summaryrefslogtreecommitdiffstats
path: root/kernel/drivers/gpu/drm/radeon/ni_dma.c
blob: ce787a9f12c01fd1f8179d54ef610c27b9be131b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
/*
 * Copyright 2010 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 "radeon.h"
#include "radeon_asic.h"
#include "radeon_trace.h"
#include "nid.h"

u32 cayman_gpu_check_soft_reset(struct radeon_device *rdev);

/*
 * DMA
 * Starting with R600, the GPU has an asynchronous
 * DMA engine.  The programming model is very similar
 * to the 3D engine (ring buffer, IBs, etc.), but the
 * DMA controller has it's own packet format that is
 * different form the PM4 format used by the 3D engine.
 * It supports copying data, writing embedded data,
 * solid fills, and a number of other things.  It also
 * has support for tiling/detiling of buffers.
 * Cayman and newer support two asynchronous DMA engines.
 */

/**
 * cayman_dma_get_rptr - get the current read pointer
 *
 * @rdev: radeon_device pointer
 * @ring: radeon ring pointer
 *
 * Get the current rptr from the hardware (cayman+).
 */
uint32_t cayman_dma_get_rptr(struct radeon_device *rdev,
			     struct radeon_ring *ring)
{
	u32 rptr, reg;

	if (rdev->wb.enabled) {
		rptr = rdev->wb.wb[ring->rptr_offs/4];
	} else {
		if (ring->idx == R600_RING_TYPE_DMA_INDEX)
			reg = DMA_RB_RPTR + DMA0_REGISTER_OFFSET;
		else
			reg = DMA_RB_RPTR + DMA1_REGISTER_OFFSET;

		rptr = RREG32(reg);
	}

	return (rptr & 0x3fffc) >> 2;
}

/**
 * cayman_dma_get_wptr - get the current write pointer
 *
 * @rdev: radeon_device pointer
 * @ring: radeon ring pointer
 *
 * Get the current wptr from the hardware (cayman+).
 */
uint32_t cayman_dma_get_wptr(struct radeon_device *rdev,
			   struct radeon_ring *ring)
{
	u32 reg;

	if (ring->idx == R600_RING_TYPE_DMA_INDEX)
		reg = DMA_RB_WPTR + DMA0_REGISTER_OFFSET;
	else
		reg = DMA_RB_WPTR + DMA1_REGISTER_OFFSET;

	return (RREG32(reg) & 0x3fffc) >> 2;
}

/**
 * cayman_dma_set_wptr - commit the write pointer
 *
 * @rdev: radeon_device pointer
 * @ring: radeon ring pointer
 *
 * Write the wptr back to the hardware (cayman+).
 */
void cayman_dma_set_wptr(struct radeon_device *rdev,
			 struct radeon_ring *ring)
{
	u32 reg;

	if (ring->idx == R600_RING_TYPE_DMA_INDEX)
		reg = DMA_RB_WPTR + DMA0_REGISTER_OFFSET;
	else
		reg = DMA_RB_WPTR + DMA1_REGISTER_OFFSET;

	WREG32(reg, (ring->wptr << 2) & 0x3fffc);
}

/**
 * cayman_dma_ring_ib_execute - Schedule an IB on the DMA engine
 *
 * @rdev: radeon_device pointer
 * @ib: IB object to schedule
 *
 * Schedule an IB in the DMA ring (cayman-SI).
 */
void cayman_dma_ring_ib_execute(struct radeon_device *rdev,
				struct radeon_ib *ib)
{
	struct radeon_ring *ring = &rdev->ring[ib->ring];
	unsigned vm_id = ib->vm ? ib->vm->ids[ib->ring].id : 0;

	if (rdev->wb.enabled) {
		u32 next_rptr = ring->wptr + 4;
		while ((next_rptr & 7) != 5)
			next_rptr++;
		next_rptr += 3;
		radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1));
		radeon_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
		radeon_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xff);
		radeon_ring_write(ring, next_rptr);
	}

	/* The indirect buffer packet must end on an 8 DW boundary in the DMA ring.
	 * Pad as necessary with NOPs.
	 */
	while ((ring->wptr & 7) != 5)
		radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0));
	radeon_ring_write(ring, DMA_IB_PACKET(DMA_PACKET_INDIRECT_BUFFER, vm_id, 0));
	radeon_ring_write(ring, (ib->gpu_addr & 0xFFFFFFE0));
	radeon_ring_write(ring, (ib->length_dw << 12) | (upper_32_bits(ib->gpu_addr) & 0xFF));

