data)
+					WREG32_SMC(ixTHM_CLK_CNTL, data);
+
+				orig = data = RREG32_SMC(ixMISC_CLK_CTRL);
+				data &= ~(MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL_MASK |
+					MISC_CLK_CTRL__ZCLK_SEL_MASK);
+				data |= (1 << MISC_CLK_CTRL__DEEP_SLEEP_CLK_SEL__SHIFT) |
+					(1 << MISC_CLK_CTRL__ZCLK_SEL__SHIFT);
+				if (orig != data)
+					WREG32_SMC(ixMISC_CLK_CTRL, data);
+
+				orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL);
+				data &= ~CG_CLKPIN_CNTL__BCLK_AS_XCLK_MASK;
+				if (orig != data)
+					WREG32_SMC(ixCG_CLKPIN_CNTL, data);
+
+				orig = data = RREG32_SMC(ixCG_CLKPIN_CNTL_2);
+				data &= ~CG_CLKPIN_CNTL_2__FORCE_BIF_REFCLK_EN_MASK;
+				if (orig != data)
+					WREG32_SMC(ixCG_CLKPIN_CNTL_2, data);
+
+				orig = data = RREG32_SMC(ixMPLL_BYPASSCLK_SEL);
+				data &= ~MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL_MASK;
+				data |= (4 << MPLL_BYPASSCLK_SEL__MPLL_CLKOUT_SEL__SHIFT);
+				if (orig != data)
+					WREG32_SMC(ixMPLL_BYPASSCLK_SEL, data);
+			}
+		}
+	} else {
+		if (orig != data)
+			WREG32_PCIE(ixPCIE_LC_CNTL, data);
+	}
+
+	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
+	data |= PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
+		PCIE_CNTL2__MST_MEM_LS_EN_MASK |
+		PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK;
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_CNTL2, data);
+
+	if (!disable_l0s) {
+		data = RREG32_PCIE(ixPCIE_LC_N_FTS_CNTL);
+		if ((data & PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) ==
+				PCIE_LC_N_FTS_CNTL__LC_N_FTS_MASK) {
+			data = RREG32_PCIE(ixPCIE_LC_STATUS1);
+			if ((data & PCIE_LC_STATUS1__LC_REVERSE_XMIT_MASK) &&
+			(data & PCIE_LC_STATUS1__LC_REVERSE_RCVR_MASK)) {
+				orig = data = RREG32_PCIE(ixPCIE_LC_CNTL);
+				data &= ~PCIE_LC_CNTL__LC_L0S_INACTIVITY_MASK;
+				if (orig != data)
+					WREG32_PCIE(ixPCIE_LC_CNTL, data);
+			}
+		}
+	}
+}
+
+static uint32_t cik_get_rev_id(struct amdgpu_device *adev)
+{
+	return (RREG32(mmCC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK)
+		>> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT;
+}
+
+static const struct amdgpu_ip_block_version bonaire_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 1,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &ci_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 8,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &dce_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 7,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &gfx_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_sdma_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 4,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v4_2_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v2_0_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version hawaii_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 1,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &ci_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 8,
+		.minor = 5,
+		.rev = 0,
+		.funcs = &dce_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 7,
+		.minor = 3,
+		.rev = 0,
+		.funcs = &gfx_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_sdma_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 4,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v4_2_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v2_0_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version kabini_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 1,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &kv_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 8,
+		.minor = 3,
+		.rev = 0,
+		.funcs = &dce_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 7,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &gfx_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_sdma_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 4,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v4_2_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v2_0_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version mullins_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 1,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &kv_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 8,
+		.minor = 3,
+		.rev = 0,
+		.funcs = &dce_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 7,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &gfx_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_sdma_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 4,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v4_2_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v2_0_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version kaveri_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 1,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &kv_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 8,
+		.minor = 1,
+		.rev = 0,
+		.funcs = &dce_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 7,
+		.minor = 1,
+		.rev = 0,
+		.funcs = &gfx_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cik_sdma_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 4,
+		.minor = 2,
+		.rev = 0,
+		.funcs = &uvd_v4_2_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v2_0_ip_funcs,
+	},
+};
+
+int cik_set_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		adev->ip_blocks = bonaire_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(bonaire_ip_blocks);
+		break;
+	case CHIP_HAWAII:
+		adev->ip_blocks = hawaii_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(hawaii_ip_blocks);
+		break;
+	case CHIP_KAVERI:
+		adev->ip_blocks = kaveri_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(kaveri_ip_blocks);
+		break;
+	case CHIP_KABINI:
+		adev->ip_blocks = kabini_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(kabini_ip_blocks);
+		break;
+	case CHIP_MULLINS:
+		adev->ip_blocks = mullins_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(mullins_ip_blocks);
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct amdgpu_asic_funcs cik_asic_funcs =
+{
+	.read_disabled_bios = &cik_read_disabled_bios,
+	.read_register = &cik_read_register,
+	.reset = &cik_asic_reset,
+	.set_vga_state = &cik_vga_set_state,
+	.get_xclk = &cik_get_xclk,
+	.set_uvd_clocks = &cik_set_uvd_clocks,
+	.set_vce_clocks = &cik_set_vce_clocks,
+	.get_cu_info = &gfx_v7_0_get_cu_info,
+	/* these should be moved to their own ip modules */
+	.get_gpu_clock_counter = &gfx_v7_0_get_gpu_clock_counter,
+	.wait_for_mc_idle = &gmc_v7_0_mc_wait_for_idle,
+};
+
+static int cik_common_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->smc_rreg = &cik_smc_rreg;
+	adev->smc_wreg = &cik_smc_wreg;
+	adev->pcie_rreg = &cik_pcie_rreg;
+	adev->pcie_wreg = &cik_pcie_wreg;
+	adev->uvd_ctx_rreg = &cik_uvd_ctx_rreg;
+	adev->uvd_ctx_wreg = &cik_uvd_ctx_wreg;
+	adev->didt_rreg = &cik_didt_rreg;
+	adev->didt_wreg = &cik_didt_wreg;
+
+	adev->asic_funcs = &cik_asic_funcs;
+
+	adev->has_uvd = true;
+
+	adev->rev_id = cik_get_rev_id(adev);
+	adev->external_rev_id = 0xFF;
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		adev->cg_flags =
+			AMDGPU_CG_SUPPORT_GFX_MGCG |
+			AMDGPU_CG_SUPPORT_GFX_MGLS |
+			/*AMDGPU_CG_SUPPORT_GFX_CGCG |*/
+			AMDGPU_CG_SUPPORT_GFX_CGLS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS_LS |
+			AMDGPU_CG_SUPPORT_GFX_CP_LS |
+			AMDGPU_CG_SUPPORT_MC_LS |
+			AMDGPU_CG_SUPPORT_MC_MGCG |
+			AMDGPU_CG_SUPPORT_SDMA_MGCG |
+			AMDGPU_CG_SUPPORT_SDMA_LS |
+			AMDGPU_CG_SUPPORT_BIF_LS |
+			AMDGPU_CG_SUPPORT_VCE_MGCG |
+			AMDGPU_CG_SUPPORT_UVD_MGCG |
+			AMDGPU_CG_SUPPORT_HDP_LS |
+			AMDGPU_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x14;
+		break;
+	case CHIP_HAWAII:
+		adev->cg_flags =
+			AMDGPU_CG_SUPPORT_GFX_MGCG |
+			AMDGPU_CG_SUPPORT_GFX_MGLS |
+			/*AMDGPU_CG_SUPPORT_GFX_CGCG |*/
+			AMDGPU_CG_SUPPORT_GFX_CGLS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS |
+			AMDGPU_CG_SUPPORT_GFX_CP_LS |
+			AMDGPU_CG_SUPPORT_MC_LS |
+			AMDGPU_CG_SUPPORT_MC_MGCG |
+			AMDGPU_CG_SUPPORT_SDMA_MGCG |
+			AMDGPU_CG_SUPPORT_SDMA_LS |
+			AMDGPU_CG_SUPPORT_BIF_LS |
+			AMDGPU_CG_SUPPORT_VCE_MGCG |
+			AMDGPU_CG_SUPPORT_UVD_MGCG |
+			AMDGPU_CG_SUPPORT_HDP_LS |
+			AMDGPU_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 0x28;
+		break;
+	case CHIP_KAVERI:
+		adev->cg_flags =
+			AMDGPU_CG_SUPPORT_GFX_MGCG |
+			AMDGPU_CG_SUPPORT_GFX_MGLS |
+			/*AMDGPU_CG_SUPPORT_GFX_CGCG |*/
+			AMDGPU_CG_SUPPORT_GFX_CGLS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS_LS |
+			AMDGPU_CG_SUPPORT_GFX_CP_LS |
+			AMDGPU_CG_SUPPORT_SDMA_MGCG |
+			AMDGPU_CG_SUPPORT_SDMA_LS |
+			AMDGPU_CG_SUPPORT_BIF_LS |
+			AMDGPU_CG_SUPPORT_VCE_MGCG |
+			AMDGPU_CG_SUPPORT_UVD_MGCG |
+			AMDGPU_CG_SUPPORT_HDP_LS |
+			AMDGPU_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags =
+			/*AMDGPU_PG_SUPPORT_GFX_PG |
+			  AMDGPU_PG_SUPPORT_GFX_SMG |
+			  AMDGPU_PG_SUPPORT_GFX_DMG |*/
+			AMDGPU_PG_SUPPORT_UVD |
+			/*AMDGPU_PG_SUPPORT_VCE |
+			  AMDGPU_PG_SUPPORT_CP |
+			  AMDGPU_PG_SUPPORT_GDS |
+			  AMDGPU_PG_SUPPORT_RLC_SMU_HS |
+			  AMDGPU_PG_SUPPORT_ACP |
+			  AMDGPU_PG_SUPPORT_SAMU |*/
+			0;
+		if (adev->pdev->device == 0x1312 ||
+			adev->pdev->device == 0x1316 ||
+			adev->pdev->device == 0x1317)
+			adev->external_rev_id = 0x41;
+		else
+			adev->external_rev_id = 0x1;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		adev->cg_flags =
+			AMDGPU_CG_SUPPORT_GFX_MGCG |
+			AMDGPU_CG_SUPPORT_GFX_MGLS |
+			/*AMDGPU_CG_SUPPORT_GFX_CGCG |*/
+			AMDGPU_CG_SUPPORT_GFX_CGLS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS |
+			AMDGPU_CG_SUPPORT_GFX_CGTS_LS |
+			AMDGPU_CG_SUPPORT_GFX_CP_LS |
+			AMDGPU_CG_SUPPORT_SDMA_MGCG |
+			AMDGPU_CG_SUPPORT_SDMA_LS |
+			AMDGPU_CG_SUPPORT_BIF_LS |
+			AMDGPU_CG_SUPPORT_VCE_MGCG |
+			AMDGPU_CG_SUPPORT_UVD_MGCG |
+			AMDGPU_CG_SUPPORT_HDP_LS |
+			AMDGPU_CG_SUPPORT_HDP_MGCG;
+		adev->pg_flags =
+			/*AMDGPU_PG_SUPPORT_GFX_PG |
+			  AMDGPU_PG_SUPPORT_GFX_SMG | */
+			AMDGPU_PG_SUPPORT_UVD |
+			/*AMDGPU_PG_SUPPORT_VCE |
+			  AMDGPU_PG_SUPPORT_CP |
+			  AMDGPU_PG_SUPPORT_GDS |
+			  AMDGPU_PG_SUPPORT_RLC_SMU_HS |
+			  AMDGPU_PG_SUPPORT_SAMU |*/
+			0;
+		if (adev->asic_type == CHIP_KABINI) {
+			if (adev->rev_id == 0)
+				adev->external_rev_id = 0x81;
+			else if (adev->rev_id == 1)
+				adev->external_rev_id = 0x82;
+			else if (adev->rev_id == 2)
+				adev->external_rev_id = 0x85;
+		} else
+			adev->external_rev_id = adev->rev_id + 0xa1;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int cik_common_sw_init(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* move the golden regs per IP block */
+	cik_init_golden_registers(adev);
+	/* enable pcie gen2/3 link */
+	cik_pcie_gen3_enable(adev);
+	/* enable aspm */
+	cik_program_aspm(adev);
+
+	return 0;
+}
+
+static int cik_common_hw_fini(void *handle)
+{
+	return 0;
+}
+
+static int cik_common_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_amdkfd_suspend(adev);
+
+	return cik_common_hw_fini(adev);
+}
+
+static int cik_common_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cik_common_hw_init(adev);
+	if (r)
+		return r;
+
+	return amdgpu_amdkfd_resume(adev);
+}
+
+static bool cik_common_is_idle(void *handle)
+{
+	return true;
+}
+
+static int cik_common_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static void cik_common_print_status(void *handle)
+{
+
+}
+
+static int cik_common_soft_reset(void *handle)
+{
+	/* XXX hard reset?? */
+	return 0;
+}
+
+static int cik_common_set_clockgating_state(void *handle,
+					    enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int cik_common_set_powergating_state(void *handle,
+					    enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs cik_common_ip_funcs = {
+	.early_init = cik_common_early_init,
+	.late_init = NULL,
+	.sw_init = cik_common_sw_init,
+	.sw_fini = cik_common_sw_fini,
+	.hw_init = cik_common_hw_init,
+	.hw_fini = cik_common_hw_fini,
+	.suspend = cik_common_suspend,
+	.resume = cik_common_resume,
+	.is_idle = cik_common_is_idle,
+	.wait_for_idle = cik_common_wait_for_idle,
+	.soft_reset = cik_common_soft_reset,
+	.print_status = cik_common_print_status,
+	.set_clockgating_state = cik_common_set_clockgating_state,
+	.set_powergating_state = cik_common_set_powergating_state,
+};
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik.h b/kernel/drivers/gpu/drm/amd/amdgpu/cik.h
new file mode 100644
index 000000000..5ebd2d7a0
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_H__
+#define __CIK_H__
+
+extern const struct amd_ip_funcs cik_common_ip_funcs;
+
+void cik_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid);
+int cik_set_ip_blocks(struct amdgpu_device *adev);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik_dpm.h b/kernel/drivers/gpu/drm/amd/amdgpu/cik_dpm.h
new file mode 100644
index 000000000..b1c8e7b44
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik_dpm.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_DPM_H__
+#define __CIK_DPM_H__
+
+extern const struct amd_ip_funcs ci_dpm_ip_funcs;
+extern const struct amd_ip_funcs kv_dpm_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.c
new file mode 100644
index 000000000..8993c50cb
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.c
@@ -0,0 +1,471 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "cikd.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void cik_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * cik_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (CIK).
+ */
+static void cik_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_cntl |= IH_CNTL__ENABLE_INTR_MASK;
+	ih_rb_cntl |= IH_RB_CNTL__RB_ENABLE_MASK;
+	WREG32(mmIH_CNTL, ih_cntl);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * cik_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (CIK).
+ */
+static void cik_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+
+	ih_rb_cntl &= ~IH_RB_CNTL__RB_ENABLE_MASK;
+	ih_cntl &= ~IH_CNTL__ENABLE_INTR_MASK;
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	WREG32(mmIH_CNTL, ih_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * cik_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (CIK).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int cik_ih_irq_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	/* disable irqs */
+	cik_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl &= ~INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK;
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl &= ~INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK;
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+
+	ih_rb_cntl = (IH_RB_CNTL__WPTR_OVERFLOW_ENABLE_MASK |
+		      IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK |
+		      (rb_bufsz << 1));
+
+	ih_rb_cntl |= IH_RB_CNTL__WPTR_WRITEBACK_ENABLE_MASK;
+
+	/* set the writeback address whether it's enabled or not */
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	/* Default settings for IH_CNTL (disabled at first) */
+	ih_cntl = (0x10 << IH_CNTL__MC_WRREQ_CREDIT__SHIFT) |
+		(0x10 << IH_CNTL__MC_WR_CLEAN_CNT__SHIFT) |
+		(0 << IH_CNTL__MC_VMID__SHIFT);
+	/* IH_CNTL__RPTR_REARM_MASK only works if msi's are enabled */
+	if (adev->irq.msi_enabled)
+		ih_cntl |= IH_CNTL__RPTR_REARM_MASK;
+	WREG32(mmIH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+
+	/* enable irqs */
+	cik_ih_enable_interrupts(adev);
+
+	return ret;
+}
+
+/**
+ * cik_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (CIK).
+ */
+static void cik_ih_irq_disable(struct amdgpu_device *adev)
+{
+	cik_ih_disable_interrupts(adev);
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * cik_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (CIK).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cik_irq_process().
+ * Returns the value of the wptr.
+ */
+static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
+		wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/*        CIK IV Ring
+ * Each IV ring entry is 128 bits:
+ * [7:0]    - interrupt source id
+ * [31:8]   - reserved
+ * [59:32]  - interrupt source data
+ * [63:60]  - reserved
+ * [71:64]  - RINGID
+ *            CP:
+ *            ME_ID [1:0], PIPE_ID[1:0], QUEUE_ID[2:0]
+ *            QUEUE_ID - for compute, which of the 8 queues owned by the dispatcher
+ *                     - for gfx, hw shader state (0=PS...5=LS, 6=CS)
+ *            ME_ID - 0 = gfx, 1 = first 4 CS pipes, 2 = second 4 CS pipes
+ *            PIPE_ID - ME0 0=3D
+ *                    - ME1&2 compute dispatcher (4 pipes each)
+ *            SDMA:
+ *            INSTANCE_ID [1:0], QUEUE_ID[1:0]
+ *            INSTANCE_ID - 0 = sdma0, 1 = sdma1
+ *            QUEUE_ID - 0 = gfx, 1 = rlc0, 2 = rlc1
+ * [79:72]  - VMID
+ * [95:80]  - PASID
+ * [127:96] - reserved
+ */
+
+ /**
+ * cik_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void cik_ih_decode_iv(struct amdgpu_device *adev,
+			     struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+	
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vm_id = (dw[2] >> 8) & 0xff;
+	entry->pas_id = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * cik_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void cik_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int cik_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cik_ih_set_interrupt_funcs(adev);
+
+	return 0;
+}
+
+static int cik_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int cik_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+
+	return 0;
+}
+
+static int cik_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cik_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int cik_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cik_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int cik_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_ih_hw_fini(adev);
+}
+
+static int cik_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_ih_hw_init(adev);
+}
+
+static bool cik_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		return false;
+
+	return true;
+}
+
+static int cik_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & SRBM_STATUS__IH_BUSY_MASK;
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void cik_ih_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "CIK IH registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+	dev_info(adev->dev, "  INTERRUPT_CNTL=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL));
+	dev_info(adev->dev, "  INTERRUPT_CNTL2=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL2));
+	dev_info(adev->dev, "  IH_CNTL=0x%08X\n",
+		 RREG32(mmIH_CNTL));
+	dev_info(adev->dev, "  IH_RB_CNTL=0x%08X\n",
+		 RREG32(mmIH_RB_CNTL));
+	dev_info(adev->dev, "  IH_RB_BASE=0x%08X\n",
+		 RREG32(mmIH_RB_BASE));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_LO=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_LO));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_HI=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_HI));
+	dev_info(adev->dev, "  IH_RB_RPTR=0x%08X\n",
+		 RREG32(mmIH_RB_RPTR));
+	dev_info(adev->dev, "  IH_RB_WPTR=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR));
+}
+
+static int cik_ih_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_IH_MASK;
+
+	if (srbm_soft_reset) {
+		cik_ih_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		cik_ih_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int cik_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int cik_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs cik_ih_ip_funcs = {
+	.early_init = cik_ih_early_init,
+	.late_init = NULL,
+	.sw_init = cik_ih_sw_init,
+	.sw_fini = cik_ih_sw_fini,
+	.hw_init = cik_ih_hw_init,
+	.hw_fini = cik_ih_hw_fini,
+	.suspend = cik_ih_suspend,
+	.resume = cik_ih_resume,
+	.is_idle = cik_ih_is_idle,
+	.wait_for_idle = cik_ih_wait_for_idle,
+	.soft_reset = cik_ih_soft_reset,
+	.print_status = cik_ih_print_status,
+	.set_clockgating_state = cik_ih_set_clockgating_state,
+	.set_powergating_state = cik_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs cik_ih_funcs = {
+	.get_wptr = cik_ih_get_wptr,
+	.decode_iv = cik_ih_decode_iv,
+	.set_rptr = cik_ih_set_rptr
+};
+
+static void cik_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &cik_ih_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.h b/kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.h
new file mode 100644
index 000000000..6b0f375ec
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_IH_H__
+#define __CIK_IH_H__
+
+extern const struct amd_ip_funcs cik_ih_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
new file mode 100644
index 000000000..5f712cedd
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -0,0 +1,1412 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "cikd.h"
+#include "cik.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	SDMA0_REGISTER_OFFSET,
+	SDMA1_REGISTER_OFFSET
+};
+
+static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev);
+static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev);
+static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev);
+static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("radeon/bonaire_sdma.bin");
+MODULE_FIRMWARE("radeon/bonaire_sdma1.bin");
+MODULE_FIRMWARE("radeon/hawaii_sdma.bin");
+MODULE_FIRMWARE("radeon/hawaii_sdma1.bin");
+MODULE_FIRMWARE("radeon/kaveri_sdma.bin");
+MODULE_FIRMWARE("radeon/kaveri_sdma1.bin");
+MODULE_FIRMWARE("radeon/kabini_sdma.bin");
+MODULE_FIRMWARE("radeon/kabini_sdma1.bin");
+MODULE_FIRMWARE("radeon/mullins_sdma.bin");
+MODULE_FIRMWARE("radeon/mullins_sdma1.bin");
+
+u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev);
+
+/*
+ * sDMA - System DMA
+ * Starting with CIK, the GPU has new asynchronous
+ * DMA engines.  These engines are used for compute
+ * and gfx.  There are two DMA engines (SDMA0, SDMA1)
+ * and each one supports 1 ring buffer used for gfx
+ * and 2 queues used for compute.
+ *
+ * The programming model is very similar to the CP
+ * (ring buffer, IBs, etc.), but sDMA has it's own
+ * packet format that is different from the PM4 format
+ * used by the CP. sDMA supports copying data, writing
+ * embedded data, solid fills, and a number of other
+ * things.  It also has support for tiling/detiling of
+ * buffers.
+ */
+
+/**
+ * cik_sdma_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int cik_sdma_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_KAVERI:
+		chip_name = "kaveri";
+		break;
+	case CHIP_KABINI:
+		chip_name = "kabini";
+		break;
+	case CHIP_MULLINS:
+		chip_name = "mullins";
+		break;
+	default: BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "radeon/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "radeon/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+	}
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "cik_sdma: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * cik_sdma_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (CIK+).
+ */
+static uint32_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	rptr = ring->adev->wb.wb[ring->rptr_offs];
+
+	return (rptr & 0x3fffc) >> 2;
+}
+
+/**
+ * cik_sdma_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (CIK+).
+ */
+static uint32_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
+
+	return (RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) & 0x3fffc) >> 2;
+}
+
+/**
+ * cik_sdma_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (CIK+).
+ */
+static void cik_sdma_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 me = (ring == &adev->sdma.instance[0].ring) ? 0 : 1;
+
+	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], (ring->wptr << 2) & 0x3fffc);
+}
+
+static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->nop |
+					  SDMA_NOP_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->nop);
+}
+
+/**
+ * cik_sdma_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (CIK).
+ */
+static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring,
+			   struct amdgpu_ib *ib)
+{
+	u32 extra_bits = (ib->vm ? ib->vm->ids[ring->idx].id : 0) & 0xf;
+	u32 next_rptr = ring->wptr + 5;
+
+	while ((next_rptr & 7) != 4)
+		next_rptr++;
+
+	next_rptr += 4;
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
+	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, 1); /* number of DWs to follow */
+	amdgpu_ring_write(ring, next_rptr);
+
+	/* IB packet must end on a 8 DW boundary */
+	cik_sdma_ring_insert_nop(ring, (12 - (ring->wptr & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_INDIRECT_BUFFER, 0, extra_bits));
+	amdgpu_ring_write(ring, ib->gpu_addr & 0xffffffe0); /* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, ib->length_dw);
+
+}
+
+/**
+ * cik_sdma_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(1) |
+			  SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */
+	u32 ref_and_mask;
+
+	if (ring == &ring->adev->sdma.instance[0].ring)
+		ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA0_MASK;
+	else
+		ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA1_MASK;
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE << 2);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
+}
+
+/**
+ * cik_sdma_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (CIK).
+ */
+static void cik_sdma_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_FENCE, 0, 0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_FENCE, 0, 0));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_TRAP, 0, 0));
+}
+
+/**
+ * cik_sdma_ring_emit_semaphore - emit a semaphore on the dma ring
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @semaphore: amdgpu semaphore object
+ * @emit_wait: wait or signal semaphore
+ *
+ * Add a DMA semaphore packet to the ring wait on or signal
+ * other rings (CIK).
+ */
+static bool cik_sdma_ring_emit_semaphore(struct amdgpu_ring *ring,
+					 struct amdgpu_semaphore *semaphore,
+					 bool emit_wait)
+{
+	u64 addr = semaphore->gpu_addr;
+	u32 extra_bits = emit_wait ? 0 : SDMA_SEMAPHORE_EXTRA_S;
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SEMAPHORE, 0, extra_bits));
+	amdgpu_ring_write(ring, addr & 0xfffffff8);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+
+	return true;
+}
+
+/**
+ * cik_sdma_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (CIK).
+ */
+static void cik_sdma_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl &= ~SDMA0_GFX_RB_CNTL__RB_ENABLE_MASK;
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], 0);
+	}
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * cik_sdma_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (CIK).
+ */
+static void cik_sdma_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * cik_sdma_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (CIK).
+ */
+static void cik_sdma_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 me_cntl;
+	int i;
+
+	if (enable == false) {
+		cik_sdma_gfx_stop(adev);
+		cik_sdma_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		me_cntl = RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]);
+		if (enable)
+			me_cntl &= ~SDMA0_F32_CNTL__HALT_MASK;
+		else
+			me_cntl |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + sdma_offsets[i], me_cntl);
+	}
+}
+
+/**
+ * cik_sdma_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	int i, j, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		wb_offset = (ring->rptr_offs * 4);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			cik_srbm_select(adev, 0, 0, 0, j);
+			/* SDMA GFX */
+			WREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i], 0);
+			WREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i], 0);
+			/* XXX SDMA RLC - todo */
+		}
+		cik_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		WREG32(mmSDMA0_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = rb_bufsz << 1;
+#ifdef __BIG_ENDIAN
+		rb_cntl |= SDMA0_GFX_RB_CNTL__RB_SWAP_ENABLE_MASK |
+			SDMA0_GFX_RB_CNTL__RPTR_WRITEBACK_SWAP_ENABLE_MASK;
+#endif
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
+		       ((adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC));
+
+		rb_cntl |= SDMA0_GFX_RB_CNTL__RPTR_WRITEBACK_ENABLE_MASK;
+
+		WREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
+
+		/* enable DMA RB */
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i],
+		       rb_cntl | SDMA0_GFX_RB_CNTL__RB_ENABLE_MASK);
+
+		ib_cntl = SDMA0_GFX_IB_CNTL__IB_ENABLE_MASK;
+#ifdef __BIG_ENDIAN
+		ib_cntl |= SDMA0_GFX_IB_CNTL__IB_SWAP_ENABLE_MASK;
+#endif
+		/* enable DMA IBs */
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		ring->ready = true;
+
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
+	}
+
+	return 0;
+}
+
+/**
+ * cik_sdma_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_rlc_resume(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+	return 0;
+}
+
+/**
+ * cik_sdma_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int cik_sdma_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	cik_sdma_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
+		for (j = 0; j < fw_size; j++)
+			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * cik_sdma_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_start(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = cik_sdma_load_microcode(adev);
+	if (r)
+		return r;
+
+	/* unhalt the MEs */
+	cik_sdma_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = cik_sdma_gfx_resume(adev);
+	if (r)
+		return r;
+	r = cik_sdma_rlc_resume(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * cik_sdma_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (CIK).
+ * Returns 0 for success, error for failure.
+ */
+static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_lock(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_wb_free(adev, index);
+		return r;
+	}
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, 1); /* number of DWs to follow */
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * cik_sdma_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (CIK).
+ * Returns 0 on success, error on failure.
+ */
+static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct fence *f = NULL;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	r = amdgpu_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(ring, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = 1;
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.length_dw = 5;
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err1;
+
+	r = fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto err1;
+	}
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
+			 ring->idx, i);
+		goto err1;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+
+err1:
+	fence_put(f);
+	amdgpu_ib_free(adev, &ib);
+err0:
+	amdgpu_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * cik_sdma_vm_copy_pages - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (CIK).
+ */
+static void cik_sdma_vm_copy_pte(struct amdgpu_ib *ib,
+				 uint64_t pe, uint64_t src,
+				 unsigned count)
+{
+	while (count) {
+		unsigned bytes = count * 8;
+		if (bytes > 0x1FFFF8)
+			bytes = 0x1FFFF8;
+
+		ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY,
+			SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+		ib->ptr[ib->length_dw++] = bytes;
+		ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+		ib->ptr[ib->length_dw++] = lower_32_bits(src);
+		ib->ptr[ib->length_dw++] = upper_32_bits(src);
+		ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+		pe += bytes;
+		src += bytes;
+		count -= bytes / 8;
+	}
+}
+
+/**
+ * cik_sdma_vm_write_pages - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update PTEs by writing them manually using sDMA (CIK).
+ */
+static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib,
+				  uint64_t pe,
+				  uint64_t addr, unsigned count,
+				  uint32_t incr, uint32_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count * 2;
+		if (ndw > 0xFFFFE)
+			ndw = 0xFFFFE;
+
+		/* for non-physically contiguous pages (system) */
+		ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_WRITE,
+			SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+		ib->ptr[ib->length_dw++] = pe;
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+		ib->ptr[ib->length_dw++] = ndw;
+		for (; ndw > 0; ndw -= 2, --count, pe += 8) {
+			if (flags & AMDGPU_PTE_SYSTEM) {
+				value = amdgpu_vm_map_gart(ib->ring->adev, addr);
+				value &= 0xFFFFFFFFFFFFF000ULL;
+			} else if (flags & AMDGPU_PTE_VALID) {
+				value = addr;
+			} else {
+				value = 0;
+			}
+			addr += incr;
+			value |= flags;
+			ib->ptr[ib->length_dw++] = value;
+			ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		}
+	}
+}
+
+/**
+ * cik_sdma_vm_set_pages - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib,
+				    uint64_t pe,
+				    uint64_t addr, unsigned count,
+				    uint32_t incr, uint32_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count;
+		if (ndw > 0x7FFFF)
+			ndw = 0x7FFFF;
+
+		if (flags & AMDGPU_PTE_VALID)
+			value = addr;
+		else
+			value = 0;
+
+		/* for physically contiguous pages (vram) */
+		ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_GENERATE_PTE_PDE, 0, 0);
+		ib->ptr[ib->length_dw++] = pe; /* dst addr */
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+		ib->ptr[ib->length_dw++] = flags; /* mask */
+		ib->ptr[ib->length_dw++] = 0;
+		ib->ptr[ib->length_dw++] = value; /* value */
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		ib->ptr[ib->length_dw++] = incr; /* increment size */
+		ib->ptr[ib->length_dw++] = 0;
+		ib->ptr[ib->length_dw++] = ndw; /* number of entries */
+
+		pe += ndw * 8;
+		addr += ndw * incr;
+		count -= ndw;
+	}
+}
+
+/**
+ * cik_sdma_vm_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void cik_sdma_vm_pad_ib(struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ib->ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+					SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0) |
+					SDMA_NOP_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+					SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0);
+}
+
+/**
+ * cik_sdma_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (CIK).
+ */
+static void cik_sdma_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vm_id, uint64_t pd_addr)
+{
+	u32 extra_bits = (SDMA_POLL_REG_MEM_EXTRA_OP(0) |
+			  SDMA_POLL_REG_MEM_EXTRA_FUNC(0)); /* always */
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
+	if (vm_id < 8) {
+		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
+	} else {
+		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
+	}
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	/* flush TLB */
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 1 << vm_id);
+
+	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_POLL_REG_MEM, 0, extra_bits));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* reference */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
+}
+
+static void cik_enable_sdma_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_SDMA_MGCG)) {
+		WREG32(mmSDMA0_CLK_CTRL + SDMA0_REGISTER_OFFSET, 0x00000100);
+		WREG32(mmSDMA0_CLK_CTRL + SDMA1_REGISTER_OFFSET, 0x00000100);
+	} else {
+		orig = data = RREG32(mmSDMA0_CLK_CTRL + SDMA0_REGISTER_OFFSET);
+		data |= 0xff000000;
+		if (data != orig)
+			WREG32(mmSDMA0_CLK_CTRL + SDMA0_REGISTER_OFFSET, data);
+
+		orig = data = RREG32(mmSDMA0_CLK_CTRL + SDMA1_REGISTER_OFFSET);
+		data |= 0xff000000;
+		if (data != orig)
+			WREG32(mmSDMA0_CLK_CTRL + SDMA1_REGISTER_OFFSET, data);
+	}
+}
+
+static void cik_enable_sdma_mgls(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_SDMA_LS)) {
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET);
+		data |= 0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET, data);
+
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET);
+		data |= 0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET, data);
+	} else {
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET);
+		data &= ~0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA0_REGISTER_OFFSET, data);
+
+		orig = data = RREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET);
+		data &= ~0x100;
+		if (orig != data)
+			WREG32(mmSDMA0_POWER_CNTL + SDMA1_REGISTER_OFFSET, data);
+	}
+}
+
+static int cik_sdma_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
+
+	cik_sdma_set_ring_funcs(adev);
+	cik_sdma_set_irq_funcs(adev);
+	cik_sdma_set_buffer_funcs(adev);
+	cik_sdma_set_vm_pte_funcs(adev);
+
+	return 0;
+}
+
+static int cik_sdma_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r, i;
+
+	r = cik_sdma_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, 224, &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, 241, &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, 247, &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 256 * 1024,
+				     SDMA_PACKET(SDMA_OPCODE_NOP, 0, 0), 0xf,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 : AMDGPU_SDMA_IRQ_TRAP1,
+				     AMDGPU_RING_TYPE_SDMA);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int cik_sdma_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	return 0;
+}
+
+static int cik_sdma_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cik_sdma_start(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int cik_sdma_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cik_sdma_enable(adev, false);
+
+	return 0;
+}
+
+static int cik_sdma_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_sdma_hw_fini(adev);
+}
+
+static int cik_sdma_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cik_sdma_hw_init(adev);
+}
+
+static bool cik_sdma_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int cik_sdma_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK);
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void cik_sdma_print_status(void *handle)
+{
+	int i, j;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "CIK SDMA registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		 RREG32(mmSRBM_STATUS2));
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		dev_info(adev->dev, "  SDMA%d_STATUS_REG=0x%08X\n",
+			 i, RREG32(mmSDMA0_STATUS_REG + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_ME_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_SEM_INCOMPLETE_TIMER_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_SEM_INCOMPLETE_TIMER_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_SEM_WAIT_FAIL_TIMER_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_IB_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_WPTR=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR_ADDR_HI=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR_ADDR_LO=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_BASE=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_BASE_HI=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i]));
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			cik_srbm_select(adev, 0, 0, 0, j);
+			dev_info(adev->dev, "  VM %d:\n", j);
+			dev_info(adev->dev, "  SDMA0_GFX_VIRTUAL_ADDR=0x%08X\n",
+				 RREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i]));
+			dev_info(adev->dev, "  SDMA0_GFX_APE1_CNTL=0x%08X\n",
+				 RREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i]));
+		}
+		cik_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+}
+
+static int cik_sdma_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+	}
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp |= SDMA0_F32_CNTL__HALT_MASK;
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+	}
+
+	if (srbm_soft_reset) {
+		cik_sdma_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		cik_sdma_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int cik_sdma_set_trap_irq_state(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *src,
+				       unsigned type,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl &= ~SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl |= SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl &= ~SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl |= SDMA0_CNTL__TRAP_ENABLE_MASK;
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int cik_sdma_process_trap_irq(struct amdgpu_device *adev,
+				     struct amdgpu_irq_src *source,
+				     struct amdgpu_iv_entry *entry)
+{
+	u8 instance_id, queue_id;
+
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (instance_id) {
+	case 0:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	case 1:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	}
+
+	return 0;
+}
+
+static int cik_sdma_process_illegal_inst_irq(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int cik_sdma_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	cik_enable_sdma_mgcg(adev, gate);
+	cik_enable_sdma_mgls(adev, gate);
+
+	return 0;
+}
+
+static int cik_sdma_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs cik_sdma_ip_funcs = {
+	.early_init = cik_sdma_early_init,
+	.late_init = NULL,
+	.sw_init = cik_sdma_sw_init,
+	.sw_fini = cik_sdma_sw_fini,
+	.hw_init = cik_sdma_hw_init,
+	.hw_fini = cik_sdma_hw_fini,
+	.suspend = cik_sdma_suspend,
+	.resume = cik_sdma_resume,
+	.is_idle = cik_sdma_is_idle,
+	.wait_for_idle = cik_sdma_wait_for_idle,
+	.soft_reset = cik_sdma_soft_reset,
+	.print_status = cik_sdma_print_status,
+	.set_clockgating_state = cik_sdma_set_clockgating_state,
+	.set_powergating_state = cik_sdma_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
+	.get_rptr = cik_sdma_ring_get_rptr,
+	.get_wptr = cik_sdma_ring_get_wptr,
+	.set_wptr = cik_sdma_ring_set_wptr,
+	.parse_cs = NULL,
+	.emit_ib = cik_sdma_ring_emit_ib,
+	.emit_fence = cik_sdma_ring_emit_fence,
+	.emit_semaphore = cik_sdma_ring_emit_semaphore,
+	.emit_vm_flush = cik_sdma_ring_emit_vm_flush,
+	.emit_hdp_flush = cik_sdma_ring_emit_hdp_flush,
+	.test_ring = cik_sdma_ring_test_ring,
+	.test_ib = cik_sdma_ring_test_ib,
+	.insert_nop = cik_sdma_ring_insert_nop,
+};
+
+static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {
+	.set = cik_sdma_set_trap_irq_state,
+	.process = cik_sdma_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs cik_sdma_illegal_inst_irq_funcs = {
+	.process = cik_sdma_process_illegal_inst_irq,
+};
+
+static void cik_sdma_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &cik_sdma_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &cik_sdma_illegal_inst_irq_funcs;
+}
+
+/**
+ * cik_sdma_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (CIK).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void cik_sdma_emit_copy_buffer(struct amdgpu_ib *ib,
+				      uint64_t src_offset,
+				      uint64_t dst_offset,
+				      uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, SDMA_COPY_SUB_OPCODE_LINEAR, 0);
+	ib->ptr[ib->length_dw++] = byte_count;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * cik_sdma_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (CIK).
+ */
+static void cik_sdma_emit_fill_buffer(struct amdgpu_ib *ib,
+				      uint32_t src_data,
+				      uint64_t dst_offset,
+				      uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_CONSTANT_FILL, 0, 0);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count;
+}
+
+static const struct amdgpu_buffer_funcs cik_sdma_buffer_funcs = {
+	.copy_max_bytes = 0x1fffff,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = cik_sdma_emit_copy_buffer,
+
+	.fill_max_bytes = 0x1fffff,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = cik_sdma_emit_fill_buffer,
+};
+
+static void cik_sdma_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &cik_sdma_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs cik_sdma_vm_pte_funcs = {
+	.copy_pte = cik_sdma_vm_copy_pte,
+	.write_pte = cik_sdma_vm_write_pte,
+	.set_pte_pde = cik_sdma_vm_set_pte_pde,
+	.pad_ib = cik_sdma_vm_pad_ib,
+};
+
+static void cik_sdma_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &cik_sdma_vm_pte_funcs;
+		adev->vm_manager.vm_pte_funcs_ring = &adev->sdma.instance[0].ring;
+		adev->vm_manager.vm_pte_funcs_ring->is_pte_ring = true;
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.h b/kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.h
new file mode 100644
index 000000000..027727c67
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cik_sdma.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CIK_SDMA_H__
+#define __CIK_SDMA_H__
+
+extern const struct amd_ip_funcs cik_sdma_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cikd.h b/kernel/drivers/gpu/drm/amd/amdgpu/cikd.h
new file mode 100644
index 000000000..7f6d457f2
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cikd.h
@@ -0,0 +1,568 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#ifndef CIK_H
+#define CIK_H
+
+#define MC_SEQ_MISC0__MT__MASK	0xf0000000
+#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
+#define MC_SEQ_MISC0__MT__DDR2   0x20000000
+#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
+#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
+#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
+#define MC_SEQ_MISC0__MT__HBM    0x60000000
+#define MC_SEQ_MISC0__MT__DDR3   0xB0000000
+
+#define CP_ME_TABLE_SIZE    96
+
+/* display controller offsets used for crtc/cur/lut/grph/viewport/etc. */
+#define CRTC0_REGISTER_OFFSET                 (0x1b7c - 0x1b7c)
+#define CRTC1_REGISTER_OFFSET                 (0x1e7c - 0x1b7c)
+#define CRTC2_REGISTER_OFFSET                 (0x417c - 0x1b7c)
+#define CRTC3_REGISTER_OFFSET                 (0x447c - 0x1b7c)
+#define CRTC4_REGISTER_OFFSET                 (0x477c - 0x1b7c)
+#define CRTC5_REGISTER_OFFSET                 (0x4a7c - 0x1b7c)
+
+#define BONAIRE_GB_ADDR_CONFIG_GOLDEN        0x12010001
+#define HAWAII_GB_ADDR_CONFIG_GOLDEN         0x12011003
+
+#define CIK_RB_BITMAP_WIDTH_PER_SH     2
+#define HAWAII_RB_BITMAP_WIDTH_PER_SH  4
+
+#define AMDGPU_NUM_OF_VMIDS	8
+
+#define		PIPEID(x)					((x) << 0)
+#define		MEID(x)						((x) << 2)
+#define		VMID(x)						((x) << 4)
+#define		QUEUEID(x)					((x) << 8)
+
+#define mmCC_DRM_ID_STRAPS				0x1559
+#define CC_DRM_ID_STRAPS__ATI_REV_ID_MASK		0xf0000000
+
+#define mmCHUB_CONTROL					0x619
+#define		BYPASS_VM					(1 << 0)
+
+#define		SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU	(0 << 5)
+
+#define mmGRPH_LUT_10BIT_BYPASS_CONTROL			0x1a02
+#define		LUT_10BIT_BYPASS_EN			(1 << 8)
+
+#       define CURSOR_MONO                    0
+#       define CURSOR_24_1                    1
+#       define CURSOR_24_8_PRE_MULT           2
+#       define CURSOR_24_8_UNPRE_MULT         3
+#       define CURSOR_URGENT_ALWAYS           0
+#       define CURSOR_URGENT_1_8              1
+#       define CURSOR_URGENT_1_4              2
+#       define CURSOR_URGENT_3_8              3
+#       define CURSOR_URGENT_1_2              4
+
+#       define GRPH_DEPTH_8BPP                0
+#       define GRPH_DEPTH_16BPP               1
+#       define GRPH_DEPTH_32BPP               2
+/* 8 BPP */
+#       define GRPH_FORMAT_INDEXED            0
+/* 16 BPP */
+#       define GRPH_FORMAT_ARGB1555           0
+#       define GRPH_FORMAT_ARGB565            1
+#       define GRPH_FORMAT_ARGB4444           2
+#       define GRPH_FORMAT_AI88               3
+#       define GRPH_FORMAT_MONO16             4
+#       define GRPH_FORMAT_BGRA5551           5
+/* 32 BPP */
+#       define GRPH_FORMAT_ARGB8888           0
+#       define GRPH_FORMAT_ARGB2101010        1
+#       define GRPH_FORMAT_32BPP_DIG          2
+#       define GRPH_FORMAT_8B_ARGB2101010     3
+#       define GRPH_FORMAT_BGRA1010102        4
+#       define GRPH_FORMAT_8B_BGRA1010102     5
+#       define GRPH_FORMAT_RGB111110          6
+#       define GRPH_FORMAT_BGR101111          7
+#       define ADDR_SURF_MACRO_TILE_ASPECT_1  0
+#       define ADDR_SURF_MACRO_TILE_ASPECT_2  1
+#       define ADDR_SURF_MACRO_TILE_ASPECT_4  2
+#       define ADDR_SURF_MACRO_TILE_ASPECT_8  3
+#       define GRPH_ARRAY_LINEAR_GENERAL      0
+#       define GRPH_ARRAY_LINEAR_ALIGNED      1
+#       define GRPH_ARRAY_1D_TILED_THIN1      2
+#       define GRPH_ARRAY_2D_TILED_THIN1      4
+#       define DISPLAY_MICRO_TILING          0
+#       define THIN_MICRO_TILING             1
+#       define DEPTH_MICRO_TILING            2
+#       define ROTATED_MICRO_TILING          4
+#       define GRPH_ENDIAN_NONE               0
+#       define GRPH_ENDIAN_8IN16              1
+#       define GRPH_ENDIAN_8IN32              2
+#       define GRPH_ENDIAN_8IN64              3
+#       define GRPH_RED_SEL_R                 0
+#       define GRPH_RED_SEL_G                 1
+#       define GRPH_RED_SEL_B                 2
+#       define GRPH_RED_SEL_A                 3
+#       define GRPH_GREEN_SEL_G               0
+#       define GRPH_GREEN_SEL_B               1
+#       define GRPH_GREEN_SEL_A               2
+#       define GRPH_GREEN_SEL_R               3
+#       define GRPH_BLUE_SEL_B                0
+#       define GRPH_BLUE_SEL_A                1
+#       define GRPH_BLUE_SEL_R                2
+#       define GRPH_BLUE_SEL_G                3
+#       define GRPH_ALPHA_SEL_A               0
+#       define GRPH_ALPHA_SEL_R               1
+#       define GRPH_ALPHA_SEL_G               2
+#       define GRPH_ALPHA_SEL_B               3
+#       define INPUT_GAMMA_USE_LUT                  0
+#       define INPUT_GAMMA_BYPASS                   1
+#       define INPUT_GAMMA_SRGB_24                  2
+#       define INPUT_GAMMA_XVYCC_222                3
+
+#       define INPUT_CSC_BYPASS                     0
+#       define INPUT_CSC_PROG_COEFF                 1
+#       define INPUT_CSC_PROG_SHARED_MATRIXA        2
+
+#       define OUTPUT_CSC_BYPASS                    0
+#       define OUTPUT_CSC_TV_RGB                    1
+#       define OUTPUT_CSC_YCBCR_601                 2
+#       define OUTPUT_CSC_YCBCR_709                 3
+#       define OUTPUT_CSC_PROG_COEFF                4
+#       define OUTPUT_CSC_PROG_SHARED_MATRIXB       5
+
+#       define DEGAMMA_BYPASS                       0
+#       define DEGAMMA_SRGB_24                      1
+#       define DEGAMMA_XVYCC_222                    2
+#       define GAMUT_REMAP_BYPASS                   0
+#       define GAMUT_REMAP_PROG_COEFF               1
+#       define GAMUT_REMAP_PROG_SHARED_MATRIXA      2
+#       define GAMUT_REMAP_PROG_SHARED_MATRIXB      3
+
+#       define REGAMMA_BYPASS                       0
+#       define REGAMMA_SRGB_24                      1
+#       define REGAMMA_XVYCC_222                    2
+#       define REGAMMA_PROG_A                       3
+#       define REGAMMA_PROG_B                       4
+
+#       define FMT_CLAMP_6BPC                0
+#       define FMT_CLAMP_8BPC                1
+#       define FMT_CLAMP_10BPC               2
+
+#       define HDMI_24BIT_DEEP_COLOR         0
+#       define HDMI_30BIT_DEEP_COLOR         1
+#       define HDMI_36BIT_DEEP_COLOR         2
+#       define HDMI_ACR_HW                   0
+#       define HDMI_ACR_32                   1
+#       define HDMI_ACR_44                   2
+#       define HDMI_ACR_48                   3
+#       define HDMI_ACR_X1                   1
+#       define HDMI_ACR_X2                   2
+#       define HDMI_ACR_X4                   4
+#       define AFMT_AVI_INFO_Y_RGB           0
+#       define AFMT_AVI_INFO_Y_YCBCR422      1
+#       define AFMT_AVI_INFO_Y_YCBCR444      2
+
+#define			NO_AUTO						0
+#define			ES_AUTO						1
+#define			GS_AUTO						2
+#define			ES_AND_GS_AUTO					3
+
+#       define ARRAY_MODE(x)					((x) << 2)
+#       define PIPE_CONFIG(x)					((x) << 6)
+#       define TILE_SPLIT(x)					((x) << 11)
+#       define MICRO_TILE_MODE_NEW(x)				((x) << 22)
+#       define SAMPLE_SPLIT(x)					((x) << 25)
+#       define BANK_WIDTH(x)					((x) << 0)
+#       define BANK_HEIGHT(x)					((x) << 2)
+#       define MACRO_TILE_ASPECT(x)				((x) << 4)
+#       define NUM_BANKS(x)					((x) << 6)
+
+#define		MSG_ENTER_RLC_SAFE_MODE      			1
+#define		MSG_EXIT_RLC_SAFE_MODE      			0
+
+/*
+ * PM4
+ */
+#define	PACKET_TYPE0	0
+#define	PACKET_TYPE1	1
+#define	PACKET_TYPE2	2
+#define	PACKET_TYPE3	3
+
+#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
+#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
+#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
+#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
+#define PACKET0(reg, n)	((PACKET_TYPE0 << 30) |				\
+			 ((reg) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+
+#define PACKET3(op, n)	((PACKET_TYPE3 << 30) |				\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC_MEM				0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x33
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_PREAMBLE				0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+		/* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - gl2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_CACHE_POLICY(x)              ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+		/* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
+#              define PACKET3_SEM_SEL_SIGNAL_TYPE   (0x1 << 20) /* 0 = increment, 1 = write 1 */
+#              define PACKET3_SEM_CLIENT_CODE	    ((x) << 24) /* 0 = CP, 1 = CB, 2 = DB */
+#              define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#              define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+#define	PACKET3_COPY_DW					0x3B
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+		/* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_OPERATION(x)               ((x) << 6)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+		/* 0 - me
+		 * 1 - pfp
+		 */
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define		INDIRECT_BUFFER_TCL2_VOLATILE           (1 << 22)
+#define		INDIRECT_BUFFER_VALID                   (1 << 23)
+#define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
+#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
+#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
+#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
+#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
+#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
+#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
+#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
+#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
+#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
+#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
+#              define PACKET3_TCL1_VOL_ACTION_ENA  (1 << 15)
+#              define PACKET3_TC_VOL_ACTION_ENA    (1 << 16) /* L2 */
+#              define PACKET3_TC_WB_ACTION_ENA     (1 << 18) /* L2 */
+#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
+#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
+#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
+#              define PACKET3_TC_ACTION_ENA        (1 << 23) /* L2 */
+#              define PACKET3_CB_ACTION_ENA        (1 << 25)
+#              define PACKET3_DB_ACTION_ENA        (1 << 26)
+#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
+#              define PACKET3_SH_KCACHE_VOL_ACTION_ENA (1 << 28)
+#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+		/* 0 - any non-TS event
+		 * 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 * 5 - EOP events
+		 * 6 - EOS events
+		 */
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define		EOP_TCL1_VOL_ACTION_EN                  (1 << 12)
+#define		EOP_TC_VOL_ACTION_EN                    (1 << 13) /* L2 */
+#define		EOP_TC_WB_ACTION_EN                     (1 << 15) /* L2 */
+#define		EOP_TCL1_ACTION_EN                      (1 << 16)
+#define		EOP_TC_ACTION_EN                        (1 << 17) /* L2 */
+#define		EOP_TCL2_VOLATILE                       (1 << 24)
+#define		EOP_CACHE_POLICY(x)                     ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define		DATA_SEL(x)                             ((x) << 29)
+		/* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit GPU counter value
+		 * 4 - send 64bit sys counter value
+		 */
+#define		INT_SEL(x)                              ((x) << 24)
+		/* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define		DST_SEL(x)                              ((x) << 16)
+		/* 0 - MC
+		 * 1 - TC/L2
+		 */
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_RELEASE_MEM				0x49
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+		/* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_SRC_VOLATILE (1 << 15)
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+		/* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_DST_VOLATILE (1 << 27)
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+		/* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_DIS_WC      (1 << 21)
+#              define PACKET3_DMA_DATA_CMD_SRC_SWAP(x) ((x) << 22)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_DST_SWAP(x) ((x) << 24)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_REWIND					0x59
+#define	PACKET3_LOAD_UCONFIG_REG			0x5E
+#define	PACKET3_LOAD_SH_REG				0x5F
+#define	PACKET3_LOAD_CONFIG_REG				0x60
+#define	PACKET3_LOAD_CONTEXT_REG			0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x0000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x0000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_SET_QUEUE_REG				0x78
+#define	PACKET3_SET_UCONFIG_REG				0x79
+#define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
+#define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SWITCH_BUFFER				0x8B
+
+/* SDMA - first instance at 0xd000, second at 0xd800 */
+#define SDMA0_REGISTER_OFFSET                             0x0 /* not a register */
+#define SDMA1_REGISTER_OFFSET                             0x200 /* not a register */
+#define SDMA_MAX_INSTANCE 2
+
+#define SDMA_PACKET(op, sub_op, e)	((((e) & 0xFFFF) << 16) |	\
+					 (((sub_op) & 0xFF) << 8) |	\
+					 (((op) & 0xFF) << 0))
+/* sDMA opcodes */
+#define	SDMA_OPCODE_NOP					  0
+#	define SDMA_NOP_COUNT(x)			  (((x) & 0x3FFF) << 16)
+#define	SDMA_OPCODE_COPY				  1
+#       define SDMA_COPY_SUB_OPCODE_LINEAR                0
+#       define SDMA_COPY_SUB_OPCODE_TILED                 1
+#       define SDMA_COPY_SUB_OPCODE_SOA                   3
+#       define SDMA_COPY_SUB_OPCODE_LINEAR_SUB_WINDOW     4
+#       define SDMA_COPY_SUB_OPCODE_TILED_SUB_WINDOW      5
+#       define SDMA_COPY_SUB_OPCODE_T2T_SUB_WINDOW        6
+#define	SDMA_OPCODE_WRITE				  2
+#       define SDMA_WRITE_SUB_OPCODE_LINEAR               0
+#       define SDMA_WRTIE_SUB_OPCODE_TILED                1
+#define	SDMA_OPCODE_INDIRECT_BUFFER			  4
+#define	SDMA_OPCODE_FENCE				  5
+#define	SDMA_OPCODE_TRAP				  6
+#define	SDMA_OPCODE_SEMAPHORE				  7
+#       define SDMA_SEMAPHORE_EXTRA_O                     (1 << 13)
+		/* 0 - increment
+		 * 1 - write 1
+		 */
+#       define SDMA_SEMAPHORE_EXTRA_S                     (1 << 14)
+		/* 0 - wait
+		 * 1 - signal
+		 */
+#       define SDMA_SEMAPHORE_EXTRA_M                     (1 << 15)
+		/* mailbox */
+#define	SDMA_OPCODE_POLL_REG_MEM			  8
+#       define SDMA_POLL_REG_MEM_EXTRA_OP(x)              ((x) << 10)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#       define SDMA_POLL_REG_MEM_EXTRA_FUNC(x)            ((x) << 12)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#       define SDMA_POLL_REG_MEM_EXTRA_M                  (1 << 15)
+		/* 0 = register
+		 * 1 = memory
+		 */
+#define	SDMA_OPCODE_COND_EXEC				  9
+#define	SDMA_OPCODE_CONSTANT_FILL			  11
+#       define SDMA_CONSTANT_FILL_EXTRA_SIZE(x)           ((x) << 14)
+		/* 0 = byte fill
+		 * 2 = DW fill
+		 */
+#define	SDMA_OPCODE_GENERATE_PTE_PDE			  12
+#define	SDMA_OPCODE_TIMESTAMP				  13
+#       define SDMA_TIMESTAMP_SUB_OPCODE_SET_LOCAL        0
+#       define SDMA_TIMESTAMP_SUB_OPCODE_GET_LOCAL        1
+#       define SDMA_TIMESTAMP_SUB_OPCODE_GET_GLOBAL       2
+#define	SDMA_OPCODE_SRBM_WRITE				  14
+#       define SDMA_SRBM_WRITE_EXTRA_BYTE_ENABLE(x)       ((x) << 12)
+		/* byte mask */
+
+#define VCE_CMD_NO_OP		0x00000000
+#define VCE_CMD_END		0x00000001
+#define VCE_CMD_IB		0x00000002
+#define VCE_CMD_FENCE		0x00000003
+#define VCE_CMD_TRAP		0x00000004
+#define VCE_CMD_IB_AUTO		0x00000005
+#define VCE_CMD_SEMAPHORE	0x00000006
+
+/* if PTR32, these are the bases for scratch and lds */
+#define	PRIVATE_BASE(x)	((x) << 0) /* scratch */
+#define	SHARED_BASE(x)	((x) << 16) /* LDS */
+
+#define KFD_CIK_SDMA_QUEUE_OFFSET	0x200
+
+/* valid for both DEFAULT_MTYPE and APE1_MTYPE */
+enum {
+	MTYPE_CACHED = 0,
+	MTYPE_NONCACHED = 3
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_ci.h b/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_ci.h
new file mode 100644
index 000000000..c3982f947
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_ci.h
@@ -0,0 +1,944 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const unsigned int ci_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0, // HOLE
+    0x00000000, // DB_DEPTH_INFO
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_DEPTH_SIZE
+    0x00000000, // DB_DEPTH_SLICE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // TA_BC_BASE_ADDR
+    0x00000000, // TA_BC_BASE_ADDR_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_HI_0
+    0x00000000, // COHER_DEST_BASE_HI_1
+    0x00000000, // COHER_DEST_BASE_HI_2
+    0x00000000, // COHER_DEST_BASE_HI_3
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+};
+static const unsigned int ci_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+    0, // HOLE
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_RINGID
+    0x00000000, // CP_VMID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0xffffffff, // VGT_MAX_VTX_INDX
+    0x00000000, // VGT_MIN_VTX_INDX
+    0x00000000, // VGT_INDX_OFFSET
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0, // HOLE
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_STENCIL_CONTROL
+    0x00000000, // DB_STENCILREFMASK
+    0x00000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0, // HOLE
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+};
+static const unsigned int ci_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+    0x00000000, // VGT_DMA_BASE_HI
+    0x00000000, // VGT_DMA_BASE
+};
+static const unsigned int ci_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0x00000000, // VGT_OUTPUT_PATH_CNTL
+    0x00000000, // VGT_HOS_CNTL
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0x00000000, // VGT_HOS_REUSE_DEPTH
+    0x00000000, // VGT_GROUP_PRIM_TYPE
+    0x00000000, // VGT_GROUP_FIRST_DECR
+    0x00000000, // VGT_GROUP_DECR
+    0x00000000, // VGT_GROUP_VECT_0_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_CNTL
+    0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+    0x00000000, // VGT_GS_MODE
+    0x00000000, // VGT_GS_ONCHIP_CNTL
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0x00000100, // VGT_GS_PER_ES
+    0x00000080, // VGT_ES_PER_GS
+    0x00000002, // VGT_GS_PER_VS
+    0x00000000, // VGT_GSVS_RING_OFFSET_1
+    0x00000000, // VGT_GSVS_RING_OFFSET_2
+    0x00000000, // VGT_GSVS_RING_OFFSET_3
+    0x00000000, // VGT_GS_OUT_PRIM_TYPE
+    0x00000000, // IA_ENHANCE
+};
+static const unsigned int ci_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // WD_ENHANCE
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int ci_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int ci_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_INSTANCE_STEP_RATE_0
+    0x00000000, // VGT_INSTANCE_STEP_RATE_1
+    0x000000ff, // IA_MULTI_VGT_PARAM
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0x00000000, // VGT_GSVS_RING_ITEMSIZE
+    0x00000000, // VGT_REUSE_OFF
+    0x00000000, // VGT_VTX_CNT_EN
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0x00000000, // VGT_GS_VERT_ITEMSIZE
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0, // HOLE
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0x00000000, // VGT_STRMOUT_CONFIG
+    0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x0000000e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+    0x00000010, // VGT_OUT_DEALLOC_CNTL
+    0x00000000, // CB_COLOR0_BASE
+    0x00000000, // CB_COLOR0_PITCH
+    0x00000000, // CB_COLOR0_SLICE
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR0_CMASK
+    0x00000000, // CB_COLOR0_CMASK_SLICE
+    0x00000000, // CB_COLOR0_FMASK
+    0x00000000, // CB_COLOR0_FMASK_SLICE
+    0x00000000, // CB_COLOR0_CLEAR_WORD0
+    0x00000000, // CB_COLOR0_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_BASE
+    0x00000000, // CB_COLOR1_PITCH
+    0x00000000, // CB_COLOR1_SLICE
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR1_CMASK
+    0x00000000, // CB_COLOR1_CMASK_SLICE
+    0x00000000, // CB_COLOR1_FMASK
+    0x00000000, // CB_COLOR1_FMASK_SLICE
+    0x00000000, // CB_COLOR1_CLEAR_WORD0
+    0x00000000, // CB_COLOR1_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_BASE
+    0x00000000, // CB_COLOR2_PITCH
+    0x00000000, // CB_COLOR2_SLICE
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR2_CMASK
+    0x00000000, // CB_COLOR2_CMASK_SLICE
+    0x00000000, // CB_COLOR2_FMASK
+    0x00000000, // CB_COLOR2_FMASK_SLICE
+    0x00000000, // CB_COLOR2_CLEAR_WORD0
+    0x00000000, // CB_COLOR2_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_BASE
+    0x00000000, // CB_COLOR3_PITCH
+    0x00000000, // CB_COLOR3_SLICE
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR3_CMASK
+    0x00000000, // CB_COLOR3_CMASK_SLICE
+    0x00000000, // CB_COLOR3_FMASK
+    0x00000000, // CB_COLOR3_FMASK_SLICE
+    0x00000000, // CB_COLOR3_CLEAR_WORD0
+    0x00000000, // CB_COLOR3_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_BASE
+    0x00000000, // CB_COLOR4_PITCH
+    0x00000000, // CB_COLOR4_SLICE
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR4_CMASK
+    0x00000000, // CB_COLOR4_CMASK_SLICE
+    0x00000000, // CB_COLOR4_FMASK
+    0x00000000, // CB_COLOR4_FMASK_SLICE
+    0x00000000, // CB_COLOR4_CLEAR_WORD0
+    0x00000000, // CB_COLOR4_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_BASE
+    0x00000000, // CB_COLOR5_PITCH
+    0x00000000, // CB_COLOR5_SLICE
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR5_CMASK
+    0x00000000, // CB_COLOR5_CMASK_SLICE
+    0x00000000, // CB_COLOR5_FMASK
+    0x00000000, // CB_COLOR5_FMASK_SLICE
+    0x00000000, // CB_COLOR5_CLEAR_WORD0
+    0x00000000, // CB_COLOR5_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_BASE
+    0x00000000, // CB_COLOR6_PITCH
+    0x00000000, // CB_COLOR6_SLICE
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR6_CMASK
+    0x00000000, // CB_COLOR6_CMASK_SLICE
+    0x00000000, // CB_COLOR6_FMASK
+    0x00000000, // CB_COLOR6_FMASK_SLICE
+    0x00000000, // CB_COLOR6_CLEAR_WORD0
+    0x00000000, // CB_COLOR6_CLEAR_WORD1
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_BASE
+    0x00000000, // CB_COLOR7_PITCH
+    0x00000000, // CB_COLOR7_SLICE
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0, // HOLE
+    0x00000000, // CB_COLOR7_CMASK
+    0x00000000, // CB_COLOR7_CMASK_SLICE
+    0x00000000, // CB_COLOR7_FMASK
+    0x00000000, // CB_COLOR7_FMASK_SLICE
+    0x00000000, // CB_COLOR7_CLEAR_WORD0
+    0x00000000, // CB_COLOR7_CLEAR_WORD1
+};
+static const struct cs_extent_def ci_SECT_CONTEXT_defs[] =
+{
+    {ci_SECT_CONTEXT_def_1, 0x0000a000, 212 },
+    {ci_SECT_CONTEXT_def_2, 0x0000a0d6, 274 },
+    {ci_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
+    {ci_SECT_CONTEXT_def_4, 0x0000a200, 157 },
+    {ci_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+    {ci_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {ci_SECT_CONTEXT_def_7, 0x0000a2a5, 233 },
+    { 0, 0, 0 }
+};
+static const struct cs_section_def ci_cs_data[] = {
+    { ci_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_defs.h b/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_defs.h
new file mode 100644
index 000000000..3eda707d7
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_defs.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2012 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef CLEARSTATE_DEFS_H
+#define CLEARSTATE_DEFS_H
+
+enum section_id {
+    SECT_NONE,
+    SECT_CONTEXT,
+    SECT_CLEAR,
+    SECT_CTRLCONST
+};
+
+struct cs_extent_def {
+    const unsigned int *extent;
+    const unsigned int reg_index;
+    const unsigned int reg_count;
+};
+
+struct cs_section_def {
+    const struct cs_extent_def *section;
+    const enum section_id id;
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_vi.h b/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_vi.h
new file mode 100644
index 000000000..1aab9bef9
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/clearstate_vi.h
@@ -0,0 +1,944 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+static const unsigned int vi_SECT_CONTEXT_def_1[] =
+{
+    0x00000000, // DB_RENDER_CONTROL
+    0x00000000, // DB_COUNT_CONTROL
+    0x00000000, // DB_DEPTH_VIEW
+    0x00000000, // DB_RENDER_OVERRIDE
+    0x00000000, // DB_RENDER_OVERRIDE2
+    0x00000000, // DB_HTILE_DATA_BASE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // DB_DEPTH_BOUNDS_MIN
+    0x00000000, // DB_DEPTH_BOUNDS_MAX
+    0x00000000, // DB_STENCIL_CLEAR
+    0x00000000, // DB_DEPTH_CLEAR
+    0x00000000, // PA_SC_SCREEN_SCISSOR_TL
+    0x40004000, // PA_SC_SCREEN_SCISSOR_BR
+    0, // HOLE
+    0x00000000, // DB_DEPTH_INFO
+    0x00000000, // DB_Z_INFO
+    0x00000000, // DB_STENCIL_INFO
+    0x00000000, // DB_Z_READ_BASE
+    0x00000000, // DB_STENCIL_READ_BASE
+    0x00000000, // DB_Z_WRITE_BASE
+    0x00000000, // DB_STENCIL_WRITE_BASE
+    0x00000000, // DB_DEPTH_SIZE
+    0x00000000, // DB_DEPTH_SLICE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // TA_BC_BASE_ADDR
+    0x00000000, // TA_BC_BASE_ADDR_HI
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // COHER_DEST_BASE_HI_0
+    0x00000000, // COHER_DEST_BASE_HI_1
+    0x00000000, // COHER_DEST_BASE_HI_2
+    0x00000000, // COHER_DEST_BASE_HI_3
+    0x00000000, // COHER_DEST_BASE_2
+    0x00000000, // COHER_DEST_BASE_3
+    0x00000000, // PA_SC_WINDOW_OFFSET
+    0x80000000, // PA_SC_WINDOW_SCISSOR_TL
+    0x40004000, // PA_SC_WINDOW_SCISSOR_BR
+    0x0000ffff, // PA_SC_CLIPRECT_RULE
+    0x00000000, // PA_SC_CLIPRECT_0_TL
+    0x40004000, // PA_SC_CLIPRECT_0_BR
+    0x00000000, // PA_SC_CLIPRECT_1_TL
+    0x40004000, // PA_SC_CLIPRECT_1_BR
+    0x00000000, // PA_SC_CLIPRECT_2_TL
+    0x40004000, // PA_SC_CLIPRECT_2_BR
+    0x00000000, // PA_SC_CLIPRECT_3_TL
+    0x40004000, // PA_SC_CLIPRECT_3_BR
+    0xaa99aaaa, // PA_SC_EDGERULE
+    0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
+    0xffffffff, // CB_TARGET_MASK
+    0xffffffff, // CB_SHADER_MASK
+    0x80000000, // PA_SC_GENERIC_SCISSOR_TL
+    0x40004000, // PA_SC_GENERIC_SCISSOR_BR
+    0x00000000, // COHER_DEST_BASE_0
+    0x00000000, // COHER_DEST_BASE_1
+    0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
+    0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
+    0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
+    0x00000000, // PA_SC_VPORT_ZMIN_0
+    0x3f800000, // PA_SC_VPORT_ZMAX_0
+    0x00000000, // PA_SC_VPORT_ZMIN_1
+    0x3f800000, // PA_SC_VPORT_ZMAX_1
+    0x00000000, // PA_SC_VPORT_ZMIN_2
+    0x3f800000, // PA_SC_VPORT_ZMAX_2
+    0x00000000, // PA_SC_VPORT_ZMIN_3
+    0x3f800000, // PA_SC_VPORT_ZMAX_3
+    0x00000000, // PA_SC_VPORT_ZMIN_4
+    0x3f800000, // PA_SC_VPORT_ZMAX_4
+    0x00000000, // PA_SC_VPORT_ZMIN_5
+    0x3f800000, // PA_SC_VPORT_ZMAX_5
+    0x00000000, // PA_SC_VPORT_ZMIN_6
+    0x3f800000, // PA_SC_VPORT_ZMAX_6
+    0x00000000, // PA_SC_VPORT_ZMIN_7
+    0x3f800000, // PA_SC_VPORT_ZMAX_7
+    0x00000000, // PA_SC_VPORT_ZMIN_8
+    0x3f800000, // PA_SC_VPORT_ZMAX_8
+    0x00000000, // PA_SC_VPORT_ZMIN_9
+    0x3f800000, // PA_SC_VPORT_ZMAX_9
+    0x00000000, // PA_SC_VPORT_ZMIN_10
+    0x3f800000, // PA_SC_VPORT_ZMAX_10
+    0x00000000, // PA_SC_VPORT_ZMIN_11
+    0x3f800000, // PA_SC_VPORT_ZMAX_11
+    0x00000000, // PA_SC_VPORT_ZMIN_12
+    0x3f800000, // PA_SC_VPORT_ZMAX_12
+    0x00000000, // PA_SC_VPORT_ZMIN_13
+    0x3f800000, // PA_SC_VPORT_ZMAX_13
+    0x00000000, // PA_SC_VPORT_ZMIN_14
+    0x3f800000, // PA_SC_VPORT_ZMAX_14
+    0x00000000, // PA_SC_VPORT_ZMIN_15
+    0x3f800000, // PA_SC_VPORT_ZMAX_15
+};
+static const unsigned int vi_SECT_CONTEXT_def_2[] =
+{
+    0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
+    0, // HOLE
+    0x00000000, // CP_PERFMON_CNTX_CNTL
+    0x00000000, // CP_RINGID
+    0x00000000, // CP_VMID
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0xffffffff, // VGT_MAX_VTX_INDX
+    0x00000000, // VGT_MIN_VTX_INDX
+    0x00000000, // VGT_INDX_OFFSET
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
+    0, // HOLE
+    0x00000000, // CB_BLEND_RED
+    0x00000000, // CB_BLEND_GREEN
+    0x00000000, // CB_BLEND_BLUE
+    0x00000000, // CB_BLEND_ALPHA
+    0x00000000, // CB_DCC_CONTROL
+    0, // HOLE
+    0x00000000, // DB_STENCIL_CONTROL
+    0x00000000, // DB_STENCILREFMASK
+    0x00000000, // DB_STENCILREFMASK_BF
+    0, // HOLE
+    0x00000000, // PA_CL_VPORT_XSCALE
+    0x00000000, // PA_CL_VPORT_XOFFSET
+    0x00000000, // PA_CL_VPORT_YSCALE
+    0x00000000, // PA_CL_VPORT_YOFFSET
+    0x00000000, // PA_CL_VPORT_ZSCALE
+    0x00000000, // PA_CL_VPORT_ZOFFSET
+    0x00000000, // PA_CL_VPORT_XSCALE_1
+    0x00000000, // PA_CL_VPORT_XOFFSET_1
+    0x00000000, // PA_CL_VPORT_YSCALE_1
+    0x00000000, // PA_CL_VPORT_YOFFSET_1
+    0x00000000, // PA_CL_VPORT_ZSCALE_1
+    0x00000000, // PA_CL_VPORT_ZOFFSET_1
+    0x00000000, // PA_CL_VPORT_XSCALE_2
+    0x00000000, // PA_CL_VPORT_XOFFSET_2
+    0x00000000, // PA_CL_VPORT_YSCALE_2
+    0x00000000, // PA_CL_VPORT_YOFFSET_2
+    0x00000000, // PA_CL_VPORT_ZSCALE_2
+    0x00000000, // PA_CL_VPORT_ZOFFSET_2
+    0x00000000, // PA_CL_VPORT_XSCALE_3
+    0x00000000, // PA_CL_VPORT_XOFFSET_3
+    0x00000000, // PA_CL_VPORT_YSCALE_3
+    0x00000000, // PA_CL_VPORT_YOFFSET_3
+    0x00000000, // PA_CL_VPORT_ZSCALE_3
+    0x00000000, // PA_CL_VPORT_ZOFFSET_3
+    0x00000000, // PA_CL_VPORT_XSCALE_4
+    0x00000000, // PA_CL_VPORT_XOFFSET_4
+    0x00000000, // PA_CL_VPORT_YSCALE_4
+    0x00000000, // PA_CL_VPORT_YOFFSET_4
+    0x00000000, // PA_CL_VPORT_ZSCALE_4
+    0x00000000, // PA_CL_VPORT_ZOFFSET_4
+    0x00000000, // PA_CL_VPORT_XSCALE_5
+    0x00000000, // PA_CL_VPORT_XOFFSET_5
+    0x00000000, // PA_CL_VPORT_YSCALE_5
+    0x00000000, // PA_CL_VPORT_YOFFSET_5
+    0x00000000, // PA_CL_VPORT_ZSCALE_5
+    0x00000000, // PA_CL_VPORT_ZOFFSET_5
+    0x00000000, // PA_CL_VPORT_XSCALE_6
+    0x00000000, // PA_CL_VPORT_XOFFSET_6
+    0x00000000, // PA_CL_VPORT_YSCALE_6
+    0x00000000, // PA_CL_VPORT_YOFFSET_6
+    0x00000000, // PA_CL_VPORT_ZSCALE_6
+    0x00000000, // PA_CL_VPORT_ZOFFSET_6
+    0x00000000, // PA_CL_VPORT_XSCALE_7
+    0x00000000, // PA_CL_VPORT_XOFFSET_7
+    0x00000000, // PA_CL_VPORT_YSCALE_7
+    0x00000000, // PA_CL_VPORT_YOFFSET_7
+    0x00000000, // PA_CL_VPORT_ZSCALE_7
+    0x00000000, // PA_CL_VPORT_ZOFFSET_7
+    0x00000000, // PA_CL_VPORT_XSCALE_8
+    0x00000000, // PA_CL_VPORT_XOFFSET_8
+    0x00000000, // PA_CL_VPORT_YSCALE_8
+    0x00000000, // PA_CL_VPORT_YOFFSET_8
+    0x00000000, // PA_CL_VPORT_ZSCALE_8
+    0x00000000, // PA_CL_VPORT_ZOFFSET_8
+    0x00000000, // PA_CL_VPORT_XSCALE_9
+    0x00000000, // PA_CL_VPORT_XOFFSET_9
+    0x00000000, // PA_CL_VPORT_YSCALE_9
+    0x00000000, // PA_CL_VPORT_YOFFSET_9
+    0x00000000, // PA_CL_VPORT_ZSCALE_9
+    0x00000000, // PA_CL_VPORT_ZOFFSET_9
+    0x00000000, // PA_CL_VPORT_XSCALE_10
+    0x00000000, // PA_CL_VPORT_XOFFSET_10
+    0x00000000, // PA_CL_VPORT_YSCALE_10
+    0x00000000, // PA_CL_VPORT_YOFFSET_10
+    0x00000000, // PA_CL_VPORT_ZSCALE_10
+    0x00000000, // PA_CL_VPORT_ZOFFSET_10
+    0x00000000, // PA_CL_VPORT_XSCALE_11
+    0x00000000, // PA_CL_VPORT_XOFFSET_11
+    0x00000000, // PA_CL_VPORT_YSCALE_11
+    0x00000000, // PA_CL_VPORT_YOFFSET_11
+    0x00000000, // PA_CL_VPORT_ZSCALE_11
+    0x00000000, // PA_CL_VPORT_ZOFFSET_11
+    0x00000000, // PA_CL_VPORT_XSCALE_12
+    0x00000000, // PA_CL_VPORT_XOFFSET_12
+    0x00000000, // PA_CL_VPORT_YSCALE_12
+    0x00000000, // PA_CL_VPORT_YOFFSET_12
+    0x00000000, // PA_CL_VPORT_ZSCALE_12
+    0x00000000, // PA_CL_VPORT_ZOFFSET_12
+    0x00000000, // PA_CL_VPORT_XSCALE_13
+    0x00000000, // PA_CL_VPORT_XOFFSET_13
+    0x00000000, // PA_CL_VPORT_YSCALE_13
+    0x00000000, // PA_CL_VPORT_YOFFSET_13
+    0x00000000, // PA_CL_VPORT_ZSCALE_13
+    0x00000000, // PA_CL_VPORT_ZOFFSET_13
+    0x00000000, // PA_CL_VPORT_XSCALE_14
+    0x00000000, // PA_CL_VPORT_XOFFSET_14
+    0x00000000, // PA_CL_VPORT_YSCALE_14
+    0x00000000, // PA_CL_VPORT_YOFFSET_14
+    0x00000000, // PA_CL_VPORT_ZSCALE_14
+    0x00000000, // PA_CL_VPORT_ZOFFSET_14
+    0x00000000, // PA_CL_VPORT_XSCALE_15
+    0x00000000, // PA_CL_VPORT_XOFFSET_15
+    0x00000000, // PA_CL_VPORT_YSCALE_15
+    0x00000000, // PA_CL_VPORT_YOFFSET_15
+    0x00000000, // PA_CL_VPORT_ZSCALE_15
+    0x00000000, // PA_CL_VPORT_ZOFFSET_15
+    0x00000000, // PA_CL_UCP_0_X
+    0x00000000, // PA_CL_UCP_0_Y
+    0x00000000, // PA_CL_UCP_0_Z
+    0x00000000, // PA_CL_UCP_0_W
+    0x00000000, // PA_CL_UCP_1_X
+    0x00000000, // PA_CL_UCP_1_Y
+    0x00000000, // PA_CL_UCP_1_Z
+    0x00000000, // PA_CL_UCP_1_W
+    0x00000000, // PA_CL_UCP_2_X
+    0x00000000, // PA_CL_UCP_2_Y
+    0x00000000, // PA_CL_UCP_2_Z
+    0x00000000, // PA_CL_UCP_2_W
+    0x00000000, // PA_CL_UCP_3_X
+    0x00000000, // PA_CL_UCP_3_Y
+    0x00000000, // PA_CL_UCP_3_Z
+    0x00000000, // PA_CL_UCP_3_W
+    0x00000000, // PA_CL_UCP_4_X
+    0x00000000, // PA_CL_UCP_4_Y
+    0x00000000, // PA_CL_UCP_4_Z
+    0x00000000, // PA_CL_UCP_4_W
+    0x00000000, // PA_CL_UCP_5_X
+    0x00000000, // PA_CL_UCP_5_Y
+    0x00000000, // PA_CL_UCP_5_Z
+    0x00000000, // PA_CL_UCP_5_W
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_CNTL_0
+    0x00000000, // SPI_PS_INPUT_CNTL_1
+    0x00000000, // SPI_PS_INPUT_CNTL_2
+    0x00000000, // SPI_PS_INPUT_CNTL_3
+    0x00000000, // SPI_PS_INPUT_CNTL_4
+    0x00000000, // SPI_PS_INPUT_CNTL_5
+    0x00000000, // SPI_PS_INPUT_CNTL_6
+    0x00000000, // SPI_PS_INPUT_CNTL_7
+    0x00000000, // SPI_PS_INPUT_CNTL_8
+    0x00000000, // SPI_PS_INPUT_CNTL_9
+    0x00000000, // SPI_PS_INPUT_CNTL_10
+    0x00000000, // SPI_PS_INPUT_CNTL_11
+    0x00000000, // SPI_PS_INPUT_CNTL_12
+    0x00000000, // SPI_PS_INPUT_CNTL_13
+    0x00000000, // SPI_PS_INPUT_CNTL_14
+    0x00000000, // SPI_PS_INPUT_CNTL_15
+    0x00000000, // SPI_PS_INPUT_CNTL_16
+    0x00000000, // SPI_PS_INPUT_CNTL_17
+    0x00000000, // SPI_PS_INPUT_CNTL_18
+    0x00000000, // SPI_PS_INPUT_CNTL_19
+    0x00000000, // SPI_PS_INPUT_CNTL_20
+    0x00000000, // SPI_PS_INPUT_CNTL_21
+    0x00000000, // SPI_PS_INPUT_CNTL_22
+    0x00000000, // SPI_PS_INPUT_CNTL_23
+    0x00000000, // SPI_PS_INPUT_CNTL_24
+    0x00000000, // SPI_PS_INPUT_CNTL_25
+    0x00000000, // SPI_PS_INPUT_CNTL_26
+    0x00000000, // SPI_PS_INPUT_CNTL_27
+    0x00000000, // SPI_PS_INPUT_CNTL_28
+    0x00000000, // SPI_PS_INPUT_CNTL_29
+    0x00000000, // SPI_PS_INPUT_CNTL_30
+    0x00000000, // SPI_PS_INPUT_CNTL_31
+    0x00000000, // SPI_VS_OUT_CONFIG
+    0, // HOLE
+    0x00000000, // SPI_PS_INPUT_ENA
+    0x00000000, // SPI_PS_INPUT_ADDR
+    0x00000000, // SPI_INTERP_CONTROL_0
+    0x00000002, // SPI_PS_IN_CONTROL
+    0, // HOLE
+    0x00000000, // SPI_BARYC_CNTL
+    0, // HOLE
+    0x00000000, // SPI_TMPRING_SIZE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // SPI_SHADER_POS_FORMAT
+    0x00000000, // SPI_SHADER_Z_FORMAT
+    0x00000000, // SPI_SHADER_COL_FORMAT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // CB_BLEND0_CONTROL
+    0x00000000, // CB_BLEND1_CONTROL
+    0x00000000, // CB_BLEND2_CONTROL
+    0x00000000, // CB_BLEND3_CONTROL
+    0x00000000, // CB_BLEND4_CONTROL
+    0x00000000, // CB_BLEND5_CONTROL
+    0x00000000, // CB_BLEND6_CONTROL
+    0x00000000, // CB_BLEND7_CONTROL
+};
+static const unsigned int vi_SECT_CONTEXT_def_3[] =
+{
+    0x00000000, // PA_CL_POINT_X_RAD
+    0x00000000, // PA_CL_POINT_Y_RAD
+    0x00000000, // PA_CL_POINT_SIZE
+    0x00000000, // PA_CL_POINT_CULL_RAD
+    0x00000000, // VGT_DMA_BASE_HI
+    0x00000000, // VGT_DMA_BASE
+};
+static const unsigned int vi_SECT_CONTEXT_def_4[] =
+{
+    0x00000000, // DB_DEPTH_CONTROL
+    0x00000000, // DB_EQAA
+    0x00000000, // CB_COLOR_CONTROL
+    0x00000000, // DB_SHADER_CONTROL
+    0x00090000, // PA_CL_CLIP_CNTL
+    0x00000004, // PA_SU_SC_MODE_CNTL
+    0x00000000, // PA_CL_VTE_CNTL
+    0x00000000, // PA_CL_VS_OUT_CNTL
+    0x00000000, // PA_CL_NANINF_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_CNTL
+    0x00000000, // PA_SU_LINE_STIPPLE_SCALE
+    0x00000000, // PA_SU_PRIM_FILTER_CNTL
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SU_POINT_SIZE
+    0x00000000, // PA_SU_POINT_MINMAX
+    0x00000000, // PA_SU_LINE_CNTL
+    0x00000000, // PA_SC_LINE_STIPPLE
+    0x00000000, // VGT_OUTPUT_PATH_CNTL
+    0x00000000, // VGT_HOS_CNTL
+    0x00000000, // VGT_HOS_MAX_TESS_LEVEL
+    0x00000000, // VGT_HOS_MIN_TESS_LEVEL
+    0x00000000, // VGT_HOS_REUSE_DEPTH
+    0x00000000, // VGT_GROUP_PRIM_TYPE
+    0x00000000, // VGT_GROUP_FIRST_DECR
+    0x00000000, // VGT_GROUP_DECR
+    0x00000000, // VGT_GROUP_VECT_0_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_CNTL
+    0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
+    0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
+    0x00000000, // VGT_GS_MODE
+    0x00000000, // VGT_GS_ONCHIP_CNTL
+    0x00000000, // PA_SC_MODE_CNTL_0
+    0x00000000, // PA_SC_MODE_CNTL_1
+    0x00000000, // VGT_ENHANCE
+    0x00000100, // VGT_GS_PER_ES
+    0x00000080, // VGT_ES_PER_GS
+    0x00000002, // VGT_GS_PER_VS
+    0x00000000, // VGT_GSVS_RING_OFFSET_1
+    0x00000000, // VGT_GSVS_RING_OFFSET_2
+    0x00000000, // VGT_GSVS_RING_OFFSET_3
+    0x00000000, // VGT_GS_OUT_PRIM_TYPE
+    0x00000000, // IA_ENHANCE
+};
+static const unsigned int vi_SECT_CONTEXT_def_5[] =
+{
+    0x00000000, // WD_ENHANCE
+    0x00000000, // VGT_PRIMITIVEID_EN
+};
+static const unsigned int vi_SECT_CONTEXT_def_6[] =
+{
+    0x00000000, // VGT_PRIMITIVEID_RESET
+};
+static const unsigned int vi_SECT_CONTEXT_def_7[] =
+{
+    0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_INSTANCE_STEP_RATE_0
+    0x00000000, // VGT_INSTANCE_STEP_RATE_1
+    0x000000ff, // IA_MULTI_VGT_PARAM
+    0x00000000, // VGT_ESGS_RING_ITEMSIZE
+    0x00000000, // VGT_GSVS_RING_ITEMSIZE
+    0x00000000, // VGT_REUSE_OFF
+    0x00000000, // VGT_VTX_CNT_EN
+    0x00000000, // DB_HTILE_SURFACE
+    0x00000000, // DB_SRESULTS_COMPARE_STATE0
+    0x00000000, // DB_SRESULTS_COMPARE_STATE1
+    0x00000000, // DB_PRELOAD_CONTROL
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
+    0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
+    0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
+    0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
+    0, // HOLE
+    0x00000000, // VGT_GS_MAX_VERT_OUT
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // VGT_TESS_DISTRIBUTION
+    0x00000000, // VGT_SHADER_STAGES_EN
+    0x00000000, // VGT_LS_HS_CONFIG
+    0x00000000, // VGT_GS_VERT_ITEMSIZE
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_1
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_2
+    0x00000000, // VGT_GS_VERT_ITEMSIZE_3
+    0x00000000, // VGT_TF_PARAM
+    0x00000000, // DB_ALPHA_TO_MASK
+    0x00000000, // VGT_DISPATCH_DRAW_INDEX
+    0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
+    0x00000000, // PA_SU_POLY_OFFSET_CLAMP
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
+    0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
+    0x00000000, // VGT_GS_INSTANCE_CNT
+    0x00000000, // VGT_STRMOUT_CONFIG
+    0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x00000000, // PA_SC_CENTROID_PRIORITY_0
+    0x00000000, // PA_SC_CENTROID_PRIORITY_1
+    0x00001000, // PA_SC_LINE_CNTL
+    0x00000000, // PA_SC_AA_CONFIG
+    0x00000005, // PA_SU_VTX_CNTL
+    0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
+    0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
+    0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
+    0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
+    0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0, // HOLE
+    0x0000001e, // VGT_VERTEX_REUSE_BLOCK_CNTL
+    0x00000020, // VGT_OUT_DEALLOC_CNTL
+    0x00000000, // CB_COLOR0_BASE
+    0x00000000, // CB_COLOR0_PITCH
+    0x00000000, // CB_COLOR0_SLICE
+    0x00000000, // CB_COLOR0_VIEW
+    0x00000000, // CB_COLOR0_INFO
+    0x00000000, // CB_COLOR0_ATTRIB
+    0x00000000, // CB_COLOR0_DCC_CONTROL
+    0x00000000, // CB_COLOR0_CMASK
+    0x00000000, // CB_COLOR0_CMASK_SLICE
+    0x00000000, // CB_COLOR0_FMASK
+    0x00000000, // CB_COLOR0_FMASK_SLICE
+    0x00000000, // CB_COLOR0_CLEAR_WORD0
+    0x00000000, // CB_COLOR0_CLEAR_WORD1
+    0x00000000, // CB_COLOR0_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR1_BASE
+    0x00000000, // CB_COLOR1_PITCH
+    0x00000000, // CB_COLOR1_SLICE
+    0x00000000, // CB_COLOR1_VIEW
+    0x00000000, // CB_COLOR1_INFO
+    0x00000000, // CB_COLOR1_ATTRIB
+    0x00000000, // CB_COLOR1_DCC_CONTROL
+    0x00000000, // CB_COLOR1_CMASK
+    0x00000000, // CB_COLOR1_CMASK_SLICE
+    0x00000000, // CB_COLOR1_FMASK
+    0x00000000, // CB_COLOR1_FMASK_SLICE
+    0x00000000, // CB_COLOR1_CLEAR_WORD0
+    0x00000000, // CB_COLOR1_CLEAR_WORD1
+    0x00000000, // CB_COLOR1_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR2_BASE
+    0x00000000, // CB_COLOR2_PITCH
+    0x00000000, // CB_COLOR2_SLICE
+    0x00000000, // CB_COLOR2_VIEW
+    0x00000000, // CB_COLOR2_INFO
+    0x00000000, // CB_COLOR2_ATTRIB
+    0x00000000, // CB_COLOR2_DCC_CONTROL
+    0x00000000, // CB_COLOR2_CMASK
+    0x00000000, // CB_COLOR2_CMASK_SLICE
+    0x00000000, // CB_COLOR2_FMASK
+    0x00000000, // CB_COLOR2_FMASK_SLICE
+    0x00000000, // CB_COLOR2_CLEAR_WORD0
+    0x00000000, // CB_COLOR2_CLEAR_WORD1
+    0x00000000, // CB_COLOR2_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR3_BASE
+    0x00000000, // CB_COLOR3_PITCH
+    0x00000000, // CB_COLOR3_SLICE
+    0x00000000, // CB_COLOR3_VIEW
+    0x00000000, // CB_COLOR3_INFO
+    0x00000000, // CB_COLOR3_ATTRIB
+    0x00000000, // CB_COLOR3_DCC_CONTROL
+    0x00000000, // CB_COLOR3_CMASK
+    0x00000000, // CB_COLOR3_CMASK_SLICE
+    0x00000000, // CB_COLOR3_FMASK
+    0x00000000, // CB_COLOR3_FMASK_SLICE
+    0x00000000, // CB_COLOR3_CLEAR_WORD0
+    0x00000000, // CB_COLOR3_CLEAR_WORD1
+    0x00000000, // CB_COLOR3_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR4_BASE
+    0x00000000, // CB_COLOR4_PITCH
+    0x00000000, // CB_COLOR4_SLICE
+    0x00000000, // CB_COLOR4_VIEW
+    0x00000000, // CB_COLOR4_INFO
+    0x00000000, // CB_COLOR4_ATTRIB
+    0x00000000, // CB_COLOR4_DCC_CONTROL
+    0x00000000, // CB_COLOR4_CMASK
+    0x00000000, // CB_COLOR4_CMASK_SLICE
+    0x00000000, // CB_COLOR4_FMASK
+    0x00000000, // CB_COLOR4_FMASK_SLICE
+    0x00000000, // CB_COLOR4_CLEAR_WORD0
+    0x00000000, // CB_COLOR4_CLEAR_WORD1
+    0x00000000, // CB_COLOR4_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR5_BASE
+    0x00000000, // CB_COLOR5_PITCH
+    0x00000000, // CB_COLOR5_SLICE
+    0x00000000, // CB_COLOR5_VIEW
+    0x00000000, // CB_COLOR5_INFO
+    0x00000000, // CB_COLOR5_ATTRIB
+    0x00000000, // CB_COLOR5_DCC_CONTROL
+    0x00000000, // CB_COLOR5_CMASK
+    0x00000000, // CB_COLOR5_CMASK_SLICE
+    0x00000000, // CB_COLOR5_FMASK
+    0x00000000, // CB_COLOR5_FMASK_SLICE
+    0x00000000, // CB_COLOR5_CLEAR_WORD0
+    0x00000000, // CB_COLOR5_CLEAR_WORD1
+    0x00000000, // CB_COLOR5_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR6_BASE
+    0x00000000, // CB_COLOR6_PITCH
+    0x00000000, // CB_COLOR6_SLICE
+    0x00000000, // CB_COLOR6_VIEW
+    0x00000000, // CB_COLOR6_INFO
+    0x00000000, // CB_COLOR6_ATTRIB
+    0x00000000, // CB_COLOR6_DCC_CONTROL
+    0x00000000, // CB_COLOR6_CMASK
+    0x00000000, // CB_COLOR6_CMASK_SLICE
+    0x00000000, // CB_COLOR6_FMASK
+    0x00000000, // CB_COLOR6_FMASK_SLICE
+    0x00000000, // CB_COLOR6_CLEAR_WORD0
+    0x00000000, // CB_COLOR6_CLEAR_WORD1
+    0x00000000, // CB_COLOR6_DCC_BASE
+    0, // HOLE
+    0x00000000, // CB_COLOR7_BASE
+    0x00000000, // CB_COLOR7_PITCH
+    0x00000000, // CB_COLOR7_SLICE
+    0x00000000, // CB_COLOR7_VIEW
+    0x00000000, // CB_COLOR7_INFO
+    0x00000000, // CB_COLOR7_ATTRIB
+    0x00000000, // CB_COLOR7_DCC_CONTROL
+    0x00000000, // CB_COLOR7_CMASK
+    0x00000000, // CB_COLOR7_CMASK_SLICE
+    0x00000000, // CB_COLOR7_FMASK
+    0x00000000, // CB_COLOR7_FMASK_SLICE
+    0x00000000, // CB_COLOR7_CLEAR_WORD0
+    0x00000000, // CB_COLOR7_CLEAR_WORD1
+};
+static const struct cs_extent_def vi_SECT_CONTEXT_defs[] =
+{
+    {vi_SECT_CONTEXT_def_1, 0x0000a000, 212 },
+    {vi_SECT_CONTEXT_def_2, 0x0000a0d6, 274 },
+    {vi_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
+    {vi_SECT_CONTEXT_def_4, 0x0000a200, 157 },
+    {vi_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
+    {vi_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
+    {vi_SECT_CONTEXT_def_7, 0x0000a2a5, 233 },
+    { 0, 0, 0 }
+};
+static const struct cs_section_def vi_cs_data[] = {
+    { vi_SECT_CONTEXT_defs, SECT_CONTEXT },
+    { 0, SECT_NONE }
+};
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
new file mode 100644
index 000000000..8035d4d6a
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
@@ -0,0 +1,2002 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include <linux/seq_file.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_atombios.h"
+#include "vid.h"
+#include "vi_dpm.h"
+#include "amdgpu_dpm.h"
+#include "cz_dpm.h"
+#include "cz_ppsmc.h"
+#include "atom.h"
+
+#include "smu/smu_8_0_d.h"
+#include "smu/smu_8_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "bif/bif_5_1_d.h"
+#include "gfx_v8_0.h"
+
+static void cz_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate);
+static void cz_dpm_powergate_vce(struct amdgpu_device *adev, bool gate);
+
+static struct cz_ps *cz_get_ps(struct amdgpu_ps *rps)
+{
+	struct cz_ps *ps = rps->ps_priv;
+
+	return ps;
+}
+
+static struct cz_power_info *cz_get_pi(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+static uint16_t cz_convert_8bit_index_to_voltage(struct amdgpu_device *adev,
+							uint16_t voltage)
+{
+	uint16_t tmp = 6200 - voltage * 25;
+
+	return tmp;
+}
+
+static void cz_construct_max_power_limits_table(struct amdgpu_device *adev,
+				struct amdgpu_clock_and_voltage_limits *table)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *dep_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (dep_table->count > 0) {
+		table->sclk = dep_table->entries[dep_table->count - 1].clk;
+		table->vddc = cz_convert_8bit_index_to_voltage(adev,
+				dep_table->entries[dep_table->count - 1].v);
+	}
+
+	table->mclk = pi->sys_info.nbp_memory_clock[0];
+
+}
+
+union igp_info {
+	struct _ATOM_INTEGRATED_SYSTEM_INFO info;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 info_7;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_8 info_8;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_9 info_9;
+};
+
+static int cz_parse_sys_info_table(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo);
+	union igp_info *igp_info;
+	u8 frev, crev;
+	u16 data_offset;
+	int i = 0;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)(mode_info->atom_context->bios +
+					      data_offset);
+
+		if (crev != 9) {
+			DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev);
+			return -EINVAL;
+		}
+		pi->sys_info.bootup_sclk =
+			le32_to_cpu(igp_info->info_9.ulBootUpEngineClock);
+		pi->sys_info.bootup_uma_clk =
+			le32_to_cpu(igp_info->info_9.ulBootUpUMAClock);
+		pi->sys_info.dentist_vco_freq =
+			le32_to_cpu(igp_info->info_9.ulDentistVCOFreq);
+		pi->sys_info.bootup_nb_voltage_index =
+			le16_to_cpu(igp_info->info_9.usBootUpNBVoltage);
+
+		if (igp_info->info_9.ucHtcTmpLmt == 0)
+			pi->sys_info.htc_tmp_lmt = 203;
+		else
+			pi->sys_info.htc_tmp_lmt = igp_info->info_9.ucHtcTmpLmt;
+
+		if (igp_info->info_9.ucHtcHystLmt == 0)
+			pi->sys_info.htc_hyst_lmt = 5;
+		else
+			pi->sys_info.htc_hyst_lmt = igp_info->info_9.ucHtcHystLmt;
+
+		if (pi->sys_info.htc_tmp_lmt <= pi->sys_info.htc_hyst_lmt) {
+			DRM_ERROR("The htcTmpLmt should be larger than htcHystLmt.\n");
+			return -EINVAL;
+		}
+
+		if (le32_to_cpu(igp_info->info_9.ulSystemConfig) & (1 << 3) &&
+				pi->enable_nb_ps_policy)
+			pi->sys_info.nb_dpm_enable = true;
+		else
+			pi->sys_info.nb_dpm_enable = false;
+
+		for (i = 0; i < CZ_NUM_NBPSTATES; i++) {
+			if (i < CZ_NUM_NBPMEMORY_CLOCK)
+				pi->sys_info.nbp_memory_clock[i] =
+				le32_to_cpu(igp_info->info_9.ulNbpStateMemclkFreq[i]);
+			pi->sys_info.nbp_n_clock[i] =
+			le32_to_cpu(igp_info->info_9.ulNbpStateNClkFreq[i]);
+		}
+
+		for (i = 0; i < CZ_MAX_DISPLAY_CLOCK_LEVEL; i++)
+			pi->sys_info.display_clock[i] =
+			le32_to_cpu(igp_info->info_9.sDispClkVoltageMapping[i].ulMaximumSupportedCLK);
+
+		for (i = 0; i < CZ_NUM_NBPSTATES; i++)
+			pi->sys_info.nbp_voltage_index[i] =
+				le32_to_cpu(igp_info->info_9.usNBPStateVoltage[i]);
+
+		if (le32_to_cpu(igp_info->info_9.ulGPUCapInfo) &
+			SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS)
+			pi->caps_enable_dfs_bypass = true;
+
+		pi->sys_info.uma_channel_number =
+			igp_info->info_9.ucUMAChannelNumber;
+
+		cz_construct_max_power_limits_table(adev,
+			&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac);
+	}
+
+	return 0;
+}
+
+static void cz_patch_voltage_values(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_uvd_clock_voltage_dependency_table *uvd_table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table *vce_table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table *acp_table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+
+	if (uvd_table->count) {
+		for (i = 0; i < uvd_table->count; i++)
+			uvd_table->entries[i].v =
+				cz_convert_8bit_index_to_voltage(adev,
+						uvd_table->entries[i].v);
+	}
+
+	if (vce_table->count) {
+		for (i = 0; i < vce_table->count; i++)
+			vce_table->entries[i].v =
+				cz_convert_8bit_index_to_voltage(adev,
+						vce_table->entries[i].v);
+	}
+
+	if (acp_table->count) {
+		for (i = 0; i < acp_table->count; i++)
+			acp_table->entries[i].v =
+				cz_convert_8bit_index_to_voltage(adev,
+						acp_table->entries[i].v);
+	}
+
+}
+
+static void cz_construct_boot_state(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	pi->boot_pl.sclk = pi->sys_info.bootup_sclk;
+	pi->boot_pl.vddc_index = pi->sys_info.bootup_nb_voltage_index;
+	pi->boot_pl.ds_divider_index = 0;
+	pi->boot_pl.ss_divider_index = 0;
+	pi->boot_pl.allow_gnb_slow = 1;
+	pi->boot_pl.force_nbp_state = 0;
+	pi->boot_pl.display_wm = 0;
+	pi->boot_pl.vce_wm = 0;
+
+}
+
+static void cz_patch_boot_state(struct amdgpu_device *adev,
+				struct cz_ps *ps)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	ps->num_levels = 1;
+	ps->levels[0] = pi->boot_pl;
+}
+
+union pplib_clock_info {
+	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
+	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
+	struct _ATOM_PPLIB_CZ_CLOCK_INFO carrizo;
+};
+
+static void cz_parse_pplib_clock_info(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps, int index,
+					union pplib_clock_info *clock_info)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct cz_ps *ps = cz_get_ps(rps);
+	struct cz_pl *pl = &ps->levels[index];
+	struct amdgpu_clock_voltage_dependency_table *table =
+			&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	pl->sclk = table->entries[clock_info->carrizo.index].clk;
+	pl->vddc_index = table->entries[clock_info->carrizo.index].v;
+
+	ps->num_levels = index + 1;
+
+	if (pi->caps_sclk_ds) {
+		pl->ds_divider_index = 5;
+		pl->ss_divider_index = 5;
+	}
+
+}
+
+static void cz_parse_pplib_non_clock_info(struct amdgpu_device *adev,
+			struct amdgpu_ps *rps,
+			struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
+			u8 table_rev)
+{
+	struct cz_ps *ps = cz_get_ps(rps);
+
+	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
+	rps->class = le16_to_cpu(non_clock_info->usClassification);
+	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
+
+	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
+		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
+		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
+	} else {
+		rps->vclk = 0;
+		rps->dclk = 0;
+	}
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
+		adev->pm.dpm.boot_ps = rps;
+		cz_patch_boot_state(adev, ps);
+	}
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		adev->pm.dpm.uvd_ps = rps;
+
+}
+
+union power_info {
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE4 pplib4;
+	struct _ATOM_PPLIB_POWERPLAYTABLE5 pplib5;
+};
+
+union pplib_power_state {
+	struct _ATOM_PPLIB_STATE v1;
+	struct _ATOM_PPLIB_STATE_V2 v2;
+};
+
+static int cz_parse_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
+	union pplib_power_state *power_state;
+	int i, j, k, non_clock_array_index, clock_array_index;
+	union pplib_clock_info *clock_info;
+	struct _StateArray *state_array;
+	struct _ClockInfoArray *clock_info_array;
+	struct _NonClockInfoArray *non_clock_info_array;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	u8 *power_state_offset;
+	struct cz_ps *ps;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				    &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	state_array = (struct _StateArray *)
+		(mode_info->atom_context->bios + data_offset +
+		le16_to_cpu(power_info->pplib.usStateArrayOffset));
+	clock_info_array = (struct _ClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
+	non_clock_info_array = (struct _NonClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
+
+	adev->pm.dpm.ps = kzalloc(sizeof(struct amdgpu_ps) *
+					state_array->ucNumEntries, GFP_KERNEL);
+
+	if (!adev->pm.dpm.ps)
+		return -ENOMEM;
+
+	power_state_offset = (u8 *)state_array->states;
+	adev->pm.dpm.platform_caps =
+			le32_to_cpu(power_info->pplib.ulPlatformCaps);
+	adev->pm.dpm.backbias_response_time =
+			le16_to_cpu(power_info->pplib.usBackbiasTime);
+	adev->pm.dpm.voltage_response_time =
+			le16_to_cpu(power_info->pplib.usVoltageTime);
+
+	for (i = 0; i < state_array->ucNumEntries; i++) {
+		power_state = (union pplib_power_state *)power_state_offset;
+		non_clock_array_index = power_state->v2.nonClockInfoIndex;
+		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
+			&non_clock_info_array->nonClockInfo[non_clock_array_index];
+
+		ps = kzalloc(sizeof(struct cz_ps), GFP_KERNEL);
+		if (ps == NULL) {
+			kfree(adev->pm.dpm.ps);
+			return -ENOMEM;
+		}
+
+		adev->pm.dpm.ps[i].ps_priv = ps;
+		k = 0;
+		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
+			clock_array_index = power_state->v2.clockInfoIndex[j];
+			if (clock_array_index >= clock_info_array->ucNumEntries)
+				continue;
+			if (k >= CZ_MAX_HARDWARE_POWERLEVELS)
+				break;
+			clock_info = (union pplib_clock_info *)
+				&clock_info_array->clockInfo[clock_array_index *
+				clock_info_array->ucEntrySize];
+			cz_parse_pplib_clock_info(adev, &adev->pm.dpm.ps[i],
+				k, clock_info);
+			k++;
+		}
+		cz_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
+					non_clock_info,
+					non_clock_info_array->ucEntrySize);
+		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
+	}
+	adev->pm.dpm.num_ps = state_array->ucNumEntries;
+
+	return 0;
+}
+
+static int cz_process_firmware_header(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = cz_read_smc_sram_dword(adev, SMU8_FIRMWARE_HEADER_LOCATION +
+				     offsetof(struct SMU8_Firmware_Header,
+				     DpmTable),
+				     &tmp, pi->sram_end);
+
+	if (ret == 0)
+		pi->dpm_table_start = tmp;
+
+	return ret;
+}
+
+static int cz_dpm_init(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi;
+	int ret, i;
+
+	pi = kzalloc(sizeof(struct cz_power_info), GFP_KERNEL);
+	if (NULL == pi)
+		return -ENOMEM;
+
+	adev->pm.dpm.priv = pi;
+
+	ret = amdgpu_get_platform_caps(adev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_parse_extended_power_table(adev);
+	if (ret)
+		return ret;
+
+	pi->sram_end = SMC_RAM_END;
+
+	/* set up DPM defaults */
+	for (i = 0; i < CZ_MAX_HARDWARE_POWERLEVELS; i++)
+		pi->active_target[i] = CZ_AT_DFLT;
+
+	pi->mgcg_cgtt_local0 = 0x0;
+	pi->mgcg_cgtt_local1 = 0x0;
+	pi->clock_slow_down_step = 25000;
+	pi->skip_clock_slow_down = 1;
+	pi->enable_nb_ps_policy = 0;
+	pi->caps_power_containment = true;
+	pi->caps_cac = true;
+	pi->didt_enabled = false;
+	if (pi->didt_enabled) {
+		pi->caps_sq_ramping = true;
+		pi->caps_db_ramping = true;
+		pi->caps_td_ramping = true;
+		pi->caps_tcp_ramping = true;
+	}
+	pi->caps_sclk_ds = true;
+	pi->voting_clients = 0x00c00033;
+	pi->auto_thermal_throttling_enabled = true;
+	pi->bapm_enabled = false;
+	pi->disable_nb_ps3_in_battery = false;
+	pi->voltage_drop_threshold = 0;
+	pi->caps_sclk_throttle_low_notification = false;
+	pi->gfx_pg_threshold = 500;
+	pi->caps_fps = true;
+	/* uvd */
+	pi->caps_uvd_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_UVD) ? true : false;
+	pi->caps_uvd_dpm = true;
+	/* vce */
+	pi->caps_vce_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_VCE) ? true : false;
+	pi->caps_vce_dpm = true;
+	/* acp */
+	pi->caps_acp_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_ACP) ? true : false;
+	pi->caps_acp_dpm = true;
+
+	pi->caps_stable_power_state = false;
+	pi->nb_dpm_enabled_by_driver = true;
+	pi->nb_dpm_enabled = false;
+	pi->caps_voltage_island = false;
+	/* flags which indicate need to upload pptable */
+	pi->need_pptable_upload = true;
+
+	ret = cz_parse_sys_info_table(adev);
+	if (ret)
+		return ret;
+
+	cz_patch_voltage_values(adev);
+	cz_construct_boot_state(adev);
+
+	ret = cz_parse_power_table(adev);
+	if (ret)
+		return ret;
+
+	ret = cz_process_firmware_header(adev);
+	if (ret)
+		return ret;
+
+	pi->dpm_enabled = true;
+	pi->uvd_dynamic_pg = false;
+
+	return 0;
+}
+
+static void cz_dpm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++)
+		kfree(adev->pm.dpm.ps[i].ps_priv);
+
+	kfree(adev->pm.dpm.ps);
+	kfree(adev->pm.dpm.priv);
+	amdgpu_free_extended_power_table(adev);
+}
+
+#define ixSMUSVI_NB_CURRENTVID 0xD8230044
+#define CURRENT_NB_VID_MASK 0xff000000
+#define CURRENT_NB_VID__SHIFT 24
+#define ixSMUSVI_GFX_CURRENTVID  0xD8230048
+#define CURRENT_GFX_VID_MASK 0xff000000
+#define CURRENT_GFX_VID__SHIFT 24
+
+static void
+cz_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev,
+					       struct seq_file *m)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	struct amdgpu_uvd_clock_voltage_dependency_table *uvd_table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table *vce_table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	u32 sclk_index = REG_GET_FIELD(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX),
+				       TARGET_AND_CURRENT_PROFILE_INDEX, CURR_SCLK_INDEX);
+	u32 uvd_index = REG_GET_FIELD(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_2),
+				      TARGET_AND_CURRENT_PROFILE_INDEX_2, CURR_UVD_INDEX);
+	u32 vce_index = REG_GET_FIELD(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_2),
+				      TARGET_AND_CURRENT_PROFILE_INDEX_2, CURR_VCE_INDEX);
+	u32 sclk, vclk, dclk, ecclk, tmp;
+	u16 vddnb, vddgfx;
+
+	if (sclk_index >= NUM_SCLK_LEVELS) {
+		seq_printf(m, "invalid sclk dpm profile %d\n", sclk_index);
+	} else {
+		sclk = table->entries[sclk_index].clk;
+		seq_printf(m, "%u sclk: %u\n", sclk_index, sclk);
+	}
+
+	tmp = (RREG32_SMC(ixSMUSVI_NB_CURRENTVID) &
+	       CURRENT_NB_VID_MASK) >> CURRENT_NB_VID__SHIFT;
+	vddnb = cz_convert_8bit_index_to_voltage(adev, (u16)tmp);
+	tmp = (RREG32_SMC(ixSMUSVI_GFX_CURRENTVID) &
+	       CURRENT_GFX_VID_MASK) >> CURRENT_GFX_VID__SHIFT;
+	vddgfx = cz_convert_8bit_index_to_voltage(adev, (u16)tmp);
+	seq_printf(m, "vddnb: %u vddgfx: %u\n", vddnb, vddgfx);
+
+	seq_printf(m, "uvd    %sabled\n", pi->uvd_power_gated ? "dis" : "en");
+	if (!pi->uvd_power_gated) {
+		if (uvd_index >= CZ_MAX_HARDWARE_POWERLEVELS) {
+			seq_printf(m, "invalid uvd dpm level %d\n", uvd_index);
+		} else {
+			vclk = uvd_table->entries[uvd_index].vclk;
+			dclk = uvd_table->entries[uvd_index].dclk;
+			seq_printf(m, "%u uvd vclk: %u dclk: %u\n", uvd_index, vclk, dclk);
+		}
+	}
+
+	seq_printf(m, "vce    %sabled\n", pi->vce_power_gated ? "dis" : "en");
+	if (!pi->vce_power_gated) {
+		if (vce_index >= CZ_MAX_HARDWARE_POWERLEVELS) {
+			seq_printf(m, "invalid vce dpm level %d\n", vce_index);
+		} else {
+			ecclk = vce_table->entries[vce_index].ecclk;
+			seq_printf(m, "%u vce ecclk: %u\n", vce_index, ecclk);
+		}
+	}
+}
+
+static void cz_dpm_print_power_state(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps)
+{
+	int i;
+	struct cz_ps *ps = cz_get_ps(rps);
+
+	amdgpu_dpm_print_class_info(rps->class, rps->class2);
+	amdgpu_dpm_print_cap_info(rps->caps);
+
+	DRM_INFO("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+	for (i = 0; i < ps->num_levels; i++) {
+		struct cz_pl *pl = &ps->levels[i];
+
+		DRM_INFO("\t\tpower level %d    sclk: %u vddc: %u\n",
+		       i, pl->sclk,
+		       cz_convert_8bit_index_to_voltage(adev, pl->vddc_index));
+	}
+
+	amdgpu_dpm_print_ps_status(adev, rps);
+}
+
+static void cz_dpm_set_funcs(struct amdgpu_device *adev);
+
+static int cz_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cz_dpm_set_funcs(adev);
+
+	return 0;
+}
+
+
+static int cz_dpm_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (amdgpu_dpm) {
+		int ret;
+		/* init the sysfs and debugfs files late */
+		ret = amdgpu_pm_sysfs_init(adev);
+		if (ret)
+			return ret;
+
+		/* powerdown unused blocks for now */
+		cz_dpm_powergate_uvd(adev, true);
+		cz_dpm_powergate_vce(adev, true);
+	}
+
+	return 0;
+}
+
+static int cz_dpm_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+	/* fix me to add thermal support TODO */
+
+	/* default to balanced state */
+	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.forced_level = AMDGPU_DPM_FORCED_LEVEL_AUTO;
+	adev->pm.default_sclk = adev->clock.default_sclk;
+	adev->pm.default_mclk = adev->clock.default_mclk;
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
+
+	if (amdgpu_dpm == 0)
+		return 0;
+
+	mutex_lock(&adev->pm.mutex);
+	ret = cz_dpm_init(adev);
+	if (ret)
+		goto dpm_init_failed;
+
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+	if (amdgpu_dpm == 1)
+		amdgpu_pm_print_power_states(adev);
+
+	mutex_unlock(&adev->pm.mutex);
+	DRM_INFO("amdgpu: dpm initialized\n");
+
+	return 0;
+
+dpm_init_failed:
+	cz_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_ERROR("amdgpu: dpm initialization failed\n");
+
+	return ret;
+}
+
+static int cz_dpm_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+	amdgpu_pm_sysfs_fini(adev);
+	cz_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+
+	return 0;
+}
+
+static void cz_reset_ap_mask(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	pi->active_process_mask = 0;
+
+}
+
+static int cz_dpm_download_pptable_from_smu(struct amdgpu_device *adev,
+							void **table)
+{
+	int ret = 0;
+
+	ret = cz_smu_download_pptable(adev, table);
+
+	return ret;
+}
+
+static int cz_dpm_upload_pptable_to_smu(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct SMU8_Fusion_ClkTable *clock_table;
+	struct atom_clock_dividers dividers;
+	void *table = NULL;
+	uint8_t i = 0;
+	int ret = 0;
+
+	struct amdgpu_clock_voltage_dependency_table *vddc_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	struct amdgpu_clock_voltage_dependency_table *vddgfx_table =
+		&adev->pm.dpm.dyn_state.vddgfx_dependency_on_sclk;
+	struct amdgpu_uvd_clock_voltage_dependency_table *uvd_table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table *vce_table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table *acp_table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+
+	if (!pi->need_pptable_upload)
+		return 0;
+
+	ret = cz_dpm_download_pptable_from_smu(adev, &table);
+	if (ret) {
+		DRM_ERROR("amdgpu: Failed to get power play table from SMU!\n");
+		return -EINVAL;
+	}
+
+	clock_table = (struct SMU8_Fusion_ClkTable *)table;
+	/* patch clock table */
+	if (vddc_table->count > CZ_MAX_HARDWARE_POWERLEVELS ||
+			vddgfx_table->count > CZ_MAX_HARDWARE_POWERLEVELS ||
+			uvd_table->count > CZ_MAX_HARDWARE_POWERLEVELS ||
+			vce_table->count > CZ_MAX_HARDWARE_POWERLEVELS ||
+			acp_table->count > CZ_MAX_HARDWARE_POWERLEVELS) {
+		DRM_ERROR("amdgpu: Invalid Clock Voltage Dependency Table!\n");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < CZ_MAX_HARDWARE_POWERLEVELS; i++) {
+
+		/* vddc sclk */
+		clock_table->SclkBreakdownTable.ClkLevel[i].GnbVid =
+			(i < vddc_table->count) ? (uint8_t)vddc_table->entries[i].v : 0;
+		clock_table->SclkBreakdownTable.ClkLevel[i].Frequency =
+			(i < vddc_table->count) ? vddc_table->entries[i].clk : 0;
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+				clock_table->SclkBreakdownTable.ClkLevel[i].Frequency,
+				false, &dividers);
+		if (ret)
+			return ret;
+		clock_table->SclkBreakdownTable.ClkLevel[i].DfsDid =
+						(uint8_t)dividers.post_divider;
+
+		/* vddgfx sclk */
+		clock_table->SclkBreakdownTable.ClkLevel[i].GfxVid =
+			(i < vddgfx_table->count) ? (uint8_t)vddgfx_table->entries[i].v : 0;
+
+		/* acp breakdown */
+		clock_table->AclkBreakdownTable.ClkLevel[i].GfxVid =
+			(i < acp_table->count) ? (uint8_t)acp_table->entries[i].v : 0;
+		clock_table->AclkBreakdownTable.ClkLevel[i].Frequency =
+			(i < acp_table->count) ? acp_table->entries[i].clk : 0;
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+				clock_table->SclkBreakdownTable.ClkLevel[i].Frequency,
+				false, &dividers);
+		if (ret)
+			return ret;
+		clock_table->AclkBreakdownTable.ClkLevel[i].DfsDid =
+						(uint8_t)dividers.post_divider;
+
+		/* uvd breakdown */
+		clock_table->VclkBreakdownTable.ClkLevel[i].GfxVid =
+			(i < uvd_table->count) ? (uint8_t)uvd_table->entries[i].v : 0;
+		clock_table->VclkBreakdownTable.ClkLevel[i].Frequency =
+			(i < uvd_table->count) ? uvd_table->entries[i].vclk : 0;
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+				clock_table->VclkBreakdownTable.ClkLevel[i].Frequency,
+				false, &dividers);
+		if (ret)
+			return ret;
+		clock_table->VclkBreakdownTable.ClkLevel[i].DfsDid =
+						(uint8_t)dividers.post_divider;
+
+		clock_table->DclkBreakdownTable.ClkLevel[i].GfxVid =
+			(i < uvd_table->count) ? (uint8_t)uvd_table->entries[i].v : 0;
+		clock_table->DclkBreakdownTable.ClkLevel[i].Frequency =
+			(i < uvd_table->count) ? uvd_table->entries[i].dclk : 0;
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+				clock_table->DclkBreakdownTable.ClkLevel[i].Frequency,
+				false, &dividers);
+		if (ret)
+			return ret;
+		clock_table->DclkBreakdownTable.ClkLevel[i].DfsDid =
+						(uint8_t)dividers.post_divider;
+
+		/* vce breakdown */
+		clock_table->EclkBreakdownTable.ClkLevel[i].GfxVid =
+			(i < vce_table->count) ? (uint8_t)vce_table->entries[i].v : 0;
+		clock_table->EclkBreakdownTable.ClkLevel[i].Frequency =
+			(i < vce_table->count) ? vce_table->entries[i].ecclk : 0;
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+				clock_table->EclkBreakdownTable.ClkLevel[i].Frequency,
+				false, &dividers);
+		if (ret)
+			return ret;
+		clock_table->EclkBreakdownTable.ClkLevel[i].DfsDid =
+						(uint8_t)dividers.post_divider;
+	}
+
+	/* its time to upload to SMU */
+	ret = cz_smu_upload_pptable(adev);
+	if (ret) {
+		DRM_ERROR("amdgpu: Failed to put power play table to SMU!\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static void cz_init_sclk_limit(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	uint32_t clock = 0, level;
+
+	if (!table || !table->count) {
+		DRM_ERROR("Invalid Voltage Dependency table.\n");
+		return;
+	}
+
+	pi->sclk_dpm.soft_min_clk = 0;
+	pi->sclk_dpm.hard_min_clk = 0;
+	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxSclkLevel);
+	level = cz_get_argument(adev);
+	if (level < table->count)
+		clock = table->entries[level].clk;
+	else {
+		DRM_ERROR("Invalid SLCK Voltage Dependency table entry.\n");
+		clock = table->entries[table->count - 1].clk;
+	}
+
+	pi->sclk_dpm.soft_max_clk = clock;
+	pi->sclk_dpm.hard_max_clk = clock;
+
+}
+
+static void cz_init_uvd_limit(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_uvd_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	uint32_t clock = 0, level;
+
+	if (!table || !table->count) {
+		DRM_ERROR("Invalid Voltage Dependency table.\n");
+		return;
+	}
+
+	pi->uvd_dpm.soft_min_clk = 0;
+	pi->uvd_dpm.hard_min_clk = 0;
+	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxUvdLevel);
+	level = cz_get_argument(adev);
+	if (level < table->count)
+		clock = table->entries[level].vclk;
+	else {
+		DRM_ERROR("Invalid UVD Voltage Dependency table entry.\n");
+		clock = table->entries[table->count - 1].vclk;
+	}
+
+	pi->uvd_dpm.soft_max_clk = clock;
+	pi->uvd_dpm.hard_max_clk = clock;
+
+}
+
+static void cz_init_vce_limit(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	uint32_t clock = 0, level;
+
+	if (!table || !table->count) {
+		DRM_ERROR("Invalid Voltage Dependency table.\n");
+		return;
+	}
+
+	pi->vce_dpm.soft_min_clk = table->entries[0].ecclk;
+	pi->vce_dpm.hard_min_clk = table->entries[0].ecclk;
+	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxEclkLevel);
+	level = cz_get_argument(adev);
+	if (level < table->count)
+		clock = table->entries[level].ecclk;
+	else {
+		/* future BIOS would fix this error */
+		DRM_ERROR("Invalid VCE Voltage Dependency table entry.\n");
+		clock = table->entries[table->count - 1].ecclk;
+	}
+
+	pi->vce_dpm.soft_max_clk = clock;
+	pi->vce_dpm.hard_max_clk = clock;
+
+}
+
+static void cz_init_acp_limit(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+	uint32_t clock = 0, level;
+
+	if (!table || !table->count) {
+		DRM_ERROR("Invalid Voltage Dependency table.\n");
+		return;
+	}
+
+	pi->acp_dpm.soft_min_clk = 0;
+	pi->acp_dpm.hard_min_clk = 0;
+	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxAclkLevel);
+	level = cz_get_argument(adev);
+	if (level < table->count)
+		clock = table->entries[level].clk;
+	else {
+		DRM_ERROR("Invalid ACP Voltage Dependency table entry.\n");
+		clock = table->entries[table->count - 1].clk;
+	}
+
+	pi->acp_dpm.soft_max_clk = clock;
+	pi->acp_dpm.hard_max_clk = clock;
+
+}
+
+static void cz_init_pg_state(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	pi->uvd_power_gated = false;
+	pi->vce_power_gated = false;
+	pi->acp_power_gated = false;
+
+}
+
+static void cz_init_sclk_threshold(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	pi->low_sclk_interrupt_threshold = 0;
+
+}
+
+static void cz_dpm_setup_asic(struct amdgpu_device *adev)
+{
+	cz_reset_ap_mask(adev);
+	cz_dpm_upload_pptable_to_smu(adev);
+	cz_init_sclk_limit(adev);
+	cz_init_uvd_limit(adev);
+	cz_init_vce_limit(adev);
+	cz_init_acp_limit(adev);
+	cz_init_pg_state(adev);
+	cz_init_sclk_threshold(adev);
+
+}
+
+static bool cz_check_smu_feature(struct amdgpu_device *adev,
+				uint32_t feature)
+{
+	uint32_t smu_feature = 0;
+	int ret;
+
+	ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_GetFeatureStatus, 0);
+	if (ret) {
+		DRM_ERROR("Failed to get SMU features from SMC.\n");
+		return false;
+	} else {
+		smu_feature = cz_get_argument(adev);
+		if (feature & smu_feature)
+			return true;
+	}
+
+	return false;
+}
+
+static bool cz_check_for_dpm_enabled(struct amdgpu_device *adev)
+{
+	if (cz_check_smu_feature(adev,
+				SMU_EnabledFeatureScoreboard_SclkDpmOn))
+		return true;
+
+	return false;
+}
+
+static void cz_program_voting_clients(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, PPCZ_VOTINGRIGHTSCLIENTS_DFLT0);
+}
+
+static void cz_clear_voting_clients(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0);
+}
+
+static int cz_start_dpm(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	if (amdgpu_dpm) {
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_EnableAllSmuFeatures, SCLK_DPM_MASK);
+		if (ret) {
+			DRM_ERROR("SMU feature: SCLK_DPM enable failed\n");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static int cz_stop_dpm(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	if (amdgpu_dpm && adev->pm.dpm_enabled) {
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_DisableAllSmuFeatures, SCLK_DPM_MASK);
+		if (ret) {
+			DRM_ERROR("SMU feature: SCLK_DPM disable failed\n");
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+static uint32_t cz_get_sclk_level(struct amdgpu_device *adev,
+				uint32_t clock, uint16_t msg)
+{
+	int i = 0;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	switch (msg) {
+	case PPSMC_MSG_SetSclkSoftMin:
+	case PPSMC_MSG_SetSclkHardMin:
+		for (i = 0; i < table->count; i++)
+			if (clock <= table->entries[i].clk)
+				break;
+		if (i == table->count)
+			i = table->count - 1;
+		break;
+	case PPSMC_MSG_SetSclkSoftMax:
+	case PPSMC_MSG_SetSclkHardMax:
+		for (i = table->count - 1; i >= 0; i--)
+			if (clock >= table->entries[i].clk)
+				break;
+		if (i < 0)
+			i = 0;
+		break;
+	default:
+		break;
+	}
+
+	return i;
+}
+
+static uint32_t cz_get_eclk_level(struct amdgpu_device *adev,
+				uint32_t clock, uint16_t msg)
+{
+	int i = 0;
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	if (table->count == 0)
+		return 0;
+
+	switch (msg) {
+	case PPSMC_MSG_SetEclkSoftMin:
+	case PPSMC_MSG_SetEclkHardMin:
+		for (i = 0; i < table->count-1; i++)
+			if (clock <= table->entries[i].ecclk)
+				break;
+		break;
+	case PPSMC_MSG_SetEclkSoftMax:
+	case PPSMC_MSG_SetEclkHardMax:
+		for (i = table->count - 1; i > 0; i--)
+			if (clock >= table->entries[i].ecclk)
+				break;
+		break;
+	default:
+		break;
+	}
+
+	return i;
+}
+
+static int cz_program_bootup_state(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	uint32_t soft_min_clk = 0;
+	uint32_t soft_max_clk = 0;
+	int ret = 0;
+
+	pi->sclk_dpm.soft_min_clk = pi->sys_info.bootup_sclk;
+	pi->sclk_dpm.soft_max_clk = pi->sys_info.bootup_sclk;
+
+	soft_min_clk = cz_get_sclk_level(adev,
+				pi->sclk_dpm.soft_min_clk,
+				PPSMC_MSG_SetSclkSoftMin);
+	soft_max_clk = cz_get_sclk_level(adev,
+				pi->sclk_dpm.soft_max_clk,
+				PPSMC_MSG_SetSclkSoftMax);
+
+	ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMin, soft_min_clk);
+	if (ret)
+		return -EINVAL;
+
+	ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMax, soft_max_clk);
+	if (ret)
+		return -EINVAL;
+
+	return 0;
+}
+
+/* TODO */
+static int cz_disable_cgpg(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+/* TODO */
+static int cz_enable_cgpg(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+/* TODO */
+static int cz_program_pt_config_registers(struct amdgpu_device *adev)
+{
+	return 0;
+}
+
+static void cz_do_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	uint32_t reg = 0;
+
+	if (pi->caps_sq_ramping) {
+		reg = RREG32_DIDT(ixDIDT_SQ_CTRL0);
+		if (enable)
+			reg = REG_SET_FIELD(reg, DIDT_SQ_CTRL0, DIDT_CTRL_EN, 1);
+		else
+			reg = REG_SET_FIELD(reg, DIDT_SQ_CTRL0, DIDT_CTRL_EN, 0);
+		WREG32_DIDT(ixDIDT_SQ_CTRL0, reg);
+	}
+	if (pi->caps_db_ramping) {
+		reg = RREG32_DIDT(ixDIDT_DB_CTRL0);
+		if (enable)
+			reg = REG_SET_FIELD(reg, DIDT_DB_CTRL0, DIDT_CTRL_EN, 1);
+		else
+			reg = REG_SET_FIELD(reg, DIDT_DB_CTRL0, DIDT_CTRL_EN, 0);
+		WREG32_DIDT(ixDIDT_DB_CTRL0, reg);
+	}
+	if (pi->caps_td_ramping) {
+		reg = RREG32_DIDT(ixDIDT_TD_CTRL0);
+		if (enable)
+			reg = REG_SET_FIELD(reg, DIDT_TD_CTRL0, DIDT_CTRL_EN, 1);
+		else
+			reg = REG_SET_FIELD(reg, DIDT_TD_CTRL0, DIDT_CTRL_EN, 0);
+		WREG32_DIDT(ixDIDT_TD_CTRL0, reg);
+	}
+	if (pi->caps_tcp_ramping) {
+		reg = RREG32_DIDT(ixDIDT_TCP_CTRL0);
+		if (enable)
+			reg = REG_SET_FIELD(reg, DIDT_SQ_CTRL0, DIDT_CTRL_EN, 1);
+		else
+			reg = REG_SET_FIELD(reg, DIDT_SQ_CTRL0, DIDT_CTRL_EN, 0);
+		WREG32_DIDT(ixDIDT_TCP_CTRL0, reg);
+	}
+
+}
+
+static int cz_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	int ret;
+
+	if (pi->caps_sq_ramping || pi->caps_db_ramping ||
+			pi->caps_td_ramping || pi->caps_tcp_ramping) {
+		if (adev->gfx.gfx_current_status != AMDGPU_GFX_SAFE_MODE) {
+			ret = cz_disable_cgpg(adev);
+			if (ret) {
+				DRM_ERROR("Pre Di/Dt disable cg/pg failed\n");
+				return -EINVAL;
+			}
+			adev->gfx.gfx_current_status = AMDGPU_GFX_SAFE_MODE;
+		}
+
+		ret = cz_program_pt_config_registers(adev);
+		if (ret) {
+			DRM_ERROR("Di/Dt config failed\n");
+			return -EINVAL;
+		}
+		cz_do_enable_didt(adev, enable);
+
+		if (adev->gfx.gfx_current_status == AMDGPU_GFX_SAFE_MODE) {
+			ret = cz_enable_cgpg(adev);
+			if (ret) {
+				DRM_ERROR("Post Di/Dt enable cg/pg failed\n");
+				return -EINVAL;
+			}
+			adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+		}
+	}
+
+	return 0;
+}
+
+/* TODO */
+static void cz_reset_acp_boot_level(struct amdgpu_device *adev)
+{
+}
+
+static void cz_update_current_ps(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct cz_ps *ps = cz_get_ps(rps);
+
+	pi->current_ps = *ps;
+	pi->current_rps = *rps;
+	pi->current_rps.ps_priv = ps;
+
+}
+
+static void cz_update_requested_ps(struct amdgpu_device *adev,
+					struct amdgpu_ps *rps)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct cz_ps *ps = cz_get_ps(rps);
+
+	pi->requested_ps = *ps;
+	pi->requested_rps = *rps;
+	pi->requested_rps.ps_priv = ps;
+
+}
+
+/* PP arbiter support needed TODO */
+static void cz_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *new_rps,
+					struct amdgpu_ps *old_rps)
+{
+	struct cz_ps *ps = cz_get_ps(new_rps);
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_and_voltage_limits *limits =
+		&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	/* 10kHz memory clock */
+	uint32_t mclk = 0;
+
+	ps->force_high = false;
+	ps->need_dfs_bypass = true;
+	pi->video_start = new_rps->dclk || new_rps->vclk ||
+				new_rps->evclk || new_rps->ecclk;
+
+	if ((new_rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) ==
+			ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
+		pi->battery_state = true;
+	else
+		pi->battery_state = false;
+
+	if (pi->caps_stable_power_state)
+		mclk = limits->mclk;
+
+	if (mclk > pi->sys_info.nbp_memory_clock[CZ_NUM_NBPMEMORY_CLOCK - 1])
+		ps->force_high = true;
+
+}
+
+static int cz_dpm_enable(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	int ret = 0;
+
+	/* renable will hang up SMU, so check first */
+	if (cz_check_for_dpm_enabled(adev))
+		return -EINVAL;
+
+	cz_program_voting_clients(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+		chip_name = "carrizo";
+		break;
+	case CHIP_STONEY:
+		chip_name = "stoney";
+		break;
+	default:
+		BUG();
+	}
+
+
+	ret = cz_start_dpm(adev);
+	if (ret) {
+		DRM_ERROR("%s DPM enable failed\n", chip_name);
+		return -EINVAL;
+	}
+
+	ret = cz_program_bootup_state(adev);
+	if (ret) {
+		DRM_ERROR("%s bootup state program failed\n", chip_name);
+		return -EINVAL;
+	}
+
+	ret = cz_enable_didt(adev, true);
+	if (ret) {
+		DRM_ERROR("%s enable di/dt failed\n", chip_name);
+		return -EINVAL;
+	}
+
+	cz_reset_acp_boot_level(adev);
+
+	cz_update_current_ps(adev, adev->pm.dpm.boot_ps);
+
+	return 0;
+}
+
+static int cz_dpm_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret = 0;
+
+	mutex_lock(&adev->pm.mutex);
+
+	/* smu init only needs to be called at startup, not resume.
+	 * It should be in sw_init, but requires the fw info gathered
+	 * in sw_init from other IP modules.
+	 */
+	ret = cz_smu_init(adev);
+	if (ret) {
+		DRM_ERROR("amdgpu: smc initialization failed\n");
+		mutex_unlock(&adev->pm.mutex);
+		return ret;
+	}
+
+	/* do the actual fw loading */
+	ret = cz_smu_start(adev);
+	if (ret) {
+		DRM_ERROR("amdgpu: smc start failed\n");
+		mutex_unlock(&adev->pm.mutex);
+		return ret;
+	}
+
+	if (!amdgpu_dpm) {
+		adev->pm.dpm_enabled = false;
+		mutex_unlock(&adev->pm.mutex);
+		return ret;
+	}
+
+	/* cz dpm setup asic */
+	cz_dpm_setup_asic(adev);
+
+	/* cz dpm enable */
+	ret = cz_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+
+	mutex_unlock(&adev->pm.mutex);
+
+	return 0;
+}
+
+static int cz_dpm_disable(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	if (!cz_check_for_dpm_enabled(adev))
+		return -EINVAL;
+
+	ret = cz_enable_didt(adev, false);
+	if (ret) {
+		DRM_ERROR("disable di/dt failed\n");
+		return -EINVAL;
+	}
+
+	/* powerup blocks */
+	cz_dpm_powergate_uvd(adev, false);
+	cz_dpm_powergate_vce(adev, false);
+
+	cz_clear_voting_clients(adev);
+	cz_stop_dpm(adev);
+	cz_update_current_ps(adev, adev->pm.dpm.boot_ps);
+
+	return 0;
+}
+
+static int cz_dpm_hw_fini(void *handle)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+
+	/* smu fini only needs to be called at teardown, not suspend.
+	 * It should be in sw_fini, but we put it here for symmetry
+	 * with smu init.
+	 */
+	cz_smu_fini(adev);
+
+	if (adev->pm.dpm_enabled) {
+		ret = cz_dpm_disable(adev);
+
+		adev->pm.dpm.current_ps =
+			adev->pm.dpm.requested_ps =
+			adev->pm.dpm.boot_ps;
+	}
+
+	adev->pm.dpm_enabled = false;
+
+	mutex_unlock(&adev->pm.mutex);
+
+	return ret;
+}
+
+static int cz_dpm_suspend(void *handle)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+
+		ret = cz_dpm_disable(adev);
+
+		adev->pm.dpm.current_ps =
+			adev->pm.dpm.requested_ps =
+			adev->pm.dpm.boot_ps;
+
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+	return ret;
+}
+
+static int cz_dpm_resume(void *handle)
+{
+	int ret = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+
+	/* do the actual fw loading */
+	ret = cz_smu_start(adev);
+	if (ret) {
+		DRM_ERROR("amdgpu: smc start failed\n");
+		mutex_unlock(&adev->pm.mutex);
+		return ret;
+	}
+
+	if (!amdgpu_dpm) {
+		adev->pm.dpm_enabled = false;
+		mutex_unlock(&adev->pm.mutex);
+		return ret;
+	}
+
+	/* cz dpm setup asic */
+	cz_dpm_setup_asic(adev);
+
+	/* cz dpm enable */
+	ret = cz_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+
+	mutex_unlock(&adev->pm.mutex);
+	/* upon resume, re-compute the clocks */
+	if (adev->pm.dpm_enabled)
+		amdgpu_pm_compute_clocks(adev);
+
+	return 0;
+}
+
+static int cz_dpm_set_clockgating_state(void *handle,
+					enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int cz_dpm_set_powergating_state(void *handle,
+					enum amd_powergating_state state)
+{
+	return 0;
+}
+
+/* borrowed from KV, need future unify */
+static int cz_dpm_get_temperature(struct amdgpu_device *adev)
+{
+	int actual_temp = 0;
+	uint32_t temp = RREG32_SMC(0xC0300E0C);
+
+	if (temp)
+		actual_temp = 1000 * ((temp / 8) - 49);
+
+	return actual_temp;
+}
+
+static int cz_dpm_pre_set_power_state(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
+	struct amdgpu_ps *new_ps = &requested_ps;
+
+	cz_update_requested_ps(adev, new_ps);
+	cz_apply_state_adjust_rules(adev, &pi->requested_rps,
+					&pi->current_rps);
+
+	return 0;
+}
+
+static int cz_dpm_update_sclk_limit(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_and_voltage_limits *limits =
+		&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	uint32_t clock, stable_ps_clock = 0;
+
+	clock = pi->sclk_dpm.soft_min_clk;
+
+	if (pi->caps_stable_power_state) {
+		stable_ps_clock = limits->sclk * 75 / 100;
+		if (clock < stable_ps_clock)
+			clock = stable_ps_clock;
+	}
+
+	if (clock != pi->sclk_dpm.soft_min_clk) {
+		pi->sclk_dpm.soft_min_clk = clock;
+		cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMin,
+				cz_get_sclk_level(adev, clock,
+					PPSMC_MSG_SetSclkSoftMin));
+	}
+
+	if (pi->caps_stable_power_state &&
+			pi->sclk_dpm.soft_max_clk != clock) {
+		pi->sclk_dpm.soft_max_clk = clock;
+		cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMax,
+				cz_get_sclk_level(adev, clock,
+					PPSMC_MSG_SetSclkSoftMax));
+	} else {
+		cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMax,
+				cz_get_sclk_level(adev,
+					pi->sclk_dpm.soft_max_clk,
+					PPSMC_MSG_SetSclkSoftMax));
+	}
+
+	return 0;
+}
+
+static int cz_dpm_set_deep_sleep_sclk_threshold(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	if (pi->caps_sclk_ds) {
+		cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetMinDeepSleepSclk,
+				CZ_MIN_DEEP_SLEEP_SCLK);
+	}
+
+	return ret;
+}
+
+/* ?? without dal support, is this still needed in setpowerstate list*/
+static int cz_dpm_set_watermark_threshold(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	cz_send_msg_to_smc_with_parameter(adev,
+			PPSMC_MSG_SetWatermarkFrequency,
+			pi->sclk_dpm.soft_max_clk);
+
+	return ret;
+}
+
+static int cz_dpm_enable_nbdpm(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	/* also depend on dal NBPStateDisableRequired */
+	if (pi->nb_dpm_enabled_by_driver && !pi->nb_dpm_enabled) {
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_EnableAllSmuFeatures,
+				NB_DPM_MASK);
+		if (ret) {
+			DRM_ERROR("amdgpu: nb dpm enable failed\n");
+			return ret;
+		}
+		pi->nb_dpm_enabled = true;
+	}
+
+	return ret;
+}
+
+static void cz_dpm_nbdpm_lm_pstate_enable(struct amdgpu_device *adev,
+							bool enable)
+{
+	if (enable)
+		cz_send_msg_to_smc(adev, PPSMC_MSG_EnableLowMemoryPstate);
+	else
+		cz_send_msg_to_smc(adev, PPSMC_MSG_DisableLowMemoryPstate);
+
+}
+
+static int cz_dpm_update_low_memory_pstate(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct cz_ps *ps = &pi->requested_ps;
+
+	if (pi->sys_info.nb_dpm_enable) {
+		if (ps->force_high)
+			cz_dpm_nbdpm_lm_pstate_enable(adev, false);
+		else
+			cz_dpm_nbdpm_lm_pstate_enable(adev, true);
+	}
+
+	return ret;
+}
+
+/* with dpm enabled */
+static int cz_dpm_set_power_state(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	cz_dpm_update_sclk_limit(adev);
+	cz_dpm_set_deep_sleep_sclk_threshold(adev);
+	cz_dpm_set_watermark_threshold(adev);
+	cz_dpm_enable_nbdpm(adev);
+	cz_dpm_update_low_memory_pstate(adev);
+
+	return ret;
+}
+
+static void cz_dpm_post_set_power_state(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_ps *ps = &pi->requested_rps;
+
+	cz_update_current_ps(adev, ps);
+
+}
+
+static int cz_dpm_force_highest(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	int ret = 0;
+
+	if (pi->sclk_dpm.soft_min_clk != pi->sclk_dpm.soft_max_clk) {
+		pi->sclk_dpm.soft_min_clk =
+			pi->sclk_dpm.soft_max_clk;
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMin,
+				cz_get_sclk_level(adev,
+					pi->sclk_dpm.soft_min_clk,
+					PPSMC_MSG_SetSclkSoftMin));
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+static int cz_dpm_force_lowest(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	int ret = 0;
+
+	if (pi->sclk_dpm.soft_max_clk != pi->sclk_dpm.soft_min_clk) {
+		pi->sclk_dpm.soft_max_clk = pi->sclk_dpm.soft_min_clk;
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMax,
+				cz_get_sclk_level(adev,
+					pi->sclk_dpm.soft_max_clk,
+					PPSMC_MSG_SetSclkSoftMax));
+		if (ret)
+			return ret;
+	}
+
+	return ret;
+}
+
+static uint32_t cz_dpm_get_max_sclk_level(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	if (!pi->max_sclk_level) {
+		cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxSclkLevel);
+		pi->max_sclk_level = cz_get_argument(adev) + 1;
+	}
+
+	if (pi->max_sclk_level > CZ_MAX_HARDWARE_POWERLEVELS) {
+		DRM_ERROR("Invalid max sclk level!\n");
+		return -EINVAL;
+	}
+
+	return pi->max_sclk_level;
+}
+
+static int cz_dpm_unforce_dpm_levels(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *dep_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	uint32_t level = 0;
+	int ret = 0;
+
+	pi->sclk_dpm.soft_min_clk = dep_table->entries[0].clk;
+	level = cz_dpm_get_max_sclk_level(adev) - 1;
+	if (level < dep_table->count)
+		pi->sclk_dpm.soft_max_clk = dep_table->entries[level].clk;
+	else
+		pi->sclk_dpm.soft_max_clk =
+			dep_table->entries[dep_table->count - 1].clk;
+
+	/* get min/max sclk soft value
+	 * notify SMU to execute */
+	ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMin,
+				cz_get_sclk_level(adev,
+					pi->sclk_dpm.soft_min_clk,
+					PPSMC_MSG_SetSclkSoftMin));
+	if (ret)
+		return ret;
+
+	ret = cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetSclkSoftMax,
+				cz_get_sclk_level(adev,
+					pi->sclk_dpm.soft_max_clk,
+					PPSMC_MSG_SetSclkSoftMax));
+	if (ret)
+		return ret;
+
+	DRM_DEBUG("DPM unforce state min=%d, max=%d.\n",
+		  pi->sclk_dpm.soft_min_clk,
+		  pi->sclk_dpm.soft_max_clk);
+
+	return 0;
+}
+
+static int cz_dpm_force_dpm_level(struct amdgpu_device *adev,
+				  enum amdgpu_dpm_forced_level level)
+{
+	int ret = 0;
+
+	switch (level) {
+	case AMDGPU_DPM_FORCED_LEVEL_HIGH:
+		ret = cz_dpm_unforce_dpm_levels(adev);
+		if (ret)
+			return ret;
+		ret = cz_dpm_force_highest(adev);
+		if (ret)
+			return ret;
+		break;
+	case AMDGPU_DPM_FORCED_LEVEL_LOW:
+		ret = cz_dpm_unforce_dpm_levels(adev);
+		if (ret)
+			return ret;
+		ret = cz_dpm_force_lowest(adev);
+		if (ret)
+			return ret;
+		break;
+	case AMDGPU_DPM_FORCED_LEVEL_AUTO:
+		ret = cz_dpm_unforce_dpm_levels(adev);
+		if (ret)
+			return ret;
+		break;
+	default:
+		break;
+	}
+
+	adev->pm.dpm.forced_level = level;
+
+	return ret;
+}
+
+/* fix me, display configuration change lists here
+ * mostly dal related*/
+static void cz_dpm_display_configuration_changed(struct amdgpu_device *adev)
+{
+}
+
+static uint32_t cz_dpm_get_sclk(struct amdgpu_device *adev, bool low)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct cz_ps *requested_state = cz_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->levels[0].sclk;
+	else
+		return requested_state->levels[requested_state->num_levels - 1].sclk;
+
+}
+
+static uint32_t cz_dpm_get_mclk(struct amdgpu_device *adev, bool low)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	return pi->sys_info.bootup_uma_clk;
+}
+
+static int cz_enable_uvd_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	int ret = 0;
+
+	if (enable && pi->caps_uvd_dpm ) {
+		pi->dpm_flags |= DPMFlags_UVD_Enabled;
+		DRM_DEBUG("UVD DPM Enabled.\n");
+
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+			PPSMC_MSG_EnableAllSmuFeatures, UVD_DPM_MASK);
+	} else {
+		pi->dpm_flags &= ~DPMFlags_UVD_Enabled;
+		DRM_DEBUG("UVD DPM Stopped\n");
+
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+			PPSMC_MSG_DisableAllSmuFeatures, UVD_DPM_MASK);
+	}
+
+	return ret;
+}
+
+static int cz_update_uvd_dpm(struct amdgpu_device *adev, bool gate)
+{
+	return cz_enable_uvd_dpm(adev, !gate);
+}
+
+
+static void cz_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	int ret;
+
+	if (pi->uvd_power_gated == gate)
+		return;
+
+	pi->uvd_power_gated = gate;
+
+	if (gate) {
+		if (pi->caps_uvd_pg) {
+			/* disable clockgating so we can properly shut down the block */
+			ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_CG_STATE_UNGATE);
+			/* shutdown the UVD block */
+			ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_PG_STATE_GATE);
+			/* XXX: check for errors */
+		}
+		cz_update_uvd_dpm(adev, gate);
+		if (pi->caps_uvd_pg)
+			/* power off the UVD block */
+			cz_send_msg_to_smc(adev, PPSMC_MSG_UVDPowerOFF);
+	} else {
+		if (pi->caps_uvd_pg) {
+			/* power on the UVD block */
+			if (pi->uvd_dynamic_pg)
+				cz_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_UVDPowerON, 1);
+			else
+				cz_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_UVDPowerON, 0);
+			/* re-init the UVD block */
+			ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_PG_STATE_UNGATE);
+			/* enable clockgating. hw will dynamically gate/ungate clocks on the fly */
+			ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_CG_STATE_GATE);
+			/* XXX: check for errors */
+		}
+		cz_update_uvd_dpm(adev, gate);
+	}
+}
+
+static int cz_enable_vce_dpm(struct amdgpu_device *adev, bool enable)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	int ret = 0;
+
+	if (enable && pi->caps_vce_dpm) {
+		pi->dpm_flags |= DPMFlags_VCE_Enabled;
+		DRM_DEBUG("VCE DPM Enabled.\n");
+
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+			PPSMC_MSG_EnableAllSmuFeatures, VCE_DPM_MASK);
+
+	} else {
+		pi->dpm_flags &= ~DPMFlags_VCE_Enabled;
+		DRM_DEBUG("VCE DPM Stopped\n");
+
+		ret = cz_send_msg_to_smc_with_parameter(adev,
+			PPSMC_MSG_DisableAllSmuFeatures, VCE_DPM_MASK);
+	}
+
+	return ret;
+}
+
+static int cz_update_vce_dpm(struct amdgpu_device *adev)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	/* Stable Pstate is enabled and we need to set the VCE DPM to highest level */
+	if (pi->caps_stable_power_state) {
+		pi->vce_dpm.hard_min_clk = table->entries[table->count-1].ecclk;
+
+	} else { /* non-stable p-state cases. without vce.Arbiter.EcclkHardMin */
+		pi->vce_dpm.hard_min_clk = table->entries[0].ecclk;
+	}
+
+	cz_send_msg_to_smc_with_parameter(adev,
+		PPSMC_MSG_SetEclkHardMin,
+		cz_get_eclk_level(adev,
+			pi->vce_dpm.hard_min_clk,
+			PPSMC_MSG_SetEclkHardMin));
+	return 0;
+}
+
+static void cz_dpm_powergate_vce(struct amdgpu_device *adev, bool gate)
+{
+	struct cz_power_info *pi = cz_get_pi(adev);
+
+	if (pi->caps_vce_pg) {
+		if (pi->vce_power_gated != gate) {
+			if (gate) {
+				/* disable clockgating so we can properly shut down the block */
+				amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_CG_STATE_UNGATE);
+				/* shutdown the VCE block */
+				amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_PG_STATE_GATE);
+
+				cz_enable_vce_dpm(adev, false);
+				/* TODO: to figure out why vce can't be poweroff. */
+				/* cz_send_msg_to_smc(adev, PPSMC_MSG_VCEPowerOFF); */
+				pi->vce_power_gated = true;
+			} else {
+				cz_send_msg_to_smc(adev, PPSMC_MSG_VCEPowerON);
+				pi->vce_power_gated = false;
+
+				/* re-init the VCE block */
+				amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_PG_STATE_UNGATE);
+				/* enable clockgating. hw will dynamically gate/ungate clocks on the fly */
+				amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_CG_STATE_GATE);
+
+				cz_update_vce_dpm(adev);
+				cz_enable_vce_dpm(adev, true);
+			}
+		} else {
+			if (! pi->vce_power_gated) {
+				cz_update_vce_dpm(adev);
+			}
+		}
+	} else { /*pi->caps_vce_pg*/
+		cz_update_vce_dpm(adev);
+		cz_enable_vce_dpm(adev, true);
+	}
+
+	return;
+}
+
+const struct amd_ip_funcs cz_dpm_ip_funcs = {
+	.early_init = cz_dpm_early_init,
+	.late_init = cz_dpm_late_init,
+	.sw_init = cz_dpm_sw_init,
+	.sw_fini = cz_dpm_sw_fini,
+	.hw_init = cz_dpm_hw_init,
+	.hw_fini = cz_dpm_hw_fini,
+	.suspend = cz_dpm_suspend,
+	.resume = cz_dpm_resume,
+	.is_idle = NULL,
+	.wait_for_idle = NULL,
+	.soft_reset = NULL,
+	.print_status = NULL,
+	.set_clockgating_state = cz_dpm_set_clockgating_state,
+	.set_powergating_state = cz_dpm_set_powergating_state,
+};
+
+static const struct amdgpu_dpm_funcs cz_dpm_funcs = {
+	.get_temperature = cz_dpm_get_temperature,
+	.pre_set_power_state = cz_dpm_pre_set_power_state,
+	.set_power_state = cz_dpm_set_power_state,
+	.post_set_power_state = cz_dpm_post_set_power_state,
+	.display_configuration_changed = cz_dpm_display_configuration_changed,
+	.get_sclk = cz_dpm_get_sclk,
+	.get_mclk = cz_dpm_get_mclk,
+	.print_power_state = cz_dpm_print_power_state,
+	.debugfs_print_current_performance_level =
+				cz_dpm_debugfs_print_current_performance_level,
+	.force_performance_level = cz_dpm_force_dpm_level,
+	.vblank_too_short = NULL,
+	.powergate_uvd = cz_dpm_powergate_uvd,
+	.powergate_vce = cz_dpm_powergate_vce,
+};
+
+static void cz_dpm_set_funcs(struct amdgpu_device *adev)
+{
+	if (NULL == adev->pm.funcs)
+		adev->pm.funcs = &cz_dpm_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.h b/kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.h
new file mode 100644
index 000000000..99e1afc89
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_dpm.h
@@ -0,0 +1,237 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CZ_DPM_H__
+#define __CZ_DPM_H__
+
+#include "smu8_fusion.h"
+
+#define CZ_AT_DFLT					30
+#define CZ_NUM_NBPSTATES				4
+#define CZ_NUM_NBPMEMORY_CLOCK				2
+#define CZ_MAX_HARDWARE_POWERLEVELS			8
+#define CZ_MAX_DISPLAY_CLOCK_LEVEL			8
+#define CZ_MAX_DISPLAYPHY_IDS				10
+
+#define PPCZ_VOTINGRIGHTSCLIENTS_DFLT0			0x3FFFC102
+
+#define SMC_RAM_END					0x40000
+
+#define DPMFlags_SCLK_Enabled				0x00000001
+#define DPMFlags_UVD_Enabled				0x00000002
+#define DPMFlags_VCE_Enabled				0x00000004
+#define DPMFlags_ACP_Enabled				0x00000008
+#define DPMFlags_ForceHighestValid			0x40000000
+#define DPMFlags_Debug					0x80000000
+
+/* Do not change the following, it is also defined in SMU8.h */
+#define SMU_EnabledFeatureScoreboard_AcpDpmOn		0x00000001
+#define SMU_EnabledFeatureScoreboard_SclkDpmOn		0x00200000
+#define SMU_EnabledFeatureScoreboard_UvdDpmOn		0x00800000
+#define SMU_EnabledFeatureScoreboard_VceDpmOn		0x01000000
+
+/* temporary solution to SetMinDeepSleepSclk
+ * should indicate by display adaptor
+ * 10k Hz unit*/
+#define CZ_MIN_DEEP_SLEEP_SCLK				800
+
+enum cz_pt_config_reg_type {
+	CZ_CONFIGREG_MMR = 0,
+	CZ_CONFIGREG_SMC_IND,
+	CZ_CONFIGREG_DIDT_IND,
+	CZ_CONFIGREG_CACHE,
+	CZ_CONFIGREG_MAX
+};
+
+struct cz_pt_config_reg {
+	uint32_t offset;
+	uint32_t mask;
+	uint32_t shift;
+	uint32_t value;
+	enum cz_pt_config_reg_type type;
+};
+
+struct cz_dpm_entry {
+	uint32_t	soft_min_clk;
+	uint32_t	hard_min_clk;
+	uint32_t	soft_max_clk;
+	uint32_t	hard_max_clk;
+};
+
+struct cz_pl {
+	uint32_t sclk;
+	uint8_t vddc_index;
+	uint8_t ds_divider_index;
+	uint8_t ss_divider_index;
+	uint8_t allow_gnb_slow;
+	uint8_t force_nbp_state;
+	uint8_t display_wm;
+	uint8_t vce_wm;
+};
+
+struct cz_ps {
+	struct cz_pl levels[CZ_MAX_HARDWARE_POWERLEVELS];
+	uint32_t num_levels;
+	bool need_dfs_bypass;
+	uint8_t dpm0_pg_nb_ps_lo;
+	uint8_t dpm0_pg_nb_ps_hi;
+	uint8_t dpmx_nb_ps_lo;
+	uint8_t dpmx_nb_ps_hi;
+	bool force_high;
+};
+
+struct cz_displayphy_entry {
+	uint8_t phy_present;
+	uint8_t active_lane_mapping;
+	uint8_t display_conf_type;
+	uint8_t num_active_lanes;
+};
+
+struct cz_displayphy_info {
+	bool phy_access_initialized;
+	struct cz_displayphy_entry entries[CZ_MAX_DISPLAYPHY_IDS];
+};
+
+struct cz_sys_info {
+	uint32_t bootup_uma_clk;
+	uint32_t bootup_sclk;
+	uint32_t dentist_vco_freq;
+	uint32_t nb_dpm_enable;
+	uint32_t nbp_memory_clock[CZ_NUM_NBPMEMORY_CLOCK];
+	uint32_t nbp_n_clock[CZ_NUM_NBPSTATES];
+	uint8_t nbp_voltage_index[CZ_NUM_NBPSTATES];
+	uint32_t display_clock[CZ_MAX_DISPLAY_CLOCK_LEVEL];
+	uint16_t bootup_nb_voltage_index;
+	uint8_t htc_tmp_lmt;
+	uint8_t htc_hyst_lmt;
+	uint32_t uma_channel_number;
+};
+
+struct cz_power_info {
+	uint32_t active_target[CZ_MAX_HARDWARE_POWERLEVELS];
+	struct cz_sys_info sys_info;
+	struct cz_pl boot_pl;
+	bool disable_nb_ps3_in_battery;
+	bool battery_state;
+	uint32_t lowest_valid;
+	uint32_t highest_valid;
+	uint16_t high_voltage_threshold;
+	/* smc offsets */
+	uint32_t sram_end;
+	uint32_t dpm_table_start;
+	uint32_t soft_regs_start;
+	/* dpm SMU tables */
+	uint8_t uvd_level_count;
+	uint8_t vce_level_count;
+	uint8_t acp_level_count;
+	uint32_t fps_high_threshold;
+	uint32_t fps_low_threshold;
+	/* dpm table */
+	uint32_t dpm_flags;
+	struct cz_dpm_entry sclk_dpm;
+	struct cz_dpm_entry uvd_dpm;
+	struct cz_dpm_entry vce_dpm;
+	struct cz_dpm_entry acp_dpm;
+
+	uint8_t uvd_boot_level;
+	uint8_t uvd_interval;
+	uint8_t vce_boot_level;
+	uint8_t vce_interval;
+	uint8_t acp_boot_level;
+	uint8_t acp_interval;
+
+	uint8_t graphics_boot_level;
+	uint8_t graphics_interval;
+	uint8_t graphics_therm_throttle_enable;
+	uint8_t graphics_voltage_change_enable;
+	uint8_t graphics_clk_slow_enable;
+	uint8_t graphics_clk_slow_divider;
+
+	uint32_t low_sclk_interrupt_threshold;
+	bool uvd_power_gated;
+	bool vce_power_gated;
+	bool acp_power_gated;
+
+	uint32_t active_process_mask;
+
+	uint32_t mgcg_cgtt_local0;
+	uint32_t mgcg_cgtt_local1;
+	uint32_t clock_slow_down_step;
+	uint32_t skip_clock_slow_down;
+	bool enable_nb_ps_policy;
+	uint32_t voting_clients;
+	uint32_t voltage_drop_threshold;
+	uint32_t gfx_pg_threshold;
+	uint32_t max_sclk_level;
+	/* flags */
+	bool didt_enabled;
+	bool video_start;
+	bool cac_enabled;
+	bool bapm_enabled;
+	bool nb_dpm_enabled_by_driver;
+	bool nb_dpm_enabled;
+	bool auto_thermal_throttling_enabled;
+	bool dpm_enabled;
+	bool need_pptable_upload;
+	/* caps */
+	bool caps_cac;
+	bool caps_power_containment;
+	bool caps_sq_ramping;
+	bool caps_db_ramping;
+	bool caps_td_ramping;
+	bool caps_tcp_ramping;
+	bool caps_sclk_throttle_low_notification;
+	bool caps_fps;
+	bool caps_uvd_dpm;
+	bool caps_uvd_pg;
+	bool caps_vce_dpm;
+	bool caps_vce_pg;
+	bool caps_acp_dpm;
+	bool caps_acp_pg;
+	bool caps_stable_power_state;
+	bool caps_enable_dfs_bypass;
+	bool caps_sclk_ds;
+	bool caps_voltage_island;
+	/* power state */
+	struct amdgpu_ps current_rps;
+	struct cz_ps current_ps;
+	struct amdgpu_ps requested_rps;
+	struct cz_ps requested_ps;
+
+	bool uvd_power_down;
+	bool vce_power_down;
+	bool acp_power_down;
+
+	bool uvd_dynamic_pg;
+};
+
+/* cz_smc.c */
+uint32_t cz_get_argument(struct amdgpu_device *adev);
+int cz_send_msg_to_smc(struct amdgpu_device *adev, uint16_t msg);
+int cz_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+			uint16_t msg, uint32_t parameter);
+int cz_read_smc_sram_dword(struct amdgpu_device *adev,
+			uint32_t smc_address, uint32_t *value, uint32_t limit);
+int cz_smu_upload_pptable(struct amdgpu_device *adev);
+int cz_smu_download_pptable(struct amdgpu_device *adev, void **table);
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.c
new file mode 100644
index 000000000..bc751bfbc
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.c
@@ -0,0 +1,452 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "vid.h"
+
+#include "oss/oss_3_0_1_d.h"
+#include "oss/oss_3_0_1_sh_mask.h"
+
+#include "bif/bif_5_1_d.h"
+#include "bif/bif_5_1_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void cz_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * cz_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VI).
