diff options
| author |   Linus Torvalds <torvalds@linux-foundation.org> | 2016-08-04 20:10:31 -0400 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2016-08-04 20:10:31 -0400 |
| commit | 0cda611386b2fcbf8bb32e9a5d82bfed4856fc36 (patch) | |
| tree | 1647e00675ab924edfb22b69ea3872db091b8900 /drivers/infiniband/hw | |
| parent | Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending (diff) | |
| parent | Merge branches 'misc' and 'rxe' into k.o/for-4.8-1 (diff) | |
| download | linux-dev-0cda611386b2fcbf8bb32e9a5d82bfed4856fc36.tar.xz linux-dev-0cda611386b2fcbf8bb32e9a5d82bfed4856fc36.zip | |
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Pull base rdma updates from Doug Ledford:
"Round one of 4.8 code: while this is mostly normal, there is a new
driver in here (the driver was hosted outside the kernel for several
years and is actually a fairly mature and well coded driver). It
amounts to 13,000 of the 16,000 lines of added code in here.
Summary:
- Updates/fixes for iw_cxgb4 driver
- Updates/fixes for mlx5 driver
- Add flow steering and RSS API
- Add hardware stats to mlx4 and mlx5 drivers
- Add firmware version API for RDMA driver use
- Add the rxe driver (this is a software RoCE driver that makes any
Ethernet device a RoCE device)
- Fixes for i40iw driver
- Support for send only multicast joins in the cma layer
- Other minor fixes"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (72 commits)
Soft RoCE driver
IB/core: Support for CMA multicast join flags
IB/sa: Add cached attribute containing SM information to SA port
IB/uverbs: Fix race between uverbs_close and remove_one
IB/mthca: Clean up error unwind flow in mthca_reset()
IB/mthca: NULL arg to pci_dev_put is OK
IB/hfi1: NULL arg to sc_return_credits is OK
IB/mlx4: Add diagnostic hardware counters
net/mlx4: Query performance and diagnostics counters
net/mlx4: Add diagnostic counters capability bit
Use smaller 512 byte messages for portmapper messages
IB/ipoib: Report SG feature regardless of HW UD CSUM capability
IB/mlx4: Don't use GFP_ATOMIC for CQ resize struct
IB/hfi1: Disable by default
IB/rdmavt: Disable by default
IB/mlx5: Fix port counter ID association to QP offset
IB/mlx5: Fix iteration overrun in GSI qps
i40iw: Add NULL check for puda buffer
i40iw: Change dup_ack_thresh to u8
i40iw: Remove unnecessary check for moving CQ head
...
Diffstat (limited to 'drivers/infiniband/hw')
37 files changed, 2084 insertions, 436 deletions
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c index 3e8431b5cad7..04bbf172abde 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_cm.c +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c @@ -1396,10 +1396,10 @@ static int pass_accept_req(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) state_set(&child_ep->com, CONNECTING); child_ep->com.tdev = tdev; child_ep->com.cm_id = NULL; - child_ep->com.local_addr.sin_family = PF_INET; + child_ep->com.local_addr.sin_family = AF_INET; child_ep->com.local_addr.sin_port = req->local_port; child_ep->com.local_addr.sin_addr.s_addr = req->local_ip; - child_ep->com.remote_addr.sin_family = PF_INET; + child_ep->com.remote_addr.sin_family = AF_INET; child_ep->com.remote_addr.sin_port = req->peer_port; child_ep->com.remote_addr.sin_addr.s_addr = req->peer_ip; get_ep(&parent_ep->com); diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index bb1a839d4d6d..3edb80644b53 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -1183,18 +1183,6 @@ static ssize_t show_rev(struct device *dev, struct device_attribute *attr, return sprintf(buf, "%d\n", iwch_dev->rdev.t3cdev_p->type); } -static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, char *buf) -{ - struct iwch_dev *iwch_dev = container_of(dev, struct iwch_dev, - ibdev.dev); - struct ethtool_drvinfo info; - struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev; - - PDBG("%s dev 0x%p\n", __func__, dev); - lldev->ethtool_ops->get_drvinfo(lldev, &info); - return sprintf(buf, "%s\n", info.fw_version); -} - static ssize_t show_hca(struct device *dev, struct device_attribute *attr, char *buf) { @@ -1334,13 +1322,11 @@ static int iwch_get_mib(struct ib_device *ibdev, struct rdma_hw_stats *stats, } 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 *iwch_class_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type, &dev_attr_board_id, }; @@ -1362,6 +1348,18 @@ static int iwch_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_dev_fw_ver_str(struct ib_device *ibdev, char *str, + size_t str_len) +{ + struct iwch_dev *iwch_dev = to_iwch_dev(ibdev); + struct ethtool_drvinfo info; + struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev; + + PDBG("%s dev 0x%p\n", __func__, iwch_dev); + lldev->ethtool_ops->get_drvinfo(lldev, &info); + snprintf(str, str_len, "%s", info.fw_version); +} + int iwch_register_device(struct iwch_dev *dev) { int ret; @@ -1437,6 +1435,7 @@ int iwch_register_device(struct iwch_dev *dev) dev->ibdev.get_hw_stats = iwch_get_mib; dev->ibdev.uverbs_abi_ver = IWCH_UVERBS_ABI_VERSION; dev->ibdev.get_port_immutable = iwch_port_immutable; + dev->ibdev.get_dev_fw_str = get_dev_fw_ver_str; dev->ibdev.iwcm = kmalloc(sizeof(struct iw_cm_verbs), GFP_KERNEL); if (!dev->ibdev.iwcm) diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c index a3a67216bce6..3aca7f6171b4 100644 --- a/drivers/infiniband/hw/cxgb4/cm.c +++ b/drivers/infiniband/hw/cxgb4/cm.c @@ -294,6 +294,25 @@ static void state_set(struct c4iw_ep_common *epc, enum c4iw_ep_state new) return; } +static int alloc_ep_skb_list(struct sk_buff_head *ep_skb_list, int size) +{ + struct sk_buff *skb; + unsigned int i; + size_t len; + + len = roundup(sizeof(union cpl_wr_size), 16); + for (i = 0; i < size; i++) { + skb = alloc_skb(len, GFP_KERNEL); + if (!skb) + goto fail; + skb_queue_tail(ep_skb_list, skb); + } + return 0; +fail: + skb_queue_purge(ep_skb_list); + return -ENOMEM; +} + static void *alloc_ep(int size, gfp_t gfp) { struct c4iw_ep_common *epc; @@ -384,6 +403,8 @@ void _c4iw_free_ep(struct kref *kref) if (ep->mpa_skb) kfree_skb(ep->mpa_skb); } + if (!skb_queue_empty(&ep->com.ep_skb_list)) + skb_queue_purge(&ep->com.ep_skb_list); kfree(ep); } @@ -620,25 +641,27 @@ static void abort_arp_failure(void *handle, struct sk_buff *skb) } } -static int send_flowc(struct c4iw_ep *ep, struct sk_buff *skb) +static int send_flowc(struct c4iw_ep *ep) { - unsigned int flowclen = 80; struct fw_flowc_wr *flowc; + struct sk_buff *skb = skb_dequeue(&ep->com.ep_skb_list); int i; u16 vlan = ep->l2t->vlan; int nparams; + if (WARN_ON(!skb)) + return -ENOMEM; + if (vlan == CPL_L2T_VLAN_NONE) nparams = 8; else nparams = 9; - skb = get_skb(skb, flowclen, GFP_KERNEL); - flowc = (struct fw_flowc_wr *)__skb_put(skb, flowclen); + flowc = (struct fw_flowc_wr *)__skb_put(skb, FLOWC_LEN); flowc->op_to_nparams = cpu_to_be32(FW_WR_OP_V(FW_FLOWC_WR) | FW_FLOWC_WR_NPARAMS_V(nparams)); - flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(flowclen, + flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(FLOWC_LEN, 16)) | FW_WR_FLOWID_V(ep->hwtid)); flowc->mnemval[0].mnemonic = FW_FLOWC_MNEM_PFNVFN; @@ -679,18 +702,16 @@ static int send_flowc(struct c4iw_ep *ep, struct sk_buff *skb) return c4iw_ofld_send(&ep->com.dev->rdev, skb); } -static int send_halfclose(struct c4iw_ep *ep, gfp_t gfp) +static int send_halfclose(struct c4iw_ep *ep) { struct cpl_close_con_req *req; - struct sk_buff *skb; + struct sk_buff *skb = skb_dequeue(&ep->com.ep_skb_list); 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__); + if (WARN_ON(!skb)) 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); @@ -701,26 +722,24 @@ static int send_halfclose(struct c4iw_ep *ep, gfp_t gfp) 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) +static int send_abort(struct c4iw_ep *ep) { struct cpl_abort_req *req; int wrlen = roundup(sizeof *req, 16); + struct sk_buff *req_skb = skb_dequeue(&ep->com.ep_skb_list); 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__); + if (WARN_ON(!req_skb)) return -ENOMEM; - } - set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); - t4_set_arp_err_handler(skb, ep, abort_arp_failure); - req = (struct cpl_abort_req *) skb_put(skb, wrlen); + + set_wr_txq(req_skb, CPL_PRIORITY_DATA, ep->txq_idx); + t4_set_arp_err_handler(req_skb, ep, abort_arp_failure); + req = (struct cpl_abort_req *)skb_put(req_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); + return c4iw_l2t_send(&ep->com.dev->rdev, req_skb, ep->l2t); } static void best_mtu(const unsigned short *mtus, unsigned short mtu, @@ -992,9 +1011,19 @@ static int send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb, 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->flags = 0; + if (crc_enabled) + mpa->flags |= MPA_CRC; + if (markers_enabled) { + mpa->flags |= MPA_MARKERS; + ep->mpa_attr.recv_marker_enabled = 1; + } else { + ep->mpa_attr.recv_marker_enabled = 0; + } + if (mpa_rev_to_use == 2) + mpa->flags |= MPA_ENHANCED_RDMA_CONN; + mpa->private_data_size = htons(ep->plen); mpa->revision = mpa_rev_to_use; if (mpa_rev_to_use == 1) { @@ -1169,8 +1198,11 @@ static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen) 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->flags = 0; + if (ep->mpa_attr.crc_enabled) + mpa->flags |= MPA_CRC; + if (ep->mpa_attr.recv_marker_enabled) + mpa->flags |= MPA_MARKERS; mpa->revision = ep->mpa_attr.version; mpa->private_data_size = htons(plen); @@ -1248,7 +1280,7 @@ static int act_establish(struct c4iw_dev *dev, struct sk_buff *skb) set_bit(ACT_ESTAB, &ep->com.history); /* start MPA negotiation */ - ret = send_flowc(ep, NULL); + ret = send_flowc(ep); if (ret) goto err; if (ep->retry_with_mpa_v1) @@ -1555,7 +1587,6 @@ static int process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb) */ __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; @@ -2004,12 +2035,17 @@ static int send_fw_act_open_req(struct c4iw_ep *ep, unsigned int atid) } /* - * Return whether a failed active open has allocated a TID + * Some of the error codes above implicitly indicate that there is no TID + * allocated with the result of an ACT_OPEN. We use this predicate to make + * that explicit. */ 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; + return (status != CPL_ERR_TCAM_PARITY && + status != CPL_ERR_TCAM_MISS && + status != CPL_ERR_TCAM_FULL && + status != CPL_ERR_CONN_EXIST_SYNRECV && + status != CPL_ERR_CONN_EXIST); } /* Returns whether a CPL status conveys negative advice. @@ -2130,6 +2166,7 @@ out: static int c4iw_reconnect(struct c4iw_ep *ep) { int err = 0; + int size = 0; struct sockaddr_in *laddr = (struct sockaddr_in *) &ep->com.cm_id->m_local_addr; struct sockaddr_in *raddr = (struct sockaddr_in *) @@ -2145,6 +2182,21 @@ static int c4iw_reconnect(struct c4iw_ep *ep) init_timer(&ep->timer); c4iw_init_wr_wait(&ep->com.wr_wait); + /* When MPA revision is different on nodes, the node with MPA_rev=2 + * tries to reconnect with MPA_rev 1 for the same EP through + * c4iw_reconnect(), where the same EP is assigned with new tid for + * further connection establishment. As we are using the same EP pointer + * for reconnect, few skbs are used during the previous c4iw_connect(), + * which leaves the EP with inadequate skbs for further + * c4iw_reconnect(), Further causing an assert BUG_ON() due to empty + * skb_list() during peer_abort(). Allocate skbs which is already used. + */ + size = (CN_MAX_CON_BUF - skb_queue_len(&ep->com.ep_skb_list)); + if (alloc_ep_skb_list(&ep->com.ep_skb_list, size)) { + err = -ENOMEM; + goto fail1; + } + /* * Allocate an active TID to initiate a TCP connection. */ @@ -2210,6 +2262,7 @@ fail2: * response of 1st connect request. */ connect_reply_upcall(ep, -ECONNRESET); +fail1: c4iw_put_ep(&ep->com); out: return err; @@ -2576,6 +2629,10 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb) if (peer_mss && child_ep->mtu > (peer_mss + hdrs)) child_ep->mtu = peer_mss + hdrs; + skb_queue_head_init(&child_ep->com.ep_skb_list); + if (alloc_ep_skb_list(&child_ep->com.ep_skb_list, CN_MAX_CON_BUF)) + goto fail; + state_set(&child_ep->com, CONNECTING); child_ep->com.dev = dev; child_ep->com.cm_id = NULL; @@ -2640,6 +2697,8 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb) (const u32 *)&sin6->sin6_addr.s6_addr, 1); } goto out; +fail: + c4iw_put_ep(&child_ep->com); reject: reject_cr(dev, hwtid, skb); if (parent_ep) @@ -2670,7 +2729,7 @@ static int pass_establish(struct c4iw_dev *dev, struct sk_buff *skb) ep->com.state = MPA_REQ_WAIT; start_ep_timer(ep); set_bit(PASS_ESTAB, &ep->com.history); - ret = send_flowc(ep, skb); + ret = send_flowc(ep); mutex_unlock(&ep->com.mutex); if (ret) c4iw_ep_disconnect(ep, 1, GFP_KERNEL); @@ -2871,10 +2930,8 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) } 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__); + rpl_skb = skb_dequeue(&ep->com.ep_skb_list); + if (WARN_ON(!rpl_skb)) { release = 1; goto out; } @@ -3011,9 +3068,9 @@ static int fw4_ack(struct c4iw_dev *dev, struct sk_buff *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); + mutex_lock(&ep->com.mutex); kfree_skb(ep->mpa_skb); ep->mpa_skb = NULL; - mutex_lock(&ep->com.mutex); if (test_bit(STOP_MPA_TIMER, &ep->com.flags)) stop_ep_timer(ep); mutex_unlock(&ep->com.mutex); @@ -3025,9 +3082,9 @@ out: int c4iw_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len) { - int err = 0; - int disconnect = 0; + int abort; struct c4iw_ep *ep = to_ep(cm_id); + PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); mutex_lock(&ep->com.mutex); @@ -3038,16 +3095,13 @@ int c4iw_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len) } set_bit(ULP_REJECT, &ep->com.history); if (mpa_rev == 0) - disconnect = 2; - else { - err = send_mpa_reject(ep, pdata, pdata_len); - disconnect = 1; - } + abort = 1; + else + abort = send_mpa_reject(ep, pdata, pdata_len); mutex_unlock(&ep->com.mutex); - if (disconnect) { - stop_ep_timer(ep); - err = c4iw_ep_disconnect(ep, disconnect == 2, GFP_KERNEL); - } + + stop_ep_timer(ep); + c4iw_ep_disconnect(ep, abort != 0, GFP_KERNEL); c4iw_put_ep(&ep->com); return 0; } @@ -3248,6 +3302,13 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) err = -ENOMEM; goto out; } + + skb_queue_head_init(&ep->com.ep_skb_list); + if (alloc_ep_skb_list(&ep->com.ep_skb_list, CN_MAX_CON_BUF)) { + err = -ENOMEM; + goto fail1; + } + init_timer(&ep->timer); ep->plen = conn_param->private_data_len; if (ep->plen) @@ -3266,7 +3327,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) if (!ep->com.qp) { PDBG("%s qpn 0x%x not found!\n", __func__, conn_param->qpn); err = -EINVAL; - goto fail1; + goto fail2; } ref_qp(ep); PDBG("%s qpn 0x%x qp %p cm_id %p\n", __func__, conn_param->qpn, @@ -3279,7 +3340,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) if (ep->atid == -1) { printk(KERN_ERR MOD "%s - cannot alloc atid.\n", __func__); err = -ENOMEM; - goto fail1; + goto fail2; } insert_handle(dev, &dev->atid_idr, ep, ep->atid); @@ -3303,7 +3364,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) if (raddr->sin_addr.s_addr == htonl(INADDR_ANY)) { err = pick_local_ipaddrs(dev, cm_id); if (err) - goto fail1; + goto fail2; } /* find a route */ @@ -3323,7 +3384,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) if (ipv6_addr_type(&raddr6->sin6_addr) == IPV6_ADDR_ANY) { err = pick_local_ip6addrs(dev, cm_id); if (err) - goto fail1; + goto fail2; } /* find a route */ @@ -3339,14 +3400,14 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) if (!ep->dst) { printk(KERN_ERR MOD "%s - cannot find route.\n", __func__); err = -EHOSTUNREACH; - goto fail2; + goto fail3; } err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, true, ep->com.dev->rdev.lldi.adapter_type, cm_id->tos); if (err) { printk(KERN_ERR MOD "%s - cannot alloc l2e.