diff options
author | 2015-08-04 12:17:53 -0700 | |
---|---|---|
committer | 2015-08-04 15:44:42 -0700 | |
commit | 9ca8dbcc65cfc63d6f5ef3312a33184e1d726e00 (patch) | |
tree | 1c9cafbcd35f783a87880a10f85d1a060db1a563 /kernel/drivers/infiniband/hw/cxgb4 | |
parent | 98260f3884f4a202f9ca5eabed40b1354c489b29 (diff) |
Add the rt linux 4.1.3-rt3 as base
Import the rt linux 4.1.3-rt3 as OPNFV kvm base.
It's from git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git linux-4.1.y-rt and
the base is:
commit 0917f823c59692d751951bf5ea699a2d1e2f26a2
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Date: Sat Jul 25 12:13:34 2015 +0200
Prepare v4.1.3-rt3
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
We lose all the git history this way and it's not good. We
should apply another opnfv project repo in future.
Change-Id: I87543d81c9df70d99c5001fbdf646b202c19f423
Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
Diffstat (limited to 'kernel/drivers/infiniband/hw/cxgb4')
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/Kconfig | 18 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/Makefile | 5 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/cm.c | 4053 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/cq.c | 1015 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/device.c | 1564 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/ev.c | 244 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/id_table.c | 112 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/iw_cxgb4.h | 1042 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/mem.c | 979 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/provider.c | 588 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/qp.c | 1886 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/resource.c | 453 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/t4.h | 685 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/t4fw_ri_api.h | 855 | ||||
-rw-r--r-- | kernel/drivers/infiniband/hw/cxgb4/user.h | 80 |
15 files changed, 13579 insertions, 0 deletions
diff --git a/kernel/drivers/infiniband/hw/cxgb4/Kconfig b/kernel/drivers/infiniband/hw/cxgb4/Kconfig new file mode 100644 index 000000000..23f38cf2c --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/Kconfig @@ -0,0 +1,18 @@ +config INFINIBAND_CXGB4 + tristate "Chelsio T4/T5 RDMA Driver" + depends on CHELSIO_T4 && INET && (IPV6 || IPV6=n) + select GENERIC_ALLOCATOR + ---help--- + This is an iWARP/RDMA driver for the Chelsio T4 and T5 + 1GbE, 10GbE adapters and T5 40GbE adapter. + + For general information about Chelsio and our products, visit + our website at <http://www.chelsio.com>. + + For customer support, please visit our customer support page at + <http://www.chelsio.com/support.html>. + + Please send feedback to <linux-bugs@chelsio.com>. + + To compile this driver as a module, choose M here: the module + will be called iw_cxgb4. diff --git a/kernel/drivers/infiniband/hw/cxgb4/Makefile b/kernel/drivers/infiniband/hw/cxgb4/Makefile new file mode 100644 index 000000000..e11cf7299 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/Makefile @@ -0,0 +1,5 @@ +ccflags-y := -Idrivers/net/ethernet/chelsio/cxgb4 + +obj-$(CONFIG_INFINIBAND_CXGB4) += iw_cxgb4.o + +iw_cxgb4-y := device.o cm.o provider.o mem.o cq.o qp.o resource.o ev.o id_table.o diff --git a/kernel/drivers/infiniband/hw/cxgb4/cm.c b/kernel/drivers/infiniband/hw/cxgb4/cm.c new file mode 100644 index 000000000..3ad8dc798 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/cm.c @@ -0,0 +1,4053 @@ +/* + * Copyright (c) 2009-2014 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#include <linux/module.h> +#include <linux/list.h> +#include <linux/workqueue.h> +#include <linux/skbuff.h> +#include <linux/timer.h> +#include <linux/notifier.h> +#include <linux/inetdevice.h> +#include <linux/ip.h> +#include <linux/tcp.h> +#include <linux/if_vlan.h> + +#include <net/neighbour.h> +#include <net/netevent.h> +#include <net/route.h> +#include <net/tcp.h> +#include <net/ip6_route.h> +#include <net/addrconf.h> + +#include <rdma/ib_addr.h> + +#include "iw_cxgb4.h" + +static char *states[] = { + "idle", + "listen", + "connecting", + "mpa_wait_req", + "mpa_req_sent", + "mpa_req_rcvd", + "mpa_rep_sent", + "fpdu_mode", + "aborting", + "closing", + "moribund", + "dead", + NULL, +}; + +static int nocong; +module_param(nocong, int, 0644); +MODULE_PARM_DESC(nocong, "Turn of congestion control (default=0)"); + +static int enable_ecn; +module_param(enable_ecn, int, 0644); +MODULE_PARM_DESC(enable_ecn, "Enable ECN (default=0/disabled)"); + +static int dack_mode = 1; +module_param(dack_mode, int, 0644); +MODULE_PARM_DESC(dack_mode, "Delayed ack mode (default=1)"); + +uint c4iw_max_read_depth = 32; +module_param(c4iw_max_read_depth, int, 0644); +MODULE_PARM_DESC(c4iw_max_read_depth, + "Per-connection max ORD/IRD (default=32)"); + +static int enable_tcp_timestamps; +module_param(enable_tcp_timestamps, int, 0644); +MODULE_PARM_DESC(enable_tcp_timestamps, "Enable tcp timestamps (default=0)"); + +static int enable_tcp_sack; +module_param(enable_tcp_sack, int, 0644); +MODULE_PARM_DESC(enable_tcp_sack, "Enable tcp SACK (default=0)"); + +static int enable_tcp_window_scaling = 1; +module_param(enable_tcp_window_scaling, int, 0644); +MODULE_PARM_DESC(enable_tcp_window_scaling, + "Enable tcp window scaling (default=1)"); + +int c4iw_debug; +module_param(c4iw_debug, int, 0644); +MODULE_PARM_DESC(c4iw_debug, "Enable debug logging (default=0)"); + +static int peer2peer = 1; +module_param(peer2peer, int, 0644); +MODULE_PARM_DESC(peer2peer, "Support peer2peer ULPs (default=1)"); + +static int p2p_type = FW_RI_INIT_P2PTYPE_READ_REQ; +module_param(p2p_type, int, 0644); +MODULE_PARM_DESC(p2p_type, "RDMAP opcode to use for the RTR message: " + "1=RDMA_READ 0=RDMA_WRITE (default 1)"); + +static int ep_timeout_secs = 60; +module_param(ep_timeout_secs, int, 0644); +MODULE_PARM_DESC(ep_timeout_secs, "CM Endpoint operation timeout " + "in seconds (default=60)"); + +static int mpa_rev = 1; +module_param(mpa_rev, int, 0644); +MODULE_PARM_DESC(mpa_rev, "MPA Revision, 0 supports amso1100, " + "1 is RFC0544 spec compliant, 2 is IETF MPA Peer Connect Draft" + " compliant (default=1)"); + +static int markers_enabled; +module_param(markers_enabled, int, 0644); +MODULE_PARM_DESC(markers_enabled, "Enable MPA MARKERS (default(0)=disabled)"); + +static int crc_enabled = 1; +module_param(crc_enabled, int, 0644); +MODULE_PARM_DESC(crc_enabled, "Enable MPA CRC (default(1)=enabled)"); + +static int rcv_win = 256 * 1024; +module_param(rcv_win, int, 0644); +MODULE_PARM_DESC(rcv_win, "TCP receive window in bytes (default=256KB)"); + +static int snd_win = 128 * 1024; +module_param(snd_win, int, 0644); +MODULE_PARM_DESC(snd_win, "TCP send window in bytes (default=128KB)"); + +static struct workqueue_struct *workq; + +static struct sk_buff_head rxq; + +static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp); +static void ep_timeout(unsigned long arg); +static void connect_reply_upcall(struct c4iw_ep *ep, int status); + +static LIST_HEAD(timeout_list); +static spinlock_t timeout_lock; + +static void deref_qp(struct c4iw_ep *ep) +{ + c4iw_qp_rem_ref(&ep->com.qp->ibqp); + clear_bit(QP_REFERENCED, &ep->com.flags); +} + +static void ref_qp(struct c4iw_ep *ep) +{ + set_bit(QP_REFERENCED, &ep->com.flags); + c4iw_qp_add_ref(&ep->com.qp->ibqp); +} + +static void start_ep_timer(struct c4iw_ep *ep) +{ + PDBG("%s ep %p\n", __func__, ep); + if (timer_pending(&ep->timer)) { + pr_err("%s timer already started! ep %p\n", + __func__, ep); + return; + } + clear_bit(TIMEOUT, &ep->com.flags); + c4iw_get_ep(&ep->com); + ep->timer.expires = jiffies + ep_timeout_secs * HZ; + ep->timer.data = (unsigned long)ep; + ep->timer.function = ep_timeout; + add_timer(&ep->timer); +} + +static int stop_ep_timer(struct c4iw_ep *ep) +{ + PDBG("%s ep %p stopping\n", __func__, ep); + del_timer_sync(&ep->timer); + if (!test_and_set_bit(TIMEOUT, &ep->com.flags)) { + c4iw_put_ep(&ep->com); + return 0; + } + return 1; +} + +static int c4iw_l2t_send(struct c4iw_rdev *rdev, struct sk_buff *skb, + struct l2t_entry *l2e) +{ + int error = 0; + + if (c4iw_fatal_error(rdev)) { + kfree_skb(skb); + PDBG("%s - device in error state - dropping\n", __func__); + return -EIO; + } + error = cxgb4_l2t_send(rdev->lldi.ports[0], skb, l2e); + if (error < 0) + kfree_skb(skb); + return error < 0 ? error : 0; +} + +int c4iw_ofld_send(struct c4iw_rdev *rdev, struct sk_buff *skb) +{ + int error = 0; + + if (c4iw_fatal_error(rdev)) { + kfree_skb(skb); + PDBG("%s - device in error state - dropping\n", __func__); + return -EIO; + } + error = cxgb4_ofld_send(rdev->lldi.ports[0], skb); + if (error < 0) + kfree_skb(skb); + return error < 0 ? error : 0; +} + +static void release_tid(struct c4iw_rdev *rdev, u32 hwtid, struct sk_buff *skb) +{ + struct cpl_tid_release *req; + + skb = get_skb(skb, sizeof *req, GFP_KERNEL); + if (!skb) + return; + req = (struct cpl_tid_release *) skb_put(skb, sizeof(*req)); + INIT_TP_WR(req, hwtid); + OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_TID_RELEASE, hwtid)); + set_wr_txq(skb, CPL_PRIORITY_SETUP, 0); + c4iw_ofld_send(rdev, skb); + return; +} + +static void set_emss(struct c4iw_ep *ep, u16 opt) +{ + ep->emss = ep->com.dev->rdev.lldi.mtus[TCPOPT_MSS_G(opt)] - + ((AF_INET == ep->com.remote_addr.ss_family) ? + sizeof(struct iphdr) : sizeof(struct ipv6hdr)) - + sizeof(struct tcphdr); + ep->mss = ep->emss; + if (TCPOPT_TSTAMP_G(opt)) + ep->emss -= round_up(TCPOLEN_TIMESTAMP, 4); + if (ep->emss < 128) + ep->emss = 128; + if (ep->emss & 7) + PDBG("Warning: misaligned mtu idx %u mss %u emss=%u\n", + TCPOPT_MSS_G(opt), ep->mss, ep->emss); + PDBG("%s mss_idx %u mss %u emss=%u\n", __func__, TCPOPT_MSS_G(opt), + ep->mss, ep->emss); +} + +static enum c4iw_ep_state state_read(struct c4iw_ep_common *epc) +{ + enum c4iw_ep_state state; + + mutex_lock(&epc->mutex); + state = epc->state; + mutex_unlock(&epc->mutex); + return state; +} + +static void __state_set(struct c4iw_ep_common *epc, enum c4iw_ep_state new) +{ + epc->state = new; +} + +static void state_set(struct c4iw_ep_common *epc, enum c4iw_ep_state new) +{ + mutex_lock(&epc->mutex); + PDBG("%s - %s -> %s\n", __func__, states[epc->state], states[new]); + __state_set(epc, new); + mutex_unlock(&epc->mutex); + return; +} + +static void *alloc_ep(int size, gfp_t gfp) +{ + struct c4iw_ep_common *epc; + + epc = kzalloc(size, gfp); + if (epc) { + kref_init(&epc->kref); + mutex_init(&epc->mutex); + c4iw_init_wr_wait(&epc->wr_wait); + } + PDBG("%s alloc ep %p\n", __func__, epc); + return epc; +} + +void _c4iw_free_ep(struct kref *kref) +{ + struct c4iw_ep *ep; + + ep = container_of(kref, struct c4iw_ep, com.kref); + PDBG("%s ep %p state %s\n", __func__, ep, states[state_read(&ep->com)]); + if (test_bit(QP_REFERENCED, &ep->com.flags)) + deref_qp(ep); + if (test_bit(RELEASE_RESOURCES, &ep->com.flags)) { + remove_handle(ep->com.dev, &ep->com.dev->hwtid_idr, ep->hwtid); + cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, ep->hwtid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + } + if (test_bit(RELEASE_MAPINFO, &ep->com.flags)) { + print_addr(&ep->com, __func__, "remove_mapinfo/mapping"); + iwpm_remove_mapinfo(&ep->com.local_addr, + &ep->com.mapped_local_addr); + iwpm_remove_mapping(&ep->com.local_addr, RDMA_NL_C4IW); + } + kfree(ep); +} + +static void release_ep_resources(struct c4iw_ep *ep) +{ + set_bit(RELEASE_RESOURCES, &ep->com.flags); + c4iw_put_ep(&ep->com); +} + +static int status2errno(int status) +{ + switch (status) { + case CPL_ERR_NONE: + return 0; + case CPL_ERR_CONN_RESET: + return -ECONNRESET; + case CPL_ERR_ARP_MISS: + return -EHOSTUNREACH; + case CPL_ERR_CONN_TIMEDOUT: + return -ETIMEDOUT; + case CPL_ERR_TCAM_FULL: + return -ENOMEM; + case CPL_ERR_CONN_EXIST: + return -EADDRINUSE; + default: + return -EIO; + } +} + +/* + * Try and reuse skbs already allocated... + */ +static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp) +{ + if (skb && !skb_is_nonlinear(skb) && !skb_cloned(skb)) { + skb_trim(skb, 0); + skb_get(skb); + skb_reset_transport_header(skb); + } else { + skb = alloc_skb(len, gfp); + } + t4_set_arp_err_handler(skb, NULL, NULL); + return skb; +} + +static struct net_device *get_real_dev(struct net_device *egress_dev) +{ + return rdma_vlan_dev_real_dev(egress_dev) ? : egress_dev; +} + +static int our_interface(struct c4iw_dev *dev, struct net_device *egress_dev) +{ + int i; + + egress_dev = get_real_dev(egress_dev); + for (i = 0; i < dev->rdev.lldi.nports; i++) + if (dev->rdev.lldi.ports[i] == egress_dev) + return 1; + return 0; +} + +static struct dst_entry *find_route6(struct c4iw_dev *dev, __u8 *local_ip, + __u8 *peer_ip, __be16 local_port, + __be16 peer_port, u8 tos, + __u32 sin6_scope_id) +{ + struct dst_entry *dst = NULL; + + if (IS_ENABLED(CONFIG_IPV6)) { + struct flowi6 fl6; + + memset(&fl6, 0, sizeof(fl6)); + memcpy(&fl6.daddr, peer_ip, 16); + memcpy(&fl6.saddr, local_ip, 16); + if (ipv6_addr_type(&fl6.daddr) & IPV6_ADDR_LINKLOCAL) + fl6.flowi6_oif = sin6_scope_id; + dst = ip6_route_output(&init_net, NULL, &fl6); + if (!dst) + goto out; + if (!our_interface(dev, ip6_dst_idev(dst)->dev) && + !(ip6_dst_idev(dst)->dev->flags & IFF_LOOPBACK)) { + dst_release(dst); + dst = NULL; + } + } + +out: + return dst; +} + +static struct dst_entry *find_route(struct c4iw_dev *dev, __be32 local_ip, + __be32 peer_ip, __be16 local_port, + __be16 peer_port, u8 tos) +{ + struct rtable *rt; + struct flowi4 fl4; + struct neighbour *n; + + rt = ip_route_output_ports(&init_net, &fl4, NULL, peer_ip, local_ip, + peer_port, local_port, IPPROTO_TCP, + tos, 0); + if (IS_ERR(rt)) + return NULL; + n = dst_neigh_lookup(&rt->dst, &peer_ip); + if (!n) + return NULL; + if (!our_interface(dev, n->dev) && + !(n->dev->flags & IFF_LOOPBACK)) { + neigh_release(n); + dst_release(&rt->dst); + return NULL; + } + neigh_release(n); + return &rt->dst; +} + +static void arp_failure_discard(void *handle, struct sk_buff *skb) +{ + PDBG("%s c4iw_dev %p\n", __func__, handle); + kfree_skb(skb); +} + +/* + * Handle an ARP failure for an active open. + */ +static void act_open_req_arp_failure(void *handle, struct sk_buff *skb) +{ + struct c4iw_ep *ep = handle; + + printk(KERN_ERR MOD "ARP failure duing connect\n"); + kfree_skb(skb); + connect_reply_upcall(ep, -EHOSTUNREACH); + state_set(&ep->com, DEAD); + remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid); + cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + c4iw_put_ep(&ep->com); +} + +/* + * Handle an ARP failure for a CPL_ABORT_REQ. Change it into a no RST variant + * and send it along. + */ +static void abort_arp_failure(void *handle, struct sk_buff *skb) +{ + struct c4iw_rdev *rdev = handle; + struct cpl_abort_req *req = cplhdr(skb); + + PDBG("%s rdev %p\n", __func__, rdev); + req->cmd = CPL_ABORT_NO_RST; + c4iw_ofld_send(rdev, skb); +} + +static void send_flowc(struct c4iw_ep *ep, struct sk_buff *skb) +{ + unsigned int flowclen = 80; + struct fw_flowc_wr *flowc; + int i; + + skb = get_skb(skb, flowclen, GFP_KERNEL); + flowc = (struct fw_flowc_wr *)__skb_put(skb, flowclen); + + flowc->op_to_nparams = cpu_to_be32(FW_WR_OP_V(FW_FLOWC_WR) | + FW_FLOWC_WR_NPARAMS_V(8)); + flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(flowclen, + 16)) | FW_WR_FLOWID_V(ep->hwtid)); + + flowc->mnemval[0].mnemonic = FW_FLOWC_MNEM_PFNVFN; + flowc->mnemval[0].val = cpu_to_be32(FW_PFVF_CMD_PFN_V + (ep->com.dev->rdev.lldi.pf)); + flowc->mnemval[1].mnemonic = FW_FLOWC_MNEM_CH; + flowc->mnemval[1].val = cpu_to_be32(ep->tx_chan); + flowc->mnemval[2].mnemonic = FW_FLOWC_MNEM_PORT; + flowc->mnemval[2].val = cpu_to_be32(ep->tx_chan); + flowc->mnemval[3].mnemonic = FW_FLOWC_MNEM_IQID; + flowc->mnemval[3].val = cpu_to_be32(ep->rss_qid); + flowc->mnemval[4].mnemonic = FW_FLOWC_MNEM_SNDNXT; + flowc->mnemval[4].val = cpu_to_be32(ep->snd_seq); + flowc->mnemval[5].mnemonic = FW_FLOWC_MNEM_RCVNXT; + flowc->mnemval[5].val = cpu_to_be32(ep->rcv_seq); + flowc->mnemval[6].mnemonic = FW_FLOWC_MNEM_SNDBUF; + flowc->mnemval[6].val = cpu_to_be32(ep->snd_win); + flowc->mnemval[7].mnemonic = FW_FLOWC_MNEM_MSS; + flowc->mnemval[7].val = cpu_to_be32(ep->emss); + /* Pad WR to 16 byte boundary */ + flowc->mnemval[8].mnemonic = 0; + flowc->mnemval[8].val = 0; + for (i = 0; i < 9; i++) { + flowc->mnemval[i].r4[0] = 0; + flowc->mnemval[i].r4[1] = 0; + flowc->mnemval[i].r4[2] = 0; + } + + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + c4iw_ofld_send(&ep->com.dev->rdev, skb); +} + +static int send_halfclose(struct c4iw_ep *ep, gfp_t gfp) +{ + struct cpl_close_con_req *req; + struct sk_buff *skb; + int wrlen = roundup(sizeof *req, 16); + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + skb = get_skb(NULL, wrlen, gfp); + if (!skb) { + printk(KERN_ERR MOD "%s - failed to alloc skb\n", __func__); + return -ENOMEM; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + t4_set_arp_err_handler(skb, NULL, arp_failure_discard); + req = (struct cpl_close_con_req *) skb_put(skb, wrlen); + memset(req, 0, wrlen); + INIT_TP_WR(req, ep->hwtid); + OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_CLOSE_CON_REQ, + ep->hwtid)); + return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); +} + +static int send_abort(struct c4iw_ep *ep, struct sk_buff *skb, gfp_t gfp) +{ + struct cpl_abort_req *req; + int wrlen = roundup(sizeof *req, 16); + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + skb = get_skb(skb, wrlen, gfp); + if (!skb) { + printk(KERN_ERR MOD "%s - failed to alloc skb.\n", + __func__); + return -ENOMEM; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + t4_set_arp_err_handler(skb, &ep->com.dev->rdev, abort_arp_failure); + req = (struct cpl_abort_req *) skb_put(skb, wrlen); + memset(req, 0, wrlen); + INIT_TP_WR(req, ep->hwtid); + OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_ABORT_REQ, ep->hwtid)); + req->cmd = CPL_ABORT_SEND_RST; + return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); +} + +/* + * c4iw_form_pm_msg - Form a port mapper message with mapping info + */ +static void c4iw_form_pm_msg(struct c4iw_ep *ep, + struct iwpm_sa_data *pm_msg) +{ + memcpy(&pm_msg->loc_addr, &ep->com.local_addr, + sizeof(ep->com.local_addr)); + memcpy(&pm_msg->rem_addr, &ep->com.remote_addr, + sizeof(ep->com.remote_addr)); +} + +/* + * c4iw_form_reg_msg - Form a port mapper message with dev info + */ +static void c4iw_form_reg_msg(struct c4iw_dev *dev, + struct iwpm_dev_data *pm_msg) +{ + memcpy(pm_msg->dev_name, dev->ibdev.name, IWPM_DEVNAME_SIZE); + memcpy(pm_msg->if_name, dev->rdev.lldi.ports[0]->name, + IWPM_IFNAME_SIZE); +} + +static void c4iw_record_pm_msg(struct c4iw_ep *ep, + struct iwpm_sa_data *pm_msg) +{ + memcpy(&ep->com.mapped_local_addr, &pm_msg->mapped_loc_addr, + sizeof(ep->com.mapped_local_addr)); + memcpy(&ep->com.mapped_remote_addr, &pm_msg->mapped_rem_addr, + sizeof(ep->com.mapped_remote_addr)); +} + +static int get_remote_addr(struct c4iw_ep *parent_ep, struct c4iw_ep *child_ep) +{ + int ret; + + print_addr(&parent_ep->com, __func__, "get_remote_addr parent_ep "); + print_addr(&child_ep->com, __func__, "get_remote_addr child_ep "); + + ret = iwpm_get_remote_info(&parent_ep->com.mapped_local_addr, + &child_ep->com.mapped_remote_addr, + &child_ep->com.remote_addr, RDMA_NL_C4IW); + if (ret) + PDBG("Unable to find remote peer addr info - err %d\n", ret); + + return ret; +} + +static void best_mtu(const unsigned short *mtus, unsigned short mtu, + unsigned int *idx, int use_ts, int ipv6) +{ + unsigned short hdr_size = (ipv6 ? + sizeof(struct ipv6hdr) : + sizeof(struct iphdr)) + + sizeof(struct tcphdr) + + (use_ts ? + round_up(TCPOLEN_TIMESTAMP, 4) : 0); + unsigned short data_size = mtu - hdr_size; + + cxgb4_best_aligned_mtu(mtus, hdr_size, data_size, 8, idx); +} + +static int send_connect(struct c4iw_ep *ep) +{ + struct cpl_act_open_req *req; + struct cpl_t5_act_open_req *t5_req; + struct cpl_act_open_req6 *req6; + struct cpl_t5_act_open_req6 *t5_req6; + struct sk_buff *skb; + u64 opt0; + u32 opt2; + unsigned int mtu_idx; + int wscale; + int wrlen; + int sizev4 = is_t4(ep->com.dev->rdev.lldi.adapter_type) ? + sizeof(struct cpl_act_open_req) : + sizeof(struct cpl_t5_act_open_req); + int sizev6 = is_t4(ep->com.dev->rdev.lldi.adapter_type) ? + sizeof(struct cpl_act_open_req6) : + sizeof(struct cpl_t5_act_open_req6); + struct sockaddr_in *la = (struct sockaddr_in *) + &ep->com.mapped_local_addr; + struct sockaddr_in *ra = (struct sockaddr_in *) + &ep->com.mapped_remote_addr; + struct sockaddr_in6 *la6 = (struct sockaddr_in6 *) + &ep->com.mapped_local_addr; + struct sockaddr_in6 *ra6 = (struct sockaddr_in6 *) + &ep->com.mapped_remote_addr; + int win; + + wrlen = (ep->com.remote_addr.ss_family == AF_INET) ? + roundup(sizev4, 16) : + roundup(sizev6, 16); + + PDBG("%s ep %p atid %u\n", __func__, ep, ep->atid); + + skb = get_skb(NULL, wrlen, GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "%s - failed to alloc skb.\n", + __func__); + return -ENOMEM; + } + set_wr_txq(skb, CPL_PRIORITY_SETUP, ep->ctrlq_idx); + + best_mtu(ep->com.dev->rdev.lldi.mtus, ep->mtu, &mtu_idx, + enable_tcp_timestamps, + (AF_INET == ep->com.remote_addr.ss_family) ? 0 : 1); + wscale = compute_wscale(rcv_win); + + /* + * Specify the largest window that will fit in opt0. The + * remainder will be specified in the rx_data_ack. + */ + win = ep->rcv_win >> 10; + if (win > RCV_BUFSIZ_M) + win = RCV_BUFSIZ_M; + + opt0 = (nocong ? NO_CONG_F : 0) | + KEEP_ALIVE_F | + DELACK_F | + WND_SCALE_V(wscale) | + MSS_IDX_V(mtu_idx) | + L2T_IDX_V(ep->l2t->idx) | + TX_CHAN_V(ep->tx_chan) | + SMAC_SEL_V(ep->smac_idx) | + DSCP_V(ep->tos) | + ULP_MODE_V(ULP_MODE_TCPDDP) | + RCV_BUFSIZ_V(win); + opt2 = RX_CHANNEL_V(0) | + CCTRL_ECN_V(enable_ecn) | + RSS_QUEUE_VALID_F | RSS_QUEUE_V(ep->rss_qid); + if (enable_tcp_timestamps) + opt2 |= TSTAMPS_EN_F; + if (enable_tcp_sack) + opt2 |= SACK_EN_F; + if (wscale && enable_tcp_window_scaling) + opt2 |= WND_SCALE_EN_F; + if (is_t5(ep->com.dev->rdev.lldi.adapter_type)) { + opt2 |= T5_OPT_2_VALID_F; + opt2 |= CONG_CNTRL_V(CONG_ALG_TAHOE); + opt2 |= T5_ISS_F; + } + t4_set_arp_err_handler(skb, ep, act_open_req_arp_failure); + + if (is_t4(ep->com.dev->rdev.lldi.adapter_type)) { + if (ep->com.remote_addr.ss_family == AF_INET) { + req = (struct cpl_act_open_req *) skb_put(skb, wrlen); + INIT_TP_WR(req, 0); + OPCODE_TID(req) = cpu_to_be32( + MK_OPCODE_TID(CPL_ACT_OPEN_REQ, + ((ep->rss_qid << 14) | ep->atid))); + req->local_port = la->sin_port; + req->peer_port = ra->sin_port; + req->local_ip = la->sin_addr.s_addr; + req->peer_ip = ra->sin_addr.s_addr; + req->opt0 = cpu_to_be64(opt0); + req->params = cpu_to_be32(cxgb4_select_ntuple( + ep->com.dev->rdev.lldi.ports[0], + ep->l2t)); + req->opt2 = cpu_to_be32(opt2); + } else { + req6 = (struct cpl_act_open_req6 *)skb_put(skb, wrlen); + + INIT_TP_WR(req6, 0); + OPCODE_TID(req6) = cpu_to_be32( + MK_OPCODE_TID(CPL_ACT_OPEN_REQ6, + ((ep->rss_qid<<14)|ep->atid))); + req6->local_port = la6->sin6_port; + req6->peer_port = ra6->sin6_port; + req6->local_ip_hi = *((__be64 *) + (la6->sin6_addr.s6_addr)); + req6->local_ip_lo = *((__be64 *) + (la6->sin6_addr.s6_addr + 8)); + req6->peer_ip_hi = *((__be64 *) + (ra6->sin6_addr.s6_addr)); + req6->peer_ip_lo = *((__be64 *) + (ra6->sin6_addr.s6_addr + 8)); + req6->opt0 = cpu_to_be64(opt0); + req6->params = cpu_to_be32(cxgb4_select_ntuple( + ep->com.dev->rdev.lldi.ports[0], + ep->l2t)); + req6->opt2 = cpu_to_be32(opt2); + } + } else { + u32 isn = (prandom_u32() & ~7UL) - 1; + + if (peer2peer) + isn += 4; + + if (ep->com.remote_addr.ss_family == AF_INET) { + t5_req = (struct cpl_t5_act_open_req *) + skb_put(skb, wrlen); + INIT_TP_WR(t5_req, 0); + OPCODE_TID(t5_req) = cpu_to_be32( + MK_OPCODE_TID(CPL_ACT_OPEN_REQ, + ((ep->rss_qid << 14) | ep->atid))); + t5_req->local_port = la->sin_port; + t5_req->peer_port = ra->sin_port; + t5_req->local_ip = la->sin_addr.s_addr; + t5_req->peer_ip = ra->sin_addr.s_addr; + t5_req->opt0 = cpu_to_be64(opt0); + t5_req->params = cpu_to_be64(FILTER_TUPLE_V( + cxgb4_select_ntuple( + ep->com.dev->rdev.lldi.ports[0], + ep->l2t))); + t5_req->rsvd = cpu_to_be32(isn); + PDBG("%s snd_isn %u\n", __func__, + be32_to_cpu(t5_req->rsvd)); + t5_req->opt2 = cpu_to_be32(opt2); + } else { + t5_req6 = (struct cpl_t5_act_open_req6 *) + skb_put(skb, wrlen); + INIT_TP_WR(t5_req6, 0); + OPCODE_TID(t5_req6) = cpu_to_be32( + MK_OPCODE_TID(CPL_ACT_OPEN_REQ6, + ((ep->rss_qid<<14)|ep->atid))); + t5_req6->local_port = la6->sin6_port; + t5_req6->peer_port = ra6->sin6_port; + t5_req6->local_ip_hi = *((__be64 *) + (la6->sin6_addr.s6_addr)); + t5_req6->local_ip_lo = *((__be64 *) + (la6->sin6_addr.s6_addr + 8)); + t5_req6->peer_ip_hi = *((__be64 *) + (ra6->sin6_addr.s6_addr)); + t5_req6->peer_ip_lo = *((__be64 *) + (ra6->sin6_addr.s6_addr + 8)); + t5_req6->opt0 = cpu_to_be64(opt0); + t5_req6->params = cpu_to_be64(FILTER_TUPLE_V( + cxgb4_select_ntuple( + ep->com.dev->rdev.lldi.ports[0], + ep->l2t))); + t5_req6->rsvd = cpu_to_be32(isn); + PDBG("%s snd_isn %u\n", __func__, + be32_to_cpu(t5_req6->rsvd)); + t5_req6->opt2 = cpu_to_be32(opt2); + } + } + + set_bit(ACT_OPEN_REQ, &ep->com.history); + return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); +} + +static void send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb, + u8 mpa_rev_to_use) +{ + int mpalen, wrlen; + struct fw_ofld_tx_data_wr *req; + struct mpa_message *mpa; + struct mpa_v2_conn_params mpa_v2_params; + + PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen); + + BUG_ON(skb_cloned(skb)); + + mpalen = sizeof(*mpa) + ep->plen; + if (mpa_rev_to_use == 2) + mpalen += sizeof(struct mpa_v2_conn_params); + wrlen = roundup(mpalen + sizeof *req, 16); + skb = get_skb(skb, wrlen, GFP_KERNEL); + if (!skb) { + connect_reply_upcall(ep, -ENOMEM); + return; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + + req = (struct fw_ofld_tx_data_wr *)skb_put(skb, wrlen); + memset(req, 0, wrlen); + req->op_to_immdlen = cpu_to_be32( + FW_WR_OP_V(FW_OFLD_TX_DATA_WR) | + FW_WR_COMPL_F | + FW_WR_IMMDLEN_V(mpalen)); + req->flowid_len16 = cpu_to_be32( + FW_WR_FLOWID_V(ep->hwtid) | + FW_WR_LEN16_V(wrlen >> 4)); + req->plen = cpu_to_be32(mpalen); + req->tunnel_to_proxy = cpu_to_be32( + FW_OFLD_TX_DATA_WR_FLUSH_F | + FW_OFLD_TX_DATA_WR_SHOVE_F); + + mpa = (struct mpa_message *)(req + 1); + memcpy(mpa->key, MPA_KEY_REQ, sizeof(mpa->key)); + mpa->flags = (crc_enabled ? MPA_CRC : 0) | + (markers_enabled ? MPA_MARKERS : 0) | + (mpa_rev_to_use == 2 ? MPA_ENHANCED_RDMA_CONN : 0); + mpa->private_data_size = htons(ep->plen); + mpa->revision = mpa_rev_to_use; + if (mpa_rev_to_use == 1) { + ep->tried_with_mpa_v1 = 1; + ep->retry_with_mpa_v1 = 0; + } + + if (mpa_rev_to_use == 2) { + mpa->private_data_size = htons(ntohs(mpa->private_data_size) + + sizeof (struct mpa_v2_conn_params)); + PDBG("%s initiator ird %u ord %u\n", __func__, ep->ird, + ep->ord); + mpa_v2_params.ird = htons((u16)ep->ird); + mpa_v2_params.ord = htons((u16)ep->ord); + + if (peer2peer) { + mpa_v2_params.ird |= htons(MPA_V2_PEER2PEER_MODEL); + if (p2p_type == FW_RI_INIT_P2PTYPE_RDMA_WRITE) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_WRITE_RTR); + else if (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_READ_RTR); + } + memcpy(mpa->private_data, &mpa_v2_params, + sizeof(struct mpa_v2_conn_params)); + + if (ep->plen) + memcpy(mpa->private_data + + sizeof(struct mpa_v2_conn_params), + ep->mpa_pkt + sizeof(*mpa), ep->plen); + } else + if (ep->plen) + memcpy(mpa->private_data, + ep->mpa_pkt + sizeof(*mpa), ep->plen); + + /* + * Reference the mpa skb. This ensures the data area + * will remain in memory until the hw acks the tx. + * Function fw4_ack() will deref it. + */ + skb_get(skb); + t4_set_arp_err_handler(skb, NULL, arp_failure_discard); + BUG_ON(ep->mpa_skb); + ep->mpa_skb = skb; + c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); + start_ep_timer(ep); + __state_set(&ep->com, MPA_REQ_SENT); + ep->mpa_attr.initiator = 1; + ep->snd_seq += mpalen; + return; +} + +static int send_mpa_reject(struct c4iw_ep *ep, const void *pdata, u8 plen) +{ + int mpalen, wrlen; + struct fw_ofld_tx_data_wr *req; + struct mpa_message *mpa; + struct sk_buff *skb; + struct mpa_v2_conn_params mpa_v2_params; + + PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen); + + mpalen = sizeof(*mpa) + plen; + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) + mpalen += sizeof(struct mpa_v2_conn_params); + wrlen = roundup(mpalen + sizeof *req, 16); + + skb = get_skb(NULL, wrlen, GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __func__); + return -ENOMEM; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + + req = (struct fw_ofld_tx_data_wr *)skb_put(skb, wrlen); + memset(req, 0, wrlen); + req->op_to_immdlen = cpu_to_be32( + FW_WR_OP_V(FW_OFLD_TX_DATA_WR) | + FW_WR_COMPL_F | + FW_WR_IMMDLEN_V(mpalen)); + req->flowid_len16 = cpu_to_be32( + FW_WR_FLOWID_V(ep->hwtid) | + FW_WR_LEN16_V(wrlen >> 4)); + req->plen = cpu_to_be32(mpalen); + req->tunnel_to_proxy = cpu_to_be32( + FW_OFLD_TX_DATA_WR_FLUSH_F | + FW_OFLD_TX_DATA_WR_SHOVE_F); + + mpa = (struct mpa_message *)(req + 1); + memset(mpa, 0, sizeof(*mpa)); + memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key)); + mpa->flags = MPA_REJECT; + mpa->revision = ep->mpa_attr.version; + mpa->private_data_size = htons(plen); + + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) { + mpa->flags |= MPA_ENHANCED_RDMA_CONN; + mpa->private_data_size = htons(ntohs(mpa->private_data_size) + + sizeof (struct mpa_v2_conn_params)); + mpa_v2_params.ird = htons(((u16)ep->ird) | + (peer2peer ? MPA_V2_PEER2PEER_MODEL : + 0)); + mpa_v2_params.ord = htons(((u16)ep->ord) | (peer2peer ? + (p2p_type == + FW_RI_INIT_P2PTYPE_RDMA_WRITE ? + MPA_V2_RDMA_WRITE_RTR : p2p_type == + FW_RI_INIT_P2PTYPE_READ_REQ ? + MPA_V2_RDMA_READ_RTR : 0) : 0)); + memcpy(mpa->private_data, &mpa_v2_params, + sizeof(struct mpa_v2_conn_params)); + + if (ep->plen) + memcpy(mpa->private_data + + sizeof(struct mpa_v2_conn_params), pdata, plen); + } else + if (plen) + memcpy(mpa->private_data, pdata, plen); + + /* + * Reference the mpa skb again. This ensures the data area + * will remain in memory until the hw acks the tx. + * Function fw4_ack() will deref it. + */ + skb_get(skb); + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + t4_set_arp_err_handler(skb, NULL, arp_failure_discard); + BUG_ON(ep->mpa_skb); + ep->mpa_skb = skb; + ep->snd_seq += mpalen; + return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); +} + +static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen) +{ + int mpalen, wrlen; + struct fw_ofld_tx_data_wr *req; + struct mpa_message *mpa; + struct sk_buff *skb; + struct mpa_v2_conn_params mpa_v2_params; + + PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen); + + mpalen = sizeof(*mpa) + plen; + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) + mpalen += sizeof(struct mpa_v2_conn_params); + wrlen = roundup(mpalen + sizeof *req, 16); + + skb = get_skb(NULL, wrlen, GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __func__); + return -ENOMEM; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + + req = (struct fw_ofld_tx_data_wr *) skb_put(skb, wrlen); + memset(req, 0, wrlen); + req->op_to_immdlen = cpu_to_be32( + FW_WR_OP_V(FW_OFLD_TX_DATA_WR) | + FW_WR_COMPL_F | + FW_WR_IMMDLEN_V(mpalen)); + req->flowid_len16 = cpu_to_be32( + FW_WR_FLOWID_V(ep->hwtid) | + FW_WR_LEN16_V(wrlen >> 4)); + req->plen = cpu_to_be32(mpalen); + req->tunnel_to_proxy = cpu_to_be32( + FW_OFLD_TX_DATA_WR_FLUSH_F | + FW_OFLD_TX_DATA_WR_SHOVE_F); + + mpa = (struct mpa_message *)(req + 1); + memset(mpa, 0, sizeof(*mpa)); + memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key)); + mpa->flags = (ep->mpa_attr.crc_enabled ? MPA_CRC : 0) | + (markers_enabled ? MPA_MARKERS : 0); + mpa->revision = ep->mpa_attr.version; + mpa->private_data_size = htons(plen); + + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) { + mpa->flags |= MPA_ENHANCED_RDMA_CONN; + mpa->private_data_size = htons(ntohs(mpa->private_data_size) + + sizeof (struct mpa_v2_conn_params)); + mpa_v2_params.ird = htons((u16)ep->ird); + mpa_v2_params.ord = htons((u16)ep->ord); + if (peer2peer && (ep->mpa_attr.p2p_type != + FW_RI_INIT_P2PTYPE_DISABLED)) { + mpa_v2_params.ird |= htons(MPA_V2_PEER2PEER_MODEL); + + if (p2p_type == FW_RI_INIT_P2PTYPE_RDMA_WRITE) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_WRITE_RTR); + else if (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_READ_RTR); + } + + memcpy(mpa->private_data, &mpa_v2_params, + sizeof(struct mpa_v2_conn_params)); + + if (ep->plen) + memcpy(mpa->private_data + + sizeof(struct mpa_v2_conn_params), pdata, plen); + } else + if (plen) + memcpy(mpa->private_data, pdata, plen); + + /* + * Reference the mpa skb. This ensures the data area + * will remain in memory until the hw acks the tx. + * Function fw4_ack() will deref it. + */ + skb_get(skb); + t4_set_arp_err_handler(skb, NULL, arp_failure_discard); + ep->mpa_skb = skb; + __state_set(&ep->com, MPA_REP_SENT); + ep->snd_seq += mpalen; + return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); +} + +static int act_establish(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct cpl_act_establish *req = cplhdr(skb); + unsigned int tid = GET_TID(req); + unsigned int atid = TID_TID_G(ntohl(req->tos_atid)); + struct tid_info *t = dev->rdev.lldi.tids; + + ep = lookup_atid(t, atid); + + PDBG("%s ep %p tid %u snd_isn %u rcv_isn %u\n", __func__, ep, tid, + be32_to_cpu(req->snd_isn), be32_to_cpu(req->rcv_isn)); + + mutex_lock(&ep->com.mutex); + dst_confirm(ep->dst); + + /* setup the hwtid for this connection */ + ep->hwtid = tid; + cxgb4_insert_tid(t, ep, tid); + insert_handle(dev, &dev->hwtid_idr, ep, ep->hwtid); + + ep->snd_seq = be32_to_cpu(req->snd_isn); + ep->rcv_seq = be32_to_cpu(req->rcv_isn); + + set_emss(ep, ntohs(req->tcp_opt)); + + /* dealloc the atid */ + remove_handle(ep->com.dev, &ep->com.dev->atid_idr, atid); + cxgb4_free_atid(t, atid); + set_bit(ACT_ESTAB, &ep->com.history); + + /* start MPA negotiation */ + send_flowc(ep, NULL); + if (ep->retry_with_mpa_v1) + send_mpa_req(ep, skb, 1); + else + send_mpa_req(ep, skb, mpa_rev); + mutex_unlock(&ep->com.mutex); + return 0; +} + +static void close_complete_upcall(struct c4iw_ep *ep, int status) +{ + struct iw_cm_event event; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CLOSE; + event.status = status; + if (ep->com.cm_id) { + PDBG("close complete delivered ep %p cm_id %p tid %u\n", + ep, ep->com.cm_id, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + ep->com.cm_id->rem_ref(ep->com.cm_id); + ep->com.cm_id = NULL; + set_bit(CLOSE_UPCALL, &ep->com.history); + } +} + +static int abort_connection(struct c4iw_ep *ep, struct sk_buff *skb, gfp_t gfp) +{ + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + __state_set(&ep->com, ABORTING); + set_bit(ABORT_CONN, &ep->com.history); + return send_abort(ep, skb, gfp); +} + +static void peer_close_upcall(struct c4iw_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_DISCONNECT; + if (ep->com.cm_id) { + PDBG("peer close delivered ep %p cm_id %p tid %u\n", + ep, ep->com.cm_id, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + set_bit(DISCONN_UPCALL, &ep->com.history); + } +} + +static void peer_abort_upcall(struct c4iw_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CLOSE; + event.status = -ECONNRESET; + if (ep->com.cm_id) { + PDBG("abort delivered ep %p cm_id %p tid %u\n", ep, + ep->com.cm_id, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + ep->com.cm_id->rem_ref(ep->com.cm_id); + ep->com.cm_id = NULL; + set_bit(ABORT_UPCALL, &ep->com.history); + } +} + +static void connect_reply_upcall(struct c4iw_ep *ep, int status) +{ + struct iw_cm_event event; + + PDBG("%s ep %p tid %u status %d\n", __func__, ep, ep->hwtid, status); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CONNECT_REPLY; + event.status = status; + memcpy(&event.local_addr, &ep->com.local_addr, + sizeof(ep->com.local_addr)); + memcpy(&event.remote_addr, &ep->com.remote_addr, + sizeof(ep->com.remote_addr)); + + if ((status == 0) || (status == -ECONNREFUSED)) { + if (!ep->tried_with_mpa_v1) { + /* this means MPA_v2 is used */ + event.private_data_len = ep->plen - + sizeof(struct mpa_v2_conn_params); + event.private_data = ep->mpa_pkt + + sizeof(struct mpa_message) + + sizeof(struct mpa_v2_conn_params); + } else { + /* this means MPA_v1 is used */ + event.private_data_len = ep->plen; + event.private_data = ep->mpa_pkt + + sizeof(struct mpa_message); + } + } + + PDBG("%s ep %p tid %u status %d\n", __func__, ep, + ep->hwtid, status); + set_bit(CONN_RPL_UPCALL, &ep->com.history); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + + if (status < 0) { + ep->com.cm_id->rem_ref(ep->com.cm_id); + ep->com.cm_id = NULL; + } +} + +static int connect_request_upcall(struct c4iw_ep *ep) +{ + struct iw_cm_event event; + int ret; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CONNECT_REQUEST; + memcpy(&event.local_addr, &ep->com.local_addr, + sizeof(ep->com.local_addr)); + memcpy(&event.remote_addr, &ep->com.remote_addr, + sizeof(ep->com.remote_addr)); + event.provider_data = ep; + if (!ep->tried_with_mpa_v1) { + /* this means MPA_v2 is used */ + event.ord = ep->ord; + event.ird = ep->ird; + event.private_data_len = ep->plen - + sizeof(struct mpa_v2_conn_params); + event.private_data = ep->mpa_pkt + sizeof(struct mpa_message) + + sizeof(struct mpa_v2_conn_params); + } else { + /* this means MPA_v1 is used. Send max supported */ + event.ord = cur_max_read_depth(ep->com.dev); + event.ird = cur_max_read_depth(ep->com.dev); + event.private_data_len = ep->plen; + event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); + } + c4iw_get_ep(&ep->com); + ret = ep->parent_ep->com.cm_id->event_handler(ep->parent_ep->com.cm_id, + &event); + if (ret) + c4iw_put_ep(&ep->com); + set_bit(CONNREQ_UPCALL, &ep->com.history); + c4iw_put_ep(&ep->parent_ep->com); + return ret; +} + +static void established_upcall(struct c4iw_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_ESTABLISHED; + event.ird = ep->ird; + event.ord = ep->ord; + if (ep->com.cm_id) { + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + set_bit(ESTAB_UPCALL, &ep->com.history); + } +} + +static int update_rx_credits(struct c4iw_ep *ep, u32 credits) +{ + struct cpl_rx_data_ack *req; + struct sk_buff *skb; + int wrlen = roundup(sizeof *req, 16); + + PDBG("%s ep %p tid %u credits %u\n", __func__, ep, ep->hwtid, credits); + skb = get_skb(NULL, wrlen, GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "update_rx_credits - cannot alloc skb!\n"); + return 0; + } + + /* + * If we couldn't specify the entire rcv window at connection setup + * due to the limit in the number of bits in the RCV_BUFSIZ field, + * then add the overage in to the credits returned. + */ + if (ep->rcv_win > RCV_BUFSIZ_M * 1024) + credits += ep->rcv_win - RCV_BUFSIZ_M * 1024; + + req = (struct cpl_rx_data_ack *) skb_put(skb, wrlen); + memset(req, 0, wrlen); + INIT_TP_WR(req, ep->hwtid); + OPCODE_TID(req) = cpu_to_be32(MK_OPCODE_TID(CPL_RX_DATA_ACK, + ep->hwtid)); + req->credit_dack = cpu_to_be32(credits | RX_FORCE_ACK_F | + RX_DACK_CHANGE_F | + RX_DACK_MODE_V(dack_mode)); + set_wr_txq(skb, CPL_PRIORITY_ACK, ep->ctrlq_idx); + c4iw_ofld_send(&ep->com.dev->rdev, skb); + return credits; +} + +#define RELAXED_IRD_NEGOTIATION 1 + +static int process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb) +{ + struct mpa_message *mpa; + struct mpa_v2_conn_params *mpa_v2_params; + u16 plen; + u16 resp_ird, resp_ord; + u8 rtr_mismatch = 0, insuff_ird = 0; + struct c4iw_qp_attributes attrs; + enum c4iw_qp_attr_mask mask; + int err; + int disconnect = 0; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + + /* + * Stop mpa timer. If it expired, then + * we ignore the MPA reply. process_timeout() + * will abort the connection. + */ + if (stop_ep_timer(ep)) + return 0; + + /* + * If we get more than the supported amount of private data + * then we must fail this connection. + */ + if (ep->mpa_pkt_len + skb->len > sizeof(ep->mpa_pkt)) { + err = -EINVAL; + goto err; + } + + /* + * copy the new data into our accumulation buffer. + */ + skb_copy_from_linear_data(skb, &(ep->mpa_pkt[ep->mpa_pkt_len]), + skb->len); + ep->mpa_pkt_len += skb->len; + + /* + * if we don't even have the mpa message, then bail. + */ + if (ep->mpa_pkt_len < sizeof(*mpa)) + return 0; + mpa = (struct mpa_message *) ep->mpa_pkt; + + /* Validate MPA header. */ + if (mpa->revision > mpa_rev) { + printk(KERN_ERR MOD "%s MPA version mismatch. Local = %d," + " Received = %d\n", __func__, mpa_rev, mpa->revision); + err = -EPROTO; + goto err; + } + if (memcmp(mpa->key, MPA_KEY_REP, sizeof(mpa->key))) { + err = -EPROTO; + goto err; + } + + plen = ntohs(mpa->private_data_size); + + /* + * Fail if there's too much private data. + */ + if (plen > MPA_MAX_PRIVATE_DATA) { + err = -EPROTO; + goto err; + } + + /* + * If plen does not account for pkt size + */ + if (ep->mpa_pkt_len > (sizeof(*mpa) + plen)) { + err = -EPROTO; + goto err; + } + + ep->plen = (u8) plen; + + /* + * If we don't have all the pdata yet, then bail. + * We'll continue process when more data arrives. + */ + if (ep->mpa_pkt_len < (sizeof(*mpa) + plen)) + return 0; + + if (mpa->flags & MPA_REJECT) { + err = -ECONNREFUSED; + goto err; + } + + /* + * If we get here we have accumulated the entire mpa + * start reply message including private data. And + * the MPA header is valid. + */ + __state_set(&ep->com, FPDU_MODE); + ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0; + ep->mpa_attr.recv_marker_enabled = markers_enabled; + ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0; + ep->mpa_attr.version = mpa->revision; + ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED; + + if (mpa->revision == 2) { + ep->mpa_attr.enhanced_rdma_conn = + mpa->flags & MPA_ENHANCED_RDMA_CONN ? 