+ */
+static void cz_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * cz_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VI).
+ */
+static void cz_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 0);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	WREG32(mmIH_CNTL, ih_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * cz_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int cz_ih_irq_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	/* disable irqs */
+	cz_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+
+	/* set the writeback address whether it's enabled or not */
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	/* Default settings for IH_CNTL (disabled at first) */
+	ih_cntl = RREG32(mmIH_CNTL);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, RPTR_REARM, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	cz_ih_enable_interrupts(adev);
+
+	return ret;
+}
+
+/**
+ * cz_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VI).
+ */
+static void cz_ih_irq_disable(struct amdgpu_device *adev)
+{
+	cz_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * cz_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VI).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cz_irq_process(VI).
+ * Returns the value of the wptr.
+ */
+static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * cz_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void cz_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+	
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vm_id = (dw[2] >> 8) & 0xff;
+	entry->pas_id = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * cz_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void cz_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int cz_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cz_ih_set_interrupt_funcs(adev);
+	return 0;
+}
+
+static int cz_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int cz_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+
+	return 0;
+}
+
+static int cz_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = cz_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int cz_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	cz_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int cz_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cz_ih_hw_fini(adev);
+}
+
+static int cz_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return cz_ih_hw_init(adev);
+}
+
+static bool cz_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+		return false;
+
+	return true;
+}
+
+static int cz_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS);
+		if (!REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void cz_ih_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "CZ IH registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+	dev_info(adev->dev, "  INTERRUPT_CNTL=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL));
+	dev_info(adev->dev, "  INTERRUPT_CNTL2=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL2));
+	dev_info(adev->dev, "  IH_CNTL=0x%08X\n",
+		 RREG32(mmIH_CNTL));
+	dev_info(adev->dev, "  IH_RB_CNTL=0x%08X\n",
+		 RREG32(mmIH_RB_CNTL));
+	dev_info(adev->dev, "  IH_RB_BASE=0x%08X\n",
+		 RREG32(mmIH_RB_BASE));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_LO=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_LO));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_HI=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_HI));
+	dev_info(adev->dev, "  IH_RB_RPTR=0x%08X\n",
+		 RREG32(mmIH_RB_RPTR));
+	dev_info(adev->dev, "  IH_RB_WPTR=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR));
+}
+
+static int cz_ih_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_IH, 1);
+
+	if (srbm_soft_reset) {
+		cz_ih_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		cz_ih_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int cz_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int cz_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	// TODO
+	return 0;
+}
+
+const struct amd_ip_funcs cz_ih_ip_funcs = {
+	.early_init = cz_ih_early_init,
+	.late_init = NULL,
+	.sw_init = cz_ih_sw_init,
+	.sw_fini = cz_ih_sw_fini,
+	.hw_init = cz_ih_hw_init,
+	.hw_fini = cz_ih_hw_fini,
+	.suspend = cz_ih_suspend,
+	.resume = cz_ih_resume,
+	.is_idle = cz_ih_is_idle,
+	.wait_for_idle = cz_ih_wait_for_idle,
+	.soft_reset = cz_ih_soft_reset,
+	.print_status = cz_ih_print_status,
+	.set_clockgating_state = cz_ih_set_clockgating_state,
+	.set_powergating_state = cz_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs cz_ih_funcs = {
+	.get_wptr = cz_ih_get_wptr,
+	.decode_iv = cz_ih_decode_iv,
+	.set_rptr = cz_ih_set_rptr
+};
+
+static void cz_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &cz_ih_funcs;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.h b/kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.h
new file mode 100644
index 000000000..fc4057a2e
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __CZ_IH_H__
+#define __CZ_IH_H__
+
+extern const struct amd_ip_funcs cz_ih_ip_funcs;
+
+#endif /* __CZ_IH_H__ */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h b/kernel/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h
new file mode 100644
index 000000000..273616ab4
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_ppsmc.h
@@ -0,0 +1,185 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef CZ_PP_SMC_H
+#define CZ_PP_SMC_H
+
+#pragma pack(push, 1)
+
+/* Fan control algorithm:*/
+#define FDO_MODE_HARDWARE 0
+#define FDO_MODE_PIECE_WISE_LINEAR 1
+
+enum FAN_CONTROL {
+    FAN_CONTROL_FUZZY,
+    FAN_CONTROL_TABLE
+};
+
+enum DPM_ARRAY {
+    DPM_ARRAY_HARD_MAX,
+    DPM_ARRAY_HARD_MIN,
+    DPM_ARRAY_SOFT_MAX,
+    DPM_ARRAY_SOFT_MIN
+};
+
+/*
+ * Return codes for driver to SMC communication.
+ * Leave these #define-s, enums might not be exactly 8-bits on the microcontroller.
+ */
+#define PPSMC_Result_OK             ((uint16_t)0x01)
+#define PPSMC_Result_NoMore         ((uint16_t)0x02)
+#define PPSMC_Result_NotNow         ((uint16_t)0x03)
+#define PPSMC_Result_Failed         ((uint16_t)0xFF)
+#define PPSMC_Result_UnknownCmd     ((uint16_t)0xFE)
+#define PPSMC_Result_UnknownVT      ((uint16_t)0xFD)
+
+#define PPSMC_isERROR(x)            ((uint16_t)0x80 & (x))
+
+/*
+ * Supported driver messages
+ */
+#define PPSMC_MSG_Test                        ((uint16_t) 0x1)
+#define PPSMC_MSG_GetFeatureStatus            ((uint16_t) 0x2)
+#define PPSMC_MSG_EnableAllSmuFeatures        ((uint16_t) 0x3)
+#define PPSMC_MSG_DisableAllSmuFeatures       ((uint16_t) 0x4)
+#define PPSMC_MSG_OptimizeBattery             ((uint16_t) 0x5)
+#define PPSMC_MSG_MaximizePerf                ((uint16_t) 0x6)
+#define PPSMC_MSG_UVDPowerOFF                 ((uint16_t) 0x7)
+#define PPSMC_MSG_UVDPowerON                  ((uint16_t) 0x8)
+#define PPSMC_MSG_VCEPowerOFF                 ((uint16_t) 0x9)
+#define PPSMC_MSG_VCEPowerON                  ((uint16_t) 0xA)
+#define PPSMC_MSG_ACPPowerOFF                 ((uint16_t) 0xB)
+#define PPSMC_MSG_ACPPowerON                  ((uint16_t) 0xC)
+#define PPSMC_MSG_SDMAPowerOFF                ((uint16_t) 0xD)
+#define PPSMC_MSG_SDMAPowerON                 ((uint16_t) 0xE)
+#define PPSMC_MSG_XDMAPowerOFF                ((uint16_t) 0xF)
+#define PPSMC_MSG_XDMAPowerON                 ((uint16_t) 0x10)
+#define PPSMC_MSG_SetMinDeepSleepSclk         ((uint16_t) 0x11)
+#define PPSMC_MSG_SetSclkSoftMin              ((uint16_t) 0x12)
+#define PPSMC_MSG_SetSclkSoftMax              ((uint16_t) 0x13)
+#define PPSMC_MSG_SetSclkHardMin              ((uint16_t) 0x14)
+#define PPSMC_MSG_SetSclkHardMax              ((uint16_t) 0x15)
+#define PPSMC_MSG_SetLclkSoftMin              ((uint16_t) 0x16)
+#define PPSMC_MSG_SetLclkSoftMax              ((uint16_t) 0x17)
+#define PPSMC_MSG_SetLclkHardMin              ((uint16_t) 0x18)
+#define PPSMC_MSG_SetLclkHardMax              ((uint16_t) 0x19)
+#define PPSMC_MSG_SetUvdSoftMin               ((uint16_t) 0x1A)
+#define PPSMC_MSG_SetUvdSoftMax               ((uint16_t) 0x1B)
+#define PPSMC_MSG_SetUvdHardMin               ((uint16_t) 0x1C)
+#define PPSMC_MSG_SetUvdHardMax               ((uint16_t) 0x1D)
+#define PPSMC_MSG_SetEclkSoftMin              ((uint16_t) 0x1E)
+#define PPSMC_MSG_SetEclkSoftMax              ((uint16_t) 0x1F)
+#define PPSMC_MSG_SetEclkHardMin              ((uint16_t) 0x20)
+#define PPSMC_MSG_SetEclkHardMax              ((uint16_t) 0x21)
+#define PPSMC_MSG_SetAclkSoftMin              ((uint16_t) 0x22)
+#define PPSMC_MSG_SetAclkSoftMax              ((uint16_t) 0x23)
+#define PPSMC_MSG_SetAclkHardMin              ((uint16_t) 0x24)
+#define PPSMC_MSG_SetAclkHardMax              ((uint16_t) 0x25)
+#define PPSMC_MSG_SetNclkSoftMin              ((uint16_t) 0x26)
+#define PPSMC_MSG_SetNclkSoftMax              ((uint16_t) 0x27)
+#define PPSMC_MSG_SetNclkHardMin              ((uint16_t) 0x28)
+#define PPSMC_MSG_SetNclkHardMax              ((uint16_t) 0x29)
+#define PPSMC_MSG_SetPstateSoftMin            ((uint16_t) 0x2A)
+#define PPSMC_MSG_SetPstateSoftMax            ((uint16_t) 0x2B)
+#define PPSMC_MSG_SetPstateHardMin            ((uint16_t) 0x2C)
+#define PPSMC_MSG_SetPstateHardMax            ((uint16_t) 0x2D)
+#define PPSMC_MSG_DisableLowMemoryPstate      ((uint16_t) 0x2E)
+#define PPSMC_MSG_EnableLowMemoryPstate       ((uint16_t) 0x2F)
+#define PPSMC_MSG_UcodeAddressLow             ((uint16_t) 0x30)
+#define PPSMC_MSG_UcodeAddressHigh            ((uint16_t) 0x31)
+#define PPSMC_MSG_UcodeLoadStatus             ((uint16_t) 0x32)
+#define PPSMC_MSG_DriverDramAddrHi            ((uint16_t) 0x33)
+#define PPSMC_MSG_DriverDramAddrLo            ((uint16_t) 0x34)
+#define PPSMC_MSG_CondExecDramAddrHi          ((uint16_t) 0x35)
+#define PPSMC_MSG_CondExecDramAddrLo          ((uint16_t) 0x36)
+#define PPSMC_MSG_LoadUcodes                  ((uint16_t) 0x37)
+#define PPSMC_MSG_DriverResetMode             ((uint16_t) 0x38)
+#define PPSMC_MSG_PowerStateNotify            ((uint16_t) 0x39)
+#define PPSMC_MSG_SetDisplayPhyConfig         ((uint16_t) 0x3A)
+#define PPSMC_MSG_GetMaxSclkLevel             ((uint16_t) 0x3B)
+#define PPSMC_MSG_GetMaxLclkLevel             ((uint16_t) 0x3C)
+#define PPSMC_MSG_GetMaxUvdLevel              ((uint16_t) 0x3D)
+#define PPSMC_MSG_GetMaxEclkLevel             ((uint16_t) 0x3E)
+#define PPSMC_MSG_GetMaxAclkLevel             ((uint16_t) 0x3F)
+#define PPSMC_MSG_GetMaxNclkLevel             ((uint16_t) 0x40)
+#define PPSMC_MSG_GetMaxPstate                ((uint16_t) 0x41)
+#define PPSMC_MSG_DramAddrHiVirtual           ((uint16_t) 0x42)
+#define PPSMC_MSG_DramAddrLoVirtual           ((uint16_t) 0x43)
+#define PPSMC_MSG_DramAddrHiPhysical          ((uint16_t) 0x44)
+#define PPSMC_MSG_DramAddrLoPhysical          ((uint16_t) 0x45)
+#define PPSMC_MSG_DramBufferSize              ((uint16_t) 0x46)
+#define PPSMC_MSG_SetMmPwrLogDramAddrHi       ((uint16_t) 0x47)
+#define PPSMC_MSG_SetMmPwrLogDramAddrLo       ((uint16_t) 0x48)
+#define PPSMC_MSG_SetClkTableAddrHi           ((uint16_t) 0x49)
+#define PPSMC_MSG_SetClkTableAddrLo           ((uint16_t) 0x4A)
+#define PPSMC_MSG_GetConservativePowerLimit   ((uint16_t) 0x4B)
+
+#define PPSMC_MSG_InitJobs                    ((uint16_t) 0x252)
+#define PPSMC_MSG_ExecuteJob                  ((uint16_t) 0x254)
+
+#define PPSMC_MSG_NBDPM_Enable                ((uint16_t) 0x140)
+#define PPSMC_MSG_NBDPM_Disable               ((uint16_t) 0x141)
+
+#define PPSMC_MSG_DPM_FPS_Mode                ((uint16_t) 0x15d)
+#define PPSMC_MSG_DPM_Activity_Mode           ((uint16_t) 0x15e)
+
+#define PPSMC_MSG_PmStatusLogStart            ((uint16_t) 0x170)
+#define PPSMC_MSG_PmStatusLogSample           ((uint16_t) 0x171)
+
+#define PPSMC_MSG_AllowLowSclkInterrupt       ((uint16_t) 0x184)
+#define PPSMC_MSG_MmPowerMonitorStart         ((uint16_t) 0x18F)
+#define PPSMC_MSG_MmPowerMonitorStop          ((uint16_t) 0x190)
+#define PPSMC_MSG_MmPowerMonitorRestart       ((uint16_t) 0x191)
+
+#define PPSMC_MSG_SetClockGateMask            ((uint16_t) 0x260)
+#define PPSMC_MSG_SetFpsThresholdLo           ((uint16_t) 0x264)
+#define PPSMC_MSG_SetFpsThresholdHi           ((uint16_t) 0x265)
+#define PPSMC_MSG_SetLowSclkIntrThreshold     ((uint16_t) 0x266)
+
+#define PPSMC_MSG_ClkTableXferToDram          ((uint16_t) 0x267)
+#define PPSMC_MSG_ClkTableXferToSmu           ((uint16_t) 0x268)
+#define PPSMC_MSG_GetAverageGraphicsActivity  ((uint16_t) 0x269)
+#define PPSMC_MSG_GetAverageGioActivity       ((uint16_t) 0x26A)
+#define PPSMC_MSG_SetLoggerBufferSize         ((uint16_t) 0x26B)
+#define PPSMC_MSG_SetLoggerAddressHigh        ((uint16_t) 0x26C)
+#define PPSMC_MSG_SetLoggerAddressLow         ((uint16_t) 0x26D)
+#define PPSMC_MSG_SetWatermarkFrequency       ((uint16_t) 0x26E)
+
+/* REMOVE LATER*/
+#define PPSMC_MSG_DPM_ForceState              ((uint16_t) 0x104)
+
+/* Feature Enable Masks*/
+#define NB_DPM_MASK             0x00000800
+#define VDDGFX_MASK             0x00800000
+#define VCE_DPM_MASK            0x00400000
+#define ACP_DPM_MASK            0x00040000
+#define UVD_DPM_MASK            0x00010000
+#define GFX_CU_PG_MASK          0x00004000
+#define SCLK_DPM_MASK           0x00080000
+
+#if !defined(SMC_MICROCODE)
+#pragma pack(pop)
+
+#endif
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_smc.c b/kernel/drivers/gpu/drm/amd/amdgpu/cz_smc.c
new file mode 100644
index 000000000..ac7fee7b7
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_smc.c
@@ -0,0 +1,1000 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "smu8.h"
+#include "smu8_fusion.h"
+#include "cz_ppsmc.h"
+#include "cz_smumgr.h"
+#include "smu_ucode_xfer_cz.h"
+#include "amdgpu_ucode.h"
+
+#include "smu/smu_8_0_d.h"
+#include "smu/smu_8_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+uint32_t cz_get_argument(struct amdgpu_device *adev)
+{
+	return RREG32(mmSMU_MP1_SRBM2P_ARG_0);
+}
+
+static struct cz_smu_private_data *cz_smu_get_priv(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv =
+			(struct cz_smu_private_data *)(adev->smu.priv);
+
+	return priv;
+}
+
+int cz_send_msg_to_smc_async(struct amdgpu_device *adev, u16 msg)
+{
+	int i;
+	u32 content = 0, tmp;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = REG_GET_FIELD(RREG32(mmSMU_MP1_SRBM2P_RESP_0),
+				SMU_MP1_SRBM2P_RESP_0, CONTENT);
+		if (content != tmp)
+			break;
+		udelay(1);
+	}
+
+	/* timeout means wrong logic*/
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	WREG32(mmSMU_MP1_SRBM2P_RESP_0, 0);
+	WREG32(mmSMU_MP1_SRBM2P_MSG_0, msg);
+
+	return 0;
+}
+
+int cz_send_msg_to_smc(struct amdgpu_device *adev, u16 msg)
+{
+	int i;
+	u32 content = 0, tmp = 0;
+
+	if (cz_send_msg_to_smc_async(adev, msg))
+		return -EINVAL;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = REG_GET_FIELD(RREG32(mmSMU_MP1_SRBM2P_RESP_0),
+				SMU_MP1_SRBM2P_RESP_0, CONTENT);
+		if (content != tmp)
+			break;
+		udelay(1);
+	}
+
+	/* timeout means wrong logic*/
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	if (PPSMC_Result_OK != tmp) {
+		dev_err(adev->dev, "SMC Failed to send Message.\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int cz_send_msg_to_smc_with_parameter_async(struct amdgpu_device *adev,
+						u16 msg, u32 parameter)
+{
+	WREG32(mmSMU_MP1_SRBM2P_ARG_0, parameter);
+	return cz_send_msg_to_smc_async(adev, msg);
+}
+
+int cz_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						u16 msg, u32 parameter)
+{
+	WREG32(mmSMU_MP1_SRBM2P_ARG_0, parameter);
+	return cz_send_msg_to_smc(adev, msg);
+}
+
+static int cz_set_smc_sram_address(struct amdgpu_device *adev,
+						u32 smc_address, u32 limit)
+{
+	if (smc_address & 3)
+		return -EINVAL;
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmMP0PUB_IND_INDEX_0, SMN_MP1_SRAM_START_ADDR + smc_address);
+
+	return 0;
+}
+
+int cz_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+						u32 *value, u32 limit)
+{
+	int ret;
+
+	ret = cz_set_smc_sram_address(adev, smc_address, limit);
+	if (ret)
+		return ret;
+
+	*value = RREG32(mmMP0PUB_IND_DATA_0);
+
+	return 0;
+}
+
+int cz_write_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+						u32 value, u32 limit)
+{
+	int ret;
+
+	ret = cz_set_smc_sram_address(adev, smc_address, limit);
+	if (ret)
+		return ret;
+
+	WREG32(mmMP0PUB_IND_DATA_0, value);
+
+	return 0;
+}
+
+static int cz_smu_request_load_fw(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+
+	uint32_t smc_addr = SMU8_FIRMWARE_HEADER_LOCATION +
+			offsetof(struct SMU8_Firmware_Header, UcodeLoadStatus);
+
+	cz_write_smc_sram_dword(adev, smc_addr, 0, smc_addr + 4);
+
+	/*prepare toc buffers*/
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_DriverDramAddrHi,
+				priv->toc_buffer.mc_addr_high);
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_DriverDramAddrLo,
+				priv->toc_buffer.mc_addr_low);
+	cz_send_msg_to_smc(adev, PPSMC_MSG_InitJobs);
+
+	/*execute jobs*/
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_ExecuteJob,
+				priv->toc_entry_aram);
+
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_ExecuteJob,
+				priv->toc_entry_power_profiling_index);
+
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_ExecuteJob,
+				priv->toc_entry_initialize_index);
+
+	return 0;
+}
+
+/*
+ *Check if the FW has been loaded, SMU will not return if loading
+ *has not finished.
+ */
+static int cz_smu_check_fw_load_finish(struct amdgpu_device *adev,
+						uint32_t fw_mask)
+{
+	int i;
+	uint32_t index = SMN_MP1_SRAM_START_ADDR +
+			SMU8_FIRMWARE_HEADER_LOCATION +
+			offsetof(struct SMU8_Firmware_Header, UcodeLoadStatus);
+
+	WREG32(mmMP0PUB_IND_INDEX, index);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (fw_mask == (RREG32(mmMP0PUB_IND_DATA) & fw_mask))
+			break;
+		udelay(1);
+	}
+
+	if (i >= adev->usec_timeout) {
+		dev_err(adev->dev,
+		"SMU check loaded firmware failed, expecting 0x%x, getting 0x%x",
+		fw_mask, RREG32(mmMP0PUB_IND_DATA));
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/*
+ * interfaces for different ip blocks to check firmware loading status
+ * 0 for success otherwise failed
+ */
+static int cz_smu_check_finished(struct amdgpu_device *adev,
+							enum AMDGPU_UCODE_ID id)
+{
+	switch (id) {
+	case AMDGPU_UCODE_ID_SDMA0:
+		if (adev->smu.fw_flags & AMDGPU_SDMA0_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_SDMA1:
+		if (adev->smu.fw_flags & AMDGPU_SDMA1_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_CP_CE:
+		if (adev->smu.fw_flags & AMDGPU_CPCE_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_CP_PFP:
+		if (adev->smu.fw_flags & AMDGPU_CPPFP_UCODE_LOADED)
+			return 0;
+	case AMDGPU_UCODE_ID_CP_ME:
+		if (adev->smu.fw_flags & AMDGPU_CPME_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC1:
+		if (adev->smu.fw_flags & AMDGPU_CPMEC1_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_CP_MEC2:
+		if (adev->smu.fw_flags & AMDGPU_CPMEC2_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_RLC_G:
+		if (adev->smu.fw_flags & AMDGPU_CPRLC_UCODE_LOADED)
+			return 0;
+		break;
+	case AMDGPU_UCODE_ID_MAXIMUM:
+	default:
+		break;
+	}
+
+	return 1;
+}
+
+static int cz_load_mec_firmware(struct amdgpu_device *adev)
+{
+	struct amdgpu_firmware_info *ucode =
+				&adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
+	uint32_t reg_data;
+	uint32_t tmp;
+
+	if (ucode->fw == NULL)
+		return -EINVAL;
+
+	/* Disable MEC parsing/prefetching */
+	tmp = RREG32(mmCP_MEC_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_CNTL, MEC_ME1_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_CNTL, MEC_ME2_HALT, 1);
+	WREG32(mmCP_MEC_CNTL, tmp);
+
+	tmp = RREG32(mmCP_CPC_IC_BASE_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, VMID, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, ATC, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0);
+	tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, MTYPE, 1);
+	WREG32(mmCP_CPC_IC_BASE_CNTL, tmp);
+
+	reg_data = lower_32_bits(ucode->mc_addr) &
+			REG_FIELD_MASK(CP_CPC_IC_BASE_LO, IC_BASE_LO);
+	WREG32(mmCP_CPC_IC_BASE_LO, reg_data);
+
+	reg_data = upper_32_bits(ucode->mc_addr) &
+			REG_FIELD_MASK(CP_CPC_IC_BASE_HI, IC_BASE_HI);
+	WREG32(mmCP_CPC_IC_BASE_HI, reg_data);
+
+	return 0;
+}
+
+int cz_smu_start(struct amdgpu_device *adev)
+{
+	int ret = 0;
+
+	uint32_t fw_to_check = UCODE_ID_RLC_G_MASK |
+				UCODE_ID_SDMA0_MASK |
+				UCODE_ID_SDMA1_MASK |
+				UCODE_ID_CP_CE_MASK |
+				UCODE_ID_CP_ME_MASK |
+				UCODE_ID_CP_PFP_MASK |
+				UCODE_ID_CP_MEC_JT1_MASK |
+				UCODE_ID_CP_MEC_JT2_MASK;
+
+	if (adev->asic_type == CHIP_STONEY)
+		fw_to_check &= ~(UCODE_ID_SDMA1_MASK | UCODE_ID_CP_MEC_JT2_MASK);
+
+	cz_smu_request_load_fw(adev);
+	ret = cz_smu_check_fw_load_finish(adev, fw_to_check);
+	if (ret)
+		return ret;
+
+	/* manually load MEC firmware for CZ */
+	if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY) {
+		ret = cz_load_mec_firmware(adev);
+		if (ret) {
+			dev_err(adev->dev, "(%d) Mec Firmware load failed\n", ret);
+			return ret;
+		}
+	}
+
+	/* setup fw load flag */
+	adev->smu.fw_flags = AMDGPU_SDMA0_UCODE_LOADED |
+				AMDGPU_SDMA1_UCODE_LOADED |
+				AMDGPU_CPCE_UCODE_LOADED |
+				AMDGPU_CPPFP_UCODE_LOADED |
+				AMDGPU_CPME_UCODE_LOADED |
+				AMDGPU_CPMEC1_UCODE_LOADED |
+				AMDGPU_CPMEC2_UCODE_LOADED |
+				AMDGPU_CPRLC_UCODE_LOADED;
+
+	if (adev->asic_type == CHIP_STONEY)
+		adev->smu.fw_flags &= ~(AMDGPU_SDMA1_UCODE_LOADED | AMDGPU_CPMEC2_UCODE_LOADED);
+
+	return ret;
+}
+
+static uint32_t cz_convert_fw_type(uint32_t fw_type)
+{
+	enum AMDGPU_UCODE_ID result = AMDGPU_UCODE_ID_MAXIMUM;
+
+	switch (fw_type) {
+	case UCODE_ID_SDMA0:
+		result = AMDGPU_UCODE_ID_SDMA0;
+		break;
+	case UCODE_ID_SDMA1:
+		result = AMDGPU_UCODE_ID_SDMA1;
+		break;
+	case UCODE_ID_CP_CE:
+		result = AMDGPU_UCODE_ID_CP_CE;
+		break;
+	case UCODE_ID_CP_PFP:
+		result = AMDGPU_UCODE_ID_CP_PFP;
+		break;
+	case UCODE_ID_CP_ME:
+		result = AMDGPU_UCODE_ID_CP_ME;
+		break;
+	case UCODE_ID_CP_MEC_JT1:
+	case UCODE_ID_CP_MEC_JT2:
+		result = AMDGPU_UCODE_ID_CP_MEC1;
+		break;
+	case UCODE_ID_RLC_G:
+		result = AMDGPU_UCODE_ID_RLC_G;
+		break;
+	default:
+		DRM_ERROR("UCode type is out of range!");
+	}
+
+	return result;
+}
+
+static uint8_t cz_smu_translate_firmware_enum_to_arg(
+			enum cz_scratch_entry firmware_enum)
+{
+	uint8_t ret = 0;
+
+	switch (firmware_enum) {
+	case CZ_SCRATCH_ENTRY_UCODE_ID_SDMA0:
+		ret = UCODE_ID_SDMA0;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_SDMA1:
+		ret = UCODE_ID_SDMA1;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_CP_CE:
+		ret = UCODE_ID_CP_CE;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_CP_PFP:
+		ret = UCODE_ID_CP_PFP;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_CP_ME:
+		ret = UCODE_ID_CP_ME;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1:
+		ret = UCODE_ID_CP_MEC_JT1;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT2:
+		ret = UCODE_ID_CP_MEC_JT2;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_GMCON_RENG:
+		ret = UCODE_ID_GMCON_RENG;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_RLC_G:
+		ret = UCODE_ID_RLC_G;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SCRATCH:
+		ret = UCODE_ID_RLC_SCRATCH;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_ARAM:
+		ret = UCODE_ID_RLC_SRM_ARAM;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_DRAM:
+		ret = UCODE_ID_RLC_SRM_DRAM;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_DMCU_ERAM:
+		ret = UCODE_ID_DMCU_ERAM;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_DMCU_IRAM:
+		ret = UCODE_ID_DMCU_IRAM;
+		break;
+	case CZ_SCRATCH_ENTRY_UCODE_ID_POWER_PROFILING:
+		ret = TASK_ARG_INIT_MM_PWR_LOG;
+		break;
+	case CZ_SCRATCH_ENTRY_DATA_ID_SDMA_HALT:
+	case CZ_SCRATCH_ENTRY_DATA_ID_SYS_CLOCKGATING:
+	case CZ_SCRATCH_ENTRY_DATA_ID_SDMA_RING_REGS:
+	case CZ_SCRATCH_ENTRY_DATA_ID_NONGFX_REINIT:
+	case CZ_SCRATCH_ENTRY_DATA_ID_SDMA_START:
+	case CZ_SCRATCH_ENTRY_DATA_ID_IH_REGISTERS:
+		ret = TASK_ARG_REG_MMIO;
+		break;
+	case CZ_SCRATCH_ENTRY_SMU8_FUSION_CLKTABLE:
+		ret = TASK_ARG_INIT_CLK_TABLE;
+		break;
+	}
+
+	return ret;
+}
+
+static int cz_smu_populate_single_firmware_entry(struct amdgpu_device *adev,
+					enum cz_scratch_entry firmware_enum,
+					struct cz_buffer_entry *entry)
+{
+	uint64_t gpu_addr;
+	uint32_t data_size;
+	uint8_t ucode_id = cz_smu_translate_firmware_enum_to_arg(firmware_enum);
+	enum AMDGPU_UCODE_ID id = cz_convert_fw_type(ucode_id);
+	struct amdgpu_firmware_info *ucode = &adev->firmware.ucode[id];
+	const struct gfx_firmware_header_v1_0 *header;
+
+	if (ucode->fw == NULL)
+		return -EINVAL;
+
+	gpu_addr  = ucode->mc_addr;
+	header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+	data_size = le32_to_cpu(header->header.ucode_size_bytes);
+
+	if ((firmware_enum == CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1) ||
+	    (firmware_enum == CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT2)) {
+		gpu_addr += le32_to_cpu(header->jt_offset) << 2;
+		data_size = le32_to_cpu(header->jt_size) << 2;
+	}
+
+	entry->mc_addr_low = lower_32_bits(gpu_addr);
+	entry->mc_addr_high = upper_32_bits(gpu_addr);
+	entry->data_size = data_size;
+	entry->firmware_ID = firmware_enum;
+
+	return 0;
+}
+
+static int cz_smu_populate_single_scratch_entry(struct amdgpu_device *adev,
+					enum cz_scratch_entry scratch_type,
+					uint32_t size_in_byte,
+					struct cz_buffer_entry *entry)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	uint64_t mc_addr = (((uint64_t) priv->smu_buffer.mc_addr_high) << 32) |
+						priv->smu_buffer.mc_addr_low;
+	mc_addr += size_in_byte;
+
+	priv->smu_buffer_used_bytes += size_in_byte;
+	entry->data_size = size_in_byte;
+	entry->kaddr = priv->smu_buffer.kaddr + priv->smu_buffer_used_bytes;
+	entry->mc_addr_low = lower_32_bits(mc_addr);
+	entry->mc_addr_high = upper_32_bits(mc_addr);
+	entry->firmware_ID = scratch_type;
+
+	return 0;
+}
+
+static int cz_smu_populate_single_ucode_load_task(struct amdgpu_device *adev,
+						enum cz_scratch_entry firmware_enum,
+						bool is_last)
+{
+	uint8_t i;
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	struct TOC *toc = (struct TOC *)priv->toc_buffer.kaddr;
+	struct SMU_Task *task = &toc->tasks[priv->toc_entry_used_count++];
+
+	task->type = TASK_TYPE_UCODE_LOAD;
+	task->arg = cz_smu_translate_firmware_enum_to_arg(firmware_enum);
+	task->next = is_last ? END_OF_TASK_LIST : priv->toc_entry_used_count;
+
+	for (i = 0; i < priv->driver_buffer_length; i++)
+		if (priv->driver_buffer[i].firmware_ID == firmware_enum)
+			break;
+
+	if (i >= priv->driver_buffer_length) {
+		dev_err(adev->dev, "Invalid Firmware Type\n");
+		return -EINVAL;
+	}
+
+	task->addr.low = priv->driver_buffer[i].mc_addr_low;
+	task->addr.high = priv->driver_buffer[i].mc_addr_high;
+	task->size_bytes = priv->driver_buffer[i].data_size;
+
+	return 0;
+}
+
+static int cz_smu_populate_single_scratch_task(struct amdgpu_device *adev,
+						enum cz_scratch_entry firmware_enum,
+						uint8_t type, bool is_last)
+{
+	uint8_t i;
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	struct TOC *toc = (struct TOC *)priv->toc_buffer.kaddr;
+	struct SMU_Task *task = &toc->tasks[priv->toc_entry_used_count++];
+
+	task->type = type;
+	task->arg = cz_smu_translate_firmware_enum_to_arg(firmware_enum);
+	task->next = is_last ? END_OF_TASK_LIST : priv->toc_entry_used_count;
+
+	for (i = 0; i < priv->scratch_buffer_length; i++)
+		if (priv->scratch_buffer[i].firmware_ID == firmware_enum)
+			break;
+
+	if (i >= priv->scratch_buffer_length) {
+		dev_err(adev->dev, "Invalid Firmware Type\n");
+		return -EINVAL;
+	}
+
+	task->addr.low = priv->scratch_buffer[i].mc_addr_low;
+	task->addr.high = priv->scratch_buffer[i].mc_addr_high;
+	task->size_bytes = priv->scratch_buffer[i].data_size;
+
+	if (CZ_SCRATCH_ENTRY_DATA_ID_IH_REGISTERS == firmware_enum) {
+		struct cz_ih_meta_data *pIHReg_restore =
+			(struct cz_ih_meta_data *)priv->scratch_buffer[i].kaddr;
+		pIHReg_restore->command =
+			METADATA_CMD_MODE0 | METADATA_PERFORM_ON_LOAD;
+	}
+
+	return 0;
+}
+
+static int cz_smu_construct_toc_for_rlc_aram_save(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	priv->toc_entry_aram = priv->toc_entry_used_count;
+	cz_smu_populate_single_scratch_task(adev,
+			CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_ARAM,
+			TASK_TYPE_UCODE_SAVE, true);
+
+	return 0;
+}
+
+static int cz_smu_construct_toc_for_vddgfx_enter(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	struct TOC *toc = (struct TOC *)priv->toc_buffer.kaddr;
+
+	toc->JobList[JOB_GFX_SAVE] = (uint8_t)priv->toc_entry_used_count;
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SCRATCH,
+				TASK_TYPE_UCODE_SAVE, false);
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_DRAM,
+				TASK_TYPE_UCODE_SAVE, true);
+
+	return 0;
+}
+
+static int cz_smu_construct_toc_for_vddgfx_exit(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	struct TOC *toc = (struct TOC *)priv->toc_buffer.kaddr;
+
+	toc->JobList[JOB_GFX_RESTORE] = (uint8_t)priv->toc_entry_used_count;
+
+	/* populate ucode */
+	if (adev->firmware.smu_load) {
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_CE, false);
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_PFP, false);
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_ME, false);
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1, false);
+		if (adev->asic_type == CHIP_STONEY) {
+			cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1, false);
+		} else {
+			cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT2, false);
+		}
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_G, false);
+	}
+
+	/* populate scratch */
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SCRATCH,
+				TASK_TYPE_UCODE_LOAD, false);
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_ARAM,
+				TASK_TYPE_UCODE_LOAD, false);
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_DRAM,
+				TASK_TYPE_UCODE_LOAD, true);
+
+	return 0;
+}
+
+static int cz_smu_construct_toc_for_power_profiling(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+
+	priv->toc_entry_power_profiling_index = priv->toc_entry_used_count;
+
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_POWER_PROFILING,
+				TASK_TYPE_INITIALIZE, true);
+	return 0;
+}
+
+static int cz_smu_construct_toc_for_bootup(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+
+	priv->toc_entry_initialize_index = priv->toc_entry_used_count;
+
+	if (adev->firmware.smu_load) {
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_SDMA0, false);
+		if (adev->asic_type == CHIP_STONEY) {
+			cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_SDMA0, false);
+		} else {
+			cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_SDMA1, false);
+		}
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_CE, false);
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_PFP, false);
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_ME, false);
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1, false);
+		if (adev->asic_type == CHIP_STONEY) {
+			cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1, false);
+		} else {
+			cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT2, false);
+		}
+		cz_smu_populate_single_ucode_load_task(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_G, true);
+	}
+
+	return 0;
+}
+
+static int cz_smu_construct_toc_for_clock_table(struct amdgpu_device *adev)
+{
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+
+	priv->toc_entry_clock_table = priv->toc_entry_used_count;
+
+	cz_smu_populate_single_scratch_task(adev,
+				CZ_SCRATCH_ENTRY_SMU8_FUSION_CLKTABLE,
+				TASK_TYPE_INITIALIZE, true);
+
+	return 0;
+}
+
+static int cz_smu_initialize_toc_empty_job_list(struct amdgpu_device *adev)
+{
+	int i;
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+	struct TOC *toc = (struct TOC *)priv->toc_buffer.kaddr;
+
+	for (i = 0; i < NUM_JOBLIST_ENTRIES; i++)
+		toc->JobList[i] = (uint8_t)IGNORE_JOB;
+
+	return 0;
+}
+
+/*
+ * cz smu uninitialization
+ */
+int cz_smu_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->smu.toc_buf);
+	amdgpu_bo_unref(&adev->smu.smu_buf);
+	kfree(adev->smu.priv);
+	adev->smu.priv = NULL;
+	if (adev->firmware.smu_load)
+		amdgpu_ucode_fini_bo(adev);
+
+	return 0;
+}
+
+int cz_smu_download_pptable(struct amdgpu_device *adev, void **table)
+{
+	uint8_t i;
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+
+	for (i = 0; i < priv->scratch_buffer_length; i++)
+		if (priv->scratch_buffer[i].firmware_ID ==
+				CZ_SCRATCH_ENTRY_SMU8_FUSION_CLKTABLE)
+			break;
+
+	if (i >= priv->scratch_buffer_length) {
+		dev_err(adev->dev, "Invalid Scratch Type\n");
+		return -EINVAL;
+	}
+
+	*table = (struct SMU8_Fusion_ClkTable *)priv->scratch_buffer[i].kaddr;
+
+	/* prepare buffer for pptable */
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetClkTableAddrHi,
+				priv->scratch_buffer[i].mc_addr_high);
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetClkTableAddrLo,
+				priv->scratch_buffer[i].mc_addr_low);
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_ExecuteJob,
+				priv->toc_entry_clock_table);
+
+	/* actual downloading */
+	cz_send_msg_to_smc(adev, PPSMC_MSG_ClkTableXferToDram);
+
+	return 0;
+}
+
+int cz_smu_upload_pptable(struct amdgpu_device *adev)
+{
+	uint8_t i;
+	struct cz_smu_private_data *priv = cz_smu_get_priv(adev);
+
+	for (i = 0; i < priv->scratch_buffer_length; i++)
+		if (priv->scratch_buffer[i].firmware_ID ==
+				CZ_SCRATCH_ENTRY_SMU8_FUSION_CLKTABLE)
+			break;
+
+	if (i >= priv->scratch_buffer_length) {
+		dev_err(adev->dev, "Invalid Scratch Type\n");
+		return -EINVAL;
+	}
+
+	/* prepare SMU */
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetClkTableAddrHi,
+				priv->scratch_buffer[i].mc_addr_high);
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_SetClkTableAddrLo,
+				priv->scratch_buffer[i].mc_addr_low);
+	cz_send_msg_to_smc_with_parameter(adev,
+				PPSMC_MSG_ExecuteJob,
+				priv->toc_entry_clock_table);
+
+	/* actual uploading */
+	cz_send_msg_to_smc(adev, PPSMC_MSG_ClkTableXferToSmu);
+
+	return 0;
+}
+
+/*
+ * cz smumgr functions initialization
+ */
+static const struct amdgpu_smumgr_funcs cz_smumgr_funcs = {
+	.check_fw_load_finish = cz_smu_check_finished,
+	.request_smu_load_fw = NULL,
+	.request_smu_specific_fw = NULL,
+};
+
+/*
+ * cz smu initialization
+ */
+int cz_smu_init(struct amdgpu_device *adev)
+{
+	int ret = -EINVAL;
+	uint64_t mc_addr = 0;
+	struct amdgpu_bo **toc_buf = &adev->smu.toc_buf;
+	struct amdgpu_bo **smu_buf = &adev->smu.smu_buf;
+	void *toc_buf_ptr = NULL;
+	void *smu_buf_ptr = NULL;
+
+	struct cz_smu_private_data *priv =
+		kzalloc(sizeof(struct cz_smu_private_data), GFP_KERNEL);
+	if (priv == NULL)
+		return -ENOMEM;
+
+	/* allocate firmware buffers */
+	if (adev->firmware.smu_load)
+		amdgpu_ucode_init_bo(adev);
+
+	adev->smu.priv = priv;
+	adev->smu.fw_flags = 0;
+	priv->toc_buffer.data_size = 4096;
+
+	priv->smu_buffer.data_size =
+				ALIGN(UCODE_ID_RLC_SCRATCH_SIZE_BYTE, 32) +
+				ALIGN(UCODE_ID_RLC_SRM_ARAM_SIZE_BYTE, 32) +
+				ALIGN(UCODE_ID_RLC_SRM_DRAM_SIZE_BYTE, 32) +
+				ALIGN(sizeof(struct SMU8_MultimediaPowerLogData), 32) +
+				ALIGN(sizeof(struct SMU8_Fusion_ClkTable), 32);
+
+	/* prepare toc buffer and smu buffer:
+	* 1. create amdgpu_bo for toc buffer and smu buffer
+	* 2. pin mc address
+	* 3. map kernel virtual address
+	*/
+	ret = amdgpu_bo_create(adev, priv->toc_buffer.data_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
+			       toc_buf);
+
+	if (ret) {
+		dev_err(adev->dev, "(%d) SMC TOC buffer allocation failed\n", ret);
+		return ret;
+	}
+
+	ret = amdgpu_bo_create(adev, priv->smu_buffer.data_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
+			       smu_buf);
+
+	if (ret) {
+		dev_err(adev->dev, "(%d) SMC Internal buffer allocation failed\n", ret);
+		return ret;
+	}
+
+	/* toc buffer reserve/pin/map */
+	ret = amdgpu_bo_reserve(adev->smu.toc_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		dev_err(adev->dev, "(%d) SMC TOC buffer reserve failed\n", ret);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.toc_buf, AMDGPU_GEM_DOMAIN_GTT, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		dev_err(adev->dev, "(%d) SMC TOC buffer pin failed\n", ret);
+		return ret;
+	}
+
+	ret = amdgpu_bo_kmap(*toc_buf, &toc_buf_ptr);
+	if (ret)
+		goto smu_init_failed;
+
+	amdgpu_bo_unreserve(adev->smu.toc_buf);
+
+	priv->toc_buffer.mc_addr_low = lower_32_bits(mc_addr);
+	priv->toc_buffer.mc_addr_high = upper_32_bits(mc_addr);
+	priv->toc_buffer.kaddr = toc_buf_ptr;
+
+	/* smu buffer reserve/pin/map */
+	ret = amdgpu_bo_reserve(adev->smu.smu_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		dev_err(adev->dev, "(%d) SMC Internal buffer reserve failed\n", ret);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.smu_buf, AMDGPU_GEM_DOMAIN_GTT, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		dev_err(adev->dev, "(%d) SMC Internal buffer pin failed\n", ret);
+		return ret;
+	}
+
+	ret = amdgpu_bo_kmap(*smu_buf, &smu_buf_ptr);
+	if (ret)
+		goto smu_init_failed;
+
+	amdgpu_bo_unreserve(adev->smu.smu_buf);
+
+	priv->smu_buffer.mc_addr_low = lower_32_bits(mc_addr);
+	priv->smu_buffer.mc_addr_high = upper_32_bits(mc_addr);
+	priv->smu_buffer.kaddr = smu_buf_ptr;
+
+	if (adev->firmware.smu_load) {
+		if (cz_smu_populate_single_firmware_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_SDMA0,
+				&priv->driver_buffer[priv->driver_buffer_length++]))
+			goto smu_init_failed;
+
+		if (adev->asic_type == CHIP_STONEY) {
+			if (cz_smu_populate_single_firmware_entry(adev,
+					CZ_SCRATCH_ENTRY_UCODE_ID_SDMA0,
+					&priv->driver_buffer[priv->driver_buffer_length++]))
+				goto smu_init_failed;
+		} else {
+			if (cz_smu_populate_single_firmware_entry(adev,
+					CZ_SCRATCH_ENTRY_UCODE_ID_SDMA1,
+					&priv->driver_buffer[priv->driver_buffer_length++]))
+				goto smu_init_failed;
+		}
+		if (cz_smu_populate_single_firmware_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_CE,
+				&priv->driver_buffer[priv->driver_buffer_length++]))
+			goto smu_init_failed;
+		if (cz_smu_populate_single_firmware_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_PFP,
+				&priv->driver_buffer[priv->driver_buffer_length++]))
+			goto smu_init_failed;
+		if (cz_smu_populate_single_firmware_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_ME,
+				&priv->driver_buffer[priv->driver_buffer_length++]))
+			goto smu_init_failed;
+		if (cz_smu_populate_single_firmware_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1,
+				&priv->driver_buffer[priv->driver_buffer_length++]))
+			goto smu_init_failed;
+		if (adev->asic_type == CHIP_STONEY) {
+			if (cz_smu_populate_single_firmware_entry(adev,
+					CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1,
+					&priv->driver_buffer[priv->driver_buffer_length++]))
+				goto smu_init_failed;
+		} else {
+			if (cz_smu_populate_single_firmware_entry(adev,
+					CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT2,
+					&priv->driver_buffer[priv->driver_buffer_length++]))
+				goto smu_init_failed;
+		}
+		if (cz_smu_populate_single_firmware_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_G,
+				&priv->driver_buffer[priv->driver_buffer_length++]))
+			goto smu_init_failed;
+	}
+
+	if (cz_smu_populate_single_scratch_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SCRATCH,
+				UCODE_ID_RLC_SCRATCH_SIZE_BYTE,
+				&priv->scratch_buffer[priv->scratch_buffer_length++]))
+		goto smu_init_failed;
+	if (cz_smu_populate_single_scratch_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_ARAM,
+				UCODE_ID_RLC_SRM_ARAM_SIZE_BYTE,
+				&priv->scratch_buffer[priv->scratch_buffer_length++]))
+		goto smu_init_failed;
+	if (cz_smu_populate_single_scratch_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_DRAM,
+				UCODE_ID_RLC_SRM_DRAM_SIZE_BYTE,
+				&priv->scratch_buffer[priv->scratch_buffer_length++]))
+		goto smu_init_failed;
+	if (cz_smu_populate_single_scratch_entry(adev,
+				CZ_SCRATCH_ENTRY_UCODE_ID_POWER_PROFILING,
+				sizeof(struct SMU8_MultimediaPowerLogData),
+				&priv->scratch_buffer[priv->scratch_buffer_length++]))
+		goto smu_init_failed;
+	if (cz_smu_populate_single_scratch_entry(adev,
+				CZ_SCRATCH_ENTRY_SMU8_FUSION_CLKTABLE,
+				sizeof(struct SMU8_Fusion_ClkTable),
+				&priv->scratch_buffer[priv->scratch_buffer_length++]))
+		goto smu_init_failed;
+
+	cz_smu_initialize_toc_empty_job_list(adev);
+	cz_smu_construct_toc_for_rlc_aram_save(adev);
+	cz_smu_construct_toc_for_vddgfx_enter(adev);
+	cz_smu_construct_toc_for_vddgfx_exit(adev);
+	cz_smu_construct_toc_for_power_profiling(adev);
+	cz_smu_construct_toc_for_bootup(adev);
+	cz_smu_construct_toc_for_clock_table(adev);
+	/* init the smumgr functions */
+	adev->smu.smumgr_funcs = &cz_smumgr_funcs;
+
+	return 0;
+
+smu_init_failed:
+	amdgpu_bo_unref(toc_buf);
+	amdgpu_bo_unref(smu_buf);
+
+	return ret;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/cz_smumgr.h b/kernel/drivers/gpu/drm/amd/amdgpu/cz_smumgr.h
new file mode 100644
index 000000000..924d355b4
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/cz_smumgr.h
@@ -0,0 +1,94 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __CZ_SMC_H__
+#define __CZ_SMC_H__
+
+#define MAX_NUM_FIRMWARE                        8
+#define MAX_NUM_SCRATCH                         11
+#define CZ_SCRATCH_SIZE_NONGFX_CLOCKGATING      1024
+#define CZ_SCRATCH_SIZE_NONGFX_GOLDENSETTING    2048
+#define CZ_SCRATCH_SIZE_SDMA_METADATA           1024
+#define CZ_SCRATCH_SIZE_IH                      ((2*256+1)*4)
+
+enum cz_scratch_entry {
+	CZ_SCRATCH_ENTRY_UCODE_ID_SDMA0 = 0,
+	CZ_SCRATCH_ENTRY_UCODE_ID_SDMA1,
+	CZ_SCRATCH_ENTRY_UCODE_ID_CP_CE,
+	CZ_SCRATCH_ENTRY_UCODE_ID_CP_PFP,
+	CZ_SCRATCH_ENTRY_UCODE_ID_CP_ME,
+	CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT1,
+	CZ_SCRATCH_ENTRY_UCODE_ID_CP_MEC_JT2,
+	CZ_SCRATCH_ENTRY_UCODE_ID_GMCON_RENG,
+	CZ_SCRATCH_ENTRY_UCODE_ID_RLC_G,
+	CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SCRATCH,
+	CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_ARAM,
+	CZ_SCRATCH_ENTRY_UCODE_ID_RLC_SRM_DRAM,
+	CZ_SCRATCH_ENTRY_UCODE_ID_DMCU_ERAM,
+	CZ_SCRATCH_ENTRY_UCODE_ID_DMCU_IRAM,
+	CZ_SCRATCH_ENTRY_UCODE_ID_POWER_PROFILING,
+	CZ_SCRATCH_ENTRY_DATA_ID_SDMA_HALT,
+	CZ_SCRATCH_ENTRY_DATA_ID_SYS_CLOCKGATING,
+	CZ_SCRATCH_ENTRY_DATA_ID_SDMA_RING_REGS,
+	CZ_SCRATCH_ENTRY_DATA_ID_NONGFX_REINIT,
+	CZ_SCRATCH_ENTRY_DATA_ID_SDMA_START,
+	CZ_SCRATCH_ENTRY_DATA_ID_IH_REGISTERS,
+	CZ_SCRATCH_ENTRY_SMU8_FUSION_CLKTABLE
+};
+
+struct cz_buffer_entry {
+	uint32_t	data_size;
+	uint32_t	mc_addr_low;
+	uint32_t	mc_addr_high;
+	void		*kaddr;
+	enum cz_scratch_entry firmware_ID;
+};
+
+struct cz_register_index_data_pair {
+	uint32_t	offset;
+	uint32_t	value;
+};
+
+struct cz_ih_meta_data {
+	uint32_t	command;
+	struct cz_register_index_data_pair register_index_value_pair[1];
+};
+
+struct cz_smu_private_data {
+	uint8_t		driver_buffer_length;
+	uint8_t		scratch_buffer_length;
+	uint16_t	toc_entry_used_count;
+	uint16_t 	toc_entry_initialize_index;
+	uint16_t	toc_entry_power_profiling_index;
+	uint16_t	toc_entry_aram;
+	uint16_t	toc_entry_ih_register_restore_task_index;
+	uint16_t	toc_entry_clock_table;
+	uint16_t	ih_register_restore_task_size;
+	uint16_t	smu_buffer_used_bytes;
+
+	struct cz_buffer_entry  toc_buffer;
+	struct cz_buffer_entry  smu_buffer;
+	struct cz_buffer_entry  driver_buffer[MAX_NUM_FIRMWARE];
+	struct cz_buffer_entry  scratch_buffer[MAX_NUM_SCRATCH];
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
new file mode 100644
index 000000000..4dcc8fba5
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -0,0 +1,3831 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "vid.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+#include "dce/dce_10_0_enum.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[] =
+{
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
+	CRTC6_REGISTER_OFFSET
+};
+
+static const u32 hpd_offsets[] =
+{
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
+};
+
+static const uint32_t dig_offsets[] = {
+	DIG0_REGISTER_OFFSET,
+	DIG1_REGISTER_OFFSET,
+	DIG2_REGISTER_OFFSET,
+	DIG3_REGISTER_OFFSET,
+	DIG4_REGISTER_OFFSET,
+	DIG5_REGISTER_OFFSET,
+	DIG6_REGISTER_OFFSET
+};
+
+static const struct {
+	uint32_t        reg;
+	uint32_t        vblank;
+	uint32_t        vline;
+	uint32_t        hpd;
+
+} interrupt_status_offsets[] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x1f311fff, 0x12300000,
+	mmHDMI_CONTROL, 0x31000111, 0x00000011,
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmDCI_CLK_CNTL, 0x00000080, 0x00000000,
+	mmFBC_DEBUG_COMP, 0x000000f0, 0x00000070,
+	mmFBC_MISC, 0x1f311fff, 0x12300000,
+	mmHDMI_CONTROL, 0x31000111, 0x00000011,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+};
+
+static void dce_v10_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_program_register_sequence(adev,
+						 fiji_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_fiji_a10,
+						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		break;
+	case CHIP_TONGA:
+		amdgpu_program_register_sequence(adev,
+						 tonga_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_tonga_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static u32 dce_v10_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v10_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static bool dce_v10_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
+{
+	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) &
+			CRTC_V_BLANK_START_END__CRTC_V_BLANK_START_MASK)
+		return true;
+	else
+		return false;
+}
+
+static bool dce_v10_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
+{
+	u32 pos1, pos2;
+
+	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	if (pos1 != pos2)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_vblank_wait - vblank wait asic callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc: crtc to wait for vblank on
+ *
+ * Wait for vblank on the requested crtc (evergreen+).
+ */
+static void dce_v10_0_vblank_wait(struct amdgpu_device *adev, int crtc)
+{
+	unsigned i = 0;
+
+	if (crtc >= adev->mode_info.num_crtc)
+		return;
+
+	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
+		return;
+
+	/* depending on when we hit vblank, we may be close to active; if so,
+	 * wait for another frame.
+	 */
+	while (dce_v10_0_is_in_vblank(adev, crtc)) {
+		if (i++ % 100 == 0) {
+			if (!dce_v10_0_is_counter_moving(adev, crtc))
+				break;
+		}
+	}
+
+	while (!dce_v10_0_is_in_vblank(adev, crtc)) {
+		if (i++ % 100 == 0) {
+			if (!dce_v10_0_is_counter_moving(adev, crtc))
+				break;
+		}
+	}
+}
+
+static u32 dce_v10_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v10_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v10_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v10_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Triggers the actual pageflip by updating the primary
+ * surface base address.
+ */
+static void dce_v10_0_page_flip(struct amdgpu_device *adev,
+			      int crtc_id, u64 crtc_base)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	/* update the primary scanout address */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(crtc_base));
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v10_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+}
+
+/**
+ * dce_v10_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v10_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	int idx;
+	bool connected = false;
+
+	switch (hpd) {
+	case AMDGPU_HPD_1:
+		idx = 0;
+		break;
+	case AMDGPU_HPD_2:
+		idx = 1;
+		break;
+	case AMDGPU_HPD_3:
+		idx = 2;
+		break;
+	case AMDGPU_HPD_4:
+		idx = 3;
+		break;
+	case AMDGPU_HPD_5:
+		idx = 4;
+		break;
+	case AMDGPU_HPD_6:
+		idx = 5;
+		break;
+	default:
+		return connected;
+	}
+
+	if (RREG32(mmDC_HPD_INT_STATUS + hpd_offsets[idx]) &
+	    DC_HPD_INT_STATUS__DC_HPD_SENSE_MASK)
+		connected = true;
+
+	return connected;
+}
+
+/**
+ * dce_v10_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v10_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v10_0_hpd_sense(adev, hpd);
+	int idx;
+
+	switch (hpd) {
+	case AMDGPU_HPD_1:
+		idx = 0;
+		break;
+	case AMDGPU_HPD_2:
+		idx = 1;
+		break;
+	case AMDGPU_HPD_3:
+		idx = 2;
+		break;
+	case AMDGPU_HPD_4:
+		idx = 3;
+		break;
+	case AMDGPU_HPD_5:
+		idx = 4;
+		break;
+	case AMDGPU_HPD_6:
+		idx = 5;
+		break;
+	default:
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[idx]);
+	if (connected)
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[idx], tmp);
+}
+
+/**
+ * dce_v10_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+	int idx;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			continue;
+		}
+
+		switch (amdgpu_connector->hpd.hpd) {
+		case AMDGPU_HPD_1:
+			idx = 0;
+			break;
+		case AMDGPU_HPD_2:
+			idx = 1;
+			break;
+		case AMDGPU_HPD_3:
+			idx = 2;
+			break;
+		case AMDGPU_HPD_4:
+			idx = 3;
+			break;
+		case AMDGPU_HPD_5:
+			idx = 4;
+			break;
+		case AMDGPU_HPD_6:
+			idx = 5;
+			break;
+		default:
+			continue;
+		}
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[idx]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 1);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[idx], tmp);
+
+		tmp = RREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[idx]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_CONNECT_INT_DELAY,
+				    AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_DISCONNECT_INT_DELAY,
+				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
+		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[idx], tmp);
+
+		dce_v10_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq,
+			       amdgpu_connector->hpd.hpd);
+	}
+}
+
+/**
+ * dce_v10_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v10_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+	int idx;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		switch (amdgpu_connector->hpd.hpd) {
+		case AMDGPU_HPD_1:
+			idx = 0;
+			break;
+		case AMDGPU_HPD_2:
+			idx = 1;
+			break;
+		case AMDGPU_HPD_3:
+			idx = 2;
+			break;
+		case AMDGPU_HPD_4:
+			idx = 3;
+			break;
+		case AMDGPU_HPD_5:
+			idx = 4;
+			break;
+		case AMDGPU_HPD_6:
+			idx = 5;
+			break;
+		default:
+			continue;
+		}
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[idx]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 0);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[idx], tmp);
+
+		amdgpu_irq_put(adev, &adev->hpd_irq,
+			       amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v10_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static bool dce_v10_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+		if (REG_GET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN)) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
+static void dce_v10_0_stop_mc_access(struct amdgpu_device *adev,
+				     struct amdgpu_mode_mc_save *save)
+{
+	u32 crtc_enabled, tmp;
+	int i;
+
+	save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+
+	/* blank the display controllers */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
+					     CRTC_CONTROL, CRTC_MASTER_EN);
+		if (crtc_enabled) {
+#if 0
+			u32 frame_count;
+			int j;
+
+			save->crtc_enabled[i] = true;
+			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
+				amdgpu_display_vblank_wait(adev, i);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
+				WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+			/* wait for the next frame */
+			frame_count = amdgpu_display_vblank_get_counter(adev, i);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
+					break;
+				udelay(1);
+			}
+			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK) == 0) {
+				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 1);
+				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK) == 0) {
+				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 1);
+				WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
+			}
+#else
+			/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+			tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+			tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
+			WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			save->crtc_enabled[i] = false;
+			/* ***** */
+#endif
+		} else {
+			save->crtc_enabled[i] = false;
+		}
+	}
+}
+
+static void dce_v10_0_resume_mc_access(struct amdgpu_device *adev,
+				       struct amdgpu_mode_mc_save *save)
+{
+	u32 tmp, frame_count;
+	int i, j;
+
+	/* update crtc base addresses */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
+		       upper_32_bits(adev->mc.vram_start));
+		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
+		       upper_32_bits(adev->mc.vram_start));
+		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
+		       (u32)adev->mc.vram_start);
+		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i],
+		       (u32)adev->mc.vram_start);
+
+		if (save->crtc_enabled[i]) {
+			tmp = RREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 3) {
+				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 3);
+				WREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK)) {
+				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 0);
+				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK)) {
+				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 0);
+				WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
+			}
+			for (j = 0; j < adev->usec_timeout; j++) {
+				tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+				if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING) == 0)
+					break;
+				udelay(1);
+			}
+			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
+			tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+			WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			/* wait for the next frame */
+			frame_count = amdgpu_display_vblank_get_counter(adev, i);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
+					break;
+				udelay(1);
+			}
+		}
+	}
+
+	WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
+	WREG32(mmVGA_MEMORY_BASE_ADDRESS, lower_32_bits(adev->mc.vram_start));
+
+	/* Unlock vga access */
+	WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
+	mdelay(1);
+	WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
+}
+
+static void dce_v10_0_set_vga_render_state(struct amdgpu_device *adev,
+					   bool render)
+{
+	u32 tmp;
+
+	/* Lockout access through VGA aperture*/
+	tmp = RREG32(mmVGA_HDP_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+	WREG32(mmVGA_HDP_CONTROL, tmp);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+}
+
+static void dce_v10_0_program_fmt(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS/eDP FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	/* not needed for analog */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
+		return;
+
+	if (bpc == 0)
+		return;
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 0);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 0);
+		}
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 1);
+		}
+		break;
+	case 10:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 2);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 2);
+		}
+		break;
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+
+/* display watermark setup */
+/**
+ * dce_v10_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
+static u32 dce_v10_0_line_buffer_adjust(struct amdgpu_device *adev,
+				       struct amdgpu_crtc *amdgpu_crtc,
+				       struct drm_display_mode *mode)
+{
+	u32 tmp, buffer_alloc, i, mem_cfg;
+	u32 pipe_offset = amdgpu_crtc->crtc_id;
+	/*
+	 * Line Buffer Setup
+	 * There are 6 line buffers, one for each display controllers.
+	 * There are 3 partitions per LB. Select the number of partitions
+	 * to enable based on the display width.  For display widths larger
+	 * than 4096, you need use to use 2 display controllers and combine
+	 * them using the stereo blender.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (mode->crtc_hdisplay < 1920) {
+			mem_cfg = 1;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 2560) {
+			mem_cfg = 2;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 4096) {
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		} else {
+			DRM_DEBUG_KMS("Mode too big for LB!\n");
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		}
+	} else {
+		mem_cfg = 1;
+		buffer_alloc = 0;
+	}
+
+	tmp = RREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mem_cfg);
+	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+	tmp = REG_SET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, buffer_alloc);
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+		if (REG_GET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED))
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (mem_cfg) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 1:
+			return 1920 * 2;
+		case 2:
+			return 2560 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce10_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v10_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_dram_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v10_0_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_data_return_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_dmif_request_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_available_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v10_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v10_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v10_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v10_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v10_0_average_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v10_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v10_0_latency_watermark(struct dce10_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v10_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+
+	b.full = dfixed_const(mc_latency + 512);
+	c.full = dfixed_const(wm->disp_clk);
+	b.full = dfixed_div(b, c);
+
+	c.full = dfixed_const(dmif_size);
+	b.full = dfixed_div(c, b);
+
+	tmp = min(dfixed_trunc(a), dfixed_trunc(b));
+
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(wm->disp_clk);
+	b.full = dfixed_div(c, b);
+	c.full = dfixed_const(wm->bytes_per_pixel);
+	b.full = dfixed_mul(b, c);
+
+	lb_fill_bw = min(tmp, dfixed_trunc(b));
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	if (dce_v10_0_average_bandwidth(wm) <=
+	    (dce_v10_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v10_0_average_bandwidth_vs_available_bandwidth(struct dce10_wm_params *wm)
+{
+	if (dce_v10_0_average_bandwidth(wm) <=
+	    (dce_v10_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v10_0_check_latency_hiding(struct dce10_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v10_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v10_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce10_wm_params wm_low, wm_high;
+	u32 pixel_period;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		pixel_period = 1000000 / (u32)mode->clock;
+		line_time = min((u32)mode->crtc_htotal * pixel_period, (u32)65535);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = mode->crtc_hdisplay * pixel_period;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_high.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v10_0_latency_watermark(&wm_high), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v10_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v10_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+
+		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = mode->crtc_hdisplay * pixel_period;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_low.num_heads = num_heads;
+
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v10_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v10_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v10_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v10_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 1);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_a);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* select wm B */
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 2);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_b);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* restore original selection */
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/**
+ * dce_v10_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v10_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	amdgpu_update_display_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		mode = &adev->mode_info.crtcs[i]->base.mode;
+		lb_size = dce_v10_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
+		dce_v10_0_program_watermarks(adev, adev->mode_info.crtcs[i],
+					    lb_size, num_heads);
+	}
+}
+
+static void dce_v10_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 offset, tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		offset = adev->mode_info.audio.pin[i].offset;
+		tmp = RREG32_AUDIO_ENDPT(offset,
+					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (((tmp &
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+}
+
+static struct amdgpu_audio_pin *dce_v10_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v10_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v10_0_afmt_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	tmp = RREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, dig->afmt->pin->id);
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset, tmp);
+}
+
+static void dce_v10_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	int interlace = 0;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		interlace = 1;
+	if (connector->latency_present[interlace]) {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, connector->video_latency[interlace]);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, connector->audio_latency[interlace]);
+	} else {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, 0);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, 0);
+	}
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v10_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	u8 *sadb = NULL;
+	int sad_count;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+				 ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    DP_CONNECTION, 0);
+	/* set HDMI mode */
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    HDMI_CONNECTION, 1);
+	if (sad_count)
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, sadb[0]);
+	else
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, 5); /* stereo */
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v10_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+	BUG_ON(!sads);
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 tmp = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    MAX_CHANNELS, sad->channels);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    DESCRIPTOR_BYTE_2, sad->byte2);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    SUPPORTED_FREQUENCIES, sad->freq);
+					max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+				    SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
+		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+	}
+
+	kfree(sads);
+}
+
+static void dce_v10_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[] =
+{
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
+};
+
+static int dce_v10_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	adev->mode_info.audio.num_pins = 7;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v10_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+/*
+ * update the N and CTS parameters for a given pixel clock rate
+ */
+static void dce_v10_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
+	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
+	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
+	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
+	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
+	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
+	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
+
+}
+
+/*
+ * build a HDMI Video Info Frame
+ */
+static void dce_v10_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
+					       void *buffer, size_t size)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint8_t *frame = buffer + 3;
+	uint8_t *header = buffer;
+
+	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
+		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
+		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
+		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
+		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
+}
+
+static void dce_v10_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	u32 dto_phase = 24 * 1000;
+	u32 dto_modulo = clock;
+	u32 tmp;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* XXX two dtos; generally use dto0 for hdmi */
+	/* Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
+	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL,
+			    amdgpu_crtc->crtc_id);
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
+	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
+	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
+}
+
+/*
+ * update the info frames with the data from the current display mode
+ */
+static void dce_v10_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	struct hdmi_avi_infoframe frame;
+	ssize_t err;
+	u32 tmp;
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (!dig->afmt->enabled)
+		return;
+
+	/* hdmi deep color mode general control packets setup, if bpc > 8 */
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio prior to setting up hw */
+	dig->afmt->pin = dce_v10_0_audio_get_pin(adev);
+	dce_v10_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v10_0_audio_set_dto(encoder, mode->clock);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp); /* send null packets when required */
+
+	WREG32(mmAFMT_AUDIO_CRC_CONTROL + dig->afmt->offset, 0x1000);
+
+	tmp = RREG32(mmHDMI_CONTROL + dig->afmt->offset);
+	switch (bpc) {
+	case 0:
+	case 6:
+	case 8:
+	case 16:
+	default:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
+			  connector->name, bpc);
+		break;
+	case 10:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 1);
+		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
+			  connector->name);
+		break;
+	case 12:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 2);
+		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
+			  connector->name);
+		break;
+	}
+	WREG32(mmHDMI_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1); /* send null packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1); /* send general control packets */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1); /* send general control packets every frame */
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable audio info frames (frames won't be set until audio is enabled) */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	/* anything other than 0 */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	WREG32(mmHDMI_GC + dig->afmt->offset, 0); /* unset HDMI_GC_AVMUTE */
+
+	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* set the default audio delay */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
+	/* should be suffient for all audio modes and small enough for all hblanks */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* allow 60958 channel status fields to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
+	if (bpc > 8)
+		/* clear SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 0);
+	else
+		/* select SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 1);
+	/* allow hw to sent ACR packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
+	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	dce_v10_0_afmt_update_ACR(encoder, mode->clock);
+
+	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
+	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
+
+	dce_v10_0_audio_write_speaker_allocation(encoder);
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset,
+	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
+
+	dce_v10_0_afmt_audio_select_pin(encoder);
+	dce_v10_0_audio_write_sad_regs(encoder);
+	dce_v10_0_audio_write_latency_fields(encoder, mode);
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	dce_v10_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable AVI info frames */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* send audio packets */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	WREG32(mmAFMT_RAMP_CONTROL0 + dig->afmt->offset, 0x00FFFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL1 + dig->afmt->offset, 0x007FFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL2 + dig->afmt->offset, 0x00000001);
+	WREG32(mmAFMT_RAMP_CONTROL3 + dig->afmt->offset, 0x00000001);
+
+	/* enable audio after to setting up hw */
+	dce_v10_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v10_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v10_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static void dce_v10_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE10 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		}
+	}
+}
+
+static void dce_v10_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v10_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	if (enable)
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
+	else
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
+}
+
+static void dce_v10_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (enable)
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
+	else
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
+}
+
+static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_framebuffer *amdgpu_fb;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *rbo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels;
+	u32 fb_swap = REG_SET_FIELD(0, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP, ENDIAN_NONE);
+	u32 pipe_config;
+	u32 tmp, viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic) {
+		amdgpu_fb = to_amdgpu_framebuffer(fb);
+		target_fb = fb;
+	}
+	else {
+		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+		target_fb = crtc->primary->fb;
+	}
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = amdgpu_fb->obj;
+	rbo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (atomic)
+		fb_location = amdgpu_bo_gpu_offset(rbo);
+	else {
+		r = amdgpu_bo_pin(rbo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(rbo);
+			return -EINVAL;
+		}
+	}
+
+	amdgpu_bo_get_tiling_flags(rbo, &tiling_flags);
+	amdgpu_bo_unreserve(rbo);
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+
+	switch (target_fb->pixel_format) {
+	case DRM_FORMAT_C8:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 0);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 2);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 5);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 4);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+			drm_get_format_name(target_fb->pixel_format));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_NUM_BANKS, num_banks);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_2D_TILED_THIN1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_TILE_SPLIT,
+					  tile_split);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_WIDTH, bankw);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_HEIGHT, bankh);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT,
+					  mtaspect);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MICRO_TILE_MODE,
+					  ADDR_SURF_MICRO_TILING_DISPLAY);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_1D_TILED_THIN1);
+	}
+
+	fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_PIPE_CONFIG,
+				  pipe_config);
+
+	dce_v10_0_vga_enable(crtc, false);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	tmp = RREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset);
+	if (bypass_lut)
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 0);
+	WREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset, tmp);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / (target_fb->bits_per_pixel / 8);
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v10_0_grph_enable(crtc, true);
+
+	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+	       target_fb->height);
+
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* pageflip setup */
+	/* make sure flip is at vb rather than hb */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
+			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, 0);
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* set pageflip to happen only at start of vblank interval (front porch) */
+	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 3);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		amdgpu_fb = to_amdgpu_framebuffer(fb);
+		rbo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		r = amdgpu_bo_reserve(rbo, false);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(rbo);
+		amdgpu_bo_unreserve(rbo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v10_0_bandwidth_update(adev);
+
+	return 0;
+}
+
+static void dce_v10_0_set_interleave(struct drm_crtc *crtc,
+				     struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	u32 tmp;
+
+	tmp = RREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset);
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 0);
+	WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v10_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int i;
+	u32 tmp;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	tmp = RREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_GRPH_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_OVL_MODE, 0);
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, PRESCALE_OVL_CONTROL, OVL_PRESCALE_BYPASS, 1);
+	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, OVL_INPUT_GAMMA_MODE, 0);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       (amdgpu_crtc->lut_r[i] << 20) |
+		       (amdgpu_crtc->lut_g[i] << 10) |
+		       (amdgpu_crtc->lut_b[i] << 0));
+	}
+
+	tmp = RREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, OVL_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, 0);
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, OVL_GAMUT_REMAP_MODE, 0);
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, GRPH_REGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, OVL_REGAMMA_MODE, 0);
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_OVL_MODE, 0);
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(mmDENORM_CONTROL + amdgpu_crtc->crtc_offset, 0);
+	/* XXX this only needs to be programmed once per crtc at startup,
+	 * not sure where the best place for it is
+	 */
+	tmp = RREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, ALPHA_CONTROL, CURSOR_ALPHA_BLND_ENA, 1);
+	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v10_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		if (dig->linkb)
+			return 1;
+		else
+			return 0;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		if (dig->linkb)
+			return 3;
+		else
+			return 2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		if (dig->linkb)
+			return 5;
+		else
+			return 4;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+		break;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v10_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ * Asic specific PLL information
+ *
+ * DCE 10.x
+ * Tonga
+ * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
+ * CI
+ * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
+ *
+ */
+static u32 dce_v10_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else {
+			/* use the same PPLL for all DP monitors */
+			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
+			if (pll != ATOM_PPLL_INVALID)
+				return pll;
+		}
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+
+	/* DCE10 has PPLL0, PPLL1, and PPLL2 */
+	pll_in_use = amdgpu_pll_get_use_mask(crtc);
+	if (!(pll_in_use & (1 << ATOM_PPLL2)))
+		return ATOM_PPLL2;
+	if (!(pll_in_use & (1 << ATOM_PPLL1)))
+		return ATOM_PPLL1;
+	if (!(pll_in_use & (1 << ATOM_PPLL0)))
+		return ATOM_PPLL0;
+	DRM_ERROR("unable to allocate a PPLL\n");
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v10_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 1);
+	else
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 0);
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v10_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v10_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v10_0_cursor_move_locked(struct drm_crtc *crtc,
+					int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	return 0;
+}
+
+static int dce_v10_0_crtc_cursor_move(struct drm_crtc *crtc,
+				      int x, int y)
+{
+	int ret;
+
+	dce_v10_0_lock_cursor(crtc, true);
+	ret = dce_v10_0_cursor_move_locked(crtc, x, y);
+	dce_v10_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				      struct drm_file *file_priv,
+				      uint32_t handle,
+				      uint32_t width,
+				      uint32_t height,
+				      int32_t hot_x,
+				      int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v10_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(crtc->dev, file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_unreference_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_unreference_unlocked(obj);
+		return ret;
+	}
+
+	amdgpu_crtc->cursor_width = width;
+	amdgpu_crtc->cursor_height = height;
+
+	dce_v10_0_lock_cursor(crtc, true);
+
+	if (hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v10_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v10_0_show_cursor(crtc);
+	dce_v10_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, false);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v10_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v10_0_lock_cursor(crtc, true);
+
+		dce_v10_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					     amdgpu_crtc->cursor_y);
+
+		dce_v10_0_show_cursor(crtc);
+
+		dce_v10_0_lock_cursor(crtc, false);
+	}
+}
+
+static void dce_v10_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t start, uint32_t size)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	int end = (start + size > 256) ? 256 : start + size, i;
+
+	/* userspace palettes are always correct as is */
+	for (i = start; i < end; i++) {
+		amdgpu_crtc->lut_r[i] = red[i] >> 6;
+		amdgpu_crtc->lut_g[i] = green[i] >> 6;
+		amdgpu_crtc->lut_b[i] = blue[i] >> 6;
+	}
+	dce_v10_0_crtc_load_lut(crtc);
+}
+
+static void dce_v10_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	destroy_workqueue(amdgpu_crtc->pflip_queue);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v10_0_crtc_funcs = {
+	.cursor_set2 = dce_v10_0_crtc_cursor_set2,
+	.cursor_move = dce_v10_0_crtc_cursor_move,
+	.gamma_set = dce_v10_0_crtc_gamma_set,
+	.set_config = amdgpu_crtc_set_config,
+	.destroy = dce_v10_0_crtc_destroy,
+	.page_flip = amdgpu_crtc_page_flip,
+};
+
+static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		dce_v10_0_vga_enable(crtc, true);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		dce_v10_0_vga_enable(crtc, false);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_vblank_post_modeset(dev, amdgpu_crtc->crtc_id);
+		dce_v10_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_vblank_pre_modeset(dev, amdgpu_crtc->crtc_id);
+		if (amdgpu_crtc->enabled) {
+			dce_v10_0_vga_enable(crtc, true);
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+			dce_v10_0_vga_enable(crtc, false);
+		}
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v10_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v10_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v10_0_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v10_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_framebuffer *amdgpu_fb;
+		struct amdgpu_bo *rbo;
+
+		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+		rbo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		r = amdgpu_bo_reserve(rbo, false);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve rbo before unpin\n");
+		else {
+			amdgpu_bo_unpin(rbo);
+			amdgpu_bo_unreserve(rbo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v10_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL0:
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+					  0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v10_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v10_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v10_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v10_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v10_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v10_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v10_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v10_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v10_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v10_0_crtc_helper_funcs = {
+	.dpms = dce_v10_0_crtc_dpms,
+	.mode_fixup = dce_v10_0_crtc_mode_fixup,
+	.mode_set = dce_v10_0_crtc_mode_set,
+	.mode_set_base = dce_v10_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v10_0_crtc_set_base_atomic,
+	.prepare = dce_v10_0_crtc_prepare,
+	.commit = dce_v10_0_crtc_commit,
+	.load_lut = dce_v10_0_crtc_load_lut,
+	.disable = dce_v10_0_crtc_disable,
+};
+
+static int dce_v10_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+	int i;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v10_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	amdgpu_crtc->pflip_queue = create_singlethread_workqueue("amdgpu-pageflip-queue");
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = 128;
+	amdgpu_crtc->max_cursor_height = 128;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	for (i = 0; i < 256; i++) {
+		amdgpu_crtc->lut_r[i] = i << 2;
+		amdgpu_crtc->lut_g[i] = i << 2;
+		amdgpu_crtc->lut_b[i] = i << 2;
+	}
+
+	switch (amdgpu_crtc->crtc_id) {
+	case 0:
+	default:
+		amdgpu_crtc->crtc_offset = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		amdgpu_crtc->crtc_offset = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		amdgpu_crtc->crtc_offset = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		amdgpu_crtc->crtc_offset = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		amdgpu_crtc->crtc_offset = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		amdgpu_crtc->crtc_offset = CRTC5_REGISTER_OFFSET;
+		break;
+	}
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v10_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v10_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v10_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v10_0_audio_endpt_wreg;
+
+	dce_v10_0_set_display_funcs(adev);
+	dce_v10_0_set_irq_funcs(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+		adev->mode_info.num_crtc = 6; /* XXX 7??? */
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 7;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, i + 1, &adev->crtc_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = 8; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, 42, &adev->hpd_irq);
+	if (r)
+		return r;
+
+	adev->mode_info.mode_config_initialized = true;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
+
+	r = amdgpu_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v10_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	dce_v10_0_afmt_init(adev);
+
+	r = dce_v10_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	return r;
+}
+
+static int dce_v10_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v10_0_audio_fini(adev);
+
+	dce_v10_0_afmt_fini(adev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v10_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v10_0_init_golden_registers(adev);
+
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_encoder_init_dig(adev);
+	amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+
+	/* initialize hpd */
+	dce_v10_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v10_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v10_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v10_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v10_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v10_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v10_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_atombios_scratch_regs_save(adev);
+
+	return dce_v10_0_hw_fini(handle);
+}
+
+static int dce_v10_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = dce_v10_0_hw_init(handle);
+
+	amdgpu_atombios_scratch_regs_restore(adev);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v10_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v10_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static void dce_v10_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "DCE 10.x registers\n");
+	/* XXX todo */
+}
+
+static int dce_v10_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (dce_v10_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		dce_v10_0_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+		dce_v10_0_print_status((void *)adev);
+	}
+	return 0;
+}
+
+static void dce_v10_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v10_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static int dce_v10_0_set_hpd_irq_state(struct amdgpu_device *adev,
+				       struct amdgpu_irq_src *source,
+				       unsigned hpd,
+				       enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 1);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_set_crtc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v10_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v10_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v10_0_set_pageflip_irq_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v10_0_pageflip_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) {
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
+						 amdgpu_crtc->pflip_status,
+						 AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if (works->event)
+		drm_send_vblank_event(adev->ddev, crtc_id, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_vblank_put(adev->ddev, amdgpu_crtc->crtc_id);
+	queue_work(amdgpu_crtc->pflip_queue, &works->unpin_work);
+
+	return 0;
+}
+
+static void dce_v10_0_hpd_int_ack(struct amdgpu_device *adev,
+				  int hpd)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+	tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_ACK, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+static void dce_v10_0_crtc_vblank_int_ack(struct amdgpu_device *adev,
+					  int crtc)
+{
+	u32 tmp;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VBLANK_STATUS, VBLANK_ACK, 1);
+	WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static void dce_v10_0_crtc_vline_int_ack(struct amdgpu_device *adev,
+					 int crtc)
+{
+	u32 tmp;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VLINE_STATUS, VLINE_ACK, 1);
+	WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static int dce_v10_0_crtc_irq(struct amdgpu_device *adev,
+			      struct amdgpu_irq_src *source,
+			      struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
+
+	switch (entry->src_data) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			dce_v10_0_crtc_vblank_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			dce_v10_0_crtc_vline_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_hpd_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask;
+	unsigned hpd;
+
+	if (entry->src_data >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
+		return 0;
+	}
+
+	hpd = entry->src_data;
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+
+	if (disp_int & mask) {
+		dce_v10_0_hpd_int_ack(adev, hpd);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+}
+
+static int dce_v10_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v10_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs dce_v10_0_ip_funcs = {
+	.early_init = dce_v10_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v10_0_sw_init,
+	.sw_fini = dce_v10_0_sw_fini,
+	.hw_init = dce_v10_0_hw_init,
+	.hw_fini = dce_v10_0_hw_fini,
+	.suspend = dce_v10_0_suspend,
+	.resume = dce_v10_0_resume,
+	.is_idle = dce_v10_0_is_idle,
+	.wait_for_idle = dce_v10_0_wait_for_idle,
+	.soft_reset = dce_v10_0_soft_reset,
+	.print_status = dce_v10_0_print_status,
+	.set_clockgating_state = dce_v10_0_set_clockgating_state,
+	.set_powergating_state = dce_v10_0_set_powergating_state,
+};
+
+static void
+dce_v10_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v10_0_set_interleave(encoder->crtc, mode);
+
+	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
+		dce_v10_0_afmt_enable(encoder, true);
+		dce_v10_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v10_0_encoder_prepare(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v10_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v10_0_program_fmt(encoder);
+}
+
+static void dce_v10_0_encoder_commit(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v10_0_encoder_disable(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
+			dce_v10_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v10_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v10_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v10_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v10_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v10_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static bool dce_v10_0_ext_mode_fixup(struct drm_encoder *encoder,
+				    const struct drm_display_mode *mode,
+				    struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static const struct drm_encoder_helper_funcs dce_v10_0_ext_helper_funcs = {
+	.dpms = dce_v10_0_ext_dpms,
+	.mode_fixup = dce_v10_0_ext_mode_fixup,
+	.prepare = dce_v10_0_ext_prepare,
+	.mode_set = dce_v10_0_ext_mode_set,
+	.commit = dce_v10_0_ext_commit,
+	.disable = dce_v10_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v10_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v10_0_encoder_prepare,
+	.mode_set = dce_v10_0_encoder_mode_set,
+	.commit = dce_v10_0_encoder_commit,
+	.disable = dce_v10_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v10_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v10_0_encoder_prepare,
+	.mode_set = dce_v10_0_encoder_mode_set,
+	.commit = dce_v10_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v10_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v10_0_encoder_funcs = {
+	.destroy = dce_v10_0_encoder_destroy,
+};
+
+static void dce_v10_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC);
+		drm_encoder_helper_add(encoder, &dce_v10_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v10_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC);
+		else
+			drm_encoder_init(dev, encoder, &dce_v10_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS);
+		drm_encoder_helper_add(encoder, &dce_v10_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
+	.set_vga_render_state = &dce_v10_0_set_vga_render_state,
+	.bandwidth_update = &dce_v10_0_bandwidth_update,
+	.vblank_get_counter = &dce_v10_0_vblank_get_counter,
+	.vblank_wait = &dce_v10_0_vblank_wait,
+	.is_display_hung = &dce_v10_0_is_display_hung,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v10_0_hpd_sense,
+	.hpd_set_polarity = &dce_v10_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v10_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v10_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v10_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v10_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+	.stop_mc_access = &dce_v10_0_stop_mc_access,
+	.resume_mc_access = &dce_v10_0_resume_mc_access,
+};
+
+static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v10_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v10_0_crtc_irq_funcs = {
+	.set = dce_v10_0_set_crtc_irq_state,
+	.process = dce_v10_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v10_0_pageflip_irq_funcs = {
+	.set = dce_v10_0_set_pageflip_irq_state,
+	.process = dce_v10_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v10_0_hpd_irq_funcs = {
+	.set = dce_v10_0_set_hpd_irq_state,
+	.process = dce_v10_0_hpd_irq,
+};
+
+static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
+	adev->crtc_irq.funcs = &dce_v10_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
+	adev->pageflip_irq.funcs = &dce_v10_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
+	adev->hpd_irq.funcs = &dce_v10_0_hpd_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h
new file mode 100644
index 000000000..1bfa48ddd
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V10_0_H__
+#define __DCE_V10_0_H__
+
+extern const struct amd_ip_funcs dce_v10_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
new file mode 100644
index 000000000..8f1e51128
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -0,0 +1,3824 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "vid.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+
+#include "dce/dce_11_0_d.h"
+#include "dce/dce_11_0_sh_mask.h"
+#include "dce/dce_11_0_enum.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[] =
+{
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
+	CRTC6_REGISTER_OFFSET
+};
+
+static const u32 hpd_offsets[] =
+{
+	HPD0_REGISTER_OFFSET,
+	HPD1_REGISTER_OFFSET,
+	HPD2_REGISTER_OFFSET,
+	HPD3_REGISTER_OFFSET,
+	HPD4_REGISTER_OFFSET,
+	HPD5_REGISTER_OFFSET
+};
+
+static const uint32_t dig_offsets[] = {
+	DIG0_REGISTER_OFFSET,
+	DIG1_REGISTER_OFFSET,
+	DIG2_REGISTER_OFFSET,
+	DIG3_REGISTER_OFFSET,
+	DIG4_REGISTER_OFFSET,
+	DIG5_REGISTER_OFFSET,
+	DIG6_REGISTER_OFFSET,
+	DIG7_REGISTER_OFFSET,
+	DIG8_REGISTER_OFFSET
+};
+
+static const struct {
+	uint32_t        reg;
+	uint32_t        vblank;
+	uint32_t        vline;
+	uint32_t        hpd;
+
+} interrupt_status_offsets[] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static const u32 cz_golden_settings_a11[] =
+{
+	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
+	mmFBC_MISC, 0x1f311fff, 0x14300000,
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmXDMA_CLOCK_GATING_CNTL, 0xffffffff, 0x00000100,
+	mmXDMA_MEM_POWER_CNTL, 0x00000101, 0x00000000,
+};
+
+static const u32 stoney_golden_settings_a11[] =
+{
+	mmCRTC_DOUBLE_BUFFER_CONTROL, 0x00010101, 0x00010000,
+	mmFBC_MISC, 0x1f311fff, 0x14302000,
+};
+
+
+static void dce_v11_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+		amdgpu_program_register_sequence(adev,
+						 cz_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 cz_golden_settings_a11,
+						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
+		break;
+	case CHIP_STONEY:
+		amdgpu_program_register_sequence(adev,
+						 stoney_golden_settings_a11,
+						 (const u32)ARRAY_SIZE(stoney_golden_settings_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+static u32 dce_v11_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v11_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static bool dce_v11_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
+{
+	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) &
+			CRTC_V_BLANK_START_END__CRTC_V_BLANK_START_MASK)
+		return true;
+	else
+		return false;
+}
+
+static bool dce_v11_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
+{
+	u32 pos1, pos2;
+
+	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	if (pos1 != pos2)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_vblank_wait - vblank wait asic callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc: crtc to wait for vblank on
+ *
+ * Wait for vblank on the requested crtc (evergreen+).
+ */
+static void dce_v11_0_vblank_wait(struct amdgpu_device *adev, int crtc)
+{
+	unsigned i = 0;
+
+	if (crtc >= adev->mode_info.num_crtc)
+		return;
+
+	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
+		return;
+
+	/* depending on when we hit vblank, we may be close to active; if so,
+	 * wait for another frame.
+	 */
+	while (dce_v11_0_is_in_vblank(adev, crtc)) {
+		if (i++ % 100 == 0) {
+			if (!dce_v11_0_is_counter_moving(adev, crtc))
+				break;
+		}
+	}
+
+	while (!dce_v11_0_is_in_vblank(adev, crtc)) {
+		if (i++ % 100 == 0) {
+			if (!dce_v11_0_is_counter_moving(adev, crtc))
+				break;
+		}
+	}
+}
+
+static u32 dce_v11_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v11_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v11_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v11_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Triggers the actual pageflip by updating the primary
+ * surface base address.
+ */
+static void dce_v11_0_page_flip(struct amdgpu_device *adev,
+			      int crtc_id, u64 crtc_base)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	/* update the scanout addresses */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(crtc_base));
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v11_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+}
+
+/**
+ * dce_v11_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v11_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	int idx;
+	bool connected = false;
+
+	switch (hpd) {
+	case AMDGPU_HPD_1:
+		idx = 0;
+		break;
+	case AMDGPU_HPD_2:
+		idx = 1;
+		break;
+	case AMDGPU_HPD_3:
+		idx = 2;
+		break;
+	case AMDGPU_HPD_4:
+		idx = 3;
+		break;
+	case AMDGPU_HPD_5:
+		idx = 4;
+		break;
+	case AMDGPU_HPD_6:
+		idx = 5;
+		break;
+	default:
+		return connected;
+	}
+
+	if (RREG32(mmDC_HPD_INT_STATUS + hpd_offsets[idx]) &
+	    DC_HPD_INT_STATUS__DC_HPD_SENSE_MASK)
+		connected = true;
+
+	return connected;
+}
+
+/**
+ * dce_v11_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v11_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v11_0_hpd_sense(adev, hpd);
+	int idx;
+
+	switch (hpd) {
+	case AMDGPU_HPD_1:
+		idx = 0;
+		break;
+	case AMDGPU_HPD_2:
+		idx = 1;
+		break;
+	case AMDGPU_HPD_3:
+		idx = 2;
+		break;
+	case AMDGPU_HPD_4:
+		idx = 3;
+		break;
+	case AMDGPU_HPD_5:
+		idx = 4;
+		break;
+	case AMDGPU_HPD_6:
+		idx = 5;
+		break;
+	default:
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[idx]);
+	if (connected)
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_POLARITY, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[idx], tmp);
+}
+
+/**
+ * dce_v11_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+	int idx;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			continue;
+		}
+
+		switch (amdgpu_connector->hpd.hpd) {
+		case AMDGPU_HPD_1:
+			idx = 0;
+			break;
+		case AMDGPU_HPD_2:
+			idx = 1;
+			break;
+		case AMDGPU_HPD_3:
+			idx = 2;
+			break;
+		case AMDGPU_HPD_4:
+			idx = 3;
+			break;
+		case AMDGPU_HPD_5:
+			idx = 4;
+			break;
+		case AMDGPU_HPD_6:
+			idx = 5;
+			break;
+		default:
+			continue;
+		}
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[idx]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 1);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[idx], tmp);
+
+		tmp = RREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[idx]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_CONNECT_INT_DELAY,
+				    AMDGPU_HPD_CONNECT_INT_DELAY_IN_MS);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_TOGGLE_FILT_CNTL,
+				    DC_HPD_DISCONNECT_INT_DELAY,
+				    AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
+		WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[idx], tmp);
+
+		dce_v11_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+/**
+ * dce_v11_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v11_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp;
+	int idx;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		switch (amdgpu_connector->hpd.hpd) {
+		case AMDGPU_HPD_1:
+			idx = 0;
+			break;
+		case AMDGPU_HPD_2:
+			idx = 1;
+			break;
+		case AMDGPU_HPD_3:
+			idx = 2;
+			break;
+		case AMDGPU_HPD_4:
+			idx = 3;
+			break;
+		case AMDGPU_HPD_5:
+			idx = 4;
+			break;
+		case AMDGPU_HPD_6:
+			idx = 5;
+			break;
+		default:
+			continue;
+		}
+
+		tmp = RREG32(mmDC_HPD_CONTROL + hpd_offsets[idx]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_CONTROL, DC_HPD_EN, 0);
+		WREG32(mmDC_HPD_CONTROL + hpd_offsets[idx], tmp);
+
+		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v11_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static bool dce_v11_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+		if (REG_GET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN)) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
+static void dce_v11_0_stop_mc_access(struct amdgpu_device *adev,
+				     struct amdgpu_mode_mc_save *save)
+{
+	u32 crtc_enabled, tmp;
+	int i;
+
+	save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+
+	/* blank the display controllers */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
+					     CRTC_CONTROL, CRTC_MASTER_EN);
+		if (crtc_enabled) {
+#if 0
+			u32 frame_count;
+			int j;
+
+			save->crtc_enabled[i] = true;
+			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
+				amdgpu_display_vblank_wait(adev, i);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
+				WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+			/* wait for the next frame */
+			frame_count = amdgpu_display_vblank_get_counter(adev, i);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
+					break;
+				udelay(1);
+			}
+			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK) == 0) {
+				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 1);
+				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmCRTC_MASTER_UPDATE_LOCK + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, CRTC_MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK) == 0) {
+				tmp = REG_SET_FIELD(tmp, CRTC_MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 1);
+				WREG32(mmCRTC_MASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
+			}
+#else
+			/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+			tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+			tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
+			WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			save->crtc_enabled[i] = false;
+			/* ***** */
+#endif
+		} else {
+			save->crtc_enabled[i] = false;
+		}
+	}
+}
+
+static void dce_v11_0_resume_mc_access(struct amdgpu_device *adev,
+				       struct amdgpu_mode_mc_save *save)
+{
+	u32 tmp, frame_count;
+	int i, j;
+
+	/* update crtc base addresses */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
+		       upper_32_bits(adev->mc.vram_start));
+		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
+		       upper_32_bits(adev->mc.vram_start));
+		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
+		       (u32)adev->mc.vram_start);
+		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i],
+		       (u32)adev->mc.vram_start);
+
+		if (save->crtc_enabled[i]) {
+			tmp = RREG32(mmCRTC_MASTER_UPDATE_MODE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, CRTC_MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 3) {
+				tmp = REG_SET_FIELD(tmp, CRTC_MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 3);
+				WREG32(mmCRTC_MASTER_UPDATE_MODE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK)) {
+				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 0);
+				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmCRTC_MASTER_UPDATE_LOCK + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, CRTC_MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK)) {
+				tmp = REG_SET_FIELD(tmp, CRTC_MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 0);
+				WREG32(mmCRTC_MASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
+			}
+			for (j = 0; j < adev->usec_timeout; j++) {
+				tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+				if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING) == 0)
+					break;
+				udelay(1);
+			}
+			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
+			tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+			WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			/* wait for the next frame */
+			frame_count = amdgpu_display_vblank_get_counter(adev, i);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
+					break;
+				udelay(1);
+			}
+		}
+	}
+
+	WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
+	WREG32(mmVGA_MEMORY_BASE_ADDRESS, lower_32_bits(adev->mc.vram_start));
+
+	/* Unlock vga access */
+	WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
+	mdelay(1);
+	WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
+}
+
+static void dce_v11_0_set_vga_render_state(struct amdgpu_device *adev,
+					   bool render)
+{
+	u32 tmp;
+
+	/* Lockout access through VGA aperture*/
+	tmp = RREG32(mmVGA_HDP_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+	WREG32(mmVGA_HDP_CONTROL, tmp);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+}
+
+static void dce_v11_0_program_fmt(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS/eDP FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	/* not needed for analog */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
+		return;
+
+	if (bpc == 0)
+		return;
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 0);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 0);
+		}
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 1);
+		}
+		break;
+	case 10:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE) {
+			/* XXX sort out optimal dither settings */
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_FRAME_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_HIGHPASS_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_RGB_RANDOM_ENABLE, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_SPATIAL_DITHER_DEPTH, 2);
+		} else {
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, 1);
+			tmp = REG_SET_FIELD(tmp, FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_DEPTH, 2);
+		}
+		break;
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+
+/* display watermark setup */
+/**
+ * dce_v11_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
+static u32 dce_v11_0_line_buffer_adjust(struct amdgpu_device *adev,
+				       struct amdgpu_crtc *amdgpu_crtc,
+				       struct drm_display_mode *mode)
+{
+	u32 tmp, buffer_alloc, i, mem_cfg;
+	u32 pipe_offset = amdgpu_crtc->crtc_id;
+	/*
+	 * Line Buffer Setup
+	 * There are 6 line buffers, one for each display controllers.
+	 * There are 3 partitions per LB. Select the number of partitions
+	 * to enable based on the display width.  For display widths larger
+	 * than 4096, you need use to use 2 display controllers and combine
+	 * them using the stereo blender.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (mode->crtc_hdisplay < 1920) {
+			mem_cfg = 1;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 2560) {
+			mem_cfg = 2;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 4096) {
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		} else {
+			DRM_DEBUG_KMS("Mode too big for LB!\n");
+			mem_cfg = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		}
+	} else {
+		mem_cfg = 1;
+		buffer_alloc = 0;
+	}
+
+	tmp = RREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, LB_MEMORY_CTRL, LB_MEMORY_CONFIG, mem_cfg);
+	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+	tmp = REG_SET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATED, buffer_alloc);
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset, tmp);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset);
+		if (REG_GET_FIELD(tmp, PIPE0_DMIF_BUFFER_CONTROL, DMIF_BUFFERS_ALLOCATION_COMPLETED))
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (mem_cfg) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 1:
+			return 1920 * 2;
+		case 2:
+			return 2560 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce10_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v11_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_dram_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v11_0_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_data_return_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_dmif_request_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_available_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v11_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v11_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v11_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v11_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v11_0_average_bandwidth(struct dce10_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v11_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v11_0_latency_watermark(struct dce10_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v11_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+
+	b.full = dfixed_const(mc_latency + 512);
+	c.full = dfixed_const(wm->disp_clk);
+	b.full = dfixed_div(b, c);
+
+	c.full = dfixed_const(dmif_size);
+	b.full = dfixed_div(c, b);
+
+	tmp = min(dfixed_trunc(a), dfixed_trunc(b));
+
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(wm->disp_clk);
+	b.full = dfixed_div(c, b);
+	c.full = dfixed_const(wm->bytes_per_pixel);
+	b.full = dfixed_mul(b, c);
+
+	lb_fill_bw = min(tmp, dfixed_trunc(b));
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce10_wm_params *wm)
+{
+	if (dce_v11_0_average_bandwidth(wm) <=
+	    (dce_v11_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v11_0_average_bandwidth_vs_available_bandwidth(struct dce10_wm_params *wm)
+{
+	if (dce_v11_0_average_bandwidth(wm) <=
+	    (dce_v11_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v11_0_check_latency_hiding(struct dce10_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v11_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v11_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce10_wm_params wm_low, wm_high;
+	u32 pixel_period;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		pixel_period = 1000000 / (u32)mode->clock;
+		line_time = min((u32)mode->crtc_htotal * pixel_period, (u32)65535);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = mode->crtc_hdisplay * pixel_period;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_high.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v11_0_latency_watermark(&wm_high), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v11_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+
+		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = mode->crtc_hdisplay * pixel_period;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_low.num_heads = num_heads;
+
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v11_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v11_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v11_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v11_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 1);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_a);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* select wm B */
+	tmp = REG_SET_FIELD(wm_mask, DPG_WATERMARK_MASK_CONTROL, URGENCY_WATERMARK_MASK, 2);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	tmp = RREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_LOW_WATERMARK, latency_watermark_b);
+	tmp = REG_SET_FIELD(tmp, DPG_PIPE_URGENCY_CONTROL, URGENCY_HIGH_WATERMARK, line_time);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	/* restore original selection */
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/**
+ * dce_v11_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v11_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	amdgpu_update_display_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		mode = &adev->mode_info.crtcs[i]->base.mode;
+		lb_size = dce_v11_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
+		dce_v11_0_program_watermarks(adev, adev->mode_info.crtcs[i],
+					    lb_size, num_heads);
+	}
+}
+
+static void dce_v11_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 offset, tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		offset = adev->mode_info.audio.pin[i].offset;
+		tmp = RREG32_AUDIO_ENDPT(offset,
+					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (((tmp &
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+}
+
+static struct amdgpu_audio_pin *dce_v11_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v11_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v11_0_afmt_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	tmp = RREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_SRC_CONTROL, AFMT_AUDIO_SRC_SELECT, dig->afmt->pin->id);
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + dig->afmt->offset, tmp);
+}
+
+static void dce_v11_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	int interlace = 0;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		interlace = 1;
+	if (connector->latency_present[interlace]) {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, connector->video_latency[interlace]);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, connector->audio_latency[interlace]);
+	} else {
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    VIDEO_LIPSYNC, 0);
+		tmp = REG_SET_FIELD(0, AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC,
+				    AUDIO_LIPSYNC, 0);
+	}
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v11_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp;
+	u8 *sadb = NULL;
+	int sad_count;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+				 ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    DP_CONNECTION, 0);
+	/* set HDMI mode */
+	tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+			    HDMI_CONNECTION, 1);
+	if (sad_count)
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, sadb[0]);
+	else
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER,
+				    SPEAKER_ALLOCATION, 5); /* stereo */
+	WREG32_AUDIO_ENDPT(dig->afmt->pin->offset,
+			   ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v11_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+	BUG_ON(!sads);
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 tmp = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    MAX_CHANNELS, sad->channels);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    DESCRIPTOR_BYTE_2, sad->byte2);
+					tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+							    SUPPORTED_FREQUENCIES, sad->freq);
+					max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		tmp = REG_SET_FIELD(tmp, AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0,
+				    SUPPORTED_FREQUENCIES_STEREO, stereo_freqs);
+		WREG32_AUDIO_ENDPT(dig->afmt->pin->offset, eld_reg_to_type[i][0], tmp);
+	}
+
+	kfree(sads);
+}
+
+static void dce_v11_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+			   enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[] =
+{
+	AUD0_REGISTER_OFFSET,
+	AUD1_REGISTER_OFFSET,
+	AUD2_REGISTER_OFFSET,
+	AUD3_REGISTER_OFFSET,
+	AUD4_REGISTER_OFFSET,
+	AUD5_REGISTER_OFFSET,
+	AUD6_REGISTER_OFFSET,
+};
+
+static int dce_v11_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	adev->mode_info.audio.num_pins = 7;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v11_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+/*
+ * update the N and CTS parameters for a given pixel clock rate
+ */
+static void dce_v11_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 tmp;
+
+	tmp = RREG32(mmHDMI_ACR_32_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_0, HDMI_ACR_CTS_32, acr.cts_32khz);
+	WREG32(mmHDMI_ACR_32_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_32_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_32_1, HDMI_ACR_N_32, acr.n_32khz);
+	WREG32(mmHDMI_ACR_32_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_44_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_0, HDMI_ACR_CTS_44, acr.cts_44_1khz);
+	WREG32(mmHDMI_ACR_44_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_44_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_44_1, HDMI_ACR_N_44, acr.n_44_1khz);
+	WREG32(mmHDMI_ACR_44_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_48_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_0, HDMI_ACR_CTS_48, acr.cts_48khz);
+	WREG32(mmHDMI_ACR_48_0 + dig->afmt->offset, tmp);
+	tmp = RREG32(mmHDMI_ACR_48_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_48_1, HDMI_ACR_N_48, acr.n_48khz);
+	WREG32(mmHDMI_ACR_48_1 + dig->afmt->offset, tmp);
+
+}
+
+/*
+ * build a HDMI Video Info Frame
+ */
+static void dce_v11_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
+					       void *buffer, size_t size)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint8_t *frame = buffer + 3;
+	uint8_t *header = buffer;
+
+	WREG32(mmAFMT_AVI_INFO0 + dig->afmt->offset,
+		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + dig->afmt->offset,
+		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + dig->afmt->offset,
+		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + dig->afmt->offset,
+		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
+}
+
+static void dce_v11_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	u32 dto_phase = 24 * 1000;
+	u32 dto_modulo = clock;
+	u32 tmp;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* XXX two dtos; generally use dto0 for hdmi */
+	/* Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	tmp = RREG32(mmDCCG_AUDIO_DTO_SOURCE);
+	tmp = REG_SET_FIELD(tmp, DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO0_SOURCE_SEL,
+			    amdgpu_crtc->crtc_id);
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, tmp);
+	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
+	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
+}
+
+/*
+ * update the info frames with the data from the current display mode
+ */
+static void dce_v11_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	struct hdmi_avi_infoframe frame;
+	ssize_t err;
+	u32 tmp;
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (!dig->afmt->enabled)
+		return;
+
+	/* hdmi deep color mode general control packets setup, if bpc > 8 */
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio prior to setting up hw */
+	dig->afmt->pin = dce_v11_0_audio_get_pin(adev);
+	dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v11_0_audio_set_dto(encoder, mode->clock);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1);
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp); /* send null packets when required */
+
+	WREG32(mmAFMT_AUDIO_CRC_CONTROL + dig->afmt->offset, 0x1000);
+
+	tmp = RREG32(mmHDMI_CONTROL + dig->afmt->offset);
+	switch (bpc) {
+	case 0:
+	case 6:
+	case 8:
+	case 16:
+	default:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 0);
+		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
+			  connector->name, bpc);
+		break;
+	case 10:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 1);
+		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
+			  connector->name);
+		break;
+	case 12:
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, HDMI_CONTROL, HDMI_DEEP_COLOR_DEPTH, 2);
+		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
+			  connector->name);
+		break;
+	}
+	WREG32(mmHDMI_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_NULL_SEND, 1); /* send null packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_SEND, 1); /* send general control packets */
+	tmp = REG_SET_FIELD(tmp, HDMI_VBI_PACKET_CONTROL, HDMI_GC_CONT, 1); /* send general control packets every frame */
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable audio info frames (frames won't be set until audio is enabled) */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AUDIO_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_INFOFRAME_CONTROL0, AFMT_AUDIO_INFO_UPDATE, 1);
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	/* anything other than 0 */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AUDIO_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	WREG32(mmHDMI_GC + dig->afmt->offset, 0); /* unset HDMI_GC_AVMUTE */
+
+	tmp = RREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* set the default audio delay */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_DELAY_EN, 1);
+	/* should be suffient for all audio modes and small enough for all hblanks */
+	tmp = REG_SET_FIELD(tmp, HDMI_AUDIO_PACKET_CONTROL, HDMI_AUDIO_PACKETS_PER_LINE, 3);
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* allow 60958 channel status fields to be updated */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_60958_CS_UPDATE, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset);
+	if (bpc > 8)
+		/* clear SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 0);
+	else
+		/* select SW CTS value */
+		tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_SOURCE, 1);
+	/* allow hw to sent ACR packets when required */
+	tmp = REG_SET_FIELD(tmp, HDMI_ACR_PACKET_CONTROL, HDMI_ACR_AUTO_SEND, 1);
+	WREG32(mmHDMI_ACR_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	dce_v11_0_afmt_update_ACR(encoder, mode->clock);
+
+	tmp = RREG32(mmAFMT_60958_0 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_0, AFMT_60958_CS_CHANNEL_NUMBER_L, 1);
+	WREG32(mmAFMT_60958_0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_1, AFMT_60958_CS_CHANNEL_NUMBER_R, 2);
+	WREG32(mmAFMT_60958_1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_60958_2 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_2, 3);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_3, 4);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_4, 5);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_5, 6);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_6, 7);
+	tmp = REG_SET_FIELD(tmp, AFMT_60958_2, AFMT_60958_CS_CHANNEL_NUMBER_7, 8);
+	WREG32(mmAFMT_60958_2 + dig->afmt->offset, tmp);
+
+	dce_v11_0_audio_write_speaker_allocation(encoder);
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + dig->afmt->offset,
+	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
+
+	dce_v11_0_afmt_audio_select_pin(encoder);
+	dce_v11_0_audio_write_sad_regs(encoder);
+	dce_v11_0_audio_write_latency_fields(encoder, mode);
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	dce_v11_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset);
+	/* enable AVI info frames */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_SEND, 1);
+	/* required for audio info values to be updated */
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL0, HDMI_AVI_INFO_CONT, 1);
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset);
+	tmp = REG_SET_FIELD(tmp, HDMI_INFOFRAME_CONTROL1, HDMI_AVI_INFO_LINE, 2);
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + dig->afmt->offset, tmp);
+
+	tmp = RREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset);
+	/* send audio packets */
+	tmp = REG_SET_FIELD(tmp, AFMT_AUDIO_PACKET_CONTROL, AFMT_AUDIO_SAMPLE_SEND, 1);
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + dig->afmt->offset, tmp);
+
+	WREG32(mmAFMT_RAMP_CONTROL0 + dig->afmt->offset, 0x00FFFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL1 + dig->afmt->offset, 0x007FFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL2 + dig->afmt->offset, 0x00000001);
+	WREG32(mmAFMT_RAMP_CONTROL3 + dig->afmt->offset, 0x00000001);
+
+	/* enable audio after to setting up hw */
+	dce_v11_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v11_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v11_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static void dce_v11_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE11 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		}
+	}
+}
+
+static void dce_v11_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v11_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	if (enable)
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
+	else
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
+}
+
+static void dce_v11_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (enable)
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
+	else
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
+}
+
+static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_framebuffer *amdgpu_fb;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *rbo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels;
+	u32 fb_swap = REG_SET_FIELD(0, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP, ENDIAN_NONE);
+	u32 pipe_config;
+	u32 tmp, viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic) {
+		amdgpu_fb = to_amdgpu_framebuffer(fb);
+		target_fb = fb;
+	}
+	else {
+		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+		target_fb = crtc->primary->fb;
+	}
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = amdgpu_fb->obj;
+	rbo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (atomic)
+		fb_location = amdgpu_bo_gpu_offset(rbo);
+	else {
+		r = amdgpu_bo_pin(rbo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(rbo);
+			return -EINVAL;
+		}
+	}
+
+	amdgpu_bo_get_tiling_flags(rbo, &tiling_flags);
+	amdgpu_bo_unreserve(rbo);
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+
+	switch (target_fb->pixel_format) {
+	case DRM_FORMAT_C8:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 0);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 2);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 5);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN16);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 0);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 1);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = REG_SET_FIELD(0, GRPH_CONTROL, GRPH_DEPTH, 2);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_FORMAT, 4);
+#ifdef __BIG_ENDIAN
+		fb_swap = REG_SET_FIELD(fb_swap, GRPH_SWAP_CNTL, GRPH_ENDIAN_SWAP,
+					ENDIAN_8IN32);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+			drm_get_format_name(target_fb->pixel_format));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_NUM_BANKS, num_banks);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_2D_TILED_THIN1);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_TILE_SPLIT,
+					  tile_split);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_WIDTH, bankw);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_BANK_HEIGHT, bankh);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MACRO_TILE_ASPECT,
+					  mtaspect);
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_MICRO_TILE_MODE,
+					  ADDR_SURF_MICRO_TILING_DISPLAY);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_ARRAY_MODE,
+					  ARRAY_1D_TILED_THIN1);
+	}
+
+	fb_format = REG_SET_FIELD(fb_format, GRPH_CONTROL, GRPH_PIPE_CONFIG,
+				  pipe_config);
+
+	dce_v11_0_vga_enable(crtc, false);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	tmp = RREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset);
+	if (bypass_lut)
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, GRPH_LUT_10BIT_BYPASS, GRPH_LUT_10BIT_BYPASS_EN, 0);
+	WREG32(mmGRPH_LUT_10BIT_BYPASS + amdgpu_crtc->crtc_offset, tmp);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / (target_fb->bits_per_pixel / 8);
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v11_0_grph_enable(crtc, true);
+
+	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+	       target_fb->height);
+
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* pageflip setup */
+	/* make sure flip is at vb rather than hb */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
+			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, 0);
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* set pageflip to happen only at start of vblank interval (front porch) */
+	WREG32(mmCRTC_MASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 3);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		amdgpu_fb = to_amdgpu_framebuffer(fb);
+		rbo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		r = amdgpu_bo_reserve(rbo, false);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(rbo);
+		amdgpu_bo_unreserve(rbo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v11_0_bandwidth_update(adev);
+
+	return 0;
+}
+
+static void dce_v11_0_set_interleave(struct drm_crtc *crtc,
+				     struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	u32 tmp;
+
+	tmp = RREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset);
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, LB_DATA_FORMAT, INTERLEAVE_EN, 0);
+	WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v11_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int i;
+	u32 tmp;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	tmp = RREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_CSC_CONTROL, INPUT_CSC_GRPH_MODE, 0);
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, PRESCALE_GRPH_CONTROL, GRPH_PRESCALE_BYPASS, 1);
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, INPUT_GAMMA_CONTROL, GRPH_INPUT_GAMMA_MODE, 0);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       (amdgpu_crtc->lut_r[i] << 20) |
+		       (amdgpu_crtc->lut_g[i] << 10) |
+		       (amdgpu_crtc->lut_b[i] << 0));
+	}
+
+	tmp = RREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, GRPH_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR_DEGAMMA_MODE, 0);
+	tmp = REG_SET_FIELD(tmp, DEGAMMA_CONTROL, CURSOR2_DEGAMMA_MODE, 0);
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, GAMUT_REMAP_CONTROL, GRPH_GAMUT_REMAP_MODE, 0);
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, REGAMMA_CONTROL, GRPH_REGAMMA_MODE, 0);
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	tmp = RREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, OUTPUT_CSC_CONTROL, OUTPUT_CSC_GRPH_MODE, 0);
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(mmDENORM_CONTROL + amdgpu_crtc->crtc_offset, 0);
+	/* XXX this only needs to be programmed once per crtc at startup,
+	 * not sure where the best place for it is
+	 */
+	tmp = RREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, ALPHA_CONTROL, CURSOR_ALPHA_BLND_ENA, 1);
+	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v11_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		if (dig->linkb)
+			return 1;
+		else
+			return 0;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		if (dig->linkb)
+			return 3;
+		else
+			return 2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		if (dig->linkb)
+			return 5;
+		else
+			return 4;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+		break;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v11_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ * Asic specific PLL information
+ *
+ * DCE 10.x
+ * Tonga
+ * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
+ * CI
+ * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
+ *
+ */
+static u32 dce_v11_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else {
+			/* use the same PPLL for all DP monitors */
+			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
+			if (pll != ATOM_PPLL_INVALID)
+				return pll;
+		}
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+
+	/* XXX need to determine what plls are available on each DCE11 part */
+	pll_in_use = amdgpu_pll_get_use_mask(crtc);
+	if (adev->asic_type == CHIP_CARRIZO || adev->asic_type == CHIP_STONEY) {
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		if (!(pll_in_use & (1 << ATOM_PPLL0)))
+			return ATOM_PPLL0;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	} else {
+		if (!(pll_in_use & (1 << ATOM_PPLL2)))
+			return ATOM_PPLL2;
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		if (!(pll_in_use & (1 << ATOM_PPLL0)))
+			return ATOM_PPLL0;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	}
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v11_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 1);
+	else
+		cur_lock = REG_SET_FIELD(cur_lock, CUR_UPDATE, CURSOR_UPDATE_LOCK, 0);
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v11_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 0);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static void dce_v11_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	u32 tmp;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	tmp = RREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_EN, 1);
+	tmp = REG_SET_FIELD(tmp, CUR_CONTROL, CURSOR_MODE, 2);
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+static int dce_v11_0_cursor_move_locked(struct drm_crtc *crtc,
+					int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	return 0;
+}
+
+static int dce_v11_0_crtc_cursor_move(struct drm_crtc *crtc,
+				      int x, int y)
+{
+	int ret;
+
+	dce_v11_0_lock_cursor(crtc, true);
+	ret = dce_v11_0_cursor_move_locked(crtc, x, y);
+	dce_v11_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				      struct drm_file *file_priv,
+				      uint32_t handle,
+				      uint32_t width,
+				      uint32_t height,
+				      int32_t hot_x,
+				      int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v11_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(crtc->dev, file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_unreference_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_unreference_unlocked(obj);
+		return ret;
+	}
+
+	amdgpu_crtc->cursor_width = width;
+	amdgpu_crtc->cursor_height = height;
+
+	dce_v11_0_lock_cursor(crtc, true);
+
+	if (hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v11_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v11_0_show_cursor(crtc);
+	dce_v11_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, false);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v11_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v11_0_lock_cursor(crtc, true);
+
+		dce_v11_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					     amdgpu_crtc->cursor_y);
+
+		dce_v11_0_show_cursor(crtc);
+
+		dce_v11_0_lock_cursor(crtc, false);
+	}
+}
+
+static void dce_v11_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t start, uint32_t size)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	int end = (start + size > 256) ? 256 : start + size, i;
+
+	/* userspace palettes are always correct as is */
+	for (i = start; i < end; i++) {
+		amdgpu_crtc->lut_r[i] = red[i] >> 6;
+		amdgpu_crtc->lut_g[i] = green[i] >> 6;
+		amdgpu_crtc->lut_b[i] = blue[i] >> 6;
+	}
+	dce_v11_0_crtc_load_lut(crtc);
+}
+
+static void dce_v11_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	destroy_workqueue(amdgpu_crtc->pflip_queue);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v11_0_crtc_funcs = {
+	.cursor_set2 = dce_v11_0_crtc_cursor_set2,
+	.cursor_move = dce_v11_0_crtc_cursor_move,
+	.gamma_set = dce_v11_0_crtc_gamma_set,
+	.set_config = amdgpu_crtc_set_config,
+	.destroy = dce_v11_0_crtc_destroy,
+	.page_flip = amdgpu_crtc_page_flip,
+};
+
+static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		dce_v11_0_vga_enable(crtc, true);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		dce_v11_0_vga_enable(crtc, false);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_vblank_post_modeset(dev, amdgpu_crtc->crtc_id);
+		dce_v11_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_vblank_pre_modeset(dev, amdgpu_crtc->crtc_id);
+		if (amdgpu_crtc->enabled) {
+			dce_v11_0_vga_enable(crtc, true);
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+			dce_v11_0_vga_enable(crtc, false);
+		}
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v11_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v11_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v11_0_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v11_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_framebuffer *amdgpu_fb;
+		struct amdgpu_bo *rbo;
+
+		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+		rbo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		r = amdgpu_bo_reserve(rbo, false);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve rbo before unpin\n");
+		else {
+			amdgpu_bo_unpin(rbo);
+			amdgpu_bo_unreserve(rbo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v11_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL0:
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+					  0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v11_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v11_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v11_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v11_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v11_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v11_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v11_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v11_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v11_0_crtc_helper_funcs = {
+	.dpms = dce_v11_0_crtc_dpms,
+	.mode_fixup = dce_v11_0_crtc_mode_fixup,
+	.mode_set = dce_v11_0_crtc_mode_set,
+	.mode_set_base = dce_v11_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v11_0_crtc_set_base_atomic,
+	.prepare = dce_v11_0_crtc_prepare,
+	.commit = dce_v11_0_crtc_commit,
+	.load_lut = dce_v11_0_crtc_load_lut,
+	.disable = dce_v11_0_crtc_disable,
+};
+
+static int dce_v11_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+	int i;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v11_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	amdgpu_crtc->pflip_queue = create_singlethread_workqueue("amdgpu-pageflip-queue");
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = 128;
+	amdgpu_crtc->max_cursor_height = 128;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	for (i = 0; i < 256; i++) {
+		amdgpu_crtc->lut_r[i] = i << 2;
+		amdgpu_crtc->lut_g[i] = i << 2;
+		amdgpu_crtc->lut_b[i] = i << 2;
+	}
+
+	switch (amdgpu_crtc->crtc_id) {
+	case 0:
+	default:
+		amdgpu_crtc->crtc_offset = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		amdgpu_crtc->crtc_offset = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		amdgpu_crtc->crtc_offset = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		amdgpu_crtc->crtc_offset = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		amdgpu_crtc->crtc_offset = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		amdgpu_crtc->crtc_offset = CRTC5_REGISTER_OFFSET;
+		break;
+	}
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v11_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v11_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v11_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v11_0_audio_endpt_wreg;
+
+	dce_v11_0_set_display_funcs(adev);
+	dce_v11_0_set_irq_funcs(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+		adev->mode_info.num_crtc = 3;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 9;
+		break;
+	case CHIP_STONEY:
+		adev->mode_info.num_crtc = 2;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 9;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, i + 1, &adev->crtc_irq);
+		if (r)
+		return r;
+	}
+
+	for (i = 8; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, 42, &adev->hpd_irq);
+	if (r)
+	return r;
+
+	adev->mode_info.mode_config_initialized = true;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
+
+	r = amdgpu_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v11_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	dce_v11_0_afmt_init(adev);
+
+	r = dce_v11_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	return r;
+}
+
+static int dce_v11_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v11_0_audio_fini(adev);
+
+	dce_v11_0_afmt_fini(adev);
+
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v11_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v11_0_init_golden_registers(adev);
+
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_crtc_powergate_init(adev);
+	amdgpu_atombios_encoder_init_dig(adev);
+	amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+
+	/* initialize hpd */
+	dce_v11_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v11_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v11_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v11_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v11_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v11_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v11_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_atombios_scratch_regs_save(adev);
+
+	return dce_v11_0_hw_fini(handle);
+}
+
+static int dce_v11_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = dce_v11_0_hw_init(handle);
+
+	amdgpu_atombios_scratch_regs_restore(adev);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v11_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v11_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static void dce_v11_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "DCE 10.x registers\n");
+	/* XXX todo */
+}
+
+static int dce_v11_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (dce_v11_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		dce_v11_0_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+		dce_v11_0_print_status((void *)adev);
+	}
+	return 0;
+}
+
+static void dce_v11_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VBLANK_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v11_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+	u32 lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 0);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc]);
+		lb_interrupt_mask = REG_SET_FIELD(lb_interrupt_mask, LB_INTERRUPT_MASK,
+						  VLINE_INTERRUPT_MASK, 1);
+		WREG32(mmLB_INTERRUPT_MASK + crtc_offsets[crtc], lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static int dce_v11_0_set_hpd_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned hpd,
+					enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 0);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+		tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_EN, 1);
+		WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_set_crtc_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v11_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	 case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v11_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v11_0_set_pageflip_irq_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v11_0_pageflip_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if(amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						 "AMDGPU_FLIP_SUBMITTED(%d)\n",
+						 amdgpu_crtc->pflip_status,
+						 AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if(works->event)
+		drm_send_vblank_event(adev->ddev, crtc_id, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_vblank_put(adev->ddev, amdgpu_crtc->crtc_id);
+	queue_work(amdgpu_crtc->pflip_queue, &works->unpin_work);
+
+	return 0;
+}
+
+static void dce_v11_0_hpd_int_ack(struct amdgpu_device *adev,
+				  int hpd)
+{
+	u32 tmp;
+
+	if (hpd >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("invalid hdp %d\n", hpd);
+		return;
+	}
+
+	tmp = RREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd]);
+	tmp = REG_SET_FIELD(tmp, DC_HPD_INT_CONTROL, DC_HPD_INT_ACK, 1);
+	WREG32(mmDC_HPD_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
+static void dce_v11_0_crtc_vblank_int_ack(struct amdgpu_device *adev,
+					  int crtc)
+{
+	u32 tmp;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VBLANK_STATUS, VBLANK_ACK, 1);
+	WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static void dce_v11_0_crtc_vline_int_ack(struct amdgpu_device *adev,
+					 int crtc)
+{
+	u32 tmp;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	tmp = RREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc]);
+	tmp = REG_SET_FIELD(tmp, LB_VLINE_STATUS, VLINE_ACK, 1);
+	WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], tmp);
+}
+
+static int dce_v11_0_crtc_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
+
+	switch (entry->src_data) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			dce_v11_0_crtc_vblank_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			dce_v11_0_crtc_vline_int_ack(adev, crtc);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_hpd_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask;
+	unsigned hpd;
+
+	if (entry->src_data >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
+		return 0;
+	}
+
+	hpd = entry->src_data;
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+
+	if (disp_int & mask) {
+		dce_v11_0_hpd_int_ack(adev, hpd);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+}
+
+static int dce_v11_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v11_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs dce_v11_0_ip_funcs = {
+	.early_init = dce_v11_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v11_0_sw_init,
+	.sw_fini = dce_v11_0_sw_fini,
+	.hw_init = dce_v11_0_hw_init,
+	.hw_fini = dce_v11_0_hw_fini,
+	.suspend = dce_v11_0_suspend,
+	.resume = dce_v11_0_resume,
+	.is_idle = dce_v11_0_is_idle,
+	.wait_for_idle = dce_v11_0_wait_for_idle,
+	.soft_reset = dce_v11_0_soft_reset,
+	.print_status = dce_v11_0_print_status,
+	.set_clockgating_state = dce_v11_0_set_clockgating_state,
+	.set_powergating_state = dce_v11_0_set_powergating_state,
+};
+
+static void
+dce_v11_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v11_0_set_interleave(encoder->crtc, mode);
+
+	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
+		dce_v11_0_afmt_enable(encoder, true);
+		dce_v11_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v11_0_encoder_prepare(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v11_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v11_0_program_fmt(encoder);
+}
+
+static void dce_v11_0_encoder_commit(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v11_0_encoder_disable(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
+			dce_v11_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v11_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v11_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v11_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v11_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v11_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static bool dce_v11_0_ext_mode_fixup(struct drm_encoder *encoder,
+				    const struct drm_display_mode *mode,
+				    struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static const struct drm_encoder_helper_funcs dce_v11_0_ext_helper_funcs = {
+	.dpms = dce_v11_0_ext_dpms,
+	.mode_fixup = dce_v11_0_ext_mode_fixup,
+	.prepare = dce_v11_0_ext_prepare,
+	.mode_set = dce_v11_0_ext_mode_set,
+	.commit = dce_v11_0_ext_commit,
+	.disable = dce_v11_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v11_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v11_0_encoder_prepare,
+	.mode_set = dce_v11_0_encoder_mode_set,
+	.commit = dce_v11_0_encoder_commit,
+	.disable = dce_v11_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v11_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v11_0_encoder_prepare,
+	.mode_set = dce_v11_0_encoder_mode_set,
+	.commit = dce_v11_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v11_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v11_0_encoder_funcs = {
+	.destroy = dce_v11_0_encoder_destroy,
+};
+
+static void dce_v11_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC);
+		drm_encoder_helper_add(encoder, &dce_v11_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v11_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC);
+		else
+			drm_encoder_init(dev, encoder, &dce_v11_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS);
+		drm_encoder_helper_add(encoder, &dce_v11_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
+	.set_vga_render_state = &dce_v11_0_set_vga_render_state,
+	.bandwidth_update = &dce_v11_0_bandwidth_update,
+	.vblank_get_counter = &dce_v11_0_vblank_get_counter,
+	.vblank_wait = &dce_v11_0_vblank_wait,
+	.is_display_hung = &dce_v11_0_is_display_hung,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v11_0_hpd_sense,
+	.hpd_set_polarity = &dce_v11_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v11_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v11_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v11_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v11_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+	.stop_mc_access = &dce_v11_0_stop_mc_access,
+	.resume_mc_access = &dce_v11_0_resume_mc_access,
+};
+
+static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v11_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v11_0_crtc_irq_funcs = {
+	.set = dce_v11_0_set_crtc_irq_state,
+	.process = dce_v11_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v11_0_pageflip_irq_funcs = {
+	.set = dce_v11_0_set_pageflip_irq_state,
+	.process = dce_v11_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v11_0_hpd_irq_funcs = {
+	.set = dce_v11_0_set_hpd_irq_state,
+	.process = dce_v11_0_hpd_irq,
+};
+
+static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
+	adev->crtc_irq.funcs = &dce_v11_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
+	adev->pageflip_irq.funcs = &dce_v11_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
+	adev->hpd_irq.funcs = &dce_v11_0_hpd_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h
new file mode 100644
index 000000000..84e4618f5
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V11_0_H__
+#define __DCE_V11_0_H__
+
+extern const struct amd_ip_funcs dce_v11_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
new file mode 100644
index 000000000..42d954dc4
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -0,0 +1,3761 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "amdgpu_i2c.h"
+#include "cikd.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "atombios_crtc.h"
+#include "atombios_encoders.h"
+#include "amdgpu_pll.h"
+#include "amdgpu_connectors.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "gca/gfx_7_2_enum.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev);
+static void dce_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+
+static const u32 crtc_offsets[6] =
+{
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET
+};
+
+static const uint32_t dig_offsets[] = {
+	CRTC0_REGISTER_OFFSET,
+	CRTC1_REGISTER_OFFSET,
+	CRTC2_REGISTER_OFFSET,
+	CRTC3_REGISTER_OFFSET,
+	CRTC4_REGISTER_OFFSET,
+	CRTC5_REGISTER_OFFSET,
+	(0x13830 - 0x7030) >> 2,
+};
+
+static const struct {
+	uint32_t	reg;
+	uint32_t	vblank;
+	uint32_t	vline;
+	uint32_t	hpd;
+
+} interrupt_status_offsets[6] = { {
+	.reg = mmDISP_INTERRUPT_STATUS,
+	.vblank = DISP_INTERRUPT_STATUS__LB_D1_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS__LB_D1_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS__DC_HPD1_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE__LB_D2_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE__DC_HPD2_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE2,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE2__LB_D3_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE2__DC_HPD3_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE3,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE3__LB_D4_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE3__DC_HPD4_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE4,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE4__LB_D5_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE4__DC_HPD5_INTERRUPT_MASK
+}, {
+	.reg = mmDISP_INTERRUPT_STATUS_CONTINUE5,
+	.vblank = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VBLANK_INTERRUPT_MASK,
+	.vline = DISP_INTERRUPT_STATUS_CONTINUE5__LB_D6_VLINE_INTERRUPT_MASK,
+	.hpd = DISP_INTERRUPT_STATUS_CONTINUE5__DC_HPD6_INTERRUPT_MASK
+} };
+
+static const uint32_t hpd_int_control_offsets[6] = {
+	mmDC_HPD1_INT_CONTROL,
+	mmDC_HPD2_INT_CONTROL,
+	mmDC_HPD3_INT_CONTROL,
+	mmDC_HPD4_INT_CONTROL,
+	mmDC_HPD5_INT_CONTROL,
+	mmDC_HPD6_INT_CONTROL,
+};
+
+static u32 dce_v8_0_audio_endpt_rreg(struct amdgpu_device *adev,
+				     u32 block_offset, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	r = RREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+
+	return r;
+}
+
+static void dce_v8_0_audio_endpt_wreg(struct amdgpu_device *adev,
+				      u32 block_offset, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->audio_endpt_idx_lock, flags);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_INDEX + block_offset, reg);
+	WREG32(mmAZALIA_F0_CODEC_ENDPOINT_DATA + block_offset, v);
+	spin_unlock_irqrestore(&adev->audio_endpt_idx_lock, flags);
+}
+
+static bool dce_v8_0_is_in_vblank(struct amdgpu_device *adev, int crtc)
+{
+	if (RREG32(mmCRTC_STATUS + crtc_offsets[crtc]) &
+			CRTC_V_BLANK_START_END__CRTC_V_BLANK_START_MASK)
+		return true;
+	else
+		return false;
+}
+
+static bool dce_v8_0_is_counter_moving(struct amdgpu_device *adev, int crtc)
+{
+	u32 pos1, pos2;
+
+	pos1 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+	pos2 = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	if (pos1 != pos2)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_vblank_wait - vblank wait asic callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc: crtc to wait for vblank on
+ *
+ * Wait for vblank on the requested crtc (evergreen+).
+ */
+static void dce_v8_0_vblank_wait(struct amdgpu_device *adev, int crtc)
+{
+	unsigned i = 0;
+
+	if (crtc >= adev->mode_info.num_crtc)
+		return;
+
+	if (!(RREG32(mmCRTC_CONTROL + crtc_offsets[crtc]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK))
+		return;
+
+	/* depending on when we hit vblank, we may be close to active; if so,
+	 * wait for another frame.
+	 */
+	while (dce_v8_0_is_in_vblank(adev, crtc)) {
+		if (i++ % 100 == 0) {
+			if (!dce_v8_0_is_counter_moving(adev, crtc))
+				break;
+		}
+	}
+
+	while (!dce_v8_0_is_in_vblank(adev, crtc)) {
+		if (i++ % 100 == 0) {
+			if (!dce_v8_0_is_counter_moving(adev, crtc))
+				break;
+		}
+	}
+}
+
+static u32 dce_v8_0_vblank_get_counter(struct amdgpu_device *adev, int crtc)
+{
+	if (crtc >= adev->mode_info.num_crtc)
+		return 0;
+	else
+		return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
+}
+
+static void dce_v8_0_pageflip_interrupt_init(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Enable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_get(adev, &adev->pageflip_irq, i);
+}
+
+static void dce_v8_0_pageflip_interrupt_fini(struct amdgpu_device *adev)
+{
+	unsigned i;
+
+	/* Disable pflip interrupts */
+	for (i = 0; i < adev->mode_info.num_crtc; i++)
+		amdgpu_irq_put(adev, &adev->pageflip_irq, i);
+}
+
+/**
+ * dce_v8_0_page_flip - pageflip callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @crtc_id: crtc to cleanup pageflip on
+ * @crtc_base: new address of the crtc (GPU MC address)
+ *
+ * Triggers the actual pageflip by updating the primary
+ * surface base address.
+ */
+static void dce_v8_0_page_flip(struct amdgpu_device *adev,
+			      int crtc_id, u64 crtc_base)
+{
+	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	/* update the primary scanout addresses */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(crtc_base));
+	/* writing to the low address triggers the update */
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(crtc_base));
+	/* post the write */
+	RREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset);
+}
+
+static int dce_v8_0_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc,
+					u32 *vbl, u32 *position)
+{
+	if ((crtc < 0) || (crtc >= adev->mode_info.num_crtc))
+		return -EINVAL;
+
+	*vbl = RREG32(mmCRTC_V_BLANK_START_END + crtc_offsets[crtc]);
+	*position = RREG32(mmCRTC_STATUS_POSITION + crtc_offsets[crtc]);
+
+	return 0;
+}
+
+/**
+ * dce_v8_0_hpd_sense - hpd sense callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Checks if a digital monitor is connected (evergreen+).
+ * Returns true if connected, false if not connected.
+ */
+static bool dce_v8_0_hpd_sense(struct amdgpu_device *adev,
+			       enum amdgpu_hpd_id hpd)
+{
+	bool connected = false;
+
+	switch (hpd) {
+	case AMDGPU_HPD_1:
+		if (RREG32(mmDC_HPD1_INT_STATUS) & DC_HPD1_INT_STATUS__DC_HPD1_SENSE_MASK)
+			connected = true;
+		break;
+	case AMDGPU_HPD_2:
+		if (RREG32(mmDC_HPD2_INT_STATUS) & DC_HPD2_INT_STATUS__DC_HPD2_SENSE_MASK)
+			connected = true;
+		break;
+	case AMDGPU_HPD_3:
+		if (RREG32(mmDC_HPD3_INT_STATUS) & DC_HPD3_INT_STATUS__DC_HPD3_SENSE_MASK)
+			connected = true;
+		break;
+	case AMDGPU_HPD_4:
+		if (RREG32(mmDC_HPD4_INT_STATUS) & DC_HPD4_INT_STATUS__DC_HPD4_SENSE_MASK)
+			connected = true;
+		break;
+	case AMDGPU_HPD_5:
+		if (RREG32(mmDC_HPD5_INT_STATUS) & DC_HPD5_INT_STATUS__DC_HPD5_SENSE_MASK)
+			connected = true;
+		break;
+	case AMDGPU_HPD_6:
+		if (RREG32(mmDC_HPD6_INT_STATUS) & DC_HPD6_INT_STATUS__DC_HPD6_SENSE_MASK)
+			connected = true;
+		break;
+	default:
+		break;
+	}
+
+	return connected;
+}
+
+/**
+ * dce_v8_0_hpd_set_polarity - hpd set polarity callback.
+ *
+ * @adev: amdgpu_device pointer
+ * @hpd: hpd (hotplug detect) pin
+ *
+ * Set the polarity of the hpd pin (evergreen+).
+ */
+static void dce_v8_0_hpd_set_polarity(struct amdgpu_device *adev,
+				      enum amdgpu_hpd_id hpd)
+{
+	u32 tmp;
+	bool connected = dce_v8_0_hpd_sense(adev, hpd);
+
+	switch (hpd) {
+	case AMDGPU_HPD_1:
+		tmp = RREG32(mmDC_HPD1_INT_CONTROL);
+		if (connected)
+			tmp &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_POLARITY_MASK;
+		else
+			tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_POLARITY_MASK;
+		WREG32(mmDC_HPD1_INT_CONTROL, tmp);
+		break;
+	case AMDGPU_HPD_2:
+		tmp = RREG32(mmDC_HPD2_INT_CONTROL);
+		if (connected)
+			tmp &= ~DC_HPD2_INT_CONTROL__DC_HPD2_INT_POLARITY_MASK;
+		else
+			tmp |= DC_HPD2_INT_CONTROL__DC_HPD2_INT_POLARITY_MASK;
+		WREG32(mmDC_HPD2_INT_CONTROL, tmp);
+		break;
+	case AMDGPU_HPD_3:
+		tmp = RREG32(mmDC_HPD3_INT_CONTROL);
+		if (connected)
+			tmp &= ~DC_HPD3_INT_CONTROL__DC_HPD3_INT_POLARITY_MASK;
+		else
+			tmp |= DC_HPD3_INT_CONTROL__DC_HPD3_INT_POLARITY_MASK;
+		WREG32(mmDC_HPD3_INT_CONTROL, tmp);
+		break;
+	case AMDGPU_HPD_4:
+		tmp = RREG32(mmDC_HPD4_INT_CONTROL);
+		if (connected)
+			tmp &= ~DC_HPD4_INT_CONTROL__DC_HPD4_INT_POLARITY_MASK;
+		else
+			tmp |= DC_HPD4_INT_CONTROL__DC_HPD4_INT_POLARITY_MASK;
+		WREG32(mmDC_HPD4_INT_CONTROL, tmp);
+		break;
+	case AMDGPU_HPD_5:
+		tmp = RREG32(mmDC_HPD5_INT_CONTROL);
+		if (connected)
+			tmp &= ~DC_HPD5_INT_CONTROL__DC_HPD5_INT_POLARITY_MASK;
+		else
+			tmp |= DC_HPD5_INT_CONTROL__DC_HPD5_INT_POLARITY_MASK;
+		WREG32(mmDC_HPD5_INT_CONTROL, tmp);
+			break;
+	case AMDGPU_HPD_6:
+		tmp = RREG32(mmDC_HPD6_INT_CONTROL);
+		if (connected)
+			tmp &= ~DC_HPD6_INT_CONTROL__DC_HPD6_INT_POLARITY_MASK;
+		else
+			tmp |= DC_HPD6_INT_CONTROL__DC_HPD6_INT_POLARITY_MASK;
+		WREG32(mmDC_HPD6_INT_CONTROL, tmp);
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * dce_v8_0_hpd_init - hpd setup callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup the hpd pins used by the card (evergreen+).
+ * Enable the pin, set the polarity, and enable the hpd interrupts.
+ */
+static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+	u32 tmp = (0x9c4 << DC_HPD1_CONTROL__DC_HPD1_CONNECTION_TIMER__SHIFT) |
+		(0xfa << DC_HPD1_CONTROL__DC_HPD1_RX_INT_TIMER__SHIFT) |
+		DC_HPD1_CONTROL__DC_HPD1_EN_MASK;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			continue;
+		}
+		switch (amdgpu_connector->hpd.hpd) {
+		case AMDGPU_HPD_1:
+			WREG32(mmDC_HPD1_CONTROL, tmp);
+			break;
+		case AMDGPU_HPD_2:
+			WREG32(mmDC_HPD2_CONTROL, tmp);
+			break;
+		case AMDGPU_HPD_3:
+			WREG32(mmDC_HPD3_CONTROL, tmp);
+			break;
+		case AMDGPU_HPD_4:
+			WREG32(mmDC_HPD4_CONTROL, tmp);
+			break;
+		case AMDGPU_HPD_5:
+			WREG32(mmDC_HPD5_CONTROL, tmp);
+			break;
+		case AMDGPU_HPD_6:
+			WREG32(mmDC_HPD6_CONTROL, tmp);
+			break;
+		default:
+			break;
+		}
+		dce_v8_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
+		amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+/**
+ * dce_v8_0_hpd_fini - hpd tear down callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down the hpd pins used by the card (evergreen+).
+ * Disable the hpd interrupts.
