diff options
Diffstat (limited to 'drivers/infiniband')
301 files changed, 25705 insertions, 16216 deletions
diff --git a/drivers/infiniband/Kconfig b/drivers/infiniband/Kconfig index ade86388434f..91b023341b77 100644 --- a/drivers/infiniband/Kconfig +++ b/drivers/infiniband/Kconfig @@ -8,7 +8,7 @@ menuconfig INFINIBAND depends on !ALPHA select IRQ_POLL select DIMLIB - ---help--- + help Core support for InfiniBand (IB). Make sure to also select any protocols you wish to use as well as drivers for your InfiniBand hardware. @@ -18,7 +18,7 @@ if INFINIBAND config INFINIBAND_USER_MAD tristate "InfiniBand userspace MAD support" depends on INFINIBAND - ---help--- + help Userspace InfiniBand Management Datagram (MAD) support. This is the kernel side of the userspace MAD support, which allows userspace processes to send and receive MADs. You will also @@ -28,7 +28,7 @@ config INFINIBAND_USER_MAD config INFINIBAND_USER_ACCESS tristate "InfiniBand userspace access (verbs and CM)" depends on MMU - ---help--- + help Userspace InfiniBand access support. This enables the kernel side of userspace verbs and the userspace communication manager (CM). This allows userspace processes @@ -37,14 +37,6 @@ config INFINIBAND_USER_ACCESS libibverbs, libibcm and a hardware driver library from rdma-core <https://github.com/linux-rdma/rdma-core>. -config INFINIBAND_EXP_LEGACY_VERBS_NEW_UAPI - bool "Allow experimental legacy verbs in new ioctl uAPI (EXPERIMENTAL)" - depends on INFINIBAND_USER_ACCESS - ---help--- - IOCTL based uAPI support for Infiniband is enabled by default for - new verbs only. This allows userspace to invoke the IOCTL based uAPI - for current legacy verbs too. - config INFINIBAND_USER_MEM bool depends on INFINIBAND_USER_ACCESS != n @@ -57,7 +49,7 @@ config INFINIBAND_ON_DEMAND_PAGING select MMU_NOTIFIER select INTERVAL_TREE default y - ---help--- + help On demand paging support for the InfiniBand subsystem. Together with driver support this allows registration of memory regions without pinning their pages, fetching the @@ -67,7 +59,7 @@ config INFINIBAND_ADDR_TRANS bool "RDMA/CM" depends on INFINIBAND default y - ---help--- + help Support for RDMA communication manager (CM). This allows for a generic connection abstraction over RDMA. @@ -75,7 +67,7 @@ config INFINIBAND_ADDR_TRANS_CONFIGFS bool depends on INFINIBAND_ADDR_TRANS && CONFIGFS_FS && !(INFINIBAND=y && CONFIGFS_FS=m) default y - ---help--- + help ConfigFS support for RDMA communication manager (CM). This allows the user to config the default GID type that the CM uses for each device, when initiaing new connections. @@ -107,6 +99,7 @@ source "drivers/infiniband/ulp/srpt/Kconfig" source "drivers/infiniband/ulp/iser/Kconfig" source "drivers/infiniband/ulp/isert/Kconfig" +source "drivers/infiniband/ulp/rtrs/Kconfig" source "drivers/infiniband/ulp/opa_vnic/Kconfig" diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile index d1b14887960e..24cb71a16a28 100644 --- a/drivers/infiniband/core/Makefile +++ b/drivers/infiniband/core/Makefile @@ -8,11 +8,11 @@ obj-$(CONFIG_INFINIBAND_USER_MAD) += ib_umad.o obj-$(CONFIG_INFINIBAND_USER_ACCESS) += ib_uverbs.o $(user_access-y) ib_core-y := packer.o ud_header.o verbs.o cq.o rw.o sysfs.o \ - device.o fmr_pool.o cache.o netlink.o \ + device.o cache.o netlink.o \ roce_gid_mgmt.o mr_pool.o addr.o sa_query.o \ multicast.o mad.o smi.o agent.o mad_rmpp.o \ nldev.o restrack.o counters.o ib_core_uverbs.o \ - trace.o + trace.o lag.o ib_core-$(CONFIG_SECURITY_INFINIBAND) += security.o ib_core-$(CONFIG_CGROUP_RDMA) += cgroup.o @@ -36,6 +36,9 @@ ib_uverbs-y := uverbs_main.o uverbs_cmd.o uverbs_marshall.o \ uverbs_std_types_flow_action.o uverbs_std_types_dm.o \ uverbs_std_types_mr.o uverbs_std_types_counters.o \ uverbs_uapi.o uverbs_std_types_device.o \ - uverbs_std_types_async_fd.o + uverbs_std_types_async_fd.o \ + uverbs_std_types_srq.o \ + uverbs_std_types_wq.o \ + uverbs_std_types_qp.o ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o ib_uverbs-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index 1753a9801b70..3a98439bba83 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -371,6 +371,8 @@ static int fetch_ha(const struct dst_entry *dst, struct rdma_dev_addr *dev_addr, (const void *)&dst_in6->sin6_addr; sa_family_t family = dst_in->sa_family; + might_sleep(); + /* If we have a gateway in IB mode then it must be an IB network */ if (has_gateway(dst, family) && dev_addr->network == RDMA_NETWORK_IB) return ib_nl_fetch_ha(dev_addr, daddr, seq, family); @@ -727,6 +729,8 @@ int roce_resolve_route_from_path(struct sa_path_rec *rec, struct rdma_dev_addr dev_addr = {}; int ret; + might_sleep(); + if (rec->roce.route_resolved) return 0; diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index a670209bbce6..5a76611e684a 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -1054,7 +1054,7 @@ int ib_get_cached_pkey(struct ib_device *device, cache = device->port_data[port_num].cache.pkey; - if (index < 0 || index >= cache->table_len) + if (!cache || index < 0 || index >= cache->table_len) ret = -EINVAL; else *pkey = cache->table[index]; @@ -1099,6 +1099,10 @@ int ib_find_cached_pkey(struct ib_device *device, read_lock_irqsave(&device->cache_lock, flags); cache = device->port_data[port_num].cache.pkey; + if (!cache) { + ret = -EINVAL; + goto err; + } *index = -1; @@ -1117,6 +1121,7 @@ int ib_find_cached_pkey(struct ib_device *device, ret = 0; } +err: read_unlock_irqrestore(&device->cache_lock, flags); return ret; @@ -1139,6 +1144,10 @@ int ib_find_exact_cached_pkey(struct ib_device *device, read_lock_irqsave(&device->cache_lock, flags); cache = device->port_data[port_num].cache.pkey; + if (!cache) { + ret = -EINVAL; + goto err; + } *index = -1; @@ -1149,6 +1158,7 @@ int ib_find_exact_cached_pkey(struct ib_device *device, break; } +err: read_unlock_irqrestore(&device->cache_lock, flags); return ret; @@ -1310,9 +1320,10 @@ struct net_device *rdma_read_gid_attr_ndev_rcu(const struct ib_gid_attr *attr) } EXPORT_SYMBOL(rdma_read_gid_attr_ndev_rcu); -static int get_lower_dev_vlan(struct net_device *lower_dev, void *data) +static int get_lower_dev_vlan(struct net_device *lower_dev, + struct netdev_nested_priv *priv) { - u16 *vlan_id = data; + u16 *vlan_id = (u16 *)priv->data; if (is_vlan_dev(lower_dev)) *vlan_id = vlan_dev_vlan_id(lower_dev); @@ -1338,6 +1349,9 @@ static int get_lower_dev_vlan(struct net_device *lower_dev, void *data) int rdma_read_gid_l2_fields(const struct ib_gid_attr *attr, u16 *vlan_id, u8 *smac) { + struct netdev_nested_priv priv = { + .data = (void *)vlan_id, + }; struct net_device *ndev; rcu_read_lock(); @@ -1358,7 +1372,7 @@ int rdma_read_gid_l2_fields(const struct ib_gid_attr *attr, * the lower vlan device for this gid entry. */ netdev_walk_all_lower_dev_rcu(attr->ndev, - get_lower_dev_vlan, vlan_id); + get_lower_dev_vlan, &priv); } } rcu_read_unlock(); @@ -1425,23 +1439,26 @@ ib_cache_update(struct ib_device *device, u8 port, bool enforce_security) goto err; } - pkey_cache = kmalloc(struct_size(pkey_cache, table, - tprops->pkey_tbl_len), - GFP_KERNEL); - if (!pkey_cache) { - ret = -ENOMEM; - goto err; - } + if (tprops->pkey_tbl_len) { + pkey_cache = kmalloc(struct_size(pkey_cache, table, + tprops->pkey_tbl_len), + GFP_KERNEL); + if (!pkey_cache) { + ret = -ENOMEM; + goto err; + } - pkey_cache->table_len = tprops->pkey_tbl_len; + pkey_cache->table_len = tprops->pkey_tbl_len; - for (i = 0; i < pkey_cache->table_len; ++i) { - ret = ib_query_pkey(device, port, i, pkey_cache->table + i); - if (ret) { - dev_warn(&device->dev, - "ib_query_pkey failed (%d) for index %d\n", - ret, i); - goto err; + for (i = 0; i < pkey_cache->table_len; ++i) { + ret = ib_query_pkey(device, port, i, + pkey_cache->table + i); + if (ret) { + dev_warn(&device->dev, + "ib_query_pkey failed (%d) for index %d\n", + ret, i); + goto err; + } } } diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index 17f14e0eafe4..fbc28f1a8b92 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -66,6 +66,8 @@ static const char * const ibcm_rej_reason_strs[] = { [IB_CM_REJ_INVALID_CLASS_VERSION] = "invalid class version", [IB_CM_REJ_INVALID_FLOW_LABEL] = "invalid flow label", [IB_CM_REJ_INVALID_ALT_FLOW_LABEL] = "invalid alt flow label", + [IB_CM_REJ_VENDOR_OPTION_NOT_SUPPORTED] = + "vendor option is not supported", }; const char *__attribute_const__ ibcm_reject_msg(int reason) @@ -81,8 +83,11 @@ const char *__attribute_const__ ibcm_reject_msg(int reason) EXPORT_SYMBOL(ibcm_reject_msg); struct cm_id_private; -static void cm_add_one(struct ib_device *device); +struct cm_work; +static int cm_add_one(struct ib_device *device); static void cm_remove_one(struct ib_device *device, void *client_data); +static void cm_process_work(struct cm_id_private *cm_id_priv, + struct cm_work *work); static int cm_send_sidr_rep_locked(struct cm_id_private *cm_id_priv, struct ib_cm_sidr_rep_param *param); static int cm_send_dreq_locked(struct cm_id_private *cm_id_priv, @@ -287,6 +292,8 @@ struct cm_id_private { struct list_head work_list; atomic_t work_count; + + struct rdma_ucm_ece ece; }; static void cm_work_handler(struct work_struct *work); @@ -474,24 +481,19 @@ static int cm_init_av_for_response(struct cm_port *port, struct ib_wc *wc, grh, &av->ah_attr); } -static int add_cm_id_to_port_list(struct cm_id_private *cm_id_priv, - struct cm_av *av, - struct cm_port *port) +static void add_cm_id_to_port_list(struct cm_id_private *cm_id_priv, + struct cm_av *av, struct cm_port *port) { unsigned long flags; - int ret = 0; spin_lock_irqsave(&cm.lock, flags); - if (&cm_id_priv->av == av) list_add_tail(&cm_id_priv->prim_list, &port->cm_priv_prim_list); else if (&cm_id_priv->alt_av == av) list_add_tail(&cm_id_priv->altr_list, &port->cm_priv_altr_list); else - ret = -EINVAL; - + WARN_ON(true); spin_unlock_irqrestore(&cm.lock, flags); - return ret; } static struct cm_port * @@ -572,12 +574,7 @@ static int cm_init_av_by_path(struct sa_path_rec *path, return ret; av->timeout = path->packet_life_time + 1; - - ret = add_cm_id_to_port_list(cm_id_priv, av, port); - if (ret) { - rdma_destroy_ah_attr(&new_ah_attr); - return ret; - } + add_cm_id_to_port_list(cm_id_priv, av, port); rdma_move_ah_attr(&av->ah_attr, &new_ah_attr); return 0; } @@ -587,11 +584,6 @@ static u32 cm_local_id(__be32 local_id) return (__force u32) (local_id ^ cm.random_id_operand); } -static void cm_free_id(__be32 local_id) -{ - xa_erase_irq(&cm.local_id_table, cm_local_id(local_id)); -} - static struct cm_id_private *cm_acquire_id(__be32 local_id, __be32 remote_id) { struct cm_id_private *cm_id_priv; @@ -698,9 +690,10 @@ static struct cm_id_private * cm_find_listen(struct ib_device *device, cm_id_priv = rb_entry(node, struct cm_id_private, service_node); if ((cm_id_priv->id.service_mask & service_id) == cm_id_priv->id.service_id && - (cm_id_priv->id.device == device)) + (cm_id_priv->id.device == device)) { + refcount_inc(&cm_id_priv->refcount); return cm_id_priv; - + } if (device < cm_id_priv->id.device) node = node->rb_left; else if (device > cm_id_priv->id.device) @@ -745,12 +738,14 @@ static struct cm_timewait_info * cm_insert_remote_id(struct cm_timewait_info return NULL; } -static struct cm_timewait_info * cm_find_remote_id(__be64 remote_ca_guid, - __be32 remote_id) +static struct cm_id_private *cm_find_remote_id(__be64 remote_ca_guid, + __be32 remote_id) { struct rb_node *node = cm.remote_id_table.rb_node; struct cm_timewait_info *timewait_info; + struct cm_id_private *res = NULL; + spin_lock_irq(&cm.lock); while (node) { timewait_info = rb_entry(node, struct cm_timewait_info, remote_id_node); @@ -762,10 +757,14 @@ static struct cm_timewait_info * cm_find_remote_id(__be64 remote_ca_guid, node = node->rb_left; else if (be64_gt(remote_ca_guid, timewait_info->remote_ca_guid)) node = node->rb_right; - else - return timewait_info; + else { + res = cm_acquire_id(timewait_info->work.local_id, + timewait_info->work.remote_id); + break; + } } - return NULL; + spin_unlock_irq(&cm.lock); + return res; } static struct cm_timewait_info * cm_insert_remote_qpn(struct cm_timewait_info @@ -917,6 +916,36 @@ static void cm_free_work(struct cm_work *work) kfree(work); } +static void cm_queue_work_unlock(struct cm_id_private *cm_id_priv, + struct cm_work *work) + __releases(&cm_id_priv->lock) +{ + bool immediate; + + /* + * To deliver the event to the user callback we have the drop the + * spinlock, however, we need to ensure that the user callback is single + * threaded and receives events in the temporal order. If there are + * already events being processed then thread new events onto a list, + * the thread currently processing will pick them up. + */ + immediate = atomic_inc_and_test(&cm_id_priv->work_count); + if (!immediate) { + list_add_tail(&work->list, &cm_id_priv->work_list); + /* + * This routine always consumes incoming reference. Once queued + * to the work_list then a reference is held by the thread + * currently running cm_process_work() and this reference is not + * needed. + */ + cm_deref_id(cm_id_priv); + } + spin_unlock_irq(&cm_id_priv->lock); + + if (immediate) + cm_process_work(cm_id_priv, work); +} + static inline int cm_convert_to_ms(int iba_time) { /* approximate conversion to ms from 4.096us x 2^iba_time */ @@ -942,8 +971,10 @@ static u8 cm_ack_timeout(u8 ca_ack_delay, u8 packet_life_time) return min(31, ack_timeout); } -static void cm_cleanup_timewait(struct cm_timewait_info *timewait_info) +static void cm_remove_remote(struct cm_id_private *cm_id_priv) { + struct cm_timewait_info *timewait_info = cm_id_priv->timewait_info; + if (timewait_info->inserted_remote_id) { rb_erase(&timewait_info->remote_id_node, &cm.remote_id_table); timewait_info->inserted_remote_id = 0; @@ -982,7 +1013,7 @@ static void cm_enter_timewait(struct cm_id_private *cm_id_priv) return; spin_lock_irqsave(&cm.lock, flags); - cm_cleanup_timewait(cm_id_priv->timewait_info); + cm_remove_remote(cm_id_priv); list_add_tail(&cm_id_priv->timewait_info->list, &cm.timewait_list); spin_unlock_irqrestore(&cm.lock, flags); @@ -1001,6 +1032,11 @@ static void cm_enter_timewait(struct cm_id_private *cm_id_priv) msecs_to_jiffies(wait_time)); spin_unlock_irqrestore(&cm.lock, flags); + /* + * The timewait_info is converted into a work and gets freed during + * cm_free_work() in cm_timewait_handler(). + */ + BUILD_BUG_ON(offsetof(struct cm_timewait_info, work) != 0); cm_id_priv->timewait_info = NULL; } @@ -1013,7 +1049,7 @@ static void cm_reset_to_idle(struct cm_id_private *cm_id_priv) cm_id_priv->id.state = IB_CM_IDLE; if (cm_id_priv->timewait_info) { spin_lock_irqsave(&cm.lock, flags); - cm_cleanup_timewait(cm_id_priv->timewait_info); + cm_remove_remote(cm_id_priv); spin_unlock_irqrestore(&cm.lock, flags); kfree(cm_id_priv->timewait_info); cm_id_priv->timewait_info = NULL; @@ -1076,7 +1112,9 @@ retest: case IB_CM_REP_SENT: case IB_CM_MRA_REP_RCVD: ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - /* Fall through */ + cm_send_rej_locked(cm_id_priv, IB_CM_REJ_CONSUMER_DEFINED, NULL, + 0, NULL, 0); + goto retest; case IB_CM_MRA_REQ_SENT: case IB_CM_REP_RCVD: case IB_CM_MRA_REP_SENT: @@ -1101,7 +1139,7 @@ retest: case IB_CM_TIMEWAIT: /* * The cm_acquire_id in cm_timewait_handler will stop working - * once we do cm_free_id() below, so just move to idle here for + * once we do xa_erase below, so just move to idle here for * consistency. */ cm_id->state = IB_CM_IDLE; @@ -1114,7 +1152,7 @@ retest: spin_lock(&cm.lock); /* Required for cleanup paths related cm_req_handler() */ if (cm_id_priv->timewait_info) { - cm_cleanup_timewait(cm_id_priv->timewait_info); + cm_remove_remote(cm_id_priv); kfree(cm_id_priv->timewait_info); cm_id_priv->timewait_info = NULL; } @@ -1131,7 +1169,7 @@ retest: spin_unlock(&cm.lock); spin_unlock_irq(&cm_id_priv->lock); - cm_free_id(cm_id->local_id); + xa_erase_irq(&cm.local_id_table, cm_local_id(cm_id->local_id)); cm_deref_id(cm_id_priv); wait_for_completion(&cm_id_priv->comp); while ((work = cm_dequeue_work(cm_id_priv)) != NULL) @@ -1287,6 +1325,13 @@ static void cm_format_mad_hdr(struct ib_mad_hdr *hdr, hdr->tid = tid; } +static void cm_format_mad_ece_hdr(struct ib_mad_hdr *hdr, __be16 attr_id, + __be64 tid, u32 attr_mod) +{ + cm_format_mad_hdr(hdr, attr_id, tid); + hdr->attr_mod = cpu_to_be32(attr_mod); +} + static void cm_format_req(struct cm_req_msg *req_msg, struct cm_id_private *cm_id_priv, struct ib_cm_req_param *param) @@ -1299,8 +1344,8 @@ static void cm_format_req(struct cm_req_msg *req_msg, pri_ext = opa_is_extended_lid(pri_path->opa.dlid, pri_path->opa.slid); - cm_format_mad_hdr(&req_msg->hdr, CM_REQ_ATTR_ID, - cm_form_tid(cm_id_priv)); + cm_format_mad_ece_hdr(&req_msg->hdr, CM_REQ_ATTR_ID, + cm_form_tid(cm_id_priv), param->ece.attr_mod); IBA_SET(CM_REQ_LOCAL_COMM_ID, req_msg, be32_to_cpu(cm_id_priv->id.local_id)); @@ -1423,6 +1468,7 @@ static void cm_format_req(struct cm_req_msg *req_msg, cm_ack_timeout(cm_id_priv->av.port->cm_dev->ack_delay, alt_path->packet_life_time)); } + IBA_SET(CM_REQ_VENDOR_ID, req_msg, param->ece.vendor_id); if (param->private_data && param->private_data_len) IBA_SET_MEM(CM_REQ_PRIVATE_DATA, req_msg, param->private_data, @@ -1779,6 +1825,9 @@ static void cm_format_req_event(struct cm_work *work, param->rnr_retry_count = IBA_GET(CM_REQ_RNR_RETRY_COUNT, req_msg); param->srq = IBA_GET(CM_REQ_SRQ, req_msg); param->ppath_sgid_attr = cm_id_priv->av.ah_attr.grh.sgid_attr; + param->ece.vendor_id = IBA_GET(CM_REQ_VENDOR_ID, req_msg); + param->ece.attr_mod = be32_to_cpu(req_msg->hdr.attr_mod); + work->cm_event.private_data = IBA_GET_MEM_PTR(CM_REQ_PRIVATE_DATA, req_msg); } @@ -1927,7 +1976,6 @@ static struct cm_id_private * cm_match_req(struct cm_work *work, struct cm_id_private *listen_cm_id_priv, *cur_cm_id_priv; struct cm_timewait_info *timewait_info; struct cm_req_msg *req_msg; - struct ib_cm_id *cm_id; req_msg = (struct cm_req_msg *)work->mad_recv_wc->recv_buf.mad; @@ -1948,7 +1996,7 @@ static struct cm_id_private * cm_match_req(struct cm_work *work, /* Check for stale connections. */ timewait_info = cm_insert_remote_qpn(cm_id_priv->timewait_info); if (timewait_info) { - cm_cleanup_timewait(cm_id_priv->timewait_info); + cm_remove_remote(cm_id_priv); cur_cm_id_priv = cm_acquire_id(timewait_info->work.local_id, timewait_info->work.remote_id); @@ -1957,8 +2005,7 @@ static struct cm_id_private * cm_match_req(struct cm_work *work, IB_CM_REJ_STALE_CONN, CM_MSG_RESPONSE_REQ, NULL, 0); if (cur_cm_id_priv) { - cm_id = &cur_cm_id_priv->id; - ib_send_cm_dreq(cm_id, NULL, 0); + ib_send_cm_dreq(&cur_cm_id_priv->id, NULL, 0); cm_deref_id(cur_cm_id_priv); } return NULL; @@ -1969,14 +2016,13 @@ static struct cm_id_private * cm_match_req(struct cm_work *work, cm_id_priv->id.device, cpu_to_be64(IBA_GET(CM_REQ_SERVICE_ID, req_msg))); if (!listen_cm_id_priv) { - cm_cleanup_timewait(cm_id_priv->timewait_info); + cm_remove_remote(cm_id_priv); spin_unlock_irq(&cm.lock); cm_issue_rej(work->port, work->mad_recv_wc, IB_CM_REJ_INVALID_SERVICE_ID, CM_MSG_RESPONSE_REQ, NULL, 0); return NULL; } - refcount_inc(&listen_cm_id_priv->refcount); spin_unlock_irq(&cm.lock); return listen_cm_id_priv; } @@ -2153,9 +2199,7 @@ static int cm_req_handler(struct cm_work *work) /* Refcount belongs to the event, pairs with cm_process_work() */ refcount_inc(&cm_id_priv->refcount); - atomic_inc(&cm_id_priv->work_count); - spin_unlock_irq(&cm_id_priv->lock); - cm_process_work(cm_id_priv, work); + cm_queue_work_unlock(cm_id_priv, work); /* * Since this ID was just created and was not made visible to other MAD * handlers until the cm_finalize_id() above we know that the @@ -2176,7 +2220,8 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg, struct cm_id_private *cm_id_priv, struct ib_cm_rep_param *param) { - cm_format_mad_hdr(&rep_msg->hdr, CM_REP_ATTR_ID, cm_id_priv->tid); + cm_format_mad_ece_hdr(&rep_msg->hdr, CM_REP_ATTR_ID, cm_id_priv->tid, + param->ece.attr_mod); IBA_SET(CM_REP_LOCAL_COMM_ID, rep_msg, be32_to_cpu(cm_id_priv->id.local_id)); IBA_SET(CM_REP_REMOTE_COMM_ID, rep_msg, @@ -2203,6 +2248,10 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg, IBA_SET(CM_REP_LOCAL_EE_CONTEXT_NUMBER, rep_msg, param->qp_num); } + IBA_SET(CM_REP_VENDOR_ID_L, rep_msg, param->ece.vendor_id); + IBA_SET(CM_REP_VENDOR_ID_M, rep_msg, param->ece.vendor_id >> 8); + IBA_SET(CM_REP_VENDOR_ID_H, rep_msg, param->ece.vendor_id >> 16); + if (param->private_data && param->private_data_len) IBA_SET_MEM(CM_REP_PRIVATE_DATA, rep_msg, param->private_data, param->private_data_len); @@ -2350,6 +2399,11 @@ static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type) param->flow_control = IBA_GET(CM_REP_END_TO_END_FLOW_CONTROL, rep_msg); param->rnr_retry_count = IBA_GET(CM_REP_RNR_RETRY_COUNT, rep_msg); param->srq = IBA_GET(CM_REP_SRQ, rep_msg); + param->ece.vendor_id = IBA_GET(CM_REP_VENDOR_ID_H, rep_msg) << 16; + param->ece.vendor_id |= IBA_GET(CM_REP_VENDOR_ID_M, rep_msg) << 8; + param->ece.vendor_id |= IBA_GET(CM_REP_VENDOR_ID_L, rep_msg); + param->ece.attr_mod = be32_to_cpu(rep_msg->hdr.attr_mod); + work->cm_event.private_data = IBA_GET_MEM_PTR(CM_REP_PRIVATE_DATA, rep_msg); } @@ -2404,7 +2458,6 @@ static int cm_rep_handler(struct cm_work *work) struct cm_rep_msg *rep_msg; int ret; struct cm_id_private *cur_cm_id_priv; - struct ib_cm_id *cm_id; struct cm_timewait_info *timewait_info; rep_msg = (struct cm_rep_msg *)work->mad_recv_wc->recv_buf.mad; @@ -2454,9 +2507,7 @@ static int cm_rep_handler(struct cm_work *work) /* Check for a stale connection. */ timewait_info = cm_insert_remote_qpn(cm_id_priv->timewait_info); if (timewait_info) { - rb_erase(&cm_id_priv->timewait_info->remote_id_node, - &cm.remote_id_table); - cm_id_priv->timewait_info->inserted_remote_id = 0; + cm_remove_remote(cm_id_priv); cur_cm_id_priv = cm_acquire_id(timewait_info->work.local_id, timewait_info->work.remote_id); @@ -2472,8 +2523,7 @@ static int cm_rep_handler(struct cm_work *work) IBA_GET(CM_REP_REMOTE_COMM_ID, rep_msg)); if (cur_cm_id_priv) { - cm_id = &cur_cm_id_priv->id; - ib_send_cm_dreq(cm_id, NULL, 0); + ib_send_cm_dreq(&cur_cm_id_priv->id, NULL, 0); cm_deref_id(cur_cm_id_priv); } @@ -2501,15 +2551,7 @@ static int cm_rep_handler(struct cm_work *work) cm_id_priv->alt_av.timeout - 1); ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; error: @@ -2520,7 +2562,6 @@ error: static int cm_establish_handler(struct cm_work *work) { struct cm_id_private *cm_id_priv; - int ret; /* See comment in cm_establish about lookup. */ cm_id_priv = cm_acquire_id(work->local_id, work->remote_id); @@ -2534,15 +2575,7 @@ static int cm_establish_handler(struct cm_work *work) } ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: cm_deref_id(cm_id_priv); @@ -2553,7 +2586,6 @@ static int cm_rtu_handler(struct cm_work *work) { struct cm_id_private *cm_id_priv; struct cm_rtu_msg *rtu_msg; - int ret; rtu_msg = (struct cm_rtu_msg *)work->mad_recv_wc->recv_buf.mad; cm_id_priv = cm_acquire_id( @@ -2576,15 +2608,7 @@ static int cm_rtu_handler(struct cm_work *work) cm_id_priv->id.state = IB_CM_ESTABLISHED; ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: cm_deref_id(cm_id_priv); @@ -2777,7 +2801,6 @@ static int cm_dreq_handler(struct cm_work *work) struct cm_id_private *cm_id_priv; struct cm_dreq_msg *dreq_msg; struct ib_mad_send_buf *msg = NULL; - int ret; dreq_msg = (struct cm_dreq_msg *)work->mad_recv_wc->recv_buf.mad; cm_id_priv = cm_acquire_id( @@ -2842,15 +2865,7 @@ static int cm_dreq_handler(struct cm_work *work) } cm_id_priv->id.state = IB_CM_DREQ_RCVD; cm_id_priv->tid = dreq_msg->hdr.tid; - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; unlock: spin_unlock_irq(&cm_id_priv->lock); @@ -2862,7 +2877,6 @@ static int cm_drep_handler(struct cm_work *work) { struct cm_id_private *cm_id_priv; struct cm_drep_msg *drep_msg; - int ret; drep_msg = (struct cm_drep_msg *)work->mad_recv_wc->recv_buf.mad; cm_id_priv = cm_acquire_id( @@ -2883,15 +2897,7 @@ static int cm_drep_handler(struct cm_work *work) cm_enter_timewait(cm_id_priv); ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: cm_deref_id(cm_id_priv); @@ -2987,24 +2993,15 @@ static void cm_format_rej_event(struct cm_work *work) static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg) { - struct cm_timewait_info *timewait_info; struct cm_id_private *cm_id_priv; __be32 remote_id; remote_id = cpu_to_be32(IBA_GET(CM_REJ_LOCAL_COMM_ID, rej_msg)); if (IBA_GET(CM_REJ_REASON, rej_msg) == IB_CM_REJ_TIMEOUT) { - spin_lock_irq(&cm.lock); - timewait_info = cm_find_remote_id( + cm_id_priv = cm_find_remote_id( *((__be64 *)IBA_GET_MEM_PTR(CM_REJ_ARI, rej_msg)), remote_id); - if (!timewait_info) { - spin_unlock_irq(&cm.lock); - return NULL; - } - cm_id_priv = - cm_acquire_id(timewait_info->work.local_id, remote_id); - spin_unlock_irq(&cm.lock); } else if (IBA_GET(CM_REJ_MESSAGE_REJECTED, rej_msg) == CM_MSG_RESPONSE_REQ) cm_id_priv = cm_acquire_id( @@ -3022,7 +3019,6 @@ static int cm_rej_handler(struct cm_work *work) { struct cm_id_private *cm_id_priv; struct cm_rej_msg *rej_msg; - int ret; rej_msg = (struct cm_rej_msg *)work->mad_recv_wc->recv_buf.mad; cm_id_priv = cm_acquire_rejected_id(rej_msg); @@ -3038,7 +3034,7 @@ static int cm_rej_handler(struct cm_work *work) case IB_CM_REP_SENT: case IB_CM_MRA_REP_RCVD: ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - /* fall through */ + fallthrough; case IB_CM_REQ_RCVD: case IB_CM_MRA_REQ_SENT: if (IBA_GET(CM_REJ_REASON, rej_msg) == IB_CM_REJ_STALE_CONN) @@ -3048,7 +3044,7 @@ static int cm_rej_handler(struct cm_work *work) break; case IB_CM_DREQ_SENT: ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); - /* fall through */ + fallthrough; case IB_CM_REP_RCVD: case IB_CM_MRA_REP_SENT: cm_enter_timewait(cm_id_priv); @@ -3062,25 +3058,16 @@ static int cm_rej_handler(struct cm_work *work) cm_enter_timewait(cm_id_priv); break; } - /* fall through */ + fallthrough; default: pr_debug("%s: local_id %d, cm_id_priv->id.state: %d\n", __func__, be32_to_cpu(cm_id_priv->id.local_id), cm_id_priv->id.state); spin_unlock_irq(&cm_id_priv->lock); - ret = -EINVAL; goto out; } - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: cm_deref_id(cm_id_priv); @@ -3129,7 +3116,7 @@ int ib_send_cm_mra(struct ib_cm_id *cm_id, msg_response = CM_MSG_RESPONSE_OTHER; break; } - /* fall through */ + fallthrough; default: pr_debug("%s: local_id %d, cm_id_priv->id.state: %d\n", __func__, be32_to_cpu(cm_id_priv->id.local_id), @@ -3190,7 +3177,7 @@ static int cm_mra_handler(struct cm_work *work) { struct cm_id_private *cm_id_priv; struct cm_mra_msg *mra_msg; - int timeout, ret; + int timeout; mra_msg = (struct cm_mra_msg *)work->mad_recv_wc->recv_buf.mad; cm_id_priv = cm_acquire_mraed_id(mra_msg); @@ -3240,7 +3227,7 @@ static int cm_mra_handler(struct cm_work *work) case IB_CM_MRA_REP_RCVD: atomic_long_inc(&work->port->counter_group[CM_RECV_DUPLICATES]. counter[CM_MRA_COUNTER]); - /* fall through */ + fallthrough; default: pr_debug("%s local_id %d, cm_id_priv->id.state: %d\n", __func__, be32_to_cpu(cm_id_priv->id.local_id), @@ -3250,15 +3237,7 @@ static int cm_mra_handler(struct cm_work *work) cm_id_priv->msg->context[1] = (void *) (unsigned long) cm_id_priv->id.state; - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: spin_unlock_irq(&cm_id_priv->lock); @@ -3393,15 +3372,7 @@ static int cm_lap_handler(struct cm_work *work) cm_id_priv->id.lap_state = IB_CM_LAP_RCVD; cm_id_priv->tid = lap_msg->hdr.tid; - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; unlock: spin_unlock_irq(&cm_id_priv->lock); @@ -3413,7 +3384,6 @@ static int cm_apr_handler(struct cm_work *work) { struct cm_id_private *cm_id_priv; struct cm_apr_msg *apr_msg; - int ret; /* Currently Alternate path messages are not supported for * RoCE link layer. @@ -3448,16 +3418,7 @@ static int cm_apr_handler(struct cm_work *work) cm_id_priv->id.lap_state = IB_CM_LAP_IDLE; ib_cancel_mad(cm_id_priv->av.port->mad_agent, cm_id_priv->msg); cm_id_priv->msg = NULL; - - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: cm_deref_id(cm_id_priv); @@ -3468,7 +3429,6 @@ static int cm_timewait_handler(struct cm_work *work) { struct cm_timewait_info *timewait_info; struct cm_id_private *cm_id_priv; - int ret; timewait_info = container_of(work, struct cm_timewait_info, work); spin_lock_irq(&cm.lock); @@ -3487,15 +3447,7 @@ static int cm_timewait_handler(struct cm_work *work) goto out; } cm_id_priv->id.state = IB_CM_IDLE; - ret = atomic_inc_and_test(&cm_id_priv->work_count); - if (!ret) - list_add_tail(&work->list, &cm_id_priv->work_list); - spin_unlock_irq(&cm_id_priv->lock); - - if (ret) - cm_process_work(cm_id_priv, work); - else - cm_deref_id(cm_id_priv); + cm_queue_work_unlock(cm_id_priv, work); return 0; out: cm_deref_id(cm_id_priv); @@ -3642,7 +3594,6 @@ static int cm_sidr_req_handler(struct cm_work *work) .status = IB_SIDR_UNSUPPORTED }); goto out; /* No match. */ } - refcount_inc(&listen_cm_id_priv->refcount); spin_unlock_irq(&cm.lock); cm_id_priv->id.cm_handler = listen_cm_id_priv->id.cm_handler; @@ -3674,8 +3625,8 @@ static void cm_format_sidr_rep(struct cm_sidr_rep_msg *sidr_rep_msg, struct cm_id_private *cm_id_priv, struct ib_cm_sidr_rep_param *param) { - cm_format_mad_hdr(&sidr_rep_msg->hdr, CM_SIDR_REP_ATTR_ID, - cm_id_priv->tid); + cm_format_mad_ece_hdr(&sidr_rep_msg->hdr, CM_SIDR_REP_ATTR_ID, + cm_id_priv->tid, param->ece.attr_mod); IBA_SET(CM_SIDR_REP_REQUESTID, sidr_rep_msg, be32_to_cpu(cm_id_priv->id.remote_id)); IBA_SET(CM_SIDR_REP_STATUS, sidr_rep_msg, param->status); @@ -3683,6 +3634,10 @@ static void cm_format_sidr_rep(struct cm_sidr_rep_msg *sidr_rep_msg, IBA_SET(CM_SIDR_REP_SERVICEID, sidr_rep_msg, be64_to_cpu(cm_id_priv->id.service_id)); IBA_SET(CM_SIDR_REP_Q_KEY, sidr_rep_msg, param->qkey); + IBA_SET(CM_SIDR_REP_VENDOR_ID_L, sidr_rep_msg, + param->ece.vendor_id & 0xFF); + IBA_SET(CM_SIDR_REP_VENDOR_ID_H, sidr_rep_msg, + (param->ece.vendor_id >> 8) & 0xFF); if (param->info && param->info_length) IBA_SET_MEM(CM_SIDR_REP_ADDITIONAL_INFORMATION, sidr_rep_msg, @@ -3721,10 +3676,12 @@ static int cm_send_sidr_rep_locked(struct cm_id_private *cm_id_priv, return ret; } cm_id_priv->id.state = IB_CM_IDLE; + spin_lock_irq(&cm.lock); if (!RB_EMPTY_NODE(&cm_id_priv->sidr_id_node)) { rb_erase(&cm_id_priv->sidr_id_node, &cm.remote_sidr_table); RB_CLEAR_NODE(&cm_id_priv->sidr_id_node); } + spin_unlock_irq(&cm.lock); return 0; } @@ -4257,7 +4214,7 @@ static int cm_init_qp_rts_attr(struct cm_id_private *cm_id_priv, qp_attr->retry_cnt = cm_id_priv->retry_count; qp_attr->rnr_retry = cm_id_priv->rnr_retry_count; qp_attr->max_rd_atomic = cm_id_priv->initiator_depth; - /* fall through */ + fallthrough; case IB_QPT_XRC_TGT: *qp_attr_mask |= IB_QP_TIMEOUT; qp_attr->timeout = cm_id_priv->av.timeout; @@ -4384,7 +4341,7 @@ static void cm_remove_port_fs(struct cm_port *port) } -static void cm_add_one(struct ib_device *ib_device) +static int cm_add_one(struct ib_device *ib_device) { struct cm_device *cm_dev; struct cm_port *port; @@ -4403,7 +4360,7 @@ static void cm_add_one(struct ib_device *ib_device) cm_dev = kzalloc(struct_size(cm_dev, port, ib_device->phys_port_cnt), GFP_KERNEL); if (!cm_dev) - return; + return -ENOMEM; cm_dev->ib_device = ib_device; cm_dev->ack_delay = ib_device->attrs.local_ca_ack_delay; @@ -4415,8 +4372,10 @@ static void cm_add_one(struct ib_device *ib_device) continue; port = kzalloc(sizeof *port, GFP_KERNEL); - if (!port) + if (!port) { + ret = -ENOMEM; goto error1; + } cm_dev->port[i-1] = port; port->cm_dev = cm_dev; @@ -4437,8 +4396,10 @@ static void cm_add_one(struct ib_device *ib_device) cm_recv_handler, port, 0); - if (IS_ERR(port->mad_agent)) + if (IS_ERR(port->mad_agent)) { + ret = PTR_ERR(port->mad_agent); goto error2; + } ret = ib_modify_port(ib_device, i, 0, &port_modify); if (ret) @@ -4447,15 +4408,17 @@ static void cm_add_one(struct ib_device *ib_device) count++; } - if (!count) + if (!count) { + ret = -EOPNOTSUPP; goto free; + } ib_set_client_data(ib_device, &cm_client, cm_dev); write_lock_irqsave(&cm.device_lock, flags); list_add_tail(&cm_dev->list, &cm.device_list); write_unlock_irqrestore(&cm.device_lock, flags); - return; + return 0; error3: ib_unregister_mad_agent(port->mad_agent); @@ -4477,6 +4440,7 @@ error1: } free: kfree(cm_dev); + return ret; } static void cm_remove_one(struct ib_device *ib_device, void *client_data) @@ -4491,9 +4455,6 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data) unsigned long flags; int i; - if (!cm_dev) - return; - write_lock_irqsave(&cm.device_lock, flags); list_del(&cm_dev->list); write_unlock_irqrestore(&cm.device_lock, flags); diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 26e6f7df247b..5888311b2119 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -91,7 +91,13 @@ const char *__attribute_const__ rdma_reject_msg(struct rdma_cm_id *id, } EXPORT_SYMBOL(rdma_reject_msg); -bool rdma_is_consumer_reject(struct rdma_cm_id *id, int reason) +/** + * rdma_is_consumer_reject - return true if the consumer rejected the connect + * request. + * @id: Communication identifier that received the REJECT event. + * @reason: Value returned in the REJECT event status field. + */ +static bool rdma_is_consumer_reject(struct rdma_cm_id *id, int reason) { if (rdma_ib_or_roce(id->device, id->port_num)) return reason == IB_CM_REJ_CONSUMER_DEFINED; @@ -102,7 +108,6 @@ bool rdma_is_consumer_reject(struct rdma_cm_id *id, int reason) WARN_ON_ONCE(1); return false; } -EXPORT_SYMBOL(rdma_is_consumer_reject); const void *rdma_consumer_reject_data(struct rdma_cm_id *id, struct rdma_cm_event *ev, u8 *data_len) @@ -148,7 +153,7 @@ struct rdma_cm_id *rdma_res_to_id(struct rdma_restrack_entry *res) } EXPORT_SYMBOL(rdma_res_to_id); -static void cma_add_one(struct ib_device *device); +static int cma_add_one(struct ib_device *device); static void cma_remove_one(struct ib_device *device, void *client_data); static struct ib_client cma_client = { @@ -423,19 +428,6 @@ static int cma_comp_exch(struct rdma_id_private *id_priv, return ret; } -static enum rdma_cm_state cma_exch(struct rdma_id_private *id_priv, - enum rdma_cm_state exch) -{ - unsigned long flags; - enum rdma_cm_state old; - - spin_lock_irqsave(&id_priv->lock, flags); - old = id_priv->state; - id_priv->state = exch; - spin_unlock_irqrestore(&id_priv->lock, flags); - return old; -} - static inline u8 cma_get_ip_ver(const struct cma_hdr *hdr) { return hdr->ip_version >> 4; @@ -479,6 +471,7 @@ static void _cma_attach_to_dev(struct rdma_id_private *id_priv, rdma_restrack_kadd(&id_priv->res); else rdma_restrack_uadd(&id_priv->res); + trace_cm_id_attach(id_priv, cma_dev->device); } static void cma_attach_to_dev(struct rdma_id_private *id_priv, @@ -883,7 +876,6 @@ struct rdma_cm_id *__rdma_create_id(struct net *net, id_priv->id.route.addr.dev_addr.net = get_net(net); id_priv->seq_num &= 0x00ffffff; - trace_cm_id_create(id_priv); return &id_priv->id; } EXPORT_SYMBOL(__rdma_create_id); @@ -1619,6 +1611,8 @@ static struct rdma_id_private *cma_find_listener( { struct rdma_id_private *id_priv, *id_priv_dev; + lockdep_assert_held(&lock); + if (!bind_list) return ERR_PTR(-EINVAL); @@ -1665,6 +1659,7 @@ cma_ib_id_from_event(struct ib_cm_id *cm_id, } } + mutex_lock(&lock); /* * Net namespace might be getting deleted while route lookup, * cm_id lookup is in progress. Therefore, perform netdevice @@ -1706,6 +1701,7 @@ cma_ib_id_from_event(struct ib_cm_id *cm_id, id_priv = cma_find_listener(bind_list, cm_id, ib_event, req, *net_dev); err: rcu_read_unlock(); + mutex_unlock(&lock); if (IS_ERR(id_priv) && *net_dev) { dev_put(*net_dev); *net_dev = NULL; @@ -1820,23 +1816,11 @@ static void cma_leave_mc_groups(struct rdma_id_private *id_priv) } } -void rdma_destroy_id(struct rdma_cm_id *id) +static void _destroy_id(struct rdma_id_private *id_priv, + enum rdma_cm_state state) { - struct rdma_id_private *id_priv; - enum rdma_cm_state state; - - id_priv = container_of(id, struct rdma_id_private, id); - trace_cm_id_destroy(id_priv); - state = cma_exch(id_priv, RDMA_CM_DESTROYING); cma_cancel_operation(id_priv, state); - /* - * Wait for any active callback to finish. New callbacks will find - * the id_priv state set to destroying and abort. - */ - mutex_lock(&id_priv->handler_mutex); - mutex_unlock(&id_priv->handler_mutex); - rdma_restrack_del(&id_priv->res); if (id_priv->cma_dev) { if (rdma_cap_ib_cm(id_priv->id.device, 1)) { @@ -1865,6 +1849,42 @@ void rdma_destroy_id(struct rdma_cm_id *id) put_net(id_priv->id.route.addr.dev_addr.net); kfree(id_priv); } + +/* + * destroy an ID from within the handler_mutex. This ensures that no other + * handlers can start running concurrently. + */ +static void destroy_id_handler_unlock(struct rdma_id_private *id_priv) + __releases(&idprv->handler_mutex) +{ + enum rdma_cm_state state; + unsigned long flags; + + trace_cm_id_destroy(id_priv); + + /* + * Setting the state to destroyed under the handler mutex provides a + * fence against calling handler callbacks. If this is invoked due to + * the failure of a handler callback then it guarentees that no future + * handlers will be called. + */ + lockdep_assert_held(&id_priv->handler_mutex); + spin_lock_irqsave(&id_priv->lock, flags); + state = id_priv->state; + id_priv->state = RDMA_CM_DESTROYING; + spin_unlock_irqrestore(&id_priv->lock, flags); + mutex_unlock(&id_priv->handler_mutex); + _destroy_id(id_priv, state); +} + +void rdma_destroy_id(struct rdma_cm_id *id) +{ + struct rdma_id_private *id_priv = + container_of(id, struct rdma_id_private, id); + + mutex_lock(&id_priv->handler_mutex); + destroy_id_handler_unlock(id_priv); +} EXPORT_SYMBOL(rdma_destroy_id); static int cma_rep_recv(struct rdma_id_private *id_priv) @@ -1906,6 +1926,9 @@ static void cma_set_rep_event_data(struct rdma_cm_event *event, event->param.conn.rnr_retry_count = rep_data->rnr_retry_count; event->param.conn.srq = rep_data->srq; event->param.conn.qp_num = rep_data->remote_qpn; + + event->ece.vendor_id = rep_data->ece.vendor_id; + event->ece.attr_mod = rep_data->ece.attr_mod; } static int cma_cm_event_handler(struct rdma_id_private *id_priv, @@ -1913,6 +1936,8 @@ static int cma_cm_event_handler(struct rdma_id_private *id_priv, { int ret; + lockdep_assert_held(&id_priv->handler_mutex); + trace_cm_event_handler(id_priv, event); ret = id_priv->id.event_handler(&id_priv->id, event); trace_cm_event_done(id_priv, event, ret); @@ -1924,7 +1949,7 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, { struct rdma_id_private *id_priv = cm_id->context; struct rdma_cm_event event = {}; - int ret = 0; + int ret; mutex_lock(&id_priv->handler_mutex); if ((ib_event->event != IB_CM_TIMEWAIT_EXIT && @@ -1960,7 +1985,8 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, event.event = RDMA_CM_EVENT_ESTABLISHED; break; case IB_CM_DREQ_ERROR: - event.status = -ETIMEDOUT; /* fall through */ + event.status = -ETIMEDOUT; + fallthrough; case IB_CM_DREQ_RECEIVED: case IB_CM_DREP_RECEIVED: if (!cma_comp_exch(id_priv, RDMA_CM_CONNECT, @@ -1993,14 +2019,12 @@ static int cma_ib_handler(struct ib_cm_id *cm_id, if (ret) { /* Destroy