1 : 0; + if (ep->mpa_attr.enhanced_rdma_conn) { + mpa_v2_params = (struct mpa_v2_conn_params *) + (ep->mpa_pkt + sizeof(*mpa)); + resp_ird = ntohs(mpa_v2_params->ird) & + MPA_V2_IRD_ORD_MASK; + resp_ord = ntohs(mpa_v2_params->ord) & + MPA_V2_IRD_ORD_MASK; + PDBG("%s responder ird %u ord %u ep ird %u ord %u\n", + __func__, resp_ird, resp_ord, ep->ird, ep->ord); + + /* + * This is a double-check. Ideally, below checks are + * not required since ird/ord stuff has been taken + * care of in c4iw_accept_cr + */ + if (ep->ird < resp_ord) { + if (RELAXED_IRD_NEGOTIATION && resp_ord <= + ep->com.dev->rdev.lldi.max_ordird_qp) + ep->ird = resp_ord; + else + insuff_ird = 1; + } else if (ep->ird > resp_ord) { + ep->ird = resp_ord; + } + if (ep->ord > resp_ird) { + if (RELAXED_IRD_NEGOTIATION) + ep->ord = resp_ird; + else + insuff_ird = 1; + } + if (insuff_ird) { + err = -ENOMEM; + ep->ird = resp_ord; + ep->ord = resp_ird; + } + + if (ntohs(mpa_v2_params->ird) & + MPA_V2_PEER2PEER_MODEL) { + if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_WRITE_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_RDMA_WRITE; + else if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_READ_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_READ_REQ; + } + } + } else if (mpa->revision == 1) + if (peer2peer) + ep->mpa_attr.p2p_type = p2p_type; + + PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, " + "xmit_marker_enabled=%d, version=%d p2p_type=%d local-p2p_type = " + "%d\n", __func__, ep->mpa_attr.crc_enabled, + ep->mpa_attr.recv_marker_enabled, + ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version, + ep->mpa_attr.p2p_type, p2p_type); + + /* + * If responder's RTR does not match with that of initiator, assign + * FW_RI_INIT_P2PTYPE_DISABLED in mpa attributes so that RTR is not + * generated when moving QP to RTS state. + * A TERM message will be sent after QP has moved to RTS state + */ + if ((ep->mpa_attr.version == 2) && peer2peer && + (ep->mpa_attr.p2p_type != p2p_type)) { + ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED; + rtr_mismatch = 1; + } + + attrs.mpa_attr = ep->mpa_attr; + attrs.max_ird = ep->ird; + attrs.max_ord = ep->ord; + attrs.llp_stream_handle = ep; + attrs.next_state = C4IW_QP_STATE_RTS; + + mask = C4IW_QP_ATTR_NEXT_STATE | + C4IW_QP_ATTR_LLP_STREAM_HANDLE | C4IW_QP_ATTR_MPA_ATTR | + C4IW_QP_ATTR_MAX_IRD | C4IW_QP_ATTR_MAX_ORD; + + /* bind QP and TID with INIT_WR */ + err = c4iw_modify_qp(ep->com.qp->rhp, + ep->com.qp, mask, &attrs, 1); + if (err) + goto err; + + /* + * If responder's RTR requirement did not match with what initiator + * supports, generate TERM message + */ + if (rtr_mismatch) { + printk(KERN_ERR "%s: RTR mismatch, sending TERM\n", __func__); + attrs.layer_etype = LAYER_MPA | DDP_LLP; + attrs.ecode = MPA_NOMATCH_RTR; + attrs.next_state = C4IW_QP_STATE_TERMINATE; + attrs.send_term = 1; + err = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + err = -ENOMEM; + disconnect = 1; + goto out; + } + + /* + * Generate TERM if initiator IRD is not sufficient for responder + * provided ORD. Currently, we do the same behaviour even when + * responder provided IRD is also not sufficient as regards to + * initiator ORD. + */ + if (insuff_ird) { + printk(KERN_ERR "%s: Insufficient IRD, sending TERM\n", + __func__); + attrs.layer_etype = LAYER_MPA | DDP_LLP; + attrs.ecode = MPA_INSUFF_IRD; + attrs.next_state = C4IW_QP_STATE_TERMINATE; + attrs.send_term = 1; + err = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + err = -ENOMEM; + disconnect = 1; + goto out; + } + goto out; +err: + __state_set(&ep->com, ABORTING); + send_abort(ep, skb, GFP_KERNEL); +out: + connect_reply_upcall(ep, err); + return disconnect; +} + +static void process_mpa_request(struct c4iw_ep *ep, struct sk_buff *skb) +{ + struct mpa_message *mpa; + struct mpa_v2_conn_params *mpa_v2_params; + u16 plen; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + + /* + * If we get more than the supported amount of private data + * then we must fail this connection. + */ + if (ep->mpa_pkt_len + skb->len > sizeof(ep->mpa_pkt)) { + (void)stop_ep_timer(ep); + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + PDBG("%s enter (%s line %u)\n", __func__, __FILE__, __LINE__); + + /* + * Copy the new data into our accumulation buffer. + */ + skb_copy_from_linear_data(skb, &(ep->mpa_pkt[ep->mpa_pkt_len]), + skb->len); + ep->mpa_pkt_len += skb->len; + + /* + * If we don't even have the mpa message, then bail. + * We'll continue process when more data arrives. + */ + if (ep->mpa_pkt_len < sizeof(*mpa)) + return; + + PDBG("%s enter (%s line %u)\n", __func__, __FILE__, __LINE__); + mpa = (struct mpa_message *) ep->mpa_pkt; + + /* + * Validate MPA Header. + */ + if (mpa->revision > mpa_rev) { + printk(KERN_ERR MOD "%s MPA version mismatch. Local = %d," + " Received = %d\n", __func__, mpa_rev, mpa->revision); + (void)stop_ep_timer(ep); + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + if (memcmp(mpa->key, MPA_KEY_REQ, sizeof(mpa->key))) { + (void)stop_ep_timer(ep); + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + plen = ntohs(mpa->private_data_size); + + /* + * Fail if there's too much private data. + */ + if (plen > MPA_MAX_PRIVATE_DATA) { + (void)stop_ep_timer(ep); + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + /* + * If plen does not account for pkt size + */ + if (ep->mpa_pkt_len > (sizeof(*mpa) + plen)) { + (void)stop_ep_timer(ep); + abort_connection(ep, skb, GFP_KERNEL); + return; + } + ep->plen = (u8) plen; + + /* + * If we don't have all the pdata yet, then bail. + */ + if (ep->mpa_pkt_len < (sizeof(*mpa) + plen)) + return; + + /* + * If we get here we have accumulated the entire mpa + * start reply message including private data. + */ + ep->mpa_attr.initiator = 0; + ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0; + ep->mpa_attr.recv_marker_enabled = markers_enabled; + ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0; + ep->mpa_attr.version = mpa->revision; + if (mpa->revision == 1) + ep->tried_with_mpa_v1 = 1; + ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED; + + if (mpa->revision == 2) { + ep->mpa_attr.enhanced_rdma_conn = + mpa->flags & MPA_ENHANCED_RDMA_CONN ? 1 : 0; + if (ep->mpa_attr.enhanced_rdma_conn) { + mpa_v2_params = (struct mpa_v2_conn_params *) + (ep->mpa_pkt + sizeof(*mpa)); + ep->ird = ntohs(mpa_v2_params->ird) & + MPA_V2_IRD_ORD_MASK; + ep->ord = ntohs(mpa_v2_params->ord) & + MPA_V2_IRD_ORD_MASK; + PDBG("%s initiator ird %u ord %u\n", __func__, ep->ird, + ep->ord); + if (ntohs(mpa_v2_params->ird) & MPA_V2_PEER2PEER_MODEL) + if (peer2peer) { + if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_WRITE_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_RDMA_WRITE; + else if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_READ_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_READ_REQ; + } + } + } else if (mpa->revision == 1) + if (peer2peer) + ep->mpa_attr.p2p_type = p2p_type; + + PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, " + "xmit_marker_enabled=%d, version=%d p2p_type=%d\n", __func__, + ep->mpa_attr.crc_enabled, ep->mpa_attr.recv_marker_enabled, + ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version, + ep->mpa_attr.p2p_type); + + /* + * If the endpoint timer already expired, then we ignore + * the start request. process_timeout() will abort + * the connection. + */ + if (!stop_ep_timer(ep)) { + __state_set(&ep->com, MPA_REQ_RCVD); + + /* drive upcall */ + mutex_lock_nested(&ep->parent_ep->com.mutex, + SINGLE_DEPTH_NESTING); + if (ep->parent_ep->com.state != DEAD) { + if (connect_request_upcall(ep)) + abort_connection(ep, skb, GFP_KERNEL); + } else { + abort_connection(ep, skb, GFP_KERNEL); + } + mutex_unlock(&ep->parent_ep->com.mutex); + } + return; +} + +static int rx_data(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct cpl_rx_data *hdr = cplhdr(skb); + unsigned int dlen = ntohs(hdr->len); + unsigned int tid = GET_TID(hdr); + struct tid_info *t = dev->rdev.lldi.tids; + __u8 status = hdr->status; + int disconnect = 0; + + ep = lookup_tid(t, tid); + if (!ep) + return 0; + PDBG("%s ep %p tid %u dlen %u\n", __func__, ep, ep->hwtid, dlen); + skb_pull(skb, sizeof(*hdr)); + skb_trim(skb, dlen); + mutex_lock(&ep->com.mutex); + + /* update RX credits */ + update_rx_credits(ep, dlen); + + switch (ep->com.state) { + case MPA_REQ_SENT: + ep->rcv_seq += dlen; + disconnect = process_mpa_reply(ep, skb); + break; + case MPA_REQ_WAIT: + ep->rcv_seq += dlen; + process_mpa_request(ep, skb); + break; + case FPDU_MODE: { + struct c4iw_qp_attributes attrs; + BUG_ON(!ep->com.qp); + if (status) + pr_err("%s Unexpected streaming data." \ + " qpid %u ep %p state %d tid %u status %d\n", + __func__, ep->com.qp->wq.sq.qid, ep, + ep->com.state, ep->hwtid, status); + attrs.next_state = C4IW_QP_STATE_TERMINATE; + c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + disconnect = 1; + break; + } + default: + break; + } + mutex_unlock(&ep->com.mutex); + if (disconnect) + c4iw_ep_disconnect(ep, 0, GFP_KERNEL); + return 0; +} + +static int abort_rpl(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct cpl_abort_rpl_rss *rpl = cplhdr(skb); + int release = 0; + unsigned int tid = GET_TID(rpl); + struct tid_info *t = dev->rdev.lldi.tids; + + ep = lookup_tid(t, tid); + if (!ep) { + printk(KERN_WARNING MOD "Abort rpl to freed endpoint\n"); + return 0; + } + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + mutex_lock(&ep->com.mutex); + switch (ep->com.state) { + case ABORTING: + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + __state_set(&ep->com, DEAD); + release = 1; + break; + default: + printk(KERN_ERR "%s ep %p state %d\n", + __func__, ep, ep->com.state); + break; + } + mutex_unlock(&ep->com.mutex); + + if (release) + release_ep_resources(ep); + return 0; +} + +static void send_fw_act_open_req(struct c4iw_ep *ep, unsigned int atid) +{ + struct sk_buff *skb; + struct fw_ofld_connection_wr *req; + unsigned int mtu_idx; + int wscale; + struct sockaddr_in *sin; + int win; + + skb = get_skb(NULL, sizeof(*req), GFP_KERNEL); + req = (struct fw_ofld_connection_wr *)__skb_put(skb, sizeof(*req)); + memset(req, 0, sizeof(*req)); + req->op_compl = htonl(WR_OP_V(FW_OFLD_CONNECTION_WR)); + req->len16_pkd = htonl(FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*req), 16))); + req->le.filter = cpu_to_be32(cxgb4_select_ntuple( + ep->com.dev->rdev.lldi.ports[0], + ep->l2t)); + sin = (struct sockaddr_in *)&ep->com.mapped_local_addr; + req->le.lport = sin->sin_port; + req->le.u.ipv4.lip = sin->sin_addr.s_addr; + sin = (struct sockaddr_in *)&ep->com.mapped_remote_addr; + req->le.pport = sin->sin_port; + req->le.u.ipv4.pip = sin->sin_addr.s_addr; + req->tcb.t_state_to_astid = + htonl(FW_OFLD_CONNECTION_WR_T_STATE_V(TCP_SYN_SENT) | + FW_OFLD_CONNECTION_WR_ASTID_V(atid)); + req->tcb.cplrxdataack_cplpassacceptrpl = + htons(FW_OFLD_CONNECTION_WR_CPLRXDATAACK_F); + req->tcb.tx_max = (__force __be32) jiffies; + req->tcb.rcv_adv = htons(1); + best_mtu(ep->com.dev->rdev.lldi.mtus, ep->mtu, &mtu_idx, + enable_tcp_timestamps, + (AF_INET == ep->com.remote_addr.ss_family) ? 0 : 1); + wscale = compute_wscale(rcv_win); + + /* + * Specify the largest window that will fit in opt0. The + * remainder will be specified in the rx_data_ack. + */ + win = ep->rcv_win >> 10; + if (win > RCV_BUFSIZ_M) + win = RCV_BUFSIZ_M; + + req->tcb.opt0 = (__force __be64) (TCAM_BYPASS_F | + (nocong ? NO_CONG_F : 0) | + KEEP_ALIVE_F | + DELACK_F | + WND_SCALE_V(wscale) | + MSS_IDX_V(mtu_idx) | + L2T_IDX_V(ep->l2t->idx) | + TX_CHAN_V(ep->tx_chan) | + SMAC_SEL_V(ep->smac_idx) | + DSCP_V(ep->tos) | + ULP_MODE_V(ULP_MODE_TCPDDP) | + RCV_BUFSIZ_V(win)); + req->tcb.opt2 = (__force __be32) (PACE_V(1) | + TX_QUEUE_V(ep->com.dev->rdev.lldi.tx_modq[ep->tx_chan]) | + RX_CHANNEL_V(0) | + CCTRL_ECN_V(enable_ecn) | + RSS_QUEUE_VALID_F | RSS_QUEUE_V(ep->rss_qid)); + if (enable_tcp_timestamps) + req->tcb.opt2 |= (__force __be32)TSTAMPS_EN_F; + if (enable_tcp_sack) + req->tcb.opt2 |= (__force __be32)SACK_EN_F; + if (wscale && enable_tcp_window_scaling) + req->tcb.opt2 |= (__force __be32)WND_SCALE_EN_F; + req->tcb.opt0 = cpu_to_be64((__force u64)req->tcb.opt0); + req->tcb.opt2 = cpu_to_be32((__force u32)req->tcb.opt2); + set_wr_txq(skb, CPL_PRIORITY_CONTROL, ep->ctrlq_idx); + set_bit(ACT_OFLD_CONN, &ep->com.history); + c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); +} + +/* + * Return whether a failed active open has allocated a TID + */ +static inline int act_open_has_tid(int status) +{ + return status != CPL_ERR_TCAM_FULL && status != CPL_ERR_CONN_EXIST && + status != CPL_ERR_ARP_MISS; +} + +/* Returns whether a CPL status conveys negative advice. + */ +static int is_neg_adv(unsigned int status) +{ + return status == CPL_ERR_RTX_NEG_ADVICE || + status == CPL_ERR_PERSIST_NEG_ADVICE || + status == CPL_ERR_KEEPALV_NEG_ADVICE; +} + +static char *neg_adv_str(unsigned int status) +{ + switch (status) { + case CPL_ERR_RTX_NEG_ADVICE: + return "Retransmit timeout"; + case CPL_ERR_PERSIST_NEG_ADVICE: + return "Persist timeout"; + case CPL_ERR_KEEPALV_NEG_ADVICE: + return "Keepalive timeout"; + default: + return "Unknown"; + } +} + +static void set_tcp_window(struct c4iw_ep *ep, struct port_info *pi) +{ + ep->snd_win = snd_win; + ep->rcv_win = rcv_win; + PDBG("%s snd_win %d rcv_win %d\n", __func__, ep->snd_win, ep->rcv_win); +} + +#define ACT_OPEN_RETRY_COUNT 2 + +static int import_ep(struct c4iw_ep *ep, int iptype, __u8 *peer_ip, + struct dst_entry *dst, struct c4iw_dev *cdev, + bool clear_mpa_v1) +{ + struct neighbour *n; + int err, step; + struct net_device *pdev; + + n = dst_neigh_lookup(dst, peer_ip); + if (!n) + return -ENODEV; + + rcu_read_lock(); + err = -ENOMEM; + if (n->dev->flags & IFF_LOOPBACK) { + if (iptype == 4) + pdev = ip_dev_find(&init_net, *(__be32 *)peer_ip); + else if (IS_ENABLED(CONFIG_IPV6)) + for_each_netdev(&init_net, pdev) { + if (ipv6_chk_addr(&init_net, + (struct in6_addr *)peer_ip, + pdev, 1)) + break; + } + else + pdev = NULL; + + if (!pdev) { + err = -ENODEV; + goto out; + } + ep->l2t = cxgb4_l2t_get(cdev->rdev.lldi.l2t, + n, pdev, 0); + if (!ep->l2t) + goto out; + ep->mtu = pdev->mtu; + ep->tx_chan = cxgb4_port_chan(pdev); + ep->smac_idx = (cxgb4_port_viid(pdev) & 0x7F) << 1; + step = cdev->rdev.lldi.ntxq / + cdev->rdev.lldi.nchan; + ep->txq_idx = cxgb4_port_idx(pdev) * step; + step = cdev->rdev.lldi.nrxq / + cdev->rdev.lldi.nchan; + ep->ctrlq_idx = cxgb4_port_idx(pdev); + ep->rss_qid = cdev->rdev.lldi.rxq_ids[ + cxgb4_port_idx(pdev) * step]; + set_tcp_window(ep, (struct port_info *)netdev_priv(pdev)); + dev_put(pdev); + } else { + pdev = get_real_dev(n->dev); + ep->l2t = cxgb4_l2t_get(cdev->rdev.lldi.l2t, + n, pdev, 0); + if (!ep->l2t) + goto out; + ep->mtu = dst_mtu(dst); + ep->tx_chan = cxgb4_port_chan(pdev); + ep->smac_idx = (cxgb4_port_viid(pdev) & 0x7F) << 1; + step = cdev->rdev.lldi.ntxq / + cdev->rdev.lldi.nchan; + ep->txq_idx = cxgb4_port_idx(pdev) * step; + ep->ctrlq_idx = cxgb4_port_idx(pdev); + step = cdev->rdev.lldi.nrxq / + cdev->rdev.lldi.nchan; + ep->rss_qid = cdev->rdev.lldi.rxq_ids[ + cxgb4_port_idx(pdev) * step]; + set_tcp_window(ep, (struct port_info *)netdev_priv(pdev)); + + if (clear_mpa_v1) { + ep->retry_with_mpa_v1 = 0; + ep->tried_with_mpa_v1 = 0; + } + } + err = 0; +out: + rcu_read_unlock(); + + neigh_release(n); + + return err; +} + +static int c4iw_reconnect(struct c4iw_ep *ep) +{ + int err = 0; + struct sockaddr_in *laddr = (struct sockaddr_in *) + &ep->com.cm_id->local_addr; + struct sockaddr_in *raddr = (struct sockaddr_in *) + &ep->com.cm_id->remote_addr; + struct sockaddr_in6 *laddr6 = (struct sockaddr_in6 *) + &ep->com.cm_id->local_addr; + struct sockaddr_in6 *raddr6 = (struct sockaddr_in6 *) + &ep->com.cm_id->remote_addr; + int iptype; + __u8 *ra; + + PDBG("%s qp %p cm_id %p\n", __func__, ep->com.qp, ep->com.cm_id); + init_timer(&ep->timer); + + /* + * Allocate an active TID to initiate a TCP connection. + */ + ep->atid = cxgb4_alloc_atid(ep->com.dev->rdev.lldi.tids, ep); + if (ep->atid == -1) { + pr_err("%s - cannot alloc atid.\n", __func__); + err = -ENOMEM; + goto fail2; + } + insert_handle(ep->com.dev, &ep->com.dev->atid_idr, ep, ep->atid); + + /* find a route */ + if (ep->com.cm_id->local_addr.ss_family == AF_INET) { + ep->dst = find_route(ep->com.dev, laddr->sin_addr.s_addr, + raddr->sin_addr.s_addr, laddr->sin_port, + raddr->sin_port, 0); + iptype = 4; + ra = (__u8 *)&raddr->sin_addr; + } else { + ep->dst = find_route6(ep->com.dev, laddr6->sin6_addr.s6_addr, + raddr6->sin6_addr.s6_addr, + laddr6->sin6_port, raddr6->sin6_port, 0, + raddr6->sin6_scope_id); + iptype = 6; + ra = (__u8 *)&raddr6->sin6_addr; + } + if (!ep->dst) { + pr_err("%s - cannot find route.\n", __func__); + err = -EHOSTUNREACH; + goto fail3; + } + err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, false); + if (err) { + pr_err("%s - cannot alloc l2e.\n", __func__); + goto fail4; + } + + PDBG("%s txq_idx %u tx_chan %u smac_idx %u rss_qid %u l2t_idx %u\n", + __func__, ep->txq_idx, ep->tx_chan, ep->smac_idx, ep->rss_qid, + ep->l2t->idx); + + state_set(&ep->com, CONNECTING); + ep->tos = 0; + + /* send connect request to rnic */ + err = send_connect(ep); + if (!err) + goto out; + + cxgb4_l2t_release(ep->l2t); +fail4: + dst_release(ep->dst); +fail3: + remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid); + cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid); +fail2: + /* + * remember to send notification to upper layer. + * We are in here so the upper layer is not aware that this is + * re-connect attempt and so, upper layer is still waiting for + * response of 1st connect request. + */ + connect_reply_upcall(ep, -ECONNRESET); + c4iw_put_ep(&ep->com); +out: + return err; +} + +static int act_open_rpl(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct cpl_act_open_rpl *rpl = cplhdr(skb); + unsigned int atid = TID_TID_G(AOPEN_ATID_G( + ntohl(rpl->atid_status))); + struct tid_info *t = dev->rdev.lldi.tids; + int status = AOPEN_STATUS_G(ntohl(rpl->atid_status)); + struct sockaddr_in *la; + struct sockaddr_in *ra; + struct sockaddr_in6 *la6; + struct sockaddr_in6 *ra6; + + ep = lookup_atid(t, atid); + la = (struct sockaddr_in *)&ep->com.mapped_local_addr; + ra = (struct sockaddr_in *)&ep->com.mapped_remote_addr; + la6 = (struct sockaddr_in6 *)&ep->com.mapped_local_addr; + ra6 = (struct sockaddr_in6 *)&ep->com.mapped_remote_addr; + + PDBG("%s ep %p atid %u status %u errno %d\n", __func__, ep, atid, + status, status2errno(status)); + + if (is_neg_adv(status)) { + PDBG("%s Connection problems for atid %u status %u (%s)\n", + __func__, atid, status, neg_adv_str(status)); + ep->stats.connect_neg_adv++; + mutex_lock(&dev->rdev.stats.lock); + dev->rdev.stats.neg_adv++; + mutex_unlock(&dev->rdev.stats.lock); + return 0; + } + + set_bit(ACT_OPEN_RPL, &ep->com.history); + + /* + * Log interesting failures. + */ + switch (status) { + case CPL_ERR_CONN_RESET: + case CPL_ERR_CONN_TIMEDOUT: + break; + case CPL_ERR_TCAM_FULL: + mutex_lock(&dev->rdev.stats.lock); + dev->rdev.stats.tcam_full++; + mutex_unlock(&dev->rdev.stats.lock); + if (ep->com.local_addr.ss_family == AF_INET && + dev->rdev.lldi.enable_fw_ofld_conn) { + send_fw_act_open_req(ep, + TID_TID_G(AOPEN_ATID_G( + ntohl(rpl->atid_status)))); + return 0; + } + break; + case CPL_ERR_CONN_EXIST: + if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) { + set_bit(ACT_RETRY_INUSE, &ep->com.history); + remove_handle(ep->com.dev, &ep->com.dev->atid_idr, + atid); + cxgb4_free_atid(t, atid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + c4iw_reconnect(ep); + return 0; + } + break; + default: + if (ep->com.local_addr.ss_family == AF_INET) { + pr_info("Active open failure - atid %u status %u errno %d %pI4:%u->%pI4:%u\n", + atid, status, status2errno(status), + &la->sin_addr.s_addr, ntohs(la->sin_port), + &ra->sin_addr.s_addr, ntohs(ra->sin_port)); + } else { + pr_info("Active open failure - atid %u status %u errno %d %pI6:%u->%pI6:%u\n", + atid, status, status2errno(status), + la6->sin6_addr.s6_addr, ntohs(la6->sin6_port), + ra6->sin6_addr.s6_addr, ntohs(ra6->sin6_port)); + } + break; + } + + connect_reply_upcall(ep, status2errno(status)); + state_set(&ep->com, DEAD); + + if (status && act_open_has_tid(status)) + cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, GET_TID(rpl)); + + remove_handle(ep->com.dev, &ep->com.dev->atid_idr, atid); + cxgb4_free_atid(t, atid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + c4iw_put_ep(&ep->com); + + return 0; +} + +static int pass_open_rpl(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_pass_open_rpl *rpl = cplhdr(skb); + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int stid = GET_TID(rpl); + struct c4iw_listen_ep *ep = lookup_stid(t, stid); + + if (!ep) { + PDBG("%s stid %d lookup failure!\n", __func__, stid); + goto out; + } + PDBG("%s ep %p status %d error %d\n", __func__, ep, + rpl->status, status2errno(rpl->status)); + c4iw_wake_up(&ep->com.wr_wait, status2errno(rpl->status)); + +out: + return 0; +} + +static int close_listsrv_rpl(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_close_listsvr_rpl *rpl = cplhdr(skb); + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int stid = GET_TID(rpl); + struct c4iw_listen_ep *ep = lookup_stid(t, stid); + + PDBG("%s ep %p\n", __func__, ep); + c4iw_wake_up(&ep->com.wr_wait, status2errno(rpl->status)); + return 0; +} + +static void accept_cr(struct c4iw_ep *ep, struct sk_buff *skb, + struct cpl_pass_accept_req *req) +{ + struct cpl_pass_accept_rpl *rpl; + unsigned int mtu_idx; + u64 opt0; + u32 opt2; + int wscale; + struct cpl_t5_pass_accept_rpl *rpl5 = NULL; + int win; + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + BUG_ON(skb_cloned(skb)); + + skb_get(skb); + rpl = cplhdr(skb); + if (is_t5(ep->com.dev->rdev.lldi.adapter_type)) { + skb_trim(skb, roundup(sizeof(*rpl5), 16)); + rpl5 = (void *)rpl; + INIT_TP_WR(rpl5, ep->hwtid); + } else { + skb_trim(skb, sizeof(*rpl)); + INIT_TP_WR(rpl, ep->hwtid); + } + OPCODE_TID(rpl) = cpu_to_be32(MK_OPCODE_TID(CPL_PASS_ACCEPT_RPL, + ep->hwtid)); + + best_mtu(ep->com.dev->rdev.lldi.mtus, ep->mtu, &mtu_idx, + enable_tcp_timestamps && req->tcpopt.tstamp, + (AF_INET == ep->com.remote_addr.ss_family) ? 0 : 1); + wscale = compute_wscale(rcv_win); + + /* + * Specify the largest window that will fit in opt0. The + * remainder will be specified in the rx_data_ack. + */ + win = ep->rcv_win >> 10; + if (win > RCV_BUFSIZ_M) + win = RCV_BUFSIZ_M; + opt0 = (nocong ? NO_CONG_F : 0) | + KEEP_ALIVE_F | + DELACK_F | + WND_SCALE_V(wscale) | + MSS_IDX_V(mtu_idx) | + L2T_IDX_V(ep->l2t->idx) | + TX_CHAN_V(ep->tx_chan) | + SMAC_SEL_V(ep->smac_idx) | + DSCP_V(ep->tos >> 2) | + ULP_MODE_V(ULP_MODE_TCPDDP) | + RCV_BUFSIZ_V(win); + opt2 = RX_CHANNEL_V(0) | + RSS_QUEUE_VALID_F | RSS_QUEUE_V(ep->rss_qid); + + if (enable_tcp_timestamps && req->tcpopt.tstamp) + opt2 |= TSTAMPS_EN_F; + if (enable_tcp_sack && req->tcpopt.sack) + opt2 |= SACK_EN_F; + if (wscale && enable_tcp_window_scaling) + opt2 |= WND_SCALE_EN_F; + if (enable_ecn) { + const struct tcphdr *tcph; + u32 hlen = ntohl(req->hdr_len); + + tcph = (const void *)(req + 1) + ETH_HDR_LEN_G(hlen) + + IP_HDR_LEN_G(hlen); + if (tcph->ece && tcph->cwr) + opt2 |= CCTRL_ECN_V(1); + } + if (is_t5(ep->com.dev->rdev.lldi.adapter_type)) { + u32 isn = (prandom_u32() & ~7UL) - 1; + opt2 |= T5_OPT_2_VALID_F; + opt2 |= CONG_CNTRL_V(CONG_ALG_TAHOE); + opt2 |= T5_ISS_F; + rpl5 = (void *)rpl; + memset(&rpl5->iss, 0, roundup(sizeof(*rpl5)-sizeof(*rpl), 16)); + if (peer2peer) + isn += 4; + rpl5->iss = cpu_to_be32(isn); + PDBG("%s iss %u\n", __func__, be32_to_cpu(rpl5->iss)); + } + + rpl->opt0 = cpu_to_be64(opt0); + rpl->opt2 = cpu_to_be32(opt2); + set_wr_txq(skb, CPL_PRIORITY_SETUP, ep->ctrlq_idx); + t4_set_arp_err_handler(skb, NULL, arp_failure_discard); + c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); + + return; +} + +static void reject_cr(struct c4iw_dev *dev, u32 hwtid, struct sk_buff *skb) +{ + PDBG("%s c4iw_dev %p tid %u\n", __func__, dev, hwtid); + BUG_ON(skb_cloned(skb)); + skb_trim(skb, sizeof(struct cpl_tid_release)); + release_tid(&dev->rdev, hwtid, skb); + return; +} + +static void get_4tuple(struct cpl_pass_accept_req *req, int *iptype, + __u8 *local_ip, __u8 *peer_ip, + __be16 *local_port, __be16 *peer_port) +{ + int eth_len = ETH_HDR_LEN_G(be32_to_cpu(req->hdr_len)); + int ip_len = IP_HDR_LEN_G(be32_to_cpu(req->hdr_len)); + struct iphdr *ip = (struct iphdr *)((u8 *)(req + 1) + eth_len); + struct ipv6hdr *ip6 = (struct ipv6hdr *)((u8 *)(req + 1) + eth_len); + struct tcphdr *tcp = (struct tcphdr *) + ((u8 *)(req + 1) + eth_len + ip_len); + + if (ip->version == 4) { + PDBG("%s saddr 0x%x daddr 0x%x sport %u dport %u\n", __func__, + ntohl(ip->saddr), ntohl(ip->daddr), ntohs(tcp->source), + ntohs(tcp->dest)); + *iptype = 4; + memcpy(peer_ip, &ip->saddr, 4); + memcpy(local_ip, &ip->daddr, 4); + } else { + PDBG("%s saddr %pI6 daddr %pI6 sport %u dport %u\n", __func__, + ip6->saddr.s6_addr, ip6->daddr.s6_addr, ntohs(tcp->source), + ntohs(tcp->dest)); + *iptype = 6; + memcpy(peer_ip, ip6->saddr.s6_addr, 16); + memcpy(local_ip, ip6->daddr.s6_addr, 16); + } + *peer_port = tcp->source; + *local_port = tcp->dest; + + return; +} + +static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *child_ep = NULL, *parent_ep; + struct cpl_pass_accept_req *req = cplhdr(skb); + unsigned int stid = PASS_OPEN_TID_G(ntohl(req->tos_stid)); + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int hwtid = GET_TID(req); + struct dst_entry *dst; + __u8 local_ip[16], peer_ip[16]; + __be16 local_port, peer_port; + int err; + u16 peer_mss = ntohs(req->tcpopt.mss); + int iptype; + unsigned short hdrs; + + parent_ep = lookup_stid(t, stid); + if (!parent_ep) { + PDBG("%s connect request on invalid stid %d\n", __func__, stid); + goto reject; + } + + if (state_read(&parent_ep->com) != LISTEN) { + printk(KERN_ERR "%s - listening ep not in LISTEN\n", + __func__); + goto reject; + } + + get_4tuple(req, &iptype, local_ip, peer_ip, &local_port, &peer_port); + + /* Find output route */ + if (iptype == 4) { + PDBG("%s parent ep %p hwtid %u laddr %pI4 raddr %pI4 lport %d rport %d peer_mss %d\n" + , __func__, parent_ep, hwtid, + local_ip, peer_ip, ntohs(local_port), + ntohs(peer_port), peer_mss); + dst = find_route(dev, *(__be32 *)local_ip, *(__be32 *)peer_ip, + local_port, peer_port, + PASS_OPEN_TOS_G(ntohl(req->tos_stid))); + } else { + PDBG("%s parent ep %p hwtid %u laddr %pI6 raddr %pI6 lport %d rport %d peer_mss %d\n" + , __func__, parent_ep, hwtid, + local_ip, peer_ip, ntohs(local_port), + ntohs(peer_port), peer_mss); + dst = find_route6(dev, local_ip, peer_ip, local_port, peer_port, + PASS_OPEN_TOS_G(ntohl(req->tos_stid)), + ((struct sockaddr_in6 *) + &parent_ep->com.local_addr)->sin6_scope_id); + } + if (!dst) { + printk(KERN_ERR MOD "%s - failed to find dst entry!\n", + __func__); + goto reject; + } + + child_ep = alloc_ep(sizeof(*child_ep), GFP_KERNEL); + if (!child_ep) { + printk(KERN_ERR MOD "%s - failed to allocate ep entry!\n", + __func__); + dst_release(dst); + goto reject; + } + + err = import_ep(child_ep, iptype, peer_ip, dst, dev, false); + if (err) { + printk(KERN_ERR MOD "%s - failed to allocate l2t entry!\n", + __func__); + dst_release(dst); + kfree(child_ep); + goto reject; + } + + hdrs = sizeof(struct iphdr) + sizeof(struct tcphdr) + + ((enable_tcp_timestamps && req->tcpopt.tstamp) ? 12 : 0); + if (peer_mss && child_ep->mtu > (peer_mss + hdrs)) + child_ep->mtu = peer_mss + hdrs; + + state_set(&child_ep->com, CONNECTING); + child_ep->com.dev = dev; + child_ep->com.cm_id = NULL; + + /* + * The mapped_local and mapped_remote addresses get setup with + * the actual 4-tuple. The local address will be based on the + * actual local address of the connection, but on the port number + * of the parent listening endpoint. The remote address is + * setup based on a query to the IWPM since we don't know what it + * originally was before mapping. If no mapping was done, then + * mapped_remote == remote, and mapped_local == local. + */ + if (iptype == 4) { + struct sockaddr_in *sin = (struct sockaddr_in *) + &child_ep->com.mapped_local_addr; + + sin->sin_family = PF_INET; + sin->sin_port = local_port; + sin->sin_addr.s_addr = *(__be32 *)local_ip; + + sin = (struct sockaddr_in *)&child_ep->com.local_addr; + sin->sin_family = PF_INET; + sin->sin_port = ((struct sockaddr_in *) + &parent_ep->com.local_addr)->sin_port; + sin->sin_addr.s_addr = *(__be32 *)local_ip; + + sin = (struct sockaddr_in *)&child_ep->com.mapped_remote_addr; + sin->sin_family = PF_INET; + sin->sin_port = peer_port; + sin->sin_addr.s_addr = *(__be32 *)peer_ip; + } else { + struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *) + &child_ep->com.mapped_local_addr; + + sin6->sin6_family = PF_INET6; + sin6->sin6_port = local_port; + memcpy(sin6->sin6_addr.s6_addr, local_ip, 16); + + sin6 = (struct sockaddr_in6 *)&child_ep->com.local_addr; + sin6->sin6_family = PF_INET6; + sin6->sin6_port = ((struct sockaddr_in6 *) + &parent_ep->com.local_addr)->sin6_port; + memcpy(sin6->sin6_addr.s6_addr, local_ip, 16); + + sin6 = (struct sockaddr_in6 *)&child_ep->com.mapped_remote_addr; + sin6->sin6_family = PF_INET6; + sin6->sin6_port = peer_port; + memcpy(sin6->sin6_addr.s6_addr, peer_ip, 16); + } + memcpy(&child_ep->com.remote_addr, &child_ep->com.mapped_remote_addr, + sizeof(child_ep->com.remote_addr)); + get_remote_addr(parent_ep, child_ep); + + c4iw_get_ep(&parent_ep->com); + child_ep->parent_ep = parent_ep; + child_ep->tos = PASS_OPEN_TOS_G(ntohl(req->tos_stid)); + child_ep->dst = dst; + child_ep->hwtid = hwtid; + + PDBG("%s tx_chan %u smac_idx %u rss_qid %u\n", __func__, + child_ep->tx_chan, child_ep->smac_idx, child_ep->rss_qid); + + init_timer(&child_ep->timer); + cxgb4_insert_tid(t, child_ep, hwtid); + insert_handle(dev, &dev->hwtid_idr, child_ep, child_ep->hwtid); + accept_cr(child_ep, skb, req); + set_bit(PASS_ACCEPT_REQ, &child_ep->com.history); + goto out; +reject: + reject_cr(dev, hwtid, skb); +out: + return 0; +} + +static int pass_establish(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct cpl_pass_establish *req = cplhdr(skb); + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int tid = GET_TID(req); + + ep = lookup_tid(t, tid); + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + ep->snd_seq = be32_to_cpu(req->snd_isn); + ep->rcv_seq = be32_to_cpu(req->rcv_isn); + + PDBG("%s ep %p hwtid %u tcp_opt 0x%02x\n", __func__, ep, tid, + ntohs(req->tcp_opt)); + + set_emss(ep, ntohs(req->tcp_opt)); + + dst_confirm(ep->dst); + state_set(&ep->com, MPA_REQ_WAIT); + start_ep_timer(ep); + send_flowc(ep, skb); + set_bit(PASS_ESTAB, &ep->com.history); + + return 0; +} + +static int peer_close(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_peer_close *hdr = cplhdr(skb); + struct c4iw_ep *ep; + struct c4iw_qp_attributes attrs; + int disconnect = 1; + int release = 0; + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int tid = GET_TID(hdr); + int ret; + + ep = lookup_tid(t, tid); + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + dst_confirm(ep->dst); + + set_bit(PEER_CLOSE, &ep->com.history); + mutex_lock(&ep->com.mutex); + switch (ep->com.state) { + case MPA_REQ_WAIT: + __state_set(&ep->com, CLOSING); + break; + case MPA_REQ_SENT: + __state_set(&ep->com, CLOSING); + connect_reply_upcall(ep, -ECONNRESET); + break; + case MPA_REQ_RCVD: + + /* + * We're gonna mark this puppy DEAD, but keep + * the reference on it until the ULP accepts or + * rejects the CR. Also wake up anyone waiting + * in rdma connection migration (see c4iw_accept_cr()). + */ + __state_set(&ep->com, CLOSING); + PDBG("waking up ep %p tid %u\n", ep, ep->hwtid); + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + break; + case MPA_REP_SENT: + __state_set(&ep->com, CLOSING); + PDBG("waking up ep %p tid %u\n", ep, ep->hwtid); + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + break; + case FPDU_MODE: + start_ep_timer(ep); + __state_set(&ep->com, CLOSING); + attrs.next_state = C4IW_QP_STATE_CLOSING; + ret = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + if (ret != -ECONNRESET) { + peer_close_upcall(ep); + disconnect = 1; + } + break; + case ABORTING: + disconnect = 0; + break; + case CLOSING: + __state_set(&ep->com, MORIBUND); + disconnect = 0; + break; + case MORIBUND: + (void)stop_ep_timer(ep); + if (ep->com.cm_id && ep->com.qp) { + attrs.next_state = C4IW_QP_STATE_IDLE; + c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + } + close_complete_upcall(ep, 0); + __state_set(&ep->com, DEAD); + release = 1; + disconnect = 0; + break; + case DEAD: + disconnect = 0; + break; + default: + BUG_ON(1); + } + mutex_unlock(&ep->com.mutex); + if (disconnect) + c4iw_ep_disconnect(ep, 0, GFP_KERNEL); + if (release) + release_ep_resources(ep); + return 0; +} + +static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_abort_req_rss *req = cplhdr(skb); + struct c4iw_ep *ep; + struct cpl_abort_rpl *rpl; + struct sk_buff *rpl_skb; + struct c4iw_qp_attributes attrs; + int ret; + int release = 0; + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int tid = GET_TID(req); + + ep = lookup_tid(t, tid); + if (is_neg_adv(req->status)) { + PDBG("%s Negative advice on abort- tid %u status %d (%s)\n", + __func__, ep->hwtid, req->status, + neg_adv_str(req->status)); + ep->stats.abort_neg_adv++; + mutex_lock(&dev->rdev.stats.lock); + dev->rdev.stats.neg_adv++; + mutex_unlock(&dev->rdev.stats.lock); + return 0; + } + PDBG("%s ep %p tid %u state %u\n", __func__, ep, ep->hwtid, + ep->com.state); + set_bit(PEER_ABORT, &ep->com.history); + + /* + * Wake up any threads in rdma_init() or rdma_fini(). + * However, this is not needed if com state is just + * MPA_REQ_SENT + */ + if (ep->com.state != MPA_REQ_SENT) + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + + mutex_lock(&ep->com.mutex); + switch (ep->com.state) { + case CONNECTING: + break; + case MPA_REQ_WAIT: + (void)stop_ep_timer(ep); + break; + case MPA_REQ_SENT: + (void)stop_ep_timer(ep); + if (mpa_rev == 1 || (mpa_rev == 2 && ep->tried_with_mpa_v1)) + connect_reply_upcall(ep, -ECONNRESET); + else { + /* + * we just don't send notification upwards because we + * want to retry with mpa_v1 without upper layers even + * knowing it. + * + * do some housekeeping so as to re-initiate the + * connection + */ + PDBG("%s: mpa_rev=%d. Retrying with mpav1\n", __func__, + mpa_rev); + ep->retry_with_mpa_v1 = 1; + } + break; + case MPA_REP_SENT: + break; + case MPA_REQ_RCVD: + break; + case MORIBUND: + case CLOSING: + stop_ep_timer(ep); + /*FALLTHROUGH*/ + case FPDU_MODE: + if (ep->com.cm_id && ep->com.qp) { + attrs.next_state = C4IW_QP_STATE_ERROR; + ret = c4iw_modify_qp(ep->com.qp->rhp, + ep->com.qp, C4IW_QP_ATTR_NEXT_STATE, + &attrs, 1); + if (ret) + printk(KERN_ERR MOD + "%s - qp <- error failed!\n", + __func__); + } + peer_abort_upcall(ep); + break; + case ABORTING: + break; + case DEAD: + PDBG("%s PEER_ABORT IN DEAD STATE!!!!\n", __func__); + mutex_unlock(&ep->com.mutex); + return 0; + default: + BUG_ON(1); + break; + } + dst_confirm(ep->dst); + if (ep->com.state != ABORTING) { + __state_set(&ep->com, DEAD); + /* we don't release if we want to retry with mpa_v1 */ + if (!ep->retry_with_mpa_v1) + release = 1; + } + mutex_unlock(&ep->com.mutex); + + rpl_skb = get_skb(skb, sizeof(*rpl), GFP_KERNEL); + if (!rpl_skb) { + printk(KERN_ERR MOD "%s - cannot allocate skb!\n", + __func__); + release = 1; + goto out; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + rpl = (struct cpl_abort_rpl *) skb_put(rpl_skb, sizeof(*rpl)); + INIT_TP_WR(rpl, ep->hwtid); + OPCODE_TID(rpl) = cpu_to_be32(MK_OPCODE_TID(CPL_ABORT_RPL, ep->hwtid)); + rpl->cmd = CPL_ABORT_NO_RST; + c4iw_ofld_send(&ep->com.dev->rdev, rpl_skb); +out: + if (release) + release_ep_resources(ep); + else if (ep->retry_with_mpa_v1) { + remove_handle(ep->com.dev, &ep->com.dev->hwtid_idr, ep->hwtid); + cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, ep->hwtid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + c4iw_reconnect(ep); + } + + return 0; +} + +static int close_con_rpl(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct c4iw_qp_attributes attrs; + struct cpl_close_con_rpl *rpl = cplhdr(skb); + int release = 0; + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int tid = GET_TID(rpl); + + ep = lookup_tid(t, tid); + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + BUG_ON(!ep); + + /* The cm_id may be null if we failed to connect */ + mutex_lock(&ep->com.mutex); + switch (ep->com.state) { + case CLOSING: + __state_set(&ep->com, MORIBUND); + break; + case MORIBUND: + (void)stop_ep_timer(ep); + if ((ep->com.cm_id) && (ep->com.qp)) { + attrs.next_state = C4IW_QP_STATE_IDLE; + c4iw_modify_qp(ep->com.qp->rhp, + ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, + &attrs, 1); + } + close_complete_upcall(ep, 0); + __state_set(&ep->com, DEAD); + release = 1; + break; + case ABORTING: + case DEAD: + break; + default: + BUG_ON(1); + break; + } + mutex_unlock(&ep->com.mutex); + if (release) + release_ep_resources(ep); + return 0; +} + +static int terminate(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_rdma_terminate *rpl = cplhdr(skb); + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int tid = GET_TID(rpl); + struct c4iw_ep *ep; + struct c4iw_qp_attributes attrs; + + ep = lookup_tid(t, tid); + BUG_ON(!ep); + + if (ep && ep->com.qp) { + printk(KERN_WARNING MOD "TERM received tid %u qpid %u\n", tid, + ep->com.qp->wq.sq.qid); + attrs.next_state = C4IW_QP_STATE_TERMINATE; + c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + } else + printk(KERN_WARNING MOD "TERM received tid %u no ep/qp\n", tid); + + return 0; +} + +/* + * Upcall from the adapter indicating data has been transmitted. + * For us its just the single MPA request or reply. We can now free + * the skb holding the mpa message. + */ +static int fw4_ack(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct c4iw_ep *ep; + struct cpl_fw4_ack *hdr = cplhdr(skb); + u8 credits = hdr->credits; + unsigned int tid = GET_TID(hdr); + struct tid_info *t = dev->rdev.lldi.tids; + + + ep = lookup_tid(t, tid); + PDBG("%s ep %p tid %u credits %u\n", __func__, ep, ep->hwtid, credits); + if (credits == 0) { + PDBG("%s 0 credit ack ep %p tid %u state %u\n", + __func__, ep, ep->hwtid, state_read(&ep->com)); + return 0; + } + + dst_confirm(ep->dst); + if (ep->mpa_skb) { + PDBG("%s last streaming msg ack ep %p tid %u state %u " + "initiator %u freeing skb\n", __func__, ep, ep->hwtid, + state_read(&ep->com), ep->mpa_attr.initiator ? 1 : 0); + kfree_skb(ep->mpa_skb); + ep->mpa_skb = NULL; + } + return 0; +} + +int c4iw_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len) +{ + int err = 0; + int disconnect = 0; + struct c4iw_ep *ep = to_ep(cm_id); + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + + mutex_lock(&ep->com.mutex); + if (ep->com.state == DEAD) { + mutex_unlock(&ep->com.mutex); + c4iw_put_ep(&ep->com); + return -ECONNRESET; + } + set_bit(ULP_REJECT, &ep->com.history); + BUG_ON(ep->com.state != MPA_REQ_RCVD); + if (mpa_rev == 0) + abort_connection(ep, NULL, GFP_KERNEL); + else { + err = send_mpa_reject(ep, pdata, pdata_len); + disconnect = 1; + } + mutex_unlock(&ep->com.mutex); + if (disconnect) + err = c4iw_ep_disconnect(ep, 0, GFP_KERNEL); + c4iw_put_ep(&ep->com); + return 0; +} + +int c4iw_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) +{ + int err; + struct c4iw_qp_attributes attrs; + enum c4iw_qp_attr_mask mask; + struct c4iw_ep *ep = to_ep(cm_id); + struct c4iw_dev *h = to_c4iw_dev(cm_id->device); + struct c4iw_qp *qp = get_qhp(h, conn_param->qpn); + + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); + + mutex_lock(&ep->com.mutex); + if (ep->com.state == DEAD) { + err = -ECONNRESET; + goto err; + } + + BUG_ON(ep->com.state != MPA_REQ_RCVD); + BUG_ON(!qp); + + set_bit(ULP_ACCEPT, &ep->com.history); + if ((conn_param->ord > cur_max_read_depth(ep->com.dev)) || + (conn_param->ird > cur_max_read_depth(ep->com.dev))) { + abort_connection(ep, NULL, GFP_KERNEL); + err = -EINVAL; + goto err; + } + + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) { + if (conn_param->ord > ep->ird) { + if (RELAXED_IRD_NEGOTIATION) { + ep->ord = ep->ird; + } else { + ep->ird = conn_param->ird; + ep->ord = conn_param->ord; + send_mpa_reject(ep, conn_param->private_data, + conn_param->private_data_len); + abort_connection(ep, NULL, GFP_KERNEL); + err = -ENOMEM; + goto err; + } + } + if (conn_param->ird < ep->ord) { + if (RELAXED_IRD_NEGOTIATION && + ep->ord <= h->rdev.lldi.max_ordird_qp) { + conn_param->ird = ep->ord; + } else { + abort_connection(ep, NULL, GFP_KERNEL); + err = -ENOMEM; + goto err; + } + } + } + ep->ird = conn_param->ird; + ep->ord = conn_param->ord; + + if (ep->mpa_attr.version == 1) { + if (peer2peer && ep->ird == 0) + ep->ird = 1; + } else { + if (peer2peer && + (ep->mpa_attr.p2p_type != FW_RI_INIT_P2PTYPE_DISABLED) && + (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ) && ep->ord == 0) + ep->ird = 1; + } + + PDBG("%s %d ird %d ord %d\n", __func__, __LINE__, ep->ird, ep->ord); + + cm_id->add_ref(cm_id); + ep->com.cm_id = cm_id; + ep->com.qp = qp; + ref_qp(ep); + + /* bind QP to EP and move to RTS */ + attrs.mpa_attr = ep->mpa_attr; + attrs.max_ird = ep->ird; + attrs.max_ord = ep->ord; + attrs.llp_stream_handle = ep; + attrs.next_state = C4IW_QP_STATE_RTS; + + /* bind QP and TID with INIT_WR */ + mask = C4IW_QP_ATTR_NEXT_STATE | + C4IW_QP_ATTR_LLP_STREAM_HANDLE | + C4IW_QP_ATTR_MPA_ATTR | + C4IW_QP_ATTR_MAX_IRD | + C4IW_QP_ATTR_MAX_ORD; + + err = c4iw_modify_qp(ep->com.qp->rhp, + ep->com.qp, mask, &attrs, 1); + if (err) + goto err1; + err = send_mpa_reply(ep, conn_param->private_data, + conn_param->private_data_len); + if (err) + goto err1; + + __state_set(&ep->com, FPDU_MODE); + established_upcall(ep); + mutex_unlock(&ep->com.mutex); + c4iw_put_ep(&ep->com); + return 0; +err1: + ep->com.cm_id = NULL; + abort_connection(ep, NULL, GFP_KERNEL); + cm_id->rem_ref(cm_id); +err: + mutex_unlock(&ep->com.mutex); + c4iw_put_ep(&ep->com); + return err; +} + +static int pick_local_ipaddrs(struct c4iw_dev *dev, struct iw_cm_id *cm_id) +{ + struct in_device *ind; + int found = 0; + struct sockaddr_in *laddr = (struct sockaddr_in *)&cm_id->local_addr; + struct sockaddr_in *raddr = (struct sockaddr_in *)&cm_id->remote_addr; + + ind = in_dev_get(dev->rdev.lldi.ports[0]); + if (!ind) + return -EADDRNOTAVAIL; + for_primary_ifa(ind) { + laddr->sin_addr.s_addr = ifa->ifa_address; + raddr->sin_addr.s_addr = ifa->ifa_address; + found = 1; + break; + } + endfor_ifa(ind); + in_dev_put(ind); + return found ? 0 : -EADDRNOTAVAIL; +} + +static int get_lladdr(struct net_device *dev, struct in6_addr *addr, + unsigned char banned_flags) +{ + struct inet6_dev *idev; + int err = -EADDRNOTAVAIL; + + rcu_read_lock(); + idev = __in6_dev_get(dev); + if (idev != NULL) { + struct inet6_ifaddr *ifp; + + read_lock_bh(&idev->lock); + list_for_each_entry(ifp, &idev->addr_list, if_list) { + if (ifp->scope == IFA_LINK && + !(ifp->flags & banned_flags)) { + memcpy(addr, &ifp->addr, 16); + err = 0; + break; + } + } + read_unlock_bh(&idev->lock); + } + rcu_read_unlock(); + return err; +} + +static int pick_local_ip6addrs(struct c4iw_dev *dev, struct iw_cm_id *cm_id) +{ + struct in6_addr uninitialized_var(addr); + struct sockaddr_in6 *la6 = (struct sockaddr_in6 *)&cm_id->local_addr; + struct sockaddr_in6 *ra6 = (struct sockaddr_in6 *)&cm_id->remote_addr; + + if (get_lladdr(dev->rdev.lldi.ports[0], &addr, IFA_F_TENTATIVE)) { + memcpy(la6->sin6_addr.s6_addr, &addr, 16); + memcpy(ra6->sin6_addr.s6_addr, &addr, 16); + return 0; + } + return -EADDRNOTAVAIL; +} + +int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) +{ + struct c4iw_dev *dev = to_c4iw_dev(cm_id->device); + struct c4iw_ep *ep; + int err = 0; + struct sockaddr_in *laddr; + struct sockaddr_in *raddr; + struct sockaddr_in6 *laddr6; + struct sockaddr_in6 *raddr6; + struct iwpm_dev_data pm_reg_msg; + struct iwpm_sa_data pm_msg; + __u8 *ra; + int iptype; + int iwpm_err = 0; + + if ((conn_param->ord > cur_max_read_depth(dev)) || + (conn_param->ird > cur_max_read_depth(dev))) { + err = -EINVAL; + goto out; + } + ep = alloc_ep(sizeof(*ep), GFP_KERNEL); + if (!ep) { + printk(KERN_ERR MOD "%s - cannot alloc ep.\n", __func__); + err = -ENOMEM; + goto out; + } + init_timer(&ep->timer); + ep->plen = conn_param->private_data_len; + if (ep->plen) + memcpy(ep->mpa_pkt + sizeof(struct mpa_message), + conn_param->private_data, ep->plen); + ep->ird = conn_param->ird; + ep->ord = conn_param->ord; + + if (peer2peer && ep->ord == 0) + ep->ord = 1; + + cm_id->add_ref(cm_id); + ep->com.dev = dev; + ep->com.cm_id = cm_id; + ep->com.qp = get_qhp(dev, conn_param->qpn); + if (!ep->com.qp) { + PDBG("%s qpn 0x%x not found!\n", __func__, conn_param->qpn); + err = -EINVAL; + goto fail1; + } + ref_qp(ep); + PDBG("%s qpn 0x%x qp %p cm_id %p\n", __func__, conn_param->qpn, + ep->com.qp, cm_id); + + /* + * Allocate an active TID to initiate a TCP connection. + */ + ep->atid = cxgb4_alloc_atid(dev->rdev.lldi.tids, ep); + if (ep->atid == -1) { + printk(KERN_ERR MOD "%s - cannot alloc atid.\n", __func__); + err = -ENOMEM; + goto fail1; + } + insert_handle(dev, &dev->atid_idr, ep, ep->atid); + + memcpy(&ep->com.local_addr, &cm_id->local_addr, + sizeof(ep->com.local_addr)); + memcpy(&ep->com.remote_addr, &cm_id->remote_addr, + sizeof(ep->com.remote_addr)); + + /* No port mapper available, go with the specified peer information */ + memcpy(&ep->com.mapped_local_addr, &cm_id->local_addr, + sizeof(ep->com.mapped_local_addr)); + memcpy(&ep->com.mapped_remote_addr, &cm_id->remote_addr, + sizeof(ep->com.mapped_remote_addr)); + + c4iw_form_reg_msg(dev, &pm_reg_msg); + iwpm_err = iwpm_register_pid(&pm_reg_msg, RDMA_NL_C4IW); + if (iwpm_err) { + PDBG("%s: Port Mapper reg pid fail (err = %d).\n", + __func__, iwpm_err); + } + if (iwpm_valid_pid() && !iwpm_err) { + c4iw_form_pm_msg(ep, &pm_msg); + iwpm_err = iwpm_add_and_query_mapping(&pm_msg, RDMA_NL_C4IW); + if (iwpm_err) + PDBG("%s: Port Mapper query fail (err = %d).\n", + __func__, iwpm_err); + else + c4iw_record_pm_msg(ep, &pm_msg); + } + if (iwpm_create_mapinfo(&ep->com.local_addr, + &ep->com.mapped_local_addr, RDMA_NL_C4IW)) { + iwpm_remove_mapping(&ep->com.local_addr, RDMA_NL_C4IW); + err = -ENOMEM; + goto fail1; + } + print_addr(&ep->com, __func__, "add_query/create_mapinfo"); + set_bit(RELEASE_MAPINFO, &ep->com.flags); + + laddr = (struct sockaddr_in *)&ep->com.mapped_local_addr; + raddr = (struct sockaddr_in *)&ep->com.mapped_remote_addr; + laddr6 = (struct sockaddr_in6 *)&ep->com.mapped_local_addr; + raddr6 = (struct sockaddr_in6 *) &ep->com.mapped_remote_addr; + + if (cm_id->remote_addr.ss_family == AF_INET) { + iptype = 4; + ra = (__u8 *)&raddr->sin_addr; + + /* + * Handle loopback requests to INADDR_ANY. + */ + if ((__force int)raddr->sin_addr.s_addr == INADDR_ANY) { + err = pick_local_ipaddrs(dev, cm_id); + if (err) + goto fail1; + } + + /* find a route */ + PDBG("%s saddr %pI4 sport 0x%x raddr %pI4 rport 0x%x\n", + __func__, &laddr->sin_addr, ntohs(laddr->sin_port), + ra, ntohs(raddr->sin_port)); + ep->dst = find_route(dev, laddr->sin_addr.s_addr, + raddr->sin_addr.s_addr, laddr->sin_port, + raddr->sin_port, 0); + } else { + iptype = 6; + ra = (__u8 *)&raddr6->sin6_addr; + + /* + * Handle loopback requests to INADDR_ANY. + */ + if (ipv6_addr_type(&raddr6->sin6_addr) == IPV6_ADDR_ANY) { + err = pick_local_ip6addrs(dev, cm_id); + if (err) + goto fail1; + } + + /* find a route */ + PDBG("%s saddr %pI6 sport 0x%x raddr %pI6 rport 0x%x\n", + __func__, laddr6->sin6_addr.s6_addr, + ntohs(laddr6->sin6_port), + raddr6->sin6_addr.s6_addr, ntohs(raddr6->sin6_port)); + ep->dst = find_route6(dev, laddr6->sin6_addr.s6_addr, + raddr6->sin6_addr.s6_addr, + laddr6->sin6_port, raddr6->sin6_port, 0, + raddr6->sin6_scope_id); + } + if (!ep->dst) { + printk(KERN_ERR MOD "%s - cannot find route.