+ */
+static void dce_v8_0_hpd_fini(struct amdgpu_device *adev)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_connector *connector;
+
+	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		switch (amdgpu_connector->hpd.hpd) {
+		case AMDGPU_HPD_1:
+			WREG32(mmDC_HPD1_CONTROL, 0);
+			break;
+		case AMDGPU_HPD_2:
+			WREG32(mmDC_HPD2_CONTROL, 0);
+			break;
+		case AMDGPU_HPD_3:
+			WREG32(mmDC_HPD3_CONTROL, 0);
+			break;
+		case AMDGPU_HPD_4:
+			WREG32(mmDC_HPD4_CONTROL, 0);
+			break;
+		case AMDGPU_HPD_5:
+			WREG32(mmDC_HPD5_CONTROL, 0);
+			break;
+		case AMDGPU_HPD_6:
+			WREG32(mmDC_HPD6_CONTROL, 0);
+			break;
+		default:
+			break;
+		}
+		amdgpu_irq_put(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
+	}
+}
+
+static u32 dce_v8_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
+{
+	return mmDC_GPIO_HPD_A;
+}
+
+static bool dce_v8_0_is_display_hung(struct amdgpu_device *adev)
+{
+	u32 crtc_hung = 0;
+	u32 crtc_status[6];
+	u32 i, j, tmp;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (RREG32(mmCRTC_CONTROL + crtc_offsets[i]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK) {
+			crtc_status[i] = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+			crtc_hung |= (1 << i);
+		}
+	}
+
+	for (j = 0; j < 10; j++) {
+		for (i = 0; i < adev->mode_info.num_crtc; i++) {
+			if (crtc_hung & (1 << i)) {
+				tmp = RREG32(mmCRTC_STATUS_HV_COUNT + crtc_offsets[i]);
+				if (tmp != crtc_status[i])
+					crtc_hung &= ~(1 << i);
+			}
+		}
+		if (crtc_hung == 0)
+			return false;
+		udelay(100);
+	}
+
+	return true;
+}
+
+static void dce_v8_0_stop_mc_access(struct amdgpu_device *adev,
+				    struct amdgpu_mode_mc_save *save)
+{
+	u32 crtc_enabled, tmp;
+	int i;
+
+	save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+
+	/* blank the display controllers */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
+					     CRTC_CONTROL, CRTC_MASTER_EN);
+		if (crtc_enabled) {
+#if 0
+			u32 frame_count;
+			int j;
+
+			save->crtc_enabled[i] = true;
+			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
+				amdgpu_display_vblank_wait(adev, i);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
+				WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
+				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			}
+			/* wait for the next frame */
+			frame_count = amdgpu_display_vblank_get_counter(adev, i);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
+					break;
+				udelay(1);
+			}
+			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK) == 0) {
+				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 1);
+				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK) == 0) {
+				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 1);
+				WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
+			}
+#else
+			/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+			tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
+			tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
+			WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			save->crtc_enabled[i] = false;
+			/* ***** */
+#endif
+		} else {
+			save->crtc_enabled[i] = false;
+		}
+	}
+}
+
+static void dce_v8_0_resume_mc_access(struct amdgpu_device *adev,
+				      struct amdgpu_mode_mc_save *save)
+{
+	u32 tmp, frame_count;
+	int i, j;
+
+	/* update crtc base addresses */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
+		       upper_32_bits(adev->mc.vram_start));
+		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
+		       upper_32_bits(adev->mc.vram_start));
+		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
+		       (u32)adev->mc.vram_start);
+		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i],
+		       (u32)adev->mc.vram_start);
+
+		if (save->crtc_enabled[i]) {
+			tmp = RREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 3) {
+				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 3);
+				WREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK)) {
+				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 0);
+				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
+			}
+			tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
+			if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK)) {
+				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 0);
+				WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
+			}
+			for (j = 0; j < adev->usec_timeout; j++) {
+				tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
+				if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING) == 0)
+					break;
+				udelay(1);
+			}
+			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
+			tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+			WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
+			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
+			/* wait for the next frame */
+			frame_count = amdgpu_display_vblank_get_counter(adev, i);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
+					break;
+				udelay(1);
+			}
+		}
+	}
+
+	WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
+	WREG32(mmVGA_MEMORY_BASE_ADDRESS, lower_32_bits(adev->mc.vram_start));
+
+	/* Unlock vga access */
+	WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
+	mdelay(1);
+	WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
+}
+
+static void dce_v8_0_set_vga_render_state(struct amdgpu_device *adev,
+					  bool render)
+{
+	u32 tmp;
+
+	/* Lockout access through VGA aperture*/
+	tmp = RREG32(mmVGA_HDP_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 0);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
+	WREG32(mmVGA_HDP_CONTROL, tmp);
+
+	/* disable VGA render */
+	tmp = RREG32(mmVGA_RENDER_CONTROL);
+	if (render)
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
+	WREG32(mmVGA_RENDER_CONTROL, tmp);
+}
+
+static void dce_v8_0_program_fmt(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	int bpc = 0;
+	u32 tmp = 0;
+	enum amdgpu_connector_dither dither = AMDGPU_FMT_DITHER_DISABLE;
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+		bpc = amdgpu_connector_get_monitor_bpc(connector);
+		dither = amdgpu_connector->dither;
+	}
+
+	/* LVDS/eDP FMT is set up by atom */
+	if (amdgpu_encoder->devices & ATOM_DEVICE_LCD_SUPPORT)
+		return;
+
+	/* not needed for analog */
+	if ((amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) ||
+	    (amdgpu_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2))
+		return;
+
+	if (bpc == 0)
+		return;
+
+	switch (bpc) {
+	case 6:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				(0 << FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH__SHIFT));
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+			(0 << FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH__SHIFT));
+		break;
+	case 8:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_RGB_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				(1 << FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH__SHIFT));
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+			(1 << FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH__SHIFT));
+		break;
+	case 10:
+		if (dither == AMDGPU_FMT_DITHER_ENABLE)
+			/* XXX sort out optimal dither settings */
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_FRAME_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_HIGHPASS_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_RGB_RANDOM_ENABLE_MASK |
+				FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_EN_MASK |
+				(2 << FMT_BIT_DEPTH_CONTROL__FMT_SPATIAL_DITHER_DEPTH__SHIFT));
+		else
+			tmp |= (FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_EN_MASK |
+			(2 << FMT_BIT_DEPTH_CONTROL__FMT_TRUNCATE_DEPTH__SHIFT));
+		break;
+	default:
+		/* not needed */
+		break;
+	}
+
+	WREG32(mmFMT_BIT_DEPTH_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+}
+
+
+/* display watermark setup */
+/**
+ * dce_v8_0_line_buffer_adjust - Set up the line buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @mode: the current display mode on the selected display
+ * controller
+ *
+ * Setup up the line buffer allocation for
+ * the selected display controller (CIK).
+ * Returns the line buffer size in pixels.
+ */
+static u32 dce_v8_0_line_buffer_adjust(struct amdgpu_device *adev,
+				       struct amdgpu_crtc *amdgpu_crtc,
+				       struct drm_display_mode *mode)
+{
+	u32 tmp, buffer_alloc, i;
+	u32 pipe_offset = amdgpu_crtc->crtc_id * 0x8;
+	/*
+	 * Line Buffer Setup
+	 * There are 6 line buffers, one for each display controllers.
+	 * There are 3 partitions per LB. Select the number of partitions
+	 * to enable based on the display width.  For display widths larger
+	 * than 4096, you need use to use 2 display controllers and combine
+	 * them using the stereo blender.
+	 */
+	if (amdgpu_crtc->base.enabled && mode) {
+		if (mode->crtc_hdisplay < 1920) {
+			tmp = 1;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 2560) {
+			tmp = 2;
+			buffer_alloc = 2;
+		} else if (mode->crtc_hdisplay < 4096) {
+			tmp = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		} else {
+			DRM_DEBUG_KMS("Mode too big for LB!\n");
+			tmp = 0;
+			buffer_alloc = (adev->flags & AMD_IS_APU) ? 2 : 4;
+		}
+	} else {
+		tmp = 1;
+		buffer_alloc = 0;
+	}
+
+	WREG32(mmLB_MEMORY_CTRL + amdgpu_crtc->crtc_offset,
+	      (tmp << LB_MEMORY_CTRL__LB_MEMORY_CONFIG__SHIFT) |
+	      (0x6B0 << LB_MEMORY_CTRL__LB_MEMORY_SIZE__SHIFT));
+
+	WREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset,
+	       (buffer_alloc << PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATED__SHIFT));
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmPIPE0_DMIF_BUFFER_CONTROL + pipe_offset) &
+		    PIPE0_DMIF_BUFFER_CONTROL__DMIF_BUFFERS_ALLOCATION_COMPLETED_MASK)
+			break;
+		udelay(1);
+	}
+
+	if (amdgpu_crtc->base.enabled && mode) {
+		switch (tmp) {
+		case 0:
+		default:
+			return 4096 * 2;
+		case 1:
+			return 1920 * 2;
+		case 2:
+			return 2560 * 2;
+		}
+	}
+
+	/* controller not enabled, so no lb used */
+	return 0;
+}
+
+/**
+ * cik_get_number_of_dram_channels - get the number of dram channels
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the number of video ram channels (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the number of dram channels
+ */
+static u32 cik_get_number_of_dram_channels(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmMC_SHARED_CHMAP);
+
+	switch ((tmp & MC_SHARED_CHMAP__NOOFCHAN_MASK) >> MC_SHARED_CHMAP__NOOFCHAN__SHIFT) {
+	case 0:
+	default:
+		return 1;
+	case 1:
+		return 2;
+	case 2:
+		return 4;
+	case 3:
+		return 8;
+	case 4:
+		return 3;
+	case 5:
+		return 6;
+	case 6:
+		return 10;
+	case 7:
+		return 12;
+	case 8:
+		return 16;
+	}
+}
+
+struct dce8_wm_params {
+	u32 dram_channels; /* number of dram channels */
+	u32 yclk;          /* bandwidth per dram data pin in kHz */
+	u32 sclk;          /* engine clock in kHz */
+	u32 disp_clk;      /* display clock in kHz */
+	u32 src_width;     /* viewport width */
+	u32 active_time;   /* active display time in ns */
+	u32 blank_time;    /* blank time in ns */
+	bool interlaced;    /* mode is interlaced */
+	fixed20_12 vsc;    /* vertical scale ratio */
+	u32 num_heads;     /* number of active crtcs */
+	u32 bytes_per_pixel; /* bytes per pixel display + overlay */
+	u32 lb_size;       /* line buffer allocated to pipe */
+	u32 vtaps;         /* vertical scaler taps */
+};
+
+/**
+ * dce_v8_0_dram_bandwidth - get the dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the raw dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_dram_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate raw DRAM Bandwidth */
+	fixed20_12 dram_efficiency; /* 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	dram_efficiency.full = dfixed_const(7);
+	dram_efficiency.full = dfixed_div(dram_efficiency, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, dram_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_dram_bandwidth_for_display - get the dram bandwidth for display
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dram bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dram bandwidth for display in MBytes/s
+ */
+static u32 dce_v8_0_dram_bandwidth_for_display(struct dce8_wm_params *wm)
+{
+	/* Calculate DRAM Bandwidth and the part allocated to display. */
+	fixed20_12 disp_dram_allocation; /* 0.3 to 0.7 */
+	fixed20_12 yclk, dram_channels, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	yclk.full = dfixed_const(wm->yclk);
+	yclk.full = dfixed_div(yclk, a);
+	dram_channels.full = dfixed_const(wm->dram_channels * 4);
+	a.full = dfixed_const(10);
+	disp_dram_allocation.full = dfixed_const(3); /* XXX worse case value 0.3 */
+	disp_dram_allocation.full = dfixed_div(disp_dram_allocation, a);
+	bandwidth.full = dfixed_mul(dram_channels, yclk);
+	bandwidth.full = dfixed_mul(bandwidth, disp_dram_allocation);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_data_return_bandwidth - get the data return bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the data return bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the data return bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_data_return_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the display Data return Bandwidth */
+	fixed20_12 return_efficiency; /* 0.8 */
+	fixed20_12 sclk, bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	sclk.full = dfixed_const(wm->sclk);
+	sclk.full = dfixed_div(sclk, a);
+	a.full = dfixed_const(10);
+	return_efficiency.full = dfixed_const(8);
+	return_efficiency.full = dfixed_div(return_efficiency, a);
+	a.full = dfixed_const(32);
+	bandwidth.full = dfixed_mul(a, sclk);
+	bandwidth.full = dfixed_mul(bandwidth, return_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_dmif_request_bandwidth - get the dmif bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the dmif bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the dmif bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_dmif_request_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the DMIF Request Bandwidth */
+	fixed20_12 disp_clk_request_efficiency; /* 0.8 */
+	fixed20_12 disp_clk, bandwidth;
+	fixed20_12 a, b;
+
+	a.full = dfixed_const(1000);
+	disp_clk.full = dfixed_const(wm->disp_clk);
+	disp_clk.full = dfixed_div(disp_clk, a);
+	a.full = dfixed_const(32);
+	b.full = dfixed_mul(a, disp_clk);
+
+	a.full = dfixed_const(10);
+	disp_clk_request_efficiency.full = dfixed_const(8);
+	disp_clk_request_efficiency.full = dfixed_div(disp_clk_request_efficiency, a);
+
+	bandwidth.full = dfixed_mul(b, disp_clk_request_efficiency);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_available_bandwidth - get the min available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the min available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the min available bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_available_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the Available bandwidth. Display can use this temporarily but not in average. */
+	u32 dram_bandwidth = dce_v8_0_dram_bandwidth(wm);
+	u32 data_return_bandwidth = dce_v8_0_data_return_bandwidth(wm);
+	u32 dmif_req_bandwidth = dce_v8_0_dmif_request_bandwidth(wm);
+
+	return min(dram_bandwidth, min(data_return_bandwidth, dmif_req_bandwidth));
+}
+
+/**
+ * dce_v8_0_average_bandwidth - get the average available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the average available bandwidth used for display (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the average available bandwidth in MBytes/s
+ */
+static u32 dce_v8_0_average_bandwidth(struct dce8_wm_params *wm)
+{
+	/* Calculate the display mode Average Bandwidth
+	 * DisplayMode should contain the source and destination dimensions,
+	 * timing, etc.
+	 */
+	fixed20_12 bpp;
+	fixed20_12 line_time;
+	fixed20_12 src_width;
+	fixed20_12 bandwidth;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1000);
+	line_time.full = dfixed_const(wm->active_time + wm->blank_time);
+	line_time.full = dfixed_div(line_time, a);
+	bpp.full = dfixed_const(wm->bytes_per_pixel);
+	src_width.full = dfixed_const(wm->src_width);
+	bandwidth.full = dfixed_mul(src_width, bpp);
+	bandwidth.full = dfixed_mul(bandwidth, wm->vsc);
+	bandwidth.full = dfixed_div(bandwidth, line_time);
+
+	return dfixed_trunc(bandwidth);
+}
+
+/**
+ * dce_v8_0_latency_watermark - get the latency watermark
+ *
+ * @wm: watermark calculation data
+ *
+ * Calculate the latency watermark (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns the latency watermark in ns
+ */
+static u32 dce_v8_0_latency_watermark(struct dce8_wm_params *wm)
+{
+	/* First calculate the latency in ns */
+	u32 mc_latency = 2000; /* 2000 ns. */
+	u32 available_bandwidth = dce_v8_0_available_bandwidth(wm);
+	u32 worst_chunk_return_time = (512 * 8 * 1000) / available_bandwidth;
+	u32 cursor_line_pair_return_time = (128 * 4 * 1000) / available_bandwidth;
+	u32 dc_latency = 40000000 / wm->disp_clk; /* dc pipe latency */
+	u32 other_heads_data_return_time = ((wm->num_heads + 1) * worst_chunk_return_time) +
+		(wm->num_heads * cursor_line_pair_return_time);
+	u32 latency = mc_latency + other_heads_data_return_time + dc_latency;
+	u32 max_src_lines_per_dst_line, lb_fill_bw, line_fill_time;
+	u32 tmp, dmif_size = 12288;
+	fixed20_12 a, b, c;
+
+	if (wm->num_heads == 0)
+		return 0;
+
+	a.full = dfixed_const(2);
+	b.full = dfixed_const(1);
+	if ((wm->vsc.full > a.full) ||
+	    ((wm->vsc.full > b.full) && (wm->vtaps >= 3)) ||
+	    (wm->vtaps >= 5) ||
+	    ((wm->vsc.full >= a.full) && wm->interlaced))
+		max_src_lines_per_dst_line = 4;
+	else
+		max_src_lines_per_dst_line = 2;
+
+	a.full = dfixed_const(available_bandwidth);
+	b.full = dfixed_const(wm->num_heads);
+	a.full = dfixed_div(a, b);
+
+	b.full = dfixed_const(mc_latency + 512);
+	c.full = dfixed_const(wm->disp_clk);
+	b.full = dfixed_div(b, c);
+
+	c.full = dfixed_const(dmif_size);
+	b.full = dfixed_div(c, b);
+
+	tmp = min(dfixed_trunc(a), dfixed_trunc(b));
+
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(wm->disp_clk);
+	b.full = dfixed_div(c, b);
+	c.full = dfixed_const(wm->bytes_per_pixel);
+	b.full = dfixed_mul(b, c);
+
+	lb_fill_bw = min(tmp, dfixed_trunc(b));
+
+	a.full = dfixed_const(max_src_lines_per_dst_line * wm->src_width * wm->bytes_per_pixel);
+	b.full = dfixed_const(1000);
+	c.full = dfixed_const(lb_fill_bw);
+	b.full = dfixed_div(c, b);
+	a.full = dfixed_div(a, b);
+	line_fill_time = dfixed_trunc(a);
+
+	if (line_fill_time < wm->active_time)
+		return latency;
+	else
+		return latency + (line_fill_time - wm->active_time);
+
+}
+
+/**
+ * dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display - check
+ * average and available dram bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * dram bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display(struct dce8_wm_params *wm)
+{
+	if (dce_v8_0_average_bandwidth(wm) <=
+	    (dce_v8_0_dram_bandwidth_for_display(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_average_bandwidth_vs_available_bandwidth - check
+ * average and available bandwidth
+ *
+ * @wm: watermark calculation data
+ *
+ * Check if the display average bandwidth fits in the display
+ * available bandwidth (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v8_0_average_bandwidth_vs_available_bandwidth(struct dce8_wm_params *wm)
+{
+	if (dce_v8_0_average_bandwidth(wm) <=
+	    (dce_v8_0_available_bandwidth(wm) / wm->num_heads))
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_check_latency_hiding - check latency hiding
+ *
+ * @wm: watermark calculation data
+ *
+ * Check latency hiding (CIK).
+ * Used for display watermark bandwidth calculations
+ * Returns true if the display fits, false if not.
+ */
+static bool dce_v8_0_check_latency_hiding(struct dce8_wm_params *wm)
+{
+	u32 lb_partitions = wm->lb_size / wm->src_width;
+	u32 line_time = wm->active_time + wm->blank_time;
+	u32 latency_tolerant_lines;
+	u32 latency_hiding;
+	fixed20_12 a;
+
+	a.full = dfixed_const(1);
+	if (wm->vsc.full > a.full)
+		latency_tolerant_lines = 1;
+	else {
+		if (lb_partitions <= (wm->vtaps + 1))
+			latency_tolerant_lines = 1;
+		else
+			latency_tolerant_lines = 2;
+	}
+
+	latency_hiding = (latency_tolerant_lines * line_time + wm->blank_time);
+
+	if (dce_v8_0_latency_watermark(wm) <= latency_hiding)
+		return true;
+	else
+		return false;
+}
+
+/**
+ * dce_v8_0_program_watermarks - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ * @amdgpu_crtc: the selected display controller
+ * @lb_size: line buffer size
+ * @num_heads: number of display controllers in use
+ *
+ * Calculate and program the display watermarks for the
+ * selected display controller (CIK).
+ */
+static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
+					struct amdgpu_crtc *amdgpu_crtc,
+					u32 lb_size, u32 num_heads)
+{
+	struct drm_display_mode *mode = &amdgpu_crtc->base.mode;
+	struct dce8_wm_params wm_low, wm_high;
+	u32 pixel_period;
+	u32 line_time = 0;
+	u32 latency_watermark_a = 0, latency_watermark_b = 0;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
+
+	if (amdgpu_crtc->base.enabled && num_heads && mode) {
+		pixel_period = 1000000 / (u32)mode->clock;
+		line_time = min((u32)mode->crtc_htotal * pixel_period, (u32)65535);
+
+		/* watermark for high clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_high.yclk =
+				amdgpu_dpm_get_mclk(adev, false) * 10;
+			wm_high.sclk =
+				amdgpu_dpm_get_sclk(adev, false) * 10;
+		} else {
+			wm_high.yclk = adev->pm.current_mclk * 10;
+			wm_high.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_high.disp_clk = mode->clock;
+		wm_high.src_width = mode->crtc_hdisplay;
+		wm_high.active_time = mode->crtc_hdisplay * pixel_period;
+		wm_high.blank_time = line_time - wm_high.active_time;
+		wm_high.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_high.interlaced = true;
+		wm_high.vsc = amdgpu_crtc->vsc;
+		wm_high.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_high.vtaps = 2;
+		wm_high.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_high.lb_size = lb_size;
+		wm_high.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_high.num_heads = num_heads;
+
+		/* set for high clocks */
+		latency_watermark_a = min(dce_v8_0_latency_watermark(&wm_high), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_high) ||
+		    !dce_v8_0_average_bandwidth_vs_available_bandwidth(&wm_high) ||
+		    !dce_v8_0_check_latency_hiding(&wm_high) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+
+		/* watermark for low clocks */
+		if (adev->pm.dpm_enabled) {
+			wm_low.yclk =
+				amdgpu_dpm_get_mclk(adev, true) * 10;
+			wm_low.sclk =
+				amdgpu_dpm_get_sclk(adev, true) * 10;
+		} else {
+			wm_low.yclk = adev->pm.current_mclk * 10;
+			wm_low.sclk = adev->pm.current_sclk * 10;
+		}
+
+		wm_low.disp_clk = mode->clock;
+		wm_low.src_width = mode->crtc_hdisplay;
+		wm_low.active_time = mode->crtc_hdisplay * pixel_period;
+		wm_low.blank_time = line_time - wm_low.active_time;
+		wm_low.interlaced = false;
+		if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+			wm_low.interlaced = true;
+		wm_low.vsc = amdgpu_crtc->vsc;
+		wm_low.vtaps = 1;
+		if (amdgpu_crtc->rmx_type != RMX_OFF)
+			wm_low.vtaps = 2;
+		wm_low.bytes_per_pixel = 4; /* XXX: get this from fb config */
+		wm_low.lb_size = lb_size;
+		wm_low.dram_channels = cik_get_number_of_dram_channels(adev);
+		wm_low.num_heads = num_heads;
+
+		/* set for low clocks */
+		latency_watermark_b = min(dce_v8_0_latency_watermark(&wm_low), (u32)65535);
+
+		/* possibly force display priority to high */
+		/* should really do this at mode validation time... */
+		if (!dce_v8_0_average_bandwidth_vs_dram_bandwidth_for_display(&wm_low) ||
+		    !dce_v8_0_average_bandwidth_vs_available_bandwidth(&wm_low) ||
+		    !dce_v8_0_check_latency_hiding(&wm_low) ||
+		    (adev->mode_info.disp_priority == 2)) {
+			DRM_DEBUG_KMS("force priority to high\n");
+		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
+	}
+
+	/* select wm A */
+	wm_mask = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp = wm_mask;
+	tmp &= ~(3 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (1 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((latency_watermark_a << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
+		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
+	/* select wm B */
+	tmp = RREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp &= ~(3 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	tmp |= (2 << DPG_WATERMARK_MASK_CONTROL__URGENCY_WATERMARK_MASK__SHIFT);
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+	WREG32(mmDPG_PIPE_URGENCY_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((latency_watermark_b << DPG_PIPE_URGENCY_CONTROL__URGENCY_LOW_WATERMARK__SHIFT) |
+		(line_time << DPG_PIPE_URGENCY_CONTROL__URGENCY_HIGH_WATERMARK__SHIFT)));
+	/* restore original selection */
+	WREG32(mmDPG_WATERMARK_MASK_CONTROL + amdgpu_crtc->crtc_offset, wm_mask);
+
+	/* save values for DPM */
+	amdgpu_crtc->line_time = line_time;
+	amdgpu_crtc->wm_high = latency_watermark_a;
+	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
+}
+
+/**
+ * dce_v8_0_bandwidth_update - program display watermarks
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Calculate and program the display watermarks and line
+ * buffer allocation (CIK).
+ */
+static void dce_v8_0_bandwidth_update(struct amdgpu_device *adev)
+{
+	struct drm_display_mode *mode = NULL;
+	u32 num_heads = 0, lb_size;
+	int i;
+
+	amdgpu_update_display_priority(adev);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i]->base.enabled)
+			num_heads++;
+	}
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		mode = &adev->mode_info.crtcs[i]->base.mode;
+		lb_size = dce_v8_0_line_buffer_adjust(adev, adev->mode_info.crtcs[i], mode);
+		dce_v8_0_program_watermarks(adev, adev->mode_info.crtcs[i],
+					    lb_size, num_heads);
+	}
+}
+
+static void dce_v8_0_audio_get_connected_pins(struct amdgpu_device *adev)
+{
+	int i;
+	u32 offset, tmp;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		offset = adev->mode_info.audio.pin[i].offset;
+		tmp = RREG32_AUDIO_ENDPT(offset,
+					 ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT);
+		if (((tmp &
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY_MASK) >>
+		AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_CONFIGURATION_DEFAULT__PORT_CONNECTIVITY__SHIFT) == 1)
+			adev->mode_info.audio.pin[i].connected = false;
+		else
+			adev->mode_info.audio.pin[i].connected = true;
+	}
+}
+
+static struct amdgpu_audio_pin *dce_v8_0_audio_get_pin(struct amdgpu_device *adev)
+{
+	int i;
+
+	dce_v8_0_audio_get_connected_pins(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		if (adev->mode_info.audio.pin[i].connected)
+			return &adev->mode_info.audio.pin[i];
+	}
+	DRM_ERROR("No connected audio pins found!\n");
+	return NULL;
+}
+
+static void dce_v8_0_afmt_audio_select_pin(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 offset;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->offset;
+
+	WREG32(mmAFMT_AUDIO_SRC_CONTROL + offset,
+	       (dig->afmt->pin->id << AFMT_AUDIO_SRC_CONTROL__AFMT_AUDIO_SRC_SELECT__SHIFT));
+}
+
+static void dce_v8_0_audio_write_latency_fields(struct drm_encoder *encoder,
+						struct drm_display_mode *mode)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 tmp = 0, offset;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE) {
+		if (connector->latency_present[1])
+			tmp =
+			(connector->video_latency[1] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(connector->audio_latency[1] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+		else
+			tmp =
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+	} else {
+		if (connector->latency_present[0])
+			tmp =
+			(connector->video_latency[0] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(connector->audio_latency[0] <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+		else
+			tmp =
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__VIDEO_LIPSYNC__SHIFT) |
+			(0 <<
+			 AZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC__AUDIO_LIPSYNC__SHIFT);
+
+	}
+	WREG32_AUDIO_ENDPT(offset, ixAZALIA_F0_CODEC_PIN_CONTROL_RESPONSE_LIPSYNC, tmp);
+}
+
+static void dce_v8_0_audio_write_speaker_allocation(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	u32 offset, tmp;
+	u8 *sadb = NULL;
+	int sad_count;
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_speaker_allocation(amdgpu_connector_edid(connector), &sadb);
+	if (sad_count < 0) {
+		DRM_ERROR("Couldn't read Speaker Allocation Data Block: %d\n", sad_count);
+		sad_count = 0;
+	}
+
+	/* program the speaker allocation */
+	tmp = RREG32_AUDIO_ENDPT(offset, ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER);
+	tmp &= ~(AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__DP_CONNECTION_MASK |
+		AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__SPEAKER_ALLOCATION_MASK);
+	/* set HDMI mode */
+	tmp |= AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__HDMI_CONNECTION_MASK;
+	if (sad_count)
+		tmp |= (sadb[0] << AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__SPEAKER_ALLOCATION__SHIFT);
+	else
+		tmp |= (5 << AZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER__SPEAKER_ALLOCATION__SHIFT); /* stereo */
+	WREG32_AUDIO_ENDPT(offset, ixAZALIA_F0_CODEC_PIN_CONTROL_CHANNEL_SPEAKER, tmp);
+
+	kfree(sadb);
+}
+
+static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	u32 offset;
+	struct drm_connector *connector;
+	struct amdgpu_connector *amdgpu_connector = NULL;
+	struct cea_sad *sads;
+	int i, sad_count;
+
+	static const u16 eld_reg_to_type[][2] = {
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0, HDMI_AUDIO_CODING_TYPE_PCM },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR1, HDMI_AUDIO_CODING_TYPE_AC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR2, HDMI_AUDIO_CODING_TYPE_MPEG1 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR3, HDMI_AUDIO_CODING_TYPE_MP3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR4, HDMI_AUDIO_CODING_TYPE_MPEG2 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR5, HDMI_AUDIO_CODING_TYPE_AAC_LC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR6, HDMI_AUDIO_CODING_TYPE_DTS },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR7, HDMI_AUDIO_CODING_TYPE_ATRAC },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR9, HDMI_AUDIO_CODING_TYPE_EAC3 },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR10, HDMI_AUDIO_CODING_TYPE_DTS_HD },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR11, HDMI_AUDIO_CODING_TYPE_MLP },
+		{ ixAZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR13, HDMI_AUDIO_CODING_TYPE_WMA_PRO },
+	};
+
+	if (!dig || !dig->afmt || !dig->afmt->pin)
+		return;
+
+	offset = dig->afmt->pin->offset;
+
+	list_for_each_entry(connector, &encoder->dev->mode_config.connector_list, head) {
+		if (connector->encoder == encoder) {
+			amdgpu_connector = to_amdgpu_connector(connector);
+			break;
+		}
+	}
+
+	if (!amdgpu_connector) {
+		DRM_ERROR("Couldn't find encoder's connector\n");
+		return;
+	}
+
+	sad_count = drm_edid_to_sad(amdgpu_connector_edid(connector), &sads);
+	if (sad_count <= 0) {
+		DRM_ERROR("Couldn't read SADs: %d\n", sad_count);
+		return;
+	}
+	BUG_ON(!sads);
+
+	for (i = 0; i < ARRAY_SIZE(eld_reg_to_type); i++) {
+		u32 value = 0;
+		u8 stereo_freqs = 0;
+		int max_channels = -1;
+		int j;
+
+		for (j = 0; j < sad_count; j++) {
+			struct cea_sad *sad = &sads[j];
+
+			if (sad->format == eld_reg_to_type[i][1]) {
+				if (sad->channels > max_channels) {
+				value = (sad->channels <<
+				 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) |
+				(sad->byte2 <<
+				 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) |
+				(sad->freq <<
+				 AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT);
+				max_channels = sad->channels;
+				}
+
+				if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM)
+					stereo_freqs |= sad->freq;
+				else
+					break;
+			}
+		}
+
+		value |= (stereo_freqs <<
+			AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES_STEREO__SHIFT);
+
+		WREG32_AUDIO_ENDPT(offset, eld_reg_to_type[i][0], value);
+	}
+
+	kfree(sads);
+}
+
+static void dce_v8_0_audio_enable(struct amdgpu_device *adev,
+				  struct amdgpu_audio_pin *pin,
+				  bool enable)
+{
+	if (!pin)
+		return;
+
+	WREG32_AUDIO_ENDPT(pin->offset, ixAZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL,
+		enable ? AZALIA_F0_CODEC_PIN_CONTROL_HOT_PLUG_CONTROL__AUDIO_ENABLED_MASK : 0);
+}
+
+static const u32 pin_offsets[7] =
+{
+	(0x1780 - 0x1780),
+	(0x1786 - 0x1780),
+	(0x178c - 0x1780),
+	(0x1792 - 0x1780),
+	(0x1798 - 0x1780),
+	(0x179d - 0x1780),
+	(0x17a4 - 0x1780),
+};
+
+static int dce_v8_0_audio_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!amdgpu_audio)
+		return 0;
+
+	adev->mode_info.audio.enabled = true;
+
+	if (adev->asic_type == CHIP_KAVERI) /* KV: 4 streams, 7 endpoints */
+		adev->mode_info.audio.num_pins = 7;
+	else if ((adev->asic_type == CHIP_KABINI) ||
+		 (adev->asic_type == CHIP_MULLINS)) /* KB/ML: 2 streams, 3 endpoints */
+		adev->mode_info.audio.num_pins = 3;
+	else if ((adev->asic_type == CHIP_BONAIRE) ||
+		 (adev->asic_type == CHIP_HAWAII))/* BN/HW: 6 streams, 7 endpoints */
+		adev->mode_info.audio.num_pins = 7;
+	else
+		adev->mode_info.audio.num_pins = 3;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		adev->mode_info.audio.pin[i].channels = -1;
+		adev->mode_info.audio.pin[i].rate = -1;
+		adev->mode_info.audio.pin[i].bits_per_sample = -1;
+		adev->mode_info.audio.pin[i].status_bits = 0;
+		adev->mode_info.audio.pin[i].category_code = 0;
+		adev->mode_info.audio.pin[i].connected = false;
+		adev->mode_info.audio.pin[i].offset = pin_offsets[i];
+		adev->mode_info.audio.pin[i].id = i;
+		/* disable audio.  it will be set up later */
+		/* XXX remove once we switch to ip funcs */
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	return 0;
+}
+
+static void dce_v8_0_audio_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	if (!adev->mode_info.audio.enabled)
+		return;
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++)
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+
+	adev->mode_info.audio.enabled = false;
+}
+
+/*
+ * update the N and CTS parameters for a given pixel clock rate
+ */
+static void dce_v8_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_afmt_acr acr = amdgpu_afmt_acr(clock);
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint32_t offset = dig->afmt->offset;
+
+	WREG32(mmHDMI_ACR_32_0 + offset, (acr.cts_32khz << HDMI_ACR_44_0__HDMI_ACR_CTS_44__SHIFT));
+	WREG32(mmHDMI_ACR_32_1 + offset, acr.n_32khz);
+
+	WREG32(mmHDMI_ACR_44_0 + offset, (acr.cts_44_1khz << HDMI_ACR_44_0__HDMI_ACR_CTS_44__SHIFT));
+	WREG32(mmHDMI_ACR_44_1 + offset, acr.n_44_1khz);
+
+	WREG32(mmHDMI_ACR_48_0 + offset, (acr.cts_48khz << HDMI_ACR_48_0__HDMI_ACR_CTS_48__SHIFT));
+	WREG32(mmHDMI_ACR_48_1 + offset, acr.n_48khz);
+}
+
+/*
+ * build a HDMI Video Info Frame
+ */
+static void dce_v8_0_afmt_update_avi_infoframe(struct drm_encoder *encoder,
+					       void *buffer, size_t size)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	uint32_t offset = dig->afmt->offset;
+	uint8_t *frame = buffer + 3;
+	uint8_t *header = buffer;
+
+	WREG32(mmAFMT_AVI_INFO0 + offset,
+		frame[0x0] | (frame[0x1] << 8) | (frame[0x2] << 16) | (frame[0x3] << 24));
+	WREG32(mmAFMT_AVI_INFO1 + offset,
+		frame[0x4] | (frame[0x5] << 8) | (frame[0x6] << 16) | (frame[0x7] << 24));
+	WREG32(mmAFMT_AVI_INFO2 + offset,
+		frame[0x8] | (frame[0x9] << 8) | (frame[0xA] << 16) | (frame[0xB] << 24));
+	WREG32(mmAFMT_AVI_INFO3 + offset,
+		frame[0xC] | (frame[0xD] << 8) | (header[1] << 24));
+}
+
+static void dce_v8_0_audio_set_dto(struct drm_encoder *encoder, u32 clock)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+	u32 dto_phase = 24 * 1000;
+	u32 dto_modulo = clock;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* XXX two dtos; generally use dto0 for hdmi */
+	/* Express [24MHz / target pixel clock] as an exact rational
+	 * number (coefficient of two integer numbers.  DCCG_AUDIO_DTOx_PHASE
+	 * is the numerator, DCCG_AUDIO_DTOx_MODULE is the denominator
+	 */
+	WREG32(mmDCCG_AUDIO_DTO_SOURCE, (amdgpu_crtc->crtc_id << DCCG_AUDIO_DTO_SOURCE__DCCG_AUDIO_DTO0_SOURCE_SEL__SHIFT));
+	WREG32(mmDCCG_AUDIO_DTO0_PHASE, dto_phase);
+	WREG32(mmDCCG_AUDIO_DTO0_MODULE, dto_modulo);
+}
+
+/*
+ * update the info frames with the data from the current display mode
+ */
+static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder,
+				  struct drm_display_mode *mode)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+	u8 buffer[HDMI_INFOFRAME_HEADER_SIZE + HDMI_AVI_INFOFRAME_SIZE];
+	struct hdmi_avi_infoframe frame;
+	uint32_t offset, val;
+	ssize_t err;
+	int bpc = 8;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (!dig->afmt->enabled)
+		return;
+	offset = dig->afmt->offset;
+
+	/* hdmi deep color mode general control packets setup, if bpc > 8 */
+	if (encoder->crtc) {
+		struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(encoder->crtc);
+		bpc = amdgpu_crtc->bpc;
+	}
+
+	/* disable audio prior to setting up hw */
+	dig->afmt->pin = dce_v8_0_audio_get_pin(adev);
+	dce_v8_0_audio_enable(adev, dig->afmt->pin, false);
+
+	dce_v8_0_audio_set_dto(encoder, mode->clock);
+
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + offset,
+	       HDMI_VBI_PACKET_CONTROL__HDMI_NULL_SEND_MASK); /* send null packets when required */
+
+	WREG32(mmAFMT_AUDIO_CRC_CONTROL + offset, 0x1000);
+
+	val = RREG32(mmHDMI_CONTROL + offset);
+	val &= ~HDMI_CONTROL__HDMI_DEEP_COLOR_ENABLE_MASK;
+	val &= ~HDMI_CONTROL__HDMI_DEEP_COLOR_DEPTH_MASK;
+
+	switch (bpc) {
+	case 0:
+	case 6:
+	case 8:
+	case 16:
+	default:
+		DRM_DEBUG("%s: Disabling hdmi deep color for %d bpc.\n",
+			  connector->name, bpc);
+		break;
+	case 10:
+		val |= HDMI_CONTROL__HDMI_DEEP_COLOR_ENABLE_MASK;
+		val |= 1 << HDMI_CONTROL__HDMI_DEEP_COLOR_DEPTH__SHIFT;
+		DRM_DEBUG("%s: Enabling hdmi deep color 30 for 10 bpc.\n",
+			  connector->name);
+		break;
+	case 12:
+		val |= HDMI_CONTROL__HDMI_DEEP_COLOR_ENABLE_MASK;
+		val |= 2 << HDMI_CONTROL__HDMI_DEEP_COLOR_DEPTH__SHIFT;
+		DRM_DEBUG("%s: Enabling hdmi deep color 36 for 12 bpc.\n",
+			  connector->name);
+		break;
+	}
+
+	WREG32(mmHDMI_CONTROL + offset, val);
+
+	WREG32(mmHDMI_VBI_PACKET_CONTROL + offset,
+	       HDMI_VBI_PACKET_CONTROL__HDMI_NULL_SEND_MASK | /* send null packets when required */
+	       HDMI_VBI_PACKET_CONTROL__HDMI_GC_SEND_MASK | /* send general control packets */
+	       HDMI_VBI_PACKET_CONTROL__HDMI_GC_CONT_MASK); /* send general control packets every frame */
+
+	WREG32(mmHDMI_INFOFRAME_CONTROL0 + offset,
+	       HDMI_INFOFRAME_CONTROL0__HDMI_AUDIO_INFO_SEND_MASK | /* enable audio info frames (frames won't be set until audio is enabled) */
+	       HDMI_INFOFRAME_CONTROL0__HDMI_AUDIO_INFO_CONT_MASK); /* required for audio info values to be updated */
+
+	WREG32(mmAFMT_INFOFRAME_CONTROL0 + offset,
+	       AFMT_INFOFRAME_CONTROL0__AFMT_AUDIO_INFO_UPDATE_MASK); /* required for audio info values to be updated */
+
+	WREG32(mmHDMI_INFOFRAME_CONTROL1 + offset,
+	       (2 << HDMI_INFOFRAME_CONTROL1__HDMI_AUDIO_INFO_LINE__SHIFT)); /* anything other than 0 */
+
+	WREG32(mmHDMI_GC + offset, 0); /* unset HDMI_GC_AVMUTE */
+
+	WREG32(mmHDMI_AUDIO_PACKET_CONTROL + offset,
+	       (1 << HDMI_AUDIO_PACKET_CONTROL__HDMI_AUDIO_DELAY_EN__SHIFT) | /* set the default audio delay */
+	       (3 << HDMI_AUDIO_PACKET_CONTROL__HDMI_AUDIO_PACKETS_PER_LINE__SHIFT)); /* should be suffient for all audio modes and small enough for all hblanks */
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL + offset,
+	       AFMT_AUDIO_PACKET_CONTROL__AFMT_60958_CS_UPDATE_MASK); /* allow 60958 channel status fields to be updated */
+
+	/* fglrx clears sth in AFMT_AUDIO_PACKET_CONTROL2 here */
+
+	if (bpc > 8)
+		WREG32(mmHDMI_ACR_PACKET_CONTROL + offset,
+		       HDMI_ACR_PACKET_CONTROL__HDMI_ACR_AUTO_SEND_MASK); /* allow hw to sent ACR packets when required */
+	else
+		WREG32(mmHDMI_ACR_PACKET_CONTROL + offset,
+		       HDMI_ACR_PACKET_CONTROL__HDMI_ACR_SOURCE_MASK | /* select SW CTS value */
+		       HDMI_ACR_PACKET_CONTROL__HDMI_ACR_AUTO_SEND_MASK); /* allow hw to sent ACR packets when required */
+
+	dce_v8_0_afmt_update_ACR(encoder, mode->clock);
+
+	WREG32(mmAFMT_60958_0 + offset,
+	       (1 << AFMT_60958_0__AFMT_60958_CS_CHANNEL_NUMBER_L__SHIFT));
+
+	WREG32(mmAFMT_60958_1 + offset,
+	       (2 << AFMT_60958_1__AFMT_60958_CS_CHANNEL_NUMBER_R__SHIFT));
+
+	WREG32(mmAFMT_60958_2 + offset,
+	       (3 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_2__SHIFT) |
+	       (4 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_3__SHIFT) |
+	       (5 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_4__SHIFT) |
+	       (6 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_5__SHIFT) |
+	       (7 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_6__SHIFT) |
+	       (8 << AFMT_60958_2__AFMT_60958_CS_CHANNEL_NUMBER_7__SHIFT));
+
+	dce_v8_0_audio_write_speaker_allocation(encoder);
+
+
+	WREG32(mmAFMT_AUDIO_PACKET_CONTROL2 + offset,
+	       (0xff << AFMT_AUDIO_PACKET_CONTROL2__AFMT_AUDIO_CHANNEL_ENABLE__SHIFT));
+
+	dce_v8_0_afmt_audio_select_pin(encoder);
+	dce_v8_0_audio_write_sad_regs(encoder);
+	dce_v8_0_audio_write_latency_fields(encoder, mode);
+
+	err = drm_hdmi_avi_infoframe_from_display_mode(&frame, mode);
+	if (err < 0) {
+		DRM_ERROR("failed to setup AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	err = hdmi_avi_infoframe_pack(&frame, buffer, sizeof(buffer));
+	if (err < 0) {
+		DRM_ERROR("failed to pack AVI infoframe: %zd\n", err);
+		return;
+	}
+
+	dce_v8_0_afmt_update_avi_infoframe(encoder, buffer, sizeof(buffer));
+
+	WREG32_OR(mmHDMI_INFOFRAME_CONTROL0 + offset,
+		  HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_SEND_MASK | /* enable AVI info frames */
+		  HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_SEND_MASK); /* required for audio info values to be updated */
+
+	WREG32_P(mmHDMI_INFOFRAME_CONTROL1 + offset,
+		 (2 << HDMI_INFOFRAME_CONTROL1__HDMI_AVI_INFO_LINE__SHIFT), /* anything other than 0 */
+		 ~HDMI_INFOFRAME_CONTROL1__HDMI_AVI_INFO_LINE_MASK);
+
+	WREG32_OR(mmAFMT_AUDIO_PACKET_CONTROL + offset,
+		  AFMT_AUDIO_PACKET_CONTROL__AFMT_AUDIO_SAMPLE_SEND_MASK); /* send audio packets */
+
+	/* it's unknown what these bits do excatly, but it's indeed quite useful for debugging */
+	WREG32(mmAFMT_RAMP_CONTROL0 + offset, 0x00FFFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL1 + offset, 0x007FFFFF);
+	WREG32(mmAFMT_RAMP_CONTROL2 + offset, 0x00000001);
+	WREG32(mmAFMT_RAMP_CONTROL3 + offset, 0x00000001);
+
+	/* enable audio after to setting up hw */
+	dce_v8_0_audio_enable(adev, dig->afmt->pin, true);
+}
+
+static void dce_v8_0_afmt_enable(struct drm_encoder *encoder, bool enable)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	if (!dig || !dig->afmt)
+		return;
+
+	/* Silent, r600_hdmi_enable will raise WARN for us */
+	if (enable && dig->afmt->enabled)
+		return;
+	if (!enable && !dig->afmt->enabled)
+		return;
+
+	if (!enable && dig->afmt->pin) {
+		dce_v8_0_audio_enable(adev, dig->afmt->pin, false);
+		dig->afmt->pin = NULL;
+	}
+
+	dig->afmt->enabled = enable;
+
+	DRM_DEBUG("%sabling AFMT interface @ 0x%04X for encoder 0x%x\n",
+		  enable ? "En" : "Dis", dig->afmt->offset, amdgpu_encoder->encoder_id);
+}
+
+static void dce_v8_0_afmt_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++)
+		adev->mode_info.afmt[i] = NULL;
+
+	/* DCE8 has audio blocks tied to DIG encoders */
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		adev->mode_info.afmt[i] = kzalloc(sizeof(struct amdgpu_afmt), GFP_KERNEL);
+		if (adev->mode_info.afmt[i]) {
+			adev->mode_info.afmt[i]->offset = dig_offsets[i];
+			adev->mode_info.afmt[i]->id = i;
+		}
+	}
+}
+
+static void dce_v8_0_afmt_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->mode_info.num_dig; i++) {
+		kfree(adev->mode_info.afmt[i]);
+		adev->mode_info.afmt[i] = NULL;
+	}
+}
+
+static const u32 vga_control_regs[6] =
+{
+	mmD1VGA_CONTROL,
+	mmD2VGA_CONTROL,
+	mmD3VGA_CONTROL,
+	mmD4VGA_CONTROL,
+	mmD5VGA_CONTROL,
+	mmD6VGA_CONTROL,
+};
+
+static void dce_v8_0_vga_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 vga_control;
+
+	vga_control = RREG32(vga_control_regs[amdgpu_crtc->crtc_id]) & ~1;
+	if (enable)
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control | 1);
+	else
+		WREG32(vga_control_regs[amdgpu_crtc->crtc_id], vga_control);
+}
+
+static void dce_v8_0_grph_enable(struct drm_crtc *crtc, bool enable)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	if (enable)
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 1);
+	else
+		WREG32(mmGRPH_ENABLE + amdgpu_crtc->crtc_offset, 0);
+}
+
+static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc,
+				     struct drm_framebuffer *fb,
+				     int x, int y, int atomic)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_framebuffer *amdgpu_fb;
+	struct drm_framebuffer *target_fb;
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *rbo;
+	uint64_t fb_location, tiling_flags;
+	uint32_t fb_format, fb_pitch_pixels;
+	u32 fb_swap = (GRPH_ENDIAN_NONE << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+	u32 pipe_config;
+	u32 tmp, viewport_w, viewport_h;
+	int r;
+	bool bypass_lut = false;
+
+	/* no fb bound */
+	if (!atomic && !crtc->primary->fb) {
+		DRM_DEBUG_KMS("No FB bound\n");
+		return 0;
+	}
+
+	if (atomic) {
+		amdgpu_fb = to_amdgpu_framebuffer(fb);
+		target_fb = fb;
+	}
+	else {
+		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+		target_fb = crtc->primary->fb;
+	}
+
+	/* If atomic, assume fb object is pinned & idle & fenced and
+	 * just update base pointers
+	 */
+	obj = amdgpu_fb->obj;
+	rbo = gem_to_amdgpu_bo(obj);
+	r = amdgpu_bo_reserve(rbo, false);
+	if (unlikely(r != 0))
+		return r;
+
+	if (atomic)
+		fb_location = amdgpu_bo_gpu_offset(rbo);
+	else {
+		r = amdgpu_bo_pin(rbo, AMDGPU_GEM_DOMAIN_VRAM, &fb_location);
+		if (unlikely(r != 0)) {
+			amdgpu_bo_unreserve(rbo);
+			return -EINVAL;
+		}
+	}
+
+	amdgpu_bo_get_tiling_flags(rbo, &tiling_flags);
+	amdgpu_bo_unreserve(rbo);
+
+	pipe_config = AMDGPU_TILING_GET(tiling_flags, PIPE_CONFIG);
+
+	switch (target_fb->pixel_format) {
+	case DRM_FORMAT_C8:
+		fb_format = ((GRPH_DEPTH_8BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_INDEXED << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+		break;
+	case DRM_FORMAT_XRGB4444:
+	case DRM_FORMAT_ARGB4444:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB1555 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_XRGB1555:
+	case DRM_FORMAT_ARGB1555:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB1555 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_BGRX5551:
+	case DRM_FORMAT_BGRA5551:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA5551 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_RGB565:
+		fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB565 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_XRGB8888:
+	case DRM_FORMAT_ARGB8888:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB8888 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		break;
+	case DRM_FORMAT_XRGB2101010:
+	case DRM_FORMAT_ARGB2101010:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_ARGB2101010 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	case DRM_FORMAT_BGRX1010102:
+	case DRM_FORMAT_BGRA1010102:
+		fb_format = ((GRPH_DEPTH_32BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) |
+			     (GRPH_FORMAT_BGRA1010102 << GRPH_CONTROL__GRPH_FORMAT__SHIFT));
+#ifdef __BIG_ENDIAN
+		fb_swap = (GRPH_ENDIAN_8IN32 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT);
+#endif
+		/* Greater 8 bpc fb needs to bypass hw-lut to retain precision */
+		bypass_lut = true;
+		break;
+	default:
+		DRM_ERROR("Unsupported screen format %s\n",
+			  drm_get_format_name(target_fb->pixel_format));
+		return -EINVAL;
+	}
+
+	if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_2D_TILED_THIN1) {
+		unsigned bankw, bankh, mtaspect, tile_split, num_banks;
+
+		bankw = AMDGPU_TILING_GET(tiling_flags, BANK_WIDTH);
+		bankh = AMDGPU_TILING_GET(tiling_flags, BANK_HEIGHT);
+		mtaspect = AMDGPU_TILING_GET(tiling_flags, MACRO_TILE_ASPECT);
+		tile_split = AMDGPU_TILING_GET(tiling_flags, TILE_SPLIT);
+		num_banks = AMDGPU_TILING_GET(tiling_flags, NUM_BANKS);
+
+		fb_format |= (num_banks << GRPH_CONTROL__GRPH_NUM_BANKS__SHIFT);
+		fb_format |= (GRPH_ARRAY_2D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
+		fb_format |= (tile_split << GRPH_CONTROL__GRPH_TILE_SPLIT__SHIFT);
+		fb_format |= (bankw << GRPH_CONTROL__GRPH_BANK_WIDTH__SHIFT);
+		fb_format |= (bankh << GRPH_CONTROL__GRPH_BANK_HEIGHT__SHIFT);
+		fb_format |= (mtaspect << GRPH_CONTROL__GRPH_MACRO_TILE_ASPECT__SHIFT);
+		fb_format |= (DISPLAY_MICRO_TILING << GRPH_CONTROL__GRPH_MICRO_TILE_MODE__SHIFT);
+	} else if (AMDGPU_TILING_GET(tiling_flags, ARRAY_MODE) == ARRAY_1D_TILED_THIN1) {
+		fb_format |= (GRPH_ARRAY_1D_TILED_THIN1 << GRPH_CONTROL__GRPH_ARRAY_MODE__SHIFT);
+	}
+
+	fb_format |= (pipe_config << GRPH_CONTROL__GRPH_PIPE_CONFIG__SHIFT);
+
+	dce_v8_0_vga_enable(crtc, false);
+
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(fb_location));
+	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32)fb_location & GRPH_PRIMARY_SURFACE_ADDRESS__GRPH_PRIMARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       (u32) fb_location & GRPH_SECONDARY_SURFACE_ADDRESS__GRPH_SECONDARY_SURFACE_ADDRESS_MASK);
+	WREG32(mmGRPH_CONTROL + amdgpu_crtc->crtc_offset, fb_format);
+	WREG32(mmGRPH_SWAP_CNTL + amdgpu_crtc->crtc_offset, fb_swap);
+
+	/*
+	 * The LUT only has 256 slots for indexing by a 8 bpc fb. Bypass the LUT
+	 * for > 8 bpc scanout to avoid truncation of fb indices to 8 msb's, to
+	 * retain the full precision throughout the pipeline.
+	 */
+	WREG32_P(mmGRPH_LUT_10BIT_BYPASS_CONTROL + amdgpu_crtc->crtc_offset,
+		 (bypass_lut ? LUT_10BIT_BYPASS_EN : 0),
+		 ~LUT_10BIT_BYPASS_EN);
+
+	if (bypass_lut)
+		DRM_DEBUG_KMS("Bypassing hardware LUT due to 10 bit fb scanout.\n");
+
+	WREG32(mmGRPH_SURFACE_OFFSET_X + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_SURFACE_OFFSET_Y + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_Y_START + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmGRPH_X_END + amdgpu_crtc->crtc_offset, target_fb->width);
+	WREG32(mmGRPH_Y_END + amdgpu_crtc->crtc_offset, target_fb->height);
+
+	fb_pitch_pixels = target_fb->pitches[0] / (target_fb->bits_per_pixel / 8);
+	WREG32(mmGRPH_PITCH + amdgpu_crtc->crtc_offset, fb_pitch_pixels);
+
+	dce_v8_0_grph_enable(crtc, true);
+
+	WREG32(mmLB_DESKTOP_HEIGHT + amdgpu_crtc->crtc_offset,
+	       target_fb->height);
+
+	x &= ~3;
+	y &= ~1;
+	WREG32(mmVIEWPORT_START + amdgpu_crtc->crtc_offset,
+	       (x << 16) | y);
+	viewport_w = crtc->mode.hdisplay;
+	viewport_h = (crtc->mode.vdisplay + 1) & ~1;
+	WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset,
+	       (viewport_w << 16) | viewport_h);
+
+	/* pageflip setup */
+	/* make sure flip is at vb rather than hb */
+	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
+	tmp &= ~GRPH_FLIP_CONTROL__GRPH_SURFACE_UPDATE_H_RETRACE_EN_MASK;
+	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
+
+	/* set pageflip to happen only at start of vblank interval (front porch) */
+	WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 3);
+
+	if (!atomic && fb && fb != crtc->primary->fb) {
+		amdgpu_fb = to_amdgpu_framebuffer(fb);
+		rbo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		r = amdgpu_bo_reserve(rbo, false);
+		if (unlikely(r != 0))
+			return r;
+		amdgpu_bo_unpin(rbo);
+		amdgpu_bo_unreserve(rbo);
+	}
+
+	/* Bytes per pixel may have changed */
+	dce_v8_0_bandwidth_update(adev);
+
+	return 0;
+}
+
+static void dce_v8_0_set_interleave(struct drm_crtc *crtc,
+				    struct drm_display_mode *mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (mode->flags & DRM_MODE_FLAG_INTERLACE)
+		WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset,
+		       LB_DATA_FORMAT__INTERLEAVE_EN__SHIFT);
+	else
+		WREG32(mmLB_DATA_FORMAT + amdgpu_crtc->crtc_offset, 0);
+}
+
+static void dce_v8_0_crtc_load_lut(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	int i;
+
+	DRM_DEBUG_KMS("%d\n", amdgpu_crtc->crtc_id);
+
+	WREG32(mmINPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_GRPH_MODE__SHIFT) |
+		(INPUT_CSC_BYPASS << INPUT_CSC_CONTROL__INPUT_CSC_OVL_MODE__SHIFT)));
+	WREG32(mmPRESCALE_GRPH_CONTROL + amdgpu_crtc->crtc_offset,
+	       PRESCALE_GRPH_CONTROL__GRPH_PRESCALE_BYPASS_MASK);
+	WREG32(mmPRESCALE_OVL_CONTROL + amdgpu_crtc->crtc_offset,
+	       PRESCALE_OVL_CONTROL__OVL_PRESCALE_BYPASS_MASK);
+	WREG32(mmINPUT_GAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__GRPH_INPUT_GAMMA_MODE__SHIFT) |
+		(INPUT_GAMMA_USE_LUT << INPUT_GAMMA_CONTROL__OVL_INPUT_GAMMA_MODE__SHIFT)));
+
+	WREG32(mmDC_LUT_CONTROL + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_BLACK_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_BLACK_OFFSET_RED + amdgpu_crtc->crtc_offset, 0);
+
+	WREG32(mmDC_LUT_WHITE_OFFSET_BLUE + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_GREEN + amdgpu_crtc->crtc_offset, 0xffff);
+	WREG32(mmDC_LUT_WHITE_OFFSET_RED + amdgpu_crtc->crtc_offset, 0xffff);
+
+	WREG32(mmDC_LUT_RW_MODE + amdgpu_crtc->crtc_offset, 0);
+	WREG32(mmDC_LUT_WRITE_EN_MASK + amdgpu_crtc->crtc_offset, 0x00000007);
+
+	WREG32(mmDC_LUT_RW_INDEX + amdgpu_crtc->crtc_offset, 0);
+	for (i = 0; i < 256; i++) {
+		WREG32(mmDC_LUT_30_COLOR + amdgpu_crtc->crtc_offset,
+		       (amdgpu_crtc->lut_r[i] << 20) |
+		       (amdgpu_crtc->lut_g[i] << 10) |
+		       (amdgpu_crtc->lut_b[i] << 0));
+	}
+
+	WREG32(mmDEGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((DEGAMMA_BYPASS << DEGAMMA_CONTROL__GRPH_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__OVL_DEGAMMA_MODE__SHIFT) |
+		(DEGAMMA_BYPASS << DEGAMMA_CONTROL__CURSOR_DEGAMMA_MODE__SHIFT)));
+	WREG32(mmGAMUT_REMAP_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__GRPH_GAMUT_REMAP_MODE__SHIFT) |
+		(GAMUT_REMAP_BYPASS << GAMUT_REMAP_CONTROL__OVL_GAMUT_REMAP_MODE__SHIFT)));
+	WREG32(mmREGAMMA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((REGAMMA_BYPASS << REGAMMA_CONTROL__GRPH_REGAMMA_MODE__SHIFT) |
+		(REGAMMA_BYPASS << REGAMMA_CONTROL__OVL_REGAMMA_MODE__SHIFT)));
+	WREG32(mmOUTPUT_CSC_CONTROL + amdgpu_crtc->crtc_offset,
+	       ((OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_GRPH_MODE__SHIFT) |
+		(OUTPUT_CSC_BYPASS << OUTPUT_CSC_CONTROL__OUTPUT_CSC_OVL_MODE__SHIFT)));
+	/* XXX match this to the depth of the crtc fmt block, move to modeset? */
+	WREG32(0x1a50 + amdgpu_crtc->crtc_offset, 0);
+	/* XXX this only needs to be programmed once per crtc at startup,
+	 * not sure where the best place for it is
+	 */
+	WREG32(mmALPHA_CONTROL + amdgpu_crtc->crtc_offset,
+	       ALPHA_CONTROL__CURSOR_ALPHA_BLND_ENA_MASK);
+}
+
+static int dce_v8_0_pick_dig_encoder(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		if (dig->linkb)
+			return 1;
+		else
+			return 0;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		if (dig->linkb)
+			return 3;
+		else
+			return 2;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		if (dig->linkb)
+			return 5;
+		else
+			return 4;
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		return 6;
+		break;
+	default:
+		DRM_ERROR("invalid encoder_id: 0x%x\n", amdgpu_encoder->encoder_id);
+		return 0;
+	}
+}
+
+/**
+ * dce_v8_0_pick_pll - Allocate a PPLL for use by the crtc.
+ *
+ * @crtc: drm crtc
+ *
+ * Returns the PPLL (Pixel PLL) to be used by the crtc.  For DP monitors
+ * a single PPLL can be used for all DP crtcs/encoders.  For non-DP
+ * monitors a dedicated PPLL must be used.  If a particular board has
+ * an external DP PLL, return ATOM_PPLL_INVALID to skip PLL programming
+ * as there is no need to program the PLL itself.  If we are not able to
+ * allocate a PLL, return ATOM_PPLL_INVALID to skip PLL programming to
+ * avoid messing up an existing monitor.
+ *
+ * Asic specific PLL information
+ *
+ * DCE 8.x
+ * KB/KV
+ * - PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP)
+ * CI
+ * - PPLL0, PPLL1, PPLL2 are available for all UNIPHY (both DP and non-DP) and DAC
+ *
+ */
+static u32 dce_v8_0_pick_pll(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	u32 pll_in_use;
+	int pll;
+
+	if (ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder))) {
+		if (adev->clock.dp_extclk)
+			/* skip PPLL programming if using ext clock */
+			return ATOM_PPLL_INVALID;
+		else {
+			/* use the same PPLL for all DP monitors */
+			pll = amdgpu_pll_get_shared_dp_ppll(crtc);
+			if (pll != ATOM_PPLL_INVALID)
+				return pll;
+		}
+	} else {
+		/* use the same PPLL for all monitors with the same clock */
+		pll = amdgpu_pll_get_shared_nondp_ppll(crtc);
+		if (pll != ATOM_PPLL_INVALID)
+			return pll;
+	}
+	/* otherwise, pick one of the plls */
+	if ((adev->asic_type == CHIP_KABINI) ||
+	    (adev->asic_type == CHIP_MULLINS)) {
+		/* KB/ML has PPLL1 and PPLL2 */
+		pll_in_use = amdgpu_pll_get_use_mask(crtc);
+		if (!(pll_in_use & (1 << ATOM_PPLL2)))
+			return ATOM_PPLL2;
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	} else {
+		/* CI/KV has PPLL0, PPLL1, and PPLL2 */
+		pll_in_use = amdgpu_pll_get_use_mask(crtc);
+		if (!(pll_in_use & (1 << ATOM_PPLL2)))
+			return ATOM_PPLL2;
+		if (!(pll_in_use & (1 << ATOM_PPLL1)))
+			return ATOM_PPLL1;
+		if (!(pll_in_use & (1 << ATOM_PPLL0)))
+			return ATOM_PPLL0;
+		DRM_ERROR("unable to allocate a PPLL\n");
+		return ATOM_PPLL_INVALID;
+	}
+	return ATOM_PPLL_INVALID;
+}
+
+static void dce_v8_0_lock_cursor(struct drm_crtc *crtc, bool lock)
+{
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	uint32_t cur_lock;
+
+	cur_lock = RREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset);
+	if (lock)
+		cur_lock |= CUR_UPDATE__CURSOR_UPDATE_LOCK_MASK;
+	else
+		cur_lock &= ~CUR_UPDATE__CURSOR_UPDATE_LOCK_MASK;
+	WREG32(mmCUR_UPDATE + amdgpu_crtc->crtc_offset, cur_lock);
+}
+
+static void dce_v8_0_hide_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+}
+
+static void dce_v8_0_show_cursor(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+
+	WREG32(mmCUR_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
+	       upper_32_bits(amdgpu_crtc->cursor_addr));
+	WREG32(mmCUR_SURFACE_ADDRESS + amdgpu_crtc->crtc_offset,
+	       lower_32_bits(amdgpu_crtc->cursor_addr));
+
+	WREG32_IDX(mmCUR_CONTROL + amdgpu_crtc->crtc_offset,
+		   CUR_CONTROL__CURSOR_EN_MASK |
+		   (CURSOR_24_8_PRE_MULT << CUR_CONTROL__CURSOR_MODE__SHIFT) |
+		   (CURSOR_URGENT_1_2 << CUR_CONTROL__CURSOR_URGENT_CONTROL__SHIFT));
+}
+
+static int dce_v8_0_cursor_move_locked(struct drm_crtc *crtc,
+				       int x, int y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct amdgpu_device *adev = crtc->dev->dev_private;
+	int xorigin = 0, yorigin = 0;
+
+	/* avivo cursor are offset into the total surface */
+	x += crtc->x;
+	y += crtc->y;
+	DRM_DEBUG("x %d y %d c->x %d c->y %d\n", x, y, crtc->x, crtc->y);
+
+	if (x < 0) {
+		xorigin = min(-x, amdgpu_crtc->max_cursor_width - 1);
+		x = 0;
+	}
+	if (y < 0) {
+		yorigin = min(-y, amdgpu_crtc->max_cursor_height - 1);
+		y = 0;
+	}
+
+	WREG32(mmCUR_POSITION + amdgpu_crtc->crtc_offset, (x << 16) | y);
+	WREG32(mmCUR_HOT_SPOT + amdgpu_crtc->crtc_offset, (xorigin << 16) | yorigin);
+	WREG32(mmCUR_SIZE + amdgpu_crtc->crtc_offset,
+	       ((amdgpu_crtc->cursor_width - 1) << 16) | (amdgpu_crtc->cursor_height - 1));
+
+	amdgpu_crtc->cursor_x = x;
+	amdgpu_crtc->cursor_y = y;
+
+	return 0;
+}
+
+static int dce_v8_0_crtc_cursor_move(struct drm_crtc *crtc,
+				     int x, int y)
+{
+	int ret;
+
+	dce_v8_0_lock_cursor(crtc, true);
+	ret = dce_v8_0_cursor_move_locked(crtc, x, y);
+	dce_v8_0_lock_cursor(crtc, false);
+
+	return ret;
+}
+
+static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
+				     struct drm_file *file_priv,
+				     uint32_t handle,
+				     uint32_t width,
+				     uint32_t height,
+				     int32_t hot_x,
+				     int32_t hot_y)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_gem_object *obj;
+	struct amdgpu_bo *aobj;
+	int ret;
+
+	if (!handle) {
+		/* turn off cursor */
+		dce_v8_0_hide_cursor(crtc);
+		obj = NULL;
+		goto unpin;
+	}
+
+	if ((width > amdgpu_crtc->max_cursor_width) ||
+	    (height > amdgpu_crtc->max_cursor_height)) {
+		DRM_ERROR("bad cursor width or height %d x %d\n", width, height);
+		return -EINVAL;
+	}
+
+	obj = drm_gem_object_lookup(crtc->dev, file_priv, handle);
+	if (!obj) {
+		DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, amdgpu_crtc->crtc_id);
+		return -ENOENT;
+	}
+
+	aobj = gem_to_amdgpu_bo(obj);
+	ret = amdgpu_bo_reserve(aobj, false);
+	if (ret != 0) {
+		drm_gem_object_unreference_unlocked(obj);
+		return ret;
+	}
+
+	ret = amdgpu_bo_pin(aobj, AMDGPU_GEM_DOMAIN_VRAM, &amdgpu_crtc->cursor_addr);
+	amdgpu_bo_unreserve(aobj);
+	if (ret) {
+		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
+		drm_gem_object_unreference_unlocked(obj);
+		return ret;
+	}
+
+	amdgpu_crtc->cursor_width = width;
+	amdgpu_crtc->cursor_height = height;
+
+	dce_v8_0_lock_cursor(crtc, true);
+
+	if (hot_x != amdgpu_crtc->cursor_hot_x ||
+	    hot_y != amdgpu_crtc->cursor_hot_y) {
+		int x, y;
+
+		x = amdgpu_crtc->cursor_x + amdgpu_crtc->cursor_hot_x - hot_x;
+		y = amdgpu_crtc->cursor_y + amdgpu_crtc->cursor_hot_y - hot_y;
+
+		dce_v8_0_cursor_move_locked(crtc, x, y);
+
+		amdgpu_crtc->cursor_hot_x = hot_x;
+		amdgpu_crtc->cursor_hot_y = hot_y;
+	}
+
+	dce_v8_0_show_cursor(crtc);
+	dce_v8_0_lock_cursor(crtc, false);
+
+unpin:
+	if (amdgpu_crtc->cursor_bo) {
+		struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
+		ret = amdgpu_bo_reserve(aobj, false);
+		if (likely(ret == 0)) {
+			amdgpu_bo_unpin(aobj);
+			amdgpu_bo_unreserve(aobj);
+		}
+		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+	}
+
+	amdgpu_crtc->cursor_bo = obj;
+	return 0;
+}
+
+static void dce_v8_0_cursor_reset(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (amdgpu_crtc->cursor_bo) {
+		dce_v8_0_lock_cursor(crtc, true);
+
+		dce_v8_0_cursor_move_locked(crtc, amdgpu_crtc->cursor_x,
+					    amdgpu_crtc->cursor_y);
+
+		dce_v8_0_show_cursor(crtc);
+
+		dce_v8_0_lock_cursor(crtc, false);
+	}
+}
+
+static void dce_v8_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t start, uint32_t size)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	int end = (start + size > 256) ? 256 : start + size, i;
+
+	/* userspace palettes are always correct as is */
+	for (i = start; i < end; i++) {
+		amdgpu_crtc->lut_r[i] = red[i] >> 6;
+		amdgpu_crtc->lut_g[i] = green[i] >> 6;
+		amdgpu_crtc->lut_b[i] = blue[i] >> 6;
+	}
+	dce_v8_0_crtc_load_lut(crtc);
+}
+
+static void dce_v8_0_crtc_destroy(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	drm_crtc_cleanup(crtc);
+	destroy_workqueue(amdgpu_crtc->pflip_queue);
+	kfree(amdgpu_crtc);
+}
+
+static const struct drm_crtc_funcs dce_v8_0_crtc_funcs = {
+	.cursor_set2 = dce_v8_0_crtc_cursor_set2,
+	.cursor_move = dce_v8_0_crtc_cursor_move,
+	.gamma_set = dce_v8_0_crtc_gamma_set,
+	.set_config = amdgpu_crtc_set_config,
+	.destroy = dce_v8_0_crtc_destroy,
+	.page_flip = amdgpu_crtc_page_flip,
+};
+
+static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
+{
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	unsigned type;
+
+	switch (mode) {
+	case DRM_MODE_DPMS_ON:
+		amdgpu_crtc->enabled = true;
+		amdgpu_atombios_crtc_enable(crtc, ATOM_ENABLE);
+		dce_v8_0_vga_enable(crtc, true);
+		amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
+		dce_v8_0_vga_enable(crtc, false);
+		/* Make sure VBLANK and PFLIP interrupts are still enabled */
+		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+		amdgpu_irq_update(adev, &adev->crtc_irq, type);
+		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
+		drm_vblank_post_modeset(dev, amdgpu_crtc->crtc_id);
+		dce_v8_0_crtc_load_lut(crtc);
+		break;
+	case DRM_MODE_DPMS_STANDBY:
+	case DRM_MODE_DPMS_SUSPEND:
+	case DRM_MODE_DPMS_OFF:
+		drm_vblank_pre_modeset(dev, amdgpu_crtc->crtc_id);
+		if (amdgpu_crtc->enabled) {
+			dce_v8_0_vga_enable(crtc, true);
+			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
+			dce_v8_0_vga_enable(crtc, false);
+		}
+		amdgpu_atombios_crtc_enable(crtc, ATOM_DISABLE);
+		amdgpu_crtc->enabled = false;
+		break;
+	}
+	/* adjust pm to dpms */
+	amdgpu_pm_compute_clocks(adev);
+}
+
+static void dce_v8_0_crtc_prepare(struct drm_crtc *crtc)
+{
+	/* disable crtc pair power gating before programming */
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_DISABLE);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_ENABLE);
+	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+}
+
+static void dce_v8_0_crtc_commit(struct drm_crtc *crtc)
+{
+	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_crtc_lock(crtc, ATOM_DISABLE);
+}
+
+static void dce_v8_0_crtc_disable(struct drm_crtc *crtc)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct amdgpu_atom_ss ss;
+	int i;
+
+	dce_v8_0_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
+	if (crtc->primary->fb) {
+		int r;
+		struct amdgpu_framebuffer *amdgpu_fb;
+		struct amdgpu_bo *rbo;
+
+		amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb);
+		rbo = gem_to_amdgpu_bo(amdgpu_fb->obj);
+		r = amdgpu_bo_reserve(rbo, false);
+		if (unlikely(r))
+			DRM_ERROR("failed to reserve rbo before unpin\n");
+		else {
+			amdgpu_bo_unpin(rbo);
+			amdgpu_bo_unreserve(rbo);
+		}
+	}
+	/* disable the GRPH */
+	dce_v8_0_grph_enable(crtc, false);
+
+	amdgpu_atombios_crtc_powergate(crtc, ATOM_ENABLE);
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		if (adev->mode_info.crtcs[i] &&
+		    adev->mode_info.crtcs[i]->enabled &&
+		    i != amdgpu_crtc->crtc_id &&
+		    amdgpu_crtc->pll_id == adev->mode_info.crtcs[i]->pll_id) {
+			/* one other crtc is using this pll don't turn
+			 * off the pll
+			 */
+			goto done;
+		}
+	}
+
+	switch (amdgpu_crtc->pll_id) {
+	case ATOM_PPLL1:
+	case ATOM_PPLL2:
+		/* disable the ppll */
+		amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+					  0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	case ATOM_PPLL0:
+		/* disable the ppll */
+		if ((adev->asic_type == CHIP_KAVERI) ||
+		    (adev->asic_type == CHIP_BONAIRE) ||
+		    (adev->asic_type == CHIP_HAWAII))
+			amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id,
+						  0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss);
+		break;
+	default:
+		break;
+	}
+done:
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+}
+
+static int dce_v8_0_crtc_mode_set(struct drm_crtc *crtc,
+				  struct drm_display_mode *mode,
+				  struct drm_display_mode *adjusted_mode,
+				  int x, int y, struct drm_framebuffer *old_fb)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+
+	if (!amdgpu_crtc->adjusted_clock)
+		return -EINVAL;
+
+	amdgpu_atombios_crtc_set_pll(crtc, adjusted_mode);
+	amdgpu_atombios_crtc_set_dtd_timing(crtc, adjusted_mode);
+	dce_v8_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+	amdgpu_atombios_crtc_overscan_setup(crtc, mode, adjusted_mode);
+	amdgpu_atombios_crtc_scaler_setup(crtc);
+	dce_v8_0_cursor_reset(crtc);
+	/* update the hw version fpr dpm */
+	amdgpu_crtc->hw_mode = *adjusted_mode;
+
+	return 0;
+}
+
+static bool dce_v8_0_crtc_mode_fixup(struct drm_crtc *crtc,
+				     const struct drm_display_mode *mode,
+				     struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+	struct drm_device *dev = crtc->dev;
+	struct drm_encoder *encoder;
+
+	/* assign the encoder to the amdgpu crtc to avoid repeated lookups later */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		if (encoder->crtc == crtc) {
+			amdgpu_crtc->encoder = encoder;
+			amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder);
+			break;
+		}
+	}
+	if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) {
+		amdgpu_crtc->encoder = NULL;
+		amdgpu_crtc->connector = NULL;
+		return false;
+	}
+	if (!amdgpu_crtc_scaling_mode_fixup(crtc, mode, adjusted_mode))
+		return false;
+	if (amdgpu_atombios_crtc_prepare_pll(crtc, adjusted_mode))
+		return false;
+	/* pick pll */
+	amdgpu_crtc->pll_id = dce_v8_0_pick_pll(crtc);
+	/* if we can't get a PPLL for a non-DP encoder, fail */
+	if ((amdgpu_crtc->pll_id == ATOM_PPLL_INVALID) &&
+	    !ENCODER_MODE_IS_DP(amdgpu_atombios_encoder_get_encoder_mode(amdgpu_crtc->encoder)))
+		return false;
+
+	return true;
+}
+
+static int dce_v8_0_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+				  struct drm_framebuffer *old_fb)
+{
+	return dce_v8_0_crtc_do_set_base(crtc, old_fb, x, y, 0);
+}
+
+static int dce_v8_0_crtc_set_base_atomic(struct drm_crtc *crtc,
+					 struct drm_framebuffer *fb,
+					 int x, int y, enum mode_set_atomic state)
+{
+       return dce_v8_0_crtc_do_set_base(crtc, fb, x, y, 1);
+}
+
+static const struct drm_crtc_helper_funcs dce_v8_0_crtc_helper_funcs = {
+	.dpms = dce_v8_0_crtc_dpms,
+	.mode_fixup = dce_v8_0_crtc_mode_fixup,
+	.mode_set = dce_v8_0_crtc_mode_set,
+	.mode_set_base = dce_v8_0_crtc_set_base,
+	.mode_set_base_atomic = dce_v8_0_crtc_set_base_atomic,
+	.prepare = dce_v8_0_crtc_prepare,
+	.commit = dce_v8_0_crtc_commit,
+	.load_lut = dce_v8_0_crtc_load_lut,
+	.disable = dce_v8_0_crtc_disable,
+};
+
+static int dce_v8_0_crtc_init(struct amdgpu_device *adev, int index)
+{
+	struct amdgpu_crtc *amdgpu_crtc;
+	int i;
+
+	amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) +
+			      (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
+	if (amdgpu_crtc == NULL)
+		return -ENOMEM;
+
+	drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_v8_0_crtc_funcs);
+
+	drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256);
+	amdgpu_crtc->crtc_id = index;
+	amdgpu_crtc->pflip_queue = create_singlethread_workqueue("amdgpu-pageflip-queue");
+	adev->mode_info.crtcs[index] = amdgpu_crtc;
+
+	amdgpu_crtc->max_cursor_width = CIK_CURSOR_WIDTH;
+	amdgpu_crtc->max_cursor_height = CIK_CURSOR_HEIGHT;
+	adev->ddev->mode_config.cursor_width = amdgpu_crtc->max_cursor_width;
+	adev->ddev->mode_config.cursor_height = amdgpu_crtc->max_cursor_height;
+
+	for (i = 0; i < 256; i++) {
+		amdgpu_crtc->lut_r[i] = i << 2;
+		amdgpu_crtc->lut_g[i] = i << 2;
+		amdgpu_crtc->lut_b[i] = i << 2;
+	}
+
+	amdgpu_crtc->crtc_offset = crtc_offsets[amdgpu_crtc->crtc_id];
+
+	amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
+	amdgpu_crtc->adjusted_clock = 0;
+	amdgpu_crtc->encoder = NULL;
+	amdgpu_crtc->connector = NULL;
+	drm_crtc_helper_add(&amdgpu_crtc->base, &dce_v8_0_crtc_helper_funcs);
+
+	return 0;
+}
+
+static int dce_v8_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->audio_endpt_rreg = &dce_v8_0_audio_endpt_rreg;
+	adev->audio_endpt_wreg = &dce_v8_0_audio_endpt_wreg;
+
+	dce_v8_0_set_display_funcs(adev);
+	dce_v8_0_set_irq_funcs(adev);
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+	case CHIP_HAWAII:
+		adev->mode_info.num_crtc = 6;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6;
+		break;
+	case CHIP_KAVERI:
+		adev->mode_info.num_crtc = 4;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 7;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		adev->mode_info.num_crtc = 2;
+		adev->mode_info.num_hpd = 6;
+		adev->mode_info.num_dig = 6; /* ? */
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int dce_v8_0_sw_init(void *handle)
+{
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = amdgpu_irq_add_id(adev, i + 1, &adev->crtc_irq);
+		if (r)
+			return r;
+	}
+
+	for (i = 8; i < 20; i += 2) {
+		r = amdgpu_irq_add_id(adev, i, &adev->pageflip_irq);
+		if (r)
+			return r;
+	}
+
+	/* HPD hotplug */
+	r = amdgpu_irq_add_id(adev, 42, &adev->hpd_irq);
+	if (r)
+		return r;
+
+	adev->mode_info.mode_config_initialized = true;
+
+	adev->ddev->mode_config.funcs = &amdgpu_mode_funcs;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	adev->ddev->mode_config.preferred_depth = 24;
+	adev->ddev->mode_config.prefer_shadow = 1;
+
+	adev->ddev->mode_config.fb_base = adev->mc.aper_base;
+
+	r = amdgpu_modeset_create_props(adev);
+	if (r)
+		return r;
+
+	adev->ddev->mode_config.max_width = 16384;
+	adev->ddev->mode_config.max_height = 16384;
+
+	/* allocate crtcs */
+	for (i = 0; i < adev->mode_info.num_crtc; i++) {
+		r = dce_v8_0_crtc_init(adev, i);
+		if (r)
+			return r;
+	}
+
+	if (amdgpu_atombios_get_connector_info_from_object_table(adev))
+		amdgpu_print_display_setup(adev->ddev);
+	else
+		return -EINVAL;
+
+	/* setup afmt */
+	dce_v8_0_afmt_init(adev);
+
+	r = dce_v8_0_audio_init(adev);
+	if (r)
+		return r;
+
+	drm_kms_helper_poll_init(adev->ddev);
+
+	return r;
+}
+
+static int dce_v8_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kfree(adev->mode_info.bios_hardcoded_edid);
+
+	drm_kms_helper_poll_fini(adev->ddev);
+
+	dce_v8_0_audio_fini(adev);
+
+	dce_v8_0_afmt_fini(adev);
+
+	drm_mode_config_cleanup(adev->ddev);
+	adev->mode_info.mode_config_initialized = false;
+
+	return 0;
+}
+
+static int dce_v8_0_hw_init(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* init dig PHYs, disp eng pll */
+	amdgpu_atombios_encoder_init_dig(adev);
+	amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
+
+	/* initialize hpd */
+	dce_v8_0_hpd_init(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v8_0_pageflip_interrupt_init(adev);
+
+	return 0;
+}
+
+static int dce_v8_0_hw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dce_v8_0_hpd_fini(adev);
+
+	for (i = 0; i < adev->mode_info.audio.num_pins; i++) {
+		dce_v8_0_audio_enable(adev, &adev->mode_info.audio.pin[i], false);
+	}
+
+	dce_v8_0_pageflip_interrupt_fini(adev);
+
+	return 0;
+}
+
+static int dce_v8_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_atombios_scratch_regs_save(adev);
+
+	return dce_v8_0_hw_fini(handle);
+}
+
+static int dce_v8_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	ret = dce_v8_0_hw_init(handle);
+
+	amdgpu_atombios_scratch_regs_restore(adev);
+
+	/* turn on the BL */
+	if (adev->mode_info.bl_encoder) {
+		u8 bl_level = amdgpu_display_backlight_get_level(adev,
+								  adev->mode_info.bl_encoder);
+		amdgpu_display_backlight_set_level(adev, adev->mode_info.bl_encoder,
+						    bl_level);
+	}
+
+	return ret;
+}
+
+static bool dce_v8_0_is_idle(void *handle)
+{
+	return true;
+}
+
+static int dce_v8_0_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static void dce_v8_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "DCE 8.x registers\n");
+	/* XXX todo */
+}
+
+static int dce_v8_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0, tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (dce_v8_0_is_display_hung(adev))
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK;
+
+	if (srbm_soft_reset) {
+		dce_v8_0_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+		dce_v8_0_print_status((void *)adev);
+	}
+	return 0;
+}
+
+static void dce_v8_0_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev,
+						     int crtc,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 reg_block, lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (crtc) {
+	case 0:
+		reg_block = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		reg_block = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		reg_block = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		reg_block = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		reg_block = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		reg_block = CRTC5_REGISTER_OFFSET;
+		break;
+	default:
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask &= ~LB_INTERRUPT_MASK__VBLANK_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask |= LB_INTERRUPT_MASK__VBLANK_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static void dce_v8_0_set_crtc_vline_interrupt_state(struct amdgpu_device *adev,
+						    int crtc,
+						    enum amdgpu_interrupt_state state)
+{
+	u32 reg_block, lb_interrupt_mask;
+
+	if (crtc >= adev->mode_info.num_crtc) {
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (crtc) {
+	case 0:
+		reg_block = CRTC0_REGISTER_OFFSET;
+		break;
+	case 1:
+		reg_block = CRTC1_REGISTER_OFFSET;
+		break;
+	case 2:
+		reg_block = CRTC2_REGISTER_OFFSET;
+		break;
+	case 3:
+		reg_block = CRTC3_REGISTER_OFFSET;
+		break;
+	case 4:
+		reg_block = CRTC4_REGISTER_OFFSET;
+		break;
+	case 5:
+		reg_block = CRTC5_REGISTER_OFFSET;
+		break;
+	default:
+		DRM_DEBUG("invalid crtc %d\n", crtc);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask &= ~LB_INTERRUPT_MASK__VLINE_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		lb_interrupt_mask = RREG32(mmLB_INTERRUPT_MASK + reg_block);
+		lb_interrupt_mask |= LB_INTERRUPT_MASK__VLINE_INTERRUPT_MASK_MASK;
+		WREG32(mmLB_INTERRUPT_MASK + reg_block, lb_interrupt_mask);
+		break;
+	default:
+		break;
+	}
+}
+
+static int dce_v8_0_set_hpd_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	u32 dc_hpd_int_cntl_reg, dc_hpd_int_cntl;
+
+	switch (type) {
+	case AMDGPU_HPD_1:
+		dc_hpd_int_cntl_reg = mmDC_HPD1_INT_CONTROL;
+		break;
+	case AMDGPU_HPD_2:
+		dc_hpd_int_cntl_reg = mmDC_HPD2_INT_CONTROL;
+		break;
+	case AMDGPU_HPD_3:
+		dc_hpd_int_cntl_reg = mmDC_HPD3_INT_CONTROL;
+		break;
+	case AMDGPU_HPD_4:
+		dc_hpd_int_cntl_reg = mmDC_HPD4_INT_CONTROL;
+		break;
+	case AMDGPU_HPD_5:
+		dc_hpd_int_cntl_reg = mmDC_HPD5_INT_CONTROL;
+		break;
+	case AMDGPU_HPD_6:
+		dc_hpd_int_cntl_reg = mmDC_HPD6_INT_CONTROL;
+		break;
+	default:
+		DRM_DEBUG("invalid hdp %d\n", type);
+		return 0;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		dc_hpd_int_cntl = RREG32(dc_hpd_int_cntl_reg);
+		dc_hpd_int_cntl &= ~DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(dc_hpd_int_cntl_reg, dc_hpd_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		dc_hpd_int_cntl = RREG32(dc_hpd_int_cntl_reg);
+		dc_hpd_int_cntl |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_EN_MASK;
+		WREG32(dc_hpd_int_cntl_reg, dc_hpd_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v8_0_set_crtc_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CRTC_IRQ_VBLANK1:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK2:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK3:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK4:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK5:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VBLANK6:
+		dce_v8_0_set_crtc_vblank_interrupt_state(adev, 5, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE1:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 0, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE2:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 1, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE3:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 2, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE4:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 3, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE5:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 4, state);
+		break;
+	case AMDGPU_CRTC_IRQ_VLINE6:
+		dce_v8_0_set_crtc_vline_interrupt_state(adev, 5, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
+			     struct amdgpu_irq_src *source,
+			     struct amdgpu_iv_entry *entry)
+{
+	unsigned crtc = entry->src_id - 1;
+	uint32_t disp_int = RREG32(interrupt_status_offsets[crtc].reg);
+	unsigned irq_type = amdgpu_crtc_idx_to_irq_type(adev, crtc);
+
+	switch (entry->src_data) {
+	case 0: /* vblank */
+		if (disp_int & interrupt_status_offsets[crtc].vblank)
+			WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], LB_VBLANK_STATUS__VBLANK_ACK_MASK);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		if (amdgpu_irq_enabled(adev, source, irq_type)) {
+			drm_handle_vblank(adev->ddev, crtc);
+		}
+		DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
+		break;
+	case 1: /* vline */
+		if (disp_int & interrupt_status_offsets[crtc].vline)
+			WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], LB_VLINE_STATUS__VLINE_ACK_MASK);
+		else
+			DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+		DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
+		break;
+	default:
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
+		break;
+	}
+
+	return 0;
+}
+
+static int dce_v8_0_set_pageflip_interrupt_state(struct amdgpu_device *adev,
+						 struct amdgpu_irq_src *src,
+						 unsigned type,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 reg;
+
+	if (type >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", type);
+		return -EINVAL;
+	}
+
+	reg = RREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type]);
+	if (state == AMDGPU_IRQ_STATE_DISABLE)
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg & ~GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+	else
+		WREG32(mmGRPH_INTERRUPT_CONTROL + crtc_offsets[type],
+		       reg | GRPH_INTERRUPT_CONTROL__GRPH_PFLIP_INT_MASK_MASK);
+
+	return 0;
+}
+
+static int dce_v8_0_pageflip_irq(struct amdgpu_device *adev,
+				struct amdgpu_irq_src *source,
+				struct amdgpu_iv_entry *entry)
+{
+	unsigned long flags;
+	unsigned crtc_id;
+	struct amdgpu_crtc *amdgpu_crtc;
+	struct amdgpu_flip_work *works;
+
+	crtc_id = (entry->src_id - 8) >> 1;
+	amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
+
+	if (crtc_id >= adev->mode_info.num_crtc) {
+		DRM_ERROR("invalid pageflip crtc %d\n", crtc_id);
+		return -EINVAL;
+	}
+
+	if (RREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id]) &
+	    GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_OCCURRED_MASK)
+		WREG32(mmGRPH_INTERRUPT_STATUS + crtc_offsets[crtc_id],
+		       GRPH_INTERRUPT_STATUS__GRPH_PFLIP_INT_CLEAR_MASK);
+
+	/* IRQ could occur when in initial stage */
+	if (amdgpu_crtc == NULL)
+		return 0;
+
+	spin_lock_irqsave(&adev->ddev->event_lock, flags);
+	works = amdgpu_crtc->pflip_works;
+	if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED){
+		DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != "
+						"AMDGPU_FLIP_SUBMITTED(%d)\n",
+						amdgpu_crtc->pflip_status,
+						AMDGPU_FLIP_SUBMITTED);
+		spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+		return 0;
+	}
+
+	/* page flip completed. clean up */
+	amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE;
+	amdgpu_crtc->pflip_works = NULL;
+
+	/* wakeup usersapce */
+	if (works->event)
+		drm_send_vblank_event(adev->ddev, crtc_id, works->event);
+
+	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
+
+	drm_vblank_put(adev->ddev, amdgpu_crtc->crtc_id);
+	queue_work(amdgpu_crtc->pflip_queue, &works->unpin_work);
+
+	return 0;
+}
+
+static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	uint32_t disp_int, mask, int_control, tmp;
+	unsigned hpd;
+
+	if (entry->src_data >= adev->mode_info.num_hpd) {
+		DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
+		return 0;
+	}
+
+	hpd = entry->src_data;
+	disp_int = RREG32(interrupt_status_offsets[hpd].reg);
+	mask = interrupt_status_offsets[hpd].hpd;
+	int_control = hpd_int_control_offsets[hpd];
+
+	if (disp_int & mask) {
+		tmp = RREG32(int_control);
+		tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+		WREG32(int_control, tmp);
+		schedule_work(&adev->hotplug_work);
+		DRM_DEBUG("IH: HPD%d\n", hpd + 1);
+	}
+
+	return 0;
+
+}
+
+static int dce_v8_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int dce_v8_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs dce_v8_0_ip_funcs = {
+	.early_init = dce_v8_0_early_init,
+	.late_init = NULL,
+	.sw_init = dce_v8_0_sw_init,
+	.sw_fini = dce_v8_0_sw_fini,
+	.hw_init = dce_v8_0_hw_init,
+	.hw_fini = dce_v8_0_hw_fini,
+	.suspend = dce_v8_0_suspend,
+	.resume = dce_v8_0_resume,
+	.is_idle = dce_v8_0_is_idle,
+	.wait_for_idle = dce_v8_0_wait_for_idle,
+	.soft_reset = dce_v8_0_soft_reset,
+	.print_status = dce_v8_0_print_status,
+	.set_clockgating_state = dce_v8_0_set_clockgating_state,
+	.set_powergating_state = dce_v8_0_set_powergating_state,
+};
+
+static void
+dce_v8_0_encoder_mode_set(struct drm_encoder *encoder,
+			  struct drm_display_mode *mode,
+			  struct drm_display_mode *adjusted_mode)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+
+	amdgpu_encoder->pixel_clock = adjusted_mode->clock;
+
+	/* need to call this here rather than in prepare() since we need some crtc info */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	/* set scaler clears this on some chips */
+	dce_v8_0_set_interleave(encoder->crtc, mode);
+
+	if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
+		dce_v8_0_afmt_enable(encoder, true);
+		dce_v8_0_afmt_setmode(encoder, adjusted_mode);
+	}
+}
+
+static void dce_v8_0_encoder_prepare(struct drm_encoder *encoder)
+{
+	struct amdgpu_device *adev = encoder->dev->dev_private;
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct drm_connector *connector = amdgpu_get_connector_for_encoder(encoder);
+
+	if ((amdgpu_encoder->active_device &
+	     (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) ||
+	    (amdgpu_encoder_get_dp_bridge_encoder_id(encoder) !=
+	     ENCODER_OBJECT_ID_NONE)) {
+		struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv;
+		if (dig) {
+			dig->dig_encoder = dce_v8_0_pick_dig_encoder(encoder);
+			if (amdgpu_encoder->active_device & ATOM_DEVICE_DFP_SUPPORT)
+				dig->afmt = adev->mode_info.afmt[dig->dig_encoder];
+		}
+	}
+
+	amdgpu_atombios_scratch_regs_lock(adev, true);
+
+	if (connector) {
+		struct amdgpu_connector *amdgpu_connector = to_amdgpu_connector(connector);
+
+		/* select the clock/data port if it uses a router */
+		if (amdgpu_connector->router.cd_valid)
+			amdgpu_i2c_router_select_cd_port(amdgpu_connector);
+
+		/* turn eDP panel on for mode set */
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP)
+			amdgpu_atombios_encoder_set_edp_panel_power(connector,
+							     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
+
+	/* this is needed for the pll/ss setup to work correctly in some cases */
+	amdgpu_atombios_encoder_set_crtc_source(encoder);
+	/* set up the FMT blocks */
+	dce_v8_0_program_fmt(encoder);
+}
+
+static void dce_v8_0_encoder_commit(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* need to call this here as we need the crtc set up */
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_ON);
+	amdgpu_atombios_scratch_regs_lock(adev, false);
+}
+
+static void dce_v8_0_encoder_disable(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	struct amdgpu_encoder_atom_dig *dig;
+
+	amdgpu_atombios_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+	if (amdgpu_atombios_encoder_is_digital(encoder)) {
+		if (amdgpu_atombios_encoder_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI)
+			dce_v8_0_afmt_enable(encoder, false);
+		dig = amdgpu_encoder->enc_priv;
+		dig->dig_encoder = -1;
+	}
+	amdgpu_encoder->active_device = 0;
+}
+
+/* these are handled by the primary encoders */
+static void dce_v8_0_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void dce_v8_0_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v8_0_ext_mode_set(struct drm_encoder *encoder,
+		      struct drm_display_mode *mode,
+		      struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void dce_v8_0_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+dce_v8_0_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static bool dce_v8_0_ext_mode_fixup(struct drm_encoder *encoder,
+				    const struct drm_display_mode *mode,
+				    struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static const struct drm_encoder_helper_funcs dce_v8_0_ext_helper_funcs = {
+	.dpms = dce_v8_0_ext_dpms,
+	.mode_fixup = dce_v8_0_ext_mode_fixup,
+	.prepare = dce_v8_0_ext_prepare,
+	.mode_set = dce_v8_0_ext_mode_set,
+	.commit = dce_v8_0_ext_commit,
+	.disable = dce_v8_0_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
+static const struct drm_encoder_helper_funcs dce_v8_0_dig_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v8_0_encoder_prepare,
+	.mode_set = dce_v8_0_encoder_mode_set,
+	.commit = dce_v8_0_encoder_commit,
+	.disable = dce_v8_0_encoder_disable,
+	.detect = amdgpu_atombios_encoder_dig_detect,
+};
+
+static const struct drm_encoder_helper_funcs dce_v8_0_dac_helper_funcs = {
+	.dpms = amdgpu_atombios_encoder_dpms,
+	.mode_fixup = amdgpu_atombios_encoder_mode_fixup,
+	.prepare = dce_v8_0_encoder_prepare,
+	.mode_set = dce_v8_0_encoder_mode_set,
+	.commit = dce_v8_0_encoder_commit,
+	.detect = amdgpu_atombios_encoder_dac_detect,
+};
+
+static void dce_v8_0_encoder_destroy(struct drm_encoder *encoder)
+{
+	struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder);
+	if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+		amdgpu_atombios_encoder_fini_backlight(amdgpu_encoder);
+	kfree(amdgpu_encoder->enc_priv);
+	drm_encoder_cleanup(encoder);
+	kfree(amdgpu_encoder);
+}
+
+static const struct drm_encoder_funcs dce_v8_0_encoder_funcs = {
+	.destroy = dce_v8_0_encoder_destroy,
+};
+
+static void dce_v8_0_encoder_add(struct amdgpu_device *adev,
+				 uint32_t encoder_enum,
+				 uint32_t supported_device,
+				 u16 caps)
+{
+	struct drm_device *dev = adev->ddev;
+	struct drm_encoder *encoder;
+	struct amdgpu_encoder *amdgpu_encoder;
+
+	/* see if we already added it */
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		amdgpu_encoder = to_amdgpu_encoder(encoder);
+		if (amdgpu_encoder->encoder_enum == encoder_enum) {
+			amdgpu_encoder->devices |= supported_device;
+			return;
+		}
+
+	}
+
+	/* add a new one */
+	amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL);
+	if (!amdgpu_encoder)
+		return;
+
+	encoder = &amdgpu_encoder->base;
+	switch (adev->mode_info.num_crtc) {
+	case 1:
+		encoder->possible_crtcs = 0x1;
+		break;
+	case 2:
+	default:
+		encoder->possible_crtcs = 0x3;
+		break;
+	case 4:
+		encoder->possible_crtcs = 0xf;
+		break;
+	case 6:
+		encoder->possible_crtcs = 0x3f;
+		break;
+	}
+
+	amdgpu_encoder->enc_priv = NULL;
+
+	amdgpu_encoder->encoder_enum = encoder_enum;
+	amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	amdgpu_encoder->devices = supported_device;
+	amdgpu_encoder->rmx_type = RMX_OFF;
+	amdgpu_encoder->underscan_type = UNDERSCAN_OFF;
+	amdgpu_encoder->is_ext_encoder = false;
+	amdgpu_encoder->caps = caps;
+
+	switch (amdgpu_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+				 DRM_MODE_ENCODER_DAC);
+		drm_encoder_helper_add(encoder, &dce_v8_0_dac_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY3:
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+			amdgpu_encoder->rmx_type = RMX_FULL;
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_lcd_info(amdgpu_encoder);
+		} else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		} else {
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS);
+			amdgpu_encoder->enc_priv = amdgpu_atombios_encoder_get_dig_info(amdgpu_encoder);
+		}
+		drm_encoder_helper_add(encoder, &dce_v8_0_dig_helper_funcs);
+		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+	case ENCODER_OBJECT_ID_TRAVIS:
+	case ENCODER_OBJECT_ID_NUTMEG:
+		/* these are handled by the primary encoders */
+		amdgpu_encoder->is_ext_encoder = true;
+		if (amdgpu_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_LVDS);
+		else if (amdgpu_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_DAC);
+		else
+			drm_encoder_init(dev, encoder, &dce_v8_0_encoder_funcs,
+					 DRM_MODE_ENCODER_TMDS);
+		drm_encoder_helper_add(encoder, &dce_v8_0_ext_helper_funcs);
+		break;
+	}
+}
+
+static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
+	.set_vga_render_state = &dce_v8_0_set_vga_render_state,
+	.bandwidth_update = &dce_v8_0_bandwidth_update,
+	.vblank_get_counter = &dce_v8_0_vblank_get_counter,
+	.vblank_wait = &dce_v8_0_vblank_wait,
+	.is_display_hung = &dce_v8_0_is_display_hung,
+	.backlight_set_level = &amdgpu_atombios_encoder_set_backlight_level,
+	.backlight_get_level = &amdgpu_atombios_encoder_get_backlight_level,
+	.hpd_sense = &dce_v8_0_hpd_sense,
+	.hpd_set_polarity = &dce_v8_0_hpd_set_polarity,
+	.hpd_get_gpio_reg = &dce_v8_0_hpd_get_gpio_reg,
+	.page_flip = &dce_v8_0_page_flip,
+	.page_flip_get_scanoutpos = &dce_v8_0_crtc_get_scanoutpos,
+	.add_encoder = &dce_v8_0_encoder_add,
+	.add_connector = &amdgpu_connector_add,
+	.stop_mc_access = &dce_v8_0_stop_mc_access,
+	.resume_mc_access = &dce_v8_0_resume_mc_access,
+};
+
+static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mode_info.funcs == NULL)
+		adev->mode_info.funcs = &dce_v8_0_display_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs dce_v8_0_crtc_irq_funcs = {
+	.set = dce_v8_0_set_crtc_interrupt_state,
+	.process = dce_v8_0_crtc_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v8_0_pageflip_irq_funcs = {
+	.set = dce_v8_0_set_pageflip_interrupt_state,
+	.process = dce_v8_0_pageflip_irq,
+};
+
+static const struct amdgpu_irq_src_funcs dce_v8_0_hpd_irq_funcs = {
+	.set = dce_v8_0_set_hpd_interrupt_state,
+	.process = dce_v8_0_hpd_irq,
+};
+
+static void dce_v8_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
+	adev->crtc_irq.funcs = &dce_v8_0_crtc_irq_funcs;
+
+	adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST;
+	adev->pageflip_irq.funcs = &dce_v8_0_pageflip_irq_funcs;
+
+	adev->hpd_irq.num_types = AMDGPU_HPD_LAST;
+	adev->hpd_irq.funcs = &dce_v8_0_hpd_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h
new file mode 100644
index 000000000..77016852b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V8_0_H__
+#define __DCE_V8_0_H__
+
+extern const struct amd_ip_funcs dce_v8_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c
new file mode 100644
index 000000000..8f9845d9a
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c
@@ -0,0 +1,181 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "fiji_smumgr.h"
+
+MODULE_FIRMWARE("amdgpu/fiji_smc.bin");
+
+static void fiji_dpm_set_funcs(struct amdgpu_device *adev);
+
+static int fiji_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	fiji_dpm_set_funcs(adev);
+
+	return 0;
+}
+
+static int fiji_dpm_init_microcode(struct amdgpu_device *adev)
+{
+	char fw_name[30] = "amdgpu/fiji_smc.bin";
+	int err;
+
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+
+out:
+	if (err) {
+		DRM_ERROR("Failed to load firmware \"%s\"", fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+}
+
+static int fiji_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = fiji_dpm_init_microcode(adev);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int fiji_dpm_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int fiji_dpm_hw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+
+	ret = fiji_smu_init(adev);
+	if (ret) {
+		DRM_ERROR("SMU initialization failed\n");
+		goto fail;
+	}
+
+	ret = fiji_smu_start(adev);
+	if (ret) {
+		DRM_ERROR("SMU start failed\n");
+		goto fail;
+	}
+
+	mutex_unlock(&adev->pm.mutex);
+	return 0;
+
+fail:
+	adev->firmware.smu_load = false;
+	mutex_unlock(&adev->pm.mutex);
+	return -EINVAL;
+}
+
+static int fiji_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	mutex_lock(&adev->pm.mutex);
+	fiji_smu_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	return 0;
+}
+
+static int fiji_dpm_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	fiji_dpm_hw_fini(adev);
+
+	return 0;
+}
+
+static int fiji_dpm_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	fiji_dpm_hw_init(adev);
+
+	return 0;
+}
+
+static int fiji_dpm_set_clockgating_state(void *handle,
+			enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int fiji_dpm_set_powergating_state(void *handle,
+			enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs fiji_dpm_ip_funcs = {
+	.early_init = fiji_dpm_early_init,
+	.late_init = NULL,
+	.sw_init = fiji_dpm_sw_init,
+	.sw_fini = fiji_dpm_sw_fini,
+	.hw_init = fiji_dpm_hw_init,
+	.hw_fini = fiji_dpm_hw_fini,
+	.suspend = fiji_dpm_suspend,
+	.resume = fiji_dpm_resume,
+	.is_idle = NULL,
+	.wait_for_idle = NULL,
+	.soft_reset = NULL,
+	.print_status = NULL,
+	.set_clockgating_state = fiji_dpm_set_clockgating_state,
+	.set_powergating_state = fiji_dpm_set_powergating_state,
+};
+
+static const struct amdgpu_dpm_funcs fiji_dpm_funcs = {
+	.get_temperature = NULL,
+	.pre_set_power_state = NULL,
+	.set_power_state = NULL,
+	.post_set_power_state = NULL,
+	.display_configuration_changed = NULL,
+	.get_sclk = NULL,
+	.get_mclk = NULL,
+	.print_power_state = NULL,
+	.debugfs_print_current_performance_level = NULL,
+	.force_performance_level = NULL,
+	.vblank_too_short = NULL,
+	.powergate_uvd = NULL,
+};
+
+static void fiji_dpm_set_funcs(struct amdgpu_device *adev)
+{
+	if (NULL == adev->pm.funcs)
+		adev->pm.funcs = &fiji_dpm_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h
new file mode 100644
index 000000000..3c4824082
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_ppsmc.h
@@ -0,0 +1,182 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef FIJI_PP_SMC_H
+#define FIJI_PP_SMC_H
+
+#pragma pack(push, 1)
+
+#define PPSMC_SWSTATE_FLAG_DC                           0x01
+#define PPSMC_SWSTATE_FLAG_UVD                          0x02
+#define PPSMC_SWSTATE_FLAG_VCE                          0x04
+
+#define PPSMC_THERMAL_PROTECT_TYPE_INTERNAL             0x00
+#define PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL             0x01
+#define PPSMC_THERMAL_PROTECT_TYPE_NONE                 0xff
+
+#define PPSMC_SYSTEMFLAG_GPIO_DC                        0x01
+#define PPSMC_SYSTEMFLAG_STEPVDDC                       0x02
+#define PPSMC_SYSTEMFLAG_GDDR5                          0x04
+
+#define PPSMC_SYSTEMFLAG_DISABLE_BABYSTEP               0x08
+
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT                  0x10
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_ANALOG           0x20
+
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_MASK              0x07
+#define PPSMC_EXTRAFLAGS_AC2DC_DONT_WAIT_FOR_VBLANK     0x08
+
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTODPMLOWSTATE   0x00
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTOINITIALSTATE  0x01
+
+#define PPSMC_DPM2FLAGS_TDPCLMP                         0x01
+#define PPSMC_DPM2FLAGS_PWRSHFT                         0x02
+#define PPSMC_DPM2FLAGS_OCP                             0x04
+
+#define PPSMC_DISPLAY_WATERMARK_LOW                     0
+#define PPSMC_DISPLAY_WATERMARK_HIGH                    1
+
+#define PPSMC_STATEFLAG_AUTO_PULSE_SKIP    		0x01
+#define PPSMC_STATEFLAG_POWERBOOST         		0x02
+#define PPSMC_STATEFLAG_PSKIP_ON_TDP_FAULT 		0x04
+#define PPSMC_STATEFLAG_POWERSHIFT         		0x08
+#define PPSMC_STATEFLAG_SLOW_READ_MARGIN   		0x10
+#define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 		0x20
+#define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS   		0x40
+
+#define FDO_MODE_HARDWARE 0
+#define FDO_MODE_PIECE_WISE_LINEAR 1
+
+enum FAN_CONTROL {
+	FAN_CONTROL_FUZZY,
+	FAN_CONTROL_TABLE
+};
+
+//Gemini Modes
+#define PPSMC_GeminiModeNone   0  //Single GPU board
+#define PPSMC_GeminiModeMaster 1  //Master GPU on a Gemini board
+#define PPSMC_GeminiModeSlave  2  //Slave GPU on a Gemini board
+
+#define PPSMC_Result_OK             			((uint16_t)0x01)
+#define PPSMC_Result_NoMore         			((uint16_t)0x02)
+#define PPSMC_Result_NotNow         			((uint16_t)0x03)
+#define PPSMC_Result_Failed         			((uint16_t)0xFF)
+#define PPSMC_Result_UnknownCmd     			((uint16_t)0xFE)
+#define PPSMC_Result_UnknownVT      			((uint16_t)0xFD)
+
+typedef uint16_t PPSMC_Result;
+
+#define PPSMC_isERROR(x) ((uint16_t)0x80 & (x))
+
+#define PPSMC_MSG_Halt                      		((uint16_t)0x10)
+#define PPSMC_MSG_Resume                    		((uint16_t)0x11)
+#define PPSMC_MSG_EnableDPMLevel            		((uint16_t)0x12)
+#define PPSMC_MSG_ZeroLevelsDisabled        		((uint16_t)0x13)
+#define PPSMC_MSG_OneLevelsDisabled         		((uint16_t)0x14)
+#define PPSMC_MSG_TwoLevelsDisabled         		((uint16_t)0x15)
+#define PPSMC_MSG_EnableThermalInterrupt    		((uint16_t)0x16)
+#define PPSMC_MSG_RunningOnAC               		((uint16_t)0x17)
+#define PPSMC_MSG_LevelUp                   		((uint16_t)0x18)
+#define PPSMC_MSG_LevelDown                 		((uint16_t)0x19)
+#define PPSMC_MSG_ResetDPMCounters          		((uint16_t)0x1a)
+#define PPSMC_MSG_SwitchToSwState           		((uint16_t)0x20)
+#define PPSMC_MSG_SwitchToSwStateLast       		((uint16_t)0x3f)
+#define PPSMC_MSG_SwitchToInitialState      		((uint16_t)0x40)
+#define PPSMC_MSG_NoForcedLevel             		((uint16_t)0x41)
+#define PPSMC_MSG_ForceHigh                 		((uint16_t)0x42)
+#define PPSMC_MSG_ForceMediumOrHigh         		((uint16_t)0x43)
+#define PPSMC_MSG_SwitchToMinimumPower      		((uint16_t)0x51)
+#define PPSMC_MSG_ResumeFromMinimumPower    		((uint16_t)0x52)
+#define PPSMC_MSG_EnableCac                 		((uint16_t)0x53)
+#define PPSMC_MSG_DisableCac                		((uint16_t)0x54)
+#define PPSMC_DPMStateHistoryStart          		((uint16_t)0x55)
+#define PPSMC_DPMStateHistoryStop           		((uint16_t)0x56)
+#define PPSMC_CACHistoryStart               		((uint16_t)0x57)
+#define PPSMC_CACHistoryStop                		((uint16_t)0x58)
+#define PPSMC_TDPClampingActive             		((uint16_t)0x59)
+#define PPSMC_TDPClampingInactive           		((uint16_t)0x5A)
+#define PPSMC_StartFanControl               		((uint16_t)0x5B)
+#define PPSMC_StopFanControl                		((uint16_t)0x5C)
+#define PPSMC_NoDisplay                     		((uint16_t)0x5D)
+#define PPSMC_HasDisplay                    		((uint16_t)0x5E)
+#define PPSMC_MSG_UVDPowerOFF               		((uint16_t)0x60)
+#define PPSMC_MSG_UVDPowerON                		((uint16_t)0x61)
+#define PPSMC_MSG_EnableULV                 		((uint16_t)0x62)
+#define PPSMC_MSG_DisableULV                		((uint16_t)0x63)
+#define PPSMC_MSG_EnterULV                  		((uint16_t)0x64)
+#define PPSMC_MSG_ExitULV                   		((uint16_t)0x65)
+#define PPSMC_PowerShiftActive              		((uint16_t)0x6A)
+#define PPSMC_PowerShiftInactive            		((uint16_t)0x6B)
+#define PPSMC_OCPActive                     		((uint16_t)0x6C)
+#define PPSMC_OCPInactive                   		((uint16_t)0x6D)
+#define PPSMC_CACLongTermAvgEnable          		((uint16_t)0x6E)
+#define PPSMC_CACLongTermAvgDisable         		((uint16_t)0x6F)
+#define PPSMC_MSG_InferredStateSweep_Start  		((uint16_t)0x70)
+#define PPSMC_MSG_InferredStateSweep_Stop   		((uint16_t)0x71)
+#define PPSMC_MSG_SwitchToLowestInfState    		((uint16_t)0x72)
+#define PPSMC_MSG_SwitchToNonInfState       		((uint16_t)0x73)
+#define PPSMC_MSG_AllStateSweep_Start       		((uint16_t)0x74)
+#define PPSMC_MSG_AllStateSweep_Stop        		((uint16_t)0x75)
+#define PPSMC_MSG_SwitchNextLowerInfState   		((uint16_t)0x76)
+#define PPSMC_MSG_SwitchNextHigherInfState  		((uint16_t)0x77)
+#define PPSMC_MSG_MclkRetrainingTest        		((uint16_t)0x78)
+#define PPSMC_MSG_ForceTDPClamping          		((uint16_t)0x79)
+#define PPSMC_MSG_CollectCAC_PowerCorreln   		((uint16_t)0x7A)
+#define PPSMC_MSG_CollectCAC_WeightCalib    		((uint16_t)0x7B)
+#define PPSMC_MSG_CollectCAC_SQonly         		((uint16_t)0x7C)
+#define PPSMC_MSG_CollectCAC_TemperaturePwr 		((uint16_t)0x7D)
+#define PPSMC_MSG_ExtremitiesTest_Start     		((uint16_t)0x7E)
+#define PPSMC_MSG_ExtremitiesTest_Stop      		((uint16_t)0x7F)
+#define PPSMC_FlushDataCache                		((uint16_t)0x80)
+#define PPSMC_FlushInstrCache               		((uint16_t)0x81)
+#define PPSMC_MSG_SetEnabledLevels          		((uint16_t)0x82)
+#define PPSMC_MSG_SetForcedLevels           		((uint16_t)0x83)
+#define PPSMC_MSG_ResetToDefaults           		((uint16_t)0x84)
+#define PPSMC_MSG_SetForcedLevelsAndJump    		((uint16_t)0x85)
+#define PPSMC_MSG_SetCACHistoryMode         		((uint16_t)0x86)
+#define PPSMC_MSG_EnableDTE                 		((uint16_t)0x87)
+#define PPSMC_MSG_DisableDTE                		((uint16_t)0x88)
+#define PPSMC_MSG_SmcSpaceSetAddress        		((uint16_t)0x89)
+#define PPSMC_MSG_SmcSpaceWriteDWordInc     		((uint16_t)0x8A)
+#define PPSMC_MSG_SmcSpaceWriteWordInc      		((uint16_t)0x8B)
+#define PPSMC_MSG_SmcSpaceWriteByteInc      		((uint16_t)0x8C)
+
+#define PPSMC_MSG_BREAK                     		((uint16_t)0xF8)
+
+#define PPSMC_MSG_Test                      		((uint16_t)0x100)
+#define PPSMC_MSG_DRV_DRAM_ADDR_HI            		((uint16_t)0x250)
+#define PPSMC_MSG_DRV_DRAM_ADDR_LO            		((uint16_t)0x251)
+#define PPSMC_MSG_SMU_DRAM_ADDR_HI            		((uint16_t)0x252)
+#define PPSMC_MSG_SMU_DRAM_ADDR_LO            		((uint16_t)0x253)
+#define PPSMC_MSG_LoadUcodes                  		((uint16_t)0x254)
+
+typedef uint16_t PPSMC_Msg;
+
+#define PPSMC_EVENT_STATUS_THERMAL          		0x00000001
+#define PPSMC_EVENT_STATUS_REGULATORHOT     		0x00000002
+#define PPSMC_EVENT_STATUS_DC               		0x00000004
+#define PPSMC_EVENT_STATUS_GPIO17           		0x00000008
+
+#pragma pack(pop)
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/fiji_smc.c b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_smc.c
new file mode 100644
index 000000000..bda1249eb
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_smc.c
@@ -0,0 +1,857 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "fiji_ppsmc.h"
+#include "fiji_smumgr.h"
+#include "smu_ucode_xfer_vi.h"
+#include "amdgpu_ucode.h"
+
+#include "smu/smu_7_1_3_d.h"
+#include "smu/smu_7_1_3_sh_mask.h"
+
+#define FIJI_SMC_SIZE 0x20000
+
+static int fiji_set_smc_sram_address(struct amdgpu_device *adev, uint32_t smc_address, uint32_t limit)
+{
+	uint32_t val;
+
+	if (smc_address & 3)
+		return -EINVAL;
+
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 0);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+
+	return 0;
+}
+
+static int fiji_copy_bytes_to_smc(struct amdgpu_device *adev, uint32_t smc_start_address, const uint8_t *src, uint32_t byte_count, uint32_t limit)
+{
+	uint32_t addr;
+	uint32_t data, orig_data;
+	int result = 0;
+	uint32_t extra_shift;
+	unsigned long flags;
+
+	if (smc_start_address & 3)
+		return -EINVAL;
+
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	while (byte_count >= 4) {
+		/* Bytes are written into the SMC addres space with the MSB first */
+		data = (src[0] << 24) + (src[1] << 16) + (src[2] << 8) + src[3];
+
+		result = fiji_set_smc_sram_address(adev, addr, limit);
+
+		if (result)
+			goto out;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	if (0 != byte_count) {
+		/* Now write odd bytes left, do a read modify write cycle */
+		data = 0;
+
+		result = fiji_set_smc_sram_address(adev, addr, limit);
+		if (result)
+			goto out;
+
+		orig_data = RREG32(mmSMC_IND_DATA_0);
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+		data |= (orig_data & ~((~0UL) << extra_shift));
+
+		result = fiji_set_smc_sram_address(adev, addr, limit);
+		if (result)
+			goto out;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+
+out:
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int fiji_program_jump_on_start(struct amdgpu_device *adev)
+{
+	static unsigned char data[] = {0xE0, 0x00, 0x80, 0x40};
+	fiji_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
+
+	return 0;
+}
+
+static bool fiji_is_smc_ram_running(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_GET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable);
+
+	return ((0 == val) && (0x20100 <= RREG32_SMC(ixSMC_PC_C)));
+}
+
+static int wait_smu_response(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32(mmSMC_RESP_0);
+		if (REG_GET_FIELD(val, SMC_RESP_0, SMC_RESP))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int fiji_send_msg_to_smc_offset(struct amdgpu_device *adev)
+{
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MSG_ARG_0, 0x20000);
+	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_Test);
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send message\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int fiji_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
+{
+	if (!fiji_is_smc_ram_running(adev))
+	{
+		return -EINVAL;;
+	}
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send message\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int fiji_send_msg_to_smc_without_waiting(struct amdgpu_device *adev,
+						PPSMC_Msg msg)
+{
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	return 0;
+}
+
+static int fiji_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						PPSMC_Msg msg,
+						uint32_t parameter)
+{
+	if (!fiji_is_smc_ram_running(adev))
+		return -EINVAL;
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return fiji_send_msg_to_smc(adev, msg);
+}
+
+static int fiji_send_msg_to_smc_with_parameter_without_waiting(
+					struct amdgpu_device *adev,
+					PPSMC_Msg msg, uint32_t parameter)
+{
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return fiji_send_msg_to_smc_without_waiting(adev, msg);
+}
+
+#if 0 /* not used yet */
+static int fiji_wait_for_smc_inactive(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	if (!fiji_is_smc_ram_running(adev))
+		return -EINVAL;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+		if (REG_GET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, cken) == 0)
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+#endif
+
+static int fiji_smu_upload_firmware_image(struct amdgpu_device *adev)
+{
+	const struct smc_firmware_header_v1_0 *hdr;
+	uint32_t ucode_size;
+	uint32_t ucode_start_address;
+	const uint8_t *src;
+	uint32_t val;
+	uint32_t byte_count;
+	uint32_t *data;
+	unsigned long flags;
+
+	if (!adev->pm.fw)
+		return -EINVAL;
+
+	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+	src = (const uint8_t *)
+		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	if (ucode_size & 3) {
+		DRM_ERROR("SMC ucode is not 4 bytes aligned\n");
+		return -EINVAL;
+	}
+
+	if (ucode_size > FIJI_SMC_SIZE) {
+		DRM_ERROR("SMC address is beyond the SMC RAM area\n");
+		return -EINVAL;
+	}
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 1);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+
+	byte_count = ucode_size;
+	data = (uint32_t *)src;
+	for (; byte_count >= 4; data++, byte_count -= 4)
+		WREG32(mmSMC_IND_DATA_0, data[0]);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 0);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return 0;
+}
+
+#if 0 /* not used yet */
+static int fiji_read_smc_sram_dword(struct amdgpu_device *adev,
+				uint32_t smc_address,
+				uint32_t *value,
+				uint32_t limit)
+{
+	int result;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	result = fiji_set_smc_sram_address(adev, smc_address, limit);
+	if (result == 0)
+		*value = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int fiji_write_smc_sram_dword(struct amdgpu_device *adev,
+				uint32_t smc_address,
+				uint32_t value,
+				uint32_t limit)
+{
+	int result;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	result = fiji_set_smc_sram_address(adev, smc_address, limit);
+	if (result == 0)
+		WREG32(mmSMC_IND_DATA_0, value);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int fiji_smu_stop_smc(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 1);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+
+	return 0;
+}
+#endif
+
+static enum AMDGPU_UCODE_ID fiji_convert_fw_type(uint32_t fw_type)
+{
+	switch (fw_type) {
+		case UCODE_ID_SDMA0:
+			return AMDGPU_UCODE_ID_SDMA0;
+		case UCODE_ID_SDMA1:
+			return AMDGPU_UCODE_ID_SDMA1;
+		case UCODE_ID_CP_CE:
+			return AMDGPU_UCODE_ID_CP_CE;
+		case UCODE_ID_CP_PFP:
+			return AMDGPU_UCODE_ID_CP_PFP;
+		case UCODE_ID_CP_ME:
+			return AMDGPU_UCODE_ID_CP_ME;
+		case UCODE_ID_CP_MEC:
+		case UCODE_ID_CP_MEC_JT1:
+		case UCODE_ID_CP_MEC_JT2:
+			return AMDGPU_UCODE_ID_CP_MEC1;
+		case UCODE_ID_RLC_G:
+			return AMDGPU_UCODE_ID_RLC_G;
+		default:
+			DRM_ERROR("ucode type is out of range!\n");
+			return AMDGPU_UCODE_ID_MAXIMUM;
+	}
+}
+
+static int fiji_smu_populate_single_firmware_entry(struct amdgpu_device *adev,
+						uint32_t fw_type,
+						struct SMU_Entry *entry)
+{
+	enum AMDGPU_UCODE_ID id = fiji_convert_fw_type(fw_type);
+	struct amdgpu_firmware_info *ucode = &adev->firmware.ucode[id];
+	const struct gfx_firmware_header_v1_0 *header = NULL;
+	uint64_t gpu_addr;
+	uint32_t data_size;
+
+	if (ucode->fw == NULL)
+		return -EINVAL;
+	gpu_addr  = ucode->mc_addr;
+	header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+	data_size = le32_to_cpu(header->header.ucode_size_bytes);
+
+	if ((fw_type == UCODE_ID_CP_MEC_JT1) ||
+		(fw_type == UCODE_ID_CP_MEC_JT2)) {
+		gpu_addr += le32_to_cpu(header->jt_offset) << 2;
+		data_size = le32_to_cpu(header->jt_size) << 2;
+	}
+
+	entry->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
+	entry->id = (uint16_t)fw_type;
+	entry->image_addr_high = upper_32_bits(gpu_addr);
+	entry->image_addr_low = lower_32_bits(gpu_addr);
+	entry->meta_data_addr_high = 0;
+	entry->meta_data_addr_low = 0;
+	entry->data_size_byte = data_size;
+	entry->num_register_entries = 0;
+
+	if (fw_type == UCODE_ID_RLC_G)
+		entry->flags = 1;
+	else
+		entry->flags = 0;
+
+	return 0;
+}
+
+static int fiji_smu_request_load_fw(struct amdgpu_device *adev)
+{
+	struct fiji_smu_private_data *private = (struct fiji_smu_private_data *)adev->smu.priv;
+	struct SMU_DRAMData_TOC *toc;
+	uint32_t fw_to_load;
+
+	WREG32_SMC(ixSOFT_REGISTERS_TABLE_28, 0);
+
+	fiji_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SMU_DRAM_ADDR_HI, private->smu_buffer_addr_high);
+	fiji_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SMU_DRAM_ADDR_LO, private->smu_buffer_addr_low);
+
+	toc = (struct SMU_DRAMData_TOC *)private->header;
+	toc->num_entries = 0;
+	toc->structure_version = 1;
+
+	if (!adev->firmware.smu_load)
+		return 0;
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_RLC_G,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for RLC\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_CE,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for CE\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_PFP,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for PFP\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_ME,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for ME\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC_JT1,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC_JT1\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC_JT2,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC_JT2\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_SDMA0,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for SDMA0\n");
+		return -EINVAL;
+	}
+
+	if (fiji_smu_populate_single_firmware_entry(adev, UCODE_ID_SDMA1,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for SDMA1\n");
+		return -EINVAL;
+	}
+
+	fiji_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DRV_DRAM_ADDR_HI, private->header_addr_high);
+	fiji_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DRV_DRAM_ADDR_LO, private->header_addr_low);
+
+	fw_to_load = UCODE_ID_RLC_G_MASK |
+			UCODE_ID_SDMA0_MASK |
+			UCODE_ID_SDMA1_MASK |
+			UCODE_ID_CP_CE_MASK |
+			UCODE_ID_CP_ME_MASK |
+			UCODE_ID_CP_PFP_MASK |
+			UCODE_ID_CP_MEC_MASK;
+
+	if (fiji_send_msg_to_smc_with_parameter_without_waiting(adev, PPSMC_MSG_LoadUcodes, fw_to_load)) {
+		DRM_ERROR("Fail to request SMU load ucode\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static uint32_t fiji_smu_get_mask_for_fw_type(uint32_t fw_type)
+{
+	switch (fw_type) {
+		case AMDGPU_UCODE_ID_SDMA0:
+			return UCODE_ID_SDMA0_MASK;
+		case AMDGPU_UCODE_ID_SDMA1:
+			return UCODE_ID_SDMA1_MASK;
+		case AMDGPU_UCODE_ID_CP_CE:
+			return UCODE_ID_CP_CE_MASK;
+		case AMDGPU_UCODE_ID_CP_PFP:
+			return UCODE_ID_CP_PFP_MASK;
+		case AMDGPU_UCODE_ID_CP_ME:
+			return UCODE_ID_CP_ME_MASK;
+		case AMDGPU_UCODE_ID_CP_MEC1:
+			return UCODE_ID_CP_MEC_MASK;
+		case AMDGPU_UCODE_ID_CP_MEC2:
+			return UCODE_ID_CP_MEC_MASK;
+		case AMDGPU_UCODE_ID_RLC_G:
+			return UCODE_ID_RLC_G_MASK;
+		default:
+			DRM_ERROR("ucode type is out of range!\n");
+			return 0;
+	}
+}
+
+static int fiji_smu_check_fw_load_finish(struct amdgpu_device *adev,
+					uint32_t fw_type)
+{
+	uint32_t fw_mask = fiji_smu_get_mask_for_fw_type(fw_type);
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (fw_mask == (RREG32_SMC(ixSOFT_REGISTERS_TABLE_28) & fw_mask))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("check firmware loading failed\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int fiji_smu_start_in_protection_mode(struct amdgpu_device *adev)
+{
+	int result;
+	uint32_t val;
+	int i;
+
+	/* Assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	result = fiji_smu_upload_firmware_image(adev);
+	if (result)
+		return result;
+
+	/* Clear status */
+	WREG32_SMC(ixSMU_STATUS, 0);
+
+	/* Enable clock */
+	val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 0);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+
+	/* De-assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 0);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	/* Set SMU Auto Start */
+	val = RREG32_SMC(ixSMU_INPUT_DATA);
+	val = REG_SET_FIELD(val, SMU_INPUT_DATA, AUTO_START, 1);
+	WREG32_SMC(ixSMU_INPUT_DATA, val);
+
+	/* Clear firmware interrupt enable flag */
+	WREG32_SMC(ixFIRMWARE_FLAGS, 0);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixRCU_UC_EVENTS);
+		if (REG_GET_FIELD(val, RCU_UC_EVENTS, INTERRUPTS_ENABLED))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("Interrupt is not enabled by firmware\n");
+		return -EINVAL;
+	}
+
+	/* Call Test SMU message with 0x20000 offset
+	 * to trigger SMU start
+	 */
+	fiji_send_msg_to_smc_offset(adev);
+	DRM_INFO("[FM]try triger smu start\n");
+	/* Wait for done bit to be set */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixSMU_STATUS);
+		if (REG_GET_FIELD(val, SMU_STATUS, SMU_DONE))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("Timeout for SMU start\n");
+		return -EINVAL;
+	}
+
+	/* Check pass/failed indicator */
+	val = RREG32_SMC(ixSMU_STATUS);
+	if (!REG_GET_FIELD(val, SMU_STATUS, SMU_PASS)) {
+		DRM_ERROR("SMU Firmware start failed\n");
+		return -EINVAL;
+	}
+	DRM_INFO("[FM]smu started\n");
+	/* Wait for firmware to initialize */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixFIRMWARE_FLAGS);
+		if(REG_GET_FIELD(val, FIRMWARE_FLAGS, INTERRUPTS_ENABLED))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("SMU firmware initialization failed\n");
+		return -EINVAL;
+	}
+	DRM_INFO("[FM]smu initialized\n");
+
+	return 0;
+}
+
+static int fiji_smu_start_in_non_protection_mode(struct amdgpu_device *adev)
+{
+	int i, result;
+	uint32_t val;
+
+	/* wait for smc boot up */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixRCU_UC_EVENTS);
+		val = REG_GET_FIELD(val, RCU_UC_EVENTS, boot_seq_done);
+		if (val)
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("SMC boot sequence is not completed\n");
+		return -EINVAL;
+	}
+
+	/* Clear firmware interrupt enable flag */
+	WREG32_SMC(ixFIRMWARE_FLAGS, 0);
+
+	/* Assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	result = fiji_smu_upload_firmware_image(adev);
+	if (result)
+		return result;
+
+	/* Set smc instruct start point at 0x0 */
+	fiji_program_jump_on_start(adev);
+
+	/* Enable clock */
+	val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 0);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+
+	/* De-assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 0);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	/* Wait for firmware to initialize */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixFIRMWARE_FLAGS);
+		if (REG_GET_FIELD(val, FIRMWARE_FLAGS, INTERRUPTS_ENABLED))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("Timeout for SMC firmware initialization\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int fiji_smu_start(struct amdgpu_device *adev)
+{
+	int result;
+	uint32_t val;
+
+	if (!fiji_is_smc_ram_running(adev)) {
+		val = RREG32_SMC(ixSMU_FIRMWARE);
+		if (!REG_GET_FIELD(val, SMU_FIRMWARE, SMU_MODE)) {
+			DRM_INFO("[FM]start smu in nonprotection mode\n");
+			result = fiji_smu_start_in_non_protection_mode(adev);
+			if (result)
+				return result;
+		} else {
+			DRM_INFO("[FM]start smu in protection mode\n");
+			result = fiji_smu_start_in_protection_mode(adev);
+			if (result)
+				return result;
+		}
+	}
+
+	return fiji_smu_request_load_fw(adev);
+}
+
+static const struct amdgpu_smumgr_funcs fiji_smumgr_funcs = {
+	.check_fw_load_finish = fiji_smu_check_fw_load_finish,
+	.request_smu_load_fw = NULL,
+	.request_smu_specific_fw = NULL,
+};
+
+int fiji_smu_init(struct amdgpu_device *adev)
+{
+	struct fiji_smu_private_data *private;
+	uint32_t image_size = ((sizeof(struct SMU_DRAMData_TOC) / 4096) + 1) * 4096;
+	uint32_t smu_internal_buffer_size = 200*4096;
+	struct amdgpu_bo **toc_buf = &adev->smu.toc_buf;
+	struct amdgpu_bo **smu_buf = &adev->smu.smu_buf;
+	uint64_t mc_addr;
+	void *toc_buf_ptr;
+	void *smu_buf_ptr;
+	int ret;
+
+	private = kzalloc(sizeof(struct fiji_smu_private_data), GFP_KERNEL);
+	if (NULL == private)
+		return -ENOMEM;
+
+	/* allocate firmware buffers */
+	if (adev->firmware.smu_load)
+		amdgpu_ucode_init_bo(adev);
+
+	adev->smu.priv = private;
+	adev->smu.fw_flags = 0;
+
+	/* Allocate FW image data structure and header buffer */
+	ret = amdgpu_bo_create(adev, image_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_VRAM,
+			       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			       NULL, NULL, toc_buf);
+	if (ret) {
+		DRM_ERROR("Failed to allocate memory for TOC buffer\n");
+		return -ENOMEM;
+	}
+
+	/* Allocate buffer for SMU internal buffer */
+	ret = amdgpu_bo_create(adev, smu_internal_buffer_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_VRAM,
+			       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			       NULL, NULL, smu_buf);
+	if (ret) {
+		DRM_ERROR("Failed to allocate memory for SMU internal buffer\n");
+		return -ENOMEM;
+	}
+
+	/* Retrieve GPU address for header buffer and internal buffer */
+	ret = amdgpu_bo_reserve(adev->smu.toc_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to reserve the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.toc_buf, AMDGPU_GEM_DOMAIN_VRAM, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to pin the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_kmap(*toc_buf, &toc_buf_ptr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to map the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	amdgpu_bo_unreserve(adev->smu.toc_buf);
+	private->header_addr_low = lower_32_bits(mc_addr);
+	private->header_addr_high = upper_32_bits(mc_addr);
+	private->header = toc_buf_ptr;
+
+	ret = amdgpu_bo_reserve(adev->smu.smu_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to reserve the SMU internal buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.smu_buf, AMDGPU_GEM_DOMAIN_VRAM, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to pin the SMU internal buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_kmap(*smu_buf, &smu_buf_ptr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to map the SMU internal buffer\n");
+		return -EINVAL;
+	}
+
+	amdgpu_bo_unreserve(adev->smu.smu_buf);
+	private->smu_buffer_addr_low = lower_32_bits(mc_addr);
+	private->smu_buffer_addr_high = upper_32_bits(mc_addr);
+
+	adev->smu.smumgr_funcs = &fiji_smumgr_funcs;
+
+	return 0;
+}
+
+int fiji_smu_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->smu.toc_buf);
+	amdgpu_bo_unref(&adev->smu.smu_buf);
+	kfree(adev->smu.priv);
+	adev->smu.priv = NULL;
+	if (adev->firmware.fw_buf)
+		amdgpu_ucode_fini_bo(adev);
+
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h
new file mode 100644
index 000000000..1cef03dee
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/fiji_smumgr.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef FIJI_SMUMGR_H
+#define FIJI_SMUMGR_H
+
+#include "fiji_ppsmc.h"
+
+int fiji_smu_init(struct amdgpu_device *adev);
+int fiji_smu_fini(struct amdgpu_device *adev);
+int fiji_smu_start(struct amdgpu_device *adev);
+
+struct fiji_smu_private_data
+{
+	uint8_t *header;
+	uint32_t smu_buffer_addr_high;
+	uint32_t smu_buffer_addr_low;
+	uint32_t header_addr_high;
+	uint32_t header_addr_low;
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
new file mode 100644
index 000000000..aa491540b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -0,0 +1,5691 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_gfx.h"
+#include "cikd.h"
+#include "cik.h"
+#include "atom.h"
+#include "amdgpu_ucode.h"
+#include "clearstate_ci.h"
+
+#include "uvd/uvd_4_2_d.h"
+
+#include "dce/dce_8_0_d.h"
+#include "dce/dce_8_0_sh_mask.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gca/gfx_7_0_d.h"
+#include "gca/gfx_7_2_enum.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#include "gmc/gmc_7_0_d.h"
+#include "gmc/gmc_7_0_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#define GFX7_NUM_GFX_RINGS     1
+#define GFX7_NUM_COMPUTE_RINGS 8
+
+static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev);
+int gfx_v7_0_get_cu_info(struct amdgpu_device *, struct amdgpu_cu_info *);
+
+MODULE_FIRMWARE("radeon/bonaire_pfp.bin");
+MODULE_FIRMWARE("radeon/bonaire_me.bin");
+MODULE_FIRMWARE("radeon/bonaire_ce.bin");
+MODULE_FIRMWARE("radeon/bonaire_rlc.bin");
+MODULE_FIRMWARE("radeon/bonaire_mec.bin");
+
+MODULE_FIRMWARE("radeon/hawaii_pfp.bin");
+MODULE_FIRMWARE("radeon/hawaii_me.bin");
+MODULE_FIRMWARE("radeon/hawaii_ce.bin");
+MODULE_FIRMWARE("radeon/hawaii_rlc.bin");
+MODULE_FIRMWARE("radeon/hawaii_mec.bin");
+
+MODULE_FIRMWARE("radeon/kaveri_pfp.bin");
+MODULE_FIRMWARE("radeon/kaveri_me.bin");
+MODULE_FIRMWARE("radeon/kaveri_ce.bin");
+MODULE_FIRMWARE("radeon/kaveri_rlc.bin");
+MODULE_FIRMWARE("radeon/kaveri_mec.bin");
+MODULE_FIRMWARE("radeon/kaveri_mec2.bin");
+
+MODULE_FIRMWARE("radeon/kabini_pfp.bin");
+MODULE_FIRMWARE("radeon/kabini_me.bin");
+MODULE_FIRMWARE("radeon/kabini_ce.bin");
+MODULE_FIRMWARE("radeon/kabini_rlc.bin");
+MODULE_FIRMWARE("radeon/kabini_mec.bin");
+
+MODULE_FIRMWARE("radeon/mullins_pfp.bin");
+MODULE_FIRMWARE("radeon/mullins_me.bin");
+MODULE_FIRMWARE("radeon/mullins_ce.bin");
+MODULE_FIRMWARE("radeon/mullins_rlc.bin");
+MODULE_FIRMWARE("radeon/mullins_mec.bin");
+
+static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
+{
+	{mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0},
+	{mmGDS_VMID1_BASE, mmGDS_VMID1_SIZE, mmGDS_GWS_VMID1, mmGDS_OA_VMID1},
+	{mmGDS_VMID2_BASE, mmGDS_VMID2_SIZE, mmGDS_GWS_VMID2, mmGDS_OA_VMID2},
+	{mmGDS_VMID3_BASE, mmGDS_VMID3_SIZE, mmGDS_GWS_VMID3, mmGDS_OA_VMID3},
+	{mmGDS_VMID4_BASE, mmGDS_VMID4_SIZE, mmGDS_GWS_VMID4, mmGDS_OA_VMID4},
+	{mmGDS_VMID5_BASE, mmGDS_VMID5_SIZE, mmGDS_GWS_VMID5, mmGDS_OA_VMID5},
+	{mmGDS_VMID6_BASE, mmGDS_VMID6_SIZE, mmGDS_GWS_VMID6, mmGDS_OA_VMID6},
+	{mmGDS_VMID7_BASE, mmGDS_VMID7_SIZE, mmGDS_GWS_VMID7, mmGDS_OA_VMID7},
+	{mmGDS_VMID8_BASE, mmGDS_VMID8_SIZE, mmGDS_GWS_VMID8, mmGDS_OA_VMID8},
+	{mmGDS_VMID9_BASE, mmGDS_VMID9_SIZE, mmGDS_GWS_VMID9, mmGDS_OA_VMID9},
+	{mmGDS_VMID10_BASE, mmGDS_VMID10_SIZE, mmGDS_GWS_VMID10, mmGDS_OA_VMID10},
+	{mmGDS_VMID11_BASE, mmGDS_VMID11_SIZE, mmGDS_GWS_VMID11, mmGDS_OA_VMID11},
+	{mmGDS_VMID12_BASE, mmGDS_VMID12_SIZE, mmGDS_GWS_VMID12, mmGDS_OA_VMID12},
+	{mmGDS_VMID13_BASE, mmGDS_VMID13_SIZE, mmGDS_GWS_VMID13, mmGDS_OA_VMID13},
+	{mmGDS_VMID14_BASE, mmGDS_VMID14_SIZE, mmGDS_GWS_VMID14, mmGDS_OA_VMID14},
+	{mmGDS_VMID15_BASE, mmGDS_VMID15_SIZE, mmGDS_GWS_VMID15, mmGDS_OA_VMID15}
+};
+
+static const u32 spectre_rlc_save_restore_register_list[] =
+{
+	(0x0e00 << 16) | (0xc12c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc140 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc150 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc15c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc168 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc170 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc178 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc204 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8228 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x829c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x869c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x98f4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x98f8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc260 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x90e8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c000 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c00c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c1c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8900 >> 2),
+	0x00000000,
+	0x3,
+	(0x0e00 << 16) | (0xc130 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc134 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc1fc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc208 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc264 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc268 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc26c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc270 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc274 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc278 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc27c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc280 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc284 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc288 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc28c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc290 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc294 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc298 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc29c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2ac  >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x301d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30238 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30250 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30254 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30258 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3025c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc99c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9834 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f00 >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f00 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f04 >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f04 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f08 >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f08 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f0c >> 2),
+	0x00000000,
+	(0x0001 << 16) | (0x30f0c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x9b7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bf0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bcc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8b24 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30a04 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a10 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a14 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a2c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc704 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc708 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc768 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc770 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc774 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc778 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc77c >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc780 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc784 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc788 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc78c >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc798 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc79c >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7a0 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7a4 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7a8 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7ac >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7b0 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc7b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9100 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c010 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92a8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92ac >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92c4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92c8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92cc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x92d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c04 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c38 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c3c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xae00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9604 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac08 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac0c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac58 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac68 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac6c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac70 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac74 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac78 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac80 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac84 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac88 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac8c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x970c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9714 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9718 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x971c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x8e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x9e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0xae00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0xbe00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88bc >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0x89c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8980 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30938 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3093c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30940 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30904 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c210 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c214 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c218 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8904 >> 2),
+	0x00000000,
+	0x5,
+	(0x0e00 << 16) | (0x8c28 >> 2),
+	(0x0e00 << 16) | (0x8c2c >> 2),
+	(0x0e00 << 16) | (0x8c30 >> 2),
+	(0x0e00 << 16) | (0x8c34 >> 2),
+	(0x0e00 << 16) | (0x9600 >> 2),
+};
+
+static const u32 kalindi_rlc_save_restore_register_list[] =
+{
+	(0x0e00 << 16) | (0xc12c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc140 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc150 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc15c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc168 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc170 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc204 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8228 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x829c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x869c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x98f4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x98f8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc260 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x90e8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c000 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c00c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c1c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xcd20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89bc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8900 >> 2),
+	0x00000000,
+	0x3,
+	(0x0e00 << 16) | (0xc130 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc134 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc1fc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc208 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc264 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc268 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc26c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc270 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc274 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc28c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc290 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc294 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc298 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2a8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc2ac >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x301d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30238 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30250 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30254 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30258 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3025c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc900 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc904 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc908 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc90c >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0xc910 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc99c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9834 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f00 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f04 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f08 >> 2),
+	0x00000000,
+	(0x0000 << 16) | (0x30f0c >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x9b7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bf0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8bcc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8b24 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30a04 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a10 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a14 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a18 >> 2),
+	0x00000000,
+	(0x0600 << 16) | (0x30a2c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc700 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc704 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc708 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xc768 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc770 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc774 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc798 >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0xc79c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9100 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c010 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c04 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c20 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c38 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8c3c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xae00 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9604 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac08 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac0c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac58 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac68 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac6c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac70 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac74 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac78 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac7c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac80 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac84 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac88 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xac8c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x970c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9714 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x9718 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x971c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x4e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x5e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x6e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x7e00 << 16) | (0x31068 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd10 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0xcd14 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88b8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88bc >> 2),
+	0x00000000,
+	(0x0400 << 16) | (0x89c0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88c8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x88d8 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8980 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30938 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3093c >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30940 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89a0 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30900 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x30904 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x89b4 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3e1fc >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c210 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c214 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x3c218 >> 2),
+	0x00000000,
+	(0x0e00 << 16) | (0x8904 >> 2),
+	0x00000000,
+	0x5,
+	(0x0e00 << 16) | (0x8c28 >> 2),
+	(0x0e00 << 16) | (0x8c2c >> 2),
+	(0x0e00 << 16) | (0x8c30 >> 2),
+	(0x0e00 << 16) | (0x8c34 >> 2),
+	(0x0e00 << 16) | (0x9600 >> 2),
+};
+
+static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev);
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer);
+static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev);
+static void gfx_v7_0_init_pg(struct amdgpu_device *adev);
+
+/*
+ * Core functions
+ */
+/**
+ * gfx_v7_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v7_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_KAVERI:
+		chip_name = "kaveri";
+		break;
+	case CHIP_KABINI:
+		chip_name = "kabini";
+		break;
+	case CHIP_MULLINS:
+		chip_name = "mullins";
+		break;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name);
+	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (err)
+		goto out;
+
+	snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name);
+	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (err)
+		goto out;
+
+	snprintf(fw_name, sizeof(fw_name), "radeon/%s_ce.bin", chip_name);
+	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (err)
+		goto out;
+
+	snprintf(fw_name, sizeof(fw_name), "radeon/%s_mec.bin", chip_name);
+	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	if (err)
+		goto out;
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		snprintf(fw_name, sizeof(fw_name), "radeon/%s_mec2.bin", chip_name);
+		err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
+		if (err)
+			goto out;
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", chip_name);
+	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "gfx7: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		release_firmware(adev->gfx.pfp_fw);
+		adev->gfx.pfp_fw = NULL;
+		release_firmware(adev->gfx.me_fw);
+		adev->gfx.me_fw = NULL;
+		release_firmware(adev->gfx.ce_fw);
+		adev->gfx.ce_fw = NULL;
+		release_firmware(adev->gfx.mec_fw);
+		adev->gfx.mec_fw = NULL;
+		release_firmware(adev->gfx.mec2_fw);
+		adev->gfx.mec2_fw = NULL;
+		release_firmware(adev->gfx.rlc_fw);
+		adev->gfx.rlc_fw = NULL;
+	}
+	return err;
+}
+
+/**
+ * gfx_v7_0_tiling_mode_table_init - init the hw tiling table
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Starting with SI, the tiling setup is done globally in a
+ * set of 32 tiling modes.  Rather than selecting each set of
+ * parameters per surface as on older asics, we just select
+ * which index in the tiling table we want to use, and the
+ * surface uses those parameters (CIK).
+ */
+static void gfx_v7_0_tiling_mode_table_init(struct amdgpu_device *adev)
+{
+	const u32 num_tile_mode_states = 32;
+	const u32 num_secondary_tile_mode_states = 16;
+	u32 reg_offset, gb_tile_moden, split_equal_to_row_size;
+
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_1KB;
+		break;
+	case 2:
+	default:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_2KB;
+		break;
+	case 4:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_4KB;
+		break;
+	}
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 7:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 12:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 17:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 23:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 30:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		break;
+	case CHIP_HAWAII:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 7:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 12:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 17:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 23:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 30:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		break;
+	case CHIP_KABINI:
+	case CHIP_KAVERI:
+	case CHIP_MULLINS:
+	default:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING) |
+						TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 7:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P2));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 12:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 17:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 23:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 30:
+				gb_tile_moden = (TILE_SPLIT(split_equal_to_row_size));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		break;
+	}
+}
+
+/**
+ * gfx_v7_0_select_se_sh - select which SE, SH to address
+ *
+ * @adev: amdgpu_device pointer
+ * @se_num: shader engine to address
+ * @sh_num: sh block to address
+ *
+ * Select which SE, SH combinations to address. Certain
+ * registers are instanced per SE or SH.  0xffffffff means
+ * broadcast to all SEs or SHs (CIK).
+ */
+void gfx_v7_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num)
+{
+	u32 data = GRBM_GFX_INDEX__INSTANCE_BROADCAST_WRITES_MASK;
+
+	if ((se_num == 0xffffffff) && (sh_num == 0xffffffff))
+		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+			GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK;
+	else if (se_num == 0xffffffff)
+		data |= GRBM_GFX_INDEX__SE_BROADCAST_WRITES_MASK |
+			(sh_num << GRBM_GFX_INDEX__SH_INDEX__SHIFT);
+	else if (sh_num == 0xffffffff)
+		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
+			(se_num << GRBM_GFX_INDEX__SE_INDEX__SHIFT);
+	else
+		data |= (sh_num << GRBM_GFX_INDEX__SH_INDEX__SHIFT) |
+			(se_num << GRBM_GFX_INDEX__SE_INDEX__SHIFT);
+	WREG32(mmGRBM_GFX_INDEX, data);
+}
+
+/**
+ * gfx_v7_0_create_bitmask - create a bitmask
+ *
+ * @bit_width: length of the mask
+ *
+ * create a variable length bit mask (CIK).
+ * Returns the bitmask.
+ */
+static u32 gfx_v7_0_create_bitmask(u32 bit_width)
+{
+	u32 i, mask = 0;
+
+	for (i = 0; i < bit_width; i++) {
+		mask <<= 1;
+		mask |= 1;
+	}
+	return mask;
+}
+
+/**
+ * gfx_v7_0_get_rb_disabled - computes the mask of disabled RBs
+ *
+ * @adev: amdgpu_device pointer
+ * @max_rb_num: max RBs (render backends) for the asic
+ * @se_num: number of SEs (shader engines) for the asic
+ * @sh_per_se: number of SH blocks per SE for the asic
+ *
+ * Calculates the bitmask of disabled RBs (CIK).
+ * Returns the disabled RB bitmask.
+ */
+static u32 gfx_v7_0_get_rb_disabled(struct amdgpu_device *adev,
+				    u32 max_rb_num_per_se,
+				    u32 sh_per_se)
+{
+	u32 data, mask;
+
+	data = RREG32(mmCC_RB_BACKEND_DISABLE);
+	if (data & 1)
+		data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
+	else
+		data = 0;
+
+	data |= RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+
+	data >>= GC_USER_RB_BACKEND_DISABLE__BACKEND_DISABLE__SHIFT;
+
+	mask = gfx_v7_0_create_bitmask(max_rb_num_per_se / sh_per_se);
+
+	return data & mask;
+}
+
+/**
+ * gfx_v7_0_setup_rb - setup the RBs on the asic
+ *
+ * @adev: amdgpu_device pointer
+ * @se_num: number of SEs (shader engines) for the asic
+ * @sh_per_se: number of SH blocks per SE for the asic
+ * @max_rb_num: max RBs (render backends) for the asic
+ *
+ * Configures per-SE/SH RB registers (CIK).
+ */
+static void gfx_v7_0_setup_rb(struct amdgpu_device *adev,
+			      u32 se_num, u32 sh_per_se,
+			      u32 max_rb_num_per_se)
+{
+	int i, j;
+	u32 data, mask;
+	u32 disabled_rbs = 0;
+	u32 enabled_rbs = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < se_num; i++) {
+		for (j = 0; j < sh_per_se; j++) {
+			gfx_v7_0_select_se_sh(adev, i, j);
+			data = gfx_v7_0_get_rb_disabled(adev, max_rb_num_per_se, sh_per_se);
+			if (adev->asic_type == CHIP_HAWAII)
+				disabled_rbs |= data << ((i * sh_per_se + j) * HAWAII_RB_BITMAP_WIDTH_PER_SH);
+			else
+				disabled_rbs |= data << ((i * sh_per_se + j) * CIK_RB_BITMAP_WIDTH_PER_SH);
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = 1;
+	for (i = 0; i < max_rb_num_per_se * se_num; i++) {
+		if (!(disabled_rbs & mask))
+			enabled_rbs |= mask;
+		mask <<= 1;
+	}
+
+	adev->gfx.config.backend_enable_mask = enabled_rbs;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < se_num; i++) {
+		gfx_v7_0_select_se_sh(adev, i, 0xffffffff);
+		data = 0;
+		for (j = 0; j < sh_per_se; j++) {
+			switch (enabled_rbs & 3) {
+			case 0:
+				if (j == 0)
+					data |= (RASTER_CONFIG_RB_MAP_3 <<
+						PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT);
+				else
+					data |= (RASTER_CONFIG_RB_MAP_0 <<
+						PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT);
+				break;
+			case 1:
+				data |= (RASTER_CONFIG_RB_MAP_0 << (i * sh_per_se + j) * 2);
+				break;
+			case 2:
+				data |= (RASTER_CONFIG_RB_MAP_3 << (i * sh_per_se + j) * 2);
+				break;
+			case 3:
+			default:
+				data |= (RASTER_CONFIG_RB_MAP_2 << (i * sh_per_se + j) * 2);
+				break;
+			}
+			enabled_rbs >>= 2;
+		}
+		WREG32(mmPA_SC_RASTER_CONFIG, data);
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * gmc_v7_0_init_compute_vmid - gart enable
+ *
+ * @rdev: amdgpu_device pointer
+ *
+ * Initialize compute vmid sh_mem registers
+ *
+ */
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+#define FIRST_COMPUTE_VMID	(8)
+#define LAST_COMPUTE_VMID	(16)
+static void gmc_v7_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_config;
+	uint32_t sh_mem_bases;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	*/
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+	sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
+	sh_mem_config |= MTYPE_NONCACHED << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT;
+	mutex_lock(&adev->srbm_mutex);
+	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+		cik_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, sh_mem_bases);
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+/**
+ * gfx_v7_0_gpu_init - setup the 3D engine
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Configures the 3D engine and tiling configuration
+ * registers so that the 3D engine is usable.