\n", __func__); - goto fail3; + goto fail4; } PDBG("%s txq_idx %u tx_chan %u smac_idx %u rss_qid %u l2t_idx %u\n", @@ -3362,13 +3423,15 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) goto out; cxgb4_l2t_release(ep->l2t); -fail3: +fail4: dst_release(ep->dst); -fail2: +fail3: remove_handle(ep->com.dev, &ep->com.dev->atid_idr, ep->atid); cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid); -fail1: +fail2: + skb_queue_purge(&ep->com.ep_skb_list); deref_cm_id(&ep->com); +fail1: c4iw_put_ep(&ep->com); out: return err; @@ -3461,6 +3524,7 @@ int c4iw_create_listen(struct iw_cm_id *cm_id, int backlog) err = -ENOMEM; goto fail1; } + skb_queue_head_init(&ep->com.ep_skb_list); PDBG("%s ep %p\n", __func__, ep); ep->com.cm_id = cm_id; ref_cm_id(&ep->com); @@ -3577,11 +3641,22 @@ int c4iw_ep_disconnect(struct c4iw_ep *ep, int abrupt, gfp_t gfp) case MPA_REQ_RCVD: case MPA_REP_SENT: case FPDU_MODE: + case CONNECTING: close = 1; if (abrupt) ep->com.state = ABORTING; else { ep->com.state = CLOSING; + + /* + * if we close before we see the fw4_ack() then we fix + * up the timer state since we're reusing it. + */ + if (ep->mpa_skb && + test_bit(STOP_MPA_TIMER, &ep->com.flags)) { + clear_bit(STOP_MPA_TIMER, &ep->com.flags); + stop_ep_timer(ep); + } start_ep_timer(ep); } set_bit(CLOSE_SENT, &ep->com.flags); @@ -3611,10 +3686,10 @@ int c4iw_ep_disconnect(struct c4iw_ep *ep, int abrupt, gfp_t gfp) if (abrupt) { set_bit(EP_DISC_ABORT, &ep->com.history); close_complete_upcall(ep, -ECONNRESET); - ret = send_abort(ep, NULL, gfp); + ret = send_abort(ep); } else { set_bit(EP_DISC_CLOSE, &ep->com.history); - ret = send_halfclose(ep, gfp); + ret = send_halfclose(ep); } if (ret) { set_bit(EP_DISC_FAIL, &ep->com.history); diff --git a/drivers/infiniband/hw/cxgb4/cq.c b/drivers/infiniband/hw/cxgb4/cq.c index b0b955724458..812ab7278b8e 100644 --- a/drivers/infiniband/hw/cxgb4/cq.c +++ b/drivers/infiniband/hw/cxgb4/cq.c @@ -33,19 +33,15 @@ #include "iw_cxgb4.h" static int destroy_cq(struct c4iw_rdev *rdev, struct t4_cq *cq, - struct c4iw_dev_ucontext *uctx) + struct c4iw_dev_ucontext *uctx, struct sk_buff *skb) { 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); @@ -863,7 +859,9 @@ int c4iw_destroy_cq(struct ib_cq *ib_cq) 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); + ucontext ? &ucontext->uctx : &chp->cq.rdev->uctx, + chp->destroy_skb); + chp->destroy_skb = NULL; kfree(chp); return 0; } @@ -879,7 +877,7 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, struct c4iw_cq *chp; struct c4iw_create_cq_resp uresp; struct c4iw_ucontext *ucontext = NULL; - int ret; + int ret, wr_len; size_t memsize, hwentries; struct c4iw_mm_entry *mm, *mm2; @@ -896,6 +894,13 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, if (!chp) return ERR_PTR(-ENOMEM); + wr_len = sizeof(struct fw_ri_res_wr) + sizeof(struct fw_ri_res); + chp->destroy_skb = alloc_skb(wr_len, GFP_KERNEL); + if (!chp->destroy_skb) { + ret = -ENOMEM; + goto err1; + } + if (ib_context) ucontext = to_c4iw_ucontext(ib_context); @@ -936,7 +941,7 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, ret = create_cq(&rhp->rdev, &chp->cq, ucontext ? &ucontext->uctx : &rhp->rdev.uctx); if (ret) - goto err1; + goto err2; chp->rhp = rhp; chp->cq.size--; /* status page */ @@ -947,15 +952,15 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, init_waitqueue_head(&chp->wait); ret = insert_handle(rhp, &rhp->cqidr, chp, chp->cq.cqid); if (ret) - goto err2; + goto err3; if (ucontext) { mm = kmalloc(sizeof *mm, GFP_KERNEL); if (!mm) - goto err3; + goto err4; mm2 = kmalloc(sizeof *mm2, GFP_KERNEL); if (!mm2) - goto err4; + goto err5; uresp.qid_mask = rhp->rdev.cqmask; uresp.cqid = chp->cq.cqid; @@ -970,7 +975,7 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, ret = ib_copy_to_udata(udata, &uresp, sizeof(uresp) - sizeof(uresp.reserved)); if (ret) - goto err5; + goto err6; mm->key = uresp.key; mm->addr = virt_to_phys(chp->cq.queue); @@ -986,15 +991,18 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, __func__, chp->cq.cqid, chp, chp->cq.size, chp->cq.memsize, (unsigned long long) chp->cq.dma_addr); return &chp->ibcq; -err5: +err6: kfree(mm2); -err4: +err5: kfree(mm); -err3: +err4: remove_handle(rhp, &rhp->cqidr, chp->cq.cqid); -err2: +err3: destroy_cq(&chp->rhp->rdev, &chp->cq, - ucontext ? &ucontext->uctx : &rhp->rdev.uctx); + ucontext ? &ucontext->uctx : &rhp->rdev.uctx, + chp->destroy_skb); +err2: + kfree_skb(chp->destroy_skb); err1: kfree(chp); return ERR_PTR(ret); diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c index ae2e8b23d2dd..071d7332ec06 100644 --- a/drivers/infiniband/hw/cxgb4/device.c +++ b/drivers/infiniband/hw/cxgb4/device.c @@ -317,7 +317,7 @@ static int qp_open(struct inode *inode, struct file *file) idr_for_each(&qpd->devp->qpidr, count_idrs, &count); spin_unlock_irq(&qpd->devp->lock); - qpd->bufsize = count * 128; + qpd->bufsize = count * 180; qpd->buf = vmalloc(qpd->bufsize); if (!qpd->buf) { kfree(qpd); diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h index f6f34a75af27..aa47e0ae80bc 100644 --- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h +++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h @@ -384,6 +384,7 @@ struct c4iw_mr { struct ib_mr ibmr; struct ib_umem *umem; struct c4iw_dev *rhp; + struct sk_buff *dereg_skb; u64 kva; struct tpt_attributes attr; u64 *mpl; @@ -400,6 +401,7 @@ static inline struct c4iw_mr *to_c4iw_mr(struct ib_mr *ibmr) struct c4iw_mw { struct ib_mw ibmw; struct c4iw_dev *rhp; + struct sk_buff *dereg_skb; u64 kva; struct tpt_attributes attr; }; @@ -412,6 +414,7 @@ static inline struct c4iw_mw *to_c4iw_mw(struct ib_mw *ibmw) struct c4iw_cq { struct ib_cq ibcq; struct c4iw_dev *rhp; + struct sk_buff *destroy_skb; struct t4_cq cq; spinlock_t lock; spinlock_t comp_handler_lock; @@ -472,7 +475,7 @@ struct c4iw_qp { struct t4_wq wq; spinlock_t lock; struct mutex mutex; - atomic_t refcnt; + struct kref kref; wait_queue_head_t wait; struct timer_list timer; int sq_sig_all; @@ -789,10 +792,29 @@ enum c4iw_ep_history { CM_ID_DEREFED = 28, }; +enum conn_pre_alloc_buffers { + CN_ABORT_REQ_BUF, + CN_ABORT_RPL_BUF, + CN_CLOSE_CON_REQ_BUF, + CN_DESTROY_BUF, + CN_FLOWC_BUF, + CN_MAX_CON_BUF +}; + +#define FLOWC_LEN 80 +union cpl_wr_size { + struct cpl_abort_req abrt_req; + struct cpl_abort_rpl abrt_rpl; + struct fw_ri_wr ri_req; + struct cpl_close_con_req close_req; + char flowc_buf[FLOWC_LEN]; +}; + struct c4iw_ep_common { struct iw_cm_id *cm_id; struct c4iw_qp *qp; struct c4iw_dev *dev; + struct sk_buff_head ep_skb_list; enum c4iw_ep_state state; struct kref kref; struct mutex mutex; diff --git a/drivers/infiniband/hw/cxgb4/mem.c b/drivers/infiniband/hw/cxgb4/mem.c index 55d0651ee4de..0b91b0f4df71 100644 --- a/drivers/infiniband/hw/cxgb4/mem.c +++ b/drivers/infiniband/hw/cxgb4/mem.c @@ -59,9 +59,9 @@ static int mr_exceeds_hw_limits(struct c4iw_dev *dev, u64 length) } static int _c4iw_write_mem_dma_aligned(struct c4iw_rdev *rdev, u32 addr, - u32 len, dma_addr_t data, int wait) + u32 len, dma_addr_t data, + int wait, struct sk_buff *skb) { - struct sk_buff *skb; struct ulp_mem_io *req; struct ulptx_sgl *sgl; u8 wr_len; @@ -74,9 +74,11 @@ static int _c4iw_write_mem_dma_aligned(struct c4iw_rdev *rdev, u32 addr, 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; + if (!skb) { + skb = alloc_skb(wr_len, GFP_KERNEL | __GFP_NOFAIL); + if (!skb) + return -ENOMEM; + } set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); req = (struct ulp_mem_io *)__skb_put(skb, wr_len); @@ -108,9 +110,8 @@ static int _c4iw_write_mem_dma_aligned(struct c4iw_rdev *rdev, u32 addr, } static int _c4iw_write_mem_inline(struct c4iw_rdev *rdev, u32 addr, u32 len, - void *data) + void *data, struct sk_buff *skb) { - struct sk_buff *skb; struct ulp_mem_io *req; struct ulptx_idata *sc; u8 wr_len, *to_dp, *from_dp; @@ -134,9 +135,11 @@ static int _c4iw_write_mem_inline(struct c4iw_rdev *rdev, u32 addr, u32 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; + if (!skb) { + skb = alloc_skb(wr_len, GFP_KERNEL | __GFP_NOFAIL); + if (!skb) + return -ENOMEM; + } set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0); req = (struct ulp_mem_io *)__skb_put(skb, wr_len); @@ -173,6 +176,7 @@ static int _c4iw_write_mem_inline(struct c4iw_rdev *rdev, u32 addr, u32 len, memset(to_dp + copy_len, 0, T4_ULPTX_MIN_IO - (copy_len % T4_ULPTX_MIN_IO)); ret = c4iw_ofld_send(rdev, skb); + skb = NULL; if (ret) return ret; len -= C4IW_MAX_INLINE_SIZE; @@ -182,7 +186,8 @@ static int _c4iw_write_mem_inline(struct c4iw_rdev *rdev, u32 addr, u32 len, return ret; } -static int _c4iw_write_mem_dma(struct c4iw_rdev *rdev, u32 addr, u32 len, void *data) +static int _c4iw_write_mem_dma(struct c4iw_rdev *rdev, u32 addr, u32 len, + void *data, struct sk_buff *skb) { u32 remain = len; u32 dmalen; @@ -205,7 +210,7 @@ static int _c4iw_write_mem_dma(struct c4iw_rdev *rdev, u32 addr, u32 len, void * dmalen = T4_ULPTX_MAX_DMA; remain -= dmalen; ret = _c4iw_write_mem_dma_aligned(rdev, addr, dmalen, daddr, - !remain); + !remain, skb); if (ret) goto out; addr += dmalen >> 5; @@ -213,7 +218,7 @@ static int _c4iw_write_mem_dma(struct c4iw_rdev *rdev, u32 addr, u32 len, void * daddr += dmalen; } if (remain) - ret = _c4iw_write_mem_inline(rdev, addr, remain, data); + ret = _c4iw_write_mem_inline(rdev, addr, remain, data, skb); out: dma_unmap_single(&rdev->lldi.pdev->dev, save, len, DMA_TO_DEVICE); return ret; @@ -224,23 +229,25 @@ out: * 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) + void *data, struct sk_buff *skb) { if (is_t5(rdev->lldi.adapter_type) && use_dsgl) { if (len > inline_threshold) { - if (_c4iw_write_mem_dma(rdev, addr, len, data)) { + if (_c4iw_write_mem_dma(rdev, addr, len, data, skb)) { 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 + data, skb); + } else { return 0; + } } else - return _c4iw_write_mem_inline(rdev, addr, len, data); + return _c4iw_write_mem_inline(rdev, addr, + len, data, skb); } else - return _c4iw_write_mem_inline(rdev, addr, len, data); + return _c4iw_write_mem_inline(rdev, addr, len, data, skb); } /* @@ -253,7 +260,8 @@ 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) + u64 len, u8 page_size, u32 pbl_size, u32 pbl_addr, + struct sk_buff *skb) { int err; struct fw_ri_tpte tpt; @@ -307,7 +315,7 @@ static int write_tpt_entry(struct c4iw_rdev *rdev, u32 reset_tpt_entry, } err = write_adapter_mem(rdev, stag_idx + (rdev->lldi.vr->stag.start >> 5), - sizeof(tpt), &tpt); + sizeof(tpt), &tpt, skb); if (reset_tpt_entry) { c4iw_put_resource(&rdev->resource.tpt_table, stag_idx); @@ -327,28 +335,29 @@ static int write_pbl(struct c4iw_rdev *rdev, __be64 *pbl, __func__, pbl_addr, rdev->lldi.vr->pbl.start, pbl_size); - err = write_adapter_mem(rdev, pbl_addr >> 5, pbl_size << 3, pbl); + err = write_adapter_mem(rdev, pbl_addr >> 5, pbl_size << 3, pbl, NULL); return err; } static int dereg_mem(struct c4iw_rdev *rdev, u32 stag, u32 pbl_size, - u32 pbl_addr) + u32 pbl_addr, struct sk_buff *skb) { return write_tpt_entry(rdev, 1, &stag, 0, 0, 0, 0, 0, 0, 0UL, 0, 0, - pbl_size, pbl_addr); + pbl_size, pbl_addr, skb); } 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); + 0UL, 0, 0, 0, 0, NULL); } -static int deallocate_window(struct c4iw_rdev *rdev, u32 stag) +static int deallocate_window(struct c4iw_rdev *rdev, u32 stag, + struct sk_buff *skb) { return write_tpt_entry(rdev, 1, &stag, 0, 0, 0, 0, 0, 0, 0UL, 0, 0, 0, - 0); + 0, skb); } static int allocate_stag(struct c4iw_rdev *rdev, u32 *stag, u32 pdid, @@ -356,7 +365,7 @@ static int allocate_stag(struct c4iw_rdev *rdev, u32 *stag, u32 pdid, { *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); + 0UL, 0, 0, pbl_size, pbl_addr, NULL); } static int finish_mem_reg(struct c4iw_mr *mhp, u32 stag) @@ -383,14 +392,16 @@ static int register_mem(struct c4iw_dev *rhp, struct c4iw_pd *php, 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); + mhp->attr.pbl_size, mhp->attr.pbl_addr, NULL); if (ret) return ret; ret = finish_mem_reg(mhp, stag); - if (ret) + if (ret) { dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, - mhp->attr.pbl_addr); + mhp->attr.pbl_addr, mhp->dereg_skb); + mhp->dereg_skb = NULL; + } return ret; } @@ -423,6 +434,12 @@ struct ib_mr *c4iw_get_dma_mr(struct ib_pd *pd, int acc) if (!mhp) return ERR_PTR(-ENOMEM); + mhp->dereg_skb = alloc_skb(SGE_MAX_WR_LEN, GFP_KERNEL); + if (!mhp->dereg_skb) { + ret = -ENOMEM; + goto err0; + } + mhp->rhp = rhp; mhp->attr.pdid = php->pdid; mhp->attr.perms = c4iw_ib_to_tpt_access(acc); @@ -435,7 +452,8 @@ struct ib_mr *c4iw_get_dma_mr(struct ib_pd *pd, int acc) 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); + mhp->attr.mw_bind_enable, 0, 0, ~0ULL, 0, 0, 0, + NULL); if (ret) goto err1; @@ -445,8 +463,10 @@ struct ib_mr *c4iw_get_dma_mr(struct ib_pd *pd, int acc) return &mhp->ibmr; err2: dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, - mhp->attr.pbl_addr); + mhp->attr.pbl_addr, mhp->dereg_skb); err1: + kfree_skb(mhp->dereg_skb); +err0: kfree(mhp); return ERR_PTR(ret); } @@ -481,11 +501,18 @@ struct ib_mr *c4iw_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, if (!mhp) return ERR_PTR(-ENOMEM); + mhp->dereg_skb = alloc_skb(SGE_MAX_WR_LEN, GFP_KERNEL); + if (!mhp->dereg_skb) { + kfree(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_skb(mhp->dereg_skb); kfree(mhp); return ERR_PTR(err); } @@ -550,6 +577,7 @@ err_pbl: err: ib_umem_release(mhp->umem); + kfree_skb(mhp->dereg_skb); kfree(mhp); return ERR_PTR(err); } @@ -572,11 +600,16 @@ struct ib_mw *c4iw_alloc_mw(struct ib_pd *pd, enum ib_mw_type type, 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->dereg_skb = alloc_skb(SGE_MAX_WR_LEN, GFP_KERNEL); + if (!mhp->dereg_skb) { + ret = -ENOMEM; + goto free_mhp; } + + ret = allocate_window(&rhp->rdev, &stag, php->pdid); + if (ret) + goto free_skb; mhp->rhp = rhp; mhp->attr.pdid = php->pdid; mhp->attr.type = FW_RI_STAG_MW; @@ -584,12 +617,19 @@ struct ib_mw *c4iw_alloc_mw(struct ib_pd *pd, enum ib_mw_type type, 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); + ret = -ENOMEM; + goto dealloc_win; } PDBG("%s mmid 0x%x mhp %p stag 0x%x\n", __func__, mmid, mhp, stag); return &(mhp->ibmw); + +dealloc_win: + deallocate_window(&rhp->rdev, mhp->attr.stag, mhp->dereg_skb); +free_skb: + kfree_skb(mhp->dereg_skb); +free_mhp: + kfree(mhp); + return ERR_PTR(ret); } int c4iw_dealloc_mw(struct ib_mw *mw) @@ -602,7 +642,8 @@ int c4iw_dealloc_mw(struct ib_mw *mw) rhp = mhp->rhp; mmid = (mw->rkey) >> 8; remove_handle(rhp, &rhp->mmidr, mmid); - deallocate_window(&rhp->rdev, mhp->attr.stag); + deallocate_window(&rhp->rdev, mhp->attr.stag, mhp->dereg_skb); + kfree_skb(mhp->dereg_skb); kfree(mhp); PDBG("%s ib_mw %p mmid 0x%x ptr %p\n", __func__, mw, mmid, mhp); return 0; @@ -666,7 +707,7 @@ struct ib_mr *c4iw_alloc_mr(struct ib_pd *pd, return &(mhp->ibmr); err3: dereg_mem(&rhp->rdev, stag, mhp->attr.pbl_size, - mhp->attr.pbl_addr); + mhp->attr.pbl_addr, mhp->dereg_skb); err2: c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr, mhp->attr.pbl_size << 3); @@ -717,7 +758,7 @@ int c4iw_dereg_mr(struct ib_mr *ib_mr) dma_free_coherent(&mhp->rhp->rdev.lldi.pdev->dev, mhp->max_mpl_len, mhp->mpl, mhp->mpl_addr); dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size, - mhp->attr.pbl_addr); + mhp->attr.pbl_addr, mhp->dereg_skb); if (mhp->attr.pbl_size) c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr, mhp->attr.pbl_size << 3); diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c index dd8a86b726d2..df127ce6b6ec 100644 --- a/drivers/infiniband/hw/cxgb4/provider.c +++ b/drivers/infiniband/hw/cxgb4/provider.c @@ -409,20 +409,6 @@ static ssize_t show_rev(struct device *dev, struct device_attribute *attr, 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) { @@ -502,13 +488,11 @@ static int c4iw_get_mib(struct ib_device *ibdev, } 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, }; @@ -530,6 +514,20 @@ static int c4iw_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_dev_fw_str(struct ib_device *dev, char *str, + size_t str_len) +{ + struct c4iw_dev *c4iw_dev = container_of(dev, struct c4iw_dev, + ibdev); + PDBG("%s dev 0x%p\n", __func__, dev); + + snprintf(str, str_len, "%u.%u.%u.