From 765d67748bcf802c4642a49cd0139787d0d80783 Mon Sep 17 00:00:00 2001 From: Steve Wise Date: Wed, 17 Feb 2016 08:15:41 -0800 Subject: IB: new common API for draining queues Add provider-specific drain_sq/drain_rq functions for providers needing special drain logic. Add static functions __ib_drain_sq() and __ib_drain_rq() which post noop WRs to the SQ or RQ and block until their completions are processed. This ensures the applications completions for work requests posted prior to the drain work request have all been processed. Add API functions ib_drain_sq(), ib_drain_rq(), and ib_drain_qp(). For the drain logic to work, the caller must: ensure there is room in the CQ(s) and QP for the drain work request and completion. allocate the CQ using ib_alloc_cq() and the CQ poll context cannot be IB_POLL_DIRECT. ensure that there are no other contexts that are posting WRs concurrently. Otherwise the drain is not guaranteed. Reviewed-by: Chuck Lever Signed-off-by: Steve Wise Signed-off-by: Doug Ledford --- drivers/infiniband/core/verbs.c | 164 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 164 insertions(+) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index 5af6d02..48dc43c 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -1657,3 +1657,167 @@ next_page: return i; } EXPORT_SYMBOL(ib_sg_to_pages); + +struct ib_drain_cqe { + struct ib_cqe cqe; + struct completion done; +}; + +static void ib_drain_qp_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct ib_drain_cqe *cqe = container_of(wc->wr_cqe, struct ib_drain_cqe, + cqe); + + complete(&cqe->done); +} + +/* + * Post a WR and block until its completion is reaped for the SQ. + */ +static void __ib_drain_sq(struct ib_qp *qp) +{ + struct ib_qp_attr attr = { .qp_state = IB_QPS_ERR }; + struct ib_drain_cqe sdrain; + struct ib_send_wr swr = {}, *bad_swr; + int ret; + + if (qp->send_cq->poll_ctx == IB_POLL_DIRECT) { + WARN_ONCE(qp->send_cq->poll_ctx == IB_POLL_DIRECT, + "IB_POLL_DIRECT poll_ctx not supported for drain\n"); + return; + } + + swr.wr_cqe = &sdrain.cqe; + sdrain.cqe.done = ib_drain_qp_done; + init_completion(&sdrain.done); + + ret = ib_modify_qp(qp, &attr, IB_QP_STATE); + if (ret) { + WARN_ONCE(ret, "failed to drain send queue: %d\n", ret); + return; + } + + ret = ib_post_send(qp, &swr, &bad_swr); + if (ret) { + WARN_ONCE(ret, "failed to drain send queue: %d\n", ret); + return; + } + + wait_for_completion(&sdrain.done); +} + +/* + * Post a WR and block until its completion is reaped for the RQ. + */ +static void __ib_drain_rq(struct ib_qp *qp) +{ + struct ib_qp_attr attr = { .qp_state = IB_QPS_ERR }; + struct ib_drain_cqe rdrain; + struct ib_recv_wr rwr = {}, *bad_rwr; + int ret; + + if (qp->recv_cq->poll_ctx == IB_POLL_DIRECT) { + WARN_ONCE(qp->recv_cq->poll_ctx == IB_POLL_DIRECT, + "IB_POLL_DIRECT poll_ctx not supported for drain\n"); + return; + } + + rwr.wr_cqe = &rdrain.cqe; + rdrain.cqe.done = ib_drain_qp_done; + init_completion(&rdrain.done); + + ret = ib_modify_qp(qp, &attr, IB_QP_STATE); + if (ret) { + WARN_ONCE(ret, "failed to drain recv queue: %d\n", ret); + return; + } + + ret = ib_post_recv(qp, &rwr, &bad_rwr); + if (ret) { + WARN_ONCE(ret, "failed to drain recv queue: %d\n", ret); + return; + } + + wait_for_completion(&rdrain.done); +} + +/** + * ib_drain_sq() - Block until all SQ CQEs have been consumed by the + * application. + * @qp: queue pair to drain + * + * If the device has a provider-specific drain function, then + * call that. Otherwise call the generic drain function + * __ib_drain_sq(). + * + * The caller must: + * + * ensure there is room in the CQ and SQ for the drain work request and + * completion. + * + * allocate the CQ using ib_alloc_cq() and the CQ poll context cannot be + * IB_POLL_DIRECT. + * + * ensure that there are no other contexts that are posting WRs concurrently. + * Otherwise the drain is not guaranteed. + */ +void ib_drain_sq(struct ib_qp *qp) +{ + if (qp->device->drain_sq) + qp->device->drain_sq(qp); + else + __ib_drain_sq(qp); +} +EXPORT_SYMBOL(ib_drain_sq); + +/** + * ib_drain_rq() - Block until all RQ CQEs have been consumed by the + * application. + * @qp: queue pair to drain + * + * If the device has a provider-specific drain function, then + * call that. Otherwise call the generic drain function + * __ib_drain_rq(). + * + * The caller must: + * + * ensure there is room in the CQ and RQ for the drain work request and + * completion. + * + * allocate the CQ using ib_alloc_cq() and the CQ poll context cannot be + * IB_POLL_DIRECT. + * + * ensure that there are no other contexts that are posting WRs concurrently. + * Otherwise the drain is not guaranteed. + */ +void ib_drain_rq(struct ib_qp *qp) +{ + if (qp->device->drain_rq) + qp->device->drain_rq(qp); + else + __ib_drain_rq(qp); +} +EXPORT_SYMBOL(ib_drain_rq); + +/** + * ib_drain_qp() - Block until all CQEs have been consumed by the + * application on both the RQ and SQ. + * @qp: queue pair to drain + * + * The caller must: + * + * ensure there is room in the CQ(s), SQ, and RQ for drain work requests + * and completions. + * + * allocate the CQs using ib_alloc_cq() and the CQ poll context cannot be + * IB_POLL_DIRECT. + * + * ensure that there are no other contexts that are posting WRs concurrently. + * Otherwise the drain is not guaranteed. + */ +void ib_drain_qp(struct ib_qp *qp) +{ + ib_drain_sq(qp); + ib_drain_rq(qp); +} +EXPORT_SYMBOL(ib_drain_qp); -- cgit v1.1 From a3100a78794175d7f2488a3155d247da3d7390e4 Mon Sep 17 00:00:00 2001 From: Marina Varshaver Date: Thu, 18 Feb 2016 18:31:05 +0200 Subject: IB/core: Add don't trap flag to flow creation Don't trap flag (i.e. IB_FLOW_ATTR_FLAGS_DONT_TRAP) indicates that QP will receive traffic, but will not steal it. When a packet matches a flow steering rule that was created with the don't trap flag, the QPs assigned to this rule will get this packet, but matching will continue to other equal/lower priority rules. This will let other QPs assigned to those rules to get the packet too. If both don't trap rule and other rules have the same priority and match the same packet, the behavior is undefined. The don't trap flag can't be set with default rule types (i.e. IB_FLOW_ATTR_ALL_DEFAULT, IB_FLOW_ATTR_MC_DEFAULT) as default rules don't have rules after them and don't trap has no meaning here. Signed-off-by: Marina Varshaver Reviewed-by: Matan Barak Reviewed-by: Yishai Hadas Signed-off-by: Doug Ledford --- drivers/infiniband/core/uverbs_cmd.