+ */
+static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 dimm00_addr_map, dimm01_addr_map, dimm10_addr_map, dimm11_addr_map;
+	u32 sh_mem_cfg;
+	u32 tmp;
+	int i;
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		adev->gfx.config.max_shader_engines = 2;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 7;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = BONAIRE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_HAWAII:
+		adev->gfx.config.max_shader_engines = 4;
+		adev->gfx.config.max_tile_pipes = 16;
+		adev->gfx.config.max_cu_per_sh = 11;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 16;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = HAWAII_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_KAVERI:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 4;
+		if ((adev->pdev->device == 0x1304) ||
+		    (adev->pdev->device == 0x1305) ||
+		    (adev->pdev->device == 0x130C) ||
+		    (adev->pdev->device == 0x130F) ||
+		    (adev->pdev->device == 0x1310) ||
+		    (adev->pdev->device == 0x1311) ||
+		    (adev->pdev->device == 0x131C)) {
+			adev->gfx.config.max_cu_per_sh = 8;
+			adev->gfx.config.max_backends_per_se = 2;
+		} else if ((adev->pdev->device == 0x1309) ||
+			   (adev->pdev->device == 0x130A) ||
+			   (adev->pdev->device == 0x130D) ||
+			   (adev->pdev->device == 0x1313) ||
+			   (adev->pdev->device == 0x131D)) {
+			adev->gfx.config.max_cu_per_sh = 6;
+			adev->gfx.config.max_backends_per_se = 2;
+		} else if ((adev->pdev->device == 0x1306) ||
+			   (adev->pdev->device == 0x1307) ||
+			   (adev->pdev->device == 0x130B) ||
+			   (adev->pdev->device == 0x130E) ||
+			   (adev->pdev->device == 0x1315) ||
+			   (adev->pdev->device == 0x131B)) {
+			adev->gfx.config.max_cu_per_sh = 4;
+			adev->gfx.config.max_backends_per_se = 1;
+		} else {
+			adev->gfx.config.max_cu_per_sh = 3;
+			adev->gfx.config.max_backends_per_se = 1;
+		}
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = BONAIRE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+	default:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_cu_per_sh = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 1;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = BONAIRE_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	}
+
+	WREG32(mmGRBM_CNTL, (0xff << GRBM_CNTL__READ_TIMEOUT__SHIFT));
+
+	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
+	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
+	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
+
+	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
+	adev->gfx.config.mem_max_burst_length_bytes = 256;
+	if (adev->flags & AMD_IS_APU) {
+		/* Get memory bank mapping mode. */
+		tmp = RREG32(mmMC_FUS_DRAM0_BANK_ADDR_MAPPING);
+		dimm00_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm01_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		tmp = RREG32(mmMC_FUS_DRAM1_BANK_ADDR_MAPPING);
+		dimm10_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm11_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		/* Validate settings in case only one DIMM installed. */
+		if ((dimm00_addr_map == 0) || (dimm00_addr_map == 3) || (dimm00_addr_map == 4) || (dimm00_addr_map > 12))
+			dimm00_addr_map = 0;
+		if ((dimm01_addr_map == 0) || (dimm01_addr_map == 3) || (dimm01_addr_map == 4) || (dimm01_addr_map > 12))
+			dimm01_addr_map = 0;
+		if ((dimm10_addr_map == 0) || (dimm10_addr_map == 3) || (dimm10_addr_map == 4) || (dimm10_addr_map > 12))
+			dimm10_addr_map = 0;
+		if ((dimm11_addr_map == 0) || (dimm11_addr_map == 3) || (dimm11_addr_map == 4) || (dimm11_addr_map > 12))
+			dimm11_addr_map = 0;
+
+		/* If DIMM Addr map is 8GB, ROW size should be 2KB. Otherwise 1KB. */
+		/* If ROW size(DIMM1) != ROW size(DMIMM0), ROW size should be larger one. */
+		if ((dimm00_addr_map == 11) || (dimm01_addr_map == 11) || (dimm10_addr_map == 11) || (dimm11_addr_map == 11))
+			adev->gfx.config.mem_row_size_in_kb = 2;
+		else
+			adev->gfx.config.mem_row_size_in_kb = 1;
+	} else {
+		tmp = (mc_arb_ramcfg & MC_ARB_RAMCFG__NOOFCOLS_MASK) >> MC_ARB_RAMCFG__NOOFCOLS__SHIFT;
+		adev->gfx.config.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024;
+		if (adev->gfx.config.mem_row_size_in_kb > 4)
+			adev->gfx.config.mem_row_size_in_kb = 4;
+	}
+	/* XXX use MC settings? */
+	adev->gfx.config.shader_engine_tile_size = 32;
+	adev->gfx.config.num_gpus = 1;
+	adev->gfx.config.multi_gpu_tile_size = 64;
+
+	/* fix up row size */
+	gb_addr_config &= ~GB_ADDR_CONFIG__ROW_SIZE_MASK;
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+	default:
+		gb_addr_config |= (0 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT);
+		break;
+	case 2:
+		gb_addr_config |= (1 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT);
+		break;
+	case 4:
+		gb_addr_config |= (2 << GB_ADDR_CONFIG__ROW_SIZE__SHIFT);
+		break;
+	}
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+
+	WREG32(mmGB_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmHDP_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmDMIF_ADDR_CALC, gb_addr_config);
+	WREG32(mmSDMA0_TILING_CONFIG + SDMA0_REGISTER_OFFSET, gb_addr_config & 0x70);
+	WREG32(mmSDMA0_TILING_CONFIG + SDMA1_REGISTER_OFFSET, gb_addr_config & 0x70);
+	WREG32(mmUVD_UDEC_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, gb_addr_config);
+	WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, gb_addr_config);
+
+	gfx_v7_0_tiling_mode_table_init(adev);
+
+	gfx_v7_0_setup_rb(adev, adev->gfx.config.max_shader_engines,
+			  adev->gfx.config.max_sh_per_se,
+			  adev->gfx.config.max_backends_per_se);
+
+	/* set HW defaults for 3D engine */
+	WREG32(mmCP_MEQ_THRESHOLDS,
+			(0x30 << CP_MEQ_THRESHOLDS__MEQ1_START__SHIFT) |
+			(0x60 << CP_MEQ_THRESHOLDS__MEQ2_START__SHIFT));
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/*
+	 * making sure that the following register writes will be broadcasted
+	 * to all the shaders
+	 */
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	sh_mem_cfg = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE, 
+				   SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < 16; i++) {
+		cik_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32(mmSH_MEM_CONFIG, sh_mem_cfg);
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, 0);
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	gmc_v7_0_init_compute_vmid(adev);
+
+	WREG32(mmSX_DEBUG_1, 0x20);
+
+	WREG32(mmTA_CNTL_AUX, 0x00010000);
+
+	tmp = RREG32(mmSPI_CONFIG_CNTL);
+	tmp |= 0x03000000;
+	WREG32(mmSPI_CONFIG_CNTL, tmp);
+
+	WREG32(mmSQ_CONFIG, 1);
+
+	WREG32(mmDB_DEBUG, 0);
+
+	tmp = RREG32(mmDB_DEBUG2) & ~0xf00fffff;
+	tmp |= 0x00000400;
+	WREG32(mmDB_DEBUG2, tmp);
+
+	tmp = RREG32(mmDB_DEBUG3) & ~0x0002021c;
+	tmp |= 0x00020200;
+	WREG32(mmDB_DEBUG3, tmp);
+
+	tmp = RREG32(mmCB_HW_CONTROL) & ~0x00010000;
+	tmp |= 0x00018208;
+	WREG32(mmCB_HW_CONTROL, tmp);
+
+	WREG32(mmSPI_CONFIG_CNTL_1, (4 << SPI_CONFIG_CNTL_1__VTX_DONE_DELAY__SHIFT));
+
+	WREG32(mmPA_SC_FIFO_SIZE,
+		((adev->gfx.config.sc_prim_fifo_size_frontend << PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
+		(adev->gfx.config.sc_prim_fifo_size_backend << PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
+		(adev->gfx.config.sc_hiz_tile_fifo_size << PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
+		(adev->gfx.config.sc_earlyz_tile_fifo_size << PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT)));
+
+	WREG32(mmVGT_NUM_INSTANCES, 1);
+
+	WREG32(mmCP_PERFMON_CNTL, 0);
+
+	WREG32(mmSQ_CONFIG, 0);
+
+	WREG32(mmPA_SC_FORCE_EOV_MAX_CNTS,
+		((4095 << PA_SC_FORCE_EOV_MAX_CNTS__FORCE_EOV_MAX_CLK_CNT__SHIFT) |
+		(255 << PA_SC_FORCE_EOV_MAX_CNTS__FORCE_EOV_MAX_REZ_CNT__SHIFT)));
+
+	WREG32(mmVGT_CACHE_INVALIDATION,
+		(VC_AND_TC << VGT_CACHE_INVALIDATION__CACHE_INVALIDATION__SHIFT) |
+		(ES_AND_GS_AUTO << VGT_CACHE_INVALIDATION__AUTO_INVLD_EN__SHIFT));
+
+	WREG32(mmVGT_GS_VERTEX_REUSE, 16);
+	WREG32(mmPA_SC_LINE_STIPPLE_STATE, 0);
+
+	WREG32(mmPA_CL_ENHANCE, PA_CL_ENHANCE__CLIP_VTX_REORDER_ENA_MASK |
+			(3 << PA_CL_ENHANCE__NUM_CLIP_SEQ__SHIFT));
+	WREG32(mmPA_SC_ENHANCE, PA_SC_ENHANCE__ENABLE_PA_SC_OUT_OF_ORDER_MASK);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	udelay(50);
+}
+
+/*
+ * GPU scratch registers helpers function.
+ */
+/**
+ * gfx_v7_0_scratch_init - setup driver info for CP scratch regs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the number and offset of the CP scratch registers.
+ * NOTE: use of CP scratch registers is a legacy inferface and
+ * is not used by default on newer asics (r6xx+).  On newer asics,
+ * memory buffers are used for fences rather than scratch regs.
+ */
+static void gfx_v7_0_scratch_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.scratch.num_reg = 7;
+	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
+	for (i = 0; i < adev->gfx.scratch.num_reg; i++) {
+		adev->gfx.scratch.free[i] = true;
+		adev->gfx.scratch.reg[i] = adev->gfx.scratch.reg_base + i;
+	}
+}
+
+/**
+ * gfx_v7_0_ring_test_ring - basic gfx ring test
+ *
+ * @adev: amdgpu_device pointer
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Allocate a scratch register and write to it using the gfx ring (CIK).
+ * Provides a basic gfx ring test to verify that the ring is working.
+ * Used by gfx_v7_0_cp_gfx_resume();
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v7_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_lock(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+			  ring->idx, scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+/**
+ * gfx_v7_0_ring_emit_hdp - emit an hdp flush on the cp
+ *
+ * @adev: amdgpu_device pointer
+ * @ridx: amdgpu ring index
+ *
+ * Emits an hdp flush on the cp.
+ */
+static void gfx_v7_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask;
+	int usepfp = ring->type == AMDGPU_RING_TYPE_COMPUTE ? 0 : 1;
+
+	if (ring->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP2_MASK << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP6_MASK << ring->pipe;
+			break;
+		default:
+			return;
+		}
+	} else {
+		ref_and_mask = GPU_HDP_FLUSH_DONE__CP0_MASK;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(1) | /* write, wait, write */
+				 WAIT_REG_MEM_FUNCTION(3) |  /* == */
+				 WAIT_REG_MEM_ENGINE(usepfp)));   /* pfp or me */
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+}
+
+/**
+ * gfx_v7_0_ring_emit_fence_gfx - emit a fence on the gfx ring
+ *
+ * @adev: amdgpu_device pointer
+ * @fence: amdgpu fence object
+ *
+ * Emits a fence sequnce number on the gfx ring and flushes
+ * GPU caches.
+ */
+static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+	/* Workaround for cache flush problems. First send a dummy EOP
+	 * event down the pipe with seq one below.
+	 */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+				DATA_SEL(1) | INT_SEL(0));
+	amdgpu_ring_write(ring, lower_32_bits(seq - 1));
+	amdgpu_ring_write(ring, upper_32_bits(seq - 1));
+
+	/* Then send the real EOP event down the pipe. */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) |
+				DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+/**
+ * gfx_v7_0_ring_emit_fence_compute - emit a fence on the compute ring
+ *
+ * @adev: amdgpu_device pointer
+ * @fence: amdgpu fence object
+ *
+ * Emits a fence sequnce number on the compute ring and flushes
+ * GPU caches.
+ */
+static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
+					     u64 addr, u64 seq,
+					     unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 5));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+/**
+ * gfx_v7_0_ring_emit_semaphore - emit a semaphore on the CP ring
+ *
+ * @ring: amdgpu ring buffer object
+ * @semaphore: amdgpu semaphore object
+ * @emit_wait: Is this a sempahore wait?
+ *
+ * Emits a semaphore signal/wait packet to the CP ring and prevents the PFP
+ * from running ahead of semaphore waits.
+ */
+static bool gfx_v7_0_ring_emit_semaphore(struct amdgpu_ring *ring,
+					 struct amdgpu_semaphore *semaphore,
+					 bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+	unsigned sel = emit_wait ? PACKET3_SEM_SEL_WAIT : PACKET3_SEM_SEL_SIGNAL;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_MEM_SEMAPHORE, 1));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) | sel);
+
+	if (emit_wait && (ring->type == AMDGPU_RING_TYPE_GFX)) {
+		/* Prevent the PFP from running ahead of the semaphore wait */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+
+	return true;
+}
+
+/*
+ * IB stuff
+ */
+/**
+ * gfx_v7_0_ring_emit_ib - emit an IB (Indirect Buffer) on the ring
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @ib: amdgpu indirect buffer object
+ *
+ * Emits an DE (drawing engine) or CE (constant engine) IB
+ * on the gfx ring.  IBs are usually generated by userspace
+ * acceleration drivers and submitted to the kernel for
+ * sheduling on the ring.  This function schedules the IB
+ * on the gfx ring for execution by the GPU.
+ */
+static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	bool need_ctx_switch = ring->current_ctx != ib->ctx;
+	u32 header, control = 0;
+	u32 next_rptr = ring->wptr + 5;
+
+	/* drop the CE preamble IB for the same context */
+	if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) && !need_ctx_switch)
+		return;
+
+	if (need_ctx_switch)
+		next_rptr += 2;
+
+	next_rptr += 4;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
+	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, next_rptr);
+
+	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
+	if (need_ctx_switch) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw |
+		(ib->vm ? (ib->vm->ids[ring->idx].id << 24) : 0);
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+			  (2 << 0) |
+#endif
+			  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	u32 header, control = 0;
+	u32 next_rptr = ring->wptr + 5;
+
+	control |= INDIRECT_BUFFER_VALID;
+	next_rptr += 4;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
+	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, next_rptr);
+
+	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw |
+			   (ib->vm ? (ib->vm->ids[ring->idx].id << 24) : 0);
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+					  (2 << 0) |
+#endif
+					  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+/**
+ * gfx_v7_0_ring_test_ib - basic ring IB test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Allocate an IB and execute it on the gfx ring (CIK).
+ * Provides a basic gfx ring test to verify that IBs are working.
+ * Returns 0 on success, error on failure.
+ */
+static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct fence *f = NULL;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(ring, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		goto err1;
+	}
+	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
+	ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START));
+	ib.ptr[2] = 0xDEADBEEF;
+	ib.length_dw = 3;
+
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err2;
+
+	r = fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto err2;
+	}
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
+			 ring->idx, i);
+		goto err2;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
+			  scratch, tmp);
+		r = -EINVAL;
+	}
+
+err2:
+	fence_put(f);
+	amdgpu_ib_free(adev, &ib);
+err1:
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+/*
+ * CP.
+ * On CIK, gfx and compute now have independant command processors.
+ *
+ * GFX
+ * Gfx consists of a single ring and can process both gfx jobs and
+ * compute jobs.  The gfx CP consists of three microengines (ME):
+ * PFP - Pre-Fetch Parser
+ * ME - Micro Engine
+ * CE - Constant Engine
+ * The PFP and ME make up what is considered the Drawing Engine (DE).
+ * The CE is an asynchronous engine used for updating buffer desciptors
+ * used by the DE so that they can be loaded into cache in parallel
+ * while the DE is processing state update packets.
+ *
+ * Compute
+ * The compute CP consists of two microengines (ME):
+ * MEC1 - Compute MicroEngine 1
+ * MEC2 - Compute MicroEngine 2
+ * Each MEC supports 4 compute pipes and each pipe supports 8 queues.
+ * The queues are exposed to userspace and are programmed directly
+ * by the compute runtime.
+ */
+/**
+ * gfx_v7_0_cp_gfx_enable - enable/disable the gfx CP MEs
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable the MEs
+ *
+ * Halts or unhalts the gfx MEs.
+ */
+static void gfx_v7_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32(mmCP_ME_CNTL, 0);
+	} else {
+		WREG32(mmCP_ME_CNTL, (CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			adev->gfx.gfx_ring[i].ready = false;
+	}
+	udelay(50);
+}
+
+/**
+ * gfx_v7_0_cp_gfx_load_microcode - load the gfx CP ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the gfx PFP, ME, and CE ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int gfx_v7_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+	adev->gfx.pfp_fw_version = le32_to_cpu(pfp_hdr->header.ucode_version);
+	adev->gfx.ce_fw_version = le32_to_cpu(ce_hdr->header.ucode_version);
+	adev->gfx.me_fw_version = le32_to_cpu(me_hdr->header.ucode_version);
+	adev->gfx.me_feature_version = le32_to_cpu(me_hdr->ucode_feature_version);
+	adev->gfx.ce_feature_version = le32_to_cpu(ce_hdr->ucode_feature_version);
+	adev->gfx.pfp_feature_version = le32_to_cpu(pfp_hdr->ucode_feature_version);
+
+	gfx_v7_0_cp_gfx_enable(adev, false);
+
+	/* PFP */
+	fw_data = (const __le32 *)
+		(adev->gfx.pfp_fw->data +
+		 le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	/* CE */
+	fw_data = (const __le32 *)
+		(adev->gfx.ce_fw->data +
+		 le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
+
+	/* ME */
+	fw_data = (const __le32 *)
+		(adev->gfx.me_fw->data +
+		 le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_ME_RAM_WADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_gfx_start - start the gfx ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enables the ring and loads the clear state context and other
+ * packets required to init the ring.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i;
+
+	/* init the CP */
+	WREG32(mmCP_MAX_CONTEXT, adev->gfx.config.max_hw_contexts - 1);
+	WREG32(mmCP_ENDIAN_SWAP, 0);
+	WREG32(mmCP_DEVICE_ID, 1);
+
+	gfx_v7_0_cp_gfx_enable(adev, true);
+
+	r = amdgpu_ring_lock(ring, gfx_v7_0_get_csb_size(adev) + 8);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	/* init the CE partitions.  CE only used for gfx on CIK */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0x8000);
+	amdgpu_ring_write(ring, 0x8000);
+
+	/* clear state buffer */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+						  PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				amdgpu_ring_write(ring, ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		amdgpu_ring_write(ring, 0x16000012);
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
+	case CHIP_KAVERI:
+		amdgpu_ring_write(ring, 0x00000000); /* XXX */
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		amdgpu_ring_write(ring, 0x00000000); /* XXX */
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
+	case CHIP_HAWAII:
+		amdgpu_ring_write(ring, 0x3a00161a);
+		amdgpu_ring_write(ring, 0x0000002e);
+		break;
+	default:
+		amdgpu_ring_write(ring, 0x00000000);
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, 0x00000316);
+	amdgpu_ring_write(ring, 0x0000000e); /* VGT_VERTEX_REUSE_BLOCK_CNTL */
+	amdgpu_ring_write(ring, 0x00000010); /* VGT_OUT_DEALLOC_CNTL */
+
+	amdgpu_ring_unlock_commit(ring);
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_gfx_resume - setup the gfx ring buffer registers
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Program the location and size of the gfx ring buffer
+ * and test it to make sure it's working.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr;
+	int r;
+
+	WREG32(mmCP_SEM_WAIT_TIMER, 0x0);
+	if (adev->asic_type != CHIP_HAWAII)
+		WREG32(mmCP_SEM_INCOMPLETE_TIMER_CNTL, 0x0);
+
+	/* Set the write pointer delay */
+	WREG32(mmCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32(mmCP_RB_VMID, 0);
+
+	WREG32(mmSCRATCH_ADDR, 0);
+
+	/* ring 0 - compute and gfx */
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = (order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
+#ifdef __BIG_ENDIAN
+	tmp |= 2 << CP_RB0_CNTL__BUF_SWAP__SHIFT;
+#endif
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmCP_RB0_CNTL, tmp | CP_RB0_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB0_WPTR, ring->wptr);
+
+	/* set the wb address wether it's enabled or not */
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	/* scratch register shadowing is no longer supported */
+	WREG32(mmSCRATCH_UMSK, 0);
+
+	mdelay(1);
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32(mmCP_RB0_BASE, rb_addr);
+	WREG32(mmCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	/* start the ring */
+	gfx_v7_0_cp_gfx_start(adev);
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	return 0;
+}
+
+static u32 gfx_v7_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	rptr = ring->adev->wb.wb[ring->rptr_offs];
+
+	return rptr;
+}
+
+static u32 gfx_v7_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 wptr;
+
+	wptr = RREG32(mmCP_RB0_WPTR);
+
+	return wptr;
+}
+
+static void gfx_v7_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmCP_RB0_WPTR, ring->wptr);
+	(void)RREG32(mmCP_RB0_WPTR);
+}
+
+static u32 gfx_v7_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	rptr = ring->adev->wb.wb[ring->rptr_offs];
+
+	return rptr;
+}
+
+static u32 gfx_v7_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	u32 wptr;
+
+	/* XXX check if swapping is necessary on BE */
+	wptr = ring->adev->wb.wb[ring->wptr_offs];
+
+	return wptr;
+}
+
+static void gfx_v7_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	adev->wb.wb[ring->wptr_offs] = ring->wptr;
+	WDOORBELL32(ring->doorbell_index, ring->wptr);
+}
+
+/**
+ * gfx_v7_0_cp_compute_enable - enable/disable the compute CP MEs
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable the MEs
+ *
+ * Halts or unhalts the compute MEs.
+ */
+static void gfx_v7_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32(mmCP_MEC_CNTL, 0);
+	} else {
+		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			adev->gfx.compute_ring[i].ready = false;
+	}
+	udelay(50);
+}
+
+/**
+ * gfx_v7_0_cp_compute_load_microcode - load the compute CP ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the compute MEC1&2 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int gfx_v7_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+	adev->gfx.mec_fw_version = le32_to_cpu(mec_hdr->header.ucode_version);
+	adev->gfx.mec_feature_version = le32_to_cpu(
+					mec_hdr->ucode_feature_version);
+
+	gfx_v7_0_cp_compute_enable(adev, false);
+
+	/* MEC1 */
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_MEC_ME1_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		const struct gfx_firmware_header_v1_0 *mec2_hdr;
+
+		if (!adev->gfx.mec2_fw)
+			return -EINVAL;
+
+		mec2_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+		amdgpu_ucode_print_gfx_hdr(&mec2_hdr->header);
+		adev->gfx.mec2_fw_version = le32_to_cpu(mec2_hdr->header.ucode_version);
+		adev->gfx.mec2_feature_version = le32_to_cpu(
+				mec2_hdr->ucode_feature_version);
+
+		/* MEC2 */
+		fw_data = (const __le32 *)
+			(adev->gfx.mec2_fw->data +
+			 le32_to_cpu(mec2_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(mec2_hdr->header.ucode_size_bytes) / 4;
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
+		for (i = 0; i < fw_size; i++)
+			WREG32(mmCP_MEC_ME2_UCODE_DATA, le32_to_cpup(fw_data++));
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
+	}
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_compute_start - start the compute queues
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the compute queues.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_compute_start(struct amdgpu_device *adev)
+{
+	gfx_v7_0_cp_compute_enable(adev, true);
+
+	return 0;
+}
+
+/**
+ * gfx_v7_0_cp_compute_fini - stop the compute queues
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute queues and tear down the driver queue
+ * info.
+ */
+static void gfx_v7_0_cp_compute_fini(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		if (ring->mqd_obj) {
+			r = amdgpu_bo_reserve(ring->mqd_obj, false);
+			if (unlikely(r != 0))
+				dev_warn(adev->dev, "(%d) reserve MQD bo failed\n", r);
+
+			amdgpu_bo_unpin(ring->mqd_obj);
+			amdgpu_bo_unreserve(ring->mqd_obj);
+
+			amdgpu_bo_unref(&ring->mqd_obj);
+			ring->mqd_obj = NULL;
+		}
+	}
+}
+
+static void gfx_v7_0_mec_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gfx.mec.hpd_eop_obj) {
+		r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
+		if (unlikely(r != 0))
+			dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
+		amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+		amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
+		adev->gfx.mec.hpd_eop_obj = NULL;
+	}
+}
+
+#define MEC_HPD_SIZE 2048
+
+static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+
+	/*
+	 * KV:    2 MEC, 4 Pipes/MEC, 8 Queues/Pipe - 64 Queues total
+	 * CI/KB: 1 MEC, 4 Pipes/MEC, 8 Queues/Pipe - 32 Queues total
+	 * Nonetheless, we assign only 1 pipe because all other pipes will
+	 * be handled by KFD
+	 */
+	adev->gfx.mec.num_mec = 1;
+	adev->gfx.mec.num_pipe = 1;
+	adev->gfx.mec.num_queue = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe * 8;
+
+	if (adev->gfx.mec.hpd_eop_obj == NULL) {
+		r = amdgpu_bo_create(adev,
+				     adev->gfx.mec.num_mec *adev->gfx.mec.num_pipe * MEC_HPD_SIZE * 2,
+				     PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
+				     &adev->gfx.mec.hpd_eop_obj);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			return r;
+		}
+	}
+
+	r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
+	if (unlikely(r != 0)) {
+		gfx_v7_0_mec_fini(adev);
+		return r;
+	}
+	r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
+			  &adev->gfx.mec.hpd_eop_gpu_addr);
+	if (r) {
+		dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
+		gfx_v7_0_mec_fini(adev);
+		return r;
+	}
+	r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
+		gfx_v7_0_mec_fini(adev);
+		return r;
+	}
+
+	/* clear memory.  Not sure if this is required or not */
+	memset(hpd, 0, adev->gfx.mec.num_mec *adev->gfx.mec.num_pipe * MEC_HPD_SIZE * 2);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+	return 0;
+}
+
+struct hqd_registers
+{
+	u32 cp_mqd_base_addr;
+	u32 cp_mqd_base_addr_hi;
+	u32 cp_hqd_active;
+	u32 cp_hqd_vmid;
+	u32 cp_hqd_persistent_state;
+	u32 cp_hqd_pipe_priority;
+	u32 cp_hqd_queue_priority;
+	u32 cp_hqd_quantum;
+	u32 cp_hqd_pq_base;
+	u32 cp_hqd_pq_base_hi;
+	u32 cp_hqd_pq_rptr;
+	u32 cp_hqd_pq_rptr_report_addr;
+	u32 cp_hqd_pq_rptr_report_addr_hi;
+	u32 cp_hqd_pq_wptr_poll_addr;
+	u32 cp_hqd_pq_wptr_poll_addr_hi;
+	u32 cp_hqd_pq_doorbell_control;
+	u32 cp_hqd_pq_wptr;
+	u32 cp_hqd_pq_control;
+	u32 cp_hqd_ib_base_addr;
+	u32 cp_hqd_ib_base_addr_hi;
+	u32 cp_hqd_ib_rptr;
+	u32 cp_hqd_ib_control;
+	u32 cp_hqd_iq_timer;
+	u32 cp_hqd_iq_rptr;
+	u32 cp_hqd_dequeue_request;
+	u32 cp_hqd_dma_offload;
+	u32 cp_hqd_sema_cmd;
+	u32 cp_hqd_msg_type;
+	u32 cp_hqd_atomic0_preop_lo;
+	u32 cp_hqd_atomic0_preop_hi;
+	u32 cp_hqd_atomic1_preop_lo;
+	u32 cp_hqd_atomic1_preop_hi;
+	u32 cp_hqd_hq_scheduler0;
+	u32 cp_hqd_hq_scheduler1;
+	u32 cp_mqd_control;
+};
+
+struct bonaire_mqd
+{
+	u32 header;
+	u32 dispatch_initiator;
+	u32 dimensions[3];
+	u32 start_idx[3];
+	u32 num_threads[3];
+	u32 pipeline_stat_enable;
+	u32 perf_counter_enable;
+	u32 pgm[2];
+	u32 tba[2];
+	u32 tma[2];
+	u32 pgm_rsrc[2];
+	u32 vmid;
+	u32 resource_limits;
+	u32 static_thread_mgmt01[2];
+	u32 tmp_ring_size;
+	u32 static_thread_mgmt23[2];
+	u32 restart[3];
+	u32 thread_trace_enable;
+	u32 reserved1;
+	u32 user_data[16];
+	u32 vgtcs_invoke_count[2];
+	struct hqd_registers queue_state;
+	u32 dequeue_cntr;
+	u32 interrupt_queue[64];
+};
+
+/**
+ * gfx_v7_0_cp_compute_resume - setup the compute queue registers
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Program the compute queues and test them to make sure they
+ * are working.
+ * Returns 0 for success, error for failure.
+ */
+static int gfx_v7_0_cp_compute_resume(struct amdgpu_device *adev)
+{
+	int r, i, j;
+	u32 tmp;
+	bool use_doorbell = true;
+	u64 hqd_gpu_addr;
+	u64 mqd_gpu_addr;
+	u64 eop_gpu_addr;
+	u64 wb_gpu_addr;
+	u32 *buf;
+	struct bonaire_mqd *mqd;
+
+	r = gfx_v7_0_cp_compute_start(adev);
+	if (r)
+		return r;
+
+	/* fix up chicken bits */
+	tmp = RREG32(mmCP_CPF_DEBUG);
+	tmp |= (1 << 23);
+	WREG32(mmCP_CPF_DEBUG, tmp);
+
+	/* init the pipes */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < (adev->gfx.mec.num_pipe * adev->gfx.mec.num_mec); i++) {
+		int me = (i < 4) ? 1 : 2;
+		int pipe = (i < 4) ? i : (i - 4);
+
+		eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr + (i * MEC_HPD_SIZE * 2);
+
+		cik_srbm_select(adev, me, pipe, 0, 0);
+
+		/* write the EOP addr */
+		WREG32(mmCP_HPD_EOP_BASE_ADDR, eop_gpu_addr >> 8);
+		WREG32(mmCP_HPD_EOP_BASE_ADDR_HI, upper_32_bits(eop_gpu_addr) >> 8);
+
+		/* set the VMID assigned */
+		WREG32(mmCP_HPD_EOP_VMID, 0);
+
+		/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+		tmp = RREG32(mmCP_HPD_EOP_CONTROL);
+		tmp &= ~CP_HPD_EOP_CONTROL__EOP_SIZE_MASK;
+		tmp |= order_base_2(MEC_HPD_SIZE / 8);
+		WREG32(mmCP_HPD_EOP_CONTROL, tmp);
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* init the queues.  Just two for now. */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		if (ring->mqd_obj == NULL) {
+			r = amdgpu_bo_create(adev,
+					     sizeof(struct bonaire_mqd),
+					     PAGE_SIZE, true,
+					     AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
+					     &ring->mqd_obj);
+			if (r) {
+				dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
+				return r;
+			}
+		}
+
+		r = amdgpu_bo_reserve(ring->mqd_obj, false);
+		if (unlikely(r != 0)) {
+			gfx_v7_0_cp_compute_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_pin(ring->mqd_obj, AMDGPU_GEM_DOMAIN_GTT,
+				  &mqd_gpu_addr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) pin MQD bo failed\n", r);
+			gfx_v7_0_cp_compute_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&buf);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map MQD bo failed\n", r);
+			gfx_v7_0_cp_compute_fini(adev);
+			return r;
+		}
+
+		/* init the mqd struct */
+		memset(buf, 0, sizeof(struct bonaire_mqd));
+
+		mqd = (struct bonaire_mqd *)buf;
+		mqd->header = 0xC0310800;
+		mqd->static_thread_mgmt01[0] = 0xffffffff;
+		mqd->static_thread_mgmt01[1] = 0xffffffff;
+		mqd->static_thread_mgmt23[0] = 0xffffffff;
+		mqd->static_thread_mgmt23[1] = 0xffffffff;
+
+		mutex_lock(&adev->srbm_mutex);
+		cik_srbm_select(adev, ring->me,
+				ring->pipe,
+				ring->queue, 0);
+
+		/* disable wptr polling */
+		tmp = RREG32(mmCP_PQ_WPTR_POLL_CNTL);
+		tmp &= ~CP_PQ_WPTR_POLL_CNTL__EN_MASK;
+		WREG32(mmCP_PQ_WPTR_POLL_CNTL, tmp);
+
+		/* enable doorbell? */
+		mqd->queue_state.cp_hqd_pq_doorbell_control =
+			RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+		if (use_doorbell)
+			mqd->queue_state.cp_hqd_pq_doorbell_control |= CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_EN_MASK;
+		else
+			mqd->queue_state.cp_hqd_pq_doorbell_control &= ~CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_EN_MASK;
+		WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL,
+		       mqd->queue_state.cp_hqd_pq_doorbell_control);
+
+		/* disable the queue if it's active */
+		mqd->queue_state.cp_hqd_dequeue_request = 0;
+		mqd->queue_state.cp_hqd_pq_rptr = 0;
+		mqd->queue_state.cp_hqd_pq_wptr= 0;
+		if (RREG32(mmCP_HQD_ACTIVE) & 1) {
+			WREG32(mmCP_HQD_DEQUEUE_REQUEST, 1);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (!(RREG32(mmCP_HQD_ACTIVE) & 1))
+					break;
+				udelay(1);
+			}
+			WREG32(mmCP_HQD_DEQUEUE_REQUEST, mqd->queue_state.cp_hqd_dequeue_request);
+			WREG32(mmCP_HQD_PQ_RPTR, mqd->queue_state.cp_hqd_pq_rptr);
+			WREG32(mmCP_HQD_PQ_WPTR, mqd->queue_state.cp_hqd_pq_wptr);
+		}
+
+		/* set the pointer to the MQD */
+		mqd->queue_state.cp_mqd_base_addr = mqd_gpu_addr & 0xfffffffc;
+		mqd->queue_state.cp_mqd_base_addr_hi = upper_32_bits(mqd_gpu_addr);
+		WREG32(mmCP_MQD_BASE_ADDR, mqd->queue_state.cp_mqd_base_addr);
+		WREG32(mmCP_MQD_BASE_ADDR_HI, mqd->queue_state.cp_mqd_base_addr_hi);
+		/* set MQD vmid to 0 */
+		mqd->queue_state.cp_mqd_control = RREG32(mmCP_MQD_CONTROL);
+		mqd->queue_state.cp_mqd_control &= ~CP_MQD_CONTROL__VMID_MASK;
+		WREG32(mmCP_MQD_CONTROL, mqd->queue_state.cp_mqd_control);
+
+		/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+		hqd_gpu_addr = ring->gpu_addr >> 8;
+		mqd->queue_state.cp_hqd_pq_base = hqd_gpu_addr;
+		mqd->queue_state.cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+		WREG32(mmCP_HQD_PQ_BASE, mqd->queue_state.cp_hqd_pq_base);
+		WREG32(mmCP_HQD_PQ_BASE_HI, mqd->queue_state.cp_hqd_pq_base_hi);
+
+		/* set up the HQD, this is similar to CP_RB0_CNTL */
+		mqd->queue_state.cp_hqd_pq_control = RREG32(mmCP_HQD_PQ_CONTROL);
+		mqd->queue_state.cp_hqd_pq_control &=
+			~(CP_HQD_PQ_CONTROL__QUEUE_SIZE_MASK |
+					CP_HQD_PQ_CONTROL__RPTR_BLOCK_SIZE_MASK);
+
+		mqd->queue_state.cp_hqd_pq_control |=
+			order_base_2(ring->ring_size / 8);
+		mqd->queue_state.cp_hqd_pq_control |=
+			(order_base_2(AMDGPU_GPU_PAGE_SIZE/8) << 8);
+#ifdef __BIG_ENDIAN
+		mqd->queue_state.cp_hqd_pq_control |=
+			2 << CP_HQD_PQ_CONTROL__ENDIAN_SWAP__SHIFT;
+#endif
+		mqd->queue_state.cp_hqd_pq_control &=
+			~(CP_HQD_PQ_CONTROL__UNORD_DISPATCH_MASK |
+				CP_HQD_PQ_CONTROL__ROQ_PQ_IB_FLIP_MASK |
+				CP_HQD_PQ_CONTROL__PQ_VOLATILE_MASK);
+		mqd->queue_state.cp_hqd_pq_control |=
+			CP_HQD_PQ_CONTROL__PRIV_STATE_MASK |
+			CP_HQD_PQ_CONTROL__KMD_QUEUE_MASK; /* assuming kernel queue control */
+		WREG32(mmCP_HQD_PQ_CONTROL, mqd->queue_state.cp_hqd_pq_control);
+
+		/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+		wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+		mqd->queue_state.cp_hqd_pq_wptr_poll_addr = wb_gpu_addr & 0xfffffffc;
+		mqd->queue_state.cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+		WREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR, mqd->queue_state.cp_hqd_pq_wptr_poll_addr);
+		WREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+		       mqd->queue_state.cp_hqd_pq_wptr_poll_addr_hi);
+
+		/* set the wb address wether it's enabled or not */
+		wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+		mqd->queue_state.cp_hqd_pq_rptr_report_addr = wb_gpu_addr & 0xfffffffc;
+		mqd->queue_state.cp_hqd_pq_rptr_report_addr_hi =
+			upper_32_bits(wb_gpu_addr) & 0xffff;
+		WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR,
+		       mqd->queue_state.cp_hqd_pq_rptr_report_addr);
+		WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		       mqd->queue_state.cp_hqd_pq_rptr_report_addr_hi);
+
+		/* enable the doorbell if requested */
+		if (use_doorbell) {
+			mqd->queue_state.cp_hqd_pq_doorbell_control =
+				RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+			mqd->queue_state.cp_hqd_pq_doorbell_control &=
+				~CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET_MASK;
+			mqd->queue_state.cp_hqd_pq_doorbell_control |=
+				(ring->doorbell_index <<
+				 CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_OFFSET__SHIFT);
+			mqd->queue_state.cp_hqd_pq_doorbell_control |=
+				CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_EN_MASK;
+			mqd->queue_state.cp_hqd_pq_doorbell_control &=
+				~(CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_SOURCE_MASK |
+				CP_HQD_PQ_DOORBELL_CONTROL__DOORBELL_HIT_MASK);
+
+		} else {
+			mqd->queue_state.cp_hqd_pq_doorbell_control = 0;
+		}
+		WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL,
+		       mqd->queue_state.cp_hqd_pq_doorbell_control);
+
+		/* read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+		ring->wptr = 0;
+		mqd->queue_state.cp_hqd_pq_wptr = ring->wptr;
+		WREG32(mmCP_HQD_PQ_WPTR, mqd->queue_state.cp_hqd_pq_wptr);
+		mqd->queue_state.cp_hqd_pq_rptr = RREG32(mmCP_HQD_PQ_RPTR);
+
+		/* set the vmid for the queue */
+		mqd->queue_state.cp_hqd_vmid = 0;
+		WREG32(mmCP_HQD_VMID, mqd->queue_state.cp_hqd_vmid);
+
+		/* activate the queue */
+		mqd->queue_state.cp_hqd_active = 1;
+		WREG32(mmCP_HQD_ACTIVE, mqd->queue_state.cp_hqd_active);
+
+		cik_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		amdgpu_bo_kunmap(ring->mqd_obj);
+		amdgpu_bo_unreserve(ring->mqd_obj);
+
+		ring->ready = true;
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			ring->ready = false;
+	}
+
+	return 0;
+}
+
+static void gfx_v7_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v7_0_cp_gfx_enable(adev, enable);
+	gfx_v7_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v7_0_cp_load_microcode(struct amdgpu_device *adev)
+{
+	int r;
+
+	r = gfx_v7_0_cp_gfx_load_microcode(adev);
+	if (r)
+		return r;
+	r = gfx_v7_0_cp_compute_load_microcode(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static void gfx_v7_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp = RREG32(mmCP_INT_CNTL_RING0);
+
+	if (enable)
+		tmp |= (CP_INT_CNTL_RING0__CNTX_BUSY_INT_ENABLE_MASK |
+				CP_INT_CNTL_RING0__CNTX_EMPTY_INT_ENABLE_MASK);
+	else
+		tmp &= ~(CP_INT_CNTL_RING0__CNTX_BUSY_INT_ENABLE_MASK |
+				CP_INT_CNTL_RING0__CNTX_EMPTY_INT_ENABLE_MASK);
+	WREG32(mmCP_INT_CNTL_RING0, tmp);
+}
+
+static int gfx_v7_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+
+	r = gfx_v7_0_cp_load_microcode(adev);
+	if (r)
+		return r;
+
+	r = gfx_v7_0_cp_gfx_resume(adev);
+	if (r)
+		return r;
+	r = gfx_v7_0_cp_compute_resume(adev);
+	if (r)
+		return r;
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+/*
+ * vm
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+/**
+ * gfx_v7_0_ring_emit_vm_flush - cik vm flush using the CP
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using the CP (CIK).
+ */
+static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vm_id, uint64_t pd_addr)
+{
+	int usepfp = (ring->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_MEM_SPACE(1) | /* memory */
+				 WAIT_REG_MEM_FUNCTION(3) | /* equal */
+				 WAIT_REG_MEM_ENGINE(usepfp)));   /* pfp or me */
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, 0xffffffff);
+	amdgpu_ring_write(ring, 4); /* poll interval */
+
+	if (usepfp) {
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
+				 WRITE_DATA_DST_SEL(0)));
+	if (vm_id < 8) {
+		amdgpu_ring_write(ring,
+				  (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
+	} else {
+		amdgpu_ring_write(ring,
+				  (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
+	}
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 1 << vm_id);
+
+	/* wait for the invalidate to complete */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(0) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* ref */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+
+	/* compute doesn't have PFP */
+	if (usepfp) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+/*
+ * RLC
+ * The RLC is a multi-purpose microengine that handles a
+ * variety of functions.
+ */
+static void gfx_v7_0_rlc_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	/* save restore block */
+	if (adev->gfx.rlc.save_restore_obj) {
+		r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, false);
+		if (unlikely(r != 0))
+			dev_warn(adev->dev, "(%d) reserve RLC sr bo failed\n", r);
+		amdgpu_bo_unpin(adev->gfx.rlc.save_restore_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
+
+		amdgpu_bo_unref(&adev->gfx.rlc.save_restore_obj);
+		adev->gfx.rlc.save_restore_obj = NULL;
+	}
+
+	/* clear state block */
+	if (adev->gfx.rlc.clear_state_obj) {
+		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
+		if (unlikely(r != 0))
+			dev_warn(adev->dev, "(%d) reserve RLC c bo failed\n", r);
+		amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+
+		amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
+		adev->gfx.rlc.clear_state_obj = NULL;
+	}
+
+	/* clear state block */
+	if (adev->gfx.rlc.cp_table_obj) {
+		r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, false);
+		if (unlikely(r != 0))
+			dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
+		amdgpu_bo_unpin(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+		amdgpu_bo_unref(&adev->gfx.rlc.cp_table_obj);
+		adev->gfx.rlc.cp_table_obj = NULL;
+	}
+}
+
+static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
+{
+	const u32 *src_ptr;
+	volatile u32 *dst_ptr;
+	u32 dws, i;
+	const struct cs_section_def *cs_data;
+	int r;
+
+	/* allocate rlc buffers */
+	if (adev->flags & AMD_IS_APU) {
+		if (adev->asic_type == CHIP_KAVERI) {
+			adev->gfx.rlc.reg_list = spectre_rlc_save_restore_register_list;
+			adev->gfx.rlc.reg_list_size =
+				(u32)ARRAY_SIZE(spectre_rlc_save_restore_register_list);
+		} else {
+			adev->gfx.rlc.reg_list = kalindi_rlc_save_restore_register_list;
+			adev->gfx.rlc.reg_list_size =
+				(u32)ARRAY_SIZE(kalindi_rlc_save_restore_register_list);
+		}
+	}
+	adev->gfx.rlc.cs_data = ci_cs_data;
+	adev->gfx.rlc.cp_table_size = CP_ME_TABLE_SIZE * 5 * 4;
+
+	src_ptr = adev->gfx.rlc.reg_list;
+	dws = adev->gfx.rlc.reg_list_size;
+	dws += (5 * 16) + 48 + 48 + 64;
+
+	cs_data = adev->gfx.rlc.cs_data;
+
+	if (src_ptr) {
+		/* save restore block */
+		if (adev->gfx.rlc.save_restore_obj == NULL) {
+			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
+					     AMDGPU_GEM_DOMAIN_VRAM,
+					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+					     NULL, NULL,
+					     &adev->gfx.rlc.save_restore_obj);
+			if (r) {
+				dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r);
+				return r;
+			}
+		}
+
+		r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, false);
+		if (unlikely(r != 0)) {
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_pin(adev->gfx.rlc.save_restore_obj, AMDGPU_GEM_DOMAIN_VRAM,
+				  &adev->gfx.rlc.save_restore_gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
+			dev_warn(adev->dev, "(%d) pin RLC sr bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+
+		r = amdgpu_bo_kmap(adev->gfx.rlc.save_restore_obj, (void **)&adev->gfx.rlc.sr_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map RLC sr bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+		/* write the sr buffer */
+		dst_ptr = adev->gfx.rlc.sr_ptr;
+		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
+			dst_ptr[i] = cpu_to_le32(src_ptr[i]);
+		amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
+	}
+
+	if (cs_data) {
+		/* clear state block */
+		adev->gfx.rlc.clear_state_size = dws = gfx_v7_0_get_csb_size(adev);
+
+		if (adev->gfx.rlc.clear_state_obj == NULL) {
+			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
+					     AMDGPU_GEM_DOMAIN_VRAM,
+					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+					     NULL, NULL,
+					     &adev->gfx.rlc.clear_state_obj);
+			if (r) {
+				dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
+				gfx_v7_0_rlc_fini(adev);
+				return r;
+			}
+		}
+		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
+		if (unlikely(r != 0)) {
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
+				  &adev->gfx.rlc.clear_state_gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+			dev_warn(adev->dev, "(%d) pin RLC c bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+
+		r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map RLC c bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+		/* set up the cs buffer */
+		dst_ptr = adev->gfx.rlc.cs_ptr;
+		gfx_v7_0_get_csb_buffer(adev, dst_ptr);
+		amdgpu_bo_kunmap(adev->gfx.rlc.clear_state_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
+	}
+
+	if (adev->gfx.rlc.cp_table_size) {
+		if (adev->gfx.rlc.cp_table_obj == NULL) {
+			r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true,
+					     AMDGPU_GEM_DOMAIN_VRAM,
+					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+					     NULL, NULL,
+					     &adev->gfx.rlc.cp_table_obj);
+			if (r) {
+				dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
+				gfx_v7_0_rlc_fini(adev);
+				return r;
+			}
+		}
+
+		r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, false);
+		if (unlikely(r != 0)) {
+			dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_pin(adev->gfx.rlc.cp_table_obj, AMDGPU_GEM_DOMAIN_VRAM,
+				  &adev->gfx.rlc.cp_table_gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+			dev_warn(adev->dev, "(%d) pin RLC cp_table bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_kmap(adev->gfx.rlc.cp_table_obj, (void **)&adev->gfx.rlc.cp_table_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map RLC cp table bo failed\n", r);
+			gfx_v7_0_rlc_fini(adev);
+			return r;
+		}
+
+		gfx_v7_0_init_cp_pg_table(adev);
+
+		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+	}
+
+	return 0;
+}
+
+static void gfx_v7_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmRLC_LB_CNTL);
+	if (enable)
+		tmp |= RLC_LB_CNTL__LOAD_BALANCE_ENABLE_MASK;
+	else
+		tmp &= ~RLC_LB_CNTL__LOAD_BALANCE_ENABLE_MASK;
+	WREG32(mmRLC_LB_CNTL, tmp);
+}
+
+static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v7_0_select_se_sh(adev, i, j);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+		}
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32(mmRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v7_0_update_rlc(struct amdgpu_device *adev, u32 rlc)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmRLC_CNTL);
+	if (tmp != rlc)
+		WREG32(mmRLC_CNTL, rlc);
+}
+
+static u32 gfx_v7_0_halt_rlc(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_CNTL);
+
+	if (data & RLC_CNTL__RLC_ENABLE_F32_MASK) {
+		u32 i;
+
+		data &= ~RLC_CNTL__RLC_ENABLE_F32_MASK;
+		WREG32(mmRLC_CNTL, data);
+
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if ((RREG32(mmRLC_GPM_STAT) & RLC_GPM_STAT__RLC_BUSY_MASK) == 0)
+				break;
+			udelay(1);
+		}
+
+		gfx_v7_0_wait_for_rlc_serdes(adev);
+	}
+
+	return orig;
+}
+
+void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	u32 tmp, i, mask;
+
+	tmp = 0x1 | (1 << 1);
+	WREG32(mmRLC_GPR_REG2, tmp);
+
+	mask = RLC_GPM_STAT__GFX_POWER_STATUS_MASK |
+		RLC_GPM_STAT__GFX_CLOCK_STATUS_MASK;
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmRLC_GPM_STAT) & mask) == mask)
+			break;
+		udelay(1);
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmRLC_GPR_REG2) & 0x1) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = 0x1 | (0 << 1);
+	WREG32(mmRLC_GPR_REG2, tmp);
+}
+
+/**
+ * gfx_v7_0_rlc_stop - stop the RLC ME
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Halt the RLC ME (MicroEngine) (CIK).
+ */
+void gfx_v7_0_rlc_stop(struct amdgpu_device *adev)
+{
+	WREG32(mmRLC_CNTL, 0);
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+
+	gfx_v7_0_wait_for_rlc_serdes(adev);
+}
+
+/**
+ * gfx_v7_0_rlc_start - start the RLC ME
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Unhalt the RLC ME (MicroEngine) (CIK).
+ */
+static void gfx_v7_0_rlc_start(struct amdgpu_device *adev)
+{
+	WREG32(mmRLC_CNTL, RLC_CNTL__RLC_ENABLE_F32_MASK);
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+	udelay(50);
+}
+
+static void gfx_v7_0_rlc_reset(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmGRBM_SOFT_RESET);
+
+	tmp |= GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+	WREG32(mmGRBM_SOFT_RESET, tmp);
+	udelay(50);
+	tmp &= ~GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+	WREG32(mmGRBM_SOFT_RESET, tmp);
+	udelay(50);
+}
+
+/**
+ * gfx_v7_0_rlc_resume - setup the RLC hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the RLC registers, load the ucode,
+ * and start the RLC (CIK).
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+	u32 tmp;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v1_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+	adev->gfx.rlc_fw_version = le32_to_cpu(hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(
+					hdr->ucode_feature_version);
+
+	gfx_v7_0_rlc_stop(adev);
+
+	/* disable CG */
+	tmp = RREG32(mmRLC_CGCG_CGLS_CTRL) & 0xfffffffc;
+	WREG32(mmRLC_CGCG_CGLS_CTRL, tmp);
+
+	gfx_v7_0_rlc_reset(adev);
+
+	gfx_v7_0_init_pg(adev);
+
+	WREG32(mmRLC_LB_CNTR_INIT, 0);
+	WREG32(mmRLC_LB_CNTR_MAX, 0x00008000);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	WREG32(mmRLC_LB_INIT_CU_MASK, 0xffffffff);
+	WREG32(mmRLC_LB_PARAMS, 0x00600408);
+	WREG32(mmRLC_LB_CNTL, 0x80000004);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	WREG32(mmRLC_MC_CNTL, 0);
+	WREG32(mmRLC_UCODE_CNTL, 0);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.rlc_fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmRLC_GPM_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	/* XXX - find out what chips support lbpw */
+	gfx_v7_0_enable_lbpw(adev, false);
+
+	if (adev->asic_type == CHIP_BONAIRE)
+		WREG32(mmRLC_DRIVER_CPDMA_STATUS, 0);
+
+	gfx_v7_0_rlc_start(adev);
+
+	return 0;
+}
+
+static void gfx_v7_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig, tmp, tmp2;
+
+	orig = data = RREG32(mmRLC_CGCG_CGLS_CTRL);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_CGCG)) {
+		gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+		tmp = gfx_v7_0_halt_rlc(adev);
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+		tmp2 = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
+			RLC_SERDES_WR_CTRL__CGCG_OVERRIDE_0_MASK |
+			RLC_SERDES_WR_CTRL__CGLS_ENABLE_MASK;
+		WREG32(mmRLC_SERDES_WR_CTRL, tmp2);
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		gfx_v7_0_update_rlc(adev, tmp);
+
+		data |= RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK;
+	} else {
+		gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+		RREG32(mmCB_CGTT_SCLK_CTRL);
+
+		data &= ~(RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK | RLC_CGCG_CGLS_CTRL__CGLS_EN_MASK);
+	}
+
+	if (orig != data)
+		WREG32(mmRLC_CGCG_CGLS_CTRL, data);
+
+}
+
+static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig, tmp = 0;
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_MGCG)) {
+		if (adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_MGLS) {
+			if (adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_CP_LS) {
+				orig = data = RREG32(mmCP_MEM_SLP_CNTL);
+				data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+				if (orig != data)
+					WREG32(mmCP_MEM_SLP_CNTL, data);
+			}
+		}
+
+		orig = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= 0x00000001;
+		data &= 0xfffffffd;
+		if (orig != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		tmp = gfx_v7_0_halt_rlc(adev);
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
+			RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_0_MASK;
+		WREG32(mmRLC_SERDES_WR_CTRL, data);
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		gfx_v7_0_update_rlc(adev, tmp);
+
+		if (adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_CGTS) {
+			orig = data = RREG32(mmCGTS_SM_CTRL_REG);
+			data &= ~CGTS_SM_CTRL_REG__SM_MODE_MASK;
+			data |= (0x2 << CGTS_SM_CTRL_REG__SM_MODE__SHIFT);
+			data |= CGTS_SM_CTRL_REG__SM_MODE_ENABLE_MASK;
+			data &= ~CGTS_SM_CTRL_REG__OVERRIDE_MASK;
+			if ((adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_MGLS) &&
+			    (adev->cg_flags & AMDGPU_CG_SUPPORT_GFX_CGTS_LS))
+				data &= ~CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK;
+			data &= ~CGTS_SM_CTRL_REG__ON_MONITOR_ADD_MASK;
+			data |= CGTS_SM_CTRL_REG__ON_MONITOR_ADD_EN_MASK;
+			data |= (0x96 << CGTS_SM_CTRL_REG__ON_MONITOR_ADD__SHIFT);
+			if (orig != data)
+				WREG32(mmCGTS_SM_CTRL_REG, data);
+		}
+	} else {
+		orig = data = RREG32(mmRLC_CGTT_MGCG_OVERRIDE);
+		data |= 0x00000003;
+		if (orig != data)
+			WREG32(mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		data = RREG32(mmRLC_MEM_SLP_CNTL);
+		if (data & RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK) {
+			data &= ~RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK;
+			WREG32(mmRLC_MEM_SLP_CNTL, data);
+		}
+
+		data = RREG32(mmCP_MEM_SLP_CNTL);
+		if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) {
+			data &= ~CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK;
+			WREG32(mmCP_MEM_SLP_CNTL, data);
+		}
+
+		orig = data = RREG32(mmCGTS_SM_CTRL_REG);
+		data |= CGTS_SM_CTRL_REG__OVERRIDE_MASK | CGTS_SM_CTRL_REG__LS_OVERRIDE_MASK;
+		if (orig != data)
+			WREG32(mmCGTS_SM_CTRL_REG, data);
+
+		tmp = gfx_v7_0_halt_rlc(adev);
+
+		mutex_lock(&adev->grbm_idx_mutex);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
+		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
+		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK | RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_1_MASK;
+		WREG32(mmRLC_SERDES_WR_CTRL, data);
+		mutex_unlock(&adev->grbm_idx_mutex);
+
+		gfx_v7_0_update_rlc(adev, tmp);
+	}
+}
+
+static void gfx_v7_0_update_cg(struct amdgpu_device *adev,
+			       bool enable)
+{
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+	/* order matters! */
+	if (enable) {
+		gfx_v7_0_enable_mgcg(adev, true);
+		gfx_v7_0_enable_cgcg(adev, true);
+	} else {
+		gfx_v7_0_enable_cgcg(adev, false);
+		gfx_v7_0_enable_mgcg(adev, false);
+	}
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+}
+
+static void gfx_v7_0_enable_sclk_slowdown_on_pu(struct amdgpu_device *adev,
+						bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_RLC_SMU_HS))
+		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_sclk_slowdown_on_pd(struct amdgpu_device *adev,
+						bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_RLC_SMU_HS))
+		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_cp_pg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_CP))
+		data &= ~0x8000;
+	else
+		data |= 0x8000;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_gds_pg(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_GDS))
+		data &= ~0x2000;
+	else
+		data |= 0x2000;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_init_cp_pg_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	volatile u32 *dst_ptr;
+	int me, i, max_me = 4;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	if (adev->asic_type == CHIP_KAVERI)
+		max_me = 5;
+
+	if (adev->gfx.rlc.cp_table_ptr == NULL)
+		return;
+
+	/* write the cp table buffer */
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i ++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+
+static void gfx_v7_0_enable_gfx_cgpg(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 data, orig;
+
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_GFX_PG)) {
+		orig = data = RREG32(mmRLC_PG_CNTL);
+		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+		if (orig != data)
+			WREG32(mmRLC_PG_CNTL, data);
+
+		orig = data = RREG32(mmRLC_AUTO_PG_CTRL);
+		data |= RLC_AUTO_PG_CTRL__AUTO_PG_EN_MASK;
+		if (orig != data)
+			WREG32(mmRLC_AUTO_PG_CTRL, data);
+	} else {
+		orig = data = RREG32(mmRLC_PG_CNTL);
+		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+		if (orig != data)
+			WREG32(mmRLC_PG_CNTL, data);
+
+		orig = data = RREG32(mmRLC_AUTO_PG_CTRL);
+		data &= ~RLC_AUTO_PG_CTRL__AUTO_PG_EN_MASK;
+		if (orig != data)
+			WREG32(mmRLC_AUTO_PG_CTRL, data);
+
+		data = RREG32(mmDB_RENDER_CONTROL);
+	}
+}
+
+static u32 gfx_v7_0_get_cu_active_bitmap(struct amdgpu_device *adev,
+					 u32 se, u32 sh)
+{
+	u32 mask = 0, tmp, tmp1;
+	int i;
+
+	gfx_v7_0_select_se_sh(adev, se, sh);
+	tmp = RREG32(mmCC_GC_SHADER_ARRAY_CONFIG);
+	tmp1 = RREG32(mmGC_USER_SHADER_ARRAY_CONFIG);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+
+	tmp &= 0xffff0000;
+
+	tmp |= tmp1;
+	tmp >>= 16;
+
+	for (i = 0; i < adev->gfx.config.max_cu_per_sh; i ++) {
+		mask <<= 1;
+		mask |= 1;
+	}
+
+	return (~tmp) & mask;
+}
+
+static void gfx_v7_0_init_ao_cu_mask(struct amdgpu_device *adev)
+{
+	uint32_t tmp, active_cu_number;
+	struct amdgpu_cu_info cu_info;
+
+	gfx_v7_0_get_cu_info(adev, &cu_info);
+	tmp = cu_info.ao_cu_mask;
+	active_cu_number = cu_info.number;
+
+	WREG32(mmRLC_PG_ALWAYS_ON_CU_MASK, tmp);
+
+	tmp = RREG32(mmRLC_MAX_PG_CU);
+	tmp &= ~RLC_MAX_PG_CU__MAX_POWERED_UP_CU_MASK;
+	tmp |= (active_cu_number << RLC_MAX_PG_CU__MAX_POWERED_UP_CU__SHIFT);
+	WREG32(mmRLC_MAX_PG_CU, tmp);
+}
+
+static void gfx_v7_0_enable_gfx_static_mgpg(struct amdgpu_device *adev,
+					    bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_GFX_SMG))
+		data |= RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v7_0_enable_gfx_dynamic_mgpg(struct amdgpu_device *adev,
+					     bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	if (enable && (adev->pg_flags & AMDGPU_PG_SUPPORT_GFX_DMG))
+		data |= RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+#define RLC_SAVE_AND_RESTORE_STARTING_OFFSET 0x90
+#define RLC_CLEAR_STATE_DESCRIPTOR_OFFSET    0x3D
+
+static void gfx_v7_0_init_gfx_cgpg(struct amdgpu_device *adev)
+{
+	u32 data, orig;
+	u32 i;
+
+	if (adev->gfx.rlc.cs_data) {
+		WREG32(mmRLC_GPM_SCRATCH_ADDR, RLC_CLEAR_STATE_DESCRIPTOR_OFFSET);
+		WREG32(mmRLC_GPM_SCRATCH_DATA, upper_32_bits(adev->gfx.rlc.clear_state_gpu_addr));
+		WREG32(mmRLC_GPM_SCRATCH_DATA, lower_32_bits(adev->gfx.rlc.clear_state_gpu_addr));
+		WREG32(mmRLC_GPM_SCRATCH_DATA, adev->gfx.rlc.clear_state_size);
+	} else {
+		WREG32(mmRLC_GPM_SCRATCH_ADDR, RLC_CLEAR_STATE_DESCRIPTOR_OFFSET);
+		for (i = 0; i < 3; i++)
+			WREG32(mmRLC_GPM_SCRATCH_DATA, 0);
+	}
+	if (adev->gfx.rlc.reg_list) {
+		WREG32(mmRLC_GPM_SCRATCH_ADDR, RLC_SAVE_AND_RESTORE_STARTING_OFFSET);
+		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
+			WREG32(mmRLC_GPM_SCRATCH_DATA, adev->gfx.rlc.reg_list[i]);
+	}
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+	data |= RLC_PG_CNTL__GFX_POWER_GATING_SRC_MASK;
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+
+	WREG32(mmRLC_SAVE_AND_RESTORE_BASE, adev->gfx.rlc.save_restore_gpu_addr >> 8);
+	WREG32(mmRLC_JUMP_TABLE_RESTORE, adev->gfx.rlc.cp_table_gpu_addr >> 8);
+
+	data = RREG32(mmCP_RB_WPTR_POLL_CNTL);
+	data &= ~CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK;
+	data |= (0x60 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
+	WREG32(mmCP_RB_WPTR_POLL_CNTL, data);
+
+	data = 0x10101010;
+	WREG32(mmRLC_PG_DELAY, data);
+
+	data = RREG32(mmRLC_PG_DELAY_2);
+	data &= ~0xff;
+	data |= 0x3;
+	WREG32(mmRLC_PG_DELAY_2, data);
+
+	data = RREG32(mmRLC_AUTO_PG_CTRL);
+	data &= ~RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD_MASK;
+	data |= (0x700 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT);
+	WREG32(mmRLC_AUTO_PG_CTRL, data);
+
+}
+
+static void gfx_v7_0_update_gfx_pg(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v7_0_enable_gfx_cgpg(adev, enable);
+	gfx_v7_0_enable_gfx_static_mgpg(adev, enable);
+	gfx_v7_0_enable_gfx_dynamic_mgpg(adev, enable);
+}
+
+static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return 0;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+	/* pa_sc_raster_config/pa_sc_raster_config1 */
+	count += 4;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev,
+				    volatile u32 *buffer)
+{
+	u32 count = 0, i;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	if (adev->gfx.rlc.cs_data == NULL)
+		return;
+	if (buffer == NULL)
+		return;
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	buffer[count++] = cpu_to_le32(0x80000000);
+	buffer[count++] = cpu_to_le32(0x80000000);
+
+	for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				buffer[count++] =
+					cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count));
+				buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					buffer[count++] = cpu_to_le32(ext->extent[i]);
+			} else {
+				return;
+			}
+		}
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		buffer[count++] = cpu_to_le32(0x16000012);
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	case CHIP_KAVERI:
+		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	case CHIP_KABINI:
+	case CHIP_MULLINS:
+		buffer[count++] = cpu_to_le32(0x00000000); /* XXX */
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	case CHIP_HAWAII:
+		buffer[count++] = cpu_to_le32(0x3a00161a);
+		buffer[count++] = cpu_to_le32(0x0000002e);
+		break;
+	default:
+		buffer[count++] = cpu_to_le32(0x00000000);
+		buffer[count++] = cpu_to_le32(0x00000000);
+		break;
+	}
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0));
+	buffer[count++] = cpu_to_le32(0);
+}
+
+static void gfx_v7_0_init_pg(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & (AMDGPU_PG_SUPPORT_GFX_PG |
+			      AMDGPU_PG_SUPPORT_GFX_SMG |
+			      AMDGPU_PG_SUPPORT_GFX_DMG |
+			      AMDGPU_PG_SUPPORT_CP |
+			      AMDGPU_PG_SUPPORT_GDS |
+			      AMDGPU_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v7_0_enable_sclk_slowdown_on_pu(adev, true);
+		gfx_v7_0_enable_sclk_slowdown_on_pd(adev, true);
+		if (adev->pg_flags & AMDGPU_PG_SUPPORT_GFX_PG) {
+			gfx_v7_0_init_gfx_cgpg(adev);
+			gfx_v7_0_enable_cp_pg(adev, true);
+			gfx_v7_0_enable_gds_pg(adev, true);
+		}
+		gfx_v7_0_init_ao_cu_mask(adev);
+		gfx_v7_0_update_gfx_pg(adev, true);
+	}
+}
+
+static void gfx_v7_0_fini_pg(struct amdgpu_device *adev)
+{
+	if (adev->pg_flags & (AMDGPU_PG_SUPPORT_GFX_PG |
+			      AMDGPU_PG_SUPPORT_GFX_SMG |
+			      AMDGPU_PG_SUPPORT_GFX_DMG |
+			      AMDGPU_PG_SUPPORT_CP |
+			      AMDGPU_PG_SUPPORT_GDS |
+			      AMDGPU_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v7_0_update_gfx_pg(adev, false);
+		if (adev->pg_flags & AMDGPU_PG_SUPPORT_GFX_PG) {
+			gfx_v7_0_enable_cp_pg(adev, false);
+			gfx_v7_0_enable_gds_pg(adev, false);
+		}
+	}
+}
+
+/**
+ * gfx_v7_0_get_gpu_clock_counter - return GPU clock counter snapshot
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Fetches a GPU clock counter snapshot (SI).
+ * Returns the 64 bit clock counter snapshot.
+ */
+uint64_t gfx_v7_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32(mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	return clock;
+}
+
+static void gfx_v7_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					  uint32_t vmid,
+					  uint32_t gds_base, uint32_t gds_size,
+					  uint32_t gws_base, uint32_t gws_size,
+					  uint32_t oa_base, uint32_t oa_size)
+{
+	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
+	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
+
+	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
+	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
+
+	oa_base = oa_base >> AMDGPU_OA_SHIFT;
+	oa_size = oa_size >> AMDGPU_OA_SHIFT;
+
+	/* GDS Base */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_base);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_base);
+
+	/* GDS Size */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_size);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_size);
+
+	/* GWS */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].gws);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].oa);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static int gfx_v7_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.num_gfx_rings = GFX7_NUM_GFX_RINGS;
+	adev->gfx.num_compute_rings = GFX7_NUM_COMPUTE_RINGS;
+	gfx_v7_0_set_ring_funcs(adev);
+	gfx_v7_0_set_irq_funcs(adev);
+	gfx_v7_0_set_gds_init(adev);
+
+	return 0;
+}
+
+static int gfx_v7_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i, r;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, 181, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, 184, &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, 185, &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	gfx_v7_0_scratch_init(adev);
+
+	r = gfx_v7_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load gfx firmware!\n");
+		return r;
+	}
+
+	r = gfx_v7_0_rlc_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init rlc BOs!\n");
+		return r;
+	}
+
+	/* allocate mec buffers */
+	r = gfx_v7_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		ring->ring_obj = NULL;
+		sprintf(ring->name, "gfx");
+		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
+				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP,
+				     AMDGPU_RING_TYPE_GFX);
+		if (r)
+			return r;
+	}
+
+	/* set up the compute queues */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		unsigned irq_type;
+
+		/* max 32 queues per MEC */
+		if ((i >= 32) || (i >= AMDGPU_MAX_COMPUTE_RINGS)) {
+			DRM_ERROR("Too many (%d) compute rings!\n", i);
+			break;
+		}
+		ring = &adev->gfx.compute_ring[i];
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + i;
+		ring->me = 1; /* first MEC */
+		ring->pipe = i / 8;
+		ring->queue = i % 8;
+		sprintf(ring->name, "comp %d.%d.%d", ring->me, ring->pipe, ring->queue);
+		irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe;
+		/* type-2 packets are deprecated on MEC, use type-3 instead */
+		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
+				     &adev->gfx.eop_irq, irq_type,
+				     AMDGPU_RING_TYPE_COMPUTE);
+		if (r)
+			return r;
+	}
+
+	/* reserve GDS, GWS and OA resource for gfx */
+	r = amdgpu_bo_create(adev, adev->gds.mem.gfx_partition_size,
+			PAGE_SIZE, true,
+			AMDGPU_GEM_DOMAIN_GDS, 0,
+			NULL, NULL, &adev->gds.gds_gfx_bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create(adev, adev->gds.gws.gfx_partition_size,
+		PAGE_SIZE, true,
+		AMDGPU_GEM_DOMAIN_GWS, 0,
+		NULL, NULL, &adev->gds.gws_gfx_bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create(adev, adev->gds.oa.gfx_partition_size,
+			PAGE_SIZE, true,
+			AMDGPU_GEM_DOMAIN_OA, 0,
+			NULL, NULL, &adev->gds.oa_gfx_bo);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gfx_v7_0_sw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_unref(&adev->gds.oa_gfx_bo);
+	amdgpu_bo_unref(&adev->gds.gws_gfx_bo);
+	amdgpu_bo_unref(&adev->gds.gds_gfx_bo);
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	gfx_v7_0_cp_compute_fini(adev);
+	gfx_v7_0_rlc_fini(adev);
+	gfx_v7_0_mec_fini(adev);
+
+	return 0;
+}
+
+static int gfx_v7_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v7_0_gpu_init(adev);
+
+	/* init rlc */
+	r = gfx_v7_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v7_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	adev->gfx.ce_ram_size = 0x8000;
+
+	return r;
+}
+
+static int gfx_v7_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v7_0_cp_enable(adev, false);
+	gfx_v7_0_rlc_stop(adev);
+	gfx_v7_0_fini_pg(adev);
+
+	return 0;
+}
+
+static int gfx_v7_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v7_0_hw_fini(adev);
+}
+
+static int gfx_v7_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v7_0_hw_init(adev);
+}
+
+static bool gfx_v7_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK)
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v7_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK;
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void gfx_v7_0_print_status(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "GFX 7.x registers\n");
+	dev_info(adev->dev, "  GRBM_STATUS=0x%08X\n",
+		RREG32(mmGRBM_STATUS));
+	dev_info(adev->dev, "  GRBM_STATUS2=0x%08X\n",
+		RREG32(mmGRBM_STATUS2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE0=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE0));
+	dev_info(adev->dev, "  GRBM_STATUS_SE1=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE1));
+	dev_info(adev->dev, "  GRBM_STATUS_SE2=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE3=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE3));
+	dev_info(adev->dev, "  CP_STAT = 0x%08x\n", RREG32(mmCP_STAT));
+	dev_info(adev->dev, "  CP_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_STALLED_STAT2 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT2));
+	dev_info(adev->dev, "  CP_STALLED_STAT3 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT3));
+	dev_info(adev->dev, "  CP_CPF_BUSY_STAT = 0x%08x\n",
+		 RREG32(mmCP_CPF_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPF_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPF_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPF_STATUS = 0x%08x\n", RREG32(mmCP_CPF_STATUS));
+	dev_info(adev->dev, "  CP_CPC_BUSY_STAT = 0x%08x\n", RREG32(mmCP_CPC_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPC_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPC_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPC_STATUS = 0x%08x\n", RREG32(mmCP_CPC_STATUS));
+
+	for (i = 0; i < 32; i++) {
+		dev_info(adev->dev, "  GB_TILE_MODE%d=0x%08X\n",
+			 i, RREG32(mmGB_TILE_MODE0 + (i * 4)));
+	}
+	for (i = 0; i < 16; i++) {
+		dev_info(adev->dev, "  GB_MACROTILE_MODE%d=0x%08X\n",
+			 i, RREG32(mmGB_MACROTILE_MODE0 + (i * 4)));
+	}
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		dev_info(adev->dev, "  se: %d\n", i);
+		gfx_v7_0_select_se_sh(adev, i, 0xffffffff);
+		dev_info(adev->dev, "  PA_SC_RASTER_CONFIG=0x%08X\n",
+			 RREG32(mmPA_SC_RASTER_CONFIG));
+		dev_info(adev->dev, "  PA_SC_RASTER_CONFIG_1=0x%08X\n",
+			 RREG32(mmPA_SC_RASTER_CONFIG_1));
+	}
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+
+	dev_info(adev->dev, "  GB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmGB_ADDR_CONFIG));
+	dev_info(adev->dev, "  HDP_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmHDP_ADDR_CONFIG));
+	dev_info(adev->dev, "  DMIF_ADDR_CALC=0x%08X\n",
+		 RREG32(mmDMIF_ADDR_CALC));
+	dev_info(adev->dev, "  SDMA0_TILING_CONFIG=0x%08X\n",
+		 RREG32(mmSDMA0_TILING_CONFIG + SDMA0_REGISTER_OFFSET));
+	dev_info(adev->dev, "  SDMA1_TILING_CONFIG=0x%08X\n",
+		 RREG32(mmSDMA0_TILING_CONFIG + SDMA1_REGISTER_OFFSET));
+	dev_info(adev->dev, "  UVD_UDEC_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DB_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DBW_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DBW_ADDR_CONFIG));
+
+	dev_info(adev->dev, "  CP_MEQ_THRESHOLDS=0x%08X\n",
+		 RREG32(mmCP_MEQ_THRESHOLDS));
+	dev_info(adev->dev, "  SX_DEBUG_1=0x%08X\n",
+		 RREG32(mmSX_DEBUG_1));
+	dev_info(adev->dev, "  TA_CNTL_AUX=0x%08X\n",
+		 RREG32(mmTA_CNTL_AUX));
+	dev_info(adev->dev, "  SPI_CONFIG_CNTL=0x%08X\n",
+		 RREG32(mmSPI_CONFIG_CNTL));
+	dev_info(adev->dev, "  SQ_CONFIG=0x%08X\n",
+		 RREG32(mmSQ_CONFIG));
+	dev_info(adev->dev, "  DB_DEBUG=0x%08X\n",
+		 RREG32(mmDB_DEBUG));
+	dev_info(adev->dev, "  DB_DEBUG2=0x%08X\n",
+		 RREG32(mmDB_DEBUG2));
+	dev_info(adev->dev, "  DB_DEBUG3=0x%08X\n",
+		 RREG32(mmDB_DEBUG3));
+	dev_info(adev->dev, "  CB_HW_CONTROL=0x%08X\n",
+		 RREG32(mmCB_HW_CONTROL));
+	dev_info(adev->dev, "  SPI_CONFIG_CNTL_1=0x%08X\n",
+		 RREG32(mmSPI_CONFIG_CNTL_1));
+	dev_info(adev->dev, "  PA_SC_FIFO_SIZE=0x%08X\n",
+		 RREG32(mmPA_SC_FIFO_SIZE));
+	dev_info(adev->dev, "  VGT_NUM_INSTANCES=0x%08X\n",
+		 RREG32(mmVGT_NUM_INSTANCES));
+	dev_info(adev->dev, "  CP_PERFMON_CNTL=0x%08X\n",
+		 RREG32(mmCP_PERFMON_CNTL));
+	dev_info(adev->dev, "  PA_SC_FORCE_EOV_MAX_CNTS=0x%08X\n",
+		 RREG32(mmPA_SC_FORCE_EOV_MAX_CNTS));
+	dev_info(adev->dev, "  VGT_CACHE_INVALIDATION=0x%08X\n",
+		 RREG32(mmVGT_CACHE_INVALIDATION));
+	dev_info(adev->dev, "  VGT_GS_VERTEX_REUSE=0x%08X\n",
+		 RREG32(mmVGT_GS_VERTEX_REUSE));
+	dev_info(adev->dev, "  PA_SC_LINE_STIPPLE_STATE=0x%08X\n",
+		 RREG32(mmPA_SC_LINE_STIPPLE_STATE));
+	dev_info(adev->dev, "  PA_CL_ENHANCE=0x%08X\n",
+		 RREG32(mmPA_CL_ENHANCE));
+	dev_info(adev->dev, "  PA_SC_ENHANCE=0x%08X\n",
+		 RREG32(mmPA_SC_ENHANCE));
+
+	dev_info(adev->dev, "  CP_ME_CNTL=0x%08X\n",
+		 RREG32(mmCP_ME_CNTL));
+	dev_info(adev->dev, "  CP_MAX_CONTEXT=0x%08X\n",
+		 RREG32(mmCP_MAX_CONTEXT));
+	dev_info(adev->dev, "  CP_ENDIAN_SWAP=0x%08X\n",
+		 RREG32(mmCP_ENDIAN_SWAP));
+	dev_info(adev->dev, "  CP_DEVICE_ID=0x%08X\n",
+		 RREG32(mmCP_DEVICE_ID));
+
+	dev_info(adev->dev, "  CP_SEM_WAIT_TIMER=0x%08X\n",
+		 RREG32(mmCP_SEM_WAIT_TIMER));
+	if (adev->asic_type != CHIP_HAWAII)
+		dev_info(adev->dev, "  CP_SEM_INCOMPLETE_TIMER_CNTL=0x%08X\n",
+			 RREG32(mmCP_SEM_INCOMPLETE_TIMER_CNTL));
+
+	dev_info(adev->dev, "  CP_RB_WPTR_DELAY=0x%08X\n",
+		 RREG32(mmCP_RB_WPTR_DELAY));
+	dev_info(adev->dev, "  CP_RB_VMID=0x%08X\n",
+		 RREG32(mmCP_RB_VMID));
+	dev_info(adev->dev, "  CP_RB0_CNTL=0x%08X\n",
+		 RREG32(mmCP_RB0_CNTL));
+	dev_info(adev->dev, "  CP_RB0_WPTR=0x%08X\n",
+		 RREG32(mmCP_RB0_WPTR));
+	dev_info(adev->dev, "  CP_RB0_RPTR_ADDR=0x%08X\n",
+		 RREG32(mmCP_RB0_RPTR_ADDR));
+	dev_info(adev->dev, "  CP_RB0_RPTR_ADDR_HI=0x%08X\n",
+		 RREG32(mmCP_RB0_RPTR_ADDR_HI));
+	dev_info(adev->dev, "  CP_RB0_CNTL=0x%08X\n",
+		 RREG32(mmCP_RB0_CNTL));
+	dev_info(adev->dev, "  CP_RB0_BASE=0x%08X\n",
+		 RREG32(mmCP_RB0_BASE));
+	dev_info(adev->dev, "  CP_RB0_BASE_HI=0x%08X\n",
+		 RREG32(mmCP_RB0_BASE_HI));
+	dev_info(adev->dev, "  CP_MEC_CNTL=0x%08X\n",
+		 RREG32(mmCP_MEC_CNTL));
+	dev_info(adev->dev, "  CP_CPF_DEBUG=0x%08X\n",
+		 RREG32(mmCP_CPF_DEBUG));
+
+	dev_info(adev->dev, "  SCRATCH_ADDR=0x%08X\n",
+		 RREG32(mmSCRATCH_ADDR));
+	dev_info(adev->dev, "  SCRATCH_UMSK=0x%08X\n",
+		 RREG32(mmSCRATCH_UMSK));
+
+	/* init the pipes */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < (adev->gfx.mec.num_pipe * adev->gfx.mec.num_mec); i++) {
+		int me = (i < 4) ? 1 : 2;
+		int pipe = (i < 4) ? i : (i - 4);
+		int queue;
+
+		dev_info(adev->dev, "  me: %d, pipe: %d\n", me, pipe);
+		cik_srbm_select(adev, me, pipe, 0, 0);
+		dev_info(adev->dev, "  CP_HPD_EOP_BASE_ADDR=0x%08X\n",
+			 RREG32(mmCP_HPD_EOP_BASE_ADDR));
+		dev_info(adev->dev, "  CP_HPD_EOP_BASE_ADDR_HI=0x%08X\n",
+			 RREG32(mmCP_HPD_EOP_BASE_ADDR_HI));
+		dev_info(adev->dev, "  CP_HPD_EOP_VMID=0x%08X\n",
+			 RREG32(mmCP_HPD_EOP_VMID));
+		dev_info(adev->dev, "  CP_HPD_EOP_CONTROL=0x%08X\n",
+			 RREG32(mmCP_HPD_EOP_CONTROL));
+
+		for (queue = 0; queue < 8; queue++) {
+			cik_srbm_select(adev, me, pipe, queue, 0);
+			dev_info(adev->dev, "  queue: %d\n", queue);
+			dev_info(adev->dev, "  CP_PQ_WPTR_POLL_CNTL=0x%08X\n",
+				 RREG32(mmCP_PQ_WPTR_POLL_CNTL));
+			dev_info(adev->dev, "  CP_HQD_PQ_DOORBELL_CONTROL=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL));
+			dev_info(adev->dev, "  CP_HQD_ACTIVE=0x%08X\n",
+				 RREG32(mmCP_HQD_ACTIVE));
+			dev_info(adev->dev, "  CP_HQD_DEQUEUE_REQUEST=0x%08X\n",
+				 RREG32(mmCP_HQD_DEQUEUE_REQUEST));
+			dev_info(adev->dev, "  CP_HQD_PQ_RPTR=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_RPTR));
+			dev_info(adev->dev, "  CP_HQD_PQ_WPTR=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_WPTR));
+			dev_info(adev->dev, "  CP_HQD_PQ_BASE=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_BASE));
+			dev_info(adev->dev, "  CP_HQD_PQ_BASE_HI=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_BASE_HI));
+			dev_info(adev->dev, "  CP_HQD_PQ_CONTROL=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_CONTROL));
+			dev_info(adev->dev, "  CP_HQD_PQ_WPTR_POLL_ADDR=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR));
+			dev_info(adev->dev, "  CP_HQD_PQ_WPTR_POLL_ADDR_HI=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR_HI));
+			dev_info(adev->dev, "  CP_HQD_PQ_RPTR_REPORT_ADDR=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR));
+			dev_info(adev->dev, "  CP_HQD_PQ_RPTR_REPORT_ADDR_HI=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI));
+			dev_info(adev->dev, "  CP_HQD_PQ_DOORBELL_CONTROL=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL));
+			dev_info(adev->dev, "  CP_HQD_PQ_WPTR=0x%08X\n",
+				 RREG32(mmCP_HQD_PQ_WPTR));
+			dev_info(adev->dev, "  CP_HQD_VMID=0x%08X\n",
+				 RREG32(mmCP_HQD_VMID));
+			dev_info(adev->dev, "  CP_MQD_BASE_ADDR=0x%08X\n",
+				 RREG32(mmCP_MQD_BASE_ADDR));
+			dev_info(adev->dev, "  CP_MQD_BASE_ADDR_HI=0x%08X\n",
+				 RREG32(mmCP_MQD_BASE_ADDR_HI));
+			dev_info(adev->dev, "  CP_MQD_CONTROL=0x%08X\n",
+				 RREG32(mmCP_MQD_CONTROL));
+		}
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	dev_info(adev->dev, "  CP_INT_CNTL_RING0=0x%08X\n",
+		 RREG32(mmCP_INT_CNTL_RING0));
+	dev_info(adev->dev, "  RLC_LB_CNTL=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTL));
+	dev_info(adev->dev, "  RLC_CNTL=0x%08X\n",
+		 RREG32(mmRLC_CNTL));
+	dev_info(adev->dev, "  RLC_CGCG_CGLS_CTRL=0x%08X\n",
+		 RREG32(mmRLC_CGCG_CGLS_CTRL));
+	dev_info(adev->dev, "  RLC_LB_CNTR_INIT=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTR_INIT));
+	dev_info(adev->dev, "  RLC_LB_CNTR_MAX=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTR_MAX));
+	dev_info(adev->dev, "  RLC_LB_INIT_CU_MASK=0x%08X\n",
+		 RREG32(mmRLC_LB_INIT_CU_MASK));
+	dev_info(adev->dev, "  RLC_LB_PARAMS=0x%08X\n",
+		 RREG32(mmRLC_LB_PARAMS));
+	dev_info(adev->dev, "  RLC_LB_CNTL=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTL));
+	dev_info(adev->dev, "  RLC_MC_CNTL=0x%08X\n",
+		 RREG32(mmRLC_MC_CNTL));
+	dev_info(adev->dev, "  RLC_UCODE_CNTL=0x%08X\n",
+		 RREG32(mmRLC_UCODE_CNTL));
+
+	if (adev->asic_type == CHIP_BONAIRE)
+		dev_info(adev->dev, "  RLC_DRIVER_CPDMA_STATUS=0x%08X\n",
+			 RREG32(mmRLC_DRIVER_CPDMA_STATUS));
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < 16; i++) {
+		cik_srbm_select(adev, 0, 0, 0, i);
+		dev_info(adev->dev, "  VM %d:\n", i);
+		dev_info(adev->dev, "  SH_MEM_CONFIG=0x%08X\n",
+			 RREG32(mmSH_MEM_CONFIG));
+		dev_info(adev->dev, "  SH_MEM_APE1_BASE=0x%08X\n",
+			 RREG32(mmSH_MEM_APE1_BASE));
+		dev_info(adev->dev, "  SH_MEM_APE1_LIMIT=0x%08X\n",
+			 RREG32(mmSH_MEM_APE1_LIMIT));
+		dev_info(adev->dev, "  SH_MEM_BASES=0x%08X\n",
+			 RREG32(mmSH_MEM_BASES));
+	}
+	cik_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static int gfx_v7_0_soft_reset(void *handle)
+{
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* GRBM_STATUS */
+	tmp = RREG32(mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK))
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_CP_MASK |
+			GRBM_SOFT_RESET__SOFT_RESET_GFX_MASK;
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_CP_MASK;
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_GRBM_MASK;
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32(mmGRBM_STATUS2);
+	if (tmp & GRBM_STATUS2__RLC_BUSY_MASK)
+		grbm_soft_reset |= GRBM_SOFT_RESET__SOFT_RESET_RLC_MASK;
+
+	/* SRBM_STATUS */
+	tmp = RREG32(mmSRBM_STATUS);
+	if (tmp & SRBM_STATUS__GRBM_RQ_PENDING_MASK)
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_GRBM_MASK;
+
+	if (grbm_soft_reset || srbm_soft_reset) {
+		gfx_v7_0_print_status((void *)adev);
+		/* disable CG/PG */
+		gfx_v7_0_fini_pg(adev);
+		gfx_v7_0_update_cg(adev, false);
+
+		/* stop the rlc */
+		gfx_v7_0_rlc_stop(adev);
+
+		/* Disable GFX parsing/prefetching */
+		WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK | CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK);
+
+		/* Disable MEC parsing/prefetching */
+		WREG32(mmCP_MEC_CNTL, CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK);
+
+		if (grbm_soft_reset) {
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+			tmp |= grbm_soft_reset;
+			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32(mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~grbm_soft_reset;
+			WREG32(mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+		}
+
+		if (srbm_soft_reset) {
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+			tmp |= srbm_soft_reset;
+			dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32(mmSRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~srbm_soft_reset;
+			WREG32(mmSRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+		}
+		/* Wait a little for things to settle down */
+		udelay(50);
+		gfx_v7_0_print_status((void *)adev);
+	}
+	return 0;
+}
+
+static void gfx_v7_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v7_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only pipe 0 of MEC1. That's why this function only
+	 * handles the setting of interrupts for this specific pipe. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = mmCP_ME1_PIPE0_INT_CNTL;
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl &= ~CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl |= CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK;
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v7_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__PRIV_REG_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v7_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *src,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl &= ~CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl |= CP_INT_CNTL_RING0__PRIV_INSTR_INT_ENABLE_MASK;
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v7_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_EOP:
+		gfx_v7_0_set_gfx_eop_interrupt_state(adev, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+		gfx_v7_0_set_compute_eop_interrupt_state(adev, 2, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v7_0_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	u8 me_id, pipe_id;
+	struct amdgpu_ring *ring;
+	int i;
+
+	DRM_DEBUG("IH: CP EOP\n");
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	switch (me_id) {
+	case 0:
+		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			if ((ring->me == me_id) & (ring->pipe == pipe_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v7_0_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v7_0_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	// XXX soft reset the gfx block only
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v7_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	gfx_v7_0_enable_gui_idle_interrupt(adev, false);
+	/* order matters! */
+	if (gate) {
+		gfx_v7_0_enable_mgcg(adev, true);
+		gfx_v7_0_enable_cgcg(adev, true);
+	} else {
+		gfx_v7_0_enable_cgcg(adev, false);
+		gfx_v7_0_enable_mgcg(adev, false);
+	}
+	gfx_v7_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static int gfx_v7_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		gate = true;
+
+	if (adev->pg_flags & (AMDGPU_PG_SUPPORT_GFX_PG |
+			      AMDGPU_PG_SUPPORT_GFX_SMG |
+			      AMDGPU_PG_SUPPORT_GFX_DMG |
+			      AMDGPU_PG_SUPPORT_CP |
+			      AMDGPU_PG_SUPPORT_GDS |
+			      AMDGPU_PG_SUPPORT_RLC_SMU_HS)) {
+		gfx_v7_0_update_gfx_pg(adev, gate);
+		if (adev->pg_flags & AMDGPU_PG_SUPPORT_GFX_PG) {
+			gfx_v7_0_enable_cp_pg(adev, gate);
+			gfx_v7_0_enable_gds_pg(adev, gate);
+		}
+	}
+
+	return 0;
+}
+
+const struct amd_ip_funcs gfx_v7_0_ip_funcs = {
+	.early_init = gfx_v7_0_early_init,
+	.late_init = NULL,
+	.sw_init = gfx_v7_0_sw_init,
+	.sw_fini = gfx_v7_0_sw_fini,
+	.hw_init = gfx_v7_0_hw_init,
+	.hw_fini = gfx_v7_0_hw_fini,
+	.suspend = gfx_v7_0_suspend,
+	.resume = gfx_v7_0_resume,
+	.is_idle = gfx_v7_0_is_idle,
+	.wait_for_idle = gfx_v7_0_wait_for_idle,
+	.soft_reset = gfx_v7_0_soft_reset,
+	.print_status = gfx_v7_0_print_status,
+	.set_clockgating_state = gfx_v7_0_set_clockgating_state,
+	.set_powergating_state = gfx_v7_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
+	.get_rptr = gfx_v7_0_ring_get_rptr_gfx,
+	.get_wptr = gfx_v7_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v7_0_ring_set_wptr_gfx,
+	.parse_cs = NULL,
+	.emit_ib = gfx_v7_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v7_0_ring_emit_fence_gfx,
+	.emit_semaphore = gfx_v7_0_ring_emit_semaphore,
+	.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v7_0_ring_test_ring,
+	.test_ib = gfx_v7_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
+	.get_rptr = gfx_v7_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v7_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v7_0_ring_set_wptr_compute,
+	.parse_cs = NULL,
+	.emit_ib = gfx_v7_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v7_0_ring_emit_fence_compute,
+	.emit_semaphore = gfx_v7_0_ring_emit_semaphore,
+	.emit_vm_flush = gfx_v7_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v7_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v7_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v7_0_ring_test_ring,
+	.test_ib = gfx_v7_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v7_0_ring_funcs_gfx;
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v7_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v7_0_eop_irq_funcs = {
+	.set = gfx_v7_0_set_eop_interrupt_state,
+	.process = gfx_v7_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v7_0_priv_reg_irq_funcs = {
+	.set = gfx_v7_0_set_priv_reg_fault_state,
+	.process = gfx_v7_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v7_0_priv_inst_irq_funcs = {
+	.set = gfx_v7_0_set_priv_inst_fault_state,
+	.process = gfx_v7_0_priv_inst_irq,
+};
+
+static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v7_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v7_0_priv_reg_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v7_0_priv_inst_irq_funcs;
+}
+
+static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev)
+{
+	/* init asci gds info */
+	adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
+	adev->gds.gws.total_size = 64;
+	adev->gds.oa.total_size = 16;
+
+	if (adev->gds.mem.total_size == 64 * 1024) {
+		adev->gds.mem.gfx_partition_size = 4096;
+		adev->gds.mem.cs_partition_size = 4096;
+
+		adev->gds.gws.gfx_partition_size = 4;
+		adev->gds.gws.cs_partition_size = 4;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 1;
+	} else {
+		adev->gds.mem.gfx_partition_size = 1024;
+		adev->gds.mem.cs_partition_size = 1024;
+
+		adev->gds.gws.gfx_partition_size = 16;
+		adev->gds.gws.cs_partition_size = 16;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 4;
+	}
+}
+
+
+int gfx_v7_0_get_cu_info(struct amdgpu_device *adev,
+								   struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			bitmap = gfx_v7_0_get_cu_active_bitmap(adev, i, j);
+			cu_info->bitmap[i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+				if (bitmap & mask) {
+					if (counter < 2)
+						ao_bitmap |= mask;
+					counter ++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+		}
+	}
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
new file mode 100644
index 000000000..c04bfbabf
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V7_0_H__
+#define __GFX_V7_0_H__
+
+extern const struct amd_ip_funcs gfx_v7_0_ip_funcs;
+
+/* XXX these shouldn't be exported */
+void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev);
+void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev);
+void gfx_v7_0_rlc_stop(struct amdgpu_device *adev);
+uint64_t gfx_v7_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+void gfx_v7_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
+int gfx_v7_0_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
new file mode 100644
index 000000000..d1054034d
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -0,0 +1,5171 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_gfx.h"
+#include "vi.h"
+#include "vid.h"
+#include "amdgpu_ucode.h"
+#include "clearstate_vi.h"
+
+#include "gmc/gmc_8_2_d.h"
+#include "gmc/gmc_8_2_sh_mask.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "gca/gfx_8_0_sh_mask.h"
+#include "gca/gfx_8_0_enum.h"
+
+#include "uvd/uvd_5_0_d.h"
+#include "uvd/uvd_5_0_sh_mask.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+
+#define GFX8_NUM_GFX_RINGS     1
+#define GFX8_NUM_COMPUTE_RINGS 8
+
+#define TOPAZ_GB_ADDR_CONFIG_GOLDEN 0x22010001
+#define CARRIZO_GB_ADDR_CONFIG_GOLDEN 0x22010001
+#define TONGA_GB_ADDR_CONFIG_GOLDEN 0x22011003
+
+#define ARRAY_MODE(x)					((x) << GB_TILE_MODE0__ARRAY_MODE__SHIFT)
+#define PIPE_CONFIG(x)					((x) << GB_TILE_MODE0__PIPE_CONFIG__SHIFT)
+#define TILE_SPLIT(x)					((x) << GB_TILE_MODE0__TILE_SPLIT__SHIFT)
+#define MICRO_TILE_MODE_NEW(x)				((x) << GB_TILE_MODE0__MICRO_TILE_MODE_NEW__SHIFT)
+#define SAMPLE_SPLIT(x)					((x) << GB_TILE_MODE0__SAMPLE_SPLIT__SHIFT)
+#define BANK_WIDTH(x)					((x) << GB_MACROTILE_MODE0__BANK_WIDTH__SHIFT)
+#define BANK_HEIGHT(x)					((x) << GB_MACROTILE_MODE0__BANK_HEIGHT__SHIFT)
+#define MACRO_TILE_ASPECT(x)				((x) << GB_MACROTILE_MODE0__MACRO_TILE_ASPECT__SHIFT)
+#define NUM_BANKS(x)					((x) << GB_MACROTILE_MODE0__NUM_BANKS__SHIFT)
+
+MODULE_FIRMWARE("amdgpu/carrizo_ce.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_pfp.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_me.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_mec.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_mec2.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/stoney_ce.bin");
+MODULE_FIRMWARE("amdgpu/stoney_pfp.bin");
+MODULE_FIRMWARE("amdgpu/stoney_me.bin");
+MODULE_FIRMWARE("amdgpu/stoney_mec.bin");
+MODULE_FIRMWARE("amdgpu/stoney_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/tonga_ce.bin");
+MODULE_FIRMWARE("amdgpu/tonga_pfp.bin");
+MODULE_FIRMWARE("amdgpu/tonga_me.bin");
+MODULE_FIRMWARE("amdgpu/tonga_mec.bin");
+MODULE_FIRMWARE("amdgpu/tonga_mec2.bin");
+MODULE_FIRMWARE("amdgpu/tonga_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/topaz_ce.bin");
+MODULE_FIRMWARE("amdgpu/topaz_pfp.bin");
+MODULE_FIRMWARE("amdgpu/topaz_me.bin");
+MODULE_FIRMWARE("amdgpu/topaz_mec.bin");
+MODULE_FIRMWARE("amdgpu/topaz_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/fiji_ce.bin");
+MODULE_FIRMWARE("amdgpu/fiji_pfp.bin");
+MODULE_FIRMWARE("amdgpu/fiji_me.bin");
+MODULE_FIRMWARE("amdgpu/fiji_mec.bin");
+MODULE_FIRMWARE("amdgpu/fiji_mec2.bin");
+MODULE_FIRMWARE("amdgpu/fiji_rlc.bin");
+
+static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] =
+{
+	{mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0},
+	{mmGDS_VMID1_BASE, mmGDS_VMID1_SIZE, mmGDS_GWS_VMID1, mmGDS_OA_VMID1},
+	{mmGDS_VMID2_BASE, mmGDS_VMID2_SIZE, mmGDS_GWS_VMID2, mmGDS_OA_VMID2},
+	{mmGDS_VMID3_BASE, mmGDS_VMID3_SIZE, mmGDS_GWS_VMID3, mmGDS_OA_VMID3},
+	{mmGDS_VMID4_BASE, mmGDS_VMID4_SIZE, mmGDS_GWS_VMID4, mmGDS_OA_VMID4},
+	{mmGDS_VMID5_BASE, mmGDS_VMID5_SIZE, mmGDS_GWS_VMID5, mmGDS_OA_VMID5},
+	{mmGDS_VMID6_BASE, mmGDS_VMID6_SIZE, mmGDS_GWS_VMID6, mmGDS_OA_VMID6},
+	{mmGDS_VMID7_BASE, mmGDS_VMID7_SIZE, mmGDS_GWS_VMID7, mmGDS_OA_VMID7},
+	{mmGDS_VMID8_BASE, mmGDS_VMID8_SIZE, mmGDS_GWS_VMID8, mmGDS_OA_VMID8},
+	{mmGDS_VMID9_BASE, mmGDS_VMID9_SIZE, mmGDS_GWS_VMID9, mmGDS_OA_VMID9},
+	{mmGDS_VMID10_BASE, mmGDS_VMID10_SIZE, mmGDS_GWS_VMID10, mmGDS_OA_VMID10},
+	{mmGDS_VMID11_BASE, mmGDS_VMID11_SIZE, mmGDS_GWS_VMID11, mmGDS_OA_VMID11},
+	{mmGDS_VMID12_BASE, mmGDS_VMID12_SIZE, mmGDS_GWS_VMID12, mmGDS_OA_VMID12},
+	{mmGDS_VMID13_BASE, mmGDS_VMID13_SIZE, mmGDS_GWS_VMID13, mmGDS_OA_VMID13},
+	{mmGDS_VMID14_BASE, mmGDS_VMID14_SIZE, mmGDS_GWS_VMID14, mmGDS_OA_VMID14},
+	{mmGDS_VMID15_BASE, mmGDS_VMID15_SIZE, mmGDS_GWS_VMID15, mmGDS_OA_VMID15}
+};
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmCB_HW_CONTROL, 0xfffdf3cf, 0x00007208,
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_FIFO_DEPTH_CNTL, 0x000003ff, 0x000000fc,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000002fb,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x0000543b,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0xa9210876,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 tonga_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x16000012,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x0000002A,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x40000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU0_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU4_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU6_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU6_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU6_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU6_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU6_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU7_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU7_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU7_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU7_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU7_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+};
+
+static const u32 fiji_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x3a00161a,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x0000002e,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x00000009,
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000ff,
+	mmVGT_RESET_DEBUG, 0x00000004, 0x00000004,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x40000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+};
+
+static const u32 golden_settings_iceland_a11[] =
+{
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmDB_DEBUG3, 0xc0000000, 0xc0000000,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x00000002,
+	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x00000000,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f1,
+	mmTCP_CHAN_STEER_HI, 0xffffffff, 0x00000000,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00000010,
+};
+
+static const u32 iceland_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x00000002,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x00000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22010001,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0xc0000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0xc0000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0xc0000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0xff000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x0f840f87,
+	mmCGTS_CU0_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x0f840f87,
+	mmCGTS_CU4_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003c,
+};
+
+static const u32 cz_golden_settings_a11[] =
+{
+	mmCB_HW_CONTROL_3, 0x00000040, 0x00000040,
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x00000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
+	mmTA_CNTL_AUX, 0x000f000f, 0x00010000,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x0000000f, 0x000000f3,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00001302
+};
+
+static const u32 cz_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x00000002,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x00000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x22010001,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmRLC_CGTT_MGCG_OVERRIDE, 0xffffffff, 0xffffffff,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCB_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_BCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_CPF_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_GDS_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_IA_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PA_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_WD_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmCGTT_PC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_RLC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SC_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SPI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQ_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SQG_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL1, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL2, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL3, 0xffffffff, 0x00000100,
+	mmCGTT_SX_CLK_CTRL4, 0xffffffff, 0x00000100,
+	mmCGTT_TCI_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_TCP_CLK_CTRL, 0xffffffff, 0x00000100,
+	mmCGTT_VGT_CLK_CTRL, 0xffffffff, 0x06000100,
+	mmDB_CGTT_CLK_CTRL_0, 0xffffffff, 0x00000100,
+	mmTA_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmTCA_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTCC_CGTT_SCLK_CTRL, 0xffffffff, 0x00000100,
+	mmTD_CGTT_CTRL, 0xffffffff, 0x00000100,
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmCGTS_CU0_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU0_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU0_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU0_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU0_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU1_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU1_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU1_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU1_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU1_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU2_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU2_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU2_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU2_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU2_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU3_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU3_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU3_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU3_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU3_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU4_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU4_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU4_TA_SQC_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU4_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU4_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU5_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU5_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU5_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU5_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU5_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU6_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU6_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU6_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU6_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU6_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_CU7_SP0_CTRL_REG, 0xffffffff, 0x00010000,
+	mmCGTS_CU7_LDS_SQ_CTRL_REG, 0xffffffff, 0x00030002,
+	mmCGTS_CU7_TA_CTRL_REG, 0xffffffff, 0x00040007,
+	mmCGTS_CU7_SP1_CTRL_REG, 0xffffffff, 0x00060005,
+	mmCGTS_CU7_TD_TCP_CTRL_REG, 0xffffffff, 0x00090008,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96e00200,
+	mmCP_RB_WPTR_POLL_CNTL, 0xffffffff, 0x00900100,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	mmCP_MEM_SLP_CNTL, 0x00000001, 0x00000001,
+};
+
+static const u32 stoney_golden_settings_a11[] =
+{
+	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
+	mmGB_GPU_ID, 0x0000000f, 0x00000000,
+	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
+	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
+	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
+  	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
+	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
+	mmTCP_ADDR_CONFIG, 0x0000000f, 0x000000f1,
+	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x10101010,
+};
+
+static const u32 stoney_golden_common_all[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmPA_SC_RASTER_CONFIG, 0xffffffff, 0x00000000,
+	mmPA_SC_RASTER_CONFIG_1, 0xffffffff, 0x00000000,
+	mmGB_ADDR_CONFIG, 0xffffffff, 0x12010001,
+	mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
+	mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
+	mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
+	mmRLC_CGCG_CGLS_CTRL, 0xffffffff, 0x0020003f,
+	mmCP_MEM_SLP_CNTL, 0xffffffff, 0x00020201,
+	mmRLC_MEM_SLP_CNTL, 0xffffffff, 0x00020201,
+	mmCGTS_SM_CTRL_REG, 0xffffffff, 0x96940200,
+	mmATC_MISC_CG, 0xffffffff, 0x000c0200,
+};
+
+static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev);
+static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev);
+
+static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_program_register_sequence(adev,
+						 iceland_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_iceland_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_iceland_a11));
+		amdgpu_program_register_sequence(adev,
+						 iceland_golden_common_all,
+						 (const u32)ARRAY_SIZE(iceland_golden_common_all));
+		break;
+	case CHIP_FIJI:
+		amdgpu_program_register_sequence(adev,
+						 fiji_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_fiji_a10,
+						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		amdgpu_program_register_sequence(adev,
+						 fiji_golden_common_all,
+						 (const u32)ARRAY_SIZE(fiji_golden_common_all));
+		break;
+
+	case CHIP_TONGA:
+		amdgpu_program_register_sequence(adev,
+						 tonga_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_tonga_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		amdgpu_program_register_sequence(adev,
+						 tonga_golden_common_all,
+						 (const u32)ARRAY_SIZE(tonga_golden_common_all));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_program_register_sequence(adev,
+						 cz_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 cz_golden_settings_a11,
+						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
+		amdgpu_program_register_sequence(adev,
+						 cz_golden_common_all,
+						 (const u32)ARRAY_SIZE(cz_golden_common_all));
+		break;
+	case CHIP_STONEY:
+		amdgpu_program_register_sequence(adev,
+						 stoney_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 stoney_golden_settings_a11,
+						 (const u32)ARRAY_SIZE(stoney_golden_settings_a11));
+		amdgpu_program_register_sequence(adev,
+						 stoney_golden_common_all,
+						 (const u32)ARRAY_SIZE(stoney_golden_common_all));
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v8_0_scratch_init(struct amdgpu_device *adev)
+{
+	int i;
+
+	adev->gfx.scratch.num_reg = 7;
+	adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
+	for (i = 0; i < adev->gfx.scratch.num_reg; i++) {
+		adev->gfx.scratch.free[i] = true;
+		adev->gfx.scratch.reg[i] = adev->gfx.scratch.reg_base + i;
+	}
+}
+
+static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	r = amdgpu_ring_lock(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		amdgpu_gfx_scratch_free(adev, scratch);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG, 1));
+	amdgpu_ring_write(ring, (scratch - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (scratch(0x%04X)=0x%08X)\n",
+			  ring->idx, scratch, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct fence *f = NULL;
+	uint32_t scratch;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	r = amdgpu_gfx_scratch_get(adev, &scratch);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get scratch reg (%d).\n", r);
+		return r;
+	}
+	WREG32(scratch, 0xCAFEDEAD);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(ring, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		goto err1;
+	}
+	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
+	ib.ptr[1] = ((scratch - PACKET3_SET_UCONFIG_REG_START));
+	ib.ptr[2] = 0xDEADBEEF;
+	ib.length_dw = 3;
+
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err2;
+
+	r = fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto err2;
+	}
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(scratch);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
+			 ring->idx, i);
+		goto err2;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
+			  scratch, tmp);
+		r = -EINVAL;
+	}
+err2:
+	fence_put(f);
+	amdgpu_ib_free(adev, &ib);
+err1:
+	amdgpu_gfx_scratch_free(adev, scratch);
+	return r;
+}
+
+static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct gfx_firmware_header_v1_0 *cp_hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		chip_name = "topaz";
+		break;
+	case CHIP_TONGA:
+		chip_name = "tonga";
+		break;
+	case CHIP_CARRIZO:
+		chip_name = "carrizo";
+		break;
+	case CHIP_FIJI:
+		chip_name = "fiji";
+		break;
+	case CHIP_STONEY:
+		chip_name = "stoney";
+		break;
+	default:
+		BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name);
+	err = request_firmware(&adev->gfx.pfp_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.pfp_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+	adev->gfx.pfp_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.pfp_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name);
+	err = request_firmware(&adev->gfx.me_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.me_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+	adev->gfx.me_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.me_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name);
+	err = request_firmware(&adev->gfx.ce_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.ce_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+	adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
+	err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.rlc_fw);
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlc_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.rlc_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name);
+	err = request_firmware(&adev->gfx.mec_fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->gfx.mec_fw);
+	if (err)
+		goto out;
+	cp_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	adev->gfx.mec_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
+	adev->gfx.mec_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
+
+	if ((adev->asic_type != CHIP_STONEY) &&
+	    (adev->asic_type != CHIP_TOPAZ)) {
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name);
+		err = request_firmware(&adev->gfx.mec2_fw, fw_name, adev->dev);
+		if (!err) {
+			err = amdgpu_ucode_validate(adev->gfx.mec2_fw);
+			if (err)
+				goto out;
+			cp_hdr = (const struct gfx_firmware_header_v1_0 *)
+				adev->gfx.mec2_fw->data;
+			adev->gfx.mec2_fw_version =
+				le32_to_cpu(cp_hdr->header.ucode_version);
+			adev->gfx.mec2_feature_version =
+				le32_to_cpu(cp_hdr->ucode_feature_version);
+		} else {
+			err = 0;
+			adev->gfx.mec2_fw = NULL;
+		}
+	}
+
+	if (adev->firmware.smu_load) {
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_PFP];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_PFP;
+		info->fw = adev->gfx.pfp_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_ME];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_ME;
+		info->fw = adev->gfx.me_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_CE];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_CE;
+		info->fw = adev->gfx.ce_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_RLC_G];
+		info->ucode_id = AMDGPU_UCODE_ID_RLC_G;
+		info->fw = adev->gfx.rlc_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC1];
+		info->ucode_id = AMDGPU_UCODE_ID_CP_MEC1;
+		info->fw = adev->gfx.mec_fw;
+		header = (const struct common_firmware_header *)info->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+
+		if (adev->gfx.mec2_fw) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CP_MEC2];
+			info->ucode_id = AMDGPU_UCODE_ID_CP_MEC2;
+			info->fw = adev->gfx.mec2_fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+
+	}
+
+out:
+	if (err) {
+		dev_err(adev->dev,
+			"gfx8: Failed to load firmware \"%s\"\n",
+			fw_name);
+		release_firmware(adev->gfx.pfp_fw);
+		adev->gfx.pfp_fw = NULL;
+		release_firmware(adev->gfx.me_fw);
+		adev->gfx.me_fw = NULL;
+		release_firmware(adev->gfx.ce_fw);
+		adev->gfx.ce_fw = NULL;
+		release_firmware(adev->gfx.rlc_fw);
+		adev->gfx.rlc_fw = NULL;
+		release_firmware(adev->gfx.mec_fw);
+		adev->gfx.mec_fw = NULL;
+		release_firmware(adev->gfx.mec2_fw);
+		adev->gfx.mec2_fw = NULL;
+	}
+	return err;
+}
+
+static void gfx_v8_0_mec_fini(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gfx.mec.hpd_eop_obj) {
+		r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
+		if (unlikely(r != 0))
+			dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
+		amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
+		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+		amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
+		adev->gfx.mec.hpd_eop_obj = NULL;
+	}
+}
+
+#define MEC_HPD_SIZE 2048
+
+static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
+{
+	int r;
+	u32 *hpd;
+
+	/*
+	 * we assign only 1 pipe because all other pipes will
+	 * be handled by KFD
+	 */
+	adev->gfx.mec.num_mec = 1;
+	adev->gfx.mec.num_pipe = 1;
+	adev->gfx.mec.num_queue = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe * 8;
+
+	if (adev->gfx.mec.hpd_eop_obj == NULL) {
+		r = amdgpu_bo_create(adev,
+				     adev->gfx.mec.num_mec *adev->gfx.mec.num_pipe * MEC_HPD_SIZE * 2,
+				     PAGE_SIZE, true,
+				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
+				     &adev->gfx.mec.hpd_eop_obj);
+		if (r) {
+			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
+			return r;
+		}
+	}
+
+	r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
+	if (unlikely(r != 0)) {
+		gfx_v8_0_mec_fini(adev);
+		return r;
+	}
+	r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
+			  &adev->gfx.mec.hpd_eop_gpu_addr);
+	if (r) {
+		dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
+		gfx_v8_0_mec_fini(adev);
+		return r;
+	}
+	r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
+	if (r) {
+		dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
+		gfx_v8_0_mec_fini(adev);
+		return r;
+	}
+
+	memset(hpd, 0, adev->gfx.mec.num_mec *adev->gfx.mec.num_pipe * MEC_HPD_SIZE * 2);
+
+	amdgpu_bo_kunmap(adev->gfx.mec.hpd_eop_obj);
+	amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
+
+	return 0;
+}
+
+static void gfx_v8_0_gpu_early_init(struct amdgpu_device *adev)
+{
+	u32 gb_addr_config;
+	u32 mc_shared_chmap, mc_arb_ramcfg;
+	u32 dimm00_addr_map, dimm01_addr_map, dimm10_addr_map, dimm11_addr_map;
+	u32 tmp;
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_cu_per_sh = 6;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TOPAZ_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_FIJI:
+		adev->gfx.config.max_shader_engines = 4;
+		adev->gfx.config.max_tile_pipes = 16;
+		adev->gfx.config.max_cu_per_sh = 16;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 4;
+		adev->gfx.config.max_texture_channel_caches = 16;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_TONGA:
+		adev->gfx.config.max_shader_engines = 4;
+		adev->gfx.config.max_tile_pipes = 8;
+		adev->gfx.config.max_cu_per_sh = 8;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 8;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_CARRIZO:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+
+		switch (adev->pdev->revision) {
+		case 0xc4:
+		case 0x84:
+		case 0xc8:
+		case 0xcc:
+		case 0xe1:
+		case 0xe3:
+			/* B10 */
+			adev->gfx.config.max_cu_per_sh = 8;
+			break;
+		case 0xc5:
+		case 0x81:
+		case 0x85:
+		case 0xc9:
+		case 0xcd:
+		case 0xe2:
+		case 0xe4:
+			/* B8 */
+			adev->gfx.config.max_cu_per_sh = 6;
+			break;
+		case 0xc6:
+		case 0xca:
+		case 0xce:
+		case 0x88:
+			/* B6 */
+			adev->gfx.config.max_cu_per_sh = 6;
+			break;
+		case 0xc7:
+		case 0x87:
+		case 0xcb:
+		case 0xe5:
+		case 0x89:
+		default:
+			/* B4 */
+			adev->gfx.config.max_cu_per_sh = 4;
+			break;
+		}
+
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = CARRIZO_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	case CHIP_STONEY:
+		adev->gfx.config.max_shader_engines = 1;
+		adev->gfx.config.max_tile_pipes = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 1;
+
+		switch (adev->pdev->revision) {
+		case 0xc0:
+		case 0xc1:
+		case 0xc2:
+		case 0xc4:
+		case 0xc8:
+		case 0xc9:
+			adev->gfx.config.max_cu_per_sh = 3;
+			break;
+		case 0xd0:
+		case 0xd1:
+		case 0xd2:
+		default:
+			adev->gfx.config.max_cu_per_sh = 2;
+			break;
+		}
+
+		adev->gfx.config.max_texture_channel_caches = 2;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 16;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = CARRIZO_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	default:
+		adev->gfx.config.max_shader_engines = 2;
+		adev->gfx.config.max_tile_pipes = 4;
+		adev->gfx.config.max_cu_per_sh = 2;
+		adev->gfx.config.max_sh_per_se = 1;
+		adev->gfx.config.max_backends_per_se = 2;
+		adev->gfx.config.max_texture_channel_caches = 4;
+		adev->gfx.config.max_gprs = 256;
+		adev->gfx.config.max_gs_threads = 32;
+		adev->gfx.config.max_hw_contexts = 8;
+
+		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
+		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
+		adev->gfx.config.sc_hiz_tile_fifo_size = 0x30;
+		adev->gfx.config.sc_earlyz_tile_fifo_size = 0x130;
+		gb_addr_config = TONGA_GB_ADDR_CONFIG_GOLDEN;
+		break;
+	}
+
+	mc_shared_chmap = RREG32(mmMC_SHARED_CHMAP);
+	adev->gfx.config.mc_arb_ramcfg = RREG32(mmMC_ARB_RAMCFG);
+	mc_arb_ramcfg = adev->gfx.config.mc_arb_ramcfg;
+
+	adev->gfx.config.num_tile_pipes = adev->gfx.config.max_tile_pipes;
+	adev->gfx.config.mem_max_burst_length_bytes = 256;
+	if (adev->flags & AMD_IS_APU) {
+		/* Get memory bank mapping mode. */
+		tmp = RREG32(mmMC_FUS_DRAM0_BANK_ADDR_MAPPING);
+		dimm00_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm01_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM0_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		tmp = RREG32(mmMC_FUS_DRAM1_BANK_ADDR_MAPPING);
+		dimm10_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM0ADDRMAP);
+		dimm11_addr_map = REG_GET_FIELD(tmp, MC_FUS_DRAM1_BANK_ADDR_MAPPING, DIMM1ADDRMAP);
+
+		/* Validate settings in case only one DIMM installed. */
+		if ((dimm00_addr_map == 0) || (dimm00_addr_map == 3) || (dimm00_addr_map == 4) || (dimm00_addr_map > 12))
+			dimm00_addr_map = 0;
+		if ((dimm01_addr_map == 0) || (dimm01_addr_map == 3) || (dimm01_addr_map == 4) || (dimm01_addr_map > 12))
+			dimm01_addr_map = 0;
+		if ((dimm10_addr_map == 0) || (dimm10_addr_map == 3) || (dimm10_addr_map == 4) || (dimm10_addr_map > 12))
+			dimm10_addr_map = 0;
+		if ((dimm11_addr_map == 0) || (dimm11_addr_map == 3) || (dimm11_addr_map == 4) || (dimm11_addr_map > 12))
+			dimm11_addr_map = 0;
+
+		/* If DIMM Addr map is 8GB, ROW size should be 2KB. Otherwise 1KB. */
+		/* If ROW size(DIMM1) != ROW size(DMIMM0), ROW size should be larger one. */
+		if ((dimm00_addr_map == 11) || (dimm01_addr_map == 11) || (dimm10_addr_map == 11) || (dimm11_addr_map == 11))
+			adev->gfx.config.mem_row_size_in_kb = 2;
+		else
+			adev->gfx.config.mem_row_size_in_kb = 1;
+	} else {
+		tmp = REG_GET_FIELD(mc_arb_ramcfg, MC_ARB_RAMCFG, NOOFCOLS);
+		adev->gfx.config.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024;
+		if (adev->gfx.config.mem_row_size_in_kb > 4)
+			adev->gfx.config.mem_row_size_in_kb = 4;
+	}
+
+	adev->gfx.config.shader_engine_tile_size = 32;
+	adev->gfx.config.num_gpus = 1;
+	adev->gfx.config.multi_gpu_tile_size = 64;
+
+	/* fix up row size */
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+	default:
+		gb_addr_config = REG_SET_FIELD(gb_addr_config, GB_ADDR_CONFIG, ROW_SIZE, 0);
+		break;
+	case 2:
+		gb_addr_config = REG_SET_FIELD(gb_addr_config, GB_ADDR_CONFIG, ROW_SIZE, 1);
+		break;
+	case 4:
+		gb_addr_config = REG_SET_FIELD(gb_addr_config, GB_ADDR_CONFIG, ROW_SIZE, 2);
+		break;
+	}
+	adev->gfx.config.gb_addr_config = gb_addr_config;
+}
+
+static int gfx_v8_0_sw_init(void *handle)
+{
+	int i, r;
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* EOP Event */
+	r = amdgpu_irq_add_id(adev, 181, &adev->gfx.eop_irq);
+	if (r)
+		return r;
+
+	/* Privileged reg */
+	r = amdgpu_irq_add_id(adev, 184, &adev->gfx.priv_reg_irq);
+	if (r)
+		return r;
+
+	/* Privileged inst */
+	r = amdgpu_irq_add_id(adev, 185, &adev->gfx.priv_inst_irq);
+	if (r)
+		return r;
+
+	adev->gfx.gfx_current_status = AMDGPU_GFX_NORMAL_MODE;
+
+	gfx_v8_0_scratch_init(adev);
+
+	r = gfx_v8_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load gfx firmware!\n");
+		return r;
+	}
+
+	r = gfx_v8_0_mec_init(adev);
+	if (r) {
+		DRM_ERROR("Failed to init MEC BOs!\n");
+		return r;
+	}
+
+	/* set up the gfx ring */
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
+		ring = &adev->gfx.gfx_ring[i];
+		ring->ring_obj = NULL;
+		sprintf(ring->name, "gfx");
+		/* no gfx doorbells on iceland */
+		if (adev->asic_type != CHIP_TOPAZ) {
+			ring->use_doorbell = true;
+			ring->doorbell_index = AMDGPU_DOORBELL_GFX_RING0;
+		}
+
+		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
+				     &adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP,
+				     AMDGPU_RING_TYPE_GFX);
+		if (r)
+			return r;
+	}
+
+	/* set up the compute queues */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		unsigned irq_type;
+
+		/* max 32 queues per MEC */
+		if ((i >= 32) || (i >= AMDGPU_MAX_COMPUTE_RINGS)) {
+			DRM_ERROR("Too many (%d) compute rings!\n", i);
+			break;
+		}
+		ring = &adev->gfx.compute_ring[i];
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		ring->doorbell_index = AMDGPU_DOORBELL_MEC_RING0 + i;
+		ring->me = 1; /* first MEC */
+		ring->pipe = i / 8;
+		ring->queue = i % 8;
+		sprintf(ring->name, "comp %d.%d.%d", ring->me, ring->pipe, ring->queue);
+		irq_type = AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP + ring->pipe;
+		/* type-2 packets are deprecated on MEC, use type-3 instead */
+		r = amdgpu_ring_init(adev, ring, 1024 * 1024,
+				     PACKET3(PACKET3_NOP, 0x3FFF), 0xf,
+				     &adev->gfx.eop_irq, irq_type,
+				     AMDGPU_RING_TYPE_COMPUTE);
+		if (r)
+			return r;
+	}
+
+	/* reserve GDS, GWS and OA resource for gfx */
+	r = amdgpu_bo_create(adev, adev->gds.mem.gfx_partition_size,
+			PAGE_SIZE, true,
+			AMDGPU_GEM_DOMAIN_GDS, 0, NULL,
+			NULL, &adev->gds.gds_gfx_bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create(adev, adev->gds.gws.gfx_partition_size,
+		PAGE_SIZE, true,
+		AMDGPU_GEM_DOMAIN_GWS, 0, NULL,
+		NULL, &adev->gds.gws_gfx_bo);
+	if (r)
+		return r;
+
+	r = amdgpu_bo_create(adev, adev->gds.oa.gfx_partition_size,
+			PAGE_SIZE, true,
+			AMDGPU_GEM_DOMAIN_OA, 0, NULL,
+			NULL, &adev->gds.oa_gfx_bo);
+	if (r)
+		return r;
+
+	adev->gfx.ce_ram_size = 0x8000;
+
+	gfx_v8_0_gpu_early_init(adev);
+
+	return 0;
+}
+
+static int gfx_v8_0_sw_fini(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_bo_unref(&adev->gds.oa_gfx_bo);
+	amdgpu_bo_unref(&adev->gds.gws_gfx_bo);
+	amdgpu_bo_unref(&adev->gds.gds_gfx_bo);
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
+
+	gfx_v8_0_mec_fini(adev);
+
+	return 0;
+}
+
+static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
+{
+	const u32 num_tile_mode_states = 32;
+	const u32 num_secondary_tile_mode_states = 16;
+	u32 reg_offset, gb_tile_moden, split_equal_to_row_size;
+
+	switch (adev->gfx.config.mem_row_size_in_kb) {
+	case 1:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_1KB;
+		break;
+	case 2:
+	default:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_2KB;
+		break;
+	case 4:
+		split_equal_to_row_size = ADDR_SURF_TILE_SPLIT_4KB;
+		break;
+	}
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P2));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 7:
+			case 12:
+			case 17:
+			case 23:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 7:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+	case CHIP_FIJI:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 7:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 12:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 17:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 23:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 30:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			case 7:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			}
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		break;
+	case CHIP_TONGA:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 7:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 12:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 17:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 23:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 30:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P4_16x16) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_1) |
+						NUM_BANKS(ADDR_SURF_4_BANK));
+				break;
+			case 7:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		break;
+	case CHIP_STONEY:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P2));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 7:
+			case 12:
+			case 17:
+			case 23:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 7:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		break;
+	case CHIP_CARRIZO:
+	default:
+		for (reg_offset = 0; reg_offset < num_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_64B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 1:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_128B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 2:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_256B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 3:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_512B) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 4:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 5:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 6:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						TILE_SPLIT(ADDR_SURF_TILE_SPLIT_2KB) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING));
+				break;
+			case 8:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_LINEAR_ALIGNED) |
+						PIPE_CONFIG(ADDR_SURF_P2));
+				break;
+			case 9:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 10:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 11:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 13:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 14:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 15:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 16:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 18:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 19:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 20:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 21:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 22:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 24:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 25:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 26:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_3D_TILED_XTHICK) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_THICK_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_1));
+				break;
+			case 27:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 28:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_2));
+				break;
+			case 29:
+				gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_TILED_THIN1) |
+						PIPE_CONFIG(ADDR_SURF_P2) |
+						MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) |
+						SAMPLE_SPLIT(ADDR_SURF_SAMPLE_SPLIT_8));
+				break;
+			case 7:
+			case 12:
+			case 17:
+			case 23:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.tile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_TILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+		for (reg_offset = 0; reg_offset < num_secondary_tile_mode_states; reg_offset++) {
+			switch (reg_offset) {
+			case 0:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 1:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 2:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 3:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 4:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 5:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 6:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 8:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_8) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 9:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_4) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 10:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_4) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 11:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_2) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 12:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_2) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 13:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_4) |
+						NUM_BANKS(ADDR_SURF_16_BANK));
+				break;
+			case 14:
+				gb_tile_moden = (BANK_WIDTH(ADDR_SURF_BANK_WIDTH_1) |
+						BANK_HEIGHT(ADDR_SURF_BANK_HEIGHT_1) |
+						MACRO_TILE_ASPECT(ADDR_SURF_MACRO_ASPECT_2) |
+						NUM_BANKS(ADDR_SURF_8_BANK));
+				break;
+			case 7:
+				/* unused idx */
+				continue;
+			default:
+				gb_tile_moden = 0;
+				break;
+			};
+			adev->gfx.config.macrotile_mode_array[reg_offset] = gb_tile_moden;
+			WREG32(mmGB_MACROTILE_MODE0 + reg_offset, gb_tile_moden);
+		}
+	}
+}
+
+static u32 gfx_v8_0_create_bitmask(u32 bit_width)
+{
+	u32 i, mask = 0;
+
+	for (i = 0; i < bit_width; i++) {
+		mask <<= 1;
+		mask |= 1;
+	}
+	return mask;
+}
+
+void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num)
+{
+	u32 data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+
+	if ((se_num == 0xffffffff) && (sh_num == 0xffffffff)) {
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1);
+	} else if (se_num == 0xffffffff) {
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1);
+	} else if (sh_num == 0xffffffff) {
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+	} else {
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
+		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
+	}
+	WREG32(mmGRBM_GFX_INDEX, data);
+}
+
+static u32 gfx_v8_0_get_rb_disabled(struct amdgpu_device *adev,
+				    u32 max_rb_num_per_se,
+				    u32 sh_per_se)
+{
+	u32 data, mask;
+
+	data = RREG32(mmCC_RB_BACKEND_DISABLE);
+	data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
+
+	data |= RREG32(mmGC_USER_RB_BACKEND_DISABLE);
+
+	data >>= GC_USER_RB_BACKEND_DISABLE__BACKEND_DISABLE__SHIFT;
+
+	mask = gfx_v8_0_create_bitmask(max_rb_num_per_se / sh_per_se);
+
+	return data & mask;
+}
+
+static void gfx_v8_0_setup_rb(struct amdgpu_device *adev,
+			      u32 se_num, u32 sh_per_se,
+			      u32 max_rb_num_per_se)
+{
+	int i, j;
+	u32 data, mask;
+	u32 disabled_rbs = 0;
+	u32 enabled_rbs = 0;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < se_num; i++) {
+		for (j = 0; j < sh_per_se; j++) {
+			gfx_v8_0_select_se_sh(adev, i, j);
+			data = gfx_v8_0_get_rb_disabled(adev,
+					      max_rb_num_per_se, sh_per_se);
+			disabled_rbs |= data << ((i * sh_per_se + j) *
+						 RB_BITMAP_WIDTH_PER_SH);
+		}
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = 1;
+	for (i = 0; i < max_rb_num_per_se * se_num; i++) {
+		if (!(disabled_rbs & mask))
+			enabled_rbs |= mask;
+		mask <<= 1;
+	}
+
+	adev->gfx.config.backend_enable_mask = enabled_rbs;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < se_num; i++) {
+		gfx_v8_0_select_se_sh(adev, i, 0xffffffff);
+		data = 0;
+		for (j = 0; j < sh_per_se; j++) {
+			switch (enabled_rbs & 3) {
+			case 0:
+				if (j == 0)
+					data |= (RASTER_CONFIG_RB_MAP_3 <<
+						 PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT);
+				else
+					data |= (RASTER_CONFIG_RB_MAP_0 <<
+						 PA_SC_RASTER_CONFIG__PKR_MAP__SHIFT);
+				break;
+			case 1:
+				data |= (RASTER_CONFIG_RB_MAP_0 <<
+					 (i * sh_per_se + j) * 2);
+				break;
+			case 2:
+				data |= (RASTER_CONFIG_RB_MAP_3 <<
+					 (i * sh_per_se + j) * 2);
+				break;
+			case 3:
+			default:
+				data |= (RASTER_CONFIG_RB_MAP_2 <<
+					 (i * sh_per_se + j) * 2);
+				break;
+			}
+			enabled_rbs >>= 2;
+		}
+		WREG32(mmPA_SC_RASTER_CONFIG, data);
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * gfx_v8_0_init_compute_vmid - gart enable
+ *
+ * @rdev: amdgpu_device pointer
+ *
+ * Initialize compute vmid sh_mem registers
+ *
+ */
+#define DEFAULT_SH_MEM_BASES	(0x6000)
+#define FIRST_COMPUTE_VMID	(8)
+#define LAST_COMPUTE_VMID	(16)
+static void gfx_v8_0_init_compute_vmid(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t sh_mem_config;
+	uint32_t sh_mem_bases;
+
+	/*
+	 * Configure apertures:
+	 * LDS:         0x60000000'00000000 - 0x60000001'00000000 (4GB)
+	 * Scratch:     0x60000001'00000000 - 0x60000002'00000000 (4GB)
+	 * GPUVM:       0x60010000'00000000 - 0x60020000'00000000 (1TB)
+	 */
+	sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
+
+	sh_mem_config = SH_MEM_ADDRESS_MODE_HSA64 <<
+			SH_MEM_CONFIG__ADDRESS_MODE__SHIFT |
+			SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
+			SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT |
+			MTYPE_CC << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT |
+			SH_MEM_CONFIG__PRIVATE_ATC_MASK;
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
+		vi_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		WREG32(mmSH_MEM_CONFIG, sh_mem_config);
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, sh_mem_bases);
+	}
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int i;
+
+	tmp = RREG32(mmGRBM_CNTL);
+	tmp = REG_SET_FIELD(tmp, GRBM_CNTL, READ_TIMEOUT, 0xff);
+	WREG32(mmGRBM_CNTL, tmp);
+
+	WREG32(mmGB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmHDP_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmDMIF_ADDR_CALC, adev->gfx.config.gb_addr_config);
+	WREG32(mmSDMA0_TILING_CONFIG + SDMA0_REGISTER_OFFSET,
+	       adev->gfx.config.gb_addr_config & 0x70);
+	WREG32(mmSDMA0_TILING_CONFIG + SDMA1_REGISTER_OFFSET,
+	       adev->gfx.config.gb_addr_config & 0x70);
+	WREG32(mmUVD_UDEC_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DB_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+	WREG32(mmUVD_UDEC_DBW_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
+
+	gfx_v8_0_tiling_mode_table_init(adev);
+
+	gfx_v8_0_setup_rb(adev, adev->gfx.config.max_shader_engines,
+				 adev->gfx.config.max_sh_per_se,
+				 adev->gfx.config.max_backends_per_se);
+
+	/* XXX SH_MEM regs */
+	/* where to put LDS, scratch, GPUVM in FSA64 space */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < 16; i++) {
+		vi_srbm_select(adev, 0, 0, 0, i);
+		/* CP and shaders */
+		if (i == 0) {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, DEFAULT_MTYPE, MTYPE_UC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, APE1_MTYPE, MTYPE_UC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			WREG32(mmSH_MEM_CONFIG, tmp);
+		} else {
+			tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, DEFAULT_MTYPE, MTYPE_NC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, APE1_MTYPE, MTYPE_NC);
+			tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, ALIGNMENT_MODE,
+					    SH_MEM_ALIGNMENT_MODE_UNALIGNED);
+			WREG32(mmSH_MEM_CONFIG, tmp);
+		}
+
+		WREG32(mmSH_MEM_APE1_BASE, 1);
+		WREG32(mmSH_MEM_APE1_LIMIT, 0);
+		WREG32(mmSH_MEM_BASES, 0);
+	}
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	gfx_v8_0_init_compute_vmid(adev);
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	/*
+	 * making sure that the following register writes will be broadcasted
+	 * to all the shaders
+	 */
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+
+	WREG32(mmPA_SC_FIFO_SIZE,
+		   (adev->gfx.config.sc_prim_fifo_size_frontend <<
+			PA_SC_FIFO_SIZE__SC_FRONTEND_PRIM_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_prim_fifo_size_backend <<
+			PA_SC_FIFO_SIZE__SC_BACKEND_PRIM_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_hiz_tile_fifo_size <<
+			PA_SC_FIFO_SIZE__SC_HIZ_TILE_FIFO_SIZE__SHIFT) |
+		   (adev->gfx.config.sc_earlyz_tile_fifo_size <<
+			PA_SC_FIFO_SIZE__SC_EARLYZ_TILE_FIFO_SIZE__SHIFT));
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+}
+
+static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
+{
+	u32 i, j, k;
+	u32 mask;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			gfx_v8_0_select_se_sh(adev, i, j);
+			for (k = 0; k < adev->usec_timeout; k++) {
+				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
+					break;
+				udelay(1);
+			}
+		}
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__GC_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC0_MASTER_BUSY_MASK |
+		RLC_SERDES_NONCU_MASTER_BUSY__TC1_MASTER_BUSY_MASK;
+	for (k = 0; k < adev->usec_timeout; k++) {
+		if ((RREG32(mmRLC_SERDES_NONCU_MASTER_BUSY) & mask) == 0)
+			break;
+		udelay(1);
+	}
+}
+
+static void gfx_v8_0_enable_gui_idle_interrupt(struct amdgpu_device *adev,
+					       bool enable)
+{
+	u32 tmp = RREG32(mmCP_INT_CNTL_RING0);
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, 1);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_BUSY_INT_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CNTX_EMPTY_INT_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, CMP_BUSY_INT_ENABLE, 0);
+		tmp = REG_SET_FIELD(tmp, CP_INT_CNTL_RING0, GFX_IDLE_INT_ENABLE, 0);
+	}
+	WREG32(mmCP_INT_CNTL_RING0, tmp);
+}
+
+void gfx_v8_0_rlc_stop(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmRLC_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 0);
+	WREG32(mmRLC_CNTL, tmp);
+
+	gfx_v8_0_enable_gui_idle_interrupt(adev, false);
+
+	gfx_v8_0_wait_for_rlc_serdes(adev);
+}
+
+static void gfx_v8_0_rlc_reset(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmGRBM_SOFT_RESET);
+
+	tmp = REG_SET_FIELD(tmp, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+	WREG32(mmGRBM_SOFT_RESET, tmp);
+	udelay(50);
+	tmp = REG_SET_FIELD(tmp, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0);
+	WREG32(mmGRBM_SOFT_RESET, tmp);
+	udelay(50);
+}
+
+static void gfx_v8_0_rlc_start(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32(mmRLC_CNTL);
+
+	tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 1);
+	WREG32(mmRLC_CNTL, tmp);
+
+	/* carrizo do enable cp interrupt after cp inited */
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v8_0_enable_gui_idle_interrupt(adev, true);
+
+	udelay(50);
+}
+
+static int gfx_v8_0_rlc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct rlc_firmware_header_v2_0 *hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.rlc_fw)
+		return -EINVAL;
+
+	hdr = (const struct rlc_firmware_header_v2_0 *)adev->gfx.rlc_fw->data;
+	amdgpu_ucode_print_rlc_hdr(&hdr->header);
+
+	fw_data = (const __le32 *)(adev->gfx.rlc_fw->data +
+			   le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+
+	WREG32(mmRLC_GPM_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmRLC_GPM_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmRLC_GPM_UCODE_ADDR, adev->gfx.rlc_fw_version);
+
+	return 0;
+}
+
+static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	gfx_v8_0_rlc_stop(adev);
+
+	/* disable CG */
+	WREG32(mmRLC_CGCG_CGLS_CTRL, 0);
+
+	/* disable PG */
+	WREG32(mmRLC_PG_CNTL, 0);
+
+	gfx_v8_0_rlc_reset(adev);
+
+	if (!adev->firmware.smu_load) {
+		/* legacy rlc firmware loading */
+		r = gfx_v8_0_rlc_load_microcode(adev);
+		if (r)
+			return r;
+	} else {
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_RLC_G);
+		if (r)
+			return -EINVAL;
+	}
+
+	gfx_v8_0_rlc_start(adev);
+
+	return 0;
+}
+
+static void gfx_v8_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+	u32 tmp = RREG32(mmCP_ME_CNTL);
+
+	if (enable) {
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 0);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 0);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 0);
+	} else {
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 1);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 1);
+		tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 1);
+		for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+			adev->gfx.gfx_ring[i].ready = false;
+	}
+	WREG32(mmCP_ME_CNTL, tmp);
+	udelay(50);
+}
+
+static int gfx_v8_0_cp_gfx_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *pfp_hdr;
+	const struct gfx_firmware_header_v1_0 *ce_hdr;
+	const struct gfx_firmware_header_v1_0 *me_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.me_fw || !adev->gfx.pfp_fw || !adev->gfx.ce_fw)
+		return -EINVAL;
+
+	pfp_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.pfp_fw->data;
+	ce_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.ce_fw->data;
+	me_hdr = (const struct gfx_firmware_header_v1_0 *)
+		adev->gfx.me_fw->data;
+
+	amdgpu_ucode_print_gfx_hdr(&pfp_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&ce_hdr->header);
+	amdgpu_ucode_print_gfx_hdr(&me_hdr->header);
+
+	gfx_v8_0_cp_gfx_enable(adev, false);
+
+	/* PFP */
+	fw_data = (const __le32 *)
+		(adev->gfx.pfp_fw->data +
+		 le32_to_cpu(pfp_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(pfp_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_PFP_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_PFP_UCODE_ADDR, adev->gfx.pfp_fw_version);
+
+	/* CE */
+	fw_data = (const __le32 *)
+		(adev->gfx.ce_fw->data +
+		 le32_to_cpu(ce_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(ce_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_CE_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_CE_UCODE_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_CE_UCODE_ADDR, adev->gfx.ce_fw_version);
+
+	/* ME */
+	fw_data = (const __le32 *)
+		(adev->gfx.me_fw->data +
+		 le32_to_cpu(me_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(me_hdr->header.ucode_size_bytes) / 4;
+	WREG32(mmCP_ME_RAM_WADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_ME_RAM_DATA, le32_to_cpup(fw_data++));
+	WREG32(mmCP_ME_RAM_WADDR, adev->gfx.me_fw_version);
+
+	return 0;
+}
+
+static u32 gfx_v8_0_get_csb_size(struct amdgpu_device *adev)
+{
+	u32 count = 0;
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+
+	/* begin clear state */
+	count += 2;
+	/* context control state */
+	count += 3;
+
+	for (sect = vi_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT)
+				count += 2 + ext->reg_count;
+			else
+				return 0;
+		}
+	}
+	/* pa_sc_raster_config/pa_sc_raster_config1 */
+	count += 4;
+	/* end clear state */
+	count += 2;
+	/* clear state */
+	count += 2;
+
+	return count;
+}
+
+static int gfx_v8_0_cp_gfx_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
+	const struct cs_section_def *sect = NULL;
+	const struct cs_extent_def *ext = NULL;
+	int r, i;
+
+	/* init the CP */
+	WREG32(mmCP_MAX_CONTEXT, adev->gfx.config.max_hw_contexts - 1);
+	WREG32(mmCP_ENDIAN_SWAP, 0);
+	WREG32(mmCP_DEVICE_ID, 1);
+
+	gfx_v8_0_cp_gfx_enable(adev, true);
+
+	r = amdgpu_ring_lock(ring, gfx_v8_0_get_csb_size(adev) + 4);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
+		return r;
+	}
+
+	/* clear state buffer */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_BEGIN_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, 0x80000000);
+	amdgpu_ring_write(ring, 0x80000000);
+
+	for (sect = vi_cs_data; sect->section != NULL; ++sect) {
+		for (ext = sect->section; ext->extent != NULL; ++ext) {
+			if (sect->id == SECT_CONTEXT) {
+				amdgpu_ring_write(ring,
+				       PACKET3(PACKET3_SET_CONTEXT_REG,
+					       ext->reg_count));
+				amdgpu_ring_write(ring,
+				       ext->reg_index - PACKET3_SET_CONTEXT_REG_START);
+				for (i = 0; i < ext->reg_count; i++)
+					amdgpu_ring_write(ring, ext->extent[i]);
+			}
+		}
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONTEXT_REG, 2));
+	amdgpu_ring_write(ring, mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START);
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		amdgpu_ring_write(ring, 0x16000012);
+		amdgpu_ring_write(ring, 0x0000002A);
+		break;
+	case CHIP_FIJI:
+		amdgpu_ring_write(ring, 0x3a00161a);
+		amdgpu_ring_write(ring, 0x0000002e);
+		break;
+	case CHIP_TOPAZ:
+	case CHIP_CARRIZO:
+		amdgpu_ring_write(ring, 0x00000002);
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
+	case CHIP_STONEY:
+		amdgpu_ring_write(ring, 0x00000000);
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
+	default:
+		BUG();
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_PREAMBLE_CNTL, 0));
+	amdgpu_ring_write(ring, PACKET3_PREAMBLE_END_CLEAR_STATE);
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CLEAR_STATE, 0));
+	amdgpu_ring_write(ring, 0);
+
+	/* init the CE partitions */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_BASE, 2));
+	amdgpu_ring_write(ring, PACKET3_BASE_INDEX(CE_PARTITION_BASE));
+	amdgpu_ring_write(ring, 0x8000);
+	amdgpu_ring_write(ring, 0x8000);
+
+	amdgpu_ring_unlock_commit(ring);
+
+	return 0;
+}
+
+static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 tmp;
+	u32 rb_bufsz;
+	u64 rb_addr, rptr_addr;
+	int r;
+
+	/* Set the write pointer delay */
+	WREG32(mmCP_RB_WPTR_DELAY, 0);
+
+	/* set the RB to use vmid 0 */
+	WREG32(mmCP_RB_VMID, 0);
+
+	/* Set ring buffer size */
+	ring = &adev->gfx.gfx_ring[0];
+	rb_bufsz = order_base_2(ring->ring_size / 8);
+	tmp = REG_SET_FIELD(0, CP_RB0_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, RB_BLKSZ, rb_bufsz - 2);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, MTYPE, 3);
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, MIN_IB_AVAILSZ, 1);
+#ifdef __BIG_ENDIAN
+	tmp = REG_SET_FIELD(tmp, CP_RB0_CNTL, BUF_SWAP, 1);
+#endif
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmCP_RB0_CNTL, tmp | CP_RB0_CNTL__RB_RPTR_WR_ENA_MASK);
+	ring->wptr = 0;
+	WREG32(mmCP_RB0_WPTR, ring->wptr);
+
+	/* set the wb address wether it's enabled or not */
+	rptr_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+	WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr));
+	WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF);
+
+	mdelay(1);
+	WREG32(mmCP_RB0_CNTL, tmp);
+
+	rb_addr = ring->gpu_addr >> 8;
+	WREG32(mmCP_RB0_BASE, rb_addr);
+	WREG32(mmCP_RB0_BASE_HI, upper_32_bits(rb_addr));
+
+	/* no gfx doorbells on iceland */
+	if (adev->asic_type != CHIP_TOPAZ) {
+		tmp = RREG32(mmCP_RB_DOORBELL_CONTROL);
+		if (ring->use_doorbell) {
+			tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+					    DOORBELL_OFFSET, ring->doorbell_index);
+			tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+					    DOORBELL_EN, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL,
+					    DOORBELL_EN, 0);
+		}
+		WREG32(mmCP_RB_DOORBELL_CONTROL, tmp);
+
+		if (adev->asic_type == CHIP_TONGA) {
+			tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
+					    DOORBELL_RANGE_LOWER,
+					    AMDGPU_DOORBELL_GFX_RING0);
+			WREG32(mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
+
+			WREG32(mmCP_RB_DOORBELL_RANGE_UPPER,
+			       CP_RB_DOORBELL_RANGE_UPPER__DOORBELL_RANGE_UPPER_MASK);
+		}
+
+	}
+
+	/* start the ring */
+	gfx_v8_0_cp_gfx_start(adev);
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	return 0;
+}
+
+static void gfx_v8_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
+{
+	int i;
+
+	if (enable) {
+		WREG32(mmCP_MEC_CNTL, 0);
+	} else {
+		WREG32(mmCP_MEC_CNTL, (CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK));
+		for (i = 0; i < adev->gfx.num_compute_rings; i++)
+			adev->gfx.compute_ring[i].ready = false;
+	}
+	udelay(50);
+}
+
+static int gfx_v8_0_cp_compute_start(struct amdgpu_device *adev)
+{
+	gfx_v8_0_cp_compute_enable(adev, true);
+
+	return 0;
+}
+
+static int gfx_v8_0_cp_compute_load_microcode(struct amdgpu_device *adev)
+{
+	const struct gfx_firmware_header_v1_0 *mec_hdr;
+	const __le32 *fw_data;
+	unsigned i, fw_size;
+
+	if (!adev->gfx.mec_fw)
+		return -EINVAL;
+
+	gfx_v8_0_cp_compute_enable(adev, false);
+
+	mec_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+	amdgpu_ucode_print_gfx_hdr(&mec_hdr->header);
+
+	fw_data = (const __le32 *)
+		(adev->gfx.mec_fw->data +
+		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
+	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
+
+	/* MEC1 */
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, 0);
+	for (i = 0; i < fw_size; i++)
+		WREG32(mmCP_MEC_ME1_UCODE_DATA, le32_to_cpup(fw_data+i));
+	WREG32(mmCP_MEC_ME1_UCODE_ADDR, adev->gfx.mec_fw_version);
+
+	/* Loading MEC2 firmware is only necessary if MEC2 should run different microcode than MEC1. */
+	if (adev->gfx.mec2_fw) {
+		const struct gfx_firmware_header_v1_0 *mec2_hdr;
+
+		mec2_hdr = (const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+		amdgpu_ucode_print_gfx_hdr(&mec2_hdr->header);
+
+		fw_data = (const __le32 *)
+			(adev->gfx.mec2_fw->data +
+			 le32_to_cpu(mec2_hdr->header.ucode_array_offset_bytes));
+		fw_size = le32_to_cpu(mec2_hdr->header.ucode_size_bytes) / 4;
+
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, 0);
+		for (i = 0; i < fw_size; i++)
+			WREG32(mmCP_MEC_ME2_UCODE_DATA, le32_to_cpup(fw_data+i));
+		WREG32(mmCP_MEC_ME2_UCODE_ADDR, adev->gfx.mec2_fw_version);
+	}
+
+	return 0;
+}
+
+struct vi_mqd {
+	uint32_t header;  /* ordinal0 */
+	uint32_t compute_dispatch_initiator;  /* ordinal1 */
+	uint32_t compute_dim_x;  /* ordinal2 */
+	uint32_t compute_dim_y;  /* ordinal3 */
+	uint32_t compute_dim_z;  /* ordinal4 */
+	uint32_t compute_start_x;  /* ordinal5 */
+	uint32_t compute_start_y;  /* ordinal6 */
+	uint32_t compute_start_z;  /* ordinal7 */
+	uint32_t compute_num_thread_x;  /* ordinal8 */
+	uint32_t compute_num_thread_y;  /* ordinal9 */
+	uint32_t compute_num_thread_z;  /* ordinal10 */
+	uint32_t compute_pipelinestat_enable;  /* ordinal11 */
+	uint32_t compute_perfcount_enable;  /* ordinal12 */
+	uint32_t compute_pgm_lo;  /* ordinal13 */
+	uint32_t compute_pgm_hi;  /* ordinal14 */
+	uint32_t compute_tba_lo;  /* ordinal15 */
+	uint32_t compute_tba_hi;  /* ordinal16 */
+	uint32_t compute_tma_lo;  /* ordinal17 */
+	uint32_t compute_tma_hi;  /* ordinal18 */
+	uint32_t compute_pgm_rsrc1;  /* ordinal19 */
+	uint32_t compute_pgm_rsrc2;  /* ordinal20 */
+	uint32_t compute_vmid;  /* ordinal21 */
+	uint32_t compute_resource_limits;  /* ordinal22 */
+	uint32_t compute_static_thread_mgmt_se0;  /* ordinal23 */
+	uint32_t compute_static_thread_mgmt_se1;  /* ordinal24 */
+	uint32_t compute_tmpring_size;  /* ordinal25 */
+	uint32_t compute_static_thread_mgmt_se2;  /* ordinal26 */
+	uint32_t compute_static_thread_mgmt_se3;  /* ordinal27 */
+	uint32_t compute_restart_x;  /* ordinal28 */
+	uint32_t compute_restart_y;  /* ordinal29 */
+	uint32_t compute_restart_z;  /* ordinal30 */
+	uint32_t compute_thread_trace_enable;  /* ordinal31 */
+	uint32_t compute_misc_reserved;  /* ordinal32 */
+	uint32_t compute_dispatch_id;  /* ordinal33 */
+	uint32_t compute_threadgroup_id;  /* ordinal34 */
+	uint32_t compute_relaunch;  /* ordinal35 */
+	uint32_t compute_wave_restore_addr_lo;  /* ordinal36 */
+	uint32_t compute_wave_restore_addr_hi;  /* ordinal37 */
+	uint32_t compute_wave_restore_control;  /* ordinal38 */
+	uint32_t reserved9;  /* ordinal39 */
+	uint32_t reserved10;  /* ordinal40 */
+	uint32_t reserved11;  /* ordinal41 */
+	uint32_t reserved12;  /* ordinal42 */
+	uint32_t reserved13;  /* ordinal43 */
+	uint32_t reserved14;  /* ordinal44 */
+	uint32_t reserved15;  /* ordinal45 */
+	uint32_t reserved16;  /* ordinal46 */
+	uint32_t reserved17;  /* ordinal47 */
+	uint32_t reserved18;  /* ordinal48 */
+	uint32_t reserved19;  /* ordinal49 */
+	uint32_t reserved20;  /* ordinal50 */
+	uint32_t reserved21;  /* ordinal51 */
+	uint32_t reserved22;  /* ordinal52 */
+	uint32_t reserved23;  /* ordinal53 */
+	uint32_t reserved24;  /* ordinal54 */
+	uint32_t reserved25;  /* ordinal55 */
+	uint32_t reserved26;  /* ordinal56 */
+	uint32_t reserved27;  /* ordinal57 */
+	uint32_t reserved28;  /* ordinal58 */
+	uint32_t reserved29;  /* ordinal59 */
+	uint32_t reserved30;  /* ordinal60 */
+	uint32_t reserved31;  /* ordinal61 */
+	uint32_t reserved32;  /* ordinal62 */
+	uint32_t reserved33;  /* ordinal63 */
+	uint32_t reserved34;  /* ordinal64 */
+	uint32_t compute_user_data_0;  /* ordinal65 */
+	uint32_t compute_user_data_1;  /* ordinal66 */
+	uint32_t compute_user_data_2;  /* ordinal67 */
+	uint32_t compute_user_data_3;  /* ordinal68 */
+	uint32_t compute_user_data_4;  /* ordinal69 */
+	uint32_t compute_user_data_5;  /* ordinal70 */
+	uint32_t compute_user_data_6;  /* ordinal71 */
+	uint32_t compute_user_data_7;  /* ordinal72 */
+	uint32_t compute_user_data_8;  /* ordinal73 */
+	uint32_t compute_user_data_9;  /* ordinal74 */
+	uint32_t compute_user_data_10;  /* ordinal75 */
+	uint32_t compute_user_data_11;  /* ordinal76 */
+	uint32_t compute_user_data_12;  /* ordinal77 */
+	uint32_t compute_user_data_13;  /* ordinal78 */
+	uint32_t compute_user_data_14;  /* ordinal79 */
+	uint32_t compute_user_data_15;  /* ordinal80 */
+	uint32_t cp_compute_csinvoc_count_lo;  /* ordinal81 */
+	uint32_t cp_compute_csinvoc_count_hi;  /* ordinal82 */
+	uint32_t reserved35;  /* ordinal83 */
+	uint32_t reserved36;  /* ordinal84 */
+	uint32_t reserved37;  /* ordinal85 */
+	uint32_t cp_mqd_query_time_lo;  /* ordinal86 */
+	uint32_t cp_mqd_query_time_hi;  /* ordinal87 */
+	uint32_t cp_mqd_connect_start_time_lo;  /* ordinal88 */
+	uint32_t cp_mqd_connect_start_time_hi;  /* ordinal89 */
+	uint32_t cp_mqd_connect_end_time_lo;  /* ordinal90 */
+	uint32_t cp_mqd_connect_end_time_hi;  /* ordinal91 */
+	uint32_t cp_mqd_connect_end_wf_count;  /* ordinal92 */
+	uint32_t cp_mqd_connect_end_pq_rptr;  /* ordinal93 */
+	uint32_t cp_mqd_connect_end_pq_wptr;  /* ordinal94 */
+	uint32_t cp_mqd_connect_end_ib_rptr;  /* ordinal95 */
+	uint32_t reserved38;  /* ordinal96 */
+	uint32_t reserved39;  /* ordinal97 */
+	uint32_t cp_mqd_save_start_time_lo;  /* ordinal98 */
+	uint32_t cp_mqd_save_start_time_hi;  /* ordinal99 */
+	uint32_t cp_mqd_save_end_time_lo;  /* ordinal100 */
+	uint32_t cp_mqd_save_end_time_hi;  /* ordinal101 */
+	uint32_t cp_mqd_restore_start_time_lo;  /* ordinal102 */
+	uint32_t cp_mqd_restore_start_time_hi;  /* ordinal103 */
+	uint32_t cp_mqd_restore_end_time_lo;  /* ordinal104 */
+	uint32_t cp_mqd_restore_end_time_hi;  /* ordinal105 */
+	uint32_t reserved40;  /* ordinal106 */
+	uint32_t reserved41;  /* ordinal107 */
+	uint32_t gds_cs_ctxsw_cnt0;  /* ordinal108 */
+	uint32_t gds_cs_ctxsw_cnt1;  /* ordinal109 */
+	uint32_t gds_cs_ctxsw_cnt2;  /* ordinal110 */
+	uint32_t gds_cs_ctxsw_cnt3;  /* ordinal111 */
+	uint32_t reserved42;  /* ordinal112 */
+	uint32_t reserved43;  /* ordinal113 */
+	uint32_t cp_pq_exe_status_lo;  /* ordinal114 */
+	uint32_t cp_pq_exe_status_hi;  /* ordinal115 */
+	uint32_t cp_packet_id_lo;  /* ordinal116 */
+	uint32_t cp_packet_id_hi;  /* ordinal117 */
+	uint32_t cp_packet_exe_status_lo;  /* ordinal118 */
+	uint32_t cp_packet_exe_status_hi;  /* ordinal119 */
+	uint32_t gds_save_base_addr_lo;  /* ordinal120 */
+	uint32_t gds_save_base_addr_hi;  /* ordinal121 */
+	uint32_t gds_save_mask_lo;  /* ordinal122 */
+	uint32_t gds_save_mask_hi;  /* ordinal123 */
+	uint32_t ctx_save_base_addr_lo;  /* ordinal124 */
+	uint32_t ctx_save_base_addr_hi;  /* ordinal125 */
+	uint32_t reserved44;  /* ordinal126 */
+	uint32_t reserved45;  /* ordinal127 */
+	uint32_t cp_mqd_base_addr_lo;  /* ordinal128 */
+	uint32_t cp_mqd_base_addr_hi;  /* ordinal129 */
+	uint32_t cp_hqd_active;  /* ordinal130 */
+	uint32_t cp_hqd_vmid;  /* ordinal131 */
+	uint32_t cp_hqd_persistent_state;  /* ordinal132 */
+	uint32_t cp_hqd_pipe_priority;  /* ordinal133 */
+	uint32_t cp_hqd_queue_priority;  /* ordinal134 */
+	uint32_t cp_hqd_quantum;  /* ordinal135 */
+	uint32_t cp_hqd_pq_base_lo;  /* ordinal136 */
+	uint32_t cp_hqd_pq_base_hi;  /* ordinal137 */
+	uint32_t cp_hqd_pq_rptr;  /* ordinal138 */
+	uint32_t cp_hqd_pq_rptr_report_addr_lo;  /* ordinal139 */
+	uint32_t cp_hqd_pq_rptr_report_addr_hi;  /* ordinal140 */
+	uint32_t cp_hqd_pq_wptr_poll_addr;  /* ordinal141 */
+	uint32_t cp_hqd_pq_wptr_poll_addr_hi;  /* ordinal142 */
+	uint32_t cp_hqd_pq_doorbell_control;  /* ordinal143 */
+	uint32_t cp_hqd_pq_wptr;  /* ordinal144 */
+	uint32_t cp_hqd_pq_control;  /* ordinal145 */
+	uint32_t cp_hqd_ib_base_addr_lo;  /* ordinal146 */
+	uint32_t cp_hqd_ib_base_addr_hi;  /* ordinal147 */
+	uint32_t cp_hqd_ib_rptr;  /* ordinal148 */
+	uint32_t cp_hqd_ib_control;  /* ordinal149 */
+	uint32_t cp_hqd_iq_timer;  /* ordinal150 */
+	uint32_t cp_hqd_iq_rptr;  /* ordinal151 */
+	uint32_t cp_hqd_dequeue_request;  /* ordinal152 */
+	uint32_t cp_hqd_dma_offload;  /* ordinal153 */
+	uint32_t cp_hqd_sema_cmd;  /* ordinal154 */
+	uint32_t cp_hqd_msg_type;  /* ordinal155 */
+	uint32_t cp_hqd_atomic0_preop_lo;  /* ordinal156 */
+	uint32_t cp_hqd_atomic0_preop_hi;  /* ordinal157 */
+	uint32_t cp_hqd_atomic1_preop_lo;  /* ordinal158 */
+	uint32_t cp_hqd_atomic1_preop_hi;  /* ordinal159 */
+	uint32_t cp_hqd_hq_status0;  /* ordinal160 */
+	uint32_t cp_hqd_hq_control0;  /* ordinal161 */
+	uint32_t cp_mqd_control;  /* ordinal162 */
+	uint32_t cp_hqd_hq_status1;  /* ordinal163 */
+	uint32_t cp_hqd_hq_control1;  /* ordinal164 */
+	uint32_t cp_hqd_eop_base_addr_lo;  /* ordinal165 */
+	uint32_t cp_hqd_eop_base_addr_hi;  /* ordinal166 */
+	uint32_t cp_hqd_eop_control;  /* ordinal167 */
+	uint32_t cp_hqd_eop_rptr;  /* ordinal168 */
+	uint32_t cp_hqd_eop_wptr;  /* ordinal169 */
+	uint32_t cp_hqd_eop_done_events;  /* ordinal170 */
+	uint32_t cp_hqd_ctx_save_base_addr_lo;  /* ordinal171 */
+	uint32_t cp_hqd_ctx_save_base_addr_hi;  /* ordinal172 */
+	uint32_t cp_hqd_ctx_save_control;  /* ordinal173 */
+	uint32_t cp_hqd_cntl_stack_offset;  /* ordinal174 */
+	uint32_t cp_hqd_cntl_stack_size;  /* ordinal175 */
+	uint32_t cp_hqd_wg_state_offset;  /* ordinal176 */
+	uint32_t cp_hqd_ctx_save_size;  /* ordinal177 */
+	uint32_t cp_hqd_gds_resource_state;  /* ordinal178 */
+	uint32_t cp_hqd_error;  /* ordinal179 */
+	uint32_t cp_hqd_eop_wptr_mem;  /* ordinal180 */
+	uint32_t cp_hqd_eop_dones;  /* ordinal181 */
+	uint32_t reserved46;  /* ordinal182 */
+	uint32_t reserved47;  /* ordinal183 */
+	uint32_t reserved48;  /* ordinal184 */
+	uint32_t reserved49;  /* ordinal185 */
+	uint32_t reserved50;  /* ordinal186 */
+	uint32_t reserved51;  /* ordinal187 */
+	uint32_t reserved52;  /* ordinal188 */
+	uint32_t reserved53;  /* ordinal189 */
+	uint32_t reserved54;  /* ordinal190 */
+	uint32_t reserved55;  /* ordinal191 */
+	uint32_t iqtimer_pkt_header;  /* ordinal192 */
+	uint32_t iqtimer_pkt_dw0;  /* ordinal193 */
+	uint32_t iqtimer_pkt_dw1;  /* ordinal194 */
+	uint32_t iqtimer_pkt_dw2;  /* ordinal195 */
+	uint32_t iqtimer_pkt_dw3;  /* ordinal196 */
+	uint32_t iqtimer_pkt_dw4;  /* ordinal197 */
+	uint32_t iqtimer_pkt_dw5;  /* ordinal198 */
+	uint32_t iqtimer_pkt_dw6;  /* ordinal199 */
+	uint32_t iqtimer_pkt_dw7;  /* ordinal200 */
+	uint32_t iqtimer_pkt_dw8;  /* ordinal201 */
+	uint32_t iqtimer_pkt_dw9;  /* ordinal202 */
+	uint32_t iqtimer_pkt_dw10;  /* ordinal203 */
+	uint32_t iqtimer_pkt_dw11;  /* ordinal204 */
+	uint32_t iqtimer_pkt_dw12;  /* ordinal205 */
+	uint32_t iqtimer_pkt_dw13;  /* ordinal206 */
+	uint32_t iqtimer_pkt_dw14;  /* ordinal207 */
+	uint32_t iqtimer_pkt_dw15;  /* ordinal208 */
+	uint32_t iqtimer_pkt_dw16;  /* ordinal209 */
+	uint32_t iqtimer_pkt_dw17;  /* ordinal210 */
+	uint32_t iqtimer_pkt_dw18;  /* ordinal211 */
+	uint32_t iqtimer_pkt_dw19;  /* ordinal212 */
+	uint32_t iqtimer_pkt_dw20;  /* ordinal213 */
+	uint32_t iqtimer_pkt_dw21;  /* ordinal214 */
+	uint32_t iqtimer_pkt_dw22;  /* ordinal215 */
+	uint32_t iqtimer_pkt_dw23;  /* ordinal216 */
+	uint32_t iqtimer_pkt_dw24;  /* ordinal217 */
+	uint32_t iqtimer_pkt_dw25;  /* ordinal218 */
+	uint32_t iqtimer_pkt_dw26;  /* ordinal219 */
+	uint32_t iqtimer_pkt_dw27;  /* ordinal220 */
+	uint32_t iqtimer_pkt_dw28;  /* ordinal221 */
+	uint32_t iqtimer_pkt_dw29;  /* ordinal222 */
+	uint32_t iqtimer_pkt_dw30;  /* ordinal223 */
+	uint32_t iqtimer_pkt_dw31;  /* ordinal224 */
+	uint32_t reserved56;  /* ordinal225 */
+	uint32_t reserved57;  /* ordinal226 */
+	uint32_t reserved58;  /* ordinal227 */
+	uint32_t set_resources_header;  /* ordinal228 */
+	uint32_t set_resources_dw1;  /* ordinal229 */
+	uint32_t set_resources_dw2;  /* ordinal230 */
+	uint32_t set_resources_dw3;  /* ordinal231 */
+	uint32_t set_resources_dw4;  /* ordinal232 */
+	uint32_t set_resources_dw5;  /* ordinal233 */
+	uint32_t set_resources_dw6;  /* ordinal234 */
+	uint32_t set_resources_dw7;  /* ordinal235 */
+	uint32_t reserved59;  /* ordinal236 */
+	uint32_t reserved60;  /* ordinal237 */
+	uint32_t reserved61;  /* ordinal238 */
+	uint32_t reserved62;  /* ordinal239 */
+	uint32_t reserved63;  /* ordinal240 */
+	uint32_t reserved64;  /* ordinal241 */
+	uint32_t reserved65;  /* ordinal242 */
+	uint32_t reserved66;  /* ordinal243 */
+	uint32_t reserved67;  /* ordinal244 */
+	uint32_t reserved68;  /* ordinal245 */
+	uint32_t reserved69;  /* ordinal246 */
+	uint32_t reserved70;  /* ordinal247 */
+	uint32_t reserved71;  /* ordinal248 */
+	uint32_t reserved72;  /* ordinal249 */
+	uint32_t reserved73;  /* ordinal250 */
+	uint32_t reserved74;  /* ordinal251 */
+	uint32_t reserved75;  /* ordinal252 */
+	uint32_t reserved76;  /* ordinal253 */
+	uint32_t reserved77;  /* ordinal254 */
+	uint32_t reserved78;  /* ordinal255 */
+
+	uint32_t reserved_t[256]; /* Reserve 256 dword buffer used by ucode */
+};
+
+static void gfx_v8_0_cp_compute_fini(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		if (ring->mqd_obj) {
+			r = amdgpu_bo_reserve(ring->mqd_obj, false);
+			if (unlikely(r != 0))
+				dev_warn(adev->dev, "(%d) reserve MQD bo failed\n", r);
+
+			amdgpu_bo_unpin(ring->mqd_obj);
+			amdgpu_bo_unreserve(ring->mqd_obj);
+
+			amdgpu_bo_unref(&ring->mqd_obj);
+			ring->mqd_obj = NULL;
+		}
+	}
+}
+
+static int gfx_v8_0_cp_compute_resume(struct amdgpu_device *adev)
+{
+	int r, i, j;
+	u32 tmp;
+	bool use_doorbell = true;
+	u64 hqd_gpu_addr;
+	u64 mqd_gpu_addr;
+	u64 eop_gpu_addr;
+	u64 wb_gpu_addr;
+	u32 *buf;
+	struct vi_mqd *mqd;
+
+	/* init the pipes */
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < (adev->gfx.mec.num_pipe * adev->gfx.mec.num_mec); i++) {
+		int me = (i < 4) ? 1 : 2;
+		int pipe = (i < 4) ? i : (i - 4);
+
+		eop_gpu_addr = adev->gfx.mec.hpd_eop_gpu_addr + (i * MEC_HPD_SIZE);
+		eop_gpu_addr >>= 8;
+
+		vi_srbm_select(adev, me, pipe, 0, 0);
+
+		/* write the EOP addr */
+		WREG32(mmCP_HQD_EOP_BASE_ADDR, eop_gpu_addr);
+		WREG32(mmCP_HQD_EOP_BASE_ADDR_HI, upper_32_bits(eop_gpu_addr));
+
+		/* set the VMID assigned */
+		WREG32(mmCP_HQD_VMID, 0);
+
+		/* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */
+		tmp = RREG32(mmCP_HQD_EOP_CONTROL);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE,
+				    (order_base_2(MEC_HPD_SIZE / 4) - 1));
+		WREG32(mmCP_HQD_EOP_CONTROL, tmp);
+	}
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+
+	/* init the queues.  Just two for now. */
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		if (ring->mqd_obj == NULL) {
+			r = amdgpu_bo_create(adev,
+					     sizeof(struct vi_mqd),
+					     PAGE_SIZE, true,
+					     AMDGPU_GEM_DOMAIN_GTT, 0, NULL,
+					     NULL, &ring->mqd_obj);
+			if (r) {
+				dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
+				return r;
+			}
+		}
+
+		r = amdgpu_bo_reserve(ring->mqd_obj, false);
+		if (unlikely(r != 0)) {
+			gfx_v8_0_cp_compute_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_pin(ring->mqd_obj, AMDGPU_GEM_DOMAIN_GTT,
+				  &mqd_gpu_addr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) pin MQD bo failed\n", r);
+			gfx_v8_0_cp_compute_fini(adev);
+			return r;
+		}
+		r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&buf);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map MQD bo failed\n", r);
+			gfx_v8_0_cp_compute_fini(adev);
+			return r;
+		}
+
+		/* init the mqd struct */
+		memset(buf, 0, sizeof(struct vi_mqd));
+
+		mqd = (struct vi_mqd *)buf;
+		mqd->header = 0xC0310800;
+		mqd->compute_pipelinestat_enable = 0x00000001;
+		mqd->compute_static_thread_mgmt_se0 = 0xffffffff;
+		mqd->compute_static_thread_mgmt_se1 = 0xffffffff;
+		mqd->compute_static_thread_mgmt_se2 = 0xffffffff;
+		mqd->compute_static_thread_mgmt_se3 = 0xffffffff;
+		mqd->compute_misc_reserved = 0x00000003;
+
+		mutex_lock(&adev->srbm_mutex);
+		vi_srbm_select(adev, ring->me,
+			       ring->pipe,
+			       ring->queue, 0);
+
+		/* disable wptr polling */
+		tmp = RREG32(mmCP_PQ_WPTR_POLL_CNTL);
+		tmp = REG_SET_FIELD(tmp, CP_PQ_WPTR_POLL_CNTL, EN, 0);
+		WREG32(mmCP_PQ_WPTR_POLL_CNTL, tmp);
+
+		mqd->cp_hqd_eop_base_addr_lo =
+			RREG32(mmCP_HQD_EOP_BASE_ADDR);
+		mqd->cp_hqd_eop_base_addr_hi =
+			RREG32(mmCP_HQD_EOP_BASE_ADDR_HI);
+
+		/* enable doorbell? */
+		tmp = RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+		if (use_doorbell) {
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+		} else {
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 0);
+		}
+		WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, tmp);
+		mqd->cp_hqd_pq_doorbell_control = tmp;
+
+		/* disable the queue if it's active */
+		mqd->cp_hqd_dequeue_request = 0;
+		mqd->cp_hqd_pq_rptr = 0;
+		mqd->cp_hqd_pq_wptr= 0;
+		if (RREG32(mmCP_HQD_ACTIVE) & 1) {
+			WREG32(mmCP_HQD_DEQUEUE_REQUEST, 1);
+			for (j = 0; j < adev->usec_timeout; j++) {
+				if (!(RREG32(mmCP_HQD_ACTIVE) & 1))
+					break;
+				udelay(1);
+			}
+			WREG32(mmCP_HQD_DEQUEUE_REQUEST, mqd->cp_hqd_dequeue_request);
+			WREG32(mmCP_HQD_PQ_RPTR, mqd->cp_hqd_pq_rptr);
+			WREG32(mmCP_HQD_PQ_WPTR, mqd->cp_hqd_pq_wptr);
+		}
+
+		/* set the pointer to the MQD */
+		mqd->cp_mqd_base_addr_lo = mqd_gpu_addr & 0xfffffffc;
+		mqd->cp_mqd_base_addr_hi = upper_32_bits(mqd_gpu_addr);
+		WREG32(mmCP_MQD_BASE_ADDR, mqd->cp_mqd_base_addr_lo);
+		WREG32(mmCP_MQD_BASE_ADDR_HI, mqd->cp_mqd_base_addr_hi);
+
+		/* set MQD vmid to 0 */
+		tmp = RREG32(mmCP_MQD_CONTROL);
+		tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0);
+		WREG32(mmCP_MQD_CONTROL, tmp);
+		mqd->cp_mqd_control = tmp;
+
+		/* set the pointer to the HQD, this is similar CP_RB0_BASE/_HI */
+		hqd_gpu_addr = ring->gpu_addr >> 8;
+		mqd->cp_hqd_pq_base_lo = hqd_gpu_addr;
+		mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr);
+		WREG32(mmCP_HQD_PQ_BASE, mqd->cp_hqd_pq_base_lo);
+		WREG32(mmCP_HQD_PQ_BASE_HI, mqd->cp_hqd_pq_base_hi);
+
+		/* set up the HQD, this is similar to CP_RB0_CNTL */
+		tmp = RREG32(mmCP_HQD_PQ_CONTROL);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE,
+				    (order_base_2(ring->ring_size / 4) - 1));
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
+			       ((order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1) << 8));
+#ifdef __BIG_ENDIAN
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ENDIAN_SWAP, 1);
+#endif
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, ROQ_PQ_IB_FLIP, 0);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
+		WREG32(mmCP_HQD_PQ_CONTROL, tmp);
+		mqd->cp_hqd_pq_control = tmp;
+
+		/* set the wb address wether it's enabled or not */
+		wb_gpu_addr = adev->wb.gpu_addr + (ring->rptr_offs * 4);
+		mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc;
+		mqd->cp_hqd_pq_rptr_report_addr_hi =
+			upper_32_bits(wb_gpu_addr) & 0xffff;
+		WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR,
+		       mqd->cp_hqd_pq_rptr_report_addr_lo);
+		WREG32(mmCP_HQD_PQ_RPTR_REPORT_ADDR_HI,
+		       mqd->cp_hqd_pq_rptr_report_addr_hi);
+
+		/* only used if CP_PQ_WPTR_POLL_CNTL.CP_PQ_WPTR_POLL_CNTL__EN_MASK=1 */
+		wb_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+		mqd->cp_hqd_pq_wptr_poll_addr = wb_gpu_addr & 0xfffffffc;
+		mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff;
+		WREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR, mqd->cp_hqd_pq_wptr_poll_addr);
+		WREG32(mmCP_HQD_PQ_WPTR_POLL_ADDR_HI,
+		       mqd->cp_hqd_pq_wptr_poll_addr_hi);
+
+		/* enable the doorbell if requested */
+		if (use_doorbell) {
+			if ((adev->asic_type == CHIP_CARRIZO) ||
+			    (adev->asic_type == CHIP_FIJI) ||
+			    (adev->asic_type == CHIP_STONEY)) {
+				WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER,
+				       AMDGPU_DOORBELL_KIQ << 2);
+				WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER,
+				       AMDGPU_DOORBELL_MEC_RING7 << 2);
+			}
+			tmp = RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL);
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL,
+					    DOORBELL_OFFSET, ring->doorbell_index);
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_SOURCE, 0);
+			tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_HIT, 0);
+			mqd->cp_hqd_pq_doorbell_control = tmp;
+
+		} else {
+			mqd->cp_hqd_pq_doorbell_control = 0;
+		}
+		WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL,
+		       mqd->cp_hqd_pq_doorbell_control);
+
+		/* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */
+		ring->wptr = 0;
+		mqd->cp_hqd_pq_wptr = ring->wptr;
+		WREG32(mmCP_HQD_PQ_WPTR, mqd->cp_hqd_pq_wptr);
+		mqd->cp_hqd_pq_rptr = RREG32(mmCP_HQD_PQ_RPTR);
+
+		/* set the vmid for the queue */
+		mqd->cp_hqd_vmid = 0;
+		WREG32(mmCP_HQD_VMID, mqd->cp_hqd_vmid);
+
+		tmp = RREG32(mmCP_HQD_PERSISTENT_STATE);
+		tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x53);
+		WREG32(mmCP_HQD_PERSISTENT_STATE, tmp);
+		mqd->cp_hqd_persistent_state = tmp;
+
+		/* activate the queue */
+		mqd->cp_hqd_active = 1;
+		WREG32(mmCP_HQD_ACTIVE, mqd->cp_hqd_active);
+
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		amdgpu_bo_kunmap(ring->mqd_obj);
+		amdgpu_bo_unreserve(ring->mqd_obj);
+	}
+
+	if (use_doorbell) {
+		tmp = RREG32(mmCP_PQ_STATUS);
+		tmp = REG_SET_FIELD(tmp, CP_PQ_STATUS, DOORBELL_ENABLE, 1);
+		WREG32(mmCP_PQ_STATUS, tmp);
+	}
+
+	r = gfx_v8_0_cp_compute_start(adev);
+	if (r)
+		return r;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
+
+		ring->ready = true;
+		r = amdgpu_ring_test_ring(ring);
+		if (r)
+			ring->ready = false;
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_cp_resume(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!(adev->flags & AMD_IS_APU))
+		gfx_v8_0_enable_gui_idle_interrupt(adev, false);
+
+	if (!adev->firmware.smu_load) {
+		/* legacy firmware loading */
+		r = gfx_v8_0_cp_gfx_load_microcode(adev);
+		if (r)
+			return r;
+
+		r = gfx_v8_0_cp_compute_load_microcode(adev);
+		if (r)
+			return r;
+	} else {
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_CP_CE);
+		if (r)
+			return -EINVAL;
+
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_CP_PFP);
+		if (r)
+			return -EINVAL;
+
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_CP_ME);
+		if (r)
+			return -EINVAL;
+
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_CP_MEC1);
+		if (r)
+			return -EINVAL;
+	}
+
+	r = gfx_v8_0_cp_gfx_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v8_0_cp_compute_resume(adev);
+	if (r)
+		return r;
+
+	gfx_v8_0_enable_gui_idle_interrupt(adev, true);
+
+	return 0;
+}
+
+static void gfx_v8_0_cp_enable(struct amdgpu_device *adev, bool enable)
+{
+	gfx_v8_0_cp_gfx_enable(adev, enable);
+	gfx_v8_0_cp_compute_enable(adev, enable);
+}
+
+static int gfx_v8_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v8_0_init_golden_registers(adev);
+
+	gfx_v8_0_gpu_init(adev);
+
+	r = gfx_v8_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	r = gfx_v8_0_cp_resume(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gfx_v8_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gfx_v8_0_cp_enable(adev, false);
+	gfx_v8_0_rlc_stop(adev);
+	gfx_v8_0_cp_compute_fini(adev);
+
+	return 0;
+}
+
+static int gfx_v8_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v8_0_hw_fini(adev);
+}
+
+static int gfx_v8_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return gfx_v8_0_hw_init(adev);
+}
+
+static bool gfx_v8_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (REG_GET_FIELD(RREG32(mmGRBM_STATUS), GRBM_STATUS, GUI_ACTIVE))
+		return false;
+	else
+		return true;
+}
+
+static int gfx_v8_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK;
+
+		if (!REG_GET_FIELD(tmp, GRBM_STATUS, GUI_ACTIVE))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void gfx_v8_0_print_status(void *handle)
+{
+	int i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "GFX 8.x registers\n");
+	dev_info(adev->dev, "  GRBM_STATUS=0x%08X\n",
+		 RREG32(mmGRBM_STATUS));
+	dev_info(adev->dev, "  GRBM_STATUS2=0x%08X\n",
+		 RREG32(mmGRBM_STATUS2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE0=0x%08X\n",
+		 RREG32(mmGRBM_STATUS_SE0));
+	dev_info(adev->dev, "  GRBM_STATUS_SE1=0x%08X\n",
+		 RREG32(mmGRBM_STATUS_SE1));
+	dev_info(adev->dev, "  GRBM_STATUS_SE2=0x%08X\n",
+		 RREG32(mmGRBM_STATUS_SE2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE3=0x%08X\n",
+		 RREG32(mmGRBM_STATUS_SE3));
+	dev_info(adev->dev, "  CP_STAT = 0x%08x\n", RREG32(mmCP_STAT));
+	dev_info(adev->dev, "  CP_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_STALLED_STAT2 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT2));
+	dev_info(adev->dev, "  CP_STALLED_STAT3 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT3));
+	dev_info(adev->dev, "  CP_CPF_BUSY_STAT = 0x%08x\n",
+		 RREG32(mmCP_CPF_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPF_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPF_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPF_STATUS = 0x%08x\n", RREG32(mmCP_CPF_STATUS));
+	dev_info(adev->dev, "  CP_CPC_BUSY_STAT = 0x%08x\n", RREG32(mmCP_CPC_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPC_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPC_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPC_STATUS = 0x%08x\n", RREG32(mmCP_CPC_STATUS));
+
+	for (i = 0; i < 32; i++) {
+		dev_info(adev->dev, "  GB_TILE_MODE%d=0x%08X\n",
+			 i, RREG32(mmGB_TILE_MODE0 + (i * 4)));
+	}
+	for (i = 0; i < 16; i++) {
+		dev_info(adev->dev, "  GB_MACROTILE_MODE%d=0x%08X\n",
+			 i, RREG32(mmGB_MACROTILE_MODE0 + (i * 4)));
+	}
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		dev_info(adev->dev, "  se: %d\n", i);
+		gfx_v8_0_select_se_sh(adev, i, 0xffffffff);
+		dev_info(adev->dev, "  PA_SC_RASTER_CONFIG=0x%08X\n",
+			 RREG32(mmPA_SC_RASTER_CONFIG));
+		dev_info(adev->dev, "  PA_SC_RASTER_CONFIG_1=0x%08X\n",
+			 RREG32(mmPA_SC_RASTER_CONFIG_1));
+	}
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+
+	dev_info(adev->dev, "  GB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmGB_ADDR_CONFIG));
+	dev_info(adev->dev, "  HDP_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmHDP_ADDR_CONFIG));
+	dev_info(adev->dev, "  DMIF_ADDR_CALC=0x%08X\n",
+		 RREG32(mmDMIF_ADDR_CALC));
+	dev_info(adev->dev, "  SDMA0_TILING_CONFIG=0x%08X\n",
+		 RREG32(mmSDMA0_TILING_CONFIG + SDMA0_REGISTER_OFFSET));
+	dev_info(adev->dev, "  SDMA1_TILING_CONFIG=0x%08X\n",
+		 RREG32(mmSDMA0_TILING_CONFIG + SDMA1_REGISTER_OFFSET));
+	dev_info(adev->dev, "  UVD_UDEC_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DB_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DBW_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DBW_ADDR_CONFIG));
+
+	dev_info(adev->dev, "  CP_MEQ_THRESHOLDS=0x%08X\n",
+		 RREG32(mmCP_MEQ_THRESHOLDS));
+	dev_info(adev->dev, "  SX_DEBUG_1=0x%08X\n",
+		 RREG32(mmSX_DEBUG_1));
+	dev_info(adev->dev, "  TA_CNTL_AUX=0x%08X\n",
+		 RREG32(mmTA_CNTL_AUX));
+	dev_info(adev->dev, "  SPI_CONFIG_CNTL=0x%08X\n",
+		 RREG32(mmSPI_CONFIG_CNTL));
+	dev_info(adev->dev, "  SQ_CONFIG=0x%08X\n",
+		 RREG32(mmSQ_CONFIG));
+	dev_info(adev->dev, "  DB_DEBUG=0x%08X\n",
+		 RREG32(mmDB_DEBUG));
+	dev_info(adev->dev, "  DB_DEBUG2=0x%08X\n",
+		 RREG32(mmDB_DEBUG2));
+	dev_info(adev->dev, "  DB_DEBUG3=0x%08X\n",
+		 RREG32(mmDB_DEBUG3));
+	dev_info(adev->dev, "  CB_HW_CONTROL=0x%08X\n",
+		 RREG32(mmCB_HW_CONTROL));
+	dev_info(adev->dev, "  SPI_CONFIG_CNTL_1=0x%08X\n",
+		 RREG32(mmSPI_CONFIG_CNTL_1));
+	dev_info(adev->dev, "  PA_SC_FIFO_SIZE=0x%08X\n",
+		 RREG32(mmPA_SC_FIFO_SIZE));
+	dev_info(adev->dev, "  VGT_NUM_INSTANCES=0x%08X\n",
+		 RREG32(mmVGT_NUM_INSTANCES));
+	dev_info(adev->dev, "  CP_PERFMON_CNTL=0x%08X\n",
+		 RREG32(mmCP_PERFMON_CNTL));
+	dev_info(adev->dev, "  PA_SC_FORCE_EOV_MAX_CNTS=0x%08X\n",
+		 RREG32(mmPA_SC_FORCE_EOV_MAX_CNTS));
+	dev_info(adev->dev, "  VGT_CACHE_INVALIDATION=0x%08X\n",
+		 RREG32(mmVGT_CACHE_INVALIDATION));
+	dev_info(adev->dev, "  VGT_GS_VERTEX_REUSE=0x%08X\n",
+		 RREG32(mmVGT_GS_VERTEX_REUSE));
+	dev_info(adev->dev, "  PA_SC_LINE_STIPPLE_STATE=0x%08X\n",
+		 RREG32(mmPA_SC_LINE_STIPPLE_STATE));
+	dev_info(adev->dev, "  PA_CL_ENHANCE=0x%08X\n",
+		 RREG32(mmPA_CL_ENHANCE));
+	dev_info(adev->dev, "  PA_SC_ENHANCE=0x%08X\n",
+		 RREG32(mmPA_SC_ENHANCE));
+
+	dev_info(adev->dev, "  CP_ME_CNTL=0x%08X\n",
+		 RREG32(mmCP_ME_CNTL));
+	dev_info(adev->dev, "  CP_MAX_CONTEXT=0x%08X\n",
+		 RREG32(mmCP_MAX_CONTEXT));
+	dev_info(adev->dev, "  CP_ENDIAN_SWAP=0x%08X\n",
+		 RREG32(mmCP_ENDIAN_SWAP));
+	dev_info(adev->dev, "  CP_DEVICE_ID=0x%08X\n",
+		 RREG32(mmCP_DEVICE_ID));
+
+	dev_info(adev->dev, "  CP_SEM_WAIT_TIMER=0x%08X\n",
+		 RREG32(mmCP_SEM_WAIT_TIMER));
+
+	dev_info(adev->dev, "  CP_RB_WPTR_DELAY=0x%08X\n",
+		 RREG32(mmCP_RB_WPTR_DELAY));
+	dev_info(adev->dev, "  CP_RB_VMID=0x%08X\n",
+		 RREG32(mmCP_RB_VMID));
+	dev_info(adev->dev, "  CP_RB0_CNTL=0x%08X\n",
+		 RREG32(mmCP_RB0_CNTL));
+	dev_info(adev->dev, "  CP_RB0_WPTR=0x%08X\n",
+		 RREG32(mmCP_RB0_WPTR));
+	dev_info(adev->dev, "  CP_RB0_RPTR_ADDR=0x%08X\n",
+		 RREG32(mmCP_RB0_RPTR_ADDR));
+	dev_info(adev->dev, "  CP_RB0_RPTR_ADDR_HI=0x%08X\n",
+		 RREG32(mmCP_RB0_RPTR_ADDR_HI));
+	dev_info(adev->dev, "  CP_RB0_CNTL=0x%08X\n",
+		 RREG32(mmCP_RB0_CNTL));
+	dev_info(adev->dev, "  CP_RB0_BASE=0x%08X\n",
+		 RREG32(mmCP_RB0_BASE));
+	dev_info(adev->dev, "  CP_RB0_BASE_HI=0x%08X\n",
+		 RREG32(mmCP_RB0_BASE_HI));
+	dev_info(adev->dev, "  CP_MEC_CNTL=0x%08X\n",
+		 RREG32(mmCP_MEC_CNTL));
+	dev_info(adev->dev, "  CP_CPF_DEBUG=0x%08X\n",
+		 RREG32(mmCP_CPF_DEBUG));
+
+	dev_info(adev->dev, "  SCRATCH_ADDR=0x%08X\n",
+		 RREG32(mmSCRATCH_ADDR));
+	dev_info(adev->dev, "  SCRATCH_UMSK=0x%08X\n",
+		 RREG32(mmSCRATCH_UMSK));
+
+	dev_info(adev->dev, "  CP_INT_CNTL_RING0=0x%08X\n",
+		 RREG32(mmCP_INT_CNTL_RING0));
+	dev_info(adev->dev, "  RLC_LB_CNTL=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTL));
+	dev_info(adev->dev, "  RLC_CNTL=0x%08X\n",
+		 RREG32(mmRLC_CNTL));
+	dev_info(adev->dev, "  RLC_CGCG_CGLS_CTRL=0x%08X\n",
+		 RREG32(mmRLC_CGCG_CGLS_CTRL));
+	dev_info(adev->dev, "  RLC_LB_CNTR_INIT=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTR_INIT));
+	dev_info(adev->dev, "  RLC_LB_CNTR_MAX=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTR_MAX));
+	dev_info(adev->dev, "  RLC_LB_INIT_CU_MASK=0x%08X\n",
+		 RREG32(mmRLC_LB_INIT_CU_MASK));
+	dev_info(adev->dev, "  RLC_LB_PARAMS=0x%08X\n",
+		 RREG32(mmRLC_LB_PARAMS));
+	dev_info(adev->dev, "  RLC_LB_CNTL=0x%08X\n",
+		 RREG32(mmRLC_LB_CNTL));
+	dev_info(adev->dev, "  RLC_MC_CNTL=0x%08X\n",
+		 RREG32(mmRLC_MC_CNTL));
+	dev_info(adev->dev, "  RLC_UCODE_CNTL=0x%08X\n",
+		 RREG32(mmRLC_UCODE_CNTL));
+
+	mutex_lock(&adev->srbm_mutex);
+	for (i = 0; i < 16; i++) {
+		vi_srbm_select(adev, 0, 0, 0, i);
+		dev_info(adev->dev, "  VM %d:\n", i);
+		dev_info(adev->dev, "  SH_MEM_CONFIG=0x%08X\n",
+			 RREG32(mmSH_MEM_CONFIG));
+		dev_info(adev->dev, "  SH_MEM_APE1_BASE=0x%08X\n",
+			 RREG32(mmSH_MEM_APE1_BASE));
+		dev_info(adev->dev, "  SH_MEM_APE1_LIMIT=0x%08X\n",
+			 RREG32(mmSH_MEM_APE1_LIMIT));
+		dev_info(adev->dev, "  SH_MEM_BASES=0x%08X\n",
+			 RREG32(mmSH_MEM_BASES));
+	}
+	vi_srbm_select(adev, 0, 0, 0, 0);
+	mutex_unlock(&adev->srbm_mutex);
+}
+
+static int gfx_v8_0_soft_reset(void *handle)
+{
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* GRBM_STATUS */
+	tmp = RREG32(mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_GFX, 1);
+	}
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) {
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1);
+	}
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32(mmGRBM_STATUS2);
+	if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY))
+		grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
+						GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+
+	/* SRBM_STATUS */
+	tmp = RREG32(mmSRBM_STATUS);
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, GRBM_RQ_PENDING))
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1);
+
+	if (grbm_soft_reset || srbm_soft_reset) {
+		gfx_v8_0_print_status((void *)adev);
+		/* stop the rlc */
+		gfx_v8_0_rlc_stop(adev);
+
+		/* Disable GFX parsing/prefetching */
+		gfx_v8_0_cp_gfx_enable(adev, false);
+
+		/* Disable MEC parsing/prefetching */
+		/* XXX todo */
+
+		if (grbm_soft_reset) {
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+			tmp |= grbm_soft_reset;
+			dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32(mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~grbm_soft_reset;
+			WREG32(mmGRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmGRBM_SOFT_RESET);
+		}
+
+		if (srbm_soft_reset) {
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+			tmp |= srbm_soft_reset;
+			dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+			WREG32(mmSRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+
+			udelay(50);
+
+			tmp &= ~srbm_soft_reset;
+			WREG32(mmSRBM_SOFT_RESET, tmp);
+			tmp = RREG32(mmSRBM_SOFT_RESET);
+		}
+		/* Wait a little for things to settle down */
+		udelay(50);
+		gfx_v8_0_print_status((void *)adev);
+	}
+	return 0;
+}
+
+/**
+ * gfx_v8_0_get_gpu_clock_counter - return GPU clock counter snapshot
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Fetches a GPU clock counter snapshot.