%u", + 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)); +} + int c4iw_register_device(struct c4iw_dev *dev) { int ret; @@ -605,6 +603,7 @@ int c4iw_register_device(struct c4iw_dev *dev) dev->ibdev.get_hw_stats = c4iw_get_mib; dev->ibdev.uverbs_abi_ver = C4IW_UVERBS_ABI_VERSION; dev->ibdev.get_port_immutable = c4iw_port_immutable; + dev->ibdev.get_dev_fw_str = get_dev_fw_str; dev->ibdev.drain_sq = c4iw_drain_sq; dev->ibdev.drain_rq = c4iw_drain_rq; diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c index e8993e49b8b3..edb1172b6f54 100644 --- a/drivers/infiniband/hw/cxgb4/qp.c +++ b/drivers/infiniband/hw/cxgb4/qp.c @@ -683,17 +683,25 @@ static int build_inv_stag(union t4_wr *wqe, struct ib_send_wr *wr, return 0; } +void _free_qp(struct kref *kref) +{ + struct c4iw_qp *qhp; + + qhp = container_of(kref, struct c4iw_qp, kref); + PDBG("%s qhp %p\n", __func__, qhp); + kfree(qhp); +} + void c4iw_qp_add_ref(struct ib_qp *qp) { PDBG("%s ib_qp %p\n", __func__, qp); - atomic_inc(&(to_c4iw_qp(qp)->refcnt)); + kref_get(&to_c4iw_qp(qp)->kref); } 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)); + kref_put(&to_c4iw_qp(qp)->kref, _free_qp); } static void add_to_fc_list(struct list_head *head, struct list_head *entry) @@ -1081,9 +1089,10 @@ static void post_terminate(struct c4iw_qp *qhp, struct t4_cqe *err_cqe, 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) + skb = skb_dequeue(&qhp->ep->com.ep_skb_list); + if (WARN_ON(!skb)) return; + set_wr_txq(skb, CPL_PRIORITY_DATA, qhp->ep->txq_idx); wqe = (struct fw_ri_wr *)__skb_put(skb, sizeof(*wqe)); @@ -1202,9 +1211,10 @@ static int rdma_fini(struct c4iw_dev *rhp, struct c4iw_qp *qhp, 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) + skb = skb_dequeue(&ep->com.ep_skb_list); + if (WARN_ON(!skb)) return -ENOMEM; + set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx); wqe = (struct fw_ri_wr *)__skb_put(skb, sizeof(*wqe)); @@ -1592,8 +1602,6 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp) 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)) @@ -1606,8 +1614,9 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp) destroy_qp(&rhp->rdev, &qhp->wq, ucontext ? &ucontext->uctx : &rhp->rdev.uctx); + c4iw_qp_rem_ref(ib_qp); + PDBG("%s ib_qp %p qpid 0x%0x\n", __func__, ib_qp, qhp->wq.sq.qid); - kfree(qhp); return 0; } @@ -1704,7 +1713,7 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs, init_completion(&qhp->rq_drained); mutex_init(&qhp->mutex); init_waitqueue_head(&qhp->wait); - atomic_set(&qhp->refcnt, 1); + kref_init(&qhp->kref); ret = insert_handle(rhp, &rhp->qpidr, qhp, qhp->wq.sq.qid); if (ret) @@ -1896,12 +1905,20 @@ int c4iw_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, return 0; } +static void move_qp_to_err(struct c4iw_qp *qp) +{ + struct c4iw_qp_attributes attrs = { .next_state = C4IW_QP_STATE_ERROR }; + + (void)c4iw_modify_qp(qp->rhp, qp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); +} + void c4iw_drain_sq(struct ib_qp *ibqp) { struct c4iw_qp *qp = to_c4iw_qp(ibqp); unsigned long flag; bool need_to_wait; + move_qp_to_err(qp); spin_lock_irqsave(&qp->lock, flag); need_to_wait = !t4_sq_empty(&qp->wq); spin_unlock_irqrestore(&qp->lock, flag); @@ -1916,6 +1933,7 @@ void c4iw_drain_rq(struct ib_qp *ibqp) unsigned long flag; bool need_to_wait; + move_qp_to_err(qp); spin_lock_irqsave(&qp->lock, flag); need_to_wait = !t4_rq_empty(&qp->wq); spin_unlock_irqrestore(&qp->lock, flag); diff --git a/drivers/infiniband/hw/hfi1/Kconfig b/drivers/infiniband/hw/hfi1/Kconfig index a925fb0db706..f846fd51b85b 100644 --- a/drivers/infiniband/hw/hfi1/Kconfig +++ b/drivers/infiniband/hw/hfi1/Kconfig @@ -3,7 +3,6 @@ config INFINIBAND_HFI1 depends on X86_64 && INFINIBAND_RDMAVT select MMU_NOTIFIER select CRC32 - default m ---help--- This is a low-level driver for Intel OPA Gen1 adapter. config HFI1_DEBUG_SDMA_ORDER diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c index c702a009608f..32c19fad12a4 100644 --- a/drivers/infiniband/hw/hfi1/file_ops.c +++ b/drivers/infiniband/hw/hfi1/file_ops.c @@ -228,7 +228,7 @@ static long hfi1_file_ioctl(struct file *fp, unsigned int cmd, sizeof(struct hfi1_base_info)); break; case HFI1_IOCTL_CREDIT_UPD: - if (uctxt && uctxt->sc) + if (uctxt) sc_return_credits(uctxt->sc); break; diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h index 4417a0fd3ef9..49a71e24a8f0 100644 --- a/drivers/infiniband/hw/hfi1/hfi.h +++ b/drivers/infiniband/hw/hfi1/hfi.h @@ -1174,6 +1174,8 @@ struct hfi1_devdata { /* 8051 firmware version helper */ #define dc8051_ver(a, b) ((a) << 8 | (b)) +#define dc8051_ver_maj(a) ((a & 0xff00) >> 8) +#define dc8051_ver_min(a) (a & 0x00ff) /* f_put_tid types */ #define PT_EXPECTED 0 diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 849c4b9399d4..dd4be3c2b225 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -1291,9 +1291,12 @@ int hfi1_verbs_send(struct rvt_qp *qp, struct hfi1_pkt_state *ps) static void hfi1_fill_device_attr(struct hfi1_devdata *dd) { struct rvt_dev_info *rdi = &dd->verbs_dev.rdi; + u16 ver = dd->dc8051_ver; memset(&rdi->dparms.props, 0, sizeof(rdi->dparms.props)); + rdi->dparms.props.fw_ver = ((u64)(dc8051_ver_maj(ver)) << 16) | + (u64)dc8051_ver_min(ver); rdi->dparms.props.device_cap_flags = IB_DEVICE_BAD_PKEY_CNTR | IB_DEVICE_BAD_QKEY_CNTR | IB_DEVICE_SHUTDOWN_PORT | IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN | @@ -1567,6 +1570,17 @@ static void init_ibport(struct hfi1_pportdata *ppd) RCU_INIT_POINTER(ibp->rvp.qp[1], NULL); } +static void hfi1_get_dev_fw_str(struct ib_device *ibdev, char *str, + size_t str_len) +{ + struct rvt_dev_info *rdi = ib_to_rvt(ibdev); + struct hfi1_ibdev *dev = dev_from_rdi(rdi); + u16 ver = dd_from_dev(dev)->dc8051_ver; + + snprintf(str, str_len, "%u.%u", dc8051_ver_maj(ver), + dc8051_ver_min(ver)); +} + /** * hfi1_register_ib_device - register our device with the infiniband core * @dd: the device data structure @@ -1613,6 +1627,7 @@ int hfi1_register_ib_device(struct hfi1_devdata *dd) /* keep process mad in the driver */ ibdev->process_mad = hfi1_process_mad; + ibdev->get_dev_fw_str = hfi1_get_dev_fw_str; strncpy(ibdev->node_desc, init_utsname()->nodename, sizeof(ibdev->node_desc)); diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c index d2fa72516960..5026dc79978a 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_cm.c +++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c @@ -1567,12 +1567,12 @@ static enum i40iw_status_code i40iw_del_multiple_qhash( ret = i40iw_manage_qhash(iwdev, cm_info, I40IW_QHASH_TYPE_TCP_SYN, I40IW_QHASH_MANAGE_TYPE_DELETE, NULL, false); - kfree(child_listen_node); - cm_parent_listen_node->cm_core->stats_listen_nodes_destroyed++; i40iw_debug(&iwdev->sc_dev, I40IW_DEBUG_CM, "freed pointer = %p\n", child_listen_node); + kfree(child_listen_node); + cm_parent_listen_node->cm_core->stats_listen_nodes_destroyed++; } spin_unlock_irqrestore(&iwdev->cm_core.listen_list_lock, flags); diff --git a/drivers/infiniband/hw/i40iw/i40iw_d.h b/drivers/infiniband/hw/i40iw/i40iw_d.h index bd942da91a27..2fac1db0e0a0 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_d.h +++ b/drivers/infiniband/hw/i40iw/i40iw_d.h @@ -1557,6 +1557,9 @@ enum i40iw_alignment { #define I40IW_RING_MOVE_TAIL(_ring) \ (_ring).tail = ((_ring).tail + 1) % (_ring).size +#define I40IW_RING_MOVE_HEAD_NOCHECK(_ring) \ + (_ring).head = ((_ring).head + 1) % (_ring).size + #define I40IW_RING_MOVE_TAIL_BY_COUNT(_ring, _count) \ (_ring).tail = ((_ring).tail + (_count)) % (_ring).size diff --git a/drivers/infiniband/hw/i40iw/i40iw_puda.c b/drivers/infiniband/hw/i40iw/i40iw_puda.c index e9c6e82af9c7..c62d354f7810 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_puda.c +++ b/drivers/infiniband/hw/i40iw/i40iw_puda.c @@ -1025,6 +1025,8 @@ static void i40iw_ieq_compl_pfpdu(struct i40iw_puda_rsrc *ieq, u16 txoffset, bufoffset; buf = i40iw_puda_get_listbuf(pbufl); + if (!buf) + return; nextseqnum = buf->seqnum + fpdu_len; txbuf->totallen = buf->hdrlen + fpdu_len; txbuf->data = (u8 *)txbuf->mem.va + buf->hdrlen; @@ -1048,6 +1050,8 @@ static void i40iw_ieq_compl_pfpdu(struct i40iw_puda_rsrc *ieq, fpdu_len -= buf->datalen; i40iw_puda_ret_bufpool(ieq, buf); buf = i40iw_puda_get_listbuf(pbufl); + if (!buf) + return; bufoffset = (u16)(buf->data - (u8 *)buf->mem.va); } while (1); diff --git a/drivers/infiniband/hw/i40iw/i40iw_type.h b/drivers/infiniband/hw/i40iw/i40iw_type.h index 16cc61720b53..2b1a04e9ca3c 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_type.h +++ b/drivers/infiniband/hw/i40iw/i40iw_type.h @@ -667,7 +667,7 @@ struct i40iw_tcp_offload_info { bool time_stamp; u8 cwnd_inc_limit; bool drop_ooo_seg; - bool dup_ack_thresh; + u8 dup_ack_thresh; u8 ttl; u8 src_mac_addr_idx; bool avoid_stretch_ack; diff --git a/drivers/infiniband/hw/i40iw/i40iw_uk.c b/drivers/infiniband/hw/i40iw/i40iw_uk.c index e35faea88c13..4d28c3cb03cc 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_uk.c +++ b/drivers/infiniband/hw/i40iw/i40iw_uk.c @@ -291,9 +291,9 @@ static enum i40iw_status_code i40iw_rdma_write(struct i40iw_qp_uk *qp, i40iw_set_fragment(wqe, 0, op_info->lo_sg_list); - for (i = 1; i < op_info->num_lo_sges; i++) { - byte_off = 32 + (i - 1) * 16; + for (i = 1, byte_off = 32; i < op_info->num_lo_sges; i++) { i40iw_set_fragment(wqe, byte_off, &op_info->lo_sg_list[i]); + byte_off += 16; } wmb(); /* make sure WQE is populated before valid bit is set */ @@ -401,9 +401,9 @@ static enum i40iw_status_code i40iw_send(struct i40iw_qp_uk *qp, i40iw_set_fragment(wqe, 0, op_info->sg_list); - for (i = 1; i < op_info->num_sges; i++) { - byte_off = 32 + (i - 1) * 16; + for (i = 1, byte_off = 32; i < op_info->num_sges; i++) { i40iw_set_fragment(wqe, byte_off, &op_info->sg_list[i]); + byte_off += 16; } wmb(); /* make sure WQE is populated before valid bit is set */ @@ -685,9 +685,9 @@ static enum i40iw_status_code i40iw_post_receive(struct i40iw_qp_uk *qp, i40iw_set_fragment(wqe, 0, info->sg_list); - for (i = 1; i < info->num_sges; i++) { - byte_off = 32 + (i - 1) * 16; + for (i = 1, byte_off = 32; i < info->num_sges; i++) { i40iw_set_fragment(wqe, byte_off, &info->sg_list[i]); + byte_off += 16; } wmb(); /* make sure WQE is populated before valid bit is set */ @@ -753,8 +753,7 @@ static enum i40iw_status_code i40iw_cq_post_entries(struct i40iw_cq_uk *cq, * @post_cq: update cq tail */ static enum i40iw_status_code i40iw_cq_poll_completion(struct i40iw_cq_uk *cq, - struct i40iw_cq_poll_info *info, - bool post_cq) + struct i40iw_cq_poll_info *info) { u64 comp_ctx, qword0, qword2, qword3, wqe_qword; u64 *cqe, *sw_wqe; @@ -762,7 +761,6 @@ static enum i40iw_status_code i40iw_cq_poll_completion(struct i40iw_cq_uk *cq, struct i40iw_ring *pring = NULL; u32 wqe_idx, q_type, array_idx = 0; enum i40iw_status_code ret_code = 0; - enum i40iw_status_code ret_code2 = 0; bool move_cq_head = true; u8 polarity; u8 addl_wqes = 0; @@ -870,19 +868,14 @@ exit: move_cq_head = false; if (move_cq_head) { - I40IW_RING_MOVE_HEAD(cq->cq_ring, ret_code2); - - if (ret_code2 && !ret_code) - ret_code = ret_code2; + I40IW_RING_MOVE_HEAD_NOCHECK(cq->cq_ring); if (I40IW_RING_GETCURRENT_HEAD(cq->cq_ring) == 0) cq->polarity ^= 1; - if (post_cq) { - I40IW_RING_MOVE_TAIL(cq->cq_ring); - set_64bit_val(cq->shadow_area, 0, - I40IW_RING_GETCURRENT_HEAD(cq->cq_ring)); - } + I40IW_RING_MOVE_TAIL(cq->cq_ring); + set_64bit_val(cq->shadow_area, 0, + I40IW_RING_GETCURRENT_HEAD(cq->cq_ring)); } else { if (info->is_srq) return ret_code; diff --git a/drivers/infiniband/hw/i40iw/i40iw_user.h b/drivers/infiniband/hw/i40iw/i40iw_user.h index 4627646fe8cd..276bcefffd7e 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_user.h +++ b/drivers/infiniband/hw/i40iw/i40iw_user.h @@ -327,7 +327,7 @@ struct i40iw_cq_ops { void (*iw_cq_request_notification)(struct i40iw_cq_uk *, enum i40iw_completion_notify); enum i40iw_status_code (*iw_cq_poll_completion)(struct i40iw_cq_uk *, - struct i40iw_cq_poll_info *, bool); + struct i40iw_cq_poll_info *); enum i40iw_status_code (*iw_cq_post_entries)(struct i40iw_cq_uk *, u8 count); void (*iw_cq_clean)(void *, struct i40iw_cq_uk *); }; diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c index 283b64c942ee..2360338877bf 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c @@ -529,7 +529,7 @@ static int i40iw_setup_kmode_qp(struct i40iw_device *iwdev, status = i40iw_get_wqe_shift(rq_size, ukinfo->max_rq_frag_cnt, 0, &rqshift); if (status) - return -ENOSYS; + return -ENOMEM; sqdepth = sq_size << sqshift; rqdepth = rq_size << rqshift; @@ -671,7 +671,7 @@ static struct ib_qp *i40iw_create_qp(struct ib_pd *ibpd, iwqp->ctx_info.qp_compl_ctx = (uintptr_t)qp; if (init_attr->qp_type != IB_QPT_RC) { - err_code = -ENOSYS; + err_code = -EINVAL; goto error; } if (iwdev->push_mode) @@ -1840,6 +1840,7 @@ struct ib_mr *i40iw_reg_phys_mr(struct ib_pd *pd, iwmr->ibmr.lkey = stag; iwmr->page_cnt = 1; iwmr->pgaddrmem[0] = addr; + iwmr->length = size; status = i40iw_hwreg_mr(iwdev, iwmr, access); if (status) { i40iw_free_stag(iwdev, stag); @@ -1863,7 +1864,7 @@ static struct ib_mr *i40iw_get_dma_mr(struct ib_pd *pd, int acc) { u64 kva = 0; - return i40iw_reg_phys_mr(pd, 0, 0xffffffffffULL, acc, &kva); + return i40iw_reg_phys_mr(pd, 0, 0, acc, &kva); } /** @@ -1975,18 +1976,6 @@ static ssize_t i40iw_show_rev(struct device *dev, } /** - * i40iw_show_fw_ver - */ -static ssize_t i40iw_show_fw_ver(struct device *dev, - struct device_attribute *attr, char *buf) -{ - u32 firmware_version = I40IW_FW_VERSION; - - return sprintf(buf, "%u.%u\n", firmware_version, - (firmware_version & 0x000000ff)); -} - -/** * i40iw_show_hca */ static ssize_t i40iw_show_hca(struct device *dev, @@ -2006,13 +1995,11 @@ static ssize_t i40iw_show_board(struct device *dev, } static DEVICE_ATTR(hw_rev, S_IRUGO, i40iw_show_rev, NULL); -static DEVICE_ATTR(fw_ver, S_IRUGO, i40iw_show_fw_ver, NULL); static DEVICE_ATTR(hca_type, S_IRUGO, i40iw_show_hca, NULL); static DEVICE_ATTR(board_id, S_IRUGO, i40iw_show_board, NULL); static struct device_attribute *i40iw_dev_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type, &dev_attr_board_id }; @@ -2091,8 +2078,12 @@ static int i40iw_post_send(struct ib_qp *ibqp, ret = ukqp->ops.iw_send(ukqp, &info, ib_wr->ex.invalidate_rkey, false); } - if (ret) - err = -EIO; + if (ret) { + if (ret == I40IW_ERR_QP_TOOMANY_WRS_POSTED) + err = -ENOMEM; + else + err = -EINVAL; + } break; case IB_WR_RDMA_WRITE: info.op_type = I40IW_OP_TYPE_RDMA_WRITE; @@ -2113,8 +2104,12 @@ static int i40iw_post_send(struct ib_qp *ibqp, ret = ukqp->ops.iw_rdma_write(ukqp, &info, false); } - if (ret) - err = -EIO; + if (ret) { + if (ret == I40IW_ERR_QP_TOOMANY_WRS_POSTED) + err = -ENOMEM; + else + err = -EINVAL; + } break; case IB_WR_RDMA_READ_WITH_INV: inv_stag = true; @@ -2132,15 +2127,19 @@ static int i40iw_post_send(struct ib_qp *ibqp, info.op.rdma_read.lo_addr.stag = ib_wr->sg_list->lkey; info.op.rdma_read.lo_addr.len = ib_wr->sg_list->length; ret = ukqp->ops.iw_rdma_read(ukqp, &info, inv_stag, false); - if (ret) - err = -EIO; + if (ret) { + if (ret == I40IW_ERR_QP_TOOMANY_WRS_POSTED) + err = -ENOMEM; + else + err = -EINVAL; + } break; case IB_WR_LOCAL_INV: info.op_type = I40IW_OP_TYPE_INV_STAG; info.op.inv_local_stag.target_stag = ib_wr->ex.invalidate_rkey; ret = ukqp->ops.iw_stag_local_invalidate(ukqp, &info, true); if (ret) - err = -EIO; + err = -ENOMEM; break; case IB_WR_REG_MR: { @@ -2174,7 +2173,7 @@ static int i40iw_post_send(struct ib_qp *ibqp, ret = dev->iw_priv_qp_ops->iw_mr_fast_register(&iwqp->sc_qp, &info, true); if (ret) - err = -EIO; + err = -ENOMEM; break; } default: @@ -2214,6 +2213,7 @@ static int i40iw_post_recv(struct ib_qp *ibqp, struct i40iw_sge sg_list[I40IW_MAX_WQ_FRAGMENT_COUNT]; enum i40iw_status_code ret = 0; unsigned long flags; + int err = 0; iwqp = (struct i40iw_qp *)ibqp; ukqp = &iwqp->sc_qp.qp_uk; @@ -2228,6 +2228,10 @@ static int i40iw_post_recv(struct ib_qp *ibqp, ret = ukqp->ops.iw_post_receive(ukqp, &post_recv); if (ret) { i40iw_pr_err(" post_recv err %d\n", ret); + if (ret == I40IW_ERR_QP_TOOMANY_WRS_POSTED) + err = -ENOMEM; + else + err = -EINVAL; *bad_wr = ib_wr; goto out; } @@ -2235,9 +2239,7 @@ static int i40iw_post_recv(struct ib_qp *ibqp, } out: spin_unlock_irqrestore(&iwqp->lock, flags); - if (ret) - return -ENOSYS; - return 0; + return err; } /** @@ -2264,7 +2266,7 @@ static int i40iw_poll_cq(struct ib_cq *ibcq, spin_lock_irqsave(&iwcq->lock, flags); while (cqe_count < num_entries) { - ret = ukcq->ops.iw_cq_poll_completion(ukcq, &cq_poll_info, true); + ret = ukcq->ops.iw_cq_poll_completion(ukcq, &cq_poll_info); if (ret == I40IW_ERR_QUEUE_EMPTY) { break; } else if (ret == I40IW_ERR_QUEUE_DESTROYED) { @@ -2437,6 +2439,15 @@ static const char * const i40iw_hw_stat_names[] = { "iwRdmaInv" }; +static void i40iw_get_dev_fw_str(struct ib_device *dev, char *str, + size_t str_len) +{ + u32 firmware_version = I40IW_FW_VERSION; + + snprintf(str, str_len, "%u.%u", firmware_version, + (firmware_version & 0x000000ff)); +} + /** * i40iw_alloc_hw_stats - Allocate a hw stats structure * @ibdev: device pointer from stack @@ -2528,7 +2539,7 @@ static int i40iw_modify_port(struct ib_device *ibdev, int port_modify_mask, struct ib_port_modify *props) { - return 0; + return -ENOSYS; } /** @@ -2660,6 +2671,7 @@ static struct i40iw_ib_device *i40iw_init_rdma_device(struct i40iw_device *iwdev memcpy(iwibdev->ibdev.iwcm->ifname, netdev->name, sizeof(iwibdev->ibdev.iwcm->ifname)); iwibdev->ibdev.get_port_immutable = i40iw_port_immutable; + iwibdev->ibdev.get_dev_fw_str = i40iw_get_dev_fw_str; iwibdev->ibdev.poll_cq = i40iw_poll_cq; iwibdev->ibdev.req_notify_cq = i40iw_req_notify_cq; iwibdev->ibdev.post_send = i40iw_post_send; @@ -2723,7 +2735,7 @@ int i40iw_register_rdma_device(struct i40iw_device *iwdev) iwdev->iwibdev = i40iw_init_rdma_device(iwdev); if (!iwdev->iwibdev) - return -ENOSYS; + return -ENOMEM; iwibdev = iwdev->iwibdev; ret = ib_register_device(&iwibdev->ibdev, NULL); @@ -2748,5 +2760,5 @@ error: kfree(iwdev->iwibdev->ibdev.iwcm); iwdev->iwibdev->ibdev.iwcm = NULL; ib_dealloc_device(&iwdev->iwibdev->ibdev); - return -ENOSYS; + return ret; } diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c index 9f8b516eb2b0..d6fc8a6e8c33 100644 --- a/drivers/infiniband/hw/mlx4/cq.c +++ b/drivers/infiniband/hw/mlx4/cq.c @@ -288,7 +288,7 @@ static int mlx4_alloc_resize_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq, if (cq->resize_buf) return -EBUSY; - cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC); + cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_KERNEL); if (!cq->resize_buf) return -ENOMEM; @@ -316,7 +316,7 @@ static int mlx4_alloc_resize_umem(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) return -EFAULT; - cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC); + cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_KERNEL); if (!cq->resize_buf) return -ENOMEM; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 42a46078d7d5..2af44c2de262 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -2025,16 +2025,6 @@ static ssize_t show_hca(struct device *device, struct device_attribute *attr, return sprintf(buf, "MT%d\n", dev->dev->persist->pdev->device); } -static ssize_t show_fw_ver(struct device *device, struct device_attribute *attr, - char *buf) -{ - struct mlx4_ib_dev *dev = - container_of(device, struct mlx4_ib_dev, ib_dev.dev); - return sprintf(buf, "%d.