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index 6ffc9c4..0f05de6 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -3085,6 +3085,14 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, !capable(CAP_NET_ADMIN)) || !capable(CAP_NET_RAW)) return -EPERM; + if (cmd.flow_attr.flags >= IB_FLOW_ATTR_FLAGS_RESERVED) + return -EINVAL; + + if ((cmd.flow_attr.flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP) && + ((cmd.flow_attr.type == IB_FLOW_ATTR_ALL_DEFAULT) || + (cmd.flow_attr.type == IB_FLOW_ATTR_MC_DEFAULT))) + return -EINVAL; + if (cmd.flow_attr.num_of_specs > IB_FLOW_SPEC_SUPPORT_LAYERS) return -EINVAL; -- cgit v1.1 From 84424a7fc793979da12992cfe5c2f5f73a3e8725 Mon Sep 17 00:00:00 2001 From: Haggai Eran Date: Mon, 29 Feb 2016 15:45:12 +0200 Subject: IB/cma: Print warning on different inner and header P_Keys Commit 4c21b5bcef73 ("IB/cma: Add net_dev and private data checks to RDMA CM") added checks for incoming RDMA CM requests that they can be matched to a netdev based on the P_Key in the BTH of the request. This behavior was reverted in commit ab3964ad2acf ("IB/cma: Use inner P_Key to determine netdev"), since the mlx5 and ipath drivers didn't send the correct value in the BTH P_Key. Since the ipath driver was removed, and the mlx5 driver can now send GSI packets on different P_Keys, we could revert the patch to let the rdma_cm module look on the BTH P_Key when deciding to what netdev a packet belongs. However, that still breaks compatibility with the older drivers. Change the behavior to print a warning when receiving a request that has a different BTH P_Key and inner payload P_Key. In the future, after users have seen the warnings and upgraded their setups, remove the warning and block these requests. Signed-off-by: Haggai Eran Signed-off-by: Doug Ledford --- drivers/infiniband/core/cma.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 9729639..7eace1f 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -1206,6 +1206,10 @@ static int cma_save_req_info(const struct ib_cm_event *ib_event, req->has_gid = true; req->service_id = req_param->primary_path->service_id; req->pkey = be16_to_cpu(req_param->primary_path->pkey); + if (req->pkey != req_param->bth_pkey) + pr_warn_ratelimited("RDMA CMA: got different BTH P_Key (0x%x) and primary path P_Key (0x%x)\n" + "RDMA CMA: in the future this may cause the request to be dropped\n", + req_param->bth_pkey, req->pkey); break; case IB_CM_SIDR_REQ_RECEIVED: req->device = sidr_param->listen_id->device; @@ -1213,6 +1217,10 @@ static int cma_save_req_info(const struct ib_cm_event *ib_event, req->has_gid = false; req->service_id = sidr_param->service_id; req->pkey = sidr_param->pkey; + if (req->pkey != sidr_param->bth_pkey) + pr_warn_ratelimited("RDMA CMA: got different BTH P_Key (0x%x) and SIDR request payload P_Key (0x%x)\n" + "RDMA CMA: in the future this may cause the request to be dropped\n", + sidr_param->bth_pkey, req->pkey); break; default: return -EINVAL; -- cgit v1.1 From b2a239df4e65fe35240ddf3e5f9f31335c90589b Mon Sep 17 00:00:00 2001 From: Matan Barak Date: Mon, 29 Feb 2016 18:05:29 +0200 Subject: IB/core: Add vendor's specific data to alloc mw Passing udata to the vendor's driver in order to pass data from the user-space driver to the kernel-space driver. This data will be used in downstream patches. Signed-off-by: Matan Barak Reviewed-by: Yishai Hadas Signed-off-by: Doug Ledford --- drivers/infiniband/core/uverbs_cmd.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index 6ffc9c4..2bf751e 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -1174,6 +1174,7 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file, struct ib_uobject *uobj; struct ib_pd *pd; struct ib_mw *mw; + struct ib_udata udata; int ret; if (out_len < sizeof(resp)) @@ -1195,7 +1196,12 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file, goto err_free; } - mw = pd->device->alloc_mw(pd, cmd.mw_type); + INIT_UDATA(&udata, buf + sizeof(cmd), + (unsigned long)cmd.response + sizeof(resp), + in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), + out_len - sizeof(resp)); + + mw = pd->device->alloc_mw(pd, cmd.mw_type, &udata); if (IS_ERR(mw)) { ret = PTR_ERR(mw); goto err_put; -- cgit v1.1 From 74a0b0a5ea6b48f8945e6e900e9c678866aa7520 Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Sun, 14 Feb 2016 17:07:47 +0200 Subject: IB/core: Avoid duplicate code Move the check on the validity of the command to a common area. Signed-off-by: Eli Cohen Signed-off-by: Leon Romanovsky Signed-off-by: Doug Ledford --- drivers/infiniband/core/uverbs_main.c | 29 +++++++++-------------------- 1 file changed, 9 insertions(+), 20 deletions(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 39680ae..08f1a7b 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -689,6 +689,7 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, struct ib_uverbs_file *file = filp->private_data; struct ib_device *ib_dev; struct ib_uverbs_cmd_hdr hdr; + __u32 command; __u32 flags; int srcu_key; ssize_t ret; @@ -707,20 +708,18 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, goto out; } + if (hdr.command & ~(__u32)(IB_USER_VERBS_CMD_FLAGS_MASK | + IB_USER_VERBS_CMD_COMMAND_MASK)) { + ret = -EINVAL; + goto out; + } + + command = hdr.command & IB_USER_VERBS_CMD_COMMAND_MASK; + flags = (hdr.command & IB_USER_VERBS_CMD_FLAGS_MASK) >> IB_USER_VERBS_CMD_FLAGS_SHIFT; if (!flags) { - __u32 command; - - if (hdr.command & ~(__u32)(IB_USER_VERBS_CMD_FLAGS_MASK | - IB_USER_VERBS_CMD_COMMAND_MASK)) { - ret = -EINVAL; - goto out; - } - - command = hdr.command & IB_USER_VERBS_CMD_COMMAND_MASK; - if (command >= ARRAY_SIZE(uverbs_cmd_table) || !uverbs_cmd_table[command]) { ret = -EINVAL; @@ -749,21 +748,11 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, hdr.out_words * 4); } else if (flags == IB_USER_VERBS_CMD_FLAG_EXTENDED) { - __u32 command; - struct ib_uverbs_ex_cmd_hdr ex_hdr; struct ib_udata ucore; struct ib_udata uhw; size_t written_count = count; - if (hdr.command & ~(__u32)(IB_USER_VERBS_CMD_FLAGS_MASK | - IB_USER_VERBS_CMD_COMMAND_MASK)) { - ret = -EINVAL; - goto out; - } - - command = hdr.command & IB_USER_VERBS_CMD_COMMAND_MASK; - if (command >= ARRAY_SIZE(uverbs_ex_cmd_table) || !uverbs_ex_cmd_table[command]) { ret = -ENOSYS; -- cgit v1.1 From 2dbd5186a39c7ef0d92045c43b4857c23cb117ab Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Sun, 14 Feb 2016 17:07:48 +0200 Subject: IB/core: IB/core: Allow legacy verbs through extended interfaces When an extended verb is an extension to a legacy verb, the original functionality is preserved. Hence we do not require each hardware driver to set the extended capability. This will allow the use of the extended verb in its simple form with drivers that do not support the extended capability. Signed-off-by: Eli Cohen Signed-off-by: Leon Romanovsky Signed-off-by: Doug Ledford --- drivers/infiniband/core/uverbs_main.c | 29 +++++++++++++++++++---------- 1 file changed, 19 insertions(+), 10 deletions(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 08f1a7b..20f0049 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -683,6 +683,21 @@ out: return ev_file; } +static int verify_command_mask(struct ib_device *ib_dev, __u32 command) +{ + u64 mask; + + if (command <= IB_USER_VERBS_CMD_OPEN_QP) + mask = ib_dev->uverbs_cmd_mask; + else + mask = ib_dev->uverbs_ex_cmd_mask; + + if (mask & ((u64)1 << command)) + return 0; + + return -1; +} + static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, size_t count, loff_t *pos) { @@ -715,6 +730,10 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, } command = hdr.command & IB_USER_VERBS_CMD_COMMAND_MASK; + if (verify_command_mask(ib_dev, command)) { + ret = -EOPNOTSUPP; + goto out; + } flags = (hdr.command & IB_USER_VERBS_CMD_FLAGS_MASK) >> IB_USER_VERBS_CMD_FLAGS_SHIFT; @@ -732,11 +751,6 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, goto out; } - if (!(ib_dev->uverbs_cmd_mask & (1ull << command))) { - ret = -ENOSYS; - goto out; - } - if (hdr.in_words * 4 != count) { ret = -EINVAL; goto out; @@ -764,11 +778,6 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, goto out; } - if (!(ib_dev->uverbs_ex_cmd_mask & (1ull << command))) { - ret = -ENOSYS; - goto out; - } - if (count < (sizeof(hdr) + sizeof(ex_hdr))) { ret = -EINVAL; goto out; -- cgit v1.1 From eaebc7d21e102d86b389c77686c7e6258dd6d479 Mon Sep 17 00:00:00 2001 From: Eli Cohen Date: Sun, 14 Feb 2016 17:07:49 +0200 Subject: IB/core: Modify conditional on ucontext existence Since we allow to call legacy verbs using their extended counterpart, the check on ucontext has to move up to a common area in case this verb is ever extended. Signed-off-by: Eli Cohen Signed-off-by: Leon Romanovsky Signed-off-by: Doug Ledford --- drivers/infiniband/core/uverbs_main.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 20f0049..8b299df 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -735,6 +735,12 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, goto out; } + if (!file->ucontext && + command != IB_USER_VERBS_CMD_GET_CONTEXT) { + ret = -EINVAL; + goto out; + } + flags = (hdr.command & IB_USER_VERBS_CMD_FLAGS_MASK) >> IB_USER_VERBS_CMD_FLAGS_SHIFT; @@ -745,12 +751,6 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, goto out; } - if (!file->ucontext && - command != IB_USER_VERBS_CMD_GET_CONTEXT) { - ret = -EINVAL; - goto out; - } - if (hdr.in_words * 4 != count) { ret = -EINVAL; goto out; -- cgit v1.1 From db9314cd351ebd734970c7afc2995a743e34aa02 Mon Sep 17 00:00:00 2001 From: Amitoj Kaur Chawla Date: Fri, 12 Feb 2016 13:16:10 +0530 Subject: IB/core: Replace memset with eth_zero_addr Use eth_zero_addr to assign the zero address to the given address array instead of memset when second argument is address of zero. The Coccinelle semantic patch used to make this change is as follows: // @eth_zero_addr@ expression e; @@ -memset(e,0x00,ETH_ALEN); +eth_zero_addr(e); // Signed-off-by: Amitoj Kaur Chawla Signed-off-by: Doug Ledford --- drivers/infiniband/core/sa_query.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index f334090..4b3ac76 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -1221,7 +1221,7 @@ static void ib_sa_path_rec_callback(struct ib_sa_query *sa_query, rec.net = NULL; rec.ifindex = 0; rec.gid_type = IB_GID_TYPE_IB; - memset(rec.dmac, 0, ETH_ALEN); + eth_zero_addr(rec.dmac); query->callback(status, &rec, query->context); } else query->callback(status, NULL, query->context); -- cgit v1.1 From aba25a3e96563083389fc7b9f7d35be6bdfe6579 Mon Sep 17 00:00:00 2001 From: Parav Pandit Date: Wed, 2 Mar 2016 00:50:29 +0530 Subject: IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit Reviewed-by: Haggai Eran Reviewed-by: Sagi Grimberg Signed-off-by: Doug Ledford --- drivers/infiniband/core/cache.c | 15 +++++++------- drivers/infiniband/core/cma.c | 14 ++++++------- drivers/infiniband/core/device.c | 28 +++++++++++++------------- drivers/infiniband/core/fmr_pool.c | 37 ++++++++++++++--------------------- drivers/infiniband/core/packer.c | 14 ++++++------- drivers/infiniband/core/sa_query.c | 11 +++++------ drivers/infiniband/core/ucm.c | 8 ++++---- drivers/infiniband/core/ucma.c | 6 +++--- drivers/infiniband/core/ud_header.c | 23 +++++++++++----------- drivers/infiniband/core/uverbs_main.c | 10 +++++----- 10 files changed, 77 insertions(+), 89 deletions(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index 53343ff..cb00d59 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -1043,8 +1043,8 @@ static void ib_cache_update(struct ib_device *device, ret = ib_query_port(device, port, tprops); if (ret) { - printk(KERN_WARNING "ib_query_port failed (%d) for %s\n", - ret, device->name); + pr_warn("ib_query_port failed (%d) for %s\n", + ret, device->name); goto err; } @@ -1067,8 +1067,8 @@ static void ib_cache_update(struct ib_device *device, for (i = 0; i < pkey_cache->table_len; ++i) { ret = ib_query_pkey(device, port, i, pkey_cache->table + i); if (ret) { - printk(KERN_WARNING "ib_query_pkey failed (%d) for %s (index %d)\n", - ret, device->name, i); + pr_warn("ib_query_pkey failed (%d) for %s (index %d)\n", + ret, device->name, i); goto err; } } @@ -1078,8 +1078,8 @@ static void ib_cache_update(struct ib_device *device, ret = ib_query_gid(device, port, i, gid_cache->table + i, NULL); if (ret) { - printk(KERN_WARNING "ib_query_gid failed (%d) for %s (index %d)\n", - ret, device->name, i); + pr_warn("ib_query_gid failed (%d) for %s (index %d)\n", + ret, device->name, i); goto err; } } @@ -1161,8 +1161,7 @@ int ib_cache_setup_one(struct ib_device *device) GFP_KERNEL); if (!device->cache.pkey_cache || !device->cache.lmc_cache) { - printk(KERN_WARNING "Couldn't allocate cache " - "for %s\n", device->name); + pr_warn("Couldn't allocate cache for %s\n", device->name); return -ENOMEM; } diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 9729639..e13121f 