+ * Returns the 64 bit clock counter snapshot.
+ */
+uint64_t gfx_v8_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+{
+	uint64_t clock;
+
+	mutex_lock(&adev->gfx.gpu_clock_mutex);
+	WREG32(mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1);
+	clock = (uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_LSB) |
+		((uint64_t)RREG32(mmRLC_GPU_CLOCK_COUNT_MSB) << 32ULL);
+	mutex_unlock(&adev->gfx.gpu_clock_mutex);
+	return clock;
+}
+
+static void gfx_v8_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
+					  uint32_t vmid,
+					  uint32_t gds_base, uint32_t gds_size,
+					  uint32_t gws_base, uint32_t gws_size,
+					  uint32_t oa_base, uint32_t oa_size)
+{
+	gds_base = gds_base >> AMDGPU_GDS_SHIFT;
+	gds_size = gds_size >> AMDGPU_GDS_SHIFT;
+
+	gws_base = gws_base >> AMDGPU_GWS_SHIFT;
+	gws_size = gws_size >> AMDGPU_GWS_SHIFT;
+
+	oa_base = oa_base >> AMDGPU_OA_SHIFT;
+	oa_size = oa_size >> AMDGPU_OA_SHIFT;
+
+	/* GDS Base */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_base);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_base);
+
+	/* GDS Size */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].mem_size);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gds_size);
+
+	/* GWS */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].gws);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, gws_size << GDS_GWS_VMID0__SIZE__SHIFT | gws_base);
+
+	/* OA */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, amdgpu_gds_reg_offset[vmid].oa);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
+}
+
+static int gfx_v8_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->gfx.num_gfx_rings = GFX8_NUM_GFX_RINGS;
+	adev->gfx.num_compute_rings = GFX8_NUM_COMPUTE_RINGS;
+	gfx_v8_0_set_ring_funcs(adev);
+	gfx_v8_0_set_irq_funcs(adev);
+	gfx_v8_0_set_gds_init(adev);
+
+	return 0;
+}
+
+static int gfx_v8_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+static int gfx_v8_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static u32 gfx_v8_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	rptr = ring->adev->wb.wb[ring->rptr_offs];
+
+	return rptr;
+}
+
+static u32 gfx_v8_0_ring_get_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 wptr;
+
+	if (ring->use_doorbell)
+		/* XXX check if swapping is necessary on BE */
+		wptr = ring->adev->wb.wb[ring->wptr_offs];
+	else
+		wptr = RREG32(mmCP_RB0_WPTR);
+
+	return wptr;
+}
+
+static void gfx_v8_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		adev->wb.wb[ring->wptr_offs] = ring->wptr;
+		WDOORBELL32(ring->doorbell_index, ring->wptr);
+	} else {
+		WREG32(mmCP_RB0_WPTR, ring->wptr);
+		(void)RREG32(mmCP_RB0_WPTR);
+	}
+}
+
+static void gfx_v8_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask, reg_mem_engine;
+
+	if (ring->type == AMDGPU_RING_TYPE_COMPUTE) {
+		switch (ring->me) {
+		case 1:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP2_MASK << ring->pipe;
+			break;
+		case 2:
+			ref_and_mask = GPU_HDP_FLUSH_DONE__CP6_MASK << ring->pipe;
+			break;
+		default:
+			return;
+		}
+		reg_mem_engine = 0;
+	} else {
+		ref_and_mask = GPU_HDP_FLUSH_DONE__CP0_MASK;
+		reg_mem_engine = WAIT_REG_MEM_ENGINE(1); /* pfp */
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(1) | /* write, wait, write */
+				 WAIT_REG_MEM_FUNCTION(3) |  /* == */
+				 reg_mem_engine));
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, ref_and_mask);
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+}
+
+static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	bool need_ctx_switch = ring->current_ctx != ib->ctx;
+	u32 header, control = 0;
+	u32 next_rptr = ring->wptr + 5;
+
+	/* drop the CE preamble IB for the same context */
+	if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) && !need_ctx_switch)
+		return;
+
+	if (need_ctx_switch)
+		next_rptr += 2;
+
+	next_rptr += 4;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
+	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, next_rptr);
+
+	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
+	if (need_ctx_switch) {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+	if (ib->flags & AMDGPU_IB_FLAG_CE)
+		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
+	else
+		header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw |
+		(ib->vm ? (ib->vm->ids[ring->idx].id << 24) : 0);
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+			  (2 << 0) |
+#endif
+			  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	u32 header, control = 0;
+	u32 next_rptr = ring->wptr + 5;
+
+	control |= INDIRECT_BUFFER_VALID;
+
+	next_rptr += 4;
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
+	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
+	amdgpu_ring_write(ring, next_rptr);
+
+	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+
+	control |= ib->length_dw |
+			   (ib->vm ? (ib->vm->ids[ring->idx].id << 24) : 0);
+
+	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring,
+#ifdef __BIG_ENDIAN
+					  (2 << 0) |
+#endif
+					  (ib->gpu_addr & 0xFFFFFFFC));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF);
+	amdgpu_ring_write(ring, control);
+}
+
+static void gfx_v8_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr,
+					 u64 seq, unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* EVENT_WRITE_EOP - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) | 
+			  DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+
+}
+
+/**
+ * gfx_v8_0_ring_emit_semaphore - emit a semaphore on the CP ring
+ *
+ * @ring: amdgpu ring buffer object
+ * @semaphore: amdgpu semaphore object
+ * @emit_wait: Is this a sempahore wait?
+ *
+ * Emits a semaphore signal/wait packet to the CP ring and prevents the PFP
+ * from running ahead of semaphore waits.
+ */
+static bool gfx_v8_0_ring_emit_semaphore(struct amdgpu_ring *ring,
+					 struct amdgpu_semaphore *semaphore,
+					 bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+	unsigned sel = emit_wait ? PACKET3_SEM_SEL_WAIT : PACKET3_SEM_SEL_SIGNAL;
+
+	if (ring->adev->asic_type == CHIP_TOPAZ ||
+	    ring->adev->asic_type == CHIP_TONGA ||
+	    ring->adev->asic_type == CHIP_FIJI)
+		/* we got a hw semaphore bug in VI TONGA, return false to switch back to sw fence wait */
+		return false;
+	else {
+		amdgpu_ring_write(ring, PACKET3(PACKET3_MEM_SEMAPHORE, 2));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, sel);
+	}
+
+	if (emit_wait && (ring->type == AMDGPU_RING_TYPE_GFX)) {
+		/* Prevent the PFP from running ahead of the semaphore wait */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+	}
+
+	return true;
+}
+
+static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					unsigned vm_id, uint64_t pd_addr)
+{
+	int usepfp = (ring->type == AMDGPU_RING_TYPE_GFX);
+	uint32_t seq = ring->fence_drv.sync_seq[ring->idx];
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_MEM_SPACE(1) | /* memory */
+				 WAIT_REG_MEM_FUNCTION(3) | /* equal */
+				 WAIT_REG_MEM_ENGINE(usepfp))); /* pfp or me */
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xffffffff);
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, 0xffffffff);
+	amdgpu_ring_write(ring, 4); /* poll interval */
+
+	if (usepfp) {
+		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
+				 WRITE_DATA_DST_SEL(0)) |
+				 WR_CONFIRM);
+	if (vm_id < 8) {
+		amdgpu_ring_write(ring,
+				  (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
+	} else {
+		amdgpu_ring_write(ring,
+				  (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
+	}
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	/* invalidate the cache */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
+	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(0) |
+				 WRITE_DATA_DST_SEL(0)));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 1 << vm_id);
+
+	/* wait for the invalidate to complete */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_WAIT_REG_MEM, 5));
+	amdgpu_ring_write(ring, (WAIT_REG_MEM_OPERATION(0) | /* wait */
+				 WAIT_REG_MEM_FUNCTION(0) |  /* always */
+				 WAIT_REG_MEM_ENGINE(0))); /* me */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* ref */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, 0x20); /* poll interval */
+
+	/* compute doesn't have PFP */
+	if (usepfp) {
+		/* sync PFP to ME, otherwise we might get invalid PFP reads */
+		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
+		amdgpu_ring_write(ring, 0x0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+		amdgpu_ring_write(ring, 0);
+	}
+}
+
+static u32 gfx_v8_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->rptr_offs];
+}
+
+static u32 gfx_v8_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
+{
+	return ring->adev->wb.wb[ring->wptr_offs];
+}
+
+static void gfx_v8_0_ring_set_wptr_compute(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	/* XXX check if swapping is necessary on BE */
+	adev->wb.wb[ring->wptr_offs] = ring->wptr;
+	WDOORBELL32(ring->doorbell_index, ring->wptr);
+}
+
+static void gfx_v8_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
+					     u64 addr, u64 seq,
+					     unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	bool int_sel = flags & AMDGPU_FENCE_FLAG_INT;
+
+	/* RELEASE_MEM - flush caches, send int */
+	amdgpu_ring_write(ring, PACKET3(PACKET3_RELEASE_MEM, 5));
+	amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN |
+				 EOP_TC_ACTION_EN |
+				 EOP_TC_WB_ACTION_EN |
+				 EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) |
+				 EVENT_INDEX(5)));
+	amdgpu_ring_write(ring, DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0));
+	amdgpu_ring_write(ring, addr & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+	amdgpu_ring_write(ring, upper_32_bits(seq));
+}
+
+static void gfx_v8_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
+						 enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    TIME_STAMP_INT_ENABLE, 0);
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl =
+			REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+				      TIME_STAMP_INT_ENABLE, 1);
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static void gfx_v8_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev,
+						     int me, int pipe,
+						     enum amdgpu_interrupt_state state)
+{
+	u32 mec_int_cntl, mec_int_cntl_reg;
+
+	/*
+	 * amdgpu controls only pipe 0 of MEC1. That's why this function only
+	 * handles the setting of interrupts for this specific pipe. All other
+	 * pipes' interrupts are set by amdkfd.
+	 */
+
+	if (me == 1) {
+		switch (pipe) {
+		case 0:
+			mec_int_cntl_reg = mmCP_ME1_PIPE0_INT_CNTL;
+			break;
+		default:
+			DRM_DEBUG("invalid pipe %d\n", pipe);
+			return;
+		}
+	} else {
+		DRM_DEBUG("invalid me %d\n", me);
+		return;
+	}
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 0);
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		mec_int_cntl = RREG32(mec_int_cntl_reg);
+		mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL,
+					     TIME_STAMP_INT_ENABLE, 1);
+		WREG32(mec_int_cntl_reg, mec_int_cntl);
+		break;
+	default:
+		break;
+	}
+}
+
+static int gfx_v8_0_set_priv_reg_fault_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *source,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    PRIV_REG_INT_ENABLE, 0);
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    PRIV_REG_INT_ENABLE, 0);
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_set_priv_inst_fault_state(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      unsigned type,
+					      enum amdgpu_interrupt_state state)
+{
+	u32 cp_int_cntl;
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    PRIV_INSTR_INT_ENABLE, 0);
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0);
+		cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0,
+					    PRIV_INSTR_INT_ENABLE, 1);
+		WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gfx_v8_0_set_eop_interrupt_state(struct amdgpu_device *adev,
+					    struct amdgpu_irq_src *src,
+					    unsigned type,
+					    enum amdgpu_interrupt_state state)
+{
+	switch (type) {
+	case AMDGPU_CP_IRQ_GFX_EOP:
+		gfx_v8_0_set_gfx_eop_interrupt_state(adev, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE3_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 1, 3, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE0_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 0, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE1_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 1, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE2_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 2, state);
+		break;
+	case AMDGPU_CP_IRQ_COMPUTE_MEC2_PIPE3_EOP:
+		gfx_v8_0_set_compute_eop_interrupt_state(adev, 2, 3, state);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v8_0_eop_irq(struct amdgpu_device *adev,
+			    struct amdgpu_irq_src *source,
+			    struct amdgpu_iv_entry *entry)
+{
+	int i;
+	u8 me_id, pipe_id, queue_id;
+	struct amdgpu_ring *ring;
+
+	DRM_DEBUG("IH: CP EOP\n");
+	me_id = (entry->ring_id & 0x0c) >> 2;
+	pipe_id = (entry->ring_id & 0x03) >> 0;
+	queue_id = (entry->ring_id & 0x70) >> 4;
+
+	switch (me_id) {
+	case 0:
+		amdgpu_fence_process(&adev->gfx.gfx_ring[0]);
+		break;
+	case 1:
+	case 2:
+		for (i = 0; i < adev->gfx.num_compute_rings; i++) {
+			ring = &adev->gfx.compute_ring[i];
+			/* Per-queue interrupt is supported for MEC starting from VI.
+			  * The interrupt can only be enabled/disabled per pipe instead of per queue.
+			  */
+			if ((ring->me == me_id) && (ring->pipe == pipe_id) && (ring->queue == queue_id))
+				amdgpu_fence_process(ring);
+		}
+		break;
+	}
+	return 0;
+}
+
+static int gfx_v8_0_priv_reg_irq(struct amdgpu_device *adev,
+				 struct amdgpu_irq_src *source,
+				 struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal register access in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int gfx_v8_0_priv_inst_irq(struct amdgpu_device *adev,
+				  struct amdgpu_irq_src *source,
+				  struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+const struct amd_ip_funcs gfx_v8_0_ip_funcs = {
+	.early_init = gfx_v8_0_early_init,
+	.late_init = NULL,
+	.sw_init = gfx_v8_0_sw_init,
+	.sw_fini = gfx_v8_0_sw_fini,
+	.hw_init = gfx_v8_0_hw_init,
+	.hw_fini = gfx_v8_0_hw_fini,
+	.suspend = gfx_v8_0_suspend,
+	.resume = gfx_v8_0_resume,
+	.is_idle = gfx_v8_0_is_idle,
+	.wait_for_idle = gfx_v8_0_wait_for_idle,
+	.soft_reset = gfx_v8_0_soft_reset,
+	.print_status = gfx_v8_0_print_status,
+	.set_clockgating_state = gfx_v8_0_set_clockgating_state,
+	.set_powergating_state = gfx_v8_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
+	.get_rptr = gfx_v8_0_ring_get_rptr_gfx,
+	.get_wptr = gfx_v8_0_ring_get_wptr_gfx,
+	.set_wptr = gfx_v8_0_ring_set_wptr_gfx,
+	.parse_cs = NULL,
+	.emit_ib = gfx_v8_0_ring_emit_ib_gfx,
+	.emit_fence = gfx_v8_0_ring_emit_fence_gfx,
+	.emit_semaphore = gfx_v8_0_ring_emit_semaphore,
+	.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v8_0_ring_test_ring,
+	.test_ib = gfx_v8_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
+	.get_rptr = gfx_v8_0_ring_get_rptr_compute,
+	.get_wptr = gfx_v8_0_ring_get_wptr_compute,
+	.set_wptr = gfx_v8_0_ring_set_wptr_compute,
+	.parse_cs = NULL,
+	.emit_ib = gfx_v8_0_ring_emit_ib_compute,
+	.emit_fence = gfx_v8_0_ring_emit_fence_compute,
+	.emit_semaphore = gfx_v8_0_ring_emit_semaphore,
+	.emit_vm_flush = gfx_v8_0_ring_emit_vm_flush,
+	.emit_gds_switch = gfx_v8_0_ring_emit_gds_switch,
+	.emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush,
+	.test_ring = gfx_v8_0_ring_test_ring,
+	.test_ib = gfx_v8_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
+		adev->gfx.gfx_ring[i].funcs = &gfx_v8_0_ring_funcs_gfx;
+
+	for (i = 0; i < adev->gfx.num_compute_rings; i++)
+		adev->gfx.compute_ring[i].funcs = &gfx_v8_0_ring_funcs_compute;
+}
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_eop_irq_funcs = {
+	.set = gfx_v8_0_set_eop_interrupt_state,
+	.process = gfx_v8_0_eop_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_priv_reg_irq_funcs = {
+	.set = gfx_v8_0_set_priv_reg_fault_state,
+	.process = gfx_v8_0_priv_reg_irq,
+};
+
+static const struct amdgpu_irq_src_funcs gfx_v8_0_priv_inst_irq_funcs = {
+	.set = gfx_v8_0_set_priv_inst_fault_state,
+	.process = gfx_v8_0_priv_inst_irq,
+};
+
+static void gfx_v8_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->gfx.eop_irq.num_types = AMDGPU_CP_IRQ_LAST;
+	adev->gfx.eop_irq.funcs = &gfx_v8_0_eop_irq_funcs;
+
+	adev->gfx.priv_reg_irq.num_types = 1;
+	adev->gfx.priv_reg_irq.funcs = &gfx_v8_0_priv_reg_irq_funcs;
+
+	adev->gfx.priv_inst_irq.num_types = 1;
+	adev->gfx.priv_inst_irq.funcs = &gfx_v8_0_priv_inst_irq_funcs;
+}
+
+static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev)
+{
+	/* init asci gds info */
+	adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
+	adev->gds.gws.total_size = 64;
+	adev->gds.oa.total_size = 16;
+
+	if (adev->gds.mem.total_size == 64 * 1024) {
+		adev->gds.mem.gfx_partition_size = 4096;
+		adev->gds.mem.cs_partition_size = 4096;
+
+		adev->gds.gws.gfx_partition_size = 4;
+		adev->gds.gws.cs_partition_size = 4;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 1;
+	} else {
+		adev->gds.mem.gfx_partition_size = 1024;
+		adev->gds.mem.cs_partition_size = 1024;
+
+		adev->gds.gws.gfx_partition_size = 16;
+		adev->gds.gws.cs_partition_size = 16;
+
+		adev->gds.oa.gfx_partition_size = 4;
+		adev->gds.oa.cs_partition_size = 4;
+	}
+}
+
+static u32 gfx_v8_0_get_cu_active_bitmap(struct amdgpu_device *adev,
+		u32 se, u32 sh)
+{
+	u32 mask = 0, tmp, tmp1;
+	int i;
+
+	gfx_v8_0_select_se_sh(adev, se, sh);
+	tmp = RREG32(mmCC_GC_SHADER_ARRAY_CONFIG);
+	tmp1 = RREG32(mmGC_USER_SHADER_ARRAY_CONFIG);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+
+	tmp &= 0xffff0000;
+
+	tmp |= tmp1;
+	tmp >>= 16;
+
+	for (i = 0; i < adev->gfx.config.max_cu_per_sh; i ++) {
+		mask <<= 1;
+		mask |= 1;
+	}
+
+	return (~tmp) & mask;
+}
+
+int gfx_v8_0_get_cu_info(struct amdgpu_device *adev,
+						 struct amdgpu_cu_info *cu_info)
+{
+	int i, j, k, counter, active_cu_number = 0;
+	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
+
+	if (!adev || !cu_info)
+		return -EINVAL;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
+		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
+			mask = 1;
+			ao_bitmap = 0;
+			counter = 0;
+			bitmap = gfx_v8_0_get_cu_active_bitmap(adev, i, j);
+			cu_info->bitmap[i][j] = bitmap;
+
+			for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) {
+				if (bitmap & mask) {
+					if (counter < 2)
+						ao_bitmap |= mask;
+					counter ++;
+				}
+				mask <<= 1;
+			}
+			active_cu_number += counter;
+			ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
+		}
+	}
+
+	cu_info->number = active_cu_number;
+	cu_info->ao_cu_mask = ao_cu_mask;
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h
new file mode 100644
index 000000000..021e05193
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V8_0_H__
+#define __GFX_V8_0_H__
+
+extern const struct amd_ip_funcs gfx_v8_0_ip_funcs;
+
+uint64_t gfx_v8_0_get_gpu_clock_counter(struct amdgpu_device *adev);
+void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
+int gfx_v8_0_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
new file mode 100644
index 000000000..272110cc1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -0,0 +1,1409 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "cikd.h"
+#include "cik.h"
+#include "gmc_v7_0.h"
+#include "amdgpu_ucode.h"
+
+#include "bif/bif_4_1_d.h"
+#include "bif/bif_4_1_sh_mask.h"
+
+#include "gmc/gmc_7_1_d.h"
+#include "gmc/gmc_7_1_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static void gmc_v7_0_set_gart_funcs(struct amdgpu_device *adev);
+static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("radeon/bonaire_mc.bin");
+MODULE_FIRMWARE("radeon/hawaii_mc.bin");
+MODULE_FIRMWARE("amdgpu/topaz_mc.bin");
+
+static const u32 golden_settings_iceland_a11[] =
+{
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static void gmc_v7_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_program_register_sequence(adev,
+						 iceland_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_iceland_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_iceland_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * gmc7_mc_wait_for_idle - wait for MC idle callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Wait for the MC (memory controller) to be idle.
+ * (evergreen+).
+ * Returns 0 if the MC is idle, -1 if not.
+ */
+int gmc_v7_0_mc_wait_for_idle(struct amdgpu_device *adev)
+{
+	unsigned i;
+	u32 tmp;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & 0x1F00;
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -1;
+}
+
+void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
+		      struct amdgpu_mode_mc_save *save)
+{
+	u32 blackout;
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_stop_mc_access(adev, save);
+
+	amdgpu_asic_wait_for_mc_idle(adev);
+
+	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
+		/* Block CPU access */
+		WREG32(mmBIF_FB_EN, 0);
+		/* blackout the MC */
+		blackout = REG_SET_FIELD(blackout,
+					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
+	}
+	/* wait for the MC to settle */
+	udelay(100);
+}
+
+void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
+			struct amdgpu_mode_mc_save *save)
+{
+	u32 tmp;
+
+	/* unblackout the MC */
+	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
+	/* allow CPU access */
+	tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
+	tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
+	WREG32(mmBIF_FB_EN, tmp);
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_resume_mc_access(adev, save);
+}
+
+/**
+ * gmc_v7_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v7_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_BONAIRE:
+		chip_name = "bonaire";
+		break;
+	case CHIP_HAWAII:
+		chip_name = "hawaii";
+		break;
+	case CHIP_TOPAZ:
+		chip_name = "topaz";
+		break;
+	case CHIP_KAVERI:
+	case CHIP_KABINI:
+		return 0;
+	default: BUG();
+	}
+
+	if (adev->asic_type == CHIP_TOPAZ)
+		snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
+	else
+		snprintf(fw_name, sizeof(fw_name), "radeon/%s_mc.bin", chip_name);
+
+	err = request_firmware(&adev->mc.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->mc.fw);
+
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "cik_mc: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		release_firmware(adev->mc.fw);
+		adev->mc.fw = NULL;
+	}
+	return err;
+}
+
+/**
+ * gmc_v7_0_mc_load_microcode - load MC ucode into the hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Load the GDDR MC ucode into the hw (CIK).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct mc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data = NULL;
+	const __le32 *io_mc_regs = NULL;
+	u32 running, blackout = 0;
+	int i, ucode_size, regs_size;
+
+	if (!adev->mc.fw)
+		return -EINVAL;
+
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;
+	amdgpu_ucode_print_mc_hdr(&hdr->header);
+
+	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
+	io_mc_regs = (const __le32 *)
+		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	fw_data = (const __le32 *)
+		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN);
+
+	if (running == 0) {
+		if (running) {
+			blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+			WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
+		}
+
+		/* reset the engine and set to writable */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
+
+		/* load mc io regs */
+		for (i = 0; i < regs_size; i++) {
+			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(io_mc_regs++));
+			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(io_mc_regs++));
+		}
+		/* load the MC ucode */
+		for (i = 0; i < ucode_size; i++)
+			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(fw_data++));
+
+		/* put the engine back into the active state */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
+
+		/* wait for training to complete */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D0))
+				break;
+			udelay(1);
+		}
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D1))
+				break;
+			udelay(1);
+		}
+
+		if (running)
+			WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout);
+	}
+
+	return 0;
+}
+
+static void gmc_v7_0_vram_gtt_location(struct amdgpu_device *adev,
+				       struct amdgpu_mc *mc)
+{
+	if (mc->mc_vram_size > 0xFFC0000000ULL) {
+		/* leave room for at least 1024M GTT */
+		dev_warn(adev->dev, "limiting VRAM\n");
+		mc->real_vram_size = 0xFFC0000000ULL;
+		mc->mc_vram_size = 0xFFC0000000ULL;
+	}
+	amdgpu_vram_location(adev, &adev->mc, 0);
+	adev->mc.gtt_base_align = 0;
+	amdgpu_gtt_location(adev, mc);
+}
+
+/**
+ * gmc_v7_0_mc_program - program the GPU memory controller
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the location of vram, gart, and AGP in the GPU's
+ * physical address space (CIK).
+ */
+static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_mc_save save;
+	u32 tmp;
+	int i, j;
+
+	/* Initialize HDP */
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		WREG32((0xb05 + j), 0x00000000);
+		WREG32((0xb06 + j), 0x00000000);
+		WREG32((0xb07 + j), 0x00000000);
+		WREG32((0xb08 + j), 0x00000000);
+		WREG32((0xb09 + j), 0x00000000);
+	}
+	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_set_vga_render_state(adev, false);
+
+	gmc_v7_0_mc_stop(adev, &save);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+	/* Update configuration */
+	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+	       adev->mc.vram_start >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+	       adev->mc.vram_end >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
+	       adev->vram_scratch.gpu_addr >> 12);
+	tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
+	tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
+	WREG32(mmMC_VM_FB_LOCATION, tmp);
+	/* XXX double check these! */
+	WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
+	WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
+	WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
+	WREG32(mmMC_VM_AGP_BASE, 0);
+	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
+	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+	gmc_v7_0_mc_resume(adev, &save);
+
+	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+
+	tmp = RREG32(mmHDP_MISC_CNTL);
+	tmp = REG_SET_FIELD(tmp, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 1);
+	WREG32(mmHDP_MISC_CNTL, tmp);
+
+	tmp = RREG32(mmHDP_HOST_PATH_CNTL);
+	WREG32(mmHDP_HOST_PATH_CNTL, tmp);
+}
+
+/**
+ * gmc_v7_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space (CIK).
+ * Returns 0 for success.
+ */
+static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int chansize, numchan;
+
+	/* Get VRAM informations */
+	tmp = RREG32(mmMC_ARB_RAMCFG);
+	if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE)) {
+		chansize = 64;
+	} else {
+		chansize = 32;
+	}
+	tmp = RREG32(mmMC_SHARED_CHMAP);
+	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+	case 0:
+	default:
+		numchan = 1;
+		break;
+	case 1:
+		numchan = 2;
+		break;
+	case 2:
+		numchan = 4;
+		break;
+	case 3:
+		numchan = 8;
+		break;
+	case 4:
+		numchan = 3;
+		break;
+	case 5:
+		numchan = 6;
+		break;
+	case 6:
+		numchan = 10;
+		break;
+	case 7:
+		numchan = 12;
+		break;
+	case 8:
+		numchan = 16;
+		break;
+	}
+	adev->mc.vram_width = numchan * chansize;
+	/* Could aper size report 0 ? */
+	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
+	/* size in MB on si */
+	adev->mc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->mc.visible_vram_size = adev->mc.aper_size;
+
+	/* unless the user had overridden it, set the gart
+	 * size equal to the 1024 or vram, whichever is larger.
+	 */
+	if (amdgpu_gart_size == -1)
+		adev->mc.gtt_size = max((1024ULL << 20), adev->mc.mc_vram_size);
+	else
+		adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20;
+
+	gmc_v7_0_vram_gtt_location(adev, &adev->mc);
+
+	return 0;
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+/**
+ * gmc_v7_0_gart_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ *
+ * Flush the TLB for the requested page table (CIK).
+ */
+static void gmc_v7_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
+					uint32_t vmid)
+{
+	/* flush hdp cache */
+	WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+/**
+ * gmc_v7_0_gart_set_pte_pde - update the page tables using MMIO
+ *
+ * @adev: amdgpu_device pointer
+ * @cpu_pt_addr: cpu address of the page table
+ * @gpu_page_idx: entry in the page table to update
+ * @addr: dst addr to write into pte/pde
+ * @flags: access flags
+ *
+ * Update the page tables using the CPU.
+ */
+static int gmc_v7_0_gart_set_pte_pde(struct amdgpu_device *adev,
+				     void *cpu_pt_addr,
+				     uint32_t gpu_page_idx,
+				     uint64_t addr,
+				     uint32_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	value = addr & 0xFFFFFFFFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+
+	return 0;
+}
+
+/**
+ * gmc_v8_0_set_fault_enable_default - update VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gmc_v7_0_set_fault_enable_default(struct amdgpu_device *adev,
+					      bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+}
+
+/**
+ * gmc_v7_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This sets up the TLBs, programs the page tables for VMID0,
+ * sets up the hw for VMIDs 1-15 which are allocated on
+ * demand, and sets up the global locations for the LDS, GDS,
+ * and GPUVM for FSA64 clients (CIK).
+ * Returns 0 for success, errors for failure.
+ */
+static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r, i;
+	u32 tmp;
+
+	if (adev->gart.robj == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_table_vram_pin(adev);
+	if (r)
+		return r;
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, EFFECTIVE_L2_QUEUE_SIZE, 7);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	WREG32(mmVM_L2_CNTL, tmp);
+	tmp = REG_SET_FIELD(0, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32(mmVM_L2_CNTL2, tmp);
+	tmp = RREG32(mmVM_L2_CNTL3);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 4);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 4);
+	WREG32(mmVM_L2_CNTL3, tmp);
+	/* setup context0 */
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
+			(u32)(adev->dummy_page.addr >> 12));
+	WREG32(mmVM_CONTEXT0_CNTL2, 0);
+	tmp = RREG32(mmVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	WREG32(mmVM_CONTEXT0_CNTL, tmp);
+
+	WREG32(0x575, 0);
+	WREG32(0x576, 0);
+	WREG32(0x577, 0);
+
+	/* empty context1-15 */
+	/* FIXME start with 4G, once using 2 level pt switch to full
+	 * vm size space
+	 */
+	/* set vm size, must be a multiple of 4 */
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
+	for (i = 1; i < 16; i++) {
+		if (i < 8)
+			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
+			       adev->gart.table_addr >> 12);
+		else
+			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
+			       adev->gart.table_addr >> 12);
+	}
+
+	/* enable context1-15 */
+	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
+	       (u32)(adev->dummy_page.addr >> 12));
+	WREG32(mmVM_CONTEXT1_CNTL2, 4);
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_BLOCK_SIZE,
+			    amdgpu_vm_block_size - 9);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		gmc_v7_0_set_fault_enable_default(adev, false);
+	else
+		gmc_v7_0_set_fault_enable_default(adev, true);
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		tmp = RREG32(mmCHUB_CONTROL);
+		tmp &= ~BYPASS_VM;
+		WREG32(mmCHUB_CONTROL, tmp);
+	}
+
+	gmc_v7_0_gart_flush_gpu_tlb(adev, 0);
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->mc.gtt_size >> 20),
+		 (unsigned long long)adev->gart.table_addr);
+	adev->gart.ready = true;
+	return 0;
+}
+
+static int gmc_v7_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj) {
+		WARN(1, "R600 PCIE GART already initialized\n");
+		return 0;
+	}
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+/**
+ * gmc_v7_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table (CIK).
+ */
+static void gmc_v7_0_gart_disable(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* Disable all tables */
+	WREG32(mmVM_CONTEXT0_CNTL, 0);
+	WREG32(mmVM_CONTEXT1_CNTL, 0);
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32(mmVM_L2_CNTL, tmp);
+	WREG32(mmVM_L2_CNTL2, 0);
+	amdgpu_gart_table_vram_unpin(adev);
+}
+
+/**
+ * gmc_v7_0_gart_fini - vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the driver GART/VM setup (CIK).
+ */
+static void gmc_v7_0_gart_fini(struct amdgpu_device *adev)
+{
+	amdgpu_gart_table_vram_free(adev);
+	amdgpu_gart_fini(adev);
+}
+
+/*
+ * vm
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+/**
+ * gmc_v7_0_vm_init - cik vm init callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Inits cik specific vm parameters (number of VMs, base of vram for
+ * VMIDs 1-15) (CIK).
+ * Returns 0 for success.
+ */
+static int gmc_v7_0_vm_init(struct amdgpu_device *adev)
+{
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.nvm = AMDGPU_NUM_OF_VMIDS;
+
+	/* base offset of vram pages */
+	if (adev->flags & AMD_IS_APU) {
+		u64 tmp = RREG32(mmMC_VM_FB_OFFSET);
+		tmp <<= 22;
+		adev->vm_manager.vram_base_offset = tmp;
+	} else
+		adev->vm_manager.vram_base_offset = 0;
+
+	return 0;
+}
+
+/**
+ * gmc_v7_0_vm_fini - cik vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down any asic specific VM setup (CIK).
+ */
+static void gmc_v7_0_vm_fini(struct amdgpu_device *adev)
+{
+}
+
+/**
+ * gmc_v7_0_vm_decode_fault - print human readable fault info
+ *
+ * @adev: amdgpu_device pointer
+ * @status: VM_CONTEXT1_PROTECTION_FAULT_STATUS register value
+ * @addr: VM_CONTEXT1_PROTECTION_FAULT_ADDR register value
+ *
+ * Print human readable fault information (CIK).
+ */
+static void gmc_v7_0_vm_decode_fault(struct amdgpu_device *adev,
+				     u32 status, u32 addr, u32 mc_client)
+{
+	u32 mc_id;
+	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
+	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
+		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+
+	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			      MEMORY_CLIENT_ID);
+
+	printk("VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, addr,
+	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     MEMORY_CLIENT_RW) ?
+	       "write" : "read", block, mc_client, mc_id);
+}
+
+
+static const u32 mc_cg_registers[] = {
+	mmMC_HUB_MISC_HUB_CG,
+	mmMC_HUB_MISC_SIP_CG,
+	mmMC_HUB_MISC_VM_CG,
+	mmMC_XPB_CLK_GAT,
+	mmATC_MISC_CG,
+	mmMC_CITF_MISC_WR_CG,
+	mmMC_CITF_MISC_RD_CG,
+	mmMC_CITF_MISC_VM_CG,
+	mmVM_L2_CG,
+};
+
+static const u32 mc_cg_ls_en[] = {
+	MC_HUB_MISC_HUB_CG__MEM_LS_ENABLE_MASK,
+	MC_HUB_MISC_SIP_CG__MEM_LS_ENABLE_MASK,
+	MC_HUB_MISC_VM_CG__MEM_LS_ENABLE_MASK,
+	MC_XPB_CLK_GAT__MEM_LS_ENABLE_MASK,
+	ATC_MISC_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_WR_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_RD_CG__MEM_LS_ENABLE_MASK,
+	MC_CITF_MISC_VM_CG__MEM_LS_ENABLE_MASK,
+	VM_L2_CG__MEM_LS_ENABLE_MASK,
+};
+
+static const u32 mc_cg_en[] = {
+	MC_HUB_MISC_HUB_CG__ENABLE_MASK,
+	MC_HUB_MISC_SIP_CG__ENABLE_MASK,
+	MC_HUB_MISC_VM_CG__ENABLE_MASK,
+	MC_XPB_CLK_GAT__ENABLE_MASK,
+	ATC_MISC_CG__ENABLE_MASK,
+	MC_CITF_MISC_WR_CG__ENABLE_MASK,
+	MC_CITF_MISC_RD_CG__ENABLE_MASK,
+	MC_CITF_MISC_VM_CG__ENABLE_MASK,
+	VM_L2_CG__ENABLE_MASK,
+};
+
+static void gmc_v7_0_enable_mc_ls(struct amdgpu_device *adev,
+				  bool enable)
+{
+	int i;
+	u32 orig, data;
+
+	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
+		orig = data = RREG32(mc_cg_registers[i]);
+		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_LS))
+			data |= mc_cg_ls_en[i];
+		else
+			data &= ~mc_cg_ls_en[i];
+		if (data != orig)
+			WREG32(mc_cg_registers[i], data);
+	}
+}
+
+static void gmc_v7_0_enable_mc_mgcg(struct amdgpu_device *adev,
+				    bool enable)
+{
+	int i;
+	u32 orig, data;
+
+	for (i = 0; i < ARRAY_SIZE(mc_cg_registers); i++) {
+		orig = data = RREG32(mc_cg_registers[i]);
+		if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_MC_MGCG))
+			data |= mc_cg_en[i];
+		else
+			data &= ~mc_cg_en[i];
+		if (data != orig)
+			WREG32(mc_cg_registers[i], data);
+	}
+}
+
+static void gmc_v7_0_enable_bif_mgls(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32_PCIE(ixPCIE_CNTL2);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_BIF_LS)) {
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 1);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 1);
+	} else {
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, MST_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, REPLAY_MEM_LS_EN, 0);
+		data = REG_SET_FIELD(data, PCIE_CNTL2, SLV_MEM_AGGRESSIVE_LS_EN, 0);
+	}
+
+	if (orig != data)
+		WREG32_PCIE(ixPCIE_CNTL2, data);
+}
+
+static void gmc_v7_0_enable_hdp_mgcg(struct amdgpu_device *adev,
+				     bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32(mmHDP_HOST_PATH_CNTL);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_MGCG))
+		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 0);
+	else
+		data = REG_SET_FIELD(data, HDP_HOST_PATH_CNTL, CLOCK_GATING_DIS, 1);
+
+	if (orig != data)
+		WREG32(mmHDP_HOST_PATH_CNTL, data);
+}
+
+static void gmc_v7_0_enable_hdp_ls(struct amdgpu_device *adev,
+				   bool enable)
+{
+	u32 orig, data;
+
+	orig = data = RREG32(mmHDP_MEM_POWER_LS);
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_HDP_LS))
+		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 1);
+	else
+		data = REG_SET_FIELD(data, HDP_MEM_POWER_LS, LS_ENABLE, 0);
+
+	if (orig != data)
+		WREG32(mmHDP_MEM_POWER_LS, data);
+}
+
+static int gmc_v7_0_convert_vram_type(int mc_seq_vram_type)
+{
+	switch (mc_seq_vram_type) {
+	case MC_SEQ_MISC0__MT__GDDR1:
+		return AMDGPU_VRAM_TYPE_GDDR1;
+	case MC_SEQ_MISC0__MT__DDR2:
+		return AMDGPU_VRAM_TYPE_DDR2;
+	case MC_SEQ_MISC0__MT__GDDR3:
+		return AMDGPU_VRAM_TYPE_GDDR3;
+	case MC_SEQ_MISC0__MT__GDDR4:
+		return AMDGPU_VRAM_TYPE_GDDR4;
+	case MC_SEQ_MISC0__MT__GDDR5:
+		return AMDGPU_VRAM_TYPE_GDDR5;
+	case MC_SEQ_MISC0__MT__HBM:
+		return AMDGPU_VRAM_TYPE_HBM;
+	case MC_SEQ_MISC0__MT__DDR3:
+		return AMDGPU_VRAM_TYPE_DDR3;
+	default:
+		return AMDGPU_VRAM_TYPE_UNKNOWN;
+	}
+}
+
+static int gmc_v7_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v7_0_set_gart_funcs(adev);
+	gmc_v7_0_set_irq_funcs(adev);
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+	} else {
+		u32 tmp = RREG32(mmMC_SEQ_MISC0);
+		tmp &= MC_SEQ_MISC0__MT__MASK;
+		adev->mc.vram_type = gmc_v7_0_convert_vram_type(tmp);
+	}
+
+	return 0;
+}
+
+static int gmc_v7_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
+}
+
+static int gmc_v7_0_sw_init(void *handle)
+{
+	int r;
+	int dma_bits;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_gem_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, 146, &adev->mc.vm_fault);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, 147, &adev->mc.vm_fault);
+	if (r)
+		return r;
+
+	/* Adjust VM size here.
+	 * Currently set to 4GB ((1 << 20) 4k pages).
+	 * Max GPUVM size for cayman and SI is 40 bits.
+	 */
+	adev->vm_manager.max_pfn = amdgpu_vm_size << 18;
+
+	/* Set the internal MC address mask
+	 * This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->mc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+
+	/* set DMA mask + need_dma32 flags.
+	 * PCIE - can handle 40-bits.
+	 * IGP - can handle 40-bits
+	 * PCI - dma32 for legacy pci gart, 40 bits on newer asics
+	 */
+	adev->need_dma32 = false;
+	dma_bits = adev->need_dma32 ? 32 : 40;
+	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		adev->need_dma32 = true;
+		dma_bits = 32;
+		printk(KERN_WARNING "amdgpu: No suitable DMA available.\n");
+	}
+	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
+		printk(KERN_WARNING "amdgpu: No coherent DMA available.\n");
+	}
+
+	r = gmc_v7_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load mc firmware!\n");
+		return r;
+	}
+
+	r = gmc_v7_0_mc_init(adev);
+	if (r)
+		return r;
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v7_0_gart_init(adev);
+	if (r)
+		return r;
+
+	if (!adev->vm_manager.enabled) {
+		r = gmc_v7_0_vm_init(adev);
+		if (r) {
+			dev_err(adev->dev, "vm manager initialization failed (%d).\n", r);
+			return r;
+		}
+		adev->vm_manager.enabled = true;
+	}
+
+	return r;
+}
+
+static int gmc_v7_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->vm_manager.enabled) {
+		amdgpu_vm_manager_fini(adev);
+		gmc_v7_0_vm_fini(adev);
+		adev->vm_manager.enabled = false;
+	}
+	gmc_v7_0_gart_fini(adev);
+	amdgpu_gem_fini(adev);
+	amdgpu_bo_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v7_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v7_0_init_golden_registers(adev);
+
+	gmc_v7_0_mc_program(adev);
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = gmc_v7_0_mc_load_microcode(adev);
+		if (r) {
+			DRM_ERROR("Failed to load MC firmware!\n");
+			return r;
+		}
+	}
+
+	r = gmc_v7_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gmc_v7_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
+	gmc_v7_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v7_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->vm_manager.enabled) {
+		amdgpu_vm_manager_fini(adev);
+		gmc_v7_0_vm_fini(adev);
+		adev->vm_manager.enabled = false;
+	}
+	gmc_v7_0_hw_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v7_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gmc_v7_0_hw_init(adev);
+	if (r)
+		return r;
+
+	if (!adev->vm_manager.enabled) {
+		r = gmc_v7_0_vm_init(adev);
+		if (r) {
+			dev_err(adev->dev, "vm manager initialization failed (%d).\n", r);
+			return r;
+		}
+		adev->vm_manager.enabled = true;
+	}
+
+	return r;
+}
+
+static bool gmc_v7_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK | SRBM_STATUS__VMC_BUSY_MASK))
+		return false;
+
+	return true;
+}
+
+static int gmc_v7_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__MCB_BUSY_MASK |
+					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+					       SRBM_STATUS__MCC_BUSY_MASK |
+					       SRBM_STATUS__MCD_BUSY_MASK |
+					       SRBM_STATUS__VMC_BUSY_MASK);
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+
+}
+
+static void gmc_v7_0_print_status(void *handle)
+{
+	int i, j;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "GMC 8.x registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS));
+	dev_info(adev->dev, "  MC_VM_MX_L1_TLB_CNTL=0x%08X\n",
+		 RREG32(mmMC_VM_MX_L1_TLB_CNTL));
+	dev_info(adev->dev, "  VM_L2_CNTL=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL));
+	dev_info(adev->dev, "  VM_L2_CNTL2=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL2));
+	dev_info(adev->dev, "  VM_L2_CNTL3=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL3));
+	dev_info(adev->dev, "  VM_CONTEXT0_PAGE_TABLE_START_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT0_PAGE_TABLE_END_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT0_CNTL2=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_CNTL2));
+	dev_info(adev->dev, "  VM_CONTEXT0_CNTL=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_CNTL));
+	dev_info(adev->dev, "  0x15D4=0x%08X\n",
+		 RREG32(0x575));
+	dev_info(adev->dev, "  0x15D8=0x%08X\n",
+		 RREG32(0x576));
+	dev_info(adev->dev, "  0x15DC=0x%08X\n",
+		 RREG32(0x577));
+	dev_info(adev->dev, "  VM_CONTEXT1_PAGE_TABLE_START_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PAGE_TABLE_END_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_CNTL2=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_CNTL2));
+	dev_info(adev->dev, "  VM_CONTEXT1_CNTL=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_CNTL));
+	for (i = 0; i < 16; i++) {
+		if (i < 8)
+			dev_info(adev->dev, "  VM_CONTEXT%d_PAGE_TABLE_BASE_ADDR=0x%08X\n",
+				 i, RREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i));
+		else
+			dev_info(adev->dev, "  VM_CONTEXT%d_PAGE_TABLE_BASE_ADDR=0x%08X\n",
+				 i, RREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8));
+	}
+	dev_info(adev->dev, "  MC_VM_SYSTEM_APERTURE_LOW_ADDR=0x%08X\n",
+		 RREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR));
+	dev_info(adev->dev, "  MC_VM_SYSTEM_APERTURE_HIGH_ADDR=0x%08X\n",
+		 RREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR));
+	dev_info(adev->dev, "  MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR=0x%08X\n",
+		 RREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR));
+	dev_info(adev->dev, "  MC_VM_FB_LOCATION=0x%08X\n",
+		 RREG32(mmMC_VM_FB_LOCATION));
+	dev_info(adev->dev, "  MC_VM_AGP_BASE=0x%08X\n",
+		 RREG32(mmMC_VM_AGP_BASE));
+	dev_info(adev->dev, "  MC_VM_AGP_TOP=0x%08X\n",
+		 RREG32(mmMC_VM_AGP_TOP));
+	dev_info(adev->dev, "  MC_VM_AGP_BOT=0x%08X\n",
+		 RREG32(mmMC_VM_AGP_BOT));
+
+	if (adev->asic_type == CHIP_KAVERI) {
+		dev_info(adev->dev, "  CHUB_CONTROL=0x%08X\n",
+			 RREG32(mmCHUB_CONTROL));
+	}
+
+	dev_info(adev->dev, "  HDP_REG_COHERENCY_FLUSH_CNTL=0x%08X\n",
+		 RREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL));
+	dev_info(adev->dev, "  HDP_NONSURFACE_BASE=0x%08X\n",
+		 RREG32(mmHDP_NONSURFACE_BASE));
+	dev_info(adev->dev, "  HDP_NONSURFACE_INFO=0x%08X\n",
+		 RREG32(mmHDP_NONSURFACE_INFO));
+	dev_info(adev->dev, "  HDP_NONSURFACE_SIZE=0x%08X\n",
+		 RREG32(mmHDP_NONSURFACE_SIZE));
+	dev_info(adev->dev, "  HDP_MISC_CNTL=0x%08X\n",
+		 RREG32(mmHDP_MISC_CNTL));
+	dev_info(adev->dev, "  HDP_HOST_PATH_CNTL=0x%08X\n",
+		 RREG32(mmHDP_HOST_PATH_CNTL));
+
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		dev_info(adev->dev, "  %d:\n", i);
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb05 + j, RREG32(0xb05 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb06 + j, RREG32(0xb06 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb07 + j, RREG32(0xb07 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb08 + j, RREG32(0xb08 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb09 + j, RREG32(0xb09 + j));
+	}
+
+	dev_info(adev->dev, "  BIF_FB_EN=0x%08X\n",
+		 RREG32(mmBIF_FB_EN));
+}
+
+static int gmc_v7_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_mode_mc_save save;
+	u32 srbm_soft_reset = 0;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK)) {
+		if (!(adev->flags & AMD_IS_APU))
+			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
+	}
+
+	if (srbm_soft_reset) {
+		gmc_v7_0_print_status((void *)adev);
+
+		gmc_v7_0_mc_stop(adev, &save);
+		if (gmc_v7_0_wait_for_idle(adev)) {
+			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
+		}
+
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		gmc_v7_0_mc_resume(adev, &save);
+		udelay(50);
+
+		gmc_v7_0_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int gmc_v7_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+	u32 bits = (VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK);
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u32 addr, status, mc_client;
+
+	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
+	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
+	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
+	/* reset addr and status */
+	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
+
+	if (!addr && !status)
+		return 0;
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
+		gmc_v7_0_set_fault_enable_default(adev, false);
+
+	dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+		entry->src_id, entry->src_data);
+	dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+		addr);
+	dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+		status);
+	gmc_v7_0_vm_decode_fault(adev, status, addr, mc_client);
+
+	return 0;
+}
+
+static int gmc_v7_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		gmc_v7_0_enable_mc_mgcg(adev, gate);
+		gmc_v7_0_enable_mc_ls(adev, gate);
+	}
+	gmc_v7_0_enable_bif_mgls(adev, gate);
+	gmc_v7_0_enable_hdp_mgcg(adev, gate);
+	gmc_v7_0_enable_hdp_ls(adev, gate);
+
+	return 0;
+}
+
+static int gmc_v7_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs gmc_v7_0_ip_funcs = {
+	.early_init = gmc_v7_0_early_init,
+	.late_init = gmc_v7_0_late_init,
+	.sw_init = gmc_v7_0_sw_init,
+	.sw_fini = gmc_v7_0_sw_fini,
+	.hw_init = gmc_v7_0_hw_init,
+	.hw_fini = gmc_v7_0_hw_fini,
+	.suspend = gmc_v7_0_suspend,
+	.resume = gmc_v7_0_resume,
+	.is_idle = gmc_v7_0_is_idle,
+	.wait_for_idle = gmc_v7_0_wait_for_idle,
+	.soft_reset = gmc_v7_0_soft_reset,
+	.print_status = gmc_v7_0_print_status,
+	.set_clockgating_state = gmc_v7_0_set_clockgating_state,
+	.set_powergating_state = gmc_v7_0_set_powergating_state,
+};
+
+static const struct amdgpu_gart_funcs gmc_v7_0_gart_funcs = {
+	.flush_gpu_tlb = gmc_v7_0_gart_flush_gpu_tlb,
+	.set_pte_pde = gmc_v7_0_gart_set_pte_pde,
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v7_0_irq_funcs = {
+	.set = gmc_v7_0_vm_fault_interrupt_state,
+	.process = gmc_v7_0_process_interrupt,
+};
+
+static void gmc_v7_0_set_gart_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gart.gart_funcs == NULL)
+		adev->gart.gart_funcs = &gmc_v7_0_gart_funcs;
+}
+
+static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->mc.vm_fault.num_types = 1;
+	adev->mc.vm_fault.funcs = &gmc_v7_0_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h
new file mode 100644
index 000000000..36fcbbc46
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V7_0_H__
+#define __GMC_V7_0_H__
+
+extern const struct amd_ip_funcs gmc_v7_0_ip_funcs;
+
+/* XXX these shouldn't be exported */
+void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
+		      struct amdgpu_mode_mc_save *save);
+void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
+			struct amdgpu_mode_mc_save *save);
+int gmc_v7_0_mc_wait_for_idle(struct amdgpu_device *adev);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
new file mode 100644
index 000000000..ba4ad00ba
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -0,0 +1,1348 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "gmc_v8_0.h"
+#include "amdgpu_ucode.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "vid.h"
+#include "vi.h"
+
+
+static void gmc_v8_0_set_gart_funcs(struct amdgpu_device *adev);
+static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/tonga_mc.bin");
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmMC_ARB_WTM_GRPWT_RD, 0x00000003, 0x00000000,
+	mmMC_HUB_RDREQ_DMIF_LIMIT, 0x0000007f, 0x00000028,
+	mmMC_HUB_WDP_UMC, 0x00007fb6, 0x00000991,
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmVM_PRT_APERTURE0_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE1_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE2_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+	mmVM_PRT_APERTURE3_LOW_ADDR, 0x0fffffff, 0x0fffffff,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
+};
+
+
+static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_program_register_sequence(adev,
+						 fiji_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_fiji_a10,
+						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		break;
+	case CHIP_TONGA:
+		amdgpu_program_register_sequence(adev,
+						 tonga_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_tonga_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_program_register_sequence(adev,
+						 cz_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		break;
+	case CHIP_STONEY:
+		amdgpu_program_register_sequence(adev,
+						 stoney_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * gmc8_mc_wait_for_idle - wait for MC idle callback.
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Wait for the MC (memory controller) to be idle.
+ * (evergreen+).
+ * Returns 0 if the MC is idle, -1 if not.
+ */
+int gmc_v8_0_mc_wait_for_idle(struct amdgpu_device *adev)
+{
+	unsigned i;
+	u32 tmp;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__VMC_BUSY_MASK |
+					       SRBM_STATUS__MCB_BUSY_MASK |
+					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+					       SRBM_STATUS__MCC_BUSY_MASK |
+					       SRBM_STATUS__MCD_BUSY_MASK |
+					       SRBM_STATUS__VMC1_BUSY_MASK);
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -1;
+}
+
+void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
+		      struct amdgpu_mode_mc_save *save)
+{
+	u32 blackout;
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_stop_mc_access(adev, save);
+
+	amdgpu_asic_wait_for_mc_idle(adev);
+
+	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
+		/* Block CPU access */
+		WREG32(mmBIF_FB_EN, 0);
+		/* blackout the MC */
+		blackout = REG_SET_FIELD(blackout,
+					 MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 1);
+		WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout);
+	}
+	/* wait for the MC to settle */
+	udelay(100);
+}
+
+void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
+			struct amdgpu_mode_mc_save *save)
+{
+	u32 tmp;
+
+	/* unblackout the MC */
+	tmp = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE, 0);
+	WREG32(mmMC_SHARED_BLACKOUT_CNTL, tmp);
+	/* allow CPU access */
+	tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
+	tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
+	WREG32(mmBIF_FB_EN, tmp);
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_resume_mc_access(adev, save);
+}
+
+/**
+ * gmc_v8_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v8_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		chip_name = "tonga";
+		break;
+	case CHIP_FIJI:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		return 0;
+	default: BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mc.bin", chip_name);
+	err = request_firmware(&adev->mc.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->mc.fw);
+
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "mc: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		release_firmware(adev->mc.fw);
+		adev->mc.fw = NULL;
+	}
+	return err;
+}
+
+/**
+ * gmc_v8_0_mc_load_microcode - load MC ucode into the hw
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Load the GDDR MC ucode into the hw (CIK).
+ * Returns 0 on success, error on failure.
+ */
+static int gmc_v8_0_mc_load_microcode(struct amdgpu_device *adev)
+{
+	const struct mc_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data = NULL;
+	const __le32 *io_mc_regs = NULL;
+	u32 running, blackout = 0;
+	int i, ucode_size, regs_size;
+
+	if (!adev->mc.fw)
+		return -EINVAL;
+
+	hdr = (const struct mc_firmware_header_v1_0 *)adev->mc.fw->data;
+	amdgpu_ucode_print_mc_hdr(&hdr->header);
+
+	adev->mc.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	regs_size = le32_to_cpu(hdr->io_debug_size_bytes) / (4 * 2);
+	io_mc_regs = (const __le32 *)
+		(adev->mc.fw->data + le32_to_cpu(hdr->io_debug_array_offset_bytes));
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+	fw_data = (const __le32 *)
+		(adev->mc.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN);
+
+	if (running == 0) {
+		if (running) {
+			blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
+			WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1);
+		}
+
+		/* reset the engine and set to writable */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010);
+
+		/* load mc io regs */
+		for (i = 0; i < regs_size; i++) {
+			WREG32(mmMC_SEQ_IO_DEBUG_INDEX, le32_to_cpup(io_mc_regs++));
+			WREG32(mmMC_SEQ_IO_DEBUG_DATA, le32_to_cpup(io_mc_regs++));
+		}
+		/* load the MC ucode */
+		for (i = 0; i < ucode_size; i++)
+			WREG32(mmMC_SEQ_SUP_PGM, le32_to_cpup(fw_data++));
+
+		/* put the engine back into the active state */
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000004);
+		WREG32(mmMC_SEQ_SUP_CNTL, 0x00000001);
+
+		/* wait for training to complete */
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D0))
+				break;
+			udelay(1);
+		}
+		for (i = 0; i < adev->usec_timeout; i++) {
+			if (REG_GET_FIELD(RREG32(mmMC_SEQ_TRAIN_WAKEUP_CNTL),
+					  MC_SEQ_TRAIN_WAKEUP_CNTL, TRAIN_DONE_D1))
+				break;
+			udelay(1);
+		}
+
+		if (running)
+			WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout);
+	}
+
+	return 0;
+}
+
+static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
+				       struct amdgpu_mc *mc)
+{
+	if (mc->mc_vram_size > 0xFFC0000000ULL) {
+		/* leave room for at least 1024M GTT */
+		dev_warn(adev->dev, "limiting VRAM\n");
+		mc->real_vram_size = 0xFFC0000000ULL;
+		mc->mc_vram_size = 0xFFC0000000ULL;
+	}
+	amdgpu_vram_location(adev, &adev->mc, 0);
+	adev->mc.gtt_base_align = 0;
+	amdgpu_gtt_location(adev, mc);
+}
+
+/**
+ * gmc_v8_0_mc_program - program the GPU memory controller
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the location of vram, gart, and AGP in the GPU's
+ * physical address space (CIK).
+ */
+static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_mc_save save;
+	u32 tmp;
+	int i, j;
+
+	/* Initialize HDP */
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		WREG32((0xb05 + j), 0x00000000);
+		WREG32((0xb06 + j), 0x00000000);
+		WREG32((0xb07 + j), 0x00000000);
+		WREG32((0xb08 + j), 0x00000000);
+		WREG32((0xb09 + j), 0x00000000);
+	}
+	WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
+
+	if (adev->mode_info.num_crtc)
+		amdgpu_display_set_vga_render_state(adev, false);
+
+	gmc_v8_0_mc_stop(adev, &save);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+	/* Update configuration */
+	WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
+	       adev->mc.vram_start >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
+	       adev->mc.vram_end >> 12);
+	WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
+	       adev->vram_scratch.gpu_addr >> 12);
+	tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
+	tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
+	WREG32(mmMC_VM_FB_LOCATION, tmp);
+	/* XXX double check these! */
+	WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
+	WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
+	WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
+	WREG32(mmMC_VM_AGP_BASE, 0);
+	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
+	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+	gmc_v8_0_mc_resume(adev, &save);
+
+	WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
+
+	tmp = RREG32(mmHDP_MISC_CNTL);
+	tmp = REG_SET_FIELD(tmp, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 1);
+	WREG32(mmHDP_MISC_CNTL, tmp);
+
+	tmp = RREG32(mmHDP_HOST_PATH_CNTL);
+	WREG32(mmHDP_HOST_PATH_CNTL, tmp);
+}
+
+/**
+ * gmc_v8_0_mc_init - initialize the memory controller driver params
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up the amount of vram, vram width, and decide how to place
+ * vram and gart within the GPU's physical address space (CIK).
+ * Returns 0 for success.
+ */
+static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	int chansize, numchan;
+
+	/* Get VRAM informations */
+	tmp = RREG32(mmMC_ARB_RAMCFG);
+	if (REG_GET_FIELD(tmp, MC_ARB_RAMCFG, CHANSIZE)) {
+		chansize = 64;
+	} else {
+		chansize = 32;
+	}
+	tmp = RREG32(mmMC_SHARED_CHMAP);
+	switch (REG_GET_FIELD(tmp, MC_SHARED_CHMAP, NOOFCHAN)) {
+	case 0:
+	default:
+		numchan = 1;
+		break;
+	case 1:
+		numchan = 2;
+		break;
+	case 2:
+		numchan = 4;
+		break;
+	case 3:
+		numchan = 8;
+		break;
+	case 4:
+		numchan = 3;
+		break;
+	case 5:
+		numchan = 6;
+		break;
+	case 6:
+		numchan = 10;
+		break;
+	case 7:
+		numchan = 12;
+		break;
+	case 8:
+		numchan = 16;
+		break;
+	}
+	adev->mc.vram_width = numchan * chansize;
+	/* Could aper size report 0 ? */
+	adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
+	adev->mc.aper_size = pci_resource_len(adev->pdev, 0);
+	/* size in MB on si */
+	adev->mc.mc_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
+	adev->mc.visible_vram_size = adev->mc.aper_size;
+
+	/* unless the user had overridden it, set the gart
+	 * size equal to the 1024 or vram, whichever is larger.
+	 */
+	if (amdgpu_gart_size == -1)
+		adev->mc.gtt_size = max((1024ULL << 20), adev->mc.mc_vram_size);
+	else
+		adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20;
+
+	gmc_v8_0_vram_gtt_location(adev, &adev->mc);
+
+	return 0;
+}
+
+/*
+ * GART
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+
+/**
+ * gmc_v8_0_gart_flush_gpu_tlb - gart tlb flush callback
+ *
+ * @adev: amdgpu_device pointer
+ * @vmid: vm instance to flush
+ *
+ * Flush the TLB for the requested page table (CIK).
+ */
+static void gmc_v8_0_gart_flush_gpu_tlb(struct amdgpu_device *adev,
+					uint32_t vmid)
+{
+	/* flush hdp cache */
+	WREG32(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
+
+	/* bits 0-15 are the VM contexts0-15 */
+	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
+}
+
+/**
+ * gmc_v8_0_gart_set_pte_pde - update the page tables using MMIO
+ *
+ * @adev: amdgpu_device pointer
+ * @cpu_pt_addr: cpu address of the page table
+ * @gpu_page_idx: entry in the page table to update
+ * @addr: dst addr to write into pte/pde
+ * @flags: access flags
+ *
+ * Update the page tables using the CPU.
+ */
+static int gmc_v8_0_gart_set_pte_pde(struct amdgpu_device *adev,
+				     void *cpu_pt_addr,
+				     uint32_t gpu_page_idx,
+				     uint64_t addr,
+				     uint32_t flags)
+{
+	void __iomem *ptr = (void *)cpu_pt_addr;
+	uint64_t value;
+
+	/*
+	 * PTE format on VI:
+	 * 63:40 reserved
+	 * 39:12 4k physical page base address
+	 * 11:7 fragment
+	 * 6 write
+	 * 5 read
+	 * 4 exe
+	 * 3 reserved
+	 * 2 snooped
+	 * 1 system
+	 * 0 valid
+	 *
+	 * PDE format on VI:
+	 * 63:59 block fragment size
+	 * 58:40 reserved
+	 * 39:1 physical base address of PTE
+	 * bits 5:1 must be 0.
+	 * 0 valid
+	 */
+	value = addr & 0x000000FFFFFFF000ULL;
+	value |= flags;
+	writeq(value, ptr + (gpu_page_idx * 8));
+
+	return 0;
+}
+
+/**
+ * gmc_v8_0_set_fault_enable_default - update VM fault handling
+ *
+ * @adev: amdgpu_device pointer
+ * @value: true redirects VM faults to the default page
+ */
+static void gmc_v8_0_set_fault_enable_default(struct amdgpu_device *adev,
+					      bool value)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL,
+			    EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+}
+
+/**
+ * gmc_v8_0_gart_enable - gart enable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This sets up the TLBs, programs the page tables for VMID0,
+ * sets up the hw for VMIDs 1-15 which are allocated on
+ * demand, and sets up the global locations for the LDS, GDS,
+ * and GPUVM for FSA64 clients (CIK).
+ * Returns 0 for success, errors for failure.
+ */
+static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
+{
+	int r, i;
+	u32 tmp;
+
+	if (adev->gart.robj == NULL) {
+		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
+		return -EINVAL;
+	}
+	r = amdgpu_gart_table_vram_pin(adev);
+	if (r)
+		return r;
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 1);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, EFFECTIVE_L2_QUEUE_SIZE, 7);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
+	WREG32(mmVM_L2_CNTL, tmp);
+	tmp = RREG32(mmVM_L2_CNTL2);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
+	WREG32(mmVM_L2_CNTL2, tmp);
+	tmp = RREG32(mmVM_L2_CNTL3);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY, 1);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 4);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 4);
+	WREG32(mmVM_L2_CNTL3, tmp);
+	/* XXX: set to enable PTE/PDE in system memory */
+	tmp = RREG32(mmVM_L2_CNTL4);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PDE_REQUEST_SNOOP, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT0_PTE_REQUEST_SNOOP, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PDE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PDE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PDE_REQUEST_SNOOP, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_PHYSICAL, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_SHARED, 0);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_SNOOP, 0);
+	WREG32(mmVM_L2_CNTL4, tmp);
+	/* setup context0 */
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
+	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
+			(u32)(adev->dummy_page.addr >> 12));
+	WREG32(mmVM_CONTEXT0_CNTL2, 0);
+	tmp = RREG32(mmVM_CONTEXT0_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT0_CNTL, RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	WREG32(mmVM_CONTEXT0_CNTL, tmp);
+
+	WREG32(mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR, 0);
+	WREG32(mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR, 0);
+	WREG32(mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET, 0);
+
+	/* empty context1-15 */
+	/* FIXME start with 4G, once using 2 level pt switch to full
+	 * vm size space
+	 */
+	/* set vm size, must be a multiple of 4 */
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR, 0);
+	WREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR, adev->vm_manager.max_pfn - 1);
+	for (i = 1; i < 16; i++) {
+		if (i < 8)
+			WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i,
+			       adev->gart.table_addr >> 12);
+		else
+			WREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8,
+			       adev->gart.table_addr >> 12);
+	}
+
+	/* enable context1-15 */
+	WREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR,
+	       (u32)(adev->dummy_page.addr >> 12));
+	WREG32(mmVM_CONTEXT1_CNTL2, 4);
+	tmp = RREG32(mmVM_CONTEXT1_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
+	tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, PAGE_TABLE_BLOCK_SIZE,
+			    amdgpu_vm_block_size - 9);
+	WREG32(mmVM_CONTEXT1_CNTL, tmp);
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS)
+		gmc_v8_0_set_fault_enable_default(adev, false);
+	else
+		gmc_v8_0_set_fault_enable_default(adev, true);
+
+	gmc_v8_0_gart_flush_gpu_tlb(adev, 0);
+	DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
+		 (unsigned)(adev->mc.gtt_size >> 20),
+		 (unsigned long long)adev->gart.table_addr);
+	adev->gart.ready = true;
+	return 0;
+}
+
+static int gmc_v8_0_gart_init(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (adev->gart.robj) {
+		WARN(1, "R600 PCIE GART already initialized\n");
+		return 0;
+	}
+	/* Initialize common gart structure */
+	r = amdgpu_gart_init(adev);
+	if (r)
+		return r;
+	adev->gart.table_size = adev->gart.num_gpu_pages * 8;
+	return amdgpu_gart_table_vram_alloc(adev);
+}
+
+/**
+ * gmc_v8_0_gart_disable - gart disable
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * This disables all VM page table (CIK).
+ */
+static void gmc_v8_0_gart_disable(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	/* Disable all tables */
+	WREG32(mmVM_CONTEXT0_CNTL, 0);
+	WREG32(mmVM_CONTEXT1_CNTL, 0);
+	/* Setup TLB control */
+	tmp = RREG32(mmMC_VM_MX_L1_TLB_CNTL);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_L1_FRAGMENT_PROCESSING, 0);
+	tmp = REG_SET_FIELD(tmp, MC_VM_MX_L1_TLB_CNTL, ENABLE_ADVANCED_DRIVER_MODEL, 0);
+	WREG32(mmMC_VM_MX_L1_TLB_CNTL, tmp);
+	/* Setup L2 cache */
+	tmp = RREG32(mmVM_L2_CNTL);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 0);
+	WREG32(mmVM_L2_CNTL, tmp);
+	WREG32(mmVM_L2_CNTL2, 0);
+	amdgpu_gart_table_vram_unpin(adev);
+}
+
+/**
+ * gmc_v8_0_gart_fini - vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tears down the driver GART/VM setup (CIK).
+ */
+static void gmc_v8_0_gart_fini(struct amdgpu_device *adev)
+{
+	amdgpu_gart_table_vram_free(adev);
+	amdgpu_gart_fini(adev);
+}
+
+/*
+ * vm
+ * VMID 0 is the physical GPU addresses as used by the kernel.
+ * VMIDs 1-15 are used for userspace clients and are handled
+ * by the amdgpu vm/hsa code.
+ */
+/**
+ * gmc_v8_0_vm_init - cik vm init callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Inits cik specific vm parameters (number of VMs, base of vram for
+ * VMIDs 1-15) (CIK).
+ * Returns 0 for success.
+ */
+static int gmc_v8_0_vm_init(struct amdgpu_device *adev)
+{
+	/*
+	 * number of VMs
+	 * VMID 0 is reserved for System
+	 * amdgpu graphics/compute will use VMIDs 1-7
+	 * amdkfd will use VMIDs 8-15
+	 */
+	adev->vm_manager.nvm = AMDGPU_NUM_OF_VMIDS;
+
+	/* base offset of vram pages */
+	if (adev->flags & AMD_IS_APU) {
+		u64 tmp = RREG32(mmMC_VM_FB_OFFSET);
+		tmp <<= 22;
+		adev->vm_manager.vram_base_offset = tmp;
+	} else
+		adev->vm_manager.vram_base_offset = 0;
+
+	return 0;
+}
+
+/**
+ * gmc_v8_0_vm_fini - cik vm fini callback
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Tear down any asic specific VM setup (CIK).
+ */
+static void gmc_v8_0_vm_fini(struct amdgpu_device *adev)
+{
+}
+
+/**
+ * gmc_v8_0_vm_decode_fault - print human readable fault info
+ *
+ * @adev: amdgpu_device pointer
+ * @status: VM_CONTEXT1_PROTECTION_FAULT_STATUS register value
+ * @addr: VM_CONTEXT1_PROTECTION_FAULT_ADDR register value
+ *
+ * Print human readable fault information (CIK).
+ */
+static void gmc_v8_0_vm_decode_fault(struct amdgpu_device *adev,
+				     u32 status, u32 addr, u32 mc_client)
+{
+	u32 mc_id;
+	u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID);
+	u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+					PROTECTIONS);
+	char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff,
+		(mc_client >> 8) & 0xff, mc_client & 0xff, 0 };
+
+	mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			      MEMORY_CLIENT_ID);
+
+	printk("VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n",
+	       protections, vmid, addr,
+	       REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS,
+			     MEMORY_CLIENT_RW) ?
+	       "write" : "read", block, mc_client, mc_id);
+}
+
+static int gmc_v8_0_convert_vram_type(int mc_seq_vram_type)
+{
+	switch (mc_seq_vram_type) {
+	case MC_SEQ_MISC0__MT__GDDR1:
+		return AMDGPU_VRAM_TYPE_GDDR1;
+	case MC_SEQ_MISC0__MT__DDR2:
+		return AMDGPU_VRAM_TYPE_DDR2;
+	case MC_SEQ_MISC0__MT__GDDR3:
+		return AMDGPU_VRAM_TYPE_GDDR3;
+	case MC_SEQ_MISC0__MT__GDDR4:
+		return AMDGPU_VRAM_TYPE_GDDR4;
+	case MC_SEQ_MISC0__MT__GDDR5:
+		return AMDGPU_VRAM_TYPE_GDDR5;
+	case MC_SEQ_MISC0__MT__HBM:
+		return AMDGPU_VRAM_TYPE_HBM;
+	case MC_SEQ_MISC0__MT__DDR3:
+		return AMDGPU_VRAM_TYPE_DDR3;
+	default:
+		return AMDGPU_VRAM_TYPE_UNKNOWN;
+	}
+}
+
+static int gmc_v8_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v8_0_set_gart_funcs(adev);
+	gmc_v8_0_set_irq_funcs(adev);
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
+	} else {
+		u32 tmp = RREG32(mmMC_SEQ_MISC0);
+		tmp &= MC_SEQ_MISC0__MT__MASK;
+		adev->mc.vram_type = gmc_v8_0_convert_vram_type(tmp);
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_late_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return amdgpu_irq_get(adev, &adev->mc.vm_fault, 0);
+}
+
+static int gmc_v8_0_sw_init(void *handle)
+{
+	int r;
+	int dma_bits;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_gem_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, 146, &adev->mc.vm_fault);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_add_id(adev, 147, &adev->mc.vm_fault);
+	if (r)
+		return r;
+
+	/* Adjust VM size here.
+	 * Currently set to 4GB ((1 << 20) 4k pages).
+	 * Max GPUVM size for cayman and SI is 40 bits.
+	 */
+	adev->vm_manager.max_pfn = amdgpu_vm_size << 18;
+
+	/* Set the internal MC address mask
+	 * This is the max address of the GPU's
+	 * internal address space.
+	 */
+	adev->mc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+
+	/* set DMA mask + need_dma32 flags.
+	 * PCIE - can handle 40-bits.
+	 * IGP - can handle 40-bits
+	 * PCI - dma32 for legacy pci gart, 40 bits on newer asics
+	 */
+	adev->need_dma32 = false;
+	dma_bits = adev->need_dma32 ? 32 : 40;
+	r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		adev->need_dma32 = true;
+		dma_bits = 32;
+		printk(KERN_WARNING "amdgpu: No suitable DMA available.\n");
+	}
+	r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
+	if (r) {
+		pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
+		printk(KERN_WARNING "amdgpu: No coherent DMA available.\n");
+	}
+
+	r = gmc_v8_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load mc firmware!\n");
+		return r;
+	}
+
+	r = gmc_v8_0_mc_init(adev);
+	if (r)
+		return r;
+
+	/* Memory manager */
+	r = amdgpu_bo_init(adev);
+	if (r)
+		return r;
+
+	r = gmc_v8_0_gart_init(adev);
+	if (r)
+		return r;
+
+	if (!adev->vm_manager.enabled) {
+		r = gmc_v8_0_vm_init(adev);
+		if (r) {
+			dev_err(adev->dev, "vm manager initialization failed (%d).\n", r);
+			return r;
+		}
+		adev->vm_manager.enabled = true;
+	}
+
+	return r;
+}
+
+static int gmc_v8_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->vm_manager.enabled) {
+		amdgpu_vm_manager_fini(adev);
+		gmc_v8_0_vm_fini(adev);
+		adev->vm_manager.enabled = false;
+	}
+	gmc_v8_0_gart_fini(adev);
+	amdgpu_gem_fini(adev);
+	amdgpu_bo_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v8_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	gmc_v8_0_init_golden_registers(adev);
+
+	gmc_v8_0_mc_program(adev);
+
+	if (adev->asic_type == CHIP_TONGA) {
+		r = gmc_v8_0_mc_load_microcode(adev);
+		if (r) {
+			DRM_ERROR("Failed to load MC firmware!\n");
+			return r;
+		}
+	}
+
+	r = gmc_v8_0_gart_enable(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int gmc_v8_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_put(adev, &adev->mc.vm_fault, 0);
+	gmc_v8_0_gart_disable(adev);
+
+	return 0;
+}
+
+static int gmc_v8_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->vm_manager.enabled) {
+		amdgpu_vm_manager_fini(adev);
+		gmc_v8_0_vm_fini(adev);
+		adev->vm_manager.enabled = false;
+	}
+	gmc_v8_0_hw_fini(adev);
+
+	return 0;
+}
+
+static int gmc_v8_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = gmc_v8_0_hw_init(adev);
+	if (r)
+		return r;
+
+	if (!adev->vm_manager.enabled) {
+		r = gmc_v8_0_vm_init(adev);
+		if (r) {
+			dev_err(adev->dev, "vm manager initialization failed (%d).\n", r);
+			return r;
+		}
+		adev->vm_manager.enabled = true;
+	}
+
+	return r;
+}
+
+static bool gmc_v8_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK | SRBM_STATUS__VMC_BUSY_MASK))
+		return false;
+
+	return true;
+}
+
+static int gmc_v8_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__MCB_BUSY_MASK |
+					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+					       SRBM_STATUS__MCC_BUSY_MASK |
+					       SRBM_STATUS__MCD_BUSY_MASK |
+					       SRBM_STATUS__VMC_BUSY_MASK |
+					       SRBM_STATUS__VMC1_BUSY_MASK);
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+
+}
+
+static void gmc_v8_0_print_status(void *handle)
+{
+	int i, j;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "GMC 8.x registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS));
+	dev_info(adev->dev, "  MC_VM_MX_L1_TLB_CNTL=0x%08X\n",
+		 RREG32(mmMC_VM_MX_L1_TLB_CNTL));
+	dev_info(adev->dev, "  VM_L2_CNTL=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL));
+	dev_info(adev->dev, "  VM_L2_CNTL2=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL2));
+	dev_info(adev->dev, "  VM_L2_CNTL3=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL3));
+	dev_info(adev->dev, "  VM_L2_CNTL4=0x%08X\n",
+		 RREG32(mmVM_L2_CNTL4));
+	dev_info(adev->dev, "  VM_CONTEXT0_PAGE_TABLE_START_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT0_PAGE_TABLE_END_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT0_CNTL2=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_CNTL2));
+	dev_info(adev->dev, "  VM_CONTEXT0_CNTL=0x%08X\n",
+		 RREG32(mmVM_CONTEXT0_CNTL));
+	dev_info(adev->dev, "  VM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR=0x%08X\n",
+		 RREG32(mmVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR));
+	dev_info(adev->dev, "  VM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR=0x%08X\n",
+		 RREG32(mmVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR));
+	dev_info(adev->dev, "  mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET=0x%08X\n",
+		 RREG32(mmVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET));
+	dev_info(adev->dev, "  VM_CONTEXT1_PAGE_TABLE_START_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PAGE_TABLE_START_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PAGE_TABLE_END_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PAGE_TABLE_END_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_CNTL2=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_CNTL2));
+	dev_info(adev->dev, "  VM_CONTEXT1_CNTL=0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_CNTL));
+	for (i = 0; i < 16; i++) {
+		if (i < 8)
+			dev_info(adev->dev, "  VM_CONTEXT%d_PAGE_TABLE_BASE_ADDR=0x%08X\n",
+				 i, RREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + i));
+		else
+			dev_info(adev->dev, "  VM_CONTEXT%d_PAGE_TABLE_BASE_ADDR=0x%08X\n",
+				 i, RREG32(mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + i - 8));
+	}
+	dev_info(adev->dev, "  MC_VM_SYSTEM_APERTURE_LOW_ADDR=0x%08X\n",
+		 RREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR));
+	dev_info(adev->dev, "  MC_VM_SYSTEM_APERTURE_HIGH_ADDR=0x%08X\n",
+		 RREG32(mmMC_VM_SYSTEM_APERTURE_HIGH_ADDR));
+	dev_info(adev->dev, "  MC_VM_SYSTEM_APERTURE_DEFAULT_ADDR=0x%08X\n",
+		 RREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR));
+	dev_info(adev->dev, "  MC_VM_FB_LOCATION=0x%08X\n",
+		 RREG32(mmMC_VM_FB_LOCATION));
+	dev_info(adev->dev, "  MC_VM_AGP_BASE=0x%08X\n",
+		 RREG32(mmMC_VM_AGP_BASE));
+	dev_info(adev->dev, "  MC_VM_AGP_TOP=0x%08X\n",
+		 RREG32(mmMC_VM_AGP_TOP));
+	dev_info(adev->dev, "  MC_VM_AGP_BOT=0x%08X\n",
+		 RREG32(mmMC_VM_AGP_BOT));
+
+	dev_info(adev->dev, "  HDP_REG_COHERENCY_FLUSH_CNTL=0x%08X\n",
+		 RREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL));
+	dev_info(adev->dev, "  HDP_NONSURFACE_BASE=0x%08X\n",
+		 RREG32(mmHDP_NONSURFACE_BASE));
+	dev_info(adev->dev, "  HDP_NONSURFACE_INFO=0x%08X\n",
+		 RREG32(mmHDP_NONSURFACE_INFO));
+	dev_info(adev->dev, "  HDP_NONSURFACE_SIZE=0x%08X\n",
+		 RREG32(mmHDP_NONSURFACE_SIZE));
+	dev_info(adev->dev, "  HDP_MISC_CNTL=0x%08X\n",
+		 RREG32(mmHDP_MISC_CNTL));
+	dev_info(adev->dev, "  HDP_HOST_PATH_CNTL=0x%08X\n",
+		 RREG32(mmHDP_HOST_PATH_CNTL));
+
+	for (i = 0, j = 0; i < 32; i++, j += 0x6) {
+		dev_info(adev->dev, "  %d:\n", i);
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb05 + j, RREG32(0xb05 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb06 + j, RREG32(0xb06 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb07 + j, RREG32(0xb07 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb08 + j, RREG32(0xb08 + j));
+		dev_info(adev->dev, "  0x%04X=0x%08X\n",
+			 0xb09 + j, RREG32(0xb09 + j));
+	}
+
+	dev_info(adev->dev, "  BIF_FB_EN=0x%08X\n",
+		 RREG32(mmBIF_FB_EN));
+}
+
+static int gmc_v8_0_soft_reset(void *handle)
+{
+	struct amdgpu_mode_mc_save save;
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+						SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK)) {
+		if (!(adev->flags & AMD_IS_APU))
+			srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset,
+							SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
+	}
+
+	if (srbm_soft_reset) {
+		gmc_v8_0_print_status((void *)adev);
+
+		gmc_v8_0_mc_stop(adev, &save);
+		if (gmc_v8_0_wait_for_idle(adev)) {
+			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
+		}
+
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		gmc_v8_0_mc_resume(adev, &save);
+		udelay(50);
+
+		gmc_v8_0_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
+					     struct amdgpu_irq_src *src,
+					     unsigned type,
+					     enum amdgpu_interrupt_state state)
+{
+	u32 tmp;
+	u32 bits = (VM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
+		    VM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK);
+
+	switch (state) {
+	case AMDGPU_IRQ_STATE_DISABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp &= ~bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	case AMDGPU_IRQ_STATE_ENABLE:
+		/* system context */
+		tmp = RREG32(mmVM_CONTEXT0_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT0_CNTL, tmp);
+		/* VMs */
+		tmp = RREG32(mmVM_CONTEXT1_CNTL);
+		tmp |= bits;
+		WREG32(mmVM_CONTEXT1_CNTL, tmp);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u32 addr, status, mc_client;
+
+	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
+	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
+	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
+	/* reset addr and status */
+	WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1);
+
+	if (!addr && !status)
+		return 0;
+
+	if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST)
+		gmc_v8_0_set_fault_enable_default(adev, false);
+
+	dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+		entry->src_id, entry->src_data);
+	dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+		addr);
+	dev_err(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+		status);
+	gmc_v8_0_vm_decode_fault(adev, status, addr, mc_client);
+
+	return 0;
+}
+
+static int gmc_v8_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int gmc_v8_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs gmc_v8_0_ip_funcs = {
+	.early_init = gmc_v8_0_early_init,
+	.late_init = gmc_v8_0_late_init,
+	.sw_init = gmc_v8_0_sw_init,
+	.sw_fini = gmc_v8_0_sw_fini,
+	.hw_init = gmc_v8_0_hw_init,
+	.hw_fini = gmc_v8_0_hw_fini,
+	.suspend = gmc_v8_0_suspend,
+	.resume = gmc_v8_0_resume,
+	.is_idle = gmc_v8_0_is_idle,
+	.wait_for_idle = gmc_v8_0_wait_for_idle,
+	.soft_reset = gmc_v8_0_soft_reset,
+	.print_status = gmc_v8_0_print_status,
+	.set_clockgating_state = gmc_v8_0_set_clockgating_state,
+	.set_powergating_state = gmc_v8_0_set_powergating_state,
+};
+
+static const struct amdgpu_gart_funcs gmc_v8_0_gart_funcs = {
+	.flush_gpu_tlb = gmc_v8_0_gart_flush_gpu_tlb,
+	.set_pte_pde = gmc_v8_0_gart_set_pte_pde,
+};
+
+static const struct amdgpu_irq_src_funcs gmc_v8_0_irq_funcs = {
+	.set = gmc_v8_0_vm_fault_interrupt_state,
+	.process = gmc_v8_0_process_interrupt,
+};
+
+static void gmc_v8_0_set_gart_funcs(struct amdgpu_device *adev)
+{
+	if (adev->gart.gart_funcs == NULL)
+		adev->gart.gart_funcs = &gmc_v8_0_gart_funcs;
+}
+
+static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->mc.vm_fault.num_types = 1;
+	adev->mc.vm_fault.funcs = &gmc_v8_0_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h
new file mode 100644
index 000000000..973436086
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V8_0_H__
+#define __GMC_V8_0_H__
+
+extern const struct amd_ip_funcs gmc_v8_0_ip_funcs;
+
+/* XXX these shouldn't be exported */
+void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
+		      struct amdgpu_mode_mc_save *save);
+void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
+			struct amdgpu_mode_mc_save *save);
+int gmc_v8_0_mc_wait_for_idle(struct amdgpu_device *adev);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
new file mode 100644
index 000000000..208d55f41
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
@@ -0,0 +1,195 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "iceland_smumgr.h"
+
+MODULE_FIRMWARE("amdgpu/topaz_smc.bin");
+
+static void iceland_dpm_set_funcs(struct amdgpu_device *adev);
+
+static int iceland_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	iceland_dpm_set_funcs(adev);
+
+	return 0;
+}
+
+static int iceland_dpm_init_microcode(struct amdgpu_device *adev)
+{
+	char fw_name[30] = "amdgpu/topaz_smc.bin";
+	int err;
+
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+
+out:
+	if (err) {
+		DRM_ERROR("Failed to load firmware \"%s\"", fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+}
+
+static int iceland_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = iceland_dpm_init_microcode(adev);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int iceland_dpm_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int iceland_dpm_hw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+
+	/* smu init only needs to be called at startup, not resume.
+	 * It should be in sw_init, but requires the fw info gathered
+	 * in sw_init from other IP modules.
+	 */
+	ret = iceland_smu_init(adev);
+	if (ret) {
+		DRM_ERROR("SMU initialization failed\n");
+		goto fail;
+	}
+
+	ret = iceland_smu_start(adev);
+	if (ret) {
+		DRM_ERROR("SMU start failed\n");
+		goto fail;
+	}
+
+	mutex_unlock(&adev->pm.mutex);
+	return 0;
+
+fail:
+	adev->firmware.smu_load = false;
+	mutex_unlock(&adev->pm.mutex);
+	return -EINVAL;
+}
+
+static int iceland_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+	/* smu fini only needs to be called at teardown, not suspend.
+	 * It should be in sw_fini, but we put it here for symmetry
+	 * with smu init.
+	 */
+	iceland_smu_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	return 0;
+}
+
+static int iceland_dpm_suspend(void *handle)
+{
+	return 0;
+}
+
+static int iceland_dpm_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+
+	ret = iceland_smu_start(adev);
+	if (ret) {
+		DRM_ERROR("SMU start failed\n");
+		goto fail;
+	}
+
+fail:
+	mutex_unlock(&adev->pm.mutex);
+	return ret;
+}
+
+static int iceland_dpm_set_clockgating_state(void *handle,
+			enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int iceland_dpm_set_powergating_state(void *handle,
+			enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs iceland_dpm_ip_funcs = {
+	.early_init = iceland_dpm_early_init,
+	.late_init = NULL,
+	.sw_init = iceland_dpm_sw_init,
+	.sw_fini = iceland_dpm_sw_fini,
+	.hw_init = iceland_dpm_hw_init,
+	.hw_fini = iceland_dpm_hw_fini,
+	.suspend = iceland_dpm_suspend,
+	.resume = iceland_dpm_resume,
+	.is_idle = NULL,
+	.wait_for_idle = NULL,
+	.soft_reset = NULL,
+	.print_status = NULL,
+	.set_clockgating_state = iceland_dpm_set_clockgating_state,
+	.set_powergating_state = iceland_dpm_set_powergating_state,
+};
+
+static const struct amdgpu_dpm_funcs iceland_dpm_funcs = {
+	.get_temperature = NULL,
+	.pre_set_power_state = NULL,
+	.set_power_state = NULL,
+	.post_set_power_state = NULL,
+	.display_configuration_changed = NULL,
+	.get_sclk = NULL,
+	.get_mclk = NULL,
+	.print_power_state = NULL,
+	.debugfs_print_current_performance_level = NULL,
+	.force_performance_level = NULL,
+	.vblank_too_short = NULL,
+	.powergate_uvd = NULL,
+};
+
+static void iceland_dpm_set_funcs(struct amdgpu_device *adev)
+{
+	if (NULL == adev->pm.funcs)
+		adev->pm.funcs = &iceland_dpm_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
new file mode 100644
index 000000000..779532d35
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
@@ -0,0 +1,450 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "vid.h"
+
+#include "oss/oss_2_4_d.h"
+#include "oss/oss_2_4_sh_mask.h"
+
+#include "bif/bif_5_1_d.h"
+#include "bif/bif_5_1_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void iceland_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * iceland_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VI).
+ */
+static void iceland_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * iceland_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VI).
+ */
+static void iceland_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+	u32 ih_cntl = RREG32(mmIH_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, ENABLE_INTR, 0);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	WREG32(mmIH_CNTL, ih_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * iceland_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int iceland_ih_irq_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
+	u64 wptr_off;
+
+	/* disable irqs */
+	iceland_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+
+	/* set the writeback address whether it's enabled or not */
+	wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	/* Default settings for IH_CNTL (disabled at first) */
+	ih_cntl = RREG32(mmIH_CNTL);
+	ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL, RPTR_REARM, 1);
+	WREG32(mmIH_CNTL, ih_cntl);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	iceland_ih_enable_interrupts(adev);
+
+	return ret;
+}
+
+/**
+ * iceland_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VI).
+ */
+static void iceland_ih_irq_disable(struct amdgpu_device *adev)
+{
+	iceland_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * iceland_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VI).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cz_irq_process(VI).
+ * Returns the value of the wptr.
+ */
+static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * iceland_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void iceland_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vm_id = (dw[2] >> 8) & 0xff;
+	entry->pas_id = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * iceland_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void iceland_ih_set_rptr(struct amdgpu_device *adev)
+{
+	WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+}
+
+static int iceland_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	iceland_ih_set_interrupt_funcs(adev);
+	return 0;
+}
+
+static int iceland_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 64 * 1024, false);
+	if (r)
+		return r;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int iceland_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+
+	return 0;
+}
+
+static int iceland_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = iceland_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int iceland_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	iceland_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int iceland_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return iceland_ih_hw_fini(adev);
+}
+
+static int iceland_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return iceland_ih_hw_init(adev);
+}
+
+static bool iceland_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+		return false;
+
+	return true;
+}
+
+static int iceland_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS);
+		if (!REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void iceland_ih_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "ICELAND IH registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+	dev_info(adev->dev, "  INTERRUPT_CNTL=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL));
+	dev_info(adev->dev, "  INTERRUPT_CNTL2=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL2));
+	dev_info(adev->dev, "  IH_CNTL=0x%08X\n",
+		 RREG32(mmIH_CNTL));
+	dev_info(adev->dev, "  IH_RB_CNTL=0x%08X\n",
+		 RREG32(mmIH_RB_CNTL));
+	dev_info(adev->dev, "  IH_RB_BASE=0x%08X\n",
+		 RREG32(mmIH_RB_BASE));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_LO=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_LO));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_HI=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_HI));
+	dev_info(adev->dev, "  IH_RB_RPTR=0x%08X\n",
+		 RREG32(mmIH_RB_RPTR));
+	dev_info(adev->dev, "  IH_RB_WPTR=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR));
+}
+
+static int iceland_ih_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_IH, 1);
+
+	if (srbm_soft_reset) {
+		iceland_ih_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		iceland_ih_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int iceland_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int iceland_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs iceland_ih_ip_funcs = {
+	.early_init = iceland_ih_early_init,
+	.late_init = NULL,
+	.sw_init = iceland_ih_sw_init,
+	.sw_fini = iceland_ih_sw_fini,
+	.hw_init = iceland_ih_hw_init,
+	.hw_fini = iceland_ih_hw_fini,
+	.suspend = iceland_ih_suspend,
+	.resume = iceland_ih_resume,
+	.is_idle = iceland_ih_is_idle,
+	.wait_for_idle = iceland_ih_wait_for_idle,
+	.soft_reset = iceland_ih_soft_reset,
+	.print_status = iceland_ih_print_status,
+	.set_clockgating_state = iceland_ih_set_clockgating_state,
+	.set_powergating_state = iceland_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs iceland_ih_funcs = {
+	.get_wptr = iceland_ih_get_wptr,
+	.decode_iv = iceland_ih_decode_iv,
+	.set_rptr = iceland_ih_set_rptr
+};
+
+static void iceland_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &iceland_ih_funcs;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.h b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.h
new file mode 100644
index 000000000..57558cddf
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ICELAND_IH_H__
+#define __ICELAND_IH_H__
+
+extern const struct amd_ip_funcs iceland_ih_ip_funcs;
+
+#endif /* __ICELAND_IH_H__ */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/iceland_sdma_pkt_open.h b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_sdma_pkt_open.h
new file mode 100644
index 000000000..ee6a041cb
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_sdma_pkt_open.h
@@ -0,0 +1,2172 @@
+/*
+ * Copyright (C) 2014  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __ICELAND_SDMA_PKT_OPEN_H_
+#define __ICELAND_SDMA_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_GEN_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+
+/*define for op field*/
+#define SDMA_PKT_HEADER_op_offset 0
+#define SDMA_PKT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_op_shift  0
+#define SDMA_PKT_HEADER_OP(x) (((x) & SDMA_PKT_HEADER_op_mask) << SDMA_PKT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_HEADER_sub_op_offset 0
+#define SDMA_PKT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_sub_op_shift  8
+#define SDMA_PKT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_HEADER_sub_op_mask) << SDMA_PKT_HEADER_sub_op_shift)
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift  14
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for dst1_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift  22
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_3_height_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_height_shift  16
+#define SDMA_PKT_COPY_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_height_mask) << SDMA_PKT_COPY_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_array_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_w_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_h_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_num_bank_mask) << SDMA_PKT_COPY_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask) << SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 11
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift)
+
+/*define for DW_6 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MIT_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift  15
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_W(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_H(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift  21
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_NUM_BANK(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift  14
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_HA(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift)
+
+/*define for src_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift)
+
+/*define for src_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift)
+
+/*define for src_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift)
+
+/*define for src_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift)
+
+/*define for src_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift)
+
+/*define for src_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift)
+
+/*define for src_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift)
+
+/*define for DW_11 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for dst_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift)
+
+/*define for dst_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift)
+
+/*define for dst_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_12_DST_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift)
+
+/*define for dst_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift)
+
+/*define for dst_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift)
+
+/*define for dst_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_12_DST_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift)
+
+/*define for dst_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift)
+
+/*define for dst_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_12_DST_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for struct_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift  22
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for linear_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift  30
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_height_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_height_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_height_mask) << SDMA_PKT_WRITE_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_WRITE_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_WRITE_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_WRITE_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask) << SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_WRITE_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask) << SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0000FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+#endif /* __ICELAND_SDMA_PKT_OPEN_H_ */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/iceland_smc.c b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_smc.c
new file mode 100644
index 000000000..090486c18
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_smc.c
@@ -0,0 +1,671 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "ppsmc.h"
+#include "iceland_smumgr.h"
+#include "smu_ucode_xfer_vi.h"
+#include "amdgpu_ucode.h"
+
+#include "smu/smu_7_1_1_d.h"
+#include "smu/smu_7_1_1_sh_mask.h"
+
+#define ICELAND_SMC_SIZE 0x20000
+
+static int iceland_set_smc_sram_address(struct amdgpu_device *adev,
+					uint32_t smc_address, uint32_t limit)
+{
+	uint32_t val;
+
+	if (smc_address & 3)
+		return -EINVAL;
+
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 0);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+
+	return 0;
+}
+
+static int iceland_copy_bytes_to_smc(struct amdgpu_device *adev,
+				     uint32_t smc_start_address,
+				     const uint8_t *src,
+				     uint32_t byte_count, uint32_t limit)
+{
+	uint32_t addr;
+	uint32_t data, orig_data;
+	int result = 0;
+	uint32_t extra_shift;
+	unsigned long flags;
+
+	if (smc_start_address & 3)
+		return -EINVAL;
+
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	while (byte_count >= 4) {
+		/* Bytes are written into the SMC addres space with the MSB first */
+		data = (src[0] << 24) + (src[1] << 16) + (src[2] << 8) + src[3];
+
+		result = iceland_set_smc_sram_address(adev, addr, limit);
+
+		if (result)
+			goto out;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	if (0 != byte_count) {
+		/* Now write odd bytes left, do a read modify write cycle */
+		data = 0;
+
+		result = iceland_set_smc_sram_address(adev, addr, limit);
+		if (result)
+			goto out;
+
+		orig_data = RREG32(mmSMC_IND_DATA_0);
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+		data |= (orig_data & ~((~0UL) << extra_shift));
+
+		result = iceland_set_smc_sram_address(adev, addr, limit);
+		if (result)
+			goto out;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+
+out:
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+void iceland_start_smc(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 0);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+}
+
+void iceland_reset_smc(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+}
+
+static int iceland_program_jump_on_start(struct amdgpu_device *adev)
+{
+	static unsigned char data[] = {0xE0, 0x00, 0x80, 0x40};
+	iceland_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
+
+	return 0;
+}
+
+void iceland_stop_smc_clock(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 1);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+}
+
+void iceland_start_smc_clock(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 0);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+}
+
+static bool iceland_is_smc_ram_running(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_GET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable);
+
+	return ((0 == val) && (0x20100 <= RREG32_SMC(ixSMC_PC_C)));
+}
+
+static int wait_smu_response(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32(mmSMC_RESP_0);
+		if (REG_GET_FIELD(val, SMC_RESP_0, SMC_RESP))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int iceland_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
+{
+	if (!iceland_is_smc_ram_running(adev))
+		return -EINVAL;
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send message\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int iceland_send_msg_to_smc_without_waiting(struct amdgpu_device *adev,
+						   PPSMC_Msg msg)
+{
+	if (!iceland_is_smc_ram_running(adev))
+		return -EINVAL;;
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	return 0;
+}
+
+static int iceland_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						  PPSMC_Msg msg,
+						  uint32_t parameter)
+{
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return iceland_send_msg_to_smc(adev, msg);
+}
+
+static int iceland_send_msg_to_smc_with_parameter_without_waiting(
+					struct amdgpu_device *adev,
+					PPSMC_Msg msg, uint32_t parameter)
+{
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return iceland_send_msg_to_smc_without_waiting(adev, msg);
+}
+
+#if 0 /* not used yet */
+static int iceland_wait_for_smc_inactive(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	if (!iceland_is_smc_ram_running(adev))
+		return -EINVAL;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+		if (REG_GET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, cken) == 0)
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+#endif
+
+static int iceland_smu_upload_firmware_image(struct amdgpu_device *adev)
+{
+	const struct smc_firmware_header_v1_0 *hdr;
+	uint32_t ucode_size;
+	uint32_t ucode_start_address;
+	const uint8_t *src;
+	uint32_t val;
+	uint32_t byte_count;
+	uint32_t data;
+	unsigned long flags;
+	int i;
+
+	if (!adev->pm.fw)
+		return -EINVAL;
+
+	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+	src = (const uint8_t *)
+		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	if (ucode_size & 3) {
+		DRM_ERROR("SMC ucode is not 4 bytes aligned\n");
+		return -EINVAL;
+	}
+
+	if (ucode_size > ICELAND_SMC_SIZE) {
+		DRM_ERROR("SMC address is beyond the SMC RAM area\n");
+		return -EINVAL;
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixRCU_UC_EVENTS);
+		if (REG_GET_FIELD(val, RCU_UC_EVENTS, boot_seq_done) == 0)
+			break;
+		udelay(1);
+	}
+	val = RREG32_SMC(ixSMC_SYSCON_MISC_CNTL);
+	WREG32_SMC(ixSMC_SYSCON_MISC_CNTL, val | 1);
+
+	iceland_stop_smc_clock(adev);
+	iceland_reset_smc(adev);
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 1);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+
+	byte_count = ucode_size;
+	while (byte_count >= 4) {
+		data = (src[0] << 24) + (src[1] << 16) + (src[2] << 8) + src[3];
+		WREG32(mmSMC_IND_DATA_0, data);
+		src += 4;
+		byte_count -= 4;
+	}
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 0);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return 0;
+}
+
+#if 0 /* not used yet */
+static int iceland_read_smc_sram_dword(struct amdgpu_device *adev,
+				       uint32_t smc_address,
+				       uint32_t *value,
+				       uint32_t limit)
+{
+	int result;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	result = iceland_set_smc_sram_address(adev, smc_address, limit);
+	if (result == 0)
+		*value = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int iceland_write_smc_sram_dword(struct amdgpu_device *adev,
+					uint32_t smc_address,
+					uint32_t value,
+					uint32_t limit)
+{
+	int result;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	result = iceland_set_smc_sram_address(adev, smc_address, limit);
+	if (result == 0)
+		WREG32(mmSMC_IND_DATA_0, value);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int iceland_smu_stop_smc(struct amdgpu_device *adev)
+{
+	iceland_reset_smc(adev);
+	iceland_stop_smc_clock(adev);
+
+	return 0;
+}
+#endif
+
+static int iceland_smu_start_smc(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	iceland_program_jump_on_start(adev);
+	iceland_start_smc_clock(adev);
+	iceland_start_smc(adev);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixFIRMWARE_FLAGS);
+		if (REG_GET_FIELD(val, FIRMWARE_FLAGS, INTERRUPTS_ENABLED) == 1)
+			break;
+		udelay(1);
+	}
+	return 0;
+}
+
+static enum AMDGPU_UCODE_ID iceland_convert_fw_type(uint32_t fw_type)
+{
+	switch (fw_type) {
+		case UCODE_ID_SDMA0:
+			return AMDGPU_UCODE_ID_SDMA0;
+		case UCODE_ID_SDMA1:
+			return AMDGPU_UCODE_ID_SDMA1;
+		case UCODE_ID_CP_CE:
+			return AMDGPU_UCODE_ID_CP_CE;
+		case UCODE_ID_CP_PFP:
+			return AMDGPU_UCODE_ID_CP_PFP;
+		case UCODE_ID_CP_ME:
+			return AMDGPU_UCODE_ID_CP_ME;
+		case UCODE_ID_CP_MEC:
+		case UCODE_ID_CP_MEC_JT1:
+			return AMDGPU_UCODE_ID_CP_MEC1;
+		case UCODE_ID_CP_MEC_JT2:
+			return AMDGPU_UCODE_ID_CP_MEC2;
+		case UCODE_ID_RLC_G:
+			return AMDGPU_UCODE_ID_RLC_G;
+		default:
+			DRM_ERROR("ucode type is out of range!\n");
+			return AMDGPU_UCODE_ID_MAXIMUM;
+	}
+}
+
+static uint32_t iceland_smu_get_mask_for_fw_type(uint32_t fw_type)
+{
+	switch (fw_type) {
+		case AMDGPU_UCODE_ID_SDMA0:
+			return UCODE_ID_SDMA0_MASK;
+		case AMDGPU_UCODE_ID_SDMA1:
+			return UCODE_ID_SDMA1_MASK;
+		case AMDGPU_UCODE_ID_CP_CE:
+			return UCODE_ID_CP_CE_MASK;
+		case AMDGPU_UCODE_ID_CP_PFP:
+			return UCODE_ID_CP_PFP_MASK;
+		case AMDGPU_UCODE_ID_CP_ME:
+			return UCODE_ID_CP_ME_MASK;
+		case AMDGPU_UCODE_ID_CP_MEC1:
+			return UCODE_ID_CP_MEC_MASK | UCODE_ID_CP_MEC_JT1_MASK;
+		case AMDGPU_UCODE_ID_CP_MEC2:
+			return UCODE_ID_CP_MEC_MASK;
+		case AMDGPU_UCODE_ID_RLC_G:
+			return UCODE_ID_RLC_G_MASK;
+		default:
+			DRM_ERROR("ucode type is out of range!\n");
+			return 0;
+	}
+}
+
+static int iceland_smu_populate_single_firmware_entry(struct amdgpu_device *adev,
+						      uint32_t fw_type,
+						      struct SMU_Entry *entry)
+{
+	enum AMDGPU_UCODE_ID id = iceland_convert_fw_type(fw_type);
+	struct amdgpu_firmware_info *ucode = &adev->firmware.ucode[id];
+	const struct gfx_firmware_header_v1_0 *header = NULL;
+	uint64_t gpu_addr;
+	uint32_t data_size;
+
+	if (ucode->fw == NULL)
+		return -EINVAL;
+
+	gpu_addr  = ucode->mc_addr;
+	header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+	data_size = le32_to_cpu(header->header.ucode_size_bytes);
+
+	entry->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
+	entry->id = (uint16_t)fw_type;
+	entry->image_addr_high = upper_32_bits(gpu_addr);
+	entry->image_addr_low = lower_32_bits(gpu_addr);
+	entry->meta_data_addr_high = 0;
+	entry->meta_data_addr_low = 0;
+	entry->data_size_byte = data_size;
+	entry->num_register_entries = 0;
+	entry->flags = 0;
+
+	return 0;
+}
+
+static int iceland_smu_request_load_fw(struct amdgpu_device *adev)
+{
+	struct iceland_smu_private_data *private = (struct iceland_smu_private_data *)adev->smu.priv;
+	struct SMU_DRAMData_TOC *toc;
+	uint32_t fw_to_load;
+
+	toc = (struct SMU_DRAMData_TOC *)private->header;
+	toc->num_entries = 0;
+	toc->structure_version = 1;
+
+	if (!adev->firmware.smu_load)
+		return 0;
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_RLC_G,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for RLC\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_CE,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for CE\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_PFP,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for PFP\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_ME,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for ME\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC_JT1,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC_JT1\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_SDMA0,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for SDMA0\n");
+		return -EINVAL;
+	}
+
+	if (iceland_smu_populate_single_firmware_entry(adev, UCODE_ID_SDMA1,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for SDMA1\n");
+		return -EINVAL;
+	}
+
+	iceland_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DRV_DRAM_ADDR_HI, private->header_addr_high);
+	iceland_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DRV_DRAM_ADDR_LO, private->header_addr_low);
+
+	fw_to_load = UCODE_ID_RLC_G_MASK |
+			UCODE_ID_SDMA0_MASK |
+			UCODE_ID_SDMA1_MASK |
+			UCODE_ID_CP_CE_MASK |
+			UCODE_ID_CP_ME_MASK |
+			UCODE_ID_CP_PFP_MASK |
+			UCODE_ID_CP_MEC_MASK |
+			UCODE_ID_CP_MEC_JT1_MASK;
+
+
+	if (iceland_send_msg_to_smc_with_parameter_without_waiting(adev, PPSMC_MSG_LoadUcodes, fw_to_load)) {
+		DRM_ERROR("Fail to request SMU load ucode\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int iceland_smu_check_fw_load_finish(struct amdgpu_device *adev,
+					    uint32_t fw_type)
+{
+	uint32_t fw_mask = iceland_smu_get_mask_for_fw_type(fw_type);
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (fw_mask == (RREG32_SMC(ixSOFT_REGISTERS_TABLE_27) & fw_mask))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("check firmware loading failed\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int iceland_smu_start(struct amdgpu_device *adev)
+{
+	int result;
+
+	result = iceland_smu_upload_firmware_image(adev);
+	if (result)
+		return result;
+	result = iceland_smu_start_smc(adev);
+	if (result)
+		return result;
+
+	return iceland_smu_request_load_fw(adev);
+}
+
+static const struct amdgpu_smumgr_funcs iceland_smumgr_funcs = {
+	.check_fw_load_finish = iceland_smu_check_fw_load_finish,
+	.request_smu_load_fw = NULL,
+	.request_smu_specific_fw = NULL,
+};
+
+int iceland_smu_init(struct amdgpu_device *adev)
+{
+	struct iceland_smu_private_data *private;
+	uint32_t image_size = ((sizeof(struct SMU_DRAMData_TOC) / 4096) + 1) * 4096;
+	struct amdgpu_bo **toc_buf = &adev->smu.toc_buf;
+	uint64_t mc_addr;
+	void *toc_buf_ptr;
+	int ret;
+
+	private = kzalloc(sizeof(struct iceland_smu_private_data), GFP_KERNEL);
+	if (NULL == private)
+		return -ENOMEM;
+
+	/* allocate firmware buffers */
+	if (adev->firmware.smu_load)
+		amdgpu_ucode_init_bo(adev);
+
+	adev->smu.priv = private;
+	adev->smu.fw_flags = 0;
+
+	/* Allocate FW image data structure and header buffer */
+	ret = amdgpu_bo_create(adev, image_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_VRAM,
+			       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			       NULL, NULL, toc_buf);
+	if (ret) {
+		DRM_ERROR("Failed to allocate memory for TOC buffer\n");
+		return -ENOMEM;
+	}
+
+	/* Retrieve GPU address for header buffer and internal buffer */
+	ret = amdgpu_bo_reserve(adev->smu.toc_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to reserve the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.toc_buf, AMDGPU_GEM_DOMAIN_VRAM, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to pin the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_kmap(*toc_buf, &toc_buf_ptr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to map the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	amdgpu_bo_unreserve(adev->smu.toc_buf);
+	private->header_addr_low = lower_32_bits(mc_addr);
+	private->header_addr_high = upper_32_bits(mc_addr);
+	private->header = toc_buf_ptr;
+
+	adev->smu.smumgr_funcs = &iceland_smumgr_funcs;
+
+	return 0;
+}
+
+int iceland_smu_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->smu.toc_buf);
+	kfree(adev->smu.priv);
+	adev->smu.priv = NULL;
+	if (adev->firmware.fw_buf)
+		amdgpu_ucode_fini_bo(adev);
+
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/iceland_smumgr.h b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_smumgr.h
new file mode 100644
index 000000000..1e0769e11
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/iceland_smumgr.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef ICELAND_SMUMGR_H
+#define ICELAND_SMUMGR_H
+
+#include "ppsmc.h"
+
+extern int iceland_smu_init(struct amdgpu_device *adev);
+extern int iceland_smu_fini(struct amdgpu_device *adev);
+extern int iceland_smu_start(struct amdgpu_device *adev);
+
+struct iceland_smu_private_data
+{
+	uint8_t *header;
+	uint8_t *mec_image;
+	uint32_t header_addr_high;
+	uint32_t header_addr_low;
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
new file mode 100644
index 000000000..7e9154c7f
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
@@ -0,0 +1,3349 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_pm.h"
+#include "cikd.h"
+#include "atom.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_dpm.h"
+#include "kv_dpm.h"
+#include "gfx_v7_0.h"
+#include <linux/seq_file.h>
+
+#include "smu/smu_7_0_0_d.h"
+#include "smu/smu_7_0_0_sh_mask.h"
+
+#include "gca/gfx_7_2_d.h"
+#include "gca/gfx_7_2_sh_mask.h"
+
+#define KV_MAX_DEEPSLEEP_DIVIDER_ID     5
+#define KV_MINIMUM_ENGINE_CLOCK         800
+#define SMC_RAM_END                     0x40000
+
+static void kv_dpm_set_dpm_funcs(struct amdgpu_device *adev);
+static void kv_dpm_set_irq_funcs(struct amdgpu_device *adev);
+static int kv_enable_nb_dpm(struct amdgpu_device *adev,
+			    bool enable);
+static void kv_init_graphics_levels(struct amdgpu_device *adev);
+static int kv_calculate_ds_divider(struct amdgpu_device *adev);
+static int kv_calculate_nbps_level_settings(struct amdgpu_device *adev);
+static int kv_calculate_dpm_settings(struct amdgpu_device *adev);
+static void kv_enable_new_levels(struct amdgpu_device *adev);
+static void kv_program_nbps_index_settings(struct amdgpu_device *adev,
+					   struct amdgpu_ps *new_rps);
+static int kv_set_enabled_level(struct amdgpu_device *adev, u32 level);
+static int kv_set_enabled_levels(struct amdgpu_device *adev);
+static int kv_force_dpm_highest(struct amdgpu_device *adev);
+static int kv_force_dpm_lowest(struct amdgpu_device *adev);
+static void kv_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *new_rps,
+					struct amdgpu_ps *old_rps);
+static int kv_set_thermal_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp);
+static int kv_init_fps_limits(struct amdgpu_device *adev);
+
+static void kv_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate);
+static void kv_dpm_powergate_vce(struct amdgpu_device *adev, bool gate);
+static void kv_dpm_powergate_samu(struct amdgpu_device *adev, bool gate);
+static void kv_dpm_powergate_acp(struct amdgpu_device *adev, bool gate);
+
+
+static u32 kv_convert_vid2_to_vid7(struct amdgpu_device *adev,
+				   struct sumo_vid_mapping_table *vid_mapping_table,
+				   u32 vid_2bit)
+{
+	struct amdgpu_clock_voltage_dependency_table *vddc_sclk_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 i;
+
+	if (vddc_sclk_table && vddc_sclk_table->count) {
+		if (vid_2bit < vddc_sclk_table->count)
+			return vddc_sclk_table->entries[vid_2bit].v;
+		else
+			return vddc_sclk_table->entries[vddc_sclk_table->count - 1].v;
+	} else {
+		for (i = 0; i < vid_mapping_table->num_entries; i++) {
+			if (vid_mapping_table->entries[i].vid_2bit == vid_2bit)
+				return vid_mapping_table->entries[i].vid_7bit;
+		}
+		return vid_mapping_table->entries[vid_mapping_table->num_entries - 1].vid_7bit;
+	}
+}
+
+static u32 kv_convert_vid7_to_vid2(struct amdgpu_device *adev,
+				   struct sumo_vid_mapping_table *vid_mapping_table,
+				   u32 vid_7bit)
+{
+	struct amdgpu_clock_voltage_dependency_table *vddc_sclk_table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 i;
+
+	if (vddc_sclk_table && vddc_sclk_table->count) {
+		for (i = 0; i < vddc_sclk_table->count; i++) {
+			if (vddc_sclk_table->entries[i].v == vid_7bit)
+				return i;
+		}
+		return vddc_sclk_table->count - 1;
+	} else {
+		for (i = 0; i < vid_mapping_table->num_entries; i++) {
+			if (vid_mapping_table->entries[i].vid_7bit == vid_7bit)
+				return vid_mapping_table->entries[i].vid_2bit;
+		}
+
+		return vid_mapping_table->entries[vid_mapping_table->num_entries - 1].vid_2bit;
+	}
+}
+
+static void sumo_take_smu_control(struct amdgpu_device *adev, bool enable)
+{
+/* This bit selects who handles display phy powergating.