\n", __func__); + err = -EHOSTUNREACH; + goto fail2; + } + + err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, true); + if (err) { + printk(KERN_ERR MOD "%s - cannot alloc l2e.\n", __func__); + goto fail3; + } + + PDBG("%s txq_idx %u tx_chan %u smac_idx %u rss_qid %u l2t_idx %u\n", + __func__, ep->txq_idx, ep->tx_chan, ep->smac_idx, ep->rss_qid, + ep->l2t->idx); + + state_set(&ep->com, CONNECTING); + ep->tos = 0; + + /* send connect request to rnic */ + err = send_connect(ep); + if (!err) + goto out; + + cxgb4_l2t_release(ep->l2t); +fail3: + dst_release(ep->dst); +fail2: + remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid); + cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid); +fail1: + cm_id->rem_ref(cm_id); + c4iw_put_ep(&ep->com); +out: + return err; +} + +static int create_server6(struct c4iw_dev *dev, struct c4iw_listen_ep *ep) +{ + int err; + struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *) + &ep->com.mapped_local_addr; + + c4iw_init_wr_wait(&ep->com.wr_wait); + err = cxgb4_create_server6(ep->com.dev->rdev.lldi.ports[0], + ep->stid, &sin6->sin6_addr, + sin6->sin6_port, + ep->com.dev->rdev.lldi.rxq_ids[0]); + if (!err) + err = c4iw_wait_for_reply(&ep->com.dev->rdev, + &ep->com.wr_wait, + 0, 0, __func__); + else if (err > 0) + err = net_xmit_errno(err); + if (err) + pr_err("cxgb4_create_server6/filter failed err %d stid %d laddr %pI6 lport %d\n", + err, ep->stid, + sin6->sin6_addr.s6_addr, ntohs(sin6->sin6_port)); + return err; +} + +static int create_server4(struct c4iw_dev *dev, struct c4iw_listen_ep *ep) +{ + int err; + struct sockaddr_in *sin = (struct sockaddr_in *) + &ep->com.mapped_local_addr; + + if (dev->rdev.lldi.enable_fw_ofld_conn) { + do { + err = cxgb4_create_server_filter( + ep->com.dev->rdev.lldi.ports[0], ep->stid, + sin->sin_addr.s_addr, sin->sin_port, 0, + ep->com.dev->rdev.lldi.rxq_ids[0], 0, 0); + if (err == -EBUSY) { + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(usecs_to_jiffies(100)); + } + } while (err == -EBUSY); + } else { + c4iw_init_wr_wait(&ep->com.wr_wait); + err = cxgb4_create_server(ep->com.dev->rdev.lldi.ports[0], + ep->stid, sin->sin_addr.s_addr, sin->sin_port, + 0, ep->com.dev->rdev.lldi.rxq_ids[0]); + if (!err) + err = c4iw_wait_for_reply(&ep->com.dev->rdev, + &ep->com.wr_wait, + 0, 0, __func__); + else if (err > 0) + err = net_xmit_errno(err); + } + if (err) + pr_err("cxgb4_create_server/filter failed err %d stid %d laddr %pI4 lport %d\n" + , err, ep->stid, + &sin->sin_addr, ntohs(sin->sin_port)); + return err; +} + +int c4iw_create_listen(struct iw_cm_id *cm_id, int backlog) +{ + int err = 0; + struct c4iw_dev *dev = to_c4iw_dev(cm_id->device); + struct c4iw_listen_ep *ep; + struct iwpm_dev_data pm_reg_msg; + struct iwpm_sa_data pm_msg; + int iwpm_err = 0; + + might_sleep(); + + ep = alloc_ep(sizeof(*ep), GFP_KERNEL); + if (!ep) { + printk(KERN_ERR MOD "%s - cannot alloc ep.\n", __func__); + err = -ENOMEM; + goto fail1; + } + PDBG("%s ep %p\n", __func__, ep); + cm_id->add_ref(cm_id); + ep->com.cm_id = cm_id; + ep->com.dev = dev; + ep->backlog = backlog; + memcpy(&ep->com.local_addr, &cm_id->local_addr, + sizeof(ep->com.local_addr)); + + /* + * Allocate a server TID. + */ + if (dev->rdev.lldi.enable_fw_ofld_conn && + ep->com.local_addr.ss_family == AF_INET) + ep->stid = cxgb4_alloc_sftid(dev->rdev.lldi.tids, + cm_id->local_addr.ss_family, ep); + else + ep->stid = cxgb4_alloc_stid(dev->rdev.lldi.tids, + cm_id->local_addr.ss_family, ep); + + if (ep->stid == -1) { + printk(KERN_ERR MOD "%s - cannot alloc stid.\n", __func__); + err = -ENOMEM; + goto fail2; + } + insert_handle(dev, &dev->stid_idr, ep, ep->stid); + + /* No port mapper available, go with the specified info */ + memcpy(&ep->com.mapped_local_addr, &cm_id->local_addr, + sizeof(ep->com.mapped_local_addr)); + + c4iw_form_reg_msg(dev, &pm_reg_msg); + iwpm_err = iwpm_register_pid(&pm_reg_msg, RDMA_NL_C4IW); + if (iwpm_err) { + PDBG("%s: Port Mapper reg pid fail (err = %d).\n", + __func__, iwpm_err); + } + if (iwpm_valid_pid() && !iwpm_err) { + memcpy(&pm_msg.loc_addr, &ep->com.local_addr, + sizeof(ep->com.local_addr)); + iwpm_err = iwpm_add_mapping(&pm_msg, RDMA_NL_C4IW); + if (iwpm_err) + PDBG("%s: Port Mapper query fail (err = %d).\n", + __func__, iwpm_err); + else + memcpy(&ep->com.mapped_local_addr, + &pm_msg.mapped_loc_addr, + sizeof(ep->com.mapped_local_addr)); + } + if (iwpm_create_mapinfo(&ep->com.local_addr, + &ep->com.mapped_local_addr, RDMA_NL_C4IW)) { + err = -ENOMEM; + goto fail3; + } + print_addr(&ep->com, __func__, "add_mapping/create_mapinfo"); + + set_bit(RELEASE_MAPINFO, &ep->com.flags); + state_set(&ep->com, LISTEN); + if (ep->com.local_addr.ss_family == AF_INET) + err = create_server4(dev, ep); + else + err = create_server6(dev, ep); + if (!err) { + cm_id->provider_data = ep; + goto out; + } + +fail3: + cxgb4_free_stid(ep->com.dev->rdev.lldi.tids, ep->stid, + ep->com.local_addr.ss_family); +fail2: + cm_id->rem_ref(cm_id); + c4iw_put_ep(&ep->com); +fail1: +out: + return err; +} + +int c4iw_destroy_listen(struct iw_cm_id *cm_id) +{ + int err; + struct c4iw_listen_ep *ep = to_listen_ep(cm_id); + + PDBG("%s ep %p\n", __func__, ep); + + might_sleep(); + state_set(&ep->com, DEAD); + if (ep->com.dev->rdev.lldi.enable_fw_ofld_conn && + ep->com.local_addr.ss_family == AF_INET) { + err = cxgb4_remove_server_filter( + ep->com.dev->rdev.lldi.ports[0], ep->stid, + ep->com.dev->rdev.lldi.rxq_ids[0], 0); + } else { + c4iw_init_wr_wait(&ep->com.wr_wait); + err = cxgb4_remove_server( + ep->com.dev->rdev.lldi.ports[0], ep->stid, + ep->com.dev->rdev.lldi.rxq_ids[0], 0); + if (err) + goto done; + err = c4iw_wait_for_reply(&ep->com.dev->rdev, &ep->com.wr_wait, + 0, 0, __func__); + } + remove_handle(ep->com.dev, &ep->com.dev->stid_idr, ep->stid); + cxgb4_free_stid(ep->com.dev->rdev.lldi.tids, ep->stid, + ep->com.local_addr.ss_family); +done: + cm_id->rem_ref(cm_id); + c4iw_put_ep(&ep->com); + return err; +} + +int c4iw_ep_disconnect(struct c4iw_ep *ep, int abrupt, gfp_t gfp) +{ + int ret = 0; + int close = 0; + int fatal = 0; + struct c4iw_rdev *rdev; + + mutex_lock(&ep->com.mutex); + + PDBG("%s ep %p state %s, abrupt %d\n", __func__, ep, + states[ep->com.state], abrupt); + + rdev = &ep->com.dev->rdev; + if (c4iw_fatal_error(rdev)) { + fatal = 1; + close_complete_upcall(ep, -EIO); + ep->com.state = DEAD; + } + switch (ep->com.state) { + case MPA_REQ_WAIT: + case MPA_REQ_SENT: + case MPA_REQ_RCVD: + case MPA_REP_SENT: + case FPDU_MODE: + close = 1; + if (abrupt) + ep->com.state = ABORTING; + else { + ep->com.state = CLOSING; + start_ep_timer(ep); + } + set_bit(CLOSE_SENT, &ep->com.flags); + break; + case CLOSING: + if (!test_and_set_bit(CLOSE_SENT, &ep->com.flags)) { + close = 1; + if (abrupt) { + (void)stop_ep_timer(ep); + ep->com.state = ABORTING; + } else + ep->com.state = MORIBUND; + } + break; + case MORIBUND: + case ABORTING: + case DEAD: + PDBG("%s ignoring disconnect ep %p state %u\n", + __func__, ep, ep->com.state); + break; + default: + BUG(); + break; + } + + if (close) { + if (abrupt) { + set_bit(EP_DISC_ABORT, &ep->com.history); + close_complete_upcall(ep, -ECONNRESET); + ret = send_abort(ep, NULL, gfp); + } else { + set_bit(EP_DISC_CLOSE, &ep->com.history); + ret = send_halfclose(ep, gfp); + } + if (ret) + fatal = 1; + } + mutex_unlock(&ep->com.mutex); + if (fatal) + release_ep_resources(ep); + return ret; +} + +static void active_ofld_conn_reply(struct c4iw_dev *dev, struct sk_buff *skb, + struct cpl_fw6_msg_ofld_connection_wr_rpl *req) +{ + struct c4iw_ep *ep; + int atid = be32_to_cpu(req->tid); + + ep = (struct c4iw_ep *)lookup_atid(dev->rdev.lldi.tids, + (__force u32) req->tid); + if (!ep) + return; + + switch (req->retval) { + case FW_ENOMEM: + set_bit(ACT_RETRY_NOMEM, &ep->com.history); + if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) { + send_fw_act_open_req(ep, atid); + return; + } + case FW_EADDRINUSE: + set_bit(ACT_RETRY_INUSE, &ep->com.history); + if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) { + send_fw_act_open_req(ep, atid); + return; + } + break; + default: + pr_info("%s unexpected ofld conn wr retval %d\n", + __func__, req->retval); + break; + } + pr_err("active ofld_connect_wr failure %d atid %d\n", + req->retval, atid); + mutex_lock(&dev->rdev.stats.lock); + dev->rdev.stats.act_ofld_conn_fails++; + mutex_unlock(&dev->rdev.stats.lock); + connect_reply_upcall(ep, status2errno(req->retval)); + state_set(&ep->com, DEAD); + remove_handle(dev, &dev->atid_idr, atid); + cxgb4_free_atid(dev->rdev.lldi.tids, atid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + c4iw_put_ep(&ep->com); +} + +static void passive_ofld_conn_reply(struct c4iw_dev *dev, struct sk_buff *skb, + struct cpl_fw6_msg_ofld_connection_wr_rpl *req) +{ + struct sk_buff *rpl_skb; + struct cpl_pass_accept_req *cpl; + int ret; + + rpl_skb = (struct sk_buff *)(unsigned long)req->cookie; + BUG_ON(!rpl_skb); + if (req->retval) { + PDBG("%s passive open failure %d\n", __func__, req->retval); + mutex_lock(&dev->rdev.stats.lock); + dev->rdev.stats.pas_ofld_conn_fails++; + mutex_unlock(&dev->rdev.stats.lock); + kfree_skb(rpl_skb); + } else { + cpl = (struct cpl_pass_accept_req *)cplhdr(rpl_skb); + OPCODE_TID(cpl) = htonl(MK_OPCODE_TID(CPL_PASS_ACCEPT_REQ, + (__force u32) htonl( + (__force u32) req->tid))); + ret = pass_accept_req(dev, rpl_skb); + if (!ret) + kfree_skb(rpl_skb); + } + return; +} + +static int deferred_fw6_msg(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_fw6_msg *rpl = cplhdr(skb); + struct cpl_fw6_msg_ofld_connection_wr_rpl *req; + + switch (rpl->type) { + case FW6_TYPE_CQE: + c4iw_ev_dispatch(dev, (struct t4_cqe *)&rpl->data[0]); + break; + case FW6_TYPE_OFLD_CONNECTION_WR_RPL: + req = (struct cpl_fw6_msg_ofld_connection_wr_rpl *)rpl->data; + switch (req->t_state) { + case TCP_SYN_SENT: + active_ofld_conn_reply(dev, skb, req); + break; + case TCP_SYN_RECV: + passive_ofld_conn_reply(dev, skb, req); + break; + default: + pr_err("%s unexpected ofld conn wr state %d\n", + __func__, req->t_state); + break; + } + break; + } + return 0; +} + +static void build_cpl_pass_accept_req(struct sk_buff *skb, int stid , u8 tos) +{ + u32 l2info; + u16 vlantag, len, hdr_len, eth_hdr_len; + u8 intf; + struct cpl_rx_pkt *cpl = cplhdr(skb); + struct cpl_pass_accept_req *req; + struct tcp_options_received tmp_opt; + struct c4iw_dev *dev; + + dev = *((struct c4iw_dev **) (skb->cb + sizeof(void *))); + /* Store values from cpl_rx_pkt in temporary location. */ + vlantag = (__force u16) cpl->vlan; + len = (__force u16) cpl->len; + l2info = (__force u32) cpl->l2info; + hdr_len = (__force u16) cpl->hdr_len; + intf = cpl->iff; + + __skb_pull(skb, sizeof(*req) + sizeof(struct rss_header)); + + /* + * We need to parse the TCP options from SYN packet. + * to generate cpl_pass_accept_req. + */ + memset(&tmp_opt, 0, sizeof(tmp_opt)); + tcp_clear_options(&tmp_opt); + tcp_parse_options(skb, &tmp_opt, 0, NULL); + + req = (struct cpl_pass_accept_req *)__skb_push(skb, sizeof(*req)); + memset(req, 0, sizeof(*req)); + req->l2info = cpu_to_be16(SYN_INTF_V(intf) | + SYN_MAC_IDX_V(RX_MACIDX_G( + (__force int) htonl(l2info))) | + SYN_XACT_MATCH_F); + eth_hdr_len = is_t4(dev->rdev.lldi.adapter_type) ? + RX_ETHHDR_LEN_G((__force int)htonl(l2info)) : + RX_T5_ETHHDR_LEN_G((__force int)htonl(l2info)); + req->hdr_len = cpu_to_be32(SYN_RX_CHAN_V(RX_CHAN_G( + (__force int) htonl(l2info))) | + TCP_HDR_LEN_V(RX_TCPHDR_LEN_G( + (__force int) htons(hdr_len))) | + IP_HDR_LEN_V(RX_IPHDR_LEN_G( + (__force int) htons(hdr_len))) | + ETH_HDR_LEN_V(RX_ETHHDR_LEN_G(eth_hdr_len))); + req->vlan = (__force __be16) vlantag; + req->len = (__force __be16) len; + req->tos_stid = cpu_to_be32(PASS_OPEN_TID_V(stid) | + PASS_OPEN_TOS_V(tos)); + req->tcpopt.mss = htons(tmp_opt.mss_clamp); + if (tmp_opt.wscale_ok) + req->tcpopt.wsf = tmp_opt.snd_wscale; + req->tcpopt.tstamp = tmp_opt.saw_tstamp; + if (tmp_opt.sack_ok) + req->tcpopt.sack = 1; + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_PASS_ACCEPT_REQ, 0)); + return; +} + +static void send_fw_pass_open_req(struct c4iw_dev *dev, struct sk_buff *skb, + __be32 laddr, __be16 lport, + __be32 raddr, __be16 rport, + u32 rcv_isn, u32 filter, u16 window, + u32 rss_qid, u8 port_id) +{ + struct sk_buff *req_skb; + struct fw_ofld_connection_wr *req; + struct cpl_pass_accept_req *cpl = cplhdr(skb); + int ret; + + req_skb = alloc_skb(sizeof(struct fw_ofld_connection_wr), GFP_KERNEL); + req = (struct fw_ofld_connection_wr *)__skb_put(req_skb, sizeof(*req)); + memset(req, 0, sizeof(*req)); + req->op_compl = htonl(WR_OP_V(FW_OFLD_CONNECTION_WR) | FW_WR_COMPL_F); + req->len16_pkd = htonl(FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*req), 16))); + req->le.version_cpl = htonl(FW_OFLD_CONNECTION_WR_CPL_F); + req->le.filter = (__force __be32) filter; + req->le.lport = lport; + req->le.pport = rport; + req->le.u.ipv4.lip = laddr; + req->le.u.ipv4.pip = raddr; + req->tcb.rcv_nxt = htonl(rcv_isn + 1); + req->tcb.rcv_adv = htons(window); + req->tcb.t_state_to_astid = + htonl(FW_OFLD_CONNECTION_WR_T_STATE_V(TCP_SYN_RECV) | + FW_OFLD_CONNECTION_WR_RCV_SCALE_V(cpl->tcpopt.wsf) | + FW_OFLD_CONNECTION_WR_ASTID_V( + PASS_OPEN_TID_G(ntohl(cpl->tos_stid)))); + + /* + * We store the qid in opt2 which will be used by the firmware + * to send us the wr response. + */ + req->tcb.opt2 = htonl(RSS_QUEUE_V(rss_qid)); + + /* + * We initialize the MSS index in TCB to 0xF. + * So that when driver sends cpl_pass_accept_rpl + * TCB picks up the correct value. If this was 0 + * TP will ignore any value > 0 for MSS index. + */ + req->tcb.opt0 = cpu_to_be64(MSS_IDX_V(0xF)); + req->cookie = (uintptr_t)skb; + + set_wr_txq(req_skb, CPL_PRIORITY_CONTROL, port_id); + ret = cxgb4_ofld_send(dev->rdev.lldi.ports[0], req_skb); + if (ret < 0) { + pr_err("%s - cxgb4_ofld_send error %d - dropping\n", __func__, + ret); + kfree_skb(skb); + kfree_skb(req_skb); + } +} + +/* + * Handler for CPL_RX_PKT message. Need to handle cpl_rx_pkt + * messages when a filter is being used instead of server to + * redirect a syn packet. When packets hit filter they are redirected + * to the offload queue and driver tries to establish the connection + * using firmware work request. + */ +static int rx_pkt(struct c4iw_dev *dev, struct sk_buff *skb) +{ + int stid; + unsigned int filter; + struct ethhdr *eh = NULL; + struct vlan_ethhdr *vlan_eh = NULL; + struct iphdr *iph; + struct tcphdr *tcph; + struct rss_header *rss = (void *)skb->data; + struct cpl_rx_pkt *cpl = (void *)skb->data; + struct cpl_pass_accept_req *req = (void *)(rss + 1); + struct l2t_entry *e; + struct dst_entry *dst; + struct c4iw_ep *lep; + u16 window; + struct port_info *pi; + struct net_device *pdev; + u16 rss_qid, eth_hdr_len; + int step; + u32 tx_chan; + struct neighbour *neigh; + + /* Drop all non-SYN packets */ + if (!(cpl->l2info & cpu_to_be32(RXF_SYN_F))) + goto reject; + + /* + * Drop all packets which did not hit the filter. + * Unlikely to happen. + */ + if (!(rss->filter_hit && rss->filter_tid)) + goto reject; + + /* + * Calculate the server tid from filter hit index from cpl_rx_pkt. + */ + stid = (__force int) cpu_to_be32((__force u32) rss->hash_val); + + lep = (struct c4iw_ep *)lookup_stid(dev->rdev.lldi.tids, stid); + if (!lep) { + PDBG("%s connect request on invalid stid %d\n", __func__, stid); + goto reject; + } + + eth_hdr_len = is_t4(dev->rdev.lldi.adapter_type) ? + RX_ETHHDR_LEN_G(htonl(cpl->l2info)) : + RX_T5_ETHHDR_LEN_G(htonl(cpl->l2info)); + if (eth_hdr_len == ETH_HLEN) { + eh = (struct ethhdr *)(req + 1); + iph = (struct iphdr *)(eh + 1); + } else { + vlan_eh = (struct vlan_ethhdr *)(req + 1); + iph = (struct iphdr *)(vlan_eh + 1); + skb->vlan_tci = ntohs(cpl->vlan); + } + + if (iph->version != 0x4) + goto reject; + + tcph = (struct tcphdr *)(iph + 1); + skb_set_network_header(skb, (void *)iph - (void *)rss); + skb_set_transport_header(skb, (void *)tcph - (void *)rss); + skb_get(skb); + + PDBG("%s lip 0x%x lport %u pip 0x%x pport %u tos %d\n", __func__, + ntohl(iph->daddr), ntohs(tcph->dest), ntohl(iph->saddr), + ntohs(tcph->source), iph->tos); + + dst = find_route(dev, iph->daddr, iph->saddr, tcph->dest, tcph->source, + iph->tos); + if (!dst) { + pr_err("%s - failed to find dst entry!\n", + __func__); + goto reject; + } + neigh = dst_neigh_lookup_skb(dst, skb); + + if (!neigh) { + pr_err("%s - failed to allocate neigh!\n", + __func__); + goto free_dst; + } + + if (neigh->dev->flags & IFF_LOOPBACK) { + pdev = ip_dev_find(&init_net, iph->daddr); + e = cxgb4_l2t_get(dev->rdev.lldi.l2t, neigh, + pdev, 0); + pi = (struct port_info *)netdev_priv(pdev); + tx_chan = cxgb4_port_chan(pdev); + dev_put(pdev); + } else { + pdev = get_real_dev(neigh->dev); + e = cxgb4_l2t_get(dev->rdev.lldi.l2t, neigh, + pdev, 0); + pi = (struct port_info *)netdev_priv(pdev); + tx_chan = cxgb4_port_chan(pdev); + } + neigh_release(neigh); + if (!e) { + pr_err("%s - failed to allocate l2t entry!\n", + __func__); + goto free_dst; + } + + step = dev->rdev.lldi.nrxq / dev->rdev.lldi.nchan; + rss_qid = dev->rdev.lldi.rxq_ids[pi->port_id * step]; + window = (__force u16) htons((__force u16)tcph->window); + + /* Calcuate filter portion for LE region. */ + filter = (__force unsigned int) cpu_to_be32(cxgb4_select_ntuple( + dev->rdev.lldi.ports[0], + e)); + + /* + * Synthesize the cpl_pass_accept_req. We have everything except the + * TID. Once firmware sends a reply with TID we update the TID field + * in cpl and pass it through the regular cpl_pass_accept_req path. + */ + build_cpl_pass_accept_req(skb, stid, iph->tos); + send_fw_pass_open_req(dev, skb, iph->daddr, tcph->dest, iph->saddr, + tcph->source, ntohl(tcph->seq), filter, window, + rss_qid, pi->port_id); + cxgb4_l2t_release(e); +free_dst: + dst_release(dst); +reject: + return 0; +} + +/* + * These are the real handlers that are called from a + * work queue. + */ +static c4iw_handler_func work_handlers[NUM_CPL_CMDS] = { + [CPL_ACT_ESTABLISH] = act_establish, + [CPL_ACT_OPEN_RPL] = act_open_rpl, + [CPL_RX_DATA] = rx_data, + [CPL_ABORT_RPL_RSS] = abort_rpl, + [CPL_ABORT_RPL] = abort_rpl, + [CPL_PASS_OPEN_RPL] = pass_open_rpl, + [CPL_CLOSE_LISTSRV_RPL] = close_listsrv_rpl, + [CPL_PASS_ACCEPT_REQ] = pass_accept_req, + [CPL_PASS_ESTABLISH] = pass_establish, + [CPL_PEER_CLOSE] = peer_close, + [CPL_ABORT_REQ_RSS] = peer_abort, + [CPL_CLOSE_CON_RPL] = close_con_rpl, + [CPL_RDMA_TERMINATE] = terminate, + [CPL_FW4_ACK] = fw4_ack, + [CPL_FW6_MSG] = deferred_fw6_msg, + [CPL_RX_PKT] = rx_pkt +}; + +static void process_timeout(struct c4iw_ep *ep) +{ + struct c4iw_qp_attributes attrs; + int abort = 1; + + mutex_lock(&ep->com.mutex); + PDBG("%s ep %p tid %u state %d\n", __func__, ep, ep->hwtid, + ep->com.state); + set_bit(TIMEDOUT, &ep->com.history); + switch (ep->com.state) { + case MPA_REQ_SENT: + __state_set(&ep->com, ABORTING); + connect_reply_upcall(ep, -ETIMEDOUT); + break; + case MPA_REQ_WAIT: + __state_set(&ep->com, ABORTING); + break; + case CLOSING: + case MORIBUND: + if (ep->com.cm_id && ep->com.qp) { + attrs.next_state = C4IW_QP_STATE_ERROR; + c4iw_modify_qp(ep->com.qp->rhp, + ep->com.qp, C4IW_QP_ATTR_NEXT_STATE, + &attrs, 1); + } + __state_set(&ep->com, ABORTING); + close_complete_upcall(ep, -ETIMEDOUT); + break; + case ABORTING: + case DEAD: + + /* + * These states are expected if the ep timed out at the same + * time as another thread was calling stop_ep_timer(). + * So we silently do nothing for these states. + */ + abort = 0; + break; + default: + WARN(1, "%s unexpected state ep %p tid %u state %u\n", + __func__, ep, ep->hwtid, ep->com.state); + abort = 0; + } + if (abort) + abort_connection(ep, NULL, GFP_KERNEL); + mutex_unlock(&ep->com.mutex); + c4iw_put_ep(&ep->com); +} + +static void process_timedout_eps(void) +{ + struct c4iw_ep *ep; + + spin_lock_irq(&timeout_lock); + while (!list_empty(&timeout_list)) { + struct list_head *tmp; + + tmp = timeout_list.next; + list_del(tmp); + tmp->next = NULL; + tmp->prev = NULL; + spin_unlock_irq(&timeout_lock); + ep = list_entry(tmp, struct c4iw_ep, entry); + process_timeout(ep); + spin_lock_irq(&timeout_lock); + } + spin_unlock_irq(&timeout_lock); +} + +static void process_work(struct work_struct *work) +{ + struct sk_buff *skb = NULL; + struct c4iw_dev *dev; + struct cpl_act_establish *rpl; + unsigned int opcode; + int ret; + + process_timedout_eps(); + while ((skb = skb_dequeue(&rxq))) { + rpl = cplhdr(skb); + dev = *((struct c4iw_dev **) (skb->cb + sizeof(void *))); + opcode = rpl->ot.opcode; + + BUG_ON(!work_handlers[opcode]); + ret = work_handlers[opcode](dev, skb); + if (!ret) + kfree_skb(skb); + process_timedout_eps(); + } +} + +static DECLARE_WORK(skb_work, process_work); + +static void ep_timeout(unsigned long arg) +{ + struct c4iw_ep *ep = (struct c4iw_ep *)arg; + int kickit = 0; + + spin_lock(&timeout_lock); + if (!test_and_set_bit(TIMEOUT, &ep->com.flags)) { + /* + * Only insert if it is not already on the list. + */ + if (!ep->entry.next) { + list_add_tail(&ep->entry, &timeout_list); + kickit = 1; + } + } + spin_unlock(&timeout_lock); + if (kickit) + queue_work(workq, &skb_work); +} + +/* + * All the CM events are handled on a work queue to have a safe context. + */ +static int sched(struct c4iw_dev *dev, struct sk_buff *skb) +{ + + /* + * Save dev in the skb->cb area. + */ + *((struct c4iw_dev **) (skb->cb + sizeof(void *))) = dev; + + /* + * Queue the skb and schedule the worker thread. + */ + skb_queue_tail(&rxq, skb); + queue_work(workq, &skb_work); + return 0; +} + +static int set_tcb_rpl(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_set_tcb_rpl *rpl = cplhdr(skb); + + if (rpl->status != CPL_ERR_NONE) { + printk(KERN_ERR MOD "Unexpected SET_TCB_RPL status %u " + "for tid %u\n", rpl->status, GET_TID(rpl)); + } + kfree_skb(skb); + return 0; +} + +static int fw6_msg(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_fw6_msg *rpl = cplhdr(skb); + struct c4iw_wr_wait *wr_waitp; + int ret; + + PDBG("%s type %u\n", __func__, rpl->type); + + switch (rpl->type) { + case FW6_TYPE_WR_RPL: + ret = (int)((be64_to_cpu(rpl->data[0]) >> 8) & 0xff); + wr_waitp = (struct c4iw_wr_wait *)(__force unsigned long) rpl->data[1]; + PDBG("%s wr_waitp %p ret %u\n", __func__, wr_waitp, ret); + if (wr_waitp) + c4iw_wake_up(wr_waitp, ret ? -ret : 0); + kfree_skb(skb); + break; + case FW6_TYPE_CQE: + case FW6_TYPE_OFLD_CONNECTION_WR_RPL: + sched(dev, skb); + break; + default: + printk(KERN_ERR MOD "%s unexpected fw6 msg type %u\n", __func__, + rpl->type); + kfree_skb(skb); + break; + } + return 0; +} + +static int peer_abort_intr(struct c4iw_dev *dev, struct sk_buff *skb) +{ + struct cpl_abort_req_rss *req = cplhdr(skb); + struct c4iw_ep *ep; + struct tid_info *t = dev->rdev.lldi.tids; + unsigned int tid = GET_TID(req); + + ep = lookup_tid(t, tid); + if (!ep) { + printk(KERN_WARNING MOD + "Abort on non-existent endpoint, tid %d\n", tid); + kfree_skb(skb); + return 0; + } + if (is_neg_adv(req->status)) { + PDBG("%s Negative advice on abort- tid %u status %d (%s)\n", + __func__, ep->hwtid, req->status, + neg_adv_str(req->status)); + ep->stats.abort_neg_adv++; + dev->rdev.stats.neg_adv++; + kfree_skb(skb); + return 0; + } + PDBG("%s ep %p tid %u state %u\n", __func__, ep, ep->hwtid, + ep->com.state); + + /* + * Wake up any threads in rdma_init() or rdma_fini(). + * However, if we are on MPAv2 and want to retry with MPAv1 + * then, don't wake up yet. + */ + if (mpa_rev == 2 && !ep->tried_with_mpa_v1) { + if (ep->com.state != MPA_REQ_SENT) + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + } else + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + sched(dev, skb); + return 0; +} + +/* + * Most upcalls from the T4 Core go to sched() to + * schedule the processing on a work queue. + */ +c4iw_handler_func c4iw_handlers[NUM_CPL_CMDS] = { + [CPL_ACT_ESTABLISH] = sched, + [CPL_ACT_OPEN_RPL] = sched, + [CPL_RX_DATA] = sched, + [CPL_ABORT_RPL_RSS] = sched, + [CPL_ABORT_RPL] = sched, + [CPL_PASS_OPEN_RPL] = sched, + [CPL_CLOSE_LISTSRV_RPL] = sched, + [CPL_PASS_ACCEPT_REQ] = sched, + [CPL_PASS_ESTABLISH] = sched, + [CPL_PEER_CLOSE] = sched, + [CPL_CLOSE_CON_RPL] = sched, + [CPL_ABORT_REQ_RSS] = peer_abort_intr, + [CPL_RDMA_TERMINATE] = sched, + [CPL_FW4_ACK] = sched, + [CPL_SET_TCB_RPL] = set_tcb_rpl, + [CPL_FW6_MSG] = fw6_msg, + [CPL_RX_PKT] = sched +}; + +int __init c4iw_cm_init(void) +{ + spin_lock_init(&timeout_lock); + skb_queue_head_init(&rxq); + + workq = create_singlethread_workqueue("iw_cxgb4"); + if (!workq) + return -ENOMEM; + + return 0; +} + +void c4iw_cm_term(void) +{ + WARN_ON(!list_empty(&timeout_list)); + flush_workqueue(workq); + destroy_workqueue(workq); +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/cq.c b/kernel/drivers/infiniband/hw/cxgb4/cq.c new file mode 100644 index 000000000..68ddb3710 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/cq.c @@ -0,0 +1,1015 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ + +#include "iw_cxgb4.h" + +static int destroy_cq(struct c4iw_rdev *rdev, struct t4_cq *cq, + struct c4iw_dev_ucontext *uctx) +{ + struct fw_ri_res_wr *res_wr; + struct fw_ri_res *res; + int wr_len; + struct c4iw_wr_wait wr_wait; + struct sk_buff *skb; + int ret; + + wr_len = sizeof *res_wr + sizeof *res; + skb = alloc_skb(wr_len, GFP_KERNEL); + if (!skb) + return -ENOMEM; + set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); + + res_wr = (struct fw_ri_res_wr *)__skb_put(skb, wr_len); + memset(res_wr, 0, wr_len); + res_wr->op_nres = cpu_to_be32( + FW_WR_OP_V(FW_RI_RES_WR) | + FW_RI_RES_WR_NRES_V(1) | + FW_WR_COMPL_F); + res_wr->len16_pkd = cpu_to_be32(DIV_ROUND_UP(wr_len, 16)); + res_wr->cookie = (uintptr_t)&wr_wait; + res = res_wr->res; + res->u.cq.restype = FW_RI_RES_TYPE_CQ; + res->u.cq.op = FW_RI_RES_OP_RESET; + res->u.cq.iqid = cpu_to_be32(cq->cqid); + + c4iw_init_wr_wait(&wr_wait); + ret = c4iw_ofld_send(rdev, skb); + if (!ret) { + ret = c4iw_wait_for_reply(rdev, &wr_wait, 0, 0, __func__); + } + + kfree(cq->sw_queue); + dma_free_coherent(&(rdev->lldi.pdev->dev), + cq->memsize, cq->queue, + dma_unmap_addr(cq, mapping)); + c4iw_put_cqid(rdev, cq->cqid, uctx); + return ret; +} + +static int create_cq(struct c4iw_rdev *rdev, struct t4_cq *cq, + struct c4iw_dev_ucontext *uctx) +{ + struct fw_ri_res_wr *res_wr; + struct fw_ri_res *res; + int wr_len; + int user = (uctx != &rdev->uctx); + struct c4iw_wr_wait wr_wait; + int ret; + struct sk_buff *skb; + + cq->cqid = c4iw_get_cqid(rdev, uctx); + if (!cq->cqid) { + ret = -ENOMEM; + goto err1; + } + + if (!user) { + cq->sw_queue = kzalloc(cq->memsize, GFP_KERNEL); + if (!cq->sw_queue) { + ret = -ENOMEM; + goto err2; + } + } + cq->queue = dma_alloc_coherent(&rdev->lldi.pdev->dev, cq->memsize, + &cq->dma_addr, GFP_KERNEL); + if (!cq->queue) { + ret = -ENOMEM; + goto err3; + } + dma_unmap_addr_set(cq, mapping, cq->dma_addr); + memset(cq->queue, 0, cq->memsize); + + /* build fw_ri_res_wr */ + wr_len = sizeof *res_wr + sizeof *res; + + skb = alloc_skb(wr_len, GFP_KERNEL); + if (!skb) { + ret = -ENOMEM; + goto err4; + } + set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); + + res_wr = (struct fw_ri_res_wr *)__skb_put(skb, wr_len); + memset(res_wr, 0, wr_len); + res_wr->op_nres = cpu_to_be32( + FW_WR_OP_V(FW_RI_RES_WR) | + FW_RI_RES_WR_NRES_V(1) | + FW_WR_COMPL_F); + res_wr->len16_pkd = cpu_to_be32(DIV_ROUND_UP(wr_len, 16)); + res_wr->cookie = (uintptr_t)&wr_wait; + res = res_wr->res; + res->u.cq.restype = FW_RI_RES_TYPE_CQ; + res->u.cq.op = FW_RI_RES_OP_WRITE; + res->u.cq.iqid = cpu_to_be32(cq->cqid); + res->u.cq.iqandst_to_iqandstindex = cpu_to_be32( + FW_RI_RES_WR_IQANUS_V(0) | + FW_RI_RES_WR_IQANUD_V(1) | + FW_RI_RES_WR_IQANDST_F | + FW_RI_RES_WR_IQANDSTINDEX_V( + rdev->lldi.ciq_ids[cq->vector])); + res->u.cq.iqdroprss_to_iqesize = cpu_to_be16( + FW_RI_RES_WR_IQDROPRSS_F | + FW_RI_RES_WR_IQPCIECH_V(2) | + FW_RI_RES_WR_IQINTCNTTHRESH_V(0) | + FW_RI_RES_WR_IQO_F | + FW_RI_RES_WR_IQESIZE_V(1)); + res->u.cq.iqsize = cpu_to_be16(cq->size); + res->u.cq.iqaddr = cpu_to_be64(cq->dma_addr); + + c4iw_init_wr_wait(&wr_wait); + + ret = c4iw_ofld_send(rdev, skb); + if (ret) + goto err4; + PDBG("%s wait_event wr_wait %p\n", __func__, &wr_wait); + ret = c4iw_wait_for_reply(rdev, &wr_wait, 0, 0, __func__); + if (ret) + goto err4; + + cq->gen = 1; + cq->rdev = rdev; + if (user) { + u32 off = (cq->cqid << rdev->cqshift) & PAGE_MASK; + + cq->ugts = (u64)rdev->bar2_pa + off; + } else if (is_t4(rdev->lldi.adapter_type)) { + cq->gts = rdev->lldi.gts_reg; + cq->qid_mask = -1U; + } else { + u32 off = ((cq->cqid << rdev->cqshift) & PAGE_MASK) + 12; + + cq->gts = rdev->bar2_kva + off; + cq->qid_mask = rdev->qpmask; + } + return 0; +err4: + dma_free_coherent(&rdev->lldi.pdev->dev, cq->memsize, cq->queue, + dma_unmap_addr(cq, mapping)); +err3: + kfree(cq->sw_queue); +err2: + c4iw_put_cqid(rdev, cq->cqid, uctx); +err1: + return ret; +} + +static void insert_recv_cqe(struct t4_wq *wq, struct t4_cq *cq) +{ + struct t4_cqe cqe; + + PDBG("%s wq %p cq %p sw_cidx %u sw_pidx %u\n", __func__, + wq, cq, cq->sw_cidx, cq->sw_pidx); + memset(&cqe, 0, sizeof(cqe)); + cqe.header = cpu_to_be32(CQE_STATUS_V(T4_ERR_SWFLUSH) | + CQE_OPCODE_V(FW_RI_SEND) | + CQE_TYPE_V(0) | + CQE_SWCQE_V(1) | + CQE_QPID_V(wq->sq.qid)); + cqe.bits_type_ts = cpu_to_be64(CQE_GENBIT_V((u64)cq->gen)); + cq->sw_queue[cq->sw_pidx] = cqe; + t4_swcq_produce(cq); +} + +int c4iw_flush_rq(struct t4_wq *wq, struct t4_cq *cq, int count) +{ + int flushed = 0; + int in_use = wq->rq.in_use - count; + + BUG_ON(in_use < 0); + PDBG("%s wq %p cq %p rq.in_use %u skip count %u\n", __func__, + wq, cq, wq->rq.in_use, count); + while (in_use--) { + insert_recv_cqe(wq, cq); + flushed++; + } + return flushed; +} + +static void insert_sq_cqe(struct t4_wq *wq, struct t4_cq *cq, + struct t4_swsqe *swcqe) +{ + struct t4_cqe cqe; + + PDBG("%s wq %p cq %p sw_cidx %u sw_pidx %u\n", __func__, + wq, cq, cq->sw_cidx, cq->sw_pidx); + memset(&cqe, 0, sizeof(cqe)); + cqe.header = cpu_to_be32(CQE_STATUS_V(T4_ERR_SWFLUSH) | + CQE_OPCODE_V(swcqe->opcode) | + CQE_TYPE_V(1) | + CQE_SWCQE_V(1) | + CQE_QPID_V(wq->sq.qid)); + CQE_WRID_SQ_IDX(&cqe) = swcqe->idx; + cqe.bits_type_ts = cpu_to_be64(CQE_GENBIT_V((u64)cq->gen)); + cq->sw_queue[cq->sw_pidx] = cqe; + t4_swcq_produce(cq); +} + +static void advance_oldest_read(struct t4_wq *wq); + +int c4iw_flush_sq(struct c4iw_qp *qhp) +{ + int flushed = 0; + struct t4_wq *wq = &qhp->wq; + struct c4iw_cq *chp = to_c4iw_cq(qhp->ibqp.send_cq); + struct t4_cq *cq = &chp->cq; + int idx; + struct t4_swsqe *swsqe; + + if (wq->sq.flush_cidx == -1) + wq->sq.flush_cidx = wq->sq.cidx; + idx = wq->sq.flush_cidx; + BUG_ON(idx >= wq->sq.size); + while (idx != wq->sq.pidx) { + swsqe = &wq->sq.sw_sq[idx]; + BUG_ON(swsqe->flushed); + swsqe->flushed = 1; + insert_sq_cqe(wq, cq, swsqe); + if (wq->sq.oldest_read == swsqe) { + BUG_ON(swsqe->opcode != FW_RI_READ_REQ); + advance_oldest_read(wq); + } + flushed++; + if (++idx == wq->sq.size) + idx = 0; + } + wq->sq.flush_cidx += flushed; + if (wq->sq.flush_cidx >= wq->sq.size) + wq->sq.flush_cidx -= wq->sq.size; + return flushed; +} + +static void flush_completed_wrs(struct t4_wq *wq, struct t4_cq *cq) +{ + struct t4_swsqe *swsqe; + int cidx; + + if (wq->sq.flush_cidx == -1) + wq->sq.flush_cidx = wq->sq.cidx; + cidx = wq->sq.flush_cidx; + BUG_ON(cidx > wq->sq.size); + + while (cidx != wq->sq.pidx) { + swsqe = &wq->sq.sw_sq[cidx]; + if (!swsqe->signaled) { + if (++cidx == wq->sq.size) + cidx = 0; + } else if (swsqe->complete) { + + BUG_ON(swsqe->flushed); + + /* + * Insert this completed cqe into the swcq. + */ + PDBG("%s moving cqe into swcq sq idx %u cq idx %u\n", + __func__, cidx, cq->sw_pidx); + swsqe->cqe.header |= htonl(CQE_SWCQE_V(1)); + cq->sw_queue[cq->sw_pidx] = swsqe->cqe; + t4_swcq_produce(cq); + swsqe->flushed = 1; + if (++cidx == wq->sq.size) + cidx = 0; + wq->sq.flush_cidx = cidx; + } else + break; + } +} + +static void create_read_req_cqe(struct t4_wq *wq, struct t4_cqe *hw_cqe, + struct t4_cqe *read_cqe) +{ + read_cqe->u.scqe.cidx = wq->sq.oldest_read->idx; + read_cqe->len = htonl(wq->sq.oldest_read->read_len); + read_cqe->header = htonl(CQE_QPID_V(CQE_QPID(hw_cqe)) | + CQE_SWCQE_V(SW_CQE(hw_cqe)) | + CQE_OPCODE_V(FW_RI_READ_REQ) | + CQE_TYPE_V(1)); + read_cqe->bits_type_ts = hw_cqe->bits_type_ts; +} + +static void advance_oldest_read(struct t4_wq *wq) +{ + + u32 rptr = wq->sq.oldest_read - wq->sq.sw_sq + 1; + + if (rptr == wq->sq.size) + rptr = 0; + while (rptr != wq->sq.pidx) { + wq->sq.oldest_read = &wq->sq.sw_sq[rptr]; + + if (wq->sq.oldest_read->opcode == FW_RI_READ_REQ) + return; + if (++rptr == wq->sq.size) + rptr = 0; + } + wq->sq.oldest_read = NULL; +} + +/* + * Move all CQEs from the HWCQ into the SWCQ. + * Deal with out-of-order and/or completions that complete + * prior unsignalled WRs. + */ +void c4iw_flush_hw_cq(struct c4iw_cq *chp) +{ + struct t4_cqe *hw_cqe, *swcqe, read_cqe; + struct c4iw_qp *qhp; + struct t4_swsqe *swsqe; + int ret; + + PDBG("%s cqid 0x%x\n", __func__, chp->cq.cqid); + ret = t4_next_hw_cqe(&chp->cq, &hw_cqe); + + /* + * This logic is similar to poll_cq(), but not quite the same + * unfortunately. Need to move pertinent HW CQEs to the SW CQ but + * also do any translation magic that poll_cq() normally does. + */ + while (!ret) { + qhp = get_qhp(chp->rhp, CQE_QPID(hw_cqe)); + + /* + * drop CQEs with no associated QP + */ + if (qhp == NULL) + goto next_cqe; + + if (CQE_OPCODE(hw_cqe) == FW_RI_TERMINATE) + goto next_cqe; + + if (CQE_OPCODE(hw_cqe) == FW_RI_READ_RESP) { + + /* If we have reached here because of async + * event or other error, and have egress error + * then drop + */ + if (CQE_TYPE(hw_cqe) == 1) + goto next_cqe; + + /* drop peer2peer RTR reads. + */ + if (CQE_WRID_STAG(hw_cqe) == 1) + goto next_cqe; + + /* + * Eat completions for unsignaled read WRs. + */ + if (!qhp->wq.sq.oldest_read->signaled) { + advance_oldest_read(&qhp->wq); + goto next_cqe; + } + + /* + * Don't write to the HWCQ, create a new read req CQE + * in local memory and move it into the swcq. + */ + create_read_req_cqe(&qhp->wq, hw_cqe, &read_cqe); + hw_cqe = &read_cqe; + advance_oldest_read(&qhp->wq); + } + + /* if its a SQ completion, then do the magic to move all the + * unsignaled and now in-order completions into the swcq. + */ + if (SQ_TYPE(hw_cqe)) { + swsqe = &qhp->wq.sq.sw_sq[CQE_WRID_SQ_IDX(hw_cqe)]; + swsqe->cqe = *hw_cqe; + swsqe->complete = 1; + flush_completed_wrs(&qhp->wq, &chp->cq); + } else { + swcqe = &chp->cq.sw_queue[chp->cq.sw_pidx]; + *swcqe = *hw_cqe; + swcqe->header |= cpu_to_be32(CQE_SWCQE_V(1)); + t4_swcq_produce(&chp->cq); + } +next_cqe: + t4_hwcq_consume(&chp->cq); + ret = t4_next_hw_cqe(&chp->cq, &hw_cqe); + } +} + +static int cqe_completes_wr(struct t4_cqe *cqe, struct t4_wq *wq) +{ + if (CQE_OPCODE(cqe) == FW_RI_TERMINATE) + return 0; + + if ((CQE_OPCODE(cqe) == FW_RI_RDMA_WRITE) && RQ_TYPE(cqe)) + return 0; + + if ((CQE_OPCODE(cqe) == FW_RI_READ_RESP) && SQ_TYPE(cqe)) + return 0; + + if (CQE_SEND_OPCODE(cqe) && RQ_TYPE(cqe) && t4_rq_empty(wq)) + return 0; + return 1; +} + +void c4iw_count_rcqes(struct t4_cq *cq, struct t4_wq *wq, int *count) +{ + struct t4_cqe *cqe; + u32 ptr; + + *count = 0; + PDBG("%s count zero %d\n", __func__, *count); + ptr = cq->sw_cidx; + while (ptr != cq->sw_pidx) { + cqe = &cq->sw_queue[ptr]; + if (RQ_TYPE(cqe) && (CQE_OPCODE(cqe) != FW_RI_READ_RESP) && + (CQE_QPID(cqe) == wq->sq.qid) && cqe_completes_wr(cqe, wq)) + (*count)++; + if (++ptr == cq->size) + ptr = 0; + } + PDBG("%s cq %p count %d\n", __func__, cq, *count); +} + +/* + * poll_cq + * + * Caller must: + * check the validity of the first CQE, + * supply the wq assicated with the qpid. + * + * credit: cq credit to return to sge. + * cqe_flushed: 1 iff the CQE is flushed. + * cqe: copy of the polled CQE. + * + * return value: + * 0 CQE returned ok. + * -EAGAIN CQE skipped, try again. + * -EOVERFLOW CQ overflow detected. + */ +static int poll_cq(struct t4_wq *wq, struct t4_cq *cq, struct t4_cqe *cqe, + u8 *cqe_flushed, u64 *cookie, u32 *credit) +{ + int ret = 0; + struct t4_cqe *hw_cqe, read_cqe; + + *cqe_flushed = 0; + *credit = 0; + ret = t4_next_cqe(cq, &hw_cqe); + if (ret) + return ret; + + PDBG("%s CQE OVF %u qpid 0x%0x genbit %u type %u status 0x%0x" + " opcode 0x%0x len 0x%0x wrid_hi_stag 0x%x wrid_low_msn 0x%x\n", + __func__, CQE_OVFBIT(hw_cqe), CQE_QPID(hw_cqe), + CQE_GENBIT(hw_cqe), CQE_TYPE(hw_cqe), CQE_STATUS(hw_cqe), + CQE_OPCODE(hw_cqe), CQE_LEN(hw_cqe), CQE_WRID_HI(hw_cqe), + CQE_WRID_LOW(hw_cqe)); + + /* + * skip cqe's not affiliated with a QP. + */ + if (wq == NULL) { + ret = -EAGAIN; + goto skip_cqe; + } + + /* + * skip hw cqe's if the wq is flushed. + */ + if (wq->flushed && !SW_CQE(hw_cqe)) { + ret = -EAGAIN; + goto skip_cqe; + } + + /* + * skip TERMINATE cqes... + */ + if (CQE_OPCODE(hw_cqe) == FW_RI_TERMINATE) { + ret = -EAGAIN; + goto skip_cqe; + } + + /* + * Gotta tweak READ completions: + * 1) the cqe doesn't contain the sq_wptr from the wr. + * 2) opcode not reflected from the wr. + * 3) read_len not reflected from the wr. + * 4) cq_type is RQ_TYPE not SQ_TYPE. + */ + if (RQ_TYPE(hw_cqe) && (CQE_OPCODE(hw_cqe) == FW_RI_READ_RESP)) { + + /* If we have reached here because of async + * event or other error, and have egress error + * then drop + */ + if (CQE_TYPE(hw_cqe) == 1) { + if (CQE_STATUS(hw_cqe)) + t4_set_wq_in_error(wq); + ret = -EAGAIN; + goto skip_cqe; + } + + /* If this is an unsolicited read response, then the read + * was generated by the kernel driver as part of peer-2-peer + * connection setup. So ignore the completion. + */ + if (CQE_WRID_STAG(hw_cqe) == 1) { + if (CQE_STATUS(hw_cqe)) + t4_set_wq_in_error(wq); + ret = -EAGAIN; + goto skip_cqe; + } + + /* + * Eat completions for unsignaled read WRs. + */ + if (!wq->sq.oldest_read->signaled) { + advance_oldest_read(wq); + ret = -EAGAIN; + goto skip_cqe; + } + + /* + * Don't write to the HWCQ, so create a new read req CQE + * in local memory. + */ + create_read_req_cqe(wq, hw_cqe, &read_cqe); + hw_cqe = &read_cqe; + advance_oldest_read(wq); + } + + if (CQE_STATUS(hw_cqe) || t4_wq_in_error(wq)) { + *cqe_flushed = (CQE_STATUS(hw_cqe) == T4_ERR_SWFLUSH); + t4_set_wq_in_error(wq); + } + + /* + * RECV completion. + */ + if (RQ_TYPE(hw_cqe)) { + + /* + * HW only validates 4 bits of MSN. So we must validate that + * the MSN in the SEND is the next expected MSN. If its not, + * then we complete this with T4_ERR_MSN and mark the wq in + * error. + */ + + if (t4_rq_empty(wq)) { + t4_set_wq_in_error(wq); + ret = -EAGAIN; + goto skip_cqe; + } + if (unlikely((CQE_WRID_MSN(hw_cqe) != (wq->rq.msn)))) { + t4_set_wq_in_error(wq); + hw_cqe->header |= htonl(CQE_STATUS_V(T4_ERR_MSN)); + goto proc_cqe; + } + goto proc_cqe; + } + + /* + * If we get here its a send completion. + * + * Handle out of order completion. These get stuffed + * in the SW SQ. Then the SW SQ is walked to move any + * now in-order completions into the SW CQ. This handles + * 2 cases: + * 1) reaping unsignaled WRs when the first subsequent + * signaled WR is completed. + * 2) out of order read completions. + */ + if (!SW_CQE(hw_cqe) && (CQE_WRID_SQ_IDX(hw_cqe) != wq->sq.cidx)) { + struct t4_swsqe *swsqe; + + PDBG("%s out of order completion going in sw_sq at idx %u\n", + __func__, CQE_WRID_SQ_IDX(hw_cqe)); + swsqe = &wq->sq.sw_sq[CQE_WRID_SQ_IDX(hw_cqe)]; + swsqe->cqe = *hw_cqe; + swsqe->complete = 1; + ret = -EAGAIN; + goto flush_wq; + } + +proc_cqe: + *cqe = *hw_cqe; + + /* + * Reap the associated WR(s) that are freed up with this + * completion. + */ + if (SQ_TYPE(hw_cqe)) { + int idx = CQE_WRID_SQ_IDX(hw_cqe); + BUG_ON(idx >= wq->sq.size); + + /* + * Account for any unsignaled completions completed by + * this signaled completion. In this case, cidx points + * to the first unsignaled one, and idx points to the + * signaled one. So adjust in_use based on this delta. + * if this is not completing any unsigned wrs, then the + * delta will be 0. Handle wrapping also! + */ + if (idx < wq->sq.cidx) + wq->sq.in_use -= wq->sq.size + idx - wq->sq.cidx; + else + wq->sq.in_use -= idx - wq->sq.cidx; + BUG_ON(wq->sq.in_use <= 0 && wq->sq.in_use >= wq->sq.size); + + wq->sq.cidx = (uint16_t)idx; + PDBG("%s completing sq idx %u\n", __func__, wq->sq.cidx); + *cookie = wq->sq.sw_sq[wq->sq.cidx].wr_id; + if (c4iw_wr_log) + c4iw_log_wr_stats(wq, hw_cqe); + t4_sq_consume(wq); + } else { + PDBG("%s completing rq idx %u\n", __func__, wq->rq.cidx); + *cookie = wq->rq.sw_rq[wq->rq.cidx].wr_id; + BUG_ON(t4_rq_empty(wq)); + if (c4iw_wr_log) + c4iw_log_wr_stats(wq, hw_cqe); + t4_rq_consume(wq); + goto skip_cqe; + } + +flush_wq: + /* + * Flush any completed cqes that are now in-order. + */ + flush_completed_wrs(wq, cq); + +skip_cqe: + if (SW_CQE(hw_cqe)) { + PDBG("%s cq %p cqid 0x%x skip sw cqe cidx %u\n", + __func__, cq, cq->cqid, cq->sw_cidx); + t4_swcq_consume(cq); + } else { + PDBG("%s cq %p cqid 0x%x skip hw cqe cidx %u\n", + __func__, cq, cq->cqid, cq->cidx); + t4_hwcq_consume(cq); + } + return ret; +} + +/* + * Get one cq entry from c4iw and map it to openib. + * + * Returns: + * 0 cqe returned + * -ENODATA EMPTY; + * -EAGAIN caller must try again + * any other -errno fatal error + */ +static int c4iw_poll_cq_one(struct c4iw_cq *chp, struct ib_wc *wc) +{ + struct c4iw_qp *qhp = NULL; + struct t4_cqe uninitialized_var(cqe), *rd_cqe; + struct t4_wq *wq; + u32 credit = 0; + u8 cqe_flushed; + u64 cookie = 0; + int ret; + + ret = t4_next_cqe(&chp->cq, &rd_cqe); + + if (ret) + return ret; + + qhp = get_qhp(chp->rhp, CQE_QPID(rd_cqe)); + if (!qhp) + wq = NULL; + else { + spin_lock(&qhp->lock); + wq = &(qhp->wq); + } + ret = poll_cq(wq, &(chp->cq), &cqe, &cqe_flushed, &cookie, &credit); + if (ret) + goto out; + + wc->wr_id = cookie; + wc->qp = &qhp->ibqp; + wc->vendor_err = CQE_STATUS(&cqe); + wc->wc_flags = 0; + + PDBG("%s qpid 0x%x type %d opcode %d status 0x%x len %u wrid hi 0x%x " + "lo 0x%x cookie 0x%llx\n", __func__, CQE_QPID(&cqe), + CQE_TYPE(&cqe), CQE_OPCODE(&cqe), CQE_STATUS(&cqe), CQE_LEN(&cqe), + CQE_WRID_HI(&cqe), CQE_WRID_LOW(&cqe), (unsigned long long)cookie); + + if (CQE_TYPE(&cqe) == 0) { + if (!CQE_STATUS(&cqe)) + wc->byte_len = CQE_LEN(&cqe); + else + wc->byte_len = 0; + wc->opcode = IB_WC_RECV; + if (CQE_OPCODE(&cqe) == FW_RI_SEND_WITH_INV || + CQE_OPCODE(&cqe) == FW_RI_SEND_WITH_SE_INV) { + wc->ex.invalidate_rkey = CQE_WRID_STAG(&cqe); + wc->wc_flags |= IB_WC_WITH_INVALIDATE; + } + } else { + switch (CQE_OPCODE(&cqe)) { + case FW_RI_RDMA_WRITE: + wc->opcode = IB_WC_RDMA_WRITE; + break; + case FW_RI_READ_REQ: + wc->opcode = IB_WC_RDMA_READ; + wc->byte_len = CQE_LEN(&cqe); + break; + case FW_RI_SEND_WITH_INV: + case FW_RI_SEND_WITH_SE_INV: + wc->opcode = IB_WC_SEND; + wc->wc_flags |= IB_WC_WITH_INVALIDATE; + break; + case FW_RI_SEND: + case FW_RI_SEND_WITH_SE: + wc->opcode = IB_WC_SEND; + break; + case FW_RI_BIND_MW: + wc->opcode = IB_WC_BIND_MW; + break; + + case FW_RI_LOCAL_INV: + wc->opcode = IB_WC_LOCAL_INV; + break; + case FW_RI_FAST_REGISTER: + wc->opcode = IB_WC_FAST_REG_MR; + break; + default: + printk(KERN_ERR MOD "Unexpected opcode %d " + "in the CQE received for QPID=0x%0x\n", + CQE_OPCODE(&cqe), CQE_QPID(&cqe)); + ret = -EINVAL; + goto out; + } + } + + if (cqe_flushed) + wc->status = IB_WC_WR_FLUSH_ERR; + else { + + switch (CQE_STATUS(&cqe)) { + case T4_ERR_SUCCESS: + wc->status = IB_WC_SUCCESS; + break; + case T4_ERR_STAG: + wc->status = IB_WC_LOC_ACCESS_ERR; + break; + case T4_ERR_PDID: + wc->status = IB_WC_LOC_PROT_ERR; + break; + case T4_ERR_QPID: + case T4_ERR_ACCESS: + wc->status = IB_WC_LOC_ACCESS_ERR; + break; + case T4_ERR_WRAP: + wc->status = IB_WC_GENERAL_ERR; + break; + case T4_ERR_BOUND: + wc->status = IB_WC_LOC_LEN_ERR; + break; + case T4_ERR_INVALIDATE_SHARED_MR: + case T4_ERR_INVALIDATE_MR_WITH_MW_BOUND: + wc->status = IB_WC_MW_BIND_ERR; + break; + case T4_ERR_CRC: + case T4_ERR_MARKER: + case T4_ERR_PDU_LEN_ERR: + case T4_ERR_OUT_OF_RQE: + case T4_ERR_DDP_VERSION: + case T4_ERR_RDMA_VERSION: + case T4_ERR_DDP_QUEUE_NUM: + case T4_ERR_MSN: + case T4_ERR_TBIT: + case T4_ERR_MO: + case T4_ERR_MSN_RANGE: + case T4_ERR_IRD_OVERFLOW: + case T4_ERR_OPCODE: + case T4_ERR_INTERNAL_ERR: + wc->status = IB_WC_FATAL_ERR; + break; + case T4_ERR_SWFLUSH: + wc->status = IB_WC_WR_FLUSH_ERR; + break; + default: + printk(KERN_ERR MOD + "Unexpected cqe_status 0x%x for QPID=0x%0x\n", + CQE_STATUS(&cqe), CQE_QPID(&cqe)); + ret = -EINVAL; + } + } +out: + if (wq) + spin_unlock(&qhp->lock); + return ret; +} + +int c4iw_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc) +{ + struct c4iw_cq *chp; + unsigned long flags; + int npolled; + int err = 0; + + chp = to_c4iw_cq(ibcq); + + spin_lock_irqsave(&chp->lock, flags); + for (npolled = 0; npolled < num_entries; ++npolled) { + do { + err = c4iw_poll_cq_one(chp, wc + npolled); + } while (err == -EAGAIN); + if (err) + break; + } + spin_unlock_irqrestore(&chp->lock, flags); + return !err || err == -ENODATA ? npolled : err; +} + +int c4iw_destroy_cq(struct ib_cq *ib_cq) +{ + struct c4iw_cq *chp; + struct c4iw_ucontext *ucontext; + + PDBG("%s ib_cq %p\n", __func__, ib_cq); + chp = to_c4iw_cq(ib_cq); + + remove_handle(chp->rhp, &chp->rhp->cqidr, chp->cq.cqid); + atomic_dec(&chp->refcnt); + wait_event(chp->wait, !atomic_read(&chp->refcnt)); + + ucontext = ib_cq->uobject ? to_c4iw_ucontext(ib_cq->uobject->context) + : NULL; + destroy_cq(&chp->rhp->rdev, &chp->cq, + ucontext ? &ucontext->uctx : &chp->cq.rdev->uctx); + kfree(chp); + return 0; +} + +struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, int entries, + int vector, struct ib_ucontext *ib_context, + struct ib_udata *udata) +{ + struct c4iw_dev *rhp; + struct c4iw_cq *chp; + struct c4iw_create_cq_resp uresp; + struct c4iw_ucontext *ucontext = NULL; + int ret; + size_t memsize, hwentries; + struct c4iw_mm_entry *mm, *mm2; + + PDBG("%s ib_dev %p entries %d\n", __func__, ibdev, entries); + + rhp = to_c4iw_dev(ibdev); + + if (vector >= rhp->rdev.lldi.nciq) + return ERR_PTR(-EINVAL); + + chp = kzalloc(sizeof(*chp), GFP_KERNEL); + if (!chp) + return ERR_PTR(-ENOMEM); + + if (ib_context) + ucontext = to_c4iw_ucontext(ib_context); + + /* account for the status page. */ + entries++; + + /* IQ needs one extra entry to differentiate full vs empty. */ + entries++; + + /* + * entries must be multiple of 16 for HW. + */ + entries = roundup(entries, 16); + + /* + * Make actual HW queue 2x to avoid cdix_inc overflows. + */ + hwentries = min(entries * 2, rhp->rdev.hw_queue.t4_max_iq_size); + + /* + * Make HW queue at least 64 entries so GTS updates aren't too + * frequent. + */ + if (hwentries < 64) + hwentries = 64; + + memsize = hwentries * sizeof *chp->cq.queue; + + /* + * memsize must be a multiple of the page size if its a user cq. + */ + if (ucontext) + memsize = roundup(memsize, PAGE_SIZE); + chp->cq.size = hwentries; + chp->cq.memsize = memsize; + chp->cq.vector = vector; + + ret = create_cq(&rhp->rdev, &chp->cq, + ucontext ? &ucontext->uctx : &rhp->rdev.uctx); + if (ret) + goto err1; + + chp->rhp = rhp; + chp->cq.size--; /* status page */ + chp->ibcq.cqe = entries - 2; + spin_lock_init(&chp->lock); + spin_lock_init(&chp->comp_handler_lock); + atomic_set(&chp->refcnt, 1); + init_waitqueue_head(&chp->wait); + ret = insert_handle(rhp, &rhp->cqidr, chp, chp->cq.cqid); + if (ret) + goto err2; + + if (ucontext) { + mm = kmalloc(sizeof *mm, GFP_KERNEL); + if (!mm) + goto err3; + mm2 = kmalloc(sizeof *mm2, GFP_KERNEL); + if (!mm2) + goto err4; + + uresp.qid_mask = rhp->rdev.cqmask; + uresp.cqid = chp->cq.cqid; + uresp.size = chp->cq.size; + uresp.memsize = chp->cq.memsize; + spin_lock(&ucontext->mmap_lock); + uresp.key = ucontext->key; + ucontext->key += PAGE_SIZE; + uresp.gts_key = ucontext->key; + ucontext->key += PAGE_SIZE; + spin_unlock(&ucontext->mmap_lock); + ret = ib_copy_to_udata(udata, &uresp, + sizeof(uresp) - sizeof(uresp.reserved)); + if (ret) + goto err5; + + mm->key = uresp.key; + mm->addr = virt_to_phys(chp->cq.queue); + mm->len = chp->cq.memsize; + insert_mmap(ucontext, mm); + + mm2->key = uresp.gts_key; + mm2->addr = chp->cq.ugts; + mm2->len = PAGE_SIZE; + insert_mmap(ucontext, mm2); + } + PDBG("%s cqid 0x%0x chp %p size %u memsize %zu, dma_addr 0x%0llx\n", + __func__, chp->cq.cqid, chp, chp->cq.size, + chp->cq.memsize, (unsigned long long) chp->cq.dma_addr); + return &chp->ibcq; +err5: + kfree(mm2); +err4: + kfree(mm); +err3: + remove_handle(rhp, &rhp->cqidr, chp->cq.cqid); +err2: + destroy_cq(&chp->rhp->rdev, &chp->cq, + ucontext ? &ucontext->uctx : &rhp->rdev.uctx); +err1: + kfree(chp); + return ERR_PTR(ret); +} + +int c4iw_resize_cq(struct ib_cq *cq, int cqe, struct ib_udata *udata) +{ + return -ENOSYS; +} + +int c4iw_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags) +{ + struct c4iw_cq *chp; + int ret; + unsigned long flag; + + chp = to_c4iw_cq(ibcq); + spin_lock_irqsave(&chp->lock, flag); + ret = t4_arm_cq(&chp->cq, + (flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED); + spin_unlock_irqrestore(&chp->lock, flag); + if (ret && !