the CM ID by returning a non-zero value. */ id_priv->cm_id.ib = NULL; - cma_exch(id_priv, RDMA_CM_DESTROYING); - mutex_unlock(&id_priv->handler_mutex); - rdma_destroy_id(&id_priv->id); + destroy_id_handler_unlock(id_priv); return ret; } out: mutex_unlock(&id_priv->handler_mutex); - return ret; + return 0; } static struct rdma_id_private * @@ -2124,6 +2148,9 @@ static void cma_set_req_event_data(struct rdma_cm_event *event, event->param.conn.rnr_retry_count = req_data->rnr_retry_count; event->param.conn.srq = req_data->srq; event->param.conn.qp_num = req_data->remote_qpn; + + event->ece.vendor_id = req_data->ece.vendor_id; + event->ece.attr_mod = req_data->ece.attr_mod; } static int cma_ib_check_req_qp_type(const struct rdma_cm_id *id, @@ -2159,7 +2186,7 @@ static int cma_ib_req_handler(struct ib_cm_id *cm_id, mutex_lock(&listen_id->handler_mutex); if (listen_id->state != RDMA_CM_LISTEN) { ret = -ECONNABORTED; - goto err1; + goto err_unlock; } offset = cma_user_data_offset(listen_id); @@ -2176,55 +2203,38 @@ static int cma_ib_req_handler(struct ib_cm_id *cm_id, } if (!conn_id) { ret = -ENOMEM; - goto err1; + goto err_unlock; } mutex_lock_nested(&conn_id->handler_mutex, SINGLE_DEPTH_NESTING); ret = cma_ib_acquire_dev(conn_id, listen_id, &req); - if (ret) - goto err2; + if (ret) { + destroy_id_handler_unlock(conn_id); + goto err_unlock; + } conn_id->cm_id.ib = cm_id; cm_id->context = conn_id; cm_id->cm_handler = cma_ib_handler; - /* - * Protect against the user destroying conn_id from another thread - * until we're done accessing it. - */ - cma_id_get(conn_id); ret = cma_cm_event_handler(conn_id, &event); - if (ret) - goto err3; - /* - * Acquire mutex to prevent user executing rdma_destroy_id() - * while we're accessing the cm_id. - */ - mutex_lock(&lock); + if (ret) { + /* Destroy the CM ID by returning a non-zero value. */ + conn_id->cm_id.ib = NULL; + mutex_unlock(&listen_id->handler_mutex); + destroy_id_handler_unlock(conn_id); + goto net_dev_put; + } + if (cma_comp(conn_id, RDMA_CM_CONNECT) && (conn_id->id.qp_type != IB_QPT_UD)) { trace_cm_send_mra(cm_id->context); ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0); } - mutex_unlock(&lock); mutex_unlock(&conn_id->handler_mutex); - mutex_unlock(&listen_id->handler_mutex); - cma_id_put(conn_id); - if (net_dev) - dev_put(net_dev); - return 0; -err3: - cma_id_put(conn_id); - /* Destroy the CM ID by returning a non-zero value. */ - conn_id->cm_id.ib = NULL; -err2: - cma_exch(conn_id, RDMA_CM_DESTROYING); - mutex_unlock(&conn_id->handler_mutex); -err1: +err_unlock: mutex_unlock(&listen_id->handler_mutex); - if (conn_id) - rdma_destroy_id(&conn_id->id); net_dev_put: if (net_dev) @@ -2324,9 +2334,7 @@ static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event) if (ret) { /* Destroy the CM ID by returning a non-zero value. */ id_priv->cm_id.iw = NULL; - cma_exch(id_priv, RDMA_CM_DESTROYING); - mutex_unlock(&id_priv->handler_mutex); - rdma_destroy_id(&id_priv->id); + destroy_id_handler_unlock(id_priv); return ret; } @@ -2373,16 +2381,16 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id, ret = rdma_translate_ip(laddr, &conn_id->id.route.addr.dev_addr); if (ret) { - mutex_unlock(&conn_id->handler_mutex); - rdma_destroy_id(new_cm_id); - goto out; + mutex_unlock(&listen_id->handler_mutex); + destroy_id_handler_unlock(conn_id); + return ret; } ret = cma_iw_acquire_dev(conn_id, listen_id); if (ret) { - mutex_unlock(&conn_id->handler_mutex); - rdma_destroy_id(new_cm_id); - goto out; + mutex_unlock(&listen_id->handler_mutex); + destroy_id_handler_unlock(conn_id); + return ret; } conn_id->cm_id.iw = cm_id; @@ -2392,25 +2400,16 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id, memcpy(cma_src_addr(conn_id), laddr, rdma_addr_size(laddr)); memcpy(cma_dst_addr(conn_id), raddr, rdma_addr_size(raddr)); - /* - * Protect against the user destroying conn_id from another thread - * until we're done accessing it. - */ - cma_id_get(conn_id); ret = cma_cm_event_handler(conn_id, &event); if (ret) { /* User wants to destroy the CM ID */ conn_id->cm_id.iw = NULL; - cma_exch(conn_id, RDMA_CM_DESTROYING); - mutex_unlock(&conn_id->handler_mutex); mutex_unlock(&listen_id->handler_mutex); - cma_id_put(conn_id); - rdma_destroy_id(&conn_id->id); + destroy_id_handler_unlock(conn_id); return ret; } mutex_unlock(&conn_id->handler_mutex); - cma_id_put(conn_id); out: mutex_unlock(&listen_id->handler_mutex); @@ -2467,6 +2466,10 @@ static int cma_listen_handler(struct rdma_cm_id *id, { struct rdma_id_private *id_priv = id->context; + /* Listening IDs are always destroyed on removal */ + if (event->event == RDMA_CM_EVENT_DEVICE_REMOVAL) + return -1; + id->context = id_priv->id.context; id->event_handler = id_priv->id.event_handler; trace_cm_event_handler(id_priv, event); @@ -2481,6 +2484,8 @@ static void cma_listen_on_dev(struct rdma_id_private *id_priv, struct net *net = id_priv->id.route.addr.dev_addr.net; int ret; + lockdep_assert_held(&lock); + if (cma_family(id_priv) == AF_IB && !rdma_cap_ib_cm(cma_dev->device, 1)) return; @@ -2640,21 +2645,21 @@ static void cma_work_handler(struct work_struct *_work) { struct cma_work *work = container_of(_work, struct cma_work, work); struct rdma_id_private *id_priv = work->id; - int destroy = 0; mutex_lock(&id_priv->handler_mutex); if (!cma_comp_exch(id_priv, work->old_state, work->new_state)) - goto out; + goto out_unlock; if (cma_cm_event_handler(id_priv, &work->event)) { - cma_exch(id_priv, RDMA_CM_DESTROYING); - destroy = 1; + cma_id_put(id_priv); + destroy_id_handler_unlock(id_priv); + goto out_free; } -out: + +out_unlock: mutex_unlock(&id_priv->handler_mutex); cma_id_put(id_priv); - if (destroy) - rdma_destroy_id(&id_priv->id); +out_free: kfree(work); } @@ -2662,23 +2667,22 @@ static void cma_ndev_work_handler(struct work_struct *_work) { struct cma_ndev_work *work = container_of(_work, struct cma_ndev_work, work); struct rdma_id_private *id_priv = work->id; - int destroy = 0; mutex_lock(&id_priv->handler_mutex); if (id_priv->state == RDMA_CM_DESTROYING || id_priv->state == RDMA_CM_DEVICE_REMOVAL) - goto out; + goto out_unlock; if (cma_cm_event_handler(id_priv, &work->event)) { - cma_exch(id_priv, RDMA_CM_DESTROYING); - destroy = 1; + cma_id_put(id_priv); + destroy_id_handler_unlock(id_priv); + goto out_free; } -out: +out_unlock: mutex_unlock(&id_priv->handler_mutex); cma_id_put(id_priv); - if (destroy) - rdma_destroy_id(&id_priv->id); +out_free: kfree(work); } @@ -2861,9 +2865,10 @@ struct iboe_prio_tc_map { bool found; }; -static int get_lower_vlan_dev_tc(struct net_device *dev, void *data) +static int get_lower_vlan_dev_tc(struct net_device *dev, + struct netdev_nested_priv *priv) { - struct iboe_prio_tc_map *map = data; + struct iboe_prio_tc_map *map = (struct iboe_prio_tc_map *)priv->data; if (is_vlan_dev(dev)) map->output_tc = get_vlan_ndev_tc(dev, map->input_prio); @@ -2882,16 +2887,18 @@ static int iboe_tos_to_sl(struct net_device *ndev, int tos) { struct iboe_prio_tc_map prio_tc_map = {}; int prio = rt_tos2priority(tos); + struct netdev_nested_priv priv; /* If VLAN device, get it directly from the VLAN netdev */ if (is_vlan_dev(ndev)) return get_vlan_ndev_tc(ndev, prio); prio_tc_map.input_prio = prio; + priv.data = (void *)&prio_tc_map; rcu_read_lock(); netdev_walk_all_lower_dev_rcu(ndev, get_lower_vlan_dev_tc, - &prio_tc_map); + &priv); rcu_read_unlock(); /* If map is found from lower device, use it; Otherwise * continue with the current netdevice to get priority to tc map. @@ -2904,6 +2911,24 @@ static int iboe_tos_to_sl(struct net_device *ndev, int tos) return 0; } +static __be32 cma_get_roce_udp_flow_label(struct rdma_id_private *id_priv) +{ + struct sockaddr_in6 *addr6; + u16 dport, sport; + u32 hash, fl; + + addr6 = (struct sockaddr_in6 *)cma_src_addr(id_priv); + fl = be32_to_cpu(addr6->sin6_flowinfo) & IB_GRH_FLOWLABEL_MASK; + if ((cma_family(id_priv) != AF_INET6) || !fl) { + dport = be16_to_cpu(cma_port(cma_dst_addr(id_priv))); + sport = be16_to_cpu(cma_port(cma_src_addr(id_priv))); + hash = (u32)sport * 31 + dport; + fl = hash & IB_GRH_FLOWLABEL_MASK; + } + + return cpu_to_be32(fl); +} + static int cma_resolve_iboe_route(struct rdma_id_private *id_priv) { struct rdma_route *route = &id_priv->id.route; @@ -2970,6 +2995,11 @@ static int cma_resolve_iboe_route(struct rdma_id_private *id_priv) goto err2; } + if (rdma_protocol_roce_udp_encap(id_priv->id.device, + id_priv->id.port_num)) + route->path_rec->flow_label = + cma_get_roce_udp_flow_label(id_priv); + cma_init_resolve_route_work(work, id_priv); queue_work(cma_wq, &work->work); @@ -3131,9 +3161,7 @@ static void addr_handler(int status, struct sockaddr *src_addr, event.event = RDMA_CM_EVENT_ADDR_RESOLVED; if (cma_cm_event_handler(id_priv, &event)) { - cma_exch(id_priv, RDMA_CM_DESTROYING); - mutex_unlock(&id_priv->handler_mutex); - rdma_destroy_id(&id_priv->id); + destroy_id_handler_unlock(id_priv); return; } out: @@ -3308,6 +3336,8 @@ static void cma_bind_port(struct rdma_bind_list *bind_list, u64 sid, mask; __be16 port; + lockdep_assert_held(&lock); + addr = cma_src_addr(id_priv); port = htons(bind_list->port); @@ -3336,6 +3366,8 @@ static int cma_alloc_port(enum rdma_ucm_port_space ps, struct rdma_bind_list *bind_list; int ret; + lockdep_assert_held(&lock); + bind_list = kzalloc(sizeof *bind_list, GFP_KERNEL); if (!bind_list) return -ENOMEM; @@ -3362,6 +3394,8 @@ static int cma_port_is_unique(struct rdma_bind_list *bind_list, struct sockaddr *saddr = cma_src_addr(id_priv); __be16 dport = cma_port(daddr); + lockdep_assert_held(&lock); + hlist_for_each_entry(cur_id, &bind_list->owners, node) { struct sockaddr *cur_daddr = cma_dst_addr(cur_id); struct sockaddr *cur_saddr = cma_src_addr(cur_id); @@ -3401,6 +3435,8 @@ static int cma_alloc_any_port(enum rdma_ucm_port_space ps, unsigned int rover; struct net *net = id_priv->id.route.addr.dev_addr.net; + lockdep_assert_held(&lock); + inet_get_local_port_range(net, &low, &high); remaining = (high - low) + 1; rover = prandom_u32() % remaining + low; @@ -3448,6 +3484,8 @@ static int cma_check_port(struct rdma_bind_list *bind_list, struct rdma_id_private *cur_id; struct sockaddr *addr, *cur_addr; + lockdep_assert_held(&lock); + addr = cma_src_addr(id_priv); hlist_for_each_entry(cur_id, &bind_list->owners, node) { if (id_priv == cur_id) @@ -3478,6 +3516,8 @@ static int cma_use_port(enum rdma_ucm_port_space ps, unsigned short snum; int ret; + lockdep_assert_held(&lock); + snum = ntohs(cma_port(cma_src_addr(id_priv))); if (snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE)) return -EACCES; @@ -3738,7 +3778,7 @@ static int cma_sidr_rep_handler(struct ib_cm_id *cm_id, struct rdma_cm_event event = {}; const struct ib_cm_sidr_rep_event_param *rep = &ib_event->param.sidr_rep_rcvd; - int ret = 0; + int ret; mutex_lock(&id_priv->handler_mutex); if (id_priv->state != RDMA_CM_CONNECT) @@ -3788,14 +3828,12 @@ static int cma_sidr_rep_handler(struct ib_cm_id *cm_id, if (ret) { /* Destroy the CM ID by returning a non-zero value. */ id_priv->cm_id.ib = NULL; - cma_exch(id_priv, RDMA_CM_DESTROYING); - mutex_unlock(&id_priv->handler_mutex); - rdma_destroy_id(&id_priv->id); + destroy_id_handler_unlock(id_priv); return ret; } out: mutex_unlock(&id_priv->handler_mutex); - return ret; + return 0; } static int cma_resolve_ib_udp(struct rdma_id_private *id_priv, @@ -3919,6 +3957,8 @@ static int cma_connect_ib(struct rdma_id_private *id_priv, req.local_cm_response_timeout = CMA_CM_RESPONSE_TIMEOUT; req.max_cm_retries = CMA_MAX_CM_RETRIES; req.srq = id_priv->srq ? 1 : 0; + req.ece.vendor_id = id_priv->ece.vendor_id; + req.ece.attr_mod = id_priv->ece.attr_mod; trace_cm_send_req(id_priv); ret = ib_send_cm_req(id_priv->cm_id.ib, &req); @@ -4008,6 +4048,27 @@ err: } EXPORT_SYMBOL(rdma_connect); +/** + * rdma_connect_ece - Initiate an active connection request with ECE data. + * @id: Connection identifier to connect. + * @conn_param: Connection information used for connected QPs. + * @ece: ECE parameters + * + * See rdma_connect() explanation. + */ +int rdma_connect_ece(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, + struct rdma_ucm_ece *ece) +{ + struct rdma_id_private *id_priv = + container_of(id, struct rdma_id_private, id); + + id_priv->ece.vendor_id = ece->vendor_id; + id_priv->ece.attr_mod = ece->attr_mod; + + return rdma_connect(id, conn_param); +} +EXPORT_SYMBOL(rdma_connect_ece); + static int cma_accept_ib(struct rdma_id_private *id_priv, struct rdma_conn_param *conn_param) { @@ -4033,6 +4094,8 @@ static int cma_accept_ib(struct rdma_id_private *id_priv, rep.flow_control = conn_param->flow_control; rep.rnr_retry_count = min_t(u8, 7, conn_param->rnr_retry_count); rep.srq = id_priv->srq ? 1 : 0; + rep.ece.vendor_id = id_priv->ece.vendor_id; + rep.ece.attr_mod = id_priv->ece.attr_mod; trace_cm_send_rep(id_priv); ret = ib_send_cm_rep(id_priv->cm_id.ib, &rep); @@ -4080,7 +4143,11 @@ static int cma_send_sidr_rep(struct rdma_id_private *id_priv, return ret; rep.qp_num = id_priv->qp_num; rep.qkey = id_priv->qkey; + + rep.ece.vendor_id = id_priv->ece.vendor_id; + rep.ece.attr_mod = id_priv->ece.attr_mod; } + rep.private_data = private_data; rep.private_data_len = private_data_len; @@ -4133,11 +4200,24 @@ int __rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, return 0; reject: cma_modify_qp_err(id_priv); - rdma_reject(id, NULL, 0); + rdma_reject(id, NULL, 0, IB_CM_REJ_CONSUMER_DEFINED); return ret; } EXPORT_SYMBOL(__rdma_accept); +int __rdma_accept_ece(struct rdma_cm_id *id, struct rdma_conn_param *conn_param, + const char *caller, struct rdma_ucm_ece *ece) +{ + struct rdma_id_private *id_priv = + container_of(id, struct rdma_id_private, id); + + id_priv->ece.vendor_id = ece->vendor_id; + id_priv->ece.attr_mod = ece->attr_mod; + + return __rdma_accept(id, conn_param, caller); +} +EXPORT_SYMBOL(__rdma_accept_ece); + int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event) { struct rdma_id_private *id_priv; @@ -4160,7 +4240,7 @@ int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event) EXPORT_SYMBOL(rdma_notify); int rdma_reject(struct rdma_cm_id *id, const void *private_data, - u8 private_data_len) + u8 private_data_len, u8 reason) { struct rdma_id_private *id_priv; int ret; @@ -4175,9 +4255,8 @@ int rdma_reject(struct rdma_cm_id *id, const void *private_data, private_data, private_data_len); } else { trace_cm_send_rej(id_priv); - ret = ib_send_cm_rej(id_priv->cm_id.ib, - IB_CM_REJ_CONSUMER_DEFINED, NULL, - 0, private_data, private_data_len); + ret = ib_send_cm_rej(id_priv->cm_id.ib, reason, NULL, 0, + private_data, private_data_len); } } else if (rdma_cap_iw_cm(id->device, id->port_num)) { ret = iw_cm_reject(id_priv->cm_id.iw, @@ -4279,9 +4358,7 @@ static int cma_ib_mc_handler(int status, struct ib_sa_multicast *multicast) rdma_destroy_ah_attr(&event.param.ud.ah_attr); if (ret) { - cma_exch(id_priv, RDMA_CM_DESTROYING); - mutex_unlock(&id_priv->handler_mutex); - rdma_destroy_id(&id_priv->id); + destroy_id_handler_unlock(id_priv); return 0; } @@ -4633,29 +4710,34 @@ static struct notifier_block cma_nb = { .notifier_call = cma_netdev_callback }; -static void cma_add_one(struct ib_device *device) +static int cma_add_one(struct ib_device *device) { struct cma_device *cma_dev; struct rdma_id_private *id_priv; unsigned int i; unsigned long supported_gids = 0; + int ret; cma_dev = kmalloc(sizeof *cma_dev, GFP_KERNEL); if (!cma_dev) - return; + return -ENOMEM; cma_dev->device = device; cma_dev->default_gid_type = kcalloc(device->phys_port_cnt, sizeof(*cma_dev->default_gid_type), GFP_KERNEL); - if (!cma_dev->default_gid_type) + if (!cma_dev->default_gid_type) { + ret = -ENOMEM; goto free_cma_dev; + } cma_dev->default_roce_tos = kcalloc(device->phys_port_cnt, sizeof(*cma_dev->default_roce_tos), GFP_KERNEL); - if (!cma_dev->default_roce_tos) + if (!cma_dev->default_roce_tos) { + ret = -ENOMEM; goto free_gid_type; + } rdma_for_each_port (device, i) { supported_gids = roce_gid_type_mask_support(device, i); @@ -4681,61 +4763,69 @@ static void cma_add_one(struct ib_device *device) mutex_unlock(&lock); trace_cm_add_one(device); - return; + return 0; free_gid_type: kfree(cma_dev->default_gid_type); free_cma_dev: kfree(cma_dev); - - return; + return ret; } -static int cma_remove_id_dev(struct rdma_id_private *id_priv) +static void cma_send_device_removal_put(struct rdma_id_private *id_priv) { - struct rdma_cm_event event = {}; + struct rdma_cm_event event = { .event = RDMA_CM_EVENT_DEVICE_REMOVAL }; enum rdma_cm_state state; - int ret = 0; - - /* Record that we want to remove the device */ - state = cma_exch(id_priv, RDMA_CM_DEVICE_REMOVAL); - if (state == RDMA_CM_DESTROYING) - return 0; + unsigned long flags; - cma_cancel_operation(id_priv, state); mutex_lock(&id_priv->handler_mutex); + /* Record that we want to remove the device */ + spin_lock_irqsave(&id_priv->lock, flags); + state = id_priv->state; + if (state == RDMA_CM_DESTROYING || state == RDMA_CM_DEVICE_REMOVAL) { + spin_unlock_irqrestore(&id_priv->lock, flags); + mutex_unlock(&id_priv->handler_mutex); + cma_id_put(id_priv); + return; + } + id_priv->state = RDMA_CM_DEVICE_REMOVAL; + spin_unlock_irqrestore(&id_priv->lock, flags); - /* Check for destruction from another callback. */ - if (!cma_comp(id_priv, RDMA_CM_DEVICE_REMOVAL)) - goto out; - - event.event = RDMA_CM_EVENT_DEVICE_REMOVAL; - ret = cma_cm_event_handler(id_priv, &event); -out: + if (cma_cm_event_handler(id_priv, &event)) { + /* + * At this point the ULP promises it won't call + * rdma_destroy_id() concurrently + */ + cma_id_put(id_priv); + mutex_unlock(&id_priv->handler_mutex); + trace_cm_id_destroy(id_priv); + _destroy_id(id_priv, state); + return; + } mutex_unlock(&id_priv->handler_mutex); - return ret; + + /* + * If this races with destroy then the thread that first assigns state + * to a destroying does the cancel. + */ + cma_cancel_operation(id_priv, state); + cma_id_put(id_priv); } static void cma_process_remove(struct cma_device *cma_dev) { - struct rdma_id_private *id_priv; - int ret; - mutex_lock(&lock); while (!list_empty(&cma_dev->id_list)) { - id_priv = list_entry(cma_dev->id_list.next, - struct rdma_id_private, list); + struct rdma_id_private *id_priv = list_first_entry( + &cma_dev->id_list, struct rdma_id_private, list); list_del(&id_priv->listen_list); list_del_init(&id_priv->list); cma_id_get(id_priv); mutex_unlock(&lock); - ret = id_priv->internal_id ? 1 : cma_remove_id_dev(id_priv); - cma_id_put(id_priv); - if (ret) - rdma_destroy_id(&id_priv->id); + cma_send_device_removal_put(id_priv); mutex_lock(&lock); } @@ -4751,9 +4841,6 @@ static void cma_remove_one(struct ib_device *device, void *client_data) trace_cm_remove_one(device); - if (!cma_dev) - return; - mutex_lock(&lock); list_del(&cma_dev->list); mutex_unlock(&lock); diff --git a/drivers/infiniband/core/cma_configfs.c b/drivers/infiniband/core/cma_configfs.c index c672a4978bfd..3c1e2ca564fe 100644 --- a/drivers/infiniband/core/cma_configfs.c +++ b/drivers/infiniband/core/cma_configfs.c @@ -322,8 +322,21 @@ fail: return ERR_PTR(err); } +static void drop_cma_dev(struct config_group *cgroup, struct config_item *item) +{ + struct config_group *group = + container_of(item, struct config_group, cg_item); + struct cma_dev_group *cma_dev_group = + container_of(group, struct cma_dev_group, device_group); + + configfs_remove_default_groups(&cma_dev_group->ports_group); + configfs_remove_default_groups(&cma_dev_group->device_group); + config_item_put(item); +} + static struct configfs_group_operations cma_subsys_group_ops = { .make_group = make_cma_dev, + .drop_item = drop_cma_dev, }; static const struct config_item_type cma_subsys_type = { diff --git a/drivers/infiniband/core/cma_priv.h b/drivers/infiniband/core/cma_priv.h index 5edcf44a9307..caece96ebcf5 100644 --- a/drivers/infiniband/core/cma_priv.h +++ b/drivers/infiniband/core/cma_priv.h @@ -95,6 +95,7 @@ struct rdma_id_private { * Internal to RDMA/core, don't use in the drivers */ struct rdma_restrack_entry res; + struct rdma_ucm_ece ece; }; #if IS_ENABLED(CONFIG_INFINIBAND_ADDR_TRANS_CONFIGFS) diff --git a/drivers/infiniband/core/cma_trace.h b/drivers/infiniband/core/cma_trace.h index 81e36bf13159..e6e20c36c538 100644 --- a/drivers/infiniband/core/cma_trace.h +++ b/drivers/infiniband/core/cma_trace.h @@ -103,23 +103,33 @@ DEFINE_CMA_FSM_EVENT(sent_drep); DEFINE_CMA_FSM_EVENT(sent_dreq); DEFINE_CMA_FSM_EVENT(id_destroy); -TRACE_EVENT(cm_id_create, +TRACE_EVENT(cm_id_attach, TP_PROTO( - const struct rdma_id_private *id_priv + const struct rdma_id_private *id_priv, + const struct ib_device *device ), - TP_ARGS(id_priv), + TP_ARGS(id_priv, device), TP_STRUCT__entry( __field(u32, cm_id) + __array(unsigned char, srcaddr, sizeof(struct sockaddr_in6)) + __array(unsigned char, dstaddr, sizeof(struct sockaddr_in6)) + __string(devname, device->name) ), TP_fast_assign( __entry->cm_id = id_priv->res.id; + memcpy(__entry->srcaddr, &id_priv->id.route.addr.src_addr, + sizeof(struct sockaddr_in6)); + memcpy(__entry->dstaddr, &id_priv->id.route.addr.dst_addr, + sizeof(struct sockaddr_in6)); + __assign_str(devname, device->name); ), - TP_printk("cm.id=%u", - __entry->cm_id + TP_printk("cm.id=%u src=%pISpc dst=%pISpc device=%s", + __entry->cm_id, __entry->srcaddr, __entry->dstaddr, + __get_str(devname) ) ); diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h index cf42acca4a3a..a1e6a67b2c4a 100644 --- a/drivers/infiniband/core/core_priv.h +++ b/drivers/infiniband/core/core_priv.h @@ -414,4 +414,7 @@ void rdma_umap_priv_init(struct rdma_umap_priv *priv, struct vm_area_struct *vma, struct rdma_user_mmap_entry *entry); +void ib_cq_pool_init(struct ib_device *dev); +void ib_cq_pool_destroy(struct ib_device *dev); + #endif /* _CORE_PRIV_H */ diff --git a/drivers/infiniband/core/counters.c b/drivers/infiniband/core/counters.c index 2257d7f7810f..636166880442 100644 --- a/drivers/infiniband/core/counters.c +++ b/drivers/infiniband/core/counters.c @@ -8,7 +8,7 @@ #include "core_priv.h" #include "restrack.h" -#define ALL_AUTO_MODE_MASKS (RDMA_COUNTER_MASK_QP_TYPE) +#define ALL_AUTO_MODE_MASKS (RDMA_COUNTER_MASK_QP_TYPE | RDMA_COUNTER_MASK_PID) static int __counter_set_mode(struct rdma_counter_mode *curr, enum rdma_nl_counter_mode new_mode, @@ -149,23 +149,13 @@ static bool auto_mode_match(struct ib_qp *qp, struct rdma_counter *counter, struct auto_mode_param *param = &counter->mode.param; bool match = true; - /* - * Ensure that counter belongs to the right PID. This operation can - * race with user space which kills the process and leaves QP and - * counters orphans. - * - * It is not a big deal because exitted task will leave both QP and - * counter in the same bucket of zombie process. Just ensure that - * process is still alive before procedding. - * - */ - if (task_pid_nr(counter->res.task) != task_pid_nr(qp->res.task) || - !task_pid_nr(qp->res.task)) - return false; - if (auto_mask & RDMA_COUNTER_MASK_QP_TYPE) match &= (param->qp_type == qp->qp_type); + if (auto_mask & RDMA_COUNTER_MASK_PID) + match &= (task_pid_nr(counter->res.task) == + task_pid_nr(qp->res.task)); + return match; } @@ -202,7 +192,7 @@ static int __rdma_counter_unbind_qp(struct ib_qp *qp) return ret; } -static void counter_history_stat_update(const struct rdma_counter *counter) +static void counter_history_stat_update(struct rdma_counter *counter) { struct ib_device *dev = counter->device; struct rdma_port_counter *port_counter; @@ -212,6 +202,8 @@ static void counter_history_stat_update(const struct rdma_counter *counter) if (!port_counter->hstats) return; + rdma_counter_query_stats(counter); + for (i = 0; i < counter->stats->num_counters; i++) port_counter->hstats->value[i] += counter->stats->value[i]; } @@ -286,7 +278,7 @@ int rdma_counter_bind_qp_auto(struct ib_qp *qp, u8 port) struct rdma_counter *counter; int ret; - if (!qp->res.valid) + if (!qp->res.valid || rdma_is_kernel_res(&qp->res)) return 0; if (!rdma_is_port_valid(dev, port)) @@ -481,7 +473,7 @@ int rdma_counter_bind_qpn(struct ib_device *dev, u8 port, goto err; } - if (counter->res.task != qp->res.task) { + if (rdma_is_kernel_res(&counter->res) != rdma_is_kernel_res(&qp->res)) { ret = -EINVAL; goto err_task; } diff --git a/drivers/infiniband/core/cq.c b/drivers/infiniband/core/cq.c index 4f25b2400694..a92fc3f90bb5 100644 --- a/drivers/infiniband/core/cq.c +++ b/drivers/infiniband/core/cq.c @@ -7,7 +7,11 @@ #include <linux/slab.h> #include <rdma/ib_verbs.h> +#include "core_priv.h" + #include <trace/events/rdma_core.h> +/* Max size for shared CQ, may require tuning */ +#define IB_MAX_SHARED_CQ_SZ 4096U /* # of WCs to poll for with a single call to ib_poll_cq */ #define IB_POLL_BATCH 16 @@ -68,6 +72,15 @@ static void rdma_dim_init(struct ib_cq *cq) INIT_WORK(&dim->work, ib_cq_rdma_dim_work); } +static void rdma_dim_destroy(struct ib_cq *cq) +{ + if (!cq->dim) + return; + + cancel_work_sync(&cq->dim->work); + kfree(cq->dim); +} + static int __poll_cq(struct ib_cq *cq, int num_entries, struct ib_wc *wc) { int rc; @@ -218,6 +231,7 @@ struct ib_cq *__ib_alloc_cq_user(struct ib_device *dev, void *private, cq->cq_context = private; cq->poll_ctx = poll_ctx; atomic_set(&cq->usecnt, 0); + cq->comp_vector = comp_vector; cq->wc = kmalloc_array(IB_POLL_BATCH, sizeof(*cq->wc), GFP_KERNEL); if (!cq->wc) @@ -261,6 +275,7 @@ struct ib_cq *__ib_alloc_cq_user(struct ib_device *dev, void *private, return cq; out_destroy_cq: + rdma_dim_destroy(cq); rdma_restrack_del(&cq->res); cq->device->ops.destroy_cq(cq, udata); out_free_wc: @@ -309,6 +324,8 @@ void ib_free_cq_user(struct ib_cq *cq, struct ib_udata *udata) { if (WARN_ON_ONCE(atomic_read(&cq->usecnt))) return; + if (WARN_ON_ONCE(cq->cqe_used)) + return; switch (cq->poll_ctx) { case IB_POLL_DIRECT: @@ -324,13 +341,177 @@ void ib_free_cq_user(struct ib_cq *cq, struct ib_udata *udata) WARN_ON_ONCE(1); } + rdma_dim_destroy(cq); trace_cq_free(cq); rdma_restrack_del(&cq->res); cq->device->ops.destroy_cq(cq, udata); - if (cq->dim) - cancel_work_sync(&cq->dim->work); - kfree(cq->dim); kfree(cq->wc); kfree(cq); } EXPORT_SYMBOL(ib_free_cq_user); + +void ib_cq_pool_init(struct ib_device *dev) +{ + unsigned int i; + + spin_lock_init(&dev->cq_pools_lock); + for (i = 0; i < ARRAY_SIZE(dev->cq_pools); i++) + INIT_LIST_HEAD(&dev->cq_pools[i]); +} + +void ib_cq_pool_destroy(struct ib_device *dev) +{ + struct ib_cq *cq, *n; + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(dev->cq_pools); i++) { + list_for_each_entry_safe(cq, n, &dev->cq_pools[i], + pool_entry) { + WARN_ON(cq->cqe_used); + cq->shared = false; + ib_free_cq(cq); + } + } +} + +static int ib_alloc_cqs(struct ib_device *dev, unsigned int nr_cqes, + enum ib_poll_context poll_ctx) +{ + LIST_HEAD(tmp_list); + unsigned int nr_cqs, i; + struct ib_cq *cq, *n; + int ret; + + if (poll_ctx > IB_POLL_LAST_POOL_TYPE) { + WARN_ON_ONCE(poll_ctx > IB_POLL_LAST_POOL_TYPE); + return -EINVAL; + } + + /* + * Allocate at least as many CQEs as requested, and otherwise + * a reasonable batch size so that we can share CQs between + * multiple users instead of allocating a larger number of CQs. + */ + nr_cqes = min_t(unsigned int, dev->attrs.max_cqe, + max(nr_cqes, IB_MAX_SHARED_CQ_SZ)); + nr_cqs = min_t(unsigned int, dev->num_comp_vectors, num_online_cpus()); + for (i = 0; i < nr_cqs; i++) { + cq = ib_alloc_cq(dev, NULL, nr_cqes, i, poll_ctx); + if (IS_ERR(cq)) { + ret = PTR_ERR(cq); + goto out_free_cqs; + } + cq->shared = true; + list_add_tail(&cq->pool_entry, &tmp_list); + } + + spin_lock_irq(&dev->cq_pools_lock); + list_splice(&tmp_list, &dev->cq_pools[poll_ctx]); + spin_unlock_irq(&dev->cq_pools_lock); + + return 0; + +out_free_cqs: + list_for_each_entry_safe(cq, n, &tmp_list, pool_entry) { + cq->shared = false; + ib_free_cq(cq); + } + return ret; +} + +/** + * ib_cq_pool_get() - Find the least used completion queue that matches + * a given cpu hint (or least used for wild card affinity) and fits + * nr_cqe. + * @dev: rdma device + * @nr_cqe: number of needed cqe entries + * @comp_vector_hint: completion vector hint (-1) for the driver to assign + * a comp vector based on internal counter + * @poll_ctx: cq polling context + * + * Finds a cq that satisfies @comp_vector_hint and @nr_cqe requirements and + * claim entries in it for us. In case there is no available cq, allocate + * a new cq with the requirements and add it to the device pool. + * IB_POLL_DIRECT cannot be used for shared cqs so it is not a valid value + * for @poll_ctx. + */ +struct ib_cq *ib_cq_pool_get(struct ib_device *dev, unsigned int nr_cqe, + int comp_vector_hint, + enum ib_poll_context poll_ctx) +{ + static unsigned int default_comp_vector; + unsigned int vector, num_comp_vectors; + struct ib_cq *cq, *found = NULL; + int ret; + + if (poll_ctx > IB_POLL_LAST_POOL_TYPE) { + WARN_ON_ONCE(poll_ctx > IB_POLL_LAST_POOL_TYPE); + return ERR_PTR(-EINVAL); + } + + num_comp_vectors = + min_t(unsigned int, dev->num_comp_vectors, num_online_cpus()); + /* Project the affinty to the device completion vector range */ + if (comp_vector_hint < 0) { + comp_vector_hint = + (READ_ONCE(default_comp_vector) + 1) % num_comp_vectors; + WRITE_ONCE(default_comp_vector, comp_vector_hint); + } + vector = comp_vector_hint % num_comp_vectors; + + /* + * Find the least used CQ with correct affinity and + * enough free CQ entries + */ + while (!found) { + spin_lock_irq(&dev->cq_pools_lock); + list_for_each_entry(cq, &dev->cq_pools[poll_ctx], + pool_entry) { + /* + * Check to see if we have found a CQ with the + * correct completion vector + */ + if (vector != cq->comp_vector) + continue; + if (cq->cqe_used + nr_cqe > cq->cqe) + continue; + found = cq; + break; + } + + if (found) { + found->cqe_used += nr_cqe; + spin_unlock_irq(&dev->cq_pools_lock); + + return found; + } + spin_unlock_irq(&dev->cq_pools_lock); + + /* + * Didn't find a match or ran out of CQs in the device + * pool, allocate a new array of CQs. + */ + ret = ib_alloc_cqs(dev, nr_cqe, poll_ctx); + if (ret) + return ERR_PTR(ret); + } + + return found; +} +EXPORT_SYMBOL(ib_cq_pool_get); + +/** + * ib_cq_pool_put - Return a CQ taken from a shared pool. + * @cq: The CQ to return. + * @nr_cqe: The max number of cqes that the user had requested. + */ +void ib_cq_pool_put(struct ib_cq *cq, unsigned int nr_cqe) +{ + if (WARN_ON_ONCE(nr_cqe > cq->cqe_used)) + return; + + spin_lock_irq(&cq->device->cq_pools_lock); + cq->cqe_used -= nr_cqe; + spin_unlock_irq(&cq->device->cq_pools_lock); +} +EXPORT_SYMBOL(ib_cq_pool_put); diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c index d0b3d35ad3e4..23ee65a9185f 100644 --- a/drivers/infiniband/core/device.c +++ b/drivers/infiniband/core/device.c @@ -272,7 +272,6 @@ static void ib_device_check_mandatory(struct ib_device *device) } mandatory_table[] = { IB_MANDATORY_FUNC(query_device), IB_MANDATORY_FUNC(query_port), - IB_MANDATORY_FUNC(query_pkey), IB_MANDATORY_FUNC(alloc_pd), IB_MANDATORY_FUNC(dealloc_pd), IB_MANDATORY_FUNC(create_qp), @@ -677,8 +676,20 @@ static int add_client_context(struct ib_device *device, if (ret) goto out; downgrade_write(&device->client_data_rwsem); - if (client->add) - client->add(device); + if (client->add) { + if (client->add(device)) { + /* + * If a client fails to add then the error code is + * ignored, but we won't call any more ops on this + * client. + */ + xa_erase(&device->client_data, client->client_id); + up_read(&device->client_data_rwsem); + ib_device_put(device); + ib_client_put(client); + return 0; + } + } /* Readers shall not see a client until add has been completed */ xa_set_mark(&device->client_data, client->client_id, @@ -1171,6 +1182,8 @@ static void setup_dma_device(struct ib_device *device) struct device *parent = device->dev.parent; WARN_ON_ONCE(device->dma_device); + +#ifdef CONFIG_DMA_OPS if (device->dev.dma_ops) { /* * The caller provided custom DMA operations. Copy the @@ -1191,7 +1204,9 @@ static void setup_dma_device(struct ib_device *device) else WARN_ON_ONCE(true); } - } else { + } else +#endif /* CONFIG_DMA_OPS */ + { /* * The caller did not provide custom DMA operations. Use the * DMA mapping operations of the parent device. @@ -1270,6 +1285,8 @@ static void disable_device(struct ib_device *device) remove_client_context(device, cid); } + ib_cq_pool_destroy(device); + /* Pairs with refcount_set in enable_device */ ib_device_put(device); wait_for_completion(&device->unreg_completion); @@ -1313,6 +1330,8 @@ static int enable_device_and_get(struct ib_device *device) goto out; } + ib_cq_pool_init(device); + down_read(&clients_rwsem); xa_for_each_marked (&clients, index, client, CLIENT_REGISTERED) { ret = add_client_context(device, client); @@ -1327,6 +1346,10 @@ out: return ret; } +static void prevent_dealloc_device(struct ib_device *ib_dev) +{ +} + /** * ib_register_device - Register an IB device with IB core * @device: Device to register @@ -1396,11 +1419,11 @@ int ib_register_device(struct ib_device *device, const char *name) * possibility for a parallel unregistration along with this * error flow. Since we have a refcount here we know any * parallel flow is stopped in disable_device and will see the - * NULL pointers, causing the responsibility to + * special dealloc_driver pointer, causing the responsibility to * ib_dealloc_device() to revert back to this thread. */ dealloc_fn = device->ops.dealloc_driver; - device->ops.dealloc_driver = NULL; + device->ops.dealloc_driver = prevent_dealloc_device; ib_device_put(device); __ib_unregister_device(device); device->ops.dealloc_driver = dealloc_fn; @@ -1448,7 +1471,8 @@ static void __ib_unregister_device(struct ib_device *ib_dev) * Drivers using the new flow may not call ib_dealloc_device except * in error unwind prior to registration success. */ - if (ib_dev->ops.dealloc_driver) { + if (ib_dev->ops.dealloc_driver && + ib_dev->ops.dealloc_driver != prevent_dealloc_device) { WARN_ON(kref_read(&ib_dev->dev.kobj.kref) <= 1); ib_dealloc_device(ib_dev); } @@ -2343,6 +2367,9 @@ int ib_query_pkey(struct ib_device *device, if (!rdma_is_port_valid(device, port_num)) return -EINVAL; + if (!device->ops.query_pkey) + return -EOPNOTSUPP; + return device->ops.query_pkey(device, port_num, index, pkey); } EXPORT_SYMBOL(ib_query_pkey); @@ -2557,7 +2584,6 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, add_gid); SET_DEVICE_OP(dev_ops, advise_mr); SET_DEVICE_OP(dev_ops, alloc_dm); - SET_DEVICE_OP(dev_ops, alloc_fmr); SET_DEVICE_OP(dev_ops, alloc_hw_stats); SET_DEVICE_OP(dev_ops, alloc_mr); SET_DEVICE_OP(dev_ops, alloc_mr_integrity); @@ -2584,7 +2610,6 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, create_wq); SET_DEVICE_OP(dev_ops, dealloc_dm); SET_DEVICE_OP(dev_ops, dealloc_driver); - SET_DEVICE_OP(dev_ops, dealloc_fmr); SET_DEVICE_OP(dev_ops, dealloc_mw); SET_DEVICE_OP(dev_ops, dealloc_pd); SET_DEVICE_OP(dev_ops, dealloc_ucontext); @@ -2605,8 +2630,14 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, drain_rq); SET_DEVICE_OP(dev_ops, drain_sq); SET_DEVICE_OP(dev_ops, enable_driver); - SET_DEVICE_OP(dev_ops, fill_res_entry); - SET_DEVICE_OP(dev_ops, fill_stat_entry); + SET_DEVICE_OP(dev_ops, fill_res_cm_id_entry); + SET_DEVICE_OP(dev_ops, fill_res_cq_entry); + SET_DEVICE_OP(dev_ops, fill_res_cq_entry_raw); + SET_DEVICE_OP(dev_ops, fill_res_mr_entry); + SET_DEVICE_OP(dev_ops, fill_res_mr_entry_raw); + SET_DEVICE_OP(dev_ops, fill_res_qp_entry); + SET_DEVICE_OP(dev_ops, fill_res_qp_entry_raw); + SET_DEVICE_OP(dev_ops, fill_stat_mr_entry); SET_DEVICE_OP(dev_ops, get_dev_fw_str); SET_DEVICE_OP(dev_ops, get_dma_mr); SET_DEVICE_OP(dev_ops, get_hw_stats); @@ -2628,7 +2659,6 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, iw_rem_ref); SET_DEVICE_OP(dev_ops, map_mr_sg); SET_DEVICE_OP(dev_ops, map_mr_sg_pi); - SET_DEVICE_OP(dev_ops, map_phys_fmr); SET_DEVICE_OP(dev_ops, mmap); SET_DEVICE_OP(dev_ops, mmap_free); SET_DEVICE_OP(dev_ops, modify_ah); @@ -2652,6 +2682,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, query_port); SET_DEVICE_OP(dev_ops, query_qp); SET_DEVICE_OP(dev_ops, query_srq); + SET_DEVICE_OP(dev_ops, query_ucontext); SET_DEVICE_OP(dev_ops, rdma_netdev_get_params); SET_DEVICE_OP(dev_ops, read_counters); SET_DEVICE_OP(dev_ops, reg_dm_mr); @@ -2662,13 +2693,14 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, resize_cq); SET_DEVICE_OP(dev_ops, set_vf_guid); SET_DEVICE_OP(dev_ops, set_vf_link_state); - SET_DEVICE_OP(dev_ops, unmap_fmr); SET_OBJ_SIZE(dev_ops, ib_ah); + SET_OBJ_SIZE(dev_ops, ib_counters); SET_OBJ_SIZE(dev_ops, ib_cq); SET_OBJ_SIZE(dev_ops, ib_pd); SET_OBJ_SIZE(dev_ops, ib_srq); SET_OBJ_SIZE(dev_ops, ib_ucontext); + SET_OBJ_SIZE(dev_ops, ib_xrcd); } EXPORT_SYMBOL(ib_set_device_ops); @@ -2721,7 +2753,7 @@ static int __init ib_core_init(void) ret = addr_init(); if (ret) { - pr_warn("Could't init IB address resolution\n"); + pr_warn("Couldn't init IB address resolution\n"); goto err_ibnl; } diff --git a/drivers/infiniband/core/fmr_pool.c b/drivers/infiniband/core/fmr_pool.c deleted file mode 100644 index e08aec427027..000000000000 --- a/drivers/infiniband/core/fmr_pool.c +++ /dev/null @@ -1,494 +0,0 @@ -/* - * Copyright (c) 2004 Topspin Communications. All rights reserved. - * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved. - * - * This software is available to you under a choice of one of two - * licenses. You may choose to be licensed under the terms of the GNU - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#include <linux/errno.h> -#include <linux/spinlock.h> -#include <linux/export.h> -#include <linux/slab.h> -#include <linux/jhash.h> -#include <linux/kthread.h> - -#include <rdma/ib_fmr_pool.h> - -#include "core_priv.h" - -#define PFX "fmr_pool: " - -enum { - IB_FMR_MAX_REMAPS = 32, - - IB_FMR_HASH_BITS = 8, - IB_FMR_HASH_SIZE = 1 << IB_FMR_HASH_BITS, - IB_FMR_HASH_MASK = IB_FMR_HASH_SIZE - 1 -}; - -/* - * If an FMR is not in use, then the list member will point to either - * its pool's free_list (if the FMR can be mapped again; that is, - * remap_count < pool->max_remaps) or its pool's dirty_list (if the - * FMR needs to be unmapped before being remapped). In either of - * these cases it is a bug if the ref_count is not 0. In other words, - * if ref_count is > 0, then the list member must not be linked into - * either free_list or dirty_list. - * - * The cache_node member is used to link the FMR into a cache bucket - * (if caching is enabled). This is independent of the reference - * count of the FMR. When a valid FMR is released, its ref_count is - * decremented, and if ref_count reaches 0, the FMR is placed in - * either free_list or dirty_list as appropriate. However, it is not - * removed from the cache and may be "revived" if a call to - * ib_fmr_register_physical() occurs before the FMR is remapped. In - * this case we just increment the ref_count and remove the FMR from - * free_list/dirty_list. - * - * Before we remap an FMR from free_list, we remove it from the cache - * (to prevent another user from obtaining a stale FMR). When an FMR - * is released, we add it to the