+ * Clear the bit to let atom handle it.
+ * Set it to let the driver handle it.
+ * For now we just let atom handle it.
+ */
+#if 0
+	u32 v = RREG32(mmDOUT_SCRATCH3);
+
+	if (enable)
+		v |= 0x4;
+	else
+		v &= 0xFFFFFFFB;
+
+	WREG32(mmDOUT_SCRATCH3, v);
+#endif
+}
+
+static u32 sumo_get_sleep_divider_from_id(u32 id)
+{
+	return 1 << id;
+}
+
+static void sumo_construct_sclk_voltage_mapping_table(struct amdgpu_device *adev,
+						      struct sumo_sclk_voltage_mapping_table *sclk_voltage_mapping_table,
+						      ATOM_AVAILABLE_SCLK_LIST *table)
+{
+	u32 i;
+	u32 n = 0;
+	u32 prev_sclk = 0;
+
+	for (i = 0; i < SUMO_MAX_HARDWARE_POWERLEVELS; i++) {
+		if (table[i].ulSupportedSCLK > prev_sclk) {
+			sclk_voltage_mapping_table->entries[n].sclk_frequency =
+				table[i].ulSupportedSCLK;
+			sclk_voltage_mapping_table->entries[n].vid_2bit =
+				table[i].usVoltageIndex;
+			prev_sclk = table[i].ulSupportedSCLK;
+			n++;
+		}
+	}
+
+	sclk_voltage_mapping_table->num_max_dpm_entries = n;
+}
+
+static void sumo_construct_vid_mapping_table(struct amdgpu_device *adev,
+					     struct sumo_vid_mapping_table *vid_mapping_table,
+					     ATOM_AVAILABLE_SCLK_LIST *table)
+{
+	u32 i, j;
+
+	for (i = 0; i < SUMO_MAX_HARDWARE_POWERLEVELS; i++) {
+		if (table[i].ulSupportedSCLK != 0) {
+			vid_mapping_table->entries[table[i].usVoltageIndex].vid_7bit =
+				table[i].usVoltageID;
+			vid_mapping_table->entries[table[i].usVoltageIndex].vid_2bit =
+				table[i].usVoltageIndex;
+		}
+	}
+
+	for (i = 0; i < SUMO_MAX_NUMBER_VOLTAGES; i++) {
+		if (vid_mapping_table->entries[i].vid_7bit == 0) {
+			for (j = i + 1; j < SUMO_MAX_NUMBER_VOLTAGES; j++) {
+				if (vid_mapping_table->entries[j].vid_7bit != 0) {
+					vid_mapping_table->entries[i] =
+						vid_mapping_table->entries[j];
+					vid_mapping_table->entries[j].vid_7bit = 0;
+					break;
+				}
+			}
+
+			if (j == SUMO_MAX_NUMBER_VOLTAGES)
+				break;
+		}
+	}
+
+	vid_mapping_table->num_entries = i;
+}
+
+static const struct kv_lcac_config_values sx_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{  1,       4,        1    },
+	{  2,       5,        1    },
+	{  3,       4,        2    },
+	{  4,       1,        1    },
+	{  5,       5,        2    },
+	{  6,       6,        1    },
+	{  7,       9,        2    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc0_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc1_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc2_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values mc3_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_values cpl_local_cac_cfg_kv[] =
+{
+	{  0,       4,        1    },
+	{  1,       4,        1    },
+	{  2,       5,        1    },
+	{  3,       4,        1    },
+	{  4,       1,        1    },
+	{  5,       5,        1    },
+	{  6,       6,        1    },
+	{  7,       9,        1    },
+	{  8,       4,        1    },
+	{  9,       2,        1    },
+	{  10,      3,        1    },
+	{  11,      6,        1    },
+	{  12,      8,        2    },
+	{  13,      1,        1    },
+	{  14,      2,        1    },
+	{  15,      3,        1    },
+	{  16,      1,        1    },
+	{  17,      4,        1    },
+	{  18,      3,        1    },
+	{  19,      1,        1    },
+	{  20,      8,        1    },
+	{  21,      5,        1    },
+	{  22,      1,        1    },
+	{  23,      1,        1    },
+	{  24,      4,        1    },
+	{  27,      6,        1    },
+	{  28,      1,        1    },
+	{ 0xffffffff }
+};
+
+static const struct kv_lcac_config_reg sx0_cac_config_reg[] =
+{
+	{ 0xc0400d00, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc0_cac_config_reg[] =
+{
+	{ 0xc0400d30, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc1_cac_config_reg[] =
+{
+	{ 0xc0400d3c, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc2_cac_config_reg[] =
+{
+	{ 0xc0400d48, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg mc3_cac_config_reg[] =
+{
+	{ 0xc0400d54, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_lcac_config_reg cpl_cac_config_reg[] =
+{
+	{ 0xc0400d80, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
+};
+
+static const struct kv_pt_config_reg didt_config_kv[] =
+{
+	{ 0x10, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x10, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x10, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x11, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x12, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x2, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x1, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x1, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x0, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x30, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x31, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x32, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x22, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x21, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x21, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x20, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x50, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x51, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x52, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x42, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x41, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x41, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x40, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x70, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x71, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x000000ff, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x0000ff00, 8, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0x00ff0000, 16, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x72, 0xff000000, 24, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x00003fff, 0, 0x4, KV_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x03ff0000, 16, 0x80, KV_CONFIGREG_DIDT_IND },
+	{ 0x62, 0x78000000, 27, 0x3, KV_CONFIGREG_DIDT_IND },
+	{ 0x61, 0x0000ffff, 0, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x61, 0xffff0000, 16, 0x3FFF, KV_CONFIGREG_DIDT_IND },
+	{ 0x60, 0x00000001, 0, 0x0, KV_CONFIGREG_DIDT_IND },
+	{ 0xFFFFFFFF }
+};
+
+static struct kv_ps *kv_get_ps(struct amdgpu_ps *rps)
+{
+	struct kv_ps *ps = rps->ps_priv;
+
+	return ps;
+}
+
+static struct kv_power_info *kv_get_pi(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = adev->pm.dpm.priv;
+
+	return pi;
+}
+
+#if 0
+static void kv_program_local_cac_table(struct amdgpu_device *adev,
+				       const struct kv_lcac_config_values *local_cac_table,
+				       const struct kv_lcac_config_reg *local_cac_reg)
+{
+	u32 i, count, data;
+	const struct kv_lcac_config_values *values = local_cac_table;
+
+	while (values->block_id != 0xffffffff) {
+		count = values->signal_id;
+		for (i = 0; i < count; i++) {
+			data = ((values->block_id << local_cac_reg->block_shift) &
+				local_cac_reg->block_mask);
+			data |= ((i << local_cac_reg->signal_shift) &
+				 local_cac_reg->signal_mask);
+			data |= ((values->t << local_cac_reg->t_shift) &
+				 local_cac_reg->t_mask);
+			data |= ((1 << local_cac_reg->enable_shift) &
+				 local_cac_reg->enable_mask);
+			WREG32_SMC(local_cac_reg->cntl, data);
+		}
+		values++;
+	}
+}
+#endif
+
+static int kv_program_pt_config_registers(struct amdgpu_device *adev,
+					  const struct kv_pt_config_reg *cac_config_regs)
+{
+	const struct kv_pt_config_reg *config_regs = cac_config_regs;
+	u32 data;
+	u32 cache = 0;
+
+	if (config_regs == NULL)
+		return -EINVAL;
+
+	while (config_regs->offset != 0xFFFFFFFF) {
+		if (config_regs->type == KV_CONFIGREG_CACHE) {
+			cache |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+		} else {
+			switch (config_regs->type) {
+			case KV_CONFIGREG_SMC_IND:
+				data = RREG32_SMC(config_regs->offset);
+				break;
+			case KV_CONFIGREG_DIDT_IND:
+				data = RREG32_DIDT(config_regs->offset);
+				break;
+			default:
+				data = RREG32(config_regs->offset);
+				break;
+			}
+
+			data &= ~config_regs->mask;
+			data |= ((config_regs->value << config_regs->shift) & config_regs->mask);
+			data |= cache;
+			cache = 0;
+
+			switch (config_regs->type) {
+			case KV_CONFIGREG_SMC_IND:
+				WREG32_SMC(config_regs->offset, data);
+				break;
+			case KV_CONFIGREG_DIDT_IND:
+				WREG32_DIDT(config_regs->offset, data);
+				break;
+			default:
+				WREG32(config_regs->offset, data);
+				break;
+			}
+		}
+		config_regs++;
+	}
+
+	return 0;
+}
+
+static void kv_do_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 data;
+
+	if (pi->caps_sq_ramping) {
+		data = RREG32_DIDT(ixDIDT_SQ_CTRL0);
+		if (enable)
+			data |= DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_SQ_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_SQ_CTRL0, data);
+	}
+
+	if (pi->caps_db_ramping) {
+		data = RREG32_DIDT(ixDIDT_DB_CTRL0);
+		if (enable)
+			data |= DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_DB_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_DB_CTRL0, data);
+	}
+
+	if (pi->caps_td_ramping) {
+		data = RREG32_DIDT(ixDIDT_TD_CTRL0);
+		if (enable)
+			data |= DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TD_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TD_CTRL0, data);
+	}
+
+	if (pi->caps_tcp_ramping) {
+		data = RREG32_DIDT(ixDIDT_TCP_CTRL0);
+		if (enable)
+			data |= DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		else
+			data &= ~DIDT_TCP_CTRL0__DIDT_CTRL_EN_MASK;
+		WREG32_DIDT(ixDIDT_TCP_CTRL0, data);
+	}
+}
+
+static int kv_enable_didt(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	if (pi->caps_sq_ramping ||
+	    pi->caps_db_ramping ||
+	    pi->caps_td_ramping ||
+	    pi->caps_tcp_ramping) {
+		gfx_v7_0_enter_rlc_safe_mode(adev);
+
+		if (enable) {
+			ret = kv_program_pt_config_registers(adev, didt_config_kv);
+			if (ret) {
+				gfx_v7_0_exit_rlc_safe_mode(adev);
+				return ret;
+			}
+		}
+
+		kv_do_enable_didt(adev, enable);
+
+		gfx_v7_0_exit_rlc_safe_mode(adev);
+	}
+
+	return 0;
+}
+
+#if 0
+static void kv_initialize_hardware_cac_manager(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->caps_cac) {
+		WREG32_SMC(ixLCAC_SX0_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_SX0_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, sx_local_cac_cfg_kv, sx0_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC0_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC0_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc0_local_cac_cfg_kv, mc0_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC1_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC1_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc1_local_cac_cfg_kv, mc1_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC2_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC2_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc2_local_cac_cfg_kv, mc2_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_MC3_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_MC3_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, mc3_local_cac_cfg_kv, mc3_cac_config_reg);
+
+		WREG32_SMC(ixLCAC_CPL_OVR_SEL, 0);
+		WREG32_SMC(ixLCAC_CPL_OVR_VAL, 0);
+		kv_program_local_cac_table(adev, cpl_local_cac_cfg_kv, cpl_cac_config_reg);
+	}
+}
+#endif
+
+static int kv_enable_smc_cac(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+
+	if (pi->caps_cac) {
+		if (enable) {
+			ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_EnableCac);
+			if (ret)
+				pi->cac_enabled = false;
+			else
+				pi->cac_enabled = true;
+		} else if (pi->cac_enabled) {
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DisableCac);
+			pi->cac_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int kv_process_firmware_header(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 tmp;
+	int ret;
+
+	ret = amdgpu_kv_read_smc_sram_dword(adev, SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, DpmTable),
+				     &tmp, pi->sram_end);
+
+	if (ret == 0)
+		pi->dpm_table_start = tmp;
+
+	ret = amdgpu_kv_read_smc_sram_dword(adev, SMU7_FIRMWARE_HEADER_LOCATION +
+				     offsetof(SMU7_Firmware_Header, SoftRegisters),
+				     &tmp, pi->sram_end);
+
+	if (ret == 0)
+		pi->soft_regs_start = tmp;
+
+	return ret;
+}
+
+static int kv_enable_dpm_voltage_scaling(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->graphics_voltage_change_enable = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsVoltageChangeEnable),
+				   &pi->graphics_voltage_change_enable,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static int kv_set_dpm_interval(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->graphics_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsInterval),
+				   &pi->graphics_interval,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static int kv_set_dpm_boot_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsBootLevel),
+				   &pi->graphics_boot_level,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static void kv_program_vc(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0x3FFFC100);
+}
+
+static void kv_clear_vc(struct amdgpu_device *adev)
+{
+	WREG32_SMC(ixCG_FREQ_TRAN_VOTING_0, 0);
+}
+
+static int kv_set_divider_value(struct amdgpu_device *adev,
+				u32 index, u32 sclk)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct atom_clock_dividers dividers;
+	int ret;
+
+	ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+						 sclk, false, &dividers);
+	if (ret)
+		return ret;
+
+	pi->graphics_level[index].SclkDid = (u8)dividers.post_div;
+	pi->graphics_level[index].SclkFrequency = cpu_to_be32(sclk);
+
+	return 0;
+}
+
+static u16 kv_convert_8bit_index_to_voltage(struct amdgpu_device *adev,
+					    u16 voltage)
+{
+	return 6200 - (voltage * 25);
+}
+
+static u16 kv_convert_2bit_index_to_voltage(struct amdgpu_device *adev,
+					    u32 vid_2bit)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 vid_8bit = kv_convert_vid2_to_vid7(adev,
+					       &pi->sys_info.vid_mapping_table,
+					       vid_2bit);
+
+	return kv_convert_8bit_index_to_voltage(adev, (u16)vid_8bit);
+}
+
+
+static int kv_set_vid(struct amdgpu_device *adev, u32 index, u32 vid)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].VoltageDownH = (u8)pi->voltage_drop_t;
+	pi->graphics_level[index].MinVddNb =
+		cpu_to_be32(kv_convert_2bit_index_to_voltage(adev, vid));
+
+	return 0;
+}
+
+static int kv_set_at(struct amdgpu_device *adev, u32 index, u32 at)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].AT = cpu_to_be16((u16)at);
+
+	return 0;
+}
+
+static void kv_dpm_power_level_enable(struct amdgpu_device *adev,
+				      u32 index, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].EnabledForActivity = enable ? 1 : 0;
+}
+
+static void kv_start_dpm(struct amdgpu_device *adev)
+{
+	u32 tmp = RREG32_SMC(ixGENERAL_PWRMGT);
+
+	tmp |= GENERAL_PWRMGT__GLOBAL_PWRMGT_EN_MASK;
+	WREG32_SMC(ixGENERAL_PWRMGT, tmp);
+
+	amdgpu_kv_smc_dpm_enable(adev, true);
+}
+
+static void kv_stop_dpm(struct amdgpu_device *adev)
+{
+	amdgpu_kv_smc_dpm_enable(adev, false);
+}
+
+static void kv_start_am(struct amdgpu_device *adev)
+{
+	u32 sclk_pwrmgt_cntl = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+
+	sclk_pwrmgt_cntl &= ~(SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK |
+			SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+	sclk_pwrmgt_cntl |= SCLK_PWRMGT_CNTL__DYNAMIC_PM_EN_MASK;
+
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, sclk_pwrmgt_cntl);
+}
+
+static void kv_reset_am(struct amdgpu_device *adev)
+{
+	u32 sclk_pwrmgt_cntl = RREG32_SMC(ixSCLK_PWRMGT_CNTL);
+
+	sclk_pwrmgt_cntl |= (SCLK_PWRMGT_CNTL__RESET_SCLK_CNT_MASK |
+			SCLK_PWRMGT_CNTL__RESET_BUSY_CNT_MASK);
+
+	WREG32_SMC(ixSCLK_PWRMGT_CNTL, sclk_pwrmgt_cntl);
+}
+
+static int kv_freeze_sclk_dpm(struct amdgpu_device *adev, bool freeze)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, freeze ?
+					PPSMC_MSG_SCLKDPM_FreezeLevel : PPSMC_MSG_SCLKDPM_UnfreezeLevel);
+}
+
+static int kv_force_lowest_valid(struct amdgpu_device *adev)
+{
+	return kv_force_dpm_lowest(adev);
+}
+
+static int kv_unforce_levels(struct amdgpu_device *adev)
+{
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_NoForcedLevel);
+	else
+		return kv_set_enabled_levels(adev);
+}
+
+static int kv_update_sclk_t(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 low_sclk_interrupt_t = 0;
+	int ret = 0;
+
+	if (pi->caps_sclk_throttle_low_notification) {
+		low_sclk_interrupt_t = cpu_to_be32(pi->low_sclk_interrupt_t);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, LowSclkInterruptT),
+					   (u8 *)&low_sclk_interrupt_t,
+					   sizeof(u32), pi->sram_end);
+	}
+	return ret;
+}
+
+static int kv_program_bootup_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].clk == pi->boot_pl.sclk)
+				break;
+		}
+
+		pi->graphics_boot_level = (u8)i;
+		kv_dpm_power_level_enable(adev, i, true);
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		if (table->num_max_dpm_entries == 0)
+			return -EINVAL;
+
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].sclk_frequency == pi->boot_pl.sclk)
+				break;
+		}
+
+		pi->graphics_boot_level = (u8)i;
+		kv_dpm_power_level_enable(adev, i, true);
+	}
+	return 0;
+}
+
+static int kv_enable_auto_thermal_throttling(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	pi->graphics_therm_throttle_enable = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsThermThrottleEnable),
+				   &pi->graphics_therm_throttle_enable,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static int kv_upload_dpm_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsLevel),
+				   (u8 *)&pi->graphics_level,
+				   sizeof(SMU7_Fusion_GraphicsLevel) * SMU7_MAX_LEVELS_GRAPHICS,
+				   pi->sram_end);
+
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, GraphicsDpmLevelCount),
+				   &pi->graphics_dpm_level_count,
+				   sizeof(u8), pi->sram_end);
+
+	return ret;
+}
+
+static u32 kv_get_clock_difference(u32 a, u32 b)
+{
+	return (a >= b) ? a - b : b - a;
+}
+
+static u32 kv_get_clk_bypass(struct amdgpu_device *adev, u32 clk)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 value;
+
+	if (pi->caps_enable_dfs_bypass) {
+		if (kv_get_clock_difference(clk, 40000) < 200)
+			value = 3;
+		else if (kv_get_clock_difference(clk, 30000) < 200)
+			value = 2;
+		else if (kv_get_clock_difference(clk, 20000) < 200)
+			value = 7;
+		else if (kv_get_clock_difference(clk, 15000) < 200)
+			value = 6;
+		else if (kv_get_clock_difference(clk, 10000) < 200)
+			value = 8;
+		else
+			value = 0;
+	} else {
+		value = 0;
+	}
+
+	return value;
+}
+
+static int kv_populate_uvd_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_uvd_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+	int ret;
+	u32 i;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->uvd_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		if (pi->high_voltage_t &&
+		    (pi->high_voltage_t < table->entries[i].v))
+			break;
+
+		pi->uvd_level[i].VclkFrequency = cpu_to_be32(table->entries[i].vclk);
+		pi->uvd_level[i].DclkFrequency = cpu_to_be32(table->entries[i].dclk);
+		pi->uvd_level[i].MinVddNb = cpu_to_be16(table->entries[i].v);
+
+		pi->uvd_level[i].VClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].vclk);
+		pi->uvd_level[i].DClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].dclk);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].vclk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->uvd_level[i].VclkDivider = (u8)dividers.post_div;
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].dclk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->uvd_level[i].DclkDivider = (u8)dividers.post_div;
+
+		pi->uvd_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, UvdLevelCount),
+				   (u8 *)&pi->uvd_level_count,
+				   sizeof(u8), pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->uvd_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, UVDInterval),
+				   &pi->uvd_interval,
+				   sizeof(u8), pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, UvdLevel),
+				   (u8 *)&pi->uvd_level,
+				   sizeof(SMU7_Fusion_UvdLevel) * SMU7_MAX_LEVELS_UVD,
+				   pi->sram_end);
+
+	return ret;
+
+}
+
+static int kv_populate_vce_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+	u32 i;
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->vce_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		if (pi->high_voltage_t &&
+		    pi->high_voltage_t < table->entries[i].v)
+			break;
+
+		pi->vce_level[i].Frequency = cpu_to_be32(table->entries[i].evclk);
+		pi->vce_level[i].MinVoltage = cpu_to_be16(table->entries[i].v);
+
+		pi->vce_level[i].ClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].evclk);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].evclk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->vce_level[i].Divider = (u8)dividers.post_div;
+
+		pi->vce_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, VceLevelCount),
+				   (u8 *)&pi->vce_level_count,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->vce_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, VCEInterval),
+				   (u8 *)&pi->vce_interval,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, VceLevel),
+				   (u8 *)&pi->vce_level,
+				   sizeof(SMU7_Fusion_ExtClkLevel) * SMU7_MAX_LEVELS_VCE,
+				   pi->sram_end);
+
+	return ret;
+}
+
+static int kv_populate_samu_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+	int ret;
+	u32 i;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->samu_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		if (pi->high_voltage_t &&
+		    pi->high_voltage_t < table->entries[i].v)
+			break;
+
+		pi->samu_level[i].Frequency = cpu_to_be32(table->entries[i].clk);
+		pi->samu_level[i].MinVoltage = cpu_to_be16(table->entries[i].v);
+
+		pi->samu_level[i].ClkBypassCntl =
+			(u8)kv_get_clk_bypass(adev, table->entries[i].clk);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].clk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->samu_level[i].Divider = (u8)dividers.post_div;
+
+		pi->samu_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, SamuLevelCount),
+				   (u8 *)&pi->samu_level_count,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->samu_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, SAMUInterval),
+				   (u8 *)&pi->samu_interval,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, SamuLevel),
+				   (u8 *)&pi->samu_level,
+				   sizeof(SMU7_Fusion_ExtClkLevel) * SMU7_MAX_LEVELS_SAMU,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return ret;
+}
+
+
+static int kv_populate_acp_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+	struct atom_clock_dividers dividers;
+	int ret;
+	u32 i;
+
+	if (table == NULL || table->count == 0)
+		return 0;
+
+	pi->acp_level_count = 0;
+	for (i = 0; i < table->count; i++) {
+		pi->acp_level[i].Frequency = cpu_to_be32(table->entries[i].clk);
+		pi->acp_level[i].MinVoltage = cpu_to_be16(table->entries[i].v);
+
+		ret = amdgpu_atombios_get_clock_dividers(adev, COMPUTE_ENGINE_PLL_PARAM,
+							 table->entries[i].clk, false, &dividers);
+		if (ret)
+			return ret;
+		pi->acp_level[i].Divider = (u8)dividers.post_div;
+
+		pi->acp_level_count++;
+	}
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, AcpLevelCount),
+				   (u8 *)&pi->acp_level_count,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	pi->acp_interval = 1;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, ACPInterval),
+				   (u8 *)&pi->acp_interval,
+				   sizeof(u8),
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_kv_copy_bytes_to_smc(adev,
+				   pi->dpm_table_start +
+				   offsetof(SMU7_Fusion_DpmTable, AcpLevel),
+				   (u8 *)&pi->acp_level,
+				   sizeof(SMU7_Fusion_ExtClkLevel) * SMU7_MAX_LEVELS_ACP,
+				   pi->sram_end);
+	if (ret)
+		return ret;
+
+	return ret;
+}
+
+static void kv_calculate_dfs_bypass_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		for (i = 0; i < pi->graphics_dpm_level_count; i++) {
+			if (pi->caps_enable_dfs_bypass) {
+				if (kv_get_clock_difference(table->entries[i].clk, 40000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 3;
+				else if (kv_get_clock_difference(table->entries[i].clk, 30000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 2;
+				else if (kv_get_clock_difference(table->entries[i].clk, 26600) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 7;
+				else if (kv_get_clock_difference(table->entries[i].clk , 20000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 6;
+				else if (kv_get_clock_difference(table->entries[i].clk , 10000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 8;
+				else
+					pi->graphics_level[i].ClkBypassCntl = 0;
+			} else {
+				pi->graphics_level[i].ClkBypassCntl = 0;
+			}
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+		for (i = 0; i < pi->graphics_dpm_level_count; i++) {
+			if (pi->caps_enable_dfs_bypass) {
+				if (kv_get_clock_difference(table->entries[i].sclk_frequency, 40000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 3;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 30000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 2;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 26600) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 7;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 20000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 6;
+				else if (kv_get_clock_difference(table->entries[i].sclk_frequency, 10000) < 200)
+					pi->graphics_level[i].ClkBypassCntl = 8;
+				else
+					pi->graphics_level[i].ClkBypassCntl = 0;
+			} else {
+				pi->graphics_level[i].ClkBypassCntl = 0;
+			}
+		}
+	}
+}
+
+static int kv_enable_ulv(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_EnableULV : PPSMC_MSG_DisableULV);
+}
+
+static void kv_reset_acp_boot_level(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->acp_boot_level = 0xff;
+}
+
+static void kv_update_current_ps(struct amdgpu_device *adev,
+				 struct amdgpu_ps *rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->current_rps = *rps;
+	pi->current_ps = *new_ps;
+	pi->current_rps.ps_priv = &pi->current_ps;
+}
+
+static void kv_update_requested_ps(struct amdgpu_device *adev,
+				   struct amdgpu_ps *rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->requested_rps = *rps;
+	pi->requested_ps = *new_ps;
+	pi->requested_rps.ps_priv = &pi->requested_ps;
+}
+
+static void kv_dpm_enable_bapm(struct amdgpu_device *adev, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	if (pi->bapm_enable) {
+		ret = amdgpu_kv_smc_bapm_enable(adev, enable);
+		if (ret)
+			DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n");
+	}
+}
+
+static int kv_dpm_enable(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	ret = kv_process_firmware_header(adev);
+	if (ret) {
+		DRM_ERROR("kv_process_firmware_header failed\n");
+		return ret;
+	}
+	kv_init_fps_limits(adev);
+	kv_init_graphics_levels(adev);
+	ret = kv_program_bootup_state(adev);
+	if (ret) {
+		DRM_ERROR("kv_program_bootup_state failed\n");
+		return ret;
+	}
+	kv_calculate_dfs_bypass_settings(adev);
+	ret = kv_upload_dpm_settings(adev);
+	if (ret) {
+		DRM_ERROR("kv_upload_dpm_settings failed\n");
+		return ret;
+	}
+	ret = kv_populate_uvd_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_uvd_table failed\n");
+		return ret;
+	}
+	ret = kv_populate_vce_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_vce_table failed\n");
+		return ret;
+	}
+	ret = kv_populate_samu_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_samu_table failed\n");
+		return ret;
+	}
+	ret = kv_populate_acp_table(adev);
+	if (ret) {
+		DRM_ERROR("kv_populate_acp_table failed\n");
+		return ret;
+	}
+	kv_program_vc(adev);
+#if 0
+	kv_initialize_hardware_cac_manager(adev);
+#endif
+	kv_start_am(adev);
+	if (pi->enable_auto_thermal_throttling) {
+		ret = kv_enable_auto_thermal_throttling(adev);
+		if (ret) {
+			DRM_ERROR("kv_enable_auto_thermal_throttling failed\n");
+			return ret;
+		}
+	}
+	ret = kv_enable_dpm_voltage_scaling(adev);
+	if (ret) {
+		DRM_ERROR("kv_enable_dpm_voltage_scaling failed\n");
+		return ret;
+	}
+	ret = kv_set_dpm_interval(adev);
+	if (ret) {
+		DRM_ERROR("kv_set_dpm_interval failed\n");
+		return ret;
+	}
+	ret = kv_set_dpm_boot_state(adev);
+	if (ret) {
+		DRM_ERROR("kv_set_dpm_boot_state failed\n");
+		return ret;
+	}
+	ret = kv_enable_ulv(adev, true);
+	if (ret) {
+		DRM_ERROR("kv_enable_ulv failed\n");
+		return ret;
+	}
+	kv_start_dpm(adev);
+	ret = kv_enable_didt(adev, true);
+	if (ret) {
+		DRM_ERROR("kv_enable_didt failed\n");
+		return ret;
+	}
+	ret = kv_enable_smc_cac(adev, true);
+	if (ret) {
+		DRM_ERROR("kv_enable_smc_cac failed\n");
+		return ret;
+	}
+
+	kv_reset_acp_boot_level(adev);
+
+	ret = amdgpu_kv_smc_bapm_enable(adev, false);
+	if (ret) {
+		DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n");
+		return ret;
+	}
+
+	kv_update_current_ps(adev, adev->pm.dpm.boot_ps);
+
+	if (adev->irq.installed &&
+	    amdgpu_is_internal_thermal_sensor(adev->pm.int_thermal_type)) {
+		ret = kv_set_thermal_temperature_range(adev, KV_TEMP_RANGE_MIN, KV_TEMP_RANGE_MAX);
+		if (ret) {
+			DRM_ERROR("kv_set_thermal_temperature_range failed\n");
+			return ret;
+		}
+		amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+		amdgpu_irq_get(adev, &adev->pm.dpm.thermal.irq,
+			       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+	}
+
+	return ret;
+}
+
+static void kv_dpm_disable(struct amdgpu_device *adev)
+{
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
+	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
+		       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
+
+	amdgpu_kv_smc_bapm_enable(adev, false);
+
+	if (adev->asic_type == CHIP_MULLINS)
+		kv_enable_nb_dpm(adev, false);
+
+	/* powerup blocks */
+	kv_dpm_powergate_acp(adev, false);
+	kv_dpm_powergate_samu(adev, false);
+	kv_dpm_powergate_vce(adev, false);
+	kv_dpm_powergate_uvd(adev, false);
+
+	kv_enable_smc_cac(adev, false);
+	kv_enable_didt(adev, false);
+	kv_clear_vc(adev);
+	kv_stop_dpm(adev);
+	kv_enable_ulv(adev, false);
+	kv_reset_am(adev);
+
+	kv_update_current_ps(adev, adev->pm.dpm.boot_ps);
+}
+
+#if 0
+static int kv_write_smc_soft_register(struct amdgpu_device *adev,
+				      u16 reg_offset, u32 value)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	return amdgpu_kv_copy_bytes_to_smc(adev, pi->soft_regs_start + reg_offset,
+				    (u8 *)&value, sizeof(u16), pi->sram_end);
+}
+
+static int kv_read_smc_soft_register(struct amdgpu_device *adev,
+				     u16 reg_offset, u32 *value)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	return amdgpu_kv_read_smc_sram_dword(adev, pi->soft_regs_start + reg_offset,
+				      value, pi->sram_end);
+}
+#endif
+
+static void kv_init_sclk_t(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->low_sclk_interrupt_t = 0;
+}
+
+static int kv_init_fps_limits(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+
+	if (pi->caps_fps) {
+		u16 tmp;
+
+		tmp = 45;
+		pi->fps_high_t = cpu_to_be16(tmp);
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, FpsHighT),
+					   (u8 *)&pi->fps_high_t,
+					   sizeof(u16), pi->sram_end);
+
+		tmp = 30;
+		pi->fps_low_t = cpu_to_be16(tmp);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, FpsLowT),
+					   (u8 *)&pi->fps_low_t,
+					   sizeof(u16), pi->sram_end);
+
+	}
+	return ret;
+}
+
+static void kv_init_powergate_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->uvd_power_gated = false;
+	pi->vce_power_gated = false;
+	pi->samu_power_gated = false;
+	pi->acp_power_gated = false;
+
+}
+
+static int kv_enable_uvd_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_UVDDPM_Enable : PPSMC_MSG_UVDDPM_Disable);
+}
+
+static int kv_enable_vce_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_VCEDPM_Enable : PPSMC_MSG_VCEDPM_Disable);
+}
+
+static int kv_enable_samu_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_SAMUDPM_Enable : PPSMC_MSG_SAMUDPM_Disable);
+}
+
+static int kv_enable_acp_dpm(struct amdgpu_device *adev, bool enable)
+{
+	return amdgpu_kv_notify_message_to_smu(adev, enable ?
+					PPSMC_MSG_ACPDPM_Enable : PPSMC_MSG_ACPDPM_Disable);
+}
+
+static int kv_update_uvd_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_uvd_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	int ret;
+	u32 mask;
+
+	if (!gate) {
+		if (table->count)
+			pi->uvd_boot_level = table->count - 1;
+		else
+			pi->uvd_boot_level = 0;
+
+		if (!pi->caps_uvd_dpm || pi->caps_stable_p_state) {
+			mask = 1 << pi->uvd_boot_level;
+		} else {
+			mask = 0x1f;
+		}
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, UvdBootLevel),
+					   (uint8_t *)&pi->uvd_boot_level,
+					   sizeof(u8), pi->sram_end);
+		if (ret)
+			return ret;
+
+		amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+						  PPSMC_MSG_UVDDPM_SetEnabledMask,
+						  mask);
+	}
+
+	return kv_enable_uvd_dpm(adev, !gate);
+}
+
+static u8 kv_get_vce_boot_level(struct amdgpu_device *adev, u32 evclk)
+{
+	u8 i;
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].evclk >= evclk)
+			break;
+	}
+
+	return i;
+}
+
+static int kv_update_vce_dpm(struct amdgpu_device *adev,
+			     struct amdgpu_ps *amdgpu_new_state,
+			     struct amdgpu_ps *amdgpu_current_state)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_vce_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	int ret;
+
+	if (amdgpu_new_state->evclk > 0 && amdgpu_current_state->evclk == 0) {
+		kv_dpm_powergate_vce(adev, false);
+		/* turn the clocks on when encoding */
+		ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						    AMD_CG_STATE_UNGATE);
+		if (ret)
+			return ret;
+		if (pi->caps_stable_p_state)
+			pi->vce_boot_level = table->count - 1;
+		else
+			pi->vce_boot_level = kv_get_vce_boot_level(adev, amdgpu_new_state->evclk);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, VceBootLevel),
+					   (u8 *)&pi->vce_boot_level,
+					   sizeof(u8),
+					   pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (pi->caps_stable_p_state)
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_VCEDPM_SetEnabledMask,
+							  (1 << pi->vce_boot_level));
+
+		kv_enable_vce_dpm(adev, true);
+	} else if (amdgpu_new_state->evclk == 0 && amdgpu_current_state->evclk > 0) {
+		kv_enable_vce_dpm(adev, false);
+		/* turn the clocks off when not encoding */
+		ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+						    AMD_CG_STATE_GATE);
+		if (ret)
+			return ret;
+		kv_dpm_powergate_vce(adev, true);
+	}
+
+	return 0;
+}
+
+static int kv_update_samu_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table;
+	int ret;
+
+	if (!gate) {
+		if (pi->caps_stable_p_state)
+			pi->samu_boot_level = table->count - 1;
+		else
+			pi->samu_boot_level = 0;
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, SamuBootLevel),
+					   (u8 *)&pi->samu_boot_level,
+					   sizeof(u8),
+					   pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (pi->caps_stable_p_state)
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_SAMUDPM_SetEnabledMask,
+							  (1 << pi->samu_boot_level));
+	}
+
+	return kv_enable_samu_dpm(adev, !gate);
+}
+
+static u8 kv_get_acp_boot_level(struct amdgpu_device *adev)
+{
+	u8 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+
+	for (i = 0; i < table->count; i++) {
+		if (table->entries[i].clk >= 0) /* XXX */
+			break;
+	}
+
+	if (i >= table->count)
+		i = table->count - 1;
+
+	return i;
+}
+
+static void kv_update_acp_boot_level(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u8 acp_boot_level;
+
+	if (!pi->caps_stable_p_state) {
+		acp_boot_level = kv_get_acp_boot_level(adev);
+		if (acp_boot_level != pi->acp_boot_level) {
+			pi->acp_boot_level = acp_boot_level;
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_ACPDPM_SetEnabledMask,
+							  (1 << pi->acp_boot_level));
+		}
+	}
+}
+
+static int kv_update_acp_dpm(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+	int ret;
+
+	if (!gate) {
+		if (pi->caps_stable_p_state)
+			pi->acp_boot_level = table->count - 1;
+		else
+			pi->acp_boot_level = kv_get_acp_boot_level(adev);
+
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   pi->dpm_table_start +
+					   offsetof(SMU7_Fusion_DpmTable, AcpBootLevel),
+					   (u8 *)&pi->acp_boot_level,
+					   sizeof(u8),
+					   pi->sram_end);
+		if (ret)
+			return ret;
+
+		if (pi->caps_stable_p_state)
+			amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+							  PPSMC_MSG_ACPDPM_SetEnabledMask,
+							  (1 << pi->acp_boot_level));
+	}
+
+	return kv_enable_acp_dpm(adev, !gate);
+}
+
+static void kv_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	if (pi->uvd_power_gated == gate)
+		return;
+
+	pi->uvd_power_gated = gate;
+
+	if (gate) {
+		if (pi->caps_uvd_pg) {
+			/* disable clockgating so we can properly shut down the block */
+			ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_CG_STATE_UNGATE);
+			/* shutdown the UVD block */
+			ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_PG_STATE_GATE);
+			/* XXX: check for errors */
+		}
+		kv_update_uvd_dpm(adev, gate);
+		if (pi->caps_uvd_pg)
+			/* power off the UVD block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_UVDPowerOFF);
+	} else {
+		if (pi->caps_uvd_pg) {
+			/* power on the UVD block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_UVDPowerON);
+			/* re-init the UVD block */
+			ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_PG_STATE_UNGATE);
+			/* enable clockgating. hw will dynamically gate/ungate clocks on the fly */
+			ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD,
+							    AMD_CG_STATE_GATE);
+			/* XXX: check for errors */
+		}
+		kv_update_uvd_dpm(adev, gate);
+	}
+}
+
+static void kv_dpm_powergate_vce(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret;
+
+	if (pi->vce_power_gated == gate)
+		return;
+
+	pi->vce_power_gated = gate;
+
+	if (gate) {
+		if (pi->caps_vce_pg) {
+			/* shutdown the VCE block */
+			ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_PG_STATE_GATE);
+			/* XXX: check for errors */
+			/* power off the VCE block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerOFF);
+		}
+	} else {
+		if (pi->caps_vce_pg) {
+			/* power on the VCE block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerON);
+			/* re-init the VCE block */
+			ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							    AMD_PG_STATE_UNGATE);
+			/* XXX: check for errors */
+		}
+	}
+}
+
+static void kv_dpm_powergate_samu(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->samu_power_gated == gate)
+		return;
+
+	pi->samu_power_gated = gate;
+
+	if (gate) {
+		kv_update_samu_dpm(adev, true);
+		if (pi->caps_samu_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_SAMPowerOFF);
+	} else {
+		if (pi->caps_samu_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_SAMPowerON);
+		kv_update_samu_dpm(adev, false);
+	}
+}
+
+static void kv_dpm_powergate_acp(struct amdgpu_device *adev, bool gate)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->acp_power_gated == gate)
+		return;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return;
+
+	pi->acp_power_gated = gate;
+
+	if (gate) {
+		kv_update_acp_dpm(adev, true);
+		if (pi->caps_acp_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_ACPPowerOFF);
+	} else {
+		if (pi->caps_acp_pg)
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_ACPPowerON);
+		kv_update_acp_dpm(adev, false);
+	}
+}
+
+static void kv_set_valid_clock_range(struct amdgpu_device *adev,
+				     struct amdgpu_ps *new_rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		for (i = 0; i < pi->graphics_dpm_level_count; i++) {
+			if ((table->entries[i].clk >= new_ps->levels[0].sclk) ||
+			    (i == (pi->graphics_dpm_level_count - 1))) {
+				pi->lowest_valid = i;
+				break;
+			}
+		}
+
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].clk <= new_ps->levels[new_ps->num_levels - 1].sclk)
+				break;
+		}
+		pi->highest_valid = i;
+
+		if (pi->lowest_valid > pi->highest_valid) {
+			if ((new_ps->levels[0].sclk - table->entries[pi->highest_valid].clk) >
+			    (table->entries[pi->lowest_valid].clk - new_ps->levels[new_ps->num_levels - 1].sclk))
+				pi->highest_valid = pi->lowest_valid;
+			else
+				pi->lowest_valid =  pi->highest_valid;
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		for (i = 0; i < (int)pi->graphics_dpm_level_count; i++) {
+			if (table->entries[i].sclk_frequency >= new_ps->levels[0].sclk ||
+			    i == (int)(pi->graphics_dpm_level_count - 1)) {
+				pi->lowest_valid = i;
+				break;
+			}
+		}
+
+		for (i = pi->graphics_dpm_level_count - 1; i > 0; i--) {
+			if (table->entries[i].sclk_frequency <=
+			    new_ps->levels[new_ps->num_levels - 1].sclk)
+				break;
+		}
+		pi->highest_valid = i;
+
+		if (pi->lowest_valid > pi->highest_valid) {
+			if ((new_ps->levels[0].sclk -
+			     table->entries[pi->highest_valid].sclk_frequency) >
+			    (table->entries[pi->lowest_valid].sclk_frequency -
+			     new_ps->levels[new_ps->num_levels -1].sclk))
+				pi->highest_valid = pi->lowest_valid;
+			else
+				pi->lowest_valid =  pi->highest_valid;
+		}
+	}
+}
+
+static int kv_update_dfs_bypass_settings(struct amdgpu_device *adev,
+					 struct amdgpu_ps *new_rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+	u8 clk_bypass_cntl;
+
+	if (pi->caps_enable_dfs_bypass) {
+		clk_bypass_cntl = new_ps->need_dfs_bypass ?
+			pi->graphics_level[pi->graphics_boot_level].ClkBypassCntl : 0;
+		ret = amdgpu_kv_copy_bytes_to_smc(adev,
+					   (pi->dpm_table_start +
+					    offsetof(SMU7_Fusion_DpmTable, GraphicsLevel) +
+					    (pi->graphics_boot_level * sizeof(SMU7_Fusion_GraphicsLevel)) +
+					    offsetof(SMU7_Fusion_GraphicsLevel, ClkBypassCntl)),
+					   &clk_bypass_cntl,
+					   sizeof(u8), pi->sram_end);
+	}
+
+	return ret;
+}
+
+static int kv_enable_nb_dpm(struct amdgpu_device *adev,
+			    bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	int ret = 0;
+
+	if (enable) {
+		if (pi->enable_nb_dpm && !pi->nb_dpm_enabled) {
+			ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_NBDPM_Enable);
+			if (ret == 0)
+				pi->nb_dpm_enabled = true;
+		}
+	} else {
+		if (pi->enable_nb_dpm && pi->nb_dpm_enabled) {
+			ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_NBDPM_Disable);
+			if (ret == 0)
+				pi->nb_dpm_enabled = false;
+		}
+	}
+
+	return ret;
+}
+
+static int kv_dpm_force_performance_level(struct amdgpu_device *adev,
+					  enum amdgpu_dpm_forced_level level)
+{
+	int ret;
+
+	if (level == AMDGPU_DPM_FORCED_LEVEL_HIGH) {
+		ret = kv_force_dpm_highest(adev);
+		if (ret)
+			return ret;
+	} else if (level == AMDGPU_DPM_FORCED_LEVEL_LOW) {
+		ret = kv_force_dpm_lowest(adev);
+		if (ret)
+			return ret;
+	} else if (level == AMDGPU_DPM_FORCED_LEVEL_AUTO) {
+		ret = kv_unforce_levels(adev);
+		if (ret)
+			return ret;
+	}
+
+	adev->pm.dpm.forced_level = level;
+
+	return 0;
+}
+
+static int kv_dpm_pre_set_power_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_ps requested_ps = *adev->pm.dpm.requested_ps;
+	struct amdgpu_ps *new_ps = &requested_ps;
+
+	kv_update_requested_ps(adev, new_ps);
+
+	kv_apply_state_adjust_rules(adev,
+				    &pi->requested_rps,
+				    &pi->current_rps);
+
+	return 0;
+}
+
+static int kv_dpm_set_power_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+	struct amdgpu_ps *old_ps = &pi->current_rps;
+	int ret;
+
+	if (pi->bapm_enable) {
+		ret = amdgpu_kv_smc_bapm_enable(adev, adev->pm.dpm.ac_power);
+		if (ret) {
+			DRM_ERROR("amdgpu_kv_smc_bapm_enable failed\n");
+			return ret;
+		}
+	}
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		if (pi->enable_dpm) {
+			kv_set_valid_clock_range(adev, new_ps);
+			kv_update_dfs_bypass_settings(adev, new_ps);
+			ret = kv_calculate_ds_divider(adev);
+			if (ret) {
+				DRM_ERROR("kv_calculate_ds_divider failed\n");
+				return ret;
+			}
+			kv_calculate_nbps_level_settings(adev);
+			kv_calculate_dpm_settings(adev);
+			kv_force_lowest_valid(adev);
+			kv_enable_new_levels(adev);
+			kv_upload_dpm_settings(adev);
+			kv_program_nbps_index_settings(adev, new_ps);
+			kv_unforce_levels(adev);
+			kv_set_enabled_levels(adev);
+			kv_force_lowest_valid(adev);
+			kv_unforce_levels(adev);
+
+			ret = kv_update_vce_dpm(adev, new_ps, old_ps);
+			if (ret) {
+				DRM_ERROR("kv_update_vce_dpm failed\n");
+				return ret;
+			}
+			kv_update_sclk_t(adev);
+			if (adev->asic_type == CHIP_MULLINS)
+				kv_enable_nb_dpm(adev, true);
+		}
+	} else {
+		if (pi->enable_dpm) {
+			kv_set_valid_clock_range(adev, new_ps);
+			kv_update_dfs_bypass_settings(adev, new_ps);
+			ret = kv_calculate_ds_divider(adev);
+			if (ret) {
+				DRM_ERROR("kv_calculate_ds_divider failed\n");
+				return ret;
+			}
+			kv_calculate_nbps_level_settings(adev);
+			kv_calculate_dpm_settings(adev);
+			kv_freeze_sclk_dpm(adev, true);
+			kv_upload_dpm_settings(adev);
+			kv_program_nbps_index_settings(adev, new_ps);
+			kv_freeze_sclk_dpm(adev, false);
+			kv_set_enabled_levels(adev);
+			ret = kv_update_vce_dpm(adev, new_ps, old_ps);
+			if (ret) {
+				DRM_ERROR("kv_update_vce_dpm failed\n");
+				return ret;
+			}
+			kv_update_acp_boot_level(adev);
+			kv_update_sclk_t(adev);
+			kv_enable_nb_dpm(adev, true);
+		}
+	}
+
+	return 0;
+}
+
+static void kv_dpm_post_set_power_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_ps *new_ps = &pi->requested_rps;
+
+	kv_update_current_ps(adev, new_ps);
+}
+
+static void kv_dpm_setup_asic(struct amdgpu_device *adev)
+{
+	sumo_take_smu_control(adev, true);
+	kv_init_powergate_state(adev);
+	kv_init_sclk_t(adev);
+}
+
+#if 0
+static void kv_dpm_reset_asic(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		kv_force_lowest_valid(adev);
+		kv_init_graphics_levels(adev);
+		kv_program_bootup_state(adev);
+		kv_upload_dpm_settings(adev);
+		kv_force_lowest_valid(adev);
+		kv_unforce_levels(adev);
+	} else {
+		kv_init_graphics_levels(adev);
+		kv_program_bootup_state(adev);
+		kv_freeze_sclk_dpm(adev, true);
+		kv_upload_dpm_settings(adev);
+		kv_freeze_sclk_dpm(adev, false);
+		kv_set_enabled_level(adev, pi->graphics_boot_level);
+	}
+}
+#endif
+
+static void kv_construct_max_power_limits_table(struct amdgpu_device *adev,
+						struct amdgpu_clock_and_voltage_limits *table)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	if (pi->sys_info.sclk_voltage_mapping_table.num_max_dpm_entries > 0) {
+		int idx = pi->sys_info.sclk_voltage_mapping_table.num_max_dpm_entries - 1;
+		table->sclk =
+			pi->sys_info.sclk_voltage_mapping_table.entries[idx].sclk_frequency;
+		table->vddc =
+			kv_convert_2bit_index_to_voltage(adev,
+							 pi->sys_info.sclk_voltage_mapping_table.entries[idx].vid_2bit);
+	}
+
+	table->mclk = pi->sys_info.nbp_memory_clock[0];
+}
+
+static void kv_patch_voltage_values(struct amdgpu_device *adev)
+{
+	int i;
+	struct amdgpu_uvd_clock_voltage_dependency_table *uvd_table =
+		&adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+	struct amdgpu_vce_clock_voltage_dependency_table *vce_table =
+		&adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table *samu_table =
+		&adev->pm.dpm.dyn_state.samu_clock_voltage_dependency_table;
+	struct amdgpu_clock_voltage_dependency_table *acp_table =
+		&adev->pm.dpm.dyn_state.acp_clock_voltage_dependency_table;
+
+	if (uvd_table->count) {
+		for (i = 0; i < uvd_table->count; i++)
+			uvd_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 uvd_table->entries[i].v);
+	}
+
+	if (vce_table->count) {
+		for (i = 0; i < vce_table->count; i++)
+			vce_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 vce_table->entries[i].v);
+	}
+
+	if (samu_table->count) {
+		for (i = 0; i < samu_table->count; i++)
+			samu_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 samu_table->entries[i].v);
+	}
+
+	if (acp_table->count) {
+		for (i = 0; i < acp_table->count; i++)
+			acp_table->entries[i].v =
+				kv_convert_8bit_index_to_voltage(adev,
+								 acp_table->entries[i].v);
+	}
+
+}
+
+static void kv_construct_boot_state(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->boot_pl.sclk = pi->sys_info.bootup_sclk;
+	pi->boot_pl.vddc_index = pi->sys_info.bootup_nb_voltage_index;
+	pi->boot_pl.ds_divider_index = 0;
+	pi->boot_pl.ss_divider_index = 0;
+	pi->boot_pl.allow_gnb_slow = 1;
+	pi->boot_pl.force_nbp_state = 0;
+	pi->boot_pl.display_wm = 0;
+	pi->boot_pl.vce_wm = 0;
+}
+
+static int kv_force_dpm_highest(struct amdgpu_device *adev)
+{
+	int ret;
+	u32 enable_mask, i;
+
+	ret = amdgpu_kv_dpm_get_enable_mask(adev, &enable_mask);
+	if (ret)
+		return ret;
+
+	for (i = SMU7_MAX_LEVELS_GRAPHICS - 1; i > 0; i--) {
+		if (enable_mask & (1 << i))
+			break;
+	}
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return amdgpu_kv_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DPM_ForceState, i);
+	else
+		return kv_set_enabled_level(adev, i);
+}
+
+static int kv_force_dpm_lowest(struct amdgpu_device *adev)
+{
+	int ret;
+	u32 enable_mask, i;
+
+	ret = amdgpu_kv_dpm_get_enable_mask(adev, &enable_mask);
+	if (ret)
+		return ret;
+
+	for (i = 0; i < SMU7_MAX_LEVELS_GRAPHICS; i++) {
+		if (enable_mask & (1 << i))
+			break;
+	}
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return amdgpu_kv_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DPM_ForceState, i);
+	else
+		return kv_set_enabled_level(adev, i);
+}
+
+static u8 kv_get_sleep_divider_id_from_clock(struct amdgpu_device *adev,
+					     u32 sclk, u32 min_sclk_in_sr)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	u32 temp;
+	u32 min = (min_sclk_in_sr > KV_MINIMUM_ENGINE_CLOCK) ?
+		min_sclk_in_sr : KV_MINIMUM_ENGINE_CLOCK;
+
+	if (sclk < min)
+		return 0;
+
+	if (!pi->caps_sclk_ds)
+		return 0;
+
+	for (i = KV_MAX_DEEPSLEEP_DIVIDER_ID; i > 0; i--) {
+		temp = sclk / sumo_get_sleep_divider_from_id(i);
+		if (temp >= min)
+			break;
+	}
+
+	return (u8)i;
+}
+
+static int kv_get_high_voltage_limit(struct amdgpu_device *adev, int *limit)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	int i;
+
+	if (table && table->count) {
+		for (i = table->count - 1; i >= 0; i--) {
+			if (pi->high_voltage_t &&
+			    (kv_convert_8bit_index_to_voltage(adev, table->entries[i].v) <=
+			     pi->high_voltage_t)) {
+				*limit = i;
+				return 0;
+			}
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		for (i = table->num_max_dpm_entries - 1; i >= 0; i--) {
+			if (pi->high_voltage_t &&
+			    (kv_convert_2bit_index_to_voltage(adev, table->entries[i].vid_2bit) <=
+			     pi->high_voltage_t)) {
+				*limit = i;
+				return 0;
+			}
+		}
+	}
+
+	*limit = 0;
+	return 0;
+}
+
+static void kv_apply_state_adjust_rules(struct amdgpu_device *adev,
+					struct amdgpu_ps *new_rps,
+					struct amdgpu_ps *old_rps)
+{
+	struct kv_ps *ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 min_sclk = 10000; /* ??? */
+	u32 sclk, mclk = 0;
+	int i, limit;
+	bool force_high;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+	u32 stable_p_state_sclk = 0;
+	struct amdgpu_clock_and_voltage_limits *max_limits =
+		&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+
+	if (new_rps->vce_active) {
+		new_rps->evclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].evclk;
+		new_rps->ecclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].ecclk;
+	} else {
+		new_rps->evclk = 0;
+		new_rps->ecclk = 0;
+	}
+
+	mclk = max_limits->mclk;
+	sclk = min_sclk;
+
+	if (pi->caps_stable_p_state) {
+		stable_p_state_sclk = (max_limits->sclk * 75) / 100;
+
+		for (i = table->count - 1; i >= 0; i++) {
+			if (stable_p_state_sclk >= table->entries[i].clk) {
+				stable_p_state_sclk = table->entries[i].clk;
+				break;
+			}
+		}
+
+		if (i > 0)
+			stable_p_state_sclk = table->entries[0].clk;
+
+		sclk = stable_p_state_sclk;
+	}
+
+	if (new_rps->vce_active) {
+		if (sclk < adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk)
+			sclk = adev->pm.dpm.vce_states[adev->pm.dpm.vce_level].sclk;
+	}
+
+	ps->need_dfs_bypass = true;
+
+	for (i = 0; i < ps->num_levels; i++) {
+		if (ps->levels[i].sclk < sclk)
+			ps->levels[i].sclk = sclk;
+	}
+
+	if (table && table->count) {
+		for (i = 0; i < ps->num_levels; i++) {
+			if (pi->high_voltage_t &&
+			    (pi->high_voltage_t <
+			     kv_convert_8bit_index_to_voltage(adev, ps->levels[i].vddc_index))) {
+				kv_get_high_voltage_limit(adev, &limit);
+				ps->levels[i].sclk = table->entries[limit].clk;
+			}
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		for (i = 0; i < ps->num_levels; i++) {
+			if (pi->high_voltage_t &&
+			    (pi->high_voltage_t <
+			     kv_convert_8bit_index_to_voltage(adev, ps->levels[i].vddc_index))) {
+				kv_get_high_voltage_limit(adev, &limit);
+				ps->levels[i].sclk = table->entries[limit].sclk_frequency;
+			}
+		}
+	}
+
+	if (pi->caps_stable_p_state) {
+		for (i = 0; i < ps->num_levels; i++) {
+			ps->levels[i].sclk = stable_p_state_sclk;
+		}
+	}
+
+	pi->video_start = new_rps->dclk || new_rps->vclk ||
+		new_rps->evclk || new_rps->ecclk;
+
+	if ((new_rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) ==
+	    ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
+		pi->battery_state = true;
+	else
+		pi->battery_state = false;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		ps->dpm0_pg_nb_ps_lo = 0x1;
+		ps->dpm0_pg_nb_ps_hi = 0x0;
+		ps->dpmx_nb_ps_lo = 0x1;
+		ps->dpmx_nb_ps_hi = 0x0;
+	} else {
+		ps->dpm0_pg_nb_ps_lo = 0x3;
+		ps->dpm0_pg_nb_ps_hi = 0x0;
+		ps->dpmx_nb_ps_lo = 0x3;
+		ps->dpmx_nb_ps_hi = 0x0;
+
+		if (pi->sys_info.nb_dpm_enable) {
+			force_high = (mclk >= pi->sys_info.nbp_memory_clock[3]) ||
+				pi->video_start || (adev->pm.dpm.new_active_crtc_count >= 3) ||
+				pi->disable_nb_ps3_in_battery;
+			ps->dpm0_pg_nb_ps_lo = force_high ? 0x2 : 0x3;
+			ps->dpm0_pg_nb_ps_hi = 0x2;
+			ps->dpmx_nb_ps_lo = force_high ? 0x2 : 0x3;
+			ps->dpmx_nb_ps_hi = 0x2;
+		}
+	}
+}
+
+static void kv_dpm_power_level_enabled_for_throttle(struct amdgpu_device *adev,
+						    u32 index, bool enable)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	pi->graphics_level[index].EnabledForThrottle = enable ? 1 : 0;
+}
+
+static int kv_calculate_ds_divider(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 sclk_in_sr = 10000; /* ??? */
+	u32 i;
+
+	if (pi->lowest_valid > pi->highest_valid)
+		return -EINVAL;
+
+	for (i = pi->lowest_valid; i <= pi->highest_valid; i++) {
+		pi->graphics_level[i].DeepSleepDivId =
+			kv_get_sleep_divider_id_from_clock(adev,
+							   be32_to_cpu(pi->graphics_level[i].SclkFrequency),
+							   sclk_in_sr);
+	}
+	return 0;
+}
+
+static int kv_calculate_nbps_level_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	bool force_high;
+	struct amdgpu_clock_and_voltage_limits *max_limits =
+		&adev->pm.dpm.dyn_state.max_clock_voltage_on_ac;
+	u32 mclk = max_limits->mclk;
+
+	if (pi->lowest_valid > pi->highest_valid)
+		return -EINVAL;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS) {
+		for (i = pi->lowest_valid; i <= pi->highest_valid; i++) {
+			pi->graphics_level[i].GnbSlow = 1;
+			pi->graphics_level[i].ForceNbPs1 = 0;
+			pi->graphics_level[i].UpH = 0;
+		}
+
+		if (!pi->sys_info.nb_dpm_enable)
+			return 0;
+
+		force_high = ((mclk >= pi->sys_info.nbp_memory_clock[3]) ||
+			      (adev->pm.dpm.new_active_crtc_count >= 3) || pi->video_start);
+
+		if (force_high) {
+			for (i = pi->lowest_valid; i <= pi->highest_valid; i++)
+				pi->graphics_level[i].GnbSlow = 0;
+		} else {
+			if (pi->battery_state)
+				pi->graphics_level[0].ForceNbPs1 = 1;
+
+			pi->graphics_level[1].GnbSlow = 0;
+			pi->graphics_level[2].GnbSlow = 0;
+			pi->graphics_level[3].GnbSlow = 0;
+			pi->graphics_level[4].GnbSlow = 0;
+		}
+	} else {
+		for (i = pi->lowest_valid; i <= pi->highest_valid; i++) {
+			pi->graphics_level[i].GnbSlow = 1;
+			pi->graphics_level[i].ForceNbPs1 = 0;
+			pi->graphics_level[i].UpH = 0;
+		}
+
+		if (pi->sys_info.nb_dpm_enable && pi->battery_state) {
+			pi->graphics_level[pi->lowest_valid].UpH = 0x28;
+			pi->graphics_level[pi->lowest_valid].GnbSlow = 0;
+			if (pi->lowest_valid != pi->highest_valid)
+				pi->graphics_level[pi->lowest_valid].ForceNbPs1 = 1;
+		}
+	}
+	return 0;
+}
+
+static int kv_calculate_dpm_settings(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+
+	if (pi->lowest_valid > pi->highest_valid)
+		return -EINVAL;
+
+	for (i = pi->lowest_valid; i <= pi->highest_valid; i++)
+		pi->graphics_level[i].DisplayWatermark = (i == pi->highest_valid) ? 1 : 0;
+
+	return 0;
+}
+
+static void kv_init_graphics_levels(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+	struct amdgpu_clock_voltage_dependency_table *table =
+		&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
+
+	if (table && table->count) {
+		u32 vid_2bit;
+
+		pi->graphics_dpm_level_count = 0;
+		for (i = 0; i < table->count; i++) {
+			if (pi->high_voltage_t &&
+			    (pi->high_voltage_t <
+			     kv_convert_8bit_index_to_voltage(adev, table->entries[i].v)))
+				break;
+
+			kv_set_divider_value(adev, i, table->entries[i].clk);
+			vid_2bit = kv_convert_vid7_to_vid2(adev,
+							   &pi->sys_info.vid_mapping_table,
+							   table->entries[i].v);
+			kv_set_vid(adev, i, vid_2bit);
+			kv_set_at(adev, i, pi->at[i]);
+			kv_dpm_power_level_enabled_for_throttle(adev, i, true);
+			pi->graphics_dpm_level_count++;
+		}
+	} else {
+		struct sumo_sclk_voltage_mapping_table *table =
+			&pi->sys_info.sclk_voltage_mapping_table;
+
+		pi->graphics_dpm_level_count = 0;
+		for (i = 0; i < table->num_max_dpm_entries; i++) {
+			if (pi->high_voltage_t &&
+			    pi->high_voltage_t <
+			    kv_convert_2bit_index_to_voltage(adev, table->entries[i].vid_2bit))
+				break;
+
+			kv_set_divider_value(adev, i, table->entries[i].sclk_frequency);
+			kv_set_vid(adev, i, table->entries[i].vid_2bit);
+			kv_set_at(adev, i, pi->at[i]);
+			kv_dpm_power_level_enabled_for_throttle(adev, i, true);
+			pi->graphics_dpm_level_count++;
+		}
+	}
+
+	for (i = 0; i < SMU7_MAX_LEVELS_GRAPHICS; i++)
+		kv_dpm_power_level_enable(adev, i, false);
+}
+
+static void kv_enable_new_levels(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i;
+
+	for (i = 0; i < SMU7_MAX_LEVELS_GRAPHICS; i++) {
+		if (i >= pi->lowest_valid && i <= pi->highest_valid)
+			kv_dpm_power_level_enable(adev, i, true);
+	}
+}
+
+static int kv_set_enabled_level(struct amdgpu_device *adev, u32 level)
+{
+	u32 new_mask = (1 << level);
+
+	return amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+						 PPSMC_MSG_SCLKDPM_SetEnabledMask,
+						 new_mask);
+}
+
+static int kv_set_enabled_levels(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 i, new_mask = 0;
+
+	for (i = pi->lowest_valid; i <= pi->highest_valid; i++)
+		new_mask |= (1 << i);
+
+	return amdgpu_kv_send_msg_to_smc_with_parameter(adev,
+						 PPSMC_MSG_SCLKDPM_SetEnabledMask,
+						 new_mask);
+}
+
+static void kv_program_nbps_index_settings(struct amdgpu_device *adev,
+					   struct amdgpu_ps *new_rps)
+{
+	struct kv_ps *new_ps = kv_get_ps(new_rps);
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 nbdpmconfig1;
+
+	if (adev->asic_type == CHIP_KABINI || adev->asic_type == CHIP_MULLINS)
+		return;
+
+	if (pi->sys_info.nb_dpm_enable) {
+		nbdpmconfig1 = RREG32_SMC(ixNB_DPM_CONFIG_1);
+		nbdpmconfig1 &= ~(NB_DPM_CONFIG_1__Dpm0PgNbPsLo_MASK |
+				NB_DPM_CONFIG_1__Dpm0PgNbPsHi_MASK |
+				NB_DPM_CONFIG_1__DpmXNbPsLo_MASK |
+				NB_DPM_CONFIG_1__DpmXNbPsHi_MASK);
+		nbdpmconfig1 |= (new_ps->dpm0_pg_nb_ps_lo << NB_DPM_CONFIG_1__Dpm0PgNbPsLo__SHIFT) |
+				(new_ps->dpm0_pg_nb_ps_hi << NB_DPM_CONFIG_1__Dpm0PgNbPsHi__SHIFT) |
+				(new_ps->dpmx_nb_ps_lo << NB_DPM_CONFIG_1__DpmXNbPsLo__SHIFT) |
+				(new_ps->dpmx_nb_ps_hi << NB_DPM_CONFIG_1__DpmXNbPsHi__SHIFT);
+		WREG32_SMC(ixNB_DPM_CONFIG_1, nbdpmconfig1);
+	}
+}
+
+static int kv_set_thermal_temperature_range(struct amdgpu_device *adev,
+					    int min_temp, int max_temp)
+{
+	int low_temp = 0 * 1000;
+	int high_temp = 255 * 1000;
+	u32 tmp;
+
+	if (low_temp < min_temp)
+		low_temp = min_temp;
+	if (high_temp > max_temp)
+		high_temp = max_temp;
+	if (high_temp < low_temp) {
+		DRM_ERROR("invalid thermal range: %d - %d\n", low_temp, high_temp);
+		return -EINVAL;
+	}
+
+	tmp = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+	tmp &= ~(CG_THERMAL_INT_CTRL__DIG_THERM_INTH_MASK |
+		CG_THERMAL_INT_CTRL__DIG_THERM_INTL_MASK);
+	tmp |= ((49 + (high_temp / 1000)) << CG_THERMAL_INT_CTRL__DIG_THERM_INTH__SHIFT) |
+		((49 + (low_temp / 1000)) << CG_THERMAL_INT_CTRL__DIG_THERM_INTL__SHIFT);
+	WREG32_SMC(ixCG_THERMAL_INT_CTRL, tmp);
+
+	adev->pm.dpm.thermal.min_temp = low_temp;
+	adev->pm.dpm.thermal.max_temp = high_temp;
+
+	return 0;
+}
+
+union igp_info {
+	struct _ATOM_INTEGRATED_SYSTEM_INFO info;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V2 info_2;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V5 info_5;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V6 info_6;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 info_7;
+	struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_8 info_8;
+};
+
+static int kv_parse_sys_info_table(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, IntegratedSystemInfo);
+	union igp_info *igp_info;
+	u8 frev, crev;
+	u16 data_offset;
+	int i;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		igp_info = (union igp_info *)(mode_info->atom_context->bios +
+					      data_offset);
+
+		if (crev != 8) {
+			DRM_ERROR("Unsupported IGP table: %d %d\n", frev, crev);
+			return -EINVAL;
+		}
+		pi->sys_info.bootup_sclk = le32_to_cpu(igp_info->info_8.ulBootUpEngineClock);
+		pi->sys_info.bootup_uma_clk = le32_to_cpu(igp_info->info_8.ulBootUpUMAClock);
+		pi->sys_info.bootup_nb_voltage_index =
+			le16_to_cpu(igp_info->info_8.usBootUpNBVoltage);
+		if (igp_info->info_8.ucHtcTmpLmt == 0)
+			pi->sys_info.htc_tmp_lmt = 203;
+		else
+			pi->sys_info.htc_tmp_lmt = igp_info->info_8.ucHtcTmpLmt;
+		if (igp_info->info_8.ucHtcHystLmt == 0)
+			pi->sys_info.htc_hyst_lmt = 5;
+		else
+			pi->sys_info.htc_hyst_lmt = igp_info->info_8.ucHtcHystLmt;
+		if (pi->sys_info.htc_tmp_lmt <= pi->sys_info.htc_hyst_lmt) {
+			DRM_ERROR("The htcTmpLmt should be larger than htcHystLmt.\n");
+		}
+
+		if (le32_to_cpu(igp_info->info_8.ulSystemConfig) & (1 << 3))
+			pi->sys_info.nb_dpm_enable = true;
+		else
+			pi->sys_info.nb_dpm_enable = false;
+
+		for (i = 0; i < KV_NUM_NBPSTATES; i++) {
+			pi->sys_info.nbp_memory_clock[i] =
+				le32_to_cpu(igp_info->info_8.ulNbpStateMemclkFreq[i]);
+			pi->sys_info.nbp_n_clock[i] =
+				le32_to_cpu(igp_info->info_8.ulNbpStateNClkFreq[i]);
+		}
+		if (le32_to_cpu(igp_info->info_8.ulGPUCapInfo) &
+		    SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS)
+			pi->caps_enable_dfs_bypass = true;
+
+		sumo_construct_sclk_voltage_mapping_table(adev,
+							  &pi->sys_info.sclk_voltage_mapping_table,
+							  igp_info->info_8.sAvail_SCLK);
+
+		sumo_construct_vid_mapping_table(adev,
+						 &pi->sys_info.vid_mapping_table,
+						 igp_info->info_8.sAvail_SCLK);
+
+		kv_construct_max_power_limits_table(adev,
+						    &adev->pm.dpm.dyn_state.max_clock_voltage_on_ac);
+	}
+	return 0;
+}
+
+union power_info {
+	struct _ATOM_POWERPLAY_INFO info;
+	struct _ATOM_POWERPLAY_INFO_V2 info_2;
+	struct _ATOM_POWERPLAY_INFO_V3 info_3;
+	struct _ATOM_PPLIB_POWERPLAYTABLE pplib;
+	struct _ATOM_PPLIB_POWERPLAYTABLE2 pplib2;
+	struct _ATOM_PPLIB_POWERPLAYTABLE3 pplib3;
+};
+
+union pplib_clock_info {
+	struct _ATOM_PPLIB_R600_CLOCK_INFO r600;
+	struct _ATOM_PPLIB_RS780_CLOCK_INFO rs780;
+	struct _ATOM_PPLIB_EVERGREEN_CLOCK_INFO evergreen;
+	struct _ATOM_PPLIB_SUMO_CLOCK_INFO sumo;
+};
+
+union pplib_power_state {
+	struct _ATOM_PPLIB_STATE v1;
+	struct _ATOM_PPLIB_STATE_V2 v2;
+};
+
+static void kv_patch_boot_state(struct amdgpu_device *adev,
+				struct kv_ps *ps)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	ps->num_levels = 1;
+	ps->levels[0] = pi->boot_pl;
+}
+
+static void kv_parse_pplib_non_clock_info(struct amdgpu_device *adev,
+					  struct amdgpu_ps *rps,
+					  struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info,
+					  u8 table_rev)
+{
+	struct kv_ps *ps = kv_get_ps(rps);
+
+	rps->caps = le32_to_cpu(non_clock_info->ulCapsAndSettings);
+	rps->class = le16_to_cpu(non_clock_info->usClassification);
+	rps->class2 = le16_to_cpu(non_clock_info->usClassification2);
+
+	if (ATOM_PPLIB_NONCLOCKINFO_VER1 < table_rev) {
+		rps->vclk = le32_to_cpu(non_clock_info->ulVCLK);
+		rps->dclk = le32_to_cpu(non_clock_info->ulDCLK);
+	} else {
+		rps->vclk = 0;
+		rps->dclk = 0;
+	}
+
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_BOOT) {
+		adev->pm.dpm.boot_ps = rps;
+		kv_patch_boot_state(adev, ps);
+	}
+	if (rps->class & ATOM_PPLIB_CLASSIFICATION_UVDSTATE)
+		adev->pm.dpm.uvd_ps = rps;
+}
+
+static void kv_parse_pplib_clock_info(struct amdgpu_device *adev,
+				      struct amdgpu_ps *rps, int index,
+					union pplib_clock_info *clock_info)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct kv_ps *ps = kv_get_ps(rps);
+	struct kv_pl *pl = &ps->levels[index];
+	u32 sclk;
+
+	sclk = le16_to_cpu(clock_info->sumo.usEngineClockLow);
+	sclk |= clock_info->sumo.ucEngineClockHigh << 16;
+	pl->sclk = sclk;
+	pl->vddc_index = clock_info->sumo.vddcIndex;
+
+	ps->num_levels = index + 1;
+
+	if (pi->caps_sclk_ds) {
+		pl->ds_divider_index = 5;
+		pl->ss_divider_index = 5;
+	}
+}
+
+static int kv_parse_power_table(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	struct _ATOM_PPLIB_NONCLOCK_INFO *non_clock_info;
+	union pplib_power_state *power_state;
+	int i, j, k, non_clock_array_index, clock_array_index;
+	union pplib_clock_info *clock_info;
+	struct _StateArray *state_array;
+	struct _ClockInfoArray *clock_info_array;
+	struct _NonClockInfoArray *non_clock_info_array;
+	union power_info *power_info;
+	int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo);
+	u16 data_offset;
+	u8 frev, crev;
+	u8 *power_state_offset;
+	struct kv_ps *ps;
+
+	if (!amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset))
+		return -EINVAL;
+	power_info = (union power_info *)(mode_info->atom_context->bios + data_offset);
+
+	amdgpu_add_thermal_controller(adev);
+
+	state_array = (struct _StateArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usStateArrayOffset));
+	clock_info_array = (struct _ClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usClockInfoArrayOffset));
+	non_clock_info_array = (struct _NonClockInfoArray *)
+		(mode_info->atom_context->bios + data_offset +
+		 le16_to_cpu(power_info->pplib.usNonClockInfoArrayOffset));
+
+	adev->pm.dpm.ps = kzalloc(sizeof(struct amdgpu_ps) *
+				  state_array->ucNumEntries, GFP_KERNEL);
+	if (!adev->pm.dpm.ps)
+		return -ENOMEM;
+	power_state_offset = (u8 *)state_array->states;
+	for (i = 0; i < state_array->ucNumEntries; i++) {
+		u8 *idx;
+		power_state = (union pplib_power_state *)power_state_offset;
+		non_clock_array_index = power_state->v2.nonClockInfoIndex;
+		non_clock_info = (struct _ATOM_PPLIB_NONCLOCK_INFO *)
+			&non_clock_info_array->nonClockInfo[non_clock_array_index];
+		ps = kzalloc(sizeof(struct kv_ps), GFP_KERNEL);
+		if (ps == NULL) {
+			kfree(adev->pm.dpm.ps);
+			return -ENOMEM;
+		}
+		adev->pm.dpm.ps[i].ps_priv = ps;
+		k = 0;
+		idx = (u8 *)&power_state->v2.clockInfoIndex[0];
+		for (j = 0; j < power_state->v2.ucNumDPMLevels; j++) {
+			clock_array_index = idx[j];
+			if (clock_array_index >= clock_info_array->ucNumEntries)
+				continue;
+			if (k >= SUMO_MAX_HARDWARE_POWERLEVELS)
+				break;
+			clock_info = (union pplib_clock_info *)
+				((u8 *)&clock_info_array->clockInfo[0] +
+				 (clock_array_index * clock_info_array->ucEntrySize));
+			kv_parse_pplib_clock_info(adev,
+						  &adev->pm.dpm.ps[i], k,
+						  clock_info);
+			k++;
+		}
+		kv_parse_pplib_non_clock_info(adev, &adev->pm.dpm.ps[i],
+					      non_clock_info,
+					      non_clock_info_array->ucEntrySize);
+		power_state_offset += 2 + power_state->v2.ucNumDPMLevels;
+	}
+	adev->pm.dpm.num_ps = state_array->ucNumEntries;
+
+	/* fill in the vce power states */
+	for (i = 0; i < AMDGPU_MAX_VCE_LEVELS; i++) {
+		u32 sclk;
+		clock_array_index = adev->pm.dpm.vce_states[i].clk_idx;
+		clock_info = (union pplib_clock_info *)
+			&clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize];
+		sclk = le16_to_cpu(clock_info->sumo.usEngineClockLow);
+		sclk |= clock_info->sumo.ucEngineClockHigh << 16;
+		adev->pm.dpm.vce_states[i].sclk = sclk;
+		adev->pm.dpm.vce_states[i].mclk = 0;
+	}
+
+	return 0;
+}
+
+static int kv_dpm_init(struct amdgpu_device *adev)
+{
+	struct kv_power_info *pi;
+	int ret, i;
+
+	pi = kzalloc(sizeof(struct kv_power_info), GFP_KERNEL);
+	if (pi == NULL)
+		return -ENOMEM;
+	adev->pm.dpm.priv = pi;
+
+	ret = amdgpu_get_platform_caps(adev);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_parse_extended_power_table(adev);
+	if (ret)
+		return ret;
+
+	for (i = 0; i < SUMO_MAX_HARDWARE_POWERLEVELS; i++)
+		pi->at[i] = TRINITY_AT_DFLT;
+
+	pi->sram_end = SMC_RAM_END;
+
+	pi->enable_nb_dpm = true;
+
+	pi->caps_power_containment = true;
+	pi->caps_cac = true;
+	pi->enable_didt = false;
+	if (pi->enable_didt) {
+		pi->caps_sq_ramping = true;
+		pi->caps_db_ramping = true;
+		pi->caps_td_ramping = true;
+		pi->caps_tcp_ramping = true;
+	}
+
+	pi->caps_sclk_ds = true;
+	pi->enable_auto_thermal_throttling = true;
+	pi->disable_nb_ps3_in_battery = false;
+	if (amdgpu_bapm == 0)
+		pi->bapm_enable = false;
+	else
+		pi->bapm_enable = true;
+	pi->voltage_drop_t = 0;
+	pi->caps_sclk_throttle_low_notification = false;
+	pi->caps_fps = false; /* true? */
+	pi->caps_uvd_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_UVD) ? true : false;
+	pi->caps_uvd_dpm = true;
+	pi->caps_vce_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_VCE) ? true : false;
+	pi->caps_samu_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_SAMU) ? true : false;
+	pi->caps_acp_pg = (adev->pg_flags & AMDGPU_PG_SUPPORT_ACP) ? true : false;
+	pi->caps_stable_p_state = false;
+
+	ret = kv_parse_sys_info_table(adev);
+	if (ret)
+		return ret;
+
+	kv_patch_voltage_values(adev);
+	kv_construct_boot_state(adev);
+
+	ret = kv_parse_power_table(adev);
+	if (ret)
+		return ret;
+
+	pi->enable_dpm = true;
+
+	return 0;
+}
+
+static void
+kv_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev,
+					       struct seq_file *m)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	u32 current_index =
+		(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
+		TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
+		TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
+	u32 sclk, tmp;
+	u16 vddc;
+
+	if (current_index >= SMU__NUM_SCLK_DPM_STATE) {
+		seq_printf(m, "invalid dpm profile %d\n", current_index);
+	} else {
+		sclk = be32_to_cpu(pi->graphics_level[current_index].SclkFrequency);
+		tmp = (RREG32_SMC(ixSMU_VOLTAGE_STATUS) &
+			SMU_VOLTAGE_STATUS__SMU_VOLTAGE_CURRENT_LEVEL_MASK) >>
+			SMU_VOLTAGE_STATUS__SMU_VOLTAGE_CURRENT_LEVEL__SHIFT;
+		vddc = kv_convert_8bit_index_to_voltage(adev, (u16)tmp);
+		seq_printf(m, "uvd    %sabled\n", pi->uvd_power_gated ? "dis" : "en");
+		seq_printf(m, "vce    %sabled\n", pi->vce_power_gated ? "dis" : "en");
+		seq_printf(m, "power level %d    sclk: %u vddc: %u\n",
+			   current_index, sclk, vddc);
+	}
+}
+
+static void
+kv_dpm_print_power_state(struct amdgpu_device *adev,
+			 struct amdgpu_ps *rps)
+{
+	int i;
+	struct kv_ps *ps = kv_get_ps(rps);
+
+	amdgpu_dpm_print_class_info(rps->class, rps->class2);
+	amdgpu_dpm_print_cap_info(rps->caps);
+	printk("\tuvd    vclk: %d dclk: %d\n", rps->vclk, rps->dclk);
+	for (i = 0; i < ps->num_levels; i++) {
+		struct kv_pl *pl = &ps->levels[i];
+		printk("\t\tpower level %d    sclk: %u vddc: %u\n",
+		       i, pl->sclk,
+		       kv_convert_8bit_index_to_voltage(adev, pl->vddc_index));
+	}
+	amdgpu_dpm_print_ps_status(adev, rps);
+}
+
+static void kv_dpm_fini(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->pm.dpm.num_ps; i++) {
+		kfree(adev->pm.dpm.ps[i].ps_priv);
+	}
+	kfree(adev->pm.dpm.ps);
+	kfree(adev->pm.dpm.priv);
+	amdgpu_free_extended_power_table(adev);
+}
+
+static void kv_dpm_display_configuration_changed(struct amdgpu_device *adev)
+{
+
+}
+
+static u32 kv_dpm_get_sclk(struct amdgpu_device *adev, bool low)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+	struct kv_ps *requested_state = kv_get_ps(&pi->requested_rps);
+
+	if (low)
+		return requested_state->levels[0].sclk;
+	else
+		return requested_state->levels[requested_state->num_levels - 1].sclk;
+}
+
+static u32 kv_dpm_get_mclk(struct amdgpu_device *adev, bool low)
+{
+	struct kv_power_info *pi = kv_get_pi(adev);
+
+	return pi->sys_info.bootup_uma_clk;
+}
+
+/* get temperature in millidegrees */
+static int kv_dpm_get_temp(struct amdgpu_device *adev)
+{
+	u32 temp;
+	int actual_temp = 0;
+
+	temp = RREG32_SMC(0xC0300E0C);
+
+	if (temp)
+		actual_temp = (temp / 8) - 49;
+	else
+		actual_temp = 0;
+
+	actual_temp = actual_temp * 1000;
+
+	return actual_temp;
+}
+
+static int kv_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	kv_dpm_set_dpm_funcs(adev);
+	kv_dpm_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int kv_dpm_late_init(void *handle)
+{
+	/* powerdown unused blocks for now */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int ret;
+
+	if (!amdgpu_dpm)
+		return 0;
+
+	/* init the sysfs and debugfs files late */
+	ret = amdgpu_pm_sysfs_init(adev);
+	if (ret)
+		return ret;
+
+	kv_dpm_powergate_acp(adev, true);
+	kv_dpm_powergate_samu(adev, true);
+	kv_dpm_powergate_vce(adev, true);
+	kv_dpm_powergate_uvd(adev, true);
+
+	return 0;
+}
+
+static int kv_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = amdgpu_irq_add_id(adev, 230, &adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	ret = amdgpu_irq_add_id(adev, 231, &adev->pm.dpm.thermal.irq);
+	if (ret)
+		return ret;
+
+	/* default to balanced state */
+	adev->pm.dpm.state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.user_state = POWER_STATE_TYPE_BALANCED;
+	adev->pm.dpm.forced_level = AMDGPU_DPM_FORCED_LEVEL_AUTO;
+	adev->pm.default_sclk = adev->clock.default_sclk;
+	adev->pm.default_mclk = adev->clock.default_mclk;
+	adev->pm.current_sclk = adev->clock.default_sclk;
+	adev->pm.current_mclk = adev->clock.default_mclk;
+	adev->pm.int_thermal_type = THERMAL_TYPE_NONE;
+
+	if (amdgpu_dpm == 0)
+		return 0;
+
+	INIT_WORK(&adev->pm.dpm.thermal.work, amdgpu_dpm_thermal_work_handler);
+	mutex_lock(&adev->pm.mutex);
+	ret = kv_dpm_init(adev);
+	if (ret)
+		goto dpm_failed;
+	adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+	if (amdgpu_dpm == 1)
+		amdgpu_pm_print_power_states(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_INFO("amdgpu: dpm initialized\n");
+
+	return 0;
+
+dpm_failed:
+	kv_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	DRM_ERROR("amdgpu: dpm initialization failed\n");
+	return ret;
+}
+
+static int kv_dpm_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+	amdgpu_pm_sysfs_fini(adev);
+	kv_dpm_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+
+	return 0;
+}
+
+static int kv_dpm_hw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+	kv_dpm_setup_asic(adev);
+	ret = kv_dpm_enable(adev);
+	if (ret)
+		adev->pm.dpm_enabled = false;
+	else
+		adev->pm.dpm_enabled = true;
+	mutex_unlock(&adev->pm.mutex);
+
+	return ret;
+}
+
+static int kv_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		kv_dpm_disable(adev);
+		mutex_unlock(&adev->pm.mutex);
+	}
+
+	return 0;
+}
+
+static int kv_dpm_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		mutex_lock(&adev->pm.mutex);
+		/* disable dpm */
+		kv_dpm_disable(adev);
+		/* reset the power state */
+		adev->pm.dpm.current_ps = adev->pm.dpm.requested_ps = adev->pm.dpm.boot_ps;
+		mutex_unlock(&adev->pm.mutex);
+	}
+	return 0;
+}
+
+static int kv_dpm_resume(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->pm.dpm_enabled) {
+		/* asic init will reset to the boot state */
+		mutex_lock(&adev->pm.mutex);
+		kv_dpm_setup_asic(adev);
+		ret = kv_dpm_enable(adev);
+		if (ret)
+			adev->pm.dpm_enabled = false;
+		else
+			adev->pm.dpm_enabled = true;
+		mutex_unlock(&adev->pm.mutex);
+		if (adev->pm.dpm_enabled)
+			amdgpu_pm_compute_clocks(adev);
+	}
+	return 0;
+}
+
+static bool kv_dpm_is_idle(void *handle)
+{
+	return true;
+}
+
+static int kv_dpm_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static void kv_dpm_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "KV/KB DPM registers\n");
+	dev_info(adev->dev, "  DIDT_SQ_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_SQ_CTRL0));
+	dev_info(adev->dev, "  DIDT_DB_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_DB_CTRL0));
+	dev_info(adev->dev, "  DIDT_TD_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_TD_CTRL0));
+	dev_info(adev->dev, "  DIDT_TCP_CTRL0=0x%08X\n",
+		 RREG32_DIDT(ixDIDT_TCP_CTRL0));
+	dev_info(adev->dev, "  LCAC_SX0_OVR_SEL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_SX0_OVR_SEL));
+	dev_info(adev->dev, "  LCAC_SX0_OVR_VAL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_SX0_OVR_VAL));
+	dev_info(adev->dev, "  LCAC_MC0_OVR_SEL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC0_OVR_SEL));
+	dev_info(adev->dev, "  LCAC_MC0_OVR_VAL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC0_OVR_VAL));
+	dev_info(adev->dev, "  LCAC_MC1_OVR_SEL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC1_OVR_SEL));
+	dev_info(adev->dev, "  LCAC_MC1_OVR_VAL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC1_OVR_VAL));
+	dev_info(adev->dev, "  LCAC_MC2_OVR_SEL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC2_OVR_SEL));
+	dev_info(adev->dev, "  LCAC_MC2_OVR_VAL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC2_OVR_VAL));
+	dev_info(adev->dev, "  LCAC_MC3_OVR_SEL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC3_OVR_SEL));
+	dev_info(adev->dev, "  LCAC_MC3_OVR_VAL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_MC3_OVR_VAL));
+	dev_info(adev->dev, "  LCAC_CPL_OVR_SEL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_CPL_OVR_SEL));
+	dev_info(adev->dev, "  LCAC_CPL_OVR_VAL=0x%08X\n",
+		 RREG32_SMC(ixLCAC_CPL_OVR_VAL));
+	dev_info(adev->dev, "  CG_FREQ_TRAN_VOTING_0=0x%08X\n",
+		 RREG32_SMC(ixCG_FREQ_TRAN_VOTING_0));
+	dev_info(adev->dev, "  GENERAL_PWRMGT=0x%08X\n",
+		 RREG32_SMC(ixGENERAL_PWRMGT));
+	dev_info(adev->dev, "  SCLK_PWRMGT_CNTL=0x%08X\n",
+		 RREG32_SMC(ixSCLK_PWRMGT_CNTL));
+	dev_info(adev->dev, "  SMC_MESSAGE_0=0x%08X\n",
+		 RREG32(mmSMC_MESSAGE_0));
+	dev_info(adev->dev, "  SMC_RESP_0=0x%08X\n",
+		 RREG32(mmSMC_RESP_0));
+	dev_info(adev->dev, "  SMC_MSG_ARG_0=0x%08X\n",
+		 RREG32(mmSMC_MSG_ARG_0));
+	dev_info(adev->dev, "  SMC_IND_INDEX_0=0x%08X\n",
+		 RREG32(mmSMC_IND_INDEX_0));
+	dev_info(adev->dev, "  SMC_IND_DATA_0=0x%08X\n",
+		 RREG32(mmSMC_IND_DATA_0));
+	dev_info(adev->dev, "  SMC_IND_ACCESS_CNTL=0x%08X\n",
+		 RREG32(mmSMC_IND_ACCESS_CNTL));
+}
+
+static int kv_dpm_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int kv_dpm_set_interrupt_state(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *src,
+				      unsigned type,
+				      enum amdgpu_interrupt_state state)
+{
+	u32 cg_thermal_int;
+
+	switch (type) {
+	case AMDGPU_THERMAL_IRQ_LOW_TO_HIGH:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTH_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	case AMDGPU_THERMAL_IRQ_HIGH_TO_LOW:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int &= ~CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			cg_thermal_int = RREG32_SMC(ixCG_THERMAL_INT_CTRL);
+			cg_thermal_int |= CG_THERMAL_INT_CTRL__THERM_INTL_MASK_MASK;
+			WREG32_SMC(ixCG_THERMAL_INT_CTRL, cg_thermal_int);
+			break;
+		default:
+			break;
+		}
+		break;
+
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int kv_dpm_process_interrupt(struct amdgpu_device *adev,
+				    struct amdgpu_irq_src *source,
+				    struct amdgpu_iv_entry *entry)
+{
+	bool queue_thermal = false;
+
+	if (entry == NULL)
+		return -EINVAL;
+
+	switch (entry->src_id) {
+	case 230: /* thermal low to high */
+		DRM_DEBUG("IH: thermal low to high\n");
+		adev->pm.dpm.thermal.high_to_low = false;
+		queue_thermal = true;
+		break;
+	case 231: /* thermal high to low */
+		DRM_DEBUG("IH: thermal high to low\n");
+		adev->pm.dpm.thermal.high_to_low = true;
+		queue_thermal = true;
+		break;
+	default:
+		break;
+	}
+
+	if (queue_thermal)
+		schedule_work(&adev->pm.dpm.thermal.work);
+
+	return 0;
+}
+
+static int kv_dpm_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int kv_dpm_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs kv_dpm_ip_funcs = {
+	.early_init = kv_dpm_early_init,
+	.late_init = kv_dpm_late_init,
+	.sw_init = kv_dpm_sw_init,
+	.sw_fini = kv_dpm_sw_fini,
+	.hw_init = kv_dpm_hw_init,
+	.hw_fini = kv_dpm_hw_fini,
+	.suspend = kv_dpm_suspend,
+	.resume = kv_dpm_resume,
+	.is_idle = kv_dpm_is_idle,
+	.wait_for_idle = kv_dpm_wait_for_idle,
+	.soft_reset = kv_dpm_soft_reset,
+	.print_status = kv_dpm_print_status,
+	.set_clockgating_state = kv_dpm_set_clockgating_state,
+	.set_powergating_state = kv_dpm_set_powergating_state,
+};
+
+static const struct amdgpu_dpm_funcs kv_dpm_funcs = {
+	.get_temperature = &kv_dpm_get_temp,
+	.pre_set_power_state = &kv_dpm_pre_set_power_state,
+	.set_power_state = &kv_dpm_set_power_state,
+	.post_set_power_state = &kv_dpm_post_set_power_state,
+	.display_configuration_changed = &kv_dpm_display_configuration_changed,
+	.get_sclk = &kv_dpm_get_sclk,
+	.get_mclk = &kv_dpm_get_mclk,
+	.print_power_state = &kv_dpm_print_power_state,
+	.debugfs_print_current_performance_level = &kv_dpm_debugfs_print_current_performance_level,
+	.force_performance_level = &kv_dpm_force_performance_level,
+	.powergate_uvd = &kv_dpm_powergate_uvd,
+	.enable_bapm = &kv_dpm_enable_bapm,
+};
+
+static void kv_dpm_set_dpm_funcs(struct amdgpu_device *adev)
+{
+	if (adev->pm.funcs == NULL)
+		adev->pm.funcs = &kv_dpm_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs kv_dpm_irq_funcs = {
+	.set = kv_dpm_set_interrupt_state,
+	.process = kv_dpm_process_interrupt,
+};
+
+static void kv_dpm_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->pm.dpm.thermal.irq.num_types = AMDGPU_THERMAL_IRQ_LAST;
+	adev->pm.dpm.thermal.irq.funcs = &kv_dpm_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.h b/kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.h
new file mode 100644
index 000000000..6df0ed413
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/kv_dpm.h
@@ -0,0 +1,229 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __KV_DPM_H__
+#define __KV_DPM_H__
+
+#define SMU__NUM_SCLK_DPM_STATE  8
+#define SMU__NUM_MCLK_DPM_LEVELS 4
+#define SMU__NUM_LCLK_DPM_LEVELS 8
+#define SMU__NUM_PCIE_DPM_LEVELS 0 /* ??? */
+#include "smu7_fusion.h"
+#include "ppsmc.h"
+
+#define SUMO_MAX_HARDWARE_POWERLEVELS 5
+
+#define SUMO_MAX_NUMBER_VOLTAGES    4
+
+struct sumo_vid_mapping_entry {
+	u16 vid_2bit;
+	u16 vid_7bit;
+};
+
+struct sumo_vid_mapping_table {
+	u32 num_entries;
+	struct sumo_vid_mapping_entry entries[SUMO_MAX_NUMBER_VOLTAGES];
+};
+
+struct sumo_sclk_voltage_mapping_entry {
+	u32 sclk_frequency;
+	u16 vid_2bit;
+	u16 rsv;
+};
+
+struct sumo_sclk_voltage_mapping_table {
+	u32 num_max_dpm_entries;
+	struct sumo_sclk_voltage_mapping_entry entries[SUMO_MAX_HARDWARE_POWERLEVELS];
+};
+
+#define TRINITY_AT_DFLT            30
+
+#define KV_NUM_NBPSTATES   4
+
+enum kv_pt_config_reg_type {
+	KV_CONFIGREG_MMR = 0,
+	KV_CONFIGREG_SMC_IND,
+	KV_CONFIGREG_DIDT_IND,
+	KV_CONFIGREG_CACHE,
+	KV_CONFIGREG_MAX
+};
+
+struct kv_pt_config_reg {
+	u32 offset;
+	u32 mask;
+	u32 shift;
+	u32 value;
+	enum kv_pt_config_reg_type type;
+};
+
+struct kv_lcac_config_values {
+	u32 block_id;
+	u32 signal_id;
+	u32 t;
+};
+
+struct kv_lcac_config_reg {
+	u32 cntl;
+	u32 block_mask;
+	u32 block_shift;
+	u32 signal_mask;
+	u32 signal_shift;
+	u32 t_mask;
+	u32 t_shift;
+	u32 enable_mask;
+	u32 enable_shift;
+};
+
+struct kv_pl {
+	u32 sclk;
+	u8 vddc_index;
+	u8 ds_divider_index;
+	u8 ss_divider_index;
+	u8 allow_gnb_slow;
+	u8 force_nbp_state;
+	u8 display_wm;
+	u8 vce_wm;
+};
+
+struct kv_ps {
+	struct kv_pl levels[SUMO_MAX_HARDWARE_POWERLEVELS];
+	u32 num_levels;
+	bool need_dfs_bypass;
+	u8 dpm0_pg_nb_ps_lo;
+	u8 dpm0_pg_nb_ps_hi;
+	u8 dpmx_nb_ps_lo;
+	u8 dpmx_nb_ps_hi;
+};
+
+struct kv_sys_info {
+	u32 bootup_uma_clk;
+	u32 bootup_sclk;
+	u32 dentist_vco_freq;
+	u32 nb_dpm_enable;
+	u32 nbp_memory_clock[KV_NUM_NBPSTATES];
+	u32 nbp_n_clock[KV_NUM_NBPSTATES];
+	u16 bootup_nb_voltage_index;
+	u8 htc_tmp_lmt;
+	u8 htc_hyst_lmt;
+	struct sumo_sclk_voltage_mapping_table sclk_voltage_mapping_table;
+	struct sumo_vid_mapping_table vid_mapping_table;
+	u32 uma_channel_number;
+};
+
+struct kv_power_info {
+	u32 at[SUMO_MAX_HARDWARE_POWERLEVELS];
+	u32 voltage_drop_t;
+	struct kv_sys_info sys_info;
+	struct kv_pl boot_pl;
+	bool enable_nb_ps_policy;
+	bool disable_nb_ps3_in_battery;
+	bool video_start;
+	bool battery_state;
+	u32 lowest_valid;
+	u32 highest_valid;
+	u16 high_voltage_t;
+	bool cac_enabled;
+	bool bapm_enable;
+	/* smc offsets */
+	u32 sram_end;
+	u32 dpm_table_start;
+	u32 soft_regs_start;
+	/* dpm SMU tables */
+	u8 graphics_dpm_level_count;
+	u8 uvd_level_count;
+	u8 vce_level_count;
+	u8 acp_level_count;
+	u8 samu_level_count;
+	u16 fps_high_t;
+	SMU7_Fusion_GraphicsLevel graphics_level[SMU__NUM_SCLK_DPM_STATE];
+	SMU7_Fusion_ACPILevel acpi_level;
+	SMU7_Fusion_UvdLevel uvd_level[SMU7_MAX_LEVELS_UVD];
+	SMU7_Fusion_ExtClkLevel vce_level[SMU7_MAX_LEVELS_VCE];
+	SMU7_Fusion_ExtClkLevel acp_level[SMU7_MAX_LEVELS_ACP];
+	SMU7_Fusion_ExtClkLevel samu_level[SMU7_MAX_LEVELS_SAMU];
+	u8 uvd_boot_level;
+	u8 vce_boot_level;
+	u8 acp_boot_level;
+	u8 samu_boot_level;
+	u8 uvd_interval;
+	u8 vce_interval;
+	u8 acp_interval;
+	u8 samu_interval;
+	u8 graphics_boot_level;
+	u8 graphics_interval;
+	u8 graphics_therm_throttle_enable;
+	u8 graphics_voltage_change_enable;
+	u8 graphics_clk_slow_enable;
+	u8 graphics_clk_slow_divider;
+	u8 fps_low_t;
+	u32 low_sclk_interrupt_t;
+	bool uvd_power_gated;
+	bool vce_power_gated;
+	bool acp_power_gated;
+	bool samu_power_gated;
+	bool nb_dpm_enabled;
+	/* flags */
+	bool enable_didt;
+	bool enable_dpm;
+	bool enable_auto_thermal_throttling;
+	bool enable_nb_dpm;
+	/* caps */
+	bool caps_cac;
+	bool caps_power_containment;
+	bool caps_sq_ramping;
+	bool caps_db_ramping;
+	bool caps_td_ramping;
+	bool caps_tcp_ramping;
+	bool caps_sclk_throttle_low_notification;
+	bool caps_fps;
+	bool caps_uvd_dpm;
+	bool caps_uvd_pg;
+	bool caps_vce_pg;
+	bool caps_samu_pg;
+	bool caps_acp_pg;
+	bool caps_stable_p_state;
+	bool caps_enable_dfs_bypass;
+	bool caps_sclk_ds;
+	struct amdgpu_ps current_rps;
+	struct kv_ps current_ps;
+	struct amdgpu_ps requested_rps;
+	struct kv_ps requested_ps;
+};
+
+/* XXX are these ok? */
+#define KV_TEMP_RANGE_MIN (90 * 1000)
+#define KV_TEMP_RANGE_MAX (120 * 1000)
+
+/* kv_smc.c */
+int amdgpu_kv_notify_message_to_smu(struct amdgpu_device *adev, u32 id);
+int amdgpu_kv_dpm_get_enable_mask(struct amdgpu_device *adev, u32 *enable_mask);
+int amdgpu_kv_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+				      PPSMC_Msg msg, u32 parameter);
+int amdgpu_kv_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+			   u32 *value, u32 limit);
+int amdgpu_kv_smc_dpm_enable(struct amdgpu_device *adev, bool enable);
+int amdgpu_kv_smc_bapm_enable(struct amdgpu_device *adev, bool enable);
+int amdgpu_kv_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/kv_smc.c b/kernel/drivers/gpu/drm/amd/amdgpu/kv_smc.c
new file mode 100644
index 000000000..e6b7b42ac
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/kv_smc.c
@@ -0,0 +1,219 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+
+#include "drmP.h"
+#include "amdgpu.h"
+#include "cikd.h"
+#include "kv_dpm.h"
+
+#include "smu/smu_7_0_0_d.h"
+#include "smu/smu_7_0_0_sh_mask.h"
+
+int amdgpu_kv_notify_message_to_smu(struct amdgpu_device *adev, u32 id)
+{
+	u32 i;
+	u32 tmp = 0;
+
+	WREG32(mmSMC_MESSAGE_0, id & SMC_MESSAGE_0__SMC_MSG_MASK);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if ((RREG32(mmSMC_RESP_0) & SMC_RESP_0__SMC_RESP_MASK) != 0)
+			break;
+		udelay(1);
+	}
+	tmp = RREG32(mmSMC_RESP_0) & SMC_RESP_0__SMC_RESP_MASK;
+
+	if (tmp != 1) {
+		if (tmp == 0xFF)
+			return -EINVAL;
+		else if (tmp == 0xFE)
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_kv_dpm_get_enable_mask(struct amdgpu_device *adev, u32 *enable_mask)
+{
+	int ret;
+
+	ret = amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_SCLKDPM_GetEnabledMask);
+
+	if (ret == 0)
+		*enable_mask = RREG32_SMC(ixSMC_SYSCON_MSG_ARG_0);
+
+	return ret;
+}
+
+int amdgpu_kv_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+				      PPSMC_Msg msg, u32 parameter)
+{
+
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return amdgpu_kv_notify_message_to_smu(adev, msg);
+}
+
+static int kv_set_smc_sram_address(struct amdgpu_device *adev,
+				   u32 smc_address, u32 limit)
+{
+	if (smc_address & 3)
+		return -EINVAL;
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+	WREG32_P(mmSMC_IND_ACCESS_CNTL, 0,
+			~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
+
+	return 0;
+}
+
+int amdgpu_kv_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+			   u32 *value, u32 limit)
+{
+	int ret;
+
+	ret = kv_set_smc_sram_address(adev, smc_address, limit);
+	if (ret)
+		return ret;
+
+	*value = RREG32(mmSMC_IND_DATA_0);
+	return 0;
+}
+
+int amdgpu_kv_smc_dpm_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DPM_Enable);
+	else
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DPM_Disable);
+}
+
+int amdgpu_kv_smc_bapm_enable(struct amdgpu_device *adev, bool enable)
+{
+	if (enable)
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_EnableBAPM);
+	else
+		return amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_DisableBAPM);
+}
+
+int amdgpu_kv_copy_bytes_to_smc(struct amdgpu_device *adev,
+			 u32 smc_start_address,
+			 const u8 *src, u32 byte_count, u32 limit)
+{
+	int ret;
+	u32 data, original_data, addr, extra_shift, t_byte, count, mask;
+
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+	t_byte = addr & 3;
+
+	/* RMW for the initial bytes */
+	if  (t_byte != 0) {
+		addr -= t_byte;
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		original_data = RREG32(mmSMC_IND_DATA_0);
+
+		data = 0;
+		mask = 0;
+		count = 4;
+		while (count > 0) {
+			if (t_byte > 0) {
+				mask = (mask << 8) | 0xff;
+				t_byte--;
+			} else if (byte_count > 0) {
+				data = (data << 8) + *src++;
+				byte_count--;
+				mask <<= 8;
+			} else {
+				data <<= 8;
+				mask = (mask << 8) | 0xff;
+			}
+			count--;
+		}
+
+		data |= original_data & mask;
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		addr += 4;
+	}
+
+	while (byte_count >= 4) {
+		/* SMC address space is BE */
+		data = (src[0] << 24) + (src[1] << 16) + (src[2] << 8) + src[3];
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	/* RMW for the final bytes */
+	if (byte_count > 0) {
+		data = 0;
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		original_data = RREG32(mmSMC_IND_DATA_0);
+
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			/* SMC address space is BE */
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+
+		data |= (original_data & ~((~0UL) << extra_shift));
+
+		ret = kv_set_smc_sram_address(adev, addr, limit);
+		if (ret)
+			return ret;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+	return 0;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/ppsmc.h b/kernel/drivers/gpu/drm/amd/amdgpu/ppsmc.h
new file mode 100644
index 000000000..7837f2ecc
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/ppsmc.h
@@ -0,0 +1,196 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef PP_SMC_H
+#define PP_SMC_H
+
+#pragma pack(push, 1)
+
+#define PPSMC_SWSTATE_FLAG_DC                           0x01
+#define PPSMC_SWSTATE_FLAG_UVD                          0x02
+#define PPSMC_SWSTATE_FLAG_VCE                          0x04
+#define PPSMC_SWSTATE_FLAG_PCIE_X1                      0x08
+
+#define PPSMC_THERMAL_PROTECT_TYPE_INTERNAL             0x00
+#define PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL             0x01
+#define PPSMC_THERMAL_PROTECT_TYPE_NONE                 0xff
+
+#define PPSMC_SYSTEMFLAG_GPIO_DC                        0x01
+#define PPSMC_SYSTEMFLAG_STEPVDDC                       0x02
+#define PPSMC_SYSTEMFLAG_GDDR5                          0x04
+#define PPSMC_SYSTEMFLAG_DISABLE_BABYSTEP               0x08
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT                  0x10
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_ANALOG           0x20
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_PROG_GPIO        0x40
+
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_MASK              0x07
+#define PPSMC_EXTRAFLAGS_AC2DC_DONT_WAIT_FOR_VBLANK     0x08
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTODPMLOWSTATE   0x00
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTOINITIALSTATE  0x01
+#define PPSMC_EXTRAFLAGS_AC2DC_GPIO5_POLARITY_HIGH      0x02
+
+#define PPSMC_DISPLAY_WATERMARK_LOW                     0
+#define PPSMC_DISPLAY_WATERMARK_HIGH                    1
+
+#define PPSMC_STATEFLAG_AUTO_PULSE_SKIP    0x01
+#define PPSMC_STATEFLAG_POWERBOOST         0x02
+#define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 0x20
+#define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS   0x40
+
+#define FDO_MODE_HARDWARE 0
+#define FDO_MODE_PIECE_WISE_LINEAR 1
+
+enum FAN_CONTROL {
+	FAN_CONTROL_FUZZY,
+	FAN_CONTROL_TABLE
+};
+
+#define PPSMC_Result_OK             ((uint8_t)0x01)
+#define PPSMC_Result_Failed         ((uint8_t)0xFF)
+
+typedef uint8_t PPSMC_Result;
+
+#define PPSMC_MSG_Halt                      ((uint8_t)0x10)
+#define PPSMC_MSG_Resume                    ((uint8_t)0x11)
+#define PPSMC_MSG_ZeroLevelsDisabled        ((uint8_t)0x13)
+#define PPSMC_MSG_OneLevelsDisabled         ((uint8_t)0x14)
+#define PPSMC_MSG_TwoLevelsDisabled         ((uint8_t)0x15)
+#define PPSMC_MSG_EnableThermalInterrupt    ((uint8_t)0x16)
+#define PPSMC_MSG_RunningOnAC               ((uint8_t)0x17)
+#define PPSMC_MSG_SwitchToSwState           ((uint8_t)0x20)
+#define PPSMC_MSG_SwitchToInitialState      ((uint8_t)0x40)
+#define PPSMC_MSG_NoForcedLevel             ((uint8_t)0x41)
+#define PPSMC_MSG_ForceHigh                 ((uint8_t)0x42)
+#define PPSMC_MSG_ForceMediumOrHigh         ((uint8_t)0x43)
+#define PPSMC_MSG_SwitchToMinimumPower      ((uint8_t)0x51)
+#define PPSMC_MSG_ResumeFromMinimumPower    ((uint8_t)0x52)
+#define PPSMC_MSG_EnableCac                 ((uint8_t)0x53)
+#define PPSMC_MSG_DisableCac                ((uint8_t)0x54)
+#define PPSMC_TDPClampingActive             ((uint8_t)0x59)
+#define PPSMC_TDPClampingInactive           ((uint8_t)0x5A)
+#define PPSMC_StartFanControl               ((uint8_t)0x5B)
+#define PPSMC_StopFanControl                ((uint8_t)0x5C)
+#define PPSMC_MSG_NoDisplay                 ((uint8_t)0x5D)
+#define PPSMC_MSG_HasDisplay                ((uint8_t)0x5E)
+#define PPSMC_MSG_UVDPowerOFF               ((uint8_t)0x60)
+#define PPSMC_MSG_UVDPowerON                ((uint8_t)0x61)
+#define PPSMC_MSG_EnableULV                 ((uint8_t)0x62)
+#define PPSMC_MSG_DisableULV                ((uint8_t)0x63)
+#define PPSMC_MSG_EnterULV                  ((uint8_t)0x64)
+#define PPSMC_MSG_ExitULV                   ((uint8_t)0x65)
+#define PPSMC_CACLongTermAvgEnable          ((uint8_t)0x6E)
+#define PPSMC_CACLongTermAvgDisable         ((uint8_t)0x6F)
+#define PPSMC_MSG_CollectCAC_PowerCorreln   ((uint8_t)0x7A)
+#define PPSMC_FlushDataCache                ((uint8_t)0x80)
+#define PPSMC_MSG_SetEnabledLevels          ((uint8_t)0x82)
+#define PPSMC_MSG_SetForcedLevels           ((uint8_t)0x83)
+#define PPSMC_MSG_ResetToDefaults           ((uint8_t)0x84)
+#define PPSMC_MSG_EnableDTE                 ((uint8_t)0x87)
+#define PPSMC_MSG_DisableDTE                ((uint8_t)0x88)
+#define PPSMC_MSG_ThrottleOVRDSCLKDS        ((uint8_t)0x96)
+#define PPSMC_MSG_CancelThrottleOVRDSCLKDS  ((uint8_t)0x97)
+
+/* CI/KV/KB */
+#define PPSMC_MSG_UVDDPM_SetEnabledMask       ((uint16_t) 0x12D)
+#define PPSMC_MSG_VCEDPM_SetEnabledMask       ((uint16_t) 0x12E)
+#define PPSMC_MSG_ACPDPM_SetEnabledMask       ((uint16_t) 0x12F)
+#define PPSMC_MSG_SAMUDPM_SetEnabledMask      ((uint16_t) 0x130)
+#define PPSMC_MSG_MCLKDPM_ForceState          ((uint16_t) 0x131)
+#define PPSMC_MSG_MCLKDPM_NoForcedLevel       ((uint16_t) 0x132)
+#define PPSMC_MSG_Thermal_Cntl_Disable        ((uint16_t) 0x133)
+#define PPSMC_MSG_Voltage_Cntl_Disable        ((uint16_t) 0x135)
+#define PPSMC_MSG_PCIeDPM_Enable              ((uint16_t) 0x136)
+#define PPSMC_MSG_PCIeDPM_Disable             ((uint16_t) 0x13d)
+#define PPSMC_MSG_ACPPowerOFF                 ((uint16_t) 0x137)
+#define PPSMC_MSG_ACPPowerON                  ((uint16_t) 0x138)
+#define PPSMC_MSG_SAMPowerOFF                 ((uint16_t) 0x139)
+#define PPSMC_MSG_SAMPowerON                  ((uint16_t) 0x13a)
+#define PPSMC_MSG_PCIeDPM_Disable             ((uint16_t) 0x13d)
+#define PPSMC_MSG_NBDPM_Enable                ((uint16_t) 0x140)
+#define PPSMC_MSG_NBDPM_Disable               ((uint16_t) 0x141)
+#define PPSMC_MSG_SCLKDPM_SetEnabledMask      ((uint16_t) 0x145)
+#define PPSMC_MSG_MCLKDPM_SetEnabledMask      ((uint16_t) 0x146)
+#define PPSMC_MSG_PCIeDPM_ForceLevel          ((uint16_t) 0x147)
+#define PPSMC_MSG_PCIeDPM_UnForceLevel        ((uint16_t) 0x148)
+#define PPSMC_MSG_EnableVRHotGPIOInterrupt    ((uint16_t) 0x14a)
+#define PPSMC_MSG_DPM_Enable                  ((uint16_t) 0x14e)
+#define PPSMC_MSG_DPM_Disable                 ((uint16_t) 0x14f)
+#define PPSMC_MSG_MCLKDPM_Enable              ((uint16_t) 0x150)
+#define PPSMC_MSG_MCLKDPM_Disable             ((uint16_t) 0x151)
+#define PPSMC_MSG_UVDDPM_Enable               ((uint16_t) 0x154)
+#define PPSMC_MSG_UVDDPM_Disable              ((uint16_t) 0x155)
+#define PPSMC_MSG_SAMUDPM_Enable              ((uint16_t) 0x156)
+#define PPSMC_MSG_SAMUDPM_Disable             ((uint16_t) 0x157)
+#define PPSMC_MSG_ACPDPM_Enable               ((uint16_t) 0x158)
+#define PPSMC_MSG_ACPDPM_Disable              ((uint16_t) 0x159)
+#define PPSMC_MSG_VCEDPM_Enable               ((uint16_t) 0x15a)
+#define PPSMC_MSG_VCEDPM_Disable              ((uint16_t) 0x15b)
+#define PPSMC_MSG_VddC_Request                ((uint16_t) 0x15f)
+#define PPSMC_MSG_SCLKDPM_GetEnabledMask      ((uint16_t) 0x162)
+#define PPSMC_MSG_PCIeDPM_SetEnabledMask      ((uint16_t) 0x167)
+#define PPSMC_MSG_TDCLimitEnable              ((uint16_t) 0x169)
+#define PPSMC_MSG_TDCLimitDisable             ((uint16_t) 0x16a)
+#define PPSMC_MSG_PkgPwrLimitEnable           ((uint16_t) 0x185)
+#define PPSMC_MSG_PkgPwrLimitDisable          ((uint16_t) 0x186)
+#define PPSMC_MSG_PkgPwrSetLimit              ((uint16_t) 0x187)
+#define PPSMC_MSG_OverDriveSetTargetTdp       ((uint16_t) 0x188)
+#define PPSMC_MSG_SCLKDPM_FreezeLevel         ((uint16_t) 0x189)
+#define PPSMC_MSG_SCLKDPM_UnfreezeLevel       ((uint16_t) 0x18A)
+#define PPSMC_MSG_MCLKDPM_FreezeLevel         ((uint16_t) 0x18B)
+#define PPSMC_MSG_MCLKDPM_UnfreezeLevel       ((uint16_t) 0x18C)
+#define PPSMC_MSG_MASTER_DeepSleep_ON         ((uint16_t) 0x18F)
+#define PPSMC_MSG_MASTER_DeepSleep_OFF        ((uint16_t) 0x190)
+#define PPSMC_MSG_Remove_DC_Clamp             ((uint16_t) 0x191)
+#define PPSMC_MSG_SetFanPwmMax                ((uint16_t) 0x19A)
+
+#define PPSMC_MSG_ENABLE_THERMAL_DPM          ((uint16_t) 0x19C)
+#define PPSMC_MSG_DISABLE_THERMAL_DPM         ((uint16_t) 0x19D)
+
+#define PPSMC_MSG_API_GetSclkFrequency        ((uint16_t) 0x200)
+#define PPSMC_MSG_API_GetMclkFrequency        ((uint16_t) 0x201)
+
+/* TN */
+#define PPSMC_MSG_DPM_Config                ((uint32_t) 0x102)
+#define PPSMC_MSG_DPM_ForceState            ((uint32_t) 0x104)
+#define PPSMC_MSG_PG_SIMD_Config            ((uint32_t) 0x108)
+#define PPSMC_MSG_Voltage_Cntl_Enable       ((uint32_t) 0x109)
+#define PPSMC_MSG_Thermal_Cntl_Enable       ((uint32_t) 0x10a)
+#define PPSMC_MSG_VCEPowerOFF               ((uint32_t) 0x10e)
+#define PPSMC_MSG_VCEPowerON                ((uint32_t) 0x10f)
+#define PPSMC_MSG_DPM_N_LevelsDisabled      ((uint32_t) 0x112)
+#define PPSMC_MSG_DCE_RemoveVoltageAdjustment   ((uint32_t) 0x11d)
+#define PPSMC_MSG_DCE_AllowVoltageAdjustment    ((uint32_t) 0x11e)
+#define PPSMC_MSG_EnableBAPM                ((uint32_t) 0x120)
+#define PPSMC_MSG_DisableBAPM               ((uint32_t) 0x121)
+#define PPSMC_MSG_UVD_DPM_Config            ((uint32_t) 0x124)
+
+#define PPSMC_MSG_DRV_DRAM_ADDR_HI            ((uint16_t) 0x250)
+#define PPSMC_MSG_DRV_DRAM_ADDR_LO            ((uint16_t) 0x251)
+#define PPSMC_MSG_SMU_DRAM_ADDR_HI            ((uint16_t) 0x252)
+#define PPSMC_MSG_SMU_DRAM_ADDR_LO            ((uint16_t) 0x253)
+#define PPSMC_MSG_LoadUcodes                  ((uint16_t) 0x254)
+
+typedef uint16_t PPSMC_Msg;
+
+#pragma pack(pop)
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
new file mode 100644
index 000000000..2cf50180c
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -0,0 +1,1418 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "vi.h"
+#include "vid.h"
+
+#include "oss/oss_2_4_d.h"
+#include "oss/oss_2_4_sh_mask.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "iceland_sdma_pkt_open.h"
+
+static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/topaz_sdma.bin");
+MODULE_FIRMWARE("amdgpu/topaz_sdma1.bin");
+
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	SDMA0_REGISTER_OFFSET,
+	SDMA1_REGISTER_OFFSET
+};
+
+static const u32 golden_settings_iceland_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+/*
+ * sDMA - System DMA
+ * Starting with CIK, the GPU has new asynchronous
+ * DMA engines.  These engines are used for compute
+ * and gfx.  There are two DMA engines (SDMA0, SDMA1)
+ * and each one supports 1 ring buffer used for gfx
+ * and 2 queues used for compute.
+ *
+ * The programming model is very similar to the CP
+ * (ring buffer, IBs, etc.), but sDMA has it's own
+ * packet format that is different from the PM4 format
+ * used by the CP. sDMA supports copying data, writing
+ * embedded data, solid fills, and a number of other
+ * things.  It also has support for tiling/detiling of
+ * buffers.
+ */
+
+static void sdma_v2_4_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_program_register_sequence(adev,
+						 iceland_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_iceland_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_iceland_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * sdma_v2_4_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v2_4_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct sdma_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		chip_name = "topaz";
+		break;
+	default: BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+		if (err)
+			goto out;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+
+		if (adev->firmware.smu_load) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+			info->fw = adev->sdma.instance[i].fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "sdma_v2_4: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * sdma_v2_4_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (VI+).
+ */
+static uint32_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = ring->adev->wb.wb[ring->rptr_offs] >> 2;
+
+	return rptr;
+}
+
+/**
+ * sdma_v2_4_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (VI+).
+ */
+static uint32_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
+	u32 wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >> 2;
+
+	return wptr;
+}
+
+/**
+ * sdma_v2_4_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (VI+).
+ */
+static void sdma_v2_4_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
+
+	WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], ring->wptr << 2);
+}
+
+static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->nop);
+}
+
+/**
+ * sdma_v2_4_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (VI).
+ */
+static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_ib *ib)
+{
+	u32 vmid = (ib->vm ? ib->vm->ids[ring->idx].id : 0) & 0xf;
+	u32 next_rptr = ring->wptr + 5;
+
+	while ((next_rptr & 7) != 2)
+		next_rptr++;
+
+	next_rptr += 6;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(ring->next_rptr_gpu_addr) & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
+	amdgpu_ring_write(ring, next_rptr);
+
+	/* IB packet must end on a 8 DW boundary */
+	sdma_v2_4_ring_insert_nop(ring, (10 - (ring->wptr & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+}
+
+/**
+ * sdma_v2_4_hdp_flush_ring_emit - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask = 0;
+
+	if (ring == &ring->adev->sdma.instance[0].ring)
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA0, 1);
+	else
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA1, 1);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE << 2);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v2_4_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (VI).
+ */
+static void sdma_v2_4_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+}
+
+/**
+ * sdma_v2_4_ring_emit_semaphore - emit a semaphore on the dma ring
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @semaphore: amdgpu semaphore object
+ * @emit_wait: wait or signal semaphore
+ *
+ * Add a DMA semaphore packet to the ring wait on or signal
+ * other rings (VI).
+ */
+static bool sdma_v2_4_ring_emit_semaphore(struct amdgpu_ring *ring,
+					  struct amdgpu_semaphore *semaphore,
+					  bool emit_wait)
+{
+	u64 addr = semaphore->gpu_addr;
+	u32 sig = emit_wait ? 0 : 1;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SEM) |
+			  SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(sig));
+	amdgpu_ring_write(ring, lower_32_bits(addr) & 0xfffffff8);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	return true;
+}
+
+/**
+ * sdma_v2_4_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (VI).
+ */
+static void sdma_v2_4_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+	}
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * sdma_v2_4_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (VI).
+ */
+static void sdma_v2_4_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v2_4_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (VI).
+ */
+static void sdma_v2_4_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (enable == false) {
+		sdma_v2_4_gfx_stop(adev);
+		sdma_v2_4_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]);
+		if (enable)
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 0);
+		else
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 1);
+		WREG32(mmSDMA0_F32_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * sdma_v2_4_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	int i, j, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		wb_offset = (ring->rptr_offs * 4);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			vi_srbm_select(adev, 0, 0, 0, j);
+			/* SDMA GFX */
+			WREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i], 0);
+			WREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i], 0);
+		}
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		WREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+					RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
+		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+		WREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
+
+		/* enable DMA RB */
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+		/* enable DMA IBs */
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		ring->ready = true;
+
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v2_4_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_rlc_resume(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+	return 0;
+}
+
+/**
+ * sdma_v2_4_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v2_4_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v2_4_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+			 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
+		for (j = 0; j < fw_size; j++)
+			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v2_4_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_start(struct amdgpu_device *adev)
+{
+	int r;
+
+	if (!adev->firmware.smu_load) {
+		r = sdma_v2_4_load_microcode(adev);
+		if (r)
+			return r;
+	} else {
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_SDMA0);
+		if (r)
+			return -EINVAL;
+		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+						AMDGPU_UCODE_ID_SDMA1);
+		if (r)
+			return -EINVAL;
+	}
+
+	/* unhalt the MEs */
+	sdma_v2_4_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v2_4_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v2_4_rlc_resume(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * sdma_v2_4_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_lock(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v2_4_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (VI).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct fence *f = NULL;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	r = amdgpu_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(ring, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err1;
+
+	r = fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto err1;
+	}
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
+			 ring->idx, i);
+		goto err1;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+
+err1:
+	fence_put(f);
+	amdgpu_ib_free(adev, &ib);
+err0:
+	amdgpu_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * sdma_v2_4_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (CIK).
+ */
+static void sdma_v2_4_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	while (count) {
+		unsigned bytes = count * 8;
+		if (bytes > 0x1FFFF8)
+			bytes = 0x1FFFF8;
+
+		ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+			SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+		ib->ptr[ib->length_dw++] = bytes;
+		ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+		ib->ptr[ib->length_dw++] = lower_32_bits(src);
+		ib->ptr[ib->length_dw++] = upper_32_bits(src);
+		ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+		pe += bytes;
+		src += bytes;
+		count -= bytes / 8;
+	}
+}
+
+/**
+ * sdma_v2_4_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update PTEs by writing them manually using sDMA (CIK).
+ */
+static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib,
+				   uint64_t pe,
+				   uint64_t addr, unsigned count,
+				   uint32_t incr, uint32_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count * 2;
+		if (ndw > 0xFFFFE)
+			ndw = 0xFFFFE;
+
+		/* for non-physically contiguous pages (system) */
+		ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+		ib->ptr[ib->length_dw++] = pe;
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+		ib->ptr[ib->length_dw++] = ndw;
+		for (; ndw > 0; ndw -= 2, --count, pe += 8) {
+			if (flags & AMDGPU_PTE_SYSTEM) {
+				value = amdgpu_vm_map_gart(ib->ring->adev, addr);
+				value &= 0xFFFFFFFFFFFFF000ULL;
+			} else if (flags & AMDGPU_PTE_VALID) {
+				value = addr;
+			} else {
+				value = 0;
+			}
+			addr += incr;
+			value |= flags;
+			ib->ptr[ib->length_dw++] = value;
+			ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		}
+	}
+}
+
+/**
+ * sdma_v2_4_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint32_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count;
+		if (ndw > 0x7FFFF)
+			ndw = 0x7FFFF;
+
+		if (flags & AMDGPU_PTE_VALID)
+			value = addr;
+		else
+			value = 0;
+
+		/* for physically contiguous pages (vram) */
+		ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE);
+		ib->ptr[ib->length_dw++] = pe; /* dst addr */
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+		ib->ptr[ib->length_dw++] = flags; /* mask */
+		ib->ptr[ib->length_dw++] = 0;
+		ib->ptr[ib->length_dw++] = value; /* value */
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		ib->ptr[ib->length_dw++] = incr; /* increment size */
+		ib->ptr[ib->length_dw++] = 0;
+		ib->ptr[ib->length_dw++] = ndw; /* number of entries */
+
+		pe += ndw * 8;
+		addr += ndw * incr;
+		count -= ndw;
+	}
+}
+
+/**
+ * sdma_v2_4_vm_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void sdma_v2_4_vm_pad_ib(struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ib->ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+/**
+ * sdma_v2_4_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (VI).
+ */
+static void sdma_v2_4_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vm_id, uint64_t pd_addr)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	if (vm_id < 8) {
+		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
+	} else {
+		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
+	}
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	/* flush TLB */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 1 << vm_id);
+
+	/* wait for flush */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(0)); /* always */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* reference */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+static int sdma_v2_4_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->sdma.num_instances = SDMA_MAX_INSTANCE;
+
+	sdma_v2_4_set_ring_funcs(adev);
+	sdma_v2_4_set_buffer_funcs(adev);
+	sdma_v2_4_set_vm_pte_funcs(adev);
+	sdma_v2_4_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int sdma_v2_4_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, 224, &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, 241, &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, 247, &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	r = sdma_v2_4_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = false;
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 256 * 1024,
+				     SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), 0xf,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 : AMDGPU_SDMA_IRQ_TRAP1,
+				     AMDGPU_RING_TYPE_SDMA);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int sdma_v2_4_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	return 0;
+}
+
+static int sdma_v2_4_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v2_4_init_golden_registers(adev);
+
+	r = sdma_v2_4_start(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v2_4_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v2_4_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v2_4_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v2_4_hw_fini(adev);
+}
+
+static int sdma_v2_4_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v2_4_hw_init(adev);
+}
+
+static bool sdma_v2_4_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
+		   SRBM_STATUS2__SDMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int sdma_v2_4_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK);
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void sdma_v2_4_print_status(void *handle)
+{
+	int i, j;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "VI SDMA registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		 RREG32(mmSRBM_STATUS2));
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		dev_info(adev->dev, "  SDMA%d_STATUS_REG=0x%08X\n",
+			 i, RREG32(mmSDMA0_STATUS_REG + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_F32_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_SEM_WAIT_FAIL_TIMER_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_IB_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_WPTR=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR_ADDR_HI=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR_ADDR_LO=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_BASE=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_BASE_HI=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i]));
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			vi_srbm_select(adev, 0, 0, 0, j);
+			dev_info(adev->dev, "  VM %d:\n", j);
+			dev_info(adev->dev, "  SDMA%d_GFX_VIRTUAL_ADDR=0x%08X\n",
+				 i, RREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i]));
+			dev_info(adev->dev, "  SDMA%d_GFX_APE1_CNTL=0x%08X\n",
+				 i, RREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i]));
+		}
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+}
+
+static int sdma_v2_4_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+	}
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+	}
+
+	if (srbm_soft_reset) {
+		sdma_v2_4_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		sdma_v2_4_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int sdma_v2_4_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *src,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v2_4_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u8 instance_id, queue_id;
+
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (instance_id) {
+	case 0:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	case 1:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v2_4_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int sdma_v2_4_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	/* XXX handled via the smc on VI */
+	return 0;
+}
+
+static int sdma_v2_4_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs sdma_v2_4_ip_funcs = {
+	.early_init = sdma_v2_4_early_init,
+	.late_init = NULL,
+	.sw_init = sdma_v2_4_sw_init,
+	.sw_fini = sdma_v2_4_sw_fini,
+	.hw_init = sdma_v2_4_hw_init,
+	.hw_fini = sdma_v2_4_hw_fini,
+	.suspend = sdma_v2_4_suspend,
+	.resume = sdma_v2_4_resume,
+	.is_idle = sdma_v2_4_is_idle,
+	.wait_for_idle = sdma_v2_4_wait_for_idle,
+	.soft_reset = sdma_v2_4_soft_reset,
+	.print_status = sdma_v2_4_print_status,
+	.set_clockgating_state = sdma_v2_4_set_clockgating_state,
+	.set_powergating_state = sdma_v2_4_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
+	.get_rptr = sdma_v2_4_ring_get_rptr,
+	.get_wptr = sdma_v2_4_ring_get_wptr,
+	.set_wptr = sdma_v2_4_ring_set_wptr,
+	.parse_cs = NULL,
+	.emit_ib = sdma_v2_4_ring_emit_ib,
+	.emit_fence = sdma_v2_4_ring_emit_fence,
+	.emit_semaphore = sdma_v2_4_ring_emit_semaphore,
+	.emit_vm_flush = sdma_v2_4_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v2_4_ring_emit_hdp_flush,
+	.test_ring = sdma_v2_4_ring_test_ring,
+	.test_ib = sdma_v2_4_ring_test_ib,
+	.insert_nop = sdma_v2_4_ring_insert_nop,
+};
+
+static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {
+	.set = sdma_v2_4_set_trap_irq_state,
+	.process = sdma_v2_4_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v2_4_illegal_inst_irq_funcs = {
+	.process = sdma_v2_4_process_illegal_inst_irq,
+};
+
+static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &sdma_v2_4_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v2_4_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v2_4_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (VI).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v2_4_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = byte_count;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v2_4_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (VI).
+ */
+static void sdma_v2_4_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v2_4_buffer_funcs = {
+	.copy_max_bytes = 0x1fffff,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v2_4_emit_copy_buffer,
+
+	.fill_max_bytes = 0x1fffff,
+	.fill_num_dw = 7,
+	.emit_fill_buffer = sdma_v2_4_emit_fill_buffer,
+};
+
+static void sdma_v2_4_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v2_4_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v2_4_vm_pte_funcs = {
+	.copy_pte = sdma_v2_4_vm_copy_pte,
+	.write_pte = sdma_v2_4_vm_write_pte,
+	.set_pte_pde = sdma_v2_4_vm_set_pte_pde,
+	.pad_ib = sdma_v2_4_vm_pad_ib,
+};
+
+static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v2_4_vm_pte_funcs;
+		adev->vm_manager.vm_pte_funcs_ring = &adev->sdma.instance[0].ring;
+		adev->vm_manager.vm_pte_funcs_ring->is_pte_ring = true;
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.h b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.h
new file mode 100644
index 000000000..07349f5ee
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V2_4_H__
+#define __SDMA_V2_4_H__
+
+extern const struct amd_ip_funcs sdma_v2_4_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
new file mode 100644
index 000000000..7253132f0
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -0,0 +1,1579 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Alex Deucher
+ */
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_ucode.h"
+#include "amdgpu_trace.h"
+#include "vi.h"
+#include "vid.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_enum.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "tonga_sdma_pkt_open.h"
+
+static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev);
+static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev);
+static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev);
+static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+
+MODULE_FIRMWARE("amdgpu/tonga_sdma.bin");
+MODULE_FIRMWARE("amdgpu/tonga_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_sdma.bin");
+MODULE_FIRMWARE("amdgpu/carrizo_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/fiji_sdma.bin");
+MODULE_FIRMWARE("amdgpu/fiji_sdma1.bin");
+MODULE_FIRMWARE("amdgpu/stoney_sdma.bin");
+
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
+{
+	SDMA0_REGISTER_OFFSET,
+	SDMA1_REGISTER_OFFSET
+};
+
+static const u32 golden_settings_tonga_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+};
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+static const u32 golden_settings_fiji_a10[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_GFX_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC0_IB_CNTL, 0x800f0111, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x800f0111, 0x00000100,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+static const u32 cz_golden_settings_a11[] =
+{
+	mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA0_GFX_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_POWER_CNTL, 0x00000800, 0x0003c800,
+	mmSDMA0_RLC0_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
+	mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
+	mmSDMA1_GFX_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA1_POWER_CNTL, 0x00000800, 0x0003c800,
+	mmSDMA1_RLC0_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA1_RLC1_IB_CNTL, 0x00000100, 0x00000100,
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
+	mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
+};
+
+static const u32 stoney_golden_settings_a11[] =
+{
+	mmSDMA0_GFX_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_POWER_CNTL, 0x00000800, 0x0003c800,
+	mmSDMA0_RLC0_IB_CNTL, 0x00000100, 0x00000100,
+	mmSDMA0_RLC1_IB_CNTL, 0x00000100, 0x00000100,
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmSDMA0_CLK_CTRL, 0xffffffff, 0x00000100,
+};
+
+/*
+ * sDMA - System DMA
+ * Starting with CIK, the GPU has new asynchronous
+ * DMA engines.  These engines are used for compute
+ * and gfx.  There are two DMA engines (SDMA0, SDMA1)
+ * and each one supports 1 ring buffer used for gfx
+ * and 2 queues used for compute.
+ *
+ * The programming model is very similar to the CP
+ * (ring buffer, IBs, etc.), but sDMA has it's own
+ * packet format that is different from the PM4 format
+ * used by the CP. sDMA supports copying data, writing
+ * embedded data, solid fills, and a number of other
+ * things.  It also has support for tiling/detiling of
+ * buffers.
+ */
+
+static void sdma_v3_0_init_golden_registers(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_FIJI:
+		amdgpu_program_register_sequence(adev,
+						 fiji_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_fiji_a10,
+						 (const u32)ARRAY_SIZE(golden_settings_fiji_a10));
+		break;
+	case CHIP_TONGA:
+		amdgpu_program_register_sequence(adev,
+						 tonga_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_tonga_a11,
+						 (const u32)ARRAY_SIZE(golden_settings_tonga_a11));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_program_register_sequence(adev,
+						 cz_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 cz_golden_settings_a11,
+						 (const u32)ARRAY_SIZE(cz_golden_settings_a11));
+		break;
+	case CHIP_STONEY:
+		amdgpu_program_register_sequence(adev,
+						 stoney_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 stoney_golden_settings_a11,
+						 (const u32)ARRAY_SIZE(stoney_golden_settings_a11));
+		break;
+	default:
+		break;
+	}
+}
+
+/**
+ * sdma_v3_0_init_microcode - load ucode images from disk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Use the firmware interface to load the ucode images into
+ * the driver (not loaded into hw).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v3_0_init_microcode(struct amdgpu_device *adev)
+{
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0, i;
+	struct amdgpu_firmware_info *info = NULL;
+	const struct common_firmware_header *header = NULL;
+	const struct sdma_firmware_header_v1_0 *hdr;
+
+	DRM_DEBUG("\n");
+
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		chip_name = "tonga";
+		break;
+	case CHIP_FIJI:
+		chip_name = "fiji";
+		break;
+	case CHIP_CARRIZO:
+		chip_name = "carrizo";
+		break;
+	case CHIP_STONEY:
+		chip_name = "stoney";
+		break;
+	default: BUG();
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (i == 0)
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma.bin", chip_name);
+		else
+			snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_sdma1.bin", chip_name);
+		err = request_firmware(&adev->sdma.instance[i].fw, fw_name, adev->dev);
+		if (err)
+			goto out;
+		err = amdgpu_ucode_validate(adev->sdma.instance[i].fw);
+		if (err)
+			goto out;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		adev->sdma.instance[i].fw_version = le32_to_cpu(hdr->header.ucode_version);
+		adev->sdma.instance[i].feature_version = le32_to_cpu(hdr->ucode_feature_version);
+		if (adev->sdma.instance[i].feature_version >= 20)
+			adev->sdma.instance[i].burst_nop = true;
+
+		if (adev->firmware.smu_load) {
+			info = &adev->firmware.ucode[AMDGPU_UCODE_ID_SDMA0 + i];
+			info->ucode_id = AMDGPU_UCODE_ID_SDMA0 + i;
+			info->fw = adev->sdma.instance[i].fw;
+			header = (const struct common_firmware_header *)info->fw->data;
+			adev->firmware.fw_size +=
+				ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+		}
+	}
+out:
+	if (err) {
+		printk(KERN_ERR
+		       "sdma_v3_0: Failed to load firmware \"%s\"\n",
+		       fw_name);
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+		}
+	}
+	return err;
+}
+
+/**
+ * sdma_v3_0_ring_get_rptr - get the current read pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current rptr from the hardware (VI+).
+ */
+static uint32_t sdma_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	u32 rptr;
+
+	/* XXX check if swapping is necessary on BE */
+	rptr = ring->adev->wb.wb[ring->rptr_offs] >> 2;
+
+	return rptr;
+}
+
+/**
+ * sdma_v3_0_ring_get_wptr - get the current write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Get the current wptr from the hardware (VI+).
+ */
+static uint32_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	u32 wptr;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		wptr = ring->adev->wb.wb[ring->wptr_offs] >> 2;
+	} else {
+		int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
+
+		wptr = RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >> 2;
+	}
+
+	return wptr;
+}
+
+/**
+ * sdma_v3_0_ring_set_wptr - commit the write pointer
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Write the wptr back to the hardware (VI+).
+ */
+static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring->use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		adev->wb.wb[ring->wptr_offs] = ring->wptr << 2;
+		WDOORBELL32(ring->doorbell_index, ring->wptr << 2);
+	} else {
+		int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1;
+
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], ring->wptr << 2);
+	}
+}
+
+static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ring);
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			amdgpu_ring_write(ring, ring->nop |
+				SDMA_PKT_NOP_HEADER_COUNT(count - 1));
+		else
+			amdgpu_ring_write(ring, ring->nop);
+}
+
+/**
+ * sdma_v3_0_ring_emit_ib - Schedule an IB on the DMA engine
+ *
+ * @ring: amdgpu ring pointer
+ * @ib: IB object to schedule
+ *
+ * Schedule an IB in the DMA ring (VI).
+ */
+static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
+				   struct amdgpu_ib *ib)
+{
+	u32 vmid = (ib->vm ? ib->vm->ids[ring->idx].id : 0) & 0xf;
+	u32 next_rptr = ring->wptr + 5;
+
+	while ((next_rptr & 7) != 2)
+		next_rptr++;
+	next_rptr += 6;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(ring->next_rptr_gpu_addr) & 0xfffffffc);
+	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
+	amdgpu_ring_write(ring, next_rptr);
+
+	/* IB packet must end on a 8 DW boundary */
+	sdma_v3_0_ring_insert_nop(ring, (10 - (ring->wptr & 7)) % 8);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_INDIRECT) |
+			  SDMA_PKT_INDIRECT_HEADER_VMID(vmid));
+	/* base must be 32 byte aligned */
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr) & 0xffffffe0);
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, ib->length_dw);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0);
+
+}
+
+/**
+ * sdma_v3_0_ring_emit_hdp_flush - emit an hdp flush on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ *
+ * Emit an hdp flush packet on the requested DMA ring.
+ */
+static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	u32 ref_and_mask = 0;
+
+	if (ring == &ring->adev->sdma.instance[0].ring)
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA0, 1);
+	else
+		ref_and_mask = REG_SET_FIELD(ref_and_mask, GPU_HDP_FLUSH_DONE, SDMA1, 1);
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(1) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(3)); /* == */
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_DONE << 2);
+	amdgpu_ring_write(ring, mmGPU_HDP_FLUSH_REQ << 2);
+	amdgpu_ring_write(ring, ref_and_mask); /* reference */
+	amdgpu_ring_write(ring, ref_and_mask); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+/**
+ * sdma_v3_0_ring_emit_fence - emit a fence on the DMA ring
+ *
+ * @ring: amdgpu ring pointer
+ * @fence: amdgpu fence object
+ *
+ * Add a DMA fence packet to the ring to write
+ * the fence seq number and DMA trap packet to generate
+ * an interrupt if needed (VI).
+ */
+static void sdma_v3_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				      unsigned flags)
+{
+	bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT;
+	/* write the fence */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, lower_32_bits(seq));
+
+	/* optionally write high bits as well */
+	if (write64bit) {
+		addr += 4;
+		amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_FENCE));
+		amdgpu_ring_write(ring, lower_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(addr));
+		amdgpu_ring_write(ring, upper_32_bits(seq));
+	}
+
+	/* generate an interrupt */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_TRAP));
+	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
+}
+
+
+/**
+ * sdma_v3_0_ring_emit_semaphore - emit a semaphore on the dma ring
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @semaphore: amdgpu semaphore object
+ * @emit_wait: wait or signal semaphore
+ *
+ * Add a DMA semaphore packet to the ring wait on or signal
+ * other rings (VI).
+ */
+static bool sdma_v3_0_ring_emit_semaphore(struct amdgpu_ring *ring,
+					  struct amdgpu_semaphore *semaphore,
+					  bool emit_wait)
+{
+	u64 addr = semaphore->gpu_addr;
+	u32 sig = emit_wait ? 0 : 1;
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SEM) |
+			  SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(sig));
+	amdgpu_ring_write(ring, lower_32_bits(addr) & 0xfffffff8);
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+
+	return true;
+}
+
+/**
+ * sdma_v3_0_gfx_stop - stop the gfx async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the gfx async dma ring buffers (VI).
+ */
+static void sdma_v3_0_gfx_stop(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *sdma0 = &adev->sdma.instance[0].ring;
+	struct amdgpu_ring *sdma1 = &adev->sdma.instance[1].ring;
+	u32 rb_cntl, ib_cntl;
+	int i;
+
+	if ((adev->mman.buffer_funcs_ring == sdma0) ||
+	    (adev->mman.buffer_funcs_ring == sdma1))
+		amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+	}
+	sdma0->ready = false;
+	sdma1->ready = false;
+}
+
+/**
+ * sdma_v3_0_rlc_stop - stop the compute async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the compute async dma queues (VI).
+ */
+static void sdma_v3_0_rlc_stop(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+}
+
+/**
+ * sdma_v3_0_ctx_switch_enable - stop the async dma engines context switch
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs context switch.
+ *
+ * Halt or unhalt the async dma engines context switch (VI).
+ */
+static void sdma_v3_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_CNTL + sdma_offsets[i]);
+		if (enable)
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, 1);
+		else
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
+					AUTO_CTXSW_ENABLE, 0);
+		WREG32(mmSDMA0_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * sdma_v3_0_enable - stop the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable/disable the DMA MEs.
+ *
+ * Halt or unhalt the async dma engines (VI).
+ */
+static void sdma_v3_0_enable(struct amdgpu_device *adev, bool enable)
+{
+	u32 f32_cntl;
+	int i;
+
+	if (enable == false) {
+		sdma_v3_0_gfx_stop(adev);
+		sdma_v3_0_rlc_stop(adev);
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		f32_cntl = RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]);
+		if (enable)
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 0);
+		else
+			f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, 1);
+		WREG32(mmSDMA0_F32_CNTL + sdma_offsets[i], f32_cntl);
+	}
+}
+
+/**
+ * sdma_v3_0_gfx_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the gfx DMA ring buffers and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	u32 rb_cntl, ib_cntl;
+	u32 rb_bufsz;
+	u32 wb_offset;
+	u32 doorbell;
+	int i, j, r;
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		wb_offset = (ring->rptr_offs * 4);
+
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			vi_srbm_select(adev, 0, 0, 0, j);
+			/* SDMA GFX */
+			WREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i], 0);
+			WREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i], 0);
+		}
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+
+		WREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i], 0);
+
+		/* Set ring buffer size in dwords */
+		rb_bufsz = order_base_2(ring->ring_size / 4);
+		rb_cntl = RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
+#ifdef __BIG_ENDIAN
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
+					RPTR_WRITEBACK_SWAP_ENABLE, 1);
+#endif
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		/* Initialize the ring buffer's read and write pointers */
+		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+
+		/* set the wb address whether it's enabled or not */
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
+		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
+		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i],
+		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
+
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
+
+		WREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i], ring->gpu_addr >> 8);
+		WREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i], ring->gpu_addr >> 40);
+
+		ring->wptr = 0;
+		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], ring->wptr << 2);
+
+		doorbell = RREG32(mmSDMA0_GFX_DOORBELL + sdma_offsets[i]);
+
+		if (ring->use_doorbell) {
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL,
+						 OFFSET, ring->doorbell_index);
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
+		} else {
+			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
+		}
+		WREG32(mmSDMA0_GFX_DOORBELL + sdma_offsets[i], doorbell);
+
+		/* enable DMA RB */
+		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
+		WREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i], rb_cntl);
+
+		ib_cntl = RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]);
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
+#ifdef __BIG_ENDIAN
+		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
+#endif
+		/* enable DMA IBs */
+		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
+
+		ring->ready = true;
+
+		r = amdgpu_ring_test_ring(ring);
+		if (r) {
+			ring->ready = false;
+			return r;
+		}
+
+		if (adev->mman.buffer_funcs_ring == ring)
+			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v3_0_rlc_resume - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the compute DMA queues and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_rlc_resume(struct amdgpu_device *adev)
+{
+	/* XXX todo */
+	return 0;
+}
+
+/**
+ * sdma_v3_0_load_microcode - load the sDMA ME ucode
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Loads the sDMA0/1 ucode.
+ * Returns 0 for success, -EINVAL if the ucode is not available.
+ */
+static int sdma_v3_0_load_microcode(struct amdgpu_device *adev)
+{
+	const struct sdma_firmware_header_v1_0 *hdr;
+	const __le32 *fw_data;
+	u32 fw_size;
+	int i, j;
+
+	/* halt the MEs */
+	sdma_v3_0_enable(adev, false);
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		if (!adev->sdma.instance[i].fw)
+			return -EINVAL;
+		hdr = (const struct sdma_firmware_header_v1_0 *)adev->sdma.instance[i].fw->data;
+		amdgpu_ucode_print_sdma_hdr(&hdr->header);
+		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
+		fw_data = (const __le32 *)
+			(adev->sdma.instance[i].fw->data +
+				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], 0);
+		for (j = 0; j < fw_size; j++)
+			WREG32(mmSDMA0_UCODE_DATA + sdma_offsets[i], le32_to_cpup(fw_data++));
+		WREG32(mmSDMA0_UCODE_ADDR + sdma_offsets[i], adev->sdma.instance[i].fw_version);
+	}
+
+	return 0;
+}
+
+/**
+ * sdma_v3_0_start - setup and start the async dma engines
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set up the DMA engines and enable them (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_start(struct amdgpu_device *adev)
+{
+	int r, i;
+
+	if (!adev->firmware.smu_load) {
+		r = sdma_v3_0_load_microcode(adev);
+		if (r)
+			return r;
+	} else {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+									 (i == 0) ?
+									 AMDGPU_UCODE_ID_SDMA0 :
+									 AMDGPU_UCODE_ID_SDMA1);
+			if (r)
+				return -EINVAL;
+		}
+	}
+
+	/* unhalt the MEs */
+	sdma_v3_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v3_0_ctx_switch_enable(adev, true);
+
+	/* start the gfx rings and rlc compute queues */
+	r = sdma_v3_0_gfx_resume(adev);
+	if (r)
+		return r;
+	r = sdma_v3_0_rlc_resume(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+/**
+ * sdma_v3_0_ring_test_ring - simple async dma engine test
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test the DMA engine by writing using it to write an
+ * value to memory. (VI).
+ * Returns 0 for success, error for failure.
+ */
+static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp;
+	u64 gpu_addr;
+
+	r = amdgpu_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+
+	r = amdgpu_ring_lock(ring, 5);
+	if (r) {
+		DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r);
+		amdgpu_wb_free(adev, index);
+		return r;
+	}
+
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
+	amdgpu_ring_write(ring, lower_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, upper_32_bits(gpu_addr));
+	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n", ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	amdgpu_wb_free(adev, index);
+
+	return r;
+}
+
+/**
+ * sdma_v3_0_ring_test_ib - test an IB on the DMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ *
+ * Test a simple IB in the DMA ring (VI).
+ * Returns 0 on success, error on failure.
+ */
+static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct amdgpu_ib ib;
+	struct fence *f = NULL;
+	unsigned i;
+	unsigned index;
+	int r;
+	u32 tmp = 0;
+	u64 gpu_addr;
+
+	r = amdgpu_wb_get(adev, &index);
+	if (r) {
+		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		return r;
+	}
+
+	gpu_addr = adev->wb.gpu_addr + (index * 4);
+	tmp = 0xCAFEDEAD;
+	adev->wb.wb[index] = cpu_to_le32(tmp);
+	memset(&ib, 0, sizeof(ib));
+	r = amdgpu_ib_get(ring, NULL, 256, &ib);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		goto err0;
+	}
+
+	ib.ptr[0] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR);
+	ib.ptr[1] = lower_32_bits(gpu_addr);
+	ib.ptr[2] = upper_32_bits(gpu_addr);
+	ib.ptr[3] = SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1);
+	ib.ptr[4] = 0xDEADBEEF;
+	ib.ptr[5] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[6] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.ptr[7] = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP);
+	ib.length_dw = 8;
+
+	r = amdgpu_sched_ib_submit_kernel_helper(adev, ring, &ib, 1, NULL,
+						 AMDGPU_FENCE_OWNER_UNDEFINED,
+						 &f);
+	if (r)
+		goto err1;
+
+	r = fence_wait(f, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto err1;
+	}
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = le32_to_cpu(adev->wb.wb[index]);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
+			 ring->idx, i);
+		goto err1;
+	} else {
+		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
+		r = -EINVAL;
+	}
+err1:
+	fence_put(f);
+	amdgpu_ib_free(adev, &ib);
+err0:
+	amdgpu_wb_free(adev, index);
+	return r;
+}
+
+/**
+ * sdma_v3_0_vm_copy_pte - update PTEs by copying them from the GART
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @src: src addr to copy from
+ * @count: number of page entries to update
+ *
+ * Update PTEs by copying them from the GART using sDMA (CIK).
+ */
+static void sdma_v3_0_vm_copy_pte(struct amdgpu_ib *ib,
+				  uint64_t pe, uint64_t src,
+				  unsigned count)
+{
+	while (count) {
+		unsigned bytes = count * 8;
+		if (bytes > 0x1FFFF8)
+			bytes = 0x1FFFF8;
+
+		ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+			SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+		ib->ptr[ib->length_dw++] = bytes;
+		ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+		ib->ptr[ib->length_dw++] = lower_32_bits(src);
+		ib->ptr[ib->length_dw++] = upper_32_bits(src);
+		ib->ptr[ib->length_dw++] = lower_32_bits(pe);
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+
+		pe += bytes;
+		src += bytes;
+		count -= bytes / 8;
+	}
+}
+
+/**
+ * sdma_v3_0_vm_write_pte - update PTEs by writing them manually
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update PTEs by writing them manually using sDMA (CIK).