}

/**
 * cayman_dma_stop - stop the async dma engines
 *
 * @rdev: radeon_device pointer
 *
 * Stop the async dma engines (cayman-SI).
 */
void cayman_dma_stop(struct radeon_device *rdev)
{
	u32 rb_cntl;

	if ((rdev->asic->copy.copy_ring_index == R600_RING_TYPE_DMA_INDEX) ||
	    (rdev->asic->copy.copy_ring_index == CAYMAN_RING_TYPE_DMA1_INDEX))
		radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size);

	/* dma0 */
	rb_cntl = RREG32(DMA_RB_CNTL + DMA0_REGISTER_OFFSET);
	rb_cntl &= ~DMA_RB_ENABLE;
	WREG32(DMA_RB_CNTL + DMA0_REGISTER_OFFSET, rb_cntl);

	/* dma1 */
	rb_cntl = RREG32(DMA_RB_CNTL + DMA1_REGISTER_OFFSET);
	rb_cntl &= ~DMA_RB_ENABLE;
	WREG32(DMA_RB_CNTL + DMA1_REGISTER_OFFSET, rb_cntl);

	rdev->ring[R600_RING_TYPE_DMA_INDEX].ready = false;
	rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX].ready = false;
}

/**
 * cayman_dma_resume - setup and start the async dma engines
 *
 * @rdev: radeon_device pointer
 *
 * Set up the DMA ring buffers and enable them. (cayman-SI).
 * Returns 0 for success, error for failure.
 */
int cayman_dma_resume(struct radeon_device *rdev)
{
	struct radeon_ring *ring;
	u32 rb_cntl, dma_cntl, ib_cntl;
	u32 rb_bufsz;
	u32 reg_offset, wb_offset;
	int i, r;

	for (i = 0; i < 2; i++) {
		if (i == 0) {
			ring = &rdev->ring[R600_RING_TYPE_DMA_INDEX];
			reg_offset = DMA0_REGISTER_OFFSET;
			wb_offset = R600_WB_DMA_RPTR_OFFSET;
		} else {
			ring = &rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX];
			reg_offset = DMA1_REGISTER_OFFSET;
			wb_offset = CAYMAN_WB_DMA1_RPTR_OFFSET;
		}

		WREG32(DMA_SEM_INCOMPLETE_TIMER_CNTL + reg_offset, 0);
		WREG32(DMA_SEM_WAIT_FAIL_TIMER_CNTL + reg_offset, 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 |= DMA_RB_SWAP_ENABLE | DMA_RPTR_WRITEBACK_SWAP_ENABLE;
#endif
		WREG32(DMA_RB_CNTL + reg_offset, rb_cntl);

		/* Initialize the ring buffer's read and write pointers */
		WREG32(DMA_RB_RPTR + reg_offset, 0);
		WREG32(DMA_RB_WPTR + reg_offset, 0);

		/* set the wb address whether it's enabled or not */
		WREG32(DMA_RB_RPTR_ADDR_HI + reg_offset,
		       upper_32_bits(rdev->wb.gpu_addr + wb_offset) & 0xFF);
		WREG32(DMA_RB_RPTR_ADDR_LO + reg_offset,
		       ((rdev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC));

		if (rdev->wb.enabled)
			rb_cntl |= DMA_RPTR_WRITEBACK_ENABLE;

		WREG32(DMA_RB_BASE + reg_offset, ring->gpu_addr >> 8);

		/* enable DMA IBs */
		ib_cntl = DMA_IB_ENABLE | CMD_VMID_FORCE;
#ifdef __BIG_ENDIAN
		ib_cntl |= DMA_IB_SWAP_ENABLE;
#endif
		WREG32(DMA_IB_CNTL + reg_offset, ib_cntl);

		dma_cntl = RREG32(DMA_CNTL + reg_offset);
		dma_cntl &= ~CTXEMPTY_INT_ENABLE;
		WREG32(DMA_CNTL + reg_offset, dma_cntl);

		ring->wptr = 0;
		WREG32(DMA_RB_WPTR + reg_offset, ring->wptr << 2);

		WREG32(DMA_RB_CNTL + reg_offset, rb_cntl | DMA_RB_ENABLE);

		ring->ready = true;

		r = radeon_ring_test(rdev, ring->idx, ring);
		if (r) {
			ring->ready = false;
			return r;
		}
	}

	if ((rdev->asic->copy.copy_ring_index == R600_RING_TYPE_DMA_INDEX) ||
	    (rdev->asic->copy.copy_ring_index == CAYMAN_RING_TYPE_DMA1_INDEX))
		radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size);

	return 0;
}

/**
 * cayman_dma_fini - tear down the async dma engines
 *
 * @rdev: radeon_device pointer
 *
 * Stop the async dma engines and free the rings (cayman-SI).
 */
void cayman_dma_fini(struct radeon_device *rdev)
{
	cayman_dma_stop(rdev);
	radeon_ring_fini(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX]);
	radeon_ring_fini(rdev, &rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX]);
}