(flags & IB_CQ_REPORT_MISSED_EVENTS)) + ret = 0; + return ret; +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/device.c b/kernel/drivers/infiniband/hw/cxgb4/device.c new file mode 100644 index 000000000..7e895d714 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/device.c @@ -0,0 +1,1564 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/debugfs.h> +#include <linux/vmalloc.h> +#include <linux/math64.h> + +#include <rdma/ib_verbs.h> + +#include "iw_cxgb4.h" + +#define DRV_VERSION "0.1" + +MODULE_AUTHOR("Steve Wise"); +MODULE_DESCRIPTION("Chelsio T4/T5 RDMA Driver"); +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_VERSION(DRV_VERSION); + +static int allow_db_fc_on_t5; +module_param(allow_db_fc_on_t5, int, 0644); +MODULE_PARM_DESC(allow_db_fc_on_t5, + "Allow DB Flow Control on T5 (default = 0)"); + +static int allow_db_coalescing_on_t5; +module_param(allow_db_coalescing_on_t5, int, 0644); +MODULE_PARM_DESC(allow_db_coalescing_on_t5, + "Allow DB Coalescing on T5 (default = 0)"); + +int c4iw_wr_log = 0; +module_param(c4iw_wr_log, int, 0444); +MODULE_PARM_DESC(c4iw_wr_log, "Enables logging of work request timing data."); + +static int c4iw_wr_log_size_order = 12; +module_param(c4iw_wr_log_size_order, int, 0444); +MODULE_PARM_DESC(c4iw_wr_log_size_order, + "Number of entries (log2) in the work request timing log."); + +struct uld_ctx { + struct list_head entry; + struct cxgb4_lld_info lldi; + struct c4iw_dev *dev; +}; + +static LIST_HEAD(uld_ctx_list); +static DEFINE_MUTEX(dev_mutex); + +#define DB_FC_RESUME_SIZE 64 +#define DB_FC_RESUME_DELAY 1 +#define DB_FC_DRAIN_THRESH 0 + +static struct dentry *c4iw_debugfs_root; + +struct c4iw_debugfs_data { + struct c4iw_dev *devp; + char *buf; + int bufsize; + int pos; +}; + +/* registered cxgb4 netlink callbacks */ +static struct ibnl_client_cbs c4iw_nl_cb_table[] = { + [RDMA_NL_IWPM_REG_PID] = {.dump = iwpm_register_pid_cb}, + [RDMA_NL_IWPM_ADD_MAPPING] = {.dump = iwpm_add_mapping_cb}, + [RDMA_NL_IWPM_QUERY_MAPPING] = {.dump = iwpm_add_and_query_mapping_cb}, + [RDMA_NL_IWPM_HANDLE_ERR] = {.dump = iwpm_mapping_error_cb}, + [RDMA_NL_IWPM_REMOTE_INFO] = {.dump = iwpm_remote_info_cb}, + [RDMA_NL_IWPM_MAPINFO] = {.dump = iwpm_mapping_info_cb}, + [RDMA_NL_IWPM_MAPINFO_NUM] = {.dump = iwpm_ack_mapping_info_cb} +}; + +static int count_idrs(int id, void *p, void *data) +{ + int *countp = data; + + *countp = *countp + 1; + return 0; +} + +static ssize_t debugfs_read(struct file *file, char __user *buf, size_t count, + loff_t *ppos) +{ + struct c4iw_debugfs_data *d = file->private_data; + + return simple_read_from_buffer(buf, count, ppos, d->buf, d->pos); +} + +void c4iw_log_wr_stats(struct t4_wq *wq, struct t4_cqe *cqe) +{ + struct wr_log_entry le; + int idx; + + if (!wq->rdev->wr_log) + return; + + idx = (atomic_inc_return(&wq->rdev->wr_log_idx) - 1) & + (wq->rdev->wr_log_size - 1); + le.poll_sge_ts = cxgb4_read_sge_timestamp(wq->rdev->lldi.ports[0]); + getnstimeofday(&le.poll_host_ts); + le.valid = 1; + le.cqe_sge_ts = CQE_TS(cqe); + if (SQ_TYPE(cqe)) { + le.qid = wq->sq.qid; + le.opcode = CQE_OPCODE(cqe); + le.post_host_ts = wq->sq.sw_sq[wq->sq.cidx].host_ts; + le.post_sge_ts = wq->sq.sw_sq[wq->sq.cidx].sge_ts; + le.wr_id = CQE_WRID_SQ_IDX(cqe); + } else { + le.qid = wq->rq.qid; + le.opcode = FW_RI_RECEIVE; + le.post_host_ts = wq->rq.sw_rq[wq->rq.cidx].host_ts; + le.post_sge_ts = wq->rq.sw_rq[wq->rq.cidx].sge_ts; + le.wr_id = CQE_WRID_MSN(cqe); + } + wq->rdev->wr_log[idx] = le; +} + +static int wr_log_show(struct seq_file *seq, void *v) +{ + struct c4iw_dev *dev = seq->private; + struct timespec prev_ts = {0, 0}; + struct wr_log_entry *lep; + int prev_ts_set = 0; + int idx, end; + +#define ts2ns(ts) div64_u64((ts) * dev->rdev.lldi.cclk_ps, 1000) + + idx = atomic_read(&dev->rdev.wr_log_idx) & + (dev->rdev.wr_log_size - 1); + end = idx - 1; + if (end < 0) + end = dev->rdev.wr_log_size - 1; + lep = &dev->rdev.wr_log[idx]; + while (idx != end) { + if (lep->valid) { + if (!prev_ts_set) { + prev_ts_set = 1; + prev_ts = lep->poll_host_ts; + } + seq_printf(seq, "%04u: sec %lu nsec %lu qid %u opcode " + "%u %s 0x%x host_wr_delta sec %lu nsec %lu " + "post_sge_ts 0x%llx cqe_sge_ts 0x%llx " + "poll_sge_ts 0x%llx post_poll_delta_ns %llu " + "cqe_poll_delta_ns %llu\n", + idx, + timespec_sub(lep->poll_host_ts, + prev_ts).tv_sec, + timespec_sub(lep->poll_host_ts, + prev_ts).tv_nsec, + lep->qid, lep->opcode, + lep->opcode == FW_RI_RECEIVE ? + "msn" : "wrid", + lep->wr_id, + timespec_sub(lep->poll_host_ts, + lep->post_host_ts).tv_sec, + timespec_sub(lep->poll_host_ts, + lep->post_host_ts).tv_nsec, + lep->post_sge_ts, lep->cqe_sge_ts, + lep->poll_sge_ts, + ts2ns(lep->poll_sge_ts - lep->post_sge_ts), + ts2ns(lep->poll_sge_ts - lep->cqe_sge_ts)); + prev_ts = lep->poll_host_ts; + } + idx++; + if (idx > (dev->rdev.wr_log_size - 1)) + idx = 0; + lep = &dev->rdev.wr_log[idx]; + } +#undef ts2ns + return 0; +} + +static int wr_log_open(struct inode *inode, struct file *file) +{ + return single_open(file, wr_log_show, inode->i_private); +} + +static ssize_t wr_log_clear(struct file *file, const char __user *buf, + size_t count, loff_t *pos) +{ + struct c4iw_dev *dev = ((struct seq_file *)file->private_data)->private; + int i; + + if (dev->rdev.wr_log) + for (i = 0; i < dev->rdev.wr_log_size; i++) + dev->rdev.wr_log[i].valid = 0; + return count; +} + +static const struct file_operations wr_log_debugfs_fops = { + .owner = THIS_MODULE, + .open = wr_log_open, + .release = single_release, + .read = seq_read, + .llseek = seq_lseek, + .write = wr_log_clear, +}; + +static int dump_qp(int id, void *p, void *data) +{ + struct c4iw_qp *qp = p; + struct c4iw_debugfs_data *qpd = data; + int space; + int cc; + + if (id != qp->wq.sq.qid) + return 0; + + space = qpd->bufsize - qpd->pos - 1; + if (space == 0) + return 1; + + if (qp->ep) { + if (qp->ep->com.local_addr.ss_family == AF_INET) { + struct sockaddr_in *lsin = (struct sockaddr_in *) + &qp->ep->com.local_addr; + struct sockaddr_in *rsin = (struct sockaddr_in *) + &qp->ep->com.remote_addr; + struct sockaddr_in *mapped_lsin = (struct sockaddr_in *) + &qp->ep->com.mapped_local_addr; + struct sockaddr_in *mapped_rsin = (struct sockaddr_in *) + &qp->ep->com.mapped_remote_addr; + + cc = snprintf(qpd->buf + qpd->pos, space, + "rc qp sq id %u rq id %u state %u " + "onchip %u ep tid %u state %u " + "%pI4:%u/%u->%pI4:%u/%u\n", + qp->wq.sq.qid, qp->wq.rq.qid, + (int)qp->attr.state, + qp->wq.sq.flags & T4_SQ_ONCHIP, + qp->ep->hwtid, (int)qp->ep->com.state, + &lsin->sin_addr, ntohs(lsin->sin_port), + ntohs(mapped_lsin->sin_port), + &rsin->sin_addr, ntohs(rsin->sin_port), + ntohs(mapped_rsin->sin_port)); + } else { + struct sockaddr_in6 *lsin6 = (struct sockaddr_in6 *) + &qp->ep->com.local_addr; + struct sockaddr_in6 *rsin6 = (struct sockaddr_in6 *) + &qp->ep->com.remote_addr; + struct sockaddr_in6 *mapped_lsin6 = + (struct sockaddr_in6 *) + &qp->ep->com.mapped_local_addr; + struct sockaddr_in6 *mapped_rsin6 = + (struct sockaddr_in6 *) + &qp->ep->com.mapped_remote_addr; + + cc = snprintf(qpd->buf + qpd->pos, space, + "rc qp sq id %u rq id %u state %u " + "onchip %u ep tid %u state %u " + "%pI6:%u/%u->%pI6:%u/%u\n", + qp->wq.sq.qid, qp->wq.rq.qid, + (int)qp->attr.state, + qp->wq.sq.flags & T4_SQ_ONCHIP, + qp->ep->hwtid, (int)qp->ep->com.state, + &lsin6->sin6_addr, + ntohs(lsin6->sin6_port), + ntohs(mapped_lsin6->sin6_port), + &rsin6->sin6_addr, + ntohs(rsin6->sin6_port), + ntohs(mapped_rsin6->sin6_port)); + } + } else + cc = snprintf(qpd->buf + qpd->pos, space, + "qp sq id %u rq id %u state %u onchip %u\n", + qp->wq.sq.qid, qp->wq.rq.qid, + (int)qp->attr.state, + qp->wq.sq.flags & T4_SQ_ONCHIP); + if (cc < space) + qpd->pos += cc; + return 0; +} + +static int qp_release(struct inode *inode, struct file *file) +{ + struct c4iw_debugfs_data *qpd = file->private_data; + if (!qpd) { + printk(KERN_INFO "%s null qpd?\n", __func__); + return 0; + } + vfree(qpd->buf); + kfree(qpd); + return 0; +} + +static int qp_open(struct inode *inode, struct file *file) +{ + struct c4iw_debugfs_data *qpd; + int ret = 0; + int count = 1; + + qpd = kmalloc(sizeof *qpd, GFP_KERNEL); + if (!qpd) { + ret = -ENOMEM; + goto out; + } + qpd->devp = inode->i_private; + qpd->pos = 0; + + spin_lock_irq(&qpd->devp->lock); + idr_for_each(&qpd->devp->qpidr, count_idrs, &count); + spin_unlock_irq(&qpd->devp->lock); + + qpd->bufsize = count * 128; + qpd->buf = vmalloc(qpd->bufsize); + if (!qpd->buf) { + ret = -ENOMEM; + goto err1; + } + + spin_lock_irq(&qpd->devp->lock); + idr_for_each(&qpd->devp->qpidr, dump_qp, qpd); + spin_unlock_irq(&qpd->devp->lock); + + qpd->buf[qpd->pos++] = 0; + file->private_data = qpd; + goto out; +err1: + kfree(qpd); +out: + return ret; +} + +static const struct file_operations qp_debugfs_fops = { + .owner = THIS_MODULE, + .open = qp_open, + .release = qp_release, + .read = debugfs_read, + .llseek = default_llseek, +}; + +static int dump_stag(int id, void *p, void *data) +{ + struct c4iw_debugfs_data *stagd = data; + int space; + int cc; + struct fw_ri_tpte tpte; + int ret; + + space = stagd->bufsize - stagd->pos - 1; + if (space == 0) + return 1; + + ret = cxgb4_read_tpte(stagd->devp->rdev.lldi.ports[0], (u32)id<<8, + (__be32 *)&tpte); + if (ret) { + dev_err(&stagd->devp->rdev.lldi.pdev->dev, + "%s cxgb4_read_tpte err %d\n", __func__, ret); + return ret; + } + cc = snprintf(stagd->buf + stagd->pos, space, + "stag: idx 0x%x valid %d key 0x%x state %d pdid %d " + "perm 0x%x ps %d len 0x%llx va 0x%llx\n", + (u32)id<<8, + FW_RI_TPTE_VALID_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_STAGKEY_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_STAGSTATE_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_PDID_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_PERM_G(ntohl(tpte.locread_to_qpid)), + FW_RI_TPTE_PS_G(ntohl(tpte.locread_to_qpid)), + ((u64)ntohl(tpte.len_hi) << 32) | ntohl(tpte.len_lo), + ((u64)ntohl(tpte.va_hi) << 32) | ntohl(tpte.va_lo_fbo)); + if (cc < space) + stagd->pos += cc; + return 0; +} + +static int stag_release(struct inode *inode, struct file *file) +{ + struct c4iw_debugfs_data *stagd = file->private_data; + if (!stagd) { + printk(KERN_INFO "%s null stagd?\n", __func__); + return 0; + } + vfree(stagd->buf); + kfree(stagd); + return 0; +} + +static int stag_open(struct inode *inode, struct file *file) +{ + struct c4iw_debugfs_data *stagd; + int ret = 0; + int count = 1; + + stagd = kmalloc(sizeof *stagd, GFP_KERNEL); + if (!stagd) { + ret = -ENOMEM; + goto out; + } + stagd->devp = inode->i_private; + stagd->pos = 0; + + spin_lock_irq(&stagd->devp->lock); + idr_for_each(&stagd->devp->mmidr, count_idrs, &count); + spin_unlock_irq(&stagd->devp->lock); + + stagd->bufsize = count * 256; + stagd->buf = vmalloc(stagd->bufsize); + if (!stagd->buf) { + ret = -ENOMEM; + goto err1; + } + + spin_lock_irq(&stagd->devp->lock); + idr_for_each(&stagd->devp->mmidr, dump_stag, stagd); + spin_unlock_irq(&stagd->devp->lock); + + stagd->buf[stagd->pos++] = 0; + file->private_data = stagd; + goto out; +err1: + kfree(stagd); +out: + return ret; +} + +static const struct file_operations stag_debugfs_fops = { + .owner = THIS_MODULE, + .open = stag_open, + .release = stag_release, + .read = debugfs_read, + .llseek = default_llseek, +}; + +static char *db_state_str[] = {"NORMAL", "FLOW_CONTROL", "RECOVERY", "STOPPED"}; + +static int stats_show(struct seq_file *seq, void *v) +{ + struct c4iw_dev *dev = seq->private; + + seq_printf(seq, " Object: %10s %10s %10s %10s\n", "Total", "Current", + "Max", "Fail"); + seq_printf(seq, " PDID: %10llu %10llu %10llu %10llu\n", + dev->rdev.stats.pd.total, dev->rdev.stats.pd.cur, + dev->rdev.stats.pd.max, dev->rdev.stats.pd.fail); + seq_printf(seq, " QID: %10llu %10llu %10llu %10llu\n", + dev->rdev.stats.qid.total, dev->rdev.stats.qid.cur, + dev->rdev.stats.qid.max, dev->rdev.stats.qid.fail); + seq_printf(seq, " TPTMEM: %10llu %10llu %10llu %10llu\n", + dev->rdev.stats.stag.total, dev->rdev.stats.stag.cur, + dev->rdev.stats.stag.max, dev->rdev.stats.stag.fail); + seq_printf(seq, " PBLMEM: %10llu %10llu %10llu %10llu\n", + dev->rdev.stats.pbl.total, dev->rdev.stats.pbl.cur, + dev->rdev.stats.pbl.max, dev->rdev.stats.pbl.fail); + seq_printf(seq, " RQTMEM: %10llu %10llu %10llu %10llu\n", + dev->rdev.stats.rqt.total, dev->rdev.stats.rqt.cur, + dev->rdev.stats.rqt.max, dev->rdev.stats.rqt.fail); + seq_printf(seq, " OCQPMEM: %10llu %10llu %10llu %10llu\n", + dev->rdev.stats.ocqp.total, dev->rdev.stats.ocqp.cur, + dev->rdev.stats.ocqp.max, dev->rdev.stats.ocqp.fail); + seq_printf(seq, " DB FULL: %10llu\n", dev->rdev.stats.db_full); + seq_printf(seq, " DB EMPTY: %10llu\n", dev->rdev.stats.db_empty); + seq_printf(seq, " DB DROP: %10llu\n", dev->rdev.stats.db_drop); + seq_printf(seq, " DB State: %s Transitions %llu FC Interruptions %llu\n", + db_state_str[dev->db_state], + dev->rdev.stats.db_state_transitions, + dev->rdev.stats.db_fc_interruptions); + seq_printf(seq, "TCAM_FULL: %10llu\n", dev->rdev.stats.tcam_full); + seq_printf(seq, "ACT_OFLD_CONN_FAILS: %10llu\n", + dev->rdev.stats.act_ofld_conn_fails); + seq_printf(seq, "PAS_OFLD_CONN_FAILS: %10llu\n", + dev->rdev.stats.pas_ofld_conn_fails); + seq_printf(seq, "NEG_ADV_RCVD: %10llu\n", dev->rdev.stats.neg_adv); + seq_printf(seq, "AVAILABLE IRD: %10u\n", dev->avail_ird); + return 0; +} + +static int stats_open(struct inode *inode, struct file *file) +{ + return single_open(file, stats_show, inode->i_private); +} + +static ssize_t stats_clear(struct file *file, const char __user *buf, + size_t count, loff_t *pos) +{ + struct c4iw_dev *dev = ((struct seq_file *)file->private_data)->private; + + mutex_lock(&dev->rdev.stats.lock); + dev->rdev.stats.pd.max = 0; + dev->rdev.stats.pd.fail = 0; + dev->rdev.stats.qid.max = 0; + dev->rdev.stats.qid.fail = 0; + dev->rdev.stats.stag.max = 0; + dev->rdev.stats.stag.fail = 0; + dev->rdev.stats.pbl.max = 0; + dev->rdev.stats.pbl.fail = 0; + dev->rdev.stats.rqt.max = 0; + dev->rdev.stats.rqt.fail = 0; + dev->rdev.stats.ocqp.max = 0; + dev->rdev.stats.ocqp.fail = 0; + dev->rdev.stats.db_full = 0; + dev->rdev.stats.db_empty = 0; + dev->rdev.stats.db_drop = 0; + dev->rdev.stats.db_state_transitions = 0; + dev->rdev.stats.tcam_full = 0; + dev->rdev.stats.act_ofld_conn_fails = 0; + dev->rdev.stats.pas_ofld_conn_fails = 0; + mutex_unlock(&dev->rdev.stats.lock); + return count; +} + +static const struct file_operations stats_debugfs_fops = { + .owner = THIS_MODULE, + .open = stats_open, + .release = single_release, + .read = seq_read, + .llseek = seq_lseek, + .write = stats_clear, +}; + +static int dump_ep(int id, void *p, void *data) +{ + struct c4iw_ep *ep = p; + struct c4iw_debugfs_data *epd = data; + int space; + int cc; + + space = epd->bufsize - epd->pos - 1; + if (space == 0) + return 1; + + if (ep->com.local_addr.ss_family == AF_INET) { + struct sockaddr_in *lsin = (struct sockaddr_in *) + &ep->com.local_addr; + struct sockaddr_in *rsin = (struct sockaddr_in *) + &ep->com.remote_addr; + struct sockaddr_in *mapped_lsin = (struct sockaddr_in *) + &ep->com.mapped_local_addr; + struct sockaddr_in *mapped_rsin = (struct sockaddr_in *) + &ep->com.mapped_remote_addr; + + cc = snprintf(epd->buf + epd->pos, space, + "ep %p cm_id %p qp %p state %d flags 0x%lx " + "history 0x%lx hwtid %d atid %d " + "conn_na %u abort_na %u " + "%pI4:%d/%d <-> %pI4:%d/%d\n", + ep, ep->com.cm_id, ep->com.qp, + (int)ep->com.state, ep->com.flags, + ep->com.history, ep->hwtid, ep->atid, + ep->stats.connect_neg_adv, + ep->stats.abort_neg_adv, + &lsin->sin_addr, ntohs(lsin->sin_port), + ntohs(mapped_lsin->sin_port), + &rsin->sin_addr, ntohs(rsin->sin_port), + ntohs(mapped_rsin->sin_port)); + } else { + struct sockaddr_in6 *lsin6 = (struct sockaddr_in6 *) + &ep->com.local_addr; + struct sockaddr_in6 *rsin6 = (struct sockaddr_in6 *) + &ep->com.remote_addr; + struct sockaddr_in6 *mapped_lsin6 = (struct sockaddr_in6 *) + &ep->com.mapped_local_addr; + struct sockaddr_in6 *mapped_rsin6 = (struct sockaddr_in6 *) + &ep->com.mapped_remote_addr; + + cc = snprintf(epd->buf + epd->pos, space, + "ep %p cm_id %p qp %p state %d flags 0x%lx " + "history 0x%lx hwtid %d atid %d " + "conn_na %u abort_na %u " + "%pI6:%d/%d <-> %pI6:%d/%d\n", + ep, ep->com.cm_id, ep->com.qp, + (int)ep->com.state, ep->com.flags, + ep->com.history, ep->hwtid, ep->atid, + ep->stats.connect_neg_adv, + ep->stats.abort_neg_adv, + &lsin6->sin6_addr, ntohs(lsin6->sin6_port), + ntohs(mapped_lsin6->sin6_port), + &rsin6->sin6_addr, ntohs(rsin6->sin6_port), + ntohs(mapped_rsin6->sin6_port)); + } + if (cc < space) + epd->pos += cc; + return 0; +} + +static int dump_listen_ep(int id, void *p, void *data) +{ + struct c4iw_listen_ep *ep = p; + struct c4iw_debugfs_data *epd = data; + int space; + int cc; + + space = epd->bufsize - epd->pos - 1; + if (space == 0) + return 1; + + if (ep->com.local_addr.ss_family == AF_INET) { + struct sockaddr_in *lsin = (struct sockaddr_in *) + &ep->com.local_addr; + struct sockaddr_in *mapped_lsin = (struct sockaddr_in *) + &ep->com.mapped_local_addr; + + cc = snprintf(epd->buf + epd->pos, space, + "ep %p cm_id %p state %d flags 0x%lx stid %d " + "backlog %d %pI4:%d/%d\n", + ep, ep->com.cm_id, (int)ep->com.state, + ep->com.flags, ep->stid, ep->backlog, + &lsin->sin_addr, ntohs(lsin->sin_port), + ntohs(mapped_lsin->sin_port)); + } else { + struct sockaddr_in6 *lsin6 = (struct sockaddr_in6 *) + &ep->com.local_addr; + struct sockaddr_in6 *mapped_lsin6 = (struct sockaddr_in6 *) + &ep->com.mapped_local_addr; + + cc = snprintf(epd->buf + epd->pos, space, + "ep %p cm_id %p state %d flags 0x%lx stid %d " + "backlog %d %pI6:%d/%d\n", + ep, ep->com.cm_id, (int)ep->com.state, + ep->com.flags, ep->stid, ep->backlog, + &lsin6->sin6_addr, ntohs(lsin6->sin6_port), + ntohs(mapped_lsin6->sin6_port)); + } + if (cc < space) + epd->pos += cc; + return 0; +} + +static int ep_release(struct inode *inode, struct file *file) +{ + struct c4iw_debugfs_data *epd = file->private_data; + if (!epd) { + pr_info("%s null qpd?\n", __func__); + return 0; + } + vfree(epd->buf); + kfree(epd); + return 0; +} + +static int ep_open(struct inode *inode, struct file *file) +{ + struct c4iw_debugfs_data *epd; + int ret = 0; + int count = 1; + + epd = kmalloc(sizeof(*epd), GFP_KERNEL); + if (!epd) { + ret = -ENOMEM; + goto out; + } + epd->devp = inode->i_private; + epd->pos = 0; + + spin_lock_irq(&epd->devp->lock); + idr_for_each(&epd->devp->hwtid_idr, count_idrs, &count); + idr_for_each(&epd->devp->atid_idr, count_idrs, &count); + idr_for_each(&epd->devp->stid_idr, count_idrs, &count); + spin_unlock_irq(&epd->devp->lock); + + epd->bufsize = count * 240; + epd->buf = vmalloc(epd->bufsize); + if (!epd->buf) { + ret = -ENOMEM; + goto err1; + } + + spin_lock_irq(&epd->devp->lock); + idr_for_each(&epd->devp->hwtid_idr, dump_ep, epd); + idr_for_each(&epd->devp->atid_idr, dump_ep, epd); + idr_for_each(&epd->devp->stid_idr, dump_listen_ep, epd); + spin_unlock_irq(&epd->devp->lock); + + file->private_data = epd; + goto out; +err1: + kfree(epd); +out: + return ret; +} + +static const struct file_operations ep_debugfs_fops = { + .owner = THIS_MODULE, + .open = ep_open, + .release = ep_release, + .read = debugfs_read, +}; + +static int setup_debugfs(struct c4iw_dev *devp) +{ + if (!devp->debugfs_root) + return -1; + + debugfs_create_file_size("qps", S_IWUSR, devp->debugfs_root, + (void *)devp, &qp_debugfs_fops, 4096); + + debugfs_create_file_size("stags", S_IWUSR, devp->debugfs_root, + (void *)devp, &stag_debugfs_fops, 4096); + + debugfs_create_file_size("stats", S_IWUSR, devp->debugfs_root, + (void *)devp, &stats_debugfs_fops, 4096); + + debugfs_create_file_size("eps", S_IWUSR, devp->debugfs_root, + (void *)devp, &ep_debugfs_fops, 4096); + + if (c4iw_wr_log) + debugfs_create_file_size("wr_log", S_IWUSR, devp->debugfs_root, + (void *)devp, &wr_log_debugfs_fops, 4096); + return 0; +} + +void c4iw_release_dev_ucontext(struct c4iw_rdev *rdev, + struct c4iw_dev_ucontext *uctx) +{ + struct list_head *pos, *nxt; + struct c4iw_qid_list *entry; + + mutex_lock(&uctx->lock); + list_for_each_safe(pos, nxt, &uctx->qpids) { + entry = list_entry(pos, struct c4iw_qid_list, entry); + list_del_init(&entry->entry); + if (!(entry->qid & rdev->qpmask)) { + c4iw_put_resource(&rdev->resource.qid_table, + entry->qid); + mutex_lock(&rdev->stats.lock); + rdev->stats.qid.cur -= rdev->qpmask + 1; + mutex_unlock(&rdev->stats.lock); + } + kfree(entry); + } + + list_for_each_safe(pos, nxt, &uctx->qpids) { + entry = list_entry(pos, struct c4iw_qid_list, entry); + list_del_init(&entry->entry); + kfree(entry); + } + mutex_unlock(&uctx->lock); +} + +void c4iw_init_dev_ucontext(struct c4iw_rdev *rdev, + struct c4iw_dev_ucontext *uctx) +{ + INIT_LIST_HEAD(&uctx->qpids); + INIT_LIST_HEAD(&uctx->cqids); + mutex_init(&uctx->lock); +} + +/* Caller takes care of locking if needed */ +static int c4iw_rdev_open(struct c4iw_rdev *rdev) +{ + int err; + + c4iw_init_dev_ucontext(rdev, &rdev->uctx); + + /* + * This implementation assumes udb_density == ucq_density! Eventually + * we might need to support this but for now fail the open. Also the + * cqid and qpid range must match for now. + */ + if (rdev->lldi.udb_density != rdev->lldi.ucq_density) { + pr_err(MOD "%s: unsupported udb/ucq densities %u/%u\n", + pci_name(rdev->lldi.pdev), rdev->lldi.udb_density, + rdev->lldi.ucq_density); + err = -EINVAL; + goto err1; + } + if (rdev->lldi.vr->qp.start != rdev->lldi.vr->cq.start || + rdev->lldi.vr->qp.size != rdev->lldi.vr->cq.size) { + pr_err(MOD "%s: unsupported qp and cq id ranges " + "qp start %u size %u cq start %u size %u\n", + pci_name(rdev->lldi.pdev), rdev->lldi.vr->qp.start, + rdev->lldi.vr->qp.size, rdev->lldi.vr->cq.size, + rdev->lldi.vr->cq.size); + err = -EINVAL; + goto err1; + } + + /* + * qpshift is the number of bits to shift the qpid left in order + * to get the correct address of the doorbell for that qp. + */ + rdev->qpshift = PAGE_SHIFT - ilog2(rdev->lldi.udb_density); + rdev->qpmask = rdev->lldi.udb_density - 1; + rdev->cqshift = PAGE_SHIFT - ilog2(rdev->lldi.ucq_density); + rdev->cqmask = rdev->lldi.ucq_density - 1; + PDBG("%s dev %s stag start 0x%0x size 0x%0x num stags %d " + "pbl start 0x%0x size 0x%0x rq start 0x%0x size 0x%0x " + "qp qid start %u size %u cq qid start %u size %u\n", + __func__, pci_name(rdev->lldi.pdev), rdev->lldi.vr->stag.start, + rdev->lldi.vr->stag.size, c4iw_num_stags(rdev), + rdev->lldi.vr->pbl.start, + rdev->lldi.vr->pbl.size, rdev->lldi.vr->rq.start, + rdev->lldi.vr->rq.size, + rdev->lldi.vr->qp.start, + rdev->lldi.vr->qp.size, + rdev->lldi.vr->cq.start, + rdev->lldi.vr->cq.size); + PDBG("udb len 0x%x udb base %p db_reg %p gts_reg %p qpshift %lu " + "qpmask 0x%x cqshift %lu cqmask 0x%x\n", + (unsigned)pci_resource_len(rdev->lldi.pdev, 2), + (void *)pci_resource_start(rdev->lldi.pdev, 2), + rdev->lldi.db_reg, + rdev->lldi.gts_reg, + rdev->qpshift, rdev->qpmask, + rdev->cqshift, rdev->cqmask); + + if (c4iw_num_stags(rdev) == 0) { + err = -EINVAL; + goto err1; + } + + rdev->stats.pd.total = T4_MAX_NUM_PD; + rdev->stats.stag.total = rdev->lldi.vr->stag.size; + rdev->stats.pbl.total = rdev->lldi.vr->pbl.size; + rdev->stats.rqt.total = rdev->lldi.vr->rq.size; + rdev->stats.ocqp.total = rdev->lldi.vr->ocq.size; + rdev->stats.qid.total = rdev->lldi.vr->qp.size; + + err = c4iw_init_resource(rdev, c4iw_num_stags(rdev), T4_MAX_NUM_PD); + if (err) { + printk(KERN_ERR MOD "error %d initializing resources\n", err); + goto err1; + } + err = c4iw_pblpool_create(rdev); + if (err) { + printk(KERN_ERR MOD "error %d initializing pbl pool\n", err); + goto err2; + } + err = c4iw_rqtpool_create(rdev); + if (err) { + printk(KERN_ERR MOD "error %d initializing rqt pool\n", err); + goto err3; + } + err = c4iw_ocqp_pool_create(rdev); + if (err) { + printk(KERN_ERR MOD "error %d initializing ocqp pool\n", err); + goto err4; + } + rdev->status_page = (struct t4_dev_status_page *) + __get_free_page(GFP_KERNEL); + if (!rdev->status_page) { + pr_err(MOD "error allocating status page\n"); + goto err4; + } + + if (c4iw_wr_log) { + rdev->wr_log = kzalloc((1 << c4iw_wr_log_size_order) * + sizeof(*rdev->wr_log), GFP_KERNEL); + if (rdev->wr_log) { + rdev->wr_log_size = 1 << c4iw_wr_log_size_order; + atomic_set(&rdev->wr_log_idx, 0); + } else { + pr_err(MOD "error allocating wr_log. Logging disabled\n"); + } + } + + rdev->status_page->db_off = 0; + + return 0; +err4: + c4iw_rqtpool_destroy(rdev); +err3: + c4iw_pblpool_destroy(rdev); +err2: + c4iw_destroy_resource(&rdev->resource); +err1: + return err; +} + +static void c4iw_rdev_close(struct c4iw_rdev *rdev) +{ + kfree(rdev->wr_log); + free_page((unsigned long)rdev->status_page); + c4iw_pblpool_destroy(rdev); + c4iw_rqtpool_destroy(rdev); + c4iw_destroy_resource(&rdev->resource); +} + +static void c4iw_dealloc(struct uld_ctx *ctx) +{ + c4iw_rdev_close(&ctx->dev->rdev); + idr_destroy(&ctx->dev->cqidr); + idr_destroy(&ctx->dev->qpidr); + idr_destroy(&ctx->dev->mmidr); + idr_destroy(&ctx->dev->hwtid_idr); + idr_destroy(&ctx->dev->stid_idr); + idr_destroy(&ctx->dev->atid_idr); + if (ctx->dev->rdev.bar2_kva) + iounmap(ctx->dev->rdev.bar2_kva); + if (ctx->dev->rdev.oc_mw_kva) + iounmap(ctx->dev->rdev.oc_mw_kva); + ib_dealloc_device(&ctx->dev->ibdev); + ctx->dev = NULL; +} + +static void c4iw_remove(struct uld_ctx *ctx) +{ + PDBG("%s c4iw_dev %p\n", __func__, ctx->dev); + c4iw_unregister_device(ctx->dev); + c4iw_dealloc(ctx); +} + +static int rdma_supported(const struct cxgb4_lld_info *infop) +{ + return infop->vr->stag.size > 0 && infop->vr->pbl.size > 0 && + infop->vr->rq.size > 0 && infop->vr->qp.size > 0 && + infop->vr->cq.size > 0; +} + +static struct c4iw_dev *c4iw_alloc(const struct cxgb4_lld_info *infop) +{ + struct c4iw_dev *devp; + int ret; + + if (!rdma_supported(infop)) { + printk(KERN_INFO MOD "%s: RDMA not supported on this device.\n", + pci_name(infop->pdev)); + return ERR_PTR(-ENOSYS); + } + if (!ocqp_supported(infop)) + pr_info("%s: On-Chip Queues not supported on this device.\n", + pci_name(infop->pdev)); + + devp = (struct c4iw_dev *)ib_alloc_device(sizeof(*devp)); + if (!devp) { + printk(KERN_ERR MOD "Cannot allocate ib device\n"); + return ERR_PTR(-ENOMEM); + } + devp->rdev.lldi = *infop; + + /* init various hw-queue params based on lld info */ + PDBG("%s: Ing. padding boundary is %d, egrsstatuspagesize = %d\n", + __func__, devp->rdev.lldi.sge_ingpadboundary, + devp->rdev.lldi.sge_egrstatuspagesize); + + devp->rdev.hw_queue.t4_eq_status_entries = + devp->rdev.lldi.sge_ingpadboundary > 64 ? 2 : 1; + devp->rdev.hw_queue.t4_max_eq_size = 65520; + devp->rdev.hw_queue.t4_max_iq_size = 65520; + devp->rdev.hw_queue.t4_max_rq_size = 8192 - + devp->rdev.hw_queue.t4_eq_status_entries - 1; + devp->rdev.hw_queue.t4_max_sq_size = + devp->rdev.hw_queue.t4_max_eq_size - + devp->rdev.hw_queue.t4_eq_status_entries - 1; + devp->rdev.hw_queue.t4_max_qp_depth = + devp->rdev.hw_queue.t4_max_rq_size; + devp->rdev.hw_queue.t4_max_cq_depth = + devp->rdev.hw_queue.t4_max_iq_size - 2; + devp->rdev.hw_queue.t4_stat_len = + devp->rdev.lldi.sge_egrstatuspagesize; + + /* + * For T5 devices, we map all of BAR2 with WC. + * For T4 devices with onchip qp mem, we map only that part + * of BAR2 with WC. + */ + devp->rdev.bar2_pa = pci_resource_start(devp->rdev.lldi.pdev, 2); + if (is_t5(devp->rdev.lldi.adapter_type)) { + devp->rdev.bar2_kva = ioremap_wc(devp->rdev.bar2_pa, + pci_resource_len(devp->rdev.lldi.pdev, 2)); + if (!devp->rdev.bar2_kva) { + pr_err(MOD "Unable to ioremap BAR2\n"); + ib_dealloc_device(&devp->ibdev); + return ERR_PTR(-EINVAL); + } + } else if (ocqp_supported(infop)) { + devp->rdev.oc_mw_pa = + pci_resource_start(devp->rdev.lldi.pdev, 2) + + pci_resource_len(devp->rdev.lldi.pdev, 2) - + roundup_pow_of_two(devp->rdev.lldi.vr->ocq.size); + devp->rdev.oc_mw_kva = ioremap_wc(devp->rdev.oc_mw_pa, + devp->rdev.lldi.vr->ocq.size); + if (!devp->rdev.oc_mw_kva) { + pr_err(MOD "Unable to ioremap onchip mem\n"); + ib_dealloc_device(&devp->ibdev); + return ERR_PTR(-EINVAL); + } + } + + PDBG(KERN_INFO MOD "ocq memory: " + "hw_start 0x%x size %u mw_pa 0x%lx mw_kva %p\n", + devp->rdev.lldi.vr->ocq.start, devp->rdev.lldi.vr->ocq.size, + devp->rdev.oc_mw_pa, devp->rdev.oc_mw_kva); + + ret = c4iw_rdev_open(&devp->rdev); + if (ret) { + printk(KERN_ERR MOD "Unable to open CXIO rdev err %d\n", ret); + ib_dealloc_device(&devp->ibdev); + return ERR_PTR(ret); + } + + idr_init(&devp->cqidr); + idr_init(&devp->qpidr); + idr_init(&devp->mmidr); + idr_init(&devp->hwtid_idr); + idr_init(&devp->stid_idr); + idr_init(&devp->atid_idr); + spin_lock_init(&devp->lock); + mutex_init(&devp->rdev.stats.lock); + mutex_init(&devp->db_mutex); + INIT_LIST_HEAD(&devp->db_fc_list); + devp->avail_ird = devp->rdev.lldi.max_ird_adapter; + + if (c4iw_debugfs_root) { + devp->debugfs_root = debugfs_create_dir( + pci_name(devp->rdev.lldi.pdev), + c4iw_debugfs_root); + setup_debugfs(devp); + } + + + return devp; +} + +static void *c4iw_uld_add(const struct cxgb4_lld_info *infop) +{ + struct uld_ctx *ctx; + static int vers_printed; + int i; + + if (!vers_printed++) + pr_info("Chelsio T4/T5 RDMA Driver - version %s\n", + DRV_VERSION); + + ctx = kzalloc(sizeof *ctx, GFP_KERNEL); + if (!ctx) { + ctx = ERR_PTR(-ENOMEM); + goto out; + } + ctx->lldi = *infop; + + PDBG("%s found device %s nchan %u nrxq %u ntxq %u nports %u\n", + __func__, pci_name(ctx->lldi.pdev), + ctx->lldi.nchan, ctx->lldi.nrxq, + ctx->lldi.ntxq, ctx->lldi.nports); + + mutex_lock(&dev_mutex); + list_add_tail(&ctx->entry, &uld_ctx_list); + mutex_unlock(&dev_mutex); + + for (i = 0; i < ctx->lldi.nrxq; i++) + PDBG("rxqid[%u] %u\n", i, ctx->lldi.rxq_ids[i]); +out: + return ctx; +} + +static inline struct sk_buff *copy_gl_to_skb_pkt(const struct pkt_gl *gl, + const __be64 *rsp, + u32 pktshift) +{ + struct sk_buff *skb; + + /* + * Allocate space for cpl_pass_accept_req which will be synthesized by + * driver. Once the driver synthesizes the request the skb will go + * through the regular cpl_pass_accept_req processing. + * The math here assumes sizeof cpl_pass_accept_req >= sizeof + * cpl_rx_pkt. + */ + skb = alloc_skb(gl->tot_len + sizeof(struct cpl_pass_accept_req) + + sizeof(struct rss_header) - pktshift, GFP_ATOMIC); + if (unlikely(!skb)) + return NULL; + + __skb_put(skb, gl->tot_len + sizeof(struct cpl_pass_accept_req) + + sizeof(struct rss_header) - pktshift); + + /* + * This skb will contain: + * rss_header from the rspq descriptor (1 flit) + * cpl_rx_pkt struct from the rspq descriptor (2 flits) + * space for the difference between the size of an + * rx_pkt and pass_accept_req cpl (1 flit) + * the packet data from the gl + */ + skb_copy_to_linear_data(skb, rsp, sizeof(struct cpl_pass_accept_req) + + sizeof(struct rss_header)); + skb_copy_to_linear_data_offset(skb, sizeof(struct rss_header) + + sizeof(struct cpl_pass_accept_req), + gl->va + pktshift, + gl->tot_len - pktshift); + return skb; +} + +static inline int recv_rx_pkt(struct c4iw_dev *dev, const struct pkt_gl *gl, + const __be64 *rsp) +{ + unsigned int opcode = *(u8 *)rsp; + struct sk_buff *skb; + + if (opcode != CPL_RX_PKT) + goto out; + + skb = copy_gl_to_skb_pkt(gl , rsp, dev->rdev.lldi.sge_pktshift); + if (skb == NULL) + goto out; + + if (c4iw_handlers[opcode] == NULL) { + pr_info("%s no handler opcode 0x%x...\n", __func__, + opcode); + kfree_skb(skb); + goto out; + } + c4iw_handlers[opcode](dev, skb); + return 1; +out: + return 0; +} + +static int c4iw_uld_rx_handler(void *handle, const __be64 *rsp, + const struct pkt_gl *gl) +{ + struct uld_ctx *ctx = handle; + struct c4iw_dev *dev = ctx->dev; + struct sk_buff *skb; + u8 opcode; + + if (gl == NULL) { + /* omit RSS and rsp_ctrl at end of descriptor */ + unsigned int len = 64 - sizeof(struct rsp_ctrl) - 8; + + skb = alloc_skb(256, GFP_ATOMIC); + if (!skb) + goto nomem; + __skb_put(skb, len); + skb_copy_to_linear_data(skb, &rsp[1], len); + } else if (gl == CXGB4_MSG_AN) { + const struct rsp_ctrl *rc = (void *)rsp; + + u32 qid = be32_to_cpu(rc->pldbuflen_qid); + c4iw_ev_handler(dev, qid); + return 0; + } else if (unlikely(*(u8 *)rsp != *(u8 *)gl->va)) { + if (recv_rx_pkt(dev, gl, rsp)) + return 0; + + pr_info("%s: unexpected FL contents at %p, " \ + "RSS %#llx, FL %#llx, len %u\n", + pci_name(ctx->lldi.pdev), gl->va, + (unsigned long long)be64_to_cpu(*rsp), + (unsigned long long)be64_to_cpu( + *(__force __be64 *)gl->va), + gl->tot_len); + + return 0; + } else { + skb = cxgb4_pktgl_to_skb(gl, 128, 128); + if (unlikely(!skb)) + goto nomem; + } + + opcode = *(u8 *)rsp; + if (c4iw_handlers[opcode]) { + c4iw_handlers[opcode](dev, skb); + } else { + pr_info("%s no handler opcode 0x%x...\n", __func__, + opcode); + kfree_skb(skb); + } + + return 0; +nomem: + return -1; +} + +static int c4iw_uld_state_change(void *handle, enum cxgb4_state new_state) +{ + struct uld_ctx *ctx = handle; + + PDBG("%s new_state %u\n", __func__, new_state); + switch (new_state) { + case CXGB4_STATE_UP: + printk(KERN_INFO MOD "%s: Up\n", pci_name(ctx->lldi.pdev)); + if (!ctx->dev) { + int ret; + + ctx->dev = c4iw_alloc(&ctx->lldi); + if (IS_ERR(ctx->dev)) { + printk(KERN_ERR MOD + "%s: initialization failed: %ld\n", + pci_name(ctx->lldi.pdev), + PTR_ERR(ctx->dev)); + ctx->dev = NULL; + break; + } + ret = c4iw_register_device(ctx->dev); + if (ret) { + printk(KERN_ERR MOD + "%s: RDMA registration failed: %d\n", + pci_name(ctx->lldi.pdev), ret); + c4iw_dealloc(ctx); + } + } + break; + case CXGB4_STATE_DOWN: + printk(KERN_INFO MOD "%s: Down\n", + pci_name(ctx->lldi.pdev)); + if (ctx->dev) + c4iw_remove(ctx); + break; + case CXGB4_STATE_START_RECOVERY: + printk(KERN_INFO MOD "%s: Fatal Error\n", + pci_name(ctx->lldi.pdev)); + if (ctx->dev) { + struct ib_event event; + + ctx->dev->rdev.flags |= T4_FATAL_ERROR; + memset(&event, 0, sizeof event); + event.event = IB_EVENT_DEVICE_FATAL; + event.device = &ctx->dev->ibdev; + ib_dispatch_event(&event); + c4iw_remove(ctx); + } + break; + case CXGB4_STATE_DETACH: + printk(KERN_INFO MOD "%s: Detach\n", + pci_name(ctx->lldi.pdev)); + if (ctx->dev) + c4iw_remove(ctx); + break; + } + return 0; +} + +static int disable_qp_db(int id, void *p, void *data) +{ + struct c4iw_qp *qp = p; + + t4_disable_wq_db(&qp->wq); + return 0; +} + +static void stop_queues(struct uld_ctx *ctx) +{ + unsigned long flags; + + spin_lock_irqsave(&ctx->dev->lock, flags); + ctx->dev->rdev.stats.db_state_transitions++; + ctx->dev->db_state = STOPPED; + if (ctx->dev->rdev.flags & T4_STATUS_PAGE_DISABLED) + idr_for_each(&ctx->dev->qpidr, disable_qp_db, NULL); + else + ctx->dev->rdev.status_page->db_off = 1; + spin_unlock_irqrestore(&ctx->dev->lock, flags); +} + +static int enable_qp_db(int id, void *p, void *data) +{ + struct c4iw_qp *qp = p; + + t4_enable_wq_db(&qp->wq); + return 0; +} + +static void resume_rc_qp(struct c4iw_qp *qp) +{ + spin_lock(&qp->lock); + t4_ring_sq_db(&qp->wq, qp->wq.sq.wq_pidx_inc, + is_t5(qp->rhp->rdev.lldi.adapter_type), NULL); + qp->wq.sq.wq_pidx_inc = 0; + t4_ring_rq_db(&qp->wq, qp->wq.rq.wq_pidx_inc, + is_t5(qp->rhp->rdev.lldi.adapter_type), NULL); + qp->wq.rq.wq_pidx_inc = 0; + spin_unlock(&qp->lock); +} + +static void resume_a_chunk(struct uld_ctx *ctx) +{ + int i; + struct c4iw_qp *qp; + + for (i = 0; i < DB_FC_RESUME_SIZE; i++) { + qp = list_first_entry(&ctx->dev->db_fc_list, struct c4iw_qp, + db_fc_entry); + list_del_init(&qp->db_fc_entry); + resume_rc_qp(qp); + if (list_empty(&ctx->dev->db_fc_list)) + break; + } +} + +static void resume_queues(struct uld_ctx *ctx) +{ + spin_lock_irq(&ctx->dev->lock); + if (ctx->dev->db_state != STOPPED) + goto out; + ctx->dev->db_state = FLOW_CONTROL; + while (1) { + if (list_empty(&ctx->dev->db_fc_list)) { + WARN_ON(ctx->dev->db_state != FLOW_CONTROL); + ctx->dev->db_state = NORMAL; + ctx->dev->rdev.stats.db_state_transitions++; + if (ctx->dev->rdev.flags & T4_STATUS_PAGE_DISABLED) { + idr_for_each(&ctx->dev->qpidr, enable_qp_db, + NULL); + } else { + ctx->dev->rdev.status_page->db_off = 0; + } + break; + } else { + if (cxgb4_dbfifo_count(ctx->dev->rdev.lldi.ports[0], 1) + < (ctx->dev->rdev.lldi.dbfifo_int_thresh << + DB_FC_DRAIN_THRESH)) { + resume_a_chunk(ctx); + } + if (!list_empty(&ctx->dev->db_fc_list)) { + spin_unlock_irq(&ctx->dev->lock); + if (DB_FC_RESUME_DELAY) { + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(DB_FC_RESUME_DELAY); + } + spin_lock_irq(&ctx->dev->lock); + if (ctx->dev->db_state != FLOW_CONTROL) + break; + } + } + } +out: + if (ctx->dev->db_state != NORMAL) + ctx->dev->rdev.stats.db_fc_interruptions++; + spin_unlock_irq(&ctx->dev->lock); +} + +struct qp_list { + unsigned idx; + struct c4iw_qp **qps; +}; + +static int add_and_ref_qp(int id, void *p, void *data) +{ + struct qp_list *qp_listp = data; + struct c4iw_qp *qp = p; + + c4iw_qp_add_ref(&qp->ibqp); + qp_listp->qps[qp_listp->idx++] = qp; + return 0; +} + +static int count_qps(int id, void *p, void *data) +{ + unsigned *countp = data; + (*countp)++; + return 0; +} + +static void deref_qps(struct qp_list *qp_list) +{ + int idx; + + for (idx = 0; idx < qp_list->idx; idx++) + c4iw_qp_rem_ref(&qp_list->qps[idx]->ibqp); +} + +static void recover_lost_dbs(struct uld_ctx *ctx, struct qp_list *qp_list) +{ + int idx; + int ret; + + for (idx = 0; idx < qp_list->idx; idx++) { + struct c4iw_qp *qp = qp_list->qps[idx]; + + spin_lock_irq(&qp->rhp->lock); + spin_lock(&qp->lock); + ret = cxgb4_sync_txq_pidx(qp->rhp->rdev.lldi.ports[0], + qp->wq.sq.qid, + t4_sq_host_wq_pidx(&qp->wq), + t4_sq_wq_size(&qp->wq)); + if (ret) { + pr_err(MOD "%s: Fatal error - " + "DB overflow recovery failed - " + "error syncing SQ qid %u\n", + pci_name(ctx->lldi.pdev), qp->wq.sq.qid); + spin_unlock(&qp->lock); + spin_unlock_irq(&qp->rhp->lock); + return; + } + qp->wq.sq.wq_pidx_inc = 0; + + ret = cxgb4_sync_txq_pidx(qp->rhp->rdev.lldi.ports[0], + qp->wq.rq.qid, + t4_rq_host_wq_pidx(&qp->wq), + t4_rq_wq_size(&qp->wq)); + + if (ret) { + pr_err(MOD "%s: Fatal error - " + "DB overflow recovery failed - " + "error syncing RQ qid %u\n", + pci_name(ctx->lldi.pdev), qp->wq.rq.qid); + spin_unlock(&qp->lock); + spin_unlock_irq(&qp->rhp->lock); + return; + } + qp->wq.rq.wq_pidx_inc = 0; + spin_unlock(&qp->lock); + spin_unlock_irq(&qp->rhp->lock); + + /* Wait for the dbfifo to drain */ + while (cxgb4_dbfifo_count(qp->rhp->rdev.lldi.ports[0], 1) > 0) { + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(usecs_to_jiffies(10)); + } + } +} + +static void recover_queues(struct uld_ctx *ctx) +{ + int count = 0; + struct qp_list qp_list; + int ret; + + /* slow everybody down */ + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(usecs_to_jiffies(1000)); + + /* flush the SGE contexts */ + ret = cxgb4_flush_eq_cache(ctx->dev->rdev.lldi.ports[0]); + if (ret) { + printk(KERN_ERR MOD "%s: Fatal error - DB overflow recovery failed\n", + pci_name(ctx->lldi.pdev)); + return; + } + + /* Count active queues so we can build a list of queues to recover */ + spin_lock_irq(&ctx->dev->lock); + WARN_ON(ctx->dev->db_state != STOPPED); + ctx->dev->db_state = RECOVERY; + idr_for_each(&ctx->dev->qpidr, count_qps, &count); + + qp_list.qps = kzalloc(count * sizeof *qp_list.qps, GFP_ATOMIC); + if (!qp_list.qps) { + printk(KERN_ERR MOD "%s: Fatal error - DB overflow recovery failed\n", + pci_name(ctx->lldi.pdev)); + spin_unlock_irq(&ctx->dev->lock); + return; + } + qp_list.idx = 0; + + /* add and ref each qp so it doesn't get freed */ + idr_for_each(&ctx->dev->qpidr, add_and_ref_qp, &qp_list); + + spin_unlock_irq(&ctx->dev->lock); + + /* now traverse the list in a safe context to recover the db state*/ + recover_lost_dbs(ctx, &qp_list); + + /* we're almost done! deref the qps and clean up */ + deref_qps(&qp_list); + kfree(qp_list.qps); + + spin_lock_irq(&ctx->dev->lock); + WARN_ON(ctx->dev->db_state != RECOVERY); + ctx->dev->db_state = STOPPED; + spin_unlock_irq(&ctx->dev->lock); +} + +static int c4iw_uld_control(void *handle, enum cxgb4_control control, ...) +{ + struct uld_ctx *ctx = handle; + + switch (control) { + case CXGB4_CONTROL_DB_FULL: + stop_queues(ctx); + ctx->dev->rdev.stats.db_full++; + break; + case CXGB4_CONTROL_DB_EMPTY: + resume_queues(ctx); + mutex_lock(&ctx->dev->rdev.stats.lock); + ctx->dev->rdev.stats.db_empty++; + mutex_unlock(&ctx->dev->rdev.stats.lock); + break; + case CXGB4_CONTROL_DB_DROP: + recover_queues(ctx); + mutex_lock(&ctx->dev->rdev.stats.lock); + ctx->dev->rdev.stats.db_drop++; + mutex_unlock(&ctx->dev->rdev.stats.lock); + break; + default: + printk(KERN_WARNING MOD "%s: unknown control cmd %u\n", + pci_name(ctx->lldi.pdev), control); + break; + } + return 0; +} + +static struct cxgb4_uld_info c4iw_uld_info = { + .name = DRV_NAME, + .add = c4iw_uld_add, + .rx_handler = c4iw_uld_rx_handler, + .state_change = c4iw_uld_state_change, + .control = c4iw_uld_control, +}; + +static int __init c4iw_init_module(void) +{ + int err; + + err = c4iw_cm_init(); + if (err) + return err; + + c4iw_debugfs_root = debugfs_create_dir(DRV_NAME, NULL); + if (!c4iw_debugfs_root) + printk(KERN_WARNING MOD + "could not create debugfs entry, continuing\n"); + + if (ibnl_add_client(RDMA_NL_C4IW, RDMA_NL_IWPM_NUM_OPS, + c4iw_nl_cb_table)) + pr_err("%s[%u]: Failed to add netlink callback\n" + , __func__, __LINE__); + + err = iwpm_init(RDMA_NL_C4IW); + if (err) { + pr_err("port mapper initialization failed with %d\n", err); + ibnl_remove_client(RDMA_NL_C4IW); + c4iw_cm_term(); + debugfs_remove_recursive(c4iw_debugfs_root); + return err; + } + + cxgb4_register_uld(CXGB4_ULD_RDMA, &c4iw_uld_info); + + return 0; +} + +static void __exit c4iw_exit_module(void) +{ + struct uld_ctx *ctx, *tmp; + + mutex_lock(&dev_mutex); + list_for_each_entry_safe(ctx, tmp, &uld_ctx_list, entry) { + if (ctx->dev) + c4iw_remove(ctx); + kfree(ctx); + } + mutex_unlock(&dev_mutex); + cxgb4_unregister_uld(CXGB4_ULD_RDMA); + iwpm_exit(RDMA_NL_C4IW); + ibnl_remove_client(RDMA_NL_C4IW); + c4iw_cm_term(); + debugfs_remove_recursive(c4iw_debugfs_root); +} + +module_init(c4iw_init_module); +module_exit(c4iw_exit_module); diff --git a/kernel/drivers/infiniband/hw/cxgb4/ev.c b/kernel/drivers/infiniband/hw/cxgb4/ev.c new file mode 100644 index 000000000..bdfac2ccb --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/ev.c @@ -0,0 +1,244 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#include <linux/slab.h> +#include <linux/mman.h> +#include <net/sock.h> + +#include "iw_cxgb4.h" + +static void print_tpte(struct c4iw_dev *dev, u32 stag) +{ + int ret; + struct fw_ri_tpte tpte; + + ret = cxgb4_read_tpte(dev->rdev.lldi.ports[0], stag, + (__be32 *)&tpte); + if (ret) { + dev_err(&dev->rdev.lldi.pdev->dev, + "%s cxgb4_read_tpte err %d\n", __func__, ret); + return; + } + PDBG("stag idx 0x%x valid %d key 0x%x state %d pdid %d " + "perm 0x%x ps %d len 0x%llx va 0x%llx\n", + stag & 0xffffff00, + FW_RI_TPTE_VALID_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_STAGKEY_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_STAGSTATE_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_PDID_G(ntohl(tpte.valid_to_pdid)), + FW_RI_TPTE_PERM_G(ntohl(tpte.locread_to_qpid)), + FW_RI_TPTE_PS_G(ntohl(tpte.locread_to_qpid)), + ((u64)ntohl(tpte.len_hi) << 32) | ntohl(tpte.len_lo), + ((u64)ntohl(tpte.va_hi) << 32) | ntohl(tpte.va_lo_fbo)); +} + +static void dump_err_cqe(struct c4iw_dev *dev, struct t4_cqe *err_cqe) +{ + __be64 *p = (void *)err_cqe; + + dev_err(&dev->rdev.lldi.pdev->dev, + "AE qpid %d opcode %d status 0x%x " + "type %d len 0x%x wrid.hi 0x%x wrid.lo 0x%x\n", + CQE_QPID(err_cqe), CQE_OPCODE(err_cqe), + CQE_STATUS(err_cqe), CQE_TYPE(err_cqe), ntohl(err_cqe->len), + CQE_WRID_HI(err_cqe), CQE_WRID_LOW(err_cqe)); + + PDBG("%016llx %016llx %016llx %016llx\n", + be64_to_cpu(p[0]), be64_to_cpu(p[1]), be64_to_cpu(p[2]), + be64_to_cpu(p[3])); + + /* + * Ingress WRITE and READ_RESP errors provide + * the offending stag, so parse and log it. + */ + if (RQ_TYPE(err_cqe) && (CQE_OPCODE(err_cqe) == FW_RI_RDMA_WRITE || + CQE_OPCODE(err_cqe) == FW_RI_READ_RESP)) + print_tpte(dev, CQE_WRID_STAG(err_cqe)); +} + +static void post_qp_event(struct c4iw_dev *dev, struct c4iw_cq *chp, + struct c4iw_qp *qhp, + struct t4_cqe *err_cqe, + enum ib_event_type ib_event) +{ + struct ib_event event; + struct c4iw_qp_attributes attrs; + unsigned long flag; + + dump_err_cqe(dev, err_cqe); + + if (qhp->attr.state == C4IW_QP_STATE_RTS) { + attrs.next_state = C4IW_QP_STATE_TERMINATE; + c4iw_modify_qp(qhp->rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, + &attrs, 0); + } + + event.event = ib_event; + event.device = chp->ibcq.device; + if (ib_event == IB_EVENT_CQ_ERR) + event.element.cq = &chp->ibcq; + else + event.element.qp = &qhp->ibqp; + if (qhp->ibqp.event_handler) + (*qhp->ibqp.event_handler)(&event, qhp->ibqp.qp_context); + + spin_lock_irqsave(&chp->comp_handler_lock, flag); + (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); + spin_unlock_irqrestore(&chp->comp_handler_lock, flag); +} + +void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe) +{ + struct c4iw_cq *chp; + struct c4iw_qp *qhp; + u32 cqid; + + spin_lock_irq(&dev->lock); + qhp = get_qhp(dev, CQE_QPID(err_cqe)); + if (!qhp) { + printk(KERN_ERR MOD "BAD AE qpid 0x%x opcode %d " + "status 0x%x type %d wrid.hi 0x%x wrid.lo 0x%x\n", + CQE_QPID(err_cqe), + CQE_OPCODE(err_cqe), CQE_STATUS(err_cqe), + CQE_TYPE(err_cqe), CQE_WRID_HI(err_cqe), + CQE_WRID_LOW(err_cqe)); + spin_unlock_irq(&dev->lock); + goto out; + } + + if (SQ_TYPE(err_cqe)) + cqid = qhp->attr.scq; + else + cqid = qhp->attr.rcq; + chp = get_chp(dev, cqid); + if (!chp) { + printk(KERN_ERR MOD "BAD AE cqid 0x%x qpid 0x%x opcode %d " + "status 0x%x type %d wrid.hi 0x%x wrid.lo 0x%x\n", + cqid, CQE_QPID(err_cqe), + CQE_OPCODE(err_cqe), CQE_STATUS(err_cqe), + CQE_TYPE(err_cqe), CQE_WRID_HI(err_cqe), + CQE_WRID_LOW(err_cqe)); + spin_unlock_irq(&dev->lock); + goto out; + } + + c4iw_qp_add_ref(&qhp->ibqp); + atomic_inc(&chp->refcnt); + spin_unlock_irq(&dev->lock); + + /* Bad incoming write */ + if (RQ_TYPE(err_cqe) && + (CQE_OPCODE(err_cqe) == FW_RI_RDMA_WRITE)) { + post_qp_event(dev, chp, qhp, err_cqe, IB_EVENT_QP_REQ_ERR); + goto done; + } + + switch (CQE_STATUS(err_cqe)) { + + /* Completion Events */ + case T4_ERR_SUCCESS: + printk(KERN_ERR MOD "AE with status 0!