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -1713,7 +1713,7 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event) event.param.conn.private_data_len = IB_CM_REJ_PRIVATE_DATA_SIZE; break; default: - printk(KERN_ERR "RDMA CMA: unexpected IB CM event: %d\n", + pr_err("RDMA CMA: unexpected IB CM event: %d\n", ib_event->event); goto out; } @@ -2186,8 +2186,8 @@ static void cma_listen_on_dev(struct rdma_id_private *id_priv, ret = rdma_listen(id, id_priv->backlog); if (ret) - printk(KERN_WARNING "RDMA CMA: cma_listen_on_dev, error %d, " - "listening on device %s\n", ret, cma_dev->device->name); + pr_warn("RDMA CMA: cma_listen_on_dev, error %d, listening on device %s\n", + ret, cma_dev->device->name); } static void cma_listen_on_all(struct rdma_id_private *id_priv) @@ -3239,7 +3239,7 @@ static int cma_sidr_rep_handler(struct ib_cm_id *cm_id, event.status = 0; break; default: - printk(KERN_ERR "RDMA CMA: unexpected IB CM event: %d\n", + pr_err("RDMA CMA: unexpected IB CM event: %d\n", ib_event->event); goto out; } @@ -4003,8 +4003,8 @@ static int cma_netdev_change(struct net_device *ndev, struct rdma_id_private *id if ((dev_addr->bound_dev_if == ndev->ifindex) && (net_eq(dev_net(ndev), dev_addr->net)) && memcmp(dev_addr->src_dev_addr, ndev->dev_addr, ndev->addr_len)) { - printk(KERN_INFO "RDMA CM addr change for ndev %s used by id %p\n", - ndev->name, &id_priv->id); + pr_info("RDMA CM addr change for ndev %s used by id %p\n", + ndev->name, &id_priv->id); work = kzalloc(sizeof *work, GFP_KERNEL); if (!work) return -ENOMEM; @@ -4287,7 +4287,7 @@ static int __init cma_init(void) goto err; if (ibnl_add_client(RDMA_NL_RDMA_CM, RDMA_NL_RDMA_CM_NUM_OPS, cma_cb_table)) - printk(KERN_WARNING "RDMA CMA: failed to add netlink callback\n"); + pr_warn("RDMA CMA: failed to add netlink callback\n"); cma_configfs_init(); return 0; diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c index 94b80a5..270c7ff 100644 --- a/drivers/infiniband/core/device.c +++ b/drivers/infiniband/core/device.c @@ -115,8 +115,8 @@ static int ib_device_check_mandatory(struct ib_device *device) for (i = 0; i < ARRAY_SIZE(mandatory_table); ++i) { if (!*(void **) ((void *) device + mandatory_table[i].offset)) { - printk(KERN_WARNING "Device %s is missing mandatory function %s\n", - device->name, mandatory_table[i].name); + pr_warn("Device %s is missing mandatory function %s\n", + device->name, mandatory_table[i].name); return -EINVAL; } } @@ -255,8 +255,8 @@ static int add_client_context(struct ib_device *device, struct ib_client *client context = kmalloc(sizeof *context, GFP_KERNEL); if (!context) { - printk(KERN_WARNING "Couldn't allocate client context for %s/%s\n", - device->name, client->name); + pr_warn("Couldn't allocate client context for %s/%s\n", + device->name, client->name); return -ENOMEM; } @@ -343,29 +343,29 @@ int ib_register_device(struct ib_device *device, ret = read_port_immutable(device); if (ret) { - printk(KERN_WARNING "Couldn't create per port immutable data %s\n", - device->name); + pr_warn("Couldn't create per port immutable data %s\n", + device->name); goto out; } ret = ib_cache_setup_one(device); if (ret) { - printk(KERN_WARNING "Couldn't set up InfiniBand P_Key/GID cache\n"); + pr_warn("Couldn't set up InfiniBand P_Key/GID cache\n"); goto out; } memset(&device->attrs, 0, sizeof(device->attrs)); ret = device->query_device(device, &device->attrs, &uhw); if (ret) { - printk(KERN_WARNING "Couldn't query the device attributes\n"); + pr_warn("Couldn't query the device attributes\n"); ib_cache_cleanup_one(device); goto out; } ret = ib_device_register_sysfs(device, port_callback); if (ret) { - printk(KERN_WARNING "Couldn't register device %s with driver model\n", - device->name); + pr_warn("Couldn't register device %s with driver model\n", + device->name); ib_cache_cleanup_one(device); goto out; } @@ -566,8 +566,8 @@ void ib_set_client_data(struct ib_device *device, struct ib_client *client, goto out; } - printk(KERN_WARNING "No client context found for %s/%s\n", - device->name, client->name); + pr_warn("No client context found for %s/%s\n", + device->name, client->name); out: spin_unlock_irqrestore(&device->client_data_lock, flags); @@ -960,13 +960,13 @@ static int __init ib_core_init(void) ret = class_register(&ib_class); if (ret) { - printk(KERN_WARNING "Couldn't create InfiniBand device class\n"); + pr_warn("Couldn't create InfiniBand device class\n"); goto err_comp; } ret = ibnl_init(); if (ret) { - printk(KERN_WARNING "Couldn't init IB netlink interface\n"); + pr_warn("Couldn't init IB netlink interface\n"); goto err_sysfs; } diff --git a/drivers/infiniband/core/fmr_pool.c b/drivers/infiniband/core/fmr_pool.c index 6ac3683..cdbb1f1 100644 --- a/drivers/infiniband/core/fmr_pool.c +++ b/drivers/infiniband/core/fmr_pool.c @@ -150,8 +150,8 @@ static void ib_fmr_batch_release(struct ib_fmr_pool *pool) #ifdef DEBUG if (fmr->ref_count !=0) { - printk(KERN_WARNING PFX "Unmapping FMR 0x%08x with ref count %d\n", - fmr, fmr->ref_count); + pr_warn(PFX "Unmapping FMR 0x%08x with ref count %d\n", + fmr, fmr->ref_count); } #endif } @@ -167,7 +167,7 @@ static void ib_fmr_batch_release(struct ib_fmr_pool *pool) ret = ib_unmap_fmr(&fmr_list); if (ret) - printk(KERN_WARNING PFX "ib_unmap_fmr returned %d\n", ret); + pr_warn(PFX "ib_unmap_fmr returned %d\n", ret); spin_lock_irq(&pool->pool_lock); list_splice(&unmap_list, &pool->free_list); @@ -222,8 +222,7 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, device = pd->device; if (!device->alloc_fmr || !device->dealloc_fmr || !device->map_phys_fmr || !device->unmap_fmr) { - printk(KERN_INFO PFX "Device %s does not support FMRs\n", - device->name); + pr_info(PFX "Device %s does not support FMRs\n", device->name); return ERR_PTR(-ENOSYS); } @@ -233,13 +232,10 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, max_remaps = device->attrs.max_map_per_fmr; pool = kmalloc(sizeof *pool, GFP_KERNEL); - if (!pool) { - printk(KERN_WARNING PFX "couldn't allocate pool struct\n"); + if (!pool) return ERR_PTR(-ENOMEM); - } pool->cache_bucket = NULL; - pool->flush_function = params->flush_function; pool->flush_arg = params->flush_arg; @@ -251,7 +247,7 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, kmalloc(IB_FMR_HASH_SIZE * sizeof *pool->cache_bucket, GFP_KERNEL); if (!pool->cache_bucket) { - printk(KERN_WARNING PFX "Failed to allocate cache in pool\n"); + pr_warn(PFX "Failed to allocate cache in