tail of the free list, so that our - * cache eviction policy is "least recently used." - * - * All manipulation of ref_count, list and cache_node is protected by - * pool_lock to maintain consistency. - */ - -struct ib_fmr_pool { - spinlock_t pool_lock; - - int pool_size; - int max_pages; - int max_remaps; - int dirty_watermark; - int dirty_len; - struct list_head free_list; - struct list_head dirty_list; - struct hlist_head *cache_bucket; - - void (*flush_function)(struct ib_fmr_pool *pool, - void * arg); - void *flush_arg; - - struct kthread_worker *worker; - struct kthread_work work; - - atomic_t req_ser; - atomic_t flush_ser; - - wait_queue_head_t force_wait; -}; - -static inline u32 ib_fmr_hash(u64 first_page) -{ - return jhash_2words((u32) first_page, (u32) (first_page >> 32), 0) & - (IB_FMR_HASH_SIZE - 1); -} - -/* Caller must hold pool_lock */ -static inline struct ib_pool_fmr *ib_fmr_cache_lookup(struct ib_fmr_pool *pool, - u64 *page_list, - int page_list_len, - u64 io_virtual_address) -{ - struct hlist_head *bucket; - struct ib_pool_fmr *fmr; - - if (!pool->cache_bucket) - return NULL; - - bucket = pool->cache_bucket + ib_fmr_hash(*page_list); - - hlist_for_each_entry(fmr, bucket, cache_node) - if (io_virtual_address == fmr->io_virtual_address && - page_list_len == fmr->page_list_len && - !memcmp(page_list, fmr->page_list, - page_list_len * sizeof *page_list)) - return fmr; - - return NULL; -} - -static void ib_fmr_batch_release(struct ib_fmr_pool *pool) -{ - int ret; - struct ib_pool_fmr *fmr; - LIST_HEAD(unmap_list); - LIST_HEAD(fmr_list); - - spin_lock_irq(&pool->pool_lock); - - list_for_each_entry(fmr, &pool->dirty_list, list) { - hlist_del_init(&fmr->cache_node); - fmr->remap_count = 0; - list_add_tail(&fmr->fmr->list, &fmr_list); - } - - list_splice_init(&pool->dirty_list, &unmap_list); - pool->dirty_len = 0; - - spin_unlock_irq(&pool->pool_lock); - - if (list_empty(&unmap_list)) { - return; - } - - ret = ib_unmap_fmr(&fmr_list); - if (ret) - pr_warn(PFX "ib_unmap_fmr returned %d\n", ret); - - spin_lock_irq(&pool->pool_lock); - list_splice(&unmap_list, &pool->free_list); - spin_unlock_irq(&pool->pool_lock); -} - -static void ib_fmr_cleanup_func(struct kthread_work *work) -{ - struct ib_fmr_pool *pool = container_of(work, struct ib_fmr_pool, work); - - ib_fmr_batch_release(pool); - atomic_inc(&pool->flush_ser); - wake_up_interruptible(&pool->force_wait); - - if (pool->flush_function) - pool->flush_function(pool, pool->flush_arg); - - if (atomic_read(&pool->flush_ser) - atomic_read(&pool->req_ser) < 0) - kthread_queue_work(pool->worker, &pool->work); -} - -/** - * ib_create_fmr_pool - Create an FMR pool - * @pd:Protection domain for FMRs - * @params:FMR pool parameters - * - * Create a pool of FMRs. Return value is pointer to new pool or - * error code if creation failed. - */ -struct ib_fmr_pool *ib_create_fmr_pool(struct ib_pd *pd, - struct ib_fmr_pool_param *params) -{ - struct ib_device *device; - struct ib_fmr_pool *pool; - int i; - int ret; - int max_remaps; - - if (!params) - return ERR_PTR(-EINVAL); - - device = pd->device; - if (!device->ops.alloc_fmr || !device->ops.dealloc_fmr || - !device->ops.map_phys_fmr || !device->ops.unmap_fmr) { - dev_info(&device->dev, "Device does not support FMRs\n"); - return ERR_PTR(-ENOSYS); - } - - if (!device->attrs.max_map_per_fmr) - max_remaps = IB_FMR_MAX_REMAPS; - else - max_remaps = device->attrs.max_map_per_fmr; - - pool = kmalloc(sizeof *pool, GFP_KERNEL); - if (!pool) - return ERR_PTR(-ENOMEM); - - pool->cache_bucket = NULL; - pool->flush_function = params->flush_function; - pool->flush_arg = params->flush_arg; - - INIT_LIST_HEAD(&pool->free_list); - INIT_LIST_HEAD(&pool->dirty_list); - - if (params->cache) { - pool->cache_bucket = - kmalloc_array(IB_FMR_HASH_SIZE, - sizeof(*pool->cache_bucket), - GFP_KERNEL); - if (!pool->cache_bucket) { - ret = -ENOMEM; - goto out_free_pool; - } - - for (i = 0; i < IB_FMR_HASH_SIZE; ++i) - INIT_HLIST_HEAD(pool->cache_bucket + i); - } - - pool->pool_size = 0; - pool->max_pages = params->max_pages_per_fmr; - pool->max_remaps = max_remaps; - pool->dirty_watermark = params->dirty_watermark; - pool->dirty_len = 0; - spin_lock_init(&pool->pool_lock); - atomic_set(&pool->req_ser, 0); - atomic_set(&pool->flush_ser, 0); - init_waitqueue_head(&pool->force_wait); - - pool->worker = - kthread_create_worker(0, "ib_fmr(%s)", dev_name(&device->dev)); - if (IS_ERR(pool->worker)) { - pr_warn(PFX "couldn't start cleanup kthread worker\n"); - ret = PTR_ERR(pool->worker); - goto out_free_pool; - } - kthread_init_work(&pool->work, ib_fmr_cleanup_func); - - { - struct ib_pool_fmr *fmr; - struct ib_fmr_attr fmr_attr = { - .max_pages = params->max_pages_per_fmr, - .max_maps = pool->max_remaps, - .page_shift = params->page_shift - }; - int bytes_per_fmr = sizeof *fmr; - - if (pool->cache_bucket) - bytes_per_fmr += params->max_pages_per_fmr * sizeof (u64); - - for (i = 0; i < params->pool_size; ++i) { - fmr = kmalloc(bytes_per_fmr, GFP_KERNEL); - if (!fmr) - goto out_fail; - - fmr->pool = pool; - fmr->remap_count = 0; - fmr->ref_count = 0; - INIT_HLIST_NODE(&fmr->cache_node); - - fmr->fmr = ib_alloc_fmr(pd, params->access, &fmr_attr); - if (IS_ERR(fmr->fmr)) { - pr_warn(PFX "fmr_create failed for FMR %d\n", - i); - kfree(fmr); - goto out_fail; - } - - list_add_tail(&fmr->list, &pool->free_list); - ++pool->pool_size; - } - } - - return pool; - - out_free_pool: - kfree(pool->cache_bucket); - kfree(pool); - - return ERR_PTR(ret); - - out_fail: - ib_destroy_fmr_pool(pool); - - return ERR_PTR(-ENOMEM); -} -EXPORT_SYMBOL(ib_create_fmr_pool); - -/** - * ib_destroy_fmr_pool - Free FMR pool - * @pool:FMR pool to free - * - * Destroy an FMR pool and free all associated resources. - */ -void ib_destroy_fmr_pool(struct ib_fmr_pool *pool) -{ - struct ib_pool_fmr *fmr; - struct ib_pool_fmr *tmp; - LIST_HEAD(fmr_list); - int i; - - kthread_destroy_worker(pool->worker); - ib_fmr_batch_release(pool); - - i = 0; - list_for_each_entry_safe(fmr, tmp, &pool->free_list, list) { - if (fmr->remap_count) { - INIT_LIST_HEAD(&fmr_list); - list_add_tail(&fmr->fmr->list, &fmr_list); - ib_unmap_fmr(&fmr_list); - } - ib_dealloc_fmr(fmr->fmr); - list_del(&fmr->list); - kfree(fmr); - ++i; - } - - if (i < pool->pool_size) - pr_warn(PFX "pool still has %d regions registered\n", - pool->pool_size - i); - - kfree(pool->cache_bucket); - kfree(pool); -} -EXPORT_SYMBOL(ib_destroy_fmr_pool); - -/** - * ib_flush_fmr_pool - Invalidate all unmapped FMRs - * @pool:FMR pool to flush - * - * Ensure that all unmapped FMRs are fully invalidated. - */ -int ib_flush_fmr_pool(struct ib_fmr_pool *pool) -{ - int serial; - struct ib_pool_fmr *fmr, *next; - - /* - * The free_list holds FMRs that may have been used - * but have not been remapped enough times to be dirty. - * Put them on the dirty list now so that the cleanup - * thread will reap them too. - */ - spin_lock_irq(&pool->pool_lock); - list_for_each_entry_safe(fmr, next, &pool->free_list, list) { - if (fmr->remap_count > 0) - list_move(&fmr->list, &pool->dirty_list); - } - spin_unlock_irq(&pool->pool_lock); - - serial = atomic_inc_return(&pool->req_ser); - kthread_queue_work(pool->worker, &pool->work); - - if (wait_event_interruptible(pool->force_wait, - atomic_read(&pool->flush_ser) - serial >= 0)) - return -EINTR; - - return 0; -} -EXPORT_SYMBOL(ib_flush_fmr_pool); - -/** - * ib_fmr_pool_map_phys - Map an FMR from an FMR pool. - * @pool_handle: FMR pool to allocate FMR from - * @page_list: List of pages to map - * @list_len: Number of pages in @page_list - * @io_virtual_address: I/O virtual address for new FMR - */ -struct ib_pool_fmr *ib_fmr_pool_map_phys(struct ib_fmr_pool *pool_handle, - u64 *page_list, - int list_len, - u64 io_virtual_address) -{ - struct ib_fmr_pool *pool = pool_handle; - struct ib_pool_fmr *fmr; - unsigned long flags; - int result; - - if (list_len < 1 || list_len > pool->max_pages) - return ERR_PTR(-EINVAL); - - spin_lock_irqsave(&pool->pool_lock, flags); - fmr = ib_fmr_cache_lookup(pool, - page_list, - list_len, - io_virtual_address); - if (fmr) { - /* found in cache */ - ++fmr->ref_count; - if (fmr->ref_count == 1) { - list_del(&fmr->list); - } - - spin_unlock_irqrestore(&pool->pool_lock, flags); - - return fmr; - } - - if (list_empty(&pool->free_list)) { - spin_unlock_irqrestore(&pool->pool_lock, flags); - return ERR_PTR(-EAGAIN); - } - - fmr = list_entry(pool->free_list.next, struct ib_pool_fmr, list); - list_del(&fmr->list); - hlist_del_init(&fmr->cache_node); - spin_unlock_irqrestore(&pool->pool_lock, flags); - - result = ib_map_phys_fmr(fmr->fmr, page_list, list_len, - io_virtual_address); - - if (result) { - spin_lock_irqsave(&pool->pool_lock, flags); - list_add(&fmr->list, &pool->free_list); - spin_unlock_irqrestore(&pool->pool_lock, flags); - - pr_warn(PFX "fmr_map returns %d\n", result); - - return ERR_PTR(result); - } - - ++fmr->remap_count; - fmr->ref_count = 1; - - if (pool->cache_bucket) { - fmr->io_virtual_address = io_virtual_address; - fmr->page_list_len = list_len; - memcpy(fmr->page_list, page_list, list_len * sizeof(*page_list)); - - spin_lock_irqsave(&pool->pool_lock, flags); - hlist_add_head(&fmr->cache_node, - pool->cache_bucket + ib_fmr_hash(fmr->page_list[0])); - spin_unlock_irqrestore(&pool->pool_lock, flags); - } - - return fmr; -} -EXPORT_SYMBOL(ib_fmr_pool_map_phys); - -/** - * ib_fmr_pool_unmap - Unmap FMR - * @fmr:FMR to unmap - * - * Unmap an FMR. The FMR mapping may remain valid until the FMR is - * reused (or until ib_flush_fmr_pool() is called). - */ -void ib_fmr_pool_unmap(struct ib_pool_fmr *fmr) -{ - struct ib_fmr_pool *pool; - unsigned long flags; - - pool = fmr->pool; - - spin_lock_irqsave(&pool->pool_lock, flags); - - --fmr->ref_count; - if (!fmr->ref_count) { - if (fmr->remap_count < pool->max_remaps) { - list_add_tail(&fmr->list, &pool->free_list); - } else { - list_add_tail(&fmr->list, &pool->dirty_list); - if (++pool->dirty_len >= pool->dirty_watermark) { - atomic_inc(&pool->req_ser); - kthread_queue_work(pool->worker, &pool->work); - } - } - } - - spin_unlock_irqrestore(&pool->pool_lock, flags); -} -EXPORT_SYMBOL(ib_fmr_pool_unmap); diff --git a/drivers/infiniband/core/lag.c b/drivers/infiniband/core/lag.c new file mode 100644 index 000000000000..7063e41eaf26 --- /dev/null +++ b/drivers/infiniband/core/lag.c @@ -0,0 +1,138 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020 Mellanox Technologies. All rights reserved. + */ + +#include <rdma/ib_verbs.h> +#include <rdma/ib_cache.h> +#include <rdma/lag.h> + +static struct sk_buff *rdma_build_skb(struct ib_device *device, + struct net_device *netdev, + struct rdma_ah_attr *ah_attr, + gfp_t flags) +{ + struct ipv6hdr *ip6h; + struct sk_buff *skb; + struct ethhdr *eth; + struct iphdr *iph; + struct udphdr *uh; + u8 smac[ETH_ALEN]; + bool is_ipv4; + int hdr_len; + + is_ipv4 = ipv6_addr_v4mapped((struct in6_addr *)ah_attr->grh.dgid.raw); + hdr_len = ETH_HLEN + sizeof(struct udphdr) + LL_RESERVED_SPACE(netdev); + hdr_len += is_ipv4 ? sizeof(struct iphdr) : sizeof(struct ipv6hdr); + + skb = alloc_skb(hdr_len, flags); + if (!skb) + return NULL; + + skb->dev = netdev; + skb_reserve(skb, hdr_len); + skb_push(skb, sizeof(struct udphdr)); + skb_reset_transport_header(skb); + uh = udp_hdr(skb); + uh->source = + htons(rdma_flow_label_to_udp_sport(ah_attr->grh.flow_label)); + uh->dest = htons(ROCE_V2_UDP_DPORT); + uh->len = htons(sizeof(struct udphdr)); + + if (is_ipv4) { + skb_push(skb, sizeof(struct iphdr)); + skb_reset_network_header(skb); + iph = ip_hdr(skb); + iph->frag_off = 0; + iph->version = 4; + iph->protocol = IPPROTO_UDP; + iph->ihl = 0x5; + iph->tot_len = htons(sizeof(struct udphdr) + sizeof(struct + iphdr)); + memcpy(&iph->saddr, ah_attr->grh.sgid_attr->gid.raw + 12, + sizeof(struct in_addr)); + memcpy(&iph->daddr, ah_attr->grh.dgid.raw + 12, + sizeof(struct in_addr)); + } else { + skb_push(skb, sizeof(struct ipv6hdr)); + skb_reset_network_header(skb); + ip6h = ipv6_hdr(skb); + ip6h->version = 6; + ip6h->nexthdr = IPPROTO_UDP; + memcpy(&ip6h->flow_lbl, &ah_attr->grh.flow_label, + sizeof(*ip6h->flow_lbl)); + memcpy(&ip6h->saddr, ah_attr->grh.sgid_attr->gid.raw, + sizeof(struct in6_addr)); + memcpy(&ip6h->daddr, ah_attr->grh.dgid.raw, + sizeof(struct in6_addr)); + } + + skb_push(skb, sizeof(struct ethhdr)); + skb_reset_mac_header(skb); + eth = eth_hdr(skb); + skb->protocol = eth->h_proto = htons(is_ipv4 ? ETH_P_IP : ETH_P_IPV6); + rdma_read_gid_l2_fields(ah_attr->grh.sgid_attr, NULL, smac); + memcpy(eth->h_source, smac, ETH_ALEN); + memcpy(eth->h_dest, ah_attr->roce.dmac, ETH_ALEN); + + return skb; +} + +static struct net_device *rdma_get_xmit_slave_udp(struct ib_device *device, + struct net_device *master, + struct rdma_ah_attr *ah_attr, + gfp_t flags) +{ + struct net_device *slave; + struct sk_buff *skb; + + skb = rdma_build_skb(device, master, ah_attr, flags); + if (!skb) + return ERR_PTR(-ENOMEM); + + rcu_read_lock(); + slave = netdev_get_xmit_slave(master, skb, + !!(device->lag_flags & + RDMA_LAG_FLAGS_HASH_ALL_SLAVES)); + if (slave) + dev_hold(slave); + rcu_read_unlock(); + kfree_skb(skb); + return slave; +} + +void rdma_lag_put_ah_roce_slave(struct net_device *xmit_slave) +{ + if (xmit_slave) + dev_put(xmit_slave); +} + +struct net_device *rdma_lag_get_ah_roce_slave(struct ib_device *device, + struct rdma_ah_attr *ah_attr, + gfp_t flags) +{ + struct net_device *slave = NULL; + struct net_device *master; + + if (!(ah_attr->type == RDMA_AH_ATTR_TYPE_ROCE && + ah_attr->grh.sgid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP && + ah_attr->grh.flow_label)) + return NULL; + + rcu_read_lock(); + master = rdma_read_gid_attr_ndev_rcu(ah_attr->grh.sgid_attr); + if (IS_ERR(master)) { + rcu_read_unlock(); + return master; + } + dev_hold(master); + rcu_read_unlock(); + + if (!netif_is_bond_master(master)) + goto put; + + slave = rdma_get_xmit_slave_udp(device, master, ah_attr, flags); +put: + dev_put(master); + return slave; +} diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c index c54db13fa9b0..9355e521d9f4 100644 --- a/drivers/infiniband/core/mad.c +++ b/drivers/infiniband/core/mad.c @@ -85,7 +85,6 @@ MODULE_PARM_DESC(send_queue_size, "Size of send queue in number of work requests module_param_named(recv_queue_size, mad_recvq_size, int, 0444); MODULE_PARM_DESC(recv_queue_size, "Size of receive queue in number of work requests"); -/* Client ID 0 is used for snoop-only clients */ static DEFINE_XARRAY_ALLOC1(ib_mad_clients); static u32 ib_mad_client_next; static struct list_head ib_mad_port_list; @@ -403,7 +402,7 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, INIT_DELAYED_WORK(&mad_agent_priv->timed_work, timeout_sends); INIT_LIST_HEAD(&mad_agent_priv->local_list); INIT_WORK(&mad_agent_priv->local_work, local_completions); - atomic_set(&mad_agent_priv->refcount, 1); + refcount_set(&mad_agent_priv->refcount, 1); init_completion(&mad_agent_priv->comp); ret2 = ib_mad_agent_security_setup(&mad_agent_priv->agent, qp_type); @@ -483,141 +482,12 @@ error1: } EXPORT_SYMBOL(ib_register_mad_agent); -static inline int is_snooping_sends(int mad_snoop_flags) -{ - return (mad_snoop_flags & - (/*IB_MAD_SNOOP_POSTED_SENDS | - IB_MAD_SNOOP_RMPP_SENDS |*/ - IB_MAD_SNOOP_SEND_COMPLETIONS /*| - IB_MAD_SNOOP_RMPP_SEND_COMPLETIONS*/)); -} - -static inline int is_snooping_recvs(int mad_snoop_flags) -{ - return (mad_snoop_flags & - (IB_MAD_SNOOP_RECVS /*| - IB_MAD_SNOOP_RMPP_RECVS*/)); -} - -static int register_snoop_agent(struct ib_mad_qp_info *qp_info, - struct ib_mad_snoop_private *mad_snoop_priv) -{ - struct ib_mad_snoop_private **new_snoop_table; - unsigned long flags; - int i; - - spin_lock_irqsave(&qp_info->snoop_lock, flags); - /* Check for empty slot in array. */ - for (i = 0; i < qp_info->snoop_table_size; i++) - if (!qp_info->snoop_table[i]) - break; - - if (i == qp_info->snoop_table_size) { - /* Grow table. */ - new_snoop_table = krealloc(qp_info->snoop_table, - sizeof mad_snoop_priv * - (qp_info->snoop_table_size + 1), - GFP_ATOMIC); - if (!new_snoop_table) { - i = -ENOMEM; - goto out; - } - - qp_info->snoop_table = new_snoop_table; - qp_info->snoop_table_size++; - } - qp_info->snoop_table[i] = mad_snoop_priv; - atomic_inc(&qp_info->snoop_count); -out: - spin_unlock_irqrestore(&qp_info->snoop_lock, flags); - return i; -} - -struct ib_mad_agent *ib_register_mad_snoop(struct ib_device *device, - u8 port_num, - enum ib_qp_type qp_type, - int mad_snoop_flags, - ib_mad_snoop_handler snoop_handler, - ib_mad_recv_handler recv_handler, - void *context) -{ - struct ib_mad_port_private *port_priv; - struct ib_mad_agent *ret; - struct ib_mad_snoop_private *mad_snoop_priv; - int qpn; - int err; - - /* Validate parameters */ - if ((is_snooping_sends(mad_snoop_flags) && !snoop_handler) || - (is_snooping_recvs(mad_snoop_flags) && !recv_handler)) { - ret = ERR_PTR(-EINVAL); - goto error1; - } - qpn = get_spl_qp_index(qp_type); - if (qpn == -1) { - ret = ERR_PTR(-EINVAL); - goto error1; - } - port_priv = ib_get_mad_port(device, port_num); - if (!port_priv) { - ret = ERR_PTR(-ENODEV); - goto error1; - } - /* Allocate structures */ - mad_snoop_priv = kzalloc(sizeof *mad_snoop_priv, GFP_KERNEL); - if (!mad_snoop_priv) { - ret = ERR_PTR(-ENOMEM); - goto error1; - } - - /* Now, fill in the various structures */ - mad_snoop_priv->qp_info = &port_priv->qp_info[qpn]; - mad_snoop_priv->agent.device = device; - mad_snoop_priv->agent.recv_handler = recv_handler; - mad_snoop_priv->agent.snoop_handler = snoop_handler; - mad_snoop_priv->agent.context = context; - mad_snoop_priv->agent.qp = port_priv->qp_info[qpn].qp; - mad_snoop_priv->agent.port_num = port_num; - mad_snoop_priv->mad_snoop_flags = mad_snoop_flags; - init_completion(&mad_snoop_priv->comp); - - err = ib_mad_agent_security_setup(&mad_snoop_priv->agent, qp_type); - if (err) { - ret = ERR_PTR(err); - goto error2; - } - - mad_snoop_priv->snoop_index = register_snoop_agent( - &port_priv->qp_info[qpn], - mad_snoop_priv); - if (mad_snoop_priv->snoop_index < 0) { - ret = ERR_PTR(mad_snoop_priv->snoop_index); - goto error3; - } - - atomic_set(&mad_snoop_priv->refcount, 1); - return &mad_snoop_priv->agent; -error3: - ib_mad_agent_security_cleanup(&mad_snoop_priv->agent); -error2: - kfree(mad_snoop_priv); -error1: - return ret; -} -EXPORT_SYMBOL(ib_register_mad_snoop); - static inline void deref_mad_agent(struct ib_mad_agent_private *mad_agent_priv) { - if (atomic_dec_and_test(&mad_agent_priv->refcount)) + if (refcount_dec_and_test(&mad_agent_priv->refcount)) complete(&mad_agent_priv->comp); } -static inline void deref_snoop_agent(struct ib_mad_snoop_private *mad_snoop_priv) -{ - if (atomic_dec_and_test(&mad_snoop_priv->refcount)) - complete(&mad_snoop_priv->comp); -} - static void unregister_mad_agent(struct ib_mad_agent_private *mad_agent_priv) { struct ib_mad_port_private *port_priv; @@ -639,10 +509,10 @@ static void unregister_mad_agent(struct ib_mad_agent_private *mad_agent_priv) xa_erase(&ib_mad_clients, mad_agent_priv->agent.hi_tid); flush_workqueue(port_priv->wq); - ib_cancel_rmpp_recvs(mad_agent_priv); deref_mad_agent(mad_agent_priv); wait_for_completion(&mad_agent_priv->comp); + ib_cancel_rmpp_recvs(mad_agent_priv); ib_mad_agent_security_cleanup(&mad_agent_priv->agent); @@ -650,25 +520,6 @@ static void unregister_mad_agent(struct ib_mad_agent_private *mad_agent_priv) kfree_rcu(mad_agent_priv, rcu); } -static void unregister_mad_snoop(struct ib_mad_snoop_private *mad_snoop_priv) -{ - struct ib_mad_qp_info *qp_info; - unsigned long flags; - - qp_info = mad_snoop_priv->qp_info; - spin_lock_irqsave(&qp_info->snoop_lock, flags); - qp_info->snoop_table[mad_snoop_priv->snoop_index] = NULL; - atomic_dec(&qp_info->snoop_count); - spin_unlock_irqrestore(&qp_info->snoop_lock, flags); - - deref_snoop_agent(mad_snoop_priv); - wait_for_completion(&mad_snoop_priv->comp); - - ib_mad_agent_security_cleanup(&mad_snoop_priv->agent); - - kfree(mad_snoop_priv); -} - /* * ib_unregister_mad_agent - Unregisters a client from using MAD services * @@ -677,20 +528,11 @@ static void unregister_mad_snoop(struct ib_mad_snoop_private *mad_snoop_priv) void ib_unregister_mad_agent(struct ib_mad_agent *mad_agent) { struct ib_mad_agent_private *mad_agent_priv; - struct ib_mad_snoop_private *mad_snoop_priv; - - /* If the TID is zero, the agent can only snoop. */ - if (mad_agent->hi_tid) { - mad_agent_priv = container_of(mad_agent, - struct ib_mad_agent_private, - agent); - unregister_mad_agent(mad_agent_priv); - } else { - mad_snoop_priv = container_of(mad_agent, - struct ib_mad_snoop_private, - agent); - unregister_mad_snoop(mad_snoop_priv); - } + + mad_agent_priv = container_of(mad_agent, + struct ib_mad_agent_private, + agent); + unregister_mad_agent(mad_agent_priv); } EXPORT_SYMBOL(ib_unregister_mad_agent); @@ -706,57 +548,6 @@ static void dequeue_mad(struct ib_mad_list_head *mad_list) spin_unlock_irqrestore(&mad_queue->lock, flags); } -static void snoop_send(struct ib_mad_qp_info *qp_info, - struct ib_mad_send_buf *send_buf, - struct ib_mad_send_wc *mad_send_wc, - int mad_snoop_flags) -{ - struct ib_mad_snoop_private *mad_snoop_priv; - unsigned long flags; - int i; - - spin_lock_irqsave(&qp_info->snoop_lock, flags); - for (i = 0; i < qp_info->snoop_table_size; i++) { - mad_snoop_priv = qp_info->snoop_table[i]; - if (!mad_snoop_priv || - !(mad_snoop_priv->mad_snoop_flags & mad_snoop_flags)) - continue; - - atomic_inc(&mad_snoop_priv->refcount); - spin_unlock_irqrestore(&qp_info->snoop_lock, flags); - mad_snoop_priv->agent.snoop_handler(&mad_snoop_priv->agent, - send_buf, mad_send_wc); - deref_snoop_agent(mad_snoop_priv); - spin_lock_irqsave(&qp_info->snoop_lock, flags); - } - spin_unlock_irqrestore(&qp_info->snoop_lock, flags); -} - -static void snoop_recv(struct ib_mad_qp_info *qp_info, - struct ib_mad_recv_wc *mad_recv_wc, - int mad_snoop_flags) -{ - struct ib_mad_snoop_private *mad_snoop_priv; - unsigned long flags; - int i; - - spin_lock_irqsave(&qp_info->snoop_lock, flags); - for (i = 0; i < qp_info->snoop_table_size; i++) { - mad_snoop_priv = qp_info->snoop_table[i]; - if (!mad_snoop_priv || - !(mad_snoop_priv->mad_snoop_flags & mad_snoop_flags)) - continue; - - atomic_inc(&mad_snoop_priv->refcount); - spin_unlock_irqrestore(&qp_info->snoop_lock, flags); - mad_snoop_priv->agent.recv_handler(&mad_snoop_priv->agent, NULL, - mad_recv_wc); - deref_snoop_agent(mad_snoop_priv); - spin_lock_irqsave(&qp_info->snoop_lock, flags); - } - spin_unlock_irqrestore(&qp_info->snoop_lock, flags); -} - static void build_smp_wc(struct ib_qp *qp, struct ib_cqe *cqe, u16 slid, u16 pkey_index, u8 port_num, struct ib_wc *wc) { @@ -927,7 +718,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, * Reference MAD agent until receive * side of local completion handled */ - atomic_inc(&mad_agent_priv->refcount); + refcount_inc(&mad_agent_priv->refcount); } else kfree(mad_priv); break; @@ -967,7 +758,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, local->return_wc_byte_len = mad_size; } /* Reference MAD agent until send side of local completion handled */ - atomic_inc(&mad_agent_priv->refcount); + refcount_inc(&mad_agent_priv->refcount); /* Queue local completion to local list */ spin_lock_irqsave(&mad_agent_priv->lock, flags); list_add_tail(&local->completion_list, &mad_agent_priv->local_list); @@ -1125,7 +916,7 @@ struct ib_mad_send_buf * ib_create_send_mad(struct ib_mad_agent *mad_agent, } mad_send_wr->send_buf.mad_agent = mad_agent; - atomic_inc(&mad_agent_priv->refcount); + refcount_inc(&mad_agent_priv->refcount); return &mad_send_wr->send_buf; } EXPORT_SYMBOL(ib_create_send_mad); @@ -1340,7 +1131,7 @@ int ib_post_send_mad(struct ib_mad_send_buf *send_buf, mad_send_wr->status = IB_WC_SUCCESS; /* Reference MAD agent until send completes */ - atomic_inc(&mad_agent_priv->refcount); + refcount_inc(&mad_agent_priv->refcount); spin_lock_irqsave(&mad_agent_priv->lock, flags); list_add_tail(&mad_send_wr->agent_list, &mad_agent_priv->send_list); @@ -1357,7 +1148,7 @@ int ib_post_send_mad(struct ib_mad_send_buf *send_buf, spin_lock_irqsave(&mad_agent_priv->lock, flags); list_del(&mad_send_wr->agent_list); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - atomic_dec(&mad_agent_priv->refcount); + deref_mad_agent(mad_agent_priv); goto error; } } @@ -1763,7 +1554,7 @@ find_mad_agent(struct ib_mad_port_private *port_priv, hi_tid = be64_to_cpu(mad_hdr->tid) >> 32; rcu_read_lock(); mad_agent = xa_load(&ib_mad_clients, hi_tid); - if (mad_agent && !atomic_inc_not_zero(&mad_agent->refcount)) + if (mad_agent && !refcount_inc_not_zero(&mad_agent->refcount)) mad_agent = NULL; rcu_read_unlock(); } else { @@ -1815,7 +1606,7 @@ find_mad_agent(struct ib_mad_port_private *port_priv, } } if (mad_agent) - atomic_inc(&mad_agent->refcount); + refcount_inc(&mad_agent->refcount); out: spin_unlock_irqrestore(&port_priv->reg_lock, flags); } @@ -2040,7 +1831,7 @@ static void ib_mad_complete_recv(struct ib_mad_agent_private *mad_agent_priv, mad_agent_priv->agent.recv_handler( &mad_agent_priv->agent, NULL, mad_recv_wc); - atomic_dec(&mad_agent_priv->refcount); + deref_mad_agent(mad_agent_priv); } else { /* not user rmpp, revert to normal behavior and * drop the mad */ @@ -2057,7 +1848,7 @@ static void ib_mad_complete_recv(struct ib_mad_agent_private *mad_agent_priv, &mad_agent_priv->agent, &mad_send_wr->send_buf, mad_recv_wc); - atomic_dec(&mad_agent_priv->refcount); + deref_mad_agent(mad_agent_priv); mad_send_wc.status = IB_WC_SUCCESS; mad_send_wc.vendor_err = 0; @@ -2289,9 +2080,6 @@ static void ib_mad_recv_done(struct ib_cq *cq, struct ib_wc *wc) recv->header.recv_wc.recv_buf.mad = (struct ib_mad *)recv->mad; recv->header.recv_wc.recv_buf.grh = &recv->grh; - if (atomic_read(&qp_info->snoop_count)) - snoop_recv(qp_info, &recv->header.recv_wc, IB_MAD_SNOOP_RECVS); - /* Validate MAD */ if (!validate_mad((const struct ib_mad_hdr *)recv->mad, qp_info, opa)) goto out; @@ -2538,9 +2326,6 @@ retry: mad_send_wc.send_buf = &mad_send_wr->send_buf; mad_send_wc.status = wc->status; mad_send_wc.vendor_err = wc->vendor_err; - if (atomic_read(&qp_info->snoop_count)) - snoop_send(qp_info, &mad_send_wr->send_buf, &mad_send_wc, - IB_MAD_SNOOP_SEND_COMPLETIONS); ib_mad_complete_send_wr(mad_send_wr, &mad_send_wc); if (queued_send_wr) { @@ -2653,7 +2438,7 @@ static void cancel_mads(struct ib_mad_agent_private *mad_agent_priv) list_del(&mad_send_wr->agent_list); mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, &mad_send_wc); - atomic_dec(&mad_agent_priv->refcount); + deref_mad_agent(mad_agent_priv); } } @@ -2782,16 +2567,12 @@ static void local_completions(struct work_struct *work) local->mad_priv->header.recv_wc.recv_buf.grh = NULL; local->mad_priv->header.recv_wc.recv_buf.mad = (struct ib_mad *)local->mad_priv->mad; - if (atomic_read(&recv_mad_agent->qp_info->snoop_count)) - snoop_recv(recv_mad_agent->qp_info, - &local->mad_priv->header.recv_wc, - IB_MAD_SNOOP_RECVS); recv_mad_agent->agent.recv_handler( &recv_mad_agent->agent, &local->mad_send_wr->send_buf, &local->mad_priv->header.recv_wc); spin_lock_irqsave(&recv_mad_agent->lock, flags); - atomic_dec(&recv_mad_agent->refcount); + deref_mad_agent(recv_mad_agent); spin_unlock_irqrestore(&recv_mad_agent->lock, flags); } @@ -2800,15 +2581,11 @@ local_send_completion: mad_send_wc.status = IB_WC_SUCCESS; mad_send_wc.vendor_err = 0; mad_send_wc.send_buf = &local->mad_send_wr->send_buf; - if (atomic_read(&mad_agent_priv->qp_info->snoop_count)) - snoop_send(mad_agent_priv->qp_info, - &local->mad_send_wr->send_buf, - &mad_send_wc, IB_MAD_SNOOP_SEND_COMPLETIONS); mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, &mad_send_wc); spin_lock_irqsave(&mad_agent_priv->lock, flags); - atomic_dec(&mad_agent_priv->refcount); + deref_mad_agent(mad_agent_priv); if (free_mad) kfree(local->mad_priv); kfree(local); @@ -2894,7 +2671,7 @@ static void timeout_sends(struct work_struct *work) mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, &mad_send_wc); - atomic_dec(&mad_agent_priv->refcount); + deref_mad_agent(mad_agent_priv); spin_lock_irqsave(&mad_agent_priv->lock, flags); } spin_unlock_irqrestore(&mad_agent_priv->lock, flags); @@ -2941,6 +2718,7 @@ static int ib_mad_post_receive_mads(struct ib_mad_qp_info *qp_info, DMA_FROM_DEVICE); if (unlikely(ib_dma_mapping_error(qp_info->port_priv->device, sg_list.addr))) { + kfree(mad_priv); ret = -ENOMEM; break; } @@ -3119,10 +2897,6 @@ static void init_mad_qp(struct ib_mad_port_private *port_priv, init_mad_queue(qp_info, &qp_info->send_queue); init_mad_queue(qp_info, &qp_info->recv_queue); INIT_LIST_HEAD(&qp_info->overflow_list); - spin_lock_init(&qp_info->snoop_lock); - qp_info->snoop_table = NULL; - qp_info->snoop_table_size = 0; - atomic_set(&qp_info->snoop_count, 0); } static int create_mad_qp(struct ib_mad_qp_info *qp_info, @@ -3166,7 +2940,6 @@ static void destroy_mad_qp(struct ib_mad_qp_info *qp_info) return; ib_destroy_qp(qp_info->qp); - kfree(qp_info->snoop_table); } /* @@ -3304,9 +3077,11 @@ static int ib_mad_port_close(struct ib_device *device, int port_num) return 0; } -static void ib_mad_init_device(struct ib_device *device) +static int ib_mad_init_device(struct ib_device *device) { int start, i; + unsigned int count = 0; + int ret; start = rdma_start_port(device); @@ -3314,17 +3089,23 @@ static void ib_mad_init_device(struct ib_device *device) if (!rdma_cap_ib_mad(device, i)) continue; - if (ib_mad_port_open(device, i)) { + ret = ib_mad_port_open(device, i); + if (ret) { dev_err(&device->dev, "Couldn't open port %d\n", i); goto error; } - if (ib_agent_port_open(device, i)) { + ret = ib_agent_port_open(device, i); + if (ret) { dev_err(&device->dev, "Couldn't open port %d for agents\n", i); goto error_agent; } + count++; } - return; + if (!count) + return -EOPNOTSUPP; + + return 0; error_agent: if (ib_mad_port_close(device, i)) @@ -3341,6 +3122,7 @@ error: if (ib_mad_port_close(device, i)) dev_err(&device->dev, "Couldn't close port %d\n", i); } + return ret; } static void ib_mad_remove_device(struct ib_device *device, void *client_data) diff --git a/drivers/infiniband/core/mad_priv.h b/drivers/infiniband/core/mad_priv.h index 403d8673a2f9..4aa16b35dad0 100644 --- a/drivers/infiniband/core/mad_priv.h +++ b/drivers/infiniband/core/mad_priv.h @@ -103,7 +103,7 @@ struct ib_mad_agent_private { struct work_struct local_work; struct list_head rmpp_list; - atomic_t refcount; + refcount_t refcount; union { struct completion comp; struct rcu_head rcu; diff --git a/drivers/infiniband/core/mad_rmpp.c b/drivers/infiniband/core/mad_rmpp.c index 5ec57abc0849..e0573e4d0404 100644 --- a/drivers/infiniband/core/mad_rmpp.c +++ b/drivers/infiniband/core/mad_rmpp.c @@ -40,8 +40,7 @@ enum rmpp_state { RMPP_STATE_ACTIVE, RMPP_STATE_TIMEOUT, - RMPP_STATE_COMPLETE, - RMPP_STATE_CANCELING + RMPP_STATE_COMPLETE }; struct mad_rmpp_recv { @@ -52,7 +51,7 @@ struct mad_rmpp_recv { struct completion comp; enum rmpp_state state; spinlock_t lock; - atomic_t refcount; + refcount_t refcount; struct ib_ah *ah; struct ib_mad_recv_wc *rmpp_wc; @@ -73,7 +72,7 @@ struct mad_rmpp_recv { static inline void deref_rmpp_recv(struct mad_rmpp_recv *rmpp_recv) { - if (atomic_dec_and_test(&rmpp_recv->refcount)) + if (refcount_dec_and_test(&rmpp_recv->refcount)) complete(&rmpp_recv->comp); } @@ -92,22 +91,18 @@ void ib_cancel_rmpp_recvs(struct ib_mad_agent_private *agent) spin_lock_irqsave(&agent->lock, flags); list_for_each_entry(rmpp_recv, &agent->rmpp_list, list) { - if (rmpp_recv->state != RMPP_STATE_COMPLETE) - ib_free_recv_mad(rmpp_recv->rmpp_wc); - rmpp_recv->state = RMPP_STATE_CANCELING; - } - spin_unlock_irqrestore(&agent->lock, flags); - - list_for_each_entry(rmpp_recv, &agent->rmpp_list, list) { cancel_delayed_work(&rmpp_recv->timeout_work); cancel_delayed_work(&rmpp_recv->cleanup_work); } + spin_unlock_irqrestore(&agent->lock, flags); flush_workqueue(agent->qp_info->port_priv->wq); list_for_each_entry_safe(rmpp_recv, temp_rmpp_recv, &agent->rmpp_list, list) { list_del(&rmpp_recv->list); + if (rmpp_recv->state != RMPP_STATE_COMPLETE) + ib_free_recv_mad(rmpp_recv->rmpp_wc); destroy_rmpp_recv(rmpp_recv); } } @@ -272,10 +267,6 @@ static void recv_cleanup_handler(struct work_struct *work) unsigned long flags; spin_lock_irqsave(&rmpp_recv->agent->lock, flags); - if (rmpp_recv->state == RMPP_STATE_CANCELING) { - spin_unlock_irqrestore(&rmpp_recv->agent->lock, flags); - return; - } list_del(&rmpp_recv->list); spin_unlock_irqrestore(&rmpp_recv->agent->lock, flags); destroy_rmpp_recv(rmpp_recv); @@ -305,7 +296,7 @@ create_rmpp_recv(struct ib_mad_agent_private *agent, INIT_DELAYED_WORK(&rmpp_recv->cleanup_work, recv_cleanup_handler); spin_lock_init(&rmpp_recv->lock); rmpp_recv->state = RMPP_STATE_ACTIVE; - atomic_set(&rmpp_recv->refcount, 1); + refcount_set(&rmpp_recv->refcount, 1); rmpp_recv->rmpp_wc = mad_recv_wc; rmpp_recv->cur_seg_buf = &mad_recv_wc->recv_buf; @@ -357,7 +348,7 @@ acquire_rmpp_recv(struct ib_mad_agent_private *agent, spin_lock_irqsave(&agent->lock, flags); rmpp_recv = find_rmpp_recv(agent, mad_recv_wc); if (rmpp_recv) - atomic_inc(&rmpp_recv->refcount); + refcount_inc(&rmpp_recv->refcount); spin_unlock_irqrestore(&agent->lock, flags); return rmpp_recv; } @@ -553,7 +544,7 @@ start_rmpp(struct ib_mad_agent_private *agent, destroy_rmpp_recv(rmpp_recv); return continue_rmpp(agent, mad_recv_wc); } - atomic_inc(&rmpp_recv->refcount); + refcount_inc(&rmpp_recv->refcount); if (get_last_flag(&mad_recv_wc->recv_buf)) { rmpp_recv->state = RMPP_STATE_COMPLETE; diff --git a/drivers/infiniband/core/multicast.c b/drivers/infiniband/core/multicast.c index 9c2d8b7f1af9..740f03ecc05d 100644 --- a/drivers/infiniband/core/multicast.c +++ b/drivers/infiniband/core/multicast.c @@ -42,7 +42,7 @@ #include <rdma/ib_cache.h> #include "sa.h" -static void mcast_add_one(struct ib_device *device); +static int mcast_add_one(struct ib_device *device); static void mcast_remove_one(struct ib_device *device, void *client_data); static struct ib_client mcast_client = { @@ -815,7 +815,7 @@ static void mcast_event_handler(struct ib_event_handler *handler, } } -static void mcast_add_one(struct ib_device *device) +static int mcast_add_one(struct ib_device *device) { struct mcast_device *dev; struct mcast_port *port; @@ -825,7 +825,7 @@ static void mcast_add_one(struct ib_device *device) dev = kmalloc(struct_size(dev, port, device->phys_port_cnt), GFP_KERNEL); if (!dev) - return; + return -ENOMEM; dev->start_port = rdma_start_port(device); dev->end_port = rdma_end_port(device); @@ -845,7 +845,7 @@ static void mcast_add_one(struct ib_device *device) if (!count) { kfree(dev); - return; + return -EOPNOTSUPP; } dev->device = device; @@ -853,6 +853,7 @@ static void mcast_add_one(struct ib_device *device) INIT_IB_EVENT_HANDLER(&dev->event_handler, device, mcast_event_handler); ib_register_event_handler(&dev->event_handler); + return 0; } static void mcast_remove_one(struct ib_device *device, void *client_data) @@ -861,9 +862,6 @@ static void mcast_remove_one(struct ib_device *device, void *client_data) struct mcast_port *port; int i; - if (!dev) - return; - ib_unregister_event_handler(&dev->event_handler); flush_workqueue(mcast_wq); diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c index e16105be2eb2..12d29d54a081 100644 --- a/drivers/infiniband/core/nldev.c +++ b/drivers/infiniband/core/nldev.c @@ -114,6 +114,7 @@ static const struct nla_policy nldev_policy[RDMA_NLDEV_ATTR_MAX] = { [RDMA_NLDEV_ATTR_RES_PS] = { .type = NLA_U32 }, [RDMA_NLDEV_ATTR_RES_QP] = { .type = NLA_NESTED }, [RDMA_NLDEV_ATTR_RES_QP_ENTRY] = { .type = NLA_NESTED }, + [RDMA_NLDEV_ATTR_RES_RAW] = { .type = NLA_BINARY }, [RDMA_NLDEV_ATTR_RES_RKEY] = { .type = NLA_U32 }, [RDMA_NLDEV_ATTR_RES_RQPN] = { .type = NLA_U32 }, [RDMA_NLDEV_ATTR_RES_RQ_PSN] = { .type = NLA_U32 }, @@ -446,27 +447,11 @@ static int fill_res_name_pid(struct sk_buff *msg, return err ? -EMSGSIZE : 0; } -static bool fill_res_entry(struct ib_device *dev, struct sk_buff *msg, - struct rdma_restrack_entry *res) +static int fill_res_qp_entry_query(struct sk_buff *msg, + struct rdma_restrack_entry *res, + struct ib_device *dev, + struct ib_qp *qp) { - if (!dev->ops.fill_res_entry) - return false; - return dev->ops.fill_res_entry(msg, res); -} - -static bool fill_stat_entry(struct ib_device *dev, struct sk_buff *msg, - struct rdma_restrack_entry *res) -{ - if (!dev->ops.fill_stat_entry) - return false; - return dev->ops.fill_stat_entry(msg, res); -} - -static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin, - struct rdma_restrack_entry *res, uint32_t port) -{ - struct ib_qp *qp = container_of(res, struct ib_qp, res); - struct ib_device *dev = qp->device; struct ib_qp_init_attr qp_init_attr; struct ib_qp_attr qp_attr; int ret; @@ -475,16 +460,6 @@ static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin, if (ret) return ret; - if (port && port != qp_attr.port_num) - return -EAGAIN; - - /* In create_qp() port is not set yet */ - if (qp_attr.port_num && - nla_put_u32(msg, RDMA_NLDEV_ATTR_PORT_INDEX, qp_attr.port_num)) - goto err; - - if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LQPN, qp->qp_num)) - goto err; if (qp->qp_type == IB_QPT_RC || qp->qp_type == IB_QPT_UC) { if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_RQPN, qp_attr.dest_qp_num)) @@ -508,19 +483,53 @@ static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin, if (nla_put_u8(msg, RDMA_NLDEV_ATTR_RES_STATE, qp_attr.qp_state)) goto err; + if (dev->ops.fill_res_qp_entry) + return dev->ops.fill_res_qp_entry(msg, qp); + return 0; + +err: return -EMSGSIZE; +} + +static int fill_res_qp_entry(struct sk_buff *msg, bool has_cap_net_admin, + struct rdma_restrack_entry *res, uint32_t port) +{ + struct ib_qp *qp = container_of(res, struct ib_qp, res); + struct ib_device *dev = qp->device; + int ret; + + if (port && port != qp->port) + return -EAGAIN; + + /* In create_qp() port is not set yet */ + if (qp->port && nla_put_u32(msg, RDMA_NLDEV_ATTR_PORT_INDEX, qp->port)) + return -EINVAL; + + ret = nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LQPN, qp->qp_num); + if (ret) + return -EMSGSIZE; + if (!rdma_is_kernel_res(res) && nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, qp->pd->res.id)) - goto err; + return -EMSGSIZE; - if (fill_res_name_pid(msg, res)) - goto err; + ret = fill_res_name_pid(msg, res); + if (ret) + return -EMSGSIZE; - if (fill_res_entry(dev, msg, res)) - goto err; + return fill_res_qp_entry_query(msg, res, dev, qp); +} - return 0; +static int fill_res_qp_raw_entry(struct sk_buff *msg, bool has_cap_net_admin, + struct rdma_restrack_entry *res, uint32_t port) +{ + struct ib_qp *qp = container_of(res, struct ib_qp, res); + struct ib_device *dev = qp->device; -err: return -EMSGSIZE; + if (port && port != qp->port) + return -EAGAIN; + if (!dev->ops.fill_res_qp_entry_raw) + return -EINVAL; + return dev->ops.fill_res_qp_entry_raw(msg, qp); } static int fill_res_cm_id_entry(struct sk_buff *msg, bool has_cap_net_admin, @@ -568,9 +577,8 @@ static int