\n"); + break; + + case T4_ERR_STAG: + case T4_ERR_PDID: + case T4_ERR_QPID: + case T4_ERR_ACCESS: + case T4_ERR_WRAP: + case T4_ERR_BOUND: + case T4_ERR_INVALIDATE_SHARED_MR: + case T4_ERR_INVALIDATE_MR_WITH_MW_BOUND: + post_qp_event(dev, chp, qhp, err_cqe, IB_EVENT_QP_ACCESS_ERR); + break; + + /* Device Fatal Errors */ + case T4_ERR_ECC: + case T4_ERR_ECC_PSTAG: + case T4_ERR_INTERNAL_ERR: + post_qp_event(dev, chp, qhp, err_cqe, IB_EVENT_DEVICE_FATAL); + break; + + /* QP Fatal Errors */ + case T4_ERR_OUT_OF_RQE: + case T4_ERR_PBL_ADDR_BOUND: + case T4_ERR_CRC: + case T4_ERR_MARKER: + case T4_ERR_PDU_LEN_ERR: + case T4_ERR_DDP_VERSION: + case T4_ERR_RDMA_VERSION: + case T4_ERR_OPCODE: + case T4_ERR_DDP_QUEUE_NUM: + case T4_ERR_MSN: + case T4_ERR_TBIT: + case T4_ERR_MO: + case T4_ERR_MSN_GAP: + case T4_ERR_MSN_RANGE: + case T4_ERR_RQE_ADDR_BOUND: + case T4_ERR_IRD_OVERFLOW: + post_qp_event(dev, chp, qhp, err_cqe, IB_EVENT_QP_FATAL); + break; + + default: + printk(KERN_ERR MOD "Unknown T4 status 0x%x QPID 0x%x\n", + CQE_STATUS(err_cqe), qhp->wq.sq.qid); + post_qp_event(dev, chp, qhp, err_cqe, IB_EVENT_QP_FATAL); + break; + } +done: + if (atomic_dec_and_test(&chp->refcnt)) + wake_up(&chp->wait); + c4iw_qp_rem_ref(&qhp->ibqp); +out: + return; +} + +int c4iw_ev_handler(struct c4iw_dev *dev, u32 qid) +{ + struct c4iw_cq *chp; + unsigned long flag; + + spin_lock_irqsave(&dev->lock, flag); + chp = get_chp(dev, qid); + if (chp) { + atomic_inc(&chp->refcnt); + spin_unlock_irqrestore(&dev->lock, flag); + t4_clear_cq_armed(&chp->cq); + spin_lock_irqsave(&chp->comp_handler_lock, flag); + (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); + spin_unlock_irqrestore(&chp->comp_handler_lock, flag); + if (atomic_dec_and_test(&chp->refcnt)) + wake_up(&chp->wait); + } else { + PDBG("%s unknown cqid 0x%x\n", __func__, qid); + spin_unlock_irqrestore(&dev->lock, flag); + } + return 0; +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/id_table.c b/kernel/drivers/infiniband/hw/cxgb4/id_table.c new file mode 100644 index 000000000..0161ae6ad --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/id_table.c @@ -0,0 +1,112 @@ +/* + * Copyright (c) 2011 Chelsio Communications. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#include <linux/kernel.h> +#include <linux/random.h> +#include "iw_cxgb4.h" + +#define RANDOM_SKIP 16 + +/* + * Trivial bitmap-based allocator. If the random flag is set, the + * allocator is designed to: + * - pseudo-randomize the id returned such that it is not trivially predictable. + * - avoid reuse of recently used id (at the expense of predictability) + */ +u32 c4iw_id_alloc(struct c4iw_id_table *alloc) +{ + unsigned long flags; + u32 obj; + + spin_lock_irqsave(&alloc->lock, flags); + + obj = find_next_zero_bit(alloc->table, alloc->max, alloc->last); + if (obj >= alloc->max) + obj = find_first_zero_bit(alloc->table, alloc->max); + + if (obj < alloc->max) { + if (alloc->flags & C4IW_ID_TABLE_F_RANDOM) + alloc->last += prandom_u32() % RANDOM_SKIP; + else + alloc->last = obj + 1; + if (alloc->last >= alloc->max) + alloc->last = 0; + set_bit(obj, alloc->table); + obj += alloc->start; + } else + obj = -1; + + spin_unlock_irqrestore(&alloc->lock, flags); + return obj; +} + +void c4iw_id_free(struct c4iw_id_table *alloc, u32 obj) +{ + unsigned long flags; + + obj -= alloc->start; + BUG_ON((int)obj < 0); + + spin_lock_irqsave(&alloc->lock, flags); + clear_bit(obj, alloc->table); + spin_unlock_irqrestore(&alloc->lock, flags); +} + +int c4iw_id_table_alloc(struct c4iw_id_table *alloc, u32 start, u32 num, + u32 reserved, u32 flags) +{ + int i; + + alloc->start = start; + alloc->flags = flags; + if (flags & C4IW_ID_TABLE_F_RANDOM) + alloc->last = prandom_u32() % RANDOM_SKIP; + else + alloc->last = 0; + alloc->max = num; + spin_lock_init(&alloc->lock); + alloc->table = kmalloc(BITS_TO_LONGS(num) * sizeof(long), + GFP_KERNEL); + if (!alloc->table) + return -ENOMEM; + + bitmap_zero(alloc->table, num); + if (!(alloc->flags & C4IW_ID_TABLE_F_EMPTY)) + for (i = 0; i < reserved; ++i) + set_bit(i, alloc->table); + + return 0; +} + +void c4iw_id_table_free(struct c4iw_id_table *alloc) +{ + kfree(alloc->table); +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/kernel/drivers/infiniband/hw/cxgb4/iw_cxgb4.h new file mode 100644 index 000000000..97bb5550a --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/iw_cxgb4.h @@ -0,0 +1,1042 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#ifndef __IW_CXGB4_H__ +#define __IW_CXGB4_H__ + +#include <linux/mutex.h> +#include <linux/list.h> +#include <linux/spinlock.h> +#include <linux/idr.h> +#include <linux/completion.h> +#include <linux/netdevice.h> +#include <linux/sched.h> +#include <linux/pci.h> +#include <linux/dma-mapping.h> +#include <linux/inet.h> +#include <linux/wait.h> +#include <linux/kref.h> +#include <linux/timer.h> +#include <linux/io.h> + +#include <asm/byteorder.h> + +#include <net/net_namespace.h> + +#include <rdma/ib_verbs.h> +#include <rdma/iw_cm.h> +#include <rdma/rdma_netlink.h> +#include <rdma/iw_portmap.h> + +#include "cxgb4.h" +#include "cxgb4_uld.h" +#include "l2t.h" +#include "user.h" + +#define DRV_NAME "iw_cxgb4" +#define MOD DRV_NAME ":" + +extern int c4iw_debug; +#define PDBG(fmt, args...) \ +do { \ + if (c4iw_debug) \ + printk(MOD fmt, ## args); \ +} while (0) + +#include "t4.h" + +#define PBL_OFF(rdev_p, a) ((a) - (rdev_p)->lldi.vr->pbl.start) +#define RQT_OFF(rdev_p, a) ((a) - (rdev_p)->lldi.vr->rq.start) + +static inline void *cplhdr(struct sk_buff *skb) +{ + return skb->data; +} + +#define C4IW_ID_TABLE_F_RANDOM 1 /* Pseudo-randomize the id's returned */ +#define C4IW_ID_TABLE_F_EMPTY 2 /* Table is initially empty */ + +struct c4iw_id_table { + u32 flags; + u32 start; /* logical minimal id */ + u32 last; /* hint for find */ + u32 max; + spinlock_t lock; + unsigned long *table; +}; + +struct c4iw_resource { + struct c4iw_id_table tpt_table; + struct c4iw_id_table qid_table; + struct c4iw_id_table pdid_table; +}; + +struct c4iw_qid_list { + struct list_head entry; + u32 qid; +}; + +struct c4iw_dev_ucontext { + struct list_head qpids; + struct list_head cqids; + struct mutex lock; +}; + +enum c4iw_rdev_flags { + T4_FATAL_ERROR = (1<<0), + T4_STATUS_PAGE_DISABLED = (1<<1), +}; + +struct c4iw_stat { + u64 total; + u64 cur; + u64 max; + u64 fail; +}; + +struct c4iw_stats { + struct mutex lock; + struct c4iw_stat qid; + struct c4iw_stat pd; + struct c4iw_stat stag; + struct c4iw_stat pbl; + struct c4iw_stat rqt; + struct c4iw_stat ocqp; + u64 db_full; + u64 db_empty; + u64 db_drop; + u64 db_state_transitions; + u64 db_fc_interruptions; + u64 tcam_full; + u64 act_ofld_conn_fails; + u64 pas_ofld_conn_fails; + u64 neg_adv; +}; + +struct c4iw_hw_queue { + int t4_eq_status_entries; + int t4_max_eq_size; + int t4_max_iq_size; + int t4_max_rq_size; + int t4_max_sq_size; + int t4_max_qp_depth; + int t4_max_cq_depth; + int t4_stat_len; +}; + +struct wr_log_entry { + struct timespec post_host_ts; + struct timespec poll_host_ts; + u64 post_sge_ts; + u64 cqe_sge_ts; + u64 poll_sge_ts; + u16 qid; + u16 wr_id; + u8 opcode; + u8 valid; +}; + +struct c4iw_rdev { + struct c4iw_resource resource; + unsigned long qpshift; + u32 qpmask; + unsigned long cqshift; + u32 cqmask; + struct c4iw_dev_ucontext uctx; + struct gen_pool *pbl_pool; + struct gen_pool *rqt_pool; + struct gen_pool *ocqp_pool; + u32 flags; + struct cxgb4_lld_info lldi; + unsigned long bar2_pa; + void __iomem *bar2_kva; + unsigned long oc_mw_pa; + void __iomem *oc_mw_kva; + struct c4iw_stats stats; + struct c4iw_hw_queue hw_queue; + struct t4_dev_status_page *status_page; + atomic_t wr_log_idx; + struct wr_log_entry *wr_log; + int wr_log_size; +}; + +static inline int c4iw_fatal_error(struct c4iw_rdev *rdev) +{ + return rdev->flags & T4_FATAL_ERROR; +} + +static inline int c4iw_num_stags(struct c4iw_rdev *rdev) +{ + return (int)(rdev->lldi.vr->stag.size >> 5); +} + +#define C4IW_WR_TO (60*HZ) + +struct c4iw_wr_wait { + struct completion completion; + int ret; +}; + +static inline void c4iw_init_wr_wait(struct c4iw_wr_wait *wr_waitp) +{ + wr_waitp->ret = 0; + init_completion(&wr_waitp->completion); +} + +static inline void c4iw_wake_up(struct c4iw_wr_wait *wr_waitp, int ret) +{ + wr_waitp->ret = ret; + complete(&wr_waitp->completion); +} + +static inline int c4iw_wait_for_reply(struct c4iw_rdev *rdev, + struct c4iw_wr_wait *wr_waitp, + u32 hwtid, u32 qpid, + const char *func) +{ + int ret; + + if (c4iw_fatal_error(rdev)) { + wr_waitp->ret = -EIO; + goto out; + } + + ret = wait_for_completion_timeout(&wr_waitp->completion, C4IW_WR_TO); + if (!ret) { + PDBG("%s - Device %s not responding (disabling device) - tid %u qpid %u\n", + func, pci_name(rdev->lldi.pdev), hwtid, qpid); + rdev->flags |= T4_FATAL_ERROR; + wr_waitp->ret = -EIO; + } +out: + if (wr_waitp->ret) + PDBG("%s: FW reply %d tid %u qpid %u\n", + pci_name(rdev->lldi.pdev), wr_waitp->ret, hwtid, qpid); + return wr_waitp->ret; +} + +enum db_state { + NORMAL = 0, + FLOW_CONTROL = 1, + RECOVERY = 2, + STOPPED = 3 +}; + +struct c4iw_dev { + struct ib_device ibdev; + struct c4iw_rdev rdev; + u32 device_cap_flags; + struct idr cqidr; + struct idr qpidr; + struct idr mmidr; + spinlock_t lock; + struct mutex db_mutex; + struct dentry *debugfs_root; + enum db_state db_state; + struct idr hwtid_idr; + struct idr atid_idr; + struct idr stid_idr; + struct list_head db_fc_list; + u32 avail_ird; +}; + +static inline struct c4iw_dev *to_c4iw_dev(struct ib_device *ibdev) +{ + return container_of(ibdev, struct c4iw_dev, ibdev); +} + +static inline struct c4iw_dev *rdev_to_c4iw_dev(struct c4iw_rdev *rdev) +{ + return container_of(rdev, struct c4iw_dev, rdev); +} + +static inline struct c4iw_cq *get_chp(struct c4iw_dev *rhp, u32 cqid) +{ + return idr_find(&rhp->cqidr, cqid); +} + +static inline struct c4iw_qp *get_qhp(struct c4iw_dev *rhp, u32 qpid) +{ + return idr_find(&rhp->qpidr, qpid); +} + +static inline struct c4iw_mr *get_mhp(struct c4iw_dev *rhp, u32 mmid) +{ + return idr_find(&rhp->mmidr, mmid); +} + +static inline int _insert_handle(struct c4iw_dev *rhp, struct idr *idr, + void *handle, u32 id, int lock) +{ + int ret; + + if (lock) { + idr_preload(GFP_KERNEL); + spin_lock_irq(&rhp->lock); + } + + ret = idr_alloc(idr, handle, id, id + 1, GFP_ATOMIC); + + if (lock) { + spin_unlock_irq(&rhp->lock); + idr_preload_end(); + } + + BUG_ON(ret == -ENOSPC); + return ret < 0 ? ret : 0; +} + +static inline int insert_handle(struct c4iw_dev *rhp, struct idr *idr, + void *handle, u32 id) +{ + return _insert_handle(rhp, idr, handle, id, 1); +} + +static inline int insert_handle_nolock(struct c4iw_dev *rhp, struct idr *idr, + void *handle, u32 id) +{ + return _insert_handle(rhp, idr, handle, id, 0); +} + +static inline void _remove_handle(struct c4iw_dev *rhp, struct idr *idr, + u32 id, int lock) +{ + if (lock) + spin_lock_irq(&rhp->lock); + idr_remove(idr, id); + if (lock) + spin_unlock_irq(&rhp->lock); +} + +static inline void remove_handle(struct c4iw_dev *rhp, struct idr *idr, u32 id) +{ + _remove_handle(rhp, idr, id, 1); +} + +static inline void remove_handle_nolock(struct c4iw_dev *rhp, + struct idr *idr, u32 id) +{ + _remove_handle(rhp, idr, id, 0); +} + +extern uint c4iw_max_read_depth; + +static inline int cur_max_read_depth(struct c4iw_dev *dev) +{ + return min(dev->rdev.lldi.max_ordird_qp, c4iw_max_read_depth); +} + +struct c4iw_pd { + struct ib_pd ibpd; + u32 pdid; + struct c4iw_dev *rhp; +}; + +static inline struct c4iw_pd *to_c4iw_pd(struct ib_pd *ibpd) +{ + return container_of(ibpd, struct c4iw_pd, ibpd); +} + +struct tpt_attributes { + u64 len; + u64 va_fbo; + enum fw_ri_mem_perms perms; + u32 stag; + u32 pdid; + u32 qpid; + u32 pbl_addr; + u32 pbl_size; + u32 state:1; + u32 type:2; + u32 rsvd:1; + u32 remote_invaliate_disable:1; + u32 zbva:1; + u32 mw_bind_enable:1; + u32 page_size:5; +}; + +struct c4iw_mr { + struct ib_mr ibmr; + struct ib_umem *umem; + struct c4iw_dev *rhp; + u64 kva; + struct tpt_attributes attr; +}; + +static inline struct c4iw_mr *to_c4iw_mr(struct ib_mr *ibmr) +{ + return container_of(ibmr, struct c4iw_mr, ibmr); +} + +struct c4iw_mw { + struct ib_mw ibmw; + struct c4iw_dev *rhp; + u64 kva; + struct tpt_attributes attr; +}; + +static inline struct c4iw_mw *to_c4iw_mw(struct ib_mw *ibmw) +{ + return container_of(ibmw, struct c4iw_mw, ibmw); +} + +struct c4iw_fr_page_list { + struct ib_fast_reg_page_list ibpl; + DEFINE_DMA_UNMAP_ADDR(mapping); + dma_addr_t dma_addr; + struct c4iw_dev *dev; + int pll_len; +}; + +static inline struct c4iw_fr_page_list *to_c4iw_fr_page_list( + struct ib_fast_reg_page_list *ibpl) +{ + return container_of(ibpl, struct c4iw_fr_page_list, ibpl); +} + +struct c4iw_cq { + struct ib_cq ibcq; + struct c4iw_dev *rhp; + struct t4_cq cq; + spinlock_t lock; + spinlock_t comp_handler_lock; + atomic_t refcnt; + wait_queue_head_t wait; +}; + +static inline struct c4iw_cq *to_c4iw_cq(struct ib_cq *ibcq) +{ + return container_of(ibcq, struct c4iw_cq, ibcq); +} + +struct c4iw_mpa_attributes { + u8 initiator; + u8 recv_marker_enabled; + u8 xmit_marker_enabled; + u8 crc_enabled; + u8 enhanced_rdma_conn; + u8 version; + u8 p2p_type; +}; + +struct c4iw_qp_attributes { + u32 scq; + u32 rcq; + u32 sq_num_entries; + u32 rq_num_entries; + u32 sq_max_sges; + u32 sq_max_sges_rdma_write; + u32 rq_max_sges; + u32 state; + u8 enable_rdma_read; + u8 enable_rdma_write; + u8 enable_bind; + u8 enable_mmid0_fastreg; + u32 max_ord; + u32 max_ird; + u32 pd; + u32 next_state; + char terminate_buffer[52]; + u32 terminate_msg_len; + u8 is_terminate_local; + struct c4iw_mpa_attributes mpa_attr; + struct c4iw_ep *llp_stream_handle; + u8 layer_etype; + u8 ecode; + u16 sq_db_inc; + u16 rq_db_inc; + u8 send_term; +}; + +struct c4iw_qp { + struct ib_qp ibqp; + struct list_head db_fc_entry; + struct c4iw_dev *rhp; + struct c4iw_ep *ep; + struct c4iw_qp_attributes attr; + struct t4_wq wq; + spinlock_t lock; + struct mutex mutex; + atomic_t refcnt; + wait_queue_head_t wait; + struct timer_list timer; + int sq_sig_all; +}; + +static inline struct c4iw_qp *to_c4iw_qp(struct ib_qp *ibqp) +{ + return container_of(ibqp, struct c4iw_qp, ibqp); +} + +struct c4iw_ucontext { + struct ib_ucontext ibucontext; + struct c4iw_dev_ucontext uctx; + u32 key; + spinlock_t mmap_lock; + struct list_head mmaps; +}; + +static inline struct c4iw_ucontext *to_c4iw_ucontext(struct ib_ucontext *c) +{ + return container_of(c, struct c4iw_ucontext, ibucontext); +} + +struct c4iw_mm_entry { + struct list_head entry; + u64 addr; + u32 key; + unsigned len; +}; + +static inline struct c4iw_mm_entry *remove_mmap(struct c4iw_ucontext *ucontext, + u32 key, unsigned len) +{ + struct list_head *pos, *nxt; + struct c4iw_mm_entry *mm; + + spin_lock(&ucontext->mmap_lock); + list_for_each_safe(pos, nxt, &ucontext->mmaps) { + + mm = list_entry(pos, struct c4iw_mm_entry, entry); + if (mm->key == key && mm->len == len) { + list_del_init(&mm->entry); + spin_unlock(&ucontext->mmap_lock); + PDBG("%s key 0x%x addr 0x%llx len %d\n", __func__, + key, (unsigned long long) mm->addr, mm->len); + return mm; + } + } + spin_unlock(&ucontext->mmap_lock); + return NULL; +} + +static inline void insert_mmap(struct c4iw_ucontext *ucontext, + struct c4iw_mm_entry *mm) +{ + spin_lock(&ucontext->mmap_lock); + PDBG("%s key 0x%x addr 0x%llx len %d\n", __func__, + mm->key, (unsigned long long) mm->addr, mm->len); + list_add_tail(&mm->entry, &ucontext->mmaps); + spin_unlock(&ucontext->mmap_lock); +} + +enum c4iw_qp_attr_mask { + C4IW_QP_ATTR_NEXT_STATE = 1 << 0, + C4IW_QP_ATTR_SQ_DB = 1<<1, + C4IW_QP_ATTR_RQ_DB = 1<<2, + C4IW_QP_ATTR_ENABLE_RDMA_READ = 1 << 7, + C4IW_QP_ATTR_ENABLE_RDMA_WRITE = 1 << 8, + C4IW_QP_ATTR_ENABLE_RDMA_BIND = 1 << 9, + C4IW_QP_ATTR_MAX_ORD = 1 << 11, + C4IW_QP_ATTR_MAX_IRD = 1 << 12, + C4IW_QP_ATTR_LLP_STREAM_HANDLE = 1 << 22, + C4IW_QP_ATTR_STREAM_MSG_BUFFER = 1 << 23, + C4IW_QP_ATTR_MPA_ATTR = 1 << 24, + C4IW_QP_ATTR_QP_CONTEXT_ACTIVATE = 1 << 25, + C4IW_QP_ATTR_VALID_MODIFY = (C4IW_QP_ATTR_ENABLE_RDMA_READ | + C4IW_QP_ATTR_ENABLE_RDMA_WRITE | + C4IW_QP_ATTR_MAX_ORD | + C4IW_QP_ATTR_MAX_IRD | + C4IW_QP_ATTR_LLP_STREAM_HANDLE | + C4IW_QP_ATTR_STREAM_MSG_BUFFER | + C4IW_QP_ATTR_MPA_ATTR | + C4IW_QP_ATTR_QP_CONTEXT_ACTIVATE) +}; + +int c4iw_modify_qp(struct c4iw_dev *rhp, + struct c4iw_qp *qhp, + enum c4iw_qp_attr_mask mask, + struct c4iw_qp_attributes *attrs, + int internal); + +enum c4iw_qp_state { + C4IW_QP_STATE_IDLE, + C4IW_QP_STATE_RTS, + C4IW_QP_STATE_ERROR, + C4IW_QP_STATE_TERMINATE, + C4IW_QP_STATE_CLOSING, + C4IW_QP_STATE_TOT +}; + +static inline int c4iw_convert_state(enum ib_qp_state ib_state) +{ + switch (ib_state) { + case IB_QPS_RESET: + case IB_QPS_INIT: + return C4IW_QP_STATE_IDLE; + case IB_QPS_RTS: + return C4IW_QP_STATE_RTS; + case IB_QPS_SQD: + return C4IW_QP_STATE_CLOSING; + case IB_QPS_SQE: + return C4IW_QP_STATE_TERMINATE; + case IB_QPS_ERR: + return C4IW_QP_STATE_ERROR; + default: + return -1; + } +} + +static inline int to_ib_qp_state(int c4iw_qp_state) +{ + switch (c4iw_qp_state) { + case C4IW_QP_STATE_IDLE: + return IB_QPS_INIT; + case C4IW_QP_STATE_RTS: + return IB_QPS_RTS; + case C4IW_QP_STATE_CLOSING: + return IB_QPS_SQD; + case C4IW_QP_STATE_TERMINATE: + return IB_QPS_SQE; + case C4IW_QP_STATE_ERROR: + return IB_QPS_ERR; + } + return IB_QPS_ERR; +} + +static inline u32 c4iw_ib_to_tpt_access(int a) +{ + return (a & IB_ACCESS_REMOTE_WRITE ? FW_RI_MEM_ACCESS_REM_WRITE : 0) | + (a & IB_ACCESS_REMOTE_READ ? FW_RI_MEM_ACCESS_REM_READ : 0) | + (a & IB_ACCESS_LOCAL_WRITE ? FW_RI_MEM_ACCESS_LOCAL_WRITE : 0) | + FW_RI_MEM_ACCESS_LOCAL_READ; +} + +static inline u32 c4iw_ib_to_tpt_bind_access(int acc) +{ + return (acc & IB_ACCESS_REMOTE_WRITE ? FW_RI_MEM_ACCESS_REM_WRITE : 0) | + (acc & IB_ACCESS_REMOTE_READ ? FW_RI_MEM_ACCESS_REM_READ : 0); +} + +enum c4iw_mmid_state { + C4IW_STAG_STATE_VALID, + C4IW_STAG_STATE_INVALID +}; + +#define C4IW_NODE_DESC "cxgb4 Chelsio Communications" + +#define MPA_KEY_REQ "MPA ID Req Frame" +#define MPA_KEY_REP "MPA ID Rep Frame" + +#define MPA_MAX_PRIVATE_DATA 256 +#define MPA_ENHANCED_RDMA_CONN 0x10 +#define MPA_REJECT 0x20 +#define MPA_CRC 0x40 +#define MPA_MARKERS 0x80 +#define MPA_FLAGS_MASK 0xE0 + +#define MPA_V2_PEER2PEER_MODEL 0x8000 +#define MPA_V2_ZERO_LEN_FPDU_RTR 0x4000 +#define MPA_V2_RDMA_WRITE_RTR 0x8000 +#define MPA_V2_RDMA_READ_RTR 0x4000 +#define MPA_V2_IRD_ORD_MASK 0x3FFF + +#define c4iw_put_ep(ep) { \ + PDBG("put_ep (via %s:%u) ep %p refcnt %d\n", __func__, __LINE__, \ + ep, atomic_read(&((ep)->kref.refcount))); \ + WARN_ON(atomic_read(&((ep)->kref.refcount)) < 1); \ + kref_put(&((ep)->kref), _c4iw_free_ep); \ +} + +#define c4iw_get_ep(ep) { \ + PDBG("get_ep (via %s:%u) ep %p, refcnt %d\n", __func__, __LINE__, \ + ep, atomic_read(&((ep)->kref.refcount))); \ + kref_get(&((ep)->kref)); \ +} +void _c4iw_free_ep(struct kref *kref); + +struct mpa_message { + u8 key[16]; + u8 flags; + u8 revision; + __be16 private_data_size; + u8 private_data[0]; +}; + +struct mpa_v2_conn_params { + __be16 ird; + __be16 ord; +}; + +struct terminate_message { + u8 layer_etype; + u8 ecode; + __be16 hdrct_rsvd; + u8 len_hdrs[0]; +}; + +#define TERM_MAX_LENGTH (sizeof(struct terminate_message) + 2 + 18 + 28) + +enum c4iw_layers_types { + LAYER_RDMAP = 0x00, + LAYER_DDP = 0x10, + LAYER_MPA = 0x20, + RDMAP_LOCAL_CATA = 0x00, + RDMAP_REMOTE_PROT = 0x01, + RDMAP_REMOTE_OP = 0x02, + DDP_LOCAL_CATA = 0x00, + DDP_TAGGED_ERR = 0x01, + DDP_UNTAGGED_ERR = 0x02, + DDP_LLP = 0x03 +}; + +enum c4iw_rdma_ecodes { + RDMAP_INV_STAG = 0x00, + RDMAP_BASE_BOUNDS = 0x01, + RDMAP_ACC_VIOL = 0x02, + RDMAP_STAG_NOT_ASSOC = 0x03, + RDMAP_TO_WRAP = 0x04, + RDMAP_INV_VERS = 0x05, + RDMAP_INV_OPCODE = 0x06, + RDMAP_STREAM_CATA = 0x07, + RDMAP_GLOBAL_CATA = 0x08, + RDMAP_CANT_INV_STAG = 0x09, + RDMAP_UNSPECIFIED = 0xff +}; + +enum c4iw_ddp_ecodes { + DDPT_INV_STAG = 0x00, + DDPT_BASE_BOUNDS = 0x01, + DDPT_STAG_NOT_ASSOC = 0x02, + DDPT_TO_WRAP = 0x03, + DDPT_INV_VERS = 0x04, + DDPU_INV_QN = 0x01, + DDPU_INV_MSN_NOBUF = 0x02, + DDPU_INV_MSN_RANGE = 0x03, + DDPU_INV_MO = 0x04, + DDPU_MSG_TOOBIG = 0x05, + DDPU_INV_VERS = 0x06 +}; + +enum c4iw_mpa_ecodes { + MPA_CRC_ERR = 0x02, + MPA_MARKER_ERR = 0x03, + MPA_LOCAL_CATA = 0x05, + MPA_INSUFF_IRD = 0x06, + MPA_NOMATCH_RTR = 0x07, +}; + +enum c4iw_ep_state { + IDLE = 0, + LISTEN, + CONNECTING, + MPA_REQ_WAIT, + MPA_REQ_SENT, + MPA_REQ_RCVD, + MPA_REP_SENT, + FPDU_MODE, + ABORTING, + CLOSING, + MORIBUND, + DEAD, +}; + +enum c4iw_ep_flags { + PEER_ABORT_IN_PROGRESS = 0, + ABORT_REQ_IN_PROGRESS = 1, + RELEASE_RESOURCES = 2, + CLOSE_SENT = 3, + TIMEOUT = 4, + QP_REFERENCED = 5, + RELEASE_MAPINFO = 6, +}; + +enum c4iw_ep_history { + ACT_OPEN_REQ = 0, + ACT_OFLD_CONN = 1, + ACT_OPEN_RPL = 2, + ACT_ESTAB = 3, + PASS_ACCEPT_REQ = 4, + PASS_ESTAB = 5, + ABORT_UPCALL = 6, + ESTAB_UPCALL = 7, + CLOSE_UPCALL = 8, + ULP_ACCEPT = 9, + ULP_REJECT = 10, + TIMEDOUT = 11, + PEER_ABORT = 12, + PEER_CLOSE = 13, + CONNREQ_UPCALL = 14, + ABORT_CONN = 15, + DISCONN_UPCALL = 16, + EP_DISC_CLOSE = 17, + EP_DISC_ABORT = 18, + CONN_RPL_UPCALL = 19, + ACT_RETRY_NOMEM = 20, + ACT_RETRY_INUSE = 21 +}; + +struct c4iw_ep_common { + struct iw_cm_id *cm_id; + struct c4iw_qp *qp; + struct c4iw_dev *dev; + enum c4iw_ep_state state; + struct kref kref; + struct mutex mutex; + struct sockaddr_storage local_addr; + struct sockaddr_storage remote_addr; + struct sockaddr_storage mapped_local_addr; + struct sockaddr_storage mapped_remote_addr; + struct c4iw_wr_wait wr_wait; + unsigned long flags; + unsigned long history; +}; + +struct c4iw_listen_ep { + struct c4iw_ep_common com; + unsigned int stid; + int backlog; +}; + +struct c4iw_ep_stats { + unsigned connect_neg_adv; + unsigned abort_neg_adv; +}; + +struct c4iw_ep { + struct c4iw_ep_common com; + struct c4iw_ep *parent_ep; + struct timer_list timer; + struct list_head entry; + unsigned int atid; + u32 hwtid; + u32 snd_seq; + u32 rcv_seq; + struct l2t_entry *l2t; + struct dst_entry *dst; + struct sk_buff *mpa_skb; + struct c4iw_mpa_attributes mpa_attr; + u8 mpa_pkt[sizeof(struct mpa_message) + MPA_MAX_PRIVATE_DATA]; + unsigned int mpa_pkt_len; + u32 ird; + u32 ord; + u32 smac_idx; + u32 tx_chan; + u32 mtu; + u16 mss; + u16 emss; + u16 plen; + u16 rss_qid; + u16 txq_idx; + u16 ctrlq_idx; + u8 tos; + u8 retry_with_mpa_v1; + u8 tried_with_mpa_v1; + unsigned int retry_count; + int snd_win; + int rcv_win; + struct c4iw_ep_stats stats; +}; + +static inline void print_addr(struct c4iw_ep_common *epc, const char *func, + const char *msg) +{ + +#define SINA(a) (&(((struct sockaddr_in *)(a))->sin_addr.s_addr)) +#define SINP(a) ntohs(((struct sockaddr_in *)(a))->sin_port) +#define SIN6A(a) (&(((struct sockaddr_in6 *)(a))->sin6_addr)) +#define SIN6P(a) ntohs(((struct sockaddr_in6 *)(a))->sin6_port) + + if (c4iw_debug) { + switch (epc->local_addr.ss_family) { + case AF_INET: + PDBG("%s %s %pI4:%u/%u <-> %pI4:%u/%u\n", + func, msg, SINA(&epc->local_addr), + SINP(&epc->local_addr), + SINP(&epc->mapped_local_addr), + SINA(&epc->remote_addr), + SINP(&epc->remote_addr), + SINP(&epc->mapped_remote_addr)); + break; + case AF_INET6: + PDBG("%s %s %pI6:%u/%u <-> %pI6:%u/%u\n", + func, msg, SIN6A(&epc->local_addr), + SIN6P(&epc->local_addr), + SIN6P(&epc->mapped_local_addr), + SIN6A(&epc->remote_addr), + SIN6P(&epc->remote_addr), + SIN6P(&epc->mapped_remote_addr)); + break; + default: + break; + } + } +#undef SINA +#undef SINP +#undef SIN6A +#undef SIN6P +} + +static inline struct c4iw_ep *to_ep(struct iw_cm_id *cm_id) +{ + return cm_id->provider_data; +} + +static inline struct c4iw_listen_ep *to_listen_ep(struct iw_cm_id *cm_id) +{ + return cm_id->provider_data; +} + +static inline int compute_wscale(int win) +{ + int wscale = 0; + + while (wscale < 14 && (65535<<wscale) < win) + wscale++; + return wscale; +} + +static inline int ocqp_supported(const struct cxgb4_lld_info *infop) +{ +#if defined(__i386__) || defined(__x86_64__) || defined(CONFIG_PPC64) + return infop->vr->ocq.size > 0; +#else + return 0; +#endif +} + +u32 c4iw_id_alloc(struct c4iw_id_table *alloc); +void c4iw_id_free(struct c4iw_id_table *alloc, u32 obj); +int c4iw_id_table_alloc(struct c4iw_id_table *alloc, u32 start, u32 num, + u32 reserved, u32 flags); +void c4iw_id_table_free(struct c4iw_id_table *alloc); + +typedef int (*c4iw_handler_func)(struct c4iw_dev *dev, struct sk_buff *skb); + +int c4iw_ep_redirect(void *ctx, struct dst_entry *old, struct dst_entry *new, + struct l2t_entry *l2t); +void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qpid, + struct c4iw_dev_ucontext *uctx); +u32 c4iw_get_resource(struct c4iw_id_table *id_table); +void c4iw_put_resource(struct c4iw_id_table *id_table, u32 entry); +int c4iw_init_resource(struct c4iw_rdev *rdev, u32 nr_tpt, u32 nr_pdid); +int c4iw_init_ctrl_qp(struct c4iw_rdev *rdev); +int c4iw_pblpool_create(struct c4iw_rdev *rdev); +int c4iw_rqtpool_create(struct c4iw_rdev *rdev); +int c4iw_ocqp_pool_create(struct c4iw_rdev *rdev); +void c4iw_pblpool_destroy(struct c4iw_rdev *rdev); +void c4iw_rqtpool_destroy(struct c4iw_rdev *rdev); +void c4iw_ocqp_pool_destroy(struct c4iw_rdev *rdev); +void c4iw_destroy_resource(struct c4iw_resource *rscp); +int c4iw_destroy_ctrl_qp(struct c4iw_rdev *rdev); +int c4iw_register_device(struct c4iw_dev *dev); +void c4iw_unregister_device(struct c4iw_dev *dev); +int __init c4iw_cm_init(void); +void c4iw_cm_term(void); +void c4iw_release_dev_ucontext(struct c4iw_rdev *rdev, + struct c4iw_dev_ucontext *uctx); +void c4iw_init_dev_ucontext(struct c4iw_rdev *rdev, + struct c4iw_dev_ucontext *uctx); +int c4iw_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc); +int c4iw_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, + struct ib_send_wr **bad_wr); +int c4iw_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr, + struct ib_recv_wr **bad_wr); +int c4iw_bind_mw(struct ib_qp *qp, struct ib_mw *mw, + struct ib_mw_bind *mw_bind); +int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param); +int c4iw_create_listen(struct iw_cm_id *cm_id, int backlog); +int c4iw_destroy_listen(struct iw_cm_id *cm_id); +int c4iw_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param); +int c4iw_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len); +void c4iw_qp_add_ref(struct ib_qp *qp); +void c4iw_qp_rem_ref(struct ib_qp *qp); +void c4iw_free_fastreg_pbl(struct ib_fast_reg_page_list *page_list); +struct ib_fast_reg_page_list *c4iw_alloc_fastreg_pbl( + struct ib_device *device, + int page_list_len); +struct ib_mr *c4iw_alloc_fast_reg_mr(struct ib_pd *pd, int pbl_depth); +int c4iw_dealloc_mw(struct ib_mw *mw); +struct ib_mw *c4iw_alloc_mw(struct ib_pd *pd, enum ib_mw_type type); +struct ib_mr *c4iw_reg_user_mr(struct ib_pd *pd, u64 start, + u64 length, u64 virt, int acc, + struct ib_udata *udata); +struct ib_mr *c4iw_get_dma_mr(struct ib_pd *pd, int acc); +struct ib_mr *c4iw_register_phys_mem(struct ib_pd *pd, + struct ib_phys_buf *buffer_list, + int num_phys_buf, + int acc, + u64 *iova_start); +int c4iw_reregister_phys_mem(struct ib_mr *mr, + int mr_rereg_mask, + struct ib_pd *pd, + struct ib_phys_buf *buffer_list, + int num_phys_buf, + int acc, u64 *iova_start); +int c4iw_dereg_mr(struct ib_mr *ib_mr); +int c4iw_destroy_cq(struct ib_cq *ib_cq); +struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, int entries, + int vector, + struct ib_ucontext *ib_context, + struct ib_udata *udata); +int c4iw_resize_cq(struct ib_cq *cq, int cqe, struct ib_udata *udata); +int c4iw_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags); +int c4iw_destroy_qp(struct ib_qp *ib_qp); +struct ib_qp *c4iw_create_qp(struct ib_pd *pd, + struct ib_qp_init_attr *attrs, + struct ib_udata *udata); +int c4iw_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, + int attr_mask, struct ib_udata *udata); +int c4iw_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, + int attr_mask, struct ib_qp_init_attr *init_attr); +struct ib_qp *c4iw_get_qp(struct ib_device *dev, int qpn); +u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size); +void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size); +u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size); +void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size); +u32 c4iw_ocqp_pool_alloc(struct c4iw_rdev *rdev, int size); +void c4iw_ocqp_pool_free(struct c4iw_rdev *rdev, u32 addr, int size); +int c4iw_ofld_send(struct c4iw_rdev *rdev, struct sk_buff *skb); +void c4iw_flush_hw_cq(struct c4iw_cq *chp); +void c4iw_count_rcqes(struct t4_cq *cq, struct t4_wq *wq, int *count); +int c4iw_ep_disconnect(struct c4iw_ep *ep, int abrupt, gfp_t gfp); +int c4iw_flush_rq(struct t4_wq *wq, struct t4_cq *cq, int count); +int c4iw_flush_sq(struct c4iw_qp *qhp); +int c4iw_ev_handler(struct c4iw_dev *rnicp, u32 qid); +u16 c4iw_rqes_posted(struct c4iw_qp *qhp); +int c4iw_post_terminate(struct c4iw_qp *qhp, struct t4_cqe *err_cqe); +u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx); +void c4iw_put_cqid(struct c4iw_rdev *rdev, u32 qid, + struct c4iw_dev_ucontext *uctx); +u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx); +void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qid, + struct c4iw_dev_ucontext *uctx); +void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe); + +extern struct cxgb4_client t4c_client; +extern c4iw_handler_func c4iw_handlers[NUM_CPL_CMDS]; +extern void c4iw_log_wr_stats(struct t4_wq *wq, struct t4_cqe *cqe); +extern int c4iw_wr_log; +extern int db_fc_threshold; +extern int db_coalescing_threshold; +extern int use_dsgl; + + +#endif diff --git a/kernel/drivers/infiniband/hw/cxgb4/mem.c b/kernel/drivers/infiniband/hw/cxgb4/mem.c new file mode 100644 index 000000000..cff815b91 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/mem.c @@ -0,0 +1,979 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ + +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <rdma/ib_umem.h> +#include <linux/atomic.h> + +#include "iw_cxgb4.h" + +int use_dsgl = 0; +module_param(use_dsgl, int, 0644); +MODULE_PARM_DESC(use_dsgl, "Use DSGL for PBL/FastReg (default=0)"); + +#define T4_ULPTX_MIN_IO 32 +#define C4IW_MAX_INLINE_SIZE 96 +#define T4_ULPTX_MAX_DMA 1024 +#define C4IW_INLINE_THRESHOLD 128 + +static int inline_threshold = C4IW_INLINE_THRESHOLD; +module_param(inline_threshold, int, 0644); +MODULE_PARM_DESC(inline_threshold, "inline vs dsgl threshold (default=128)"); + +static int mr_exceeds_hw_limits(struct c4iw_dev *dev, u64 length) +{ + return (is_t4(dev->rdev.lldi.adapter_type) || + is_t5(dev->rdev.lldi.adapter_type)) && + length >= 8*1024*1024*1024ULL; +} + +static int _c4iw_write_mem_dma_aligned(struct c4iw_rdev *rdev, u32 addr, + u32 len, dma_addr_t data, int wait) +{ + struct sk_buff *skb; + struct ulp_mem_io *req; + struct ulptx_sgl *sgl; + u8 wr_len; + int ret = 0; + struct c4iw_wr_wait wr_wait; + + addr &= 0x7FFFFFF; + + if (wait) + c4iw_init_wr_wait(&wr_wait); + wr_len = roundup(sizeof(*req) + sizeof(*sgl), 16); + + skb = alloc_skb(wr_len, GFP_KERNEL); + if (!skb) + return -ENOMEM; + set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); + + req = (struct ulp_mem_io *)__skb_put(skb, wr_len); + memset(req, 0, wr_len); + INIT_ULPTX_WR(req, wr_len, 0, 0); + req->wr.wr_hi = cpu_to_be32(FW_WR_OP_V(FW_ULPTX_WR) | + (wait ? FW_WR_COMPL_F : 0)); + req->wr.wr_lo = wait ? (__force __be64)(unsigned long) &wr_wait : 0L; + req->wr.wr_mid = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(wr_len, 16))); + req->cmd = cpu_to_be32(ULPTX_CMD_V(ULP_TX_MEM_WRITE)); + req->cmd |= cpu_to_be32(T5_ULP_MEMIO_ORDER_V(1)); + req->dlen = cpu_to_be32(ULP_MEMIO_DATA_LEN_V(len>>5)); + req->len16 = cpu_to_be32(DIV_ROUND_UP(wr_len-sizeof(req->wr), 16)); + req->lock_addr = cpu_to_be32(ULP_MEMIO_ADDR_V(addr)); + + sgl = (struct ulptx_sgl *)(req + 1); + sgl->cmd_nsge = cpu_to_be32(ULPTX_CMD_V(ULP_TX_SC_DSGL) | + ULPTX_NSGE_V(1)); + sgl->len0 = cpu_to_be32(len); + sgl->addr0 = cpu_to_be64(data); + + ret = c4iw_ofld_send(rdev, skb); + if (ret) + return ret; + if (wait) + ret = c4iw_wait_for_reply(rdev, &wr_wait, 0, 0, __func__); + return ret; +} + +static int _c4iw_write_mem_inline(struct c4iw_rdev *rdev, u32 addr, u32 len, + void *data) +{ + struct sk_buff *skb; + struct ulp_mem_io *req; + struct ulptx_idata *sc; + u8 wr_len, *to_dp, *from_dp; + int copy_len, num_wqe, i, ret = 0; + struct c4iw_wr_wait wr_wait; + __be32 cmd = cpu_to_be32(ULPTX_CMD_V(ULP_TX_MEM_WRITE)); + + if (is_t4(rdev->lldi.adapter_type)) + cmd |= cpu_to_be32(ULP_MEMIO_ORDER_F); + else + cmd |= cpu_to_be32(T5_ULP_MEMIO_IMM_F); + + addr &= 0x7FFFFFF; + PDBG("%s addr 0x%x len %u\n", __func__, addr, len); + num_wqe = DIV_ROUND_UP(len, C4IW_MAX_INLINE_SIZE); + c4iw_init_wr_wait(&wr_wait); + for (i = 0; i < num_wqe; i++) { + + copy_len = len > C4IW_MAX_INLINE_SIZE ? C4IW_MAX_INLINE_SIZE : + len; + wr_len = roundup(sizeof *req + sizeof *sc + + roundup(copy_len, T4_ULPTX_MIN_IO), 16); + + skb = alloc_skb(wr_len, GFP_KERNEL); + if (!skb) + return -ENOMEM; + set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); + + req = (struct ulp_mem_io *)__skb_put(skb, wr_len); + memset(req, 0, wr_len); + INIT_ULPTX_WR(req, wr_len, 0, 0); + + if (i == (num_wqe-1)) { + req->wr.wr_hi = cpu_to_be32(FW_WR_OP_V(FW_ULPTX_WR) | + FW_WR_COMPL_F); + req->wr.wr_lo = (__force __be64)&wr_wait; + } else + req->wr.wr_hi = cpu_to_be32(FW_WR_OP_V(FW_ULPTX_WR)); + req->wr.wr_mid = cpu_to_be32( + FW_WR_LEN16_V(DIV_ROUND_UP(wr_len, 16))); + + req->cmd = cmd; + req->dlen = cpu_to_be32(ULP_MEMIO_DATA_LEN_V( + DIV_ROUND_UP(copy_len, T4_ULPTX_MIN_IO))); + req->len16 = cpu_to_be32(DIV_ROUND_UP(wr_len-sizeof(req->wr), + 16)); + req->lock_addr = cpu_to_be32(ULP_MEMIO_ADDR_V(addr + i * 3)); + + sc = (struct ulptx_idata *)(req + 1); + sc->cmd_more = cpu_to_be32(ULPTX_CMD_V(ULP_TX_SC_IMM)); + sc->len = cpu_to_be32(roundup(copy_len, T4_ULPTX_MIN_IO)); + + to_dp = (u8 *)(sc + 1); + from_dp = (u8 *)data + i * C4IW_MAX_INLINE_SIZE; + if (data) + memcpy(to_dp, from_dp, copy_len); + else + memset(to_dp, 0, copy_len); + if (copy_len % T4_ULPTX_MIN_IO) + memset(to_dp + copy_len, 0, T4_ULPTX_MIN_IO - + (copy_len % T4_ULPTX_MIN_IO)); + ret = c4iw_ofld_send(rdev, skb); + if (ret) + return ret; + len -= C4IW_MAX_INLINE_SIZE; + } + + ret = c4iw_wait_for_reply(rdev, &wr_wait, 0, 0, __func__); + return ret; +} + +static int _c4iw_write_mem_dma(struct c4iw_rdev *rdev, u32 addr, u32 len, void *data) +{ + u32 remain = len; + u32 dmalen; + int ret = 0; + dma_addr_t daddr; + dma_addr_t save; + + daddr = dma_map_single(&rdev->lldi.pdev->dev, data, len, DMA_TO_DEVICE); + if (dma_mapping_error(&rdev->lldi.pdev->dev, daddr)) + return -1; + save = daddr; + + while (remain > inline_threshold) { + if (remain < T4_ULPTX_MAX_DMA) { + if (remain & ~T4_ULPTX_MIN_IO) + dmalen = remain & ~(T4_ULPTX_MIN_IO-1); + else + dmalen = remain; + } else + dmalen = T4_ULPTX_MAX_DMA; + remain -= dmalen; + ret = _c4iw_write_mem_dma_aligned(rdev, addr, dmalen, daddr, + !remain); + if (ret) + goto out; + addr += dmalen >> 5; + data += dmalen; + daddr += dmalen; + } + if (remain) + ret = _c4iw_write_mem_inline(rdev, addr, remain, data); +out: + dma_unmap_single(&rdev->lldi.pdev->dev, save, len, DMA_TO_DEVICE); + return ret; +} + +/* + * write len bytes of data into addr (32B aligned address) + * If data is NULL, clear len byte of memory to zero. + */ +static int write_adapter_mem(struct c4iw_rdev *rdev, u32 addr, u32 len, + void *data) +{ + if (is_t5(rdev->lldi.adapter_type) && use_dsgl) { + if (len > inline_threshold) { + if (_c4iw_write_mem_dma(rdev, addr, len, data)) { + printk_ratelimited(KERN_WARNING + "%s: dma map" + " failure (non fatal)\n", + pci_name(rdev->lldi.pdev)); + return _c4iw_write_mem_inline(rdev, addr, len, + data); + } else + return 0; + } else + return _c4iw_write_mem_inline(rdev, addr, len, data); + } else + return _c4iw_write_mem_inline(rdev, addr, len, data); +} + +/* + * Build and write a TPT entry. + * IN: stag key, pdid, perm, bind_enabled, zbva, to, len, page_size, + * pbl_size and pbl_addr + * OUT: stag index + */ +static int write_tpt_entry(struct c4iw_rdev *rdev, u32 reset_tpt_entry, + u32 *stag, u8 stag_state, u32 pdid, + enum fw_ri_stag_type type, enum fw_ri_mem_perms perm, + int bind_enabled, u32 zbva, u64 to, + u64 len, u8 page_size, u32 pbl_size, u32 pbl_addr) +{ + int err; + struct fw_ri_tpte tpt; + u32 stag_idx; + static atomic_t key; + + if (c4iw_fatal_error(rdev)) + return -EIO; + + stag_state = stag_state > 0; + stag_idx = (*stag) >> 8; + + if ((!reset_tpt_entry) && (*stag == T4_STAG_UNSET)) { + stag_idx = c4iw_get_resource(&rdev->resource.tpt_table); + if (!stag_idx) { + mutex_lock(&rdev->stats.lock); + rdev->stats.stag.fail++; + mutex_unlock(&rdev->stats.lock); + return -ENOMEM; + } + mutex_lock(&rdev->stats.lock); + rdev->stats.stag.cur += 32; + if (rdev->stats.stag.cur > rdev->stats.stag.max) + rdev->stats.stag.max = rdev->stats.stag.cur; + mutex_unlock(&rdev->stats.lock); + *stag = (stag_idx << 8) | (atomic_inc_return(&key) & 0xff); + } + PDBG("%s stag_state 0x%0x type 0x%0x pdid 0x%0x, stag_idx 0x%x\n", + __func__, stag_state, type, pdid, stag_idx); + + /* write TPT entry */ + if (reset_tpt_entry) + memset(&tpt, 0, sizeof(tpt)); + else { + tpt.valid_to_pdid = cpu_to_be32(FW_RI_TPTE_VALID_F | + FW_RI_TPTE_STAGKEY_V((*stag & FW_RI_TPTE_STAGKEY_M)) | + FW_RI_TPTE_STAGSTATE_V(stag_state) | + FW_RI_TPTE_STAGTYPE_V(type) | FW_RI_TPTE_PDID_V(pdid)); + tpt.locread_to_qpid = cpu_to_be32(FW_RI_TPTE_PERM_V(perm) | + (bind_enabled ? FW_RI_TPTE_MWBINDEN_F : 0) | + FW_RI_TPTE_ADDRTYPE_V((zbva ? FW_RI_ZERO_BASED_TO : + FW_RI_VA_BASED_TO))| + FW_RI_TPTE_PS_V(page_size)); + tpt.nosnoop_pbladdr = !pbl_size ? 