pool\n"); ret = -ENOMEM; goto out_free_pool; } @@ -275,7 +271,7 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, "ib_fmr(%s)", device->name); if (IS_ERR(pool->thread)) { - printk(KERN_WARNING PFX "couldn't start cleanup thread\n"); + pr_warn(PFX "couldn't start cleanup thread\n"); ret = PTR_ERR(pool->thread); goto out_free_pool; } @@ -294,11 +290,8 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, for (i = 0; i < params->pool_size; ++i) { fmr = kmalloc(bytes_per_fmr, GFP_KERNEL); - if (!fmr) { - printk(KERN_WARNING PFX "failed to allocate fmr " - "struct for FMR %d\n", i); + if (!fmr) goto out_fail; - } fmr->pool = pool; fmr->remap_count = 0; @@ -307,8 +300,8 @@ struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, fmr->fmr = ib_alloc_fmr(pd, params->access, &fmr_attr); if (IS_ERR(fmr->fmr)) { - printk(KERN_WARNING PFX "fmr_create failed " - "for FMR %d\n", i); + pr_warn(PFX "fmr_create failed for FMR %d\n", + i); kfree(fmr); goto out_fail; } @@ -363,8 +356,8 @@ void ib_destroy_fmr_pool(struct ib_fmr_pool *pool) } if (i < pool->pool_size) - printk(KERN_WARNING PFX "pool still has %d regions registered\n", - pool->pool_size - i); + pr_warn(PFX "pool still has %d regions registered\n", + pool->pool_size - i); kfree(pool->cache_bucket); kfree(pool); @@ -463,7 +456,7 @@ struct ib_pool_fmr *ib_fmr_pool_map_phys(struct ib_fmr_pool *pool_handle, list_add(&fmr->list, &pool->free_list); spin_unlock_irqrestore(&pool->pool_lock, flags); - printk(KERN_WARNING PFX "fmr_map returns %d\n", result); + pr_warn(PFX "fmr_map returns %d\n", result); return ERR_PTR(result); } @@ -517,8 +510,8 @@ int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr) #ifdef DEBUG if (fmr->ref_count < 0) - printk(KERN_WARNING PFX "FMR %p has ref count %d < 0\n", - fmr, fmr->ref_count); + pr_warn(PFX "FMR %p has ref count %d < 0\n", + fmr, fmr->ref_count); #endif spin_unlock_irqrestore(&pool->pool_lock, flags); diff --git a/drivers/infiniband/core/packer.c b/drivers/infiniband/core/packer.c index 1b65986..19b1ee3 100644 --- a/drivers/infiniband/core/packer.c +++ b/drivers/infiniband/core/packer.c @@ -44,7 +44,7 @@ static u64 value_read(int offset, int size, void *structure) case 4: return be32_to_cpup((__be32 *) (structure + offset)); case 8: return be64_to_cpup((__be64 *) (structure + offset)); default: - printk(KERN_WARNING "Field size %d bits not handled\n", size * 8); + pr_warn("Field size %d bits not handled\n", size * 8); return 0; } } @@ -104,9 +104,8 @@ void ib_pack(const struct ib_field *desc, } else { if (desc[i].offset_bits % 8 || desc[i].size_bits % 8) { - printk(KERN_WARNING "Structure field %s of size %d " - "bits is not byte-aligned\n", - desc[i].field_name, desc[i].size_bits); + pr_warn("Structure field %s of size %d bits is not byte-aligned\n", + desc[i].field_name, desc[i].size_bits); } if (desc[i].struct_size_bytes) @@ -132,7 +131,7 @@ static void value_write(int offset, int size, u64 val, void *structure) case 32: *(__be32 *) (structure + offset) = cpu_to_be32(val); break; case 64: *(__be64 *) (structure + offset) = cpu_to_be64(val); break; default: - printk(KERN_WARNING "Field size %d bits not handled\n", size * 8); + pr_warn("Field size %d bits not handled\n", size * 8); } } @@ -188,9 +187,8 @@ void ib_unpack(const struct ib_field *desc, } else { if (desc[i].offset_bits % 8 || desc[i].size_bits % 8) { - printk(KERN_WARNING "Structure field %s of size %d " - "bits is not byte-aligned\n", - desc[i].field_name, desc[i].size_bits); + pr_warn("Structure field %s of size %d bits is not byte-aligned\n", + desc[i].field_name, desc[i].size_bits); } memcpy(structure + desc[i].struct_offset_bytes, diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index 4b3ac76..8e3bf6c 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -864,13 +864,12 @@ static void update_sm_ah(struct work_struct *work) struct ib_ah_attr ah_attr; if (ib_query_port(port->agent->device, port->port_num, &port_attr)) { - printk(KERN_WARNING "Couldn't query port\n"); + pr_warn("Couldn't query port\n"); return; } new_ah = kmalloc(sizeof *new_ah, GFP_KERNEL); if (!new_ah) { - printk(KERN_WARNING "Couldn't allocate new SM AH\n"); return; } @@ -880,7 +879,7 @@ static void update_sm_ah(struct work_struct *work) new_ah->pkey_index = 0; if (ib_find_pkey(port->agent->device, port->port_num, IB_DEFAULT_PKEY_FULL, &new_ah->pkey_index)) - printk(KERN_ERR "Couldn't find index for default PKey\n"); + pr_err("Couldn't find index for default PKey\n"); memset(&ah_attr, 0, sizeof ah_attr); ah_attr.dlid = port_attr.sm_lid; @@ -889,7 +888,7 @@ static void update_sm_ah(struct work_struct *work) new_ah->ah = ib_create_ah(port->agent->qp->pd, &ah_attr); if (IS_ERR(new_ah->ah)) { - printk(KERN_WARNING "Couldn't create new SM AH\n"); + pr_warn("Couldn't create new SM AH\n"); kfree(new_ah); return; } @@ -1800,13 +1799,13 @@ static int __init ib_sa_init(void) ret = ib_register_client(&sa_client); if (ret) { - printk(KERN_ERR "Couldn't register ib_sa client\n"); + pr_err("Couldn't register ib_sa client\n"); goto err1; } ret = mcast_init(); if (ret) { - printk(KERN_ERR "Couldn't initialize multicast handling\n"); + pr_err("Couldn't initialize multicast handling\n"); goto err2; } diff --git a/drivers/infiniband/core/ucm.c b/drivers/infiniband/core/ucm.c index 6b4e8a0..4a9aa04 100644 --- a/drivers/infiniband/core/ucm.c +++ b/drivers/infiniband/core/ucm.c @@ -1234,7 +1234,7 @@ static int find_overflow_devnum(void) ret = alloc_chrdev_region(&overflow_maj, 0, IB_UCM_MAX_DEVICES, "infiniband_cm"); if (ret) { - printk(KERN_ERR "ucm: couldn't register dynamic device number\n"); + pr_err("ucm: couldn't register dynamic device number\n"); return ret; } } @@ -1329,19 +1329,19 @@ static int __init ib_ucm_init(void) ret = register_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_MAX_DEVICES, "infiniband_cm"); if (ret) { - printk(KERN_ERR "ucm: couldn't register device number\n"); + pr_err("ucm: couldn't register device number\n"); goto error1; } ret = class_create_file(&cm_class, &class_attr_abi_version.attr); if (ret) { - printk(KERN_ERR "ucm: couldn't create abi_version attribute\n"); + pr_err("ucm: couldn't create abi_version attribute\n"); goto error2; } ret = ib_register_client(&ucm_client); if (ret) { - printk(KERN_ERR "ucm: couldn't register client\n"); + pr_err("ucm: couldn't register client\n"); goto error3; } return 0; diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 8b5a934..dd3bcce 