fill_res_cm_id_entry(struct sk_buff *msg, bool has_cap_net_admin, if (fill_res_name_pid(msg, res)) goto err; - if (fill_res_entry(dev, msg, res)) - goto err; - + if (dev->ops.fill_res_cm_id_entry) + return dev->ops.fill_res_cm_id_entry(msg, cm_id); return 0; err: return -EMSGSIZE; @@ -583,35 +591,42 @@ static int fill_res_cq_entry(struct sk_buff *msg, bool has_cap_net_admin, struct ib_device *dev = cq->device; if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CQE, cq->cqe)) - goto err; + return -EMSGSIZE; if (nla_put_u64_64bit(msg, RDMA_NLDEV_ATTR_RES_USECNT, atomic_read(&cq->usecnt), RDMA_NLDEV_ATTR_PAD)) - goto err; + return -EMSGSIZE; /* Poll context is only valid for kernel CQs */ if (rdma_is_kernel_res(res) && nla_put_u8(msg, RDMA_NLDEV_ATTR_RES_POLL_CTX, cq->poll_ctx)) - goto err; + return -EMSGSIZE; if (nla_put_u8(msg, RDMA_NLDEV_ATTR_DEV_DIM, (cq->dim != NULL))) - goto err; + return -EMSGSIZE; if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CQN, res->id)) - goto err; + return -EMSGSIZE; if (!rdma_is_kernel_res(res) && nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_CTXN, cq->uobject->uevent.uobject.context->res.id)) - goto err; + return -EMSGSIZE; if (fill_res_name_pid(msg, res)) - goto err; + return -EMSGSIZE; - if (fill_res_entry(dev, msg, res)) - goto err; + return (dev->ops.fill_res_cq_entry) ? + dev->ops.fill_res_cq_entry(msg, cq) : 0; +} - return 0; +static int fill_res_cq_raw_entry(struct sk_buff *msg, bool has_cap_net_admin, + struct rdma_restrack_entry *res, uint32_t port) +{ + struct ib_cq *cq = container_of(res, struct ib_cq, res); + struct ib_device *dev = cq->device; -err: return -EMSGSIZE; + if (!dev->ops.fill_res_cq_entry_raw) + return -EINVAL; + return dev->ops.fill_res_cq_entry_raw(msg, cq); } static int fill_res_mr_entry(struct sk_buff *msg, bool has_cap_net_admin, @@ -622,38 +637,45 @@ static int fill_res_mr_entry(struct sk_buff *msg, bool has_cap_net_admin, if (has_cap_net_admin) { if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_RKEY, mr->rkey)) - goto err; + return -EMSGSIZE; if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LKEY, mr->lkey)) - goto err; + return -EMSGSIZE; } if (nla_put_u64_64bit(msg, RDMA_NLDEV_ATTR_RES_MRLEN, mr->length, RDMA_NLDEV_ATTR_PAD)) - goto err; + return -EMSGSIZE; if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_MRN, res->id)) - goto err; + return -EMSGSIZE; if (!rdma_is_kernel_res(res) && nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, mr->pd->res.id)) - goto err; + return -EMSGSIZE; if (fill_res_name_pid(msg, res)) - goto err; + return -EMSGSIZE; - if (fill_res_entry(dev, msg, res)) - goto err; + return (dev->ops.fill_res_mr_entry) ? + dev->ops.fill_res_mr_entry(msg, mr) : + 0; +} - return 0; +static int fill_res_mr_raw_entry(struct sk_buff *msg, bool has_cap_net_admin, + struct rdma_restrack_entry *res, uint32_t port) +{ + struct ib_mr *mr = container_of(res, struct ib_mr, res); + struct ib_device *dev = mr->pd->device; -err: return -EMSGSIZE; + if (!dev->ops.fill_res_mr_entry_raw) + return -EINVAL; + return dev->ops.fill_res_mr_entry_raw(msg, mr); } static int fill_res_pd_entry(struct sk_buff *msg, bool has_cap_net_admin, struct rdma_restrack_entry *res, uint32_t port) { struct ib_pd *pd = container_of(res, struct ib_pd, res); - struct ib_device *dev = pd->device; if (has_cap_net_admin) { if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY, @@ -676,13 +698,7 @@ static int fill_res_pd_entry(struct sk_buff *msg, bool has_cap_net_admin, pd->uobject->context->res.id)) goto err; - if (fill_res_name_pid(msg, res)) - goto err; - - if (fill_res_entry(dev, msg, res)) - goto err; - - return 0; + return fill_res_name_pid(msg, res); err: return -EMSGSIZE; } @@ -695,11 +711,16 @@ static int fill_stat_counter_mode(struct sk_buff *msg, if (nla_put_u32(msg, RDMA_NLDEV_ATTR_STAT_MODE, m->mode)) return -EMSGSIZE; - if (m->mode == RDMA_COUNTER_MODE_AUTO) + if (m->mode == RDMA_COUNTER_MODE_AUTO) { if ((m->mask & RDMA_COUNTER_MASK_QP_TYPE) && nla_put_u8(msg, RDMA_NLDEV_ATTR_RES_TYPE, m->param.qp_type)) return -EMSGSIZE; + if ((m->mask & RDMA_COUNTER_MASK_PID) && + fill_res_name_pid(msg, &counter->res)) + return -EMSGSIZE; + } + return 0; } @@ -738,9 +759,6 @@ static int fill_stat_counter_qps(struct sk_buff *msg, xa_lock(&rt->xa); xa_for_each(&rt->xa, id, res) { qp = container_of(res, struct ib_qp, res); - if (qp->qp_type == IB_QPT_RAW_PACKET && !capable(CAP_NET_RAW)) - continue; - if (!qp->counter || (qp->counter->id != counter->id)) continue; @@ -793,9 +811,8 @@ static int fill_stat_mr_entry(struct sk_buff *msg, bool has_cap_net_admin, if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_MRN, res->id)) goto err; - if (fill_stat_entry(dev, msg, res)) - goto err; - + if (dev->ops.fill_stat_mr_entry) + return dev->ops.fill_stat_mr_entry(msg, mr); return 0; err: @@ -840,7 +857,6 @@ static int fill_res_counter_entry(struct sk_buff *msg, bool has_cap_net_admin, if (nla_put_u32(msg, RDMA_NLDEV_ATTR_PORT_INDEX, counter->port) || nla_put_u32(msg, RDMA_NLDEV_ATTR_STAT_COUNTER_ID, counter->id) || - fill_res_name_pid(msg, &counter->res) || fill_stat_counter_mode(msg, counter) || fill_stat_counter_qps(msg, counter) || fill_stat_counter_hwcounters(msg, counter)) @@ -1177,7 +1193,6 @@ static int nldev_res_get_dumpit(struct sk_buff *skb, struct nldev_fill_res_entry { enum rdma_nldev_attr nldev_attr; - enum rdma_nldev_command nldev_cmd; u8 flags; u32 entry; u32 id; @@ -1189,40 +1204,34 @@ enum nldev_res_flags { static const struct nldev_fill_res_entry fill_entries[RDMA_RESTRACK_MAX] = { [RDMA_RESTRACK_QP] = { - .nldev_cmd = RDMA_NLDEV_CMD_RES_QP_GET, .nldev_attr = RDMA_NLDEV_ATTR_RES_QP, .entry = RDMA_NLDEV_ATTR_RES_QP_ENTRY, .id = RDMA_NLDEV_ATTR_RES_LQPN, }, [RDMA_RESTRACK_CM_ID] = { - .nldev_cmd = RDMA_NLDEV_CMD_RES_CM_ID_GET, .nldev_attr = RDMA_NLDEV_ATTR_RES_CM_ID, .entry = RDMA_NLDEV_ATTR_RES_CM_ID_ENTRY, .id = RDMA_NLDEV_ATTR_RES_CM_IDN, }, [RDMA_RESTRACK_CQ] = { - .nldev_cmd = RDMA_NLDEV_CMD_RES_CQ_GET, .nldev_attr = RDMA_NLDEV_ATTR_RES_CQ, .flags = NLDEV_PER_DEV, .entry = RDMA_NLDEV_ATTR_RES_CQ_ENTRY, .id = RDMA_NLDEV_ATTR_RES_CQN, }, [RDMA_RESTRACK_MR] = { - .nldev_cmd = RDMA_NLDEV_CMD_RES_MR_GET, .nldev_attr = RDMA_NLDEV_ATTR_RES_MR, .flags = NLDEV_PER_DEV, .entry = RDMA_NLDEV_ATTR_RES_MR_ENTRY, .id = RDMA_NLDEV_ATTR_RES_MRN, }, [RDMA_RESTRACK_PD] = { - .nldev_cmd = RDMA_NLDEV_CMD_RES_PD_GET, .nldev_attr = RDMA_NLDEV_ATTR_RES_PD, .flags = NLDEV_PER_DEV, .entry = RDMA_NLDEV_ATTR_RES_PD_ENTRY, .id = RDMA_NLDEV_ATTR_RES_PDN, }, [RDMA_RESTRACK_COUNTER] = { - .nldev_cmd = RDMA_NLDEV_CMD_STAT_GET, .nldev_attr = RDMA_NLDEV_ATTR_STAT_COUNTER, .entry = RDMA_NLDEV_ATTR_STAT_COUNTER_ENTRY, .id = RDMA_NLDEV_ATTR_STAT_COUNTER_ID, @@ -1281,7 +1290,8 @@ static int res_get_common_doit(struct sk_buff *skb, struct nlmsghdr *nlh, } nlh = nlmsg_put(msg, NETLINK_CB(skb).portid, nlh->nlmsg_seq, - RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, fe->nldev_cmd), + RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, + RDMA_NL_GET_OP(nlh->nlmsg_type)), 0, 0); if (fill_nldev_handle(msg, device)) { @@ -1359,7 +1369,8 @@ static int res_get_common_dumpit(struct sk_buff *skb, } nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq, - RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, fe->nldev_cmd), + RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, + RDMA_NL_GET_OP(cb->nlh->nlmsg_type)), 0, NLM_F_MULTI); if (fill_nldev_handle(skb, device)) { @@ -1441,26 +1452,29 @@ err_index: return ret; } -#define RES_GET_FUNCS(name, type) \ - static int nldev_res_get_##name##_dumpit(struct sk_buff *skb, \ +#define RES_GET_FUNCS(name, type) \ + static int nldev_res_get_##name##_dumpit(struct sk_buff *skb, \ struct netlink_callback *cb) \ - { \ - return res_get_common_dumpit(skb, cb, type, \ - fill_res_##name##_entry); \ - } \ - static int nldev_res_get_##name##_doit(struct sk_buff *skb, \ - struct nlmsghdr *nlh, \ + { \ + return res_get_common_dumpit(skb, cb, type, \ + fill_res_##name##_entry); \ + } \ + static int nldev_res_get_##name##_doit(struct sk_buff *skb, \ + struct nlmsghdr *nlh, \ struct netlink_ext_ack *extack) \ - { \ - return res_get_common_doit(skb, nlh, extack, type, \ - fill_res_##name##_entry); \ + { \ + return res_get_common_doit(skb, nlh, extack, type, \ + fill_res_##name##_entry); \ } RES_GET_FUNCS(qp, RDMA_RESTRACK_QP); +RES_GET_FUNCS(qp_raw, RDMA_RESTRACK_QP); RES_GET_FUNCS(cm_id, RDMA_RESTRACK_CM_ID); RES_GET_FUNCS(cq, RDMA_RESTRACK_CQ); +RES_GET_FUNCS(cq_raw, RDMA_RESTRACK_CQ); RES_GET_FUNCS(pd, RDMA_RESTRACK_PD); RES_GET_FUNCS(mr, RDMA_RESTRACK_MR); +RES_GET_FUNCS(mr_raw, RDMA_RESTRACK_MR); RES_GET_FUNCS(counter, RDMA_RESTRACK_COUNTER); static LIST_HEAD(link_ops); @@ -2145,6 +2159,21 @@ static const struct rdma_nl_cbs nldev_cb_table[RDMA_NLDEV_NUM_OPS] = { .doit = nldev_stat_del_doit, .flags = RDMA_NL_ADMIN_PERM, }, + [RDMA_NLDEV_CMD_RES_QP_GET_RAW] = { + .doit = nldev_res_get_qp_raw_doit, + .dump = nldev_res_get_qp_raw_dumpit, + .flags = RDMA_NL_ADMIN_PERM, + }, + [RDMA_NLDEV_CMD_RES_CQ_GET_RAW] = { + .doit = nldev_res_get_cq_raw_doit, + .dump = nldev_res_get_cq_raw_dumpit, + .flags = RDMA_NL_ADMIN_PERM, + }, + [RDMA_NLDEV_CMD_RES_MR_GET_RAW] = { + .doit = nldev_res_get_mr_raw_doit, + .dump = nldev_res_get_mr_raw_dumpit, + .flags = RDMA_NL_ADMIN_PERM, + }, }; void __init nldev_init(void) diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c index e0a5e897e4b1..6d3ed7c6e19e 100644 --- a/drivers/infiniband/core/rdma_core.c +++ b/drivers/infiniband/core/rdma_core.c @@ -130,6 +130,17 @@ static int uverbs_destroy_uobject(struct ib_uobject *uobj, lockdep_assert_held(&ufile->hw_destroy_rwsem); assert_uverbs_usecnt(uobj, UVERBS_LOOKUP_WRITE); + if (reason == RDMA_REMOVE_ABORT_HWOBJ) { + reason = RDMA_REMOVE_ABORT; + ret = uobj->uapi_object->type_class->destroy_hw(uobj, reason, + attrs); + /* + * Drivers are not permitted to ignore RDMA_REMOVE_ABORT, see + * ib_is_destroy_retryable, cleanup_retryable == false here. + */ + WARN_ON(ret); + } + if (reason == RDMA_REMOVE_ABORT) { WARN_ON(!list_empty(&uobj->list)); WARN_ON(!uobj->context); @@ -459,40 +470,46 @@ static struct ib_uobject * alloc_begin_fd_uobject(const struct uverbs_api_object *obj, struct uverbs_attr_bundle *attrs) { - const struct uverbs_obj_fd_type *fd_type = - container_of(obj->type_attrs, struct uverbs_obj_fd_type, type); + const struct uverbs_obj_fd_type *fd_type; int new_fd; - struct ib_uobject *uobj; + struct ib_uobject *uobj, *ret; struct file *filp; + uobj = alloc_uobj(attrs, obj); + if (IS_ERR(uobj)) + return uobj; + + fd_type = + container_of(obj->type_attrs, struct uverbs_obj_fd_type, type); if (WARN_ON(fd_type->fops->release != &uverbs_uobject_fd_release && - fd_type->fops->release != &uverbs_async_event_release)) - return ERR_PTR(-EINVAL); + fd_type->fops->release != &uverbs_async_event_release)) { + ret = ERR_PTR(-EINVAL); + goto err_fd; + } new_fd = get_unused_fd_flags(O_CLOEXEC); - if (new_fd < 0) - return ERR_PTR(new_fd); - - uobj = alloc_uobj(attrs, obj); - if (IS_ERR(uobj)) + if (new_fd < 0) { + ret = ERR_PTR(new_fd); goto err_fd; + } /* Note that uverbs_uobject_fd_release() is called during abort */ filp = anon_inode_getfile(fd_type->name, fd_type->fops, NULL, fd_type->flags); if (IS_ERR(filp)) { - uverbs_uobject_put(uobj); - uobj = ERR_CAST(filp); - goto err_fd; + ret = ERR_CAST(filp); + goto err_getfile; } uobj->object = filp; uobj->id = new_fd; return uobj; -err_fd: +err_getfile: put_unused_fd(new_fd); - return uobj; +err_fd: + uverbs_uobject_put(uobj); + return ret; } struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj, @@ -632,9 +649,6 @@ void rdma_alloc_commit_uobject(struct ib_uobject *uobj, { struct ib_uverbs_file *ufile = attrs->ufile; - /* alloc_commit consumes the uobj kref */ - uobj->uapi_object->type_class->alloc_commit(uobj); - /* kref is held so long as the uobj is on the uobj list. */ uverbs_uobject_get(uobj); spin_lock_irq(&ufile->uobjects_lock); @@ -644,6 +658,9 @@ void rdma_alloc_commit_uobject(struct ib_uobject *uobj, /* matches atomic_set(-1) in alloc_uobj */ atomic_set(&uobj->usecnt, 0); + /* alloc_commit consumes the uobj kref */ + uobj->uapi_object->type_class->alloc_commit(uobj); + /* Matches the down_read in rdma_alloc_begin_uobject */ up_read(&ufile->hw_destroy_rwsem); } @@ -653,11 +670,15 @@ void rdma_alloc_commit_uobject(struct ib_uobject *uobj, * object and anything else connected to uobj before calling this. */ void rdma_alloc_abort_uobject(struct ib_uobject *uobj, - struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs, + bool hw_obj_valid) { struct ib_uverbs_file *ufile = uobj->ufile; - uverbs_destroy_uobject(uobj, RDMA_REMOVE_ABORT, attrs); + uverbs_destroy_uobject(uobj, + hw_obj_valid ? RDMA_REMOVE_ABORT_HWOBJ : + RDMA_REMOVE_ABORT, + attrs); /* Matches the down_read in rdma_alloc_begin_uobject */ up_read(&ufile->hw_destroy_rwsem); @@ -927,8 +948,8 @@ uverbs_get_uobject_from_file(u16 object_id, enum uverbs_obj_access access, } void uverbs_finalize_object(struct ib_uobject *uobj, - enum uverbs_obj_access access, bool commit, - struct uverbs_attr_bundle *attrs) + enum uverbs_obj_access access, bool hw_obj_valid, + bool commit, struct uverbs_attr_bundle *attrs) { /* * refcounts should be handled at the object level and not at the @@ -951,7 +972,7 @@ void uverbs_finalize_object(struct ib_uobject *uobj, if (commit) rdma_alloc_commit_uobject(uobj, attrs); else - rdma_alloc_abort_uobject(uobj, attrs); + rdma_alloc_abort_uobject(uobj, attrs, hw_obj_valid); break; default: WARN_ON(true); diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h index 33978e0f1262..33706dad6c0f 100644 --- a/drivers/infiniband/core/rdma_core.h +++ b/drivers/infiniband/core/rdma_core.h @@ -64,8 +64,8 @@ uverbs_get_uobject_from_file(u16 object_id, enum uverbs_obj_access access, s64 id, struct uverbs_attr_bundle *attrs); void uverbs_finalize_object(struct ib_uobject *uobj, - enum uverbs_obj_access access, bool commit, - struct uverbs_attr_bundle *attrs); + enum uverbs_obj_access access, bool hw_obj_valid, + bool commit, struct uverbs_attr_bundle *attrs); int uverbs_output_written(const struct uverbs_attr_bundle *bundle, size_t idx); @@ -159,6 +159,9 @@ extern const struct uapi_definition uverbs_def_obj_dm[]; extern const struct uapi_definition uverbs_def_obj_flow_action[]; extern const struct uapi_definition uverbs_def_obj_intf[]; extern const struct uapi_definition uverbs_def_obj_mr[]; +extern const struct uapi_definition uverbs_def_obj_qp[]; +extern const struct uapi_definition uverbs_def_obj_srq[]; +extern const struct uapi_definition uverbs_def_obj_wq[]; extern const struct uapi_definition uverbs_def_write_intf[]; static inline const struct uverbs_api_write_method * diff --git a/drivers/infiniband/core/roce_gid_mgmt.c b/drivers/infiniband/core/roce_gid_mgmt.c index 2860def84f4d..6b8364bb032d 100644 --- a/drivers/infiniband/core/roce_gid_mgmt.c +++ b/drivers/infiniband/core/roce_gid_mgmt.c @@ -531,10 +531,11 @@ struct upper_list { struct net_device *upper; }; -static int netdev_upper_walk(struct net_device *upper, void *data) +static int netdev_upper_walk(struct net_device *upper, + struct netdev_nested_priv *priv) { struct upper_list *entry = kmalloc(sizeof(*entry), GFP_ATOMIC); - struct list_head *upper_list = data; + struct list_head *upper_list = (struct list_head *)priv->data; if (!entry) return 0; @@ -553,12 +554,14 @@ static void handle_netdev_upper(struct ib_device *ib_dev, u8 port, struct net_device *ndev)) { struct net_device *ndev = cookie; + struct netdev_nested_priv priv; struct upper_list *upper_iter; struct upper_list *upper_temp; LIST_HEAD(upper_list); + priv.data = &upper_list; rcu_read_lock(); - netdev_walk_all_upper_dev_rcu(ndev, netdev_upper_walk, &upper_list); + netdev_walk_all_upper_dev_rcu(ndev, netdev_upper_walk, &priv); rcu_read_unlock(); handle_netdev(ib_dev, port, ndev); diff --git a/drivers/infiniband/core/rw.c b/drivers/infiniband/core/rw.c index 557efbf29197..13f43ab7220b 100644 --- a/drivers/infiniband/core/rw.c +++ b/drivers/infiniband/core/rw.c @@ -129,7 +129,7 @@ static int rdma_rw_init_mr_wrs(struct rdma_rw_ctx *ctx, struct ib_qp *qp, qp->integrity_en); int i, j, ret = 0, count = 0; - ctx->nr_ops = (sg_cnt + pages_per_mr - 1) / pages_per_mr; + ctx->nr_ops = DIV_ROUND_UP(sg_cnt, pages_per_mr); ctx->reg = kcalloc(ctx->nr_ops, sizeof(*ctx->reg), GFP_KERNEL); if (!ctx->reg) { ret = -ENOMEM; @@ -510,7 +510,6 @@ struct ib_send_wr *rdma_rw_ctx_wrs(struct rdma_rw_ctx *ctx, struct ib_qp *qp, switch (ctx->type) { case RDMA_RW_SIG_MR: case RDMA_RW_MR: - /* fallthrough */ for (i = 0; i < ctx->nr_ops; i++) { rdma_rw_update_lkey(&ctx->reg[i], ctx->reg[i].wr.wr.opcode != diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index 74e0058fcf9e..8c930bf1df89 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -174,7 +174,7 @@ static const struct nla_policy ib_nl_policy[LS_NLA_TYPE_MAX] = { }; -static void ib_sa_add_one(struct ib_device *device); +static int ib_sa_add_one(struct ib_device *device); static void ib_sa_remove_one(struct ib_device *device, void *client_data); static struct ib_client sa_client = { @@ -190,7 +190,7 @@ static u32 tid; #define PATH_REC_FIELD(field) \ .struct_offset_bytes = offsetof(struct sa_path_rec, field), \ - .struct_size_bytes = sizeof((struct sa_path_rec *)0)->field, \ + .struct_size_bytes = sizeof_field(struct sa_path_rec, field), \ .field_name = "sa_path_rec:" #field static const struct ib_field path_rec_table[] = { @@ -292,7 +292,7 @@ static const struct ib_field path_rec_table[] = { .struct_offset_bytes = \ offsetof(struct sa_path_rec, field), \ .struct_size_bytes = \ - sizeof((struct sa_path_rec *)0)->field, \ + sizeof_field(struct sa_path_rec, field), \ .field_name = "sa_path_rec:" #field static const struct ib_field opa_path_rec_table[] = { @@ -420,7 +420,7 @@ static const struct ib_field opa_path_rec_table[] = { #define MCMEMBER_REC_FIELD(field) \ .struct_offset_bytes = offsetof(struct ib_sa_mcmember_rec, field), \ - .struct_size_bytes = sizeof ((struct ib_sa_mcmember_rec *) 0)->field, \ + .struct_size_bytes = sizeof_field(struct ib_sa_mcmember_rec, field), \ .field_name = "sa_mcmember_rec:" #field static const struct ib_field mcmember_rec_table[] = { @@ -504,7 +504,7 @@ static const struct ib_field mcmember_rec_table[] = { #define SERVICE_REC_FIELD(field) \ .struct_offset_bytes = offsetof(struct ib_sa_service_rec, field), \ - .struct_size_bytes = sizeof ((struct ib_sa_service_rec *) 0)->field, \ + .struct_size_bytes = sizeof_field(struct ib_sa_service_rec, field), \ .field_name = "sa_service_rec:" #field static const struct ib_field service_rec_table[] = { @@ -552,7 +552,7 @@ static const struct ib_field service_rec_table[] = { #define CLASSPORTINFO_REC_FIELD(field) \ .struct_offset_bytes = offsetof(struct ib_class_port_info, field), \ - .struct_size_bytes = sizeof((struct ib_class_port_info *)0)->field, \ + .struct_size_bytes = sizeof_field(struct ib_class_port_info, field), \ .field_name = "ib_class_port_info:" #field static const struct ib_field ib_classport_info_rec_table[] = { @@ -630,7 +630,7 @@ static const struct ib_field ib_classport_info_rec_table[] = { .struct_offset_bytes =\ offsetof(struct opa_class_port_info, field), \ .struct_size_bytes = \ - sizeof((struct opa_class_port_info *)0)->field, \ + sizeof_field(struct opa_class_port_info, field), \ .field_name = "opa_class_port_info:" #field static const struct ib_field opa_classport_info_rec_table[] = { @@ -710,7 +710,7 @@ static const struct ib_field opa_classport_info_rec_table[] = { #define GUIDINFO_REC_FIELD(field) \ .struct_offset_bytes = offsetof(struct ib_sa_guidinfo_rec, field), \ - .struct_size_bytes = sizeof((struct ib_sa_guidinfo_rec *) 0)->field, \ + .struct_size_bytes = sizeof_field(struct ib_sa_guidinfo_rec, field), \ .field_name = "sa_guidinfo_rec:" #field static const struct ib_field guidinfo_rec_table[] = { @@ -829,13 +829,20 @@ static int ib_nl_get_path_rec_attrs_len(ib_sa_comp_mask comp_mask) return len; } -static int ib_nl_send_msg(struct ib_sa_query *query, gfp_t gfp_mask) +static int ib_nl_make_request(struct ib_sa_query *query, gfp_t gfp_mask) { struct sk_buff *skb = NULL; struct nlmsghdr *nlh; void *data; struct ib_sa_mad *mad; int len; + unsigned long flags; + unsigned long delay; + gfp_t gfp_flag; + int ret; + + INIT_LIST_HEAD(&query->list); + query->seq = (u32)atomic_inc_return(&ib_nl_sa_request_seq); mad = query->mad_buf->mad; len = ib_nl_get_path_rec_attrs_len(mad->sa_hdr.comp_mask); @@ -860,36 +867,25 @@ static int ib_nl_send_msg(struct ib_sa_query *query, gfp_t gfp_mask) /* Repair the nlmsg header length */ nlmsg_end(skb, nlh); - return rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_LS, gfp_mask); -} + gfp_flag = ((gfp_mask & GFP_ATOMIC) == GFP_ATOMIC) ? GFP_ATOMIC : + GFP_NOWAIT; -static int ib_nl_make_request(struct ib_sa_query *query, gfp_t gfp_mask) -{ - unsigned long flags; - unsigned long delay; - int ret; + spin_lock_irqsave(&ib_nl_request_lock, flags); + ret = rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_LS, gfp_flag); - INIT_LIST_HEAD(&query->list); - query->seq = (u32)atomic_inc_return(&ib_nl_sa_request_seq); + if (ret) + goto out; - /* Put the request on the list first.*/ - spin_lock_irqsave(&ib_nl_request_lock, flags); + /* Put the request on the list.*/ delay = msecs_to_jiffies(sa_local_svc_timeout_ms); query->timeout = delay + jiffies; list_add_tail(&query->list, &ib_nl_request_list); /* Start the timeout if this is the only request */ if (ib_nl_request_list.next == &query->list) queue_delayed_work(ib_nl_wq, &ib_nl_timed_work, delay); - spin_unlock_irqrestore(&ib_nl_request_lock, flags); - ret = ib_nl_send_msg(query, gfp_mask); - if (ret) { - ret = -EIO; - /* Remove the request */ - spin_lock_irqsave(&ib_nl_request_lock, flags); - list_del(&query->list); - spin_unlock_irqrestore(&ib_nl_request_lock, flags); - } +out: + spin_unlock_irqrestore(&ib_nl_request_lock, flags); return ret; } @@ -1412,17 +1408,13 @@ void ib_sa_pack_path(struct sa_path_rec *rec, void *attribute) EXPORT_SYMBOL(ib_sa_pack_path); static bool ib_sa_opa_pathrecord_support(struct ib_sa_client *client, - struct ib_device *device, + struct ib_sa_device *sa_dev, u8 port_num) { - struct ib_sa_device *sa_dev = ib_get_client_data(device, &sa_client); struct ib_sa_port *port; unsigned long flags; bool ret = false; - if (!sa_dev) - return ret; - port = &sa_dev->port[port_num - sa_dev->start_port]; spin_lock_irqsave(&port->classport_lock, flags); if (!port->classport_info.valid) @@ -1450,8 +1442,8 @@ enum opa_pr_supported { * query is possible. */ static int opa_pr_query_possible(struct ib_sa_client *client, - struct ib_device *device, - u8 port_num, + struct ib_sa_device *sa_dev, + struct ib_device *device, u8 port_num, struct sa_path_rec *rec) { struct ib_port_attr port_attr; @@ -1459,7 +1451,7 @@ static int opa_pr_query_possible(struct ib_sa_client *client, if (ib_query_port(device, port_num, &port_attr)) return PR_NOT_SUPPORTED; - if (ib_sa_opa_pathrecord_support(client, device, port_num)) + if (ib_sa_opa_pathrecord_support(client, sa_dev, port_num)) return PR_OPA_SUPPORTED; if (port_attr.lid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) @@ -1574,7 +1566,8 @@ int ib_sa_path_rec_get(struct ib_sa_client *client, query->sa_query.port = port; if (rec->rec_type == SA_PATH_REC_TYPE_OPA) { - status = opa_pr_query_possible(client, device, port_num, rec); + status = opa_pr_query_possible(client, sa_dev, device, port_num, + rec); if (status == PR_NOT_SUPPORTED) { ret = -EINVAL; goto err1; @@ -2325,18 +2318,19 @@ static void ib_sa_event(struct ib_event_handler *handler, } } -static void ib_sa_add_one(struct ib_device *device) +static int ib_sa_add_one(struct ib_device *device) { struct ib_sa_device *sa_dev; int s, e, i; int count = 0; + int ret; s = rdma_start_port(device); e = rdma_end_port(device); sa_dev = kzalloc(struct_size(sa_dev, port, e - s + 1), GFP_KERNEL); if (!sa_dev) - return; + return -ENOMEM; sa_dev->start_port = s; sa_dev->end_port = e; @@ -2356,8 +2350,10 @@ static void ib_sa_add_one(struct ib_device *device) ib_register_mad_agent(device, i + s, IB_QPT_GSI, NULL, 0, send_handler, recv_handler, sa_dev, 0); - if (IS_ERR(sa_dev->port[i].agent)) + if (IS_ERR(sa_dev->port[i].agent)) { + ret = PTR_ERR(sa_dev->port[i].agent); goto err; + } INIT_WORK(&sa_dev->port[i].update_task, update_sm_ah); INIT_DELAYED_WORK(&sa_dev->port[i].ib_cpi_work, @@ -2366,8 +2362,10 @@ static void ib_sa_add_one(struct ib_device *device) count++; } - if (!count) + if (!count) { + ret = -EOPNOTSUPP; goto free; + } ib_set_client_data(device, &sa_client, sa_dev); @@ -2386,7 +2384,7 @@ static void ib_sa_add_one(struct ib_device *device) update_sm_ah(&sa_dev->port[i].update_task); } - return; + return 0; err: while (--i >= 0) { @@ -2395,7 +2393,7 @@ err: } free: kfree(sa_dev); - return; + return ret; } static void ib_sa_remove_one(struct ib_device *device, void *client_data) @@ -2403,9 +2401,6 @@ static void ib_sa_remove_one(struct ib_device *device, void *client_data) struct ib_sa_device *sa_dev = client_data; int i; - if (!sa_dev) - return; - ib_unregister_event_handler(&sa_dev->event_handler); flush_workqueue(ib_wq); diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c index 087682e6969e..c11e50510e49 100644 --- a/drivers/infiniband/core/sysfs.c +++ b/drivers/infiniband/core/sysfs.c @@ -58,7 +58,7 @@ struct ib_port { struct ib_device *ibdev; struct gid_attr_group *gid_attr_group; struct attribute_group gid_group; - struct attribute_group pkey_group; + struct attribute_group *pkey_group; struct attribute_group *pma_table; struct attribute_group *hw_stats_ag; struct rdma_hw_stats *hw_stats; @@ -681,11 +681,16 @@ static void ib_port_release(struct kobject *kobj) kfree(p->gid_group.attrs); } - if (p->pkey_group.attrs) { - for (i = 0; (a = p->pkey_group.attrs[i]); ++i) - kfree(a); + if (p->pkey_group) { + if (p->pkey_group->attrs) { + for (i = 0; (a = p->pkey_group->attrs[i]); ++i) + kfree(a); + + kfree(p->pkey_group->attrs); + } - kfree(p->pkey_group.attrs); + kfree(p->pkey_group); + p->pkey_group = NULL; } kfree(p); @@ -1058,8 +1063,7 @@ static int add_port(struct ib_core_device *coredev, int port_num) coredev->ports_kobj, "%d", port_num); if (ret) { - kfree(p); - return ret; + goto err_put; } p->gid_attr_group = kzalloc(sizeof(*p->gid_attr_group), GFP_KERNEL); @@ -1072,8 +1076,7 @@ static int add_port(struct ib_core_device *coredev, int port_num) ret = kobject_init_and_add(&p->gid_attr_group->kobj, &gid_attr_type, &p->kobj, "gid_attrs"); if (ret) { - kfree(p->gid_attr_group); - goto err_put; + goto err_put_gid_attrs; } if (device->ops.process_mad && is_full_dev) { @@ -1120,17 +1123,26 @@ static int add_port(struct ib_core_device *coredev, int port_num) if (ret) goto err_free_gid_type; - p->pkey_group.name = "pkeys"; - p->pkey_group.attrs = alloc_group_attrs(show_port_pkey, - attr.pkey_tbl_len); - if (!p->pkey_group.attrs) { - ret = -ENOMEM; - goto err_remove_gid_type; + if (attr.pkey_tbl_len) { + p->pkey_group = kzalloc(sizeof(*p->pkey_group), GFP_KERNEL); + if (!p->pkey_group) { + ret = -ENOMEM; + goto err_remove_gid_type; + } + + p->pkey_group->name = "pkeys"; + p->pkey_group->attrs = alloc_group_attrs(show_port_pkey, + attr.pkey_tbl_len); + if (!p->pkey_group->attrs) { + ret = -ENOMEM; + goto err_free_pkey_group; + } + + ret = sysfs_create_group(&p->kobj, p->pkey_group); + if (ret) + goto err_free_pkey; } - ret = sysfs_create_group(&p->kobj, &p->pkey_group); - if (ret) - goto err_free_pkey; if (device->ops.init_port && is_full_dev) { ret = device->ops.init_port(device, port_num, &p->kobj); @@ -1152,14 +1164,20 @@ static int add_port(struct ib_core_device *coredev, int port_num) return 0; err_remove_pkey: - sysfs_remove_group(&p->kobj, &p->pkey_group); + if (p->pkey_group) + sysfs_remove_group(&p->kobj, p->pkey_group); err_free_pkey: - for (i = 0; i < attr.pkey_tbl_len; ++i) - kfree(p->pkey_group.attrs[i]); + if (p->pkey_group) { + for (i = 0; i < attr.pkey_tbl_len; ++i) + kfree(p->pkey_group->attrs[i]); - kfree(p->pkey_group.attrs); - p->pkey_group.attrs = NULL; + kfree(p->pkey_group->attrs); + p->pkey_group->attrs = NULL; + } + +err_free_pkey_group: + kfree(p->pkey_group); err_remove_gid_type: sysfs_remove_group(&p->gid_attr_group->kobj, @@ -1319,7 +1337,8 @@ void ib_free_port_attrs(struct ib_core_device *coredev) if (port->pma_table) sysfs_remove_group(p, port->pma_table); - sysfs_remove_group(p, &port->pkey_group); + if (port->pkey_group) + sysfs_remove_group(p, port->pkey_group); sysfs_remove_group(p, &port->gid_group); sysfs_remove_group(&port->gid_attr_group->kobj, &port->gid_attr_group->ndev); @@ -1404,8 +1423,10 @@ int ib_port_register_module_stat(struct ib_device *device, u8 port_num, ret = kobject_init_and_add(kobj, ktype, &port->kobj, "%s", name); - if (ret) + if (ret) { + kobject_put(kobj); return ret; + } } return 0; diff --git a/drivers/infiniband/core/trace.c b/drivers/infiniband/core/trace.c index 6c3514beac4d..31e7860d35bf 100644 --- a/drivers/infiniband/core/trace.c +++ b/drivers/infiniband/core/trace.c @@ -9,6 +9,4 @@ #define CREATE_TRACE_POINTS -#include <rdma/ib_verbs.h> - #include <trace/events/rdma_core.h> diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 16b6cf57fa85..1d184ea05eba 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -52,6 +52,7 @@ #include <rdma/rdma_cm_ib.h> #include <rdma/ib_addr.h> #include <rdma/ib.h> +#include <rdma/ib_cm.h> #include <rdma/rdma_netlink.h> #include "core_priv.h" @@ -360,6 +361,9 @@ static int ucma_event_handler(struct rdma_cm_id *cm_id, ucma_copy_conn_event(&uevent->resp.param.conn, &event->param.conn); + uevent->resp.ece.vendor_id = event->ece.vendor_id; + uevent->resp.ece.attr_mod = event->ece.attr_mod; + if (event->event == RDMA_CM_EVENT_CONNECT_REQUEST) { if (!ctx->backlog) { ret = -ENOMEM; @@ -404,7 +408,8 @@ static ssize_t ucma_get_event(struct ucma_file *file, const char __user *inbuf, * Old 32 bit user space does not send the 4 byte padding in the * reserved field. We don't care, allow it to keep working. */ - if (out_len < sizeof(uevent->resp) - sizeof(uevent->resp.reserved)) + if (out_len < sizeof(uevent->resp) - sizeof(uevent->resp.reserved) - + sizeof(uevent->resp.ece)) return -ENOSPC; if (copy_from_user(&cmd, inbuf, sizeof(cmd))) @@ -789,7 +794,7 @@ static void ucma_copy_ib_route(struct rdma_ucm_query_route_resp *resp, case 2: ib_copy_path_rec_to_user(&resp->ib_route[1], &route->path_rec[1]); - /* fall through */ + fallthrough; case 1: ib_copy_path_rec_to_user(&resp->ib_route[0], &route->path_rec[0]); @@ -815,7 +820,7 @@ static void ucma_copy_iboe_route(struct rdma_ucm_query_route_resp *resp, case 2: ib_copy_path_rec_to_user(&resp->ib_route[1], &route->path_rec[1]); - /* fall through */ + fallthrough; case 1: ib_copy_path_rec_to_user(&resp->ib_route[0], &route->path_rec[0]); @@ -845,7 +850,7 @@ static ssize_t ucma_query_route(struct ucma_file *file, struct sockaddr *addr; int ret = 0; - if (out_len < sizeof(resp)) + if (out_len < offsetof(struct rdma_ucm_query_route_resp, ibdev_index)) return -ENOSPC; if (copy_from_user(&cmd, inbuf, sizeof(cmd))) @@ -869,6 +874,7 @@ static ssize_t ucma_query_route(struct ucma_file *file, goto out; resp.node_guid = (__force __u64) ctx->cm_id->device->node_guid; + resp.ibdev_index = ctx->cm_id->device->index; resp.port_num = ctx->cm_id->port_num; if (rdma_cap_ib_sa(ctx->cm_id->device, ctx->cm_id->port_num)) @@ -880,8 +886,8 @@ static ssize_t ucma_query_route(struct ucma_file *file, out: mutex_unlock(&ctx->mutex); - if (copy_to_user(u64_to_user_ptr(cmd.response), - &resp, sizeof(resp))) + if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, + min_t(size_t, out_len, sizeof(resp)))) ret = -EFAULT; ucma_put_ctx(ctx); @@ -895,6 +901,7 @@ static void ucma_query_device_addr(struct rdma_cm_id *cm_id, return; resp->node_guid = (__force __u64) cm_id->device->node_guid; + resp->ibdev_index = cm_id->device->index; resp->port_num = cm_id->port_num; resp->pkey = (__force __u16) cpu_to_be16( ib_addr_get_pkey(&cm_id->route.addr.dev_addr)); @@ -907,7 +914,7 @@ static ssize_t ucma_query_addr(struct ucma_context *ctx, struct sockaddr *addr; int ret = 0; - if (out_len < sizeof(resp)) + if (out_len < offsetof(struct rdma_ucm_query_addr_resp, ibdev_index)) return -ENOSPC; memset(&resp, 0, sizeof resp); @@ -922,7 +929,7 @@ static ssize_t ucma_query_addr(struct ucma_context *ctx, ucma_query_device_addr(ctx->cm_id, &resp); - if (copy_to_user(response, &resp, sizeof(resp))) + if (copy_to_user(response, &resp, min_t(size_t, out_len, sizeof(resp)))) ret = -EFAULT; return ret; @@ -974,7 +981,7 @@ static ssize_t ucma_query_gid(struct ucma_context *ctx, struct sockaddr_ib *addr; int ret = 0; - if (out_len < sizeof(resp)) + if (out_len < offsetof(struct rdma_ucm_query_addr_resp, ibdev_index)) return -ENOSPC; memset(&resp, 0, sizeof resp); @@ -1007,7 +1014,7 @@ static ssize_t ucma_query_gid(struct ucma_context *ctx, &ctx->cm_id->route.addr.dst_addr); } - if (copy_to_user(response, &resp, sizeof(resp))) + if (copy_to_user(response, &resp, min_t(size_t, out_len, sizeof(resp)))) ret = -EFAULT; return ret; @@ -1070,12 +1077,17 @@ static void ucma_copy_conn_param(struct rdma_cm_id *id, static ssize_t ucma_connect(struct ucma_file *file, const char __user *inbuf, int in_len, int out_len) { - struct rdma_ucm_connect cmd; struct rdma_conn_param conn_param; + struct rdma_ucm_ece ece = {}; + struct rdma_ucm_connect cmd; struct ucma_context *ctx; + size_t in_size; int ret; - if (copy_from_user(&cmd, inbuf, sizeof(cmd))) + if (in_len < offsetofend(typeof(cmd), reserved)) + return -EINVAL; + in_size = min_t(size_t, in_len, sizeof(cmd)); + if (copy_from_user(&cmd, inbuf, in_size)) return -EFAULT; if (!cmd.conn_param.valid) @@ -1086,8 +1098,13 @@ static ssize_t ucma_connect(struct ucma_file *file, const char __user *inbuf, return PTR_ERR(ctx); ucma_copy_conn_param(ctx->cm_id, &conn_param, &cmd.conn_param); + if (offsetofend(typeof(cmd), ece) <= in_size) { + ece.vendor_id = cmd.ece.vendor_id; + ece.attr_mod = cmd.ece.attr_mod; + } + mutex_lock(&ctx->mutex); - ret = rdma_connect(ctx->cm_id, &conn_param); + ret = rdma_connect_ece(ctx->cm_id, &conn_param, &ece); mutex_unlock(&ctx->mutex); ucma_put_ctx(ctx); return ret; @@ -1121,28 +1138,38 @@ static ssize_t ucma_accept(struct ucma_file *file, const char __user *inbuf, { struct rdma_ucm_accept cmd; struct rdma_conn_param conn_param; + struct rdma_ucm_ece ece = {}; struct ucma_context *ctx; + size_t in_size; int ret; - if (copy_from_user(&cmd, inbuf, sizeof(cmd))) + if (in_len < offsetofend(typeof(cmd), reserved)) + return -EINVAL; + in_size = min_t(size_t, in_len, sizeof(cmd)); + if (copy_from_user(&cmd, inbuf, in_size)) return -EFAULT; ctx = ucma_get_ctx_dev(file, cmd.id); if (IS_ERR(ctx)) return PTR_ERR(ctx); + if (offsetofend(typeof(cmd), ece) <= in_size) { + ece.vendor_id = cmd.ece.vendor_id; + ece.attr_mod = cmd.ece.attr_mod; + } + if (cmd.conn_param.valid) { ucma_copy_conn_param(ctx->cm_id, &conn_param, &cmd.conn_param); mutex_lock(&file->mut); mutex_lock(&ctx->mutex); - ret = __rdma_accept(ctx->cm_id, &conn_param, NULL); + ret = __rdma_accept_ece(ctx->cm_id, &conn_param, NULL, &ece); mutex_unlock(&ctx->mutex); if (!ret) ctx->uid = cmd.uid; mutex_unlock(&file->mut); } else { mutex_lock(&ctx->mutex); - ret = __rdma_accept(ctx->cm_id, NULL, NULL); + ret = __rdma_accept_ece(ctx->cm_id, NULL, NULL, &ece); mutex_unlock(&ctx->mutex); } ucma_put_ctx(ctx); @@ -1159,12 +1186,24 @@ static ssize_t ucma_reject(struct ucma_file *file, const char __user *inbuf, if (copy_from_user(&cmd, inbuf, sizeof(cmd))) return -EFAULT; + if (!cmd.reason) + cmd.reason = IB_CM_REJ_CONSUMER_DEFINED; + + switch (cmd.reason) { + case IB_CM_REJ_CONSUMER_DEFINED: + case IB_CM_REJ_VENDOR_OPTION_NOT_SUPPORTED: + break; + default: + return -EINVAL; + } + ctx = ucma_get_ctx_dev(file, cmd.id); if (IS_ERR(ctx)) return PTR_ERR(ctx); mutex_lock(&ctx->mutex); - ret = rdma_reject(ctx->cm_id, cmd.private_data, cmd.private_data_len); + ret = rdma_reject(ctx->cm_id, cmd.private_data, cmd.private_data_len, + cmd.reason); mutex_unlock(&ctx->mutex); ucma_put_ctx(ctx); return ret; diff --git a/drivers/infiniband/core/ud_header.c b/drivers/infiniband/core/ud_header.c index 29a45d2f8898..d65d541b9a25 100644 --- a/drivers/infiniband/core/ud_header.c +++ b/drivers/infiniband/core/ud_header.c @@ -41,7 +41,7 @@ #define STRUCT_FIELD(header, field) \ .struct_offset_bytes = offsetof(struct ib_unpacked_ ## header, field), \ - .struct_size_bytes = sizeof ((struct ib_unpacked_ ## header *) 0)->field, \ + .struct_size_bytes = sizeof_field(struct ib_unpacked_ ## header, field), \ .field_name = #header ":" #field static const struct ib_field lrh_table[] = { diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c index 82455a1392f1..831bff8d52e5 100644 --- a/drivers/infiniband/core/umem.c +++ b/drivers/infiniband/core/umem.c @@ -261,6 +261,7 @@ struct ib_umem *ib_umem_get(struct ib_device *device, unsigned long addr, sg = umem->sg_head.sgl; while (npages) { + cond_resched(); ret = pin_user_pages_fast(cur_base, min_t(unsigned long, npages, PAGE_SIZE / diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c index 3b1e627d9a8d..cc6b4befde7c 100644 --- a/drivers/infiniband/core/umem_odp.c +++ b/drivers/infiniband/core/umem_odp.c @@ -152,6 +152,7 @@ EXPORT_SYMBOL(ib_umem_odp_alloc_implicit); * ib_alloc_implicit_odp_umem() * @addr: The starting userspace VA * @size: The length of the userspace VA + * @ops: MMU interval ops, currently only @invalidate */ struct ib_umem_odp * ib_umem_odp_alloc_child(struct ib_umem_odp *root, unsigned long addr, @@ -213,6 +214,7 @@ EXPORT_SYMBOL(ib_umem_odp_alloc_child); * @addr: userspace virtual address to start at * @size: length of region to pin * @access: IB_ACCESS_xxx flags for memory being pinned + * @ops: MMU interval ops, currently only @invalidate * * The driver should use when the access flags indicate ODP memory. It avoids * pinning, instead, stores the mm for future page fault handling in @@ -429,7 +431,7 @@ int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 user_virt, ALIGN(bcnt, PAGE_SIZE) / PAGE_SIZE, PAGE_SIZE / sizeof(struct page *)); - down_read(&owning_mm->mmap_sem); + mmap_read_lock(owning_mm); /* * Note: this might result in redundent page getting. We can * avoid this by checking dma_list to be 0 before calling @@ -437,10 +439,10 @@ int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 user_virt, * complex (and doesn't gain us much performance in most use * cases). */ - npages = get_user_pages_remote(owning_process, owning_mm, + npages = get_user_pages_remote(owning_mm, user_virt, gup_num_pages, flags, local_page_list, NULL, NULL); - up_read(&owning_mm->mmap_sem); + mmap_read_unlock(owning_mm); if (npages < 0) { if (npages != -EAGAIN) diff --git a/drivers/infiniband/core/user_mad.c b/drivers/infiniband/core/user_mad.c index da229eab5903..b0d0b522cc76 100644 --- a/drivers/infiniband/core/user_mad.c +++ b/drivers/infiniband/core/user_mad.c @@ -142,7 +142,7 @@ static dev_t dynamic_issm_dev; static DEFINE_IDA(umad_ida); -static void ib_umad_add_one(struct ib_device *device); +static int ib_umad_add_one(struct ib_device *device); static void ib_umad_remove_one(struct ib_device *device, void *client_data); static void ib_umad_dev_free(struct kref *kref) @@ -1352,37 +1352,41 @@ static void ib_umad_kill_port(struct ib_umad_port *port) put_device(&port->dev); } -static void ib_umad_add_one(struct ib_device *device) +static int ib_umad_add_one(struct ib_device *device) { struct ib_umad_device *umad_dev; int s, e, i; int count = 0; + int ret; s = rdma_start_port(device); e = rdma_end_port(device); umad_dev = kzalloc(struct_size(umad_dev, ports, e - s + 1), GFP_KERNEL); if (!umad_dev) - return; + return -ENOMEM; kref_init(&umad_dev->kref); for (i = s; i <= e; ++i) { if (!rdma_cap_ib_mad(device, i)) continue; - if (ib_umad_init_port(device, i, umad_dev, - &umad_dev->ports[i - s])) + ret = ib_umad_init_port(device, i, umad_dev, + &umad_dev->ports[i - s]); + if (ret) goto err; count++; } - if (!count) + if (!count) { + ret = -EOPNOTSUPP; goto free; + } ib_set_client_data(device, &umad_client, umad_dev); - return; + return 0; err: while (--i >= s) { @@ -1394,6 +1398,7 @@ err: free: /* balances kref_init */ ib_umad_dev_put(umad_dev); + return ret; } static void ib_umad_remove_one(struct ib_device *device, void *client_data) @@ -1401,9 +1406,6 @@ static void ib_umad_remove_one(struct ib_device *device, void *client_data) struct ib_umad_device *umad_dev = client_data; unsigned int i; - if (!umad_dev) - return; - rdma_for_each_port (device, i) { if (rdma_cap_ib_mad(device, i)) ib_umad_kill_port( diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h index 3d189c7ee59e..53a10479958b 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -142,7 +142,7 @@ struct ib_uverbs_file { * ucontext_lock held */ struct ib_ucontext *ucontext; - struct ib_uverbs_async_event_file *async_file; + struct ib_uverbs_async_event_file *default_async_file; struct list_head list; /* @@ -180,6 +180,7 @@ struct ib_uverbs_mcast_entry { struct ib_uevent_object { struct ib_uobject uobject; + struct ib_uverbs_async_event_file *event_file; /* List member for ib_uverbs_async_event_file list */ struct list_head event_list; u32 events_reported; @@ -296,6 +297,24 @@ static inline u32 make_port_cap_flags(const struct ib_port_attr *attr) return res; } +static inline struct ib_uverbs_async_event_file * +ib_uverbs_get_async_event(struct uverbs_attr_bundle *attrs, + u16 id) +{ + struct ib_uobject *async_ev_file_uobj; + struct ib_uverbs_async_event_file *async_ev_file; + + async_ev_file_uobj = uverbs_attr_get_uobject(attrs, id); + if (IS_ERR(async_ev_file_uobj)) + async_ev_file = READ_ONCE(attrs->ufile->default_async_file); + else + async_ev_file = container_of(async_ev_file_uobj, + struct ib_uverbs_async_event_file, + uobj); + if (async_ev_file) + uverbs_uobject_get(&async_ev_file->uobj); + return async_ev_file; +} void copy_port_attr_to_resp(struct ib_port_attr *attr, struct ib_uverbs_query_port_resp *resp, diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index 060b4ebbd2ba..2fbc583d5bdd 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -311,7 +311,7 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs) return 0; err_uobj: - rdma_alloc_abort_uobject(uobj, attrs); + rdma_alloc_abort_uobject(uobj, attrs, false); err_ucontext: kfree(attrs->context); attrs->context = NULL; @@ -356,8 +356,6 @@ static void copy_query_dev_fields(struct ib_ucontext *ucontext, resp->max_mcast_qp_attach = attr->max_mcast_qp_attach; resp->max_total_mcast_qp_attach = attr->max_total_mcast_qp_attach; resp->max_ah = attr->max_ah; - resp->max_fmr = attr->max_fmr; - resp->max_map_per_fmr = attr->max_map_per_fmr; resp->max_srq = attr->max_srq; resp->max_srq_wr = attr->max_srq_wr; resp->max_srq_sge = attr->max_srq_sge; @@ -417,8 +415,8 @@ static int ib_uverbs_query_port(struct uverbs_attr_bundle *attrs) static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs) { + struct ib_uverbs_alloc_pd_resp resp = {}; struct ib_uverbs_alloc_pd cmd; - struct ib_uverbs_alloc_pd_resp resp; struct ib_uobject *uobj; struct ib_pd *pd; int ret; @@ -440,29 +438,20 @@ static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs) pd->device = ib_dev; pd->uobject = uobj; - pd->__internal_mr = NULL; atomic_set(&pd->usecnt, 0); pd->res.type = RDMA_RESTRACK_PD; ret = ib_dev->ops.alloc_pd(pd, &attrs->driver_udata); if (ret) goto err_alloc; - - uobj->object = pd; - memset(&resp, 0, sizeof resp); - resp.pd_handle = uobj->id; rdma_restrack_uadd(&pd->res); - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_copy; + uobj->object = pd; + uobj_finalize_uobj_create(uobj, attrs); - rdma_alloc_commit_uobject(uobj, attrs); - return 0; + resp.pd_handle = uobj->id; + return uverbs_response(attrs, &resp, sizeof(resp)); -err_copy: - ib_dealloc_pd_user(pd, uverbs_get_cleared_udata(attrs)); - pd = NULL; err_alloc: kfree(pd); err: @@ -570,15 +559,15 @@ static void xrcd_table_delete(struct ib_uverbs_device *dev, static int ib_uverbs_open_xrcd(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_device *ibudev = attrs->ufile->device; + struct ib_uverbs_open_xrcd_resp resp = {}; struct ib_uverbs_open_xrcd cmd; - struct ib_uverbs_open_xrcd_resp resp; struct ib_uxrcd_object *obj; struct ib_xrcd *xrcd = NULL; - struct fd f = {NULL, 0}; struct inode *inode = NULL; - int ret = 0; int new_xrcd = 0; struct ib_device *ib_dev; + struct fd f = {}; + int ret; ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) @@ -616,24 +605,16 @@ static int ib_uverbs_open_xrcd(struct uverbs_attr_bundle *attrs) } if (!xrcd) { - xrcd = ib_dev->ops.alloc_xrcd(ib_dev, &attrs->driver_udata); + xrcd = ib_alloc_xrcd_user(ib_dev, inode, &attrs->driver_udata); if (IS_ERR(xrcd)) { ret = PTR_ERR(xrcd); goto err; } - - xrcd->inode = inode; - xrcd->device = ib_dev; - atomic_set(&xrcd->usecnt, 0); - mutex_init(&xrcd->tgt_qp_mutex); - INIT_LIST_HEAD(&xrcd->tgt_qp_list); new_xrcd = 1; } atomic_set(&obj->refcnt, 0); obj->uobject.object = xrcd; - memset(&resp, 0, sizeof resp); - resp.xrcd_handle = obj->uobject.id; if (inode) { if (new_xrcd) { @@ -645,27 +626,17 @@ static int ib_uverbs_open_xrcd(struct uverbs_attr_bundle *attrs) atomic_inc(&xrcd->usecnt); } - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_copy; - if (f.file) fdput(f); mutex_unlock(&ibudev->xrcd_tree_mutex); + uobj_finalize_uobj_create(&obj->uobject, attrs); - rdma_alloc_commit_uobject(&obj->uobject, attrs); - return 0; - -err_copy: - if (inode) { - if (new_xrcd) - xrcd_table_delete(ibudev, inode); - atomic_dec(&xrcd->usecnt); - } + resp.xrcd_handle = obj->uobject.id; + return uverbs_response(attrs, &resp, sizeof(resp)); err_dealloc_xrcd: - ib_dealloc_xrcd(xrcd, uverbs_get_cleared_udata(attrs)); + ib_dealloc_xrcd_user(xrcd, uverbs_get_cleared_udata(attrs)); err: uobj_alloc_abort(&obj->uobject, attrs); @@ -703,7 +674,7 @@ int ib_uverbs_dealloc_xrcd(struct ib_uobject *uobject, struct ib_xrcd *xrcd, if (inode && !atomic_dec_and_test(&xrcd->usecnt)) return 0; - ret = ib_dealloc_xrcd(xrcd, &attrs->driver_udata); + ret = ib_dealloc_xrcd_user(xrcd, &attrs->driver_udata); if (ib_is_destroy_retryable(ret, why, uobject)) { atomic_inc(&xrcd->usecnt); @@ -718,8 +689,8 @@ int ib_uverbs_dealloc_xrcd(struct ib_uobject *uobject, struct ib_xrcd *xrcd, static int ib_uverbs_reg_mr(struct uverbs_attr_bundle *attrs) { + struct ib_uverbs_reg_mr_resp resp = {}; struct ib_uverbs_reg_mr cmd; - struct ib_uverbs_reg_mr_resp resp; struct ib_uobject *uobj; struct ib_pd *pd; struct ib_mr *mr; @@ -772,30 +743,20 @@ static int ib_uverbs_reg_mr(struct uverbs_attr_bundle *attrs) mr->uobject = uobj; atomic_inc(&pd->usecnt); mr->res.type = RDMA_RESTRACK_MR; + mr->iova = cmd.hca_va; rdma_restrack_uadd(&mr->res); uobj->object = mr; - - memset(&resp, 0, sizeof resp); - resp.lkey = mr->lkey; - resp.rkey = mr->rkey; - resp.mr_handle = uobj->id; - - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_copy; - uobj_put_obj_read(pd); + uobj_finalize_uobj_create(uobj, attrs); - rdma_alloc_commit_uobject(uobj, attrs); - return 0; - -err_copy: - ib_dereg_mr_user(mr, uverbs_get_cleared_udata(attrs)); + resp.lkey = mr->lkey; + resp.rkey = mr->rkey; + resp.mr_handle = uobj->id; + return uverbs_response(attrs, &resp, sizeof(resp)); err_put: uobj_put_obj_read(pd); - err_free: uobj_alloc_abort(uobj, attrs); return ret; @@ -863,6 +824,9 @@ static int ib_uverbs_rereg_mr(struct uverbs_attr_bundle *attrs) atomic_dec(&old_pd->usecnt); } + if (cmd.flags & IB_MR_REREG_TRANS) + mr->iova = cmd.hca_va; + memset(&resp, 0, sizeof(resp)); resp.lkey = mr->lkey; resp.rkey = mr->rkey; @@ -932,21 +896,13 @@ static int ib_uverbs_alloc_mw(struct uverbs_attr_bundle *attrs) atomic_inc(&pd->usecnt); uobj->object = mw; + uobj_put_obj_read(pd); + uobj_finalize_uobj_create(uobj, attrs); - memset(&resp, 0, sizeof(resp)); - resp.rkey = mw->rkey; + resp.rkey = mw->rkey; resp.mw_handle = uobj->id; + return uverbs_response(attrs, &resp, sizeof(resp)); - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_copy; - - uobj_put_obj_read(pd); - rdma_alloc_commit_uobject(uobj, attrs); - return 0; - -err_copy: - uverbs_dealloc_mw(mw); err_put: uobj_put_obj_read(pd); err_free: @@ -983,40 +939,33 @@ static int ib_uverbs_create_comp_channel(struct uverbs_attr_bundle *attrs) if (IS_ERR(uobj)) return PTR_ERR(uobj); - resp.fd = uobj->id; - ev_file = container_of(uobj, struct ib_uverbs_completion_event_file, uobj); ib_uverbs_init_event_queue(&ev_file->ev_queue); + uobj_finalize_uobj_create(uobj, attrs); - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) { - uobj_alloc_abort(uobj, attrs); - return ret; - } - - rdma_alloc_commit_uobject(uobj, attrs); - return 0; + resp.fd = uobj->id; + return uverbs_response(attrs, &resp, sizeof(resp)); } -static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs, - struct ib_uverbs_ex_create_cq *cmd) +static int create_cq(struct uverbs_attr_bundle *attrs, + struct ib_uverbs_ex_create_cq *cmd) { struct ib_ucq_object *obj; struct ib_uverbs_completion_event_file *ev_file = NULL; struct ib_cq *cq; int ret; - struct ib_uverbs_ex_create_cq_resp resp; + struct ib_uverbs_ex_create_cq_resp resp = {}; struct ib_cq_init_attr attr = {}; struct ib_device *ib_dev; if (cmd->comp_vector >= attrs->ufile->device->num_comp_vectors) - return ERR_PTR(-EINVAL); + return -EINVAL; obj = (struct ib_ucq_object *)uobj_alloc(UVERBS_OBJECT_CQ, attrs, &ib_dev); if (IS_ERR(obj)) - return obj; + return PTR_ERR(obj); if (cmd->comp_channel >= 0) { ev_file = ib_uverbs_lookup_comp_file(cmd->comp_channel, attrs); @@ -1045,47 +994,38 @@ static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs, cq->event_handler = ib_uverbs_cq_event_handler; cq->cq_context = ev_file ? &ev_file->ev_queue : NULL; atomic_set(&cq->usecnt, 0); + cq->res.type = RDMA_RESTRACK_CQ; ret = ib_dev->ops.create_cq(cq, &attr, &attrs->driver_udata); if (ret) goto err_free; + rdma_restrack_uadd(&cq->res); obj->uevent.uobject.object = cq; - memset(&resp, 0, sizeof resp); + obj->uevent.event_file = READ_ONCE(attrs->ufile->default_async_file); + if (obj->uevent.event_file) + uverbs_uobject_get(&obj->uevent.event_file->uobj); + uobj_finalize_uobj_create(&obj->uevent.uobject, attrs); + resp.base.cq_handle = obj->uevent.uobject.id; - resp.base.cqe = cq->cqe; + resp.base.cqe = cq->cqe; resp.response_length = uverbs_response_length(attrs, sizeof(resp)); + return uverbs_response(attrs, &resp, sizeof(resp)); - cq->res.type = RDMA_RESTRACK_CQ; - rdma_restrack_uadd(&cq->res); - - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_cb; - - rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs); - return obj; - -err_cb: - ib_destroy_cq_user(cq, uverbs_get_cleared_udata(attrs)); - cq = NULL; err_free: kfree(cq); err_file: if (ev_file) ib_uverbs_release_ucq(ev_file, obj); - err: uobj_alloc_abort(&obj->uevent.uobject, attrs); - - return ERR_PTR(ret); + return ret; } static int ib_uverbs_create_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_cq cmd; struct ib_uverbs_ex_create_cq cmd_ex; - struct ib_ucq_object *obj; int ret; ret = uverbs_request(attrs, &cmd, sizeof(cmd)); @@ -1098,14 +1038,12 @@ static int ib_uverbs_create_cq(struct uverbs_attr_bundle *attrs) cmd_ex.comp_vector = cmd.comp_vector; cmd_ex.comp_channel = cmd.comp_channel; - obj = create_cq(attrs, &cmd_ex); - return PTR_ERR_OR_ZERO(obj); + return create_cq(attrs, &cmd_ex); } static int ib_uverbs_ex_create_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_ex_create_cq cmd; - struct ib_ucq_object *obj; int ret; ret = uverbs_request(attrs, &cmd, sizeof(cmd)); @@ -1118,8 +1056,7 @@ static int ib_uverbs_ex_create_cq(struct uverbs_attr_bundle *attrs) if (cmd.reserved) return -EINVAL; - obj = create_cq(attrs, &cmd); - return PTR_ERR_OR_ZERO(obj); + return create_cq(attrs, &cmd); } static int ib_uverbs_resize_cq(struct uverbs_attr_bundle *attrs) @@ -1127,7 +1064,7 @@ static int ib_uverbs_resize_cq(struct uverbs_attr_bundle *attrs) struct ib_uverbs_resize_cq cmd; struct ib_uverbs_resize_cq_resp resp = {}; struct ib_cq *cq; - int ret = -EINVAL; + int ret; ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) @@ -1294,7 +1231,7 @@ static int create_qp(struct uverbs_attr_bundle *attrs, struct ib_srq *srq = NULL; struct ib_qp *qp; struct ib_qp_init_attr attr = {}; - struct ib_uverbs_ex_create_qp_resp resp; + struct ib_uverbs_ex_create_qp_resp resp = {}; int ret; struct ib_rwq_ind_table *ind_tbl = NULL; bool has_sq = true; @@ -1460,20 +1397,9 @@ static int create_qp(struct uverbs_attr_bundle *attrs, } obj->uevent.uobject.object = qp; - - memset(&resp, 0, sizeof resp); - resp.base.qpn = qp->qp_num; - resp.base.qp_handle = obj->uevent.uobject.id; - resp.base.max_recv_sge = attr.cap.max_recv_sge; - resp.base.max_send_sge = attr.cap.max_send_sge; - resp.base.max_recv_wr = attr.cap.max_recv_wr; - resp.base.max_send_wr = attr.cap.max_send_wr; - resp.base.max_inline_data = attr.cap.max_inline_data; - resp.response_length = uverbs_response_length(attrs, sizeof(resp)); - - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_cb; + obj->uevent.event_file = READ_ONCE(attrs->ufile->default_async_file); + if (obj->uevent.event_file) + uverbs_uobject_get(&obj->uevent.event_file->uobj); if (xrcd) { obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, @@ -1495,9 +1421,18 @@ static int create_qp(struct uverbs_attr_bundle *attrs, UVERBS_LOOKUP_READ); if (ind_tbl) uobj_put_obj_read(ind_tbl); + uobj_finalize_uobj_create(&obj->uevent.uobject, attrs); + + resp.base.qpn = qp->qp_num; + resp.base.qp_handle = obj->uevent.uobject.id; + resp.base.max_recv_sge = attr.cap.max_recv_sge; + resp.base.max_send_sge = attr.cap.max_send_sge; + resp.base.max_recv_wr = attr.cap.max_recv_wr; + resp.base.max_send_wr = attr.cap.max_send_wr; + resp.base.max_inline_data = attr.cap.max_inline_data; + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); + return uverbs_response(attrs, &resp, sizeof(resp)); - rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs); - return 0; err_cb: ib_destroy_qp_user(qp, uverbs_get_cleared_udata(attrs)); @@ -1570,14 +1505,14 @@ static int ib_uverbs_ex_create_qp(struct uverbs_attr_bundle *attrs) static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs) { + struct ib_uverbs_create_qp_resp resp = {}; struct ib_uverbs_open_qp cmd; - struct ib_uverbs_create_qp_resp resp; struct ib_uqp_object *obj; struct ib_xrcd *xrcd; - struct ib_uobject *uninitialized_var(xrcd_uobj); struct ib_qp *qp; struct ib_qp_open_attr attr = {}; int ret; + struct ib_uobject *xrcd_uobj; struct ib_device *ib_dev; ret = uverbs_request(attrs, &cmd, sizeof(cmd)); @@ -1617,24 +1552,16 @@ static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs) obj->uevent.uobject.object = qp; obj->uevent.uobject.user_handle = cmd.user_handle; - memset(&resp, 0, sizeof resp); - resp.qpn = qp->qp_num; - resp.qp_handle = obj->uevent.uobject.id; - - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_destroy; - obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, uobject); atomic_inc(&obj->uxrcd->refcnt); qp->uobject = obj; uobj_put_read(xrcd_uobj); + uobj_finalize_uobj_create(&obj->uevent.uobject, attrs); - rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs); - return 0; + resp.qpn = qp->qp_num; + resp.qp_handle = obj->uevent.uobject.id; + return uverbs_response(attrs, &resp, sizeof(resp)); -err_destroy: - ib_destroy_qp_user(qp, uverbs_get_cleared_udata(attrs)); err_xrcd: uobj_put_read(xrcd_uobj); err_put: @@ -1970,7 +1897,7 @@ static int ib_uverbs_ex_modify_qp(struct uverbs_attr_bundle *attrs) * Last bit is reserved for extending the attr_mask by * using another field. */ - BUILD_BUG_ON(IB_USER_LAST_QP_ATTR_MASK == (1 << 31)); + BUILD_BUG_ON(IB_USER_LAST_QP_ATTR_MASK == (1ULL << 31)); if (cmd.base.attr_mask & ~((IB_USER_LAST_QP_ATTR_MASK << 1) - 1)) @@ -2470,24 +2397,14 @@ static int ib_uverbs_create_ah(struct uverbs_attr_bundle *attrs) ah->uobject = uobj; uobj->user_handle = cmd.user_handle; uobj->object = ah; - - resp.ah_handle = uobj->id; - - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_copy; - uobj_put_obj_read(pd); - rdma_alloc_commit_uobject(uobj, attrs); - return 0; + uobj_finalize_uobj_create(uobj, attrs); -err_copy: - rdma_destroy_ah_user(ah, RDMA_DESTROY_AH_SLEEPABLE, - uverbs_get_cleared_udata(attrs)); + resp.ah_handle = uobj->id; + return uverbs_response(attrs, &resp, sizeof(resp)); err_put: uobj_put_obj_read(pd); - err: uobj_alloc_abort(uobj, attrs); return ret; @@ -2954,11 +2871,11 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs) wq_init_attr.cq = cq; wq_init_attr.max_sge = cmd.max_sge; wq_init_attr.max_wr = cmd.max_wr; - wq_init_attr.wq_context = attrs->ufile; wq_init_attr.wq_type = cmd.wq_type; wq_init_attr.event_handler = ib_uverbs_wq_event_handler; wq_init_attr.create_flags = cmd.create_flags; INIT_LIST_HEAD(&obj->uevent.event_list); + obj->uevent.uobject.user_handle = cmd.user_handle; wq = pd->device->ops.create_wq(pd, &wq_init_attr, &attrs->driver_udata); if (IS_ERR(wq)) { @@ -2972,31 +2889,25 @@ static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs) wq->cq = cq; wq->pd = pd; wq->device = pd->device; - wq->wq_context = wq_init_attr.wq_context; atomic_set(&wq->usecnt, 0); atomic_inc(&pd->usecnt); atomic_inc(&cq->usecnt); - wq->uobject = obj; - obj->uevent.uobject.object = wq; + obj->uevent.event_file = READ_ONCE(attrs->ufile->default_async_file); + if (obj->uevent.event_file) + uverbs_uobject_get(&obj->uevent.event_file->uobj); + + uobj_put_obj_read(pd); + rdma_lookup_put_uobject(&cq->uobject->uevent.uobject, + UVERBS_LOOKUP_READ); + uobj_finalize_uobj_create(&obj->uevent.uobject, attrs); - memset(&resp, 0, sizeof(resp)); resp.wq_handle = obj->uevent.uobject.id; resp.max_sge = wq_init_attr.max_sge; resp.max_wr = wq_init_attr.max_wr; resp.wqn = wq->wq_num; resp.response_length = uverbs_response_length(attrs, sizeof(resp)); - err = uverbs_response(attrs, &resp, sizeof(resp)); - if (err) - goto err_copy; - - uobj_put_obj_read(pd); - rdma_lookup_put_uobject(&cq->uobject->uevent.uobject, - UVERBS_LOOKUP_READ); - rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs); - return 0; + return uverbs_response(attrs, &resp, sizeof(resp)); -err_copy: - ib_destroy_wq(wq, uverbs_get_cleared_udata(attrs)); err_put_cq: rdma_lookup_put_uobject(&cq->uobject->uevent.uobject, UVERBS_LOOKUP_READ); @@ -3081,7 +2992,7 @@ static int ib_uverbs_ex_create_rwq_ind_table(struct uverbs_attr_bundle *attrs) struct ib_wq **wqs = NULL; u32 *wqs_handles = NULL; struct ib_wq *wq = NULL; - int i, j, num_read_wqs; + int i, num_read_wqs; u32 num_wq_handles; struct uverbs_req_iter iter; struct ib_device *ib_dev; @@ -3127,6 +3038,7 @@ static int ib_uverbs_ex_create_rwq_ind_table(struct uverbs_attr_bundle *attrs) } wqs[num_read_wqs] = wq; + atomic_inc(&wqs[num_read_wqs]->usecnt); } uobj = uobj_alloc(UVERBS_OBJECT_RWQ_IND_TBL, attrs, &ib_dev); @@ -3154,33 +3066,24 @@ static int ib_uverbs_ex_create_rwq_ind_table(struct uverbs_attr_bundle *attrs) atomic_set(&rwq_ind_tbl->usecnt, 0); for (i = 0; i < num_wq_handles; i++) - atomic_inc(&wqs[i]->usecnt); + rdma_lookup_put_uobject(&wqs[i]->uobject->uevent.uobject, + UVERBS_LOOKUP_READ); + kfree(wqs_handles); + uobj_finalize_uobj_create(uobj, attrs); resp.ind_tbl_handle = uobj->id; resp.ind_tbl_num = rwq_ind_tbl->ind_tbl_num; resp.response_length = uverbs_response_length(attrs, sizeof(resp)); + return uverbs_response(attrs, &resp, sizeof(resp)); - err = uverbs_response(attrs, &resp, sizeof(resp)); - if (err) - goto err_copy; - - kfree(wqs_handles); - - for (j = 0; j < num_read_wqs; j++) - rdma_lookup_put_uobject(&wqs[j]->uobject->uevent.uobject, - UVERBS_LOOKUP_READ); - - rdma_alloc_commit_uobject(uobj, attrs); - return 0; - -err_copy: - ib_destroy_rwq_ind_table(rwq_ind_tbl); err_uobj: uobj_alloc_abort(uobj, attrs); put_wqs: - for (j = 0; j < num_read_wqs; j++) - rdma_lookup_put_uobject(&wqs[j]->uobject->uevent.uobject, + for (i = 0; i < num_read_wqs; i++) { + rdma_lookup_put_uobject(&wqs[i]->uobject->uevent.uobject, UVERBS_LOOKUP_READ); + atomic_dec(&wqs[i]->usecnt); + } err_free: kfree(wqs_handles); kfree(wqs); @@ -3206,7 +3109,7 @@ static int ib_uverbs_ex_destroy_rwq_ind_table(struct uverbs_attr_bundle *attrs) static int ib_uverbs_ex_create_flow(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_flow cmd; - struct ib_uverbs_create_flow_resp resp; + struct ib_uverbs_create_flow_resp resp = {}; struct ib_uobject *uobj; struct ib_flow *flow_id; struct ib_uverbs_flow_attr *kern_flow_attr; @@ -3339,23 +3242,17 @@ static int ib_uverbs_ex_create_flow(struct uverbs_attr_bundle *attrs) ib_set_flow(uobj, flow_id, qp, qp->device, uflow_res); - memset(&resp, 0, sizeof(resp)); - resp.flow_handle = uobj->id; - - err = uverbs_response(attrs, &resp, sizeof(resp)); - if (err) - goto err_copy; - rdma_lookup_put_uobject(&qp->uobject->uevent.uobject, UVERBS_LOOKUP_READ); kfree(flow_attr); + if (cmd.flow_attr.num_of_specs) kfree(kern_flow_attr); - rdma_alloc_commit_uobject(uobj, attrs); - return 0; -err_copy: - if (!qp->device->ops.destroy_flow(flow_id)) - atomic_dec(&qp->usecnt); + uobj_finalize_uobj_create(uobj, attrs); + + resp.flow_handle = uobj->id; + return uverbs_response(attrs, &resp, sizeof(resp)); + err_free: ib_uverbs_flow_resources_free(uflow_res); err_free_flow_attr: @@ -3390,13 +3287,13 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs, struct ib_uverbs_create_xsrq *cmd, struct ib_udata *udata) { - struct ib_uverbs_create_srq_resp resp; + struct ib_uverbs_create_srq_resp resp = {}; struct ib_usrq_object *obj; struct ib_pd *pd; struct ib_srq *srq; - struct ib_uobject *uninitialized_var(xrcd_uobj); struct ib_srq_init_attr attr; int ret; + struct ib_uobject *xrcd_uobj; struct ib_device *ib_dev; obj = (struct ib_usrq_object *)uobj_alloc(UVERBS_OBJECT_SRQ, attrs, @@ -3441,58 +3338,29 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs, } attr.event_handler = ib_uverbs_srq_event_handler; - attr.srq_context = attrs->ufile; attr.srq_type = cmd->srq_type; attr.attr.max_wr = cmd->max_wr; attr.attr.max_sge = cmd->max_sge; attr.attr.srq_limit = cmd->srq_limit; INIT_LIST_HEAD(&obj->uevent.event_list); + obj->uevent.uobject.user_handle = cmd->user_handle; - srq = rdma_zalloc_drv_obj(ib_dev, ib_srq); - if (!srq) { - ret = -ENOMEM; - goto err_put; - } - - srq->device = pd->device; - srq->pd = pd; - srq->srq_type = cmd->srq_type; - srq->uobject = obj; - srq->event_handler = attr.event_handler; - srq->srq_context = attr.srq_context; - - ret = pd->device->ops.create_srq(srq, &attr, udata); - if (ret) - goto err_free; - - if (ib_srq_has_cq(cmd->srq_type)) { - srq->ext.cq = attr.ext.cq; - atomic_inc(&attr.ext.cq->usecnt); - } - - if (cmd->srq_type == IB_SRQT_XRC) { - srq->ext.xrc.xrcd = attr.ext.xrc.xrcd; - atomic_inc(&attr.ext.xrc.xrcd->usecnt); + srq = ib_create_srq_user(pd, &attr, obj, udata); + if (IS_ERR(srq)) { + ret = PTR_ERR(srq); + goto err_put_pd; } - atomic_inc(&pd->usecnt); - atomic_set(&srq->usecnt, 0); - obj->uevent.uobject.object = srq; obj->uevent.uobject.user_handle = cmd->user_handle; + obj->uevent.event_file = READ_ONCE(attrs->ufile->default_async_file); + if (obj->uevent.event_file) + uverbs_uobject_get(&obj->uevent.event_file->uobj); - memset(&resp, 0, sizeof resp); - resp.srq_handle = obj->uevent.uobject.id; - resp.max_wr = attr.attr.max_wr; - resp.max_sge = attr.attr.max_sge; if (cmd->srq_type == IB_SRQT_XRC) resp.srqn = srq->ext.xrc.srq_num; - ret = uverbs_response(attrs, &resp, sizeof(resp)); - if (ret) - goto err_copy; - if (cmd->srq_type == IB_SRQT_XRC) uobj_put_read(xrcd_uobj); @@ -3501,18 +3369,15 @@ static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs, UVERBS_LOOKUP_READ); uobj_put_obj_read(pd); - rdma_alloc_commit_uobject(&obj->uevent.uobject, attrs); - return 0; + uobj_finalize_uobj_create(&obj->uevent.uobject, attrs); -err_copy: - ib_destroy_srq_user(srq, uverbs_get_cleared_udata(attrs)); - /* It was released in ib_destroy_srq_user */ - srq = NULL; -err_free: - kfree(srq); -err_put: - uobj_put_obj_read(pd); + resp.srq_handle = obj->uevent.uobject.id; + resp.max_wr = attr.attr.max_wr; + resp.max_sge = attr.attr.max_sge; + return uverbs_response(attrs, &resp, sizeof(resp)); +err_put_pd: + uobj_put_obj_read(pd); err_put_cq: if (ib_srq_has_cq(cmd->srq_type)) rdma_lookup_put_uobject(&attr.ext.cq->uobject->uevent.uobject, @@ -3751,7 +3616,7 @@ static int ib_uverbs_ex_modify_cq(struct uverbs_attr_bundle *attrs) #define UAPI_DEF_WRITE_IO(req, resp) \ .write.has_resp = 1 + \ BUILD_BUG_ON_ZERO(offsetof(req, response) != 0) + \ - BUILD_BUG_ON_ZERO(sizeof(((req *)0)->response) != \ + BUILD_BUG_ON_ZERO(sizeof_field(req, response) != \ sizeof(u64)), \ .write.req_size = sizeof(req), .write.resp_size = sizeof(resp) diff --git a/drivers/infiniband/core/uverbs_ioctl.c b/drivers/infiniband/core/uverbs_ioctl.c index 538affbc517e..e47c5949013f 100644 --- a/drivers/infiniband/core/uverbs_ioctl.c +++ b/drivers/infiniband/core/uverbs_ioctl.c @@ -58,6 +58,7 @@ struct bundle_priv { DECLARE_BITMAP(uobj_finalize, UVERBS_API_ATTR_BKEY_LEN); DECLARE_BITMAP(spec_finalize, UVERBS_API_ATTR_BKEY_LEN); + DECLARE_BITMAP(uobj_hw_obj_valid, UVERBS_API_ATTR_BKEY_LEN); /* * Must be last. bundle ends in a flex array which overlaps @@ -136,7 +137,7 @@ EXPORT_SYMBOL(_uverbs_alloc); static bool uverbs_is_attr_cleared(const struct ib_uverbs_attr *uattr, u16 len) { - if (uattr->len > sizeof(((struct ib_uverbs_attr *)0)->data)) + if (uattr->len > sizeof_field(struct ib_uverbs_attr, data)) return ib_is_buffer_cleared(u64_to_user_ptr(uattr->data) + len, uattr->len - len); @@ -230,7 +231,8 @@ static void uverbs_free_idrs_array(const struct uverbs_api_attr *attr_uapi, for (i = 0; i != attr->len; i++) uverbs_finalize_object(attr->uobjects[i], - spec->u2.objs_arr.access, commit, attrs); + spec->u2.objs_arr.access, false, commit, + attrs); } static int uverbs_process_attr(struct bundle_priv *pbundle, @@ -257,7 +259,7 @@ static int uverbs_process_attr(struct bundle_priv *pbundle, return -EOPNOTSUPP; e->ptr_attr.enum_id = uattr->attr_data.enum_data.elem_id; - /* fall through */ + fallthrough; case UVERBS_ATTR_TYPE_PTR_IN: /* Ensure that any data provided by userspace beyond the known * struct is zero. Userspace that knows how to use some future @@ -269,7 +271,7 @@ static int uverbs_process_attr(struct bundle_priv *pbundle, !uverbs_is_attr_cleared(uattr, val_spec->u.ptr.len)) return -EOPNOTSUPP; - /* fall through */ + fallthrough; case UVERBS_ATTR_TYPE_PTR_OUT: if (uattr->len < val_spec->u.ptr.min_len || (!val_spec->zero_trailing && @@ -502,7 +504,9 @@ static void bundle_destroy(struct bundle_priv *pbundle, bool commit) uverbs_finalize_object( attr->obj_attr.uobject, - attr->obj_attr.attr_elm->spec.u.obj.access, commit, + attr->obj_attr.attr_elm->spec.u.obj.access, + test_bit(i, pbundle->uobj_hw_obj_valid), + commit, &pbundle->bundle); } @@ -590,6 +594,8 @@ static int ib_uverbs_cmd_verbs(struct ib_uverbs_file *ufile, sizeof(pbundle->bundle.attr_present)); memset(pbundle->uobj_finalize, 0, sizeof(pbundle->uobj_finalize)); memset(pbundle->spec_finalize, 0, sizeof(pbundle->spec_finalize)); + memset(pbundle->uobj_hw_obj_valid, 0, + sizeof(pbundle->uobj_hw_obj_valid)); ret = ib_uverbs_run_method(pbundle, hdr->num_attrs); bundle_destroy(pbundle, ret == 0); @@ -784,3 +790,16 @@ int uverbs_copy_to_struct_or_zero(const struct uverbs_attr_bundle *bundle, } return uverbs_copy_to(bundle, idx, from, size); } +EXPORT_SYMBOL(uverbs_copy_to_struct_or_zero); + +/* Once called an abort will call through to the type's destroy_hw() */ +void uverbs_finalize_uobj_create(const struct uverbs_attr_bundle *bundle, + u16 idx) +{ + struct bundle_priv *pbundle = + container_of(bundle, struct bundle_priv, bundle); + + __set_bit(uapi_bkey_attr(uapi_key_attr(idx)), + pbundle->uobj_hw_obj_valid); +} +EXPORT_SYMBOL(uverbs_finalize_uobj_create); diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 1bab8de14757..37794d88b1f3 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -75,7 +75,7 @@ static dev_t dynamic_uverbs_dev; static struct class *uverbs_class; static DEFINE_IDA(uverbs_ida); -static void ib_uverbs_add_one(struct ib_device *device); +static int ib_uverbs_add_one(struct ib_device *device); static void ib_uverbs_remove_one(struct ib_device *device, void *client_data); /* @@ -146,8 +146,7 @@ void ib_uverbs_release_ucq(struct ib_uverbs_completion_event_file *ev_file, void ib_uverbs_release_uevent(struct ib_uevent_object *uobj) { - struct ib_uverbs_async_event_file *async_file = - READ_ONCE(uobj->uobject.ufile->async_file); + struct ib_uverbs_async_event_file *async_file = uobj->event_file; struct ib_uverbs_event *evt, *tmp; if (!async_file) @@ -159,6 +158,7 @@ void ib_uverbs_release_uevent(struct ib_uevent_object *uobj) kfree(evt); } spin_unlock_irq(&async_file->ev_queue.lock); + uverbs_uobject_put(&async_file->uobj); } void ib_uverbs_detach_umcast(struct ib_qp *qp, @@ -197,8 +197,8 @@ void ib_uverbs_release_file(struct kref *ref) if (atomic_dec_and_test(&file->device->refcount)) ib_uverbs_comp_dev(file->device); - if (file->async_file) - uverbs_uobject_put(&file->async_file->uobj); + if (file->default_async_file) + uverbs_uobject_put(&file->default_async_file->uobj); put_device(&file->device->dev); if (file->disassociate_page) @@ -296,6 +296,8 @@ static __poll_t ib_uverbs_event_poll(struct ib_uverbs_event_queue *ev_queue, spin_lock_irq(&ev_queue->lock); if (!list_empty(&ev_queue->event_list)) pollflags = EPOLLIN | EPOLLRDNORM; + else if (ev_queue->is_closed) + pollflags = EPOLLERR; spin_unlock_irq(&ev_queue->lock); return pollflags; @@ -425,7 +427,7 @@ void ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file, static void uverbs_uobj_event(struct ib_uevent_object *eobj, struct ib_event *event) { - ib_uverbs_async_handler(READ_ONCE(eobj->uobject.ufile->async_file), + ib_uverbs_async_handler(eobj->event_file, eobj->uobject.user_handle, event->event, &eobj->event_list, &eobj->events_reported); } @@ -482,10 +484,10 @@ void ib_uverbs_init_async_event_file( /* The first async_event_file becomes the default one for the file. */ mutex_lock(&uverbs_file->ucontext_lock); - if (!uverbs_file->async_file) { + if (!uverbs_file->default_async_file) { /* Pairs with the put in ib_uverbs_release_file */ uverbs_uobject_get(&async_file->uobj); - smp_store_release(&uverbs_file->async_file, async_file); + smp_store_release(&uverbs_file->default_async_file, async_file); } mutex_unlock(&uverbs_file->ucontext_lock); @@ -599,6 +601,7 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, memset(bundle.attr_present, 0, sizeof(bundle.attr_present)); bundle.ufile = file; bundle.context = NULL; /* only valid if bundle has uobject */ + bundle.uobject = NULL; if (!method_elm->is_ex) { size_t in_len = hdr.in_words * 4 - sizeof(hdr); size_t out_len = hdr.out_words * 4; @@ -662,6 +665,9 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, } ret = method_elm->handler(&bundle); + if (bundle.uobject) + uverbs_finalize_object(bundle.uobject, UVERBS_ACCESS_NEW, true, + !ret, &bundle); out_unlock: srcu_read_unlock(&file->device->disassociate_srcu, srcu_key); return (ret) ? : count; @@ -833,12 +839,12 @@ void uverbs_user_mmap_disassociate(struct ib_uverbs_file *ufile) return; /* - * The umap_lock is nested under mmap_sem since it used within + * The umap_lock is nested under mmap_lock since it used within * the vma_ops callbacks, so we have to clean the list one mm * at a time to get the lock ordering right. Typically there * will only be one mm, so no big deal. */ - down_read(&mm->mmap_sem); + mmap_read_lock(mm); if (!mmget_still_valid(mm)) goto skip_mm; mutex_lock(&ufile->umap_lock); @@ -860,7 +866,7 @@ void uverbs_user_mmap_disassociate(struct ib_uverbs_file *ufile) } mutex_unlock(&ufile->umap_lock); skip_mm: - up_read(&mm->mmap_sem); + mmap_read_unlock(mm); mmput(mm); } } @@ -1092,7 +1098,7 @@ static int ib_uverbs_create_uapi(struct ib_device *device, return 0; } -static void ib_uverbs_add_one(struct ib_device *device) +static int ib_uverbs_add_one(struct ib_device *device) { int devnum; dev_t base; @@ -1100,16 +1106,16 @@ static void ib_uverbs_add_one(struct ib_device *device) int ret; if (!device->ops.alloc_ucontext) - return; + return -EOPNOTSUPP; uverbs_dev = kzalloc(sizeof(*uverbs_dev), GFP_KERNEL); if (!uverbs_dev) - return; + return -ENOMEM; ret = init_srcu_struct(&uverbs_dev->disassociate_srcu); if (ret) { kfree(uverbs_dev); - return; + return -ENOMEM; } device_initialize(&uverbs_dev->dev); @@ -1129,15 +1135,18 @@ static void ib_uverbs_add_one(struct ib_device *device) devnum = ida_alloc_max(&uverbs_ida, IB_UVERBS_MAX_DEVICES - 1, GFP_KERNEL); - if (devnum < 0) + if (devnum < 0) { + ret = -ENOMEM; goto err; + } uverbs_dev->devnum = devnum; if (devnum >= IB_UVERBS_NUM_FIXED_MINOR) base = dynamic_uverbs_dev + devnum - IB_UVERBS_NUM_FIXED_MINOR; else base = IB_UVERBS_BASE_DEV + devnum; - if (ib_uverbs_create_uapi(device, uverbs_dev)) + ret = ib_uverbs_create_uapi(device, uverbs_dev); + if (ret) goto err_uapi; uverbs_dev->dev.devt = base; @@ -1152,7 +1161,7 @@ static void ib_uverbs_add_one(struct ib_device *device) goto err_uapi; ib_set_client_data(device, &uverbs_client, uverbs_dev); - return; + return 0; err_uapi: ida_free(&uverbs_ida, devnum); @@ -1161,7 +1170,7 @@ err: ib_uverbs_comp_dev(uverbs_dev); wait_for_completion(&uverbs_dev->comp); put_device(&uverbs_dev->dev); - return; + return ret; } static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev, @@ -1201,9 +1210,6 @@ static void ib_uverbs_remove_one(struct ib_device *device, void *client_data) struct ib_uverbs_device *uverbs_dev = client_data; int wait_clients = 1; - if (!uverbs_dev) - return; - cdev_device_del(&uverbs_dev->cdev, &uverbs_dev->dev); ida_free(&uverbs_ida, uverbs_dev->devnum); diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c index 3abfc63225cb..08c39cfb1bd9 100644 --- a/drivers/infiniband/core/uverbs_std_types.c +++ b/drivers/infiniband/core/uverbs_std_types.c @@ -75,40 +75,6 @@ static int uverbs_free_mw(struct ib_uobject *uobject, return uverbs_dealloc_mw((struct ib_mw *)uobject->object); } -static int uverbs_free_qp(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct ib_qp *qp = uobject->object; - struct ib_uqp_object *uqp = - container_of(uobject, struct ib_uqp_object, uevent.uobject); - int ret; - - /* - * If this is a user triggered destroy then do not allow destruction - * until the user cleans up all the mcast bindings. Unlike in other - * places we forcibly clean up the mcast attachments for !DESTROY - * because the mcast attaches are not ubojects and will not be - * destroyed by anything else during cleanup processing. - */ - if (why == RDMA_REMOVE_DESTROY) { - if (!list_empty(&uqp->mcast_list)) - return -EBUSY; - } else if (qp == qp->real_qp) { - ib_uverbs_detach_umcast(qp, uqp); - } - - ret = ib_destroy_qp_user(qp, &attrs->driver_udata); - if (ib_is_destroy_retryable(ret, why, uobject)) - return ret; - - if (uqp->uxrcd) - atomic_dec(&uqp->uxrcd->refcnt); - - ib_uverbs_release_uevent(&uqp->uevent); - return ret; -} - static int uverbs_free_rwq_ind_tbl(struct ib_uobject *uobject, enum rdma_remove_reason why, struct uverbs_attr_bundle *attrs) @@ -125,48 +91,6 @@ static int uverbs_free_rwq_ind_tbl(struct ib_uobject *uobject, return ret; } -static int uverbs_free_wq(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct ib_wq *wq = uobject->object; - struct ib_uwq_object *uwq = - container_of(uobject, struct ib_uwq_object, uevent.uobject); - int ret; - - ret = ib_destroy_wq(wq, &attrs->driver_udata); - if (ib_is_destroy_retryable(ret, why, uobject)) - return ret; - - ib_uverbs_release_uevent(&uwq->uevent); - return ret; -} - -static int uverbs_free_srq(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct ib_srq *srq = uobject->object; - struct ib_uevent_object *uevent = - container_of(uobject, struct ib_uevent_object, uobject); - enum ib_srq_type srq_type = srq->srq_type; - int ret; - - ret = ib_destroy_srq_user(srq, &attrs->driver_udata); - if (ib_is_destroy_retryable(ret, why, uobject)) - return ret; - - if (srq_type == IB_SRQT_XRC) { - struct ib_usrq_object *us = - container_of(uevent, struct ib_usrq_object, uevent); - - atomic_dec(&us->uxrcd->refcnt); - } - - ib_uverbs_release_uevent(uevent); - return ret; -} - static int uverbs_free_xrcd(struct ib_uobject *uobject, enum rdma_remove_reason why, struct uverbs_attr_bundle *attrs) @@ -252,10 +176,6 @@ DECLARE_UVERBS_NAMED_OBJECT( "[infinibandevent]", O_RDONLY)); -DECLARE_UVERBS_NAMED_OBJECT( - UVERBS_OBJECT_QP, - UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uqp_object), uverbs_free_qp)); - DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_MW_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_MW_HANDLE, @@ -267,11 +187,6 @@ DECLARE_UVERBS_NAMED_OBJECT(UVERBS_OBJECT_MW, UVERBS_TYPE_ALLOC_IDR(uverbs_free_mw), &UVERBS_METHOD(UVERBS_METHOD_MW_DESTROY)); -DECLARE_UVERBS_NAMED_OBJECT( - UVERBS_OBJECT_SRQ, - UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_usrq_object), - uverbs_free_srq)); - DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_AH_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_AH_HANDLE, @@ -296,10 +211,6 @@ DECLARE_UVERBS_NAMED_OBJECT( uverbs_free_flow), &UVERBS_METHOD(UVERBS_METHOD_FLOW_DESTROY)); -DECLARE_UVERBS_NAMED_OBJECT( - UVERBS_OBJECT_WQ, - UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uwq_object), uverbs_free_wq)); - DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_RWQ_IND_TBL_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_RWQ_IND_TBL_HANDLE, @@ -340,18 +251,12 @@ const struct uapi_definition