%d.%d\n", (int) (dev->dev->caps.fw_ver >> 32), - (int) (dev->dev->caps.fw_ver >> 16) & 0xffff, - (int) dev->dev->caps.fw_ver & 0xffff); -} - static ssize_t show_rev(struct device *device, struct device_attribute *attr, char *buf) { @@ -2053,17 +2043,204 @@ static ssize_t show_board(struct device *device, struct device_attribute *attr, } 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 *mlx4_class_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type, &dev_attr_board_id }; +struct diag_counter { + const char *name; + u32 offset; +}; + +#define DIAG_COUNTER(_name, _offset) \ + { .name = #_name, .offset = _offset } + +static const struct diag_counter diag_basic[] = { + DIAG_COUNTER(rq_num_lle, 0x00), + DIAG_COUNTER(sq_num_lle, 0x04), + DIAG_COUNTER(rq_num_lqpoe, 0x08), + DIAG_COUNTER(sq_num_lqpoe, 0x0C), + DIAG_COUNTER(rq_num_lpe, 0x18), + DIAG_COUNTER(sq_num_lpe, 0x1C), + DIAG_COUNTER(rq_num_wrfe, 0x20), + DIAG_COUNTER(sq_num_wrfe, 0x24), + DIAG_COUNTER(sq_num_mwbe, 0x2C), + DIAG_COUNTER(sq_num_bre, 0x34), + DIAG_COUNTER(sq_num_rire, 0x44), + DIAG_COUNTER(rq_num_rire, 0x48), + DIAG_COUNTER(sq_num_rae, 0x4C), + DIAG_COUNTER(rq_num_rae, 0x50), + DIAG_COUNTER(sq_num_roe, 0x54), + DIAG_COUNTER(sq_num_tree, 0x5C), + DIAG_COUNTER(sq_num_rree, 0x64), + DIAG_COUNTER(rq_num_rnr, 0x68), + DIAG_COUNTER(sq_num_rnr, 0x6C), + DIAG_COUNTER(rq_num_oos, 0x100), + DIAG_COUNTER(sq_num_oos, 0x104), +}; + +static const struct diag_counter diag_ext[] = { + DIAG_COUNTER(rq_num_dup, 0x130), + DIAG_COUNTER(sq_num_to, 0x134), +}; + +static const struct diag_counter diag_device_only[] = { + DIAG_COUNTER(num_cqovf, 0x1A0), + DIAG_COUNTER(rq_num_udsdprd, 0x118), +}; + +static struct rdma_hw_stats *mlx4_ib_alloc_hw_stats(struct ib_device *ibdev, + u8 port_num) +{ + struct mlx4_ib_dev *dev = to_mdev(ibdev); + struct mlx4_ib_diag_counters *diag = dev->diag_counters; + + if (!diag[!!port_num].name) + return NULL; + + return rdma_alloc_hw_stats_struct(diag[!!port_num].name, + diag[!!port_num].num_counters, + RDMA_HW_STATS_DEFAULT_LIFESPAN); +} + +static int mlx4_ib_get_hw_stats(struct ib_device *ibdev, + struct rdma_hw_stats *stats, + u8 port, int index) +{ + struct mlx4_ib_dev *dev = to_mdev(ibdev); + struct mlx4_ib_diag_counters *diag = dev->diag_counters; + u32 hw_value[ARRAY_SIZE(diag_device_only) + + ARRAY_SIZE(diag_ext) + ARRAY_SIZE(diag_basic)] = {}; + int ret; + int i; + + ret = mlx4_query_diag_counters(dev->dev, + MLX4_OP_MOD_QUERY_TRANSPORT_CI_ERRORS, + diag[!!port].offset, hw_value, + diag[!!port].num_counters, port); + + if (ret) + return ret; + + for (i = 0; i < diag[!!port].num_counters; i++) + stats->value[i] = hw_value[i]; + + return diag[!!port].num_counters; +} + +static int __mlx4_ib_alloc_diag_counters(struct mlx4_ib_dev *ibdev, + const char ***name, + u32 **offset, + u32 *num, + bool port) +{ + u32 num_counters; + + num_counters = ARRAY_SIZE(diag_basic); + + if (ibdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_DIAG_PER_PORT) + num_counters += ARRAY_SIZE(diag_ext); + + if (!port) + num_counters += ARRAY_SIZE(diag_device_only); + + *name = kcalloc(num_counters, sizeof(**name), GFP_KERNEL); + if (!*name) + return -ENOMEM; + + *offset = kcalloc(num_counters, sizeof(**offset), GFP_KERNEL); + if (!*offset) + goto err_name; + + *num = num_counters; + + return 0; + +err_name: + kfree(*name); + return -ENOMEM; +} + +static void mlx4_ib_fill_diag_counters(struct mlx4_ib_dev *ibdev, + const char **name, + u32 *offset, + bool port) +{ + int i; + int j; + + for (i = 0, j = 0; i < ARRAY_SIZE(diag_basic); i++, j++) { + name[i] = diag_basic[i].name; + offset[i] = diag_basic[i].offset; + } + + if (ibdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_DIAG_PER_PORT) { + for (i = 0; i < ARRAY_SIZE(diag_ext); i++, j++) { + name[j] = diag_ext[i].name; + offset[j] = diag_ext[i].offset; + } + } + + if (!port) { + for (i = 0; i < ARRAY_SIZE(diag_device_only); i++, j++) { + name[j] = diag_device_only[i].name; + offset[j] = diag_device_only[i].offset; + } + } +} + +static int mlx4_ib_alloc_diag_counters(struct mlx4_ib_dev *ibdev) +{ + struct mlx4_ib_diag_counters *diag = ibdev->diag_counters; + int i; + int ret; + bool per_port = !!(ibdev->dev->caps.flags2 & + MLX4_DEV_CAP_FLAG2_DIAG_PER_PORT); + + for (i = 0; i < MLX4_DIAG_COUNTERS_TYPES; i++) { + /* i == 1 means we are building port counters */ + if (i && !per_port) + continue; + + ret = __mlx4_ib_alloc_diag_counters(ibdev, &diag[i].name, + &diag[i].offset, + &diag[i].num_counters, i); + if (ret) + goto err_alloc; + + mlx4_ib_fill_diag_counters(ibdev, diag[i].name, + diag[i].offset, i); + } + + ibdev->ib_dev.get_hw_stats = mlx4_ib_get_hw_stats; + ibdev->ib_dev.alloc_hw_stats = mlx4_ib_alloc_hw_stats; + + return 0; + +err_alloc: + if (i) { + kfree(diag[i - 1].name); + kfree(diag[i - 1].offset); + } + + return ret; +} + +static void mlx4_ib_diag_cleanup(struct mlx4_ib_dev *ibdev) +{ + int i; + + for (i = 0; i < MLX4_DIAG_COUNTERS_TYPES; i++) { + kfree(ibdev->diag_counters[i].offset); + kfree(ibdev->diag_counters[i].name); + } +} + #define MLX4_IB_INVALID_MAC ((u64)-1) static void mlx4_ib_update_qps(struct mlx4_ib_dev *ibdev, struct net_device *dev, @@ -2280,6 +2457,17 @@ static int mlx4_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_fw_ver_str(struct ib_device *device, char *str, + size_t str_len) +{ + struct mlx4_ib_dev *dev = + container_of(device, struct mlx4_ib_dev, ib_dev); + snprintf(str, str_len, "%d.%d.%d", + (int) (dev->dev->caps.fw_ver >> 32), + (int) (dev->dev->caps.fw_ver >> 16) & 0xffff, + (int) dev->dev->caps.fw_ver & 0xffff); +} + static void *mlx4_ib_add(struct mlx4_dev *dev) { struct mlx4_ib_dev *ibdev; @@ -2413,6 +2601,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) ibdev->ib_dev.detach_mcast = mlx4_ib_mcg_detach; ibdev->ib_dev.process_mad = mlx4_ib_process_mad; ibdev->ib_dev.get_port_immutable = mlx4_port_immutable; + ibdev->ib_dev.get_dev_fw_str = get_fw_ver_str; ibdev->ib_dev.disassociate_ucontext = mlx4_ib_disassociate_ucontext; if (!mlx4_is_slave(ibdev->dev)) { @@ -2555,9 +2744,12 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) for (j = 1; j <= ibdev->dev->caps.num_ports; j++) atomic64_set(&iboe->mac[j - 1], ibdev->dev->caps.def_mac[j]); - if (ib_register_device(&ibdev->ib_dev, NULL)) + if (mlx4_ib_alloc_diag_counters(ibdev)) goto err_steer_free_bitmap; + if (ib_register_device(&ibdev->ib_dev, NULL)) + goto err_diag_counters; + if (mlx4_ib_mad_init(ibdev)) goto err_reg; @@ -2623,6 +2815,9 @@ err_mad: err_reg: ib_unregister_device(&ibdev->ib_dev); +err_diag_counters: + mlx4_ib_diag_cleanup(ibdev); + err_steer_free_bitmap: kfree(ibdev->ib_uc_qpns_bitmap); @@ -2726,6 +2921,7 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr) mlx4_ib_close_sriov(ibdev); mlx4_ib_mad_cleanup(ibdev); ib_unregister_device(&ibdev->ib_dev); + mlx4_ib_diag_cleanup(ibdev); if (ibdev->iboe.nb.notifier_call) { if (unregister_netdevice_notifier(&ibdev->iboe.nb)) pr_warn("failure unregistering notifier\n"); diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 29acda249612..7c5832ede4bd 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -549,6 +549,14 @@ struct mlx4_ib_counters { u32 default_counter; }; +#define MLX4_DIAG_COUNTERS_TYPES 2 + +struct mlx4_ib_diag_counters { + const char **name; + u32 *offset; + u32 num_counters; +}; + struct mlx4_ib_dev { struct ib_device ib_dev; struct mlx4_dev *dev; @@ -585,6 +593,7 @@ struct mlx4_ib_dev { /* protect resources needed as part of reset flow */ spinlock_t reset_flow_resource_lock; struct list_head qp_list; + struct mlx4_ib_diag_counters diag_counters[MLX4_DIAG_COUNTERS_TYPES]; }; struct ib_event_work { diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c index 9c0e67bd2ba7..308a358e5b46 100644 --- a/drivers/infiniband/hw/mlx5/cq.c +++ b/drivers/infiniband/hw/mlx5/cq.c @@ -424,6 +424,83 @@ static void get_sig_err_item(struct mlx5_sig_err_cqe *cqe, item->key = be32_to_cpu(cqe->mkey); } +static void sw_send_comp(struct mlx5_ib_qp *qp, int num_entries, + struct ib_wc *wc, int *npolled) +{ + struct mlx5_ib_wq *wq; + unsigned int cur; + unsigned int idx; + int np; + int i; + + wq = &qp->sq; + cur = wq->head - wq->tail; + np = *npolled; + + if (cur == 0) + return; + + for (i = 0; i < cur && np < num_entries; i++) { + idx = wq->last_poll & (wq->wqe_cnt - 1); + wc->wr_id = wq->wrid[idx]; + wc->status = IB_WC_WR_FLUSH_ERR; + wc->vendor_err = MLX5_CQE_SYNDROME_WR_FLUSH_ERR; + wq->tail++; + np++; + wc->qp = &qp->ibqp; + wc++; + wq->last_poll = wq->w_list[idx].next; + } + *npolled = np; +} + +static void sw_recv_comp(struct mlx5_ib_qp *qp, int num_entries, + struct ib_wc *wc, int *npolled) +{ + struct mlx5_ib_wq *wq; + unsigned int cur; + int np; + int i; + + wq = &qp->rq; + cur = wq->head - wq->tail; + np = *npolled; + + if (cur == 0) + return; + + for (i = 0; i < cur && np < num_entries; i++) { + wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; + wc->status = IB_WC_WR_FLUSH_ERR; + wc->vendor_err = MLX5_CQE_SYNDROME_WR_FLUSH_ERR; + wq->tail++; + np++; + wc->qp = &qp->ibqp; + wc++; + } + *npolled = np; +} + +static void mlx5_ib_poll_sw_comp(struct mlx5_ib_cq *cq, int num_entries, + struct ib_wc *wc, int *npolled) +{ + struct mlx5_ib_qp *qp; + + *npolled = 0; + /* Find uncompleted WQEs belonging to that cq and retrun mmics ones */ + list_for_each_entry(qp, &cq->list_send_qp, cq_send_list) { + sw_send_comp(qp, num_entries, wc + *npolled, npolled); + if (*npolled >= num_entries) + return; + } + + list_for_each_entry(qp, &cq->list_recv_qp, cq_recv_list) { + sw_recv_comp(qp, num_entries, wc + *npolled, npolled); + if (*npolled >= num_entries) + return; + } +} + static int mlx5_poll_one(struct mlx5_ib_cq *cq, struct mlx5_ib_qp **cur_qp, struct ib_wc *wc) @@ -594,12 +671,18 @@ int mlx5_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc) { struct mlx5_ib_cq *cq = to_mcq(ibcq); struct mlx5_ib_qp *cur_qp = NULL; + struct mlx5_ib_dev *dev = to_mdev(cq->ibcq.device); + struct mlx5_core_dev *mdev = dev->mdev; unsigned long flags; int soft_polled = 0; int npolled; int err = 0; spin_lock_irqsave(&cq->lock, flags); + if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + mlx5_ib_poll_sw_comp(cq, num_entries, wc, &npolled); + goto out; + } if (unlikely(!list_empty(&cq->wc_list))) soft_polled = poll_soft_wc(cq, num_entries, wc); @@ -612,7 +695,7 @@ int mlx5_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc) if (npolled) mlx5_cq_set_ci(&cq->mcq); - +out: spin_unlock_irqrestore(&cq->lock, flags); if (err == 0 || err == -EAGAIN) @@ -843,6 +926,8 @@ struct ib_cq *mlx5_ib_create_cq(struct ib_device *ibdev, cq->resize_buf = NULL; cq->resize_umem = NULL; cq->create_flags = attr->flags; + INIT_LIST_HEAD(&cq->list_send_qp); + INIT_LIST_HEAD(&cq->list_recv_qp); if (context) { err = create_cq_user(dev, udata, context, cq, entries, diff --git a/drivers/infiniband/hw/mlx5/gsi.c b/drivers/infiniband/hw/mlx5/gsi.c index 53e03c8ede79..79e6309460dc 100644 --- a/drivers/infiniband/hw/mlx5/gsi.c +++ b/drivers/infiniband/hw/mlx5/gsi.c @@ -69,15 +69,6 @@ static bool mlx5_ib_deth_sqpn_cap(struct mlx5_ib_dev *dev) return MLX5_CAP_GEN(dev->mdev, set_deth_sqpn); } -static u32 next_outstanding(struct mlx5_ib_gsi_qp *gsi, u32 index) -{ - return ++index % gsi->cap.max_send_wr; -} - -#define for_each_outstanding_wr(gsi, index) \ - for (index = gsi->outstanding_ci; index != gsi->outstanding_pi; \ - index = next_outstanding(gsi, index)) - /* Call with gsi->lock locked */ static void generate_completions(struct mlx5_ib_gsi_qp *gsi) { @@ -85,8 +76,9 @@ static void generate_completions(struct mlx5_ib_gsi_qp *gsi) struct mlx5_ib_gsi_wr *wr; u32 index; - for_each_outstanding_wr(gsi, index) { - wr = &gsi->outstanding_wrs[index]; + for (index = gsi->outstanding_ci; index != gsi->outstanding_pi; + index++) { + wr = &gsi->outstanding_wrs[index % gsi->cap.max_send_wr]; if (!wr->completed) break; @@ -430,8 +422,9 @@ static int mlx5_ib_add_outstanding_wr(struct mlx5_ib_gsi_qp *gsi, return -ENOMEM; } - gsi_wr = &gsi->outstanding_wrs[gsi->outstanding_pi]; - gsi->outstanding_pi = next_outstanding(gsi, gsi->outstanding_pi); + gsi_wr = &gsi->outstanding_wrs[gsi->outstanding_pi % + gsi->cap.max_send_wr]; + gsi->outstanding_pi++; if (!wc) { memset(&gsi_wr->wc, 0, sizeof(gsi_wr->wc)); diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index dad63f038bb8..a84bb766fc62 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -42,11 +42,13 @@ #include <asm/pat.h> #endif #include <linux/sched.h> +#include <linux/delay.h> #include <rdma/ib_user_verbs.h> #include <rdma/ib_addr.h> #include <rdma/ib_cache.h> #include <linux/mlx5/port.h> #include <linux/mlx5/vport.h> +#include <linux/list.h> #include <rdma/ib_smi.h> #include <rdma/ib_umem.h> #include <linux/in.h> @@ -457,8 +459,17 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, int max_rq_sg; int max_sq_sg; u64 min_page_size = 1ull << MLX5_CAP_GEN(mdev, log_pg_sz); + struct mlx5_ib_query_device_resp resp = {}; + size_t resp_len; + u64 max_tso; - if (uhw->inlen || uhw->outlen) + resp_len = sizeof(resp.comp_mask) + sizeof(resp.response_length); + if (uhw->outlen && uhw->outlen < resp_len) + return -EINVAL; + else + resp.response_length = resp_len; + + if (uhw->inlen && !ib_is_udata_cleared(uhw, 0, uhw->inlen)) return -EINVAL; memset(props, 0, sizeof(*props)); @@ -511,10 +522,21 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, if (MLX5_CAP_GEN(mdev, block_lb_mc)) props->device_cap_flags |= IB_DEVICE_BLOCK_MULTICAST_LOOPBACK; - if (MLX5_CAP_GEN(dev->mdev, eth_net_offloads) && - (MLX5_CAP_ETH(dev->mdev, csum_cap))) + if (MLX5_CAP_GEN(dev->mdev, eth_net_offloads)) { + if (MLX5_CAP_ETH(mdev, csum_cap)) props->device_cap_flags |= IB_DEVICE_RAW_IP_CSUM; + if (field_avail(typeof(resp), tso_caps, uhw->outlen)) { + max_tso = MLX5_CAP_ETH(mdev, max_lso_cap); + if (max_tso) { + resp.tso_caps.max_tso = 1 << max_tso; + resp.tso_caps.supported_qpts |= + 1 << IB_QPT_RAW_PACKET; + resp.response_length += sizeof(resp.tso_caps); + } + } + } + if (MLX5_CAP_GEN(mdev, ipoib_basic_offloads)) { props->device_cap_flags |= IB_DEVICE_UD_IP_CSUM; props->device_cap_flags |= IB_DEVICE_UD_TSO; @@ -576,6 +598,13 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, if (!mlx5_core_is_pf(mdev)) props->device_cap_flags |= IB_DEVICE_VIRTUAL_FUNCTION; + if (uhw->outlen) { + err = ib_copy_to_udata(uhw, &resp, resp.response_length); + + if (err) + return err; + } + return 0; } @@ -983,6 +1012,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, goto out_uars; } + INIT_LIST_HEAD(&context->vma_private_list); INIT_LIST_HEAD(&context->db_page_list); mutex_init(&context->db_page_mutex); @@ -992,6 +1022,11 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, if (field_avail(typeof(resp), cqe_version, udata->outlen)) resp.response_length += sizeof(resp.cqe_version); + if (field_avail(typeof(resp), cmds_supp_uhw, udata->outlen)) { + resp.cmds_supp_uhw |= MLX5_USER_CMDS_SUPP_UHW_QUERY_DEVICE; + resp.response_length += sizeof(resp.cmds_supp_uhw); + } + /* * We don't want to expose information from the PCI bar that is located * after 4096 bytes, so if the arch only supports larger pages, let's @@ -1006,8 +1041,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, offsetof(struct mlx5_init_seg, internal_timer_h) % PAGE_SIZE; resp.response_length += sizeof(resp.hca_core_clock_offset) + - sizeof(resp.reserved2) + - sizeof(resp.reserved3); + sizeof(resp.reserved2); } err = ib_copy_to_udata(udata, &resp, resp.response_length); @@ -1086,6 +1120,125 @@ static int get_index(unsigned long offset) return get_arg(offset); } +static void mlx5_ib_vma_open(struct vm_area_struct *area) +{ + /* vma_open is called when a new VMA is created on top of our VMA. This + * is done through either mremap flow or split_vma (usually due to + * mlock, madvise, munmap, etc.) We do not support a clone of the VMA, + * as this VMA is strongly hardware related. Therefore we set the + * vm_ops of the newly created/cloned VMA to NULL, to prevent it from + * calling us again and trying to do incorrect actions. We assume that + * the original VMA size is exactly a single page, and therefore all + * "splitting" operation will not happen to it. + */ + area->vm_ops = NULL; +} + +static void mlx5_ib_vma_close(struct vm_area_struct *area) +{ + struct mlx5_ib_vma_private_data *mlx5_ib_vma_priv_data; + + /* It's guaranteed that all VMAs opened on a FD are closed before the + * file itself is closed, therefore no sync is needed with the regular + * closing flow. (e.g. mlx5 ib_dealloc_ucontext) + * However need a sync with accessing the vma as part of + * mlx5_ib_disassociate_ucontext. + * The