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -314,7 +314,7 @@ static void ucma_removal_event_handler(struct rdma_cm_id *cm_id) } } if (!event_found) - printk(KERN_ERR "ucma_removal_event_handler: warning: connect request event wasn't found\n"); + pr_err("ucma_removal_event_handler: warning: connect request event wasn't found\n"); } static int ucma_event_handler(struct rdma_cm_id *cm_id, @@ -1716,13 +1716,13 @@ static int __init ucma_init(void) ret = device_create_file(ucma_misc.this_device, &dev_attr_abi_version); if (ret) { - printk(KERN_ERR "rdma_ucm: couldn't create abi_version attr\n"); + pr_err("rdma_ucm: couldn't create abi_version attr\n"); goto err1; } ucma_ctl_table_hdr = register_net_sysctl(&init_net, "net/rdma_ucm", ucma_ctl_table); if (!ucma_ctl_table_hdr) { - printk(KERN_ERR "rdma_ucm: couldn't register sysctl paths\n"); + pr_err("rdma_ucm: couldn't register sysctl paths\n"); ret = -ENOMEM; goto err2; } diff --git a/drivers/infiniband/core/ud_header.c b/drivers/infiniband/core/ud_header.c index 2116132..29a45d2 100644 --- a/drivers/infiniband/core/ud_header.c +++ b/drivers/infiniband/core/ud_header.c @@ -479,8 +479,8 @@ int ib_ud_header_unpack(void *buf, buf += IB_LRH_BYTES; if (header->lrh.link_version != 0) { - printk(KERN_WARNING "Invalid LRH.link_version %d\n", - header->lrh.link_version); + pr_warn("Invalid LRH.link_version %d\n", + header->lrh.link_version); return -EINVAL; } @@ -496,20 +496,20 @@ int ib_ud_header_unpack(void *buf, buf += IB_GRH_BYTES; if (header->grh.ip_version != 6) { - printk(KERN_WARNING "Invalid GRH.ip_version %d\n", - header->grh.ip_version); + pr_warn("Invalid GRH.ip_version %d\n", + header->grh.ip_version); return -EINVAL; } if (header->grh.next_header != 0x1b) { - printk(KERN_WARNING "Invalid GRH.next_header 0x%02x\n", - header->grh.next_header); + pr_warn("Invalid GRH.next_header 0x%02x\n", + header->grh.next_header); return -EINVAL; } break; default: - printk(KERN_WARNING "Invalid LRH.link_next_header %d\n", - header->lrh.link_next_header); + pr_warn("Invalid LRH.link_next_header %d\n", + header->lrh.link_next_header); return -EINVAL; } @@ -525,14 +525,13 @@ int ib_ud_header_unpack(void *buf, header->immediate_present = 1; break; default: - printk(KERN_WARNING "Invalid BTH.opcode 0x%02x\n", - header->bth.opcode); + pr_warn("Invalid BTH.opcode 0x%02x\n", header->bth.opcode); return -EINVAL; } if (header->bth.transport_header_version != 0) { - printk(KERN_WARNING "Invalid BTH.transport_header_version %d\n", - header->bth.transport_header_version); + pr_warn("Invalid BTH.transport_header_version %d\n", + header->bth.transport_header_version); return -EINVAL; } diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 8b299df..28ba2cc 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -1056,7 +1056,7 @@ static int find_overflow_devnum(void) ret = alloc_chrdev_region(&overflow_maj, 0, IB_UVERBS_MAX_DEVICES, "infiniband_verbs"); if (ret) { - printk(KERN_ERR "user_verbs: couldn't register dynamic device number\n"); + pr_err("user_verbs: couldn't register dynamic device number\n"); return ret; } } @@ -1277,14 +1277,14 @@ static int __init ib_uverbs_init(void) ret = register_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES, "infiniband_verbs"); if (ret) { - printk(KERN_ERR "user_verbs: couldn't register device number\n"); + pr_err("user_verbs: couldn't register device number\n"); goto out; } uverbs_class = class_create(THIS_MODULE, "infiniband_verbs"); if (IS_ERR(uverbs_class)) { ret = PTR_ERR(uverbs_class); - printk(KERN_ERR "user_verbs: couldn't create class infiniband_verbs\n"); + pr_err("user_verbs: couldn't create class infiniband_verbs\n"); goto out_chrdev; } @@ -1292,13 +1292,13 @@ static int __init ib_uverbs_init(void) ret = class_create_file(uverbs_class, &class_attr_abi_version.attr); if (ret) { - printk(KERN_ERR "user_verbs: couldn't create abi_version attribute\n"); + pr_err("user_verbs: couldn't create abi_version attribute\n"); goto out_class; } ret = ib_register_client(&uverbs_client); if (ret) { - printk(KERN_ERR "user_verbs: couldn't register client\n"); + pr_err("user_verbs: couldn't register client\n"); goto out_class; } -- cgit v1.1 From f5aa9159a418726d74b67c8815ffd2739afb4c7a Mon Sep 17 00:00:00 2001 From: Sagi Grimberg Date: Mon, 29 Feb 2016 19:07:32 +0200 Subject: IB/core: Add arbitrary sg_list support Devices that are capable in registering SG lists with gaps can now expose it in the core to ULPs using a new device capability IB_DEVICE_SG_GAPS_REG (in a new field device_cap_flags_ex in the device attributes as we ran out of bits), and a new mr_type IB_MR_TYPE_SG_GAPS_REG which allocates a memory region which is capable of handling SG lists with gaps. Signed-off-by: Sagi Grimberg Signed-off-by: Doug Ledford --- drivers/infiniband/core/verbs.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index 5af6d02..16f3fb1 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -1567,6 +1567,8 @@ EXPORT_SYMBOL(ib_check_mr_status); * - The last sg element is allowed to have length less than page_size. * - If sg_nents total byte length exceeds the mr max_num_sge * page_size * then only max_num_sg entries will be mapped. + * - If the MR was allocated with type IB_MR_TYPE_SG_GAPS_REG, non of these + * constraints holds and the page_size argument is ignored. * * Returns the number of sg elements that were mapped to the memory region. * -- cgit v1.1 From b493d91d333e867a043f7ff1397bcba6e2d0dda2 Mon Sep 17 00:00:00 2001 From: Faisal Latif Date: Fri, 26 Feb 2016 09:18:00 -0600 Subject: iwcm: common code for port mapper moved port mapper related code from drivers into common code Signed-off-by: Mustafa Ismail Signed-off-by: Tatyana E. Nikolova Signed-off-by: Faisal Latif Reviewed-by: Steve Wise Tested-by: Steve Wise Signed-off-by: Doug Ledford --- drivers/infiniband/core/iwcm.c | 190 ++++++++++++++++++++++++++++++++----- drivers/infiniband/core/iwpm_msg.c | 2 +- 2 files changed, 165 insertions(+), 27 deletions(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/iwcm.c b/drivers/infiniband/core/iwcm.c index ff9163d..e28a160 100644 --- a/drivers/infiniband/core/iwcm.c +++ b/drivers/infiniband/core/iwcm.c @@ -50,6 +50,8 @@ #include #include +#include +#include #include "iwcm.h" @@ -57,6 +59,16 @@ MODULE_AUTHOR("Tom Tucker"); MODULE_DESCRIPTION("iWARP CM"); MODULE_LICENSE("Dual BSD/GPL"); +static struct ibnl_client_cbs iwcm_nl_cb_table[] = { + [RDMA_NL_IWPM_REG_PID] = {.dump = iwpm_register_pid_cb}, + [RDMA_NL_IWPM_ADD_MAPPING] = {.dump = iwpm_add_mapping_cb}, + [RDMA_NL_IWPM_QUERY_MAPPING] = {.dump = iwpm_add_and_query_mapping_cb}, + [RDMA_NL_IWPM_REMOTE_INFO] = {.dump = iwpm_remote_info_cb}, + [RDMA_NL_IWPM_HANDLE_ERR] = {.dump = iwpm_mapping_error_cb}, + [RDMA_NL_IWPM_MAPINFO] = {.dump = iwpm_mapping_info_cb}, + [RDMA_NL_IWPM_MAPINFO_NUM] = {.dump = iwpm_ack_mapping_info_cb} +}; + static struct workqueue_struct *iwcm_wq; struct iwcm_work { struct work_struct work; @@ -402,6 +414,11 @@ static void destroy_cm_id(struct iw_cm_id *cm_id) } spin_unlock_irqrestore(&cm_id_priv->lock, flags); + if (cm_id->mapped) { + iwpm_remove_mapinfo(&cm_id->local_addr, &cm_id->m_local_addr); + iwpm_remove_mapping(&cm_id->local_addr, RDMA_NL_IWCM); + } + (void)iwcm_deref_id(cm_id_priv); } @@ -426,6 +443,97 @@ void iw_destroy_cm_id(struct iw_cm_id *cm_id) } EXPORT_SYMBOL(iw_destroy_cm_id); +/** + * iw_cm_check_wildcard - If IP address is 0 then use original + * @pm_addr: sockaddr containing the ip to check for wildcard + * @cm_addr: sockaddr containing the actual IP address + * @cm_outaddr: sockaddr to set IP addr which leaving port + * + * Checks the pm_addr for wildcard and then sets cm_outaddr's + * IP to the actual (cm_addr). + */ +static void iw_cm_check_wildcard(struct sockaddr_storage *pm_addr, + struct sockaddr_storage *cm_addr, + struct sockaddr_storage *cm_outaddr) +{ + if (pm_addr->ss_family == AF_INET) { + struct sockaddr_in *pm4_addr = (struct sockaddr_in *)pm_addr; + + if (pm4_addr->sin_addr.s_addr == INADDR_ANY) { + struct sockaddr_in *cm4_addr = + (struct sockaddr_in *)cm_addr; + struct sockaddr_in *cm4_outaddr = + (struct sockaddr_in *)cm_outaddr; + + cm4_outaddr->sin_addr = cm4_addr->sin_addr; + } + } else { + struct sockaddr_in6 *pm6_addr = (struct sockaddr_in6 *)pm_addr; + + if (ipv6_addr_type(&pm6_addr->sin6_addr) == IPV6_ADDR_ANY) { + struct sockaddr_in6 *cm6_addr = + (struct sockaddr_in6 *)cm_addr; + struct sockaddr_in6 *cm6_outaddr = + (struct sockaddr_in6 *)cm_outaddr; + + cm6_outaddr->sin6_addr = cm6_addr->sin6_addr; + } + } +} + +/** + * iw_cm_map - Use portmapper to map the ports + * @cm_id: connection manager pointer + * @active: Indicates the active side when true + * returns nonzero for error only if iwpm_create_mapinfo() fails + * + * Tries to add a mapping for a port using the Portmapper. If + * successful in mapping the IP/Port it will check the remote + * mapped IP address for a wildcard IP address and replace the + * zero IP address with the remote_addr. + */ +static int iw_cm_map(struct iw_cm_id *cm_id, bool active) +{ + struct iwpm_dev_data pm_reg_msg; + struct iwpm_sa_data pm_msg; + int status; + + cm_id->m_local_addr = cm_id->local_addr; + cm_id->m_remote_addr = cm_id->remote_addr; + + memcpy(pm_reg_msg.dev_name, cm_id->device->name, + sizeof(pm_reg_msg.dev_name)); + memcpy(pm_reg_msg.if_name, cm_id->device->iwcm->ifname, + sizeof(pm_reg_msg.if_name)); + + if (iwpm_register_pid(&pm_reg_msg, RDMA_NL_IWCM) || + !iwpm_valid_pid()) + return 0; + + cm_id->mapped = true; + pm_msg.loc_addr = cm_id->local_addr; + pm_msg.rem_addr = cm_id->remote_addr; + if (active) + status = iwpm_add_and_query_mapping(&pm_msg, + RDMA_NL_IWCM); + else + status = iwpm_add_mapping(&pm_msg, RDMA_NL_IWCM); + + if (!status) { + cm_id->m_local_addr = pm_msg.mapped_loc_addr; + if (active) { + cm_id->m_remote_addr = pm_msg.mapped_rem_addr; + iw_cm_check_wildcard(&pm_msg.mapped_rem_addr, + &cm_id->remote_addr, + &cm_id->m_remote_addr); + } + } + + return iwpm_create_mapinfo(&cm_id->local_addr, + &cm_id->m_local_addr, + RDMA_NL_IWCM); +} + /* * CM_ID <-- LISTEN * @@ -452,7 +560,9 @@ int iw_cm_listen(struct iw_cm_id *cm_id, int backlog) case IW_CM_STATE_IDLE: cm_id_priv->state = IW_CM_STATE_LISTEN; spin_unlock_irqrestore(&cm_id_priv->lock, flags); - ret = cm_id->device->iwcm->create_listen(cm_id, backlog); + ret = iw_cm_map(cm_id, false); + if (!ret) + ret = cm_id->device->iwcm->create_listen(cm_id, backlog); if (ret) cm_id_priv->state = IW_CM_STATE_IDLE; spin_lock_irqsave(&cm_id_priv->lock, flags); @@ -582,39 +692,37 @@ int iw_cm_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *iw_param) spin_lock_irqsave(&cm_id_priv->lock, flags); if (cm_id_priv->state != IW_CM_STATE_IDLE) { - spin_unlock_irqrestore(&cm_id_priv->lock, flags); - clear_bit(IWCM_F_CONNECT_WAIT, &cm_id_priv->flags); - wake_up_all(&cm_id_priv->connect_wait); - return -EINVAL; + ret = -EINVAL; + goto err; } /* Get the ib_qp given the QPN */ qp = cm_id->device->iwcm->get_qp(cm_id->device, iw_param->qpn); if (!qp) { - spin_unlock_irqrestore(&cm_id_priv->lock, flags); - clear_bit(IWCM_F_CONNECT_WAIT, &cm_id_priv->flags); - wake_up_all(&cm_id_priv->connect_wait); - return -EINVAL; + ret = -EINVAL; + goto err; } cm_id->device->iwcm->add_ref(qp); cm_id_priv->qp = qp; cm_id_priv->state = IW_CM_STATE_CONN_SENT; spin_unlock_irqrestore(&cm_id_priv->lock, flags); - ret = cm_id->device->iwcm->connect(cm_id, iw_param); - if (ret) { - spin_lock_irqsave(&cm_id_priv->lock, flags); - if (cm_id_priv->qp) { - cm_id->device->iwcm->rem_ref(qp); - cm_id_priv->qp = NULL; - } - spin_unlock_irqrestore(&cm_id_priv->lock, flags); - BUG_ON(cm_id_priv->state != IW_CM_STATE_CONN_SENT); - cm_id_priv->state = IW_CM_STATE_IDLE; - clear_bit(IWCM_F_CONNECT_WAIT, &cm_id_priv->flags); - wake_up_all(&cm_id_priv->connect_wait); - } + ret = iw_cm_map(cm_id, true); + if (!ret) + ret = cm_id->device->iwcm->connect(cm_id, iw_param); + if (!ret) + return 0; /* success */ + spin_lock_irqsave(&cm_id_priv->lock, flags); + if (cm_id_priv->qp) { + cm_id->device->iwcm->rem_ref(qp); + cm_id_priv->qp = NULL; + } + cm_id_priv->state = IW_CM_STATE_IDLE; +err: + spin_unlock_irqrestore(&cm_id_priv->lock, flags); + clear_bit(IWCM_F_CONNECT_WAIT, &cm_id_priv->flags); + wake_up_all(&cm_id_priv->connect_wait); return ret; } EXPORT_SYMBOL(iw_cm_connect); @@ -656,8 +764,23 @@ static void cm_conn_req_handler(struct iwcm_id_private *listen_id_priv, goto out; cm_id->provider_data = iw_event->provider_data; - cm_id->local_addr = iw_event->local_addr; - cm_id->remote_addr = iw_event->remote_addr; + cm_id->m_local_addr = iw_event->local_addr; + cm_id->m_remote_addr = iw_event->remote_addr; + cm_id->local_addr = listen_id_priv->id.local_addr; + + ret = iwpm_get_remote_info(&listen_id_priv->id.m_local_addr, + &iw_event->remote_addr, + &cm_id->remote_addr, + RDMA_NL_IWCM); + if (ret) { + cm_id->remote_addr = iw_event->remote_addr; + } else { + iw_cm_check_wildcard(&listen_id_priv->id.m_local_addr, + &iw_event->local_addr, + &cm_id->local_addr); + iw_event->local_addr = cm_id->local_addr; + iw_event->remote_addr = cm_id->remote_addr; + } cm_id_priv = container_of(cm_id, struct iwcm_id_private, id); cm_id_priv->state = IW_CM_STATE_CONN_RECV; @@ -753,8 +876,10 @@ static int cm_conn_rep_handler(struct iwcm_id_private *cm_id_priv, clear_bit(IWCM_F_CONNECT_WAIT, &cm_id_priv->flags); BUG_ON(cm_id_priv->state != IW_CM_STATE_CONN_SENT); if (iw_event->status == 0) { - cm_id_priv->id.local_addr = iw_event->local_addr; - cm_id_priv->id.remote_addr = iw_event->remote_addr; + cm_id_priv->id.m_local_addr = iw_event->local_addr; + cm_id_priv->id.m_remote_addr = iw_event->remote_addr; + iw_event->local_addr = cm_id_priv->id.local_addr; + iw_event->remote_addr = cm_id_priv->id.remote_addr; cm_id_priv->state = IW_CM_STATE_ESTABLISHED; } else { /* REJECTED or RESET */ @@ -1044,6 +1169,17 @@ EXPORT_SYMBOL(iw_cm_init_qp_attr); static int __init iw_cm_init(void) { + int ret; + + ret = iwpm_init(RDMA_NL_IWCM); + if (ret) + pr_err("iw_cm: couldn't init iwpm\n"); + + ret = ibnl_add_client(RDMA_NL_IWCM, RDMA_NL_IWPM_NUM_OPS, + iwcm_nl_cb_table); + if (ret) + pr_err("iw_cm: couldn't register netlink callbacks\n"); + iwcm_wq = create_singlethread_workqueue("iw_cm_wq"); if (!iwcm_wq) return -ENOMEM; @@ -1063,6 +1199,8 @@ static void __exit iw_cm_cleanup(void) { unregister_net_sysctl_table(iwcm_ctl_table_hdr); destroy_workqueue(iwcm_wq); + ibnl_remove_client(RDMA_NL_IWCM); + iwpm_exit(RDMA_NL_IWCM); } module_init(iw_cm_init); diff --git a/drivers/infiniband/core/iwpm_msg.c b/drivers/infiniband/core/iwpm_msg.c index 22a3abe..c2b4ce6 100644 --- a/drivers/infiniband/core/iwpm_msg.c +++ b/drivers/infiniband/core/iwpm_msg.c @@ -88,7 +88,7 @@ int iwpm_register_pid(struct iwpm_dev_data *pm_msg, u8 nl_client) ret = ibnl_put_attr(skb, nlh, sizeof(u32), &msg_seq, IWPM_NLA_REG_PID_SEQ); if (ret) goto pid_query_error; - ret = ibnl_put_attr(skb, nlh, IWPM_IFNAME_SIZE, + ret = ibnl_put_attr(skb, nlh, IFNAMSIZ, pm_msg->if_name, IWPM_NLA_REG_IF_NAME); if (ret) goto pid_query_error; -- cgit v1.1 From dafb5587178afe8abf85f3ae91bbc88de9e54782 Mon Sep 17 00:00:00 2001 From: Faisal Latif Date: Fri, 26 Feb 2016 09:18:05 -0600 Subject: iwpm: crash fix for large connections test During large connection test, there is a crash at wake_up() in the callback as waitq is not yet initialized. Callback can happen before iwpm_wait_complete_req() is called to initialize waitq. To resolve, using signaling semaphore instead of waitq. Signed-off-by: Mustafa Ismail Reviewed-by: Tatyana E Nikolova Signed-off-by: Faisal Latif Reviewed-by: Steve Wise Tested-by: Steve Wise Signed-off-by: Doug Ledford --- drivers/infiniband/core/iwpm_msg.c | 10 +++++----- drivers/infiniband/core/iwpm_util.c | 14 +++++++------- drivers/infiniband/core/iwpm_util.h | 2 +- 3 files changed, 13 insertions(+), 13 deletions(-) (limited to 'drivers/infiniband/core') diff --git a/drivers/infiniband/core/iwpm_msg.c b/drivers/infiniband/core/iwpm_msg.c index c2b4ce6..43e3fa2 100644 --- a/drivers/infiniband/core/iwpm_msg.c +++ b/drivers/infiniband/core/iwpm_msg.c @@ -89,7 +89,7 @@ int iwpm_register_pid(struct iwpm_dev_data *pm_msg, u8 nl_client) if (ret) goto pid_query_error; ret = ibnl_put_attr(skb, nlh, IFNAMSIZ, - pm_msg->if_name, IWPM_NLA_REG_IF_NAME); + pm_msg->if_name, IWPM_NLA_REG_IF_NAME); if (ret) goto pid_query_error; ret = ibnl_put_attr(skb, nlh, IWPM_DEVNAME_SIZE, @@ -394,7 +394,7 @@ register_pid_response_exit: /* always for found nlmsg_request */ kref_put(&nlmsg_request->kref, iwpm_free_nlmsg_request); barrier(); - wake_up(&nlmsg_request->waitq); + up(&nlmsg_request->sem); return 0; } EXPORT_SYMBOL(iwpm_register_pid_cb); @@ -463,7 +463,7 @@ add_mapping_response_exit: /* always for found request */ kref_put(&nlmsg_request->kref, iwpm_free_nlmsg_request); barrier(); - wake_up(&nlmsg_request->waitq); + up(&nlmsg_request->sem); return 0; } EXPORT_SYMBOL(iwpm_add_mapping_cb); @@ -555,7 +555,7 @@ query_mapping_response_exit: /* always for found request */ kref_put(&nlmsg_request->kref, iwpm_free_nlmsg_request); barrier(); - wake_up(&nlmsg_request->waitq); + up(&nlmsg_request->sem); return 0; } EXPORT_SYMBOL(iwpm_add_and_query_mapping_cb); @@ -749,7 +749,7 @@ int iwpm_mapping_error_cb(struct sk_buff *skb, struct netlink_callback *cb) /* always for found request */ kref_put(&nlmsg_request->kref, iwpm_free_nlmsg_request); barrier(); - wake_up(&nlmsg_request->waitq); + up(&nlmsg_request->sem); return 0; } EXPORT_SYMBOL(iwpm_mapping_error_cb); diff --git a/drivers/infiniband/core/iwpm_util.c b/drivers/infiniband/core/iwpm_util.c index 5fb089e..9b2bf2f 100644 --- a/drivers/infiniband/core/iwpm_util.c +++ b/drivers/infiniband/core/iwpm_util.c @@ -254,9 +254,9 @@ void iwpm_add_remote_info(struct iwpm_remote_info *rem_info) } int iwpm_get_remote_info(struct sockaddr_storage *mapped_loc_addr, - struct sockaddr_storage *mapped_rem_addr, - struct sockaddr_storage *remote_addr, - u8 nl_client) + struct sockaddr_storage *mapped_rem_addr, + struct sockaddr_storage *remote_addr, + u8 nl_client) { struct hlist_node *tmp_hlist_node; struct hlist_head *hash_bucket_head; @@ -322,6 +322,8 @@ struct iwpm_nlmsg_request *iwpm_get_nlmsg_request(__u32 nlmsg_seq, nlmsg_request->nl_client = nl_client; nlmsg_request->request_done = 0; nlmsg_request->err_code = 0; + sema_init(&nlmsg_request->sem, 1); + down(&nlmsg_request->sem); return nlmsg_request; } @@ -364,11 +366,9 @@ struct iwpm_nlmsg_request *iwpm_find_nlmsg_request(__u32 echo_seq) int iwpm_wait_complete_req(struct iwpm_nlmsg_request *nlmsg_request) { int ret; - init_waitqueue_head(&nlmsg_request->waitq); - ret = wait_event_timeout(nlmsg_request->waitq, - (nlmsg_request->request_done != 0), IWPM_NL_TIMEOUT); - if (!ret) { + ret = down_timeout(&nlmsg_request->sem, IWPM_NL_TIMEOUT); + if (ret) { ret = -EINVAL; pr_info("%s: Timeout %d sec for netlink request (seq = %u)\n", __func__, (IWPM_NL_TIMEOUT/HZ), nlmsg_request->nlmsg_seq); diff --git a/drivers/infiniband/core/iwpm_util.h b/drivers/infiniband/core/iwpm_util.h index b7b9e19..af1fc14 100644 --- a/drivers/infiniband/core/iwpm_util.h +++ b/drivers/infiniband/core/iwpm_util.h @@ -69,7 +69,7 @@ struct iwpm_nlmsg_request { u8 nl_client; u8 request_done; u16 err_code; - wait_queue_head_t waitq; + struct semaphore sem; struct kref kref; }; -- cgit v1.1