uverbs_def_obj_intf[] = { UAPI_DEF_OBJ_NEEDS_FN(dealloc_pd)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_COMP_CHANNEL, UAPI_DEF_OBJ_NEEDS_FN(dealloc_pd)), - UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_QP, - UAPI_DEF_OBJ_NEEDS_FN(destroy_qp)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_AH, UAPI_DEF_OBJ_NEEDS_FN(destroy_ah)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MW, UAPI_DEF_OBJ_NEEDS_FN(dealloc_mw)), - UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_SRQ, - UAPI_DEF_OBJ_NEEDS_FN(destroy_srq)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_FLOW, UAPI_DEF_OBJ_NEEDS_FN(destroy_flow)), - UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_WQ, - UAPI_DEF_OBJ_NEEDS_FN(destroy_wq)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED( UVERBS_OBJECT_RWQ_IND_TBL, UAPI_DEF_OBJ_NEEDS_FN(destroy_rwq_ind_table)), diff --git a/drivers/infiniband/core/uverbs_std_types_counters.c b/drivers/infiniband/core/uverbs_std_types_counters.c index 9f013304e677..c7e7438752bc 100644 --- a/drivers/infiniband/core/uverbs_std_types_counters.c +++ b/drivers/infiniband/core/uverbs_std_types_counters.c @@ -46,7 +46,9 @@ static int uverbs_free_counters(struct ib_uobject *uobject, if (ret) return ret; - return counters->device->ops.destroy_counters(counters); + counters->device->ops.destroy_counters(counters); + kfree(counters); + return 0; } static int UVERBS_HANDLER(UVERBS_METHOD_COUNTERS_CREATE)( @@ -66,20 +68,19 @@ static int UVERBS_HANDLER(UVERBS_METHOD_COUNTERS_CREATE)( if (!ib_dev->ops.create_counters) return -EOPNOTSUPP; - counters = ib_dev->ops.create_counters(ib_dev, attrs); - if (IS_ERR(counters)) { - ret = PTR_ERR(counters); - goto err_create_counters; - } + counters = rdma_zalloc_drv_obj(ib_dev, ib_counters); + if (!counters) + return -ENOMEM; counters->device = ib_dev; counters->uobject = uobj; uobj->object = counters; atomic_set(&counters->usecnt, 0); - return 0; + ret = ib_dev->ops.create_counters(counters, attrs); + if (ret) + kfree(counters); -err_create_counters: return ret; } diff --git a/drivers/infiniband/core/uverbs_std_types_cq.c b/drivers/infiniband/core/uverbs_std_types_cq.c index da4110a0eea2..b1c7dacc02de 100644 --- a/drivers/infiniband/core/uverbs_std_types_cq.c +++ b/drivers/infiniband/core/uverbs_std_types_cq.c @@ -100,6 +100,9 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)( uverbs_uobject_get(ev_file_uobj); } + obj->uevent.event_file = ib_uverbs_get_async_event( + attrs, UVERBS_ATTR_CREATE_CQ_EVENT_FD); + if (attr.comp_vector >= attrs->ufile->device->num_comp_vectors) { ret = -EINVAL; goto err_event_file; @@ -129,19 +132,17 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)( obj->uevent.uobject.object = cq; obj->uevent.uobject.user_handle = user_handle; rdma_restrack_uadd(&cq->res); + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_CQ_HANDLE); ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_CQ_RESP_CQE, &cq->cqe, sizeof(cq->cqe)); - if (ret) - goto err_cq; + return ret; - return 0; -err_cq: - ib_destroy_cq_user(cq, uverbs_get_cleared_udata(attrs)); - cq = NULL; err_free: kfree(cq); err_event_file: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); if (ev_file) uverbs_uobject_put(ev_file_uobj); return ret; @@ -171,6 +172,10 @@ DECLARE_UVERBS_NAMED_METHOD( UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_CQ_RESP_CQE, UVERBS_ATTR_TYPE(u32), UA_MANDATORY), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_CQ_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), UVERBS_ATTR_UHW()); static int UVERBS_HANDLER(UVERBS_METHOD_CQ_DESTROY)( @@ -202,11 +207,8 @@ DECLARE_UVERBS_NAMED_METHOD( DECLARE_UVERBS_NAMED_OBJECT( UVERBS_OBJECT_CQ, UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_ucq_object), uverbs_free_cq), - -#if IS_ENABLED(CONFIG_INFINIBAND_EXP_LEGACY_VERBS_NEW_UAPI) &UVERBS_METHOD(UVERBS_METHOD_CQ_CREATE), &UVERBS_METHOD(UVERBS_METHOD_CQ_DESTROY) -#endif ); const struct uapi_definition uverbs_def_obj_cq[] = { diff --git a/drivers/infiniband/core/uverbs_std_types_device.c b/drivers/infiniband/core/uverbs_std_types_device.c index ae4a59d6f9b1..75df2094a010 100644 --- a/drivers/infiniband/core/uverbs_std_types_device.c +++ b/drivers/infiniband/core/uverbs_std_types_device.c @@ -38,7 +38,12 @@ static int UVERBS_HANDLER(UVERBS_METHOD_INVOKE_WRITE)( attrs->ucore.outlen < method_elm->resp_size) return -ENOSPC; - return method_elm->handler(attrs); + attrs->uobject = NULL; + rc = method_elm->handler(attrs); + if (attrs->uobject) + uverbs_finalize_object(attrs->uobject, UVERBS_ACCESS_NEW, true, + !rc, attrs); + return rc; } DECLARE_UVERBS_NAMED_METHOD(UVERBS_METHOD_INVOKE_WRITE, @@ -229,6 +234,37 @@ static int UVERBS_HANDLER(UVERBS_METHOD_GET_CONTEXT)( return 0; } +static int UVERBS_HANDLER(UVERBS_METHOD_QUERY_CONTEXT)( + struct uverbs_attr_bundle *attrs) +{ + u64 core_support = IB_UVERBS_CORE_SUPPORT_OPTIONAL_MR_ACCESS; + struct ib_ucontext *ucontext; + struct ib_device *ib_dev; + u32 num_comp; + int ret; + + ucontext = ib_uverbs_get_ucontext(attrs); + if (IS_ERR(ucontext)) + return PTR_ERR(ucontext); + ib_dev = ucontext->device; + + if (!ib_dev->ops.query_ucontext) + return -EOPNOTSUPP; + + num_comp = attrs->ufile->device->num_comp_vectors; + ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_CONTEXT_NUM_COMP_VECTORS, + &num_comp, sizeof(num_comp)); + if (IS_UVERBS_COPY_ERR(ret)) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_CONTEXT_CORE_SUPPORT, + &core_support, sizeof(core_support)); + if (IS_UVERBS_COPY_ERR(ret)) + return ret; + + return ucontext->device->ops.query_ucontext(ucontext, attrs); +} + DECLARE_UVERBS_NAMED_METHOD( UVERBS_METHOD_GET_CONTEXT, UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_GET_CONTEXT_NUM_COMP_VECTORS, @@ -238,6 +274,13 @@ DECLARE_UVERBS_NAMED_METHOD( UVERBS_ATTR_UHW()); DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_QUERY_CONTEXT, + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_CONTEXT_NUM_COMP_VECTORS, + UVERBS_ATTR_TYPE(u32), UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_CONTEXT_CORE_SUPPORT, + UVERBS_ATTR_TYPE(u64), UA_OPTIONAL)); + +DECLARE_UVERBS_NAMED_METHOD( UVERBS_METHOD_INFO_HANDLES, /* Also includes any device specific object ids */ UVERBS_ATTR_CONST_IN(UVERBS_ATTR_INFO_OBJECT_ID, @@ -260,7 +303,8 @@ DECLARE_UVERBS_GLOBAL_METHODS(UVERBS_OBJECT_DEVICE, &UVERBS_METHOD(UVERBS_METHOD_GET_CONTEXT), &UVERBS_METHOD(UVERBS_METHOD_INVOKE_WRITE), &UVERBS_METHOD(UVERBS_METHOD_INFO_HANDLES), - &UVERBS_METHOD(UVERBS_METHOD_QUERY_PORT)); + &UVERBS_METHOD(UVERBS_METHOD_QUERY_PORT), + &UVERBS_METHOD(UVERBS_METHOD_QUERY_CONTEXT)); const struct uapi_definition uverbs_def_obj_device[] = { UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_DEVICE), diff --git a/drivers/infiniband/core/uverbs_std_types_mr.c b/drivers/infiniband/core/uverbs_std_types_mr.c index c1286a52dc84..9b22bb553e8b 100644 --- a/drivers/infiniband/core/uverbs_std_types_mr.c +++ b/drivers/infiniband/core/uverbs_std_types_mr.c @@ -69,7 +69,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_ADVISE_MR)( num_sge = uverbs_attr_ptr_get_array_size( attrs, UVERBS_ATTR_ADVISE_MR_SGE_LIST, sizeof(struct ib_sge)); - if (num_sge < 0) + if (num_sge <= 0) return num_sge; sg_list = uverbs_attr_get_alloced_ptr(attrs, @@ -136,22 +136,46 @@ static int UVERBS_HANDLER(UVERBS_METHOD_DM_MR_REG)( uobj->object = mr; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_REG_DM_MR_HANDLE); + ret = uverbs_copy_to(attrs, UVERBS_ATTR_REG_DM_MR_RESP_LKEY, &mr->lkey, sizeof(mr->lkey)); if (ret) - goto err_dereg; + return ret; ret = uverbs_copy_to(attrs, UVERBS_ATTR_REG_DM_MR_RESP_RKEY, &mr->rkey, sizeof(mr->rkey)); + return ret; +} + +static int UVERBS_HANDLER(UVERBS_METHOD_QUERY_MR)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_mr *mr = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_QUERY_MR_HANDLE); + int ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_MR_RESP_LKEY, &mr->lkey, + sizeof(mr->lkey)); if (ret) - goto err_dereg; + return ret; - return 0; + ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_MR_RESP_RKEY, + &mr->rkey, sizeof(mr->rkey)); -err_dereg: - ib_dereg_mr_user(mr, uverbs_get_cleared_udata(attrs)); + if (ret) + return ret; - return ret; + ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_MR_RESP_LENGTH, + &mr->length, sizeof(mr->length)); + + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_QUERY_MR_RESP_IOVA, + &mr->iova, sizeof(mr->iova)); + + return IS_UVERBS_COPY_ERR(ret) ? ret : 0; } DECLARE_UVERBS_NAMED_METHOD( @@ -172,6 +196,25 @@ DECLARE_UVERBS_NAMED_METHOD( UA_ALLOC_AND_COPY)); DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_QUERY_MR, + UVERBS_ATTR_IDR(UVERBS_ATTR_QUERY_MR_HANDLE, + UVERBS_OBJECT_MR, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_MR_RESP_RKEY, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_MR_RESP_LKEY, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_MR_RESP_LENGTH, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_QUERY_MR_RESP_IOVA, + UVERBS_ATTR_TYPE(u64), + UA_OPTIONAL)); + +DECLARE_UVERBS_NAMED_METHOD( UVERBS_METHOD_DM_MR_REG, UVERBS_ATTR_IDR(UVERBS_ATTR_REG_DM_MR_HANDLE, UVERBS_OBJECT_MR, @@ -212,7 +255,8 @@ DECLARE_UVERBS_NAMED_OBJECT( UVERBS_TYPE_ALLOC_IDR(uverbs_free_mr), &UVERBS_METHOD(UVERBS_METHOD_DM_MR_REG), &UVERBS_METHOD(UVERBS_METHOD_MR_DESTROY), - &UVERBS_METHOD(UVERBS_METHOD_ADVISE_MR)); + &UVERBS_METHOD(UVERBS_METHOD_ADVISE_MR), + &UVERBS_METHOD(UVERBS_METHOD_QUERY_MR)); const struct uapi_definition uverbs_def_obj_mr[] = { UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MR, diff --git a/drivers/infiniband/core/uverbs_std_types_qp.c b/drivers/infiniband/core/uverbs_std_types_qp.c new file mode 100644 index 000000000000..3bf8dcdfe7eb --- /dev/null +++ b/drivers/infiniband/core/uverbs_std_types_qp.c @@ -0,0 +1,401 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include <rdma/uverbs_std_types.h> +#include "rdma_core.h" +#include "uverbs.h" +#include "core_priv.h" + +static int uverbs_free_qp(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct ib_qp *qp = uobject->object; + struct ib_uqp_object *uqp = + container_of(uobject, struct ib_uqp_object, uevent.uobject); + int ret; + + /* + * If this is a user triggered destroy then do not allow destruction + * until the user cleans up all the mcast bindings. Unlike in other + * places we forcibly clean up the mcast attachments for !DESTROY + * because the mcast attaches are not ubojects and will not be + * destroyed by anything else during cleanup processing. + */ + if (why == RDMA_REMOVE_DESTROY) { + if (!list_empty(&uqp->mcast_list)) + return -EBUSY; + } else if (qp == qp->real_qp) { + ib_uverbs_detach_umcast(qp, uqp); + } + + ret = ib_destroy_qp_user(qp, &attrs->driver_udata); + if (ib_is_destroy_retryable(ret, why, uobject)) + return ret; + + if (uqp->uxrcd) + atomic_dec(&uqp->uxrcd->refcnt); + + ib_uverbs_release_uevent(&uqp->uevent); + return ret; +} + +static int check_creation_flags(enum ib_qp_type qp_type, + u32 create_flags) +{ + create_flags &= ~IB_UVERBS_QP_CREATE_SQ_SIG_ALL; + + if (!create_flags || qp_type == IB_QPT_DRIVER) + return 0; + + if (qp_type != IB_QPT_RAW_PACKET && qp_type != IB_QPT_UD) + return -EINVAL; + + if ((create_flags & IB_UVERBS_QP_CREATE_SCATTER_FCS || + create_flags & IB_UVERBS_QP_CREATE_CVLAN_STRIPPING) && + qp_type != IB_QPT_RAW_PACKET) + return -EINVAL; + + return 0; +} + +static void set_caps(struct ib_qp_init_attr *attr, + struct ib_uverbs_qp_cap *cap, bool req) +{ + if (req) { + attr->cap.max_send_wr = cap->max_send_wr; + attr->cap.max_recv_wr = cap->max_recv_wr; + attr->cap.max_send_sge = cap->max_send_sge; + attr->cap.max_recv_sge = cap->max_recv_sge; + attr->cap.max_inline_data = cap->max_inline_data; + } else { + cap->max_send_wr = attr->cap.max_send_wr; + cap->max_recv_wr = attr->cap.max_recv_wr; + cap->max_send_sge = attr->cap.max_send_sge; + cap->max_recv_sge = attr->cap.max_recv_sge; + cap->max_inline_data = attr->cap.max_inline_data; + } +} + +static int UVERBS_HANDLER(UVERBS_METHOD_QP_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uqp_object *obj = container_of( + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_QP_HANDLE), + typeof(*obj), uevent.uobject); + struct ib_qp_init_attr attr = {}; + struct ib_uverbs_qp_cap cap = {}; + struct ib_rwq_ind_table *rwq_ind_tbl = NULL; + struct ib_qp *qp; + struct ib_pd *pd = NULL; + struct ib_srq *srq = NULL; + struct ib_cq *recv_cq = NULL; + struct ib_cq *send_cq = NULL; + struct ib_xrcd *xrcd = NULL; + struct ib_uobject *xrcd_uobj = NULL; + struct ib_device *device; + u64 user_handle; + int ret; + + ret = uverbs_copy_from_or_zero(&cap, attrs, + UVERBS_ATTR_CREATE_QP_CAP); + if (!ret) + ret = uverbs_copy_from(&user_handle, attrs, + UVERBS_ATTR_CREATE_QP_USER_HANDLE); + if (!ret) + ret = uverbs_get_const(&attr.qp_type, attrs, + UVERBS_ATTR_CREATE_QP_TYPE); + if (ret) + return ret; + + switch (attr.qp_type) { + case IB_QPT_XRC_TGT: + if (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_PD_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE)) + return -EINVAL; + + xrcd_uobj = uverbs_attr_get_uobject(attrs, + UVERBS_ATTR_CREATE_QP_XRCD_HANDLE); + if (IS_ERR(xrcd_uobj)) + return PTR_ERR(xrcd_uobj); + + xrcd = (struct ib_xrcd *)xrcd_uobj->object; + if (!xrcd) + return -EINVAL; + device = xrcd->device; + break; + case IB_UVERBS_QPT_RAW_PACKET: + if (!capable(CAP_NET_RAW)) + return -EPERM; + fallthrough; + case IB_UVERBS_QPT_RC: + case IB_UVERBS_QPT_UC: + case IB_UVERBS_QPT_UD: + case IB_UVERBS_QPT_XRC_INI: + case IB_UVERBS_QPT_DRIVER: + if (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_XRCD_HANDLE) || + (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE) && + attr.qp_type == IB_QPT_XRC_INI)) + return -EINVAL; + + pd = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_PD_HANDLE); + if (IS_ERR(pd)) + return PTR_ERR(pd); + + rwq_ind_tbl = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE); + if (!IS_ERR(rwq_ind_tbl)) { + if (cap.max_recv_wr || cap.max_recv_sge || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE) || + uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE)) + return -EINVAL; + + /* send_cq is optinal */ + if (cap.max_send_wr) { + send_cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE); + if (IS_ERR(send_cq)) + return PTR_ERR(send_cq); + } + attr.rwq_ind_tbl = rwq_ind_tbl; + } else { + send_cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE); + if (IS_ERR(send_cq)) + return PTR_ERR(send_cq); + + if (attr.qp_type != IB_QPT_XRC_INI) { + recv_cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE); + if (IS_ERR(recv_cq)) + return PTR_ERR(recv_cq); + } + } + + device = pd->device; + break; + default: + return -EINVAL; + } + + ret = uverbs_get_flags32(&attr.create_flags, attrs, + UVERBS_ATTR_CREATE_QP_FLAGS, + IB_UVERBS_QP_CREATE_BLOCK_MULTICAST_LOOPBACK | + IB_UVERBS_QP_CREATE_SCATTER_FCS | + IB_UVERBS_QP_CREATE_CVLAN_STRIPPING | + IB_UVERBS_QP_CREATE_PCI_WRITE_END_PADDING | + IB_UVERBS_QP_CREATE_SQ_SIG_ALL); + if (ret) + return ret; + + ret = check_creation_flags(attr.qp_type, attr.create_flags); + if (ret) + return ret; + + if (uverbs_attr_is_valid(attrs, + UVERBS_ATTR_CREATE_QP_SOURCE_QPN)) { + ret = uverbs_copy_from(&attr.source_qpn, attrs, + UVERBS_ATTR_CREATE_QP_SOURCE_QPN); + if (ret) + return ret; + attr.create_flags |= IB_QP_CREATE_SOURCE_QPN; + } + + srq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_QP_SRQ_HANDLE); + if (!IS_ERR(srq)) { + if ((srq->srq_type == IB_SRQT_XRC && + attr.qp_type != IB_QPT_XRC_TGT) || + (srq->srq_type != IB_SRQT_XRC && + attr.qp_type == IB_QPT_XRC_TGT)) + return -EINVAL; + attr.srq = srq; + } + + obj->uevent.event_file = ib_uverbs_get_async_event(attrs, + UVERBS_ATTR_CREATE_QP_EVENT_FD); + INIT_LIST_HEAD(&obj->uevent.event_list); + INIT_LIST_HEAD(&obj->mcast_list); + obj->uevent.uobject.user_handle = user_handle; + attr.event_handler = ib_uverbs_qp_event_handler; + attr.send_cq = send_cq; + attr.recv_cq = recv_cq; + attr.xrcd = xrcd; + if (attr.create_flags & IB_UVERBS_QP_CREATE_SQ_SIG_ALL) { + /* This creation bit is uverbs one, need to mask before + * calling drivers. It was added to prevent an extra user attr + * only for that when using ioctl. + */ + attr.create_flags &= ~IB_UVERBS_QP_CREATE_SQ_SIG_ALL; + attr.sq_sig_type = IB_SIGNAL_ALL_WR; + } else { + attr.sq_sig_type = IB_SIGNAL_REQ_WR; + } + + set_caps(&attr, &cap, true); + mutex_init(&obj->mcast_lock); + + if (attr.qp_type == IB_QPT_XRC_TGT) + qp = ib_create_qp(pd, &attr); + else + qp = _ib_create_qp(device, pd, &attr, &attrs->driver_udata, + obj); + + if (IS_ERR(qp)) { + ret = PTR_ERR(qp); + goto err_put; + } + + if (attr.qp_type != IB_QPT_XRC_TGT) { + atomic_inc(&pd->usecnt); + if (attr.send_cq) + atomic_inc(&attr.send_cq->usecnt); + if (attr.recv_cq) + atomic_inc(&attr.recv_cq->usecnt); + if (attr.srq) + atomic_inc(&attr.srq->usecnt); + if (attr.rwq_ind_tbl) + atomic_inc(&attr.rwq_ind_tbl->usecnt); + } else { + obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, + uobject); + atomic_inc(&obj->uxrcd->refcnt); + /* It is done in _ib_create_qp for other QP types */ + qp->uobject = obj; + } + + obj->uevent.uobject.object = qp; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_QP_HANDLE); + + if (attr.qp_type != IB_QPT_XRC_TGT) { + ret = ib_create_qp_security(qp, device); + if (ret) + return ret; + } + + set_caps(&attr, &cap, false); + ret = uverbs_copy_to_struct_or_zero(attrs, + UVERBS_ATTR_CREATE_QP_RESP_CAP, &cap, + sizeof(cap)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_QP_RESP_QP_NUM, + &qp->qp_num, + sizeof(qp->qp_num)); + + return ret; +err_put: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); + return ret; +}; + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_QP_CREATE, + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_HANDLE, + UVERBS_OBJECT_QP, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_XRCD_HANDLE, + UVERBS_OBJECT_XRCD, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_SRQ_HANDLE, + UVERBS_OBJECT_SRQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_SEND_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_RECV_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_QP_IND_TABLE_HANDLE, + UVERBS_OBJECT_RWQ_IND_TBL, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_QP_USER_HANDLE, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_QP_CAP, + UVERBS_ATTR_STRUCT(struct ib_uverbs_qp_cap, + max_inline_data), + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(UVERBS_ATTR_CREATE_QP_TYPE, + enum ib_uverbs_qp_type, + UA_MANDATORY), + UVERBS_ATTR_FLAGS_IN(UVERBS_ATTR_CREATE_QP_FLAGS, + enum ib_uverbs_qp_create_flags, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_QP_SOURCE_QPN, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_QP_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_QP_RESP_CAP, + UVERBS_ATTR_STRUCT(struct ib_uverbs_qp_cap, + max_inline_data), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_QP_RESP_QP_NUM, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_UHW()); + +static int UVERBS_HANDLER(UVERBS_METHOD_QP_DESTROY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_QP_HANDLE); + struct ib_uqp_object *obj = + container_of(uobj, struct ib_uqp_object, uevent.uobject); + struct ib_uverbs_destroy_qp_resp resp = { + .events_reported = obj->uevent.events_reported + }; + + return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_QP_RESP, &resp, + sizeof(resp)); +} + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_QP_DESTROY, + UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_QP_HANDLE, + UVERBS_OBJECT_QP, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_DESTROY_QP_RESP, + UVERBS_ATTR_TYPE(struct ib_uverbs_destroy_qp_resp), + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_OBJECT( + UVERBS_OBJECT_QP, + UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uqp_object), uverbs_free_qp), + &UVERBS_METHOD(UVERBS_METHOD_QP_CREATE), + &UVERBS_METHOD(UVERBS_METHOD_QP_DESTROY)); + +const struct uapi_definition uverbs_def_obj_qp[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_QP, + UAPI_DEF_OBJ_NEEDS_FN(destroy_qp)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_srq.c b/drivers/infiniband/core/uverbs_std_types_srq.c new file mode 100644 index 000000000000..c0ecbba26bf4 --- /dev/null +++ b/drivers/infiniband/core/uverbs_std_types_srq.c @@ -0,0 +1,234 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include <rdma/uverbs_std_types.h> +#include "rdma_core.h" +#include "uverbs.h" + +static int uverbs_free_srq(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct ib_srq *srq = uobject->object; + struct ib_uevent_object *uevent = + container_of(uobject, struct ib_uevent_object, uobject); + enum ib_srq_type srq_type = srq->srq_type; + int ret; + + ret = ib_destroy_srq_user(srq, &attrs->driver_udata); + if (ib_is_destroy_retryable(ret, why, uobject)) + return ret; + + if (srq_type == IB_SRQT_XRC) { + struct ib_usrq_object *us = + container_of(uobject, struct ib_usrq_object, + uevent.uobject); + + atomic_dec(&us->uxrcd->refcnt); + } + + ib_uverbs_release_uevent(uevent); + return ret; +} + +static int UVERBS_HANDLER(UVERBS_METHOD_SRQ_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_usrq_object *obj = container_of( + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_SRQ_HANDLE), + typeof(*obj), uevent.uobject); + struct ib_pd *pd = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_CREATE_SRQ_PD_HANDLE); + struct ib_srq_init_attr attr = {}; + struct ib_uobject *xrcd_uobj; + struct ib_srq *srq; + u64 user_handle; + int ret; + + ret = uverbs_copy_from(&attr.attr.max_sge, attrs, + UVERBS_ATTR_CREATE_SRQ_MAX_SGE); + if (!ret) + ret = uverbs_copy_from(&attr.attr.max_wr, attrs, + UVERBS_ATTR_CREATE_SRQ_MAX_WR); + if (!ret) + ret = uverbs_copy_from(&attr.attr.srq_limit, attrs, + UVERBS_ATTR_CREATE_SRQ_LIMIT); + if (!ret) + ret = uverbs_copy_from(&user_handle, attrs, + UVERBS_ATTR_CREATE_SRQ_USER_HANDLE); + if (!ret) + ret = uverbs_get_const(&attr.srq_type, attrs, + UVERBS_ATTR_CREATE_SRQ_TYPE); + if (ret) + return ret; + + if (ib_srq_has_cq(attr.srq_type)) { + attr.ext.cq = uverbs_attr_get_obj(attrs, + UVERBS_ATTR_CREATE_SRQ_CQ_HANDLE); + if (IS_ERR(attr.ext.cq)) + return PTR_ERR(attr.ext.cq); + } + + switch (attr.srq_type) { + case IB_UVERBS_SRQT_XRC: + xrcd_uobj = uverbs_attr_get_uobject(attrs, + UVERBS_ATTR_CREATE_SRQ_XRCD_HANDLE); + if (IS_ERR(xrcd_uobj)) + return PTR_ERR(xrcd_uobj); + + attr.ext.xrc.xrcd = (struct ib_xrcd *)xrcd_uobj->object; + if (!attr.ext.xrc.xrcd) + return -EINVAL; + obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, + uobject); + atomic_inc(&obj->uxrcd->refcnt); + break; + case IB_UVERBS_SRQT_TM: + ret = uverbs_copy_from(&attr.ext.tag_matching.max_num_tags, + attrs, + UVERBS_ATTR_CREATE_SRQ_MAX_NUM_TAGS); + if (ret) + return ret; + break; + case IB_UVERBS_SRQT_BASIC: + break; + default: + return -EINVAL; + } + + obj->uevent.event_file = ib_uverbs_get_async_event(attrs, + UVERBS_ATTR_CREATE_SRQ_EVENT_FD); + INIT_LIST_HEAD(&obj->uevent.event_list); + attr.event_handler = ib_uverbs_srq_event_handler; + obj->uevent.uobject.user_handle = user_handle; + + srq = ib_create_srq_user(pd, &attr, obj, &attrs->driver_udata); + if (IS_ERR(srq)) { + ret = PTR_ERR(srq); + goto err; + } + + obj->uevent.uobject.object = srq; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_SRQ_HANDLE); + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_SRQ_RESP_MAX_WR, + &attr.attr.max_wr, + sizeof(attr.attr.max_wr)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_SRQ_RESP_MAX_SGE, + &attr.attr.max_sge, + sizeof(attr.attr.max_sge)); + if (ret) + return ret; + + if (attr.srq_type == IB_SRQT_XRC) { + ret = uverbs_copy_to(attrs, + UVERBS_ATTR_CREATE_SRQ_RESP_SRQ_NUM, + &srq->ext.xrc.srq_num, + sizeof(srq->ext.xrc.srq_num)); + if (ret) + return ret; + } + + return 0; +err: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); + if (attr.srq_type == IB_SRQT_XRC) + atomic_dec(&obj->uxrcd->refcnt); + return ret; +}; + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_SRQ_CREATE, + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_HANDLE, + UVERBS_OBJECT_SRQ, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(UVERBS_ATTR_CREATE_SRQ_TYPE, + enum ib_uverbs_srq_type, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_USER_HANDLE, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_LIMIT, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_XRCD_HANDLE, + UVERBS_OBJECT_XRCD, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_SRQ_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_SRQ_MAX_NUM_TAGS, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_SRQ_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_SRQ_RESP_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_SRQ_RESP_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_SRQ_RESP_SRQ_NUM, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_UHW()); + +static int UVERBS_HANDLER(UVERBS_METHOD_SRQ_DESTROY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_SRQ_HANDLE); + struct ib_usrq_object *obj = + container_of(uobj, struct ib_usrq_object, uevent.uobject); + struct ib_uverbs_destroy_srq_resp resp = { + .events_reported = obj->uevent.events_reported + }; + + return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_SRQ_RESP, &resp, + sizeof(resp)); +} + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_SRQ_DESTROY, + UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_SRQ_HANDLE, + UVERBS_OBJECT_SRQ, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_DESTROY_SRQ_RESP, + UVERBS_ATTR_TYPE(struct ib_uverbs_destroy_srq_resp), + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_OBJECT( + UVERBS_OBJECT_SRQ, + UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_usrq_object), + uverbs_free_srq), + &UVERBS_METHOD(UVERBS_METHOD_SRQ_CREATE), + &UVERBS_METHOD(UVERBS_METHOD_SRQ_DESTROY) +); + +const struct uapi_definition uverbs_def_obj_srq[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_SRQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_srq)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_wq.c b/drivers/infiniband/core/uverbs_std_types_wq.c new file mode 100644 index 000000000000..cad842ede077 --- /dev/null +++ b/drivers/infiniband/core/uverbs_std_types_wq.c @@ -0,0 +1,194 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include <rdma/uverbs_std_types.h> +#include "rdma_core.h" +#include "uverbs.h" + +static int uverbs_free_wq(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct ib_wq *wq = uobject->object; + struct ib_uwq_object *uwq = + container_of(uobject, struct ib_uwq_object, uevent.uobject); + int ret; + + ret = ib_destroy_wq(wq, &attrs->driver_udata); + if (ib_is_destroy_retryable(ret, why, uobject)) + return ret; + + ib_uverbs_release_uevent(&uwq->uevent); + return ret; +} + +static int UVERBS_HANDLER(UVERBS_METHOD_WQ_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uwq_object *obj = container_of( + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_WQ_HANDLE), + typeof(*obj), uevent.uobject); + struct ib_pd *pd = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_CREATE_WQ_PD_HANDLE); + struct ib_cq *cq = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_CREATE_WQ_CQ_HANDLE); + struct ib_wq_init_attr wq_init_attr = {}; + struct ib_wq *wq; + u64 user_handle; + int ret; + + ret = uverbs_get_flags32(&wq_init_attr.create_flags, attrs, + UVERBS_ATTR_CREATE_WQ_FLAGS, + IB_UVERBS_WQ_FLAGS_CVLAN_STRIPPING | + IB_UVERBS_WQ_FLAGS_SCATTER_FCS | + IB_UVERBS_WQ_FLAGS_DELAY_DROP | + IB_UVERBS_WQ_FLAGS_PCI_WRITE_END_PADDING); + if (!ret) + ret = uverbs_copy_from(&wq_init_attr.max_sge, attrs, + UVERBS_ATTR_CREATE_WQ_MAX_SGE); + if (!ret) + ret = uverbs_copy_from(&wq_init_attr.max_wr, attrs, + UVERBS_ATTR_CREATE_WQ_MAX_WR); + if (!ret) + ret = uverbs_copy_from(&user_handle, attrs, + UVERBS_ATTR_CREATE_WQ_USER_HANDLE); + if (!ret) + ret = uverbs_get_const(&wq_init_attr.wq_type, attrs, + UVERBS_ATTR_CREATE_WQ_TYPE); + if (ret) + return ret; + + if (wq_init_attr.wq_type != IB_WQT_RQ) + return -EINVAL; + + obj->uevent.event_file = ib_uverbs_get_async_event(attrs, + UVERBS_ATTR_CREATE_WQ_EVENT_FD); + obj->uevent.uobject.user_handle = user_handle; + INIT_LIST_HEAD(&obj->uevent.event_list); + wq_init_attr.event_handler = ib_uverbs_wq_event_handler; + wq_init_attr.wq_context = attrs->ufile; + wq_init_attr.cq = cq; + + wq = pd->device->ops.create_wq(pd, &wq_init_attr, &attrs->driver_udata); + if (IS_ERR(wq)) { + ret = PTR_ERR(wq); + goto err; + } + + obj->uevent.uobject.object = wq; + wq->wq_type = wq_init_attr.wq_type; + wq->cq = cq; + wq->pd = pd; + wq->device = pd->device; + wq->wq_context = wq_init_attr.wq_context; + atomic_set(&wq->usecnt, 0); + atomic_inc(&pd->usecnt); + atomic_inc(&cq->usecnt); + wq->uobject = obj; + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_CREATE_WQ_HANDLE); + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_WQ_RESP_MAX_WR, + &wq_init_attr.max_wr, + sizeof(wq_init_attr.max_wr)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_WQ_RESP_MAX_SGE, + &wq_init_attr.max_sge, + sizeof(wq_init_attr.max_sge)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_CREATE_WQ_RESP_WQ_NUM, + &wq->wq_num, + sizeof(wq->wq_num)); + return ret; + +err: + if (obj->uevent.event_file) + uverbs_uobject_put(&obj->uevent.event_file->uobj); + return ret; +}; + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_WQ_CREATE, + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_WQ_HANDLE, + UVERBS_OBJECT_WQ, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_WQ_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(UVERBS_ATTR_CREATE_WQ_TYPE, + enum ib_wq_type, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_WQ_USER_HANDLE, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_WQ_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_CREATE_WQ_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_FLAGS_IN(UVERBS_ATTR_CREATE_WQ_FLAGS, + enum ib_uverbs_wq_flags, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_CREATE_WQ_CQ_HANDLE, + UVERBS_OBJECT_CQ, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_FD(UVERBS_ATTR_CREATE_WQ_EVENT_FD, + UVERBS_OBJECT_ASYNC_EVENT, + UVERBS_ACCESS_READ, + UA_OPTIONAL), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_WQ_RESP_MAX_WR, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_WQ_RESP_MAX_SGE, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_CREATE_WQ_RESP_WQ_NUM, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_UHW()); + +static int UVERBS_HANDLER(UVERBS_METHOD_WQ_DESTROY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_WQ_HANDLE); + struct ib_uwq_object *obj = + container_of(uobj, struct ib_uwq_object, uevent.uobject); + + return uverbs_copy_to(attrs, UVERBS_ATTR_DESTROY_WQ_RESP, + &obj->uevent.events_reported, + sizeof(obj->uevent.events_reported)); +} + +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_WQ_DESTROY, + UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_WQ_HANDLE, + UVERBS_OBJECT_WQ, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_DESTROY_WQ_RESP, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY)); + + +DECLARE_UVERBS_NAMED_OBJECT( + UVERBS_OBJECT_WQ, + UVERBS_TYPE_ALLOC_IDR_SZ(sizeof(struct ib_uwq_object), uverbs_free_wq), + &UVERBS_METHOD(UVERBS_METHOD_WQ_CREATE), + &UVERBS_METHOD(UVERBS_METHOD_WQ_DESTROY) +); + +const struct uapi_definition uverbs_def_obj_wq[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_WQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_wq)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c index 3f121ac31e0a..5addc8fae3f3 100644 --- a/drivers/infiniband/core/uverbs_uapi.c +++ b/drivers/infiniband/core/uverbs_uapi.c @@ -634,6 +634,9 @@ static const struct uapi_definition uverbs_core_api[] = { UAPI_DEF_CHAIN(uverbs_def_obj_flow_action), UAPI_DEF_CHAIN(uverbs_def_obj_intf), UAPI_DEF_CHAIN(uverbs_def_obj_mr), + UAPI_DEF_CHAIN(uverbs_def_obj_qp), + UAPI_DEF_CHAIN(uverbs_def_obj_srq), + UAPI_DEF_CHAIN(uverbs_def_obj_wq), UAPI_DEF_CHAIN(uverbs_def_write_intf), {}, }; diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index 56a71337112c..307886737646 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -50,6 +50,7 @@ #include <rdma/ib_cache.h> #include <rdma/ib_addr.h> #include <rdma/rw.h> +#include <rdma/lag.h> #include "core_priv.h" #include <trace/events/rdma_core.h> @@ -500,8 +501,10 @@ rdma_update_sgid_attr(struct rdma_ah_attr *ah_attr, static struct ib_ah *_rdma_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr, u32 flags, - struct ib_udata *udata) + struct ib_udata *udata, + struct net_device *xmit_slave) { + struct rdma_ah_init_attr init_attr = {}; struct ib_device *device = pd->device; struct ib_ah *ah; int ret; @@ -521,8 +524,11 @@ static struct ib_ah *_rdma_create_ah(struct ib_pd *pd, ah->pd = pd; ah->type = ah_attr->type; ah->sgid_attr = rdma_update_sgid_attr(ah_attr, NULL); + init_attr.ah_attr = ah_attr; + init_attr.flags = flags; + init_attr.xmit_slave = xmit_slave; - ret = device->ops.create_ah(ah, ah_attr, flags, udata); + ret = device->ops.create_ah(ah, &init_attr, udata); if (ret) { kfree(ah); return ERR_PTR(ret); @@ -547,15 +553,22 @@ struct ib_ah *rdma_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr, u32 flags) { const struct ib_gid_attr *old_sgid_attr; + struct net_device *slave; struct ib_ah *ah; int ret; ret = rdma_fill_sgid_attr(pd->device, ah_attr, &old_sgid_attr); if (ret) return ERR_PTR(ret); - - ah = _rdma_create_ah(pd, ah_attr, flags, NULL); - + slave = rdma_lag_get_ah_roce_slave(pd->device, ah_attr, + (flags & RDMA_CREATE_AH_SLEEPABLE) ? + GFP_KERNEL : GFP_ATOMIC); + if (IS_ERR(slave)) { + rdma_unfill_sgid_attr(ah_attr, old_sgid_attr); + return (void *)slave; + } + ah = _rdma_create_ah(pd, ah_attr, flags, NULL, slave); + rdma_lag_put_ah_roce_slave(slave); rdma_unfill_sgid_attr(ah_attr, old_sgid_attr); return ah; } @@ -594,7 +607,8 @@ struct ib_ah *rdma_create_user_ah(struct ib_pd *pd, } } - ah = _rdma_create_ah(pd, ah_attr, RDMA_CREATE_AH_SLEEPABLE, udata); + ah = _rdma_create_ah(pd, ah_attr, RDMA_CREATE_AH_SLEEPABLE, + udata, NULL); out: rdma_unfill_sgid_attr(ah_attr, old_sgid_attr); @@ -967,15 +981,29 @@ EXPORT_SYMBOL(rdma_destroy_ah_user); /* Shared receive queues */ -struct ib_srq *ib_create_srq(struct ib_pd *pd, - struct ib_srq_init_attr *srq_init_attr) +/** + * ib_create_srq_user - Creates a SRQ associated with the specified protection + * domain. + * @pd: The protection domain associated with the SRQ. + * @srq_init_attr: A list of initial attributes required to create the + * SRQ. If SRQ creation succeeds, then the attributes are updated to + * the actual capabilities of the created SRQ. + * @uobject: uobject pointer if this is not a kernel SRQ + * @udata: udata pointer if this is not a kernel SRQ + * + * srq_attr->max_wr and srq_attr->max_sge are read the determine the + * requested size of the SRQ, and set to the actual values allocated + * on return. If ib_create_srq() succeeds, then max_wr and max_sge + * will always be at least as large as the requested values. + */ +struct ib_srq *ib_create_srq_user(struct ib_pd *pd, + struct ib_srq_init_attr *srq_init_attr, + struct ib_usrq_object *uobject, + struct ib_udata *udata) { struct ib_srq *srq; int ret; - if (!pd->device->ops.create_srq) - return ERR_PTR(-EOPNOTSUPP); - srq = rdma_zalloc_drv_obj(pd->device, ib_srq); if (!srq) return ERR_PTR(-ENOMEM); @@ -985,6 +1013,7 @@ struct ib_srq *ib_create_srq(struct ib_pd *pd, srq->event_handler = srq_init_attr->event_handler; srq->srq_context = srq_init_attr->srq_context; srq->srq_type = srq_init_attr->srq_type; + srq->uobject = uobject; if (ib_srq_has_cq(srq->srq_type)) { srq->ext.cq = srq_init_attr->ext.cq; @@ -996,7 +1025,7 @@ struct ib_srq *ib_create_srq(struct ib_pd *pd, } atomic_inc(&pd->usecnt); - ret = pd->device->ops.create_srq(srq, srq_init_attr, NULL); + ret = pd->device->ops.create_srq(srq, srq_init_attr, udata); if (ret) { atomic_dec(&srq->pd->usecnt); if (srq->srq_type == IB_SRQT_XRC) @@ -1009,7 +1038,7 @@ struct ib_srq *ib_create_srq(struct ib_pd *pd, return srq; } -EXPORT_SYMBOL(ib_create_srq); +EXPORT_SYMBOL(ib_create_srq_user); int ib_modify_srq(struct ib_srq *srq, struct ib_srq_attr *srq_attr, @@ -1061,13 +1090,6 @@ static void __ib_shared_qp_event_handler(struct ib_event *event, void *context) spin_unlock_irqrestore(&qp->device->qp_open_list_lock, flags); } -static void __ib_insert_xrcd_qp(struct ib_xrcd *xrcd, struct ib_qp *qp) -{ - mutex_lock(&xrcd->tgt_qp_mutex); - list_add(&qp->xrcd_list, &xrcd->tgt_qp_list); - mutex_unlock(&xrcd->tgt_qp_mutex); -} - static struct ib_qp *__ib_open_qp(struct ib_qp *real_qp, void (*event_handler)(struct ib_event *, void *), void *qp_context) @@ -1110,16 +1132,15 @@ struct ib_qp *ib_open_qp(struct ib_xrcd *xrcd, if (qp_open_attr->qp_type != IB_QPT_XRC_TGT) return ERR_PTR(-EINVAL); - qp = ERR_PTR(-EINVAL); - mutex_lock(&xrcd->tgt_qp_mutex); - list_for_each_entry(real_qp, &xrcd->tgt_qp_list, xrcd_list) { - if (real_qp->qp_num == qp_open_attr->qp_num) { - qp = __ib_open_qp(real_qp, qp_open_attr->event_handler, - qp_open_attr->qp_context); - break; - } + down_read(&xrcd->tgt_qps_rwsem); + real_qp = xa_load(&xrcd->tgt_qps, qp_open_attr->qp_num); + if (!real_qp) { + up_read(&xrcd->tgt_qps_rwsem); + return ERR_PTR(-EINVAL); } - mutex_unlock(&xrcd->tgt_qp_mutex); + qp = __ib_open_qp(real_qp, qp_open_attr->event_handler, + qp_open_attr->qp_context); + up_read(&xrcd->tgt_qps_rwsem); return qp; } EXPORT_SYMBOL(ib_open_qp); @@ -1128,6 +1149,7 @@ static struct ib_qp *create_xrc_qp_user(struct ib_qp *qp, struct ib_qp_init_attr *qp_init_attr) { struct ib_qp *real_qp = qp; + int err; qp->event_handler = __ib_shared_qp_event_handler; qp->qp_context = qp; @@ -1143,7 +1165,12 @@ static struct ib_qp *create_xrc_qp_user(struct ib_qp *qp, if (IS_ERR(qp)) return qp; - __ib_insert_xrcd_qp(qp_init_attr->xrcd, real_qp); + err = xa_err(xa_store(&qp_init_attr->xrcd->tgt_qps, real_qp->qp_num, + real_qp, GFP_KERNEL)); + if (err) { + ib_close_qp(qp); + return ERR_PTR(err); + } return qp; } @@ -1633,11 +1660,35 @@ static int _ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr, const struct ib_gid_attr *old_sgid_attr_alt_av; int ret; + attr->xmit_slave = NULL; if (attr_mask & IB_QP_AV) { ret = rdma_fill_sgid_attr(qp->device, &attr->ah_attr, &old_sgid_attr_av); if (ret) return ret; + + if (attr->ah_attr.type == RDMA_AH_ATTR_TYPE_ROCE && + is_qp_type_connected(qp)) { + struct net_device *slave; + + /* + * If the user provided the qp_attr then we have to + * resolve it. Kerne users have to provide already + * resolved rdma_ah_attr's. + */ + if (udata) { + ret = ib_resolve_eth_dmac(qp->device, + &attr->ah_attr); + if (ret) + goto out_av; + } + slave = rdma_lag_get_ah_roce_slave(qp->device, + &attr->ah_attr, + GFP_KERNEL); + if (IS_ERR(slave)) + goto out_av; + attr->xmit_slave = slave; + } } if (attr_mask & IB_QP_ALT_PATH) { /* @@ -1659,23 +1710,11 @@ static int _ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr, if (!