0 : cpu_to_be32( + FW_RI_TPTE_PBLADDR_V(PBL_OFF(rdev, pbl_addr)>>3)); + tpt.len_lo = cpu_to_be32((u32)(len & 0xffffffffUL)); + tpt.va_hi = cpu_to_be32((u32)(to >> 32)); + tpt.va_lo_fbo = cpu_to_be32((u32)(to & 0xffffffffUL)); + tpt.dca_mwbcnt_pstag = cpu_to_be32(0); + tpt.len_hi = cpu_to_be32((u32)(len >> 32)); + } + err = write_adapter_mem(rdev, stag_idx + + (rdev->lldi.vr->stag.start >> 5), + sizeof(tpt), &tpt); + + if (reset_tpt_entry) { + c4iw_put_resource(&rdev->resource.tpt_table, stag_idx); + mutex_lock(&rdev->stats.lock); + rdev->stats.stag.cur -= 32; + mutex_unlock(&rdev->stats.lock); + } + return err; +} + +static int write_pbl(struct c4iw_rdev *rdev, __be64 *pbl, + u32 pbl_addr, u32 pbl_size) +{ + int err; + + PDBG("%s *pdb_addr 0x%x, pbl_base 0x%x, pbl_size %d\n", + __func__, pbl_addr, rdev->lldi.vr->pbl.start, + pbl_size); + + err = write_adapter_mem(rdev, pbl_addr >> 5, pbl_size << 3, pbl); + return err; +} + +static int dereg_mem(struct c4iw_rdev *rdev, u32 stag, u32 pbl_size, + u32 pbl_addr) +{ + return write_tpt_entry(rdev, 1, &stag, 0, 0, 0, 0, 0, 0, 0UL, 0, 0, + pbl_size, pbl_addr); +} + +static int allocate_window(struct c4iw_rdev *rdev, u32 * stag, u32 pdid) +{ + *stag = T4_STAG_UNSET; + return write_tpt_entry(rdev, 0, stag, 0, pdid, FW_RI_STAG_MW, 0, 0, 0, + 0UL, 0, 0, 0, 0); +} + +static int deallocate_window(struct c4iw_rdev *rdev, u32 stag) +{ + return write_tpt_entry(rdev, 1, &stag, 0, 0, 0, 0, 0, 0, 0UL, 0, 0, 0, + 0); +} + +static int allocate_stag(struct c4iw_rdev *rdev, u32 *stag, u32 pdid, + u32 pbl_size, u32 pbl_addr) +{ + *stag = T4_STAG_UNSET; + return write_tpt_entry(rdev, 0, stag, 0, pdid, FW_RI_STAG_NSMR, 0, 0, 0, + 0UL, 0, 0, pbl_size, pbl_addr); +} + +static int finish_mem_reg(struct c4iw_mr *mhp, u32 stag) +{ + u32 mmid; + + mhp->attr.state = 1; + mhp->attr.stag = stag; + mmid = stag >> 8; + mhp->ibmr.rkey = mhp->ibmr.lkey = stag; + PDBG("%s mmid 0x%x mhp %p\n", __func__, mmid, mhp); + return insert_handle(mhp->rhp, &mhp->rhp->mmidr, mhp, mmid); +} + +static int register_mem(struct c4iw_dev *rhp, struct c4iw_pd *php, + struct c4iw_mr *mhp, int shift) +{ + u32 stag = T4_STAG_UNSET; + int ret; + + ret = write_tpt_entry(&rhp->rdev, 0, &stag, 1, mhp->attr.pdid, + FW_RI_STAG_NSMR, mhp->attr.len ? + mhp->attr.perms : 0, + mhp->attr.mw_bind_enable, mhp->attr.zbva, + mhp->attr.va_fbo, mhp->attr.len ? + mhp->attr.len : -1, shift - 12, + mhp->attr.pbl_size, mhp->attr.pbl_addr); + if (ret) + return ret; + + ret = finish_mem_reg(mhp, stag); + if (ret) + dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, + mhp->attr.pbl_addr); + return ret; +} + +static int reregister_mem(struct c4iw_dev *rhp, struct c4iw_pd *php, + struct c4iw_mr *mhp, int shift, int npages) +{ + u32 stag; + int ret; + + if (npages > mhp->attr.pbl_size) + return -ENOMEM; + + stag = mhp->attr.stag; + ret = write_tpt_entry(&rhp->rdev, 0, &stag, 1, mhp->attr.pdid, + FW_RI_STAG_NSMR, mhp->attr.perms, + mhp->attr.mw_bind_enable, mhp->attr.zbva, + mhp->attr.va_fbo, mhp->attr.len, shift - 12, + mhp->attr.pbl_size, mhp->attr.pbl_addr); + if (ret) + return ret; + + ret = finish_mem_reg(mhp, stag); + if (ret) + dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, + mhp->attr.pbl_addr); + + return ret; +} + +static int alloc_pbl(struct c4iw_mr *mhp, int npages) +{ + mhp->attr.pbl_addr = c4iw_pblpool_alloc(&mhp->rhp->rdev, + npages << 3); + + if (!mhp->attr.pbl_addr) + return -ENOMEM; + + mhp->attr.pbl_size = npages; + + return 0; +} + +static int build_phys_page_list(struct ib_phys_buf *buffer_list, + int num_phys_buf, u64 *iova_start, + u64 *total_size, int *npages, + int *shift, __be64 **page_list) +{ + u64 mask; + int i, j, n; + + mask = 0; + *total_size = 0; + for (i = 0; i < num_phys_buf; ++i) { + if (i != 0 && buffer_list[i].addr & ~PAGE_MASK) + return -EINVAL; + if (i != 0 && i != num_phys_buf - 1 && + (buffer_list[i].size & ~PAGE_MASK)) + return -EINVAL; + *total_size += buffer_list[i].size; + if (i > 0) + mask |= buffer_list[i].addr; + else + mask |= buffer_list[i].addr & PAGE_MASK; + if (i != num_phys_buf - 1) + mask |= buffer_list[i].addr + buffer_list[i].size; + else + mask |= (buffer_list[i].addr + buffer_list[i].size + + PAGE_SIZE - 1) & PAGE_MASK; + } + + if (*total_size > 0xFFFFFFFFULL) + return -ENOMEM; + + /* Find largest page shift we can use to cover buffers */ + for (*shift = PAGE_SHIFT; *shift < 27; ++(*shift)) + if ((1ULL << *shift) & mask) + break; + + buffer_list[0].size += buffer_list[0].addr & ((1ULL << *shift) - 1); + buffer_list[0].addr &= ~0ull << *shift; + + *npages = 0; + for (i = 0; i < num_phys_buf; ++i) + *npages += (buffer_list[i].size + + (1ULL << *shift) - 1) >> *shift; + + if (!*npages) + return -EINVAL; + + *page_list = kmalloc(sizeof(u64) * *npages, GFP_KERNEL); + if (!*page_list) + return -ENOMEM; + + n = 0; + for (i = 0; i < num_phys_buf; ++i) + for (j = 0; + j < (buffer_list[i].size + (1ULL << *shift) - 1) >> *shift; + ++j) + (*page_list)[n++] = cpu_to_be64(buffer_list[i].addr + + ((u64) j << *shift)); + + PDBG("%s va 0x%llx mask 0x%llx shift %d len %lld pbl_size %d\n", + __func__, (unsigned long long)*iova_start, + (unsigned long long)mask, *shift, (unsigned long long)*total_size, + *npages); + + return 0; + +} + +int c4iw_reregister_phys_mem(struct ib_mr *mr, int mr_rereg_mask, + struct ib_pd *pd, struct ib_phys_buf *buffer_list, + int num_phys_buf, int acc, u64 *iova_start) +{ + + struct c4iw_mr mh, *mhp; + struct c4iw_pd *php; + struct c4iw_dev *rhp; + __be64 *page_list = NULL; + int shift = 0; + u64 total_size; + int npages; + int ret; + + PDBG("%s ib_mr %p ib_pd %p\n", __func__, mr, pd); + + /* There can be no memory windows */ + if (atomic_read(&mr->usecnt)) + return -EINVAL; + + mhp = to_c4iw_mr(mr); + rhp = mhp->rhp; + php = to_c4iw_pd(mr->pd); + + /* make sure we are on the same adapter */ + if (rhp != php->rhp) + return -EINVAL; + + memcpy(&mh, mhp, sizeof *mhp); + + if (mr_rereg_mask & IB_MR_REREG_PD) + php = to_c4iw_pd(pd); + if (mr_rereg_mask & IB_MR_REREG_ACCESS) { + mh.attr.perms = c4iw_ib_to_tpt_access(acc); + mh.attr.mw_bind_enable = (acc & IB_ACCESS_MW_BIND) == + IB_ACCESS_MW_BIND; + } + if (mr_rereg_mask & IB_MR_REREG_TRANS) { + ret = build_phys_page_list(buffer_list, num_phys_buf, + iova_start, + &total_size, &npages, + &shift, &page_list); + if (ret) + return ret; + } + + if (mr_exceeds_hw_limits(rhp, total_size)) { + kfree(page_list); + return -EINVAL; + } + + ret = reregister_mem(rhp, php, &mh, shift, npages); + kfree(page_list); + if (ret) + return ret; + if (mr_rereg_mask & IB_MR_REREG_PD) + mhp->attr.pdid = php->pdid; + if (mr_rereg_mask & IB_MR_REREG_ACCESS) + mhp->attr.perms = c4iw_ib_to_tpt_access(acc); + if (mr_rereg_mask & IB_MR_REREG_TRANS) { + mhp->attr.zbva = 0; + mhp->attr.va_fbo = *iova_start; + mhp->attr.page_size = shift - 12; + mhp->attr.len = (u32) total_size; + mhp->attr.pbl_size = npages; + } + + return 0; +} + +struct ib_mr *c4iw_register_phys_mem(struct ib_pd *pd, + struct ib_phys_buf *buffer_list, + int num_phys_buf, int acc, u64 *iova_start) +{ + __be64 *page_list; + int shift; + u64 total_size; + int npages; + struct c4iw_dev *rhp; + struct c4iw_pd *php; + struct c4iw_mr *mhp; + int ret; + + PDBG("%s ib_pd %p\n", __func__, pd); + php = to_c4iw_pd(pd); + rhp = php->rhp; + + mhp = kzalloc(sizeof(*mhp), GFP_KERNEL); + if (!mhp) + return ERR_PTR(-ENOMEM); + + mhp->rhp = rhp; + + /* First check that we have enough alignment */ + if ((*iova_start & ~PAGE_MASK) != (buffer_list[0].addr & ~PAGE_MASK)) { + ret = -EINVAL; + goto err; + } + + if (num_phys_buf > 1 && + ((buffer_list[0].addr + buffer_list[0].size) & ~PAGE_MASK)) { + ret = -EINVAL; + goto err; + } + + ret = build_phys_page_list(buffer_list, num_phys_buf, iova_start, + &total_size, &npages, &shift, + &page_list); + if (ret) + goto err; + + if (mr_exceeds_hw_limits(rhp, total_size)) { + kfree(page_list); + ret = -EINVAL; + goto err; + } + + ret = alloc_pbl(mhp, npages); + if (ret) { + kfree(page_list); + goto err; + } + + ret = write_pbl(&mhp->rhp->rdev, page_list, mhp->attr.pbl_addr, + npages); + kfree(page_list); + if (ret) + goto err_pbl; + + mhp->attr.pdid = php->pdid; + mhp->attr.zbva = 0; + + mhp->attr.perms = c4iw_ib_to_tpt_access(acc); + mhp->attr.va_fbo = *iova_start; + mhp->attr.page_size = shift - 12; + + mhp->attr.len = (u32) total_size; + mhp->attr.pbl_size = npages; + ret = register_mem(rhp, php, mhp, shift); + if (ret) + goto err_pbl; + + return &mhp->ibmr; + +err_pbl: + c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr, + mhp->attr.pbl_size << 3); + +err: + kfree(mhp); + return ERR_PTR(ret); + +} + +struct ib_mr *c4iw_get_dma_mr(struct ib_pd *pd, int acc) +{ + struct c4iw_dev *rhp; + struct c4iw_pd *php; + struct c4iw_mr *mhp; + int ret; + u32 stag = T4_STAG_UNSET; + + PDBG("%s ib_pd %p\n", __func__, pd); + php = to_c4iw_pd(pd); + rhp = php->rhp; + + mhp = kzalloc(sizeof(*mhp), GFP_KERNEL); + if (!mhp) + return ERR_PTR(-ENOMEM); + + mhp->rhp = rhp; + mhp->attr.pdid = php->pdid; + mhp->attr.perms = c4iw_ib_to_tpt_access(acc); + mhp->attr.mw_bind_enable = (acc&IB_ACCESS_MW_BIND) == IB_ACCESS_MW_BIND; + mhp->attr.zbva = 0; + mhp->attr.va_fbo = 0; + mhp->attr.page_size = 0; + mhp->attr.len = ~0ULL; + mhp->attr.pbl_size = 0; + + ret = write_tpt_entry(&rhp->rdev, 0, &stag, 1, php->pdid, + FW_RI_STAG_NSMR, mhp->attr.perms, + mhp->attr.mw_bind_enable, 0, 0, ~0ULL, 0, 0, 0); + if (ret) + goto err1; + + ret = finish_mem_reg(mhp, stag); + if (ret) + goto err2; + return &mhp->ibmr; +err2: + dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, + mhp->attr.pbl_addr); +err1: + kfree(mhp); + return ERR_PTR(ret); +} + +struct ib_mr *c4iw_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, + u64 virt, int acc, struct ib_udata *udata) +{ + __be64 *pages; + int shift, n, len; + int i, k, entry; + int err = 0; + struct scatterlist *sg; + struct c4iw_dev *rhp; + struct c4iw_pd *php; + struct c4iw_mr *mhp; + + PDBG("%s ib_pd %p\n", __func__, pd); + + if (length == ~0ULL) + return ERR_PTR(-EINVAL); + + if ((length + start) < start) + return ERR_PTR(-EINVAL); + + php = to_c4iw_pd(pd); + rhp = php->rhp; + + if (mr_exceeds_hw_limits(rhp, length)) + return ERR_PTR(-EINVAL); + + mhp = kzalloc(sizeof(*mhp), GFP_KERNEL); + if (!mhp) + return ERR_PTR(-ENOMEM); + + mhp->rhp = rhp; + + mhp->umem = ib_umem_get(pd->uobject->context, start, length, acc, 0); + if (IS_ERR(mhp->umem)) { + err = PTR_ERR(mhp->umem); + kfree(mhp); + return ERR_PTR(err); + } + + shift = ffs(mhp->umem->page_size) - 1; + + n = mhp->umem->nmap; + err = alloc_pbl(mhp, n); + if (err) + goto err; + + pages = (__be64 *) __get_free_page(GFP_KERNEL); + if (!pages) { + err = -ENOMEM; + goto err_pbl; + } + + i = n = 0; + + for_each_sg(mhp->umem->sg_head.sgl, sg, mhp->umem->nmap, entry) { + len = sg_dma_len(sg) >> shift; + for (k = 0; k < len; ++k) { + pages[i++] = cpu_to_be64(sg_dma_address(sg) + + mhp->umem->page_size * k); + if (i == PAGE_SIZE / sizeof *pages) { + err = write_pbl(&mhp->rhp->rdev, + pages, + mhp->attr.pbl_addr + (n << 3), i); + if (err) + goto pbl_done; + n += i; + i = 0; + } + } + } + + if (i) + err = write_pbl(&mhp->rhp->rdev, pages, + mhp->attr.pbl_addr + (n << 3), i); + +pbl_done: + free_page((unsigned long) pages); + if (err) + goto err_pbl; + + mhp->attr.pdid = php->pdid; + mhp->attr.zbva = 0; + mhp->attr.perms = c4iw_ib_to_tpt_access(acc); + mhp->attr.va_fbo = virt; + mhp->attr.page_size = shift - 12; + mhp->attr.len = length; + + err = register_mem(rhp, php, mhp, shift); + if (err) + goto err_pbl; + + return &mhp->ibmr; + +err_pbl: + c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr, + mhp->attr.pbl_size << 3); + +err: + ib_umem_release(mhp->umem); + kfree(mhp); + return ERR_PTR(err); +} + +struct ib_mw *c4iw_alloc_mw(struct ib_pd *pd, enum ib_mw_type type) +{ + struct c4iw_dev *rhp; + struct c4iw_pd *php; + struct c4iw_mw *mhp; + u32 mmid; + u32 stag = 0; + int ret; + + if (type != IB_MW_TYPE_1) + return ERR_PTR(-EINVAL); + + php = to_c4iw_pd(pd); + rhp = php->rhp; + mhp = kzalloc(sizeof(*mhp), GFP_KERNEL); + if (!mhp) + return ERR_PTR(-ENOMEM); + ret = allocate_window(&rhp->rdev, &stag, php->pdid); + if (ret) { + kfree(mhp); + return ERR_PTR(ret); + } + mhp->rhp = rhp; + mhp->attr.pdid = php->pdid; + mhp->attr.type = FW_RI_STAG_MW; + mhp->attr.stag = stag; + mmid = (stag) >> 8; + mhp->ibmw.rkey = stag; + if (insert_handle(rhp, &rhp->mmidr, mhp, mmid)) { + deallocate_window(&rhp->rdev, mhp->attr.stag); + kfree(mhp); + return ERR_PTR(-ENOMEM); + } + PDBG("%s mmid 0x%x mhp %p stag 0x%x\n", __func__, mmid, mhp, stag); + return &(mhp->ibmw); +} + +int c4iw_dealloc_mw(struct ib_mw *mw) +{ + struct c4iw_dev *rhp; + struct c4iw_mw *mhp; + u32 mmid; + + mhp = to_c4iw_mw(mw); + rhp = mhp->rhp; + mmid = (mw->rkey) >> 8; + remove_handle(rhp, &rhp->mmidr, mmid); + deallocate_window(&rhp->rdev, mhp->attr.stag); + kfree(mhp); + PDBG("%s ib_mw %p mmid 0x%x ptr %p\n", __func__, mw, mmid, mhp); + return 0; +} + +struct ib_mr *c4iw_alloc_fast_reg_mr(struct ib_pd *pd, int pbl_depth) +{ + struct c4iw_dev *rhp; + struct c4iw_pd *php; + struct c4iw_mr *mhp; + u32 mmid; + u32 stag = 0; + int ret = 0; + + php = to_c4iw_pd(pd); + rhp = php->rhp; + mhp = kzalloc(sizeof(*mhp), GFP_KERNEL); + if (!mhp) { + ret = -ENOMEM; + goto err; + } + + mhp->rhp = rhp; + ret = alloc_pbl(mhp, pbl_depth); + if (ret) + goto err1; + mhp->attr.pbl_size = pbl_depth; + ret = allocate_stag(&rhp->rdev, &stag, php->pdid, + mhp->attr.pbl_size, mhp->attr.pbl_addr); + if (ret) + goto err2; + mhp->attr.pdid = php->pdid; + mhp->attr.type = FW_RI_STAG_NSMR; + mhp->attr.stag = stag; + mhp->attr.state = 1; + mmid = (stag) >> 8; + mhp->ibmr.rkey = mhp->ibmr.lkey = stag; + if (insert_handle(rhp, &rhp->mmidr, mhp, mmid)) { + ret = -ENOMEM; + goto err3; + } + + PDBG("%s mmid 0x%x mhp %p stag 0x%x\n", __func__, mmid, mhp, stag); + return &(mhp->ibmr); +err3: + dereg_mem(&rhp->rdev, stag, mhp->attr.pbl_size, + mhp->attr.pbl_addr); +err2: + c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr, + mhp->attr.pbl_size << 3); +err1: + kfree(mhp); +err: + return ERR_PTR(ret); +} + +struct ib_fast_reg_page_list *c4iw_alloc_fastreg_pbl(struct ib_device *device, + int page_list_len) +{ + struct c4iw_fr_page_list *c4pl; + struct c4iw_dev *dev = to_c4iw_dev(device); + dma_addr_t dma_addr; + int pll_len = roundup(page_list_len * sizeof(u64), 32); + + c4pl = kmalloc(sizeof(*c4pl), GFP_KERNEL); + if (!c4pl) + return ERR_PTR(-ENOMEM); + + c4pl->ibpl.page_list = dma_alloc_coherent(&dev->rdev.lldi.pdev->dev, + pll_len, &dma_addr, + GFP_KERNEL); + if (!c4pl->ibpl.page_list) { + kfree(c4pl); + return ERR_PTR(-ENOMEM); + } + dma_unmap_addr_set(c4pl, mapping, dma_addr); + c4pl->dma_addr = dma_addr; + c4pl->dev = dev; + c4pl->pll_len = pll_len; + + PDBG("%s c4pl %p pll_len %u page_list %p dma_addr %pad\n", + __func__, c4pl, c4pl->pll_len, c4pl->ibpl.page_list, + &c4pl->dma_addr); + + return &c4pl->ibpl; +} + +void c4iw_free_fastreg_pbl(struct ib_fast_reg_page_list *ibpl) +{ + struct c4iw_fr_page_list *c4pl = to_c4iw_fr_page_list(ibpl); + + PDBG("%s c4pl %p pll_len %u page_list %p dma_addr %pad\n", + __func__, c4pl, c4pl->pll_len, c4pl->ibpl.page_list, + &c4pl->dma_addr); + + dma_free_coherent(&c4pl->dev->rdev.lldi.pdev->dev, + c4pl->pll_len, + c4pl->ibpl.page_list, dma_unmap_addr(c4pl, mapping)); + kfree(c4pl); +} + +int c4iw_dereg_mr(struct ib_mr *ib_mr) +{ + struct c4iw_dev *rhp; + struct c4iw_mr *mhp; + u32 mmid; + + PDBG("%s ib_mr %p\n", __func__, ib_mr); + /* There can be no memory windows */ + if (atomic_read(&ib_mr->usecnt)) + return -EINVAL; + + mhp = to_c4iw_mr(ib_mr); + rhp = mhp->rhp; + mmid = mhp->attr.stag >> 8; + remove_handle(rhp, &rhp->mmidr, mmid); + dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, + mhp->attr.pbl_addr); + if (mhp->attr.pbl_size) + c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr, + mhp->attr.pbl_size << 3); + if (mhp->kva) + kfree((void *) (unsigned long) mhp->kva); + if (mhp->umem) + ib_umem_release(mhp->umem); + PDBG("%s mmid 0x%x ptr %p\n", __func__, mmid, mhp); + kfree(mhp); + return 0; +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/provider.c b/kernel/drivers/infiniband/hw/cxgb4/provider.c new file mode 100644 index 000000000..66bd6a2ad --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/provider.c @@ -0,0 +1,588 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/device.h> +#include <linux/netdevice.h> +#include <linux/etherdevice.h> +#include <linux/delay.h> +#include <linux/errno.h> +#include <linux/list.h> +#include <linux/spinlock.h> +#include <linux/ethtool.h> +#include <linux/rtnetlink.h> +#include <linux/inetdevice.h> +#include <linux/io.h> + +#include <asm/irq.h> +#include <asm/byteorder.h> + +#include <rdma/iw_cm.h> +#include <rdma/ib_verbs.h> +#include <rdma/ib_smi.h> +#include <rdma/ib_umem.h> +#include <rdma/ib_user_verbs.h> + +#include "iw_cxgb4.h" + +static int fastreg_support = 1; +module_param(fastreg_support, int, 0644); +MODULE_PARM_DESC(fastreg_support, "Advertise fastreg support (default=1)"); + +static struct ib_ah *c4iw_ah_create(struct ib_pd *pd, + struct ib_ah_attr *ah_attr) +{ + return ERR_PTR(-ENOSYS); +} + +static int c4iw_ah_destroy(struct ib_ah *ah) +{ + return -ENOSYS; +} + +static int c4iw_multicast_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) +{ + return -ENOSYS; +} + +static int c4iw_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) +{ + return -ENOSYS; +} + +static int c4iw_process_mad(struct ib_device *ibdev, int mad_flags, + u8 port_num, struct ib_wc *in_wc, + struct ib_grh *in_grh, struct ib_mad *in_mad, + struct ib_mad *out_mad) +{ + return -ENOSYS; +} + +static int c4iw_dealloc_ucontext(struct ib_ucontext *context) +{ + struct c4iw_dev *rhp = to_c4iw_dev(context->device); + struct c4iw_ucontext *ucontext = to_c4iw_ucontext(context); + struct c4iw_mm_entry *mm, *tmp; + + PDBG("%s context %p\n", __func__, context); + list_for_each_entry_safe(mm, tmp, &ucontext->mmaps, entry) + kfree(mm); + c4iw_release_dev_ucontext(&rhp->rdev, &ucontext->uctx); + kfree(ucontext); + return 0; +} + +static struct ib_ucontext *c4iw_alloc_ucontext(struct ib_device *ibdev, + struct ib_udata *udata) +{ + struct c4iw_ucontext *context; + struct c4iw_dev *rhp = to_c4iw_dev(ibdev); + static int warned; + struct c4iw_alloc_ucontext_resp uresp; + int ret = 0; + struct c4iw_mm_entry *mm = NULL; + + PDBG("%s ibdev %p\n", __func__, ibdev); + context = kzalloc(sizeof(*context), GFP_KERNEL); + if (!context) { + ret = -ENOMEM; + goto err; + } + + c4iw_init_dev_ucontext(&rhp->rdev, &context->uctx); + INIT_LIST_HEAD(&context->mmaps); + spin_lock_init(&context->mmap_lock); + + if (udata->outlen < sizeof(uresp) - sizeof(uresp.reserved)) { + if (!warned++) + pr_err(MOD "Warning - downlevel libcxgb4 (non-fatal), device status page disabled."); + rhp->rdev.flags |= T4_STATUS_PAGE_DISABLED; + } else { + mm = kmalloc(sizeof(*mm), GFP_KERNEL); + if (!mm) { + ret = -ENOMEM; + goto err_free; + } + + uresp.status_page_size = PAGE_SIZE; + + spin_lock(&context->mmap_lock); + uresp.status_page_key = context->key; + context->key += PAGE_SIZE; + spin_unlock(&context->mmap_lock); + + ret = ib_copy_to_udata(udata, &uresp, + sizeof(uresp) - sizeof(uresp.reserved)); + if (ret) + goto err_mm; + + mm->key = uresp.status_page_key; + mm->addr = virt_to_phys(rhp->rdev.status_page); + mm->len = PAGE_SIZE; + insert_mmap(context, mm); + } + return &context->ibucontext; +err_mm: + kfree(mm); +err_free: + kfree(context); +err: + return ERR_PTR(ret); +} + +static int c4iw_mmap(struct ib_ucontext *context, struct vm_area_struct *vma) +{ + int len = vma->vm_end - vma->vm_start; + u32 key = vma->vm_pgoff << PAGE_SHIFT; + struct c4iw_rdev *rdev; + int ret = 0; + struct c4iw_mm_entry *mm; + struct c4iw_ucontext *ucontext; + u64 addr; + + PDBG("%s pgoff 0x%lx key 0x%x len %d\n", __func__, vma->vm_pgoff, + key, len); + + if (vma->vm_start & (PAGE_SIZE-1)) + return -EINVAL; + + rdev = &(to_c4iw_dev(context->device)->rdev); + ucontext = to_c4iw_ucontext(context); + + mm = remove_mmap(ucontext, key, len); + if (!mm) + return -EINVAL; + addr = mm->addr; + kfree(mm); + + if ((addr >= pci_resource_start(rdev->lldi.pdev, 0)) && + (addr < (pci_resource_start(rdev->lldi.pdev, 0) + + pci_resource_len(rdev->lldi.pdev, 0)))) { + + /* + * MA_SYNC register... + */ + vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); + ret = io_remap_pfn_range(vma, vma->vm_start, + addr >> PAGE_SHIFT, + len, vma->vm_page_prot); + } else if ((addr >= pci_resource_start(rdev->lldi.pdev, 2)) && + (addr < (pci_resource_start(rdev->lldi.pdev, 2) + + pci_resource_len(rdev->lldi.pdev, 2)))) { + + /* + * Map user DB or OCQP memory... + */ + if (addr >= rdev->oc_mw_pa) + vma->vm_page_prot = t4_pgprot_wc(vma->vm_page_prot); + else { + if (is_t5(rdev->lldi.adapter_type)) + vma->vm_page_prot = + t4_pgprot_wc(vma->vm_page_prot); + else + vma->vm_page_prot = + pgprot_noncached(vma->vm_page_prot); + } + ret = io_remap_pfn_range(vma, vma->vm_start, + addr >> PAGE_SHIFT, + len, vma->vm_page_prot); + } else { + + /* + * Map WQ or CQ contig dma memory... + */ + ret = remap_pfn_range(vma, vma->vm_start, + addr >> PAGE_SHIFT, + len, vma->vm_page_prot); + } + + return ret; +} + +static int c4iw_deallocate_pd(struct ib_pd *pd) +{ + struct c4iw_dev *rhp; + struct c4iw_pd *php; + + php = to_c4iw_pd(pd); + rhp = php->rhp; + PDBG("%s ibpd %p pdid 0x%x\n", __func__, pd, php->pdid); + c4iw_put_resource(&rhp->rdev.resource.pdid_table, php->pdid); + mutex_lock(&rhp->rdev.stats.lock); + rhp->rdev.stats.pd.cur--; + mutex_unlock(&rhp->rdev.stats.lock); + kfree(php); + return 0; +} + +static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev, + struct ib_ucontext *context, + struct ib_udata *udata) +{ + struct c4iw_pd *php; + u32 pdid; + struct c4iw_dev *rhp; + + PDBG("%s ibdev %p\n", __func__, ibdev); + rhp = (struct c4iw_dev *) ibdev; + pdid = c4iw_get_resource(&rhp->rdev.resource.pdid_table); + if (!pdid) + return ERR_PTR(-EINVAL); + php = kzalloc(sizeof(*php), GFP_KERNEL); + if (!php) { + c4iw_put_resource(&rhp->rdev.resource.pdid_table, pdid); + return ERR_PTR(-ENOMEM); + } + php->pdid = pdid; + php->rhp = rhp; + if (context) { + if (ib_copy_to_udata(udata, &php->pdid, sizeof(u32))) { + c4iw_deallocate_pd(&php->ibpd); + return ERR_PTR(-EFAULT); + } + } + mutex_lock(&rhp->rdev.stats.lock); + rhp->rdev.stats.pd.cur++; + if (rhp->rdev.stats.pd.cur > rhp->rdev.stats.pd.max) + rhp->rdev.stats.pd.max = rhp->rdev.stats.pd.cur; + mutex_unlock(&rhp->rdev.stats.lock); + PDBG("%s pdid 0x%0x ptr 0x%p\n", __func__, pdid, php); + return &php->ibpd; +} + +static int c4iw_query_pkey(struct ib_device *ibdev, u8 port, u16 index, + u16 *pkey) +{ + PDBG("%s ibdev %p\n", __func__, ibdev); + *pkey = 0; + return 0; +} + +static int c4iw_query_gid(struct ib_device *ibdev, u8 port, int index, + union ib_gid *gid) +{ + struct c4iw_dev *dev; + + PDBG("%s ibdev %p, port %d, index %d, gid %p\n", + __func__, ibdev, port, index, gid); + dev = to_c4iw_dev(ibdev); + BUG_ON(port == 0); + memset(&(gid->raw[0]), 0, sizeof(gid->raw)); + memcpy(&(gid->raw[0]), dev->rdev.lldi.ports[port-1]->dev_addr, 6); + return 0; +} + +static int c4iw_query_device(struct ib_device *ibdev, + struct ib_device_attr *props) +{ + + struct c4iw_dev *dev; + PDBG("%s ibdev %p\n", __func__, ibdev); + + dev = to_c4iw_dev(ibdev); + memset(props, 0, sizeof *props); + memcpy(&props->sys_image_guid, dev->rdev.lldi.ports[0]->dev_addr, 6); + props->hw_ver = CHELSIO_CHIP_RELEASE(dev->rdev.lldi.adapter_type); + props->fw_ver = dev->rdev.lldi.fw_vers; + props->device_cap_flags = dev->device_cap_flags; + props->page_size_cap = T4_PAGESIZE_MASK; + props->vendor_id = (u32)dev->rdev.lldi.pdev->vendor; + props->vendor_part_id = (u32)dev->rdev.lldi.pdev->device; + props->max_mr_size = T4_MAX_MR_SIZE; + props->max_qp = dev->rdev.lldi.vr->qp.size / 2; + props->max_qp_wr = dev->rdev.hw_queue.t4_max_qp_depth; + props->max_sge = T4_MAX_RECV_SGE; + props->max_sge_rd = 1; + props->max_res_rd_atom = dev->rdev.lldi.max_ird_adapter; + props->max_qp_rd_atom = min(dev->rdev.lldi.max_ordird_qp, + c4iw_max_read_depth); + props->max_qp_init_rd_atom = props->max_qp_rd_atom; + props->max_cq = dev->rdev.lldi.vr->qp.size; + props->max_cqe = dev->rdev.hw_queue.t4_max_cq_depth; + props->max_mr = c4iw_num_stags(&dev->rdev); + props->max_pd = T4_MAX_NUM_PD; + props->local_ca_ack_delay = 0; + props->max_fast_reg_page_list_len = t4_max_fr_depth(use_dsgl); + + return 0; +} + +static int c4iw_query_port(struct ib_device *ibdev, u8 port, + struct ib_port_attr *props) +{ + struct c4iw_dev *dev; + struct net_device *netdev; + struct in_device *inetdev; + + PDBG("%s ibdev %p\n", __func__, ibdev); + + dev = to_c4iw_dev(ibdev); + netdev = dev->rdev.lldi.ports[port-1]; + + memset(props, 0, sizeof(struct ib_port_attr)); + props->max_mtu = IB_MTU_4096; + if (netdev->mtu >= 4096) + props->active_mtu = IB_MTU_4096; + else if (netdev->mtu >= 2048) + props->active_mtu = IB_MTU_2048; + else if (netdev->mtu >= 1024) + props->active_mtu = IB_MTU_1024; + else if (netdev->mtu >= 512) + props->active_mtu = IB_MTU_512; + else + props->active_mtu = IB_MTU_256; + + if (!netif_carrier_ok(netdev)) + props->state = IB_PORT_DOWN; + else { + inetdev = in_dev_get(netdev); + if (inetdev) { + if (inetdev->ifa_list) + props->state = IB_PORT_ACTIVE; + else + props->state = IB_PORT_INIT; + in_dev_put(inetdev); + } else + props->state = IB_PORT_INIT; + } + + props->port_cap_flags = + IB_PORT_CM_SUP | + IB_PORT_SNMP_TUNNEL_SUP | + IB_PORT_REINIT_SUP | + IB_PORT_DEVICE_MGMT_SUP | + IB_PORT_VENDOR_CLASS_SUP | IB_PORT_BOOT_MGMT_SUP; + props->gid_tbl_len = 1; + props->pkey_tbl_len = 1; + props->active_width = 2; + props->active_speed = IB_SPEED_DDR; + props->max_msg_sz = -1; + + return 0; +} + +static ssize_t show_rev(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev, + ibdev.dev); + PDBG("%s dev 0x%p\n", __func__, dev); + return sprintf(buf, "%d\n", + CHELSIO_CHIP_RELEASE(c4iw_dev->rdev.lldi.adapter_type)); +} + +static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev, + ibdev.dev); + PDBG("%s dev 0x%p\n", __func__, dev); + + return sprintf(buf, "%u.%u.%u.%u\n", + FW_HDR_FW_VER_MAJOR_G(c4iw_dev->rdev.lldi.fw_vers), + FW_HDR_FW_VER_MINOR_G(c4iw_dev->rdev.lldi.fw_vers), + FW_HDR_FW_VER_MICRO_G(c4iw_dev->rdev.lldi.fw_vers), + FW_HDR_FW_VER_BUILD_G(c4iw_dev->rdev.lldi.fw_vers)); +} + +static ssize_t show_hca(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev, + ibdev.dev); + struct ethtool_drvinfo info; + struct net_device *lldev = c4iw_dev->rdev.lldi.ports[0]; + + PDBG("%s dev 0x%p\n", __func__, dev); + lldev->ethtool_ops->get_drvinfo(lldev, &info); + return sprintf(buf, "%s\n", info.driver); +} + +static ssize_t show_board(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev, + ibdev.dev); + PDBG("%s dev 0x%p\n", __func__, dev); + return sprintf(buf, "%x.%x\n", c4iw_dev->rdev.lldi.pdev->vendor, + c4iw_dev->rdev.lldi.pdev->device); +} + +static int c4iw_get_mib(struct ib_device *ibdev, + union rdma_protocol_stats *stats) +{ + struct tp_tcp_stats v4, v6; + struct c4iw_dev *c4iw_dev = to_c4iw_dev(ibdev); + + cxgb4_get_tcp_stats(c4iw_dev->rdev.lldi.pdev, &v4, &v6); + memset(stats, 0, sizeof *stats); + stats->iw.tcpInSegs = v4.tcpInSegs + v6.tcpInSegs; + stats->iw.tcpOutSegs = v4.tcpOutSegs + v6.tcpOutSegs; + stats->iw.tcpRetransSegs = v4.tcpRetransSegs + v6.tcpRetransSegs; + stats->iw.tcpOutRsts = v4.tcpOutRsts + v6.tcpOutSegs; + + return 0; +} + +static DEVICE_ATTR(hw_rev, S_IRUGO, show_rev, NULL); +static DEVICE_ATTR(fw_ver, S_IRUGO, show_fw_ver, NULL); +static DEVICE_ATTR(hca_type, S_IRUGO, show_hca, NULL); +static DEVICE_ATTR(board_id, S_IRUGO, show_board, NULL); + +static struct device_attribute *c4iw_class_attributes[] = { + &dev_attr_hw_rev, + &dev_attr_fw_ver, + &dev_attr_hca_type, + &dev_attr_board_id, +}; + +int c4iw_register_device(struct c4iw_dev *dev) +{ + int ret; + int i; + + PDBG("%s c4iw_dev %p\n", __func__, dev); + BUG_ON(!dev->rdev.lldi.ports[0]); + strlcpy(dev->ibdev.name, "cxgb4_%d", IB_DEVICE_NAME_MAX); + memset(&dev->ibdev.node_guid, 0, sizeof(dev->ibdev.node_guid)); + memcpy(&dev->ibdev.node_guid, dev->rdev.lldi.ports[0]->dev_addr, 6); + dev->ibdev.owner = THIS_MODULE; + dev->device_cap_flags = IB_DEVICE_LOCAL_DMA_LKEY | IB_DEVICE_MEM_WINDOW; + if (fastreg_support) + dev->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS; + dev->ibdev.local_dma_lkey = 0; + dev->ibdev.uverbs_cmd_mask = + (1ull << IB_USER_VERBS_CMD_GET_CONTEXT) | + (1ull << IB_USER_VERBS_CMD_QUERY_DEVICE) | + (1ull << IB_USER_VERBS_CMD_QUERY_PORT) | + (1ull << IB_USER_VERBS_CMD_ALLOC_PD) | + (1ull << IB_USER_VERBS_CMD_DEALLOC_PD) | + (1ull << IB_USER_VERBS_CMD_REG_MR) | + (1ull << IB_USER_VERBS_CMD_DEREG_MR) | + (1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL) | + (1ull << IB_USER_VERBS_CMD_CREATE_CQ) | + (1ull << IB_USER_VERBS_CMD_DESTROY_CQ) | + (1ull << IB_USER_VERBS_CMD_REQ_NOTIFY_CQ) | + (1ull << IB_USER_VERBS_CMD_CREATE_QP) | + (1ull << IB_USER_VERBS_CMD_MODIFY_QP) | + (1ull << IB_USER_VERBS_CMD_QUERY_QP) | + (1ull << IB_USER_VERBS_CMD_POLL_CQ) | + (1ull << IB_USER_VERBS_CMD_DESTROY_QP) | + (1ull << IB_USER_VERBS_CMD_POST_SEND) | + (1ull << IB_USER_VERBS_CMD_POST_RECV); + dev->ibdev.node_type = RDMA_NODE_RNIC; + memcpy(dev->ibdev.node_desc, C4IW_NODE_DESC, sizeof(C4IW_NODE_DESC)); + dev->ibdev.phys_port_cnt = dev->rdev.lldi.nports; + dev->ibdev.num_comp_vectors = dev->rdev.lldi.nciq; + dev->ibdev.dma_device = &(dev->rdev.lldi.pdev->dev); + dev->ibdev.query_device = c4iw_query_device; + dev->ibdev.query_port = c4iw_query_port; + dev->ibdev.query_pkey = c4iw_query_pkey; + dev->ibdev.query_gid = c4iw_query_gid; + dev->ibdev.alloc_ucontext = c4iw_alloc_ucontext; + dev->ibdev.dealloc_ucontext = c4iw_dealloc_ucontext; + dev->ibdev.mmap = c4iw_mmap; + dev->ibdev.alloc_pd = c4iw_allocate_pd; + dev->ibdev.dealloc_pd = c4iw_deallocate_pd; + dev->ibdev.create_ah = c4iw_ah_create; + dev->ibdev.destroy_ah = c4iw_ah_destroy; + dev->ibdev.create_qp = c4iw_create_qp; + dev->ibdev.modify_qp = c4iw_ib_modify_qp; + dev->ibdev.query_qp = c4iw_ib_query_qp; + dev->ibdev.destroy_qp = c4iw_destroy_qp; + dev->ibdev.create_cq = c4iw_create_cq; + dev->ibdev.destroy_cq = c4iw_destroy_cq; + dev->ibdev.resize_cq = c4iw_resize_cq; + dev->ibdev.poll_cq = c4iw_poll_cq; + dev->ibdev.get_dma_mr = c4iw_get_dma_mr; + dev->ibdev.reg_phys_mr = c4iw_register_phys_mem; + dev->ibdev.rereg_phys_mr = c4iw_reregister_phys_mem; + dev->ibdev.reg_user_mr = c4iw_reg_user_mr; + dev->ibdev.dereg_mr = c4iw_dereg_mr; + dev->ibdev.alloc_mw = c4iw_alloc_mw; + dev->ibdev.bind_mw = c4iw_bind_mw; + dev->ibdev.dealloc_mw = c4iw_dealloc_mw; + dev->ibdev.alloc_fast_reg_mr = c4iw_alloc_fast_reg_mr; + dev->ibdev.alloc_fast_reg_page_list = c4iw_alloc_fastreg_pbl; + dev->ibdev.free_fast_reg_page_list = c4iw_free_fastreg_pbl; + dev->ibdev.attach_mcast = c4iw_multicast_attach; + dev->ibdev.detach_mcast = c4iw_multicast_detach; + dev->ibdev.process_mad = c4iw_process_mad; + dev->ibdev.req_notify_cq = c4iw_arm_cq; + dev->ibdev.post_send = c4iw_post_send; + dev->ibdev.post_recv = c4iw_post_receive; + dev->ibdev.get_protocol_stats = c4iw_get_mib; + dev->ibdev.uverbs_abi_ver = C4IW_UVERBS_ABI_VERSION; + + dev->ibdev.iwcm = kmalloc(sizeof(struct iw_cm_verbs), GFP_KERNEL); + if (!dev->ibdev.iwcm) + return -ENOMEM; + + dev->ibdev.iwcm->connect = c4iw_connect; + dev->ibdev.iwcm->accept = c4iw_accept_cr; + dev->ibdev.iwcm->reject = c4iw_reject_cr; + dev->ibdev.iwcm->create_listen = c4iw_create_listen; + dev->ibdev.iwcm->destroy_listen = c4iw_destroy_listen; + dev->ibdev.iwcm->add_ref = c4iw_qp_add_ref; + dev->ibdev.iwcm->rem_ref = c4iw_qp_rem_ref; + dev->ibdev.iwcm->get_qp = c4iw_get_qp; + + ret = ib_register_device(&dev->ibdev, NULL); + if (ret) + goto bail1; + + for (i = 0; i < ARRAY_SIZE(c4iw_class_attributes); ++i) { + ret = device_create_file(&dev->ibdev.dev, + c4iw_class_attributes[i]); + if (ret) + goto bail2; + } + return 0; +bail2: + ib_unregister_device(&dev->ibdev); +bail1: + kfree(dev->ibdev.iwcm); + return ret; +} + +void c4iw_unregister_device(struct c4iw_dev *dev) +{ + int i; + + PDBG("%s c4iw_dev %p\n", __func__, dev); + for (i = 0; i < ARRAY_SIZE(c4iw_class_attributes); ++i) + device_remove_file(&dev->ibdev.dev, + c4iw_class_attributes[i]); + ib_unregister_device(&dev->ibdev); + kfree(dev->ibdev.iwcm); + return; +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/qp.c b/kernel/drivers/infiniband/hw/cxgb4/qp.c new file mode 100644 index 000000000..389ced335 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/qp.c @@ -0,0 +1,1886 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ + +#include <linux/module.h> + +#include "iw_cxgb4.h" + +static int db_delay_usecs = 1; +module_param(db_delay_usecs, int, 0644); +MODULE_PARM_DESC(db_delay_usecs, "Usecs to delay awaiting db fifo to drain"); + +static int ocqp_support = 1; +module_param(ocqp_support, int, 0644); +MODULE_PARM_DESC(ocqp_support, "Support on-chip SQs (default=1)"); + +int db_fc_threshold = 1000; +module_param(db_fc_threshold, int, 0644); +MODULE_PARM_DESC(db_fc_threshold, + "QP count/threshold that triggers" + " automatic db flow control mode (default = 1000)"); + +int db_coalescing_threshold; +module_param(db_coalescing_threshold, int, 0644); +MODULE_PARM_DESC(db_coalescing_threshold, + "QP count/threshold that triggers" + " disabling db coalescing (default = 0)"); + +static int max_fr_immd = T4_MAX_FR_IMMD; +module_param(max_fr_immd, int, 0644); +MODULE_PARM_DESC(max_fr_immd, "fastreg threshold for using DSGL instead of immedate"); + +static int alloc_ird(struct c4iw_dev *dev, u32 ird) +{ + int ret = 0; + + spin_lock_irq(&dev->lock); + if (ird <= dev->avail_ird) + dev->avail_ird -= ird; + else + ret = -ENOMEM; + spin_unlock_irq(&dev->lock); + + if (ret) + dev_warn(&dev->rdev.lldi.pdev->dev, + "device IRD resources exhausted\n"); + + return ret; +} + +static void free_ird(struct c4iw_dev *dev, int ird) +{ + spin_lock_irq(&dev->lock); + dev->avail_ird += ird; + spin_unlock_irq(&dev->lock); +} + +static void set_state(struct c4iw_qp *qhp, enum c4iw_qp_state state) +{ + unsigned long flag; + spin_lock_irqsave(&qhp->lock, flag); + qhp->attr.state = state; + spin_unlock_irqrestore(&qhp->lock, flag); +} + +static void dealloc_oc_sq(struct c4iw_rdev *rdev, struct t4_sq *sq) +{ + c4iw_ocqp_pool_free(rdev, sq->dma_addr, sq->memsize); +} + +static void dealloc_host_sq(struct c4iw_rdev *rdev, struct t4_sq *sq) +{ + dma_free_coherent(&(rdev->lldi.pdev->dev), sq->memsize, sq->queue, + pci_unmap_addr(sq, mapping)); +} + +static void dealloc_sq(struct c4iw_rdev *rdev, struct t4_sq *sq) +{ + if (t4_sq_onchip(sq)) + dealloc_oc_sq(rdev, sq); + else + dealloc_host_sq(rdev, sq); +} + +static int alloc_oc_sq(struct c4iw_rdev *rdev, struct t4_sq *sq) +{ + if (!ocqp_support || !ocqp_supported(&rdev->lldi)) + return -ENOSYS; + sq->dma_addr = c4iw_ocqp_pool_alloc(rdev, sq->memsize); + if (!sq->dma_addr) + return -ENOMEM; + sq->phys_addr = rdev->oc_mw_pa + sq->dma_addr - + rdev->lldi.vr->ocq.start; + sq->queue = (__force union t4_wr *)(rdev->oc_mw_kva + sq->dma_addr - + rdev->lldi.vr->ocq.start); + sq->flags |= T4_SQ_ONCHIP; + return 0; +} + +static int alloc_host_sq(struct c4iw_rdev *rdev, struct t4_sq *sq) +{ + sq->queue = dma_alloc_coherent(&(rdev->lldi.pdev->dev), sq->memsize, + &(sq->dma_addr), GFP_KERNEL); + if (!sq->queue) + return -ENOMEM; + sq->phys_addr = virt_to_phys(sq->queue); + pci_unmap_addr_set(sq, mapping, sq->dma_addr); + return 0; +} + +static int alloc_sq(struct c4iw_rdev *rdev, struct t4_sq *sq, int user) +{ + int ret = -ENOSYS; + if (user) + ret = alloc_oc_sq(rdev, sq); + if (ret) + ret = alloc_host_sq(rdev, sq); + return ret; +} + +static int destroy_qp(struct c4iw_rdev *rdev, struct t4_wq *wq, + struct c4iw_dev_ucontext *uctx) +{ + /* + * uP clears EQ contexts when the connection exits rdma mode, + * so no need to post a RESET WR for these EQs. + */ + dma_free_coherent(&(rdev->lldi.pdev->dev), + wq->rq.memsize, wq->rq.queue, + dma_unmap_addr(&wq->rq, mapping)); + dealloc_sq(rdev, &wq->sq); + c4iw_rqtpool_free(rdev, wq->rq.rqt_hwaddr, wq->rq.rqt_size); + kfree(wq->rq.sw_rq); + kfree(wq->sq.sw_sq); + c4iw_put_qpid(rdev, wq->rq.qid, uctx); + c4iw_put_qpid(rdev, wq->sq.qid, uctx); + return 0; +} + +static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq, + struct t4_cq *rcq, struct t4_cq *scq, + struct c4iw_dev_ucontext *uctx) +{ + int user = (uctx != &rdev->uctx); + struct fw_ri_res_wr *res_wr; + struct fw_ri_res *res; + int wr_len; + struct c4iw_wr_wait wr_wait; + struct sk_buff *skb; + int ret = 0; + int eqsize; + + wq->sq.qid = c4iw_get_qpid(rdev, uctx); + if (!wq->sq.qid) + return -ENOMEM; + + wq->rq.qid = c4iw_get_qpid(rdev, uctx); + if (!wq->rq.qid) { + ret = -ENOMEM; + goto free_sq_qid; + } + + if (!user) { + wq->sq.sw_sq = kzalloc(wq->sq.size * sizeof *wq->sq.sw_sq, + GFP_KERNEL); + if (!wq->sq.sw_sq) { + ret = -ENOMEM; + goto free_rq_qid; + } + + wq->rq.sw_rq = kzalloc(wq->rq.size * sizeof *wq->rq.sw_rq, + GFP_KERNEL); + if (!wq->rq.sw_rq) { + ret = -ENOMEM; + goto free_sw_sq; + } + } + + /* + * RQT must be a power of 2 and at least 16 deep. + */ + wq->rq.rqt_size = roundup_pow_of_two(max_t(u16, wq->rq.size, 16)); + wq->rq.rqt_hwaddr = c4iw_rqtpool_alloc(rdev, wq->rq.rqt_size); + if (!wq->rq.rqt_hwaddr) { + ret = -ENOMEM; + goto free_sw_rq; + } + + ret = alloc_sq(rdev, &wq->sq, user); + if (ret) + goto free_hwaddr; + memset(wq->sq.queue, 0, wq->sq.memsize); + dma_unmap_addr_set(&wq->sq, mapping, wq->sq.dma_addr); + + wq->rq.queue = dma_alloc_coherent(&(rdev->lldi.pdev->dev), + wq->rq.memsize, &(wq->rq.dma_addr), + GFP_KERNEL); + if (!wq->rq.queue) { + ret = -ENOMEM; + goto free_sq; + } + PDBG("%s sq base va 0x%p pa 0x%llx rq base va 0x%p pa 0x%llx\n", + __func__, wq->sq.queue, + (unsigned long long)virt_to_phys(wq->sq.queue), + wq->rq.queue, + (unsigned long long)virt_to_phys(wq->rq.queue)); + memset(wq->rq.queue, 0, wq->rq.memsize); + dma_unmap_addr_set(&wq->rq, mapping, wq->rq.dma_addr); + + wq->db = rdev->lldi.db_reg; + wq->gts = rdev->lldi.gts_reg; + if (user || is_t5(rdev->lldi.adapter_type)) { + u32 off; + + off = (wq->sq.qid << rdev->qpshift) & PAGE_MASK; + if (user) { + wq->sq.udb = (u64 __iomem *)(rdev->bar2_pa + off); + } else { + off += 128 * (wq->sq.qid & rdev->qpmask) + 8; + wq->sq.udb = (u64 __iomem *)(rdev->bar2_kva + off); + } + off = (wq->rq.qid << rdev->qpshift) & PAGE_MASK; + if (user) { + wq->rq.udb = (u64 __iomem *)(rdev->bar2_pa + off); + } else { + off += 128 * (wq->rq.qid & rdev->qpmask) + 8; + wq->rq.udb = (u64 __iomem *)(rdev->bar2_kva + off); + } + } + wq->rdev = rdev; + wq->rq.msn = 1; + + /* build fw_ri_res_wr */ + wr_len = sizeof *res_wr + 2 * sizeof *res; + + skb = alloc_skb(wr_len, GFP_KERNEL); + if (!skb) { + ret = -ENOMEM; + goto free_dma; + } + set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); + + res_wr = (struct fw_ri_res_wr *)__skb_put(skb, wr_len); + memset(res_wr, 0, wr_len); + res_wr->op_nres = cpu_to_be32( + FW_WR_OP_V(FW_RI_RES_WR) | + FW_RI_RES_WR_NRES_V(2) | + FW_WR_COMPL_F); + res_wr->len16_pkd = cpu_to_be32(DIV_ROUND_UP(wr_len, 16)); + res_wr->cookie = (uintptr_t)&wr_wait; + res = res_wr->res; + res->u.sqrq.restype = FW_RI_RES_TYPE_SQ; + res->u.sqrq.op = FW_RI_RES_OP_WRITE; + + /* + * eqsize is the number of 64B entries plus the status page size. + */ + eqsize = wq->sq.size * T4_SQ_NUM_SLOTS + + rdev->hw_queue.t4_eq_status_entries; + + res->u.sqrq.fetchszm_to_iqid = cpu_to_be32( + FW_RI_RES_WR_HOSTFCMODE_V(0) | /* no host cidx updates */ + FW_RI_RES_WR_CPRIO_V(0) | /* don't keep in chip cache */ + FW_RI_RES_WR_PCIECHN_V(0) | /* set by uP at ri_init time */ + (t4_sq_onchip(&wq->sq) ? FW_RI_RES_WR_ONCHIP_F : 0) | + FW_RI_RES_WR_IQID_V(scq->cqid)); + res->u.sqrq.dcaen_to_eqsize = cpu_to_be32( + FW_RI_RES_WR_DCAEN_V(0) | + FW_RI_RES_WR_DCACPU_V(0) | + FW_RI_RES_WR_FBMIN_V(2) | + FW_RI_RES_WR_FBMAX_V(2) | + FW_RI_RES_WR_CIDXFTHRESHO_V(0) | + FW_RI_RES_WR_CIDXFTHRESH_V(0) | + FW_RI_RES_WR_EQSIZE_V(eqsize)); + res->u.sqrq.eqid = cpu_to_be32(wq->sq.qid); + res->u.sqrq.eqaddr = cpu_to_be64(wq->sq.dma_addr); + res++; + res->u.sqrq.restype = FW_RI_RES_TYPE_RQ; + res->u.sqrq.op = FW_RI_RES_OP_WRITE; + + /* + * eqsize is the number of 64B entries plus the status page size. + */ + eqsize = wq->rq.size * T4_RQ_NUM_SLOTS + + rdev->hw_queue.t4_eq_status_entries; + res->u.sqrq.fetchszm_to_iqid = cpu_to_be32( + FW_RI_RES_WR_HOSTFCMODE_V(0) | /* no host cidx updates */ + FW_RI_RES_WR_CPRIO_V(0) | /* don't keep in chip cache */ + FW_RI_RES_WR_PCIECHN_V(0) | /* set by uP at ri_init time */ + FW_RI_RES_WR_IQID_V(rcq->cqid)); + res->u.sqrq.dcaen_to_eqsize = cpu_to_be32( + FW_RI_RES_WR_DCAEN_V(0) | + FW_RI_RES_WR_DCACPU_V(0) | + FW_RI_RES_WR_FBMIN_V(2) | + FW_RI_RES_WR_FBMAX_V(2) | + FW_RI_RES_WR_CIDXFTHRESHO_V(0) | + FW_RI_RES_WR_CIDXFTHRESH_V(0) | + FW_RI_RES_WR_EQSIZE_V(eqsize)); + res->u.sqrq.eqid = cpu_to_be32(wq->rq.qid); + res->u.sqrq.eqaddr = cpu_to_be64(wq->rq.dma_addr); + + c4iw_init_wr_wait(&wr_wait); + + ret = c4iw_ofld_send(rdev, skb); + if (ret) + goto free_dma; + ret = c4iw_wait_for_reply(rdev, &wr_wait, 0, wq->sq.qid, __func__); + if (ret) + goto free_dma; + + PDBG("%s sqid 0x%x rqid 0x%x kdb 0x%p squdb 0x%lx rqudb 0x%lx\n", + __func__, wq->sq.qid, wq->rq.qid, wq->db, + (__force unsigned long) wq->sq.udb, + (__force unsigned long) wq->rq.udb); + + return 0; +free_dma: + dma_free_coherent(&(rdev->lldi.pdev->dev), + wq->rq.memsize, wq->rq.queue, + dma_unmap_addr(&wq->rq, mapping)); +free_sq: + dealloc_sq(rdev, &wq->sq); +free_hwaddr: + c4iw_rqtpool_free(rdev, wq->rq.rqt_hwaddr, wq->rq.rqt_size); +free_sw_rq: + kfree(wq->rq.sw_rq); +free_sw_sq: + kfree(wq->sq.sw_sq); +free_rq_qid: + c4iw_put_qpid(rdev, wq->rq.qid, uctx); +free_sq_qid: + c4iw_put_qpid(rdev, wq->sq.qid, uctx); + return ret; +} + +static int build_immd(struct t4_sq *sq, struct fw_ri_immd *immdp, + struct ib_send_wr *wr, int max, u32 *plenp) +{ + u8 *dstp, *srcp; + u32 plen = 0; + int i; + int rem, len; + + dstp = (u8 *)immdp->data; + for (i = 0; i < wr->num_sge; i++) { + if ((plen + wr->sg_list[i].length) > max) + return -EMSGSIZE; + srcp = (u8 *)(unsigned long)wr->sg_list[i].addr; + plen += wr->sg_list[i].length; + rem = wr->sg_list[i].length; + while (rem) { + if (dstp == (u8 *)&sq->queue[sq->size]) + dstp = (u8 *)sq->queue; + if (rem <= (u8 *)&sq->queue[sq->size] - dstp) + len = rem; + else + len = (u8 *)&sq->queue[sq->size] - dstp; + memcpy(dstp, srcp, len); + dstp += len; + srcp += len; + rem -= len; + } + } + len = roundup(plen + sizeof *immdp, 16) - (plen + sizeof *immdp); + if (len) + memset(dstp, 0, len); + immdp->op = FW_RI_DATA_IMMD; + immdp->r1 = 0; + immdp->r2 = 0; + immdp->immdlen = cpu_to_be32(plen); + *plenp = plen; + return 0; +} + +static int build_isgl(__be64 *queue_start, __be64 *queue_end, + struct fw_ri_isgl *isglp, struct ib_sge *sg_list, + int num_sge, u32 *plenp) + +{ + int i; + u32 plen = 0; + __be64 *flitp = (__be64 *)isglp->sge; + + for (i = 0; i < num_sge; i++) { + if ((plen + sg_list[i].length) < plen) + return -EMSGSIZE; + plen += sg_list[i].length; + *flitp = cpu_to_be64(((u64)sg_list[i].lkey << 32) | + sg_list[i].length); + if (++flitp == queue_end) + flitp = queue_start; + *flitp = cpu_to_be64(sg_list[i].addr); + if (++flitp == queue_end) + flitp = queue_start; + } + *flitp = (__force __be64)0; + isglp->op = FW_RI_DATA_ISGL; + isglp->r1 = 0; + isglp->nsge = cpu_to_be16(num_sge); + isglp->r2 = 0; + if (plenp) + *plenp = plen; + return 0; +} + +static int build_rdma_send(struct t4_sq *sq, union t4_wr *wqe, + struct ib_send_wr *wr, u8 *len16) +{ + u32 plen; + int size; + int ret; + + if (wr->num_sge > T4_MAX_SEND_SGE) + return -EINVAL; + switch (wr->opcode) { + case IB_WR_SEND: + if (wr->send_flags & IB_SEND_SOLICITED) + wqe->send.sendop_pkd = cpu_to_be32( + FW_RI_SEND_WR_SENDOP_V(FW_RI_SEND_WITH_SE)); + else + wqe->send.sendop_pkd = cpu_to_be32( + FW_RI_SEND_WR_SENDOP_V(FW_RI_SEND)); + wqe->send.stag_inv = 0; + break; + case IB_WR_SEND_WITH_INV: + if (wr->send_flags & IB_SEND_SOLICITED) + wqe->send.sendop_pkd = cpu_to_be32( + FW_RI_SEND_WR_SENDOP_V(FW_RI_SEND_WITH_SE_INV)); + else + wqe->send.sendop_pkd = cpu_to_be32( + FW_RI_SEND_WR_SENDOP_V(FW_RI_SEND_WITH_INV)); + wqe->send.stag_inv = cpu_to_be32(wr->ex.invalidate_rkey); + break; + + default: + return -EINVAL; + } + wqe->send.r3 = 0; + wqe->send.r4 = 0; + + plen = 0; + if (wr->num_sge) { + if (wr->send_flags & IB_SEND_INLINE) { + ret = build_immd(sq, wqe->send.u.immd_src, wr, + T4_MAX_SEND_INLINE, &plen); + if (ret) + return ret; + size = sizeof wqe->send + sizeof(struct fw_ri_immd) + + plen; + } else { + ret = build_isgl((__be64 *)sq->queue, + (__be64 *)&sq->queue[sq->size], + wqe->send.u.isgl_src, + wr->sg_list, wr->num_sge, &plen); + if (ret) + return ret; + size = sizeof wqe->send + sizeof(struct fw_ri_isgl) + + wr->num_sge * sizeof(struct fw_ri_sge); + } + } else { + wqe->send.u.immd_src[0].op = FW_RI_DATA_IMMD; + wqe->send.u.immd_src[0].r1 = 0; + wqe->send.u.immd_src[0].r2 = 0; + wqe->send.u.immd_src[0].immdlen = 0; + size = sizeof wqe->send + sizeof(struct fw_ri_immd); + plen = 0; + } + *len16 = DIV_ROUND_UP(size, 16); + wqe->send.plen = cpu_to_be32(plen); + return 0; +} + +static int build_rdma_write(struct t4_sq *sq, union t4_wr *wqe, + struct ib_send_wr *wr, u8 *len16) +{ + u32 plen; + int size; + int ret; + + if (wr->num_sge > T4_MAX_SEND_SGE) + return -EINVAL; + wqe->write.r2 = 0; + wqe->write.stag_sink = cpu_to_be32(wr->wr.rdma.rkey); + wqe->write.to_sink = cpu_to_be64(wr->wr.rdma.remote_addr); + if (wr->num_sge) { + if (wr->send_flags & IB_SEND_INLINE) { + ret = build_immd(sq, wqe->write.u.immd_src, wr, + T4_MAX_WRITE_INLINE, &plen); + if (ret) + return ret; + size = sizeof wqe->write + sizeof(struct fw_ri_immd) + + plen; + } else { + ret = build_isgl((__be64 *)sq->queue, + (__be64 *)&sq->queue[sq->size], + wqe->write.u.isgl_src, + wr->sg_list, wr->num_sge, &plen); + if (ret) + return ret; + size = sizeof wqe->write + sizeof(struct fw_ri_isgl) + + wr->num_sge * sizeof(struct fw_ri_sge); + } + } else { + wqe->write.u.immd_src[0].op = FW_RI_DATA_IMMD; + wqe->write.u.immd_src[0].r1 = 0; + wqe->write.u.immd_src[0].r2 = 0; + wqe->write.u.immd_src[0].immdlen = 0; + size = sizeof wqe->write + sizeof(struct fw_ri_immd); + plen = 0; + } + *len16 = DIV_ROUND_UP(size, 16); + wqe->write.plen = cpu_to_be32(plen); + return 0; +} + +static int build_rdma_read(union t4_wr *wqe, struct