close operation is usually called under mm->mmap_sem except when + * process is exiting. + * The exiting case is handled explicitly as part of + * mlx5_ib_disassociate_ucontext. + */ + mlx5_ib_vma_priv_data = (struct mlx5_ib_vma_private_data *)area->vm_private_data; + + /* setting the vma context pointer to null in the mlx5_ib driver's + * private data, to protect a race condition in + * mlx5_ib_disassociate_ucontext(). + */ + mlx5_ib_vma_priv_data->vma = NULL; + list_del(&mlx5_ib_vma_priv_data->list); + kfree(mlx5_ib_vma_priv_data); +} + +static const struct vm_operations_struct mlx5_ib_vm_ops = { + .open = mlx5_ib_vma_open, + .close = mlx5_ib_vma_close +}; + +static int mlx5_ib_set_vma_data(struct vm_area_struct *vma, + struct mlx5_ib_ucontext *ctx) +{ + struct mlx5_ib_vma_private_data *vma_prv; + struct list_head *vma_head = &ctx->vma_private_list; + + vma_prv = kzalloc(sizeof(*vma_prv), GFP_KERNEL); + if (!vma_prv) + return -ENOMEM; + + vma_prv->vma = vma; + vma->vm_private_data = vma_prv; + vma->vm_ops = &mlx5_ib_vm_ops; + + list_add(&vma_prv->list, vma_head); + + return 0; +} + +static void mlx5_ib_disassociate_ucontext(struct ib_ucontext *ibcontext) +{ + int ret; + struct vm_area_struct *vma; + struct mlx5_ib_vma_private_data *vma_private, *n; + struct mlx5_ib_ucontext *context = to_mucontext(ibcontext); + struct task_struct *owning_process = NULL; + struct mm_struct *owning_mm = NULL; + + owning_process = get_pid_task(ibcontext->tgid, PIDTYPE_PID); + if (!owning_process) + return; + + owning_mm = get_task_mm(owning_process); + if (!owning_mm) { + pr_info("no mm, disassociate ucontext is pending task termination\n"); + while (1) { + put_task_struct(owning_process); + usleep_range(1000, 2000); + owning_process = get_pid_task(ibcontext->tgid, + PIDTYPE_PID); + if (!owning_process || + owning_process->state == TASK_DEAD) { + pr_info("disassociate ucontext done, task was terminated\n"); + /* in case task was dead need to release the + * task struct. + */ + if (owning_process) + put_task_struct(owning_process); + return; + } + } + } + + /* need to protect from a race on closing the vma as part of + * mlx5_ib_vma_close. + */ + down_read(&owning_mm->mmap_sem); + list_for_each_entry_safe(vma_private, n, &context->vma_private_list, + list) { + vma = vma_private->vma; + ret = zap_vma_ptes(vma, vma->vm_start, + PAGE_SIZE); + WARN_ONCE(ret, "%s: zap_vma_ptes failed", __func__); + /* context going to be destroyed, should + * not access ops any more. + */ + vma->vm_ops = NULL; + list_del(&vma_private->list); + kfree(vma_private); + } + up_read(&owning_mm->mmap_sem); + mmput(owning_mm); + put_task_struct(owning_process); +} + static inline char *mmap_cmd2str(enum mlx5_ib_mmap_cmd cmd) { switch (cmd) { @@ -1101,8 +1254,10 @@ static inline char *mmap_cmd2str(enum mlx5_ib_mmap_cmd cmd) } static int uar_mmap(struct mlx5_ib_dev *dev, enum mlx5_ib_mmap_cmd cmd, - struct vm_area_struct *vma, struct mlx5_uuar_info *uuari) + struct vm_area_struct *vma, + struct mlx5_ib_ucontext *context) { + struct mlx5_uuar_info *uuari = &context->uuari; int err; unsigned long idx; phys_addr_t pfn, pa; @@ -1152,14 +1307,13 @@ static int uar_mmap(struct mlx5_ib_dev *dev, enum mlx5_ib_mmap_cmd cmd, mlx5_ib_dbg(dev, "mapped %s at 0x%lx, PA %pa\n", mmap_cmd2str(cmd), vma->vm_start, &pa); - return 0; + return mlx5_ib_set_vma_data(vma, context); } static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma) { struct mlx5_ib_ucontext *context = to_mucontext(ibcontext); struct mlx5_ib_dev *dev = to_mdev(ibcontext->device); - struct mlx5_uuar_info *uuari = &context->uuari; unsigned long command; phys_addr_t pfn; @@ -1168,7 +1322,7 @@ static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vm case MLX5_IB_MMAP_WC_PAGE: case MLX5_IB_MMAP_NC_PAGE: case MLX5_IB_MMAP_REGULAR_PAGE: - return uar_mmap(dev, command, vma, uuari); + return uar_mmap(dev, command, vma, context); case MLX5_IB_MMAP_GET_CONTIGUOUS_PAGES: return -ENOSYS; @@ -1331,6 +1485,32 @@ static int parse_flow_attr(u32 *match_c, u32 *match_v, &ib_spec->ipv4.val.dst_ip, sizeof(ib_spec->ipv4.val.dst_ip)); break; + case IB_FLOW_SPEC_IPV6: + if (ib_spec->size != sizeof(ib_spec->ipv6)) + return -EINVAL; + + MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, + ethertype, 0xffff); + MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, + ethertype, ETH_P_IPV6); + + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, + src_ipv4_src_ipv6.ipv6_layout.ipv6), + &ib_spec->ipv6.mask.src_ip, + sizeof(ib_spec->ipv6.mask.src_ip)); + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, + src_ipv4_src_ipv6.ipv6_layout.ipv6), + &ib_spec->ipv6.val.src_ip, + sizeof(ib_spec->ipv6.val.src_ip)); + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c, + dst_ipv4_dst_ipv6.ipv6_layout.ipv6), + &ib_spec->ipv6.mask.dst_ip, + sizeof(ib_spec->ipv6.mask.dst_ip)); + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v, + dst_ipv4_dst_ipv6.ipv6_layout.ipv6), + &ib_spec->ipv6.val.dst_ip, + sizeof(ib_spec->ipv6.val.dst_ip)); + break; case IB_FLOW_SPEC_TCP: if (ib_spec->size != sizeof(ib_spec->tcp_udp)) return -EINVAL; @@ -1801,15 +1981,6 @@ static ssize_t show_hca(struct device *device, struct device_attribute *attr, return sprintf(buf, "MT%d\n", dev->mdev->pdev->device); } -static ssize_t show_fw_ver(struct device *device, struct device_attribute *attr, - char *buf) -{ - struct mlx5_ib_dev *dev = - container_of(device, struct mlx5_ib_dev, ib_dev.dev); - return sprintf(buf, "%d.%d.%04d\n", fw_rev_maj(dev->mdev), - fw_rev_min(dev->mdev), fw_rev_sub(dev->mdev)); -} - static ssize_t show_rev(struct device *device, struct device_attribute *attr, char *buf) { @@ -1828,7 +1999,6 @@ static ssize_t show_board(struct device *device, struct device_attribute *attr, } 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 DEVICE_ATTR(fw_pages, S_IRUGO, show_fw_pages, NULL); @@ -1836,7 +2006,6 @@ static DEVICE_ATTR(reg_pages, S_IRUGO, show_reg_pages, NULL); static struct device_attribute *mlx5_class_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type, &dev_attr_board_id, &dev_attr_fw_pages, @@ -1854,6 +2023,65 @@ static void pkey_change_handler(struct work_struct *work) mutex_unlock(&ports->devr->mutex); } +static void mlx5_ib_handle_internal_error(struct mlx5_ib_dev *ibdev) +{ + struct mlx5_ib_qp *mqp; + struct mlx5_ib_cq *send_mcq, *recv_mcq; + struct mlx5_core_cq *mcq; + struct list_head cq_armed_list; + unsigned long flags_qp; + unsigned long flags_cq; + unsigned long flags; + + INIT_LIST_HEAD(&cq_armed_list); + + /* Go over qp list reside on that ibdev, sync with create/destroy qp.*/ + spin_lock_irqsave(&ibdev->reset_flow_resource_lock, flags); + list_for_each_entry(mqp, &ibdev->qp_list, qps_list) { + spin_lock_irqsave(&mqp->sq.lock, flags_qp); + if (mqp->sq.tail != mqp->sq.head) { + send_mcq = to_mcq(mqp->ibqp.send_cq); + spin_lock_irqsave(&send_mcq->lock, flags_cq); + if (send_mcq->mcq.comp && + mqp->ibqp.send_cq->comp_handler) { + if (!send_mcq->mcq.reset_notify_added) { + send_mcq->mcq.reset_notify_added = 1; + list_add_tail(&send_mcq->mcq.reset_notify, + &cq_armed_list); + } + } + spin_unlock_irqrestore(&send_mcq->lock, flags_cq); + } + spin_unlock_irqrestore(&mqp->sq.lock, flags_qp); + spin_lock_irqsave(&mqp->rq.lock, flags_qp); + /* no handling is needed for SRQ */ + if (!mqp->ibqp.srq) { + if (mqp->rq.tail != mqp->rq.head) { + recv_mcq = to_mcq(mqp->ibqp.recv_cq); + spin_lock_irqsave(&recv_mcq->lock, flags_cq); + if (recv_mcq->mcq.comp && + mqp->ibqp.recv_cq->comp_handler) { + if (!recv_mcq->mcq.reset_notify_added) { + recv_mcq->mcq.reset_notify_added = 1; + list_add_tail(&recv_mcq->mcq.reset_notify, + &cq_armed_list); + } + } + spin_unlock_irqrestore(&recv_mcq->lock, + flags_cq); + } + } + spin_unlock_irqrestore(&mqp->rq.lock, flags_qp); + } + /*At that point all inflight post send were put to be executed as of we + * lock/unlock above locks Now need to arm all involved CQs. + */ + list_for_each_entry(mcq, &cq_armed_list, reset_notify) { + mcq->comp(mcq); + } + spin_unlock_irqrestore(&ibdev->reset_flow_resource_lock, flags); +} + static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context, enum mlx5_dev_event event, unsigned long param) { @@ -1866,6 +2094,7 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context, case MLX5_DEV_EVENT_SYS_ERROR: ibdev->ib_active = false; ibev.event = IB_EVENT_DEVICE_FATAL; + mlx5_ib_handle_internal_error(ibdev); break; case MLX5_DEV_EVENT_PORT_UP: @@ -2272,6 +2501,15 @@ static int mlx5_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_dev_fw_str(struct ib_device *ibdev, char *str, + size_t str_len) +{ + struct mlx5_ib_dev *dev = + container_of(ibdev, struct mlx5_ib_dev, ib_dev); + snprintf(str, str_len, "%d.%d.%04d", fw_rev_maj(dev->mdev), + fw_rev_min(dev->mdev), fw_rev_sub(dev->mdev)); +} + static int mlx5_enable_roce(struct mlx5_ib_dev *dev) { int err; @@ -2298,6 +2536,113 @@ static void mlx5_disable_roce(struct mlx5_ib_dev *dev) unregister_netdevice_notifier(&dev->roce.nb); } +static void mlx5_ib_dealloc_q_counters(struct mlx5_ib_dev *dev) +{ + unsigned int i; + + for (i = 0; i < dev->num_ports; i++) + mlx5_core_dealloc_q_counter(dev->mdev, + dev->port[i].q_cnt_id); +} + +static int mlx5_ib_alloc_q_counters(struct mlx5_ib_dev *dev) +{ + int i; + int ret; + + for (i = 0; i < dev->num_ports; i++) { + ret = mlx5_core_alloc_q_counter(dev->mdev, + &dev->port[i].q_cnt_id); + if (ret) { + mlx5_ib_warn(dev, + "couldn't allocate queue counter for port %d, err %d\n", + i + 1, ret); + goto dealloc_counters; + } + } + + return 0; + +dealloc_counters: + while (--i >= 0) + mlx5_core_dealloc_q_counter(dev->mdev, + dev->port[i].q_cnt_id); + + return ret; +} + +static const char * const names[] = { + "rx_write_requests", + "rx_read_requests", + "rx_atomic_requests", + "out_of_buffer", + "out_of_sequence", + "duplicate_request", + "rnr_nak_retry_err", + "packet_seq_err", + "implied_nak_seq_err", + "local_ack_timeout_err", +}; + +static const size_t stats_offsets[] = { + MLX5_BYTE_OFF(query_q_counter_out, rx_write_requests), + MLX5_BYTE_OFF(query_q_counter_out, rx_read_requests), + MLX5_BYTE_OFF(query_q_counter_out, rx_atomic_requests), + MLX5_BYTE_OFF(query_q_counter_out, out_of_buffer), + MLX5_BYTE_OFF(query_q_counter_out, out_of_sequence), + MLX5_BYTE_OFF(query_q_counter_out, duplicate_request), + MLX5_BYTE_OFF(query_q_counter_out, rnr_nak_retry_err), + MLX5_BYTE_OFF(query_q_counter_out, packet_seq_err), + MLX5_BYTE_OFF(query_q_counter_out, implied_nak_seq_err), + MLX5_BYTE_OFF(query_q_counter_out, local_ack_timeout_err), +}; + +static struct rdma_hw_stats *mlx5_ib_alloc_hw_stats(struct ib_device *ibdev, + u8 port_num) +{ + BUILD_BUG_ON(ARRAY_SIZE(names) != ARRAY_SIZE(stats_offsets)); + + /* We support only per port stats */ + if (port_num == 0) + return NULL; + + return rdma_alloc_hw_stats_struct(names, ARRAY_SIZE(names), + RDMA_HW_STATS_DEFAULT_LIFESPAN); +} + +static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, + struct rdma_hw_stats *stats, + u8 port, int index) +{ + struct mlx5_ib_dev *dev = to_mdev(ibdev); + int outlen = MLX5_ST_SZ_BYTES(query_q_counter_out); + void *out; + __be32 val; + int ret; + int i; + + if (!port || !stats) + return -ENOSYS; + + out = mlx5_vzalloc(outlen); + if (!out) + return -ENOMEM; + + ret = mlx5_core_query_q_counter(dev->mdev, + dev->port[port - 1].q_cnt_id, 0, + out, outlen); + if (ret) + goto free; + + for (i = 0; i < ARRAY_SIZE(names); i++) { + val = *(__be32 *)(out + stats_offsets[i]); + stats->value[i] = (u64)be32_to_cpu(val); + } +free: + kvfree(out); + return ARRAY_SIZE(names); +} + static void *mlx5_ib_add(struct mlx5_core_dev *mdev) { struct mlx5_ib_dev *dev; @@ -2320,10 +2665,15 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) dev->mdev = mdev; + dev->port = kcalloc(MLX5_CAP_GEN(mdev, num_ports), sizeof(*dev->port), + GFP_KERNEL); + if (!dev->port) + goto err_dealloc; + rwlock_init(&dev->roce.netdev_lock); err = get_port_caps(dev); if (err) - goto err_dealloc; + goto err_free_port; if (mlx5_use_mad_ifc(dev)) get_ext_port_caps(dev); @@ -2418,6 +2768,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) dev->ib_dev.map_mr_sg = mlx5_ib_map_mr_sg; dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status; dev->ib_dev.get_port_immutable = mlx5_port_immutable; + dev->ib_dev.get_dev_fw_str = get_dev_fw_str; if (mlx5_core_is_pf(mdev)) { dev->ib_dev.get_vf_config = mlx5_ib_get_vf_config; dev->ib_dev.set_vf_link_state = mlx5_ib_set_vf_link_state; @@ -2425,6 +2776,8 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) dev->ib_dev.set_vf_guid = mlx5_ib_set_vf_guid; } + dev->ib_dev.disassociate_ucontext = mlx5_ib_disassociate_ucontext; + mlx5_ib_internal_fill_odp_caps(dev); if (MLX5_CAP_GEN(mdev, imaicl)) { @@ -2435,6 +2788,12 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) (1ull << IB_USER_VERBS_CMD_DEALLOC_MW); } + if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt) && + MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) { + dev->ib_dev.get_hw_stats = mlx5_ib_get_hw_stats; + dev->ib_dev.alloc_hw_stats = mlx5_ib_alloc_hw_stats; + } + if (MLX5_CAP_GEN(mdev, xrc)) { dev->ib_dev.alloc_xrcd = mlx5_ib_alloc_xrcd; dev->ib_dev.dealloc_xrcd = mlx5_ib_dealloc_xrcd; @@ -2447,9 +2806,19 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) IB_LINK_LAYER_ETHERNET) { dev->ib_dev.create_flow = mlx5_ib_create_flow; dev->ib_dev.destroy_flow = mlx5_ib_destroy_flow; + dev->ib_dev.create_wq = mlx5_ib_create_wq; + dev->ib_dev.modify_wq = mlx5_ib_modify_wq; + dev->ib_dev.destroy_wq = mlx5_ib_destroy_wq; + dev->ib_dev.create_rwq_ind_table = mlx5_ib_create_rwq_ind_table; + dev->ib_dev.destroy_rwq_ind_table = mlx5_ib_destroy_rwq_ind_table; dev->ib_dev.uverbs_ex_cmd_mask |= (1ull << IB_USER_VERBS_EX_CMD_CREATE_FLOW) | - (1ull << IB_USER_VERBS_EX_CMD_DESTROY_FLOW); + (1ull << IB_USER_VERBS_EX_CMD_DESTROY_FLOW) | + (1ull << IB_USER_VERBS_EX_CMD_CREATE_WQ) | + (1ull << IB_USER_VERBS_EX_CMD_MODIFY_WQ) | + (1ull << IB_USER_VERBS_EX_CMD_DESTROY_WQ) | + (1ull << IB_USER_VERBS_EX_CMD_CREATE_RWQ_IND_TBL) | + (1ull << IB_USER_VERBS_EX_CMD_DESTROY_RWQ_IND_TBL); } err = init_node_data(dev); if (err) @@ -2457,6 +2826,8 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) mutex_init(&dev->flow_db.lock); mutex_init(&dev->cap_mask_mutex); + INIT_LIST_HEAD(&dev->qp_list); + spin_lock_init(&dev->reset_flow_resource_lock); if (ll == IB_LINK_LAYER_ETHERNET) { err = mlx5_enable_roce(dev); @@ -2472,10 +2843,14 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) if (err) goto err_rsrc; - err = ib_register_device(&dev->ib_dev, NULL); + err = mlx5_ib_alloc_q_counters(dev); if (err) goto err_odp; + err = ib_register_device(&dev->ib_dev, NULL); + if (err) + goto err_q_cnt; + err = create_umr_res(dev); if (err) goto err_dev; @@ -2497,6 +2872,9 @@ err_umrc: err_dev: ib_unregister_device(&dev->ib_dev); +err_q_cnt: + mlx5_ib_dealloc_q_counters(dev); + err_odp: mlx5_ib_odp_remove_one(dev); @@ -2507,6 +2885,9 @@ err_disable_roce: if (ll == IB_LINK_LAYER_ETHERNET) mlx5_disable_roce(dev); +err_free_port: + kfree(dev->port); + err_dealloc: ib_dealloc_device((struct ib_device *)dev); @@ -2519,11 +2900,13 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context) enum rdma_link_layer ll = mlx5_ib_port_link_layer(&dev->ib_dev, 1); ib_unregister_device(&dev->ib_dev); + mlx5_ib_dealloc_q_counters(dev); destroy_umrc_res(dev); mlx5_ib_odp_remove_one(dev); destroy_dev_resources(&dev->devr); if (ll == IB_LINK_LAYER_ETHERNET) mlx5_disable_roce(dev); + kfree(dev->port); ib_dealloc_device(&dev->ib_dev); } diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index c4a9825828bc..372385d0f993 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -105,6 +105,11 @@ enum { MLX5_CQE_VERSION_V1, }; +struct mlx5_ib_vma_private_data { + struct list_head list; + struct vm_area_struct *vma; +}; + struct mlx5_ib_ucontext { struct ib_ucontext ibucontext; struct list_head db_page_list; @@ -116,6 +121,7 @@ struct mlx5_ib_ucontext { u8 cqe_version; /* Transport Domain number */ u32 tdn; + struct list_head vma_private_list; }; static inline struct mlx5_ib_ucontext *to_mucontext(struct ib_ucontext *ibucontext) @@ -217,12 +223,41 @@ struct mlx5_ib_wq { void *qend; }; +struct mlx5_ib_rwq { + struct ib_wq ibwq; + u32 rqn; + u32 rq_num_pas; + u32 log_rq_stride; + u32 log_rq_size; + u32 rq_page_offset; + u32 log_page_size; + struct ib_umem *umem; + size_t buf_size; + unsigned int page_shift; + int create_type; + struct mlx5_db db; + u32 user_index; + u32 wqe_count; + u32 wqe_shift; + int wq_sig; +}; + enum { MLX5_QP_USER, MLX5_QP_KERNEL, MLX5_QP_EMPTY }; +enum { + MLX5_WQ_USER, + MLX5_WQ_KERNEL +}; + +struct mlx5_ib_rwq_ind_table { + struct ib_rwq_ind_table ib_rwq_ind_tbl; + u32 rqtn; +}; + /* * Connect-IB can trigger up to four concurrent pagefaults * per-QP. @@ -266,6 +301,10 @@ struct mlx5_ib_qp_trans { u8 resp_depth; }; +struct mlx5_ib_rss_qp { + u32 tirn; +}; + struct mlx5_ib_rq { struct mlx5_ib_qp_base