/**
 * cayman_dma_is_lockup - Check if the DMA engine is locked up
 *
 * @rdev: radeon_device pointer
 * @ring: radeon_ring structure holding ring information
 *
 * Check if the async DMA engine is locked up.
 * Returns true if the engine appears to be locked up, false if not.
 */
bool cayman_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring)
{
	u32 reset_mask = cayman_gpu_check_soft_reset(rdev);
	u32 mask;

	if (ring->idx == R600_RING_TYPE_DMA_INDEX)
		mask = RADEON_RESET_DMA;
	else
		mask = RADEON_RESET_DMA1;

	if (!(reset_mask & mask)) {
		radeon_ring_lockup_update(rdev, ring);
		return false;
	}
	return radeon_ring_test_lockup(rdev, ring);
}

/**
 * cayman_dma_vm_copy_pages - update PTEs by copying them from the GART
 *
 * @rdev: radeon_device pointer
 * @ib: indirect buffer to fill with commands
 * @pe: addr of the page entry
 * @src: src addr where to copy from
 * @count: number of page entries to update
 *
 * Update PTEs by copying them from the GART using the DMA (cayman/TN).
 */
void cayman_dma_vm_copy_pages(struct radeon_device *rdev,
			      struct radeon_ib *ib,
			      uint64_t pe, uint64_t src,
			      unsigned count)
{
	unsigned ndw;

	while (count) {
		ndw = count * 2;
		if (ndw > 0xFFFFE)
			ndw = 0xFFFFE;

		ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_COPY,
						      0, 0, ndw);
		ib->ptr[ib->length_dw++] = lower_32_bits(pe);
		ib->ptr[ib->length_dw++] = lower_32_bits(src);
		ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
		ib->ptr[ib->length_dw++] = upper_32_bits(src) & 0xff;

		pe += ndw * 4;
		src += ndw * 4;
		count -= ndw / 2;
	}
}

/**
 * cayman_dma_vm_write_pages - update PTEs by writing them manually
 *
 * @rdev: radeon_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
 *
 * Update PTEs by writing them manually using the DMA (cayman/TN).
 */
void cayman_dma_vm_write_pages(struct radeon_device *rdev,
			       struct radeon_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++] = DMA_PACKET(DMA_PACKET_WRITE,
						      0, 0, ndw);
		ib->ptr[ib->length_dw++] = pe;
		ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
		for (; ndw > 0; ndw -= 2, --count, pe += 8) {
			if (flags & R600_PTE_SYSTEM) {
				value = radeon_vm_map_gart(rdev, addr);
			} else if (flags & R600_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);
		}
	}
}

/**
 * cayman_dma_vm_set_pages - update the page tables using the DMA
 *
 * @rdev: radeon_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
 *
 * Update the page tables using the DMA (cayman/TN).
 */
void cayman_dma_vm_set_pages(struct radeon_device *rdev,
			     struct radeon_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;

		if (flags & R600_PTE_VALID)
			value = addr;
		else
			value = 0;

		/* for physically contiguous pages (vram) */
		ib->ptr[ib->length_dw++] = DMA_PTE_PDE_PACKET(ndw);
		ib->ptr[ib->length_dw++] = pe; /* dst addr */
		ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
		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;

		pe += ndw * 4;
		addr += (ndw / 2) * incr;
		count -= ndw / 2;
	}
}

/**
 * cayman_dma_vm_pad_ib - pad the IB to the required number of dw
 *
 * @ib: indirect buffer to fill with padding
 *
 */
void cayman_dma_vm_pad_ib(struct radeon_ib *ib)
{
	while (ib->length_dw & 0x7)
		ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0);
}

void cayman_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring,
			 unsigned vm_id, uint64_t pd_addr)
{
	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0));
	radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2));
	radeon_ring_write(ring, pd_addr >> 12);

	/* flush hdp cache */
	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0));
	radeon_ring_write(ring, (0xf << 16) | (HDP_MEM_COHERENCY_FLUSH_CNTL >> 2));
	radeon_ring_write(ring, 1);

	/* bits 0-7 are the VM contexts0-7 */
	radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0));
	radeon_ring_write(ring, (0xf << 16) | (VM_INVALIDATE_REQUEST >> 2));
	radeon_ring_write(ring, 1 << vm_id);

	/* wait for invalidate to complete */
	radeon_ring_write(ring, DMA_SRBM_READ_PACKET);
	radeon_ring_write(ring, (0xff << 20) | (VM_INVALIDATE_REQUEST >> 2));
	radeon_ring_write(ring, 0); /* mask */
	radeon_ring_write(ring, 0); /* value */
}