ib_send_wr *wr, u8 *len16) +{ + if (wr->num_sge > 1) + return -EINVAL; + if (wr->num_sge) { + wqe->read.stag_src = cpu_to_be32(wr->wr.rdma.rkey); + wqe->read.to_src_hi = cpu_to_be32((u32)(wr->wr.rdma.remote_addr + >> 32)); + wqe->read.to_src_lo = cpu_to_be32((u32)wr->wr.rdma.remote_addr); + wqe->read.stag_sink = cpu_to_be32(wr->sg_list[0].lkey); + wqe->read.plen = cpu_to_be32(wr->sg_list[0].length); + wqe->read.to_sink_hi = cpu_to_be32((u32)(wr->sg_list[0].addr + >> 32)); + wqe->read.to_sink_lo = cpu_to_be32((u32)(wr->sg_list[0].addr)); + } else { + wqe->read.stag_src = cpu_to_be32(2); + wqe->read.to_src_hi = 0; + wqe->read.to_src_lo = 0; + wqe->read.stag_sink = cpu_to_be32(2); + wqe->read.plen = 0; + wqe->read.to_sink_hi = 0; + wqe->read.to_sink_lo = 0; + } + wqe->read.r2 = 0; + wqe->read.r5 = 0; + *len16 = DIV_ROUND_UP(sizeof wqe->read, 16); + return 0; +} + +static int build_rdma_recv(struct c4iw_qp *qhp, union t4_recv_wr *wqe, + struct ib_recv_wr *wr, u8 *len16) +{ + int ret; + + ret = build_isgl((__be64 *)qhp->wq.rq.queue, + (__be64 *)&qhp->wq.rq.queue[qhp->wq.rq.size], + &wqe->recv.isgl, wr->sg_list, wr->num_sge, NULL); + if (ret) + return ret; + *len16 = DIV_ROUND_UP(sizeof wqe->recv + + wr->num_sge * sizeof(struct fw_ri_sge), 16); + return 0; +} + +static int build_fastreg(struct t4_sq *sq, union t4_wr *wqe, + struct ib_send_wr *wr, u8 *len16, u8 t5dev) +{ + + struct fw_ri_immd *imdp; + __be64 *p; + int i; + int pbllen = roundup(wr->wr.fast_reg.page_list_len * sizeof(u64), 32); + int rem; + + if (wr->wr.fast_reg.page_list_len > + t4_max_fr_depth(use_dsgl)) + return -EINVAL; + + wqe->fr.qpbinde_to_dcacpu = 0; + wqe->fr.pgsz_shift = wr->wr.fast_reg.page_shift - 12; + wqe->fr.addr_type = FW_RI_VA_BASED_TO; + wqe->fr.mem_perms = c4iw_ib_to_tpt_access(wr->wr.fast_reg.access_flags); + wqe->fr.len_hi = 0; + wqe->fr.len_lo = cpu_to_be32(wr->wr.fast_reg.length); + wqe->fr.stag = cpu_to_be32(wr->wr.fast_reg.rkey); + wqe->fr.va_hi = cpu_to_be32(wr->wr.fast_reg.iova_start >> 32); + wqe->fr.va_lo_fbo = cpu_to_be32(wr->wr.fast_reg.iova_start & + 0xffffffff); + + if (t5dev && use_dsgl && (pbllen > max_fr_immd)) { + struct c4iw_fr_page_list *c4pl = + to_c4iw_fr_page_list(wr->wr.fast_reg.page_list); + struct fw_ri_dsgl *sglp; + + for (i = 0; i < wr->wr.fast_reg.page_list_len; i++) { + wr->wr.fast_reg.page_list->page_list[i] = (__force u64) + cpu_to_be64((u64) + wr->wr.fast_reg.page_list->page_list[i]); + } + + sglp = (struct fw_ri_dsgl *)(&wqe->fr + 1); + sglp->op = FW_RI_DATA_DSGL; + sglp->r1 = 0; + sglp->nsge = cpu_to_be16(1); + sglp->addr0 = cpu_to_be64(c4pl->dma_addr); + sglp->len0 = cpu_to_be32(pbllen); + + *len16 = DIV_ROUND_UP(sizeof(wqe->fr) + sizeof(*sglp), 16); + } else { + imdp = (struct fw_ri_immd *)(&wqe->fr + 1); + imdp->op = FW_RI_DATA_IMMD; + imdp->r1 = 0; + imdp->r2 = 0; + imdp->immdlen = cpu_to_be32(pbllen); + p = (__be64 *)(imdp + 1); + rem = pbllen; + for (i = 0; i < wr->wr.fast_reg.page_list_len; i++) { + *p = cpu_to_be64( + (u64)wr->wr.fast_reg.page_list->page_list[i]); + rem -= sizeof(*p); + if (++p == (__be64 *)&sq->queue[sq->size]) + p = (__be64 *)sq->queue; + } + BUG_ON(rem < 0); + while (rem) { + *p = 0; + rem -= sizeof(*p); + if (++p == (__be64 *)&sq->queue[sq->size]) + p = (__be64 *)sq->queue; + } + *len16 = DIV_ROUND_UP(sizeof(wqe->fr) + sizeof(*imdp) + + pbllen, 16); + } + return 0; +} + +static int build_inv_stag(union t4_wr *wqe, struct ib_send_wr *wr, + u8 *len16) +{ + wqe->inv.stag_inv = cpu_to_be32(wr->ex.invalidate_rkey); + wqe->inv.r2 = 0; + *len16 = DIV_ROUND_UP(sizeof wqe->inv, 16); + return 0; +} + +void c4iw_qp_add_ref(struct ib_qp *qp) +{ + PDBG("%s ib_qp %p\n", __func__, qp); + atomic_inc(&(to_c4iw_qp(qp)->refcnt)); +} + +void c4iw_qp_rem_ref(struct ib_qp *qp) +{ + PDBG("%s ib_qp %p\n", __func__, qp); + if (atomic_dec_and_test(&(to_c4iw_qp(qp)->refcnt))) + wake_up(&(to_c4iw_qp(qp)->wait)); +} + +static void add_to_fc_list(struct list_head *head, struct list_head *entry) +{ + if (list_empty(entry)) + list_add_tail(entry, head); +} + +static int ring_kernel_sq_db(struct c4iw_qp *qhp, u16 inc) +{ + unsigned long flags; + + spin_lock_irqsave(&qhp->rhp->lock, flags); + spin_lock(&qhp->lock); + if (qhp->rhp->db_state == NORMAL) + t4_ring_sq_db(&qhp->wq, inc, + is_t5(qhp->rhp->rdev.lldi.adapter_type), NULL); + else { + add_to_fc_list(&qhp->rhp->db_fc_list, &qhp->db_fc_entry); + qhp->wq.sq.wq_pidx_inc += inc; + } + spin_unlock(&qhp->lock); + spin_unlock_irqrestore(&qhp->rhp->lock, flags); + return 0; +} + +static int ring_kernel_rq_db(struct c4iw_qp *qhp, u16 inc) +{ + unsigned long flags; + + spin_lock_irqsave(&qhp->rhp->lock, flags); + spin_lock(&qhp->lock); + if (qhp->rhp->db_state == NORMAL) + t4_ring_rq_db(&qhp->wq, inc, + is_t5(qhp->rhp->rdev.lldi.adapter_type), NULL); + else { + add_to_fc_list(&qhp->rhp->db_fc_list, &qhp->db_fc_entry); + qhp->wq.rq.wq_pidx_inc += inc; + } + spin_unlock(&qhp->lock); + spin_unlock_irqrestore(&qhp->rhp->lock, flags); + return 0; +} + +int c4iw_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, + struct ib_send_wr **bad_wr) +{ + int err = 0; + u8 len16 = 0; + enum fw_wr_opcodes fw_opcode = 0; + enum fw_ri_wr_flags fw_flags; + struct c4iw_qp *qhp; + union t4_wr *wqe = NULL; + u32 num_wrs; + struct t4_swsqe *swsqe; + unsigned long flag; + u16 idx = 0; + + qhp = to_c4iw_qp(ibqp); + spin_lock_irqsave(&qhp->lock, flag); + if (t4_wq_in_error(&qhp->wq)) { + spin_unlock_irqrestore(&qhp->lock, flag); + return -EINVAL; + } + num_wrs = t4_sq_avail(&qhp->wq); + if (num_wrs == 0) { + spin_unlock_irqrestore(&qhp->lock, flag); + return -ENOMEM; + } + while (wr) { + if (num_wrs == 0) { + err = -ENOMEM; + *bad_wr = wr; + break; + } + wqe = (union t4_wr *)((u8 *)qhp->wq.sq.queue + + qhp->wq.sq.wq_pidx * T4_EQ_ENTRY_SIZE); + + fw_flags = 0; + if (wr->send_flags & IB_SEND_SOLICITED) + fw_flags |= FW_RI_SOLICITED_EVENT_FLAG; + if (wr->send_flags & IB_SEND_SIGNALED || qhp->sq_sig_all) + fw_flags |= FW_RI_COMPLETION_FLAG; + swsqe = &qhp->wq.sq.sw_sq[qhp->wq.sq.pidx]; + switch (wr->opcode) { + case IB_WR_SEND_WITH_INV: + case IB_WR_SEND: + if (wr->send_flags & IB_SEND_FENCE) + fw_flags |= FW_RI_READ_FENCE_FLAG; + fw_opcode = FW_RI_SEND_WR; + if (wr->opcode == IB_WR_SEND) + swsqe->opcode = FW_RI_SEND; + else + swsqe->opcode = FW_RI_SEND_WITH_INV; + err = build_rdma_send(&qhp->wq.sq, wqe, wr, &len16); + break; + case IB_WR_RDMA_WRITE: + fw_opcode = FW_RI_RDMA_WRITE_WR; + swsqe->opcode = FW_RI_RDMA_WRITE; + err = build_rdma_write(&qhp->wq.sq, wqe, wr, &len16); + break; + case IB_WR_RDMA_READ: + case IB_WR_RDMA_READ_WITH_INV: + fw_opcode = FW_RI_RDMA_READ_WR; + swsqe->opcode = FW_RI_READ_REQ; + if (wr->opcode == IB_WR_RDMA_READ_WITH_INV) + fw_flags = FW_RI_RDMA_READ_INVALIDATE; + else + fw_flags = 0; + err = build_rdma_read(wqe, wr, &len16); + if (err) + break; + swsqe->read_len = wr->sg_list[0].length; + if (!qhp->wq.sq.oldest_read) + qhp->wq.sq.oldest_read = swsqe; + break; + case IB_WR_FAST_REG_MR: + fw_opcode = FW_RI_FR_NSMR_WR; + swsqe->opcode = FW_RI_FAST_REGISTER; + err = build_fastreg(&qhp->wq.sq, wqe, wr, &len16, + is_t5( + qhp->rhp->rdev.lldi.adapter_type) ? + 1 : 0); + break; + case IB_WR_LOCAL_INV: + if (wr->send_flags & IB_SEND_FENCE) + fw_flags |= FW_RI_LOCAL_FENCE_FLAG; + fw_opcode = FW_RI_INV_LSTAG_WR; + swsqe->opcode = FW_RI_LOCAL_INV; + err = build_inv_stag(wqe, wr, &len16); + break; + default: + PDBG("%s post of type=%d TBD!\n", __func__, + wr->opcode); + err = -EINVAL; + } + if (err) { + *bad_wr = wr; + break; + } + swsqe->idx = qhp->wq.sq.pidx; + swsqe->complete = 0; + swsqe->signaled = (wr->send_flags & IB_SEND_SIGNALED) || + qhp->sq_sig_all; + swsqe->flushed = 0; + swsqe->wr_id = wr->wr_id; + if (c4iw_wr_log) { + swsqe->sge_ts = cxgb4_read_sge_timestamp( + qhp->rhp->rdev.lldi.ports[0]); + getnstimeofday(&swsqe->host_ts); + } + + init_wr_hdr(wqe, qhp->wq.sq.pidx, fw_opcode, fw_flags, len16); + + PDBG("%s cookie 0x%llx pidx 0x%x opcode 0x%x read_len %u\n", + __func__, (unsigned long long)wr->wr_id, qhp->wq.sq.pidx, + swsqe->opcode, swsqe->read_len); + wr = wr->next; + num_wrs--; + t4_sq_produce(&qhp->wq, len16); + idx += DIV_ROUND_UP(len16*16, T4_EQ_ENTRY_SIZE); + } + if (!qhp->rhp->rdev.status_page->db_off) { + t4_ring_sq_db(&qhp->wq, idx, + is_t5(qhp->rhp->rdev.lldi.adapter_type), wqe); + spin_unlock_irqrestore(&qhp->lock, flag); + } else { + spin_unlock_irqrestore(&qhp->lock, flag); + ring_kernel_sq_db(qhp, idx); + } + return err; +} + +int c4iw_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr, + struct ib_recv_wr **bad_wr) +{ + int err = 0; + struct c4iw_qp *qhp; + union t4_recv_wr *wqe = NULL; + u32 num_wrs; + u8 len16 = 0; + unsigned long flag; + u16 idx = 0; + + qhp = to_c4iw_qp(ibqp); + spin_lock_irqsave(&qhp->lock, flag); + if (t4_wq_in_error(&qhp->wq)) { + spin_unlock_irqrestore(&qhp->lock, flag); + return -EINVAL; + } + num_wrs = t4_rq_avail(&qhp->wq); + if (num_wrs == 0) { + spin_unlock_irqrestore(&qhp->lock, flag); + return -ENOMEM; + } + while (wr) { + if (wr->num_sge > T4_MAX_RECV_SGE) { + err = -EINVAL; + *bad_wr = wr; + break; + } + wqe = (union t4_recv_wr *)((u8 *)qhp->wq.rq.queue + + qhp->wq.rq.wq_pidx * + T4_EQ_ENTRY_SIZE); + if (num_wrs) + err = build_rdma_recv(qhp, wqe, wr, &len16); + else + err = -ENOMEM; + if (err) { + *bad_wr = wr; + break; + } + + qhp->wq.rq.sw_rq[qhp->wq.rq.pidx].wr_id = wr->wr_id; + if (c4iw_wr_log) { + qhp->wq.rq.sw_rq[qhp->wq.rq.pidx].sge_ts = + cxgb4_read_sge_timestamp( + qhp->rhp->rdev.lldi.ports[0]); + getnstimeofday( + &qhp->wq.rq.sw_rq[qhp->wq.rq.pidx].host_ts); + } + + wqe->recv.opcode = FW_RI_RECV_WR; + wqe->recv.r1 = 0; + wqe->recv.wrid = qhp->wq.rq.pidx; + wqe->recv.r2[0] = 0; + wqe->recv.r2[1] = 0; + wqe->recv.r2[2] = 0; + wqe->recv.len16 = len16; + PDBG("%s cookie 0x%llx pidx %u\n", __func__, + (unsigned long long) wr->wr_id, qhp->wq.rq.pidx); + t4_rq_produce(&qhp->wq, len16); + idx += DIV_ROUND_UP(len16*16, T4_EQ_ENTRY_SIZE); + wr = wr->next; + num_wrs--; + } + if (!qhp->rhp->rdev.status_page->db_off) { + t4_ring_rq_db(&qhp->wq, idx, + is_t5(qhp->rhp->rdev.lldi.adapter_type), wqe); + spin_unlock_irqrestore(&qhp->lock, flag); + } else { + spin_unlock_irqrestore(&qhp->lock, flag); + ring_kernel_rq_db(qhp, idx); + } + return err; +} + +int c4iw_bind_mw(struct ib_qp *qp, struct ib_mw *mw, struct ib_mw_bind *mw_bind) +{ + return -ENOSYS; +} + +static inline void build_term_codes(struct t4_cqe *err_cqe, u8 *layer_type, + u8 *ecode) +{ + int status; + int tagged; + int opcode; + int rqtype; + int send_inv; + + if (!err_cqe) { + *layer_type = LAYER_RDMAP|DDP_LOCAL_CATA; + *ecode = 0; + return; + } + + status = CQE_STATUS(err_cqe); + opcode = CQE_OPCODE(err_cqe); + rqtype = RQ_TYPE(err_cqe); + send_inv = (opcode == FW_RI_SEND_WITH_INV) || + (opcode == FW_RI_SEND_WITH_SE_INV); + tagged = (opcode == FW_RI_RDMA_WRITE) || + (rqtype && (opcode == FW_RI_READ_RESP)); + + switch (status) { + case T4_ERR_STAG: + if (send_inv) { + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_OP; + *ecode = RDMAP_CANT_INV_STAG; + } else { + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_INV_STAG; + } + break; + case T4_ERR_PDID: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + if ((opcode == FW_RI_SEND_WITH_INV) || + (opcode == FW_RI_SEND_WITH_SE_INV)) + *ecode = RDMAP_CANT_INV_STAG; + else + *ecode = RDMAP_STAG_NOT_ASSOC; + break; + case T4_ERR_QPID: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_STAG_NOT_ASSOC; + break; + case T4_ERR_ACCESS: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_ACC_VIOL; + break; + case T4_ERR_WRAP: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_TO_WRAP; + break; + case T4_ERR_BOUND: + if (tagged) { + *layer_type = LAYER_DDP|DDP_TAGGED_ERR; + *ecode = DDPT_BASE_BOUNDS; + } else { + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_PROT; + *ecode = RDMAP_BASE_BOUNDS; + } + break; + case T4_ERR_INVALIDATE_SHARED_MR: + case T4_ERR_INVALIDATE_MR_WITH_MW_BOUND: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_OP; + *ecode = RDMAP_CANT_INV_STAG; + break; + case T4_ERR_ECC: + case T4_ERR_ECC_PSTAG: + case T4_ERR_INTERNAL_ERR: + *layer_type = LAYER_RDMAP|RDMAP_LOCAL_CATA; + *ecode = 0; + break; + case T4_ERR_OUT_OF_RQE: + *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; + *ecode = DDPU_INV_MSN_NOBUF; + break; + case T4_ERR_PBL_ADDR_BOUND: + *layer_type = LAYER_DDP|DDP_TAGGED_ERR; + *ecode = DDPT_BASE_BOUNDS; + break; + case T4_ERR_CRC: + *layer_type = LAYER_MPA|DDP_LLP; + *ecode = MPA_CRC_ERR; + break; + case T4_ERR_MARKER: + *layer_type = LAYER_MPA|DDP_LLP; + *ecode = MPA_MARKER_ERR; + break; + case T4_ERR_PDU_LEN_ERR: + *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; + *ecode = DDPU_MSG_TOOBIG; + break; + case T4_ERR_DDP_VERSION: + if (tagged) { + *layer_type = LAYER_DDP|DDP_TAGGED_ERR; + *ecode = DDPT_INV_VERS; + } else { + *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; + *ecode = DDPU_INV_VERS; + } + break; + case T4_ERR_RDMA_VERSION: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_OP; + *ecode = RDMAP_INV_VERS; + break; + case T4_ERR_OPCODE: + *layer_type = LAYER_RDMAP|RDMAP_REMOTE_OP; + *ecode = RDMAP_INV_OPCODE; + break; + case T4_ERR_DDP_QUEUE_NUM: + *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; + *ecode = DDPU_INV_QN; + break; + case T4_ERR_MSN: + case T4_ERR_MSN_GAP: + case T4_ERR_MSN_RANGE: + case T4_ERR_IRD_OVERFLOW: + *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; + *ecode = DDPU_INV_MSN_RANGE; + break; + case T4_ERR_TBIT: + *layer_type = LAYER_DDP|DDP_LOCAL_CATA; + *ecode = 0; + break; + case T4_ERR_MO: + *layer_type = LAYER_DDP|DDP_UNTAGGED_ERR; + *ecode = DDPU_INV_MO; + break; + default: + *layer_type = LAYER_RDMAP|DDP_LOCAL_CATA; + *ecode = 0; + break; + } +} + +static void post_terminate(struct c4iw_qp *qhp, struct t4_cqe *err_cqe, + gfp_t gfp) +{ + struct fw_ri_wr *wqe; + struct sk_buff *skb; + struct terminate_message *term; + + PDBG("%s qhp %p qid 0x%x tid %u\n", __func__, qhp, qhp->wq.sq.qid, + qhp->ep->hwtid); + + skb = alloc_skb(sizeof *wqe, gfp); + if (!skb) + return; + set_wr_txq(skb, CPL_PRIORITY_DATA, qhp->ep->txq_idx); + + wqe = (struct fw_ri_wr *)__skb_put(skb, sizeof(*wqe)); + memset(wqe, 0, sizeof *wqe); + wqe->op_compl = cpu_to_be32(FW_WR_OP_V(FW_RI_INIT_WR)); + wqe->flowid_len16 = cpu_to_be32( + FW_WR_FLOWID_V(qhp->ep->hwtid) | + FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*wqe), 16))); + + wqe->u.terminate.type = FW_RI_TYPE_TERMINATE; + wqe->u.terminate.immdlen = cpu_to_be32(sizeof *term); + term = (struct terminate_message *)wqe->u.terminate.termmsg; + if (qhp->attr.layer_etype == (LAYER_MPA|DDP_LLP)) { + term->layer_etype = qhp->attr.layer_etype; + term->ecode = qhp->attr.ecode; + } else + build_term_codes(err_cqe, &term->layer_etype, &term->ecode); + c4iw_ofld_send(&qhp->rhp->rdev, skb); +} + +/* + * Assumes qhp lock is held. + */ +static void __flush_qp(struct c4iw_qp *qhp, struct c4iw_cq *rchp, + struct c4iw_cq *schp) +{ + int count; + int rq_flushed, sq_flushed; + unsigned long flag; + + PDBG("%s qhp %p rchp %p schp %p\n", __func__, qhp, rchp, schp); + + /* locking hierarchy: cq lock first, then qp lock. */ + spin_lock_irqsave(&rchp->lock, flag); + spin_lock(&qhp->lock); + + if (qhp->wq.flushed) { + spin_unlock(&qhp->lock); + spin_unlock_irqrestore(&rchp->lock, flag); + return; + } + qhp->wq.flushed = 1; + + c4iw_flush_hw_cq(rchp); + c4iw_count_rcqes(&rchp->cq, &qhp->wq, &count); + rq_flushed = c4iw_flush_rq(&qhp->wq, &rchp->cq, count); + spin_unlock(&qhp->lock); + spin_unlock_irqrestore(&rchp->lock, flag); + + /* locking hierarchy: cq lock first, then qp lock. */ + spin_lock_irqsave(&schp->lock, flag); + spin_lock(&qhp->lock); + if (schp != rchp) + c4iw_flush_hw_cq(schp); + sq_flushed = c4iw_flush_sq(qhp); + spin_unlock(&qhp->lock); + spin_unlock_irqrestore(&schp->lock, flag); + + if (schp == rchp) { + if (t4_clear_cq_armed(&rchp->cq) && + (rq_flushed || sq_flushed)) { + spin_lock_irqsave(&rchp->comp_handler_lock, flag); + (*rchp->ibcq.comp_handler)(&rchp->ibcq, + rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + } + } else { + if (t4_clear_cq_armed(&rchp->cq) && rq_flushed) { + spin_lock_irqsave(&rchp->comp_handler_lock, flag); + (*rchp->ibcq.comp_handler)(&rchp->ibcq, + rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + } + if (t4_clear_cq_armed(&schp->cq) && sq_flushed) { + spin_lock_irqsave(&schp->comp_handler_lock, flag); + (*schp->ibcq.comp_handler)(&schp->ibcq, + schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, flag); + } + } +} + +static void flush_qp(struct c4iw_qp *qhp) +{ + struct c4iw_cq *rchp, *schp; + unsigned long flag; + + rchp = to_c4iw_cq(qhp->ibqp.recv_cq); + schp = to_c4iw_cq(qhp->ibqp.send_cq); + + t4_set_wq_in_error(&qhp->wq); + if (qhp->ibqp.uobject) { + t4_set_cq_in_error(&rchp->cq); + spin_lock_irqsave(&rchp->comp_handler_lock, flag); + (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + if (schp != rchp) { + t4_set_cq_in_error(&schp->cq); + spin_lock_irqsave(&schp->comp_handler_lock, flag); + (*schp->ibcq.comp_handler)(&schp->ibcq, + schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, flag); + } + return; + } + __flush_qp(qhp, rchp, schp); +} + +static int rdma_fini(struct c4iw_dev *rhp, struct c4iw_qp *qhp, + struct c4iw_ep *ep) +{ + struct fw_ri_wr *wqe; + int ret; + struct sk_buff *skb; + + PDBG("%s qhp %p qid 0x%x tid %u\n", __func__, qhp, qhp->wq.sq.qid, + ep->hwtid); + + skb = alloc_skb(sizeof *wqe, GFP_KERNEL); + if (!skb) + return -ENOMEM; + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); + + wqe = (struct fw_ri_wr *)__skb_put(skb, sizeof(*wqe)); + memset(wqe, 0, sizeof *wqe); + wqe->op_compl = cpu_to_be32( + FW_WR_OP_V(FW_RI_INIT_WR) | + FW_WR_COMPL_F); + wqe->flowid_len16 = cpu_to_be32( + FW_WR_FLOWID_V(ep->hwtid) | + FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*wqe), 16))); + wqe->cookie = (uintptr_t)&ep->com.wr_wait; + + wqe->u.fini.type = FW_RI_TYPE_FINI; + ret = c4iw_ofld_send(&rhp->rdev, skb); + if (ret) + goto out; + + ret = c4iw_wait_for_reply(&rhp->rdev, &ep->com.wr_wait, qhp->ep->hwtid, + qhp->wq.sq.qid, __func__); +out: + PDBG("%s ret %d\n", __func__, ret); + return ret; +} + +static void build_rtr_msg(u8 p2p_type, struct fw_ri_init *init) +{ + PDBG("%s p2p_type = %d\n", __func__, p2p_type); + memset(&init->u, 0, sizeof init->u); + switch (p2p_type) { + case FW_RI_INIT_P2PTYPE_RDMA_WRITE: + init->u.write.opcode = FW_RI_RDMA_WRITE_WR; + init->u.write.stag_sink = cpu_to_be32(1); + init->u.write.to_sink = cpu_to_be64(1); + init->u.write.u.immd_src[0].op = FW_RI_DATA_IMMD; + init->u.write.len16 = DIV_ROUND_UP(sizeof init->u.write + + sizeof(struct fw_ri_immd), + 16); + break; + case FW_RI_INIT_P2PTYPE_READ_REQ: + init->u.write.opcode = FW_RI_RDMA_READ_WR; + init->u.read.stag_src = cpu_to_be32(1); + init->u.read.to_src_lo = cpu_to_be32(1); + init->u.read.stag_sink = cpu_to_be32(1); + init->u.read.to_sink_lo = cpu_to_be32(1); + init->u.read.len16 = DIV_ROUND_UP(sizeof init->u.read, 16); + break; + } +} + +static int rdma_init(struct c4iw_dev *rhp, struct c4iw_qp *qhp) +{ + struct fw_ri_wr *wqe; + int ret; + struct sk_buff *skb; + + PDBG("%s qhp %p qid 0x%x tid %u ird %u ord %u\n", __func__, qhp, + qhp->wq.sq.qid, qhp->ep->hwtid, qhp->ep->ird, qhp->ep->ord); + + skb = alloc_skb(sizeof *wqe, GFP_KERNEL); + if (!skb) { + ret = -ENOMEM; + goto out; + } + ret = alloc_ird(rhp, qhp->attr.max_ird); + if (ret) { + qhp->attr.max_ird = 0; + kfree_skb(skb); + goto out; + } + set_wr_txq(skb, CPL_PRIORITY_DATA, qhp->ep->txq_idx); + + wqe = (struct fw_ri_wr *)__skb_put(skb, sizeof(*wqe)); + memset(wqe, 0, sizeof *wqe); + wqe->op_compl = cpu_to_be32( + FW_WR_OP_V(FW_RI_INIT_WR) | + FW_WR_COMPL_F); + wqe->flowid_len16 = cpu_to_be32( + FW_WR_FLOWID_V(qhp->ep->hwtid) | + FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*wqe), 16))); + + wqe->cookie = (uintptr_t)&qhp->ep->com.wr_wait; + + wqe->u.init.type = FW_RI_TYPE_INIT; + wqe->u.init.mpareqbit_p2ptype = + FW_RI_WR_MPAREQBIT_V(qhp->attr.mpa_attr.initiator) | + FW_RI_WR_P2PTYPE_V(qhp->attr.mpa_attr.p2p_type); + wqe->u.init.mpa_attrs = FW_RI_MPA_IETF_ENABLE; + if (qhp->attr.mpa_attr.recv_marker_enabled) + wqe->u.init.mpa_attrs |= FW_RI_MPA_RX_MARKER_ENABLE; + if (qhp->attr.mpa_attr.xmit_marker_enabled) + wqe->u.init.mpa_attrs |= FW_RI_MPA_TX_MARKER_ENABLE; + if (qhp->attr.mpa_attr.crc_enabled) + wqe->u.init.mpa_attrs |= FW_RI_MPA_CRC_ENABLE; + + wqe->u.init.qp_caps = FW_RI_QP_RDMA_READ_ENABLE | + FW_RI_QP_RDMA_WRITE_ENABLE | + FW_RI_QP_BIND_ENABLE; + if (!qhp->ibqp.uobject) + wqe->u.init.qp_caps |= FW_RI_QP_FAST_REGISTER_ENABLE | + FW_RI_QP_STAG0_ENABLE; + wqe->u.init.nrqe = cpu_to_be16(t4_rqes_posted(&qhp->wq)); + wqe->u.init.pdid = cpu_to_be32(qhp->attr.pd); + wqe->u.init.qpid = cpu_to_be32(qhp->wq.sq.qid); + wqe->u.init.sq_eqid = cpu_to_be32(qhp->wq.sq.qid); + wqe->u.init.rq_eqid = cpu_to_be32(qhp->wq.rq.qid); + wqe->u.init.scqid = cpu_to_be32(qhp->attr.scq); + wqe->u.init.rcqid = cpu_to_be32(qhp->attr.rcq); + wqe->u.init.ord_max = cpu_to_be32(qhp->attr.max_ord); + wqe->u.init.ird_max = cpu_to_be32(qhp->attr.max_ird); + wqe->u.init.iss = cpu_to_be32(qhp->ep->snd_seq); + wqe->u.init.irs = cpu_to_be32(qhp->ep->rcv_seq); + wqe->u.init.hwrqsize = cpu_to_be32(qhp->wq.rq.rqt_size); + wqe->u.init.hwrqaddr = cpu_to_be32(qhp->wq.rq.rqt_hwaddr - + rhp->rdev.lldi.vr->rq.start); + if (qhp->attr.mpa_attr.initiator) + build_rtr_msg(qhp->attr.mpa_attr.p2p_type, &wqe->u.init); + + ret = c4iw_ofld_send(&rhp->rdev, skb); + if (ret) + goto err1; + + ret = c4iw_wait_for_reply(&rhp->rdev, &qhp->ep->com.wr_wait, + qhp->ep->hwtid, qhp->wq.sq.qid, __func__); + if (!ret) + goto out; +err1: + free_ird(rhp, qhp->attr.max_ird); +out: + PDBG("%s ret %d\n", __func__, ret); + return ret; +} + +int c4iw_modify_qp(struct c4iw_dev *rhp, struct c4iw_qp *qhp, + enum c4iw_qp_attr_mask mask, + struct c4iw_qp_attributes *attrs, + int internal) +{ + int ret = 0; + struct c4iw_qp_attributes newattr = qhp->attr; + int disconnect = 0; + int terminate = 0; + int abort = 0; + int free = 0; + struct c4iw_ep *ep = NULL; + + PDBG("%s qhp %p sqid 0x%x rqid 0x%x ep %p state %d -> %d\n", __func__, + qhp, qhp->wq.sq.qid, qhp->wq.rq.qid, qhp->ep, qhp->attr.state, + (mask & C4IW_QP_ATTR_NEXT_STATE) ? attrs->next_state : -1); + + mutex_lock(&qhp->mutex); + + /* Process attr changes if in IDLE */ + if (mask & C4IW_QP_ATTR_VALID_MODIFY) { + if (qhp->attr.state != C4IW_QP_STATE_IDLE) { + ret = -EIO; + goto out; + } + if (mask & C4IW_QP_ATTR_ENABLE_RDMA_READ) + newattr.enable_rdma_read = attrs->enable_rdma_read; + if (mask & C4IW_QP_ATTR_ENABLE_RDMA_WRITE) + newattr.enable_rdma_write = attrs->enable_rdma_write; + if (mask & C4IW_QP_ATTR_ENABLE_RDMA_BIND) + newattr.enable_bind = attrs->enable_bind; + if (mask & C4IW_QP_ATTR_MAX_ORD) { + if (attrs->max_ord > c4iw_max_read_depth) { + ret = -EINVAL; + goto out; + } + newattr.max_ord = attrs->max_ord; + } + if (mask & C4IW_QP_ATTR_MAX_IRD) { + if (attrs->max_ird > cur_max_read_depth(rhp)) { + ret = -EINVAL; + goto out; + } + newattr.max_ird = attrs->max_ird; + } + qhp->attr = newattr; + } + + if (mask & C4IW_QP_ATTR_SQ_DB) { + ret = ring_kernel_sq_db(qhp, attrs->sq_db_inc); + goto out; + } + if (mask & C4IW_QP_ATTR_RQ_DB) { + ret = ring_kernel_rq_db(qhp, attrs->rq_db_inc); + goto out; + } + + if (!(mask & C4IW_QP_ATTR_NEXT_STATE)) + goto out; + if (qhp->attr.state == attrs->next_state) + goto out; + + switch (qhp->attr.state) { + case C4IW_QP_STATE_IDLE: + switch (attrs->next_state) { + case C4IW_QP_STATE_RTS: + if (!(mask & C4IW_QP_ATTR_LLP_STREAM_HANDLE)) { + ret = -EINVAL; + goto out; + } + if (!(mask & C4IW_QP_ATTR_MPA_ATTR)) { + ret = -EINVAL; + goto out; + } + qhp->attr.mpa_attr = attrs->mpa_attr; + qhp->attr.llp_stream_handle = attrs->llp_stream_handle; + qhp->ep = qhp->attr.llp_stream_handle; + set_state(qhp, C4IW_QP_STATE_RTS); + + /* + * Ref the endpoint here and deref when we + * disassociate the endpoint from the QP. This + * happens in CLOSING->IDLE transition or *->ERROR + * transition. + */ + c4iw_get_ep(&qhp->ep->com); + ret = rdma_init(rhp, qhp); + if (ret) + goto err; + break; + case C4IW_QP_STATE_ERROR: + set_state(qhp, C4IW_QP_STATE_ERROR); + flush_qp(qhp); + break; + default: + ret = -EINVAL; + goto out; + } + break; + case C4IW_QP_STATE_RTS: + switch (attrs->next_state) { + case C4IW_QP_STATE_CLOSING: + BUG_ON(atomic_read(&qhp->ep->com.kref.refcount) < 2); + t4_set_wq_in_error(&qhp->wq); + set_state(qhp, C4IW_QP_STATE_CLOSING); + ep = qhp->ep; + if (!internal) { + abort = 0; + disconnect = 1; + c4iw_get_ep(&qhp->ep->com); + } + ret = rdma_fini(rhp, qhp, ep); + if (ret) + goto err; + break; + case C4IW_QP_STATE_TERMINATE: + t4_set_wq_in_error(&qhp->wq); + set_state(qhp, C4IW_QP_STATE_TERMINATE); + qhp->attr.layer_etype = attrs->layer_etype; + qhp->attr.ecode = attrs->ecode; + ep = qhp->ep; + if (!internal) { + c4iw_get_ep(&qhp->ep->com); + terminate = 1; + disconnect = 1; + } else { + terminate = qhp->attr.send_term; + ret = rdma_fini(rhp, qhp, ep); + if (ret) + goto err; + } + break; + case C4IW_QP_STATE_ERROR: + t4_set_wq_in_error(&qhp->wq); + set_state(qhp, C4IW_QP_STATE_ERROR); + if (!internal) { + abort = 1; + disconnect = 1; + ep = qhp->ep; + c4iw_get_ep(&qhp->ep->com); + } + goto err; + break; + default: + ret = -EINVAL; + goto out; + } + break; + case C4IW_QP_STATE_CLOSING: + if (!internal) { + ret = -EINVAL; + goto out; + } + switch (attrs->next_state) { + case C4IW_QP_STATE_IDLE: + flush_qp(qhp); + set_state(qhp, C4IW_QP_STATE_IDLE); + qhp->attr.llp_stream_handle = NULL; + c4iw_put_ep(&qhp->ep->com); + qhp->ep = NULL; + wake_up(&qhp->wait); + break; + case C4IW_QP_STATE_ERROR: + goto err; + default: + ret = -EINVAL; + goto err; + } + break; + case C4IW_QP_STATE_ERROR: + if (attrs->next_state != C4IW_QP_STATE_IDLE) { + ret = -EINVAL; + goto out; + } + if (!t4_sq_empty(&qhp->wq) || !t4_rq_empty(&qhp->wq)) { + ret = -EINVAL; + goto out; + } + set_state(qhp, C4IW_QP_STATE_IDLE); + break; + case C4IW_QP_STATE_TERMINATE: + if (!internal) { + ret = -EINVAL; + goto out; + } + goto err; + break; + default: + printk(KERN_ERR "%s in a bad state %d\n", + __func__, qhp->attr.state); + ret = -EINVAL; + goto err; + break; + } + goto out; +err: + PDBG("%s disassociating ep %p qpid 0x%x\n", __func__, qhp->ep, + qhp->wq.sq.qid); + + /* disassociate the LLP connection */ + qhp->attr.llp_stream_handle = NULL; + if (!ep) + ep = qhp->ep; + qhp->ep = NULL; + set_state(qhp, C4IW_QP_STATE_ERROR); + free = 1; + abort = 1; + BUG_ON(!ep); + flush_qp(qhp); + wake_up(&qhp->wait); +out: + mutex_unlock(&qhp->mutex); + + if (terminate) + post_terminate(qhp, NULL, internal ? GFP_ATOMIC : GFP_KERNEL); + + /* + * If disconnect is 1, then we need to initiate a disconnect + * on the EP. This can be a normal close (RTS->CLOSING) or + * an abnormal close (RTS/CLOSING->ERROR). + */ + if (disconnect) { + c4iw_ep_disconnect(ep, abort, internal ? GFP_ATOMIC : + GFP_KERNEL); + c4iw_put_ep(&ep->com); + } + + /* + * If free is 1, then we've disassociated the EP from the QP + * and we need to dereference the EP. + */ + if (free) + c4iw_put_ep(&ep->com); + PDBG("%s exit state %d\n", __func__, qhp->attr.state); + return ret; +} + +int c4iw_destroy_qp(struct ib_qp *ib_qp) +{ + struct c4iw_dev *rhp; + struct c4iw_qp *qhp; + struct c4iw_qp_attributes attrs; + struct c4iw_ucontext *ucontext; + + qhp = to_c4iw_qp(ib_qp); + rhp = qhp->rhp; + + attrs.next_state = C4IW_QP_STATE_ERROR; + if (qhp->attr.state == C4IW_QP_STATE_TERMINATE) + c4iw_modify_qp(rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + else + c4iw_modify_qp(rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 0); + wait_event(qhp->wait, !qhp->ep); + + remove_handle(rhp, &rhp->qpidr, qhp->wq.sq.qid); + atomic_dec(&qhp->refcnt); + wait_event(qhp->wait, !atomic_read(&qhp->refcnt)); + + spin_lock_irq(&rhp->lock); + if (!list_empty(&qhp->db_fc_entry)) + list_del_init(&qhp->db_fc_entry); + spin_unlock_irq(&rhp->lock); + free_ird(rhp, qhp->attr.max_ird); + + ucontext = ib_qp->uobject ? + to_c4iw_ucontext(ib_qp->uobject->context) : NULL; + destroy_qp(&rhp->rdev, &qhp->wq, + ucontext ? &ucontext->uctx : &rhp->rdev.uctx); + + PDBG("%s ib_qp %p qpid 0x%0x\n", __func__, ib_qp, qhp->wq.sq.qid); + kfree(qhp); + return 0; +} + +struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs, + struct ib_udata *udata) +{ + struct c4iw_dev *rhp; + struct c4iw_qp *qhp; + struct c4iw_pd *php; + struct c4iw_cq *schp; + struct c4iw_cq *rchp; + struct c4iw_create_qp_resp uresp; + unsigned int sqsize, rqsize; + struct c4iw_ucontext *ucontext; + int ret; + struct c4iw_mm_entry *mm1, *mm2, *mm3, *mm4, *mm5 = NULL; + + PDBG("%s ib_pd %p\n", __func__, pd); + + if (attrs->qp_type != IB_QPT_RC) + return ERR_PTR(-EINVAL); + + php = to_c4iw_pd(pd); + rhp = php->rhp; + schp = get_chp(rhp, ((struct c4iw_cq *)attrs->send_cq)->cq.cqid); + rchp = get_chp(rhp, ((struct c4iw_cq *)attrs->recv_cq)->cq.cqid); + if (!schp || !rchp) + return ERR_PTR(-EINVAL); + + if (attrs->cap.max_inline_data > T4_MAX_SEND_INLINE) + return ERR_PTR(-EINVAL); + + if (attrs->cap.max_recv_wr > rhp->rdev.hw_queue.t4_max_rq_size) + return ERR_PTR(-E2BIG); + rqsize = attrs->cap.max_recv_wr + 1; + if (rqsize < 8) + rqsize = 8; + + if (attrs->cap.max_send_wr > rhp->rdev.hw_queue.t4_max_sq_size) + return ERR_PTR(-E2BIG); + sqsize = attrs->cap.max_send_wr + 1; + if (sqsize < 8) + sqsize = 8; + + ucontext = pd->uobject ? to_c4iw_ucontext(pd->uobject->context) : NULL; + + qhp = kzalloc(sizeof(*qhp), GFP_KERNEL); + if (!qhp) + return ERR_PTR(-ENOMEM); + qhp->wq.sq.size = sqsize; + qhp->wq.sq.memsize = + (sqsize + rhp->rdev.hw_queue.t4_eq_status_entries) * + sizeof(*qhp->wq.sq.queue) + 16 * sizeof(__be64); + qhp->wq.sq.flush_cidx = -1; + qhp->wq.rq.size = rqsize; + qhp->wq.rq.memsize = + (rqsize + rhp->rdev.hw_queue.t4_eq_status_entries) * + sizeof(*qhp->wq.rq.queue); + + if (ucontext) { + qhp->wq.sq.memsize = roundup(qhp->wq.sq.memsize, PAGE_SIZE); + qhp->wq.rq.memsize = roundup(qhp->wq.rq.memsize, PAGE_SIZE); + } + + ret = create_qp(&rhp->rdev, &qhp->wq, &schp->cq, &rchp->cq, + ucontext ? &ucontext->uctx : &rhp->rdev.uctx); + if (ret) + goto err1; + + attrs->cap.max_recv_wr = rqsize - 1; + attrs->cap.max_send_wr = sqsize - 1; + attrs->cap.max_inline_data = T4_MAX_SEND_INLINE; + + qhp->rhp = rhp; + qhp->attr.pd = php->pdid; + qhp->attr.scq = ((struct c4iw_cq *) attrs->send_cq)->cq.cqid; + qhp->attr.rcq = ((struct c4iw_cq *) attrs->recv_cq)->cq.cqid; + qhp->attr.sq_num_entries = attrs->cap.max_send_wr; + qhp->attr.rq_num_entries = attrs->cap.max_recv_wr; + qhp->attr.sq_max_sges = attrs->cap.max_send_sge; + qhp->attr.sq_max_sges_rdma_write = attrs->cap.max_send_sge; + qhp->attr.rq_max_sges = attrs->cap.max_recv_sge; + qhp->attr.state = C4IW_QP_STATE_IDLE; + qhp->attr.next_state = C4IW_QP_STATE_IDLE; + qhp->attr.enable_rdma_read = 1; + qhp->attr.enable_rdma_write = 1; + qhp->attr.enable_bind = 1; + qhp->attr.max_ord = 0; + qhp->attr.max_ird = 0; + qhp->sq_sig_all = attrs->sq_sig_type == IB_SIGNAL_ALL_WR; + spin_lock_init(&qhp->lock); + mutex_init(&qhp->mutex); + init_waitqueue_head(&qhp->wait); + atomic_set(&qhp->refcnt, 1); + + ret = insert_handle(rhp, &rhp->qpidr, qhp, qhp->wq.sq.qid); + if (ret) + goto err2; + + if (udata) { + mm1 = kmalloc(sizeof *mm1, GFP_KERNEL); + if (!mm1) { + ret = -ENOMEM; + goto err3; + } + mm2 = kmalloc(sizeof *mm2, GFP_KERNEL); + if (!mm2) { + ret = -ENOMEM; + goto err4; + } + mm3 = kmalloc(sizeof *mm3, GFP_KERNEL); + if (!mm3) { + ret = -ENOMEM; + goto err5; + } + mm4 = kmalloc(sizeof *mm4, GFP_KERNEL); + if (!mm4) { + ret = -ENOMEM; + goto err6; + } + if (t4_sq_onchip(&qhp->wq.sq)) { + mm5 = kmalloc(sizeof *mm5, GFP_KERNEL); + if (!mm5) { + ret = -ENOMEM; + goto err7; + } + uresp.flags = C4IW_QPF_ONCHIP; + } else + uresp.flags = 0; + uresp.qid_mask = rhp->rdev.qpmask; + uresp.sqid = qhp->wq.sq.qid; + uresp.sq_size = qhp->wq.sq.size; + uresp.sq_memsize = qhp->wq.sq.memsize; + uresp.rqid = qhp->wq.rq.qid; + uresp.rq_size = qhp->wq.rq.size; + uresp.rq_memsize = qhp->wq.rq.memsize; + spin_lock(&ucontext->mmap_lock); + if (mm5) { + uresp.ma_sync_key = ucontext->key; + ucontext->key += PAGE_SIZE; + } else { + uresp.ma_sync_key = 0; + } + uresp.sq_key = ucontext->key; + ucontext->key += PAGE_SIZE; + uresp.rq_key = ucontext->key; + ucontext->key += PAGE_SIZE; + uresp.sq_db_gts_key = ucontext->key; + ucontext->key += PAGE_SIZE; + uresp.rq_db_gts_key = ucontext->key; + ucontext->key += PAGE_SIZE; + spin_unlock(&ucontext->mmap_lock); + ret = ib_copy_to_udata(udata, &uresp, sizeof uresp); + if (ret) + goto err8; + mm1->key = uresp.sq_key; + mm1->addr = qhp->wq.sq.phys_addr; + mm1->len = PAGE_ALIGN(qhp->wq.sq.memsize); + insert_mmap(ucontext, mm1); + mm2->key = uresp.rq_key; + mm2->addr = virt_to_phys(qhp->wq.rq.queue); + mm2->len = PAGE_ALIGN(qhp->wq.rq.memsize); + insert_mmap(ucontext, mm2); + mm3->key = uresp.sq_db_gts_key; + mm3->addr = (__force unsigned long)qhp->wq.sq.udb; + mm3->len = PAGE_SIZE; + insert_mmap(ucontext, mm3); + mm4->key = uresp.rq_db_gts_key; + mm4->addr = (__force unsigned long)qhp->wq.rq.udb; + mm4->len = PAGE_SIZE; + insert_mmap(ucontext, mm4); + if (mm5) { + mm5->key = uresp.ma_sync_key; + mm5->addr = (pci_resource_start(rhp->rdev.lldi.pdev, 0) + + PCIE_MA_SYNC_A) & PAGE_MASK; + mm5->len = PAGE_SIZE; + insert_mmap(ucontext, mm5); + } + } + qhp->ibqp.qp_num = qhp->wq.sq.qid; + init_timer(&(qhp->timer)); + INIT_LIST_HEAD(&qhp->db_fc_entry); + PDBG("%s sq id %u size %u memsize %zu num_entries %u " + "rq id %u size %u memsize %zu num_entries %u\n", __func__, + qhp->wq.sq.qid, qhp->wq.sq.size, qhp->wq.sq.memsize, + attrs->cap.max_send_wr, qhp->wq.rq.qid, qhp->wq.rq.size, + qhp->wq.rq.memsize, attrs->cap.max_recv_wr); + return &qhp->ibqp; +err8: + kfree(mm5); +err7: + kfree(mm4); +err6: + kfree(mm3); +err5: + kfree(mm2); +err4: + kfree(mm1); +err3: + remove_handle(rhp, &rhp->qpidr, qhp->wq.sq.qid); +err2: + destroy_qp(&rhp->rdev, &qhp->wq, + ucontext ? &ucontext->uctx : &rhp->rdev.uctx); +err1: + kfree(qhp); + return ERR_PTR(ret); +} + +int c4iw_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, + int attr_mask, struct ib_udata *udata) +{ + struct c4iw_dev *rhp; + struct c4iw_qp *qhp; + enum c4iw_qp_attr_mask mask = 0; + struct c4iw_qp_attributes attrs; + + PDBG("%s ib_qp %p\n", __func__, ibqp); + + /* iwarp does not support the RTR state */ + if ((attr_mask & IB_QP_STATE) && (attr->qp_state == IB_QPS_RTR)) + attr_mask &= ~IB_QP_STATE; + + /* Make sure we still have something left to do */ + if (!attr_mask) + return 0; + + memset(&attrs, 0, sizeof attrs); + qhp = to_c4iw_qp(ibqp); + rhp = qhp->rhp; + + attrs.next_state = c4iw_convert_state(attr->qp_state); + attrs.enable_rdma_read = (attr->qp_access_flags & + IB_ACCESS_REMOTE_READ) ? 1 : 0; + attrs.enable_rdma_write = (attr->qp_access_flags & + IB_ACCESS_REMOTE_WRITE) ? 1 : 0; + attrs.enable_bind = (attr->qp_access_flags & IB_ACCESS_MW_BIND) ? 1 : 0; + + + mask |= (attr_mask & IB_QP_STATE) ? C4IW_QP_ATTR_NEXT_STATE : 0; + mask |= (attr_mask & IB_QP_ACCESS_FLAGS) ? + (C4IW_QP_ATTR_ENABLE_RDMA_READ | + C4IW_QP_ATTR_ENABLE_RDMA_WRITE | + C4IW_QP_ATTR_ENABLE_RDMA_BIND) : 0; + + /* + * Use SQ_PSN and RQ_PSN to pass in IDX_INC values for + * ringing the queue db when we're in DB_FULL mode. + * Only allow this on T4 devices. + */ + attrs.sq_db_inc = attr->sq_psn; + attrs.rq_db_inc = attr->rq_psn; + mask |= (attr_mask & IB_QP_SQ_PSN) ? C4IW_QP_ATTR_SQ_DB : 0; + mask |= (attr_mask & IB_QP_RQ_PSN) ? C4IW_QP_ATTR_RQ_DB : 0; + if (is_t5(to_c4iw_qp(ibqp)->rhp->rdev.lldi.adapter_type) && + (mask & (C4IW_QP_ATTR_SQ_DB|C4IW_QP_ATTR_RQ_DB))) + return -EINVAL; + + return c4iw_modify_qp(rhp, qhp, mask, &attrs, 0); +} + +struct ib_qp *c4iw_get_qp(struct ib_device *dev, int qpn) +{ + PDBG("%s ib_dev %p qpn 0x%x\n", __func__, dev, qpn); + return (struct ib_qp *)get_qhp(to_c4iw_dev(dev), qpn); +} + +int c4iw_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, + int attr_mask, struct ib_qp_init_attr *init_attr) +{ + struct c4iw_qp *qhp = to_c4iw_qp(ibqp); + + memset(attr, 0, sizeof *attr); + memset(init_attr, 0, sizeof *init_attr); + attr->qp_state = to_ib_qp_state(qhp->attr.state); + init_attr->cap.max_send_wr = qhp->attr.sq_num_entries; + init_attr->cap.max_recv_wr = qhp->attr.rq_num_entries; + init_attr->cap.max_send_sge = qhp->attr.sq_max_sges; + init_attr->cap.max_recv_sge = qhp->attr.sq_max_sges; + init_attr->cap.max_inline_data = T4_MAX_SEND_INLINE; + init_attr->sq_sig_type = qhp->sq_sig_all ? IB_SIGNAL_ALL_WR : 0; + return 0; +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/resource.c b/kernel/drivers/infiniband/hw/cxgb4/resource.c new file mode 100644 index 000000000..67df71a70 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/resource.c @@ -0,0 +1,453 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +/* Crude resource management */ +#include <linux/spinlock.h> +#include <linux/genalloc.h> +#include <linux/ratelimit.h> +#include "iw_cxgb4.h" + +static int c4iw_init_qid_table(struct c4iw_rdev *rdev) +{ + u32 i; + + if (c4iw_id_table_alloc(&rdev->resource.qid_table, + rdev->lldi.vr->qp.start, + rdev->lldi.vr->qp.size, + rdev->lldi.vr->qp.size, 0)) + return -ENOMEM; + + for (i = rdev->lldi.vr->qp.start; + i < rdev->lldi.vr->qp.start + rdev->lldi.vr->qp.size; i++) + if (!(i & rdev->qpmask)) + c4iw_id_free(&rdev->resource.qid_table, i); + return 0; +} + +/* nr_* must be power of 2 */ +int c4iw_init_resource(struct c4iw_rdev *rdev, u32 nr_tpt, u32 nr_pdid) +{ + int err = 0; + err = c4iw_id_table_alloc(&rdev->resource.tpt_table, 0, nr_tpt, 1, + C4IW_ID_TABLE_F_RANDOM); + if (err) + goto tpt_err; + err = c4iw_init_qid_table(rdev); + if (err) + goto qid_err; + err = c4iw_id_table_alloc(&rdev->resource.pdid_table, 0, + nr_pdid, 1, 0); + if (err) + goto pdid_err; + return 0; + pdid_err: + c4iw_id_table_free(&rdev->resource.qid_table); + qid_err: + c4iw_id_table_free(&rdev->resource.tpt_table); + tpt_err: + return -ENOMEM; +} + +/* + * returns 0 if no resource available + */ +u32 c4iw_get_resource(struct c4iw_id_table *id_table) +{ + u32 entry; + entry = c4iw_id_alloc(id_table); + if (entry == (u32)(-1)) + return 0; + return entry; +} + +void c4iw_put_resource(struct c4iw_id_table *id_table, u32 entry) +{ + PDBG("%s entry 0x%x\n", __func__, entry); + c4iw_id_free(id_table, entry); +} + +u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx) +{ + struct c4iw_qid_list *entry; + u32 qid; + int i; + + mutex_lock(&uctx->lock); + if (!list_empty(&uctx->cqids)) { + entry = list_entry(uctx->cqids.next, struct c4iw_qid_list, + entry); + list_del(&entry->entry); + qid = entry->qid; + kfree(entry); + } else { + qid = c4iw_get_resource(&rdev->resource.qid_table); + if (!qid) + goto out; + mutex_lock(&rdev->stats.lock); + rdev->stats.qid.cur += rdev->qpmask + 1; + mutex_unlock(&rdev->stats.lock); + for (i = qid+1; i & rdev->qpmask; i++) { + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + goto out; + entry->qid = i; + list_add_tail(&entry->entry, &uctx->cqids); + } + + /* + * now put the same ids on the qp list since they all + * map to the same db/gts page. + */ + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + goto out; + entry->qid = qid; + list_add_tail(&entry->entry, &uctx->qpids); + for (i = qid+1; i & rdev->qpmask; i++) { + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + goto out; + entry->qid = i; + list_add_tail(&entry->entry, &uctx->qpids); + } + } +out: + mutex_unlock(&uctx->lock); + PDBG("%s qid 0x%x\n", __func__, qid); + mutex_lock(&rdev->stats.lock); + if (rdev->stats.qid.cur > rdev->stats.qid.max) + rdev->stats.qid.max = rdev->stats.qid.cur; + mutex_unlock(&rdev->stats.lock); + return qid; +} + +void c4iw_put_cqid(struct c4iw_rdev *rdev, u32 qid, + struct c4iw_dev_ucontext *uctx) +{ + struct c4iw_qid_list *entry; + + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + return; + PDBG("%s qid 0x%x\n", __func__, qid); + entry->qid = qid; + mutex_lock(&uctx->lock); + list_add_tail(&entry->entry, &uctx->cqids); + mutex_unlock(&uctx->lock); +} + +u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx) +{ + struct c4iw_qid_list *entry; + u32 qid; + int i; + + mutex_lock(&uctx->lock); + if (!list_empty(&uctx->qpids)) { + entry = list_entry(uctx->qpids.next, struct c4iw_qid_list, + entry); + list_del(&entry->entry); + qid = entry->qid; + kfree(entry); + } else { + qid = c4iw_get_resource(&rdev->resource.qid_table); + if (!qid) { + mutex_lock(&rdev->stats.lock); + rdev->stats.qid.fail++; + mutex_unlock(&rdev->stats.lock); + goto out; + } + mutex_lock(&rdev->stats.lock); + rdev->stats.qid.cur += rdev->qpmask + 1; + mutex_unlock(&rdev->stats.lock); + for (i = qid+1; i & rdev->qpmask; i++) { + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + goto out; + entry->qid = i; + list_add_tail(&entry->entry, &uctx->qpids); + } + + /* + * now put the same ids on the cq list since they all + * map to the same db/gts page. + */ + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + goto out; + entry->qid = qid; + list_add_tail(&entry->entry, &uctx->cqids); + for (i = qid; i & rdev->qpmask; i++) { + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + goto out; + entry->qid = i; + list_add_tail(&entry->entry, &uctx->cqids); + } + } +out: + mutex_unlock(&uctx->lock); + PDBG("%s qid 0x%x\n", __func__, qid); + mutex_lock(&rdev->stats.lock); + if (rdev->stats.qid.cur > rdev->stats.qid.max) + rdev->stats.qid.max = rdev->stats.qid.cur; + mutex_unlock(&rdev->stats.lock); + return qid; +} + +void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qid, + struct c4iw_dev_ucontext *uctx) +{ + struct c4iw_qid_list *entry; + + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + return; + PDBG("%s qid 0x%x\n", __func__, qid); + entry->qid = qid; + mutex_lock(&uctx->lock); + list_add_tail(&entry->entry, &uctx->qpids); + mutex_unlock(&uctx->lock); +} + +void c4iw_destroy_resource(struct c4iw_resource *rscp) +{ + c4iw_id_table_free(&rscp->tpt_table); + c4iw_id_table_free(&rscp->qid_table); + c4iw_id_table_free(&rscp->pdid_table); +} + +/* + * PBL Memory Manager. Uses Linux generic allocator. + */ + +#define MIN_PBL_SHIFT 8 /* 256B == min PBL size (32 entries) */ + +u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size) +{ + unsigned long addr = gen_pool_alloc(rdev->pbl_pool, size); + PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size); + mutex_lock(&rdev->stats.lock); + if (addr) { + rdev->stats.pbl.cur += roundup(size, 1 << MIN_PBL_SHIFT); + if (rdev->stats.pbl.cur > rdev->stats.pbl.max) + rdev->stats.pbl.max = rdev->stats.pbl.cur; + } else + rdev->stats.pbl.fail++; + mutex_unlock(&rdev->stats.lock); + return (u32)addr; +} + +void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size) +{ + PDBG("%s addr 0x%x size %d\n", __func__, addr, size); + mutex_lock(&rdev->stats.lock); + rdev->stats.pbl.cur -= roundup(size, 1 << MIN_PBL_SHIFT); + mutex_unlock(&rdev->stats.lock); + gen_pool_free(rdev->pbl_pool, (unsigned long)addr, size); +} + +int c4iw_pblpool_create(struct c4iw_rdev *rdev) +{ + unsigned pbl_start, pbl_chunk, pbl_top; + + rdev->pbl_pool = gen_pool_create(MIN_PBL_SHIFT, -1); + if (!rdev->pbl_pool) + return -ENOMEM; + + pbl_start = rdev->lldi.vr->pbl.start; + pbl_chunk = rdev->lldi.vr->pbl.size; + pbl_top = pbl_start + pbl_chunk; + + while (pbl_start < pbl_top) { + pbl_chunk = min(pbl_top - pbl_start + 1, pbl_chunk); + if (gen_pool_add(rdev->pbl_pool, pbl_start, pbl_chunk, -1)) { + PDBG("%s failed to add PBL chunk (%x/%x)\n", + __func__, pbl_start, pbl_chunk); + if (pbl_chunk <= 1024 << MIN_PBL_SHIFT) { + printk(KERN_WARNING MOD + "Failed to add all PBL chunks (%x/%x)\n", + pbl_start, + pbl_top - pbl_start); + return 0; + } + pbl_chunk >>= 1; + } else { + PDBG("%s added PBL chunk (%x/%x)\n", + __func__, pbl_start, pbl_chunk); + pbl_start += pbl_chunk; + } + } + + return 0; +} + +void c4iw_pblpool_destroy(struct c4iw_rdev *rdev) +{ + gen_pool_destroy(rdev->pbl_pool); +} + +/* + * RQT Memory Manager. Uses Linux generic allocator. + */ + +#define MIN_RQT_SHIFT 10 /* 1KB == min RQT size (16 entries) */ + +u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size) +{ + unsigned long addr = gen_pool_alloc(rdev->rqt_pool, size << 6); + PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size << 6); + if (!addr) + pr_warn_ratelimited(MOD "%s: Out of RQT memory\n", + pci_name(rdev->lldi.pdev)); + mutex_lock(&rdev->stats.lock); + if (addr) { + rdev->stats.rqt.cur += roundup(size << 6, 1 << MIN_RQT_SHIFT); + if (rdev->stats.rqt.cur > rdev->stats.rqt.max) + rdev->stats.rqt.max = rdev->stats.rqt.cur; + } else + rdev->stats.rqt.fail++; + mutex_unlock(&rdev->stats.lock); + return (u32)addr; +} + +void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size) +{ + PDBG("%s addr 0x%x size %d\n", __func__, addr, size << 6); + mutex_lock(&rdev->stats.lock); + rdev->stats.rqt.cur -= roundup(size << 6, 1 << MIN_RQT_SHIFT); + mutex_unlock(&rdev->stats.lock); + gen_pool_free(rdev->rqt_pool, (unsigned long)addr, size << 6); +} + +int c4iw_rqtpool_create(struct c4iw_rdev *rdev) +{ + unsigned rqt_start, rqt_chunk, rqt_top; + + rdev->rqt_pool = gen_pool_create(MIN_RQT_SHIFT, -1); + if (!rdev->rqt_pool) + return -ENOMEM; + + rqt_start = rdev->lldi.vr->rq.start; + rqt_chunk = rdev->lldi.vr->rq.size; + rqt_top = rqt_start + rqt_chunk; + + while (rqt_start < rqt_top) { + rqt_chunk = min(rqt_top - rqt_start + 1, rqt_chunk); + if (gen_pool_add(rdev->rqt_pool, rqt_start, rqt_chunk, -1)) { + PDBG("%s failed to add RQT chunk (%x/%x)\n", + __func__, rqt_start, rqt_chunk); + if (rqt_chunk <= 1024 << MIN_RQT_SHIFT) { + printk(KERN_WARNING MOD + "Failed to add all RQT chunks (%x/%x)\n", + rqt_start, rqt_top - rqt_start); + return 0; + } + rqt_chunk >>= 1; + } else { + PDBG("%s added RQT chunk (%x/%x)\n", + __func__, rqt_start, rqt_chunk); + rqt_start += rqt_chunk; + } + } + return 0; +} + +void c4iw_rqtpool_destroy(struct c4iw_rdev *rdev) +{ + gen_pool_destroy(rdev->rqt_pool); +} + +/* + * On-Chip QP Memory. + */ +#define MIN_OCQP_SHIFT 12 /* 4KB == min ocqp size */ + +u32 c4iw_ocqp_pool_alloc(struct c4iw_rdev *rdev, int size) +{ + unsigned long addr = gen_pool_alloc(rdev->ocqp_pool, size); + PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size); + if (addr) { + mutex_lock(&rdev->stats.lock); + rdev->stats.ocqp.cur += roundup(size, 1 << MIN_OCQP_SHIFT); + if (rdev->stats.ocqp.cur > rdev->stats.ocqp.max) + rdev->stats.ocqp.max = rdev->stats.ocqp.cur; + mutex_unlock(&rdev->stats.lock); + } + return (u32)addr; +} + +void c4iw_ocqp_pool_free(struct c4iw_rdev *rdev, u32 addr, int size) +{ + PDBG("%s addr 0x%x size %d\n", __func__, addr, size); + mutex_lock(&rdev->stats.lock); + rdev->stats.ocqp.cur -= roundup(size, 1 << MIN_OCQP_SHIFT); + mutex_unlock(&rdev->stats.lock); + gen_pool_free(rdev->ocqp_pool, (unsigned long)addr, size); +} + +int c4iw_ocqp_pool_create(struct c4iw_rdev *rdev) +{ + unsigned start, chunk, top; + + rdev->ocqp_pool = gen_pool_create(MIN_OCQP_SHIFT, -1); + if (!rdev->ocqp_pool) + return -ENOMEM; + + start = rdev->lldi.vr->ocq.start; + chunk = rdev->lldi.vr->ocq.size; + top = start + chunk; + + while (start < top) { + chunk = min(top - start + 1, chunk); + if (gen_pool_add(rdev->ocqp_pool, start, chunk, -1)) { + PDBG("%s failed to add OCQP chunk (%x/%x)\n", + __func__, start, chunk); + if (chunk <= 1024 << MIN_OCQP_SHIFT) { + printk(KERN_WARNING MOD + "Failed to add all OCQP chunks (%x/%x)\n", + start, top - start); + return 0; + } + chunk >>= 1; + } else { + PDBG("%s added OCQP chunk (%x/%x)\n", + __func__, start, chunk); + start += chunk; + } + } + return 0; +} + +void c4iw_ocqp_pool_destroy(struct c4iw_rdev *rdev) +{ + gen_pool_destroy(rdev->ocqp_pool); +} diff --git a/kernel/drivers/infiniband/hw/cxgb4/t4.h b/kernel/drivers/infiniband/hw/cxgb4/t4.h new file mode 100644 index 000000000..7f2a6c244 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/t4.h @@ -0,0 +1,685 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#ifndef __T4_H__ +#define __T4_H__ + +#include "t4_hw.h" +#include "t4_regs.h" +#include "t4_msg.h" +#include "t4fw_ri_api.h" + +#define T4_MAX_NUM_PD 65536 +#define T4_MAX_MR_SIZE (~0ULL) +#define T4_PAGESIZE_MASK 0xffff000 /* 4KB-128MB */ +#define T4_STAG_UNSET 0xffffffff +#define T4_FW_MAJ 0 +#define PCIE_MA_SYNC_A 0x30b4 + +struct t4_status_page { + __be32 rsvd1; /* flit 0 - hw owns */ + __be16 rsvd2; + __be16 qid; + __be16 cidx; + __be16 pidx; + u8 qp_err; /* flit 1 - sw owns */ + u8 db_off; + u8 pad; + u16 host_wq_pidx; + u16 host_cidx; + u16 host_pidx; +}; + +#define T4_EQ_ENTRY_SIZE 64 + +#define T4_SQ_NUM_SLOTS 5 +#define T4_SQ_NUM_BYTES (T4_EQ_ENTRY_SIZE * T4_SQ_NUM_SLOTS) +#define T4_MAX_SEND_SGE ((T4_SQ_NUM_BYTES - sizeof(struct fw_ri_send_wr) - \ + sizeof(struct fw_ri_isgl)) / sizeof(struct fw_ri_sge)) +#define T4_MAX_SEND_INLINE ((T4_SQ_NUM_BYTES - sizeof(struct fw_ri_send_wr) - \ + sizeof(struct fw_ri_immd))) +#define T4_MAX_WRITE_INLINE ((T4_SQ_NUM_BYTES - \ + sizeof(struct fw_ri_rdma_write_wr) - \ + sizeof(struct fw_ri_immd))) +#define T4_MAX_WRITE_SGE ((T4_SQ_NUM_BYTES - \ + sizeof(struct fw_ri_rdma_write_wr) - \ + sizeof(struct fw_ri_isgl)) / sizeof(struct fw_ri_sge)) +#define T4_MAX_FR_IMMD ((T4_SQ_NUM_BYTES - sizeof(struct fw_ri_fr_nsmr_wr) - \ + sizeof(struct fw_ri_immd)) & ~31UL) +#define T4_MAX_FR_IMMD_DEPTH (T4_MAX_FR_IMMD / sizeof(u64)) +#define T4_MAX_FR_DSGL 1024 +#define T4_MAX_FR_DSGL_DEPTH (T4_MAX_FR_DSGL / sizeof(u64)) + +static inline int t4_max_fr_depth(int use_dsgl) +{ + return use_dsgl ? T4_MAX_FR_DSGL_DEPTH : T4_MAX_FR_IMMD_DEPTH; +} + +#define T4_RQ_NUM_SLOTS 2 +#define T4_RQ_NUM_BYTES (T4_EQ_ENTRY_SIZE * T4_RQ_NUM_SLOTS) +#define T4_MAX_RECV_SGE 4 + +union t4_wr { + struct fw_ri_res_wr res; + struct fw_ri_wr ri; + struct fw_ri_rdma_write_wr write; + struct fw_ri_send_wr send; + struct fw_ri_rdma_read_wr read; + struct fw_ri_bind_mw_wr bind; + struct fw_ri_fr_nsmr_wr fr; + struct fw_ri_inv_lstag_wr inv; + struct t4_status_page status; + __be64 flits[T4_EQ_ENTRY_SIZE / sizeof(__be64) * T4_SQ_NUM_SLOTS]; +}; + +union t4_recv_wr { + struct fw_ri_recv_wr recv; + struct t4_status_page status; + __be64 flits[T4_EQ_ENTRY_SIZE / sizeof(__be64) * T4_RQ_NUM_SLOTS]; +}; + +static inline void init_wr_hdr(union t4_wr *wqe, u16 wrid, + enum fw_wr_opcodes opcode, u8 flags, u8 len16) +{ + wqe->send.opcode = (u8)opcode; + wqe->send.flags = flags; + wqe->send.wrid = wrid; + wqe->send.r1[0] = 0; + wqe->send.r1[1] = 0; + wqe->send.r1[2] = 0; + wqe->send.len16 = len16; +} + +/* CQE/AE status codes */ +#define T4_ERR_SUCCESS 0x0 +#define T4_ERR_STAG 0x1 /* STAG invalid: either the */ + /* STAG is offlimt, being 0, */ + /* or STAG_key mismatch */ +#define T4_ERR_PDID 0x2 /* PDID mismatch */ +#define T4_ERR_QPID 0x3 /* QPID mismatch */ +#define T4_ERR_ACCESS 0x4 /* Invalid access right */ +#define T4_ERR_WRAP 0x5 /* Wrap error */ +#define T4_ERR_BOUND 0x6 /* base and bounds voilation */ +#define T4_ERR_INVALIDATE_SHARED_MR 0x7 /* attempt to invalidate a */ + /* shared memory region */ +#define T4_ERR_INVALIDATE_MR_WITH_MW_BOUND 0x8 /* attempt to invalidate a */ + /* shared memory region */ +#define T4_ERR_ECC 0x9 /* ECC error detected */ +#define T4_ERR_ECC_PSTAG 0xA /* ECC error detected when */ + /* reading PSTAG for a MW */ + /* Invalidate */ +#define T4_ERR_PBL_ADDR_BOUND 0xB /* pbl addr out of bounds: */ + /* software error */ +#define T4_ERR_SWFLUSH 0xC /* SW FLUSHED */ +#define T4_ERR_CRC 0x10 /* CRC error */ +#define T4_ERR_MARKER 0x11 /* Marker error */ +#define T4_ERR_PDU_LEN_ERR 0x12 /* invalid PDU length */ +#define T4_ERR_OUT_OF_RQE 0x13 /* out of RQE */ +#define T4_ERR_DDP_VERSION 0x14 /* wrong DDP version */ +#define T4_ERR_RDMA_VERSION 0x15 /* wrong RDMA version */ +#define T4_ERR_OPCODE 0x16 /* invalid rdma opcode */ +#define T4_ERR_DDP_QUEUE_NUM 0x17 /* invalid ddp queue number */ +#define T4_ERR_MSN 0x18 /* MSN error */ +#define T4_ERR_TBIT 0x19 /* tag bit not set correctly */ +#define T4_ERR_MO 0x1A /* MO not 0 for TERMINATE */ + /* or READ_REQ */ +#define T4_ERR_MSN_GAP 0x1B +#define T4_ERR_MSN_RANGE 0x1C +#define T4_ERR_IRD_OVERFLOW 0x1D +#define T4_ERR_RQE_ADDR_BOUND 0x1E /* RQE addr out of bounds: */ + /* software error */ +#define T4_ERR_INTERNAL_ERR 0x1F /* internal error (opcode */ + /* mismatch) */ +/* + * CQE defs + */ +struct t4_cqe { + __be32 header; + __be32 len; + union { + struct { + __be32 stag; + __be32 msn; + } rcqe; + struct { + u32 nada1; + u16 nada2; + u16 cidx; + } scqe; + struct { + __be32 wrid_hi; + __be32 wrid_low; + } gen; + } u; + __be64 reserved; + __be64 bits_type_ts; +}; + +/* macros for flit 0 of the cqe */ + +#define CQE_QPID_S 12 +#define CQE_QPID_M 0xFFFFF +#define CQE_QPID_G(x) ((((x) >> CQE_QPID_S)) & CQE_QPID_M) +#define CQE_QPID_V(x) ((x)<<CQE_QPID_S) + +#define CQE_SWCQE_S 11 +#define CQE_SWCQE_M 0x1 +#define CQE_SWCQE_G(x) ((((x) >> CQE_SWCQE_S)) & CQE_SWCQE_M) +#define CQE_SWCQE_V(x) ((x)<<CQE_SWCQE_S) + +#define CQE_STATUS_S 5 +#define CQE_STATUS_M 0x1F +#define CQE_STATUS_G(x) ((((x) >> CQE_STATUS_S)) & CQE_STATUS_M) +#define CQE_STATUS_V(x) ((x)<<CQE_STATUS_S) + +#define CQE_TYPE_S 4 +#define CQE_TYPE_M 0x1 +#define CQE_TYPE_G(x) ((((x) >> CQE_TYPE_S)) & CQE_TYPE_M) +#define CQE_TYPE_V(x) ((x)<<CQE_TYPE_S) + +#define CQE_OPCODE_S 0 +#define CQE_OPCODE_M 0xF +#define CQE_OPCODE_G(x) ((((x) >> CQE_OPCODE_S)) & CQE_OPCODE_M) +#define CQE_OPCODE_V(x) ((x)<<CQE_OPCODE_S) + +#define SW_CQE(x) (CQE_SWCQE_G(be32_to_cpu((x)->header))) +#define CQE_QPID(x) (CQE_QPID_G(be32_to_cpu((x)->header))) +#define CQE_TYPE(x) (CQE_TYPE_G(be32_to_cpu((x)->header))) +#define SQ_TYPE(x) (CQE_TYPE((x))) +#define RQ_TYPE(x) (!CQE_TYPE((x))) +#define CQE_STATUS(x) (CQE_STATUS_G(be32_to_cpu((x)->header))) +#define CQE_OPCODE(x) (CQE_OPCODE_G(be32_to_cpu((x)->header))) + +#define CQE_SEND_OPCODE(x)( \ + (CQE_OPCODE_G(be32_to_cpu((x)->header)) == FW_RI_SEND) || \ + (CQE_OPCODE_G(be32_to_cpu((x)->header)) == FW_RI_SEND_WITH_SE) || \ + (CQE_OPCODE_G(be32_to_cpu((x)->header)) == FW_RI_SEND_WITH_INV) || \ + (CQE_OPCODE_G(be32_to_cpu((x)->header)) == FW_RI_SEND_WITH_SE_INV)) + +#define CQE_LEN(x) (be32_to_cpu((x)->len)) + +/* used for RQ completion processing */ +#define CQE_WRID_STAG(x) (be32_to_cpu((x)->u.rcqe.stag)) +#define CQE_WRID_MSN(x) (be32_to_cpu((x)->u.rcqe.msn)) + +/* used for SQ completion processing */ +#define CQE_WRID_SQ_IDX(x) ((x)->u.scqe.cidx) + +/* generic accessor macros */ +#define CQE_WRID_HI(x) (be32_to_cpu((x)->u.gen.wrid_hi)) +#define CQE_WRID_LOW(x) (be32_to_cpu((x)->u.gen.wrid_low)) + +/* macros for flit 3 of the cqe */ +#define CQE_GENBIT_S 63 +#define CQE_GENBIT_M 0x1 +#define CQE_GENBIT_G(x) (((x) >> CQE_GENBIT_S) & CQE_GENBIT_M) +#define CQE_GENBIT_V(x) ((x)<<CQE_GENBIT_S) + +#define CQE_OVFBIT_S 62 +#define CQE_OVFBIT_M 0x1 +#define CQE_OVFBIT_G(x) ((((x) >> CQE_OVFBIT_S)) & CQE_OVFBIT_M) + +#define CQE_IQTYPE_S 60 +#define CQE_IQTYPE_M 0x3 +#define CQE_IQTYPE_G(x) ((((x) >> CQE_IQTYPE_S)) & CQE_IQTYPE_M) + +#define CQE_TS_M 0x0fffffffffffffffULL +#define CQE_TS_G(x) ((x) & CQE_TS_M) + +#define CQE_OVFBIT(x) ((unsigned)CQE_OVFBIT_G(be64_to_cpu((x)->bits_type_ts))) +#define CQE_GENBIT(x) ((unsigned)CQE_GENBIT_G(be64_to_cpu((x)->bits_type_ts))) +#define CQE_TS(x) (CQE_TS_G(be64_to_cpu((x)->bits_type_ts))) + +struct t4_swsqe { + u64 wr_id; + struct t4_cqe cqe; + int read_len; + int opcode; + int complete; + int signaled; + u16 idx; + int flushed; + struct timespec host_ts; + u64 sge_ts; +}; + +static inline pgprot_t t4_pgprot_wc(pgprot_t prot) +{ +#if defined(__i386__) || defined(__x86_64__) || defined(CONFIG_PPC64) + return pgprot_writecombine(prot); +#else + return pgprot_noncached(prot); +#endif +} + +enum { + T4_SQ_ONCHIP = (1<<0), +}; + +struct t4_sq { + union t4_wr *queue; + dma_addr_t dma_addr; + DEFINE_DMA_UNMAP_ADDR(mapping); + unsigned long phys_addr; + struct t4_swsqe *sw_sq; + struct t4_swsqe *oldest_read; + u64 __iomem *udb; + size_t memsize; + u32 qid; + u16 in_use; + u16 size; + u16 cidx; + u16 pidx; + u16 wq_pidx; + u16 wq_pidx_inc; + u16 flags; + short flush_cidx; +}; + +struct t4_swrqe { + u64 wr_id; + struct timespec host_ts; + u64 sge_ts; +}; + +struct t4_rq { + union t4_recv_wr *queue; + dma_addr_t dma_addr; + DEFINE_DMA_UNMAP_ADDR(mapping); + struct t4_swrqe *sw_rq; + u64 __iomem *udb; + size_t memsize; + u32 qid; + u32 msn; + u32 rqt_hwaddr; + u16 rqt_size; + u16 in_use; + u16 size; + u16 cidx; + u16 pidx; + u16 wq_pidx; + u16 wq_pidx_inc; +}; + +struct t4_wq { + struct t4_sq sq; + struct t4_rq rq; + void __iomem *db; + void __iomem *gts; + struct c4iw_rdev *rdev; + int flushed; +}; + +static inline int t4_rqes_posted(struct t4_wq *wq) +{ + return wq->rq.in_use; +} + +static inline int t4_rq_empty(struct t4_wq *wq) +{ + return wq->rq.in_use == 0; +} + +static inline int t4_rq_full(struct t4_wq *wq) +{ + return wq->rq.in_use == (wq->rq.size - 1); +} + +static inline u32 t4_rq_avail(struct t4_wq *wq) +{ + return wq->rq.size - 1 - wq->rq.in_use; +} + +static inline void t4_rq_produce(struct t4_wq *wq, u8 len16) +{ + wq->rq.in_use++; + if (++wq->rq.pidx == wq->rq.size) + wq->rq.pidx = 0; + wq->rq.wq_pidx += DIV_ROUND_UP(len16*16, T4_EQ_ENTRY_SIZE); + if (wq->rq.wq_pidx >= wq->rq.size * T4_RQ_NUM_SLOTS) + wq->rq.wq_pidx %= wq->rq.size * T4_RQ_NUM_SLOTS; +} + +static inline void t4_rq_consume(struct t4_wq *wq) +{ + wq->rq.in_use--; + wq->rq.msn++; + if (++wq->rq.cidx == wq->rq.size) + wq->rq.cidx = 0; +} + +static inline u16 t4_rq_host_wq_pidx(struct t4_wq *wq) +{ + return wq->rq.queue[wq->rq.size].status.host_wq_pidx; +} + +static inline u16 t4_rq_wq_size(struct t4_wq *wq) +{ + return wq->rq.size * T4_RQ_NUM_SLOTS; +} + +static inline int t4_sq_onchip(struct t4_sq *sq) +{ + return sq->flags & T4_SQ_ONCHIP; +} + +static inline int t4_sq_empty(struct t4_wq *wq) +{ + return wq->sq.in_use == 0; +} + +static inline int t4_sq_full(struct t4_wq *wq) +{ + return wq->sq.in_use == (wq->sq.size - 1); +} + +static inline u32 t4_sq_avail(struct t4_wq *wq) +{ + return wq->sq.size - 1 - wq->sq.in_use; +} + +static inline void t4_sq_produce(struct t4_wq *wq, u8 len16) +{ + wq->sq.in_use++; + if (++wq->sq.pidx == wq->sq.size) + wq->sq.pidx = 0; + wq->sq.wq_pidx += DIV_ROUND_UP(len16*16, T4_EQ_ENTRY_SIZE); + if (wq->sq.wq_pidx >= wq->sq.size * T4_SQ_NUM_SLOTS) + wq->sq.wq_pidx %= wq->sq.size * T4_SQ_NUM_SLOTS; +} + +static inline void t4_sq_consume(struct t4_wq *wq) +{ + BUG_ON(wq->sq.in_use < 1); + if (wq->sq.cidx == wq->sq.flush_cidx) + wq->sq.flush_cidx = -1; + wq->sq.in_use--; + if (++wq->sq.cidx == wq->sq.size) + wq->sq.cidx = 0; +} + +static inline u16 t4_sq_host_wq_pidx(struct t4_wq *wq) +{ + return wq->sq.queue[wq->sq.size].status.host_wq_pidx; +} + +static inline u16 t4_sq_wq_size(struct t4_wq *wq) +{ + return wq->sq.size * T4_SQ_NUM_SLOTS; +} + +/* This function copies 64 byte coalesced work request to memory + * mapped BAR2 space. For coalesced WRs, the SGE fetches data + * from the FIFO instead of from Host. + */ +static inline void pio_copy(u64 __iomem *dst, u64 *src) +{ + int count = 8; + + while (count) { + writeq(*src, dst); + src++; + dst++; + count--; + } +} + +static inline void t4_ring_sq_db(struct t4_wq *wq, u16 inc, u8 t5, + union t4_wr *wqe) +{ + + /* Flush host queue memory writes. */ + wmb(); + if (t5) { + if (inc == 1 && wqe) { + PDBG("%s: WC wq->sq.pidx = %d\n", + __func__, wq->sq.pidx); + pio_copy(wq->sq.udb + 7, (void *)wqe); + } else { + PDBG("%s: DB wq->sq.pidx = %d\n", + __func__, wq->sq.pidx); + writel(PIDX_T5_V(inc), wq->sq.udb); + } + + /* Flush user doorbell area writes. */ + wmb(); + return; + } + writel(QID_V(wq->sq.qid) | PIDX_V(inc), wq->db); +} + +static inline void t4_ring_rq_db(struct t4_wq *wq, u16 inc, u8 t5, + union t4_recv_wr *wqe) +{ + + /* Flush host queue memory writes. */ + wmb(); + if (t5) { + if (inc == 1 && wqe) { + PDBG("%s: WC wq->rq.pidx = %d\n", + __func__, wq->rq.pidx); + pio_copy(wq->rq.udb + 7, (void *)wqe); + } else { + PDBG("%s: DB wq->rq.pidx = %d\n", + __func__, wq->rq.pidx); + writel(PIDX_T5_V(inc), wq->rq.udb); + } + + /* Flush user doorbell area writes. */ + wmb(); + return; + } + writel(QID_V(wq->rq.qid) | PIDX_V(inc), wq->db); +} + +static inline int t4_wq_in_error(struct t4_wq *wq) +{ + return wq->rq.queue[wq->rq.size].status.qp_err; +} + +static inline void t4_set_wq_in_error(struct t4_wq *wq) +{ + wq->rq.queue[wq->rq.size].status.qp_err = 1; +} + +static inline void t4_disable_wq_db(struct t4_wq *wq) +{ + wq->rq.queue[wq->rq.size].status.db_off = 1; +} + +static inline void t4_enable_wq_db(struct t4_wq *wq) +{ + wq->rq.queue[wq->rq.size].status.db_off = 0; +} + +static inline int t4_wq_db_enabled(struct t4_wq *wq) +{ + return !wq->rq.queue[wq->rq.size].status.db_off; +} + +enum t4_cq_flags { + CQ_ARMED = 1, +}; + +struct t4_cq { + struct t4_cqe *queue; + dma_addr_t dma_addr; + DEFINE_DMA_UNMAP_ADDR(mapping); + struct t4_cqe *sw_queue; + void __iomem *gts; + struct c4iw_rdev *rdev; + u64 ugts; + size_t memsize; + __be64 bits_type_ts; + u32 cqid; + u32 qid_mask; + int vector; + u16 size; /* including status page */ + u16 cidx; + u16 sw_pidx; + u16 sw_cidx; + u16 sw_in_use; + u16 cidx_inc; + u8 gen; + u8 error; + unsigned long flags; +}; + +static inline int t4_clear_cq_armed(struct t4_cq *cq) +{ + return test_and_clear_bit(CQ_ARMED, &cq->flags); +} + +static inline int t4_arm_cq(struct t4_cq *cq, int se) +{ + u32 val; + + set_bit(CQ_ARMED, &cq->flags); + while (cq->cidx_inc > CIDXINC_M) { + val = SEINTARM_V(0) | CIDXINC_V(CIDXINC_M) | TIMERREG_V(7) | + INGRESSQID_V(cq->cqid & cq->qid_mask); + writel(val, cq->gts); + cq->cidx_inc -= CIDXINC_M; + } + val = SEINTARM_V(se) | CIDXINC_V(cq->cidx_inc) | TIMERREG_V(6) | + INGRESSQID_V(cq->cqid & cq->qid_mask); + writel(val, cq->gts); + cq->cidx_inc = 0; + return 0; +} + +static inline void t4_swcq_produce(struct t4_cq *cq) +{ + cq->sw_in_use++; + if (cq->sw_in_use == cq->size) { + PDBG("%s cxgb4 sw cq overflow cqid %u\n", __func__, cq->cqid); + cq->error = 1; + BUG_ON(1); + } + if (++cq->sw_pidx == cq->size) + cq->sw_pidx = 0; +} + +static inline void t4_swcq_consume(struct t4_cq *cq) +{ + BUG_ON(cq->sw_in_use < 1); + cq->sw_in_use--; + if (++cq->sw_cidx == cq->size) + cq->sw_cidx = 0; +} + +static inline void t4_hwcq_consume(struct t4_cq *cq) +{ + cq->bits_type_ts = cq->queue[cq->cidx].bits_type_ts; + if (++cq->cidx_inc == (cq->size >> 4) || cq->cidx_inc == CIDXINC_M) { + u32 val; + + val = SEINTARM_V(0) | CIDXINC_V(cq->cidx_inc) | TIMERREG_V(7) | + INGRESSQID_V(cq->cqid & cq->qid_mask); + writel(val, cq->gts); + cq->cidx_inc = 0; + } + if (++cq->cidx == cq->size) { + cq->cidx = 0; + cq->gen ^= 1; + } +} + +static inline int t4_valid_cqe(struct t4_cq *cq, struct t4_cqe *cqe) +{ + return (CQE_GENBIT(cqe) == cq->gen); +} + +static inline int t4_next_hw_cqe(struct t4_cq *cq, struct t4_cqe **cqe) +{ + int ret; + u16 prev_cidx; + + if (cq->cidx == 0) + prev_cidx = cq->size - 1; + else + prev_cidx = cq->cidx - 1; + + if (cq->queue[prev_cidx].bits_type_ts != cq->bits_type_ts) { + ret = -EOVERFLOW; + cq->error = 1; + printk(KERN_ERR MOD "cq overflow cqid %u\n", cq->cqid); + BUG_ON(1); + } else if (t4_valid_cqe(cq, &cq->queue[cq->cidx])) { + + /* Ensure CQE is flushed to memory */ + rmb(); + *cqe = &cq->queue[cq->cidx]; + ret = 0; + } else + ret = -ENODATA; + return ret; +} + +static inline struct t4_cqe *t4_next_sw_cqe(struct t4_cq *cq) +{ + if (cq->sw_in_use == cq->size) { + PDBG("%s cxgb4 sw cq overflow cqid %u\n", __func__, cq->cqid); + cq->error = 1; + BUG_ON(1); + return NULL; + } + if (cq->sw_in_use) + return &cq->sw_queue[cq->sw_cidx]; + return NULL; +} + +static inline int t4_next_cqe(struct t4_cq *cq, struct t4_cqe **cqe) +{ + int ret = 0; + + if (cq->error) + ret = -ENODATA; + else if (cq->sw_in_use) + *cqe = &cq->sw_queue[cq->sw_cidx]; + else + ret = t4_next_hw_cqe(cq, cqe); + return ret; +} + +static inline int t4_cq_in_error(struct t4_cq *cq) +{ + return ((struct t4_status_page *)&cq->queue[cq->size])->qp_err; +} + +static inline void t4_set_cq_in_error(struct t4_cq *cq) +{ + ((struct t4_status_page *)&cq->queue[cq->size])->qp_err = 1; +} +#endif + +struct t4_dev_status_page { + u8 db_off; +}; diff --git a/kernel/drivers/infiniband/hw/cxgb4/t4fw_ri_api.h b/kernel/drivers/infiniband/hw/cxgb4/t4fw_ri_api.h new file mode 100644 index 000000000..343e8daf2 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/t4fw_ri_api.h @@ -0,0 +1,855 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#ifndef _T4FW_RI_API_H_ +#define _T4FW_RI_API_H_ + +#include "t4fw_api.h" + +enum fw_ri_wr_opcode { + FW_RI_RDMA_WRITE = 0x0, /* IETF RDMAP v1.0 ... */ + FW_RI_READ_REQ = 0x1, + FW_RI_READ_RESP = 0x2, + FW_RI_SEND = 0x3, + FW_RI_SEND_WITH_INV = 0x4, + FW_RI_SEND_WITH_SE = 0x5, + FW_RI_SEND_WITH_SE_INV = 0x6, + FW_RI_TERMINATE = 0x7, + FW_RI_RDMA_INIT = 0x8, /* CHELSIO RI specific ... */ + FW_RI_BIND_MW = 0x9, + FW_RI_FAST_REGISTER = 0xa, + FW_RI_LOCAL_INV = 0xb, + FW_RI_QP_MODIFY = 0xc, + FW_RI_BYPASS = 0xd, + FW_RI_RECEIVE = 0xe, + + FW_RI_SGE_EC_CR_RETURN = 0xf +}; + +enum fw_ri_wr_flags { + FW_RI_COMPLETION_FLAG = 0x01, + FW_RI_NOTIFICATION_FLAG = 0x02, + FW_RI_SOLICITED_EVENT_FLAG = 0x04, + FW_RI_READ_FENCE_FLAG = 0x08, + FW_RI_LOCAL_FENCE_FLAG = 0x10, + FW_RI_RDMA_READ_INVALIDATE = 0x20 +}; + +enum fw_ri_mpa_attrs { + FW_RI_MPA_RX_MARKER_ENABLE = 0x01, + FW_RI_MPA_TX_MARKER_ENABLE = 0x02, + FW_RI_MPA_CRC_ENABLE = 0x04, + FW_RI_MPA_IETF_ENABLE = 0x08 +}; + +enum fw_ri_qp_caps { + FW_RI_QP_RDMA_READ_ENABLE = 0x01, + FW_RI_QP_RDMA_WRITE_ENABLE = 0x02, + FW_RI_QP_BIND_ENABLE = 0x04, + FW_RI_QP_FAST_REGISTER_ENABLE = 0x08, + FW_RI_QP_STAG0_ENABLE = 0x10 +}; + +enum fw_ri_addr_type { + FW_RI_ZERO_BASED_TO = 0x00, + FW_RI_VA_BASED_TO = 0x01 +}; + +enum fw_ri_mem_perms { + FW_RI_MEM_ACCESS_REM_WRITE = 0x01, + FW_RI_MEM_ACCESS_REM_READ = 0x02, + FW_RI_MEM_ACCESS_REM = 0x03, + FW_RI_MEM_ACCESS_LOCAL_WRITE = 0x04, + FW_RI_MEM_ACCESS_LOCAL_READ = 0x08, + FW_RI_MEM_ACCESS_LOCAL = 0x0C +}; + +enum fw_ri_stag_type { + FW_RI_STAG_NSMR = 0x00, + FW_RI_STAG_SMR = 0x01, + FW_RI_STAG_MW = 0x02, + FW_RI_STAG_MW_RELAXED = 0x03 +}; + +enum fw_ri_data_op { + FW_RI_DATA_IMMD = 0x81, + FW_RI_DATA_DSGL = 0x82, + FW_RI_DATA_ISGL = 0x83 +}; + +enum fw_ri_sgl_depth { + FW_RI_SGL_DEPTH_MAX_SQ = 16, + FW_RI_SGL_DEPTH_MAX_RQ = 4 +}; + +struct fw_ri_dsge_pair { + __be32 len[2]; + __be64 addr[2]; +}; + +struct fw_ri_dsgl { + __u8 op; + __u8 r1; + __be16 nsge; + __be32 len0; + __be64 addr0; +#ifndef C99_NOT_SUPPORTED + struct fw_ri_dsge_pair sge[0]; +#endif +}; + +struct fw_ri_sge { + __be32 stag; + __be32 len; + __be64 to; +}; + +struct fw_ri_isgl { + __u8 op; + __u8 r1; + __be16 nsge; + __be32 r2; +#ifndef C99_NOT_SUPPORTED + struct fw_ri_sge sge[0]; +#endif +}; + +struct fw_ri_immd { + __u8 op; + __u8 r1; + __be16 r2; + __be32 immdlen; +#ifndef C99_NOT_SUPPORTED + __u8 data[0]; +#endif +}; + +struct fw_ri_tpte { + __be32 valid_to_pdid; + __be32 locread_to_qpid; + __be32 nosnoop_pbladdr; + __be32 len_lo; + __be32 va_hi; + __be32 va_lo_fbo; + __be32 dca_mwbcnt_pstag; + __be32 len_hi; +}; + +#define FW_RI_TPTE_VALID_S 31 +#define FW_RI_TPTE_VALID_M 0x1 +#define FW_RI_TPTE_VALID_V(x) ((x) << FW_RI_TPTE_VALID_S) +#define FW_RI_TPTE_VALID_G(x) \ + (((x) >> FW_RI_TPTE_VALID_S) & FW_RI_TPTE_VALID_M) +#define FW_RI_TPTE_VALID_F FW_RI_TPTE_VALID_V(1U) + +#define FW_RI_TPTE_STAGKEY_S 23 +#define FW_RI_TPTE_STAGKEY_M 0xff +#define FW_RI_TPTE_STAGKEY_V(x) ((x) << FW_RI_TPTE_STAGKEY_S) +#define FW_RI_TPTE_STAGKEY_G(x) \ + (((x) >> FW_RI_TPTE_STAGKEY_S) & FW_RI_TPTE_STAGKEY_M) + +#define FW_RI_TPTE_STAGSTATE_S 22 +#define FW_RI_TPTE_STAGSTATE_M 0x1 +#define FW_RI_TPTE_STAGSTATE_V(x) ((x) << FW_RI_TPTE_STAGSTATE_S) +#define FW_RI_TPTE_STAGSTATE_G(x) \ + (((x) >> FW_RI_TPTE_STAGSTATE_S) & FW_RI_TPTE_STAGSTATE_M) +#define FW_RI_TPTE_STAGSTATE_F FW_RI_TPTE_STAGSTATE_V(1U) + +#define FW_RI_TPTE_STAGTYPE_S 20 +#define FW_RI_TPTE_STAGTYPE_M 0x3 +#define FW_RI_TPTE_STAGTYPE_V(x) ((x) << FW_RI_TPTE_STAGTYPE_S) +#define FW_RI_TPTE_STAGTYPE_G(x) \ + (((x) >> FW_RI_TPTE_STAGTYPE_S) & FW_RI_TPTE_STAGTYPE_M) + +#define FW_RI_TPTE_PDID_S 0 +#define FW_RI_TPTE_PDID_M 0xfffff +#define FW_RI_TPTE_PDID_V(x) ((x) << FW_RI_TPTE_PDID_S) +#define FW_RI_TPTE_PDID_G(x) \ + (((x) >> FW_RI_TPTE_PDID_S) & FW_RI_TPTE_PDID_M) + +#define FW_RI_TPTE_PERM_S 28 +#define FW_RI_TPTE_PERM_M 0xf +#define FW_RI_TPTE_PERM_V(x) ((x) << FW_RI_TPTE_PERM_S) +#define FW_RI_TPTE_PERM_G(x) \ + (((x) >> FW_RI_TPTE_PERM_S) & FW_RI_TPTE_PERM_M) + +#define FW_RI_TPTE_REMINVDIS_S 27 +#define FW_RI_TPTE_REMINVDIS_M 0x1 +#define FW_RI_TPTE_REMINVDIS_V(x) ((x) << FW_RI_TPTE_REMINVDIS_S) +#define FW_RI_TPTE_REMINVDIS_G(x) \ + (((x) >> FW_RI_TPTE_REMINVDIS_S) & FW_RI_TPTE_REMINVDIS_M) +#define FW_RI_TPTE_REMINVDIS_F FW_RI_TPTE_REMINVDIS_V(1U) + +#define FW_RI_TPTE_ADDRTYPE_S 26 +#define FW_RI_TPTE_ADDRTYPE_M 1 +#define FW_RI_TPTE_ADDRTYPE_V(x) ((x) << FW_RI_TPTE_ADDRTYPE_S) +#define FW_RI_TPTE_ADDRTYPE_G(x) \ + (((x) >> FW_RI_TPTE_ADDRTYPE_S) & FW_RI_TPTE_ADDRTYPE_M) +#define FW_RI_TPTE_ADDRTYPE_F FW_RI_TPTE_ADDRTYPE_V(1U) + +#define FW_RI_TPTE_MWBINDEN_S 25 +#define FW_RI_TPTE_MWBINDEN_M 0x1 +#define FW_RI_TPTE_MWBINDEN_V(x) ((x) << FW_RI_TPTE_MWBINDEN_S) +#define FW_RI_TPTE_MWBINDEN_G(x) \ + (((x) >> FW_RI_TPTE_MWBINDEN_S) & FW_RI_TPTE_MWBINDEN_M) +#define FW_RI_TPTE_MWBINDEN_F FW_RI_TPTE_MWBINDEN_V(1U) + +#define FW_RI_TPTE_PS_S 20 +#define FW_RI_TPTE_PS_M 0x1f +#define FW_RI_TPTE_PS_V(x) ((x) << FW_RI_TPTE_PS_S) +#define FW_RI_TPTE_PS_G(x) \ + (((x) >> FW_RI_TPTE_PS_S) & FW_RI_TPTE_PS_M) + +#define FW_RI_TPTE_QPID_S 0 +#define FW_RI_TPTE_QPID_M 0xfffff +#define FW_RI_TPTE_QPID_V(x) ((x) << FW_RI_TPTE_QPID_S) +#define FW_RI_TPTE_QPID_G(x) \ + (((x) >> FW_RI_TPTE_QPID_S) & FW_RI_TPTE_QPID_M) + +#define FW_RI_TPTE_NOSNOOP_S 30 +#define FW_RI_TPTE_NOSNOOP_M 0x1 +#define FW_RI_TPTE_NOSNOOP_V(x) ((x) << FW_RI_TPTE_NOSNOOP_S) +#define FW_RI_TPTE_NOSNOOP_G(x) \ + (((x) >> FW_RI_TPTE_NOSNOOP_S) & FW_RI_TPTE_NOSNOOP_M) +#define FW_RI_TPTE_NOSNOOP_F FW_RI_TPTE_NOSNOOP_V(1U) + +#define FW_RI_TPTE_PBLADDR_S 0 +#define FW_RI_TPTE_PBLADDR_M 0x1fffffff +#define FW_RI_TPTE_PBLADDR_V(x) ((x) << FW_RI_TPTE_PBLADDR_S) +#define FW_RI_TPTE_PBLADDR_G(x) \ + (((x) >> FW_RI_TPTE_PBLADDR_S) & FW_RI_TPTE_PBLADDR_M) + +#define FW_RI_TPTE_DCA_S 24 +#define FW_RI_TPTE_DCA_M 0x1f +#define FW_RI_TPTE_DCA_V(x) ((x) << FW_RI_TPTE_DCA_S) +#define FW_RI_TPTE_DCA_G(x) \ + (((x) >> FW_RI_TPTE_DCA_S) & FW_RI_TPTE_DCA_M) + +#define FW_RI_TPTE_MWBCNT_PSTAG_S 0 +#define FW_RI_TPTE_MWBCNT_PSTAG_M 0xffffff +#define FW_RI_TPTE_MWBCNT_PSTAT_V(x) \ + ((x) << FW_RI_TPTE_MWBCNT_PSTAG_S) +#define FW_RI_TPTE_MWBCNT_PSTAG_G(x) \ + (((x) >> FW_RI_TPTE_MWBCNT_PSTAG_S) & FW_RI_TPTE_MWBCNT_PSTAG_M) + +enum fw_ri_res_type { + FW_RI_RES_TYPE_SQ, + FW_RI_RES_TYPE_RQ, + FW_RI_RES_TYPE_CQ, +}; + +enum fw_ri_res_op { + FW_RI_RES_OP_WRITE, + FW_RI_RES_OP_RESET, +}; + +struct fw_ri_res { + union fw_ri_restype { + struct fw_ri_res_sqrq { + __u8 restype; + __u8 op; + __be16 r3; + __be32 eqid; + __be32 r4[2]; + __be32 fetchszm_to_iqid; + __be32 dcaen_to_eqsize; + __be64 eqaddr; + } sqrq; + struct fw_ri_res_cq { + __u8 restype; + __u8 op; + __be16 r3; + __be32 iqid; + __be32 r4[2]; + __be32 iqandst_to_iqandstindex; + __be16 iqdroprss_to_iqesize; + __be16 iqsize; + __be64 iqaddr; + __be32 iqns_iqro; + __be32 r6_lo; + __be64 r7; + } cq; + } u; +}; + +struct fw_ri_res_wr { + __be32 op_nres; + __be32 len16_pkd; + __u64 cookie; +#ifndef C99_NOT_SUPPORTED + struct fw_ri_res res[0]; +#endif +}; + +#define FW_RI_RES_WR_NRES_S 0 +#define FW_RI_RES_WR_NRES_M 0xff +#define FW_RI_RES_WR_NRES_V(x) ((x) << FW_RI_RES_WR_NRES_S) +#define FW_RI_RES_WR_NRES_G(x) \ + (((x) >> FW_RI_RES_WR_NRES_S) & FW_RI_RES_WR_NRES_M) + +#define FW_RI_RES_WR_FETCHSZM_S 26 +#define FW_RI_RES_WR_FETCHSZM_M 0x1 +#define FW_RI_RES_WR_FETCHSZM_V(x) ((x) << FW_RI_RES_WR_FETCHSZM_S) +#define FW_RI_RES_WR_FETCHSZM_G(x) \ + (((x) >> FW_RI_RES_WR_FETCHSZM_S) & FW_RI_RES_WR_FETCHSZM_M) +#define FW_RI_RES_WR_FETCHSZM_F FW_RI_RES_WR_FETCHSZM_V(1U) + +#define FW_RI_RES_WR_STATUSPGNS_S 25 +#define FW_RI_RES_WR_STATUSPGNS_M 0x1 +#define FW_RI_RES_WR_STATUSPGNS_V(x) ((x) << FW_RI_RES_WR_STATUSPGNS_S) +#define FW_RI_RES_WR_STATUSPGNS_G(x) \ + (((x) >> FW_RI_RES_WR_STATUSPGNS_S) & FW_RI_RES_WR_STATUSPGNS_M) +#define FW_RI_RES_WR_STATUSPGNS_F FW_RI_RES_WR_STATUSPGNS_V(1U) + +#define FW_RI_RES_WR_STATUSPGRO_S 24 +#define FW_RI_RES_WR_STATUSPGRO_M 0x1 +#define FW_RI_RES_WR_STATUSPGRO_V(x) ((x) << FW_RI_RES_WR_STATUSPGRO_S) +#define FW_RI_RES_WR_STATUSPGRO_G(x) \ + (((x) >> FW_RI_RES_WR_STATUSPGRO_S) & FW_RI_RES_WR_STATUSPGRO_M) +#define FW_RI_RES_WR_STATUSPGRO_F FW_RI_RES_WR_STATUSPGRO_V(1U) + +#define FW_RI_RES_WR_FETCHNS_S 23 +#define FW_RI_RES_WR_FETCHNS_M 0x1 +#define FW_RI_RES_WR_FETCHNS_V(x) ((x) << FW_RI_RES_WR_FETCHNS_S) +#define FW_RI_RES_WR_FETCHNS_G(x) \ + (((x) >> FW_RI_RES_WR_FETCHNS_S) & FW_RI_RES_WR_FETCHNS_M) +#define FW_RI_RES_WR_FETCHNS_F FW_RI_RES_WR_FETCHNS_V(1U) + +#define FW_RI_RES_WR_FETCHRO_S 22 +#define FW_RI_RES_WR_FETCHRO_M 0x1 +#define FW_RI_RES_WR_FETCHRO_V(x) ((x) << FW_RI_RES_WR_FETCHRO_S) +#define FW_RI_RES_WR_FETCHRO_G(x) \ + (((x) >> FW_RI_RES_WR_FETCHRO_S) & FW_RI_RES_WR_FETCHRO_M) +#define FW_RI_RES_WR_FETCHRO_F FW_RI_RES_WR_FETCHRO_V(1U) + +#define FW_RI_RES_WR_HOSTFCMODE_S 20 +#define FW_RI_RES_WR_HOSTFCMODE_M 0x3 +#define FW_RI_RES_WR_HOSTFCMODE_V(x) ((x) << FW_RI_RES_WR_HOSTFCMODE_S) +#define FW_RI_RES_WR_HOSTFCMODE_G(x) \ + (((x) >> FW_RI_RES_WR_HOSTFCMODE_S) & FW_RI_RES_WR_HOSTFCMODE_M) + +#define FW_RI_RES_WR_CPRIO_S 19 +#define FW_RI_RES_WR_CPRIO_M 0x1 +#define FW_RI_RES_WR_CPRIO_V(x) ((x) << FW_RI_RES_WR_CPRIO_S) +#define FW_RI_RES_WR_CPRIO_G(x) \ + (((x) >> FW_RI_RES_WR_CPRIO_S) & FW_RI_RES_WR_CPRIO_M) +#define FW_RI_RES_WR_CPRIO_F FW_RI_RES_WR_CPRIO_V(1U) + +#define FW_RI_RES_WR_ONCHIP_S 18 +#define FW_RI_RES_WR_ONCHIP_M 0x1 +#define FW_RI_RES_WR_ONCHIP_V(x) ((x) << FW_RI_RES_WR_ONCHIP_S) +#define FW_RI_RES_WR_ONCHIP_G(x) \ + (((x) >> FW_RI_RES_WR_ONCHIP_S) & FW_RI_RES_WR_ONCHIP_M) +#define FW_RI_RES_WR_ONCHIP_F FW_RI_RES_WR_ONCHIP_V(1U) + +#define FW_RI_RES_WR_PCIECHN_S 16 +#define FW_RI_RES_WR_PCIECHN_M 0x3 +#define FW_RI_RES_WR_PCIECHN_V(x) ((x) << FW_RI_RES_WR_PCIECHN_S) +#define FW_RI_RES_WR_PCIECHN_G(x) \ + (((x) >> FW_RI_RES_WR_PCIECHN_S) & FW_RI_RES_WR_PCIECHN_M) + +#define FW_RI_RES_WR_IQID_S 0 +#define FW_RI_RES_WR_IQID_M 0xffff +#define FW_RI_RES_WR_IQID_V(x) ((x) << FW_RI_RES_WR_IQID_S) +#define FW_RI_RES_WR_IQID_G(x) \ + (((x) >> FW_RI_RES_WR_IQID_S) & FW_RI_RES_WR_IQID_M) + +#define FW_RI_RES_WR_DCAEN_S 31 +#define FW_RI_RES_WR_DCAEN_M 0x1 +#define FW_RI_RES_WR_DCAEN_V(x) ((x) << FW_RI_RES_WR_DCAEN_S) +#define FW_RI_RES_WR_DCAEN_G(x) \ + (((x) >> FW_RI_RES_WR_DCAEN_S) & FW_RI_RES_WR_DCAEN_M) +#define FW_RI_RES_WR_DCAEN_F FW_RI_RES_WR_DCAEN_V(1U) + +#define FW_RI_RES_WR_DCACPU_S 26 +#define FW_RI_RES_WR_DCACPU_M 0x1f +#define FW_RI_RES_WR_DCACPU_V(x) ((x) << FW_RI_RES_WR_DCACPU_S) +#define FW_RI_RES_WR_DCACPU_G(x) \ + (((x) >> FW_RI_RES_WR_DCACPU_S) & FW_RI_RES_WR_DCACPU_M) + +#define FW_RI_RES_WR_FBMIN_S 23 +#define FW_RI_RES_WR_FBMIN_M 0x7 +#define FW_RI_RES_WR_FBMIN_V(x) ((x) << FW_RI_RES_WR_FBMIN_S) +#define FW_RI_RES_WR_FBMIN_G(x) \ + (((x) >> FW_RI_RES_WR_FBMIN_S) & FW_RI_RES_WR_FBMIN_M) + +#define FW_RI_RES_WR_FBMAX_S 20 +#define FW_RI_RES_WR_FBMAX_M 0x7 +#define FW_RI_RES_WR_FBMAX_V(x) ((x) << FW_RI_RES_WR_FBMAX_S) +#define FW_RI_RES_WR_FBMAX_G(x) \ + (((x) >> FW_RI_RES_WR_FBMAX_S) & FW_RI_RES_WR_FBMAX_M) + +#define FW_RI_RES_WR_CIDXFTHRESHO_S 19 +#define FW_RI_RES_WR_CIDXFTHRESHO_M 0x1 +#define FW_RI_RES_WR_CIDXFTHRESHO_V(x) ((x) << FW_RI_RES_WR_CIDXFTHRESHO_S) +#define FW_RI_RES_WR_CIDXFTHRESHO_G(x) \ + (((x) >> FW_RI_RES_WR_CIDXFTHRESHO_S) & FW_RI_RES_WR_CIDXFTHRESHO_M) +#define FW_RI_RES_WR_CIDXFTHRESHO_F FW_RI_RES_WR_CIDXFTHRESHO_V(1U) + +#define FW_RI_RES_WR_CIDXFTHRESH_S 16 +#define FW_RI_RES_WR_CIDXFTHRESH_M 0x7 +#define FW_RI_RES_WR_CIDXFTHRESH_V(x) ((x) << FW_RI_RES_WR_CIDXFTHRESH_S) +#define FW_RI_RES_WR_CIDXFTHRESH_G(x) \ + (((x) >> FW_RI_RES_WR_CIDXFTHRESH_S) & FW_RI_RES_WR_CIDXFTHRESH_M) + +#define FW_RI_RES_WR_EQSIZE_S 0 +#define FW_RI_RES_WR_EQSIZE_M 0xffff +#define FW_RI_RES_WR_EQSIZE_V(x) ((x) << FW_RI_RES_WR_EQSIZE_S) +#define FW_RI_RES_WR_EQSIZE_G(x) \ + (((x) >> FW_RI_RES_WR_EQSIZE_S) & FW_RI_RES_WR_EQSIZE_M) + +#define FW_RI_RES_WR_IQANDST_S 15 +#define FW_RI_RES_WR_IQANDST_M 0x1 +#define FW_RI_RES_WR_IQANDST_V(x) ((x) << FW_RI_RES_WR_IQANDST_S) +#define FW_RI_RES_WR_IQANDST_G(x) \ + (((x) >> FW_RI_RES_WR_IQANDST_S) & FW_RI_RES_WR_IQANDST_M) +#define FW_RI_RES_WR_IQANDST_F FW_RI_RES_WR_IQANDST_V(1U) + +#define FW_RI_RES_WR_IQANUS_S 14 +#define FW_RI_RES_WR_IQANUS_M 0x1 +#define FW_RI_RES_WR_IQANUS_V(x) ((x) << FW_RI_RES_WR_IQANUS_S) +#define FW_RI_RES_WR_IQANUS_G(x) \ + (((x) >> FW_RI_RES_WR_IQANUS_S) & FW_RI_RES_WR_IQANUS_M) +#define FW_RI_RES_WR_IQANUS_F FW_RI_RES_WR_IQANUS_V(1U) + +#define FW_RI_RES_WR_IQANUD_S 12 +#define FW_RI_RES_WR_IQANUD_M 0x3 +#define FW_RI_RES_WR_IQANUD_V(x) ((x) << FW_RI_RES_WR_IQANUD_S) +#define FW_RI_RES_WR_IQANUD_G(x) \ + (((x) >> FW_RI_RES_WR_IQANUD_S) & FW_RI_RES_WR_IQANUD_M) + +#define FW_RI_RES_WR_IQANDSTINDEX_S 0 +#define FW_RI_RES_WR_IQANDSTINDEX_M 0xfff +#define FW_RI_RES_WR_IQANDSTINDEX_V(x) ((x) << FW_RI_RES_WR_IQANDSTINDEX_S) +#define FW_RI_RES_WR_IQANDSTINDEX_G(x) \ + (((x) >> FW_RI_RES_WR_IQANDSTINDEX_S) & FW_RI_RES_WR_IQANDSTINDEX_M) + +#define FW_RI_RES_WR_IQDROPRSS_S 15 +#define FW_RI_RES_WR_IQDROPRSS_M 0x1 +#define FW_RI_RES_WR_IQDROPRSS_V(x) ((x) << FW_RI_RES_WR_IQDROPRSS_S) +#define FW_RI_RES_WR_IQDROPRSS_G(x) \ + (((x) >> FW_RI_RES_WR_IQDROPRSS_S) & FW_RI_RES_WR_IQDROPRSS_M) +#define FW_RI_RES_WR_IQDROPRSS_F FW_RI_RES_WR_IQDROPRSS_V(1U) + +#define FW_RI_RES_WR_IQGTSMODE_S 14 +#define FW_RI_RES_WR_IQGTSMODE_M 0x1 +#define FW_RI_RES_WR_IQGTSMODE_V(x) ((x) << FW_RI_RES_WR_IQGTSMODE_S) +#define FW_RI_RES_WR_IQGTSMODE_G(x) \ + (((x) >> FW_RI_RES_WR_IQGTSMODE_S) & FW_RI_RES_WR_IQGTSMODE_M) +#define FW_RI_RES_WR_IQGTSMODE_F FW_RI_RES_WR_IQGTSMODE_V(1U) + +#define FW_RI_RES_WR_IQPCIECH_S 12 +#define FW_RI_RES_WR_IQPCIECH_M 0x3 +#define FW_RI_RES_WR_IQPCIECH_V(x) ((x) << FW_RI_RES_WR_IQPCIECH_S) +#define FW_RI_RES_WR_IQPCIECH_G(x) \ + (((x) >> FW_RI_RES_WR_IQPCIECH_S) & FW_RI_RES_WR_IQPCIECH_M) + +#define FW_RI_RES_WR_IQDCAEN_S 11 +#define FW_RI_RES_WR_IQDCAEN_M 0x1 +#define FW_RI_RES_WR_IQDCAEN_V(x) ((x) << FW_RI_RES_WR_IQDCAEN_S) +#define FW_RI_RES_WR_IQDCAEN_G(x) \ + (((x) >> FW_RI_RES_WR_IQDCAEN_S) & FW_RI_RES_WR_IQDCAEN_M) +#define FW_RI_RES_WR_IQDCAEN_F FW_RI_RES_WR_IQDCAEN_V(1U) + +#define FW_RI_RES_WR_IQDCACPU_S 6 +#define FW_RI_RES_WR_IQDCACPU_M 0x1f +#define FW_RI_RES_WR_IQDCACPU_V(x) ((x) << FW_RI_RES_WR_IQDCACPU_S) +#define FW_RI_RES_WR_IQDCACPU_G(x) \ + (((x) >> FW_RI_RES_WR_IQDCACPU_S) & FW_RI_RES_WR_IQDCACPU_M) + +#define FW_RI_RES_WR_IQINTCNTTHRESH_S 4 +#define FW_RI_RES_WR_IQINTCNTTHRESH_M 0x3 +#define FW_RI_RES_WR_IQINTCNTTHRESH_V(x) \ + ((x) << FW_RI_RES_WR_IQINTCNTTHRESH_S) +#define FW_RI_RES_WR_IQINTCNTTHRESH_G(x) \ + (((x) >> FW_RI_RES_WR_IQINTCNTTHRESH_S) & FW_RI_RES_WR_IQINTCNTTHRESH_M) + +#define FW_RI_RES_WR_IQO_S 3 +#define FW_RI_RES_WR_IQO_M 0x1 +#define FW_RI_RES_WR_IQO_V(x) ((x) << FW_RI_RES_WR_IQO_S) +#define FW_RI_RES_WR_IQO_G(x) \ + (((x) >> FW_RI_RES_WR_IQO_S) & FW_RI_RES_WR_IQO_M) +#define FW_RI_RES_WR_IQO_F FW_RI_RES_WR_IQO_V(1U) + +#define FW_RI_RES_WR_IQCPRIO_S 2 +#define FW_RI_RES_WR_IQCPRIO_M 0x1 +#define FW_RI_RES_WR_IQCPRIO_V(x) ((x) << FW_RI_RES_WR_IQCPRIO_S) +#define FW_RI_RES_WR_IQCPRIO_G(x) \ + (((x) >> FW_RI_RES_WR_IQCPRIO_S) & FW_RI_RES_WR_IQCPRIO_M) +#define FW_RI_RES_WR_IQCPRIO_F FW_RI_RES_WR_IQCPRIO_V(1U) + +#define FW_RI_RES_WR_IQESIZE_S 0 +#define FW_RI_RES_WR_IQESIZE_M 0x3 +#define FW_RI_RES_WR_IQESIZE_V(x) ((x) << FW_RI_RES_WR_IQESIZE_S) +#define FW_RI_RES_WR_IQESIZE_G(x) \ + (((x) >> FW_RI_RES_WR_IQESIZE_S) & FW_RI_RES_WR_IQESIZE_M) + +#define FW_RI_RES_WR_IQNS_S 31 +#define FW_RI_RES_WR_IQNS_M 0x1 +#define FW_RI_RES_WR_IQNS_V(x) ((x) << FW_RI_RES_WR_IQNS_S) +#define FW_RI_RES_WR_IQNS_G(x) \ + (((x) >> FW_RI_RES_WR_IQNS_S) & FW_RI_RES_WR_IQNS_M) +#define FW_RI_RES_WR_IQNS_F FW_RI_RES_WR_IQNS_V(1U) + +#define FW_RI_RES_WR_IQRO_S 30 +#define FW_RI_RES_WR_IQRO_M 0x1 +#define FW_RI_RES_WR_IQRO_V(x) ((x) << FW_RI_RES_WR_IQRO_S) +#define FW_RI_RES_WR_IQRO_G(x) \ + (((x) >> FW_RI_RES_WR_IQRO_S) & FW_RI_RES_WR_IQRO_M) +#define FW_RI_RES_WR_IQRO_F FW_RI_RES_WR_IQRO_V(1U) + +struct fw_ri_rdma_write_wr { + __u8 opcode; + __u8 flags; + __u16 wrid; + __u8 r1[3]; + __u8 len16; + __be64 r2; + __be32 plen; + __be32 stag_sink; + __be64 to_sink; +#ifndef C99_NOT_SUPPORTED + union { + struct fw_ri_immd immd_src[0]; + struct fw_ri_isgl isgl_src[0]; + } u; +#endif +}; + +struct fw_ri_send_wr { + __u8 opcode; + __u8 flags; + __u16 wrid; + __u8 r1[3]; + __u8 len16; + __be32 sendop_pkd; + __be32 stag_inv; + __be32 plen; + __be32 r3; + __be64 r4; +#ifndef C99_NOT_SUPPORTED + union { + struct fw_ri_immd immd_src[0]; + struct fw_ri_isgl isgl_src[0]; + } u; +#endif +}; + +#define FW_RI_SEND_WR_SENDOP_S 0 +#define FW_RI_SEND_WR_SENDOP_M 0xf +#define FW_RI_SEND_WR_SENDOP_V(x) ((x) << FW_RI_SEND_WR_SENDOP_S) +#define FW_RI_SEND_WR_SENDOP_G(x) \ + (((x) >> FW_RI_SEND_WR_SENDOP_S) & FW_RI_SEND_WR_SENDOP_M) + +struct fw_ri_rdma_read_wr { + __u8 opcode; + __u8 flags; + __u16 wrid; + __u8 r1[3]; + __u8 len16; + __be64 r2; + __be32 stag_sink; + __be32 to_sink_hi; + __be32 to_sink_lo; + __be32 plen; + __be32 stag_src; + __be32 to_src_hi; + __be32 to_src_lo; + __be32 r5; +}; + +struct fw_ri_recv_wr { + __u8 opcode; + __u8 r1; + __u16 wrid; + __u8 r2[3]; + __u8 len16; + struct fw_ri_isgl isgl; +}; + +struct fw_ri_bind_mw_wr { + __u8 opcode; + __u8 flags; + __u16 wrid; + __u8 r1[3]; + __u8 len16; + __u8 qpbinde_to_dcacpu; + __u8 pgsz_shift; + __u8 addr_type; + __u8 mem_perms; + __be32 stag_mr; + __be32 stag_mw; + __be32 r3; + __be64 len_mw; + __be64 va_fbo; + __be64 r4; +}; + +#define FW_RI_BIND_MW_WR_QPBINDE_S 6 +#define FW_RI_BIND_MW_WR_QPBINDE_M 0x1 +#define FW_RI_BIND_MW_WR_QPBINDE_V(x) ((x) << FW_RI_BIND_MW_WR_QPBINDE_S) +#define FW_RI_BIND_MW_WR_QPBINDE_G(x) \ + (((x) >> FW_RI_BIND_MW_WR_QPBINDE_S) & FW_RI_BIND_MW_WR_QPBINDE_M) +#define FW_RI_BIND_MW_WR_QPBINDE_F FW_RI_BIND_MW_WR_QPBINDE_V(1U) + +#define FW_RI_BIND_MW_WR_NS_S 5 +#define FW_RI_BIND_MW_WR_NS_M 0x1 +#define FW_RI_BIND_MW_WR_NS_V(x) ((x) << FW_RI_BIND_MW_WR_NS_S) +#define FW_RI_BIND_MW_WR_NS_G(x) \ + (((x) >> FW_RI_BIND_MW_WR_NS_S) & FW_RI_BIND_MW_WR_NS_M) +#define FW_RI_BIND_MW_WR_NS_F FW_RI_BIND_MW_WR_NS_V(1U) + +#define FW_RI_BIND_MW_WR_DCACPU_S 0 +#define FW_RI_BIND_MW_WR_DCACPU_M 0x1f +#define FW_RI_BIND_MW_WR_DCACPU_V(x) ((x) << FW_RI_BIND_MW_WR_DCACPU_S) +#define FW_RI_BIND_MW_WR_DCACPU_G(x) \ + (((x) >> FW_RI_BIND_MW_WR_DCACPU_S) & FW_RI_BIND_MW_WR_DCACPU_M) + +struct fw_ri_fr_nsmr_wr { + __u8 opcode; + __u8 flags; + __u16 wrid; + __u8 r1[3]; + __u8 len16; + __u8 qpbinde_to_dcacpu; + __u8 pgsz_shift; + __u8 addr_type; + __u8 mem_perms; + __be32 stag; + __be32 len_hi; + __be32 len_lo; + __be32 va_hi; + __be32 va_lo_fbo; +}; + +#define FW_RI_FR_NSMR_WR_QPBINDE_S 6 +#define FW_RI_FR_NSMR_WR_QPBINDE_M 0x1 +#define FW_RI_FR_NSMR_WR_QPBINDE_V(x) ((x) << FW_RI_FR_NSMR_WR_QPBINDE_S) +#define FW_RI_FR_NSMR_WR_QPBINDE_G(x) \ + (((x) >> FW_RI_FR_NSMR_WR_QPBINDE_S) & FW_RI_FR_NSMR_WR_QPBINDE_M) +#define FW_RI_FR_NSMR_WR_QPBINDE_F FW_RI_FR_NSMR_WR_QPBINDE_V(1U) + +#define FW_RI_FR_NSMR_WR_NS_S 5 +#define FW_RI_FR_NSMR_WR_NS_M 0x1 +#define FW_RI_FR_NSMR_WR_NS_V(x) ((x) << FW_RI_FR_NSMR_WR_NS_S) +#define FW_RI_FR_NSMR_WR_NS_G(x) \ + (((x) >> FW_RI_FR_NSMR_WR_NS_S) & FW_RI_FR_NSMR_WR_NS_M) +#define FW_RI_FR_NSMR_WR_NS_F FW_RI_FR_NSMR_WR_NS_V(1U) + +#define FW_RI_FR_NSMR_WR_DCACPU_S 0 +#define FW_RI_FR_NSMR_WR_DCACPU_M 0x1f +#define FW_RI_FR_NSMR_WR_DCACPU_V(x) ((x) << FW_RI_FR_NSMR_WR_DCACPU_S) +#define FW_RI_FR_NSMR_WR_DCACPU_G(x) \ + (((x) >> FW_RI_FR_NSMR_WR_DCACPU_S) & FW_RI_FR_NSMR_WR_DCACPU_M) + +struct fw_ri_inv_lstag_wr { + __u8 opcode; + __u8 flags; + __u16 wrid; + __u8 r1[3]; + __u8 len16; + __be32 r2; + __be32 stag_inv; +}; + +enum fw_ri_type { + FW_RI_TYPE_INIT, + FW_RI_TYPE_FINI, + FW_RI_TYPE_TERMINATE +}; + +enum fw_ri_init_p2ptype { + FW_RI_INIT_P2PTYPE_RDMA_WRITE = FW_RI_RDMA_WRITE, + FW_RI_INIT_P2PTYPE_READ_REQ = FW_RI_READ_REQ, + FW_RI_INIT_P2PTYPE_SEND = FW_RI_SEND, + FW_RI_INIT_P2PTYPE_SEND_WITH_INV = FW_RI_SEND_WITH_INV, + FW_RI_INIT_P2PTYPE_SEND_WITH_SE = FW_RI_SEND_WITH_SE, + FW_RI_INIT_P2PTYPE_SEND_WITH_SE_INV = FW_RI_SEND_WITH_SE_INV, + FW_RI_INIT_P2PTYPE_DISABLED = 0xf, +}; + +struct fw_ri_wr { + __be32 op_compl; + __be32 flowid_len16; + __u64 cookie; + union fw_ri { + struct fw_ri_init { + __u8 type; + __u8 mpareqbit_p2ptype; + __u8 r4[2]; + __u8 mpa_attrs; + __u8 qp_caps; + __be16 nrqe; + __be32 pdid; + __be32 qpid; + __be32 sq_eqid; + __be32 rq_eqid; + __be32 scqid; + __be32 rcqid; + __be32 ord_max; + __be32 ird_max; + __be32 iss; + __be32 irs; + __be32 hwrqsize; + __be32 hwrqaddr; + __be64 r5; + union fw_ri_init_p2p { + struct fw_ri_rdma_write_wr write; + struct fw_ri_rdma_read_wr read; + struct fw_ri_send_wr send; + } u; + } init; + struct fw_ri_fini { + __u8 type; + __u8 r3[7]; + __be64 r4; + } fini; + struct fw_ri_terminate { + __u8 type; + __u8 r3[3]; + __be32 immdlen; + __u8 termmsg[40]; + } terminate; + } u; +}; + +#define FW_RI_WR_MPAREQBIT_S 7 +#define FW_RI_WR_MPAREQBIT_M 0x1 +#define FW_RI_WR_MPAREQBIT_V(x) ((x) << FW_RI_WR_MPAREQBIT_S) +#define FW_RI_WR_MPAREQBIT_G(x) \ + (((x) >> FW_RI_WR_MPAREQBIT_S) & FW_RI_WR_MPAREQBIT_M) +#define FW_RI_WR_MPAREQBIT_F FW_RI_WR_MPAREQBIT_V(1U) + +#define FW_RI_WR_P2PTYPE_S 0 +#define FW_RI_WR_P2PTYPE_M 0xf +#define FW_RI_WR_P2PTYPE_V(x) ((x) << FW_RI_WR_P2PTYPE_S) +#define FW_RI_WR_P2PTYPE_G(x) \ + (((x) >> FW_RI_WR_P2PTYPE_S) & FW_RI_WR_P2PTYPE_M) + +struct tcp_options { + __be16 mss; + __u8 wsf; +#if defined(__LITTLE_ENDIAN_BITFIELD) + __u8:4; + __u8 unknown:1; + __u8:1; + __u8 sack:1; + __u8 tstamp:1; +#else + __u8 tstamp:1; + __u8 sack:1; + __u8:1; + __u8 unknown:1; + __u8:4; +#endif +}; + +struct cpl_pass_accept_req { + union opcode_tid ot; + __be16 rsvd; + __be16 len; + __be32 hdr_len; + __be16 vlan; + __be16 l2info; + __be32 tos_stid; + struct tcp_options tcpopt; +}; + +/* cpl_pass_accept_req.hdr_len fields */ +#define SYN_RX_CHAN_S 0 +#define SYN_RX_CHAN_M 0xF +#define SYN_RX_CHAN_V(x) ((x) << SYN_RX_CHAN_S) +#define SYN_RX_CHAN_G(x) (((x) >> SYN_RX_CHAN_S) & SYN_RX_CHAN_M) + +#define TCP_HDR_LEN_S 10 +#define TCP_HDR_LEN_M 0x3F +#define TCP_HDR_LEN_V(x) ((x) << TCP_HDR_LEN_S) +#define TCP_HDR_LEN_G(x) (((x) >> TCP_HDR_LEN_S) & TCP_HDR_LEN_M) + +#define IP_HDR_LEN_S 16 +#define IP_HDR_LEN_M 0x3FF +#define IP_HDR_LEN_V(x) ((x) << IP_HDR_LEN_S) +#define IP_HDR_LEN_G(x) (((x) >> IP_HDR_LEN_S) & IP_HDR_LEN_M) + +#define ETH_HDR_LEN_S 26 +#define ETH_HDR_LEN_M 0x1F +#define ETH_HDR_LEN_V(x) ((x) << ETH_HDR_LEN_S) +#define ETH_HDR_LEN_G(x) (((x) >> ETH_HDR_LEN_S) & ETH_HDR_LEN_M) + +/* cpl_pass_accept_req.l2info fields */ +#define SYN_MAC_IDX_S 0 +#define SYN_MAC_IDX_M 0x1FF +#define SYN_MAC_IDX_V(x) ((x) << SYN_MAC_IDX_S) +#define SYN_MAC_IDX_G(x) (((x) >> SYN_MAC_IDX_S) & SYN_MAC_IDX_M) + +#define SYN_XACT_MATCH_S 9 +#define SYN_XACT_MATCH_V(x) ((x) << SYN_XACT_MATCH_S) +#define SYN_XACT_MATCH_F SYN_XACT_MATCH_V(1U) + +#define SYN_INTF_S 12 +#define SYN_INTF_M 0xF +#define SYN_INTF_V(x) ((x) << SYN_INTF_S) +#define SYN_INTF_G(x) (((x) >> SYN_INTF_S) & SYN_INTF_M) + +struct ulptx_idata { + __be32 cmd_more; + __be32 len; +}; + +#define ULPTX_NSGE_S 0 +#define ULPTX_NSGE_M 0xFFFF +#define ULPTX_NSGE_V(x) ((x) << ULPTX_NSGE_S) + +#define RX_DACK_MODE_S 29 +#define RX_DACK_MODE_M 0x3 +#define RX_DACK_MODE_V(x) ((x) << RX_DACK_MODE_S) +#define RX_DACK_MODE_G(x) (((x) >> RX_DACK_MODE_S) & RX_DACK_MODE_M) + +#define RX_DACK_CHANGE_S 31 +#define RX_DACK_CHANGE_V(x) ((x) << RX_DACK_CHANGE_S) +#define RX_DACK_CHANGE_F RX_DACK_CHANGE_V(1U) + +enum { /* TCP congestion control algorithms */ + CONG_ALG_RENO, + CONG_ALG_TAHOE, + CONG_ALG_NEWRENO, + CONG_ALG_HIGHSPEED +}; + +#define CONG_CNTRL_S 14 +#define CONG_CNTRL_M 0x3 +#define CONG_CNTRL_V(x) ((x) << CONG_CNTRL_S) +#define CONG_CNTRL_G(x) (((x) >> CONG_CNTRL_S) & CONG_CNTRL_M) + +#define T5_ISS_S 18 +#define T5_ISS_V(x) ((x) << T5_ISS_S) +#define T5_ISS_F T5_ISS_V(1U) + +#endif /* _T4FW_RI_API_H_ */ diff --git a/kernel/drivers/infiniband/hw/cxgb4/user.h b/kernel/drivers/infiniband/hw/cxgb4/user.h new file mode 100644 index 000000000..cbd0ce170 --- /dev/null +++ b/kernel/drivers/infiniband/hw/cxgb4/user.h @@ -0,0 +1,80 @@ +/* + * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * 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. + */ +#ifndef __C4IW_USER_H__ +#define __C4IW_USER_H__ + +#define C4IW_UVERBS_ABI_VERSION 2 + +/* + * Make sure that all structs defined in this file remain laid out so + * that they pack the same way on 32-bit and 64-bit architectures (to + * avoid incompatibility between 32-bit userspace and 64-bit kernels). + * In particular do not use pointer types -- pass pointers in __u64 + * instead. + */ +struct c4iw_create_cq_resp { + __u64 key; + __u64 gts_key; + __u64 memsize; + __u32 cqid; + __u32 size; + __u32 qid_mask; + __u32 reserved; /* explicit padding (optional for i386) */ +}; + + +enum { + C4IW_QPF_ONCHIP = (1<<0) +}; + +struct c4iw_create_qp_resp { + __u64 ma_sync_key; + __u64 sq_key; + __u64 rq_key; + __u64 sq_db_gts_key; + __u64 rq_db_gts_key; + __u64 sq_memsize; + __u64 rq_memsize; + __u32 sqid; + __u32 rqid; + __u32 sq_size; + __u32 rq_size; + __u32 qid_mask; + __u32 flags; +}; + +struct c4iw_alloc_ucontext_resp { + __u64 status_page_key; + __u32 status_page_size; + __u32 reserved; /* explicit padding (optional for i386) */ +}; +#endif |