base; struct mlx5_ib_wq *rq; @@ -294,6 +333,7 @@ struct mlx5_ib_qp { union { struct mlx5_ib_qp_trans trans_qp; struct mlx5_ib_raw_packet_qp raw_packet_qp; + struct mlx5_ib_rss_qp rss_qp; }; struct mlx5_buf buf; @@ -340,6 +380,9 @@ struct mlx5_ib_qp { spinlock_t disable_page_faults_lock; struct mlx5_ib_pfault pagefaults[MLX5_IB_PAGEFAULT_CONTEXTS]; #endif + struct list_head qps_list; + struct list_head cq_recv_list; + struct list_head cq_send_list; }; struct mlx5_ib_cq_buf { @@ -401,6 +444,8 @@ struct mlx5_ib_cq { struct mlx5_ib_cq_buf *resize_buf; struct ib_umem *resize_umem; int cqe_size; + struct list_head list_send_qp; + struct list_head list_recv_qp; u32 create_flags; struct list_head wc_list; enum ib_cq_notify_flags notify_flags; @@ -546,6 +591,10 @@ struct mlx5_ib_resources { struct mutex mutex; }; +struct mlx5_ib_port { + u16 q_cnt_id; +}; + struct mlx5_roce { /* Protect mlx5_ib_get_netdev from invoking dev_hold() with a NULL * netdev pointer @@ -581,6 +630,11 @@ struct mlx5_ib_dev { struct srcu_struct mr_srcu; #endif struct mlx5_ib_flow_db flow_db; + /* protect resources needed as part of reset flow */ + spinlock_t reset_flow_resource_lock; + struct list_head qp_list; + /* Array with num_ports elements */ + struct mlx5_ib_port *port; }; static inline struct mlx5_ib_cq *to_mibcq(struct mlx5_core_cq *mcq) @@ -628,6 +682,16 @@ static inline struct mlx5_ib_qp *to_mqp(struct ib_qp *ibqp) return container_of(ibqp, struct mlx5_ib_qp, ibqp); } +static inline struct mlx5_ib_rwq *to_mrwq(struct ib_wq *ibwq) +{ + return container_of(ibwq, struct mlx5_ib_rwq, ibwq); +} + +static inline struct mlx5_ib_rwq_ind_table *to_mrwq_ind_table(struct ib_rwq_ind_table *ib_rwq_ind_tbl) +{ + return container_of(ib_rwq_ind_tbl, struct mlx5_ib_rwq_ind_table, ib_rwq_ind_tbl); +} + static inline struct mlx5_ib_srq *to_mibsrq(struct mlx5_core_srq *msrq) { return container_of(msrq, struct mlx5_ib_srq, msrq); @@ -762,6 +826,16 @@ int mlx5_mr_cache_cleanup(struct mlx5_ib_dev *dev); int mlx5_mr_ib_cont_pages(struct ib_umem *umem, u64 addr, int *count, int *shift); int mlx5_ib_check_mr_status(struct ib_mr *ibmr, u32 check_mask, struct ib_mr_status *mr_status); +struct ib_wq *mlx5_ib_create_wq(struct ib_pd *pd, + struct ib_wq_init_attr *init_attr, + struct ib_udata *udata); +int mlx5_ib_destroy_wq(struct ib_wq *wq); +int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr, + u32 wq_attr_mask, struct ib_udata *udata); +struct ib_rwq_ind_table *mlx5_ib_create_rwq_ind_table(struct ib_device *device, + struct ib_rwq_ind_table_init_attr *init_attr, + struct ib_udata *udata); +int mlx5_ib_destroy_rwq_ind_table(struct ib_rwq_ind_table *wq_ind_table); #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING extern struct workqueue_struct *mlx5_ib_page_fault_wq; diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index 8cf2ce50511f..4b021305c321 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -1193,12 +1193,16 @@ error: static int unreg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) { + struct mlx5_core_dev *mdev = dev->mdev; struct umr_common *umrc = &dev->umrc; struct mlx5_ib_umr_context umr_context; struct mlx5_umr_wr umrwr = {}; struct ib_send_wr *bad; int err; + if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) + return 0; + mlx5_ib_init_umr_context(&umr_context); umrwr.wr.wr_cqe = &umr_context.cqe; diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index ce0a7ab35a22..0dd7d93cac95 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -77,6 +77,10 @@ struct mlx5_wqe_eth_pad { u8 rsvd0[16]; }; +static void get_cqs(enum ib_qp_type qp_type, + struct ib_cq *ib_send_cq, struct ib_cq *ib_recv_cq, + struct mlx5_ib_cq **send_cq, struct mlx5_ib_cq **recv_cq); + static int is_qp0(enum ib_qp_type qp_type) { return qp_type == IB_QPT_SMI; @@ -609,6 +613,11 @@ static int to_mlx5_st(enum ib_qp_type type) } } +static void mlx5_ib_lock_cqs(struct mlx5_ib_cq *send_cq, + struct mlx5_ib_cq *recv_cq); +static void mlx5_ib_unlock_cqs(struct mlx5_ib_cq *send_cq, + struct mlx5_ib_cq *recv_cq); + static int uuarn_to_uar_index(struct mlx5_uuar_info *uuari, int uuarn) { return uuari->uars[uuarn / MLX5_BF_REGS_PER_PAGE].index; @@ -649,6 +658,71 @@ err_umem: return err; } +static void destroy_user_rq(struct ib_pd *pd, struct mlx5_ib_rwq *rwq) +{ + struct mlx5_ib_ucontext *context; + + context = to_mucontext(pd->uobject->context); + mlx5_ib_db_unmap_user(context, &rwq->db); + if (rwq->umem) + ib_umem_release(rwq->umem); +} + +static int create_user_rq(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct mlx5_ib_rwq *rwq, + struct mlx5_ib_create_wq *ucmd) +{ + struct mlx5_ib_ucontext *context; + int page_shift = 0; + int npages; + u32 offset = 0; + int ncont = 0; + int err; + + if (!ucmd->buf_addr) + return -EINVAL; + + context = to_mucontext(pd->uobject->context); + rwq->umem = ib_umem_get(pd->uobject->context, ucmd->buf_addr, + rwq->buf_size, 0, 0); + if (IS_ERR(rwq->umem)) { + mlx5_ib_dbg(dev, "umem_get failed\n"); + err = PTR_ERR(rwq->umem); + return err; + } + + mlx5_ib_cont_pages(rwq->umem, ucmd->buf_addr, &npages, &page_shift, + &ncont, NULL); + err = mlx5_ib_get_buf_offset(ucmd->buf_addr, page_shift, + &rwq->rq_page_offset); + if (err) { + mlx5_ib_warn(dev, "bad offset\n"); + goto err_umem; + } + + rwq->rq_num_pas = ncont; + rwq->page_shift = page_shift; + rwq->log_page_size = page_shift - MLX5_ADAPTER_PAGE_SHIFT; + rwq->wq_sig = !!(ucmd->flags & MLX5_WQ_FLAG_SIGNATURE); + + mlx5_ib_dbg(dev, "addr 0x%llx, size %zd, npages %d, page_shift %d, ncont %d, offset %d\n", + (unsigned long long)ucmd->buf_addr, rwq->buf_size, + npages, page_shift, ncont, offset); + + err = mlx5_ib_db_map_user(context, ucmd->db_addr, &rwq->db); + if (err) { + mlx5_ib_dbg(dev, "map failed\n"); + goto err_umem; + } + + rwq->create_type = MLX5_WQ_USER; + return 0; + +err_umem: + ib_umem_release(rwq->umem); + return err; +} + static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_qp *qp, struct ib_udata *udata, struct ib_qp_init_attr *attr, @@ -1201,6 +1275,187 @@ static void raw_packet_qp_copy_info(struct mlx5_ib_qp *qp, rq->doorbell = &qp->db; } +static void destroy_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) +{ + mlx5_core_destroy_tir(dev->mdev, qp->rss_qp.tirn); +} + +static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + struct ib_pd *pd, + struct ib_qp_init_attr *init_attr, + struct ib_udata *udata) +{ + struct ib_uobject *uobj = pd->uobject; + struct ib_ucontext *ucontext = uobj->context; + struct mlx5_ib_ucontext *mucontext = to_mucontext(ucontext); + struct mlx5_ib_create_qp_resp resp = {}; + int inlen; + int err; + u32 *in; + void *tirc; + void *hfso; + u32 selected_fields = 0; + size_t min_resp_len; + u32 tdn = mucontext->tdn; + struct mlx5_ib_create_qp_rss ucmd = {}; + size_t required_cmd_sz; + + if (init_attr->qp_type != IB_QPT_RAW_PACKET) + return -EOPNOTSUPP; + + if (init_attr->create_flags || init_attr->send_cq) + return -EINVAL; + + min_resp_len = offsetof(typeof(resp), uuar_index) + sizeof(resp.uuar_index); + if (udata->outlen < min_resp_len) + return -EINVAL; + + required_cmd_sz = offsetof(typeof(ucmd), reserved1) + sizeof(ucmd.reserved1); + if (udata->inlen < required_cmd_sz) { + mlx5_ib_dbg(dev, "invalid inlen\n"); + return -EINVAL; + } + + if (udata->inlen > sizeof(ucmd) && + !ib_is_udata_cleared(udata, sizeof(ucmd), + udata->inlen - sizeof(ucmd))) { + mlx5_ib_dbg(dev, "inlen is not supported\n"); + return -EOPNOTSUPP; + } + + if (ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen))) { + mlx5_ib_dbg(dev, "copy failed\n"); + return -EFAULT; + } + + if (ucmd.comp_mask) { + mlx5_ib_dbg(dev, "invalid comp mask\n"); + return -EOPNOTSUPP; + } + + if (memchr_inv(ucmd.reserved, 0, sizeof(ucmd.reserved)) || ucmd.reserved1) { + mlx5_ib_dbg(dev, "invalid reserved\n"); + return -EOPNOTSUPP; + } + + err = ib_copy_to_udata(udata, &resp, min_resp_len); + if (err) { + mlx5_ib_dbg(dev, "copy failed\n"); + return -EINVAL; + } + + inlen = MLX5_ST_SZ_BYTES(create_tir_in); + in = mlx5_vzalloc(inlen); + if (!in) + return -ENOMEM; + + tirc = MLX5_ADDR_OF(create_tir_in, in, ctx); + MLX5_SET(tirc, tirc, disp_type, + MLX5_TIRC_DISP_TYPE_INDIRECT); + MLX5_SET(tirc, tirc, indirect_table, + init_attr->rwq_ind_tbl->ind_tbl_num); + MLX5_SET(tirc, tirc, transport_domain, tdn); + + hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_outer); + switch (ucmd.rx_hash_function) { + case MLX5_RX_HASH_FUNC_TOEPLITZ: + { + void *rss_key = MLX5_ADDR_OF(tirc, tirc, rx_hash_toeplitz_key); + size_t len = MLX5_FLD_SZ_BYTES(tirc, rx_hash_toeplitz_key); + + if (len != ucmd.rx_key_len) { + err = -EINVAL; + goto err; + } + + MLX5_SET(tirc, tirc, rx_hash_fn, MLX5_RX_HASH_FN_TOEPLITZ); + MLX5_SET(tirc, tirc, rx_hash_symmetric, 1); + memcpy(rss_key, ucmd.rx_hash_key, len); + break; + } + default: + err = -EOPNOTSUPP; + goto err; + } + + if (!ucmd.rx_hash_fields_mask) { + /* special case when this TIR serves as steering entry without hashing */ + if (!init_attr->rwq_ind_tbl->log_ind_tbl_size) + goto create_tir; + err = -EINVAL; + goto err; + } + + if (((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV4) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV4)) && + ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV6) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV6))) { + err = -EINVAL; + goto err; + } + + /* If none of IPV4 & IPV6 SRC/DST was set - this bit field is ignored */ + if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV4) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV4)) + MLX5_SET(rx_hash_field_select, hfso, l3_prot_type, + MLX5_L3_PROT_TYPE_IPV4); + else if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV6) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV6)) + MLX5_SET(rx_hash_field_select, hfso, l3_prot_type, + MLX5_L3_PROT_TYPE_IPV6); + + if (((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) && + ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP))) { + err = -EINVAL; + goto err; + } + + /* If none of TCP & UDP SRC/DST was set - this bit field is ignored */ + if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) + MLX5_SET(rx_hash_field_select, hfso, l4_prot_type, + MLX5_L4_PROT_TYPE_TCP); + else if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) + MLX5_SET(rx_hash_field_select, hfso, l4_prot_type, + MLX5_L4_PROT_TYPE_UDP); + + if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV4) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_IPV6)) + selected_fields |= MLX5_HASH_FIELD_SEL_SRC_IP; + + if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV4) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_IPV6)) + selected_fields |= MLX5_HASH_FIELD_SEL_DST_IP; + + if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP)) + selected_fields |= MLX5_HASH_FIELD_SEL_L4_SPORT; + + if ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP) || + (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) + selected_fields |= MLX5_HASH_FIELD_SEL_L4_DPORT; + + MLX5_SET(rx_hash_field_select, hfso, selected_fields, selected_fields); + +create_tir: + err = mlx5_core_create_tir(dev->mdev, in, inlen, &qp->rss_qp.tirn); + + if (err) + goto err; + + kvfree(in); + /* qpn is reserved for that QP */ + qp->trans_qp.base.mqp.qpn = 0; + return 0; + +err: + kvfree(in); + return err; +} + static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata, struct mlx5_ib_qp *qp) @@ -1211,6 +1466,9 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_create_qp_resp resp; struct mlx5_create_qp_mbox_in *in; struct mlx5_ib_create_qp ucmd; + struct mlx5_ib_cq *send_cq; + struct mlx5_ib_cq *recv_cq; + unsigned long flags; int inlen = sizeof(*in); int err; u32 uidx = MLX5_IB_DEFAULT_UIDX; @@ -1227,6 +1485,14 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, spin_lock_init(&qp->sq.lock); spin_lock_init(&qp->rq.lock); + if (init_attr->rwq_ind_tbl) { + if (!udata) + return -ENOSYS; + + err = create_rss_raw_qp_tir(dev, qp, pd, init_attr, udata); + return err; + } + if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) { if (!MLX5_CAP_GEN(mdev, block_lb_mc)) { mlx5_ib_dbg(dev, "block multicast loopback isn't supported\n"); @@ -1460,6 +1726,23 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, base->container_mibqp = qp; base->mqp.event = mlx5_ib_qp_event; + get_cqs(init_attr->qp_type, init_attr->send_cq, init_attr->recv_cq, + &send_cq, &recv_cq); + spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); + mlx5_ib_lock_cqs(send_cq, recv_cq); + /* Maintain device to QPs access, needed for further handling via reset + * flow + */ + list_add_tail(&qp->qps_list, &dev->qp_list); + /* Maintain CQ to QPs access, needed for further handling via reset flow + */ + if (send_cq) + list_add_tail(&qp->cq_send_list, &send_cq->list_send_qp); + if (recv_cq) + list_add_tail(&qp->cq_recv_list, &recv_cq->list_recv_qp); + mlx5_ib_unlock_cqs(send_cq, recv_cq); + spin_unlock_irqrestore(&dev->reset_flow_resource_lock, flags); + return 0; err_create: @@ -1478,23 +1761,23 @@ static void mlx5_ib_lock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *recv if (send_cq) { if (recv_cq) { if (send_cq->mcq.cqn < recv_cq->mcq.cqn) { - spin_lock_irq(&send_cq->lock); + spin_lock(&send_cq->lock); spin_lock_nested(&recv_cq->lock, SINGLE_DEPTH_NESTING); } else if (send_cq->mcq.cqn == recv_cq->mcq.cqn) { - spin_lock_irq(&send_cq->lock); + spin_lock(&send_cq->lock); __acquire(&recv_cq->lock); } else { - spin_lock_irq(&recv_cq->lock); + spin_lock(&recv_cq->lock); spin_lock_nested(&send_cq->lock, SINGLE_DEPTH_NESTING); } } else { - spin_lock_irq(&send_cq->lock); + spin_lock(&send_cq->lock); __acquire(&recv_cq->lock); } } else if (recv_cq) { - spin_lock_irq(&recv_cq->lock); + spin_lock(&recv_cq->lock); __acquire(&send_cq->lock); } else { __acquire(&send_cq->lock); @@ -1509,21 +1792,21 @@ static void mlx5_ib_unlock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *re if (recv_cq) { if (send_cq->mcq.cqn < recv_cq->mcq.cqn) { spin_unlock(&recv_cq->lock); - spin_unlock_irq(&send_cq->lock); + spin_unlock(&send_cq->lock); } else if (send_cq->mcq.cqn == recv_cq->mcq.cqn) { __release(&recv_cq->lock); - spin_unlock_irq(&send_cq->lock); + spin_unlock(&send_cq->lock); } else { spin_unlock(&send_cq->lock); - spin_unlock_irq(&recv_cq->lock); + spin_unlock(&recv_cq->lock); } } else { __release(&recv_cq->lock); - spin_unlock_irq(&send_cq->lock); + spin_unlock(&send_cq->lock); } } else if (recv_cq) { __release(&send_cq->lock); - spin_unlock_irq(&recv_cq->lock); + spin_unlock(&recv_cq->lock); } else { __release(&recv_cq->lock); __release(&send_cq->lock); @@ -1535,17 +1818,18 @@ static struct mlx5_ib_pd *get_pd(struct mlx5_ib_qp *qp) return to_mpd(qp->ibqp.pd); } -static void get_cqs(struct mlx5_ib_qp *qp, +static void get_cqs(enum ib_qp_type qp_type, + struct ib_cq *ib_send_cq, struct ib_cq *ib_recv_cq, struct mlx5_ib_cq **send_cq, struct mlx5_ib_cq **recv_cq) { - switch (qp->ibqp.qp_type) { + switch (qp_type) { case IB_QPT_XRC_TGT: *send_cq = NULL; *recv_cq = NULL; break; case MLX5_IB_QPT_REG_UMR: case IB_QPT_XRC_INI: - *send_cq = to_mcq(qp->ibqp.send_cq); + *send_cq = ib_send_cq ? to_mcq(ib_send_cq) : NULL; *recv_cq = NULL; break; @@ -1557,8 +1841,8 @@ static void get_cqs(struct mlx5_ib_qp *qp, case IB_QPT_RAW_IPV6: case IB_QPT_RAW_ETHERTYPE: case IB_QPT_RAW_PACKET: - *send_cq = to_mcq(qp->ibqp.send_cq); - *recv_cq = to_mcq(qp->ibqp.recv_cq); + *send_cq = ib_send_cq ? to_mcq(ib_send_cq) : NULL; + *recv_cq = ib_recv_cq ? to_mcq(ib_recv_cq) : NULL; break; case IB_QPT_MAX: @@ -1577,8 +1861,14 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) struct mlx5_ib_cq *send_cq, *recv_cq; struct mlx5_ib_qp_base *base = &qp->trans_qp.base; struct mlx5_modify_qp_mbox_in *in; + unsigned long flags; int err; + if (qp->ibqp.rwq_ind_tbl) { + destroy_rss_raw_qp_tir(dev, qp); + return; + } + base = qp->ibqp.qp_type == IB_QPT_RAW_PACKET ? &qp->raw_packet_qp.rq.base : &qp->trans_qp.base; @@ -1602,17 +1892,28 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) base->mqp.qpn); } - get_cqs(qp, &send_cq, &recv_cq); + get_cqs(qp->ibqp.qp_type, qp->ibqp.send_cq, qp->ibqp.recv_cq, + &send_cq, &recv_cq); + + spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); + mlx5_ib_lock_cqs(send_cq, recv_cq); + /* del from lists under both locks above to protect reset flow paths */ + list_del(&qp->qps_list); + if (send_cq) + list_del(&qp->cq_send_list); + + if (recv_cq) + list_del(&qp->cq_recv_list); if (qp->create_type == MLX5_QP_KERNEL) { - mlx5_ib_lock_cqs(send_cq, recv_cq); __mlx5_ib_cq_clean(recv_cq, base->mqp.qpn, qp->ibqp.srq ? to_msrq(qp->ibqp.srq) : NULL); if (send_cq != recv_cq) __mlx5_ib_cq_clean(send_cq, base->mqp.qpn, NULL); - mlx5_ib_unlock_cqs(send_cq, recv_cq); } + mlx5_ib_unlock_cqs(send_cq, recv_cq); + spin_unlock_irqrestore(&dev->reset_flow_resource_lock, flags); if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET) { destroy_raw_packet_qp(dev, qp); @@ -2300,7 +2601,8 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, } pd = get_pd(qp); - get_cqs(qp, &send_cq, &recv_cq); + get_cqs(qp->ibqp.qp_type, qp->ibqp.send_cq, qp->ibqp.recv_cq, + &send_cq, &recv_cq); context->flags_pd = cpu_to_be32(pd ? pd->pdn : to_mpd(dev->devr.p0)->pdn); context->cqn_send = send_cq ? cpu_to_be32(send_cq->mcq.cqn) : 0; @@ -2349,6 +2651,15 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, else sqd_event = 0; + if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { + u8 port_num = (attr_mask & IB_QP_PORT ? attr->port_num : + qp->port) - 1; + struct mlx5_ib_port *mibport = &dev->port[port_num]; + + context->qp_counter_set_usr_page |= + cpu_to_be32((u32)(mibport->q_cnt_id) << 24); + } + if (!ibqp->uobject && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) context->sq_crq_size |= cpu_to_be16(1 << 4); @@ -2439,6 +2750,9 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int port; enum rdma_link_layer ll = IB_LINK_LAYER_UNSPECIFIED; + if (ibqp->rwq_ind_tbl) + return -ENOSYS; + if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_modify_qp(ibqp, attr, attr_mask); @@ -3397,6 +3711,7 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, { struct mlx5_wqe_ctrl_seg *ctrl = NULL; /* compiler warning */ struct mlx5_ib_dev *dev = to_mdev(ibqp->device); + struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_qp *qp; struct mlx5_ib_mr *mr; struct mlx5_wqe_data_seg *dpseg; @@ -3424,6 +3739,13 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, spin_lock_irqsave(&qp->sq.lock, flags); + if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + err = -EIO; + *bad_wr = wr; + nreq = 0; + goto out; + } + for (nreq = 0; wr; nreq++, wr = wr->next) { if (unlikely(wr->opcode >= ARRAY_SIZE(mlx5_ib_opcode))) { mlx5_ib_warn(dev, "\n"); @@ -3725,6 +4047,8 @@ int mlx5_ib_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr, struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_wqe_data_seg *scat; struct mlx5_rwqe_sig *sig; + struct mlx5_ib_dev *dev = to_mdev(ibqp->device); + struct mlx5_core_dev *mdev = dev->mdev; unsigned long flags; int err = 0; int nreq; @@ -3736,6 +4060,13 @@ int mlx5_ib_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr, spin_lock_irqsave(&qp->rq.lock, flags); + if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + err = -EIO; + *bad_wr = wr; + nreq = 0; + goto out; + } + ind = qp->rq.head & (qp->rq.wqe_cnt - 1); for (nreq = 0; wr; nreq++, wr = wr->next) { @@ -4055,6 +4386,9 @@ int mlx5_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int err = 0; u8 raw_packet_qp_state; + if (ibqp->rwq_ind_tbl) + return -ENOSYS; + if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_query_qp(ibqp, qp_attr, qp_attr_mask, qp_init_attr); @@ -4164,3 +4498,322 @@ int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd) return 0; } + +static int create_rq(struct mlx5_ib_rwq *rwq, struct ib_pd *pd, + struct ib_wq_init_attr *init_attr) +{ + struct mlx5_ib_dev *dev; + __be64 *rq_pas0; + void *in; + void *rqc; + void *wq; + int inlen; + int err; + + dev = to_mdev(pd->device); + + inlen = MLX5_ST_SZ_BYTES(create_rq_in) + sizeof(u64) * rwq->rq_num_pas; + in = mlx5_vzalloc(inlen); + if (!in) + return -ENOMEM; + + rqc = MLX5_ADDR_OF(create_rq_in, in, ctx); + MLX5_SET(rqc, rqc, mem_rq_type, + MLX5_RQC_MEM_RQ_TYPE_MEMORY_RQ_INLINE); + MLX5_SET(rqc, rqc, user_index, rwq->user_index); + MLX5_SET(rqc, rqc, cqn, to_mcq(init_attr->cq)->mcq.cqn); + MLX5_SET(rqc, rqc, state, MLX5_RQC_STATE_RST); + MLX5_SET(rqc, rqc, flush_in_error_en, 1); + wq = MLX5_ADDR_OF(rqc, rqc, wq); + MLX5_SET(wq, wq, wq_type, MLX5_WQ_TYPE_CYCLIC); + MLX5_SET(wq, wq, end_padding_mode, MLX5_WQ_END_PAD_MODE_ALIGN); + MLX5_SET(wq, wq, log_wq_stride, rwq->log_rq_stride); + MLX5_SET(wq, wq, log_wq_sz, rwq->log_rq_size); + MLX5_SET(wq, wq, pd, to_mpd(pd)->pdn); + MLX5_SET(wq, wq, page_offset, rwq->rq_page_offset); + MLX5_SET(wq, wq, log_wq_pg_sz, rwq->log_page_size); + MLX5_SET(wq, wq, wq_signature, rwq->wq_sig); + MLX5_SET64(wq, wq, dbr_addr, rwq->db.dma); + rq_pas0 = (__be64 *)MLX5_ADDR_OF(wq, wq, pas); + mlx5_ib_populate_pas(dev, rwq->umem, rwq->page_shift, rq_pas0, 0); + err = mlx5_core_create_rq(dev->mdev, in, inlen, &rwq->rqn); + kvfree(in); + return err; +} + +static int set_user_rq_size(struct mlx5_ib_dev *dev, + struct ib_wq_init_attr *wq_init_attr, + struct mlx5_ib_create_wq *ucmd, + struct mlx5_ib_rwq *rwq) +{ + /* Sanity check RQ size before proceeding */ + if (wq_init_attr->max_wr > (1 << MLX5_CAP_GEN(dev->mdev, log_max_wq_sz))) + return -EINVAL; + + if (!ucmd->rq_wqe_count) + return -EINVAL; + + rwq->wqe_count = ucmd->rq_wqe_count; + rwq->wqe_shift = ucmd->rq_wqe_shift; + rwq->buf_size = (rwq->wqe_count << rwq->wqe_shift); + rwq->log_rq_stride = rwq->wqe_shift; + rwq->log_rq_size = ilog2(rwq->wqe_count); + return 0; +} + +static int prepare_user_rq(struct ib_pd *pd, + struct ib_wq_init_attr *init_attr, + struct ib_udata *udata, + struct mlx5_ib_rwq *rwq) +{ + struct mlx5_ib_dev *dev = to_mdev(pd->device); + struct mlx5_ib_create_wq ucmd = {}; + int err; + size_t required_cmd_sz; + + required_cmd_sz = offsetof(typeof(ucmd), reserved) + sizeof(ucmd.reserved); + if (udata->inlen < required_cmd_sz) { + mlx5_ib_dbg(dev, "invalid inlen\n"); + return -EINVAL; + } + + if (udata->inlen > sizeof(ucmd) && + !ib_is_udata_cleared(udata, sizeof(ucmd), + udata->inlen - sizeof(ucmd))) { + mlx5_ib_dbg(dev, "inlen is not supported\n"); + return -EOPNOTSUPP; + } + + if (ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen))) { + mlx5_ib_dbg(dev, "copy failed\n"); + return -EFAULT; + } + + if (ucmd.comp_mask) { + mlx5_ib_dbg(dev, "invalid comp mask\n"); + return -EOPNOTSUPP; + } + + if (ucmd.reserved) { + mlx5_ib_dbg(dev, "invalid reserved\n"); + return -EOPNOTSUPP; + } + + err = set_user_rq_size(dev, init_attr, &ucmd, rwq); + if (err) { + mlx5_ib_dbg(dev, "err %d\n", err); + return err; + } + + err = create_user_rq(dev, pd, rwq, &ucmd); + if (err) { + mlx5_ib_dbg(dev, "err %d\n", err); + if (err) + return err; + } + + rwq->user_index = ucmd.user_index; + return 0; +} + +struct ib_wq *mlx5_ib_create_wq(struct ib_pd *pd, + struct ib_wq_init_attr *init_attr, + struct ib_udata *udata) +{ + struct mlx5_ib_dev *dev; + struct mlx5_ib_rwq *rwq; + struct mlx5_ib_create_wq_resp resp = {}; + size_t min_resp_len; + int err; + + if (!udata) + return ERR_PTR(-ENOSYS); + + min_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved); + if (udata->outlen && udata->outlen < min_resp_len) + return ERR_PTR(-EINVAL); + + dev = to_mdev(pd->device); + switch (init_attr->wq_type) { + case IB_WQT_RQ: + rwq = kzalloc(sizeof(*rwq), GFP_KERNEL); + if (!rwq) + return ERR_PTR(-ENOMEM); + err = prepare_user_rq(pd, init_attr, udata, rwq); + if (err) + goto err; + err = create_rq(rwq, pd, init_attr); + if (err) + goto err_user_rq; + break; + default: + mlx5_ib_dbg(dev, "unsupported wq type %d\n", + init_attr->wq_type); + return ERR_PTR(-EINVAL); + } + + rwq->ibwq.wq_num = rwq->rqn; + rwq->ibwq.state = IB_WQS_RESET; + if (udata->outlen) { + resp.response_length = offsetof(typeof(resp), response_length) + + sizeof(resp.response_length); + err = ib_copy_to_udata(udata, &resp, resp.response_length); + if (err) + goto err_copy; + } + + return &rwq->ibwq; + +err_copy: + mlx5_core_destroy_rq(dev->mdev, rwq->rqn); +err_user_rq: + destroy_user_rq(pd, rwq); +err: + kfree(rwq); + return ERR_PTR(err); +} + +int mlx5_ib_destroy_wq(struct ib_wq *wq) +{ + struct mlx5_ib_dev *dev = to_mdev(wq->device); + struct mlx5_ib_rwq *rwq = to_mrwq(wq); + + mlx5_core_destroy_rq(dev->mdev, rwq->rqn); + destroy_user_rq(wq->pd, rwq); + kfree(rwq); + + return 0; +} + +struct ib_rwq_ind_table *mlx5_ib_create_rwq_ind_table(struct ib_device *device, + struct ib_rwq_ind_table_init_attr *init_attr, + struct ib_udata *udata) +{ + struct mlx5_ib_dev *dev = to_mdev(device); + struct mlx5_ib_rwq_ind_table *rwq_ind_tbl; + int sz = 1 << init_attr->log_ind_tbl_size; + struct mlx5_ib_create_rwq_ind_tbl_resp resp = {}; + size_t min_resp_len; + int inlen; + int err; + int i; + u32 *in; + void *rqtc; + + if (udata->inlen > 0 && + !ib_is_udata_cleared(udata, 0, + udata->inlen)) + return ERR_PTR(-EOPNOTSUPP); + + min_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved); + if (udata->outlen && udata->outlen < min_resp_len) + return ERR_PTR(-EINVAL); + + rwq_ind_tbl = kzalloc(sizeof(*rwq_ind_tbl), GFP_KERNEL); + if (!rwq_ind_tbl) + return ERR_PTR(-ENOMEM); + + inlen = MLX5_ST_SZ_BYTES(create_rqt_in) + sizeof(u32) * sz; + in = mlx5_vzalloc(inlen); + if (!in) { + err = -ENOMEM; + goto err; + } + + rqtc = MLX5_ADDR_OF(create_rqt_in, in, rqt_context); + + MLX5_SET(rqtc, rqtc, rqt_actual_size, sz); + MLX5_SET(rqtc, rqtc, rqt_max_size, sz); + + for (i = 0; i < sz; i++) + MLX5_SET(rqtc, rqtc, rq_num[i], init_attr->ind_tbl[i]->wq_num); + + err = mlx5_core_create_rqt(dev->mdev, in, inlen, &rwq_ind_tbl->rqtn); + kvfree(in); + + if (err) + goto err; + + rwq_ind_tbl->ib_rwq_ind_tbl.ind_tbl_num = rwq_ind_tbl->rqtn; + if (udata->outlen) { + resp.response_length = offsetof(typeof(resp), response_length) + + sizeof(resp.response_length); + err = ib_copy_to_udata(udata, &resp, resp.response_length); + if (err) + goto err_copy; + } + + return &rwq_ind_tbl->ib_rwq_ind_tbl; + +err_copy: + mlx5_core_destroy_rqt(dev->mdev, rwq_ind_tbl->rqtn); +err: + kfree(rwq_ind_tbl); + return ERR_PTR(err); +} + +int mlx5_ib_destroy_rwq_ind_table(struct ib_rwq_ind_table *ib_rwq_ind_tbl) +{ + struct mlx5_ib_rwq_ind_table *rwq_ind_tbl = to_mrwq_ind_table(ib_rwq_ind_tbl); + struct mlx5_ib_dev *dev = to_mdev(ib_rwq_ind_tbl->device); + + mlx5_core_destroy_rqt(dev->mdev, rwq_ind_tbl->rqtn); + + kfree(rwq_ind_tbl); + return 0; +} + +int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr, + u32 wq_attr_mask, struct ib_udata *udata) +{ + struct mlx5_ib_dev *dev = to_mdev(wq->device); + struct mlx5_ib_rwq *rwq = to_mrwq(wq); + struct mlx5_ib_modify_wq ucmd = {}; + size_t required_cmd_sz; + int curr_wq_state; + int wq_state; + int inlen; + int err; + void *rqc; + void *in; + + required_cmd_sz = offsetof(typeof(ucmd), reserved) + sizeof(ucmd.reserved); + if (udata->inlen < required_cmd_sz) + return -EINVAL; + + if (udata->inlen > sizeof(ucmd) && + !ib_is_udata_cleared(udata, sizeof(ucmd), + udata->inlen - sizeof(ucmd))) + return -EOPNOTSUPP; + + if (ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen))) + return -EFAULT; + + if (ucmd.comp_mask || ucmd.reserved) + return -EOPNOTSUPP; + + inlen = MLX5_ST_SZ_BYTES(modify_rq_in); + in = mlx5_vzalloc(inlen); + if (!in) + return -ENOMEM; + + rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx); + + curr_wq_state = (wq_attr_mask & IB_WQ_CUR_STATE) ? + wq_attr->curr_wq_state : wq->state; + wq_state = (wq_attr_mask & IB_WQ_STATE) ? + wq_attr->wq_state : curr_wq_state; + if (curr_wq_state == IB_WQS_ERR) + curr_wq_state = MLX5_RQC_STATE_ERR; + if (wq_state == IB_WQS_ERR) + wq_state = MLX5_RQC_STATE_ERR; + MLX5_SET(modify_rq_in, in, rq_state, curr_wq_state); + MLX5_SET(rqc, rqc, state, wq_state); + + err = mlx5_core_modify_rq(dev->mdev, rwq->rqn, in, inlen); + kvfree(in); + if (!err) + rwq->ibwq.state = (wq_state == MLX5_RQC_STATE_ERR) ? IB_WQS_ERR : wq_state; + + return err; +} diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c index 3b2ddd64a371..ed6ac52355f1 100644 --- a/drivers/infiniband/hw/mlx5/srq.c +++ b/drivers/infiniband/hw/mlx5/srq.c @@ -74,14 +74,12 @@ static void mlx5_ib_srq_event(struct mlx5_core_srq *srq, enum mlx5_event type) } static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq, - struct mlx5_create_srq_mbox_in **in, - struct ib_udata *udata, int buf_size, int *inlen, - int is_xrc) + struct mlx5_srq_attr *in, + struct ib_udata *udata, int buf_size) { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_create_srq ucmd = {}; size_t ucmdlen; - void *xsrqc; int err; int npages; int page_shift; @@ -104,7 +102,7 @@ static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq, udata->inlen - sizeof(ucmd))) return -EINVAL; - if (is_xrc) { + if (in->type == IB_SRQT_XRC) { err = get_srq_user_index(to_mucontext(pd->uobject->context), &ucmd, udata->inlen, &uidx); if (err) @@ -130,14 +128,13 @@ static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq, goto err_umem; } - *inlen = sizeof(**in) + sizeof(*(*in)->pas) * ncont; - *in = mlx5_vzalloc(*inlen); - if (!(*in)) { + in->pas = mlx5_vzalloc(sizeof(*in->pas) * ncont); + if (!in->pas) { err = -ENOMEM; goto err_umem; } - mlx5_ib_populate_pas(dev, srq->umem, page_shift, (*in)->pas, 0); + mlx5_ib_populate_pas(dev, srq->umem, page_shift, in->pas, 0); err = mlx5_ib_db_map_user(to_mucontext(pd->uobject->context), ucmd.db_addr, &srq->db); @@ -146,20 +143,16 @@ static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq, goto err_in; } - (*in)->ctx.log_pg_sz = page_shift - MLX5_ADAPTER_PAGE_SHIFT; - (*in)->ctx.pgoff_cqn = cpu_to_be32(offset << 26); - - if ((MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1) && - is_xrc){ - xsrqc = MLX5_ADDR_OF(create_xrc_srq_in, *in, - xrc_srq_context_entry); - MLX5_SET(xrc_srqc, xsrqc, user_index, uidx); - } + in->log_page_size = page_shift - MLX5_ADAPTER_PAGE_SHIFT; + in->page_offset = offset; + if (MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1 && + in->type == IB_SRQT_XRC) + in->user_index = uidx; return 0; err_in: - kvfree(*in); + kvfree(in->pas); err_umem: ib_umem_release(srq->umem); @@ -168,15 +161,13 @@ err_umem: } static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, - struct mlx5_create_srq_mbox_in **in, int buf_size, - int *inlen, int is_xrc) + struct mlx5_srq_attr *in, int buf_size) { int err; int i; struct mlx5_wqe_srq_next_seg *next; int page_shift; int npages; - void *xsrqc; err = mlx5_db_alloc(dev->mdev, &srq->db); if (err) { @@ -204,13 +195,12 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, npages = DIV_ROUND_UP(srq->buf.npages, 1 << (page_shift - PAGE_SHIFT)); mlx5_ib_dbg(dev, "buf_size %d, page_shift %d, npages %d, calc npages %d\n", buf_size, page_shift, srq->buf.npages, npages); - *inlen = sizeof(**in) + sizeof(*(*in)->pas) * npages; - *in = mlx5_vzalloc(*inlen); - if (!*in) { + in->pas = mlx5_vzalloc(sizeof(*in->pas) * npages); + if (!in->pas) { err = -ENOMEM; goto err_buf; } - mlx5_fill_page_array(&srq->buf, (*in)->pas); + mlx5_fill_page_array(&srq->buf, in->pas); srq->wrid = kmalloc(srq->msrq.max * sizeof(u64), GFP_KERNEL); if (!srq->wrid) { @@ -221,20 +211,15 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, } srq->wq_sig = !!srq_signature; - (*in)->ctx.log_pg_sz = page_shift - MLX5_ADAPTER_PAGE_SHIFT; - - if ((MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1) && - is_xrc){ - xsrqc = MLX5_ADDR_OF(create_xrc_srq_in, *in, - xrc_srq_context_entry); - /* 0xffffff means we ask to work with cqe version 0 */ - MLX5_SET(xrc_srqc, xsrqc, user_index, MLX5_IB_DEFAULT_UIDX); - } + in->log_page_size = page_shift - MLX5_ADAPTER_PAGE_SHIFT; + if (MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1 && + in->type == IB_SRQT_XRC) + in->user_index = MLX5_IB_DEFAULT_UIDX; return 0; err_in: - kvfree(*in); + kvfree(in->pas); err_buf: mlx5_buf_free(dev->mdev, &srq->buf); @@ -267,10 +252,7 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, int desc_size; int buf_size; int err; - struct mlx5_create_srq_mbox_in *uninitialized_var(in); - int uninitialized_var(inlen); - int is_xrc; - u32 flgs, xrcdn; + struct mlx5_srq_attr in = {0}; __u32 max_srq_wqes = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz); /* Sanity check SRQ size before proceeding */ @@ -302,14 +284,10 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, desc_size, init_attr->attr.max_wr, srq->msrq.max, srq->msrq.max_gs, srq->msrq.max_avail_gather); - is_xrc = (init_attr->srq_type == IB_SRQT_XRC); - if (pd->uobject) - err = create_srq_user(pd, srq, &in, udata, buf_size, &inlen, - is_xrc); + err = create_srq_user(pd, srq, &in, udata, buf_size); else - err = create_srq_kernel(dev, srq, &in, buf_size, &inlen, - is_xrc); + err = create_srq_kernel(dev, srq, &in, buf_size); if (err) { mlx5_ib_warn(dev, "create srq %s failed, err %d\n", @@ -317,23 +295,23 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, goto err_srq; } - in->ctx.state_log_sz = ilog2(srq->msrq.max); - flgs = ((srq->msrq.wqe_shift - 4) | (is_xrc << 5) | (srq->wq_sig << 7)) << 24; - xrcdn = 0; - if (is_xrc) { - xrcdn = to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn; - in->ctx.pgoff_cqn |= cpu_to_be32(to_mcq(init_attr->ext.xrc.cq)->mcq.cqn); + in.type = init_attr->srq_type; + in.log_size = ilog2(srq->msrq.max); + in.wqe_shift = srq->msrq.wqe_shift - 4; + if (srq->wq_sig) + in.flags |= MLX5_SRQ_FLAG_WQ_SIG; + if (init_attr->srq_type == IB_SRQT_XRC) { + in.xrcd = to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn; + in.cqn = to_mcq(init_attr->ext.xrc.cq)->mcq.cqn; } else if (init_attr->srq_type == IB_SRQT_BASIC) { - xrcdn = to_mxrcd(dev->devr.x0)->xrcdn; - in->ctx.pgoff_cqn |= cpu_to_be32(to_mcq(dev->devr.c0)->mcq.cqn); + in.xrcd = to_mxrcd(dev->devr.x0)->xrcdn; + in.cqn = to_mcq(dev->devr.c0)->mcq.cqn; } - in->ctx.flags_xrcd = cpu_to_be32((flgs & 0xFF000000) | (xrcdn & 0xFFFFFF)); - - in->ctx.pd = cpu_to_be32(to_mpd(pd)->pdn); - in->ctx.db_record = cpu_to_be64(srq->db.dma); - err = mlx5_core_create_srq(dev->mdev, &srq->msrq, in, inlen, is_xrc); - kvfree(in); + in.pd = to_mpd(pd)->pdn; + in.db_record = srq->db.dma; + err = mlx5_core_create_srq(dev->mdev, &srq->msrq, &in); + kvfree(in.pas); if (err) { mlx5_ib_dbg(dev, "create SRQ failed, err %d\n", err); goto err_usr_kern_srq; @@ -401,7 +379,7 @@ int mlx5_ib_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr) struct mlx5_ib_dev *dev = to_mdev(ibsrq->device); struct mlx5_ib_srq *srq = to_msrq(ibsrq); int ret; - struct mlx5_query_srq_mbox_out *out; + struct mlx5_srq_attr *out; out = kzalloc(sizeof(*out), GFP_KERNEL); if (!out) @@ -411,7 +389,7 @@ int mlx5_ib_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr) if (ret) goto out_box; - srq_attr->srq_limit = be16_to_cpu(out->ctx.lwm); + srq_attr->srq_limit = out->lwm; srq_attr->max_wr = srq->msrq.max - 1; srq_attr->max_sge = srq->msrq.max_gs; @@ -458,6 +436,8 @@ int mlx5_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr, struct mlx5_ib_srq *srq = to_msrq(ibsrq); struct mlx5_wqe_srq_next_seg *next; struct mlx5_wqe_data_seg *scat; + struct mlx5_ib_dev *dev = to_mdev(ibsrq->device); + struct mlx5_core_dev *mdev = dev->mdev; unsigned long flags; int err = 0; int nreq; @@ -465,6 +445,12 @@ int mlx5_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr, spin_lock_irqsave(&srq->lock, flags); + if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + err = -EIO; + *bad_wr = wr; + goto out; + } + for (nreq = 0; wr; nreq++, wr = wr->next) { if (unlikely(wr->num_sge > srq->msrq.max_gs)) { err = -EINVAL; @@ -507,7 +493,7 @@ int mlx5_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr, *srq->db.db = cpu_to_be32(srq->wqe_ctr); } - +out: spin_unlock_irqrestore(&srq->lock, flags); return err; diff --git a/drivers/infiniband/hw/mlx5/user.h b/drivers/infiniband/hw/mlx5/user.h index 61bc308bb802..188dac4301b5 100644 --- a/drivers/infiniband/hw/mlx5/user.h +++ b/drivers/infiniband/hw/mlx5/user.h @@ -46,6 +46,10 @@ enum { MLX5_SRQ_FLAG_SIGNATURE = 1 << 0, }; +enum { + MLX5_WQ_FLAG_SIGNATURE = 1 << 0, +}; + /* Increment this value if any changes that break userspace ABI * compatibility are made. @@ -79,6 +83,10 @@ enum mlx5_ib_alloc_ucontext_resp_mask { MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_CORE_CLOCK_OFFSET = 1UL << 0, }; +enum mlx5_user_cmds_supp_uhw { + MLX5_USER_CMDS_SUPP_UHW_QUERY_DEVICE = 1 << 0, +}; + struct mlx5_ib_alloc_ucontext_resp { __u32 qp_tab_size; __u32 bf_reg_size; @@ -94,8 +102,8 @@ struct mlx5_ib_alloc_ucontext_resp { __u32 comp_mask; __u32 response_length; __u8 cqe_version; - __u8 reserved2; - __u16 reserved3; + __u8 cmds_supp_uhw; + __u16 reserved2; __u64 hca_core_clock_offset; }; @@ -103,6 +111,22 @@ struct mlx5_ib_alloc_pd_resp { __u32 pdn; }; +struct mlx5_ib_tso_caps { + __u32 max_tso; /* Maximum tso payload size in bytes */ + + /* Corresponding bit will be set if qp type from + * 'enum ib_qp_type' is supported, e.g. + * supported_qpts |= 1 << IB_QPT_UD + */ + __u32 supported_qpts; +}; + +struct mlx5_ib_query_device_resp { + __u32 comp_mask; + __u32 response_length; + struct mlx5_ib_tso_caps tso_caps; +}; + struct mlx5_ib_create_cq { __u64 buf_addr; __u64 db_addr; @@ -148,6 +172,40 @@ struct mlx5_ib_create_qp { __u64 sq_buf_addr; }; +/* RX Hash function flags */ +enum mlx5_rx_hash_function_flags { + MLX5_RX_HASH_FUNC_TOEPLITZ = 1 << 0, +}; + +/* + * RX Hash flags, these flags allows to set which incoming packet's field should + * participates in RX Hash. Each flag represent certain packet's field, + * when the flag is set the field that is represented by the flag will + * participate in RX Hash calculation. + * Note: *IPV4 and *IPV6 flags can't be enabled together on the same QP + * and *TCP and *UDP flags can't be enabled together on the same QP. +*/ +enum mlx5_rx_hash_fields { + MLX5_RX_HASH_SRC_IPV4 = 1 << 0, + MLX5_RX_HASH_DST_IPV4 = 1 << 1, + MLX5_RX_HASH_SRC_IPV6 = 1 << 2, + MLX5_RX_HASH_DST_IPV6 = 1 << 3, + MLX5_RX_HASH_SRC_PORT_TCP = 1 << 4, + MLX5_RX_HASH_DST_PORT_TCP = 1 << 5, + MLX5_RX_HASH_SRC_PORT_UDP = 1 << 6, + MLX5_RX_HASH_DST_PORT_UDP = 1 << 7 +}; + +struct mlx5_ib_create_qp_rss { + __u64 rx_hash_fields_mask; /* enum mlx5_rx_hash_fields */ + __u8 rx_hash_function; /* enum mlx5_rx_hash_function_flags */ + __u8 rx_key_len; /* valid only for Toeplitz */ + __u8 reserved[6]; + __u8 rx_hash_key[128]; /* valid only for Toeplitz */ + __u32 comp_mask; + __u32 reserved1; +}; + struct mlx5_ib_create_qp_resp { __u32 uuar_index; }; @@ -159,6 +217,32 @@ struct mlx5_ib_alloc_mw { __u16 reserved2; }; +struct mlx5_ib_create_wq { + __u64 buf_addr; + __u64 db_addr; + __u32 rq_wqe_count; + __u32 rq_wqe_shift; + __u32 user_index; + __u32 flags; + __u32 comp_mask; + __u32 reserved; +}; + +struct mlx5_ib_create_wq_resp { + __u32 response_length; + __u32 reserved; +}; + +struct mlx5_ib_create_rwq_ind_tbl_resp { + __u32 response_length; + __u32 reserved; +}; + +struct mlx5_ib_modify_wq { + __u32 comp_mask; + __u32 reserved; +}; + static inline int get_qp_user_index(struct mlx5_ib_ucontext *ucontext, struct mlx5_ib_create_qp *ucmd, int inlen, diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c index 9866c35cc977..da2335f7f7c3 100644 --- a/drivers/infiniband/hw/mthca/mthca_provider.c +++ b/drivers/infiniband/hw/mthca/mthca_provider.c @@ -1081,16 +1081,6 @@ static ssize_t show_rev(struct device *device, struct device_attribute *attr, return sprintf(buf, "%x\n", dev->rev_id); } -static ssize_t show_fw_ver(struct device *device, struct device_attribute *attr, - char *buf) -{ - struct mthca_dev *dev = - container_of(device, struct mthca_dev, ib_dev.dev); - return sprintf(buf, "%d.%d.%d\n", (int) (dev->fw_ver >> 32), - (int) (dev->fw_ver >> 16) & 0xffff, - (int) dev->fw_ver & 0xffff); -} - static ssize_t show_hca(struct device *device, struct device_attribute *attr, char *buf) { @@ -1120,13 +1110,11 @@ static ssize_t show_board(struct device *device, struct device_attribute *attr, } 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 *mthca_dev_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type, &dev_attr_board_id }; @@ -1187,6 +1175,17 @@ static int mthca_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_dev_fw_str(struct ib_device *device, char *str, + size_t str_len) +{ + struct mthca_dev *dev = + container_of(device, struct mthca_dev, ib_dev); + snprintf(str, str_len, "%d.%d.%d", + (int) (dev->fw_ver >> 32), + (int) (dev->fw_ver >> 16) & 0xffff, + (int) dev->fw_ver & 0xffff); +} + int mthca_register_device(struct mthca_dev *dev) { int ret; @@ -1266,6 +1265,7 @@ int mthca_register_device(struct mthca_dev *dev) dev->ib_dev.reg_user_mr = mthca_reg_user_mr; dev->ib_dev.dereg_mr = mthca_dereg_mr; dev->ib_dev.get_port_immutable = mthca_port_immutable; + dev->ib_dev.get_dev_fw_str = get_dev_fw_str; if (dev->mthca_flags & MTHCA_FLAG_FMR) { dev->ib_dev.alloc_fmr = mthca_alloc_fmr; diff --git a/drivers/infiniband/hw/mthca/mthca_reset.c b/drivers/infiniband/hw/mthca/mthca_reset.c index 74c6a9426047..6727af27c017 100644 --- a/drivers/infiniband/hw/mthca/mthca_reset.c +++ b/drivers/infiniband/hw/mthca/mthca_reset.c @@ -98,7 +98,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENOMEM; mthca_err(mdev, "Couldn't allocate memory to save HCA " "PCI header, aborting.\n"); - goto out; + goto put_dev; } for (i = 0; i < 64; ++i) { @@ -108,7 +108,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENODEV; mthca_err(mdev, "Couldn't save HCA " "PCI header, aborting.\n"); - goto out; + goto free_hca; } } @@ -121,7 +121,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENOMEM; mthca_err(mdev, "Couldn't allocate memory to save HCA " "bridge PCI header, aborting.\n"); - goto out; + goto free_hca; } for (i = 0; i < 64; ++i) { @@ -131,7 +131,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENODEV; mthca_err(mdev, "Couldn't save HCA bridge " "PCI header, aborting.\n"); - goto out; + goto free_bh; } } bridge_pcix_cap = pci_find_capability(bridge, PCI_CAP_ID_PCIX); @@ -139,7 +139,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENODEV; mthca_err(mdev, "Couldn't locate HCA bridge " "PCI-X capability, aborting.\n"); - goto out; + goto free_bh; } } @@ -152,7 +152,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENOMEM; mthca_err(mdev, "Couldn't map HCA reset register, " "aborting.\n"); - goto out; + goto free_bh; } writel(MTHCA_RESET_VALUE, reset); @@ -172,7 +172,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENODEV; mthca_err(mdev, "Couldn't access HCA after reset, " "aborting.\n"); - goto out; + goto free_bh; } if (v != 0xffffffff) @@ -184,7 +184,7 @@ int mthca_reset(struct mthca_dev *mdev) err = -ENODEV; mthca_err(mdev, "PCI device did not come back after reset, " "aborting.\n"); - goto out; + goto free_bh; } good: @@ -195,14 +195,14 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA bridge Upstream " "split transaction control, aborting.\n"); - goto out; + goto free_bh; } if (pci_write_config_dword(bridge, bridge_pcix_cap + 0xc, bridge_header[(bridge_pcix_cap + 0xc) / 4])) { err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA bridge Downstream " "split transaction control, aborting.\n"); - goto out; + goto free_bh; } /* * Bridge control register is at 0x3e, so we'll @@ -216,7 +216,7 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA bridge reg %x, " "aborting.\n", i); - goto out; + goto free_bh; } } @@ -225,7 +225,7 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA bridge COMMAND, " "aborting.\n"); - goto out; + goto free_bh; } } @@ -235,7 +235,7 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA PCI-X " "command register, aborting.\n"); - goto out; + goto free_bh; } } @@ -246,7 +246,7 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA PCI Express " "Device Control register, aborting.\n"); - goto out; + goto free_bh; } linkctl = hca_header[(hca_pcie_cap + PCI_EXP_LNKCTL) / 4]; if (pcie_capability_write_word(mdev->pdev, PCI_EXP_LNKCTL, @@ -254,7 +254,7 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA PCI Express " "Link control register, aborting.\n"); - goto out; + goto free_bh; } } @@ -266,7 +266,7 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA reg %x, " "aborting.\n", i); - goto out; + goto free_bh; } } @@ -275,14 +275,12 @@ good: err = -ENODEV; mthca_err(mdev, "Couldn't restore HCA COMMAND, " "aborting.\n"); - goto out; } - -out: - if (bridge) - pci_dev_put(bridge); +free_bh: kfree(bridge_header); +free_hca: kfree(hca_header); - +put_dev: + pci_dev_put(bridge); return err; } diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c index 464d6da5fe91..bd69125731c1 100644 --- a/drivers/infiniband/hw/nes/nes_verbs.c +++ b/drivers/infiniband/hw/nes/nes_verbs.c @@ -2606,23 +2606,6 @@ static ssize_t show_rev(struct device *dev, struct device_attribute *attr, /** - * show_fw_ver - */ -static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, - char *buf) -{ - struct nes_ib_device *nesibdev = - container_of(dev, struct nes_ib_device, ibdev.dev); - struct nes_vnic *nesvnic = nesibdev->nesvnic; - - nes_debug(NES_DBG_INIT, "\n"); - return sprintf(buf, "%u.%u\n", - (nesvnic->nesdev->nesadapter->firmware_version >> 16), - (nesvnic->nesdev->nesadapter->firmware_version & 0x000000ff)); -} - - -/** * show_hca */ static ssize_t show_hca(struct device *dev, struct device_attribute *attr, @@ -2645,13 +2628,11 @@ static ssize_t show_board(struct device *dev, struct device_attribute *attr, 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 *nes_dev_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type, &dev_attr_board_id }; @@ -3703,6 +3684,19 @@ static int nes_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_dev_fw_str(struct ib_device *dev, char *str, + size_t str_len) +{ + struct nes_ib_device *nesibdev = + container_of(dev, struct nes_ib_device, ibdev); + struct nes_vnic *nesvnic = nesibdev->nesvnic; + + nes_debug(NES_DBG_INIT, "\n"); + snprintf(str, str_len, "%u.%u", + (nesvnic->nesdev->nesadapter->firmware_version >> 16), + (nesvnic->nesdev->nesadapter->firmware_version & 0x000000ff)); +} + /** * nes_init_ofa_device */ @@ -3802,6 +3796,7 @@ struct nes_ib_device *nes_init_ofa_device(struct net_device *netdev) nesibdev->ibdev.iwcm->create_listen = nes_create_listen; nesibdev->ibdev.iwcm->destroy_listen = nes_destroy_listen; nesibdev->ibdev.get_port_immutable = nes_port_immutable; + nesibdev->ibdev.get_dev_fw_str = get_dev_fw_str; memcpy(nesibdev->ibdev.iwcm->ifname, netdev->name, sizeof(nesibdev->ibdev.iwcm->ifname)); diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c index 3d75f65ce87e..07d0c6c5b046 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c @@ -107,6 +107,14 @@ static int ocrdma_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void get_dev_fw_str(struct ib_device *device, char *str, + size_t str_len) +{ + struct ocrdma_dev *dev = get_ocrdma_dev(device); + + snprintf(str, str_len, "%s", &dev->attr.fw_ver[0]); +} + static int ocrdma_register_device(struct ocrdma_dev *dev) { strlcpy(dev->ibdev.name, "ocrdma%d", IB_DEVICE_NAME_MAX); @@ -193,6 +201,7 @@ static int ocrdma_register_device(struct ocrdma_dev *dev) dev->ibdev.process_mad = ocrdma_process_mad; dev->ibdev.get_port_immutable = ocrdma_port_immutable; + dev->ibdev.get_dev_fw_str = get_dev_fw_str; if (ocrdma_get_asic_type(dev) == OCRDMA_ASIC_GEN_SKH_R) { dev->ibdev.uverbs_cmd_mask |= @@ -262,14 +271,6 @@ static ssize_t show_rev(struct device *device, struct device_attribute *attr, return scnprintf(buf, PAGE_SIZE, "0x%x\n", dev->nic_info.pdev->vendor); } -static ssize_t show_fw_ver(struct device *device, struct device_attribute *attr, - char *buf) -{ - struct ocrdma_dev *dev = dev_get_drvdata(device); - - return scnprintf(buf, PAGE_SIZE, "%s\n", &dev->attr.fw_ver[0]); -} - static ssize_t show_hca_type(struct device *device, struct device_attribute *attr, char *buf) { @@ -279,12 +280,10 @@ static ssize_t show_hca_type(struct device *device, } 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_type, NULL); static struct device_attribute *ocrdma_attributes[] = { &dev_attr_hw_rev, - &dev_attr_fw_ver, &dev_attr_hca_type }; diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c index 565c881a44ba..c229b9f4a52d 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_main.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c @@ -331,6 +331,21 @@ static int usnic_port_immutable(struct ib_device *ibdev, u8 port_num, return 0; } +static void usnic_get_dev_fw_str(struct ib_device *device, + char *str, + size_t str_len) +{ + struct usnic_ib_dev *us_ibdev = + container_of(device, struct usnic_ib_dev, ib_dev); + struct ethtool_drvinfo info; + + mutex_lock(&us_ibdev->usdev_lock); + us_ibdev->netdev->ethtool_ops->get_drvinfo(us_ibdev->netdev, &info); + mutex_unlock(&us_ibdev->usdev_lock); + + snprintf(str, str_len, "%s", info.fw_version); +} + /* Start of PF discovery section */ static void *usnic_ib_device_add(struct pci_dev *dev) { @@ -414,6 +429,7 @@ static void *usnic_ib_device_add(struct pci_dev *dev) us_ibdev->ib_dev.req_notify_cq = usnic_ib_req_notify_cq; us_ibdev->ib_dev.get_dma_mr = usnic_ib_get_dma_mr; us_ibdev->ib_dev.get_port_immutable = usnic_port_immutable; + us_ibdev->ib_dev.get_dev_fw_str = usnic_get_dev_fw_str; if (ib_register_device(&us_ibdev->ib_dev, NULL)) diff --git a/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c b/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c index 3412ea06116e..80ef3f8998c8 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_sysfs.c @@ -45,21 +45,6 @@ #include "usnic_ib_verbs.h" #include "usnic_log.h" -static ssize_t usnic_ib_show_fw_ver(struct device *device, - struct device_attribute *attr, - char *buf) -{ - struct usnic_ib_dev *us_ibdev = - container_of(device, struct usnic_ib_dev, ib_dev.dev); - struct ethtool_drvinfo info; - - mutex_lock(&us_ibdev->usdev_lock); - us_ibdev->netdev->ethtool_ops->get_drvinfo(us_ibdev->netdev, &info); - mutex_unlock(&us_ibdev->usdev_lock); - - return scnprintf(buf, PAGE_SIZE, "%s\n", info.fw_version); -} - static ssize_t usnic_ib_show_board(struct device *device, struct device_attribute *attr, char *buf) @@ -192,7 +177,6 @@ usnic_ib_show_cq_per_vf(struct device *device, struct device_attribute *attr, us_ibdev->vf_res_cnt[USNIC_VNIC_RES_TYPE_CQ]); } -static DEVICE_ATTR(fw_ver, S_IRUGO, usnic_ib_show_fw_ver, NULL); static DEVICE_ATTR(board_id, S_IRUGO, usnic_ib_show_board, NULL); static DEVICE_ATTR(config, S_IRUGO, usnic_ib_show_config, NULL); static DEVICE_ATTR(iface, S_IRUGO, usnic_ib_show_iface, NULL); @@ -201,7 +185,6 @@ static DEVICE_ATTR(qp_per_vf, S_IRUGO, usnic_ib_show_qp_per_vf, NULL); static DEVICE_ATTR(cq_per_vf, S_IRUGO, usnic_ib_show_cq_per_vf, NULL); static struct device_attribute *usnic_class_attributes[] = { - &dev_attr_fw_ver, &dev_attr_board_id, &dev_attr_config, &dev_attr_iface, |