+ */
+static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib,
+				   uint64_t pe,
+				   uint64_t addr, unsigned count,
+				   uint32_t incr, uint32_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count * 2;
+		if (ndw > 0xFFFFE)
+			ndw = 0xFFFFE;
+
+		/* for non-physically contiguous pages (system) */
+		ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
+			SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+		ib->ptr[ib->length_dw++] = pe;
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+		ib->ptr[ib->length_dw++] = ndw;
+		for (; ndw > 0; ndw -= 2, --count, pe += 8) {
+			if (flags & AMDGPU_PTE_SYSTEM) {
+				value = amdgpu_vm_map_gart(ib->ring->adev, addr);
+				value &= 0xFFFFFFFFFFFFF000ULL;
+			} else if (flags & AMDGPU_PTE_VALID) {
+				value = addr;
+			} else {
+				value = 0;
+			}
+			addr += incr;
+			value |= flags;
+			ib->ptr[ib->length_dw++] = value;
+			ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		}
+	}
+}
+
+/**
+ * sdma_v3_0_vm_set_pte_pde - update the page tables using sDMA
+ *
+ * @ib: indirect buffer to fill with commands
+ * @pe: addr of the page entry
+ * @addr: dst addr to write into pe
+ * @count: number of page entries to update
+ * @incr: increase next addr by incr bytes
+ * @flags: access flags
+ *
+ * Update the page tables using sDMA (CIK).
+ */
+static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib,
+				     uint64_t pe,
+				     uint64_t addr, unsigned count,
+				     uint32_t incr, uint32_t flags)
+{
+	uint64_t value;
+	unsigned ndw;
+
+	while (count) {
+		ndw = count;
+		if (ndw > 0x7FFFF)
+			ndw = 0x7FFFF;
+
+		if (flags & AMDGPU_PTE_VALID)
+			value = addr;
+		else
+			value = 0;
+
+		/* for physically contiguous pages (vram) */
+		ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE);
+		ib->ptr[ib->length_dw++] = pe; /* dst addr */
+		ib->ptr[ib->length_dw++] = upper_32_bits(pe);
+		ib->ptr[ib->length_dw++] = flags; /* mask */
+		ib->ptr[ib->length_dw++] = 0;
+		ib->ptr[ib->length_dw++] = value; /* value */
+		ib->ptr[ib->length_dw++] = upper_32_bits(value);
+		ib->ptr[ib->length_dw++] = incr; /* increment size */
+		ib->ptr[ib->length_dw++] = 0;
+		ib->ptr[ib->length_dw++] = ndw; /* number of entries */
+
+		pe += ndw * 8;
+		addr += ndw * incr;
+		count -= ndw;
+	}
+}
+
+/**
+ * sdma_v3_0_vm_pad_ib - pad the IB to the required number of dw
+ *
+ * @ib: indirect buffer to fill with padding
+ *
+ */
+static void sdma_v3_0_vm_pad_ib(struct amdgpu_ib *ib)
+{
+	struct amdgpu_sdma_instance *sdma = amdgpu_get_sdma_instance(ib->ring);
+	u32 pad_count;
+	int i;
+
+	pad_count = (8 - (ib->length_dw & 0x7)) % 8;
+	for (i = 0; i < pad_count; i++)
+		if (sdma && sdma->burst_nop && (i == 0))
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP) |
+				SDMA_PKT_NOP_HEADER_COUNT(pad_count - 1);
+		else
+			ib->ptr[ib->length_dw++] =
+				SDMA_PKT_HEADER_OP(SDMA_OP_NOP);
+}
+
+/**
+ * sdma_v3_0_ring_emit_vm_flush - cik vm flush using sDMA
+ *
+ * @ring: amdgpu_ring pointer
+ * @vm: amdgpu_vm pointer
+ *
+ * Update the page table base and flush the VM TLB
+ * using sDMA (VI).
+ */
+static void sdma_v3_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vm_id, uint64_t pd_addr)
+{
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	if (vm_id < 8) {
+		amdgpu_ring_write(ring, (mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id));
+	} else {
+		amdgpu_ring_write(ring, (mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8));
+	}
+	amdgpu_ring_write(ring, pd_addr >> 12);
+
+	/* flush TLB */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_SRBM_WRITE) |
+			  SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(0xf));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST);
+	amdgpu_ring_write(ring, 1 << vm_id);
+
+	/* wait for flush */
+	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_POLL_REGMEM) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(0) |
+			  SDMA_PKT_POLL_REGMEM_HEADER_FUNC(0)); /* always */
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, 0); /* reference */
+	amdgpu_ring_write(ring, 0); /* mask */
+	amdgpu_ring_write(ring, SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(0xfff) |
+			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
+}
+
+static int sdma_v3_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		adev->sdma.num_instances = 1;
+		break;
+	default:
+		adev->sdma.num_instances = SDMA_MAX_INSTANCE;
+		break;
+	}
+
+	sdma_v3_0_set_ring_funcs(adev);
+	sdma_v3_0_set_buffer_funcs(adev);
+	sdma_v3_0_set_vm_pte_funcs(adev);
+	sdma_v3_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int sdma_v3_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r, i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* SDMA trap event */
+	r = amdgpu_irq_add_id(adev, 224, &adev->sdma.trap_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, 241, &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	/* SDMA Privileged inst */
+	r = amdgpu_irq_add_id(adev, 247, &adev->sdma.illegal_inst_irq);
+	if (r)
+		return r;
+
+	r = sdma_v3_0_init_microcode(adev);
+	if (r) {
+		DRM_ERROR("Failed to load sdma firmware!\n");
+		return r;
+	}
+
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
+		ring->ring_obj = NULL;
+		ring->use_doorbell = true;
+		ring->doorbell_index = (i == 0) ?
+			AMDGPU_DOORBELL_sDMA_ENGINE0 : AMDGPU_DOORBELL_sDMA_ENGINE1;
+
+		sprintf(ring->name, "sdma%d", i);
+		r = amdgpu_ring_init(adev, ring, 256 * 1024,
+				     SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), 0xf,
+				     &adev->sdma.trap_irq,
+				     (i == 0) ?
+				     AMDGPU_SDMA_IRQ_TRAP0 : AMDGPU_SDMA_IRQ_TRAP1,
+				     AMDGPU_RING_TYPE_SDMA);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static int sdma_v3_0_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
+
+	return 0;
+}
+
+static int sdma_v3_0_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v3_0_init_golden_registers(adev);
+
+	r = sdma_v3_0_start(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int sdma_v3_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	sdma_v3_0_ctx_switch_enable(adev, false);
+	sdma_v3_0_enable(adev, false);
+
+	return 0;
+}
+
+static int sdma_v3_0_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v3_0_hw_fini(adev);
+}
+
+static int sdma_v3_0_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return sdma_v3_0_hw_init(adev);
+}
+
+static bool sdma_v3_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & (SRBM_STATUS2__SDMA_BUSY_MASK |
+		   SRBM_STATUS2__SDMA1_BUSY_MASK))
+	    return false;
+
+	return true;
+}
+
+static int sdma_v3_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmSRBM_STATUS2) & (SRBM_STATUS2__SDMA_BUSY_MASK |
+				SRBM_STATUS2__SDMA1_BUSY_MASK);
+
+		if (!tmp)
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void sdma_v3_0_print_status(void *handle)
+{
+	int i, j;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "VI SDMA registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		 RREG32(mmSRBM_STATUS2));
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		dev_info(adev->dev, "  SDMA%d_STATUS_REG=0x%08X\n",
+			 i, RREG32(mmSDMA0_STATUS_REG + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_F32_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_F32_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_SEM_WAIT_FAIL_TIMER_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_IB_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_CNTL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_CNTL + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_WPTR=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR_ADDR_HI=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_RPTR_ADDR_LO=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_RPTR_ADDR_LO + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_BASE=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_BASE + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_RB_BASE_HI=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_RB_BASE_HI + sdma_offsets[i]));
+		dev_info(adev->dev, "  SDMA%d_GFX_DOORBELL=0x%08X\n",
+			 i, RREG32(mmSDMA0_GFX_DOORBELL + sdma_offsets[i]));
+		mutex_lock(&adev->srbm_mutex);
+		for (j = 0; j < 16; j++) {
+			vi_srbm_select(adev, 0, 0, 0, j);
+			dev_info(adev->dev, "  VM %d:\n", j);
+			dev_info(adev->dev, "  SDMA%d_GFX_VIRTUAL_ADDR=0x%08X\n",
+				 i, RREG32(mmSDMA0_GFX_VIRTUAL_ADDR + sdma_offsets[i]));
+			dev_info(adev->dev, "  SDMA%d_GFX_APE1_CNTL=0x%08X\n",
+				 i, RREG32(mmSDMA0_GFX_APE1_CNTL + sdma_offsets[i]));
+		}
+		vi_srbm_select(adev, 0, 0, 0, 0);
+		mutex_unlock(&adev->srbm_mutex);
+	}
+}
+
+static int sdma_v3_0_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS2);
+
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK;
+	}
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0);
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+		srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK;
+	}
+
+	if (srbm_soft_reset) {
+		sdma_v3_0_print_status((void *)adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		sdma_v3_0_print_status((void *)adev);
+	}
+
+	return 0;
+}
+
+static int sdma_v3_0_set_trap_irq_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	u32 sdma_cntl;
+
+	switch (type) {
+	case AMDGPU_SDMA_IRQ_TRAP0:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA0_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	case AMDGPU_SDMA_IRQ_TRAP1:
+		switch (state) {
+		case AMDGPU_IRQ_STATE_DISABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 0);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		case AMDGPU_IRQ_STATE_ENABLE:
+			sdma_cntl = RREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET);
+			sdma_cntl = REG_SET_FIELD(sdma_cntl, SDMA0_CNTL, TRAP_ENABLE, 1);
+			WREG32(mmSDMA0_CNTL + SDMA1_REGISTER_OFFSET, sdma_cntl);
+			break;
+		default:
+			break;
+		}
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v3_0_process_trap_irq(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	u8 instance_id, queue_id;
+
+	instance_id = (entry->ring_id & 0x3) >> 0;
+	queue_id = (entry->ring_id & 0xc) >> 2;
+	DRM_DEBUG("IH: SDMA trap\n");
+	switch (instance_id) {
+	case 0:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[0].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	case 1:
+		switch (queue_id) {
+		case 0:
+			amdgpu_fence_process(&adev->sdma.instance[1].ring);
+			break;
+		case 1:
+			/* XXX compute */
+			break;
+		case 2:
+			/* XXX compute */
+			break;
+		}
+		break;
+	}
+	return 0;
+}
+
+static int sdma_v3_0_process_illegal_inst_irq(struct amdgpu_device *adev,
+					      struct amdgpu_irq_src *source,
+					      struct amdgpu_iv_entry *entry)
+{
+	DRM_ERROR("Illegal instruction in SDMA command stream\n");
+	schedule_work(&adev->reset_work);
+	return 0;
+}
+
+static int sdma_v3_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int sdma_v3_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs sdma_v3_0_ip_funcs = {
+	.early_init = sdma_v3_0_early_init,
+	.late_init = NULL,
+	.sw_init = sdma_v3_0_sw_init,
+	.sw_fini = sdma_v3_0_sw_fini,
+	.hw_init = sdma_v3_0_hw_init,
+	.hw_fini = sdma_v3_0_hw_fini,
+	.suspend = sdma_v3_0_suspend,
+	.resume = sdma_v3_0_resume,
+	.is_idle = sdma_v3_0_is_idle,
+	.wait_for_idle = sdma_v3_0_wait_for_idle,
+	.soft_reset = sdma_v3_0_soft_reset,
+	.print_status = sdma_v3_0_print_status,
+	.set_clockgating_state = sdma_v3_0_set_clockgating_state,
+	.set_powergating_state = sdma_v3_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
+	.get_rptr = sdma_v3_0_ring_get_rptr,
+	.get_wptr = sdma_v3_0_ring_get_wptr,
+	.set_wptr = sdma_v3_0_ring_set_wptr,
+	.parse_cs = NULL,
+	.emit_ib = sdma_v3_0_ring_emit_ib,
+	.emit_fence = sdma_v3_0_ring_emit_fence,
+	.emit_semaphore = sdma_v3_0_ring_emit_semaphore,
+	.emit_vm_flush = sdma_v3_0_ring_emit_vm_flush,
+	.emit_hdp_flush = sdma_v3_0_ring_emit_hdp_flush,
+	.test_ring = sdma_v3_0_ring_test_ring,
+	.test_ib = sdma_v3_0_ring_test_ib,
+	.insert_nop = sdma_v3_0_ring_insert_nop,
+};
+
+static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < adev->sdma.num_instances; i++)
+		adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {
+	.set = sdma_v3_0_set_trap_irq_state,
+	.process = sdma_v3_0_process_trap_irq,
+};
+
+static const struct amdgpu_irq_src_funcs sdma_v3_0_illegal_inst_irq_funcs = {
+	.process = sdma_v3_0_process_illegal_inst_irq,
+};
+
+static void sdma_v3_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_LAST;
+	adev->sdma.trap_irq.funcs = &sdma_v3_0_trap_irq_funcs;
+	adev->sdma.illegal_inst_irq.funcs = &sdma_v3_0_illegal_inst_irq_funcs;
+}
+
+/**
+ * sdma_v3_0_emit_copy_buffer - copy buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_offset: src GPU address
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Copy GPU buffers using the DMA engine (VI).
+ * Used by the amdgpu ttm implementation to move pages if
+ * registered as the asic copy callback.
+ */
+static void sdma_v3_0_emit_copy_buffer(struct amdgpu_ib *ib,
+				       uint64_t src_offset,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) |
+		SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR);
+	ib->ptr[ib->length_dw++] = byte_count;
+	ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */
+	ib->ptr[ib->length_dw++] = lower_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(src_offset);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+}
+
+/**
+ * sdma_v3_0_emit_fill_buffer - fill buffer using the sDMA engine
+ *
+ * @ring: amdgpu_ring structure holding ring information
+ * @src_data: value to write to buffer
+ * @dst_offset: dst GPU address
+ * @byte_count: number of bytes to xfer
+ *
+ * Fill GPU buffers using the DMA engine (VI).
+ */
+static void sdma_v3_0_emit_fill_buffer(struct amdgpu_ib *ib,
+				       uint32_t src_data,
+				       uint64_t dst_offset,
+				       uint32_t byte_count)
+{
+	ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_CONST_FILL);
+	ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset);
+	ib->ptr[ib->length_dw++] = src_data;
+	ib->ptr[ib->length_dw++] = byte_count;
+}
+
+static const struct amdgpu_buffer_funcs sdma_v3_0_buffer_funcs = {
+	.copy_max_bytes = 0x1fffff,
+	.copy_num_dw = 7,
+	.emit_copy_buffer = sdma_v3_0_emit_copy_buffer,
+
+	.fill_max_bytes = 0x1fffff,
+	.fill_num_dw = 5,
+	.emit_fill_buffer = sdma_v3_0_emit_fill_buffer,
+};
+
+static void sdma_v3_0_set_buffer_funcs(struct amdgpu_device *adev)
+{
+	if (adev->mman.buffer_funcs == NULL) {
+		adev->mman.buffer_funcs = &sdma_v3_0_buffer_funcs;
+		adev->mman.buffer_funcs_ring = &adev->sdma.instance[0].ring;
+	}
+}
+
+static const struct amdgpu_vm_pte_funcs sdma_v3_0_vm_pte_funcs = {
+	.copy_pte = sdma_v3_0_vm_copy_pte,
+	.write_pte = sdma_v3_0_vm_write_pte,
+	.set_pte_pde = sdma_v3_0_vm_set_pte_pde,
+	.pad_ib = sdma_v3_0_vm_pad_ib,
+};
+
+static void sdma_v3_0_set_vm_pte_funcs(struct amdgpu_device *adev)
+{
+	if (adev->vm_manager.vm_pte_funcs == NULL) {
+		adev->vm_manager.vm_pte_funcs = &sdma_v3_0_vm_pte_funcs;
+		adev->vm_manager.vm_pte_funcs_ring = &adev->sdma.instance[0].ring;
+		adev->vm_manager.vm_pte_funcs_ring->is_pte_ring = true;
+	}
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.h
new file mode 100644
index 000000000..0cb9698a3
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SDMA_V3_0_H__
+#define __SDMA_V3_0_H__
+
+extern const struct amd_ip_funcs sdma_v3_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu7.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu7.h
new file mode 100644
index 000000000..75a380a15
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu7.h
@@ -0,0 +1,170 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU7_H
+#define SMU7_H
+
+#pragma pack(push, 1)
+
+#define SMU7_CONTEXT_ID_SMC        1
+#define SMU7_CONTEXT_ID_VBIOS      2
+
+
+#define SMU7_CONTEXT_ID_SMC        1
+#define SMU7_CONTEXT_ID_VBIOS      2
+
+#define SMU7_MAX_LEVELS_VDDC            8
+#define SMU7_MAX_LEVELS_VDDCI           4
+#define SMU7_MAX_LEVELS_MVDD            4
+#define SMU7_MAX_LEVELS_VDDNB           8
+
+#define SMU7_MAX_LEVELS_GRAPHICS        SMU__NUM_SCLK_DPM_STATE   // SCLK + SQ DPM + ULV
+#define SMU7_MAX_LEVELS_MEMORY          SMU__NUM_MCLK_DPM_LEVELS   // MCLK Levels DPM
+#define SMU7_MAX_LEVELS_GIO             SMU__NUM_LCLK_DPM_LEVELS  // LCLK Levels
+#define SMU7_MAX_LEVELS_LINK            SMU__NUM_PCIE_DPM_LEVELS  // PCIe speed and number of lanes.
+#define SMU7_MAX_LEVELS_UVD             8   // VCLK/DCLK levels for UVD.
+#define SMU7_MAX_LEVELS_VCE             8   // ECLK levels for VCE.
+#define SMU7_MAX_LEVELS_ACP             8   // ACLK levels for ACP.
+#define SMU7_MAX_LEVELS_SAMU            8   // SAMCLK levels for SAMU.
+#define SMU7_MAX_ENTRIES_SMIO           32  // Number of entries in SMIO table.
+
+#define DPM_NO_LIMIT 0
+#define DPM_NO_UP 1
+#define DPM_GO_DOWN 2
+#define DPM_GO_UP 3
+
+#define SMU7_FIRST_DPM_GRAPHICS_LEVEL    0
+#define SMU7_FIRST_DPM_MEMORY_LEVEL      0
+
+#define GPIO_CLAMP_MODE_VRHOT      1
+#define GPIO_CLAMP_MODE_THERM      2
+#define GPIO_CLAMP_MODE_DC         4
+
+#define SCRATCH_B_TARG_PCIE_INDEX_SHIFT 0
+#define SCRATCH_B_TARG_PCIE_INDEX_MASK  (0x7<<SCRATCH_B_TARG_PCIE_INDEX_SHIFT)
+#define SCRATCH_B_CURR_PCIE_INDEX_SHIFT 3
+#define SCRATCH_B_CURR_PCIE_INDEX_MASK  (0x7<<SCRATCH_B_CURR_PCIE_INDEX_SHIFT)
+#define SCRATCH_B_TARG_UVD_INDEX_SHIFT  6
+#define SCRATCH_B_TARG_UVD_INDEX_MASK   (0x7<<SCRATCH_B_TARG_UVD_INDEX_SHIFT)
+#define SCRATCH_B_CURR_UVD_INDEX_SHIFT  9
+#define SCRATCH_B_CURR_UVD_INDEX_MASK   (0x7<<SCRATCH_B_CURR_UVD_INDEX_SHIFT)
+#define SCRATCH_B_TARG_VCE_INDEX_SHIFT  12
+#define SCRATCH_B_TARG_VCE_INDEX_MASK   (0x7<<SCRATCH_B_TARG_VCE_INDEX_SHIFT)
+#define SCRATCH_B_CURR_VCE_INDEX_SHIFT  15
+#define SCRATCH_B_CURR_VCE_INDEX_MASK   (0x7<<SCRATCH_B_CURR_VCE_INDEX_SHIFT)
+#define SCRATCH_B_TARG_ACP_INDEX_SHIFT  18
+#define SCRATCH_B_TARG_ACP_INDEX_MASK   (0x7<<SCRATCH_B_TARG_ACP_INDEX_SHIFT)
+#define SCRATCH_B_CURR_ACP_INDEX_SHIFT  21
+#define SCRATCH_B_CURR_ACP_INDEX_MASK   (0x7<<SCRATCH_B_CURR_ACP_INDEX_SHIFT)
+#define SCRATCH_B_TARG_SAMU_INDEX_SHIFT 24
+#define SCRATCH_B_TARG_SAMU_INDEX_MASK  (0x7<<SCRATCH_B_TARG_SAMU_INDEX_SHIFT)
+#define SCRATCH_B_CURR_SAMU_INDEX_SHIFT 27
+#define SCRATCH_B_CURR_SAMU_INDEX_MASK  (0x7<<SCRATCH_B_CURR_SAMU_INDEX_SHIFT)
+
+
+struct SMU7_PIDController
+{
+    uint32_t Ki;
+    int32_t LFWindupUL;
+    int32_t LFWindupLL;
+    uint32_t StatePrecision;
+    uint32_t LfPrecision;
+    uint32_t LfOffset;
+    uint32_t MaxState;
+    uint32_t MaxLfFraction;
+    uint32_t StateShift;
+};
+
+typedef struct SMU7_PIDController SMU7_PIDController;
+
+// -------------------------------------------------------------------------------------------------------------------------
+#define SMU7_MAX_PCIE_LINK_SPEEDS 3 /* 0:Gen1 1:Gen2 2:Gen3 */
+
+#define SMU7_SCLK_DPM_CONFIG_MASK                        0x01
+#define SMU7_VOLTAGE_CONTROLLER_CONFIG_MASK              0x02
+#define SMU7_THERMAL_CONTROLLER_CONFIG_MASK              0x04
+#define SMU7_MCLK_DPM_CONFIG_MASK                        0x08
+#define SMU7_UVD_DPM_CONFIG_MASK                         0x10
+#define SMU7_VCE_DPM_CONFIG_MASK                         0x20
+#define SMU7_ACP_DPM_CONFIG_MASK                         0x40
+#define SMU7_SAMU_DPM_CONFIG_MASK                        0x80
+#define SMU7_PCIEGEN_DPM_CONFIG_MASK                    0x100
+
+#define SMU7_ACP_MCLK_HANDSHAKE_DISABLE                  0x00000001
+#define SMU7_ACP_SCLK_HANDSHAKE_DISABLE                  0x00000002
+#define SMU7_UVD_MCLK_HANDSHAKE_DISABLE                  0x00000100
+#define SMU7_UVD_SCLK_HANDSHAKE_DISABLE                  0x00000200
+#define SMU7_VCE_MCLK_HANDSHAKE_DISABLE                  0x00010000
+#define SMU7_VCE_SCLK_HANDSHAKE_DISABLE                  0x00020000
+
+struct SMU7_Firmware_Header
+{
+    uint32_t Digest[5];
+    uint32_t Version;
+    uint32_t HeaderSize;
+    uint32_t Flags;
+    uint32_t EntryPoint;
+    uint32_t CodeSize;
+    uint32_t ImageSize;
+
+    uint32_t Rtos;
+    uint32_t SoftRegisters;
+    uint32_t DpmTable;
+    uint32_t FanTable;
+    uint32_t CacConfigTable;
+    uint32_t CacStatusTable;
+
+    uint32_t mcRegisterTable;
+
+    uint32_t mcArbDramTimingTable;
+
+    uint32_t PmFuseTable;
+    uint32_t Globals;
+    uint32_t Reserved[42];
+    uint32_t Signature;
+};
+
+typedef struct SMU7_Firmware_Header SMU7_Firmware_Header;
+
+#define SMU7_FIRMWARE_HEADER_LOCATION 0x20000
+
+enum  DisplayConfig {
+    PowerDown = 1,
+    DP54x4,
+    DP54x2,
+    DP54x1,
+    DP27x4,
+    DP27x2,
+    DP27x1,
+    HDMI297,
+    HDMI162,
+    LVDS,
+    DP324x4,
+    DP324x2,
+    DP324x1
+};
+
+#pragma pack(pop)
+
+#endif
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu7_discrete.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu7_discrete.h
new file mode 100644
index 000000000..0b0b404ff
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu7_discrete.h
@@ -0,0 +1,514 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU7_DISCRETE_H
+#define SMU7_DISCRETE_H
+
+#include "smu7.h"
+
+#pragma pack(push, 1)
+
+#define SMU7_DTE_ITERATIONS 5
+#define SMU7_DTE_SOURCES 3
+#define SMU7_DTE_SINKS 1
+#define SMU7_NUM_CPU_TES 0
+#define SMU7_NUM_GPU_TES 1
+#define SMU7_NUM_NON_TES 2
+
+struct SMU7_SoftRegisters
+{
+    uint32_t        RefClockFrequency;
+    uint32_t        PmTimerP;
+    uint32_t        FeatureEnables;
+    uint32_t        PreVBlankGap;
+    uint32_t        VBlankTimeout;
+    uint32_t        TrainTimeGap;
+
+    uint32_t        MvddSwitchTime;
+    uint32_t        LongestAcpiTrainTime;
+    uint32_t        AcpiDelay;
+    uint32_t        G5TrainTime;
+    uint32_t        DelayMpllPwron;
+    uint32_t        VoltageChangeTimeout;
+    uint32_t        HandshakeDisables;
+
+    uint8_t         DisplayPhy1Config;
+    uint8_t         DisplayPhy2Config;
+    uint8_t         DisplayPhy3Config;
+    uint8_t         DisplayPhy4Config;
+
+    uint8_t         DisplayPhy5Config;
+    uint8_t         DisplayPhy6Config;
+    uint8_t         DisplayPhy7Config;
+    uint8_t         DisplayPhy8Config;
+
+    uint32_t        AverageGraphicsA;
+    uint32_t        AverageMemoryA;
+    uint32_t        AverageGioA;
+
+    uint8_t         SClkDpmEnabledLevels;
+    uint8_t         MClkDpmEnabledLevels;
+    uint8_t         LClkDpmEnabledLevels;
+    uint8_t         PCIeDpmEnabledLevels;
+
+    uint8_t         UVDDpmEnabledLevels;
+    uint8_t         SAMUDpmEnabledLevels;
+    uint8_t         ACPDpmEnabledLevels;
+    uint8_t         VCEDpmEnabledLevels;
+
+    uint32_t        DRAM_LOG_ADDR_H;
+    uint32_t        DRAM_LOG_ADDR_L;
+    uint32_t        DRAM_LOG_PHY_ADDR_H;
+    uint32_t        DRAM_LOG_PHY_ADDR_L;
+    uint32_t        DRAM_LOG_BUFF_SIZE;
+    uint32_t        UlvEnterC;
+    uint32_t        UlvTime;
+    uint32_t        Reserved[3];
+
+};
+
+typedef struct SMU7_SoftRegisters SMU7_SoftRegisters;
+
+struct SMU7_Discrete_VoltageLevel
+{
+    uint16_t    Voltage;
+    uint16_t    StdVoltageHiSidd;
+    uint16_t    StdVoltageLoSidd;
+    uint8_t     Smio;
+    uint8_t     padding;
+};
+
+typedef struct SMU7_Discrete_VoltageLevel SMU7_Discrete_VoltageLevel;
+
+struct SMU7_Discrete_GraphicsLevel
+{
+    uint32_t    Flags;
+    uint32_t    MinVddc;
+    uint32_t    MinVddcPhases;
+
+    uint32_t    SclkFrequency;
+
+    uint8_t     padding1[2];
+    uint16_t    ActivityLevel;
+
+    uint32_t    CgSpllFuncCntl3;
+    uint32_t    CgSpllFuncCntl4;
+    uint32_t    SpllSpreadSpectrum;
+    uint32_t    SpllSpreadSpectrum2;
+    uint32_t    CcPwrDynRm;
+    uint32_t    CcPwrDynRm1;
+    uint8_t     SclkDid;
+    uint8_t     DisplayWatermark;
+    uint8_t     EnabledForActivity;
+    uint8_t     EnabledForThrottle;
+    uint8_t     UpH;
+    uint8_t     DownH;
+    uint8_t     VoltageDownH;
+    uint8_t     PowerThrottle;
+    uint8_t     DeepSleepDivId;
+    uint8_t     padding[3];
+};
+
+typedef struct SMU7_Discrete_GraphicsLevel SMU7_Discrete_GraphicsLevel;
+
+struct SMU7_Discrete_ACPILevel
+{
+    uint32_t    Flags;
+    uint32_t    MinVddc;
+    uint32_t    MinVddcPhases;
+    uint32_t    SclkFrequency;
+    uint8_t     SclkDid;
+    uint8_t     DisplayWatermark;
+    uint8_t     DeepSleepDivId;
+    uint8_t     padding;
+    uint32_t    CgSpllFuncCntl;
+    uint32_t    CgSpllFuncCntl2;
+    uint32_t    CgSpllFuncCntl3;
+    uint32_t    CgSpllFuncCntl4;
+    uint32_t    SpllSpreadSpectrum;
+    uint32_t    SpllSpreadSpectrum2;
+    uint32_t    CcPwrDynRm;
+    uint32_t    CcPwrDynRm1;
+};
+
+typedef struct SMU7_Discrete_ACPILevel SMU7_Discrete_ACPILevel;
+
+struct SMU7_Discrete_Ulv
+{
+    uint32_t    CcPwrDynRm;
+    uint32_t    CcPwrDynRm1;
+    uint16_t    VddcOffset;
+    uint8_t     VddcOffsetVid;
+    uint8_t     VddcPhase;
+    uint32_t    Reserved;
+};
+
+typedef struct SMU7_Discrete_Ulv SMU7_Discrete_Ulv;
+
+struct SMU7_Discrete_MemoryLevel
+{
+    uint32_t    MinVddc;
+    uint32_t    MinVddcPhases;
+    uint32_t    MinVddci;
+    uint32_t    MinMvdd;
+
+    uint32_t    MclkFrequency;
+
+    uint8_t     EdcReadEnable;
+    uint8_t     EdcWriteEnable;
+    uint8_t     RttEnable;
+    uint8_t     StutterEnable;
+
+    uint8_t     StrobeEnable;
+    uint8_t     StrobeRatio;
+    uint8_t     EnabledForThrottle;
+    uint8_t     EnabledForActivity;
+
+    uint8_t     UpH;
+    uint8_t     DownH;
+    uint8_t     VoltageDownH;
+    uint8_t     padding;
+
+    uint16_t    ActivityLevel;
+    uint8_t     DisplayWatermark;
+    uint8_t     padding1;
+
+    uint32_t    MpllFuncCntl;
+    uint32_t    MpllFuncCntl_1;
+    uint32_t    MpllFuncCntl_2;
+    uint32_t    MpllAdFuncCntl;
+    uint32_t    MpllDqFuncCntl;
+    uint32_t    MclkPwrmgtCntl;
+    uint32_t    DllCntl;
+    uint32_t    MpllSs1;
+    uint32_t    MpllSs2;
+};
+
+typedef struct SMU7_Discrete_MemoryLevel SMU7_Discrete_MemoryLevel;
+
+struct SMU7_Discrete_LinkLevel
+{
+    uint8_t     PcieGenSpeed;
+    uint8_t     PcieLaneCount;
+    uint8_t     EnabledForActivity;
+    uint8_t     Padding;
+    uint32_t    DownT;
+    uint32_t    UpT;
+    uint32_t    Reserved;
+};
+
+typedef struct SMU7_Discrete_LinkLevel SMU7_Discrete_LinkLevel;
+
+
+struct SMU7_Discrete_MCArbDramTimingTableEntry
+{
+    uint32_t McArbDramTiming;
+    uint32_t McArbDramTiming2;
+    uint8_t  McArbBurstTime;
+    uint8_t  padding[3];
+};
+
+typedef struct SMU7_Discrete_MCArbDramTimingTableEntry SMU7_Discrete_MCArbDramTimingTableEntry;
+
+struct SMU7_Discrete_MCArbDramTimingTable
+{
+    SMU7_Discrete_MCArbDramTimingTableEntry entries[SMU__NUM_SCLK_DPM_STATE][SMU__NUM_MCLK_DPM_LEVELS];
+};
+
+typedef struct SMU7_Discrete_MCArbDramTimingTable SMU7_Discrete_MCArbDramTimingTable;
+
+struct SMU7_Discrete_UvdLevel
+{
+    uint32_t VclkFrequency;
+    uint32_t DclkFrequency;
+    uint16_t MinVddc;
+    uint8_t  MinVddcPhases;
+    uint8_t  VclkDivider;
+    uint8_t  DclkDivider;
+    uint8_t  padding[3];
+};
+
+typedef struct SMU7_Discrete_UvdLevel SMU7_Discrete_UvdLevel;
+
+struct SMU7_Discrete_ExtClkLevel
+{
+    uint32_t Frequency;
+    uint16_t MinVoltage;
+    uint8_t  MinPhases;
+    uint8_t  Divider;
+};
+
+typedef struct SMU7_Discrete_ExtClkLevel SMU7_Discrete_ExtClkLevel;
+
+struct SMU7_Discrete_StateInfo
+{
+    uint32_t SclkFrequency;
+    uint32_t MclkFrequency;
+    uint32_t VclkFrequency;
+    uint32_t DclkFrequency;
+    uint32_t SamclkFrequency;
+    uint32_t AclkFrequency;
+    uint32_t EclkFrequency;
+    uint16_t MvddVoltage;
+    uint16_t padding16;
+    uint8_t  DisplayWatermark;
+    uint8_t  McArbIndex;
+    uint8_t  McRegIndex;
+    uint8_t  SeqIndex;
+    uint8_t  SclkDid;
+    int8_t   SclkIndex;
+    int8_t   MclkIndex;
+    uint8_t  PCIeGen;
+
+};
+
+typedef struct SMU7_Discrete_StateInfo SMU7_Discrete_StateInfo;
+
+
+struct SMU7_Discrete_DpmTable
+{
+    SMU7_PIDController                  GraphicsPIDController;
+    SMU7_PIDController                  MemoryPIDController;
+    SMU7_PIDController                  LinkPIDController;
+
+    uint32_t                            SystemFlags;
+
+
+    uint32_t                            SmioMaskVddcVid;
+    uint32_t                            SmioMaskVddcPhase;
+    uint32_t                            SmioMaskVddciVid;
+    uint32_t                            SmioMaskMvddVid;
+
+    uint32_t                            VddcLevelCount;
+    uint32_t                            VddciLevelCount;
+    uint32_t                            MvddLevelCount;
+
+    SMU7_Discrete_VoltageLevel          VddcLevel               [SMU7_MAX_LEVELS_VDDC];
+//    SMU7_Discrete_VoltageLevel          VddcStandardReference   [SMU7_MAX_LEVELS_VDDC];
+    SMU7_Discrete_VoltageLevel          VddciLevel              [SMU7_MAX_LEVELS_VDDCI];
+    SMU7_Discrete_VoltageLevel          MvddLevel               [SMU7_MAX_LEVELS_MVDD];
+
+    uint8_t                             GraphicsDpmLevelCount;
+    uint8_t                             MemoryDpmLevelCount;
+    uint8_t                             LinkLevelCount;
+    uint8_t                             UvdLevelCount;
+    uint8_t                             VceLevelCount;
+    uint8_t                             AcpLevelCount;
+    uint8_t                             SamuLevelCount;
+    uint8_t                             MasterDeepSleepControl;
+    uint32_t                            Reserved[5];
+//    uint32_t                            SamuDefaultLevel;
+
+    SMU7_Discrete_GraphicsLevel         GraphicsLevel           [SMU7_MAX_LEVELS_GRAPHICS];
+    SMU7_Discrete_MemoryLevel           MemoryACPILevel;
+    SMU7_Discrete_MemoryLevel           MemoryLevel             [SMU7_MAX_LEVELS_MEMORY];
+    SMU7_Discrete_LinkLevel             LinkLevel               [SMU7_MAX_LEVELS_LINK];
+    SMU7_Discrete_ACPILevel             ACPILevel;
+    SMU7_Discrete_UvdLevel              UvdLevel                [SMU7_MAX_LEVELS_UVD];
+    SMU7_Discrete_ExtClkLevel           VceLevel                [SMU7_MAX_LEVELS_VCE];
+    SMU7_Discrete_ExtClkLevel           AcpLevel                [SMU7_MAX_LEVELS_ACP];
+    SMU7_Discrete_ExtClkLevel           SamuLevel               [SMU7_MAX_LEVELS_SAMU];
+    SMU7_Discrete_Ulv                   Ulv;
+
+    uint32_t                            SclkStepSize;
+    uint32_t                            Smio                    [SMU7_MAX_ENTRIES_SMIO];
+
+    uint8_t                             UvdBootLevel;
+    uint8_t                             VceBootLevel;
+    uint8_t                             AcpBootLevel;
+    uint8_t                             SamuBootLevel;
+
+    uint8_t                             UVDInterval;
+    uint8_t                             VCEInterval;
+    uint8_t                             ACPInterval;
+    uint8_t                             SAMUInterval;
+
+    uint8_t                             GraphicsBootLevel;
+    uint8_t                             GraphicsVoltageChangeEnable;
+    uint8_t                             GraphicsThermThrottleEnable;
+    uint8_t                             GraphicsInterval;
+
+    uint8_t                             VoltageInterval;
+    uint8_t                             ThermalInterval;
+    uint16_t                            TemperatureLimitHigh;
+
+    uint16_t                            TemperatureLimitLow;
+    uint8_t                             MemoryBootLevel;
+    uint8_t                             MemoryVoltageChangeEnable;
+
+    uint8_t                             MemoryInterval;
+    uint8_t                             MemoryThermThrottleEnable;
+    uint16_t                            VddcVddciDelta;
+
+    uint16_t                            VoltageResponseTime;
+    uint16_t                            PhaseResponseTime;
+
+    uint8_t                             PCIeBootLinkLevel;
+    uint8_t                             PCIeGenInterval;
+    uint8_t                             DTEInterval;
+    uint8_t                             DTEMode;
+
+    uint8_t                             SVI2Enable;
+    uint8_t                             VRHotGpio;
+    uint8_t                             AcDcGpio;
+    uint8_t                             ThermGpio;
+
+    uint16_t                            PPM_PkgPwrLimit;
+    uint16_t                            PPM_TemperatureLimit;
+
+    uint16_t                            DefaultTdp;
+    uint16_t                            TargetTdp;
+
+    uint16_t                            FpsHighT;
+    uint16_t                            FpsLowT;
+
+    uint16_t                            BAPMTI_R  [SMU7_DTE_ITERATIONS][SMU7_DTE_SOURCES][SMU7_DTE_SINKS];
+    uint16_t                            BAPMTI_RC [SMU7_DTE_ITERATIONS][SMU7_DTE_SOURCES][SMU7_DTE_SINKS];
+
+    uint8_t                             DTEAmbientTempBase;
+    uint8_t                             DTETjOffset;
+    uint8_t                             GpuTjMax;
+    uint8_t                             GpuTjHyst;
+
+    uint16_t                            BootVddc;
+    uint16_t                            BootVddci;
+
+    uint16_t                            BootMVdd;
+    uint16_t                            padding;
+
+    uint32_t                            BAPM_TEMP_GRADIENT;
+
+    uint32_t                            LowSclkInterruptT;
+};
+
+typedef struct SMU7_Discrete_DpmTable SMU7_Discrete_DpmTable;
+
+#define SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE 16
+#define SMU7_DISCRETE_MC_REGISTER_ARRAY_SET_COUNT SMU7_MAX_LEVELS_MEMORY
+
+struct SMU7_Discrete_MCRegisterAddress
+{
+    uint16_t s0;
+    uint16_t s1;
+};
+
+typedef struct SMU7_Discrete_MCRegisterAddress SMU7_Discrete_MCRegisterAddress;
+
+struct SMU7_Discrete_MCRegisterSet
+{
+    uint32_t value[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
+};
+
+typedef struct SMU7_Discrete_MCRegisterSet SMU7_Discrete_MCRegisterSet;
+
+struct SMU7_Discrete_MCRegisters
+{
+    uint8_t                             last;
+    uint8_t                             reserved[3];
+    SMU7_Discrete_MCRegisterAddress     address[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
+    SMU7_Discrete_MCRegisterSet         data[SMU7_DISCRETE_MC_REGISTER_ARRAY_SET_COUNT];
+};
+
+typedef struct SMU7_Discrete_MCRegisters SMU7_Discrete_MCRegisters;
+
+struct SMU7_Discrete_FanTable
+{
+	uint16_t FdoMode;
+	int16_t  TempMin;
+	int16_t  TempMed;
+	int16_t  TempMax;
+	int16_t  Slope1;
+	int16_t  Slope2;
+	int16_t  FdoMin;
+	int16_t  HystUp;
+	int16_t  HystDown;
+	int16_t  HystSlope;
+	int16_t  TempRespLim;
+	int16_t  TempCurr;
+	int16_t  SlopeCurr;
+	int16_t  PwmCurr;
+	uint32_t RefreshPeriod;
+	int16_t  FdoMax;
+	uint8_t  TempSrc;
+	int8_t   Padding;
+};
+
+typedef struct SMU7_Discrete_FanTable SMU7_Discrete_FanTable;
+
+
+struct SMU7_Discrete_PmFuses {
+  // dw0-dw1
+  uint8_t BapmVddCVidHiSidd[8];
+
+  // dw2-dw3
+  uint8_t BapmVddCVidLoSidd[8];
+
+  // dw4-dw5
+  uint8_t VddCVid[8];
+
+  // dw6
+  uint8_t SviLoadLineEn;
+  uint8_t SviLoadLineVddC;
+  uint8_t SviLoadLineTrimVddC;
+  uint8_t SviLoadLineOffsetVddC;
+
+  // dw7
+  uint16_t TDC_VDDC_PkgLimit;
+  uint8_t TDC_VDDC_ThrottleReleaseLimitPerc;
+  uint8_t TDC_MAWt;
+
+  // dw8
+  uint8_t TdcWaterfallCtl;
+  uint8_t LPMLTemperatureMin;
+  uint8_t LPMLTemperatureMax;
+  uint8_t Reserved;
+
+  // dw9-dw10
+  uint8_t BapmVddCVidHiSidd2[8];
+
+  // dw11-dw12
+  int16_t FuzzyFan_ErrorSetDelta;
+  int16_t FuzzyFan_ErrorRateSetDelta;
+  int16_t FuzzyFan_PwmSetDelta;
+  uint16_t CalcMeasPowerBlend;
+
+  // dw13-dw16
+  uint8_t GnbLPML[16];
+
+  // dw17
+  uint8_t GnbLPMLMaxVid;
+  uint8_t GnbLPMLMinVid;
+  uint8_t Reserved1[2];
+
+  // dw18
+  uint16_t BapmVddCBaseLeakageHiSidd;
+  uint16_t BapmVddCBaseLeakageLoSidd;
+};
+
+typedef struct SMU7_Discrete_PmFuses SMU7_Discrete_PmFuses;
+
+
+#pragma pack(pop)
+
+#endif
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu7_fusion.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu7_fusion.h
new file mode 100644
index 000000000..78ada9ffd
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu7_fusion.h
@@ -0,0 +1,300 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU7_FUSION_H
+#define SMU7_FUSION_H
+
+#include "smu7.h"
+
+#pragma pack(push, 1)
+
+#define SMU7_DTE_ITERATIONS 5
+#define SMU7_DTE_SOURCES 5
+#define SMU7_DTE_SINKS 3
+#define SMU7_NUM_CPU_TES 2
+#define SMU7_NUM_GPU_TES 1
+#define SMU7_NUM_NON_TES 2
+
+// All 'soft registers' should be uint32_t.
+struct SMU7_SoftRegisters
+{
+    uint32_t        RefClockFrequency;
+    uint32_t        PmTimerP;
+    uint32_t        FeatureEnables;
+    uint32_t        HandshakeDisables;
+
+    uint8_t         DisplayPhy1Config;
+    uint8_t         DisplayPhy2Config;
+    uint8_t         DisplayPhy3Config;
+    uint8_t         DisplayPhy4Config;
+
+    uint8_t         DisplayPhy5Config;
+    uint8_t         DisplayPhy6Config;
+    uint8_t         DisplayPhy7Config;
+    uint8_t         DisplayPhy8Config;
+
+    uint32_t        AverageGraphicsA;
+    uint32_t        AverageMemoryA;
+    uint32_t        AverageGioA;
+
+    uint8_t         SClkDpmEnabledLevels;
+    uint8_t         MClkDpmEnabledLevels;
+    uint8_t         LClkDpmEnabledLevels;
+    uint8_t         PCIeDpmEnabledLevels;
+
+    uint8_t         UVDDpmEnabledLevels;
+    uint8_t         SAMUDpmEnabledLevels;
+    uint8_t         ACPDpmEnabledLevels;
+    uint8_t         VCEDpmEnabledLevels;
+
+    uint32_t        DRAM_LOG_ADDR_H;
+    uint32_t        DRAM_LOG_ADDR_L;
+    uint32_t        DRAM_LOG_PHY_ADDR_H;
+    uint32_t        DRAM_LOG_PHY_ADDR_L;
+    uint32_t        DRAM_LOG_BUFF_SIZE;
+    uint32_t        UlvEnterC;
+    uint32_t        UlvTime;
+    uint32_t        Reserved[3];
+
+};
+
+typedef struct SMU7_SoftRegisters SMU7_SoftRegisters;
+
+struct SMU7_Fusion_GraphicsLevel
+{
+    uint32_t    MinVddNb;
+
+    uint32_t    SclkFrequency;
+
+    uint8_t     Vid;
+    uint8_t     VidOffset;
+    uint16_t    AT;
+
+    uint8_t     PowerThrottle;
+    uint8_t     GnbSlow;
+    uint8_t     ForceNbPs1;
+    uint8_t     SclkDid;
+
+    uint8_t     DisplayWatermark;
+    uint8_t     EnabledForActivity;
+    uint8_t     EnabledForThrottle;
+    uint8_t     UpH;
+
+    uint8_t     DownH;
+    uint8_t     VoltageDownH;
+    uint8_t     DeepSleepDivId;
+
+    uint8_t     ClkBypassCntl;
+
+    uint32_t    reserved;
+};
+
+typedef struct SMU7_Fusion_GraphicsLevel SMU7_Fusion_GraphicsLevel;
+
+struct SMU7_Fusion_GIOLevel
+{
+    uint8_t     EnabledForActivity;
+    uint8_t     LclkDid;
+    uint8_t     Vid;
+    uint8_t     VoltageDownH;
+
+    uint32_t    MinVddNb;
+
+    uint16_t    ResidencyCounter;
+    uint8_t     UpH;
+    uint8_t     DownH;
+
+    uint32_t    LclkFrequency;
+
+    uint8_t     ActivityLevel;
+    uint8_t     EnabledForThrottle;
+
+    uint8_t     ClkBypassCntl;
+
+    uint8_t     padding;
+};
+
+typedef struct SMU7_Fusion_GIOLevel SMU7_Fusion_GIOLevel;
+
+// UVD VCLK/DCLK state (level) definition.
+struct SMU7_Fusion_UvdLevel
+{
+    uint32_t VclkFrequency;
+    uint32_t DclkFrequency;
+    uint16_t MinVddNb;
+    uint8_t  VclkDivider;
+    uint8_t  DclkDivider;
+
+    uint8_t     VClkBypassCntl;
+    uint8_t     DClkBypassCntl;
+
+    uint8_t     padding[2];
+
+};
+
+typedef struct SMU7_Fusion_UvdLevel SMU7_Fusion_UvdLevel;
+
+// Clocks for other external blocks (VCE, ACP, SAMU).
+struct SMU7_Fusion_ExtClkLevel
+{
+    uint32_t Frequency;
+    uint16_t MinVoltage;
+    uint8_t  Divider;
+    uint8_t  ClkBypassCntl;
+
+    uint32_t Reserved;
+};
+typedef struct SMU7_Fusion_ExtClkLevel SMU7_Fusion_ExtClkLevel;
+
+struct SMU7_Fusion_ACPILevel
+{
+    uint32_t    Flags;
+    uint32_t    MinVddNb;
+    uint32_t    SclkFrequency;
+    uint8_t     SclkDid;
+    uint8_t     GnbSlow;
+    uint8_t     ForceNbPs1;
+    uint8_t     DisplayWatermark;
+    uint8_t     DeepSleepDivId;
+    uint8_t     padding[3];
+};
+
+typedef struct SMU7_Fusion_ACPILevel SMU7_Fusion_ACPILevel;
+
+struct SMU7_Fusion_NbDpm
+{
+    uint8_t DpmXNbPsHi;
+    uint8_t DpmXNbPsLo;
+    uint8_t Dpm0PgNbPsHi;
+    uint8_t Dpm0PgNbPsLo;
+    uint8_t EnablePsi1;
+    uint8_t SkipDPM0;
+    uint8_t SkipPG;
+    uint8_t Hysteresis;
+    uint8_t EnableDpmPstatePoll;
+    uint8_t padding[3];
+};
+
+typedef struct SMU7_Fusion_NbDpm SMU7_Fusion_NbDpm;
+
+struct SMU7_Fusion_StateInfo
+{
+    uint32_t SclkFrequency;
+    uint32_t LclkFrequency;
+    uint32_t VclkFrequency;
+    uint32_t DclkFrequency;
+    uint32_t SamclkFrequency;
+    uint32_t AclkFrequency;
+    uint32_t EclkFrequency;
+    uint8_t  DisplayWatermark;
+    uint8_t  McArbIndex;
+    int8_t   SclkIndex;
+    int8_t   MclkIndex;
+};
+
+typedef struct SMU7_Fusion_StateInfo SMU7_Fusion_StateInfo;
+
+struct SMU7_Fusion_DpmTable
+{
+    uint32_t                            SystemFlags;
+
+    SMU7_PIDController                  GraphicsPIDController;
+    SMU7_PIDController                  GioPIDController;
+
+    uint8_t                            GraphicsDpmLevelCount;
+    uint8_t                            GIOLevelCount;
+    uint8_t                            UvdLevelCount;
+    uint8_t                            VceLevelCount;
+
+    uint8_t                            AcpLevelCount;
+    uint8_t                            SamuLevelCount;
+    uint16_t                           FpsHighT;
+
+    SMU7_Fusion_GraphicsLevel         GraphicsLevel           [SMU__NUM_SCLK_DPM_STATE];
+    SMU7_Fusion_ACPILevel             ACPILevel;
+    SMU7_Fusion_UvdLevel              UvdLevel                [SMU7_MAX_LEVELS_UVD];
+    SMU7_Fusion_ExtClkLevel           VceLevel                [SMU7_MAX_LEVELS_VCE];
+    SMU7_Fusion_ExtClkLevel           AcpLevel                [SMU7_MAX_LEVELS_ACP];
+    SMU7_Fusion_ExtClkLevel           SamuLevel               [SMU7_MAX_LEVELS_SAMU];
+
+    uint8_t                           UvdBootLevel;
+    uint8_t                           VceBootLevel;
+    uint8_t                           AcpBootLevel;
+    uint8_t                           SamuBootLevel;
+    uint8_t                           UVDInterval;
+    uint8_t                           VCEInterval;
+    uint8_t                           ACPInterval;
+    uint8_t                           SAMUInterval;
+
+    uint8_t                           GraphicsBootLevel;
+    uint8_t                           GraphicsInterval;
+    uint8_t                           GraphicsThermThrottleEnable;
+    uint8_t                           GraphicsVoltageChangeEnable;
+
+    uint8_t                           GraphicsClkSlowEnable;
+    uint8_t                           GraphicsClkSlowDivider;
+    uint16_t                          FpsLowT;
+
+    uint32_t                          DisplayCac;
+    uint32_t                          LowSclkInterruptT;
+
+    uint32_t                          DRAM_LOG_ADDR_H;
+    uint32_t                          DRAM_LOG_ADDR_L;
+    uint32_t                          DRAM_LOG_PHY_ADDR_H;
+    uint32_t                          DRAM_LOG_PHY_ADDR_L;
+    uint32_t                          DRAM_LOG_BUFF_SIZE;
+
+};
+
+struct SMU7_Fusion_GIODpmTable
+{
+
+    SMU7_Fusion_GIOLevel              GIOLevel                [SMU7_MAX_LEVELS_GIO];
+
+    SMU7_PIDController                GioPIDController;
+
+    uint32_t                          GIOLevelCount;
+
+    uint8_t                           Enable;
+    uint8_t                           GIOVoltageChangeEnable;
+    uint8_t                           GIOBootLevel;
+    uint8_t                           padding;
+    uint8_t                           padding1[2];
+    uint8_t                           TargetState;
+    uint8_t                           CurrenttState;
+    uint8_t                           ThrottleOnHtc;
+    uint8_t                           ThermThrottleStatus;
+    uint8_t                           ThermThrottleTempSelect;
+    uint8_t                           ThermThrottleEnable;
+    uint16_t                          TemperatureLimitHigh;
+    uint16_t                          TemperatureLimitLow;
+
+};
+
+typedef struct SMU7_Fusion_DpmTable SMU7_Fusion_DpmTable;
+typedef struct SMU7_Fusion_GIODpmTable SMU7_Fusion_GIODpmTable;
+
+#pragma pack(pop)
+
+#endif
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu8.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu8.h
new file mode 100644
index 000000000..d758d07b6
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu8.h
@@ -0,0 +1,72 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU8_H
+#define SMU8_H
+
+#pragma pack(push, 1)
+
+#define ENABLE_DEBUG_FEATURES
+
+struct SMU8_Firmware_Header {
+	uint32_t Version;
+	uint32_t ImageSize;
+	uint32_t CodeSize;
+	uint32_t HeaderSize;
+	uint32_t EntryPoint;
+	uint32_t Rtos;
+	uint32_t UcodeLoadStatus;
+	uint32_t DpmTable;
+	uint32_t FanTable;
+	uint32_t PmFuseTable;
+	uint32_t Globals;
+	uint32_t Reserved[20];
+	uint32_t Signature;
+};
+
+struct SMU8_MultimediaPowerLogData {
+	uint32_t avgTotalPower;
+	uint32_t avgGpuPower;
+	uint32_t avgUvdPower;
+	uint32_t avgVcePower;
+
+	uint32_t avgSclk;
+	uint32_t avgDclk;
+	uint32_t avgVclk;
+	uint32_t avgEclk;
+
+	uint32_t startTimeHi;
+	uint32_t startTimeLo;
+
+	uint32_t endTimeHi;
+	uint32_t endTimeLo;
+};
+
+#define SMU8_FIRMWARE_HEADER_LOCATION 0x1FF80
+#define SMU8_UNBCSR_START_ADDR 0xC0100000
+
+#define SMN_MP1_SRAM_START_ADDR 0x10000000
+
+#pragma pack(pop)
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h
new file mode 100644
index 000000000..5c9cc3c0b
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu8_fusion.h
@@ -0,0 +1,127 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU8_FUSION_H
+#define SMU8_FUSION_H
+
+#include "smu8.h"
+
+#pragma pack(push, 1)
+
+#define SMU8_MAX_CUS 2
+#define SMU8_PSMS_PER_CU 4
+#define SMU8_CACS_PER_CU 4
+
+struct SMU8_GfxCuPgScoreboard {
+    uint8_t Enabled;
+    uint8_t spare[3];
+};
+
+struct SMU8_Port80MonitorTable {
+	uint32_t MmioAddress;
+	uint32_t MemoryBaseHi;
+	uint32_t MemoryBaseLo;
+	uint16_t MemoryBufferSize;
+	uint16_t MemoryPosition;
+	uint16_t PollingInterval;
+	uint8_t  EnableCsrShadow;
+	uint8_t  EnableDramShadow;
+};
+
+/* Clock Table Definitions */
+#define NUM_SCLK_LEVELS     8
+#define NUM_LCLK_LEVELS     8
+#define NUM_UVD_LEVELS      8
+#define NUM_ECLK_LEVELS     8
+#define NUM_ACLK_LEVELS     8
+
+struct SMU8_Fusion_ClkLevel {
+	uint8_t		GnbVid;
+	uint8_t		GfxVid;
+	uint8_t		DfsDid;
+	uint8_t		DeepSleepDid;
+	uint32_t	DfsBypass;
+	uint32_t	Frequency;
+};
+
+struct SMU8_Fusion_SclkBreakdownTable {
+	struct SMU8_Fusion_ClkLevel ClkLevel[NUM_SCLK_LEVELS];
+	struct SMU8_Fusion_ClkLevel DpmOffLevel;
+	/* SMU8_Fusion_ClkLevel PwrOffLevel; */
+	uint32_t    SclkValidMask;
+	uint32_t    MaxSclkIndex;
+};
+
+struct SMU8_Fusion_LclkBreakdownTable {
+	struct SMU8_Fusion_ClkLevel ClkLevel[NUM_LCLK_LEVELS];
+	struct SMU8_Fusion_ClkLevel DpmOffLevel;
+    /* SMU8_Fusion_ClkLevel PwrOffLevel; */
+	uint32_t    LclkValidMask;
+	uint32_t    MaxLclkIndex;
+};
+
+struct SMU8_Fusion_EclkBreakdownTable {
+	struct SMU8_Fusion_ClkLevel ClkLevel[NUM_ECLK_LEVELS];
+	struct SMU8_Fusion_ClkLevel DpmOffLevel;
+	struct SMU8_Fusion_ClkLevel PwrOffLevel;
+	uint32_t    EclkValidMask;
+	uint32_t    MaxEclkIndex;
+};
+
+struct SMU8_Fusion_VclkBreakdownTable {
+	struct SMU8_Fusion_ClkLevel ClkLevel[NUM_UVD_LEVELS];
+	struct SMU8_Fusion_ClkLevel DpmOffLevel;
+	struct SMU8_Fusion_ClkLevel PwrOffLevel;
+	uint32_t    VclkValidMask;
+	uint32_t    MaxVclkIndex;
+};
+
+struct SMU8_Fusion_DclkBreakdownTable {
+	struct SMU8_Fusion_ClkLevel ClkLevel[NUM_UVD_LEVELS];
+	struct SMU8_Fusion_ClkLevel DpmOffLevel;
+	struct SMU8_Fusion_ClkLevel PwrOffLevel;
+	uint32_t    DclkValidMask;
+	uint32_t    MaxDclkIndex;
+};
+
+struct SMU8_Fusion_AclkBreakdownTable {
+	struct SMU8_Fusion_ClkLevel ClkLevel[NUM_ACLK_LEVELS];
+	struct SMU8_Fusion_ClkLevel DpmOffLevel;
+	struct SMU8_Fusion_ClkLevel PwrOffLevel;
+	uint32_t    AclkValidMask;
+	uint32_t    MaxAclkIndex;
+};
+
+
+struct SMU8_Fusion_ClkTable {
+	struct SMU8_Fusion_SclkBreakdownTable SclkBreakdownTable;
+	struct SMU8_Fusion_LclkBreakdownTable LclkBreakdownTable;
+	struct SMU8_Fusion_EclkBreakdownTable EclkBreakdownTable;
+	struct SMU8_Fusion_VclkBreakdownTable VclkBreakdownTable;
+	struct SMU8_Fusion_DclkBreakdownTable DclkBreakdownTable;
+	struct SMU8_Fusion_AclkBreakdownTable AclkBreakdownTable;
+};
+
+#pragma pack(pop)
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h
new file mode 100644
index 000000000..f8ba071f3
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_cz.h
@@ -0,0 +1,147 @@
+// CZ Ucode Loading Definitions
+#ifndef SMU_UCODE_XFER_CZ_H
+#define SMU_UCODE_XFER_CZ_H
+
+#define NUM_JOBLIST_ENTRIES      32
+
+#define TASK_TYPE_NO_ACTION      0
+#define TASK_TYPE_UCODE_LOAD     1
+#define TASK_TYPE_UCODE_SAVE     2
+#define TASK_TYPE_REG_LOAD       3
+#define TASK_TYPE_REG_SAVE       4
+#define TASK_TYPE_INITIALIZE     5
+
+#define TASK_ARG_REG_SMCIND      0
+#define TASK_ARG_REG_MMIO        1
+#define TASK_ARG_REG_FCH         2
+#define TASK_ARG_REG_UNB         3
+
+#define TASK_ARG_INIT_MM_PWR_LOG 0
+#define TASK_ARG_INIT_CLK_TABLE  1
+
+#define JOB_GFX_SAVE             0
+#define JOB_GFX_RESTORE          1
+#define JOB_FCH_SAVE             2
+#define JOB_FCH_RESTORE          3
+#define JOB_UNB_SAVE             4
+#define JOB_UNB_RESTORE          5
+#define JOB_GMC_SAVE             6
+#define JOB_GMC_RESTORE          7
+#define JOB_GNB_SAVE             8
+#define JOB_GNB_RESTORE          9
+
+#define IGNORE_JOB               0xff
+#define END_OF_TASK_LIST     (uint16_t)0xffff
+
+// Size of DRAM regions (in bytes) requested by SMU:
+#define SMU_DRAM_REQ_MM_PWR_LOG 48 
+
+#define UCODE_ID_SDMA0           0
+#define UCODE_ID_SDMA1           1
+#define UCODE_ID_CP_CE           2
+#define UCODE_ID_CP_PFP          3
+#define UCODE_ID_CP_ME           4
+#define UCODE_ID_CP_MEC_JT1      5
+#define UCODE_ID_CP_MEC_JT2      6
+#define UCODE_ID_GMCON_RENG      7
+#define UCODE_ID_RLC_G           8
+#define UCODE_ID_RLC_SCRATCH     9
+#define UCODE_ID_RLC_SRM_ARAM    10
+#define UCODE_ID_RLC_SRM_DRAM    11
+#define UCODE_ID_DMCU_ERAM       12
+#define UCODE_ID_DMCU_IRAM       13
+
+#define UCODE_ID_SDMA0_MASK           0x00000001       
+#define UCODE_ID_SDMA1_MASK           0x00000002        
+#define UCODE_ID_CP_CE_MASK           0x00000004      
+#define UCODE_ID_CP_PFP_MASK          0x00000008         
+#define UCODE_ID_CP_ME_MASK           0x00000010          
+#define UCODE_ID_CP_MEC_JT1_MASK      0x00000020             
+#define UCODE_ID_CP_MEC_JT2_MASK      0x00000040          
+#define UCODE_ID_GMCON_RENG_MASK      0x00000080            
+#define UCODE_ID_RLC_G_MASK           0x00000100           
+#define UCODE_ID_RLC_SCRATCH_MASK     0x00000200         
+#define UCODE_ID_RLC_SRM_ARAM_MASK    0x00000400                
+#define UCODE_ID_RLC_SRM_DRAM_MASK    0x00000800                 
+#define UCODE_ID_DMCU_ERAM_MASK       0x00001000             
+#define UCODE_ID_DMCU_IRAM_MASK       0x00002000              
+
+#define UCODE_ID_SDMA0_SIZE_BYTE           10368        
+#define UCODE_ID_SDMA1_SIZE_BYTE           10368          
+#define UCODE_ID_CP_CE_SIZE_BYTE           8576        
+#define UCODE_ID_CP_PFP_SIZE_BYTE          16768           
+#define UCODE_ID_CP_ME_SIZE_BYTE           16768            
+#define UCODE_ID_CP_MEC_JT1_SIZE_BYTE      384               
+#define UCODE_ID_CP_MEC_JT2_SIZE_BYTE      384            
+#define UCODE_ID_GMCON_RENG_SIZE_BYTE      4096              
+#define UCODE_ID_RLC_G_SIZE_BYTE           2048             
+#define UCODE_ID_RLC_SCRATCH_SIZE_BYTE     132           
+#define UCODE_ID_RLC_SRM_ARAM_SIZE_BYTE    8192                  
+#define UCODE_ID_RLC_SRM_DRAM_SIZE_BYTE    4096                   
+#define UCODE_ID_DMCU_ERAM_SIZE_BYTE       24576               
+#define UCODE_ID_DMCU_IRAM_SIZE_BYTE       1024                 
+
+#define NUM_UCODES               14
+
+typedef struct {
+	uint32_t high;
+	uint32_t low;
+} data_64_t;
+
+struct SMU_Task {
+    uint8_t type;
+    uint8_t arg;
+    uint16_t next;
+    data_64_t addr;
+    uint32_t size_bytes;
+};
+typedef struct SMU_Task SMU_Task;
+
+struct TOC {
+    uint8_t JobList[NUM_JOBLIST_ENTRIES];
+    SMU_Task tasks[1];
+};
+
+// META DATA COMMAND Definitions
+#define METADATA_CMD_MODE0         0x00000103 
+#define METADATA_CMD_MODE1         0x00000113 
+#define METADATA_CMD_MODE2         0x00000123 
+#define METADATA_CMD_MODE3         0x00000133
+#define METADATA_CMD_DELAY         0x00000203
+#define METADATA_CMD_CHNG_REGSPACE 0x00000303
+#define METADATA_PERFORM_ON_SAVE   0x00001000
+#define METADATA_PERFORM_ON_LOAD   0x00002000
+#define METADATA_CMD_ARG_MASK      0xFFFF0000
+#define METADATA_CMD_ARG_SHIFT     16
+
+// Simple register addr/data fields
+struct SMU_MetaData_Mode0 {
+    uint32_t register_address;
+    uint32_t register_data;
+};
+typedef struct SMU_MetaData_Mode0 SMU_MetaData_Mode0;
+
+// Register addr/data with mask
+struct SMU_MetaData_Mode1 {
+    uint32_t register_address;
+    uint32_t register_mask;
+    uint32_t register_data;
+};
+typedef struct SMU_MetaData_Mode1 SMU_MetaData_Mode1;
+
+struct SMU_MetaData_Mode2 {
+    uint32_t register_address;
+    uint32_t register_mask;
+    uint32_t target_value;
+};
+typedef struct SMU_MetaData_Mode2 SMU_MetaData_Mode2;
+
+// Always write data (even on a save operation)
+struct SMU_MetaData_Mode3 {
+    uint32_t register_address;
+    uint32_t register_mask;
+    uint32_t register_data;
+};
+typedef struct SMU_MetaData_Mode3 SMU_MetaData_Mode3;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_vi.h b/kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_vi.h
new file mode 100644
index 000000000..c24a81eeb
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/smu_ucode_xfer_vi.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef SMU_UCODE_XFER_VI_H
+#define SMU_UCODE_XFER_VI_H
+
+#define SMU_DRAMData_TOC_VERSION  1
+#define MAX_IH_REGISTER_COUNT     65535
+#define SMU_DIGEST_SIZE_BYTES     20
+#define SMU_FB_SIZE_BYTES         1048576
+#define SMU_MAX_ENTRIES           12
+
+#define UCODE_ID_SMU              0
+#define UCODE_ID_SDMA0            1
+#define UCODE_ID_SDMA1            2
+#define UCODE_ID_CP_CE            3
+#define UCODE_ID_CP_PFP           4
+#define UCODE_ID_CP_ME            5
+#define UCODE_ID_CP_MEC           6
+#define UCODE_ID_CP_MEC_JT1       7
+#define UCODE_ID_CP_MEC_JT2       8
+#define UCODE_ID_GMCON_RENG       9
+#define UCODE_ID_RLC_G            10
+#define UCODE_ID_IH_REG_RESTORE   11
+#define UCODE_ID_VBIOS            12
+#define UCODE_ID_MISC_METADATA    13
+#define UCODE_ID_RLC_SCRATCH      32
+#define UCODE_ID_RLC_SRM_ARAM     33
+#define UCODE_ID_RLC_SRM_DRAM     34
+#define UCODE_ID_MEC_STORAGE      35
+#define UCODE_ID_VBIOS_PARAMETERS 36
+#define UCODE_META_DATA           0xFF
+
+#define UCODE_ID_SMU_MASK             0x00000001
+#define UCODE_ID_SDMA0_MASK           0x00000002
+#define UCODE_ID_SDMA1_MASK           0x00000004
+#define UCODE_ID_CP_CE_MASK           0x00000008
+#define UCODE_ID_CP_PFP_MASK          0x00000010
+#define UCODE_ID_CP_ME_MASK           0x00000020
+#define UCODE_ID_CP_MEC_MASK          0x00000040
+#define UCODE_ID_CP_MEC_JT1_MASK      0x00000080
+#define UCODE_ID_CP_MEC_JT2_MASK      0x00000100
+#define UCODE_ID_GMCON_RENG_MASK      0x00000200
+#define UCODE_ID_RLC_G_MASK           0x00000400
+#define UCODE_ID_IH_REG_RESTORE_MASK  0x00000800
+#define UCODE_ID_VBIOS_MASK           0x00001000
+
+#define UCODE_FLAG_UNHALT_MASK   0x1
+
+struct SMU_Entry {
+#ifndef __BIG_ENDIAN
+	uint16_t id;
+	uint16_t version;
+	uint32_t image_addr_high;
+	uint32_t image_addr_low;
+	uint32_t meta_data_addr_high;
+	uint32_t meta_data_addr_low;
+	uint32_t data_size_byte;
+	uint16_t flags;
+	uint16_t num_register_entries;
+#else
+	uint16_t version;
+	uint16_t id;
+	uint32_t image_addr_high;
+	uint32_t image_addr_low;
+	uint32_t meta_data_addr_high;
+	uint32_t meta_data_addr_low;
+	uint32_t data_size_byte;
+	uint16_t num_register_entries;
+	uint16_t flags;
+#endif
+};
+
+struct SMU_DRAMData_TOC {
+	uint32_t structure_version;
+	uint32_t num_entries;
+	struct SMU_Entry entry[SMU_MAX_ENTRIES];
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c
new file mode 100644
index 000000000..63d6cb3c1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c
@@ -0,0 +1,181 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "tonga_smumgr.h"
+
+MODULE_FIRMWARE("amdgpu/tonga_smc.bin");
+
+static void tonga_dpm_set_funcs(struct amdgpu_device *adev);
+
+static int tonga_dpm_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	tonga_dpm_set_funcs(adev);
+
+	return 0;
+}
+
+static int tonga_dpm_init_microcode(struct amdgpu_device *adev)
+{
+	char fw_name[30] = "amdgpu/tonga_smc.bin";
+	int err;
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+
+out:
+	if (err) {
+		DRM_ERROR("Failed to load firmware \"%s\"", fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+}
+
+static int tonga_dpm_sw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	ret = tonga_dpm_init_microcode(adev);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int tonga_dpm_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int tonga_dpm_hw_init(void *handle)
+{
+	int ret;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+
+	/* smu init only needs to be called at startup, not resume.
+	 * It should be in sw_init, but requires the fw info gathered
+	 * in sw_init from other IP modules.
+	 */
+	ret = tonga_smu_init(adev);
+	if (ret) {
+		DRM_ERROR("SMU initialization failed\n");
+		goto fail;
+	}
+
+	ret = tonga_smu_start(adev);
+	if (ret) {
+		DRM_ERROR("SMU start failed\n");
+		goto fail;
+	}
+
+	mutex_unlock(&adev->pm.mutex);
+	return 0;
+
+fail:
+	adev->firmware.smu_load = false;
+	mutex_unlock(&adev->pm.mutex);
+	return -EINVAL;
+}
+
+static int tonga_dpm_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	mutex_lock(&adev->pm.mutex);
+	/* smu fini only needs to be called at teardown, not suspend.
+	 * It should be in sw_fini, but we put it here for symmetry
+	 * with smu init.
+	 */
+	tonga_smu_fini(adev);
+	mutex_unlock(&adev->pm.mutex);
+	return 0;
+}
+
+static int tonga_dpm_suspend(void *handle)
+{
+	return tonga_dpm_hw_fini(handle);
+}
+
+static int tonga_dpm_resume(void *handle)
+{
+	return tonga_dpm_hw_init(handle);
+}
+
+static int tonga_dpm_set_clockgating_state(void *handle,
+			enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int tonga_dpm_set_powergating_state(void *handle,
+			enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs tonga_dpm_ip_funcs = {
+	.early_init = tonga_dpm_early_init,
+	.late_init = NULL,
+	.sw_init = tonga_dpm_sw_init,
+	.sw_fini = tonga_dpm_sw_fini,
+	.hw_init = tonga_dpm_hw_init,
+	.hw_fini = tonga_dpm_hw_fini,
+	.suspend = tonga_dpm_suspend,
+	.resume = tonga_dpm_resume,
+	.is_idle = NULL,
+	.wait_for_idle = NULL,
+	.soft_reset = NULL,
+	.print_status = NULL,
+	.set_clockgating_state = tonga_dpm_set_clockgating_state,
+	.set_powergating_state = tonga_dpm_set_powergating_state,
+};
+
+static const struct amdgpu_dpm_funcs tonga_dpm_funcs = {
+	.get_temperature = NULL,
+	.pre_set_power_state = NULL,
+	.set_power_state = NULL,
+	.post_set_power_state = NULL,
+	.display_configuration_changed = NULL,
+	.get_sclk = NULL,
+	.get_mclk = NULL,
+	.print_power_state = NULL,
+	.debugfs_print_current_performance_level = NULL,
+	.force_performance_level = NULL,
+	.vblank_too_short = NULL,
+	.powergate_uvd = NULL,
+};
+
+static void tonga_dpm_set_funcs(struct amdgpu_device *adev)
+{
+	if (NULL == adev->pm.funcs)
+		adev->pm.funcs = &tonga_dpm_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
new file mode 100644
index 000000000..743c37283
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
@@ -0,0 +1,473 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+#include "vid.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "bif/bif_5_1_d.h"
+#include "bif/bif_5_1_sh_mask.h"
+
+/*
+ * Interrupts
+ * Starting with r6xx, interrupts are handled via a ring buffer.
+ * Ring buffers are areas of GPU accessible memory that the GPU
+ * writes interrupt vectors into and the host reads vectors out of.
+ * There is a rptr (read pointer) that determines where the
+ * host is currently reading, and a wptr (write pointer)
+ * which determines where the GPU has written.  When the
+ * pointers are equal, the ring is idle.  When the GPU
+ * writes vectors to the ring buffer, it increments the
+ * wptr.  When there is an interrupt, the host then starts
+ * fetching commands and processing them until the pointers are
+ * equal again at which point it updates the rptr.
+ */
+
+static void tonga_ih_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * tonga_ih_enable_interrupts - Enable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Enable the interrupt ring buffer (VI).
+ */
+static void tonga_ih_enable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	adev->irq.ih.enabled = true;
+}
+
+/**
+ * tonga_ih_disable_interrupts - Disable the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable the interrupt ring buffer (VI).
+ */
+static void tonga_ih_disable_interrupts(struct amdgpu_device *adev)
+{
+	u32 ih_rb_cntl = RREG32(mmIH_RB_CNTL);
+
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 0);
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+	adev->irq.ih.enabled = false;
+	adev->irq.ih.rptr = 0;
+}
+
+/**
+ * tonga_ih_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it (VI).
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int tonga_ih_irq_init(struct amdgpu_device *adev)
+{
+	int ret = 0;
+	int rb_bufsz;
+	u32 interrupt_cntl, ih_rb_cntl, ih_doorbell_rtpr;
+	u64 wptr_off;
+
+	/* disable irqs */
+	tonga_ih_disable_interrupts(adev);
+
+	/* setup interrupt control */
+	WREG32(mmINTERRUPT_CNTL2, adev->dummy_page.addr >> 8);
+	interrupt_cntl = RREG32(mmINTERRUPT_CNTL);
+	/* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
+	 * INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
+	 */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_DUMMY_RD_OVERRIDE, 0);
+	/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
+	interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL, IH_REQ_NONSNOOP_EN, 0);
+	WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
+
+	/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+	if (adev->irq.ih.use_bus_addr)
+		WREG32(mmIH_RB_BASE, adev->irq.ih.rb_dma_addr >> 8);
+	else
+		WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
+
+	rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
+	ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+	/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
+	ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+	if (adev->irq.msi_enabled)
+		ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM, 1);
+
+	WREG32(mmIH_RB_CNTL, ih_rb_cntl);
+
+	/* set the writeback address whether it's enabled or not */
+	if (adev->irq.ih.use_bus_addr)
+		wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
+	else
+		wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
+	WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
+	WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
+
+	/* set rptr, wptr to 0 */
+	WREG32(mmIH_RB_RPTR, 0);
+	WREG32(mmIH_RB_WPTR, 0);
+
+	ih_doorbell_rtpr = RREG32(mmIH_DOORBELL_RPTR);
+	if (adev->irq.ih.use_doorbell) {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 OFFSET, adev->irq.ih.doorbell_index);
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 ENABLE, 1);
+	} else {
+		ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
+						 ENABLE, 0);
+	}
+	WREG32(mmIH_DOORBELL_RPTR, ih_doorbell_rtpr);
+
+	pci_set_master(adev->pdev);
+
+	/* enable interrupts */
+	tonga_ih_enable_interrupts(adev);
+
+	return ret;
+}
+
+/**
+ * tonga_ih_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw (VI).
+ */
+static void tonga_ih_irq_disable(struct amdgpu_device *adev)
+{
+	tonga_ih_disable_interrupts(adev);
+
+	/* Wait and acknowledge irq */
+	mdelay(1);
+}
+
+/**
+ * tonga_ih_get_wptr - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer (VI).  Also check for
+ * ring buffer overflow and deal with it.
+ * Used by cz_irq_process(VI).
+ * Returns the value of the wptr.
+ */
+static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
+{
+	u32 wptr, tmp;
+
+	if (adev->irq.ih.use_bus_addr)
+		wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
+	else
+		wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
+
+	if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
+		wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+		/* When a ring buffer overflow happen start parsing interrupt
+		 * from the last not overwritten vector (wptr + 16). Hopefully
+		 * this should allow us to catchup.
+		 */
+		dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
+			wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
+		adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
+		tmp = RREG32(mmIH_RB_CNTL);
+		tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+		WREG32(mmIH_RB_CNTL, tmp);
+	}
+	return (wptr & adev->irq.ih.ptr_mask);
+}
+
+/**
+ * tonga_ih_decode_iv - decode an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Decodes the interrupt vector at the current rptr
+ * position and also advance the position.
+ */
+static void tonga_ih_decode_iv(struct amdgpu_device *adev,
+				 struct amdgpu_iv_entry *entry)
+{
+	/* wptr/rptr are in bytes! */
+	u32 ring_index = adev->irq.ih.rptr >> 2;
+	uint32_t dw[4];
+
+	dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
+	dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
+	dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
+	dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
+
+	entry->src_id = dw[0] & 0xff;
+	entry->src_data = dw[1] & 0xfffffff;
+	entry->ring_id = dw[2] & 0xff;
+	entry->vm_id = (dw[2] >> 8) & 0xff;
+	entry->pas_id = (dw[2] >> 16) & 0xffff;
+
+	/* wptr/rptr are in bytes! */
+	adev->irq.ih.rptr += 16;
+}
+
+/**
+ * tonga_ih_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set the IH ring buffer rptr.
+ */
+static void tonga_ih_set_rptr(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih.use_doorbell) {
+		/* XXX check if swapping is necessary on BE */
+		if (adev->irq.ih.use_bus_addr)
+			adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
+		else
+			adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
+		WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
+	} else {
+		WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
+	}
+}
+
+static int tonga_ih_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	tonga_ih_set_interrupt_funcs(adev);
+	return 0;
+}
+
+static int tonga_ih_sw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_ih_ring_init(adev, 4 * 1024, true);
+	if (r)
+		return r;
+
+	adev->irq.ih.use_doorbell = true;
+	adev->irq.ih.doorbell_index = AMDGPU_DOORBELL_IH;
+
+	r = amdgpu_irq_init(adev);
+
+	return r;
+}
+
+static int tonga_ih_sw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	amdgpu_irq_fini(adev);
+	amdgpu_ih_ring_fini(adev);
+
+	return 0;
+}
+
+static int tonga_ih_hw_init(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = tonga_ih_irq_init(adev);
+	if (r)
+		return r;
+
+	return 0;
+}
+
+static int tonga_ih_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	tonga_ih_irq_disable(adev);
+
+	return 0;
+}
+
+static int tonga_ih_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return tonga_ih_hw_fini(adev);
+}
+
+static int tonga_ih_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return tonga_ih_hw_init(adev);
+}
+
+static bool tonga_ih_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+		return false;
+
+	return true;
+}
+
+static int tonga_ih_wait_for_idle(void *handle)
+{
+	unsigned i;
+	u32 tmp;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		/* read MC_STATUS */
+		tmp = RREG32(mmSRBM_STATUS);
+		if (!REG_GET_FIELD(tmp, SRBM_STATUS, IH_BUSY))
+			return 0;
+		udelay(1);
+	}
+	return -ETIMEDOUT;
+}
+
+static void tonga_ih_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "TONGA IH registers\n");
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+	dev_info(adev->dev, "  INTERRUPT_CNTL=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL));
+	dev_info(adev->dev, "  INTERRUPT_CNTL2=0x%08X\n",
+		 RREG32(mmINTERRUPT_CNTL2));
+	dev_info(adev->dev, "  IH_CNTL=0x%08X\n",
+		 RREG32(mmIH_CNTL));
+	dev_info(adev->dev, "  IH_RB_CNTL=0x%08X\n",
+		 RREG32(mmIH_RB_CNTL));
+	dev_info(adev->dev, "  IH_RB_BASE=0x%08X\n",
+		 RREG32(mmIH_RB_BASE));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_LO=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_LO));
+	dev_info(adev->dev, "  IH_RB_WPTR_ADDR_HI=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR_ADDR_HI));
+	dev_info(adev->dev, "  IH_RB_RPTR=0x%08X\n",
+		 RREG32(mmIH_RB_RPTR));
+	dev_info(adev->dev, "  IH_RB_WPTR=0x%08X\n",
+		 RREG32(mmIH_RB_WPTR));
+}
+
+static int tonga_ih_soft_reset(void *handle)
+{
+	u32 srbm_soft_reset = 0;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET,
+						SOFT_RESET_IH, 1);
+
+	if (srbm_soft_reset) {
+		tonga_ih_print_status(adev);
+
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		/* Wait a little for things to settle down */
+		udelay(50);
+
+		tonga_ih_print_status(adev);
+	}
+
+	return 0;
+}
+
+static int tonga_ih_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int tonga_ih_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs tonga_ih_ip_funcs = {
+	.early_init = tonga_ih_early_init,
+	.late_init = NULL,
+	.sw_init = tonga_ih_sw_init,
+	.sw_fini = tonga_ih_sw_fini,
+	.hw_init = tonga_ih_hw_init,
+	.hw_fini = tonga_ih_hw_fini,
+	.suspend = tonga_ih_suspend,
+	.resume = tonga_ih_resume,
+	.is_idle = tonga_ih_is_idle,
+	.wait_for_idle = tonga_ih_wait_for_idle,
+	.soft_reset = tonga_ih_soft_reset,
+	.print_status = tonga_ih_print_status,
+	.set_clockgating_state = tonga_ih_set_clockgating_state,
+	.set_powergating_state = tonga_ih_set_powergating_state,
+};
+
+static const struct amdgpu_ih_funcs tonga_ih_funcs = {
+	.get_wptr = tonga_ih_get_wptr,
+	.decode_iv = tonga_ih_decode_iv,
+	.set_rptr = tonga_ih_set_rptr
+};
+
+static void tonga_ih_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+	if (adev->irq.ih_funcs == NULL)
+		adev->irq.ih_funcs = &tonga_ih_funcs;
+}
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.h b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.h
new file mode 100644
index 000000000..7392d70fa
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ih.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __TONGA_IH_H__
+#define __TONGA_IH_H__
+
+extern const struct amd_ip_funcs tonga_ih_ip_funcs;
+
+#endif /* __CZ_IH_H__ */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h
new file mode 100644
index 000000000..811781f69
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_ppsmc.h
@@ -0,0 +1,198 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef TONGA_PP_SMC_H
+#define TONGA_PP_SMC_H
+
+#pragma pack(push, 1)
+
+#define PPSMC_SWSTATE_FLAG_DC                           0x01
+#define PPSMC_SWSTATE_FLAG_UVD                          0x02
+#define PPSMC_SWSTATE_FLAG_VCE                          0x04
+#define PPSMC_SWSTATE_FLAG_PCIE_X1                      0x08
+
+#define PPSMC_THERMAL_PROTECT_TYPE_INTERNAL             0x00
+#define PPSMC_THERMAL_PROTECT_TYPE_EXTERNAL             0x01
+#define PPSMC_THERMAL_PROTECT_TYPE_NONE                 0xff 
+
+#define PPSMC_SYSTEMFLAG_GPIO_DC                        0x01 
+#define PPSMC_SYSTEMFLAG_STEPVDDC                       0x02 
+#define PPSMC_SYSTEMFLAG_GDDR5                          0x04 
+
+#define PPSMC_SYSTEMFLAG_DISABLE_BABYSTEP               0x08 
+
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT                  0x10 
+#define PPSMC_SYSTEMFLAG_REGULATOR_HOT_ANALOG           0x20 
+#define PPSMC_SYSTEMFLAG_12CHANNEL                      0x40
+
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_MASK              0x07
+#define PPSMC_EXTRAFLAGS_AC2DC_DONT_WAIT_FOR_VBLANK     0x08
+
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTODPMLOWSTATE   0x00
+#define PPSMC_EXTRAFLAGS_AC2DC_ACTION_GOTOINITIALSTATE  0x01
+
+#define PPSMC_EXTRAFLAGS_AC2DC_GPIO5_POLARITY_HIGH      0x10
+#define PPSMC_EXTRAFLAGS_DRIVER_TO_GPIO17               0x20
+#define PPSMC_EXTRAFLAGS_PCC_TO_GPIO17                  0x40
+
+#define PPSMC_DPM2FLAGS_TDPCLMP                         0x01 
+#define PPSMC_DPM2FLAGS_PWRSHFT                         0x02 
+#define PPSMC_DPM2FLAGS_OCP                             0x04 
+
+#define PPSMC_DISPLAY_WATERMARK_LOW                     0
+#define PPSMC_DISPLAY_WATERMARK_HIGH                    1
+
+#define PPSMC_STATEFLAG_AUTO_PULSE_SKIP    		0x01
+#define PPSMC_STATEFLAG_POWERBOOST         		0x02
+#define PPSMC_STATEFLAG_PSKIP_ON_TDP_FAULT 		0x04
+#define PPSMC_STATEFLAG_POWERSHIFT         		0x08
+#define PPSMC_STATEFLAG_SLOW_READ_MARGIN   		0x10
+#define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 		0x20
+#define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS   		0x40
+
+#define FDO_MODE_HARDWARE 0
+#define FDO_MODE_PIECE_WISE_LINEAR 1
+
+enum FAN_CONTROL {
+	FAN_CONTROL_FUZZY,
+	FAN_CONTROL_TABLE
+};
+
+#define PPSMC_Result_OK             			((uint16_t)0x01)
+#define PPSMC_Result_NoMore         			((uint16_t)0x02)
+#define PPSMC_Result_NotNow         			((uint16_t)0x03)
+#define PPSMC_Result_Failed         			((uint16_t)0xFF) 
+#define PPSMC_Result_UnknownCmd     			((uint16_t)0xFE) 
+#define PPSMC_Result_UnknownVT      			((uint16_t)0xFD) 
+
+typedef uint16_t PPSMC_Result;
+
+#define PPSMC_isERROR(x) ((uint16_t)0x80 & (x))
+
+#define PPSMC_MSG_Halt                      		((uint16_t)0x10)
+#define PPSMC_MSG_Resume                    		((uint16_t)0x11)
+#define PPSMC_MSG_EnableDPMLevel            		((uint16_t)0x12)
+#define PPSMC_MSG_ZeroLevelsDisabled        		((uint16_t)0x13)
+#define PPSMC_MSG_OneLevelsDisabled         		((uint16_t)0x14)
+#define PPSMC_MSG_TwoLevelsDisabled         		((uint16_t)0x15)
+#define PPSMC_MSG_EnableThermalInterrupt    		((uint16_t)0x16)
+#define PPSMC_MSG_RunningOnAC               		((uint16_t)0x17)
+#define PPSMC_MSG_LevelUp                   		((uint16_t)0x18)
+#define PPSMC_MSG_LevelDown                 		((uint16_t)0x19)
+#define PPSMC_MSG_ResetDPMCounters          		((uint16_t)0x1a)
+#define PPSMC_MSG_SwitchToSwState           		((uint16_t)0x20)
+#define PPSMC_MSG_SwitchToSwStateLast       		((uint16_t)0x3f)
+#define PPSMC_MSG_SwitchToInitialState      		((uint16_t)0x40)
+#define PPSMC_MSG_NoForcedLevel             		((uint16_t)0x41)
+#define PPSMC_MSG_ForceHigh                 		((uint16_t)0x42)
+#define PPSMC_MSG_ForceMediumOrHigh         		((uint16_t)0x43)
+#define PPSMC_MSG_SwitchToMinimumPower      		((uint16_t)0x51)
+#define PPSMC_MSG_ResumeFromMinimumPower    		((uint16_t)0x52)
+#define PPSMC_MSG_EnableCac                 		((uint16_t)0x53)
+#define PPSMC_MSG_DisableCac                		((uint16_t)0x54)
+#define PPSMC_DPMStateHistoryStart          		((uint16_t)0x55)
+#define PPSMC_DPMStateHistoryStop           		((uint16_t)0x56)
+#define PPSMC_CACHistoryStart               		((uint16_t)0x57)
+#define PPSMC_CACHistoryStop                		((uint16_t)0x58)
+#define PPSMC_TDPClampingActive             		((uint16_t)0x59)
+#define PPSMC_TDPClampingInactive           		((uint16_t)0x5A)
+#define PPSMC_StartFanControl               		((uint16_t)0x5B)
+#define PPSMC_StopFanControl                		((uint16_t)0x5C)
+#define PPSMC_NoDisplay                     		((uint16_t)0x5D)
+#define PPSMC_HasDisplay                    		((uint16_t)0x5E)
+#define PPSMC_MSG_UVDPowerOFF               		((uint16_t)0x60)
+#define PPSMC_MSG_UVDPowerON                		((uint16_t)0x61)
+#define PPSMC_MSG_EnableULV                 		((uint16_t)0x62)
+#define PPSMC_MSG_DisableULV                		((uint16_t)0x63)
+#define PPSMC_MSG_EnterULV                  		((uint16_t)0x64)
+#define PPSMC_MSG_ExitULV                   		((uint16_t)0x65)
+#define PPSMC_PowerShiftActive              		((uint16_t)0x6A)
+#define PPSMC_PowerShiftInactive            		((uint16_t)0x6B)
+#define PPSMC_OCPActive                     		((uint16_t)0x6C)
+#define PPSMC_OCPInactive                   		((uint16_t)0x6D)
+#define PPSMC_CACLongTermAvgEnable          		((uint16_t)0x6E)
+#define PPSMC_CACLongTermAvgDisable         		((uint16_t)0x6F)
+#define PPSMC_MSG_InferredStateSweep_Start  		((uint16_t)0x70)
+#define PPSMC_MSG_InferredStateSweep_Stop   		((uint16_t)0x71)
+#define PPSMC_MSG_SwitchToLowestInfState    		((uint16_t)0x72)
+#define PPSMC_MSG_SwitchToNonInfState       		((uint16_t)0x73)
+#define PPSMC_MSG_AllStateSweep_Start       		((uint16_t)0x74)
+#define PPSMC_MSG_AllStateSweep_Stop        		((uint16_t)0x75)
+#define PPSMC_MSG_SwitchNextLowerInfState   		((uint16_t)0x76)
+#define PPSMC_MSG_SwitchNextHigherInfState  		((uint16_t)0x77)
+#define PPSMC_MSG_MclkRetrainingTest        		((uint16_t)0x78)
+#define PPSMC_MSG_ForceTDPClamping          		((uint16_t)0x79)
+#define PPSMC_MSG_CollectCAC_PowerCorreln   		((uint16_t)0x7A)
+#define PPSMC_MSG_CollectCAC_WeightCalib    		((uint16_t)0x7B)
+#define PPSMC_MSG_CollectCAC_SQonly         		((uint16_t)0x7C)
+#define PPSMC_MSG_CollectCAC_TemperaturePwr 		((uint16_t)0x7D)
+#define PPSMC_MSG_ExtremitiesTest_Start     		((uint16_t)0x7E)
+#define PPSMC_MSG_ExtremitiesTest_Stop      		((uint16_t)0x7F)
+#define PPSMC_FlushDataCache                		((uint16_t)0x80)
+#define PPSMC_FlushInstrCache               		((uint16_t)0x81)
+#define PPSMC_MSG_SetEnabledLevels          		((uint16_t)0x82)
+#define PPSMC_MSG_SetForcedLevels           		((uint16_t)0x83)
+#define PPSMC_MSG_ResetToDefaults           		((uint16_t)0x84)
+#define PPSMC_MSG_SetForcedLevelsAndJump    		((uint16_t)0x85)
+#define PPSMC_MSG_SetCACHistoryMode         		((uint16_t)0x86)
+#define PPSMC_MSG_EnableDTE                 		((uint16_t)0x87)
+#define PPSMC_MSG_DisableDTE                		((uint16_t)0x88)
+#define PPSMC_MSG_SmcSpaceSetAddress        		((uint16_t)0x89)
+#define PPSMC_MSG_SmcSpaceWriteDWordInc     		((uint16_t)0x8A)
+#define PPSMC_MSG_SmcSpaceWriteWordInc      		((uint16_t)0x8B)
+#define PPSMC_MSG_SmcSpaceWriteByteInc      		((uint16_t)0x8C)
+#define PPSMC_MSG_ChangeNearTDPLimit        		((uint16_t)0x90)
+#define PPSMC_MSG_ChangeSafePowerLimit      		((uint16_t)0x91)
+#define PPSMC_MSG_DPMStateSweepStart        		((uint16_t)0x92)
+#define PPSMC_MSG_DPMStateSweepStop         		((uint16_t)0x93)
+#define PPSMC_MSG_OVRDDisableSCLKDS         		((uint16_t)0x94)
+#define PPSMC_MSG_CancelDisableOVRDSCLKDS   		((uint16_t)0x95)
+#define PPSMC_MSG_ThrottleOVRDSCLKDS        		((uint16_t)0x96)
+#define PPSMC_MSG_CancelThrottleOVRDSCLKDS  		((uint16_t)0x97)
+#define PPSMC_MSG_GPIO17                    		((uint16_t)0x98)
+#define PPSMC_MSG_API_SetSvi2Volt_Vddc      		((uint16_t)0x99)
+#define PPSMC_MSG_API_SetSvi2Volt_Vddci     		((uint16_t)0x9A)
+#define PPSMC_MSG_API_SetSvi2Volt_Mvdd      		((uint16_t)0x9B)
+#define PPSMC_MSG_API_GetSvi2Volt_Vddc      		((uint16_t)0x9C)
+#define PPSMC_MSG_API_GetSvi2Volt_Vddci     		((uint16_t)0x9D)
+#define PPSMC_MSG_API_GetSvi2Volt_Mvdd      		((uint16_t)0x9E)
+
+#define PPSMC_MSG_BREAK                     		((uint16_t)0xF8)
+
+#define PPSMC_MSG_Test                      		((uint16_t)0x100)
+#define PPSMC_MSG_DRV_DRAM_ADDR_HI            		((uint16_t)0x250)
+#define PPSMC_MSG_DRV_DRAM_ADDR_LO            		((uint16_t)0x251)
+#define PPSMC_MSG_SMU_DRAM_ADDR_HI            		((uint16_t)0x252)
+#define PPSMC_MSG_SMU_DRAM_ADDR_LO            		((uint16_t)0x253)
+#define PPSMC_MSG_LoadUcodes                  		((uint16_t)0x254)
+
+typedef uint16_t PPSMC_Msg;
+
+#define PPSMC_EVENT_STATUS_THERMAL          		0x00000001
+#define PPSMC_EVENT_STATUS_REGULATORHOT     		0x00000002
+#define PPSMC_EVENT_STATUS_DC               		0x00000004
+#define PPSMC_EVENT_STATUS_GPIO17           		0x00000008
+
+#pragma pack(pop)
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_sdma_pkt_open.h b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_sdma_pkt_open.h
new file mode 100644
index 000000000..e5ebd0842
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_sdma_pkt_open.h
@@ -0,0 +1,2245 @@
+/*
+ * Copyright (C) 2014  Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __TONGA_SDMA_PKT_OPEN_H_
+#define __TONGA_SDMA_PKT_OPEN_H_
+
+#define SDMA_OP_NOP  0
+#define SDMA_OP_COPY  1
+#define SDMA_OP_WRITE  2
+#define SDMA_OP_INDIRECT  4
+#define SDMA_OP_FENCE  5
+#define SDMA_OP_TRAP  6
+#define SDMA_OP_SEM  7
+#define SDMA_OP_POLL_REGMEM  8
+#define SDMA_OP_COND_EXE  9
+#define SDMA_OP_ATOMIC  10
+#define SDMA_OP_CONST_FILL  11
+#define SDMA_OP_GEN_PTEPDE  12
+#define SDMA_OP_TIMESTAMP  13
+#define SDMA_OP_SRBM_WRITE  14
+#define SDMA_OP_PRE_EXE  15
+#define SDMA_SUBOP_TIMESTAMP_SET  0
+#define SDMA_SUBOP_TIMESTAMP_GET  1
+#define SDMA_SUBOP_TIMESTAMP_GET_GLOBAL  2
+#define SDMA_SUBOP_COPY_LINEAR  0
+#define SDMA_SUBOP_COPY_LINEAR_SUB_WIND  4
+#define SDMA_SUBOP_COPY_TILED  1
+#define SDMA_SUBOP_COPY_TILED_SUB_WIND  5
+#define SDMA_SUBOP_COPY_T2T_SUB_WIND  6
+#define SDMA_SUBOP_COPY_SOA  3
+#define SDMA_SUBOP_WRITE_LINEAR  0
+#define SDMA_SUBOP_WRITE_TILED  1
+
+/*define for op field*/
+#define SDMA_PKT_HEADER_op_offset 0
+#define SDMA_PKT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_op_shift  0
+#define SDMA_PKT_HEADER_OP(x) (((x) & SDMA_PKT_HEADER_op_mask) << SDMA_PKT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_HEADER_sub_op_offset 0
+#define SDMA_PKT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_HEADER_sub_op_shift  8
+#define SDMA_PKT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_HEADER_sub_op_mask) << SDMA_PKT_HEADER_sub_op_shift)
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_offset 5
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_offset 6
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_BROADCAST_LINEAR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_sub_op_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_HEADER_broadcast_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_offset 1
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_COUNT_count_shift)
+
+/*define for PARAMETER word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift  8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift  14
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST2_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst2_ha_shift)
+
+/*define for dst1_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift  16
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_sw_shift)
+
+/*define for dst1_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift  22
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_DST1_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_dst1_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift  24
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_SW(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_offset 2
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift  30
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_SRC_HA(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_PARAMETER_src_ha_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_offset 3
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_offset 4
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DST1_ADDR_LO word*/
+/*define for dst1_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_offset 5
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_DST1_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_LO_dst1_addr_31_0_shift)
+
+/*define for DST1_ADDR_HI word*/
+/*define for dst1_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_offset 6
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_DST1_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST1_ADDR_HI_dst1_addr_63_32_shift)
+
+/*define for DST2_ADDR_LO word*/
+/*define for dst2_addr_31_0 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_offset 7
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_DST2_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_LO_dst2_addr_31_0_shift)
+
+/*define for DST2_ADDR_HI word*/
+/*define for dst2_addr_63_32 field*/
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_offset 8
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift  0
+#define SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_DST2_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_mask) << SDMA_PKT_COPY_BROADCAST_LINEAR_DST2_ADDR_HI_dst2_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_LINEAR_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_sub_op_shift)
+
+/*define for elementsize field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_offset 0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask   0x00000007
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift  29
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_ELEMENTSIZE(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_HEADER_elementsize_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_z_shift)
+
+/*define for src_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_offset 4
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_SRC_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_4_src_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_5_src_slice_pitch_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_offset 6
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_offset 7
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_8 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_offset 8
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_DST_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_8_dst_y_shift)
+
+/*define for DW_9 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_z_shift)
+
+/*define for dst_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_offset 9
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_DST_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_9_dst_pitch_shift)
+
+/*define for DW_10 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_offset 10
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_10_dst_slice_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_X(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_offset 11
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_RECT_Y(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_11_rect_y_shift)
+
+/*define for DW_12 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift  0
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_RECT_Z(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift  16
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_sw_shift)
+
+/*define for dst_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift  22
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_DST_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_dst_ha_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift  24
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_SW(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_sw_shift)
+
+/*define for src_ha field*/
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_offset 12
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask   0x00000001
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift  30
+#define SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_SRC_HA(x) (((x) & SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_mask) << SDMA_PKT_COPY_LINEAR_SUBWIN_DW_12_src_ha_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_op_mask) << SDMA_PKT_COPY_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_TILED_DW_3_height_offset 3
+#define SDMA_PKT_COPY_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_3_height_shift  16
+#define SDMA_PKT_COPY_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_COPY_TILED_DW_3_height_mask) << SDMA_PKT_COPY_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_COPY_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_element_size_mask) << SDMA_PKT_COPY_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_array_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mit_mode_mask) << SDMA_PKT_COPY_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_w_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_bank_h_mask) << SDMA_PKT_COPY_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_num_bank_mask) << SDMA_PKT_COPY_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_COPY_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_DW_5_pipe_config_mask) << SDMA_PKT_COPY_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_TILED_DW_6_x_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_x_shift  0
+#define SDMA_PKT_COPY_TILED_DW_6_X(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_x_mask) << SDMA_PKT_COPY_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_TILED_DW_6_y_offset 6
+#define SDMA_PKT_COPY_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_DW_6_y_shift  16
+#define SDMA_PKT_COPY_TILED_DW_6_Y(x) (((x) & SDMA_PKT_COPY_TILED_DW_6_y_mask) << SDMA_PKT_COPY_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_TILED_DW_7_z_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_DW_7_z_shift  0
+#define SDMA_PKT_COPY_TILED_DW_7_Z(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_z_mask) << SDMA_PKT_COPY_TILED_DW_7_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_DW_7_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_linear_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_offset 7
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_DW_7_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_DW_7_tile_sw_mask) << SDMA_PKT_COPY_TILED_DW_7_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_offset 8
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_offset 9
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_TILED_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_TILED_COUNT_count_offset 11
+#define SDMA_PKT_COPY_TILED_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_TILED_COUNT_count_shift  0
+#define SDMA_PKT_COPY_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_TILED_COUNT_count_mask) << SDMA_PKT_COPY_TILED_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_L2T_BROADCAST packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_sub_op_shift)
+
+/*define for videocopy field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_VIDEOCOPY(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_videocopy_shift)
+
+/*define for broadcast field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_offset 0
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift  27
+#define SDMA_PKT_COPY_L2T_BROADCAST_HEADER_BROADCAST(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_mask) << SDMA_PKT_COPY_L2T_BROADCAST_HEADER_broadcast_shift)
+
+/*define for TILED_ADDR_LO_0 word*/
+/*define for tiled_addr0_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_offset 1
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_TILED_ADDR0_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_0_tiled_addr0_31_0_shift)
+
+/*define for TILED_ADDR_HI_0 word*/
+/*define for tiled_addr0_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_offset 2
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_TILED_ADDR0_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_0_tiled_addr0_63_32_shift)
+
+/*define for TILED_ADDR_LO_1 word*/
+/*define for tiled_addr1_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_offset 3
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_TILED_ADDR1_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_LO_1_tiled_addr1_31_0_shift)
+
+/*define for TILED_ADDR_HI_1 word*/
+/*define for tiled_addr1_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_offset 4
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_TILED_ADDR1_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_TILED_ADDR_HI_1_tiled_addr1_63_32_shift)
+
+/*define for DW_5 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_offset 5
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_5_HEIGHT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_5_height_shift)
+
+/*define for DW_6 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_offset 6
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_6_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_6_slice_pitch_shift)
+
+/*define for DW_7 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift  3
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MIT_MODE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift  15
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_W(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift  18
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_BANK_H(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift  21
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_NUM_BANK(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_offset 7
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift  26
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_7_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_7_pipe_config_shift)
+
+/*define for DW_8 word*/
+/*define for x field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_X(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_offset 8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_8_Y(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_8_y_shift)
+
+/*define for DW_9 word*/
+/*define for z field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_offset 9
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask   0x00000FFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_9_Z(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_9_z_shift)
+
+/*define for DW_10 word*/
+/*define for dst2_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift  8
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_sw_shift)
+
+/*define for dst2_ha field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask   0x00000001
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift  14
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_DST2_HA(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_dst2_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift  16
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_offset 10
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift  24
+#define SDMA_PKT_COPY_L2T_BROADCAST_DW_10_TILE_SW(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_mask) << SDMA_PKT_COPY_L2T_BROADCAST_DW_10_tile_sw_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_offset 11
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_offset 12
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for LINEAR_PITCH word*/
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_offset 13
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask   0x0007FFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_mask) << SDMA_PKT_COPY_L2T_BROADCAST_LINEAR_PITCH_linear_pitch_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_offset 14
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask   0x000FFFFF
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift  0
+#define SDMA_PKT_COPY_L2T_BROADCAST_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_mask) << SDMA_PKT_COPY_L2T_BROADCAST_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_T2T packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_op_shift  0
+#define SDMA_PKT_COPY_T2T_HEADER_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_op_mask) << SDMA_PKT_COPY_T2T_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_T2T_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_T2T_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_T2T_HEADER_sub_op_mask) << SDMA_PKT_COPY_T2T_HEADER_sub_op_shift)
+
+/*define for SRC_ADDR_LO word*/
+/*define for src_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_offset 1
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_LO_SRC_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_LO_src_addr_31_0_shift)
+
+/*define for SRC_ADDR_HI word*/
+/*define for src_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_offset 2
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_SRC_ADDR_HI_SRC_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_mask) << SDMA_PKT_COPY_T2T_SRC_ADDR_HI_src_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for src_x field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_x_mask) << SDMA_PKT_COPY_T2T_DW_3_src_x_shift)
+
+/*define for src_y field*/
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_offset 3
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_3_src_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_3_SRC_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_3_src_y_mask) << SDMA_PKT_COPY_T2T_DW_3_src_y_shift)
+
+/*define for DW_4 word*/
+/*define for src_z field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_4_src_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_z_mask) << SDMA_PKT_COPY_T2T_DW_4_src_z_shift)
+
+/*define for src_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_4_SRC_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_4_src_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for src_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_offset 5
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_5_SRC_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_5_src_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for src_element_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift  0
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_element_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_element_size_shift)
+
+/*define for src_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_array_mode_shift)
+
+/*define for src_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mit_mode_shift)
+
+/*define for src_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_6_src_tilesplit_size_shift)
+
+/*define for src_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_w_shift)
+
+/*define for src_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_6_src_bank_h_shift)
+
+/*define for src_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_6_src_num_bank_shift)
+
+/*define for src_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_6_src_mat_aspt_shift)
+
+/*define for src_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_offset 6
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_6_SRC_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_6_src_pipe_config_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_offset 7
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_offset 8
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_COPY_T2T_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_COPY_T2T_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for dst_x field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_9_DST_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_x_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_x_shift)
+
+/*define for dst_y field*/
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_offset 9
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_9_dst_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_9_DST_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_9_dst_y_mask) << SDMA_PKT_COPY_T2T_DW_9_dst_y_shift)
+
+/*define for DW_10 word*/
+/*define for dst_z field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_10_DST_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_z_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_z_shift)
+
+/*define for dst_pitch_in_tile field*/
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_offset 10
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_T2T_DW_10_DST_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_mask) << SDMA_PKT_COPY_T2T_DW_10_dst_pitch_in_tile_shift)
+
+/*define for DW_11 word*/
+/*define for dst_slice_pitch field*/
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_offset 11
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift  0
+#define SDMA_PKT_COPY_T2T_DW_11_DST_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_mask) << SDMA_PKT_COPY_T2T_DW_11_dst_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for dst_array_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift  3
+#define SDMA_PKT_COPY_T2T_DW_12_DST_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_array_mode_shift)
+
+/*define for dst_mit_mode field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift  8
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MIT_MODE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mit_mode_shift)
+
+/*define for dst_tilesplit_size field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_T2T_DW_12_DST_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_tilesplit_size_shift)
+
+/*define for dst_bank_w field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift  15
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_W(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_w_shift)
+
+/*define for dst_bank_h field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift  18
+#define SDMA_PKT_COPY_T2T_DW_12_DST_BANK_H(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_bank_h_shift)
+
+/*define for dst_num_bank field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift  21
+#define SDMA_PKT_COPY_T2T_DW_12_DST_NUM_BANK(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_num_bank_shift)
+
+/*define for dst_mat_aspt field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift  24
+#define SDMA_PKT_COPY_T2T_DW_12_DST_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_mat_aspt_shift)
+
+/*define for dst_pipe_config field*/
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_offset 12
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift  26
+#define SDMA_PKT_COPY_T2T_DW_12_DST_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_mask) << SDMA_PKT_COPY_T2T_DW_12_dst_pipe_config_shift)
+
+/*define for DW_13 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_x_shift  0
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_X(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_x_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_offset 13
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_T2T_DW_13_rect_y_shift  16
+#define SDMA_PKT_COPY_T2T_DW_13_RECT_Y(x) (((x) & SDMA_PKT_COPY_T2T_DW_13_rect_y_mask) << SDMA_PKT_COPY_T2T_DW_13_rect_y_shift)
+
+/*define for DW_14 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_T2T_DW_14_rect_z_shift  0
+#define SDMA_PKT_COPY_T2T_DW_14_RECT_Z(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_rect_z_mask) << SDMA_PKT_COPY_T2T_DW_14_rect_z_shift)
+
+/*define for dst_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift  16
+#define SDMA_PKT_COPY_T2T_DW_14_DST_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_dst_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_dst_sw_shift)
+
+/*define for src_sw field*/
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_offset 14
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_mask   0x00000003
+#define SDMA_PKT_COPY_T2T_DW_14_src_sw_shift  24
+#define SDMA_PKT_COPY_T2T_DW_14_SRC_SW(x) (((x) & SDMA_PKT_COPY_T2T_DW_14_src_sw_mask) << SDMA_PKT_COPY_T2T_DW_14_src_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_TILED_SUBWIN packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_TILED_SUBWIN_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_HEADER_detile_shift)
+
+/*define for TILED_ADDR_LO word*/
+/*define for tiled_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_offset 1
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_TILED_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_LO_tiled_addr_31_0_shift)
+
+/*define for TILED_ADDR_HI word*/
+/*define for tiled_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_offset 2
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_TILED_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_TILED_ADDR_HI_tiled_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for tiled_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_x_shift)
+
+/*define for tiled_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_offset 3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_3_TILED_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_3_tiled_y_shift)
+
+/*define for DW_4 word*/
+/*define for tiled_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_TILED_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_tiled_z_shift)
+
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_offset 4
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask   0x00000FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_4_PITCH_IN_TILE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_4_pitch_in_tile_shift)
+
+/*define for DW_5 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_offset 5
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_5_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_5_slice_pitch_shift)
+
+/*define for DW_6 word*/
+/*define for element_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ELEMENT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask   0x0000000F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift  3
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_ARRAY_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift  8
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MIT_MODE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift  11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift  15
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_W(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift  18
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_BANK_H(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift  21
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_NUM_BANK(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_MAT_ASPT(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_offset 6
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask   0x0000001F
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift  26
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_6_PIPE_CONFIG(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_6_pipe_config_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_offset 7
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_offset 8
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_TILED_SUBWIN_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+/*define for DW_9 word*/
+/*define for linear_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_x_shift)
+
+/*define for linear_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_offset 9
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_9_LINEAR_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_9_linear_y_shift)
+
+/*define for DW_10 word*/
+/*define for linear_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_z_shift)
+
+/*define for linear_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_offset 10
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_10_LINEAR_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_10_linear_pitch_shift)
+
+/*define for DW_11 word*/
+/*define for linear_slice_pitch field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_offset 11
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask   0x0FFFFFFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_11_LINEAR_SLICE_PITCH(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_11_linear_slice_pitch_shift)
+
+/*define for DW_12 word*/
+/*define for rect_x field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_X(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_x_shift)
+
+/*define for rect_y field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_offset 12
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask   0x00003FFF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_12_RECT_Y(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_12_rect_y_shift)
+
+/*define for DW_13 word*/
+/*define for rect_z field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask   0x000007FF
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift  0
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_RECT_Z(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_rect_z_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift  16
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_linear_sw_shift)
+
+/*define for tile_sw field*/
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_offset 13
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask   0x00000003
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift  24
+#define SDMA_PKT_COPY_TILED_SUBWIN_DW_13_TILE_SW(x) (((x) & SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_mask) << SDMA_PKT_COPY_TILED_SUBWIN_DW_13_tile_sw_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COPY_STRUCT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_op_shift  0
+#define SDMA_PKT_COPY_STRUCT_HEADER_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift  8
+#define SDMA_PKT_COPY_STRUCT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_sub_op_mask) << SDMA_PKT_COPY_STRUCT_HEADER_sub_op_shift)
+
+/*define for detile field*/
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_offset 0
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_HEADER_detile_shift  31
+#define SDMA_PKT_COPY_STRUCT_HEADER_DETILE(x) (((x) & SDMA_PKT_COPY_STRUCT_HEADER_detile_mask) << SDMA_PKT_COPY_STRUCT_HEADER_detile_shift)
+
+/*define for SB_ADDR_LO word*/
+/*define for sb_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_offset 1
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_SB_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_LO_sb_addr_31_0_shift)
+
+/*define for SB_ADDR_HI word*/
+/*define for sb_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_offset 2
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_SB_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_SB_ADDR_HI_sb_addr_63_32_shift)
+
+/*define for START_INDEX word*/
+/*define for start_index field*/
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_offset 3
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift  0
+#define SDMA_PKT_COPY_STRUCT_START_INDEX_START_INDEX(x) (((x) & SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_mask) << SDMA_PKT_COPY_STRUCT_START_INDEX_start_index_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_offset 4
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_COUNT_count_shift  0
+#define SDMA_PKT_COPY_STRUCT_COUNT_COUNT(x) (((x) & SDMA_PKT_COPY_STRUCT_COUNT_count_mask) << SDMA_PKT_COPY_STRUCT_COUNT_count_shift)
+
+/*define for DW_5 word*/
+/*define for stride field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_mask   0x000007FF
+#define SDMA_PKT_COPY_STRUCT_DW_5_stride_shift  0
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRIDE(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_stride_mask) << SDMA_PKT_COPY_STRUCT_DW_5_stride_shift)
+
+/*define for struct_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift  16
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_sw_shift)
+
+/*define for struct_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift  22
+#define SDMA_PKT_COPY_STRUCT_DW_5_STRUCT_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_struct_ha_shift)
+
+/*define for linear_sw field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask   0x00000003
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift  24
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_SW(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_sw_shift)
+
+/*define for linear_ha field*/
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_offset 5
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask   0x00000001
+#define SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift  30
+#define SDMA_PKT_COPY_STRUCT_DW_5_LINEAR_HA(x) (((x) & SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_mask) << SDMA_PKT_COPY_STRUCT_DW_5_linear_ha_shift)
+
+/*define for LINEAR_ADDR_LO word*/
+/*define for linear_addr_31_0 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_offset 6
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_LINEAR_ADDR_31_0(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_LO_linear_addr_31_0_shift)
+
+/*define for LINEAR_ADDR_HI word*/
+/*define for linear_addr_63_32 field*/
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_offset 7
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift  0
+#define SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_LINEAR_ADDR_63_32(x) (((x) & SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_mask) << SDMA_PKT_COPY_STRUCT_LINEAR_ADDR_HI_linear_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_UNTILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_UNTILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_UNTILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_UNTILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_UNTILED_DW_3_count_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_count_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_count_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_offset 3
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift  24
+#define SDMA_PKT_WRITE_UNTILED_DW_3_SW(x) (((x) & SDMA_PKT_WRITE_UNTILED_DW_3_sw_mask) << SDMA_PKT_WRITE_UNTILED_DW_3_sw_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_offset 4
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_UNTILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_UNTILED_DATA0_data0_mask) << SDMA_PKT_WRITE_UNTILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_TILED packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_TILED_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_TILED_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_TILED_HEADER_sub_op_mask) << SDMA_PKT_WRITE_TILED_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_TILED_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_TILED_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DW_3 word*/
+/*define for pitch_in_tile field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask   0x000007FF
+#define SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_3_PITCH_IN_TILE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_mask) << SDMA_PKT_WRITE_TILED_DW_3_pitch_in_tile_shift)
+
+/*define for height field*/
+#define SDMA_PKT_WRITE_TILED_DW_3_height_offset 3
+#define SDMA_PKT_WRITE_TILED_DW_3_height_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_3_height_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_3_HEIGHT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_3_height_mask) << SDMA_PKT_WRITE_TILED_DW_3_height_shift)
+
+/*define for DW_4 word*/
+/*define for slice_pitch field*/
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_offset 4
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_4_SLICE_PITCH(x) (((x) & SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_mask) << SDMA_PKT_WRITE_TILED_DW_4_slice_pitch_shift)
+
+/*define for DW_5 word*/
+/*define for element_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_element_size_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_5_ELEMENT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_element_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_element_size_shift)
+
+/*define for array_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask   0x0000000F
+#define SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift  3
+#define SDMA_PKT_WRITE_TILED_DW_5_ARRAY_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_array_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_array_mode_shift)
+
+/*define for mit_mode field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift  8
+#define SDMA_PKT_WRITE_TILED_DW_5_MIT_MODE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mit_mode_mask) << SDMA_PKT_WRITE_TILED_DW_5_mit_mode_shift)
+
+/*define for tilesplit_size field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask   0x00000007
+#define SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift  11
+#define SDMA_PKT_WRITE_TILED_DW_5_TILESPLIT_SIZE(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_mask) << SDMA_PKT_WRITE_TILED_DW_5_tilesplit_size_shift)
+
+/*define for bank_w field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift  15
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_W(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_w_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_w_shift)
+
+/*define for bank_h field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift  18
+#define SDMA_PKT_WRITE_TILED_DW_5_BANK_H(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_bank_h_mask) << SDMA_PKT_WRITE_TILED_DW_5_bank_h_shift)
+
+/*define for num_bank field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift  21
+#define SDMA_PKT_WRITE_TILED_DW_5_NUM_BANK(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_num_bank_mask) << SDMA_PKT_WRITE_TILED_DW_5_num_bank_shift)
+
+/*define for mat_aspt field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_5_MAT_ASPT(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_mask) << SDMA_PKT_WRITE_TILED_DW_5_mat_aspt_shift)
+
+/*define for pipe_config field*/
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_offset 5
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask   0x0000001F
+#define SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift  26
+#define SDMA_PKT_WRITE_TILED_DW_5_PIPE_CONFIG(x) (((x) & SDMA_PKT_WRITE_TILED_DW_5_pipe_config_mask) << SDMA_PKT_WRITE_TILED_DW_5_pipe_config_shift)
+
+/*define for DW_6 word*/
+/*define for x field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_x_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_x_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_x_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_6_X(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_x_mask) << SDMA_PKT_WRITE_TILED_DW_6_x_shift)
+
+/*define for y field*/
+#define SDMA_PKT_WRITE_TILED_DW_6_y_offset 6
+#define SDMA_PKT_WRITE_TILED_DW_6_y_mask   0x00003FFF
+#define SDMA_PKT_WRITE_TILED_DW_6_y_shift  16
+#define SDMA_PKT_WRITE_TILED_DW_6_Y(x) (((x) & SDMA_PKT_WRITE_TILED_DW_6_y_mask) << SDMA_PKT_WRITE_TILED_DW_6_y_shift)
+
+/*define for DW_7 word*/
+/*define for z field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_z_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_z_mask   0x00000FFF
+#define SDMA_PKT_WRITE_TILED_DW_7_z_shift  0
+#define SDMA_PKT_WRITE_TILED_DW_7_Z(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_z_mask) << SDMA_PKT_WRITE_TILED_DW_7_z_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_offset 7
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_mask   0x00000003
+#define SDMA_PKT_WRITE_TILED_DW_7_sw_shift  24
+#define SDMA_PKT_WRITE_TILED_DW_7_SW(x) (((x) & SDMA_PKT_WRITE_TILED_DW_7_sw_mask) << SDMA_PKT_WRITE_TILED_DW_7_sw_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_TILED_COUNT_count_offset 8
+#define SDMA_PKT_WRITE_TILED_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_WRITE_TILED_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_TILED_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_TILED_COUNT_count_mask) << SDMA_PKT_WRITE_TILED_COUNT_count_shift)
+
+/*define for DATA0 word*/
+/*define for data0 field*/
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_offset 9
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_TILED_DATA0_data0_shift  0
+#define SDMA_PKT_WRITE_TILED_DATA0_DATA0(x) (((x) & SDMA_PKT_WRITE_TILED_DATA0_data0_mask) << SDMA_PKT_WRITE_TILED_DATA0_data0_shift)
+
+
+/*
+** Definitions for SDMA_PKT_WRITE_INCR packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_op_shift  0
+#define SDMA_PKT_WRITE_INCR_HEADER_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_offset 0
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift  8
+#define SDMA_PKT_WRITE_INCR_HEADER_SUB_OP(x) (((x) & SDMA_PKT_WRITE_INCR_HEADER_sub_op_mask) << SDMA_PKT_WRITE_INCR_HEADER_sub_op_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_WRITE_INCR_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_WRITE_INCR_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for MASK_DW0 word*/
+/*define for mask_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_offset 3
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW0_MASK_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_mask) << SDMA_PKT_WRITE_INCR_MASK_DW0_mask_dw0_shift)
+
+/*define for MASK_DW1 word*/
+/*define for mask_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_offset 4
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_MASK_DW1_MASK_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_mask) << SDMA_PKT_WRITE_INCR_MASK_DW1_mask_dw1_shift)
+
+/*define for INIT_DW0 word*/
+/*define for init_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_offset 5
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW0_INIT_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_mask) << SDMA_PKT_WRITE_INCR_INIT_DW0_init_dw0_shift)
+
+/*define for INIT_DW1 word*/
+/*define for init_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_offset 6
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INIT_DW1_INIT_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_mask) << SDMA_PKT_WRITE_INCR_INIT_DW1_init_dw1_shift)
+
+/*define for INCR_DW0 word*/
+/*define for incr_dw0 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_offset 7
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW0_INCR_DW0(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_mask) << SDMA_PKT_WRITE_INCR_INCR_DW0_incr_dw0_shift)
+
+/*define for INCR_DW1 word*/
+/*define for incr_dw1 field*/
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_offset 8
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask   0xFFFFFFFF
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift  0
+#define SDMA_PKT_WRITE_INCR_INCR_DW1_INCR_DW1(x) (((x) & SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_mask) << SDMA_PKT_WRITE_INCR_INCR_DW1_incr_dw1_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_WRITE_INCR_COUNT_count_offset 9
+#define SDMA_PKT_WRITE_INCR_COUNT_count_mask   0x0007FFFF
+#define SDMA_PKT_WRITE_INCR_COUNT_count_shift  0
+#define SDMA_PKT_WRITE_INCR_COUNT_COUNT(x) (((x) & SDMA_PKT_WRITE_INCR_COUNT_count_mask) << SDMA_PKT_WRITE_INCR_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_INDIRECT packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_INDIRECT_HEADER_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_op_shift  0
+#define SDMA_PKT_INDIRECT_HEADER_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_op_mask) << SDMA_PKT_INDIRECT_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_INDIRECT_HEADER_sub_op_shift  8
+#define SDMA_PKT_INDIRECT_HEADER_SUB_OP(x) (((x) & SDMA_PKT_INDIRECT_HEADER_sub_op_mask) << SDMA_PKT_INDIRECT_HEADER_sub_op_shift)
+
+/*define for vmid field*/
+#define SDMA_PKT_INDIRECT_HEADER_vmid_offset 0
+#define SDMA_PKT_INDIRECT_HEADER_vmid_mask   0x0000000F
+#define SDMA_PKT_INDIRECT_HEADER_vmid_shift  16
+#define SDMA_PKT_INDIRECT_HEADER_VMID(x) (((x) & SDMA_PKT_INDIRECT_HEADER_vmid_mask) << SDMA_PKT_INDIRECT_HEADER_vmid_shift)
+
+/*define for BASE_LO word*/
+/*define for ib_base_31_0 field*/
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_offset 1
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift  0
+#define SDMA_PKT_INDIRECT_BASE_LO_IB_BASE_31_0(x) (((x) & SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_mask) << SDMA_PKT_INDIRECT_BASE_LO_ib_base_31_0_shift)
+
+/*define for BASE_HI word*/
+/*define for ib_base_63_32 field*/
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_offset 2
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift  0
+#define SDMA_PKT_INDIRECT_BASE_HI_IB_BASE_63_32(x) (((x) & SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_mask) << SDMA_PKT_INDIRECT_BASE_HI_ib_base_63_32_shift)
+
+/*define for IB_SIZE word*/
+/*define for ib_size field*/
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_offset 3
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask   0x000FFFFF
+#define SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift  0
+#define SDMA_PKT_INDIRECT_IB_SIZE_IB_SIZE(x) (((x) & SDMA_PKT_INDIRECT_IB_SIZE_ib_size_mask) << SDMA_PKT_INDIRECT_IB_SIZE_ib_size_shift)
+
+/*define for CSA_ADDR_LO word*/
+/*define for csa_addr_31_0 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_offset 4
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_LO_CSA_ADDR_31_0(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_LO_csa_addr_31_0_shift)
+
+/*define for CSA_ADDR_HI word*/
+/*define for csa_addr_63_32 field*/
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_offset 5
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift  0
+#define SDMA_PKT_INDIRECT_CSA_ADDR_HI_CSA_ADDR_63_32(x) (((x) & SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_mask) << SDMA_PKT_INDIRECT_CSA_ADDR_HI_csa_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SEMAPHORE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_op_shift  0
+#define SDMA_PKT_SEMAPHORE_HEADER_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SEMAPHORE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_sub_op_mask) << SDMA_PKT_SEMAPHORE_HEADER_sub_op_shift)
+
+/*define for write_one field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_write_one_shift  29
+#define SDMA_PKT_SEMAPHORE_HEADER_WRITE_ONE(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_write_one_mask) << SDMA_PKT_SEMAPHORE_HEADER_write_one_shift)
+
+/*define for signal field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_signal_shift  30
+#define SDMA_PKT_SEMAPHORE_HEADER_SIGNAL(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_signal_mask) << SDMA_PKT_SEMAPHORE_HEADER_signal_shift)
+
+/*define for mailbox field*/
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_offset 0
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask   0x00000001
+#define SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift  31
+#define SDMA_PKT_SEMAPHORE_HEADER_MAILBOX(x) (((x) & SDMA_PKT_SEMAPHORE_HEADER_mailbox_mask) << SDMA_PKT_SEMAPHORE_HEADER_mailbox_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_SEMAPHORE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_SEMAPHORE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_SEMAPHORE_ADDR_HI_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_FENCE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_FENCE_HEADER_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_op_shift  0
+#define SDMA_PKT_FENCE_HEADER_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_op_mask) << SDMA_PKT_FENCE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_FENCE_HEADER_sub_op_offset 0
+#define SDMA_PKT_FENCE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_FENCE_HEADER_sub_op_shift  8
+#define SDMA_PKT_FENCE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_FENCE_HEADER_sub_op_mask) << SDMA_PKT_FENCE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_FENCE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_FENCE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_FENCE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_FENCE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_FENCE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_FENCE_ADDR_HI_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_FENCE_DATA_data_offset 3
+#define SDMA_PKT_FENCE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_FENCE_DATA_data_shift  0
+#define SDMA_PKT_FENCE_DATA_DATA(x) (((x) & SDMA_PKT_FENCE_DATA_data_mask) << SDMA_PKT_FENCE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_SRBM_WRITE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_op_shift  0
+#define SDMA_PKT_SRBM_WRITE_HEADER_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift  8
+#define SDMA_PKT_SRBM_WRITE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_sub_op_mask) << SDMA_PKT_SRBM_WRITE_HEADER_sub_op_shift)
+
+/*define for byte_en field*/
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_offset 0
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask   0x0000000F
+#define SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift  28
+#define SDMA_PKT_SRBM_WRITE_HEADER_BYTE_EN(x) (((x) & SDMA_PKT_SRBM_WRITE_HEADER_byte_en_mask) << SDMA_PKT_SRBM_WRITE_HEADER_byte_en_shift)
+
+/*define for ADDR word*/
+/*define for addr field*/
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_offset 1
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_mask   0x0000FFFF
+#define SDMA_PKT_SRBM_WRITE_ADDR_addr_shift  0
+#define SDMA_PKT_SRBM_WRITE_ADDR_ADDR(x) (((x) & SDMA_PKT_SRBM_WRITE_ADDR_addr_mask) << SDMA_PKT_SRBM_WRITE_ADDR_addr_shift)
+
+/*define for DATA word*/
+/*define for data field*/
+#define SDMA_PKT_SRBM_WRITE_DATA_data_offset 2
+#define SDMA_PKT_SRBM_WRITE_DATA_data_mask   0xFFFFFFFF
+#define SDMA_PKT_SRBM_WRITE_DATA_data_shift  0
+#define SDMA_PKT_SRBM_WRITE_DATA_DATA(x) (((x) & SDMA_PKT_SRBM_WRITE_DATA_data_mask) << SDMA_PKT_SRBM_WRITE_DATA_data_shift)
+
+
+/*
+** Definitions for SDMA_PKT_PRE_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_op_shift  0
+#define SDMA_PKT_PRE_EXE_HEADER_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_op_mask) << SDMA_PKT_PRE_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_PRE_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_sub_op_mask) << SDMA_PKT_PRE_EXE_HEADER_sub_op_shift)
+
+/*define for dev_sel field*/
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_offset 0
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask   0x000000FF
+#define SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift  16
+#define SDMA_PKT_PRE_EXE_HEADER_DEV_SEL(x) (((x) & SDMA_PKT_PRE_EXE_HEADER_dev_sel_mask) << SDMA_PKT_PRE_EXE_HEADER_dev_sel_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_offset 1
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_PRE_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_PRE_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_COND_EXE packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_COND_EXE_HEADER_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_op_shift  0
+#define SDMA_PKT_COND_EXE_HEADER_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_op_mask) << SDMA_PKT_COND_EXE_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_offset 0
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_COND_EXE_HEADER_sub_op_shift  8
+#define SDMA_PKT_COND_EXE_HEADER_SUB_OP(x) (((x) & SDMA_PKT_COND_EXE_HEADER_sub_op_mask) << SDMA_PKT_COND_EXE_HEADER_sub_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_mask) << SDMA_PKT_COND_EXE_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_COND_EXE_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_mask) << SDMA_PKT_COND_EXE_ADDR_HI_addr_63_32_shift)
+
+/*define for REFERENCE word*/
+/*define for reference field*/
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_offset 3
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_mask   0xFFFFFFFF
+#define SDMA_PKT_COND_EXE_REFERENCE_reference_shift  0
+#define SDMA_PKT_COND_EXE_REFERENCE_REFERENCE(x) (((x) & SDMA_PKT_COND_EXE_REFERENCE_reference_mask) << SDMA_PKT_COND_EXE_REFERENCE_reference_shift)
+
+/*define for EXEC_COUNT word*/
+/*define for exec_count field*/
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_offset 4
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask   0x00003FFF
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift  0
+#define SDMA_PKT_COND_EXE_EXEC_COUNT_EXEC_COUNT(x) (((x) & SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_mask) << SDMA_PKT_COND_EXE_EXEC_COUNT_exec_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_CONSTANT_FILL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_op_shift  0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift  8
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sub_op_shift)
+
+/*define for sw field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift  16
+#define SDMA_PKT_CONSTANT_FILL_HEADER_SW(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_sw_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_sw_shift)
+
+/*define for fillsize field*/
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_offset 0
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask   0x00000003
+#define SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift  30
+#define SDMA_PKT_CONSTANT_FILL_HEADER_FILLSIZE(x) (((x) & SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_mask) << SDMA_PKT_CONSTANT_FILL_HEADER_fillsize_shift)
+
+/*define for DST_ADDR_LO word*/
+/*define for dst_addr_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_offset 1
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_DST_ADDR_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_LO_dst_addr_31_0_shift)
+
+/*define for DST_ADDR_HI word*/
+/*define for dst_addr_63_32 field*/
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_offset 2
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_DST_ADDR_63_32(x) (((x) & SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_mask) << SDMA_PKT_CONSTANT_FILL_DST_ADDR_HI_dst_addr_63_32_shift)
+
+/*define for DATA word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_offset 3
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift  0
+#define SDMA_PKT_CONSTANT_FILL_DATA_SRC_DATA_31_0(x) (((x) & SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_mask) << SDMA_PKT_CONSTANT_FILL_DATA_src_data_31_0_shift)
+
+/*define for COUNT word*/
+/*define for count field*/
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_offset 4
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_mask   0x003FFFFF
+#define SDMA_PKT_CONSTANT_FILL_COUNT_count_shift  0
+#define SDMA_PKT_CONSTANT_FILL_COUNT_COUNT(x) (((x) & SDMA_PKT_CONSTANT_FILL_COUNT_count_mask) << SDMA_PKT_CONSTANT_FILL_COUNT_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_POLL_REGMEM packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_op_shift  0
+#define SDMA_PKT_POLL_REGMEM_HEADER_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift  8
+#define SDMA_PKT_POLL_REGMEM_HEADER_SUB_OP(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_sub_op_mask) << SDMA_PKT_POLL_REGMEM_HEADER_sub_op_shift)
+
+/*define for hdp_flush field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift  26
+#define SDMA_PKT_POLL_REGMEM_HEADER_HDP_FLUSH(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_mask) << SDMA_PKT_POLL_REGMEM_HEADER_hdp_flush_shift)
+
+/*define for func field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_mask   0x00000007
+#define SDMA_PKT_POLL_REGMEM_HEADER_func_shift  28
+#define SDMA_PKT_POLL_REGMEM_HEADER_FUNC(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_func_mask) << SDMA_PKT_POLL_REGMEM_HEADER_func_shift)
+
+/*define for mem_poll field*/
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_offset 0
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask   0x00000001
+#define SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift  31
+#define SDMA_PKT_POLL_REGMEM_HEADER_MEM_POLL(x) (((x) & SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_mask) << SDMA_PKT_POLL_REGMEM_HEADER_mem_poll_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_mask) << SDMA_PKT_POLL_REGMEM_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_POLL_REGMEM_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_mask) << SDMA_PKT_POLL_REGMEM_ADDR_HI_addr_63_32_shift)
+
+/*define for VALUE word*/
+/*define for value field*/
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_offset 3
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_VALUE_value_shift  0
+#define SDMA_PKT_POLL_REGMEM_VALUE_VALUE(x) (((x) & SDMA_PKT_POLL_REGMEM_VALUE_value_mask) << SDMA_PKT_POLL_REGMEM_VALUE_value_shift)
+
+/*define for MASK word*/
+/*define for mask field*/
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_offset 4
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_mask   0xFFFFFFFF
+#define SDMA_PKT_POLL_REGMEM_MASK_mask_shift  0
+#define SDMA_PKT_POLL_REGMEM_MASK_MASK(x) (((x) & SDMA_PKT_POLL_REGMEM_MASK_mask_mask) << SDMA_PKT_POLL_REGMEM_MASK_mask_shift)
+
+/*define for DW5 word*/
+/*define for interval field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_mask   0x0000FFFF
+#define SDMA_PKT_POLL_REGMEM_DW5_interval_shift  0
+#define SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_interval_mask) << SDMA_PKT_POLL_REGMEM_DW5_interval_shift)
+
+/*define for retry_count field*/
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_offset 5
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask   0x00000FFF
+#define SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift  16
+#define SDMA_PKT_POLL_REGMEM_DW5_RETRY_COUNT(x) (((x) & SDMA_PKT_POLL_REGMEM_DW5_retry_count_mask) << SDMA_PKT_POLL_REGMEM_DW5_retry_count_shift)
+
+
+/*
+** Definitions for SDMA_PKT_ATOMIC packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_ATOMIC_HEADER_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_ATOMIC_HEADER_op_shift  0
+#define SDMA_PKT_ATOMIC_HEADER_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_op_mask) << SDMA_PKT_ATOMIC_HEADER_op_shift)
+
+/*define for loop field*/
+#define SDMA_PKT_ATOMIC_HEADER_loop_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_loop_mask   0x00000001
+#define SDMA_PKT_ATOMIC_HEADER_loop_shift  16
+#define SDMA_PKT_ATOMIC_HEADER_LOOP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_loop_mask) << SDMA_PKT_ATOMIC_HEADER_loop_shift)
+
+/*define for atomic_op field*/
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_offset 0
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_mask   0x0000007F
+#define SDMA_PKT_ATOMIC_HEADER_atomic_op_shift  25
+#define SDMA_PKT_ATOMIC_HEADER_ATOMIC_OP(x) (((x) & SDMA_PKT_ATOMIC_HEADER_atomic_op_mask) << SDMA_PKT_ATOMIC_HEADER_atomic_op_shift)
+
+/*define for ADDR_LO word*/
+/*define for addr_31_0 field*/
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_offset 1
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_LO_ADDR_31_0(x) (((x) & SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_mask) << SDMA_PKT_ATOMIC_ADDR_LO_addr_31_0_shift)
+
+/*define for ADDR_HI word*/
+/*define for addr_63_32 field*/
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_offset 2
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift  0
+#define SDMA_PKT_ATOMIC_ADDR_HI_ADDR_63_32(x) (((x) & SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_mask) << SDMA_PKT_ATOMIC_ADDR_HI_addr_63_32_shift)
+
+/*define for SRC_DATA_LO word*/
+/*define for src_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_offset 3
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_LO_SRC_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_mask) << SDMA_PKT_ATOMIC_SRC_DATA_LO_src_data_31_0_shift)
+
+/*define for SRC_DATA_HI word*/
+/*define for src_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_offset 4
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_SRC_DATA_HI_SRC_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_mask) << SDMA_PKT_ATOMIC_SRC_DATA_HI_src_data_63_32_shift)
+
+/*define for CMP_DATA_LO word*/
+/*define for cmp_data_31_0 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_offset 5
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_LO_CMP_DATA_31_0(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_mask) << SDMA_PKT_ATOMIC_CMP_DATA_LO_cmp_data_31_0_shift)
+
+/*define for CMP_DATA_HI word*/
+/*define for cmp_data_63_32 field*/
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_offset 6
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift  0
+#define SDMA_PKT_ATOMIC_CMP_DATA_HI_CMP_DATA_63_32(x) (((x) & SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_mask) << SDMA_PKT_ATOMIC_CMP_DATA_HI_cmp_data_63_32_shift)
+
+/*define for LOOP_INTERVAL word*/
+/*define for loop_interval field*/
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_offset 7
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask   0x00001FFF
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift  0
+#define SDMA_PKT_ATOMIC_LOOP_INTERVAL_LOOP_INTERVAL(x) (((x) & SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_mask) << SDMA_PKT_ATOMIC_LOOP_INTERVAL_loop_interval_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_SET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_SET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_SET_HEADER_sub_op_shift)
+
+/*define for INIT_DATA_LO word*/
+/*define for init_data_31_0 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_offset 1
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_INIT_DATA_31_0(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_LO_init_data_31_0_shift)
+
+/*define for INIT_DATA_HI word*/
+/*define for init_data_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_INIT_DATA_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_mask) << SDMA_PKT_TIMESTAMP_SET_INIT_DATA_HI_init_data_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TIMESTAMP_GET_GLOBAL packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_offset 0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift  8
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_HEADER_sub_op_shift)
+
+/*define for WRITE_ADDR_LO word*/
+/*define for write_addr_31_3 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_offset 1
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask   0x1FFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift  3
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_WRITE_ADDR_31_3(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_LO_write_addr_31_3_shift)
+
+/*define for WRITE_ADDR_HI word*/
+/*define for write_addr_63_32 field*/
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_offset 2
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask   0xFFFFFFFF
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift  0
+#define SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_WRITE_ADDR_63_32(x) (((x) & SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_mask) << SDMA_PKT_TIMESTAMP_GET_GLOBAL_WRITE_ADDR_HI_write_addr_63_32_shift)
+
+
+/*
+** Definitions for SDMA_PKT_TRAP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_TRAP_HEADER_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_op_shift  0
+#define SDMA_PKT_TRAP_HEADER_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_op_mask) << SDMA_PKT_TRAP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_TRAP_HEADER_sub_op_offset 0
+#define SDMA_PKT_TRAP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_TRAP_HEADER_sub_op_shift  8
+#define SDMA_PKT_TRAP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_TRAP_HEADER_sub_op_mask) << SDMA_PKT_TRAP_HEADER_sub_op_shift)
+
+/*define for INT_CONTEXT word*/
+/*define for int_context field*/
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_offset 1
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask   0x0FFFFFFF
+#define SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift  0
+#define SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(x) (((x) & SDMA_PKT_TRAP_INT_CONTEXT_int_context_mask) << SDMA_PKT_TRAP_INT_CONTEXT_int_context_shift)
+
+
+/*
+** Definitions for SDMA_PKT_NOP packet
+*/
+
+/*define for HEADER word*/
+/*define for op field*/
+#define SDMA_PKT_NOP_HEADER_op_offset 0
+#define SDMA_PKT_NOP_HEADER_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_op_shift  0
+#define SDMA_PKT_NOP_HEADER_OP(x) (((x) & SDMA_PKT_NOP_HEADER_op_mask) << SDMA_PKT_NOP_HEADER_op_shift)
+
+/*define for sub_op field*/
+#define SDMA_PKT_NOP_HEADER_sub_op_offset 0
+#define SDMA_PKT_NOP_HEADER_sub_op_mask   0x000000FF
+#define SDMA_PKT_NOP_HEADER_sub_op_shift  8
+#define SDMA_PKT_NOP_HEADER_SUB_OP(x) (((x) & SDMA_PKT_NOP_HEADER_sub_op_mask) << SDMA_PKT_NOP_HEADER_sub_op_shift)
+
+/*define for count field*/
+#define SDMA_PKT_NOP_HEADER_count_offset 0
+#define SDMA_PKT_NOP_HEADER_count_mask   0x00003FFF
+#define SDMA_PKT_NOP_HEADER_count_shift  16
+#define SDMA_PKT_NOP_HEADER_COUNT(x) (((x) & SDMA_PKT_NOP_HEADER_count_mask) << SDMA_PKT_NOP_HEADER_count_shift)
+
+#endif /* __TONGA_SDMA_PKT_OPEN_H_ */
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_smc.c b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_smc.c
new file mode 100644
index 000000000..5421309c1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_smc.c
@@ -0,0 +1,856 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "tonga_ppsmc.h"
+#include "tonga_smumgr.h"
+#include "smu_ucode_xfer_vi.h"
+#include "amdgpu_ucode.h"
+
+#include "smu/smu_7_1_2_d.h"
+#include "smu/smu_7_1_2_sh_mask.h"
+
+#define TONGA_SMC_SIZE 0x20000
+
+static int tonga_set_smc_sram_address(struct amdgpu_device *adev, uint32_t smc_address, uint32_t limit)
+{
+	uint32_t val;
+
+	if (smc_address & 3)
+		return -EINVAL;
+
+	if ((smc_address + 3) > limit)
+		return -EINVAL;
+
+	WREG32(mmSMC_IND_INDEX_0, smc_address);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 0);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+
+	return 0;
+}
+
+static int tonga_copy_bytes_to_smc(struct amdgpu_device *adev, uint32_t smc_start_address, const uint8_t *src, uint32_t byte_count, uint32_t limit)
+{
+	uint32_t addr;
+	uint32_t data, orig_data;
+	int result = 0;
+	uint32_t extra_shift;
+	unsigned long flags;
+
+	if (smc_start_address & 3)
+		return -EINVAL;
+
+	if ((smc_start_address + byte_count) > limit)
+		return -EINVAL;
+
+	addr = smc_start_address;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	while (byte_count >= 4) {
+		/* Bytes are written into the SMC addres space with the MSB first */
+		data = (src[0] << 24) + (src[1] << 16) + (src[2] << 8) + src[3];
+
+		result = tonga_set_smc_sram_address(adev, addr, limit);
+
+		if (result)
+			goto out;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+
+		src += 4;
+		byte_count -= 4;
+		addr += 4;
+	}
+
+	if (0 != byte_count) {
+		/* Now write odd bytes left, do a read modify write cycle */
+		data = 0;
+
+		result = tonga_set_smc_sram_address(adev, addr, limit);
+		if (result)
+			goto out;
+
+		orig_data = RREG32(mmSMC_IND_DATA_0);
+		extra_shift = 8 * (4 - byte_count);
+
+		while (byte_count > 0) {
+			data = (data << 8) + *src++;
+			byte_count--;
+		}
+
+		data <<= extra_shift;
+		data |= (orig_data & ~((~0UL) << extra_shift));
+
+		result = tonga_set_smc_sram_address(adev, addr, limit);
+		if (result)
+			goto out;
+
+		WREG32(mmSMC_IND_DATA_0, data);
+	}
+
+out:
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int tonga_program_jump_on_start(struct amdgpu_device *adev)
+{
+	static unsigned char data[] = {0xE0, 0x00, 0x80, 0x40};
+	tonga_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
+
+	return 0;
+}
+
+static bool tonga_is_smc_ram_running(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_GET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable);
+
+	return ((0 == val) && (0x20100 <= RREG32_SMC(ixSMC_PC_C)));
+}
+
+static int wait_smu_response(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32(mmSMC_RESP_0);
+		if (REG_GET_FIELD(val, SMC_RESP_0, SMC_RESP))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int tonga_send_msg_to_smc_offset(struct amdgpu_device *adev)
+{
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MSG_ARG_0, 0x20000);
+	WREG32(mmSMC_MESSAGE_0, PPSMC_MSG_Test);
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send message\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int tonga_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
+{
+	if (!tonga_is_smc_ram_running(adev))
+	{
+		return -EINVAL;;
+	}
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send message\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int tonga_send_msg_to_smc_without_waiting(struct amdgpu_device *adev,
+						PPSMC_Msg msg)
+{
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MESSAGE_0, msg);
+
+	return 0;
+}
+
+static int tonga_send_msg_to_smc_with_parameter(struct amdgpu_device *adev,
+						PPSMC_Msg msg,
+						uint32_t parameter)
+{
+	if (!tonga_is_smc_ram_running(adev))
+		return -EINVAL;
+
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return tonga_send_msg_to_smc(adev, msg);
+}
+
+static int tonga_send_msg_to_smc_with_parameter_without_waiting(
+					struct amdgpu_device *adev,
+					PPSMC_Msg msg, uint32_t parameter)
+{
+	if (wait_smu_response(adev)) {
+		DRM_ERROR("Failed to send previous message\n");
+		return -EINVAL;
+	}
+
+	WREG32(mmSMC_MSG_ARG_0, parameter);
+
+	return tonga_send_msg_to_smc_without_waiting(adev, msg);
+}
+
+#if 0 /* not used yet */
+static int tonga_wait_for_smc_inactive(struct amdgpu_device *adev)
+{
+	int i;
+	uint32_t val;
+
+	if (!tonga_is_smc_ram_running(adev))
+		return -EINVAL;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+		if (REG_GET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, cken) == 0)
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -EINVAL;
+
+	return 0;
+}
+#endif
+
+static int tonga_smu_upload_firmware_image(struct amdgpu_device *adev)
+{
+	const struct smc_firmware_header_v1_0 *hdr;
+	uint32_t ucode_size;
+	uint32_t ucode_start_address;
+	const uint8_t *src;
+	uint32_t val;
+	uint32_t byte_count;
+	uint32_t *data;
+	unsigned long flags;
+
+	if (!adev->pm.fw)
+		return -EINVAL;
+
+	hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+	ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
+	ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
+	src = (const uint8_t *)
+		(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	if (ucode_size & 3) {
+		DRM_ERROR("SMC ucode is not 4 bytes aligned\n");
+		return -EINVAL;
+	}
+
+	if (ucode_size > TONGA_SMC_SIZE) {
+		DRM_ERROR("SMC address is beyond the SMC RAM area\n");
+		return -EINVAL;
+	}
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 1);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+
+	byte_count = ucode_size;
+	data = (uint32_t *)src;
+	for (; byte_count >= 4; data++, byte_count -= 4)
+		WREG32(mmSMC_IND_DATA_0, data[0]);
+
+	val = RREG32(mmSMC_IND_ACCESS_CNTL);
+	val = REG_SET_FIELD(val, SMC_IND_ACCESS_CNTL, AUTO_INCREMENT_IND_0, 0);
+	WREG32(mmSMC_IND_ACCESS_CNTL, val);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+
+	return 0;
+}
+
+#if 0 /* not used yet */
+static int tonga_read_smc_sram_dword(struct amdgpu_device *adev,
+				uint32_t smc_address,
+				uint32_t *value,
+				uint32_t limit)
+{
+	int result;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	result = tonga_set_smc_sram_address(adev, smc_address, limit);
+	if (result == 0)
+		*value = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int tonga_write_smc_sram_dword(struct amdgpu_device *adev,
+				uint32_t smc_address,
+				uint32_t value,
+				uint32_t limit)
+{
+	int result;
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	result = tonga_set_smc_sram_address(adev, smc_address, limit);
+	if (result == 0)
+		WREG32(mmSMC_IND_DATA_0, value);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return result;
+}
+
+static int tonga_smu_stop_smc(struct amdgpu_device *adev)
+{
+	uint32_t val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 1);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+
+	return 0;
+}
+#endif
+
+static enum AMDGPU_UCODE_ID tonga_convert_fw_type(uint32_t fw_type)
+{
+	switch (fw_type) {
+		case UCODE_ID_SDMA0:
+			return AMDGPU_UCODE_ID_SDMA0;
+		case UCODE_ID_SDMA1:
+			return AMDGPU_UCODE_ID_SDMA1;
+		case UCODE_ID_CP_CE:
+			return AMDGPU_UCODE_ID_CP_CE;
+		case UCODE_ID_CP_PFP:
+			return AMDGPU_UCODE_ID_CP_PFP;
+		case UCODE_ID_CP_ME:
+			return AMDGPU_UCODE_ID_CP_ME;
+		case UCODE_ID_CP_MEC:
+		case UCODE_ID_CP_MEC_JT1:
+			return AMDGPU_UCODE_ID_CP_MEC1;
+		case UCODE_ID_CP_MEC_JT2:
+			return AMDGPU_UCODE_ID_CP_MEC2;
+		case UCODE_ID_RLC_G:
+			return AMDGPU_UCODE_ID_RLC_G;
+		default:
+			DRM_ERROR("ucode type is out of range!\n");
+			return AMDGPU_UCODE_ID_MAXIMUM;
+	}
+}
+
+static int tonga_smu_populate_single_firmware_entry(struct amdgpu_device *adev,
+						uint32_t fw_type,
+						struct SMU_Entry *entry)
+{
+	enum AMDGPU_UCODE_ID id = tonga_convert_fw_type(fw_type);
+	struct amdgpu_firmware_info *ucode = &adev->firmware.ucode[id];
+	const struct gfx_firmware_header_v1_0 *header = NULL;
+	uint64_t gpu_addr;
+	uint32_t data_size;
+
+	if (ucode->fw == NULL)
+		return -EINVAL;
+
+	gpu_addr  = ucode->mc_addr;
+	header = (const struct gfx_firmware_header_v1_0 *)ucode->fw->data;
+	data_size = le32_to_cpu(header->header.ucode_size_bytes);
+
+	if ((fw_type == UCODE_ID_CP_MEC_JT1) ||
+		(fw_type == UCODE_ID_CP_MEC_JT2)) {
+		gpu_addr += le32_to_cpu(header->jt_offset) << 2;
+		data_size = le32_to_cpu(header->jt_size) << 2;
+	}
+
+	entry->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
+	entry->id = (uint16_t)fw_type;
+	entry->image_addr_high = upper_32_bits(gpu_addr);
+	entry->image_addr_low = lower_32_bits(gpu_addr);
+	entry->meta_data_addr_high = 0;
+	entry->meta_data_addr_low = 0;
+	entry->data_size_byte = data_size;
+	entry->num_register_entries = 0;
+
+	if (fw_type == UCODE_ID_RLC_G)
+		entry->flags = 1;
+	else
+		entry->flags = 0;
+
+	return 0;
+}
+
+static int tonga_smu_request_load_fw(struct amdgpu_device *adev)
+{
+	struct tonga_smu_private_data *private = (struct tonga_smu_private_data *)adev->smu.priv;
+	struct SMU_DRAMData_TOC *toc;
+	uint32_t fw_to_load;
+
+	WREG32_SMC(ixSOFT_REGISTERS_TABLE_28, 0);
+
+	tonga_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SMU_DRAM_ADDR_HI, private->smu_buffer_addr_high);
+	tonga_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_SMU_DRAM_ADDR_LO, private->smu_buffer_addr_low);
+
+	toc = (struct SMU_DRAMData_TOC *)private->header;
+	toc->num_entries = 0;
+	toc->structure_version = 1;
+
+	if (!adev->firmware.smu_load)
+		return 0;
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_RLC_G,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for RLC\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_CE,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for CE\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_PFP,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for PFP\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_ME,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for ME\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC_JT1,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC_JT1\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_CP_MEC_JT2,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for MEC_JT2\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_SDMA0,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for SDMA0\n");
+		return -EINVAL;
+	}
+
+	if (tonga_smu_populate_single_firmware_entry(adev, UCODE_ID_SDMA1,
+			&toc->entry[toc->num_entries++])) {
+		DRM_ERROR("Failed to get firmware entry for SDMA1\n");
+		return -EINVAL;
+	}
+
+	tonga_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DRV_DRAM_ADDR_HI, private->header_addr_high);
+	tonga_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_DRV_DRAM_ADDR_LO, private->header_addr_low);
+
+	fw_to_load = UCODE_ID_RLC_G_MASK |
+			UCODE_ID_SDMA0_MASK |
+			UCODE_ID_SDMA1_MASK |
+			UCODE_ID_CP_CE_MASK |
+			UCODE_ID_CP_ME_MASK |
+			UCODE_ID_CP_PFP_MASK |
+			UCODE_ID_CP_MEC_MASK;
+
+	if (tonga_send_msg_to_smc_with_parameter_without_waiting(adev, PPSMC_MSG_LoadUcodes, fw_to_load)) {
+		DRM_ERROR("Fail to request SMU load ucode\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static uint32_t tonga_smu_get_mask_for_fw_type(uint32_t fw_type)
+{
+	switch (fw_type) {
+		case AMDGPU_UCODE_ID_SDMA0:
+			return UCODE_ID_SDMA0_MASK;
+		case AMDGPU_UCODE_ID_SDMA1:
+			return UCODE_ID_SDMA1_MASK;
+		case AMDGPU_UCODE_ID_CP_CE:
+			return UCODE_ID_CP_CE_MASK;
+		case AMDGPU_UCODE_ID_CP_PFP:
+			return UCODE_ID_CP_PFP_MASK;
+		case AMDGPU_UCODE_ID_CP_ME:
+			return UCODE_ID_CP_ME_MASK;
+		case AMDGPU_UCODE_ID_CP_MEC1:
+			return UCODE_ID_CP_MEC_MASK;
+		case AMDGPU_UCODE_ID_CP_MEC2:
+			return UCODE_ID_CP_MEC_MASK;
+		case AMDGPU_UCODE_ID_RLC_G:
+			return UCODE_ID_RLC_G_MASK;
+		default:
+			DRM_ERROR("ucode type is out of range!\n");
+			return 0;
+	}
+}
+
+static int tonga_smu_check_fw_load_finish(struct amdgpu_device *adev,
+					uint32_t fw_type)
+{
+	uint32_t fw_mask = tonga_smu_get_mask_for_fw_type(fw_type);
+	int i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (fw_mask == (RREG32_SMC(ixSOFT_REGISTERS_TABLE_28) & fw_mask))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("check firmware loading failed\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int tonga_smu_start_in_protection_mode(struct amdgpu_device *adev)
+{
+	int result;
+	uint32_t val;
+	int i;
+
+	/* Assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	result = tonga_smu_upload_firmware_image(adev);
+	if (result)
+		return result;
+
+	/* Clear status */
+	WREG32_SMC(ixSMU_STATUS, 0);
+
+	/* Enable clock */
+	val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 0);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+
+	/* De-assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 0);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	/* Set SMU Auto Start */
+	val = RREG32_SMC(ixSMU_INPUT_DATA);
+	val = REG_SET_FIELD(val, SMU_INPUT_DATA, AUTO_START, 1);
+	WREG32_SMC(ixSMU_INPUT_DATA, val);
+
+	/* Clear firmware interrupt enable flag */
+	WREG32_SMC(ixFIRMWARE_FLAGS, 0);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixRCU_UC_EVENTS);
+		if (REG_GET_FIELD(val, RCU_UC_EVENTS, INTERRUPTS_ENABLED))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("Interrupt is not enabled by firmware\n");
+		return -EINVAL;
+	}
+
+	/* Call Test SMU message with 0x20000 offset
+	 * to trigger SMU start
+	 */
+	tonga_send_msg_to_smc_offset(adev);
+
+	/* Wait for done bit to be set */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixSMU_STATUS);
+		if (REG_GET_FIELD(val, SMU_STATUS, SMU_DONE))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("Timeout for SMU start\n");
+		return -EINVAL;
+	}
+
+	/* Check pass/failed indicator */
+	val = RREG32_SMC(ixSMU_STATUS);
+	if (!REG_GET_FIELD(val, SMU_STATUS, SMU_PASS)) {
+		DRM_ERROR("SMU Firmware start failed\n");
+		return -EINVAL;
+	}
+
+	/* Wait for firmware to initialize */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixFIRMWARE_FLAGS);
+		if(REG_GET_FIELD(val, FIRMWARE_FLAGS, INTERRUPTS_ENABLED))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("SMU firmware initialization failed\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int tonga_smu_start_in_non_protection_mode(struct amdgpu_device *adev)
+{
+	int i, result;
+	uint32_t val;
+
+	/* wait for smc boot up */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixRCU_UC_EVENTS);
+		val = REG_GET_FIELD(val, RCU_UC_EVENTS, boot_seq_done);
+		if (val)
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("SMC boot sequence is not completed\n");
+		return -EINVAL;
+	}
+
+	/* Clear firmware interrupt enable flag */
+	WREG32_SMC(ixFIRMWARE_FLAGS, 0);
+
+	/* Assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 1);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	result = tonga_smu_upload_firmware_image(adev);
+	if (result)
+		return result;
+
+	/* Set smc instruct start point at 0x0 */
+	tonga_program_jump_on_start(adev);
+
+	/* Enable clock */
+	val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	val = REG_SET_FIELD(val, SMC_SYSCON_CLOCK_CNTL_0, ck_disable, 0);
+	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
+
+	/* De-assert reset */
+	val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
+	val = REG_SET_FIELD(val, SMC_SYSCON_RESET_CNTL, rst_reg, 0);
+	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
+
+	/* Wait for firmware to initialize */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		val = RREG32_SMC(ixFIRMWARE_FLAGS);
+		if (REG_GET_FIELD(val, FIRMWARE_FLAGS, INTERRUPTS_ENABLED))
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout) {
+		DRM_ERROR("Timeout for SMC firmware initialization\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int tonga_smu_start(struct amdgpu_device *adev)
+{
+	int result;
+	uint32_t val;
+
+	if (!tonga_is_smc_ram_running(adev)) {
+		val = RREG32_SMC(ixSMU_FIRMWARE);
+		if (!REG_GET_FIELD(val, SMU_FIRMWARE, SMU_MODE)) {
+			result = tonga_smu_start_in_non_protection_mode(adev);
+			if (result)
+				return result;
+		} else {
+			result = tonga_smu_start_in_protection_mode(adev);
+			if (result)
+				return result;
+		}
+	}
+
+	return tonga_smu_request_load_fw(adev);
+}
+
+static const struct amdgpu_smumgr_funcs tonga_smumgr_funcs = {
+	.check_fw_load_finish = tonga_smu_check_fw_load_finish,
+	.request_smu_load_fw = NULL,
+	.request_smu_specific_fw = NULL,
+};
+
+int tonga_smu_init(struct amdgpu_device *adev)
+{
+	struct tonga_smu_private_data *private;
+	uint32_t image_size = ((sizeof(struct SMU_DRAMData_TOC) / 4096) + 1) * 4096;
+	uint32_t smu_internal_buffer_size = 200*4096;
+	struct amdgpu_bo **toc_buf = &adev->smu.toc_buf;
+	struct amdgpu_bo **smu_buf = &adev->smu.smu_buf;
+	uint64_t mc_addr;
+	void *toc_buf_ptr;
+	void *smu_buf_ptr;
+	int ret;
+
+	private = kzalloc(sizeof(struct tonga_smu_private_data), GFP_KERNEL);
+	if (NULL == private)
+		return -ENOMEM;
+
+	/* allocate firmware buffers */
+	if (adev->firmware.smu_load)
+		amdgpu_ucode_init_bo(adev);
+
+	adev->smu.priv = private;
+	adev->smu.fw_flags = 0;
+
+	/* Allocate FW image data structure and header buffer */
+	ret = amdgpu_bo_create(adev, image_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_VRAM,
+			       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			       NULL, NULL, toc_buf);
+	if (ret) {
+		DRM_ERROR("Failed to allocate memory for TOC buffer\n");
+		return -ENOMEM;
+	}
+
+	/* Allocate buffer for SMU internal buffer */
+	ret = amdgpu_bo_create(adev, smu_internal_buffer_size, PAGE_SIZE,
+			       true, AMDGPU_GEM_DOMAIN_VRAM,
+			       AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+			       NULL, NULL, smu_buf);
+	if (ret) {
+		DRM_ERROR("Failed to allocate memory for SMU internal buffer\n");
+		return -ENOMEM;
+	}
+
+	/* Retrieve GPU address for header buffer and internal buffer */
+	ret = amdgpu_bo_reserve(adev->smu.toc_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to reserve the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.toc_buf, AMDGPU_GEM_DOMAIN_VRAM, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to pin the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_kmap(*toc_buf, &toc_buf_ptr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.toc_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to map the TOC buffer\n");
+		return -EINVAL;
+	}
+
+	amdgpu_bo_unreserve(adev->smu.toc_buf);
+	private->header_addr_low = lower_32_bits(mc_addr);
+	private->header_addr_high = upper_32_bits(mc_addr);
+	private->header = toc_buf_ptr;
+
+	ret = amdgpu_bo_reserve(adev->smu.smu_buf, false);
+	if (ret) {
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to reserve the SMU internal buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_pin(adev->smu.smu_buf, AMDGPU_GEM_DOMAIN_VRAM, &mc_addr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to pin the SMU internal buffer\n");
+		return -EINVAL;
+	}
+
+	ret = amdgpu_bo_kmap(*smu_buf, &smu_buf_ptr);
+	if (ret) {
+		amdgpu_bo_unreserve(adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.smu_buf);
+		amdgpu_bo_unref(&adev->smu.toc_buf);
+		DRM_ERROR("Failed to map the SMU internal buffer\n");
+		return -EINVAL;
+	}
+
+	amdgpu_bo_unreserve(adev->smu.smu_buf);
+	private->smu_buffer_addr_low = lower_32_bits(mc_addr);
+	private->smu_buffer_addr_high = upper_32_bits(mc_addr);
+
+	adev->smu.smumgr_funcs = &tonga_smumgr_funcs;
+
+	return 0;
+}
+
+int tonga_smu_fini(struct amdgpu_device *adev)
+{
+	amdgpu_bo_unref(&adev->smu.toc_buf);
+	amdgpu_bo_unref(&adev->smu.smu_buf);
+	kfree(adev->smu.priv);
+	adev->smu.priv = NULL;
+	if (adev->firmware.fw_buf)
+		amdgpu_ucode_fini_bo(adev);
+
+	return 0;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h
new file mode 100644
index 000000000..c031ff99f
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/tonga_smumgr.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef TONGA_SMUMGR_H
+#define TONGA_SMUMGR_H
+
+#include "tonga_ppsmc.h"
+
+int tonga_smu_init(struct amdgpu_device *adev);
+int tonga_smu_fini(struct amdgpu_device *adev);
+int tonga_smu_start(struct amdgpu_device *adev);
+
+struct tonga_smu_private_data
+{
+	uint8_t *header;
+	uint32_t smu_buffer_addr_high;
+	uint32_t smu_buffer_addr_low;
+	uint32_t header_addr_high;
+	uint32_t header_addr_low;
+};
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
new file mode 100644
index 000000000..5e9f73af8
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
@@ -0,0 +1,905 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "cikd.h"
+
+#include "uvd/uvd_4_2_d.h"
+#include "uvd/uvd_4_2_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static void uvd_v4_2_mc_resume(struct amdgpu_device *adev);
+static void uvd_v4_2_init_cg(struct amdgpu_device *adev);
+static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v4_2_start(struct amdgpu_device *adev);
+static void uvd_v4_2_stop(struct amdgpu_device *adev);
+
+/**
+ * uvd_v4_2_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint32_t uvd_v4_2_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v4_2_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint32_t uvd_v4_2_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v4_2_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v4_2_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, ring->wptr);
+}
+
+static int uvd_v4_2_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v4_2_set_ring_funcs(adev);
+	uvd_v4_2_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v4_2_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, 124, &adev->uvd.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->uvd.ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 4096, CP_PACKET2, 0xf,
+			     &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD);
+
+	return r;
+}
+
+static int uvd_v4_2_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * uvd_v4_2_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v4_2_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	uint32_t tmp;
+	int r;
+
+	/* raise clocks while booting up the VCPU */
+	amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+
+	r = uvd_v4_2_start(adev);
+	if (r)
+		goto done;
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	r = amdgpu_ring_lock(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_unlock_commit(ring);
+
+done:
+	/* lower clocks again */
+	amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+
+	if (!r)
+		DRM_INFO("UVD initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * uvd_v4_2_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v4_2_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+
+	uvd_v4_2_stop(adev);
+	ring->ready = false;
+
+	return 0;
+}
+
+static int uvd_v4_2_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	r = uvd_v4_2_hw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int uvd_v4_2_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	r = uvd_v4_2_hw_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * uvd_v4_2_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v4_2_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	uint32_t rb_bufsz;
+	int i, j, r;
+
+	/* disable byte swapping */
+	u32 lmi_swap_cntl = 0;
+	u32 mp_swap_cntl = 0;
+
+	uvd_v4_2_mc_resume(adev);
+
+	/* disable clock gating */
+	WREG32(mmUVD_CGC_GATE, 0);
+
+	/* disable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+			     (1 << 21) | (1 << 9) | (1 << 20));
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+
+	/* enable UMC */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	/* boot up the VCPU */
+	WREG32(mmUVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+
+	/* enable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 3<<1, ~(3 << 1));
+
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* programm the 4GB memory segment for rptr and ring buffer */
+	WREG32(mmUVD_LMI_EXT40_ADDR, upper_32_bits(ring->gpu_addr) |
+				   (0x7 << 16) | (0x1 << 31));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0x0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, ring->wptr);
+
+	/* set the ring address */
+	WREG32(mmUVD_RBC_RB_BASE, ring->gpu_addr);
+
+	/* Set ring buffer size */
+	rb_bufsz = order_base_2(ring->ring_size);
+	rb_bufsz = (0x1 << 8) | rb_bufsz;
+	WREG32_P(mmUVD_RBC_RB_CNTL, rb_bufsz, ~0x11f1f);
+
+	return 0;
+}
+
+/**
+ * uvd_v4_2_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v4_2_stop(struct amdgpu_device *adev)
+{
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put VCPU into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* disable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL, 0x0);
+
+	/* Unstall UMC and register bus */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+}
+
+/**
+ * uvd_v4_2_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v4_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v4_2_ring_emit_semaphore - emit semaphore command
+ *
+ * @ring: amdgpu_ring pointer
+ * @semaphore: semaphore to emit commands for
+ * @emit_wait: true if we should emit a wait command
+ *
+ * Emit a semaphore command (either wait or signal) to the UVD ring.
+ */
+static bool uvd_v4_2_ring_emit_semaphore(struct amdgpu_ring *ring,
+					 struct amdgpu_semaphore *semaphore,
+					 bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_ADDR_LOW, 0));
+	amdgpu_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_ADDR_HIGH, 0));
+	amdgpu_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CMD, 0));
+	amdgpu_ring_write(ring, 0x80 | (emit_wait ? 1 : 0));
+
+	return true;
+}
+
+/**
+ * uvd_v4_2_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_lock(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v4_2_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_BASE, 0));
+	amdgpu_ring_write(ring, ib->gpu_addr);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * uvd_v4_2_ring_test_ib - test ib execution
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully execute an IB
+ */
+static int uvd_v4_2_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct fence *fence = NULL;
+	int r;
+
+	r = amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to raise UVD clocks (%d).\n", r);
+		return r;
+	}
+
+	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_uvd_get_destroy_msg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
+		goto error;
+	}
+
+	r = fence_wait(fence, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto error;
+	}
+	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+error:
+	fence_put(fence);
+	amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+	return r;
+}
+
+/**
+ * uvd_v4_2_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v4_2_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t addr;
+	uint32_t size;
+
+	/* programm the VCPU memory controller bits 0-27 */
+	addr = (adev->uvd.gpu_addr + AMDGPU_UVD_FIRMWARE_OFFSET) >> 3;
+	size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4) >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	addr += size;
+	size = AMDGPU_UVD_STACK_SIZE >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	addr += size;
+	size = AMDGPU_UVD_HEAP_SIZE >> 3;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, addr);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+
+	/* bits 28-31 */
+	addr = (adev->uvd.gpu_addr >> 28) & 0xF;
+	WREG32(mmUVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0));
+
+	/* bits 32-39 */
+	addr = (adev->uvd.gpu_addr >> 32) & 0xFF;
+	WREG32(mmUVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31));
+
+	uvd_v4_2_init_cg(adev);
+}
+
+static void uvd_v4_2_enable_mgcg(struct amdgpu_device *adev,
+				 bool enable)
+{
+	u32 orig, data;
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_UVD_MGCG)) {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data = 0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	} else {
+		data = RREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL);
+		data &= ~0xfff;
+		WREG32_UVD_CTX(ixUVD_CGC_MEM_CTRL, data);
+
+		orig = data = RREG32(mmUVD_CGC_CTRL);
+		data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		if (orig != data)
+			WREG32(mmUVD_CGC_CTRL, data);
+	}
+}
+
+static void uvd_v4_2_set_dcm(struct amdgpu_device *adev,
+			     bool sw_mode)
+{
+	u32 tmp, tmp2;
+
+	tmp = RREG32(mmUVD_CGC_CTRL);
+	tmp &= ~(UVD_CGC_CTRL__CLK_OFF_DELAY_MASK | UVD_CGC_CTRL__CLK_GATE_DLY_TIMER_MASK);
+	tmp |= UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK |
+		(1 << UVD_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT) |
+		(4 << UVD_CGC_CTRL__CLK_OFF_DELAY__SHIFT);
+
+	if (sw_mode) {
+		tmp &= ~0x7ffff800;
+		tmp2 = UVD_CGC_CTRL2__DYN_OCLK_RAMP_EN_MASK |
+			UVD_CGC_CTRL2__DYN_RCLK_RAMP_EN_MASK |
+			(7 << UVD_CGC_CTRL2__GATER_DIV_ID__SHIFT);
+	} else {
+		tmp |= 0x7ffff800;
+		tmp2 = 0;
+	}
+
+	WREG32(mmUVD_CGC_CTRL, tmp);
+	WREG32_UVD_CTX(ixUVD_CGC_CTRL2, tmp2);
+}
+
+static void uvd_v4_2_init_cg(struct amdgpu_device *adev)
+{
+	bool hw_mode = true;
+
+	if (hw_mode) {
+		uvd_v4_2_set_dcm(adev, false);
+	} else {
+		u32 tmp = RREG32(mmUVD_CGC_CTRL);
+		tmp &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK;
+		WREG32(mmUVD_CGC_CTRL, tmp);
+	}
+}
+
+static bool uvd_v4_2_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v4_2_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int uvd_v4_2_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v4_2_stop(adev);
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK,
+			~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	return uvd_v4_2_start(adev);
+}
+
+static void uvd_v4_2_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	dev_info(adev->dev, "UVD 4.2 registers\n");
+	dev_info(adev->dev, "  UVD_SEMA_ADDR_LOW=0x%08X\n",
+		 RREG32(mmUVD_SEMA_ADDR_LOW));
+	dev_info(adev->dev, "  UVD_SEMA_ADDR_HIGH=0x%08X\n",
+		 RREG32(mmUVD_SEMA_ADDR_HIGH));
+	dev_info(adev->dev, "  UVD_SEMA_CMD=0x%08X\n",
+		 RREG32(mmUVD_SEMA_CMD));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_CMD=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_CMD));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_DATA0=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_DATA0));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_DATA1=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_DATA1));
+	dev_info(adev->dev, "  UVD_ENGINE_CNTL=0x%08X\n",
+		 RREG32(mmUVD_ENGINE_CNTL));
+	dev_info(adev->dev, "  UVD_UDEC_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DB_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DBW_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DBW_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_SEMA_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_CNTL));
+	dev_info(adev->dev, "  UVD_LMI_EXT40_ADDR=0x%08X\n",
+		 RREG32(mmUVD_LMI_EXT40_ADDR));
+	dev_info(adev->dev, "  UVD_CTX_INDEX=0x%08X\n",
+		 RREG32(mmUVD_CTX_INDEX));
+	dev_info(adev->dev, "  UVD_CTX_DATA=0x%08X\n",
+		 RREG32(mmUVD_CTX_DATA));
+	dev_info(adev->dev, "  UVD_CGC_GATE=0x%08X\n",
+		 RREG32(mmUVD_CGC_GATE));
+	dev_info(adev->dev, "  UVD_CGC_CTRL=0x%08X\n",
+		 RREG32(mmUVD_CGC_CTRL));
+	dev_info(adev->dev, "  UVD_LMI_CTRL2=0x%08X\n",
+		 RREG32(mmUVD_LMI_CTRL2));
+	dev_info(adev->dev, "  UVD_MASTINT_EN=0x%08X\n",
+		 RREG32(mmUVD_MASTINT_EN));
+	dev_info(adev->dev, "  UVD_LMI_ADDR_EXT=0x%08X\n",
+		 RREG32(mmUVD_LMI_ADDR_EXT));
+	dev_info(adev->dev, "  UVD_LMI_CTRL=0x%08X\n",
+		 RREG32(mmUVD_LMI_CTRL));
+	dev_info(adev->dev, "  UVD_LMI_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmUVD_LMI_SWAP_CNTL));
+	dev_info(adev->dev, "  UVD_MP_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmUVD_MP_SWAP_CNTL));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXA0=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXA0));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXA1=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXA1));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXB0=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXB0));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXB1=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXB1));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUX=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUX));
+	dev_info(adev->dev, "  UVD_MPC_SET_ALU=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_ALU));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET0=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET0));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE0=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE0));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET1=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET1));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE1=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE1));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET2=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET2));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE2=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE2));
+	dev_info(adev->dev, "  UVD_VCPU_CNTL=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CNTL));
+	dev_info(adev->dev, "  UVD_SOFT_RESET=0x%08X\n",
+		 RREG32(mmUVD_SOFT_RESET));
+	dev_info(adev->dev, "  UVD_RBC_IB_BASE=0x%08X\n",
+		 RREG32(mmUVD_RBC_IB_BASE));
+	dev_info(adev->dev, "  UVD_RBC_IB_SIZE=0x%08X\n",
+		 RREG32(mmUVD_RBC_IB_SIZE));
+	dev_info(adev->dev, "  UVD_RBC_RB_BASE=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_BASE));
+	dev_info(adev->dev, "  UVD_RBC_RB_RPTR=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_RPTR));
+	dev_info(adev->dev, "  UVD_RBC_RB_WPTR=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_WPTR));
+	dev_info(adev->dev, "  UVD_RBC_RB_WPTR_CNTL=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_WPTR_CNTL));
+	dev_info(adev->dev, "  UVD_RBC_RB_CNTL=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_CNTL));
+	dev_info(adev->dev, "  UVD_STATUS=0x%08X\n",
+		 RREG32(mmUVD_STATUS));
+	dev_info(adev->dev, "  UVD_SEMA_TIMEOUT_STATUS=0x%08X\n",
+		 RREG32(mmUVD_SEMA_TIMEOUT_STATUS));
+	dev_info(adev->dev, "  UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_CONTEXT_ID=0x%08X\n",
+		 RREG32(mmUVD_CONTEXT_ID));
+}
+
+static int uvd_v4_2_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v4_2_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: UVD TRAP\n");
+	amdgpu_fence_process(&adev->uvd.ring);
+	return 0;
+}
+
+static int uvd_v4_2_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	uvd_v4_2_enable_mgcg(adev, gate);
+
+	return 0;
+}
+
+static int uvd_v4_2_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	 struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v4_2_stop(adev);
+		return 0;
+	} else {
+		return uvd_v4_2_start(adev);
+	}
+}
+
+const struct amd_ip_funcs uvd_v4_2_ip_funcs = {
+	.early_init = uvd_v4_2_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v4_2_sw_init,
+	.sw_fini = uvd_v4_2_sw_fini,
+	.hw_init = uvd_v4_2_hw_init,
+	.hw_fini = uvd_v4_2_hw_fini,
+	.suspend = uvd_v4_2_suspend,
+	.resume = uvd_v4_2_resume,
+	.is_idle = uvd_v4_2_is_idle,
+	.wait_for_idle = uvd_v4_2_wait_for_idle,
+	.soft_reset = uvd_v4_2_soft_reset,
+	.print_status = uvd_v4_2_print_status,
+	.set_clockgating_state = uvd_v4_2_set_clockgating_state,
+	.set_powergating_state = uvd_v4_2_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
+	.get_rptr = uvd_v4_2_ring_get_rptr,
+	.get_wptr = uvd_v4_2_ring_get_wptr,
+	.set_wptr = uvd_v4_2_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_ib = uvd_v4_2_ring_emit_ib,
+	.emit_fence = uvd_v4_2_ring_emit_fence,
+	.emit_semaphore = uvd_v4_2_ring_emit_semaphore,
+	.test_ring = uvd_v4_2_ring_test_ring,
+	.test_ib = uvd_v4_2_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.ring.funcs = &uvd_v4_2_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v4_2_irq_funcs = {
+	.set = uvd_v4_2_set_interrupt_state,
+	.process = uvd_v4_2_process_interrupt,
+};
+
+static void uvd_v4_2_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.irq.num_types = 1;
+	adev->uvd.irq.funcs = &uvd_v4_2_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.h b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.h
new file mode 100644
index 000000000..0a615dd50
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V4_2_H__
+#define __UVD_V4_2_H__
+
+extern const struct amd_ip_funcs uvd_v4_2_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
new file mode 100644
index 000000000..38864f562
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
@@ -0,0 +1,844 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "vid.h"
+#include "uvd/uvd_5_0_d.h"
+#include "uvd/uvd_5_0_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v5_0_start(struct amdgpu_device *adev);
+static void uvd_v5_0_stop(struct amdgpu_device *adev);
+
+/**
+ * uvd_v5_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint32_t uvd_v5_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v5_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint32_t uvd_v5_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v5_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v5_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, ring->wptr);
+}
+
+static int uvd_v5_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v5_0_set_ring_funcs(adev);
+	uvd_v5_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v5_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	int r;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, 124, &adev->uvd.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->uvd.ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 4096, CP_PACKET2, 0xf,
+			     &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD);
+
+	return r;
+}
+
+static int uvd_v5_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * uvd_v5_0_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v5_0_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	uint32_t tmp;
+	int r;
+
+	/* raise clocks while booting up the VCPU */
+	amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+
+	r = uvd_v5_0_start(adev);
+	if (r)
+		goto done;
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	r = amdgpu_ring_lock(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_unlock_commit(ring);
+
+done:
+	/* lower clocks again */
+	amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+
+	if (!r)
+		DRM_INFO("UVD initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * uvd_v5_0_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v5_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+
+	uvd_v5_0_stop(adev);
+	ring->ready = false;
+
+	return 0;
+}
+
+static int uvd_v5_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	r = uvd_v5_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int uvd_v5_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	r = uvd_v5_0_hw_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * uvd_v5_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v5_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t offset;
+	uint32_t size;
+
+	/* programm memory controller bits 0-27 */
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->uvd.gpu_addr));
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->uvd.gpu_addr));
+
+	offset = AMDGPU_UVD_FIRMWARE_OFFSET;
+	size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	offset += size;
+	size = AMDGPU_UVD_STACK_SIZE;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	offset += size;
+	size = AMDGPU_UVD_HEAP_SIZE;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+}
+
+/**
+ * uvd_v5_0_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v5_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	uint32_t mp_swap_cntl;
+	int i, j, r;
+
+	/*disable DPG */
+	WREG32_P(mmUVD_POWER_STATUS, 0, ~(1 << 2));
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+	mp_swap_cntl = 0;
+
+	uvd_v5_0_mc_resume(adev);
+
+	/* disable clock gating */
+	WREG32(mmUVD_CGC_GATE, 0);
+
+	/* disable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+
+	/* stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+			     (1 << 21) | (1 << 9) | (1 << 20));
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+
+	/* enable UMC */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	/* boot up the VCPU */
+	WREG32(mmUVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmUVD_SOFT_RESET, 0, ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable master interrupt */
+	WREG32_P(mmUVD_MASTINT_EN, 3 << 1, ~(3 << 1));
+
+	/* clear the bit 4 of UVD_STATUS */
+	WREG32_P(mmUVD_STATUS, 0, ~(2 << 1));
+
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+			lower_32_bits(ring->gpu_addr));
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+			upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, ring->wptr);
+
+	WREG32_P(mmUVD_RBC_RB_CNTL, 0, ~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+	return 0;
+}
+
+/**
+ * uvd_v5_0_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v5_0_stop(struct amdgpu_device *adev)
+{
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put VCPU into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* disable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL, 0x0);
+
+	/* Unstall UMC and register bus */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+}
+
+/**
+ * uvd_v5_0_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v5_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v5_0_ring_emit_semaphore - emit semaphore command
+ *
+ * @ring: amdgpu_ring pointer
+ * @semaphore: semaphore to emit commands for
+ * @emit_wait: true if we should emit a wait command
+ *
+ * Emit a semaphore command (either wait or signal) to the UVD ring.
+ */
+static bool uvd_v5_0_ring_emit_semaphore(struct amdgpu_ring *ring,
+					 struct amdgpu_semaphore *semaphore,
+					 bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_ADDR_LOW, 0));
+	amdgpu_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_ADDR_HIGH, 0));
+	amdgpu_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CMD, 0));
+	amdgpu_ring_write(ring, 0x80 | (emit_wait ? 1 : 0));
+
+	return true;
+}
+
+/**
+ * uvd_v5_0_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_lock(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v5_0_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH, 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * uvd_v5_0_ring_test_ib - test ib execution
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully execute an IB
+ */
+static int uvd_v5_0_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	struct fence *fence = NULL;
+	int r;
+
+	r = amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to raise UVD clocks (%d).\n", r);
+		return r;
+	}
+
+	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_uvd_get_destroy_msg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
+		goto error;
+	}
+
+	r = fence_wait(fence, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto error;
+	}
+	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+error:
+	fence_put(fence);
+	amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+	return r;
+}
+
+static bool uvd_v5_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v5_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int uvd_v5_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v5_0_stop(adev);
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK,
+			~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	return uvd_v5_0_start(adev);
+}
+
+static void uvd_v5_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	dev_info(adev->dev, "UVD 5.0 registers\n");
+	dev_info(adev->dev, "  UVD_SEMA_ADDR_LOW=0x%08X\n",
+		 RREG32(mmUVD_SEMA_ADDR_LOW));
+	dev_info(adev->dev, "  UVD_SEMA_ADDR_HIGH=0x%08X\n",
+		 RREG32(mmUVD_SEMA_ADDR_HIGH));
+	dev_info(adev->dev, "  UVD_SEMA_CMD=0x%08X\n",
+		 RREG32(mmUVD_SEMA_CMD));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_CMD=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_CMD));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_DATA0=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_DATA0));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_DATA1=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_DATA1));
+	dev_info(adev->dev, "  UVD_ENGINE_CNTL=0x%08X\n",
+		 RREG32(mmUVD_ENGINE_CNTL));
+	dev_info(adev->dev, "  UVD_UDEC_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DB_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DBW_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DBW_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_SEMA_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_CNTL));
+	dev_info(adev->dev, "  UVD_LMI_EXT40_ADDR=0x%08X\n",
+		 RREG32(mmUVD_LMI_EXT40_ADDR));
+	dev_info(adev->dev, "  UVD_CTX_INDEX=0x%08X\n",
+		 RREG32(mmUVD_CTX_INDEX));
+	dev_info(adev->dev, "  UVD_CTX_DATA=0x%08X\n",
+		 RREG32(mmUVD_CTX_DATA));
+	dev_info(adev->dev, "  UVD_CGC_GATE=0x%08X\n",
+		 RREG32(mmUVD_CGC_GATE));
+	dev_info(adev->dev, "  UVD_CGC_CTRL=0x%08X\n",
+		 RREG32(mmUVD_CGC_CTRL));
+	dev_info(adev->dev, "  UVD_LMI_CTRL2=0x%08X\n",
+		 RREG32(mmUVD_LMI_CTRL2));
+	dev_info(adev->dev, "  UVD_MASTINT_EN=0x%08X\n",
+		 RREG32(mmUVD_MASTINT_EN));
+	dev_info(adev->dev, "  UVD_LMI_ADDR_EXT=0x%08X\n",
+		 RREG32(mmUVD_LMI_ADDR_EXT));
+	dev_info(adev->dev, "  UVD_LMI_CTRL=0x%08X\n",
+		 RREG32(mmUVD_LMI_CTRL));
+	dev_info(adev->dev, "  UVD_LMI_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmUVD_LMI_SWAP_CNTL));
+	dev_info(adev->dev, "  UVD_MP_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmUVD_MP_SWAP_CNTL));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXA0=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXA0));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXA1=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXA1));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXB0=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXB0));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXB1=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXB1));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUX=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUX));
+	dev_info(adev->dev, "  UVD_MPC_SET_ALU=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_ALU));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET0=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET0));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE0=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE0));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET1=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET1));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE1=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE1));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET2=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET2));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE2=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE2));
+	dev_info(adev->dev, "  UVD_VCPU_CNTL=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CNTL));
+	dev_info(adev->dev, "  UVD_SOFT_RESET=0x%08X\n",
+		 RREG32(mmUVD_SOFT_RESET));
+	dev_info(adev->dev, "  UVD_LMI_RBC_IB_64BIT_BAR_LOW=0x%08X\n",
+		 RREG32(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW));
+	dev_info(adev->dev, "  UVD_LMI_RBC_IB_64BIT_BAR_HIGH=0x%08X\n",
+		 RREG32(mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH));
+	dev_info(adev->dev, "  UVD_RBC_IB_SIZE=0x%08X\n",
+		 RREG32(mmUVD_RBC_IB_SIZE));
+	dev_info(adev->dev, "  UVD_LMI_RBC_RB_64BIT_BAR_LOW=0x%08X\n",
+		 RREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW));
+	dev_info(adev->dev, "  UVD_LMI_RBC_RB_64BIT_BAR_HIGH=0x%08X\n",
+		 RREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH));
+	dev_info(adev->dev, "  UVD_RBC_RB_RPTR=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_RPTR));
+	dev_info(adev->dev, "  UVD_RBC_RB_WPTR=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_WPTR));
+	dev_info(adev->dev, "  UVD_RBC_RB_WPTR_CNTL=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_WPTR_CNTL));
+	dev_info(adev->dev, "  UVD_RBC_RB_CNTL=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_CNTL));
+	dev_info(adev->dev, "  UVD_STATUS=0x%08X\n",
+		 RREG32(mmUVD_STATUS));
+	dev_info(adev->dev, "  UVD_SEMA_TIMEOUT_STATUS=0x%08X\n",
+		 RREG32(mmUVD_SEMA_TIMEOUT_STATUS));
+	dev_info(adev->dev, "  UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_CONTEXT_ID=0x%08X\n",
+		 RREG32(mmUVD_CONTEXT_ID));
+}
+
+static int uvd_v5_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v5_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: UVD TRAP\n");
+	amdgpu_fence_process(&adev->uvd.ring);
+	return 0;
+}
+
+static int uvd_v5_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int uvd_v5_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v5_0_stop(adev);
+		return 0;
+	} else {
+		return uvd_v5_0_start(adev);
+	}
+}
+
+const struct amd_ip_funcs uvd_v5_0_ip_funcs = {
+	.early_init = uvd_v5_0_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v5_0_sw_init,
+	.sw_fini = uvd_v5_0_sw_fini,
+	.hw_init = uvd_v5_0_hw_init,
+	.hw_fini = uvd_v5_0_hw_fini,
+	.suspend = uvd_v5_0_suspend,
+	.resume = uvd_v5_0_resume,
+	.is_idle = uvd_v5_0_is_idle,
+	.wait_for_idle = uvd_v5_0_wait_for_idle,
+	.soft_reset = uvd_v5_0_soft_reset,
+	.print_status = uvd_v5_0_print_status,
+	.set_clockgating_state = uvd_v5_0_set_clockgating_state,
+	.set_powergating_state = uvd_v5_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
+	.get_rptr = uvd_v5_0_ring_get_rptr,
+	.get_wptr = uvd_v5_0_ring_get_wptr,
+	.set_wptr = uvd_v5_0_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_ib = uvd_v5_0_ring_emit_ib,
+	.emit_fence = uvd_v5_0_ring_emit_fence,
+	.emit_semaphore = uvd_v5_0_ring_emit_semaphore,
+	.test_ring = uvd_v5_0_ring_test_ring,
+	.test_ib = uvd_v5_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.ring.funcs = &uvd_v5_0_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v5_0_irq_funcs = {
+	.set = uvd_v5_0_set_interrupt_state,
+	.process = uvd_v5_0_process_interrupt,
+};
+
+static void uvd_v5_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.irq.num_types = 1;
+	adev->uvd.irq.funcs = &uvd_v5_0_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.h
new file mode 100644
index 000000000..e3b3c49fa
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V5_0_H__
+#define __UVD_V5_0_H__
+
+extern const struct amd_ip_funcs uvd_v5_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
new file mode 100644
index 000000000..121915bbc
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
@@ -0,0 +1,828 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_uvd.h"
+#include "vid.h"
+#include "uvd/uvd_6_0_d.h"
+#include "uvd/uvd_6_0_sh_mask.h"
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev);
+static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev);
+static int uvd_v6_0_start(struct amdgpu_device *adev);
+static void uvd_v6_0_stop(struct amdgpu_device *adev);
+
+/**
+ * uvd_v6_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint32_t uvd_v6_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_RPTR);
+}
+
+/**
+ * uvd_v6_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint32_t uvd_v6_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	return RREG32(mmUVD_RBC_RB_WPTR);
+}
+
+/**
+ * uvd_v6_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void uvd_v6_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	WREG32(mmUVD_RBC_RB_WPTR, ring->wptr);
+}
+
+static int uvd_v6_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v6_0_set_ring_funcs(adev);
+	uvd_v6_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int uvd_v6_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* UVD TRAP */
+	r = amdgpu_irq_add_id(adev, 124, &adev->uvd.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_init(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->uvd.ring;
+	sprintf(ring->name, "uvd");
+	r = amdgpu_ring_init(adev, ring, 4096, CP_PACKET2, 0xf,
+			     &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD);
+
+	return r;
+}
+
+static int uvd_v6_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_uvd_suspend(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_uvd_sw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * uvd_v6_0_hw_init - start and test UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int uvd_v6_0_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	uint32_t tmp;
+	int r;
+
+	r = uvd_v6_0_start(adev);
+	if (r)
+		goto done;
+
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		goto done;
+	}
+
+	r = amdgpu_ring_lock(ring, 10);
+	if (r) {
+		DRM_ERROR("amdgpu: ring failed to lock UVD ring (%d).\n", r);
+		goto done;
+	}
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_TIMEOUT_STATUS, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CNTL, 0));
+	amdgpu_ring_write(ring, 3);
+
+	amdgpu_ring_unlock_commit(ring);
+
+done:
+	if (!r)
+		DRM_INFO("UVD initialized successfully.\n");
+
+	return r;
+}
+
+/**
+ * uvd_v6_0_hw_fini - stop the hardware block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop the UVD block, mark ring as not ready any more
+ */
+static int uvd_v6_0_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+
+	uvd_v6_0_stop(adev);
+	ring->ready = false;
+
+	return 0;
+}
+
+static int uvd_v6_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* Skip this for APU for now */
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_uvd_suspend(adev);
+		if (r)
+			return r;
+	}
+	r = uvd_v6_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int uvd_v6_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* Skip this for APU for now */
+	if (!(adev->flags & AMD_IS_APU)) {
+		r = amdgpu_uvd_resume(adev);
+		if (r)
+			return r;
+	}
+	r = uvd_v6_0_hw_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+/**
+ * uvd_v6_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Let the UVD memory controller know it's offsets
+ */
+static void uvd_v6_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t offset;
+	uint32_t size;
+
+	/* programm memory controller bits 0-27 */
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+			lower_32_bits(adev->uvd.gpu_addr));
+	WREG32(mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+			upper_32_bits(adev->uvd.gpu_addr));
+
+	offset = AMDGPU_UVD_FIRMWARE_OFFSET;
+	size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
+	WREG32(mmUVD_VCPU_CACHE_OFFSET0, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE0, size);
+
+	offset += size;
+	size = AMDGPU_UVD_STACK_SIZE;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET1, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE1, size);
+
+	offset += size;
+	size = AMDGPU_UVD_HEAP_SIZE;
+	WREG32(mmUVD_VCPU_CACHE_OFFSET2, offset >> 3);
+	WREG32(mmUVD_VCPU_CACHE_SIZE2, size);
+}
+
+/**
+ * uvd_v6_0_start - start UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the UVD block
+ */
+static int uvd_v6_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &adev->uvd.ring;
+	uint32_t rb_bufsz, tmp;
+	uint32_t lmi_swap_cntl;
+	uint32_t mp_swap_cntl;
+	int i, j, r;
+
+	/*disable DPG */
+	WREG32_P(mmUVD_POWER_STATUS, 0, ~(1 << 2));
+
+	/* disable byte swapping */
+	lmi_swap_cntl = 0;
+	mp_swap_cntl = 0;
+
+	uvd_v6_0_mc_resume(adev);
+
+	/* disable clock gating */
+	WREG32(mmUVD_CGC_GATE, 0);
+
+	/* disable interupt */
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+
+	/* stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(mmUVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+			     (1 << 21) | (1 << 9) | (1 << 20));
+
+#ifdef __BIG_ENDIAN
+	/* swap (8 in 32) RB and IB */
+	lmi_swap_cntl = 0xa;
+	mp_swap_cntl = 0;
+#endif
+	WREG32(mmUVD_LMI_SWAP_CNTL, lmi_swap_cntl);
+	WREG32(mmUVD_MP_SWAP_CNTL, mp_swap_cntl);
+
+	WREG32(mmUVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXA1, 0x0);
+	WREG32(mmUVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(mmUVD_MPC_SET_MUXB1, 0x0);
+	WREG32(mmUVD_MPC_SET_ALU, 0);
+	WREG32(mmUVD_MPC_SET_MUX, 0x88);
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+
+	/* enable UMC */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	/* boot up the VCPU */
+	WREG32(mmUVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmUVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK,
+				~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmUVD_SOFT_RESET, 0,
+			 ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable master interrupt */
+	WREG32_P(mmUVD_MASTINT_EN, 3 << 1, ~(3 << 1));
+
+	/* clear the bit 4 of UVD_STATUS */
+	WREG32_P(mmUVD_STATUS, 0, ~(2 << 1));
+
+	rb_bufsz = order_base_2(ring->ring_size);
+	tmp = 0;
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, tmp);
+
+	/* set the write pointer delay */
+	WREG32(mmUVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32(mmUVD_RBC_RB_RPTR_ADDR, (upper_32_bits(ring->gpu_addr) >> 2));
+
+	/* programm the RB_BASE for ring buffer */
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
+			lower_32_bits(ring->gpu_addr));
+	WREG32(mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
+			upper_32_bits(ring->gpu_addr));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(mmUVD_RBC_RB_RPTR, 0);
+
+	ring->wptr = RREG32(mmUVD_RBC_RB_RPTR);
+	WREG32(mmUVD_RBC_RB_WPTR, ring->wptr);
+
+	WREG32_P(mmUVD_RBC_RB_CNTL, 0, ~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK);
+
+	return 0;
+}
+
+/**
+ * uvd_v6_0_stop - stop UVD block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the UVD block
+ */
+static void uvd_v6_0_stop(struct amdgpu_device *adev)
+{
+	/* force RBC into idle state */
+	WREG32(mmUVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Stall UMC and register bus before resetting VCPU */
+	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	mdelay(1);
+
+	/* put VCPU into reset */
+	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
+	mdelay(5);
+
+	/* disable VCPU clock */
+	WREG32(mmUVD_VCPU_CNTL, 0x0);
+
+	/* Unstall UMC and register bus */
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+}
+
+/**
+ * uvd_v6_0_ring_emit_fence - emit an fence & trap command
+ *
+ * @ring: amdgpu_ring pointer
+ * @fence: fence to emit
+ *
+ * Write a fence and a trap command to the ring.
+ */
+static void uvd_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+				     unsigned flags)
+{
+	WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, addr & 0xffffffff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr) & 0xff);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 2);
+}
+
+/**
+ * uvd_v6_0_ring_emit_semaphore - emit semaphore command
+ *
+ * @ring: amdgpu_ring pointer
+ * @semaphore: semaphore to emit commands for
+ * @emit_wait: true if we should emit a wait command
+ *
+ * Emit a semaphore command (either wait or signal) to the UVD ring.
+ */
+static bool uvd_v6_0_ring_emit_semaphore(struct amdgpu_ring *ring,
+					 struct amdgpu_semaphore *semaphore,
+					 bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_ADDR_LOW, 0));
+	amdgpu_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_ADDR_HIGH, 0));
+	amdgpu_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_SEMA_CMD, 0));
+	amdgpu_ring_write(ring, 0x80 | (emit_wait ? 1 : 0));
+
+	return true;
+}
+
+/**
+ * uvd_v6_0_ring_test_ring - register write test
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully write to the context register
+ */
+static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(mmUVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = amdgpu_ring_lock(ring, 3);
+	if (r) {
+		DRM_ERROR("amdgpu: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_CONTEXT_ID, 0));
+	amdgpu_ring_write(ring, 0xDEADBEEF);
+	amdgpu_ring_unlock_commit(ring);
+	for (i = 0; i < adev->usec_timeout; i++) {
+		tmp = RREG32(mmUVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < adev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("amdgpu: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
+/**
+ * uvd_v6_0_ring_emit_ib - execute indirect buffer
+ *
+ * @ring: amdgpu_ring pointer
+ * @ib: indirect buffer to execute
+ *
+ * Write ring commands to execute the indirect buffer
+ */
+static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
+				  struct amdgpu_ib *ib)
+{
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
+	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH, 0));
+	amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_SIZE, 0));
+	amdgpu_ring_write(ring, ib->length_dw);
+}
+
+/**
+ * uvd_v6_0_ring_test_ib - test ib execution
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully execute an IB
+ */
+static int uvd_v6_0_ring_test_ib(struct amdgpu_ring *ring)
+{
+	struct fence *fence = NULL;
+	int r;
+
+	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_uvd_get_destroy_msg(ring, 1, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
+		goto error;
+	}
+
+	r = fence_wait(fence, false);
+	if (r) {
+		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+		goto error;
+	}
+	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+error:
+	fence_put(fence);
+	return r;
+}
+
+static bool uvd_v6_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK);
+}
+
+static int uvd_v6_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int uvd_v6_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	uvd_v6_0_stop(adev);
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK,
+			~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK);
+	mdelay(5);
+
+	return uvd_v6_0_start(adev);
+}
+
+static void uvd_v6_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	dev_info(adev->dev, "UVD 6.0 registers\n");
+	dev_info(adev->dev, "  UVD_SEMA_ADDR_LOW=0x%08X\n",
+		 RREG32(mmUVD_SEMA_ADDR_LOW));
+	dev_info(adev->dev, "  UVD_SEMA_ADDR_HIGH=0x%08X\n",
+		 RREG32(mmUVD_SEMA_ADDR_HIGH));
+	dev_info(adev->dev, "  UVD_SEMA_CMD=0x%08X\n",
+		 RREG32(mmUVD_SEMA_CMD));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_CMD=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_CMD));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_DATA0=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_DATA0));
+	dev_info(adev->dev, "  UVD_GPCOM_VCPU_DATA1=0x%08X\n",
+		 RREG32(mmUVD_GPCOM_VCPU_DATA1));
+	dev_info(adev->dev, "  UVD_ENGINE_CNTL=0x%08X\n",
+		 RREG32(mmUVD_ENGINE_CNTL));
+	dev_info(adev->dev, "  UVD_UDEC_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DB_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DB_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_UDEC_DBW_ADDR_CONFIG=0x%08X\n",
+		 RREG32(mmUVD_UDEC_DBW_ADDR_CONFIG));
+	dev_info(adev->dev, "  UVD_SEMA_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_CNTL));
+	dev_info(adev->dev, "  UVD_LMI_EXT40_ADDR=0x%08X\n",
+		 RREG32(mmUVD_LMI_EXT40_ADDR));
+	dev_info(adev->dev, "  UVD_CTX_INDEX=0x%08X\n",
+		 RREG32(mmUVD_CTX_INDEX));
+	dev_info(adev->dev, "  UVD_CTX_DATA=0x%08X\n",
+		 RREG32(mmUVD_CTX_DATA));
+	dev_info(adev->dev, "  UVD_CGC_GATE=0x%08X\n",
+		 RREG32(mmUVD_CGC_GATE));
+	dev_info(adev->dev, "  UVD_CGC_CTRL=0x%08X\n",
+		 RREG32(mmUVD_CGC_CTRL));
+	dev_info(adev->dev, "  UVD_LMI_CTRL2=0x%08X\n",
+		 RREG32(mmUVD_LMI_CTRL2));
+	dev_info(adev->dev, "  UVD_MASTINT_EN=0x%08X\n",
+		 RREG32(mmUVD_MASTINT_EN));
+	dev_info(adev->dev, "  UVD_LMI_ADDR_EXT=0x%08X\n",
+		 RREG32(mmUVD_LMI_ADDR_EXT));
+	dev_info(adev->dev, "  UVD_LMI_CTRL=0x%08X\n",
+		 RREG32(mmUVD_LMI_CTRL));
+	dev_info(adev->dev, "  UVD_LMI_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmUVD_LMI_SWAP_CNTL));
+	dev_info(adev->dev, "  UVD_MP_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmUVD_MP_SWAP_CNTL));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXA0=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXA0));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXA1=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXA1));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXB0=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXB0));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUXB1=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUXB1));
+	dev_info(adev->dev, "  UVD_MPC_SET_MUX=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_MUX));
+	dev_info(adev->dev, "  UVD_MPC_SET_ALU=0x%08X\n",
+		 RREG32(mmUVD_MPC_SET_ALU));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET0=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET0));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE0=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE0));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET1=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET1));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE1=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE1));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_OFFSET2=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_OFFSET2));
+	dev_info(adev->dev, "  UVD_VCPU_CACHE_SIZE2=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CACHE_SIZE2));
+	dev_info(adev->dev, "  UVD_VCPU_CNTL=0x%08X\n",
+		 RREG32(mmUVD_VCPU_CNTL));
+	dev_info(adev->dev, "  UVD_SOFT_RESET=0x%08X\n",
+		 RREG32(mmUVD_SOFT_RESET));
+	dev_info(adev->dev, "  UVD_RBC_IB_SIZE=0x%08X\n",
+		 RREG32(mmUVD_RBC_IB_SIZE));
+	dev_info(adev->dev, "  UVD_RBC_RB_RPTR=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_RPTR));
+	dev_info(adev->dev, "  UVD_RBC_RB_WPTR=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_WPTR));
+	dev_info(adev->dev, "  UVD_RBC_RB_WPTR_CNTL=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_WPTR_CNTL));
+	dev_info(adev->dev, "  UVD_RBC_RB_CNTL=0x%08X\n",
+		 RREG32(mmUVD_RBC_RB_CNTL));
+	dev_info(adev->dev, "  UVD_STATUS=0x%08X\n",
+		 RREG32(mmUVD_STATUS));
+	dev_info(adev->dev, "  UVD_SEMA_TIMEOUT_STATUS=0x%08X\n",
+		 RREG32(mmUVD_SEMA_TIMEOUT_STATUS));
+	dev_info(adev->dev, "  UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL=0x%08X\n",
+		 RREG32(mmUVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL));
+	dev_info(adev->dev, "  UVD_CONTEXT_ID=0x%08X\n",
+		 RREG32(mmUVD_CONTEXT_ID));
+}
+
+static int uvd_v6_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	// TODO
+	return 0;
+}
+
+static int uvd_v6_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: UVD TRAP\n");
+	amdgpu_fence_process(&adev->uvd.ring);
+	return 0;
+}
+
+static int uvd_v6_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int uvd_v6_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the UVD block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE) {
+		uvd_v6_0_stop(adev);
+		return 0;
+	} else {
+		return uvd_v6_0_start(adev);
+	}
+}
+
+const struct amd_ip_funcs uvd_v6_0_ip_funcs = {
+	.early_init = uvd_v6_0_early_init,
+	.late_init = NULL,
+	.sw_init = uvd_v6_0_sw_init,
+	.sw_fini = uvd_v6_0_sw_fini,
+	.hw_init = uvd_v6_0_hw_init,
+	.hw_fini = uvd_v6_0_hw_fini,
+	.suspend = uvd_v6_0_suspend,
+	.resume = uvd_v6_0_resume,
+	.is_idle = uvd_v6_0_is_idle,
+	.wait_for_idle = uvd_v6_0_wait_for_idle,
+	.soft_reset = uvd_v6_0_soft_reset,
+	.print_status = uvd_v6_0_print_status,
+	.set_clockgating_state = uvd_v6_0_set_clockgating_state,
+	.set_powergating_state = uvd_v6_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs uvd_v6_0_ring_funcs = {
+	.get_rptr = uvd_v6_0_ring_get_rptr,
+	.get_wptr = uvd_v6_0_ring_get_wptr,
+	.set_wptr = uvd_v6_0_ring_set_wptr,
+	.parse_cs = amdgpu_uvd_ring_parse_cs,
+	.emit_ib = uvd_v6_0_ring_emit_ib,
+	.emit_fence = uvd_v6_0_ring_emit_fence,
+	.emit_semaphore = uvd_v6_0_ring_emit_semaphore,
+	.test_ring = uvd_v6_0_ring_test_ring,
+	.test_ib = uvd_v6_0_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.ring.funcs = &uvd_v6_0_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs uvd_v6_0_irq_funcs = {
+	.set = uvd_v6_0_set_interrupt_state,
+	.process = uvd_v6_0_process_interrupt,
+};
+
+static void uvd_v6_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->uvd.irq.num_types = 1;
+	adev->uvd.irq.funcs = &uvd_v6_0_irq_funcs;
+}
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.h
new file mode 100644
index 000000000..6b92a2352
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __UVD_V6_0_H__
+#define __UVD_V6_0_H__
+
+extern const struct amd_ip_funcs uvd_v6_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
new file mode 100644
index 000000000..52ac7a8f1
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
@@ -0,0 +1,663 @@
+/*
+ * Copyright 2013 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_vce.h"
+#include "cikd.h"
+
+#include "vce/vce_2_0_d.h"
+#include "vce/vce_2_0_sh_mask.h"
+
+#include "oss/oss_2_0_d.h"
+#include "oss/oss_2_0_sh_mask.h"
+
+#define VCE_V2_0_FW_SIZE	(256 * 1024)
+#define VCE_V2_0_STACK_SIZE	(64 * 1024)
+#define VCE_V2_0_DATA_SIZE	(23552 * AMDGPU_MAX_VCE_HANDLES)
+
+static void vce_v2_0_mc_resume(struct amdgpu_device *adev);
+static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev);
+static void vce_v2_0_set_irq_funcs(struct amdgpu_device *adev);
+
+/**
+ * vce_v2_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint32_t vce_v2_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vce.ring[0])
+		return RREG32(mmVCE_RB_RPTR);
+	else
+		return RREG32(mmVCE_RB_RPTR2);
+}
+
+/**
+ * vce_v2_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint32_t vce_v2_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vce.ring[0])
+		return RREG32(mmVCE_RB_WPTR);
+	else
+		return RREG32(mmVCE_RB_WPTR2);
+}
+
+/**
+ * vce_v2_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vce_v2_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vce.ring[0])
+		WREG32(mmVCE_RB_WPTR, ring->wptr);
+	else
+		WREG32(mmVCE_RB_WPTR2, ring->wptr);
+}
+
+/**
+ * vce_v2_0_start - start VCE block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the VCE block
+ */
+static int vce_v2_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int i, j, r;
+
+	vce_v2_0_mc_resume(adev);
+
+	/* set BUSY flag */
+	WREG32_P(mmVCE_STATUS, 1, ~1);
+
+	ring = &adev->vce.ring[0];
+	WREG32(mmVCE_RB_RPTR, ring->wptr);
+	WREG32(mmVCE_RB_WPTR, ring->wptr);
+	WREG32(mmVCE_RB_BASE_LO, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE, ring->ring_size / 4);
+
+	ring = &adev->vce.ring[1];
+	WREG32(mmVCE_RB_RPTR2, ring->wptr);
+	WREG32(mmVCE_RB_WPTR2, ring->wptr);
+	WREG32(mmVCE_RB_BASE_LO2, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4);
+
+	WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK, ~VCE_VCPU_CNTL__CLK_EN_MASK);
+
+	WREG32_P(mmVCE_SOFT_RESET,
+		 VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+		 ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+
+	mdelay(100);
+
+	WREG32_P(mmVCE_SOFT_RESET, 0, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(mmVCE_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
+		WREG32_P(mmVCE_SOFT_RESET, VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmVCE_SOFT_RESET, 0, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+		r = -1;
+	}
+
+	/* clear BUSY flag */
+	WREG32_P(mmVCE_STATUS, 0, ~1);
+
+	if (r) {
+		DRM_ERROR("VCE not responding, giving up!!!\n");
+		return r;
+	}
+
+	return 0;
+}
+
+static int vce_v2_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	vce_v2_0_set_ring_funcs(adev);
+	vce_v2_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int vce_v2_0_sw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* VCE */
+	r = amdgpu_irq_add_id(adev, 167, &adev->vce.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_sw_init(adev, VCE_V2_0_FW_SIZE +
+		VCE_V2_0_STACK_SIZE + VCE_V2_0_DATA_SIZE);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->vce.ring[0];
+	sprintf(ring->name, "vce0");
+	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
+	if (r)
+		return r;
+
+	ring = &adev->vce.ring[1];
+	sprintf(ring->name, "vce1");
+	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v2_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_sw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v2_0_hw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v2_0_start(adev);
+	if (r)
+		return r;
+
+	ring = &adev->vce.ring[0];
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	ring = &adev->vce.ring[1];
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	DRM_INFO("VCE initialized successfully.\n");
+
+	return 0;
+}
+
+static int vce_v2_0_hw_fini(void *handle)
+{
+	return 0;
+}
+
+static int vce_v2_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v2_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v2_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	r = vce_v2_0_hw_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static void vce_v2_0_set_sw_cg(struct amdgpu_device *adev, bool gated)
+{
+	u32 tmp;
+
+	if (gated) {
+		tmp = RREG32(mmVCE_CLOCK_GATING_B);
+		tmp |= 0xe70000;
+		WREG32(mmVCE_CLOCK_GATING_B, tmp);
+
+		tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+		tmp |= 0xff000000;
+		WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+		tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+		tmp &= ~0x3fc;
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp);
+
+		WREG32(mmVCE_CGTT_CLK_OVERRIDE, 0);
+    } else {
+		tmp = RREG32(mmVCE_CLOCK_GATING_B);
+		tmp |= 0xe7;
+		tmp &= ~0xe70000;
+		WREG32(mmVCE_CLOCK_GATING_B, tmp);
+
+		tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+		tmp |= 0x1fe000;
+		tmp &= ~0xff000000;
+		WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+		tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+		tmp |= 0x3fc;
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp);
+	}
+}
+
+static void vce_v2_0_set_dyn_cg(struct amdgpu_device *adev, bool gated)
+{
+	u32 orig, tmp;
+
+	tmp = RREG32(mmVCE_CLOCK_GATING_B);
+	tmp &= ~0x00060006;
+	if (gated) {
+		tmp |= 0xe10000;
+	} else {
+		tmp |= 0xe1;
+		tmp &= ~0xe10000;
+	}
+	WREG32(mmVCE_CLOCK_GATING_B, tmp);
+
+	orig = tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+	tmp &= ~0x1fe000;
+	tmp &= ~0xff000000;
+	if (tmp != orig)
+		WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+	orig = tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING);
+	tmp &= ~0x3fc;
+	if (tmp != orig)
+		WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp);
+
+	if (gated)
+		WREG32(mmVCE_CGTT_CLK_OVERRIDE, 0);
+}
+
+static void vce_v2_0_disable_cg(struct amdgpu_device *adev)
+{
+	WREG32(mmVCE_CGTT_CLK_OVERRIDE, 7);
+}
+
+static void vce_v2_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
+{
+	bool sw_cg = false;
+
+	if (enable && (adev->cg_flags & AMDGPU_CG_SUPPORT_VCE_MGCG)) {
+		if (sw_cg)
+			vce_v2_0_set_sw_cg(adev, true);
+		else
+			vce_v2_0_set_dyn_cg(adev, true);
+	} else {
+		vce_v2_0_disable_cg(adev);
+
+		if (sw_cg)
+			vce_v2_0_set_sw_cg(adev, false);
+		else
+			vce_v2_0_set_dyn_cg(adev, false);
+	}
+}
+
+static void vce_v2_0_init_cg(struct amdgpu_device *adev)
+{
+	u32 tmp;
+
+	tmp = RREG32(mmVCE_CLOCK_GATING_A);
+	tmp &= ~0xfff;
+	tmp |= ((0 << 0) | (4 << 4));
+	tmp |= 0x40000;
+	WREG32(mmVCE_CLOCK_GATING_A, tmp);
+
+	tmp = RREG32(mmVCE_UENC_CLOCK_GATING);
+	tmp &= ~0xfff;
+	tmp |= ((0 << 0) | (4 << 4));
+	WREG32(mmVCE_UENC_CLOCK_GATING, tmp);
+
+	tmp = RREG32(mmVCE_CLOCK_GATING_B);
+	tmp |= 0x10;
+	tmp &= ~0x100000;
+	WREG32(mmVCE_CLOCK_GATING_B, tmp);
+}
+
+static void vce_v2_0_mc_resume(struct amdgpu_device *adev)
+{
+	uint64_t addr = adev->vce.gpu_addr;
+	uint32_t size;
+
+	WREG32_P(mmVCE_CLOCK_GATING_A, 0, ~(1 << 16));
+	WREG32_P(mmVCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000);
+	WREG32_P(mmVCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F);
+	WREG32(mmVCE_CLOCK_GATING_B, 0xf7);
+
+	WREG32(mmVCE_LMI_CTRL, 0x00398000);
+	WREG32_P(mmVCE_LMI_CACHE_CTRL, 0x0, ~0x1);
+	WREG32(mmVCE_LMI_SWAP_CNTL, 0);
+	WREG32(mmVCE_LMI_SWAP_CNTL1, 0);
+	WREG32(mmVCE_LMI_VM_CTRL, 0);
+
+	addr += AMDGPU_VCE_FIRMWARE_OFFSET;
+	size = VCE_V2_0_FW_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET0, addr & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE0, size);
+
+	addr += size;
+	size = VCE_V2_0_STACK_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET1, addr & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE1, size);
+
+	addr += size;
+	size = VCE_V2_0_DATA_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET2, addr & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE2, size);
+
+	WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100);
+
+	WREG32_P(mmVCE_SYS_INT_EN, VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK,
+		 ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+
+	vce_v2_0_init_cg(adev);
+}
+
+static bool vce_v2_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return !(RREG32(mmSRBM_STATUS2) & SRBM_STATUS2__VCE_BUSY_MASK);
+}
+
+static int vce_v2_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS2) & SRBM_STATUS2__VCE_BUSY_MASK))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int vce_v2_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_VCE_MASK,
+			~SRBM_SOFT_RESET__SOFT_RESET_VCE_MASK);
+	mdelay(5);
+
+	return vce_v2_0_start(adev);
+}
+
+static void vce_v2_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "VCE 2.0 registers\n");
+	dev_info(adev->dev, "  VCE_STATUS=0x%08X\n",
+		 RREG32(mmVCE_STATUS));
+	dev_info(adev->dev, "  VCE_VCPU_CNTL=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CNTL));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_OFFSET0=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_OFFSET0));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_SIZE0=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_SIZE0));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_OFFSET1=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_OFFSET1));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_SIZE1=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_SIZE1));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_OFFSET2=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_OFFSET2));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_SIZE2=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_SIZE2));
+	dev_info(adev->dev, "  VCE_SOFT_RESET=0x%08X\n",
+		 RREG32(mmVCE_SOFT_RESET));
+	dev_info(adev->dev, "  VCE_RB_BASE_LO2=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_LO2));
+	dev_info(adev->dev, "  VCE_RB_BASE_HI2=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_HI2));
+	dev_info(adev->dev, "  VCE_RB_SIZE2=0x%08X\n",
+		 RREG32(mmVCE_RB_SIZE2));
+	dev_info(adev->dev, "  VCE_RB_RPTR2=0x%08X\n",
+		 RREG32(mmVCE_RB_RPTR2));
+	dev_info(adev->dev, "  VCE_RB_WPTR2=0x%08X\n",
+		 RREG32(mmVCE_RB_WPTR2));
+	dev_info(adev->dev, "  VCE_RB_BASE_LO=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_LO));
+	dev_info(adev->dev, "  VCE_RB_BASE_HI=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_HI));
+	dev_info(adev->dev, "  VCE_RB_SIZE=0x%08X\n",
+		 RREG32(mmVCE_RB_SIZE));
+	dev_info(adev->dev, "  VCE_RB_RPTR=0x%08X\n",
+		 RREG32(mmVCE_RB_RPTR));
+	dev_info(adev->dev, "  VCE_RB_WPTR=0x%08X\n",
+		 RREG32(mmVCE_RB_WPTR));
+	dev_info(adev->dev, "  VCE_CLOCK_GATING_A=0x%08X\n",
+		 RREG32(mmVCE_CLOCK_GATING_A));
+	dev_info(adev->dev, "  VCE_CLOCK_GATING_B=0x%08X\n",
+		 RREG32(mmVCE_CLOCK_GATING_B));
+	dev_info(adev->dev, "  VCE_CGTT_CLK_OVERRIDE=0x%08X\n",
+		 RREG32(mmVCE_CGTT_CLK_OVERRIDE));
+	dev_info(adev->dev, "  VCE_UENC_CLOCK_GATING=0x%08X\n",
+		 RREG32(mmVCE_UENC_CLOCK_GATING));
+	dev_info(adev->dev, "  VCE_UENC_REG_CLOCK_GATING=0x%08X\n",
+		 RREG32(mmVCE_UENC_REG_CLOCK_GATING));
+	dev_info(adev->dev, "  VCE_SYS_INT_EN=0x%08X\n",
+		 RREG32(mmVCE_SYS_INT_EN));
+	dev_info(adev->dev, "  VCE_LMI_CTRL2=0x%08X\n",
+		 RREG32(mmVCE_LMI_CTRL2));
+	dev_info(adev->dev, "  VCE_LMI_CTRL=0x%08X\n",
+		 RREG32(mmVCE_LMI_CTRL));
+	dev_info(adev->dev, "  VCE_LMI_VM_CTRL=0x%08X\n",
+		 RREG32(mmVCE_LMI_VM_CTRL));
+	dev_info(adev->dev, "  VCE_LMI_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmVCE_LMI_SWAP_CNTL));
+	dev_info(adev->dev, "  VCE_LMI_SWAP_CNTL1=0x%08X\n",
+		 RREG32(mmVCE_LMI_SWAP_CNTL1));
+	dev_info(adev->dev, "  VCE_LMI_CACHE_CTRL=0x%08X\n",
+		 RREG32(mmVCE_LMI_CACHE_CTRL));
+}
+
+static int vce_v2_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	uint32_t val = 0;
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE)
+		val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
+
+	WREG32_P(mmVCE_SYS_INT_EN, val, ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+	return 0;
+}
+
+static int vce_v2_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCE\n");
+	switch (entry->src_data) {
+	case 0:
+		amdgpu_fence_process(&adev->vce.ring[0]);
+		break;
+	case 1:
+		amdgpu_fence_process(&adev->vce.ring[1]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data);
+		break;
+	}
+
+	return 0;
+}
+
+static int vce_v2_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	bool gate = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_CG_STATE_GATE)
+		gate = true;
+
+	vce_v2_0_enable_mgcg(adev, gate);
+
+	return 0;
+}
+
+static int vce_v2_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCE block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		/* XXX do we need a vce_v2_0_stop()? */
+		return 0;
+	else
+		return vce_v2_0_start(adev);
+}
+
+const struct amd_ip_funcs vce_v2_0_ip_funcs = {
+	.early_init = vce_v2_0_early_init,
+	.late_init = NULL,
+	.sw_init = vce_v2_0_sw_init,
+	.sw_fini = vce_v2_0_sw_fini,
+	.hw_init = vce_v2_0_hw_init,
+	.hw_fini = vce_v2_0_hw_fini,
+	.suspend = vce_v2_0_suspend,
+	.resume = vce_v2_0_resume,
+	.is_idle = vce_v2_0_is_idle,
+	.wait_for_idle = vce_v2_0_wait_for_idle,
+	.soft_reset = vce_v2_0_soft_reset,
+	.print_status = vce_v2_0_print_status,
+	.set_clockgating_state = vce_v2_0_set_clockgating_state,
+	.set_powergating_state = vce_v2_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs vce_v2_0_ring_funcs = {
+	.get_rptr = vce_v2_0_ring_get_rptr,
+	.get_wptr = vce_v2_0_ring_get_wptr,
+	.set_wptr = vce_v2_0_ring_set_wptr,
+	.parse_cs = amdgpu_vce_ring_parse_cs,
+	.emit_ib = amdgpu_vce_ring_emit_ib,
+	.emit_fence = amdgpu_vce_ring_emit_fence,
+	.emit_semaphore = amdgpu_vce_ring_emit_semaphore,
+	.test_ring = amdgpu_vce_ring_test_ring,
+	.test_ib = amdgpu_vce_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.ring[0].funcs = &vce_v2_0_ring_funcs;
+	adev->vce.ring[1].funcs = &vce_v2_0_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs vce_v2_0_irq_funcs = {
+	.set = vce_v2_0_set_interrupt_state,
+	.process = vce_v2_0_process_interrupt,
+};
+
+static void vce_v2_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.irq.num_types = 1;
+	adev->vce.irq.funcs = &vce_v2_0_irq_funcs;
+};
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.h
new file mode 100644
index 000000000..0d2ae8a01
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v2_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCE_V2_0_H__
+#define __VCE_V2_0_H__
+
+extern const struct amd_ip_funcs vce_v2_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
new file mode 100644
index 000000000..370c6c9d8
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -0,0 +1,679 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * Authors: Christian König <christian.koenig@amd.com>
+ */
+
+#include <linux/firmware.h>
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include "amdgpu_vce.h"
+#include "vid.h"
+#include "vce/vce_3_0_d.h"
+#include "vce/vce_3_0_sh_mask.h"
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+#include "gca/gfx_8_0_d.h"
+#include "smu/smu_7_1_2_d.h"
+#include "smu/smu_7_1_2_sh_mask.h"
+
+#define GRBM_GFX_INDEX__VCE_INSTANCE__SHIFT	0x04
+#define GRBM_GFX_INDEX__VCE_INSTANCE_MASK	0x10
+#define mmVCE_LMI_VCPU_CACHE_40BIT_BAR0 	0x8616
+#define mmVCE_LMI_VCPU_CACHE_40BIT_BAR1 	0x8617
+#define mmVCE_LMI_VCPU_CACHE_40BIT_BAR2 	0x8618
+
+#define VCE_V3_0_FW_SIZE	(384 * 1024)
+#define VCE_V3_0_STACK_SIZE	(64 * 1024)
+#define VCE_V3_0_DATA_SIZE	((16 * 1024 * AMDGPU_MAX_VCE_HANDLES) + (52 * 1024))
+
+static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx);
+static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev);
+static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+
+/**
+ * vce_v3_0_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint32_t vce_v3_0_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vce.ring[0])
+		return RREG32(mmVCE_RB_RPTR);
+	else
+		return RREG32(mmVCE_RB_RPTR2);
+}
+
+/**
+ * vce_v3_0_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint32_t vce_v3_0_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vce.ring[0])
+		return RREG32(mmVCE_RB_WPTR);
+	else
+		return RREG32(mmVCE_RB_WPTR2);
+}
+
+/**
+ * vce_v3_0_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void vce_v3_0_ring_set_wptr(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+
+	if (ring == &adev->vce.ring[0])
+		WREG32(mmVCE_RB_WPTR, ring->wptr);
+	else
+		WREG32(mmVCE_RB_WPTR2, ring->wptr);
+}
+
+/**
+ * vce_v3_0_start - start VCE block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the VCE block
+ */
+static int vce_v3_0_start(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring;
+	int idx, i, j, r;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (idx = 0; idx < 2; ++idx) {
+
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		if(idx == 0)
+			WREG32_P(mmGRBM_GFX_INDEX, 0,
+				~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
+		else
+			WREG32_P(mmGRBM_GFX_INDEX,
+				GRBM_GFX_INDEX__VCE_INSTANCE_MASK,
+				~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
+
+		vce_v3_0_mc_resume(adev, idx);
+
+		/* set BUSY flag */
+		WREG32_P(mmVCE_STATUS, 1, ~1);
+		if (adev->asic_type >= CHIP_STONEY)
+			WREG32_P(mmVCE_VCPU_CNTL, 1, ~0x200001);
+		else
+			WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK,
+				~VCE_VCPU_CNTL__CLK_EN_MASK);
+
+		WREG32_P(mmVCE_SOFT_RESET,
+			 VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+			 ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+
+		mdelay(100);
+
+		WREG32_P(mmVCE_SOFT_RESET, 0,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+
+		for (i = 0; i < 10; ++i) {
+			uint32_t status;
+			for (j = 0; j < 100; ++j) {
+				status = RREG32(mmVCE_STATUS);
+				if (status & 2)
+					break;
+				mdelay(10);
+			}
+			r = 0;
+			if (status & 2)
+				break;
+
+			DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
+			WREG32_P(mmVCE_SOFT_RESET,
+				VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+			mdelay(10);
+			WREG32_P(mmVCE_SOFT_RESET, 0,
+				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+			mdelay(10);
+			r = -1;
+		}
+
+		/* clear BUSY flag */
+		WREG32_P(mmVCE_STATUS, 0, ~1);
+
+		if (r) {
+			DRM_ERROR("VCE not responding, giving up!!!\n");
+			mutex_unlock(&adev->grbm_idx_mutex);
+			return r;
+		}
+	}
+
+	WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
+	mutex_unlock(&adev->grbm_idx_mutex);
+
+	ring = &adev->vce.ring[0];
+	WREG32(mmVCE_RB_RPTR, ring->wptr);
+	WREG32(mmVCE_RB_WPTR, ring->wptr);
+	WREG32(mmVCE_RB_BASE_LO, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE, ring->ring_size / 4);
+
+	ring = &adev->vce.ring[1];
+	WREG32(mmVCE_RB_RPTR2, ring->wptr);
+	WREG32(mmVCE_RB_WPTR2, ring->wptr);
+	WREG32(mmVCE_RB_BASE_LO2, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4);
+
+	return 0;
+}
+
+#define ixVCE_HARVEST_FUSE_MACRO__ADDRESS     0xC0014074
+#define VCE_HARVEST_FUSE_MACRO__SHIFT       27
+#define VCE_HARVEST_FUSE_MACRO__MASK        0x18000000
+
+static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev)
+{
+	u32 tmp;
+	unsigned ret;
+
+	/* Fiji, Stoney are single pipe */
+	if ((adev->asic_type == CHIP_FIJI) ||
+	    (adev->asic_type == CHIP_STONEY)){
+		ret = AMDGPU_VCE_HARVEST_VCE1;
+		return ret;
+	}
+
+	/* Tonga and CZ are dual or single pipe */
+	if (adev->flags & AMD_IS_APU)
+		tmp = (RREG32_SMC(ixVCE_HARVEST_FUSE_MACRO__ADDRESS) &
+		       VCE_HARVEST_FUSE_MACRO__MASK) >>
+			VCE_HARVEST_FUSE_MACRO__SHIFT;
+	else
+		tmp = (RREG32_SMC(ixCC_HARVEST_FUSES) &
+		       CC_HARVEST_FUSES__VCE_DISABLE_MASK) >>
+			CC_HARVEST_FUSES__VCE_DISABLE__SHIFT;
+
+	switch (tmp) {
+	case 1:
+		ret = AMDGPU_VCE_HARVEST_VCE0;
+		break;
+	case 2:
+		ret = AMDGPU_VCE_HARVEST_VCE1;
+		break;
+	case 3:
+		ret = AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1;
+		break;
+	default:
+		ret = 0;
+	}
+
+	return ret;
+}
+
+static int vce_v3_0_early_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	adev->vce.harvest_config = vce_v3_0_get_harvest_config(adev);
+
+	if ((adev->vce.harvest_config &
+	     (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1)) ==
+	    (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1))
+		return -ENOENT;
+
+	vce_v3_0_set_ring_funcs(adev);
+	vce_v3_0_set_irq_funcs(adev);
+
+	return 0;
+}
+
+static int vce_v3_0_sw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	struct amdgpu_ring *ring;
+	int r;
+
+	/* VCE */
+	r = amdgpu_irq_add_id(adev, 167, &adev->vce.irq);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_sw_init(adev, VCE_V3_0_FW_SIZE +
+		(VCE_V3_0_STACK_SIZE + VCE_V3_0_DATA_SIZE) * 2);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	ring = &adev->vce.ring[0];
+	sprintf(ring->name, "vce0");
+	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
+	if (r)
+		return r;
+
+	ring = &adev->vce.ring[1];
+	sprintf(ring->name, "vce1");
+	r = amdgpu_ring_init(adev, ring, 4096, VCE_CMD_NO_OP, 0xf,
+			     &adev->vce.irq, 0, AMDGPU_RING_TYPE_VCE);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v3_0_sw_fini(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_sw_fini(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v3_0_hw_init(void *handle)
+{
+	struct amdgpu_ring *ring;
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v3_0_start(adev);
+	if (r)
+		return r;
+
+	ring = &adev->vce.ring[0];
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	ring = &adev->vce.ring[1];
+	ring->ready = true;
+	r = amdgpu_ring_test_ring(ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	DRM_INFO("VCE initialized successfully.\n");
+
+	return 0;
+}
+
+static int vce_v3_0_hw_fini(void *handle)
+{
+	return 0;
+}
+
+static int vce_v3_0_suspend(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v3_0_hw_fini(adev);
+	if (r)
+		return r;
+
+	r = amdgpu_vce_suspend(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static int vce_v3_0_resume(void *handle)
+{
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = amdgpu_vce_resume(adev);
+	if (r)
+		return r;
+
+	r = vce_v3_0_hw_init(adev);
+	if (r)
+		return r;
+
+	return r;
+}
+
+static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx)
+{
+	uint32_t offset, size;
+
+	WREG32_P(mmVCE_CLOCK_GATING_A, 0, ~(1 << 16));
+	WREG32_P(mmVCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000);
+	WREG32_P(mmVCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F);
+	WREG32(mmVCE_CLOCK_GATING_B, 0xf7);
+
+	WREG32(mmVCE_LMI_CTRL, 0x00398000);
+	WREG32_P(mmVCE_LMI_CACHE_CTRL, 0x0, ~0x1);
+	WREG32(mmVCE_LMI_SWAP_CNTL, 0);
+	WREG32(mmVCE_LMI_SWAP_CNTL1, 0);
+	WREG32(mmVCE_LMI_VM_CTRL, 0);
+	if (adev->asic_type >= CHIP_STONEY) {
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR0, (adev->vce.gpu_addr >> 8));
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR1, (adev->vce.gpu_addr >> 8));
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR2, (adev->vce.gpu_addr >> 8));
+	} else
+		WREG32(mmVCE_LMI_VCPU_CACHE_40BIT_BAR, (adev->vce.gpu_addr >> 8));
+	offset = AMDGPU_VCE_FIRMWARE_OFFSET;
+	size = VCE_V3_0_FW_SIZE;
+	WREG32(mmVCE_VCPU_CACHE_OFFSET0, offset & 0x7fffffff);
+	WREG32(mmVCE_VCPU_CACHE_SIZE0, size);
+
+	if (idx == 0) {
+		offset += size;
+		size = VCE_V3_0_STACK_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET1, offset & 0x7fffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE1, size);
+		offset += size;
+		size = VCE_V3_0_DATA_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET2, offset & 0x7fffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE2, size);
+	} else {
+		offset += size + VCE_V3_0_STACK_SIZE + VCE_V3_0_DATA_SIZE;
+		size = VCE_V3_0_STACK_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET1, offset & 0xfffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE1, size);
+		offset += size;
+		size = VCE_V3_0_DATA_SIZE;
+		WREG32(mmVCE_VCPU_CACHE_OFFSET2, offset & 0xfffffff);
+		WREG32(mmVCE_VCPU_CACHE_SIZE2, size);
+	}
+
+	WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100);
+
+	WREG32_P(mmVCE_SYS_INT_EN, VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK,
+		 ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+}
+
+static bool vce_v3_0_is_idle(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 mask = 0;
+	int idx;
+
+	for (idx = 0; idx < 2; ++idx) {
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		if (idx == 0)
+			mask |= SRBM_STATUS2__VCE0_BUSY_MASK;
+		else
+			mask |= SRBM_STATUS2__VCE1_BUSY_MASK;
+	}
+
+	return !(RREG32(mmSRBM_STATUS2) & mask);
+}
+
+static int vce_v3_0_wait_for_idle(void *handle)
+{
+	unsigned i;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 mask = 0;
+	int idx;
+
+	for (idx = 0; idx < 2; ++idx) {
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		if (idx == 0)
+			mask |= SRBM_STATUS2__VCE0_BUSY_MASK;
+		else
+			mask |= SRBM_STATUS2__VCE1_BUSY_MASK;
+	}
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (!(RREG32(mmSRBM_STATUS2) & mask))
+			return 0;
+	}
+	return -ETIMEDOUT;
+}
+
+static int vce_v3_0_soft_reset(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	u32 mask = 0;
+	int idx;
+
+	for (idx = 0; idx < 2; ++idx) {
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		if (idx == 0)
+			mask |= SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK;
+		else
+			mask |= SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK;
+	}
+	WREG32_P(mmSRBM_SOFT_RESET, mask,
+		 ~(SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK |
+		   SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK));
+	mdelay(5);
+
+	return vce_v3_0_start(adev);
+}
+
+static void vce_v3_0_print_status(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	dev_info(adev->dev, "VCE 3.0 registers\n");
+	dev_info(adev->dev, "  VCE_STATUS=0x%08X\n",
+		 RREG32(mmVCE_STATUS));
+	dev_info(adev->dev, "  VCE_VCPU_CNTL=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CNTL));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_OFFSET0=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_OFFSET0));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_SIZE0=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_SIZE0));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_OFFSET1=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_OFFSET1));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_SIZE1=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_SIZE1));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_OFFSET2=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_OFFSET2));
+	dev_info(adev->dev, "  VCE_VCPU_CACHE_SIZE2=0x%08X\n",
+		 RREG32(mmVCE_VCPU_CACHE_SIZE2));
+	dev_info(adev->dev, "  VCE_SOFT_RESET=0x%08X\n",
+		 RREG32(mmVCE_SOFT_RESET));
+	dev_info(adev->dev, "  VCE_RB_BASE_LO2=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_LO2));
+	dev_info(adev->dev, "  VCE_RB_BASE_HI2=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_HI2));
+	dev_info(adev->dev, "  VCE_RB_SIZE2=0x%08X\n",
+		 RREG32(mmVCE_RB_SIZE2));
+	dev_info(adev->dev, "  VCE_RB_RPTR2=0x%08X\n",
+		 RREG32(mmVCE_RB_RPTR2));
+	dev_info(adev->dev, "  VCE_RB_WPTR2=0x%08X\n",
+		 RREG32(mmVCE_RB_WPTR2));
+	dev_info(adev->dev, "  VCE_RB_BASE_LO=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_LO));
+	dev_info(adev->dev, "  VCE_RB_BASE_HI=0x%08X\n",
+		 RREG32(mmVCE_RB_BASE_HI));
+	dev_info(adev->dev, "  VCE_RB_SIZE=0x%08X\n",
+		 RREG32(mmVCE_RB_SIZE));
+	dev_info(adev->dev, "  VCE_RB_RPTR=0x%08X\n",
+		 RREG32(mmVCE_RB_RPTR));
+	dev_info(adev->dev, "  VCE_RB_WPTR=0x%08X\n",
+		 RREG32(mmVCE_RB_WPTR));
+	dev_info(adev->dev, "  VCE_CLOCK_GATING_A=0x%08X\n",
+		 RREG32(mmVCE_CLOCK_GATING_A));
+	dev_info(adev->dev, "  VCE_CLOCK_GATING_B=0x%08X\n",
+		 RREG32(mmVCE_CLOCK_GATING_B));
+	dev_info(adev->dev, "  VCE_UENC_CLOCK_GATING=0x%08X\n",
+		 RREG32(mmVCE_UENC_CLOCK_GATING));
+	dev_info(adev->dev, "  VCE_UENC_REG_CLOCK_GATING=0x%08X\n",
+		 RREG32(mmVCE_UENC_REG_CLOCK_GATING));
+	dev_info(adev->dev, "  VCE_SYS_INT_EN=0x%08X\n",
+		 RREG32(mmVCE_SYS_INT_EN));
+	dev_info(adev->dev, "  VCE_LMI_CTRL2=0x%08X\n",
+		 RREG32(mmVCE_LMI_CTRL2));
+	dev_info(adev->dev, "  VCE_LMI_CTRL=0x%08X\n",
+		 RREG32(mmVCE_LMI_CTRL));
+	dev_info(adev->dev, "  VCE_LMI_VM_CTRL=0x%08X\n",
+		 RREG32(mmVCE_LMI_VM_CTRL));
+	dev_info(adev->dev, "  VCE_LMI_SWAP_CNTL=0x%08X\n",
+		 RREG32(mmVCE_LMI_SWAP_CNTL));
+	dev_info(adev->dev, "  VCE_LMI_SWAP_CNTL1=0x%08X\n",
+		 RREG32(mmVCE_LMI_SWAP_CNTL1));
+	dev_info(adev->dev, "  VCE_LMI_CACHE_CTRL=0x%08X\n",
+		 RREG32(mmVCE_LMI_CACHE_CTRL));
+}
+
+static int vce_v3_0_set_interrupt_state(struct amdgpu_device *adev,
+					struct amdgpu_irq_src *source,
+					unsigned type,
+					enum amdgpu_interrupt_state state)
+{
+	uint32_t val = 0;
+
+	if (state == AMDGPU_IRQ_STATE_ENABLE)
+		val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
+
+	WREG32_P(mmVCE_SYS_INT_EN, val, ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+	return 0;
+}
+
+static int vce_v3_0_process_interrupt(struct amdgpu_device *adev,
+				      struct amdgpu_irq_src *source,
+				      struct amdgpu_iv_entry *entry)
+{
+	DRM_DEBUG("IH: VCE\n");
+
+	WREG32_P(mmVCE_SYS_INT_STATUS,
+		VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK,
+		~VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK);
+
+	switch (entry->src_data) {
+	case 0:
+		amdgpu_fence_process(&adev->vce.ring[0]);
+		break;
+	case 1:
+		amdgpu_fence_process(&adev->vce.ring[1]);
+		break;
+	default:
+		DRM_ERROR("Unhandled interrupt: %d %d\n",
+			  entry->src_id, entry->src_data);
+		break;
+	}
+
+	return 0;
+}
+
+static int vce_v3_0_set_clockgating_state(void *handle,
+					  enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int vce_v3_0_set_powergating_state(void *handle,
+					  enum amd_powergating_state state)
+{
+	/* This doesn't actually powergate the VCE block.
+	 * That's done in the dpm code via the SMC.  This
+	 * just re-inits the block as necessary.  The actual
+	 * gating still happens in the dpm code.  We should
+	 * revisit this when there is a cleaner line between
+	 * the smc and the hw blocks
+	 */
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (state == AMD_PG_STATE_GATE)
+		/* XXX do we need a vce_v3_0_stop()? */
+		return 0;
+	else
+		return vce_v3_0_start(adev);
+}
+
+const struct amd_ip_funcs vce_v3_0_ip_funcs = {
+	.early_init = vce_v3_0_early_init,
+	.late_init = NULL,
+	.sw_init = vce_v3_0_sw_init,
+	.sw_fini = vce_v3_0_sw_fini,
+	.hw_init = vce_v3_0_hw_init,
+	.hw_fini = vce_v3_0_hw_fini,
+	.suspend = vce_v3_0_suspend,
+	.resume = vce_v3_0_resume,
+	.is_idle = vce_v3_0_is_idle,
+	.wait_for_idle = vce_v3_0_wait_for_idle,
+	.soft_reset = vce_v3_0_soft_reset,
+	.print_status = vce_v3_0_print_status,
+	.set_clockgating_state = vce_v3_0_set_clockgating_state,
+	.set_powergating_state = vce_v3_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs vce_v3_0_ring_funcs = {
+	.get_rptr = vce_v3_0_ring_get_rptr,
+	.get_wptr = vce_v3_0_ring_get_wptr,
+	.set_wptr = vce_v3_0_ring_set_wptr,
+	.parse_cs = amdgpu_vce_ring_parse_cs,
+	.emit_ib = amdgpu_vce_ring_emit_ib,
+	.emit_fence = amdgpu_vce_ring_emit_fence,
+	.emit_semaphore = amdgpu_vce_ring_emit_semaphore,
+	.test_ring = amdgpu_vce_ring_test_ring,
+	.test_ib = amdgpu_vce_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+};
+
+static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.ring[0].funcs = &vce_v3_0_ring_funcs;
+	adev->vce.ring[1].funcs = &vce_v3_0_ring_funcs;
+}
+
+static const struct amdgpu_irq_src_funcs vce_v3_0_irq_funcs = {
+	.set = vce_v3_0_set_interrupt_state,
+	.process = vce_v3_0_process_interrupt,
+};
+
+static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+	adev->vce.irq.num_types = 1;
+	adev->vce.irq.funcs = &vce_v3_0_irq_funcs;
+};
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.h b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.h
new file mode 100644
index 000000000..b45af65da
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vce_v3_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCE_V3_0_H__
+#define __VCE_V3_0_H__
+
+extern const struct amd_ip_funcs vce_v3_0_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vi.c b/kernel/drivers/gpu/drm/amd/amdgpu/vi.c
new file mode 100644
index 000000000..3e9cbe398
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -0,0 +1,1547 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include "drmP.h"
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "amdgpu_ih.h"
+#include "amdgpu_uvd.h"
+#include "amdgpu_vce.h"
+#include "amdgpu_ucode.h"
+#include "atom.h"
+
+#include "gmc/gmc_8_1_d.h"
+#include "gmc/gmc_8_1_sh_mask.h"
+
+#include "oss/oss_3_0_d.h"
+#include "oss/oss_3_0_sh_mask.h"
+
+#include "bif/bif_5_0_d.h"
+#include "bif/bif_5_0_sh_mask.h"
+
+#include "gca/gfx_8_0_d.h"
+#include "gca/gfx_8_0_sh_mask.h"
+
+#include "smu/smu_7_1_1_d.h"
+#include "smu/smu_7_1_1_sh_mask.h"
+
+#include "uvd/uvd_5_0_d.h"
+#include "uvd/uvd_5_0_sh_mask.h"
+
+#include "vce/vce_3_0_d.h"
+#include "vce/vce_3_0_sh_mask.h"
+
+#include "dce/dce_10_0_d.h"
+#include "dce/dce_10_0_sh_mask.h"
+
+#include "vid.h"
+#include "vi.h"
+#include "vi_dpm.h"
+#include "gmc_v8_0.h"
+#include "gmc_v7_0.h"
+#include "gfx_v8_0.h"
+#include "sdma_v2_4.h"
+#include "sdma_v3_0.h"
+#include "dce_v10_0.h"
+#include "dce_v11_0.h"
+#include "iceland_ih.h"
+#include "tonga_ih.h"
+#include "cz_ih.h"
+#include "uvd_v5_0.h"
+#include "uvd_v6_0.h"
+#include "vce_v3_0.h"
+
+/*
+ * Indirect registers accessor
+ */
+static u32 vi_pcie_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	r = RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+static void vi_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(mmPCIE_INDEX, reg);
+	(void)RREG32(mmPCIE_INDEX);
+	WREG32(mmPCIE_DATA, v);
+	(void)RREG32(mmPCIE_DATA);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}
+
+static u32 vi_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	r = RREG32(mmSMC_IND_DATA_0);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void vi_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmSMC_IND_INDEX_0, (reg));
+	WREG32(mmSMC_IND_DATA_0, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+/* smu_8_0_d.h */
+#define mmMP0PUB_IND_INDEX                                                      0x180
+#define mmMP0PUB_IND_DATA                                                       0x181
+
+static u32 cz_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmMP0PUB_IND_INDEX, (reg));
+	r = RREG32(mmMP0PUB_IND_DATA);
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+	return r;
+}
+
+static void cz_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->smc_idx_lock, flags);
+	WREG32(mmMP0PUB_IND_INDEX, (reg));
+	WREG32(mmMP0PUB_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
+static u32 vi_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	r = RREG32(mmUVD_CTX_DATA);
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+	return r;
+}
+
+static void vi_uvd_ctx_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->uvd_ctx_idx_lock, flags);
+	WREG32(mmUVD_CTX_INDEX, ((reg) & 0x1ff));
+	WREG32(mmUVD_CTX_DATA, (v));
+	spin_unlock_irqrestore(&adev->uvd_ctx_idx_lock, flags);
+}
+
+static u32 vi_didt_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	r = RREG32(mmDIDT_IND_DATA);
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+	return r;
+}
+
+static void vi_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->didt_idx_lock, flags);
+	WREG32(mmDIDT_IND_INDEX, (reg));
+	WREG32(mmDIDT_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
+}
+
+static const u32 tonga_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, 0xC060000C,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 fiji_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, 0xC060000C,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 iceland_mgcg_cgcg_init[] =
+{
+	mmPCIE_INDEX, 0xffffffff, ixPCIE_CNTL2,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmSMC_IND_INDEX_4, 0xffffffff, ixCGTT_ROM_CLK_CTRL0,
+	mmSMC_IND_DATA_4, 0xc0000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 cz_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
+	mmPCIE_INDEX, 0xffffffff, 0x0140001c,
+	mmPCIE_DATA, 0x000f0000, 0x00000000,
+	mmCGTT_DRM_CLK_CTRL0, 0xff000fff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xc0000fff, 0x00000104,
+};
+
+static const u32 stoney_mgcg_cgcg_init[] =
+{
+	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00000100,
+	mmHDP_XDP_CGTT_BLK_CTRL, 0xffffffff, 0x00000104,
+	mmHDP_HOST_PATH_CNTL, 0xffffffff, 0x0f000027,
+};
+
+static void vi_init_golden_registers(struct amdgpu_device *adev)
+{
+	/* Some of the registers might be dependent on GRBM_GFX_INDEX */
+	mutex_lock(&adev->grbm_idx_mutex);
+
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		amdgpu_program_register_sequence(adev,
+						 iceland_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(iceland_mgcg_cgcg_init));
+		break;
+	case CHIP_FIJI:
+		amdgpu_program_register_sequence(adev,
+						 fiji_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(fiji_mgcg_cgcg_init));
+		break;
+	case CHIP_TONGA:
+		amdgpu_program_register_sequence(adev,
+						 tonga_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(tonga_mgcg_cgcg_init));
+		break;
+	case CHIP_CARRIZO:
+		amdgpu_program_register_sequence(adev,
+						 cz_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(cz_mgcg_cgcg_init));
+		break;
+	case CHIP_STONEY:
+		amdgpu_program_register_sequence(adev,
+						 stoney_mgcg_cgcg_init,
+						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		break;
+	default:
+		break;
+	}
+	mutex_unlock(&adev->grbm_idx_mutex);
+}
+
+/**
+ * vi_get_xclk - get the xclk
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns the reference clock used by the gfx engine
+ * (VI).
+ */
+static u32 vi_get_xclk(struct amdgpu_device *adev)
+{
+	u32 reference_clock = adev->clock.spll.reference_freq;
+	u32 tmp;
+
+	if (adev->flags & AMD_IS_APU)
+		return reference_clock;
+
+	tmp = RREG32_SMC(ixCG_CLKPIN_CNTL_2);
+	if (REG_GET_FIELD(tmp, CG_CLKPIN_CNTL_2, MUX_TCLK_TO_XCLK))
+		return 1000;
+
+	tmp = RREG32_SMC(ixCG_CLKPIN_CNTL);
+	if (REG_GET_FIELD(tmp, CG_CLKPIN_CNTL, XTALIN_DIVIDE))
+		return reference_clock / 4;
+
+	return reference_clock;
+}
+
+/**
+ * vi_srbm_select - select specific register instances
+ *
+ * @adev: amdgpu_device pointer
+ * @me: selected ME (micro engine)
+ * @pipe: pipe
+ * @queue: queue
+ * @vmid: VMID
+ *
+ * Switches the currently active registers instances.  Some
+ * registers are instanced per VMID, others are instanced per
+ * me/pipe/queue combination.
+ */
+void vi_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid)
+{
+	u32 srbm_gfx_cntl = 0;
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, PIPEID, pipe);
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, MEID, me);
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, VMID, vmid);
+	srbm_gfx_cntl = REG_SET_FIELD(srbm_gfx_cntl, SRBM_GFX_CNTL, QUEUEID, queue);
+	WREG32(mmSRBM_GFX_CNTL, srbm_gfx_cntl);
+}
+
+static void vi_vga_set_state(struct amdgpu_device *adev, bool state)
+{
+	/* todo */
+}
+
+static bool vi_read_disabled_bios(struct amdgpu_device *adev)
+{
+	u32 bus_cntl;
+	u32 d1vga_control = 0;
+	u32 d2vga_control = 0;
+	u32 vga_render_control = 0;
+	u32 rom_cntl;
+	bool r;
+
+	bus_cntl = RREG32(mmBUS_CNTL);
+	if (adev->mode_info.num_crtc) {
+		d1vga_control = RREG32(mmD1VGA_CONTROL);
+		d2vga_control = RREG32(mmD2VGA_CONTROL);
+		vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
+	}
+	rom_cntl = RREG32_SMC(ixROM_CNTL);
+
+	/* enable the rom */
+	WREG32(mmBUS_CNTL, (bus_cntl & ~BUS_CNTL__BIOS_ROM_DIS_MASK));
+	if (adev->mode_info.num_crtc) {
+		/* Disable VGA mode */
+		WREG32(mmD1VGA_CONTROL,
+		       (d1vga_control & ~(D1VGA_CONTROL__D1VGA_MODE_ENABLE_MASK |
+					  D1VGA_CONTROL__D1VGA_TIMING_SELECT_MASK)));
+		WREG32(mmD2VGA_CONTROL,
+		       (d2vga_control & ~(D2VGA_CONTROL__D2VGA_MODE_ENABLE_MASK |
+					  D2VGA_CONTROL__D2VGA_TIMING_SELECT_MASK)));
+		WREG32(mmVGA_RENDER_CONTROL,
+		       (vga_render_control & ~VGA_RENDER_CONTROL__VGA_VSTATUS_CNTL_MASK));
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl | ROM_CNTL__SCK_OVERWRITE_MASK);
+
+	r = amdgpu_read_bios(adev);
+
+	/* restore regs */
+	WREG32(mmBUS_CNTL, bus_cntl);
+	if (adev->mode_info.num_crtc) {
+		WREG32(mmD1VGA_CONTROL, d1vga_control);
+		WREG32(mmD2VGA_CONTROL, d2vga_control);
+		WREG32(mmVGA_RENDER_CONTROL, vga_render_control);
+	}
+	WREG32_SMC(ixROM_CNTL, rom_cntl);
+	return r;
+}
+static struct amdgpu_allowed_register_entry tonga_allowed_read_registers[] = {
+	{mmGB_MACROTILE_MODE7, true},
+};
+
+static struct amdgpu_allowed_register_entry cz_allowed_read_registers[] = {
+	{mmGB_TILE_MODE7, true},
+	{mmGB_TILE_MODE12, true},
+	{mmGB_TILE_MODE17, true},
+	{mmGB_TILE_MODE23, true},
+	{mmGB_MACROTILE_MODE7, true},
+};
+
+static struct amdgpu_allowed_register_entry vi_allowed_read_registers[] = {
+	{mmGRBM_STATUS, false},
+	{mmGRBM_STATUS2, false},
+	{mmGRBM_STATUS_SE0, false},
+	{mmGRBM_STATUS_SE1, false},
+	{mmGRBM_STATUS_SE2, false},
+	{mmGRBM_STATUS_SE3, false},
+	{mmSRBM_STATUS, false},
+	{mmSRBM_STATUS2, false},
+	{mmSRBM_STATUS3, false},
+	{mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET, false},
+	{mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET, false},
+	{mmCP_STAT, false},
+	{mmCP_STALLED_STAT1, false},
+	{mmCP_STALLED_STAT2, false},
+	{mmCP_STALLED_STAT3, false},
+	{mmCP_CPF_BUSY_STAT, false},
+	{mmCP_CPF_STALLED_STAT1, false},
+	{mmCP_CPF_STATUS, false},
+	{mmCP_CPC_BUSY_STAT, false},
+	{mmCP_CPC_STALLED_STAT1, false},
+	{mmCP_CPC_STATUS, false},
+	{mmGB_ADDR_CONFIG, false},
+	{mmMC_ARB_RAMCFG, false},
+	{mmGB_TILE_MODE0, false},
+	{mmGB_TILE_MODE1, false},
+	{mmGB_TILE_MODE2, false},
+	{mmGB_TILE_MODE3, false},
+	{mmGB_TILE_MODE4, false},
+	{mmGB_TILE_MODE5, false},
+	{mmGB_TILE_MODE6, false},
+	{mmGB_TILE_MODE7, false},
+	{mmGB_TILE_MODE8, false},
+	{mmGB_TILE_MODE9, false},
+	{mmGB_TILE_MODE10, false},
+	{mmGB_TILE_MODE11, false},
+	{mmGB_TILE_MODE12, false},
+	{mmGB_TILE_MODE13, false},
+	{mmGB_TILE_MODE14, false},
+	{mmGB_TILE_MODE15, false},
+	{mmGB_TILE_MODE16, false},
+	{mmGB_TILE_MODE17, false},
+	{mmGB_TILE_MODE18, false},
+	{mmGB_TILE_MODE19, false},
+	{mmGB_TILE_MODE20, false},
+	{mmGB_TILE_MODE21, false},
+	{mmGB_TILE_MODE22, false},
+	{mmGB_TILE_MODE23, false},
+	{mmGB_TILE_MODE24, false},
+	{mmGB_TILE_MODE25, false},
+	{mmGB_TILE_MODE26, false},
+	{mmGB_TILE_MODE27, false},
+	{mmGB_TILE_MODE28, false},
+	{mmGB_TILE_MODE29, false},
+	{mmGB_TILE_MODE30, false},
+	{mmGB_TILE_MODE31, false},
+	{mmGB_MACROTILE_MODE0, false},
+	{mmGB_MACROTILE_MODE1, false},
+	{mmGB_MACROTILE_MODE2, false},
+	{mmGB_MACROTILE_MODE3, false},
+	{mmGB_MACROTILE_MODE4, false},
+	{mmGB_MACROTILE_MODE5, false},
+	{mmGB_MACROTILE_MODE6, false},
+	{mmGB_MACROTILE_MODE7, false},
+	{mmGB_MACROTILE_MODE8, false},
+	{mmGB_MACROTILE_MODE9, false},
+	{mmGB_MACROTILE_MODE10, false},
+	{mmGB_MACROTILE_MODE11, false},
+	{mmGB_MACROTILE_MODE12, false},
+	{mmGB_MACROTILE_MODE13, false},
+	{mmGB_MACROTILE_MODE14, false},
+	{mmGB_MACROTILE_MODE15, false},
+	{mmCC_RB_BACKEND_DISABLE, false, true},
+	{mmGC_USER_RB_BACKEND_DISABLE, false, true},
+	{mmGB_BACKEND_MAP, false, false},
+	{mmPA_SC_RASTER_CONFIG, false, true},
+	{mmPA_SC_RASTER_CONFIG_1, false, true},
+};
+
+static uint32_t vi_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
+					 u32 sh_num, u32 reg_offset)
+{
+	uint32_t val;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		gfx_v8_0_select_se_sh(adev, se_num, sh_num);
+
+	val = RREG32(reg_offset);
+
+	if (se_num != 0xffffffff || sh_num != 0xffffffff)
+		gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	mutex_unlock(&adev->grbm_idx_mutex);
+	return val;
+}
+
+static int vi_read_register(struct amdgpu_device *adev, u32 se_num,
+			    u32 sh_num, u32 reg_offset, u32 *value)
+{
+	struct amdgpu_allowed_register_entry *asic_register_table = NULL;
+	struct amdgpu_allowed_register_entry *asic_register_entry;
+	uint32_t size, i;
+
+	*value = 0;
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		asic_register_table = tonga_allowed_read_registers;
+		size = ARRAY_SIZE(tonga_allowed_read_registers);
+		break;
+	case CHIP_FIJI:
+	case CHIP_TONGA:
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		asic_register_table = cz_allowed_read_registers;
+		size = ARRAY_SIZE(cz_allowed_read_registers);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (asic_register_table) {
+		for (i = 0; i < size; i++) {
+			asic_register_entry = asic_register_table + i;
+			if (reg_offset != asic_register_entry->reg_offset)
+				continue;
+			if (!asic_register_entry->untouched)
+				*value = asic_register_entry->grbm_indexed ?
+					vi_read_indexed_register(adev, se_num,
+								 sh_num, reg_offset) :
+					RREG32(reg_offset);
+			return 0;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(vi_allowed_read_registers); i++) {
+		if (reg_offset != vi_allowed_read_registers[i].reg_offset)
+			continue;
+
+		if (!vi_allowed_read_registers[i].untouched)
+			*value = vi_allowed_read_registers[i].grbm_indexed ?
+				vi_read_indexed_register(adev, se_num,
+							 sh_num, reg_offset) :
+				RREG32(reg_offset);
+		return 0;
+	}
+	return -EINVAL;
+}
+
+static void vi_print_gpu_status_regs(struct amdgpu_device *adev)
+{
+	dev_info(adev->dev, "  GRBM_STATUS=0x%08X\n",
+		RREG32(mmGRBM_STATUS));
+	dev_info(adev->dev, "  GRBM_STATUS2=0x%08X\n",
+		RREG32(mmGRBM_STATUS2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE0=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE0));
+	dev_info(adev->dev, "  GRBM_STATUS_SE1=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE1));
+	dev_info(adev->dev, "  GRBM_STATUS_SE2=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE2));
+	dev_info(adev->dev, "  GRBM_STATUS_SE3=0x%08X\n",
+		RREG32(mmGRBM_STATUS_SE3));
+	dev_info(adev->dev, "  SRBM_STATUS=0x%08X\n",
+		RREG32(mmSRBM_STATUS));
+	dev_info(adev->dev, "  SRBM_STATUS2=0x%08X\n",
+		RREG32(mmSRBM_STATUS2));
+	dev_info(adev->dev, "  SDMA0_STATUS_REG   = 0x%08X\n",
+		RREG32(mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET));
+	if (adev->sdma.num_instances > 1) {
+		dev_info(adev->dev, "  SDMA1_STATUS_REG   = 0x%08X\n",
+			RREG32(mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET));
+	}
+	dev_info(adev->dev, "  CP_STAT = 0x%08x\n", RREG32(mmCP_STAT));
+	dev_info(adev->dev, "  CP_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_STALLED_STAT2 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT2));
+	dev_info(adev->dev, "  CP_STALLED_STAT3 = 0x%08x\n",
+		 RREG32(mmCP_STALLED_STAT3));
+	dev_info(adev->dev, "  CP_CPF_BUSY_STAT = 0x%08x\n",
+		 RREG32(mmCP_CPF_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPF_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPF_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPF_STATUS = 0x%08x\n", RREG32(mmCP_CPF_STATUS));
+	dev_info(adev->dev, "  CP_CPC_BUSY_STAT = 0x%08x\n", RREG32(mmCP_CPC_BUSY_STAT));
+	dev_info(adev->dev, "  CP_CPC_STALLED_STAT1 = 0x%08x\n",
+		 RREG32(mmCP_CPC_STALLED_STAT1));
+	dev_info(adev->dev, "  CP_CPC_STATUS = 0x%08x\n", RREG32(mmCP_CPC_STATUS));
+}
+
+/**
+ * vi_gpu_check_soft_reset - check which blocks are busy
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Check which blocks are busy and return the relevant reset
+ * mask to be used by vi_gpu_soft_reset().
+ * Returns a mask of the blocks to be reset.
+ */
+u32 vi_gpu_check_soft_reset(struct amdgpu_device *adev)
+{
+	u32 reset_mask = 0;
+	u32 tmp;
+
+	/* GRBM_STATUS */
+	tmp = RREG32(mmGRBM_STATUS);
+	if (tmp & (GRBM_STATUS__PA_BUSY_MASK | GRBM_STATUS__SC_BUSY_MASK |
+		   GRBM_STATUS__BCI_BUSY_MASK | GRBM_STATUS__SX_BUSY_MASK |
+		   GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK |
+		   GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK |
+		   GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK |
+		   GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK))
+		reset_mask |= AMDGPU_RESET_GFX;
+
+	if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK))
+		reset_mask |= AMDGPU_RESET_CP;
+
+	/* GRBM_STATUS2 */
+	tmp = RREG32(mmGRBM_STATUS2);
+	if (tmp & GRBM_STATUS2__RLC_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_RLC;
+
+	if (tmp & (GRBM_STATUS2__CPF_BUSY_MASK |
+		   GRBM_STATUS2__CPC_BUSY_MASK |
+		   GRBM_STATUS2__CPG_BUSY_MASK))
+		reset_mask |= AMDGPU_RESET_CP;
+
+	/* SRBM_STATUS2 */
+	tmp = RREG32(mmSRBM_STATUS2);
+	if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_DMA;
+
+	if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_DMA1;
+
+	/* SRBM_STATUS */
+	tmp = RREG32(mmSRBM_STATUS);
+
+	if (tmp & SRBM_STATUS__IH_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_IH;
+
+	if (tmp & SRBM_STATUS__SEM_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_SEM;
+
+	if (tmp & SRBM_STATUS__GRBM_RQ_PENDING_MASK)
+		reset_mask |= AMDGPU_RESET_GRBM;
+
+	if (adev->asic_type != CHIP_TOPAZ) {
+		if (tmp & (SRBM_STATUS__UVD_RQ_PENDING_MASK |
+			   SRBM_STATUS__UVD_BUSY_MASK))
+			reset_mask |= AMDGPU_RESET_UVD;
+	}
+
+	if (tmp & SRBM_STATUS__VMC_BUSY_MASK)
+		reset_mask |= AMDGPU_RESET_VMC;
+
+	if (tmp & (SRBM_STATUS__MCB_BUSY_MASK | SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
+		   SRBM_STATUS__MCC_BUSY_MASK | SRBM_STATUS__MCD_BUSY_MASK))
+		reset_mask |= AMDGPU_RESET_MC;
+
+	/* SDMA0_STATUS_REG */
+	tmp = RREG32(mmSDMA0_STATUS_REG + SDMA0_REGISTER_OFFSET);
+	if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+		reset_mask |= AMDGPU_RESET_DMA;
+
+	/* SDMA1_STATUS_REG */
+	if (adev->sdma.num_instances > 1) {
+		tmp = RREG32(mmSDMA0_STATUS_REG + SDMA1_REGISTER_OFFSET);
+		if (!(tmp & SDMA0_STATUS_REG__IDLE_MASK))
+			reset_mask |= AMDGPU_RESET_DMA1;
+	}
+#if 0
+	/* VCE_STATUS */
+	if (adev->asic_type != CHIP_TOPAZ) {
+		tmp = RREG32(mmVCE_STATUS);
+		if (tmp & VCE_STATUS__VCPU_REPORT_RB0_BUSY_MASK)
+			reset_mask |= AMDGPU_RESET_VCE;
+		if (tmp & VCE_STATUS__VCPU_REPORT_RB1_BUSY_MASK)
+			reset_mask |= AMDGPU_RESET_VCE1;
+
+	}
+
+	if (adev->asic_type != CHIP_TOPAZ) {
+		if (amdgpu_display_is_display_hung(adev))
+			reset_mask |= AMDGPU_RESET_DISPLAY;
+	}
+#endif
+
+	/* Skip MC reset as it's mostly likely not hung, just busy */
+	if (reset_mask & AMDGPU_RESET_MC) {
+		DRM_DEBUG("MC busy: 0x%08X, clearing.\n", reset_mask);
+		reset_mask &= ~AMDGPU_RESET_MC;
+	}
+
+	return reset_mask;
+}
+
+/**
+ * vi_gpu_soft_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ * @reset_mask: mask of which blocks to reset
+ *
+ * Soft reset the blocks specified in @reset_mask.
+ */
+static void vi_gpu_soft_reset(struct amdgpu_device *adev, u32 reset_mask)
+{
+	struct amdgpu_mode_mc_save save;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
+
+	if (reset_mask == 0)
+		return;
+
+	dev_info(adev->dev, "GPU softreset: 0x%08X\n", reset_mask);
+
+	vi_print_gpu_status_regs(adev);
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_ADDR   0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR));
+	dev_info(adev->dev, "  VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
+		 RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS));
+
+	/* disable CG/PG */
+
+	/* stop the rlc */
+	//XXX
+	//gfx_v8_0_rlc_stop(adev);
+
+	/* Disable GFX parsing/prefetching */
+	tmp = RREG32(mmCP_ME_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 1);
+	WREG32(mmCP_ME_CNTL, tmp);
+
+	/* Disable MEC parsing/prefetching */
+	tmp = RREG32(mmCP_MEC_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_CNTL, MEC_ME1_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_CNTL, MEC_ME2_HALT, 1);
+	WREG32(mmCP_MEC_CNTL, tmp);
+
+	if (reset_mask & AMDGPU_RESET_DMA) {
+		/* sdma0 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 1);
+		WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+	}
+	if (reset_mask & AMDGPU_RESET_DMA1) {
+		/* sdma1 */
+		tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+		tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 1);
+		WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+	}
+
+	gmc_v8_0_mc_stop(adev, &save);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
+	}
+
+	if (reset_mask & (AMDGPU_RESET_GFX | AMDGPU_RESET_COMPUTE | AMDGPU_RESET_CP)) {
+		grbm_soft_reset =
+			REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		grbm_soft_reset =
+			REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX, 1);
+	}
+
+	if (reset_mask & AMDGPU_RESET_CP) {
+		grbm_soft_reset =
+			REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP, 1);
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1);
+	}
+
+	if (reset_mask & AMDGPU_RESET_DMA)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA, 1);
+
+	if (reset_mask & AMDGPU_RESET_DMA1)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1, 1);
+
+	if (reset_mask & AMDGPU_RESET_DISPLAY)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_DC, 1);
+
+	if (reset_mask & AMDGPU_RESET_RLC)
+		grbm_soft_reset =
+			REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1);
+
+	if (reset_mask & AMDGPU_RESET_SEM)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SEM, 1);
+
+	if (reset_mask & AMDGPU_RESET_IH)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_IH, 1);
+
+	if (reset_mask & AMDGPU_RESET_GRBM)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1);
+
+	if (reset_mask & AMDGPU_RESET_VMC)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VMC, 1);
+
+	if (reset_mask & AMDGPU_RESET_UVD)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_UVD, 1);
+
+	if (reset_mask & AMDGPU_RESET_VCE)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1);
+
+	if (reset_mask & AMDGPU_RESET_VCE)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1);
+
+	if (!(adev->flags & AMD_IS_APU)) {
+		if (reset_mask & AMDGPU_RESET_MC)
+		srbm_soft_reset =
+			REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_MC, 1);
+	}
+
+	if (grbm_soft_reset) {
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+		tmp |= grbm_soft_reset;
+		dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~grbm_soft_reset;
+		WREG32(mmGRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmGRBM_SOFT_RESET);
+	}
+
+	if (srbm_soft_reset) {
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+		tmp |= srbm_soft_reset;
+		dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp);
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+
+		udelay(50);
+
+		tmp &= ~srbm_soft_reset;
+		WREG32(mmSRBM_SOFT_RESET, tmp);
+		tmp = RREG32(mmSRBM_SOFT_RESET);
+	}
+
+	/* Wait a little for things to settle down */
+	udelay(50);
+
+	gmc_v8_0_mc_resume(adev, &save);
+	udelay(50);
+
+	vi_print_gpu_status_regs(adev);
+}
+
+static void vi_gpu_pci_config_reset(struct amdgpu_device *adev)
+{
+	struct amdgpu_mode_mc_save save;
+	u32 tmp, i;
+
+	dev_info(adev->dev, "GPU pci config reset\n");
+
+	/* disable dpm? */
+
+	/* disable cg/pg */
+
+	/* Disable GFX parsing/prefetching */
+	tmp = RREG32(mmCP_ME_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, ME_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, 1);
+	WREG32(mmCP_ME_CNTL, tmp);
+
+	/* Disable MEC parsing/prefetching */
+	tmp = RREG32(mmCP_MEC_CNTL);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_CNTL, MEC_ME1_HALT, 1);
+	tmp = REG_SET_FIELD(tmp, CP_MEC_CNTL, MEC_ME2_HALT, 1);
+	WREG32(mmCP_MEC_CNTL, tmp);
+
+	/* Disable GFX parsing/prefetching */
+	WREG32(mmCP_ME_CNTL, CP_ME_CNTL__ME_HALT_MASK |
+		CP_ME_CNTL__PFP_HALT_MASK | CP_ME_CNTL__CE_HALT_MASK);
+
+	/* Disable MEC parsing/prefetching */
+	WREG32(mmCP_MEC_CNTL,
+			CP_MEC_CNTL__MEC_ME1_HALT_MASK | CP_MEC_CNTL__MEC_ME2_HALT_MASK);
+
+	/* sdma0 */
+	tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET);
+	tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 1);
+	WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp);
+
+	/* sdma1 */
+	tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET);
+	tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 1);
+	WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp);
+
+	/* XXX other engines? */
+
+	/* halt the rlc, disable cp internal ints */
+	//XXX
+	//gfx_v8_0_rlc_stop(adev);
+
+	udelay(50);
+
+	/* disable mem access */
+	gmc_v8_0_mc_stop(adev, &save);
+	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+		dev_warn(adev->dev, "Wait for MC idle timed out !\n");
+	}
+
+	/* disable BM */
+	pci_clear_master(adev->pdev);
+	/* reset */
+	amdgpu_pci_config_reset(adev);
+
+	udelay(100);
+
+	/* wait for asic to come out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff)
+			break;
+		udelay(1);
+	}
+
+}
+
+static void vi_set_bios_scratch_engine_hung(struct amdgpu_device *adev, bool hung)
+{
+	u32 tmp = RREG32(mmBIOS_SCRATCH_3);
+
+	if (hung)
+		tmp |= ATOM_S3_ASIC_GUI_ENGINE_HUNG;
+	else
+		tmp &= ~ATOM_S3_ASIC_GUI_ENGINE_HUNG;
+
+	WREG32(mmBIOS_SCRATCH_3, tmp);
+}
+
+/**
+ * vi_asic_reset - soft reset GPU
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Look up which blocks are hung and attempt
+ * to reset them.
+ * Returns 0 for success.
+ */
+static int vi_asic_reset(struct amdgpu_device *adev)
+{
+	u32 reset_mask;
+
+	reset_mask = vi_gpu_check_soft_reset(adev);
+
+	if (reset_mask)
+		vi_set_bios_scratch_engine_hung(adev, true);
+
+	/* try soft reset */
+	vi_gpu_soft_reset(adev, reset_mask);
+
+	reset_mask = vi_gpu_check_soft_reset(adev);
+
+	/* try pci config reset */
+	if (reset_mask && amdgpu_hard_reset)
+		vi_gpu_pci_config_reset(adev);
+
+	reset_mask = vi_gpu_check_soft_reset(adev);
+
+	if (!reset_mask)
+		vi_set_bios_scratch_engine_hung(adev, false);
+
+	return 0;
+}
+
+static int vi_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
+			u32 cntl_reg, u32 status_reg)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+	uint32_t tmp;
+
+	r = amdgpu_atombios_get_clock_dividers(adev,
+					       COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK,
+					       clock, false, &dividers);
+	if (r)
+		return r;
+
+	tmp = RREG32_SMC(cntl_reg);
+	tmp &= ~(CG_DCLK_CNTL__DCLK_DIR_CNTL_EN_MASK |
+		CG_DCLK_CNTL__DCLK_DIVIDER_MASK);
+	tmp |= dividers.post_divider;
+	WREG32_SMC(cntl_reg, tmp);
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32_SMC(status_reg) & CG_DCLK_STATUS__DCLK_STATUS_MASK)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static int vi_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
+{
+	int r;
+
+	r = vi_set_uvd_clock(adev, vclk, ixCG_VCLK_CNTL, ixCG_VCLK_STATUS);
+	if (r)
+		return r;
+
+	r = vi_set_uvd_clock(adev, dclk, ixCG_DCLK_CNTL, ixCG_DCLK_STATUS);
+
+	return 0;
+}
+
+static int vi_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
+{
+	/* todo */
+
+	return 0;
+}
+
+static void vi_pcie_gen3_enable(struct amdgpu_device *adev)
+{
+	u32 mask;
+	int ret;
+
+	if (pci_is_root_bus(adev->pdev->bus))
+		return;
+
+	if (amdgpu_pcie_gen2 == 0)
+		return;
+
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	ret = drm_pcie_get_speed_cap_mask(adev->ddev, &mask);
+	if (ret != 0)
+		return;
+
+	if (!(mask & (DRM_PCIE_SPEED_50 | DRM_PCIE_SPEED_80)))
+		return;
+
+	/* todo */
+}
+
+static void vi_program_aspm(struct amdgpu_device *adev)
+{
+
+	if (amdgpu_aspm == 0)
+		return;
+
+	/* todo */
+}
+
+static void vi_enable_doorbell_aperture(struct amdgpu_device *adev,
+					bool enable)
+{
+	u32 tmp;
+
+	/* not necessary on CZ */
+	if (adev->flags & AMD_IS_APU)
+		return;
+
+	tmp = RREG32(mmBIF_DOORBELL_APER_EN);
+	if (enable)
+		tmp = REG_SET_FIELD(tmp, BIF_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, 1);
+	else
+		tmp = REG_SET_FIELD(tmp, BIF_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN, 0);
+
+	WREG32(mmBIF_DOORBELL_APER_EN, tmp);
+}
+
+/* topaz has no DCE, UVD, VCE */
+static const struct amdgpu_ip_block_version topaz_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vi_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 7,
+		.minor = 4,
+		.rev = 0,
+		.funcs = &gmc_v7_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 2,
+		.minor = 4,
+		.rev = 0,
+		.funcs = &iceland_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 1,
+		.rev = 0,
+		.funcs = &iceland_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gfx_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 2,
+		.minor = 4,
+		.rev = 0,
+		.funcs = &sdma_v2_4_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version tonga_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vi_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &tonga_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 1,
+		.rev = 0,
+		.funcs = &tonga_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 10,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &dce_v10_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gfx_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &sdma_v3_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 5,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &uvd_v5_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v3_0_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version fiji_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vi_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 8,
+		.minor = 5,
+		.rev = 0,
+		.funcs = &gmc_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &tonga_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 7,
+		.minor = 1,
+		.rev = 0,
+		.funcs = &fiji_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 10,
+		.minor = 1,
+		.rev = 0,
+		.funcs = &dce_v10_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gfx_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &sdma_v3_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 6,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &uvd_v6_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v3_0_ip_funcs,
+	},
+};
+
+static const struct amdgpu_ip_block_version cz_ip_blocks[] =
+{
+	/* ORDER MATTERS! */
+	{
+		.type = AMD_IP_BLOCK_TYPE_COMMON,
+		.major = 2,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vi_common_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GMC,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gmc_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_IH,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cz_ih_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SMC,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &cz_dpm_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_DCE,
+		.major = 11,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &dce_v11_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_GFX,
+		.major = 8,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &gfx_v8_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_SDMA,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &sdma_v3_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_UVD,
+		.major = 6,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &uvd_v6_0_ip_funcs,
+	},
+	{
+		.type = AMD_IP_BLOCK_TYPE_VCE,
+		.major = 3,
+		.minor = 0,
+		.rev = 0,
+		.funcs = &vce_v3_0_ip_funcs,
+	},
+};
+
+int vi_set_ip_blocks(struct amdgpu_device *adev)
+{
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		adev->ip_blocks = topaz_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(topaz_ip_blocks);
+		break;
+	case CHIP_FIJI:
+		adev->ip_blocks = fiji_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(fiji_ip_blocks);
+		break;
+	case CHIP_TONGA:
+		adev->ip_blocks = tonga_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(tonga_ip_blocks);
+		break;
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		adev->ip_blocks = cz_ip_blocks;
+		adev->num_ip_blocks = ARRAY_SIZE(cz_ip_blocks);
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+#define ATI_REV_ID_FUSE_MACRO__ADDRESS      0xC0014044
+#define ATI_REV_ID_FUSE_MACRO__SHIFT        9
+#define ATI_REV_ID_FUSE_MACRO__MASK         0x00001E00
+
+static uint32_t vi_get_rev_id(struct amdgpu_device *adev)
+{
+	if (adev->asic_type == CHIP_TOPAZ)
+		return (RREG32(mmPCIE_EFUSE4) & PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID_MASK)
+			>> PCIE_EFUSE4__STRAP_BIF_ATI_REV_ID__SHIFT;
+	else if (adev->flags & AMD_IS_APU)
+		return (RREG32_SMC(ATI_REV_ID_FUSE_MACRO__ADDRESS) & ATI_REV_ID_FUSE_MACRO__MASK)
+			>> ATI_REV_ID_FUSE_MACRO__SHIFT;
+	else
+		return (RREG32(mmCC_DRM_ID_STRAPS) & CC_DRM_ID_STRAPS__ATI_REV_ID_MASK)
+			>> CC_DRM_ID_STRAPS__ATI_REV_ID__SHIFT;
+}
+
+static const struct amdgpu_asic_funcs vi_asic_funcs =
+{
+	.read_disabled_bios = &vi_read_disabled_bios,
+	.read_register = &vi_read_register,
+	.reset = &vi_asic_reset,
+	.set_vga_state = &vi_vga_set_state,
+	.get_xclk = &vi_get_xclk,
+	.set_uvd_clocks = &vi_set_uvd_clocks,
+	.set_vce_clocks = &vi_set_vce_clocks,
+	.get_cu_info = &gfx_v8_0_get_cu_info,
+	/* these should be moved to their own ip modules */
+	.get_gpu_clock_counter = &gfx_v8_0_get_gpu_clock_counter,
+	.wait_for_mc_idle = &gmc_v8_0_mc_wait_for_idle,
+};
+
+static int vi_common_early_init(void *handle)
+{
+	bool smc_enabled = false;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	if (adev->flags & AMD_IS_APU) {
+		adev->smc_rreg = &cz_smc_rreg;
+		adev->smc_wreg = &cz_smc_wreg;
+	} else {
+		adev->smc_rreg = &vi_smc_rreg;
+		adev->smc_wreg = &vi_smc_wreg;
+	}
+	adev->pcie_rreg = &vi_pcie_rreg;
+	adev->pcie_wreg = &vi_pcie_wreg;
+	adev->uvd_ctx_rreg = &vi_uvd_ctx_rreg;
+	adev->uvd_ctx_wreg = &vi_uvd_ctx_wreg;
+	adev->didt_rreg = &vi_didt_rreg;
+	adev->didt_wreg = &vi_didt_wreg;
+
+	adev->asic_funcs = &vi_asic_funcs;
+
+	if (amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_SMC) &&
+		(amdgpu_ip_block_mask & (1 << AMD_IP_BLOCK_TYPE_SMC)))
+		smc_enabled = true;
+
+	adev->rev_id = vi_get_rev_id(adev);
+	adev->external_rev_id = 0xFF;
+	switch (adev->asic_type) {
+	case CHIP_TOPAZ:
+		adev->has_uvd = false;
+		adev->cg_flags = 0;
+		adev->pg_flags = 0;
+		adev->external_rev_id = 0x1;
+		break;
+	case CHIP_FIJI:
+		adev->has_uvd = true;
+		adev->cg_flags = 0;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x3c;
+		break;
+	case CHIP_TONGA:
+		adev->has_uvd = true;
+		adev->cg_flags = 0;
+		adev->pg_flags = 0;
+		adev->external_rev_id = adev->rev_id + 0x14;
+		break;
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		adev->has_uvd = true;
+		adev->cg_flags = 0;
+		/* Disable UVD pg */
+		adev->pg_flags = /* AMDGPU_PG_SUPPORT_UVD | */AMDGPU_PG_SUPPORT_VCE;
+		adev->external_rev_id = adev->rev_id + 0x1;
+		break;
+	default:
+		/* FIXME: not supported yet */
+		return -EINVAL;
+	}
+
+	if (amdgpu_smc_load_fw && smc_enabled)
+		adev->firmware.smu_load = true;
+
+	return 0;
+}
+
+static int vi_common_sw_init(void *handle)
+{
+	return 0;
+}
+
+static int vi_common_sw_fini(void *handle)
+{
+	return 0;
+}
+
+static int vi_common_hw_init(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* move the golden regs per IP block */
+	vi_init_golden_registers(adev);
+	/* enable pcie gen2/3 link */
+	vi_pcie_gen3_enable(adev);
+	/* enable aspm */
+	vi_program_aspm(adev);
+	/* enable the doorbell aperture */
+	vi_enable_doorbell_aperture(adev, true);
+
+	return 0;
+}
+
+static int vi_common_hw_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	/* enable the doorbell aperture */
+	vi_enable_doorbell_aperture(adev, false);
+
+	return 0;
+}
+
+static int vi_common_suspend(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return vi_common_hw_fini(adev);
+}
+
+static int vi_common_resume(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	return vi_common_hw_init(adev);
+}
+
+static bool vi_common_is_idle(void *handle)
+{
+	return true;
+}
+
+static int vi_common_wait_for_idle(void *handle)
+{
+	return 0;
+}
+
+static void vi_common_print_status(void *handle)
+{
+	return;
+}
+
+static int vi_common_soft_reset(void *handle)
+{
+	return 0;
+}
+
+static int vi_common_set_clockgating_state(void *handle,
+					    enum amd_clockgating_state state)
+{
+	return 0;
+}
+
+static int vi_common_set_powergating_state(void *handle,
+					    enum amd_powergating_state state)
+{
+	return 0;
+}
+
+const struct amd_ip_funcs vi_common_ip_funcs = {
+	.early_init = vi_common_early_init,
+	.late_init = NULL,
+	.sw_init = vi_common_sw_init,
+	.sw_fini = vi_common_sw_fini,
+	.hw_init = vi_common_hw_init,
+	.hw_fini = vi_common_hw_fini,
+	.suspend = vi_common_suspend,
+	.resume = vi_common_resume,
+	.is_idle = vi_common_is_idle,
+	.wait_for_idle = vi_common_wait_for_idle,
+	.soft_reset = vi_common_soft_reset,
+	.print_status = vi_common_print_status,
+	.set_clockgating_state = vi_common_set_clockgating_state,
+	.set_powergating_state = vi_common_set_powergating_state,
+};
+
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vi.h b/kernel/drivers/gpu/drm/amd/amdgpu/vi.h
new file mode 100644
index 000000000..502094042
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vi.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VI_H__
+#define __VI_H__
+
+extern const struct amd_ip_funcs vi_common_ip_funcs;
+
+void vi_srbm_select(struct amdgpu_device *adev,
+		    u32 me, u32 pipe, u32 queue, u32 vmid);
+int vi_set_ip_blocks(struct amdgpu_device *adev);
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vi_dpm.h b/kernel/drivers/gpu/drm/amd/amdgpu/vi_dpm.h
new file mode 100644
index 000000000..fc120ba18
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vi_dpm.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VI_DPM_H__
+#define __VI_DPM_H__
+
+extern const struct amd_ip_funcs cz_dpm_ip_funcs;
+int cz_smu_init(struct amdgpu_device *adev);
+int cz_smu_start(struct amdgpu_device *adev);
+int cz_smu_fini(struct amdgpu_device *adev);
+
+extern const struct amd_ip_funcs tonga_dpm_ip_funcs;
+extern const struct amd_ip_funcs fiji_dpm_ip_funcs;
+extern const struct amd_ip_funcs iceland_dpm_ip_funcs;
+
+#endif
diff --git a/kernel/drivers/gpu/drm/amd/amdgpu/vid.h b/kernel/drivers/gpu/drm/amd/amdgpu/vid.h
new file mode 100644
index 000000000..d98aa9d82
--- /dev/null
+++ b/kernel/drivers/gpu/drm/amd/amdgpu/vid.h
@@ -0,0 +1,373 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef VI_H
+#define VI_H
+
+#define SDMA0_REGISTER_OFFSET                             0x0 /* not a register */
+#define SDMA1_REGISTER_OFFSET                             0x200 /* not a register */
+#define SDMA_MAX_INSTANCE 2
+
+/* crtc instance offsets */
+#define CRTC0_REGISTER_OFFSET                 (0x1b9c - 0x1b9c)
+#define CRTC1_REGISTER_OFFSET                 (0x1d9c - 0x1b9c)
+#define CRTC2_REGISTER_OFFSET                 (0x1f9c - 0x1b9c)
+#define CRTC3_REGISTER_OFFSET                 (0x419c - 0x1b9c)
+#define CRTC4_REGISTER_OFFSET                 (0x439c - 0x1b9c)
+#define CRTC5_REGISTER_OFFSET                 (0x459c - 0x1b9c)
+#define CRTC6_REGISTER_OFFSET                 (0x479c - 0x1b9c)
+
+/* dig instance offsets */
+#define DIG0_REGISTER_OFFSET                 (0x4a00 - 0x4a00)
+#define DIG1_REGISTER_OFFSET                 (0x4b00 - 0x4a00)
+#define DIG2_REGISTER_OFFSET                 (0x4c00 - 0x4a00)
+#define DIG3_REGISTER_OFFSET                 (0x4d00 - 0x4a00)
+#define DIG4_REGISTER_OFFSET                 (0x4e00 - 0x4a00)
+#define DIG5_REGISTER_OFFSET                 (0x4f00 - 0x4a00)
+#define DIG6_REGISTER_OFFSET                 (0x5400 - 0x4a00)
+#define DIG7_REGISTER_OFFSET                 (0x5600 - 0x4a00)
+#define DIG8_REGISTER_OFFSET                 (0x5700 - 0x4a00)
+
+/* audio endpt instance offsets */
+#define AUD0_REGISTER_OFFSET                 (0x17a8 - 0x17a8)
+#define AUD1_REGISTER_OFFSET                 (0x17ac - 0x17a8)
+#define AUD2_REGISTER_OFFSET                 (0x17b0 - 0x17a8)
+#define AUD3_REGISTER_OFFSET                 (0x17b4 - 0x17a8)
+#define AUD4_REGISTER_OFFSET                 (0x17b8 - 0x17a8)
+#define AUD5_REGISTER_OFFSET                 (0x17bc - 0x17a8)
+#define AUD6_REGISTER_OFFSET                 (0x17c4 - 0x17a8)
+
+/* hpd instance offsets */
+#define HPD0_REGISTER_OFFSET                 (0x1898 - 0x1898)
+#define HPD1_REGISTER_OFFSET                 (0x18a0 - 0x1898)
+#define HPD2_REGISTER_OFFSET                 (0x18a8 - 0x1898)
+#define HPD3_REGISTER_OFFSET                 (0x18b0 - 0x1898)
+#define HPD4_REGISTER_OFFSET                 (0x18b8 - 0x1898)
+#define HPD5_REGISTER_OFFSET                 (0x18c0 - 0x1898)
+
+#define AMDGPU_NUM_OF_VMIDS			8
+
+#define		PIPEID(x)					((x) << 0)
+#define		MEID(x)						((x) << 2)
+#define		VMID(x)						((x) << 4)
+#define		QUEUEID(x)					((x) << 8)
+
+#define RB_BITMAP_WIDTH_PER_SH     2
+
+#define MC_SEQ_MISC0__MT__MASK	0xf0000000
+#define MC_SEQ_MISC0__MT__GDDR1  0x10000000
+#define MC_SEQ_MISC0__MT__DDR2   0x20000000
+#define MC_SEQ_MISC0__MT__GDDR3  0x30000000
+#define MC_SEQ_MISC0__MT__GDDR4  0x40000000
+#define MC_SEQ_MISC0__MT__GDDR5  0x50000000
+#define MC_SEQ_MISC0__MT__HBM    0x60000000
+#define MC_SEQ_MISC0__MT__DDR3   0xB0000000
+
+/*
+ * PM4
+ */
+#define	PACKET_TYPE0	0
+#define	PACKET_TYPE1	1
+#define	PACKET_TYPE2	2
+#define	PACKET_TYPE3	3
+
+#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
+#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
+#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
+#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
+#define PACKET0(reg, n)	((PACKET_TYPE0 << 30) |				\
+			 ((reg) & 0xFFFF) |			\
+			 ((n) & 0x3FFF) << 16)
+#define CP_PACKET2			0x80000000
+#define		PACKET2_PAD_SHIFT		0
+#define		PACKET2_PAD_MASK		(0x3fffffff << 0)
+
+#define PACKET2(v)	(CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
+
+#define PACKET3(op, n)	((PACKET_TYPE3 << 30) |				\
+			 (((op) & 0xFF) << 8) |				\
+			 ((n) & 0x3FFF) << 16)
+
+#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
+
+/* Packet 3 types */
+#define	PACKET3_NOP					0x10
+#define	PACKET3_SET_BASE				0x11
+#define		PACKET3_BASE_INDEX(x)                  ((x) << 0)
+#define			CE_PARTITION_BASE		3
+#define	PACKET3_CLEAR_STATE				0x12
+#define	PACKET3_INDEX_BUFFER_SIZE			0x13
+#define	PACKET3_DISPATCH_DIRECT				0x15
+#define	PACKET3_DISPATCH_INDIRECT			0x16
+#define	PACKET3_ATOMIC_GDS				0x1D
+#define	PACKET3_ATOMIC_MEM				0x1E
+#define	PACKET3_OCCLUSION_QUERY				0x1F
+#define	PACKET3_SET_PREDICATION				0x20
+#define	PACKET3_REG_RMW					0x21
+#define	PACKET3_COND_EXEC				0x22
+#define	PACKET3_PRED_EXEC				0x23
+#define	PACKET3_DRAW_INDIRECT				0x24
+#define	PACKET3_DRAW_INDEX_INDIRECT			0x25
+#define	PACKET3_INDEX_BASE				0x26
+#define	PACKET3_DRAW_INDEX_2				0x27
+#define	PACKET3_CONTEXT_CONTROL				0x28
+#define	PACKET3_INDEX_TYPE				0x2A
+#define	PACKET3_DRAW_INDIRECT_MULTI			0x2C
+#define	PACKET3_DRAW_INDEX_AUTO				0x2D
+#define	PACKET3_NUM_INSTANCES				0x2F
+#define	PACKET3_DRAW_INDEX_MULTI_AUTO			0x30
+#define	PACKET3_INDIRECT_BUFFER_CONST			0x33
+#define	PACKET3_STRMOUT_BUFFER_UPDATE			0x34
+#define	PACKET3_DRAW_INDEX_OFFSET_2			0x35
+#define	PACKET3_DRAW_PREAMBLE				0x36
+#define	PACKET3_WRITE_DATA				0x37
+#define		WRITE_DATA_DST_SEL(x)                   ((x) << 8)
+		/* 0 - register
+		 * 1 - memory (sync - via GRBM)
+		 * 2 - gl2
+		 * 3 - gds
+		 * 4 - reserved
+		 * 5 - memory (async - direct)
+		 */
+#define		WR_ONE_ADDR                             (1 << 16)
+#define		WR_CONFIRM                              (1 << 20)
+#define		WRITE_DATA_CACHE_POLICY(x)              ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 */
+#define		WRITE_DATA_ENGINE_SEL(x)                ((x) << 30)
+		/* 0 - me
+		 * 1 - pfp
+		 * 2 - ce
+		 */
+#define	PACKET3_DRAW_INDEX_INDIRECT_MULTI		0x38
+#define	PACKET3_MEM_SEMAPHORE				0x39
+#              define PACKET3_SEM_USE_MAILBOX       (0x1 << 16)
+#              define PACKET3_SEM_SEL_SIGNAL_TYPE   (0x1 << 20) /* 0 = increment, 1 = write 1 */
+#              define PACKET3_SEM_CLIENT_CODE	    ((x) << 24) /* 0 = CP, 1 = CB, 2 = DB */
+#              define PACKET3_SEM_SEL_SIGNAL	    (0x6 << 29)
+#              define PACKET3_SEM_SEL_WAIT	    (0x7 << 29)
+#define	PACKET3_WAIT_REG_MEM				0x3C
+#define		WAIT_REG_MEM_FUNCTION(x)                ((x) << 0)
+		/* 0 - always
+		 * 1 - <
+		 * 2 - <=
+		 * 3 - ==
+		 * 4 - !=
+		 * 5 - >=
+		 * 6 - >
+		 */
+#define		WAIT_REG_MEM_MEM_SPACE(x)               ((x) << 4)
+		/* 0 - reg
+		 * 1 - mem
+		 */
+#define		WAIT_REG_MEM_OPERATION(x)               ((x) << 6)
+		/* 0 - wait_reg_mem
+		 * 1 - wr_wait_wr_reg
+		 */
+#define		WAIT_REG_MEM_ENGINE(x)                  ((x) << 8)
+		/* 0 - me
+		 * 1 - pfp
+		 */
+#define	PACKET3_INDIRECT_BUFFER				0x3F
+#define		INDIRECT_BUFFER_TCL2_VOLATILE           (1 << 22)
+#define		INDIRECT_BUFFER_VALID                   (1 << 23)
+#define		INDIRECT_BUFFER_CACHE_POLICY(x)         ((x) << 28)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define	PACKET3_COPY_DATA				0x40
+#define	PACKET3_PFP_SYNC_ME				0x42
+#define	PACKET3_SURFACE_SYNC				0x43
+#              define PACKET3_DEST_BASE_0_ENA      (1 << 0)
+#              define PACKET3_DEST_BASE_1_ENA      (1 << 1)
+#              define PACKET3_CB0_DEST_BASE_ENA    (1 << 6)
+#              define PACKET3_CB1_DEST_BASE_ENA    (1 << 7)
+#              define PACKET3_CB2_DEST_BASE_ENA    (1 << 8)
+#              define PACKET3_CB3_DEST_BASE_ENA    (1 << 9)
+#              define PACKET3_CB4_DEST_BASE_ENA    (1 << 10)
+#              define PACKET3_CB5_DEST_BASE_ENA    (1 << 11)
+#              define PACKET3_CB6_DEST_BASE_ENA    (1 << 12)
+#              define PACKET3_CB7_DEST_BASE_ENA    (1 << 13)
+#              define PACKET3_DB_DEST_BASE_ENA     (1 << 14)
+#              define PACKET3_TCL1_VOL_ACTION_ENA  (1 << 15)
+#              define PACKET3_TC_VOL_ACTION_ENA    (1 << 16) /* L2 */
+#              define PACKET3_TC_WB_ACTION_ENA     (1 << 18) /* L2 */
+#              define PACKET3_DEST_BASE_2_ENA      (1 << 19)
+#              define PACKET3_DEST_BASE_3_ENA      (1 << 21)
+#              define PACKET3_TCL1_ACTION_ENA      (1 << 22)
+#              define PACKET3_TC_ACTION_ENA        (1 << 23) /* L2 */
+#              define PACKET3_CB_ACTION_ENA        (1 << 25)
+#              define PACKET3_DB_ACTION_ENA        (1 << 26)
+#              define PACKET3_SH_KCACHE_ACTION_ENA (1 << 27)
+#              define PACKET3_SH_KCACHE_VOL_ACTION_ENA (1 << 28)
+#              define PACKET3_SH_ICACHE_ACTION_ENA (1 << 29)
+#define	PACKET3_COND_WRITE				0x45
+#define	PACKET3_EVENT_WRITE				0x46
+#define		EVENT_TYPE(x)                           ((x) << 0)
+#define		EVENT_INDEX(x)                          ((x) << 8)
+		/* 0 - any non-TS event
+		 * 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
+		 * 2 - SAMPLE_PIPELINESTAT
+		 * 3 - SAMPLE_STREAMOUTSTAT*
+		 * 4 - *S_PARTIAL_FLUSH
+		 * 5 - EOP events
+		 * 6 - EOS events
+		 */
+#define	PACKET3_EVENT_WRITE_EOP				0x47
+#define		EOP_TCL1_VOL_ACTION_EN                  (1 << 12)
+#define		EOP_TC_VOL_ACTION_EN                    (1 << 13) /* L2 */
+#define		EOP_TC_WB_ACTION_EN                     (1 << 15) /* L2 */
+#define		EOP_TCL1_ACTION_EN                      (1 << 16)
+#define		EOP_TC_ACTION_EN                        (1 << 17) /* L2 */
+#define		EOP_TCL2_VOLATILE                       (1 << 24)
+#define		EOP_CACHE_POLICY(x)                     ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#define		DATA_SEL(x)                             ((x) << 29)
+		/* 0 - discard
+		 * 1 - send low 32bit data
+		 * 2 - send 64bit data
+		 * 3 - send 64bit GPU counter value
+		 * 4 - send 64bit sys counter value
+		 */
+#define		INT_SEL(x)                              ((x) << 24)
+		/* 0 - none
+		 * 1 - interrupt only (DATA_SEL = 0)
+		 * 2 - interrupt when data write is confirmed
+		 */
+#define		DST_SEL(x)                              ((x) << 16)
+		/* 0 - MC
+		 * 1 - TC/L2
+		 */
+#define	PACKET3_EVENT_WRITE_EOS				0x48
+#define	PACKET3_RELEASE_MEM				0x49
+#define	PACKET3_PREAMBLE_CNTL				0x4A
+#              define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE     (2 << 28)
+#              define PACKET3_PREAMBLE_END_CLEAR_STATE       (3 << 28)
+#define	PACKET3_DMA_DATA				0x50
+/* 1. header
+ * 2. CONTROL
+ * 3. SRC_ADDR_LO or DATA [31:0]
+ * 4. SRC_ADDR_HI [31:0]
+ * 5. DST_ADDR_LO [31:0]
+ * 6. DST_ADDR_HI [7:0]
+ * 7. COMMAND [30:21] | BYTE_COUNT [20:0]
+ */
+/* CONTROL */
+#              define PACKET3_DMA_DATA_ENGINE(x)     ((x) << 0)
+		/* 0 - ME
+		 * 1 - PFP
+		 */
+#              define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_SRC_VOLATILE (1 << 15)
+#              define PACKET3_DMA_DATA_DST_SEL(x)  ((x) << 20)
+		/* 0 - DST_ADDR using DAS
+		 * 1 - GDS
+		 * 3 - DST_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
+		/* 0 - LRU
+		 * 1 - Stream
+		 * 2 - Bypass
+		 */
+#              define PACKET3_DMA_DATA_DST_VOLATILE (1 << 27)
+#              define PACKET3_DMA_DATA_SRC_SEL(x)  ((x) << 29)
+		/* 0 - SRC_ADDR using SAS
+		 * 1 - GDS
+		 * 2 - DATA
+		 * 3 - SRC_ADDR using L2
+		 */
+#              define PACKET3_DMA_DATA_CP_SYNC     (1 << 31)
+/* COMMAND */
+#              define PACKET3_DMA_DATA_DIS_WC      (1 << 21)
+#              define PACKET3_DMA_DATA_CMD_SRC_SWAP(x) ((x) << 22)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_DST_SWAP(x) ((x) << 24)
+		/* 0 - none
+		 * 1 - 8 in 16
+		 * 2 - 8 in 32
+		 * 3 - 8 in 64
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAS     (1 << 26)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_DAS     (1 << 27)
+		/* 0 - memory
+		 * 1 - register
+		 */
+#              define PACKET3_DMA_DATA_CMD_SAIC    (1 << 28)
+#              define PACKET3_DMA_DATA_CMD_DAIC    (1 << 29)
+#              define PACKET3_DMA_DATA_CMD_RAW_WAIT  (1 << 30)
+#define	PACKET3_AQUIRE_MEM				0x58
+#define	PACKET3_REWIND					0x59
+#define	PACKET3_LOAD_UCONFIG_REG			0x5E
+#define	PACKET3_LOAD_SH_REG				0x5F
+#define	PACKET3_LOAD_CONFIG_REG				0x60
+#define	PACKET3_LOAD_CONTEXT_REG			0x61
+#define	PACKET3_SET_CONFIG_REG				0x68
+#define		PACKET3_SET_CONFIG_REG_START			0x00002000
+#define		PACKET3_SET_CONFIG_REG_END			0x00002c00
+#define	PACKET3_SET_CONTEXT_REG				0x69
+#define		PACKET3_SET_CONTEXT_REG_START			0x0000a000
+#define		PACKET3_SET_CONTEXT_REG_END			0x0000a400
+#define	PACKET3_SET_CONTEXT_REG_INDIRECT		0x73
+#define	PACKET3_SET_SH_REG				0x76
+#define		PACKET3_SET_SH_REG_START			0x00002c00
+#define		PACKET3_SET_SH_REG_END				0x00003000
+#define	PACKET3_SET_SH_REG_OFFSET			0x77
+#define	PACKET3_SET_QUEUE_REG				0x78
+#define	PACKET3_SET_UCONFIG_REG				0x79
+#define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
+#define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define	PACKET3_SCRATCH_RAM_WRITE			0x7D
+#define	PACKET3_SCRATCH_RAM_READ			0x7E
+#define	PACKET3_LOAD_CONST_RAM				0x80
+#define	PACKET3_WRITE_CONST_RAM				0x81
+#define	PACKET3_DUMP_CONST_RAM				0x83
+#define	PACKET3_INCREMENT_CE_COUNTER			0x84
+#define	PACKET3_INCREMENT_DE_COUNTER			0x85
+#define	PACKET3_WAIT_ON_CE_COUNTER			0x86
+#define	PACKET3_WAIT_ON_DE_COUNTER_DIFF			0x88
+#define	PACKET3_SWITCH_BUFFER				0x8B
+
+#define VCE_CMD_NO_OP		0x00000000
+#define VCE_CMD_END		0x00000001
+#define VCE_CMD_IB		0x00000002
+#define VCE_CMD_FENCE		0x00000003
+#define VCE_CMD_TRAP		0x00000004
+#define VCE_CMD_IB_AUTO 	0x00000005
+#define VCE_CMD_SEMAPHORE	0x00000006
+
+#endif
-- 
cgit 1.2.3-korg