(rdma_protocol_ib(qp->device, attr->alt_ah_attr.port_num) && rdma_protocol_ib(qp->device, port))) { - ret = EINVAL; + ret = -EINVAL; goto out; } } - /* - * If the user provided the qp_attr then we have to resolve it. Kernel - * users have to provide already resolved rdma_ah_attr's - */ - if (udata && (attr_mask & IB_QP_AV) && - attr->ah_attr.type == RDMA_AH_ATTR_TYPE_ROCE && - is_qp_type_connected(qp)) { - ret = ib_resolve_eth_dmac(qp->device, &attr->ah_attr); - if (ret) - goto out; - } - if (rdma_ib_or_roce(qp->device, port)) { if (attr_mask & IB_QP_RQ_PSN && attr->rq_psn & ~0xffffff) { dev_warn(&qp->device->dev, @@ -1717,8 +1756,10 @@ out: if (attr_mask & IB_QP_ALT_PATH) rdma_unfill_sgid_attr(&attr->alt_ah_attr, old_sgid_attr_alt_av); out_av: - if (attr_mask & IB_QP_AV) + if (attr_mask & IB_QP_AV) { + rdma_lag_put_ah_roce_slave(attr->xmit_slave); rdma_unfill_sgid_attr(&attr->ah_attr, old_sgid_attr_av); + } return ret; } @@ -1760,7 +1801,7 @@ int ib_get_eth_speed(struct ib_device *dev, u8 port_num, u8 *speed, u8 *width) dev_put(netdev); - if (!rc) { + if (!rc && lksettings.base.speed != (u32)SPEED_UNKNOWN) { netdev_speed = lksettings.base.speed; } else { netdev_speed = SPEED_1000; @@ -1844,21 +1885,18 @@ static int __ib_destroy_shared_qp(struct ib_qp *qp) real_qp = qp->real_qp; xrcd = real_qp->xrcd; - - mutex_lock(&xrcd->tgt_qp_mutex); + down_write(&xrcd->tgt_qps_rwsem); ib_close_qp(qp); if (atomic_read(&real_qp->usecnt) == 0) - list_del(&real_qp->xrcd_list); + xa_erase(&xrcd->tgt_qps, real_qp->qp_num); else real_qp = NULL; - mutex_unlock(&xrcd->tgt_qp_mutex); + up_write(&xrcd->tgt_qps_rwsem); if (real_qp) { ret = ib_destroy_qp(real_qp); if (!ret) atomic_dec(&xrcd->usecnt); - else - __ib_insert_xrcd_qp(xrcd, real_qp); } return 0; @@ -1962,6 +2000,9 @@ EXPORT_SYMBOL(__ib_create_cq); int rdma_set_cq_moderation(struct ib_cq *cq, u16 cq_count, u16 cq_period) { + if (cq->shared) + return -EOPNOTSUPP; + return cq->device->ops.modify_cq ? cq->device->ops.modify_cq(cq, cq_count, cq_period) : -EOPNOTSUPP; @@ -1970,6 +2011,9 @@ EXPORT_SYMBOL(rdma_set_cq_moderation); int ib_destroy_cq_user(struct ib_cq *cq, struct ib_udata *udata) { + if (WARN_ON_ONCE(cq->shared)) + return -EOPNOTSUPP; + if (atomic_read(&cq->usecnt)) return -EBUSY; @@ -1982,6 +2026,9 @@ EXPORT_SYMBOL(ib_destroy_cq_user); int ib_resize_cq(struct ib_cq *cq, int cqe) { + if (cq->shared) + return -EOPNOTSUPP; + return cq->device->ops.resize_cq ? cq->device->ops.resize_cq(cq, cqe, NULL) : -EOPNOTSUPP; } @@ -2025,6 +2072,9 @@ int ib_advise_mr(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice, if (!pd->device->ops.advise_mr) return -EOPNOTSUPP; + if (!num_sge) + return 0; + return pd->device->ops.advise_mr(pd, advice, flags, sg_list, num_sge, NULL); } @@ -2052,11 +2102,10 @@ int ib_dereg_mr_user(struct ib_mr *mr, struct ib_udata *udata) EXPORT_SYMBOL(ib_dereg_mr_user); /** - * ib_alloc_mr_user() - Allocates a memory region + * ib_alloc_mr() - Allocates a memory region * @pd: protection domain associated with the region * @mr_type: memory region type * @max_num_sg: maximum sg entries available for registration. - * @udata: user data or null for kernel objects * * Notes: * Memory registeration page/sg lists must not exceed max_num_sg. @@ -2064,8 +2113,8 @@ EXPORT_SYMBOL(ib_dereg_mr_user); * max_num_sg * used_page_size. * */ -struct ib_mr *ib_alloc_mr_user(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) +struct ib_mr *ib_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, + u32 max_num_sg) { struct ib_mr *mr; @@ -2080,25 +2129,26 @@ struct ib_mr *ib_alloc_mr_user(struct ib_pd *pd, enum ib_mr_type mr_type, goto out; } - mr = pd->device->ops.alloc_mr(pd, mr_type, max_num_sg, udata); - if (!IS_ERR(mr)) { - mr->device = pd->device; - mr->pd = pd; - mr->dm = NULL; - mr->uobject = NULL; - atomic_inc(&pd->usecnt); - mr->need_inval = false; - mr->res.type = RDMA_RESTRACK_MR; - rdma_restrack_kadd(&mr->res); - mr->type = mr_type; - mr->sig_attrs = NULL; - } + mr = pd->device->ops.alloc_mr(pd, mr_type, max_num_sg); + if (IS_ERR(mr)) + goto out; + + mr->device = pd->device; + mr->pd = pd; + mr->dm = NULL; + mr->uobject = NULL; + atomic_inc(&pd->usecnt); + mr->need_inval = false; + mr->res.type = RDMA_RESTRACK_MR; + rdma_restrack_kadd(&mr->res); + mr->type = mr_type; + mr->sig_attrs = NULL; out: trace_mr_alloc(pd, mr_type, max_num_sg, mr); return mr; } -EXPORT_SYMBOL(ib_alloc_mr_user); +EXPORT_SYMBOL(ib_alloc_mr); /** * ib_alloc_mr_integrity() - Allocates an integrity memory region @@ -2160,54 +2210,6 @@ out: } EXPORT_SYMBOL(ib_alloc_mr_integrity); -/* "Fast" memory regions */ - -struct ib_fmr *ib_alloc_fmr(struct ib_pd *pd, - int mr_access_flags, - struct ib_fmr_attr *fmr_attr) -{ - struct ib_fmr *fmr; - - if (!pd->device->ops.alloc_fmr) - return ERR_PTR(-EOPNOTSUPP); - - fmr = pd->device->ops.alloc_fmr(pd, mr_access_flags, fmr_attr); - if (!IS_ERR(fmr)) { - fmr->device = pd->device; - fmr->pd = pd; - atomic_inc(&pd->usecnt); - } - - return fmr; -} -EXPORT_SYMBOL(ib_alloc_fmr); - -int ib_unmap_fmr(struct list_head *fmr_list) -{ - struct ib_fmr *fmr; - - if (list_empty(fmr_list)) - return 0; - - fmr = list_entry(fmr_list->next, struct ib_fmr, list); - return fmr->device->ops.unmap_fmr(fmr_list); -} -EXPORT_SYMBOL(ib_unmap_fmr); - -int ib_dealloc_fmr(struct ib_fmr *fmr) -{ - struct ib_pd *pd; - int ret; - - pd = fmr->pd; - ret = fmr->device->ops.dealloc_fmr(fmr); - if (!ret) - atomic_dec(&pd->usecnt); - - return ret; -} -EXPORT_SYMBOL(ib_dealloc_fmr); - /* Multicast groups */ static bool is_valid_mcast_lid(struct ib_qp *qp, u16 lid) @@ -2284,45 +2286,57 @@ int ib_detach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid) } EXPORT_SYMBOL(ib_detach_mcast); -struct ib_xrcd *__ib_alloc_xrcd(struct ib_device *device, const char *caller) +/** + * ib_alloc_xrcd_user - Allocates an XRC domain. + * @device: The device on which to allocate the XRC domain. + * @inode: inode to connect XRCD + * @udata: Valid user data or NULL for kernel object + */ +struct ib_xrcd *ib_alloc_xrcd_user(struct ib_device *device, + struct inode *inode, struct ib_udata *udata) { struct ib_xrcd *xrcd; + int ret; if (!device->ops.alloc_xrcd) return ERR_PTR(-EOPNOTSUPP); - xrcd = device->ops.alloc_xrcd(device, NULL); - if (!IS_ERR(xrcd)) { - xrcd->device = device; - xrcd->inode = NULL; - atomic_set(&xrcd->usecnt, 0); - mutex_init(&xrcd->tgt_qp_mutex); - INIT_LIST_HEAD(&xrcd->tgt_qp_list); - } + xrcd = rdma_zalloc_drv_obj(device, ib_xrcd); + if (!xrcd) + return ERR_PTR(-ENOMEM); + xrcd->device = device; + xrcd->inode = inode; + atomic_set(&xrcd->usecnt, 0); + init_rwsem(&xrcd->tgt_qps_rwsem); + xa_init(&xrcd->tgt_qps); + + ret = device->ops.alloc_xrcd(xrcd, udata); + if (ret) + goto err; return xrcd; +err: + kfree(xrcd); + return ERR_PTR(ret); } -EXPORT_SYMBOL(__ib_alloc_xrcd); +EXPORT_SYMBOL(ib_alloc_xrcd_user); -int ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata) +/** + * ib_dealloc_xrcd_user - Deallocates an XRC domain. + * @xrcd: The XRC domain to deallocate. + * @udata: Valid user data or NULL for kernel object + */ +int ib_dealloc_xrcd_user(struct ib_xrcd *xrcd, struct ib_udata *udata) { - struct ib_qp *qp; - int ret; - if (atomic_read(&xrcd->usecnt)) return -EBUSY; - while (!list_empty(&xrcd->tgt_qp_list)) { - qp = list_entry(xrcd->tgt_qp_list.next, struct ib_qp, xrcd_list); - ret = ib_destroy_qp(qp); - if (ret) - return ret; - } - mutex_destroy(&xrcd->tgt_qp_mutex); - - return xrcd->device->ops.dealloc_xrcd(xrcd, udata); + WARN_ON(!xa_empty(&xrcd->tgt_qps)); + xrcd->device->ops.dealloc_xrcd(xrcd, udata); + kfree(xrcd); + return 0; } -EXPORT_SYMBOL(ib_dealloc_xrcd); +EXPORT_SYMBOL(ib_dealloc_xrcd_user); /** * ib_create_wq - Creates a WQ associated with the specified protection @@ -2406,45 +2420,6 @@ int ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr, EXPORT_SYMBOL(ib_modify_wq); /* - * ib_create_rwq_ind_table - Creates a RQ Indirection Table. - * @device: The device on which to create the rwq indirection table. - * @ib_rwq_ind_table_init_attr: A list of initial attributes required to - * create the Indirection Table. - * - * Note: The life time of ib_rwq_ind_table_init_attr->ind_tbl is not less - * than the created ib_rwq_ind_table object and the caller is responsible - * for its memory allocation/free. - */ -struct ib_rwq_ind_table *ib_create_rwq_ind_table(struct ib_device *device, - struct ib_rwq_ind_table_init_attr *init_attr) -{ - struct ib_rwq_ind_table *rwq_ind_table; - int i; - u32 table_size; - - if (!device->ops.create_rwq_ind_table) - return ERR_PTR(-EOPNOTSUPP); - - table_size = (1 << init_attr->log_ind_tbl_size); - rwq_ind_table = device->ops.create_rwq_ind_table(device, - init_attr, NULL); - if (IS_ERR(rwq_ind_table)) - return rwq_ind_table; - - rwq_ind_table->ind_tbl = init_attr->ind_tbl; - rwq_ind_table->log_ind_tbl_size = init_attr->log_ind_tbl_size; - rwq_ind_table->device = device; - rwq_ind_table->uobject = NULL; - atomic_set(&rwq_ind_table->usecnt, 0); - - for (i = 0; i < table_size; i++) - atomic_inc(&rwq_ind_table->ind_tbl[i]->usecnt); - - return rwq_ind_table; -} -EXPORT_SYMBOL(ib_create_rwq_ind_table); - -/* * ib_destroy_rwq_ind_table - Destroys the specified Indirection Table. * @wq_ind_table: The Indirection Table to destroy. */ @@ -2574,6 +2549,7 @@ EXPORT_SYMBOL(ib_map_mr_sg_pi); * @page_size: page vector desired page size * * Constraints: + * * - The first sg element is allowed to have an offset. * - Each sg element must either be aligned to page_size or virtually * contiguous to the previous element. In case an sg element has a @@ -2607,10 +2583,12 @@ EXPORT_SYMBOL(ib_map_mr_sg); * @mr: memory region * @sgl: dma mapped scatterlist * @sg_nents: number of entries in sg - * @sg_offset_p: IN: start offset in bytes into sg - * OUT: offset in bytes for element n of the sg of the first + * @sg_offset_p: ==== ======================================================= + * IN start offset in bytes into sg + * OUT offset in bytes for element n of the sg of the first * byte that has not been processed where n is the return * value of this function. + * ==== ======================================================= * @set_page: driver page assignment function pointer * * Core service helper for drivers to convert the largest diff --git a/drivers/infiniband/hw/bnxt_re/Kconfig b/drivers/infiniband/hw/bnxt_re/Kconfig index b83f1cc38c52..0feac5132ce1 100644 --- a/drivers/infiniband/hw/bnxt_re/Kconfig +++ b/drivers/infiniband/hw/bnxt_re/Kconfig @@ -5,7 +5,7 @@ config INFINIBAND_BNXT_RE depends on ETHERNET && NETDEVICES && PCI && INET && DCB select NET_VENDOR_BROADCOM select BNXT - ---help--- + help This driver supports Broadcom NetXtreme-E 10/25/40/50 gigabit RoCE HCAs. To compile this driver as a module, choose M here: the module will be called bnxt_re. diff --git a/drivers/infiniband/hw/bnxt_re/hw_counters.c b/drivers/infiniband/hw/bnxt_re/hw_counters.c index 3421a0b15983..5f5408cdf008 100644 --- a/drivers/infiniband/hw/bnxt_re/hw_counters.c +++ b/drivers/infiniband/hw/bnxt_re/hw_counters.c @@ -132,7 +132,7 @@ int bnxt_re_ib_get_hw_stats(struct ib_device *ibdev, stats->value[BNXT_RE_RECOVERABLE_ERRORS] = le64_to_cpu(bnxt_re_stats->tx_bcast_pkts); stats->value[BNXT_RE_RX_DROPS] = - le64_to_cpu(bnxt_re_stats->rx_drop_pkts); + le64_to_cpu(bnxt_re_stats->rx_error_pkts); stats->value[BNXT_RE_RX_DISCARDS] = le64_to_cpu(bnxt_re_stats->rx_discard_pkts); stats->value[BNXT_RE_RX_PKTS] = diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index 95f6d493d1b9..1d7a9ca5240c 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -177,9 +177,6 @@ int bnxt_re_query_device(struct ib_device *ibdev, ib_attr->max_total_mcast_qp_attach = 0; ib_attr->max_ah = dev_attr->max_ah; - ib_attr->max_fmr = 0; - ib_attr->max_map_per_fmr = 0; - ib_attr->max_srq = dev_attr->max_srq; ib_attr->max_srq_wr = dev_attr->max_srq_wqes; ib_attr->max_srq_sge = dev_attr->max_srq_sges; @@ -631,11 +628,12 @@ static u8 bnxt_re_stack_to_dev_nw_type(enum rdma_network_type ntype) return nw_type; } -int bnxt_re_create_ah(struct ib_ah *ib_ah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata) +int bnxt_re_create_ah(struct ib_ah *ib_ah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { struct ib_pd *ib_pd = ib_ah->pd; struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd); + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; const struct ib_global_route *grh = rdma_ah_read_grh(ah_attr); struct bnxt_re_dev *rdev = pd->rdev; const struct ib_gid_attr *sgid_attr; @@ -673,7 +671,8 @@ int bnxt_re_create_ah(struct ib_ah *ib_ah, struct rdma_ah_attr *ah_attr, memcpy(ah->qplib_ah.dmac, ah_attr->roce.dmac, ETH_ALEN); rc = bnxt_qplib_create_ah(&rdev->qplib_res, &ah->qplib_ah, - !(flags & RDMA_CREATE_AH_SLEEPABLE)); + !(init_attr->flags & + RDMA_CREATE_AH_SLEEPABLE)); if (rc) { ibdev_err(&rdev->ibdev, "Failed to allocate HW AH"); return rc; @@ -753,12 +752,6 @@ static int bnxt_re_destroy_gsi_sqp(struct bnxt_re_qp *qp) gsi_sqp = rdev->gsi_ctx.gsi_sqp; gsi_sah = rdev->gsi_ctx.gsi_sah; - /* remove from active qp list */ - mutex_lock(&rdev->qp_lock); - list_del(&gsi_sqp->list); - mutex_unlock(&rdev->qp_lock); - atomic_dec(&rdev->qp_count); - ibdev_dbg(&rdev->ibdev, "Destroy the shadow AH\n"); bnxt_qplib_destroy_ah(&rdev->qplib_res, &gsi_sah->qplib_ah, @@ -773,6 +766,12 @@ static int bnxt_re_destroy_gsi_sqp(struct bnxt_re_qp *qp) } bnxt_qplib_free_qp_res(&rdev->qplib_res, &gsi_sqp->qplib_qp); + /* remove from active qp list */ + mutex_lock(&rdev->qp_lock); + list_del(&gsi_sqp->list); + mutex_unlock(&rdev->qp_lock); + atomic_dec(&rdev->qp_count); + kfree(rdev->gsi_ctx.sqp_tbl); kfree(gsi_sah); kfree(gsi_sqp); @@ -793,11 +792,6 @@ int bnxt_re_destroy_qp(struct ib_qp *ib_qp, struct ib_udata *udata) unsigned int flags; int rc; - mutex_lock(&rdev->qp_lock); - list_del(&qp->list); - mutex_unlock(&rdev->qp_lock); - atomic_dec(&rdev->qp_count); - bnxt_qplib_flush_cqn_wq(&qp->qplib_qp); rc = bnxt_qplib_destroy_qp(&rdev->qplib_res, &qp->qplib_qp); @@ -820,6 +814,11 @@ int bnxt_re_destroy_qp(struct ib_qp *ib_qp, struct ib_udata *udata) goto sh_fail; } + mutex_lock(&rdev->qp_lock); + list_del(&qp->list); + mutex_unlock(&rdev->qp_lock); + atomic_dec(&rdev->qp_count); + ib_umem_release(qp->rumem); ib_umem_release(qp->sumem); @@ -843,27 +842,95 @@ static u8 __from_ib_qp_type(enum ib_qp_type type) } } +static u16 bnxt_re_setup_rwqe_size(struct bnxt_qplib_qp *qplqp, + int rsge, int max) +{ + if (qplqp->wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) + rsge = max; + return bnxt_re_get_rwqe_size(rsge); +} + +static u16 bnxt_re_get_wqe_size(int ilsize, int nsge) +{ + u16 wqe_size, calc_ils; + + wqe_size = bnxt_re_get_swqe_size(nsge); + if (ilsize) { + calc_ils = sizeof(struct sq_send_hdr) + ilsize; + wqe_size = max_t(u16, calc_ils, wqe_size); + wqe_size = ALIGN(wqe_size, sizeof(struct sq_send_hdr)); + } + return wqe_size; +} + +static int bnxt_re_setup_swqe_size(struct bnxt_re_qp *qp, + struct ib_qp_init_attr *init_attr) +{ + struct bnxt_qplib_dev_attr *dev_attr; + struct bnxt_qplib_qp *qplqp; + struct bnxt_re_dev *rdev; + struct bnxt_qplib_q *sq; + int align, ilsize; + + rdev = qp->rdev; + qplqp = &qp->qplib_qp; + sq = &qplqp->sq; + dev_attr = &rdev->dev_attr; + + align = sizeof(struct sq_send_hdr); + ilsize = ALIGN(init_attr->cap.max_inline_data, align); + + sq->wqe_size = bnxt_re_get_wqe_size(ilsize, sq->max_sge); + if (sq->wqe_size > bnxt_re_get_swqe_size(dev_attr->max_qp_sges)) + return -EINVAL; + /* For gen p4 and gen p5 backward compatibility mode + * wqe size is fixed to 128 bytes + */ + if (sq->wqe_size < bnxt_re_get_swqe_size(dev_attr->max_qp_sges) && + qplqp->wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) + sq->wqe_size = bnxt_re_get_swqe_size(dev_attr->max_qp_sges); + + if (init_attr->cap.max_inline_data) { + qplqp->max_inline_data = sq->wqe_size - + sizeof(struct sq_send_hdr); + init_attr->cap.max_inline_data = qplqp->max_inline_data; + if (qplqp->wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) + sq->max_sge = qplqp->max_inline_data / + sizeof(struct sq_sge); + } + + return 0; +} + static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd, struct bnxt_re_qp *qp, struct ib_udata *udata) { + struct bnxt_qplib_qp *qplib_qp; + struct bnxt_re_ucontext *cntx; struct bnxt_re_qp_req ureq; - struct bnxt_qplib_qp *qplib_qp = &qp->qplib_qp; - struct ib_umem *umem; int bytes = 0, psn_sz; - struct bnxt_re_ucontext *cntx = rdma_udata_to_drv_context( - udata, struct bnxt_re_ucontext, ib_uctx); + struct ib_umem *umem; + int psn_nume; + qplib_qp = &qp->qplib_qp; + cntx = rdma_udata_to_drv_context(udata, struct bnxt_re_ucontext, + ib_uctx); if (ib_copy_from_udata(&ureq, udata, sizeof(ureq))) return -EFAULT; - bytes = (qplib_qp->sq.max_wqe * BNXT_QPLIB_MAX_SQE_ENTRY_SIZE); + bytes = (qplib_qp->sq.max_wqe * qplib_qp->sq.wqe_size); /* Consider mapping PSN search memory only for RC QPs. */ if (qplib_qp->type == CMDQ_CREATE_QP_TYPE_RC) { psn_sz = bnxt_qplib_is_chip_gen_p5(rdev->chip_ctx) ? - sizeof(struct sq_psn_search_ext) : - sizeof(struct sq_psn_search); - bytes += (qplib_qp->sq.max_wqe * psn_sz); + sizeof(struct sq_psn_search_ext) : + sizeof(struct sq_psn_search); + psn_nume = (qplib_qp->wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) ? + qplib_qp->sq.max_wqe : + ((qplib_qp->sq.max_wqe * qplib_qp->sq.wqe_size) / + sizeof(struct bnxt_qplib_sge)); + bytes += (psn_nume * psn_sz); } + bytes = PAGE_ALIGN(bytes); umem = ib_umem_get(&rdev->ibdev, ureq.qpsva, bytes, IB_ACCESS_LOCAL_WRITE); @@ -879,7 +946,7 @@ static int bnxt_re_init_user_qp(struct bnxt_re_dev *rdev, struct bnxt_re_pd *pd, qplib_qp->qp_handle = ureq.qp_handle; if (!qp->qplib_qp.srq) { - bytes = (qplib_qp->rq.max_wqe * BNXT_QPLIB_MAX_RQE_ENTRY_SIZE); + bytes = (qplib_qp->rq.max_wqe * qplib_qp->rq.wqe_size); bytes = PAGE_ALIGN(bytes); umem = ib_umem_get(&rdev->ibdev, ureq.qprva, bytes, IB_ACCESS_LOCAL_WRITE); @@ -976,6 +1043,7 @@ static struct bnxt_re_qp *bnxt_re_create_shadow_qp qp->qplib_qp.sig_type = true; /* Shadow QP SQ depth should be same as QP1 RQ depth */ + qp->qplib_qp.sq.wqe_size = bnxt_re_get_wqe_size(0, 6); qp->qplib_qp.sq.max_wqe = qp1_qp->rq.max_wqe; qp->qplib_qp.sq.max_sge = 2; /* Q full delta can be 1 since it is internal QP */ @@ -986,6 +1054,7 @@ static struct bnxt_re_qp *bnxt_re_create_shadow_qp qp->qplib_qp.scq = qp1_qp->scq; qp->qplib_qp.rcq = qp1_qp->rcq; + qp->qplib_qp.rq.wqe_size = bnxt_re_get_rwqe_size(6); qp->qplib_qp.rq.max_wqe = qp1_qp->rq.max_wqe; qp->qplib_qp.rq.max_sge = qp1_qp->rq.max_sge; /* Q full delta can be 1 since it is internal QP */ @@ -1021,10 +1090,12 @@ static int bnxt_re_init_rq_attr(struct bnxt_re_qp *qp, struct bnxt_qplib_dev_attr *dev_attr; struct bnxt_qplib_qp *qplqp; struct bnxt_re_dev *rdev; + struct bnxt_qplib_q *rq; int entries; rdev = qp->rdev; qplqp = &qp->qplib_qp; + rq = &qplqp->rq; dev_attr = &rdev->dev_attr; if (init_attr->srq) { @@ -1036,23 +1107,23 @@ static int bnxt_re_init_rq_attr(struct bnxt_re_qp *qp, return -EINVAL; } qplqp->srq = &srq->qplib_srq; - qplqp->rq.max_wqe = 0; + rq->max_wqe = 0; } else { + rq->max_sge = init_attr->cap.max_recv_sge; + if (rq->max_sge > dev_attr->max_qp_sges) + rq->max_sge = dev_attr->max_qp_sges; + init_attr->cap.max_recv_sge = rq->max_sge; + rq->wqe_size = bnxt_re_setup_rwqe_size(qplqp, rq->max_sge, + dev_attr->max_qp_sges); /* Allocate 1 more than what's provided so posting max doesn't * mean empty. */ entries = roundup_pow_of_two(init_attr->cap.max_recv_wr + 1); - qplqp->rq.max_wqe = min_t(u32, entries, - dev_attr->max_qp_wqes + 1); - - qplqp->rq.q_full_delta = qplqp->rq.max_wqe - - init_attr->cap.max_recv_wr; - qplqp->rq.max_sge = init_attr->cap.max_recv_sge; - if (qplqp->rq.max_sge > dev_attr->max_qp_sges) - qplqp->rq.max_sge = dev_attr->max_qp_sges; + rq->max_wqe = min_t(u32, entries, dev_attr->max_qp_wqes + 1); + rq->q_full_delta = 0; + rq->sg_info.pgsize = PAGE_SIZE; + rq->sg_info.pgshft = PAGE_SHIFT; } - qplqp->rq.sg_info.pgsize = PAGE_SIZE; - qplqp->rq.sg_info.pgshft = PAGE_SHIFT; return 0; } @@ -1067,38 +1138,48 @@ static void bnxt_re_adjust_gsi_rq_attr(struct bnxt_re_qp *qp) qplqp = &qp->qplib_qp; dev_attr = &rdev->dev_attr; - qplqp->rq.max_sge = dev_attr->max_qp_sges; - if (qplqp->rq.max_sge > dev_attr->max_qp_sges) + if (!bnxt_qplib_is_chip_gen_p5(rdev->chip_ctx)) { qplqp->rq.max_sge = dev_attr->max_qp_sges; - qplqp->rq.max_sge = 6; + if (qplqp->rq.max_sge > dev_attr->max_qp_sges) + qplqp->rq.max_sge = dev_attr->max_qp_sges; + qplqp->rq.max_sge = 6; + } } -static void bnxt_re_init_sq_attr(struct bnxt_re_qp *qp, - struct ib_qp_init_attr *init_attr, - struct ib_udata *udata) +static int bnxt_re_init_sq_attr(struct bnxt_re_qp *qp, + struct ib_qp_init_attr *init_attr, + struct ib_udata *udata) { struct bnxt_qplib_dev_attr *dev_attr; struct bnxt_qplib_qp *qplqp; struct bnxt_re_dev *rdev; + struct bnxt_qplib_q *sq; int entries; + int diff; + int rc; rdev = qp->rdev; qplqp = &qp->qplib_qp; + sq = &qplqp->sq; dev_attr = &rdev->dev_attr; - qplqp->sq.max_sge = init_attr->cap.max_send_sge; - if (qplqp->sq.max_sge > dev_attr->max_qp_sges) - qplqp->sq.max_sge = dev_attr->max_qp_sges; - /* - * Change the SQ depth if user has requested minimum using - * configfs. Only supported for kernel consumers - */ + sq->max_sge = init_attr->cap.max_send_sge; + if (sq->max_sge > dev_attr->max_qp_sges) { + sq->max_sge = dev_attr->max_qp_sges; + init_attr->cap.max_send_sge = sq->max_sge; + } + + rc = bnxt_re_setup_swqe_size(qp, init_attr); + if (rc) + return rc; + entries = init_attr->cap.max_send_wr; /* Allocate 128 + 1 more than what's provided */ - entries = roundup_pow_of_two(entries + BNXT_QPLIB_RESERVED_QP_WRS + 1); - qplqp->sq.max_wqe = min_t(u32, entries, dev_attr->max_qp_wqes + - BNXT_QPLIB_RESERVED_QP_WRS + 1); - qplqp->sq.q_full_delta = BNXT_QPLIB_RESERVED_QP_WRS + 1; + diff = (qplqp->wqe_mode == BNXT_QPLIB_WQE_MODE_VARIABLE) ? + 0 : BNXT_QPLIB_RESERVED_QP_WRS; + entries = roundup_pow_of_two(entries + diff + 1); + sq->max_wqe = min_t(u32, entries, dev_attr->max_qp_wqes + diff + 1); + sq->q_full_delta = diff + 1; /* * Reserving one slot for Phantom WQE. Application can * post one extra entry in this case. But allowing this to avoid @@ -1107,6 +1188,8 @@ static void bnxt_re_init_sq_attr(struct bnxt_re_qp *qp, qplqp->sq.q_full_delta -= 1; qplqp->sq.sg_info.pgsize = PAGE_SIZE; qplqp->sq.sg_info.pgshft = PAGE_SHIFT; + + return 0; } static void bnxt_re_adjust_gsi_sq_attr(struct bnxt_re_qp *qp, @@ -1121,13 +1204,16 @@ static void bnxt_re_adjust_gsi_sq_attr(struct bnxt_re_qp *qp, qplqp = &qp->qplib_qp; dev_attr = &rdev->dev_attr; - entries = roundup_pow_of_two(init_attr->cap.max_send_wr + 1); - qplqp->sq.max_wqe = min_t(u32, entries, dev_attr->max_qp_wqes + 1); - qplqp->sq.q_full_delta = qplqp->sq.max_wqe - - init_attr->cap.max_send_wr; - qplqp->sq.max_sge++; /* Need one extra sge to put UD header */ - if (qplqp->sq.max_sge > dev_attr->max_qp_sges) - qplqp->sq.max_sge = dev_attr->max_qp_sges; + if (!bnxt_qplib_is_chip_gen_p5(rdev->chip_ctx)) { + entries = roundup_pow_of_two(init_attr->cap.max_send_wr + 1); + qplqp->sq.max_wqe = min_t(u32, entries, + dev_attr->max_qp_wqes + 1); + qplqp->sq.q_full_delta = qplqp->sq.max_wqe - + init_attr->cap.max_send_wr; + qplqp->sq.max_sge++; /* Need one extra sge to put UD header */ + if (qplqp->sq.max_sge > dev_attr->max_qp_sges) + qplqp->sq.max_sge = dev_attr->max_qp_sges; + } } static int bnxt_re_init_qp_type(struct bnxt_re_dev *rdev, @@ -1179,6 +1265,7 @@ static int bnxt_re_init_qp_attr(struct bnxt_re_qp *qp, struct bnxt_re_pd *pd, goto out; } qplqp->type = (u8)qptype; + qplqp->wqe_mode = rdev->chip_ctx->modes.wqe_mode; if (init_attr->qp_type == IB_QPT_RC) { qplqp->max_rd_atomic = dev_attr->max_qp_rd_atom; @@ -1222,7 +1309,9 @@ static int bnxt_re_init_qp_attr(struct bnxt_re_qp *qp, struct bnxt_re_pd *pd, bnxt_re_adjust_gsi_rq_attr(qp); /* Setup SQ */ - bnxt_re_init_sq_attr(qp, init_attr, udata); + rc = bnxt_re_init_sq_attr(qp, init_attr, udata); + if (rc) + goto out; if (init_attr->qp_type == IB_QPT_GSI) bnxt_re_adjust_gsi_sq_attr(qp, init_attr); @@ -1511,7 +1600,7 @@ static int bnxt_re_init_user_srq(struct bnxt_re_dev *rdev, if (ib_copy_from_udata(&ureq, udata, sizeof(ureq))) return -EFAULT; - bytes = (qplib_srq->max_wqe * BNXT_QPLIB_MAX_RQE_ENTRY_SIZE); + bytes = (qplib_srq->max_wqe * qplib_srq->wqe_size); bytes = PAGE_ALIGN(bytes); umem = ib_umem_get(&rdev->ibdev, ureq.srqva, bytes, IB_ACCESS_LOCAL_WRITE); @@ -1534,15 +1623,20 @@ int bnxt_re_create_srq(struct ib_srq *ib_srq, struct ib_srq_init_attr *srq_init_attr, struct ib_udata *udata) { - struct ib_pd *ib_pd = ib_srq->pd; - struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd); - struct bnxt_re_dev *rdev = pd->rdev; - struct bnxt_qplib_dev_attr *dev_attr = &rdev->dev_attr; - struct bnxt_re_srq *srq = - container_of(ib_srq, struct bnxt_re_srq, ib_srq); + struct bnxt_qplib_dev_attr *dev_attr; struct bnxt_qplib_nq *nq = NULL; + struct bnxt_re_dev *rdev; + struct bnxt_re_srq *srq; + struct bnxt_re_pd *pd; + struct ib_pd *ib_pd; int rc, entries; + ib_pd = ib_srq->pd; + pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd); + rdev = pd->rdev; + dev_attr = &rdev->dev_attr; + srq = container_of(ib_srq, struct bnxt_re_srq, ib_srq); + if (srq_init_attr->attr.max_wr >= dev_attr->max_srq_wqes) { ibdev_err(&rdev->ibdev, "Create CQ failed - max exceeded"); rc = -EINVAL; @@ -1563,9 +1657,11 @@ int bnxt_re_create_srq(struct ib_srq *ib_srq, entries = roundup_pow_of_two(srq_init_attr->attr.max_wr + 1); if (entries > dev_attr->max_srq_wqes + 1) entries = dev_attr->max_srq_wqes + 1; - srq->qplib_srq.max_wqe = entries; + srq->qplib_srq.max_sge = srq_init_attr->attr.max_sge; + srq->qplib_srq.wqe_size = + bnxt_re_get_rwqe_size(srq->qplib_srq.max_sge); srq->qplib_srq.threshold = srq_init_attr->attr.srq_limit; srq->srq_limit = srq_init_attr->attr.srq_limit; srq->qplib_srq.eventq_hw_ring_id = rdev->nq[0].ring_id; @@ -2561,7 +2657,7 @@ int bnxt_re_post_send(struct ib_qp *ib_qp, const struct ib_send_wr *wr, default: break; } - /* fall through */ + fallthrough; case IB_WR_SEND_WITH_INV: rc = bnxt_re_build_send_wqe(qp, wr, &wqe); break; @@ -3168,6 +3264,19 @@ static void bnxt_re_process_res_rawqp1_wc(struct ib_wc *wc, wc->wc_flags |= IB_WC_GRH; } +static bool bnxt_re_check_if_vlan_valid(struct bnxt_re_dev *rdev, + u16 vlan_id) +{ + /* + * Check if the vlan is configured in the host. If not configured, it + * can be a transparent VLAN. So dont report the vlan id. + */ + if (!__vlan_find_dev_deep_rcu(rdev->netdev, + htons(ETH_P_8021Q), vlan_id)) + return false; + return true; +} + static bool bnxt_re_is_vlan_pkt(struct bnxt_qplib_cqe *orig_cqe, u16 *vid, u8 *sl) { @@ -3236,9 +3345,11 @@ static void bnxt_re_process_res_shadow_qp_wc(struct bnxt_re_qp *gsi_sqp, wc->src_qp = orig_cqe->src_qp; memcpy(wc->smac, orig_cqe->smac, ETH_ALEN); if (bnxt_re_is_vlan_pkt(orig_cqe, &vlan_id, &sl)) { - wc->vlan_id = vlan_id; - wc->sl = sl; - wc->wc_flags |= IB_WC_WITH_VLAN; + if (bnxt_re_check_if_vlan_valid(rdev, vlan_id)) { + wc->vlan_id = vlan_id; + wc->sl = sl; + wc->wc_flags |= IB_WC_WITH_VLAN; + } } wc->port_num = 1; wc->vendor_err = orig_cqe->status; @@ -3559,7 +3670,7 @@ int bnxt_re_map_mr_sg(struct ib_mr *ib_mr, struct scatterlist *sg, int sg_nents, } struct ib_mr *bnxt_re_alloc_mr(struct ib_pd *ib_pd, enum ib_mr_type type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd); struct bnxt_re_dev *rdev = pd->rdev; diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.h b/drivers/infiniband/hw/bnxt_re/ib_verbs.h index 23d972da5652..1daeb30e06fd 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.h +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.h @@ -122,12 +122,6 @@ struct bnxt_re_frpl { u64 *page_list; }; -struct bnxt_re_fmr { - struct bnxt_re_dev *rdev; - struct ib_fmr ib_fmr; - struct bnxt_qplib_mrw qplib_fmr; -}; - struct bnxt_re_mw { struct bnxt_re_dev *rdev; struct ib_mw ib_mw; @@ -142,6 +136,16 @@ struct bnxt_re_ucontext { spinlock_t sh_lock; /* protect shpg */ }; +static inline u16 bnxt_re_get_swqe_size(int nsge) +{ + return sizeof(struct sq_send_hdr) + nsge * sizeof(struct sq_sge); +} + +static inline u16 bnxt_re_get_rwqe_size(int nsge) +{ + return sizeof(struct rq_wqe_hdr) + (nsge * sizeof(struct sq_sge)); +} + int bnxt_re_query_device(struct ib_device *ibdev, struct ib_device_attr *ib_attr, struct ib_udata *udata); @@ -160,7 +164,7 @@ enum rdma_link_layer bnxt_re_get_link_layer(struct ib_device *ibdev, u8 port_num); int bnxt_re_alloc_pd(struct ib_pd *pd, struct ib_udata *udata); void bnxt_re_dealloc_pd(struct ib_pd *pd, struct ib_udata *udata); -int bnxt_re_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, u32 flags, +int bnxt_re_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); int bnxt_re_modify_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr); int bnxt_re_query_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr); @@ -197,7 +201,7 @@ struct ib_mr *bnxt_re_get_dma_mr(struct ib_pd *pd, int mr_access_flags); int bnxt_re_map_mr_sg(struct ib_mr *ib_mr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); struct ib_mr *bnxt_re_alloc_mr(struct ib_pd *ib_pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int bnxt_re_dereg_mr(struct ib_mr *mr, struct ib_udata *udata); struct ib_mw *bnxt_re_alloc_mw(struct ib_pd *ib_pd, enum ib_mw_type type, struct ib_udata *udata); diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c index b12fbc857f94..53aee5a42ab8 100644 --- a/drivers/infiniband/hw/bnxt_re/main.c +++ b/drivers/infiniband/hw/bnxt_re/main.c @@ -82,6 +82,15 @@ static void bnxt_re_remove_device(struct bnxt_re_dev *rdev); static void bnxt_re_dealloc_driver(struct ib_device *ib_dev); static void bnxt_re_stop_irq(void *handle); +static void bnxt_re_set_drv_mode(struct bnxt_re_dev *rdev, u8 mode) +{ + struct bnxt_qplib_chip_ctx *cctx; + + cctx = rdev->chip_ctx; + cctx->modes.wqe_mode = bnxt_qplib_is_chip_gen_p5(rdev->chip_ctx) ? + mode : BNXT_QPLIB_WQE_MODE_STATIC; +} + static void bnxt_re_destroy_chip_ctx(struct bnxt_re_dev *rdev) { struct bnxt_qplib_chip_ctx *chip_ctx; @@ -97,7 +106,7 @@ static void bnxt_re_destroy_chip_ctx(struct bnxt_re_dev *rdev) kfree(chip_ctx); } -static int bnxt_re_setup_chip_ctx(struct bnxt_re_dev *rdev) +static int bnxt_re_setup_chip_ctx(struct bnxt_re_dev *rdev, u8 wqe_mode) { struct bnxt_qplib_chip_ctx *chip_ctx; struct bnxt_en_dev *en_dev; @@ -117,6 +126,7 @@ static int bnxt_re_setup_chip_ctx(struct bnxt_re_dev *rdev) rdev->qplib_res.cctx = rdev->chip_ctx; rdev->rcfw.res = &rdev->qplib_res; + bnxt_re_set_drv_mode(rdev, wqe_mode); return 0; } @@ -811,7 +821,8 @@ static int bnxt_re_handle_qp_async_event(struct creq_qp_event *qp_event, struct ib_event event; unsigned int flags; - if (qp->qplib_qp.state == CMDQ_MODIFY_QP_NEW_STATE_ERR) { + if (qp->qplib_qp.state == CMDQ_MODIFY_QP_NEW_STATE_ERR && + rdma_is_kernel_res(&qp->ib_qp.res)) { flags = bnxt_re_lock_cqs(qp); bnxt_qplib_add_flush_qp(&qp->qplib_qp); bnxt_re_unlock_cqs(qp, flags); @@ -998,7 +1009,6 @@ static void bnxt_re_free_res(struct bnxt_re_dev *rdev) static int bnxt_re_alloc_res(struct bnxt_re_dev *rdev) { struct bnxt_re_ring_attr rattr = {}; - struct bnxt_qplib_ctx *qplib_ctx; int num_vec_created = 0; int rc = 0, i; u8 type; @@ -1021,13 +1031,11 @@ static int bnxt_re_alloc_res(struct bnxt_re_dev *rdev) if (rc) goto dealloc_res; - qplib_ctx = &rdev->qplib_ctx; for (i = 0; i < rdev->num_msix - 1; i++) { struct bnxt_qplib_nq *nq; nq = &rdev->nq[i]; - nq->hwq.max_elements = (qplib_ctx->cq_count + - qplib_ctx->srqc_count + 2); + nq->hwq.max_elements = BNXT_QPLIB_NQE_MAX_CNT; rc = bnxt_qplib_alloc_nq(&rdev->qplib_res, &rdev->nq[i]); if (rc) { ibdev_err(&rdev->ibdev, "Alloc Failed NQ%d rc:%#x", @@ -1386,7 +1394,7 @@ static void bnxt_re_worker(struct work_struct *work) schedule_delayed_work(&rdev->worker, msecs_to_jiffies(30000)); } -static int bnxt_re_dev_init(struct bnxt_re_dev *rdev) +static int bnxt_re_dev_init(struct bnxt_re_dev *rdev, u8 wqe_mode) { struct bnxt_qplib_creq_ctx *creq; struct bnxt_re_ring_attr rattr; @@ -1406,7 +1414,7 @@ static int bnxt_re_dev_init(struct bnxt_re_dev *rdev) } set_bit(BNXT_RE_FLAG_NETDEV_REGISTERED, &rdev->flags); - rc = bnxt_re_setup_chip_ctx(rdev); + rc = bnxt_re_setup_chip_ctx(rdev, wqe_mode); if (rc) { ibdev_err(&rdev->ibdev, "Failed to get chip context\n"); return -EINVAL; @@ -1585,7 +1593,7 @@ static void bnxt_re_remove_device(struct bnxt_re_dev *rdev) } static int bnxt_re_add_device(struct bnxt_re_dev **rdev, - struct net_device *netdev) + struct net_device *netdev, u8 wqe_mode) { int rc; @@ -1599,7 +1607,7 @@ static int bnxt_re_add_device(struct bnxt_re_dev **rdev, } pci_dev_get((*rdev)->en_dev->pdev); - rc = bnxt_re_dev_init(*rdev); + rc = bnxt_re_dev_init(*rdev, wqe_mode); if (rc) { pci_dev_put((*rdev)->en_dev->pdev); bnxt_re_dev_unreg(*rdev); @@ -1711,7 +1719,8 @@ static int bnxt_re_netdev_event(struct notifier_block *notifier, case NETDEV_REGISTER: if (rdev) break; - rc = bnxt_re_add_device(&rdev, real_dev); + rc = bnxt_re_add_device(&rdev, real_dev, + BNXT_QPLIB_WQE_MODE_STATIC); if (!rc) sch_work = true; release = false; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c index 899a5d2c100e..f78da54a0bc5 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c @@ -178,11 +178,11 @@ static void bnxt_qplib_free_qp_hdr_buf(struct bnxt_qplib_res *res, if (qp->rq_hdr_buf) dma_free_coherent(&res->pdev->dev, - rq->hwq.max_elements * qp->rq_hdr_buf_size, + rq->max_wqe * qp->rq_hdr_buf_size, qp->rq_hdr_buf, qp->rq_hdr_buf_map); if (qp->sq_hdr_buf) dma_free_coherent(&res->pdev->dev, - sq->hwq.max_elements * qp->sq_hdr_buf_size, + sq->max_wqe * qp->sq_hdr_buf_size, qp->sq_hdr_buf, qp->sq_hdr_buf_map); qp->rq_hdr_buf = NULL; qp->sq_hdr_buf = NULL; @@ -199,10 +199,9 @@ static int bnxt_qplib_alloc_qp_hdr_buf(struct bnxt_qplib_res *res, struct bnxt_qplib_q *sq = &qp->sq; int rc = 0; - if (qp->sq_hdr_buf_size && sq->hwq.max_elements) { + if (qp->sq_hdr_buf_size && sq->max_wqe) { qp->sq_hdr_buf = dma_alloc_coherent(&res->pdev->dev, - sq->hwq.max_elements * - qp->sq_hdr_buf_size, + sq->max_wqe * qp->sq_hdr_buf_size, &qp->sq_hdr_buf_map, GFP_KERNEL); if (!qp->sq_hdr_buf) { rc = -ENOMEM; @@ -212,9 +211,9 @@ static int bnxt_qplib_alloc_qp_hdr_buf(struct bnxt_qplib_res *res, } } - if (qp->rq_hdr_buf_size && rq->hwq.max_elements) { + if (qp->rq_hdr_buf_size && rq->max_wqe) { qp->rq_hdr_buf = dma_alloc_coherent(&res->pdev->dev, - rq->hwq.max_elements * + rq->max_wqe * qp->rq_hdr_buf_size, &qp->rq_hdr_buf_map, GFP_KERNEL); @@ -300,12 +299,12 @@ static void bnxt_qplib_service_nq(unsigned long data) { struct bnxt_qplib_nq *nq = (struct bnxt_qplib_nq *)data; struct bnxt_qplib_hwq *hwq = &nq->hwq; - struct nq_base *nqe, **nq_ptr; - struct bnxt_qplib_cq *cq; - int num_cqne_processed = 0; int num_srqne_processed = 0; + int num_cqne_processed = 0; + struct bnxt_qplib_cq *cq; int budget = nq->budget; u32 sw_cons, raw_cons; + struct nq_base *nqe; uintptr_t q_handle; u16 type; @@ -314,8 +313,7 @@ static void bnxt_qplib_service_nq(unsigned long data) raw_cons = hwq->cons; while (budget--) { sw_cons = HWQ_CMP(raw_cons, hwq); - nq_ptr = (struct nq_base **)hwq->pbl_ptr; - nqe = &nq_ptr[NQE_PG(sw_cons)][NQE_IDX(sw_cons)]; + nqe = bnxt_qplib_get_qe(hwq, sw_cons, NULL); if (!NQE_CMP_VALID(nqe, raw_cons, hwq->max_elements)) break; @@ -392,13 +390,11 @@ static irqreturn_t bnxt_qplib_nq_irq(int irq, void *dev_instance) { struct bnxt_qplib_nq *nq = dev_instance; struct bnxt_qplib_hwq *hwq = &nq->hwq; - struct nq_base **nq_ptr; u32 sw_cons; /* Prefetch the NQ element */ sw_cons = HWQ_CMP(hwq->cons, hwq); - nq_ptr = (struct nq_base **)nq->hwq.pbl_ptr; - prefetch(&nq_ptr[NQE_PG(sw_cons)][NQE_IDX(sw_cons)]); + prefetch(bnxt_qplib_get_qe(hwq, sw_cons, NULL)); /* Fan out to CPU affinitized kthreads? */ tasklet_schedule(&nq->nq_tasklet); @@ -612,12 +608,13 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res, struct cmdq_create_srq req; struct bnxt_qplib_pbl *pbl; u16 cmd_flags = 0; + u16 pg_sz_lvl; int rc, idx; hwq_attr.res = res; hwq_attr.sginfo = &srq->sg_info; hwq_attr.depth = srq->max_wqe; - hwq_attr.stride = BNXT_QPLIB_MAX_RQE_ENTRY_SIZE; + hwq_attr.stride = srq->wqe_size; hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&srq->hwq, &hwq_attr); if (rc) @@ -638,22 +635,11 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res, req.srq_size = cpu_to_le16((u16)srq->hwq.max_elements); pbl = &srq->hwq.pbl[PBL_LVL_0]; - req.pg_size_lvl = cpu_to_le16((((u16)srq->hwq.level & - CMDQ_CREATE_SRQ_LVL_MASK) << - CMDQ_CREATE_SRQ_LVL_SFT) | - (pbl->pg_size == ROCE_PG_SIZE_4K ? - CMDQ_CREATE_SRQ_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? - CMDQ_CREATE_SRQ_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? - CMDQ_CREATE_SRQ_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? - CMDQ_CREATE_SRQ_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? - CMDQ_CREATE_SRQ_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? - CMDQ_CREATE_SRQ_PG_SIZE_PG_1G : - CMDQ_CREATE_SRQ_PG_SIZE_PG_4K)); + pg_sz_lvl = ((u16)bnxt_qplib_base_pg_size(&srq->hwq) << + CMDQ_CREATE_SRQ_PG_SIZE_SFT); + pg_sz_lvl |= (srq->hwq.level & CMDQ_CREATE_SRQ_LVL_MASK) << + CMDQ_CREATE_SRQ_LVL_SFT; + req.pg_size_lvl = cpu_to_le16(pg_sz_lvl); req.pbl = cpu_to_le64(pbl->pg_map_arr[0]); req.pd_id = cpu_to_le32(srq->pd->id); req.eventq_id = cpu_to_le16(srq->eventq_hw_ring_id); @@ -674,6 +660,7 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res, srq->dbinfo.hwq = &srq->hwq; srq->dbinfo.xid = srq->id; srq->dbinfo.db = srq->dpi->dbr; + srq->dbinfo.max_slot = 1; srq->dbinfo.priv_db = res->dpi_tbl.dbr_bar_reg_iomem; if (srq->threshold) bnxt_qplib_armen_db(&srq->dbinfo, DBC_DBC_TYPE_SRQ_ARMENA); @@ -740,7 +727,7 @@ int bnxt_qplib_post_srq_recv(struct bnxt_qplib_srq *srq, struct bnxt_qplib_swqe *wqe) { struct bnxt_qplib_hwq *srq_hwq = &srq->hwq; - struct rq_wqe *srqe, **srqe_ptr; + struct rq_wqe *srqe; struct sq_sge *hw_sge; u32 sw_prod, sw_cons, count = 0; int i, rc = 0, next; @@ -758,9 +745,8 @@ int bnxt_qplib_post_srq_recv(struct bnxt_qplib_srq *srq, spin_unlock(&srq_hwq->lock); sw_prod = HWQ_CMP(srq_hwq->prod, srq_hwq); - srqe_ptr = (struct rq_wqe **)srq_hwq->pbl_ptr; - srqe = &srqe_ptr[RQE_PG(sw_prod)][RQE_IDX(sw_prod)]; - memset(srqe, 0, BNXT_QPLIB_MAX_RQE_ENTRY_SIZE); + srqe = bnxt_qplib_get_qe(srq_hwq, sw_prod, NULL); + memset(srqe, 0, srq->wqe_size); /* Calculate wqe_size16 and data_len */ for (i = 0, hw_sge = (struct sq_sge *)srqe->data; i < wqe->num_sge; i++, hw_sge++) { @@ -798,6 +784,28 @@ done: } /* QP */ + +static int bnxt_qplib_alloc_init_swq(struct bnxt_qplib_q *que) +{ + int rc = 0; + int indx; + + que->swq = kcalloc(que->max_wqe, sizeof(*que->swq), GFP_KERNEL); + if (!que->swq) { + rc = -ENOMEM; + goto out; + } + + que->swq_start = 0; + que->swq_last = que->max_wqe - 1; + for (indx = 0; indx < que->max_wqe; indx++) + que->swq[indx].next_idx = indx + 1; + que->swq[que->swq_last].next_idx = 0; /* Make it circular */ + que->swq_last = 0; +out: + return rc; +} + int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) { struct bnxt_qplib_hwq_attr hwq_attr = {}; @@ -809,6 +817,8 @@ int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) struct bnxt_qplib_pbl *pbl; u16 cmd_flags = 0; u32 qp_flags = 0; + u8 pg_sz_lvl; + u32 tbl_indx; int rc; RCFW_CMD_PREP(req, CREATE_QP1, cmd_flags); @@ -821,98 +831,63 @@ int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) /* SQ */ hwq_attr.res = res; hwq_attr.sginfo = &sq->sg_info; - hwq_attr.depth = sq->max_wqe; - hwq_attr.stride = BNXT_QPLIB_MAX_SQE_ENTRY_SIZE; + hwq_attr.stride = sizeof(struct sq_sge); + hwq_attr.depth = bnxt_qplib_get_depth(sq); hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&sq->hwq, &hwq_attr); if (rc) goto exit; - sq->swq = kcalloc(sq->hwq.max_elements, sizeof(*sq->swq), GFP_KERNEL); - if (!sq->swq) { - rc = -ENOMEM; + rc = bnxt_qplib_alloc_init_swq(sq); + if (rc) goto fail_sq; - } + + req.sq_size = cpu_to_le32(bnxt_qplib_set_sq_size(sq, qp->wqe_mode)); pbl = &sq->hwq.pbl[PBL_LVL_0]; req.sq_pbl = cpu_to_le64(pbl->pg_map_arr[0]); - req.sq_pg_size_sq_lvl = - ((sq->hwq.level & CMDQ_CREATE_QP1_SQ_LVL_MASK) - << CMDQ_CREATE_QP1_SQ_LVL_SFT) | - (pbl->pg_size == ROCE_PG_SIZE_4K ? - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_1G : - CMDQ_CREATE_QP1_SQ_PG_SIZE_PG_4K); - - if (qp->scq) - req.scq_cid = cpu_to_le32(qp->scq->id); - - qp_flags |= CMDQ_CREATE_QP1_QP_FLAGS_RESERVED_LKEY_ENABLE; + pg_sz_lvl = (bnxt_qplib_base_pg_size(&sq->hwq) << + CMDQ_CREATE_QP1_SQ_PG_SIZE_SFT); + pg_sz_lvl |= (sq->hwq.level & CMDQ_CREATE_QP1_SQ_LVL_MASK); + req.sq_pg_size_sq_lvl = pg_sz_lvl; + req.sq_fwo_sq_sge = + cpu_to_le16((sq->max_sge & CMDQ_CREATE_QP1_SQ_SGE_MASK) << + CMDQ_CREATE_QP1_SQ_SGE_SFT); + req.scq_cid = cpu_to_le32(qp->scq->id); /* RQ */ if (rq->max_wqe) { hwq_attr.res = res; hwq_attr.sginfo = &rq->sg_info; - hwq_attr.stride = BNXT_QPLIB_MAX_RQE_ENTRY_SIZE; - hwq_attr.depth = qp->rq.max_wqe; + hwq_attr.stride = sizeof(struct sq_sge); + hwq_attr.depth = bnxt_qplib_get_depth(rq); hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&rq->hwq, &hwq_attr); if (rc) - goto fail_sq; - - rq->swq = kcalloc(rq->hwq.max_elements, sizeof(*rq->swq), - GFP_KERNEL); - if (!rq->swq) { - rc = -ENOMEM; + goto sq_swq; + rc = bnxt_qplib_alloc_init_swq(rq); + if (rc) goto fail_rq; - } + req.rq_size = cpu_to_le32(rq->max_wqe); pbl = &rq->hwq.pbl[PBL_LVL_0]; req.rq_pbl = cpu_to_le64(pbl->pg_map_arr[0]); - req.rq_pg_size_rq_lvl = - ((rq->hwq.level & CMDQ_CREATE_QP1_RQ_LVL_MASK) << - CMDQ_CREATE_QP1_RQ_LVL_SFT) | - (pbl->pg_size == ROCE_PG_SIZE_4K ? - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_1G : - CMDQ_CREATE_QP1_RQ_PG_SIZE_PG_4K); - if (qp->rcq) - req.rcq_cid = cpu_to_le32(qp->rcq->id); - } - + pg_sz_lvl = (bnxt_qplib_base_pg_size(&rq->hwq) << + CMDQ_CREATE_QP1_RQ_PG_SIZE_SFT); + pg_sz_lvl |= (rq->hwq.level & CMDQ_CREATE_QP1_RQ_LVL_MASK); + req.rq_pg_size_rq_lvl = pg_sz_lvl; + req.rq_fwo_rq_sge = + cpu_to_le16((rq->max_sge & + CMDQ_CREATE_QP1_RQ_SGE_MASK) << + CMDQ_CREATE_QP1_RQ_SGE_SFT); + } + req.rcq_cid = cpu_to_le32(qp->rcq->id); /* Header buffer - allow hdr_buf pass in */ rc = bnxt_qplib_alloc_qp_hdr_buf(res, qp); if (rc) { rc = -ENOMEM; - goto fail; + goto rq_rwq; } + qp_flags |= CMDQ_CREATE_QP1_QP_FLAGS_RESERVED_LKEY_ENABLE; req.qp_flags = cpu_to_le32(qp_flags); - req.sq_size = cpu_to_le32(sq->hwq.max_elements); - req.rq_size = cpu_to_le32(rq->hwq.max_elements); - - req.sq_fwo_sq_sge = - cpu_to_le16((sq->max_sge & CMDQ_CREATE_QP1_SQ_SGE_MASK) << - CMDQ_CREATE_QP1_SQ_SGE_SFT); - req.rq_fwo_rq_sge = - cpu_to_le16((rq->max_sge & CMDQ_CREATE_QP1_RQ_SGE_MASK) << - CMDQ_CREATE_QP1_RQ_SGE_SFT); - req.pd_id = cpu_to_le32(qp->pd->id); rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, @@ -926,46 +901,68 @@ int bnxt_qplib_create_qp1(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) sq->dbinfo.hwq = &sq->hwq; sq->dbinfo.xid = qp->id; sq->dbinfo.db = qp->dpi->dbr; + sq->dbinfo.max_slot = bnxt_qplib_set_sq_max_slot(qp->wqe_mode); if (rq->max_wqe) { rq->dbinfo.hwq = &rq->hwq; rq->dbinfo.xid = qp->id; rq->dbinfo.db = qp->dpi->dbr; + rq->dbinfo.max_slot = bnxt_qplib_set_rq_max_slot(rq->wqe_size); } - rcfw->qp_tbl[qp->id].qp_id = qp->id; - rcfw->qp_tbl[qp->id].qp_handle = (void *)qp; + tbl_indx = map_qp_id_to_tbl_indx(qp->id, rcfw); + rcfw->qp_tbl[tbl_indx].qp_id = qp->id; + rcfw->qp_tbl[tbl_indx].qp_handle = (void *)qp; return 0; fail: bnxt_qplib_free_qp_hdr_buf(res, qp); +rq_rwq: + kfree(rq->swq); fail_rq: bnxt_qplib_free_hwq(res, &rq->hwq); - kfree(rq->swq); +sq_swq: + kfree(sq->swq); fail_sq: bnxt_qplib_free_hwq(res, &sq->hwq); - kfree(sq->swq); exit: return rc; } +static void bnxt_qplib_init_psn_ptr(struct bnxt_qplib_qp *qp, int size) +{ + struct bnxt_qplib_hwq *hwq; + struct bnxt_qplib_q *sq; + u64 fpsne, psn_pg; + u16 indx_pad = 0; + + sq = &qp->sq; + hwq = &sq->hwq; + /* First psn entry */ + fpsne = (u64)bnxt_qplib_get_qe(hwq, hwq->depth, &psn_pg); + if (!IS_ALIGNED(fpsne, PAGE_SIZE)) + indx_pad = (fpsne & ~PAGE_MASK) / size; + hwq->pad_pgofft = indx_pad; + hwq->pad_pg = (u64 *)psn_pg; + hwq->pad_stride = size; +} + int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) { struct bnxt_qplib_rcfw *rcfw = res->rcfw; struct bnxt_qplib_hwq_attr hwq_attr = {}; - unsigned long int psn_search, poff = 0; struct bnxt_qplib_sg_info sginfo = {}; - struct sq_psn_search **psn_search_ptr; struct bnxt_qplib_q *sq = &qp->sq; struct bnxt_qplib_q *rq = &qp->rq; - int i, rc, req_size, psn_sz = 0; - struct sq_send **hw_sq_send_ptr; struct creq_create_qp_resp resp; + int rc, req_size, psn_sz = 0; struct bnxt_qplib_hwq *xrrq; - u16 cmd_flags = 0, max_ssge; - struct cmdq_create_qp req; struct bnxt_qplib_pbl *pbl; + struct cmdq_create_qp req; + u16 cmd_flags = 0; u32 qp_flags = 0; - u16 max_rsge; + u8 pg_sz_lvl; + u32 tbl_indx; + u16 nsge; RCFW_CMD_PREP(req, CREATE_QP, cmd_flags); @@ -983,142 +980,78 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) hwq_attr.res = res; hwq_attr.sginfo = &sq->sg_info; - hwq_attr.stride = BNXT_QPLIB_MAX_SQE_ENTRY_SIZE; - hwq_attr.depth = sq->max_wqe; + hwq_attr.stride = sizeof(struct sq_sge); + hwq_attr.depth = bnxt_qplib_get_depth(sq); hwq_attr.aux_stride = psn_sz; - hwq_attr.aux_depth = hwq_attr.depth; + hwq_attr.aux_depth = bnxt_qplib_set_sq_size(sq, qp->wqe_mode); hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&sq->hwq, &hwq_attr); if (rc) goto exit; - sq->swq = kcalloc(sq->hwq.max_elements, sizeof(*sq->swq), GFP_KERNEL); - if (!sq->swq) { - rc = -ENOMEM; + rc = bnxt_qplib_alloc_init_swq(sq); + if (rc) goto fail_sq; - } - hw_sq_send_ptr = (struct sq_send **)sq->hwq.pbl_ptr; - if (psn_sz) { - psn_search_ptr = (struct sq_psn_search **) - &hw_sq_send_ptr[get_sqe_pg - (sq->hwq.max_elements)]; - psn_search = (unsigned long int) - &hw_sq_send_ptr[get_sqe_pg(sq->hwq.max_elements)] - [get_sqe_idx(sq->hwq.max_elements)]; - if (psn_search & ~PAGE_MASK) { - /* If the psn_search does not start on a page boundary, - * then calculate the offset - */ - poff = (psn_search & ~PAGE_MASK) / - BNXT_QPLIB_MAX_PSNE_ENTRY_SIZE; - } - for (i = 0; i < sq->hwq.max_elements; i++) { - sq->swq[i].psn_search = - &psn_search_ptr[get_psne_pg(i + poff)] - [get_psne_idx(i + poff)]; - /*psns_ext will be used only for P5 chips. */ - sq->swq[i].psn_ext = - (struct sq_psn_search_ext *) - &psn_search_ptr[get_psne_pg(i + poff)] - [get_psne_idx(i + poff)]; - } - } + + if (psn_sz) + bnxt_qplib_init_psn_ptr(qp, psn_sz); + + req.sq_size = cpu_to_le32(bnxt_qplib_set_sq_size(sq, qp->wqe_mode)); pbl = &sq->hwq.pbl[PBL_LVL_0]; req.sq_pbl = cpu_to_le64(pbl->pg_map_arr[0]); - req.sq_pg_size_sq_lvl = - ((sq->hwq.level & CMDQ_CREATE_QP_SQ_LVL_MASK) - << CMDQ_CREATE_QP_SQ_LVL_SFT) | - (pbl->pg_size == ROCE_PG_SIZE_4K ? - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_1G : - CMDQ_CREATE_QP_SQ_PG_SIZE_PG_4K); - - if (qp->scq) - req.scq_cid = cpu_to_le32(qp->scq->id); - - qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_RESERVED_LKEY_ENABLE; - qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_FR_PMR_ENABLED; - if (qp->sig_type) - qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_FORCE_COMPLETION; + pg_sz_lvl = (bnxt_qplib_base_pg_size(&sq->hwq) << + CMDQ_CREATE_QP_SQ_PG_SIZE_SFT); + pg_sz_lvl |= (sq->hwq.level & CMDQ_CREATE_QP_SQ_LVL_MASK); + req.sq_pg_size_sq_lvl = pg_sz_lvl; + req.sq_fwo_sq_sge = + cpu_to_le16(((sq->max_sge & CMDQ_CREATE_QP_SQ_SGE_MASK) << + CMDQ_CREATE_QP_SQ_SGE_SFT) | 0); + req.scq_cid = cpu_to_le32(qp->scq->id); /* RQ */ - if (rq->max_wqe) { + if (!qp->srq) { hwq_attr.res = res; hwq_attr.sginfo = &rq->sg_info; - hwq_attr.stride = BNXT_QPLIB_MAX_RQE_ENTRY_SIZE; - hwq_attr.depth = rq->max_wqe; + hwq_attr.stride = sizeof(struct sq_sge); + hwq_attr.depth = bnxt_qplib_get_depth(rq); hwq_attr.aux_stride = 0; hwq_attr.aux_depth = 0; hwq_attr.type = HWQ_TYPE_QUEUE; rc = bnxt_qplib_alloc_init_hwq(&rq->hwq, &hwq_attr); if (rc) - goto fail_sq; - - rq->swq = kcalloc(rq->hwq.max_elements, sizeof(*rq->swq), - GFP_KERNEL); - if (!rq->swq) { - rc = -ENOMEM; + goto sq_swq; + rc = bnxt_qplib_alloc_init_swq(rq); + if (rc) goto fail_rq; - } + + req.rq_size = cpu_to_le32(rq->max_wqe); pbl = &rq->hwq.pbl[PBL_LVL_0]; req.rq_pbl = cpu_to_le64(pbl->pg_map_arr[0]); - req.rq_pg_size_rq_lvl = - ((rq->hwq.level & CMDQ_CREATE_QP_RQ_LVL_MASK) << - CMDQ_CREATE_QP_RQ_LVL_SFT) | - (pbl->pg_size == ROCE_PG_SIZE_4K ? - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_1G : - CMDQ_CREATE_QP_RQ_PG_SIZE_PG_4K); + pg_sz_lvl = (bnxt_qplib_base_pg_size(&rq->hwq) << + CMDQ_CREATE_QP_RQ_PG_SIZE_SFT); + pg_sz_lvl |= (rq->hwq.level & CMDQ_CREATE_QP_RQ_LVL_MASK); + req.rq_pg_size_rq_lvl = pg_sz_lvl; + nsge = (qp->wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) ? + 6 : rq->max_sge; + req.rq_fwo_rq_sge = + cpu_to_le16(((nsge & + CMDQ_CREATE_QP_RQ_SGE_MASK) << + CMDQ_CREATE_QP_RQ_SGE_SFT) | 0); } else { /* SRQ */ - if (qp->srq) { - qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_SRQ_USED; - req.srq_cid = cpu_to_le32(qp->srq->id); - } + qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_SRQ_USED; + req.srq_cid = cpu_to_le32(qp->srq->id); } + req.rcq_cid = cpu_to_le32(qp->rcq->id); - if (qp->rcq) - req.rcq_cid = cpu_to_le32(qp->rcq->id); + qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_RESERVED_LKEY_ENABLE; + qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_FR_PMR_ENABLED; + if (qp->sig_type) + qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_FORCE_COMPLETION; + if (qp->wqe_mode == BNXT_QPLIB_WQE_MODE_VARIABLE) + qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_VARIABLE_SIZED_WQE_ENABLED; req.qp_flags = cpu_to_le32(qp_flags); - req.sq_size = cpu_to_le32(sq->hwq.max_elements); - req.rq_size = cpu_to_le32(rq->hwq.max_elements); - qp->sq_hdr_buf = NULL; - qp->rq_hdr_buf = NULL; - rc = bnxt_qplib_alloc_qp_hdr_buf(res, qp); - if (rc) - goto fail_rq; - - /* CTRL-22434: Irrespective of the requested SGE count on the SQ - * always create the QP with max send sges possible if the requested - * inline size is greater than 0. - */ - max_ssge = qp->max_inline_data ? 6 : sq->max_sge; - req.sq_fwo_sq_sge = cpu_to_le16( - ((max_ssge & CMDQ_CREATE_QP_SQ_SGE_MASK) - << CMDQ_CREATE_QP_SQ_SGE_SFT) | 0); - max_rsge = bnxt_qplib_is_chip_gen_p5(res->cctx) ? 6 : rq->max_sge; - req.rq_fwo_rq_sge = cpu_to_le16( - ((max_rsge & CMDQ_CREATE_QP_RQ_SGE_MASK) - << CMDQ_CREATE_QP_RQ_SGE_SFT) | 0); /* ORRQ and IRRQ */ if (psn_sz) { xrrq = &qp->orrq; @@ -1139,7 +1072,7 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) hwq_attr.type = HWQ_TYPE_CTX; rc = bnxt_qplib_alloc_init_hwq(xrrq, &hwq_attr); if (rc) - goto fail_buf_free; + goto rq_swq; pbl = &xrrq->pbl[PBL_LVL_0]; req.orrq_addr = cpu_to_le64(pbl->pg_map_arr[0]); @@ -1174,30 +1107,30 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp) sq->dbinfo.hwq = &sq->hwq; sq->dbinfo.xid = qp->id; sq->dbinfo.db = qp->dpi->dbr; + sq->dbinfo.max_slot = bnxt_qplib_set_sq_max_slot(qp->wqe_mode); if (rq->max_wqe) { rq->dbinfo.hwq = &rq->hwq; rq->dbinfo.xid = qp->id; rq->dbinfo.db = qp->dpi->dbr; + rq->dbinfo.max_slot = bnxt_qplib_set_rq_max_slot(rq->wqe_size); } - rcfw->qp_tbl[qp->id].qp_id = qp->id; - rcfw->qp_tbl[qp->id].qp_handle = (void *)qp; + tbl_indx = map_qp_id_to_tbl_indx(qp->id, rcfw); + rcfw->qp_tbl[tbl_indx].qp_id = qp->id; + rcfw->qp_tbl[tbl_indx].qp_handle = (void *)qp; return 0; - fail: - if (qp->irrq.max_elements) - bnxt_qplib_free_hwq(res, &qp->irrq); + bnxt_qplib_free_hwq(res, &qp->irrq); fail_orrq: - if (qp->orrq.max_elements) - bnxt_qplib_free_hwq(res, &qp->orrq); -fail_buf_free: - bnxt_qplib_free_qp_hdr_buf(res, qp); + bnxt_qplib_free_hwq(res, &qp->orrq); +rq_swq: + kfree(rq->swq); fail_rq: bnxt_qplib_free_hwq(res, &rq->hwq); - kfree(rq->swq); +sq_swq: + kfree(sq->swq); fail_sq: bnxt_qplib_free_hwq(res, &sq->hwq); - kfree(sq->swq); exit: return rc; } @@ -1483,12 +1416,11 @@ bail: static void __clean_cq(struct bnxt_qplib_cq *cq, u64 qp) { struct bnxt_qplib_hwq *cq_hwq = &cq->hwq; - struct cq_base *hw_cqe, **hw_cqe_ptr; + struct cq_base *hw_cqe; int i; for (i = 0; i < cq_hwq->max_elements; i++) { - hw_cqe_ptr = (struct cq_base **)cq_hwq->pbl_ptr; - hw_cqe = &hw_cqe_ptr[CQE_PG(i)][CQE_IDX(i)]; + hw_cqe = bnxt_qplib_get_qe(cq_hwq, i, NULL); if (!CQE_CMP_VALID(hw_cqe, i, cq_hwq->max_elements)) continue; /* @@ -1529,10 +1461,12 @@ int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res, struct cmdq_destroy_qp req; struct creq_destroy_qp_resp resp; u16 cmd_flags = 0; + u32 tbl_indx; int rc; - rcfw->qp_tbl[qp->id].qp_id = BNXT_QPLIB_QP_ID_INVALID; - rcfw->qp_tbl[qp->id].qp_handle = NULL; + tbl_indx = map_qp_id_to_tbl_indx(qp->id, rcfw); + rcfw->qp_tbl[tbl_indx].qp_id = BNXT_QPLIB_QP_ID_INVALID; + rcfw->qp_tbl[tbl_indx].qp_handle = NULL; RCFW_CMD_PREP(req, DESTROY_QP, cmd_flags); @@ -1540,8 +1474,8 @@ int bnxt_qplib_destroy_qp(struct bnxt_qplib_res *res, rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, (void *)&resp, NULL, 0); if (rc) { - rcfw->qp_tbl[qp->id].qp_id = qp->id; - rcfw->qp_tbl[qp->id].qp_handle = qp; + rcfw->qp_tbl[tbl_indx].qp_id = qp->id; + rcfw->qp_tbl[tbl_indx].qp_handle = qp; return rc; } @@ -1574,7 +1508,7 @@ void *bnxt_qplib_get_qp1_sq_buf(struct bnxt_qplib_qp *qp, memset(sge, 0, sizeof(*sge)); if (qp->sq_hdr_buf) { - sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq); + sw_prod = sq->swq_start; sge->addr = (dma_addr_t)(qp->sq_hdr_buf_map + sw_prod * qp->sq_hdr_buf_size); sge->lkey = 0xFFFFFFFF; @@ -1588,7 +1522,7 @@ u32 bnxt_qplib_get_rq_prod_index(struct bnxt_qplib_qp *qp) { struct bnxt_qplib_q *rq = &qp->rq; - return HWQ_CMP(rq->hwq.prod, &rq->hwq); + return rq->swq_start; } dma_addr_t bnxt_qplib_get_qp_buf_from_index(struct bnxt_qplib_qp *qp, u32 index) @@ -1605,7 +1539,7 @@ void *bnxt_qplib_get_qp1_rq_buf(struct bnxt_qplib_qp *qp, memset(sge, 0, sizeof(*sge)); if (qp->rq_hdr_buf) { - sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); + sw_prod = rq->swq_start; sge->addr = (dma_addr_t)(qp->rq_hdr_buf_map + sw_prod * qp->rq_hdr_buf_size); sge->lkey = 0xFFFFFFFF; @@ -1615,6 +1549,146 @@ void *bnxt_qplib_get_qp1_rq_buf(struct bnxt_qplib_qp *qp, return NULL; } +static void bnxt_qplib_fill_psn_search(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_swqe *wqe, + struct bnxt_qplib_swq *swq) +{ + struct sq_psn_search_ext *psns_ext; + struct sq_psn_search *psns; + u32 flg_npsn; + u32 op_spsn; + + if (!swq->psn_search) + return; + psns = swq->psn_search; + psns_ext = swq->psn_ext; + + op_spsn = ((swq->start_psn << SQ_PSN_SEARCH_START_PSN_SFT) & + SQ_PSN_SEARCH_START_PSN_MASK); + op_spsn |= ((wqe->type << SQ_PSN_SEARCH_OPCODE_SFT) & + SQ_PSN_SEARCH_OPCODE_MASK); + flg_npsn = ((swq->next_psn << SQ_PSN_SEARCH_NEXT_PSN_SFT) & + SQ_PSN_SEARCH_NEXT_PSN_MASK); + + if (bnxt_qplib_is_chip_gen_p5(qp->cctx)) { + psns_ext->opcode_start_psn = cpu_to_le32(op_spsn); + psns_ext->flags_next_psn = cpu_to_le32(flg_npsn); + psns_ext->start_slot_idx = cpu_to_le16(swq->slot_idx); + } else { + psns->opcode_start_psn = cpu_to_le32(op_spsn); + psns->flags_next_psn = cpu_to_le32(flg_npsn); + } +} + +static int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_swqe *wqe, + u16 *idx) +{ + struct bnxt_qplib_hwq *hwq; + int len, t_len, offt; + bool pull_dst = true; + void *il_dst = NULL; + void *il_src = NULL; + int t_cplen, cplen; + int indx; + + hwq = &qp->sq.hwq; + t_len = 0; + for (indx = 0; indx < wqe->num_sge; indx++) { + len = wqe->sg_list[indx].size; + il_src = (void *)wqe->sg_list[indx].addr; + t_len += len; + if (t_len > qp->max_inline_data) + goto bad; + while (len) { + if (pull_dst) { + pull_dst = false; + il_dst = bnxt_qplib_get_prod_qe(hwq, *idx); + (*idx)++; + t_cplen = 0; + offt = 0; + } + cplen = min_t(int, len, sizeof(struct sq_sge)); + cplen = min_t(int, cplen, + (sizeof(struct sq_sge) - offt)); + memcpy(il_dst, il_src, cplen); + t_cplen += cplen; + il_src += cplen; + il_dst += cplen; + offt += cplen; + len -= cplen; + if (t_cplen == sizeof(struct sq_sge)) + pull_dst = true; + } + } + + return t_len; +bad: + return -ENOMEM; +} + +static u32 bnxt_qplib_put_sges(struct bnxt_qplib_hwq *hwq, + struct bnxt_qplib_sge *ssge, + u16 nsge, u16 *idx) +{ + struct sq_sge *dsge; + int indx, len = 0; + + for (indx = 0; indx < nsge; indx++, (*idx)++) { + dsge = bnxt_qplib_get_prod_qe(hwq, *idx); + dsge->va_or_pa = cpu_to_le64(ssge[indx].addr); + dsge->l_key = cpu_to_le32(ssge[indx].lkey); + dsge->size = cpu_to_le32(ssge[indx].size); + len += ssge[indx].size; + } + + return len; +} + +static u16 bnxt_qplib_required_slots(struct bnxt_qplib_qp *qp, + struct bnxt_qplib_swqe *wqe, + u16 *wqe_sz, u16 *qdf, u8 mode) +{ + u32 ilsize, bytes; + u16 nsge; + u16 slot; + + nsge = wqe->num_sge; + /* Adding sq_send_hdr is a misnomer, for rq also hdr size is same. */ + bytes = sizeof(struct sq_send_hdr) + nsge * sizeof(struct sq_sge); + if (wqe->flags & BNXT_QPLIB_SWQE_FLAGS_INLINE) { + ilsize = bnxt_qplib_calc_ilsize(wqe, qp->max_inline_data); + bytes = ALIGN(ilsize, sizeof(struct sq_sge)); + bytes += sizeof(struct sq_send_hdr); + } + + *qdf = __xlate_qfd(qp->sq.q_full_delta, bytes); + slot = bytes >> 4; + *wqe_sz = slot; + if (mode == BNXT_QPLIB_WQE_MODE_STATIC) + slot = 8; + return slot; +} + +static void bnxt_qplib_pull_psn_buff(struct bnxt_qplib_q *sq, + struct bnxt_qplib_swq *swq) +{ + struct bnxt_qplib_hwq *hwq; + u32 pg_num, pg_indx; + void *buff; + u32 tail; + + hwq = &sq->hwq; + if (!hwq->pad_pg) + return; + tail = swq->slot_idx / sq->dbinfo.max_slot; + pg_num = (tail + hwq->pad_pgofft) / (PAGE_SIZE / hwq->pad_stride); + pg_indx = (tail + hwq->pad_pgofft) % (PAGE_SIZE / hwq->pad_stride); + buff = (void *)(hwq->pad_pg[pg_num] + pg_indx * hwq->pad_stride); + swq->psn_ext = buff; + swq->psn_search = buff; +} + void bnxt_qplib_post_send_db(struct bnxt_qplib_qp *qp) { struct bnxt_qplib_q *sq = &qp->sq; @@ -1625,124 +1699,114 @@ void bnxt_qplib_post_send_db(struct bnxt_qplib_qp *qp) int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, struct bnxt_qplib_swqe *wqe) { + struct bnxt_qplib_nq_work *nq_work = NULL; + int i, rc = 0, data_len = 0, pkt_num = 0; struct bnxt_qplib_q *sq = &qp->sq; + struct bnxt_qplib_hwq *hwq; struct bnxt_qplib_swq *swq; - struct sq_send *hw_sq_send_hdr, **hw_sq_send_ptr; - struct sq_sge *hw_sge; - struct bnxt_qplib_nq_work *nq_work = NULL; bool sch_handler = false; - u32 sw_prod; - u8 wqe_size16; - int i, rc = 0, data_len = 0, pkt_num = 0; + u16 wqe_sz, qdf = 0; + void *base_hdr; + void *ext_hdr; __le32 temp32; - - if (qp->state != CMDQ_MODIFY_QP_NEW_STATE_RTS) { - if (qp->state == CMDQ_MODIFY_QP_NEW_STATE_ERR) { - sch_handler = true; - dev_dbg(&sq->hwq.pdev->dev, - "%s Error QP. Scheduling for poll_cq\n", - __func__); - goto queue_err; - } + u32 wqe_idx; + u32 slots; + u16 idx; + + hwq = &sq->hwq; + if (qp->state != CMDQ_MODIFY_QP_NEW_STATE_RTS && + qp->state != CMDQ_MODIFY_QP_NEW_STATE_ERR) { + dev_err(&hwq->pdev->dev, + "QPLIB: FP: QP (0x%x) is in the 0x%x state", + qp->id, qp->state); + rc = -EINVAL; + goto done; } - if (bnxt_qplib_queue_full(sq)) { - dev_err(&sq->hwq.pdev->dev, + slots = bnxt_qplib_required_slots(qp, wqe, &wqe_sz, &qdf, qp->wqe_mode); + if (bnxt_qplib_queue_full(sq, slots + qdf)) { + dev_err(&hwq->pdev->dev, "prod = %#x cons = %#x qdepth = %#x delta = %#x\n", - sq->hwq.prod, sq->hwq.cons, sq->hwq.max_elements, - sq->q_full_delta); + hwq->prod, hwq->cons, hwq->depth, sq->q_full_delta); rc = -ENOMEM; goto done; } - sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq); - swq = &sq->swq[sw_prod]; + + swq = bnxt_qplib_get_swqe(sq, &wqe_idx); + bnxt_qplib_pull_psn_buff(sq, swq); + + idx = 0; + swq->slot_idx = hwq->prod; + swq->slots = slots; swq->wr_id = wqe->wr_id; swq->type = wqe->type; swq->flags = wqe->flags; + swq->start_psn = sq->psn & BTH_PSN_MASK; if (qp->sig_type) swq->flags |= SQ_SEND_FLAGS_SIGNAL_COMP; - swq->start_psn = sq->psn & BTH_PSN_MASK; - hw_sq_send_ptr = (struct sq_send **)sq->hwq.pbl_ptr; - hw_sq_send_hdr = &hw_sq_send_ptr[get_sqe_pg(sw_prod)] - [get_sqe_idx(sw_prod)]; + if (qp->state == CMDQ_MODIFY_QP_NEW_STATE_ERR) { + sch_handler = true; + dev_dbg(&hwq->pdev->dev, + "%s Error QP. Scheduling for poll_cq\n", __func__); + goto queue_err; + } - memset(hw_sq_send_hdr, 0, BNXT_QPLIB_MAX_SQE_ENTRY_SIZE); + base_hdr = bnxt_qplib_get_prod_qe(hwq, idx++); + ext_hdr = bnxt_qplib_get_prod_qe(hwq, idx++); + memset(base_hdr, 0, sizeof(struct sq_sge)); + memset(ext_hdr, 0, sizeof(struct sq_sge)); - if (wqe->flags & BNXT_QPLIB_SWQE_FLAGS_INLINE) { + if (wqe->flags & BNXT_QPLIB_SWQE_FLAGS_INLINE) /* Copy the inline data */ - if (wqe->inline_len > BNXT_QPLIB_SWQE_MAX_INLINE_LENGTH) { - dev_warn(&sq->hwq.pdev->dev, - "Inline data length > 96 detected\n"); - data_len = BNXT_QPLIB_SWQE_MAX_INLINE_LENGTH; - } else { - data_len = wqe->inline_len; - } - memcpy(hw_sq_send_hdr->data, wqe->inline_data, data_len); - wqe_size16 = (data_len + 15) >> 4; - } else { - for (i = 0, hw_sge = (struct sq_sge *)hw_sq_send_hdr->data; - i < wqe->num_sge; i++, hw_sge++) { - hw_sge->va_or_pa = cpu_to_le64(wqe->sg_list[i].addr); - hw_sge->l_key = cpu_to_le32(wqe->sg_list[i].lkey); - hw_sge->size = cpu_to_le32(wqe->sg_list[i].size); - data_len += wqe->sg_list[i].size; - } - /* Each SGE entry = 1 WQE size16 */ - wqe_size16 = wqe->num_sge; - /* HW requires wqe size has room for atleast one SGE even if - * none was supplied by ULP - */ - if (!wqe->num_sge) - wqe_size16++; - } - + data_len = bnxt_qplib_put_inline(qp, wqe, &idx); + else + data_len = bnxt_qplib_put_sges(hwq, wqe->sg_list, wqe->num_sge, + &idx); + if (data_len < 0) + goto queue_err; /* Specifics */ switch (wqe->type) { case BNXT_QPLIB_SWQE_TYPE_SEND: if (qp->type == CMDQ_CREATE_QP1_TYPE_GSI) { + struct sq_send_raweth_qp1_hdr *sqe = base_hdr; + struct sq_raw_ext_hdr *ext_sqe = ext_hdr; /* Assemble info for Raw Ethertype QPs */ - struct sq_send_raweth_qp1 *sqe = - (struct sq_send_raweth_qp1 *)hw_sq_send_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; - sqe->wqe_size = wqe_size16 + - ((offsetof(typeof(*sqe), data) + 15) >> 4); + sqe->wqe_size = wqe_sz; sqe->cfa_action = cpu_to_le16(wqe->rawqp1.cfa_action); sqe->lflags = cpu_to_le16(wqe->rawqp1.lflags); sqe->length = cpu_to_le32(data_len); - sqe->cfa_meta = cpu_to_le32((wqe->rawqp1.cfa_meta & + ext_sqe->cfa_meta = cpu_to_le32((wqe->rawqp1.cfa_meta & SQ_SEND_RAWETH_QP1_CFA_META_VLAN_VID_MASK) << SQ_SEND_RAWETH_QP1_CFA_META_VLAN_VID_SFT); break; } - /* fall thru */ + fallthrough; case BNXT_QPLIB_SWQE_TYPE_SEND_WITH_IMM: case BNXT_QPLIB_SWQE_TYPE_SEND_WITH_INV: { - struct sq_send *sqe = (struct sq_send *)hw_sq_send_hdr; + struct sq_ud_ext_hdr *ext_sqe = ext_hdr; + struct sq_send_hdr *sqe = base_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; - sqe->wqe_size = wqe_size16 + - ((offsetof(typeof(*sqe), data) + 15) >> 4); - sqe->inv_key_or_imm_data = cpu_to_le32( - wqe->send.inv_key); + sqe->wqe_size = wqe_sz; + sqe->inv_key_or_imm_data = cpu_to_le32(wqe->send.inv_key); if (qp->type == CMDQ_CREATE_QP_TYPE_UD || qp->type == CMDQ_CREATE_QP_TYPE_GSI) { sqe->q_key = cpu_to_le32(wqe->send.q_key); - sqe->dst_qp = cpu_to_le32( - wqe->send.dst_qp & SQ_SEND_DST_QP_MASK); sqe->length = cpu_to_le32(data_len); - sqe->avid = cpu_to_le32(wqe->send.avid & - SQ_SEND_AVID_MASK); sq->psn = (sq->psn + 1) & BTH_PSN_MASK; + ext_sqe->dst_qp = cpu_to_le32(wqe->send.dst_qp & + SQ_SEND_DST_QP_MASK); + ext_sqe->avid = cpu_to_le32(wqe->send.avid & + SQ_SEND_AVID_MASK); } else { sqe->length = cpu_to_le32(data_len); - sqe->dst_qp = 0; - sqe->avid = 0; if (qp->mtu) pkt_num = (data_len + qp->mtu - 1) / qp->mtu; if (!pkt_num) @@ -1755,16 +1819,16 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, case BNXT_QPLIB_SWQE_TYPE_RDMA_WRITE_WITH_IMM: case BNXT_QPLIB_SWQE_TYPE_RDMA_READ: { - struct sq_rdma *sqe = (struct sq_rdma *)hw_sq_send_hdr; + struct sq_rdma_ext_hdr *ext_sqe = ext_hdr; + struct sq_rdma_hdr *sqe = base_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; - sqe->wqe_size = wqe_size16 + - ((offsetof(typeof(*sqe), data) + 15) >> 4); + sqe->wqe_size = wqe_sz; sqe->imm_data = cpu_to_le32(wqe->rdma.inv_key); sqe->length = cpu_to_le32((u32)data_len); - sqe->remote_va = cpu_to_le64(wqe->rdma.remote_va); - sqe->remote_key = cpu_to_le32(wqe->rdma.r_key); + ext_sqe->remote_va = cpu_to_le64(wqe->rdma.remote_va); + ext_sqe->remote_key = cpu_to_le32(wqe->rdma.r_key); if (qp->mtu) pkt_num = (data_len + qp->mtu - 1) / qp->mtu; if (!pkt_num) @@ -1775,14 +1839,15 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, case BNXT_QPLIB_SWQE_TYPE_ATOMIC_CMP_AND_SWP: case BNXT_QPLIB_SWQE_TYPE_ATOMIC_FETCH_AND_ADD: { - struct sq_atomic *sqe = (struct sq_atomic *)hw_sq_send_hdr; + struct sq_atomic_ext_hdr *ext_sqe = ext_hdr; + struct sq_atomic_hdr *sqe = base_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; sqe->remote_key = cpu_to_le32(wqe->atomic.r_key); sqe->remote_va = cpu_to_le64(wqe->atomic.remote_va); - sqe->swap_data = cpu_to_le64(wqe->atomic.swap_data); - sqe->cmp_data = cpu_to_le64(wqe->atomic.cmp_data); + ext_sqe->swap_data = cpu_to_le64(wqe->atomic.swap_data); + ext_sqe->cmp_data = cpu_to_le64(wqe->atomic.cmp_data); if (qp->mtu) pkt_num = (data_len + qp->mtu - 1) / qp->mtu; if (!pkt_num) @@ -1792,8 +1857,7 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, } case BNXT_QPLIB_SWQE_TYPE_LOCAL_INV: { - struct sq_localinvalidate *sqe = - (struct sq_localinvalidate *)hw_sq_send_hdr; + struct sq_localinvalidate *sqe = base_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; @@ -1803,7 +1867,8 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, } case BNXT_QPLIB_SWQE_TYPE_FAST_REG_MR: { - struct sq_fr_pmr *sqe = (struct sq_fr_pmr *)hw_sq_send_hdr; + struct sq_fr_pmr_ext_hdr *ext_sqe = ext_hdr; + struct sq_fr_pmr_hdr *sqe = base_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; @@ -1827,14 +1892,15 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, wqe->frmr.pbl_ptr[i] = cpu_to_le64( wqe->frmr.page_list[i] | PTU_PTE_VALID); - sqe->pblptr = cpu_to_le64(wqe->frmr.pbl_dma_ptr); - sqe->va = cpu_to_le64(wqe->frmr.va); + ext_sqe->pblptr = cpu_to_le64(wqe->frmr.pbl_dma_ptr); + ext_sqe->va = cpu_to_le64(wqe->frmr.va); break; } case BNXT_QPLIB_SWQE_TYPE_BIND_MW: { - struct sq_bind *sqe = (struct sq_bind *)hw_sq_send_hdr; + struct sq_bind_ext_hdr *ext_sqe = ext_hdr; + struct sq_bind_hdr *sqe = base_hdr; sqe->wqe_type = wqe->type; sqe->flags = wqe->flags; @@ -1843,9 +1909,8 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, (wqe->bind.zero_based ? SQ_BIND_ZERO_BASED : 0); sqe->parent_l_key = cpu_to_le32(wqe->bind.parent_l_key); sqe->l_key = cpu_to_le32(wqe->bind.r_key); - sqe->va = cpu_to_le64(wqe->bind.va); - temp32 = cpu_to_le32(wqe->bind.length); - memcpy(&sqe->length, &temp32, sizeof(wqe->bind.length)); + ext_sqe->va = cpu_to_le64(wqe->bind.va); + ext_sqe->length_lo = cpu_to_le32(wqe->bind.length); break; } default: @@ -1854,43 +1919,11 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp, goto done; } swq->next_psn = sq->psn & BTH_PSN_MASK; - if (swq->psn_search) { - u32 opcd_spsn; - u32 flg_npsn; - - opcd_spsn = ((swq->start_psn << SQ_PSN_SEARCH_START_PSN_SFT) & - SQ_PSN_SEARCH_START_PSN_MASK); - opcd_spsn |= ((wqe->type << SQ_PSN_SEARCH_OPCODE_SFT) & - SQ_PSN_SEARCH_OPCODE_MASK); - flg_npsn = ((swq->next_psn << SQ_PSN_SEARCH_NEXT_PSN_SFT) & - SQ_PSN_SEARCH_NEXT_PSN_MASK); - if (bnxt_qplib_is_chip_gen_p5(qp->cctx)) { - swq->psn_ext->opcode_start_psn = - cpu_to_le32(opcd_spsn); - swq->psn_ext->flags_next_psn = - cpu_to_le32(flg_npsn); - } else { - swq->psn_search->opcode_start_psn = - cpu_to_le32(opcd_spsn); - swq->psn_search->flags_next_psn = - cpu_to_le32(flg_npsn); - } - } + bnxt_qplib_fill_psn_search(qp, wqe, swq); queue_err: - if (sch_handler) { - /* Store the ULP info in the software structures */ - sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq); - swq = &sq->swq[sw_prod]; - swq->wr_id = wqe->wr_id; - swq->type = wqe->type; - swq->flags = wqe->flags; - if (qp->sig_type) - swq->flags |= SQ_SEND_FLAGS_SIGNAL_COMP; - swq->start_psn = sq->psn & BTH_PSN_MASK; - } - sq->hwq.prod++; + bnxt_qplib_swq_mod_start(sq, wqe_idx); + bnxt_qplib_hwq_incr_prod(hwq, swq->slots); qp->wqe_cnt++; - done: if (sch_handler) { nq_work = kzalloc(sizeof(*nq_work), GFP_ATOMIC); @@ -1900,7 +1933,7 @@ done: INIT_WORK(&nq_work->work, bnxt_qpn_cqn_sched_task); queue_work(qp->scq->nq->cqn_wq, &nq_work->work); } else { - dev_err(&sq->hwq.pdev->dev, + dev_err(&hwq->pdev->dev, "FP: Failed to allocate SQ nq_work!\n"); rc = -ENOMEM; } @@ -1918,62 +1951,67 @@ void bnxt_qplib_post_recv_db(struct bnxt_qplib_qp *qp) int bnxt_qplib_post_recv(struct bnxt_qplib_qp *qp, struct bnxt_qplib_swqe *wqe) { - struct bnxt_qplib_q *rq = &qp->rq; - struct rq_wqe *rqe, **rqe_ptr; - struct sq_sge *hw_sge; struct bnxt_qplib_nq_work *nq_work = NULL; + struct bnxt_qplib_q *rq = &qp->rq; + struct rq_wqe_hdr *base_hdr; + struct rq_ext_hdr *ext_hdr; + struct bnxt_qplib_hwq *hwq; + struct bnxt_qplib_swq *swq; bool sch_handler = false; - u32 sw_prod; - int i, rc = 0; + u16 wqe_sz, idx; + u32 wqe_idx; + int rc = 0; - if (qp->state == CMDQ_MODIFY_QP_NEW_STATE_ERR) { - sch_handler = true; - dev_dbg(&rq->hwq.pdev->dev, - "%s: Error QP. Scheduling for poll_cq\n", __func__); - goto queue_err; + hwq = &rq->hwq; + if (qp->state == CMDQ_MODIFY_QP_NEW_STATE_RESET) { + dev_err(&hwq->pdev->dev, + "QPLIB: FP: QP (0x%x) is in the 0x%x state", + qp->id, qp->state); + rc = -EINVAL; + goto done; } - if (bnxt_qplib_queue_full(rq)) { - dev_err(&rq->hwq.pdev->dev, + + if (bnxt_qplib_queue_full(rq, rq->dbinfo.max_slot)) { + dev_err(&hwq->pdev->dev, "FP: QP (0x%x) RQ is full!\n", qp->id); rc = -EINVAL; goto done; } - sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); - rq->swq[sw_prod].wr_id = wqe->wr_id; - - rqe_ptr = (struct rq_wqe **)rq->hwq.pbl_ptr; - rqe = &rqe_ptr[RQE_PG(sw_prod)][RQE_IDX(sw_prod)]; - memset(rqe, 0, BNXT_QPLIB_MAX_RQE_ENTRY_SIZE); + swq = bnxt_qplib_get_swqe(rq, &wqe_idx); + swq->wr_id = wqe->wr_id; + swq->slots = rq->dbinfo.max_slot; - /* Calculate wqe_size16 and data_len */ - for (i = 0, hw_sge = (struct sq_sge *)rqe->data; - i < wqe->num_sge; i++, hw_sge++) { - hw_sge->va_or_pa = cpu_to_le64(wqe->sg_list[i].addr); - hw_sge->l_key = cpu_to_le32(wqe->sg_list[i].lkey); - hw_sge->size = cpu_to_le32(wqe->sg_list[i].size); + if (qp->state == CMDQ_MODIFY_QP_NEW_STATE_ERR) { + sch_handler = true; + dev_dbg(&hwq->pdev->dev, + "%s: Error QP. Scheduling for poll_cq\n", __func__); + goto queue_err; } - rqe->wqe_type = wqe->type; - rqe->flags = wqe->flags; - rqe->wqe_size = wqe->num_sge + - ((offsetof(typeof(*rqe), data) + 15) >> 4); - /* HW requires wqe size has room for atleast one SGE even if none - * was supplied by ULP - */ - if (!wqe->num_sge) - rqe->wqe_size++; - - /* Supply the rqe->wr_id index to the wr_id_tbl for now */ - rqe->wr_id[0] = cpu_to_le32(sw_prod); + idx = 0; + base_hdr = bnxt_qplib_get_prod_qe(hwq, idx++); + ext_hdr = bnxt_qplib_get_prod_qe(hwq, idx++); + memset(base_hdr, 0, sizeof(struct sq_sge)); + memset(ext_hdr, 0, sizeof(struct sq_sge)); + wqe_sz = (sizeof(struct rq_wqe_hdr) + + wqe->num_sge * sizeof(struct sq_sge)) >> 4; + bnxt_qplib_put_sges(hwq, wqe->sg_list, wqe->num_sge, &idx); + if (!wqe->num_sge) { + struct sq_sge *sge; + + sge = bnxt_qplib_get_prod_qe(hwq, idx++); + sge->size = 0; + wqe_sz++; + } + base_hdr->wqe_type = wqe->type; + base_hdr->flags = wqe->flags; + base_hdr->wqe_size = wqe_sz; + base_hdr->wr_id[0] = cpu_to_le32(wqe_idx); queue_err: - if (sch_handler) { - /* Store the ULP info in the software structures */ - sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); - rq->swq[sw_prod].wr_id = wqe->wr_id; - } - - rq->hwq.prod++; + bnxt_qplib_swq_mod_start(rq, wqe_idx); + bnxt_qplib_hwq_incr_prod(hwq, swq->slots); +done: if (sch_handler) { nq_work = kzalloc(sizeof(*nq_work), GFP_ATOMIC); if (nq_work) { @@ -1982,12 +2020,12 @@ queue_err: INIT_WORK(&nq_work->work, bnxt_qpn_cqn_sched_task); queue_work(qp->rcq->nq->cqn_wq, &nq_work->work); } else { - dev_err(&rq->hwq.pdev->dev, + dev_err(&hwq->pdev->dev, "FP: Failed to allocate RQ nq_work!\n"); rc = -ENOMEM; } } -done: + return rc; } @@ -1997,9 +2035,10 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq) struct bnxt_qplib_rcfw *rcfw = res->rcfw; struct bnxt_qplib_hwq_attr hwq_attr = {}; struct creq_create_cq_resp resp; - struct cmdq_create_cq req; struct bnxt_qplib_pbl *pbl; + struct cmdq_create_cq req; u16 cmd_flags = 0; + u32 pg_sz_lvl; int rc; hwq_attr.res = res; @@ -2020,22 +2059,13 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq) } req.dpi = cpu_to_le32(cq->dpi->dpi); req.cq_handle = cpu_to_le64(cq->cq_handle); - req.cq_size = cpu_to_le32(cq->hwq.max_elements); pbl = &cq->hwq.pbl[PBL_LVL_0]; - req.pg_size_lvl = cpu_to_le32( - ((cq->hwq.level & CMDQ_CREATE_CQ_LVL_MASK) << - CMDQ_CREATE_CQ_LVL_SFT) | - (pbl->pg_size == ROCE_PG_SIZE_4K ? CMDQ_CREATE_CQ_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? CMDQ_CREATE_CQ_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? CMDQ_CREATE_CQ_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? CMDQ_CREATE_CQ_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? CMDQ_CREATE_CQ_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? CMDQ_CREATE_CQ_PG_SIZE_PG_1G : - CMDQ_CREATE_CQ_PG_SIZE_PG_4K)); - + pg_sz_lvl = (bnxt_qplib_base_pg_size(&cq->hwq) << + CMDQ_CREATE_CQ_PG_SIZE_SFT); + pg_sz_lvl |= (cq->hwq.level & CMDQ_CREATE_CQ_LVL_MASK); + req.pg_size_lvl = cpu_to_le32(pg_sz_lvl); req.pbl = cpu_to_le64(pbl->pg_map_arr[0]); - req.cq_fco_cnq_id = cpu_to_le32( (cq->cnq_hw_ring_id & CMDQ_CREATE_CQ_CNQ_ID_MASK) << CMDQ_CREATE_CQ_CNQ_ID_SFT); @@ -2093,20 +2123,19 @@ int bnxt_qplib_destroy_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq) static int __flush_sq(struct bnxt_qplib_q *sq, struct bnxt_qplib_qp *qp, struct bnxt_qplib_cqe **pcqe, int *budget) { - u32 sw_prod, sw_cons; struct bnxt_qplib_cqe *cqe; + u32 start, last; int rc = 0; /* Now complete all outstanding SQEs with FLUSHED_ERR */ - sw_prod = HWQ_CMP(sq->hwq.prod, &sq->hwq); + start = sq->swq_start; cqe = *pcqe; while (*budget) { - sw_cons = HWQ_CMP(sq->hwq.cons, &sq->hwq); - if (sw_cons == sw_prod) { + last = sq->swq_last; + if (start == last) break; - } /* Skip the FENCE WQE completions */ - if (sq->swq[sw_cons].wr_id == BNXT_QPLIB_FENCE_WRID) { + if (sq->swq[last].wr_id == BNXT_QPLIB_FENCE_WRID) { bnxt_qplib_cancel_phantom_processing(qp); goto skip_compl; } @@ -2114,16 +2143,17 @@ static int __flush_sq(struct bnxt_qplib_q *sq, struct bnxt_qplib_qp *qp, cqe->status = CQ_REQ_STATUS_WORK_REQUEST_FLUSHED_ERR; cqe->opcode = CQ_BASE_CQE_TYPE_REQ; cqe->qp_handle = (u64)(unsigned long)qp; - cqe->wr_id = sq->swq[sw_cons].wr_id; + cqe->wr_id = sq->swq[last].wr_id; cqe->src_qp = qp->id; - cqe->type = sq->swq[sw_cons].type; + cqe->type = sq->swq[last].type; cqe++; (*budget)--; skip_compl: - sq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&sq->hwq, sq->swq[last].slots); + sq->swq_last = sq->swq[last].next_idx; } *pcqe = cqe; - if (!(*budget) && HWQ_CMP(sq->hwq.cons, &sq->hwq) != sw_prod) + if (!(*budget) && sq->swq_last != start) /* Out of budget */ rc = -EAGAIN; @@ -2134,9 +2164,9 @@ static int __flush_rq(struct bnxt_qplib_q *rq, struct bnxt_qplib_qp *qp, struct bnxt_qplib_cqe **pcqe, int *budget) { struct bnxt_qplib_cqe *cqe; - u32 sw_prod, sw_cons; - int rc = 0; + u32 start, last; int opcode = 0; + int rc = 0; switch (qp->type) { case CMDQ_CREATE_QP1_TYPE_GSI: @@ -2152,24 +2182,25 @@ static int __flush_rq(struct bnxt_qplib_q *rq, struct bnxt_qplib_qp *qp, } /* Flush the rest of the RQ */ - sw_prod = HWQ_CMP(rq->hwq.prod, &rq->hwq); + start = rq->swq_start; cqe = *pcqe; while (*budget) { - sw_cons = HWQ_CMP(rq->hwq.cons, &rq->hwq); - if (sw_cons == sw_prod) + last = rq->swq_last; + if (last == start) break; memset(cqe, 0, sizeof(*cqe)); cqe->status = CQ_RES_RC_STATUS_WORK_REQUEST_FLUSHED_ERR; cqe->opcode = opcode; cqe->qp_handle = (unsigned long)qp; - cqe->wr_id = rq->swq[sw_cons].wr_id; + cqe->wr_id = rq->swq[last].wr_id; cqe++; (*budget)--; - rq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&rq->hwq, rq->swq[last].slots); + rq->swq_last = rq->swq[last].next_idx; } *pcqe = cqe; - if (!*budget && HWQ_CMP(rq->hwq.cons, &rq->hwq) != sw_prod) + if (!*budget && rq->swq_last != start) /* Out of budget */ rc = -EAGAIN; @@ -2192,20 +2223,20 @@ void bnxt_qplib_mark_qp_error(void *qp_handle) * CQE is track from sw_cq_cons to max_element but valid only if VALID=1 */ static int do_wa9060(struct bnxt_qplib_qp *qp, struct bnxt_qplib_cq *cq, - u32 cq_cons, u32 sw_sq_cons, u32 cqe_sq_cons) + u32 cq_cons, u32 swq_last, u32 cqe_sq_cons) { - struct bnxt_qplib_q *sq = &qp->sq; - struct bnxt_qplib_swq *swq; u32 peek_sw_cq_cons, peek_raw_cq_cons, peek_sq_cons_idx; - struct cq_base *peek_hwcqe, **peek_hw_cqe_ptr; + struct bnxt_qplib_q *sq = &qp->sq; struct cq_req *peek_req_hwcqe; struct bnxt_qplib_qp *peek_qp; struct bnxt_qplib_q *peek_sq; + struct bnxt_qplib_swq *swq; + struct cq_base *peek_hwcqe; int i, rc = 0; /* Normal mode */ /* Check for the psn_search marking before completing */ - swq = &sq->swq[sw_sq_cons]; + swq = &sq->swq[swq_last]; if (swq->psn_search && le32_to_cpu(swq->psn_search->flags_next_psn) & 0x80000000) { /* Unmark */ @@ -2214,7 +2245,7 @@ static int do_wa9060(struct bnxt_qplib_qp *qp, struct bnxt_qplib_cq *cq, & ~0x80000000); dev_dbg(&cq->hwq.pdev->dev, "FP: Process Req cq_cons=0x%x qp=0x%x sq cons sw=0x%x cqe=0x%x marked!\n", - cq_cons, qp->id, sw_sq_cons, cqe_sq_cons); + cq_cons, qp->id, swq_last, cqe_sq_cons); sq->condition = true; sq->send_phantom = true; @@ -2230,9 +2261,8 @@ static int do_wa9060(struct bnxt_qplib_qp *qp, struct bnxt_qplib_cq *cq, i = cq->hwq.max_elements; while (i--) { peek_sw_cq_cons = HWQ_CMP((peek_sw_cq_cons), &cq->hwq); - peek_hw_cqe_ptr = (struct cq_base **)cq->hwq.pbl_ptr; - peek_hwcqe = &peek_hw_cqe_ptr[CQE_PG(peek_sw_cq_cons)] - [CQE_IDX(peek_sw_cq_cons)]; + peek_hwcqe = bnxt_qplib_get_qe(&cq->hwq, + peek_sw_cq_cons, NULL); /* If the next hwcqe is VALID */ if (CQE_CMP_VALID(peek_hwcqe, peek_raw_cq_cons, cq->hwq.max_elements)) { @@ -2252,9 +2282,10 @@ static int do_wa9060(struct bnxt_qplib_qp *qp, struct bnxt_qplib_cq *cq, le64_to_cpu (peek_req_hwcqe->qp_handle)); peek_sq = &peek_qp->sq; - peek_sq_cons_idx = HWQ_CMP(le16_to_cpu( - peek_req_hwcqe->sq_cons_idx) - 1 - , &sq->hwq); + peek_sq_cons_idx = + ((le16_to_cpu( + peek_req_hwcqe->sq_cons_idx) + - 1) % sq->max_wqe); /* If the hwcqe's sq's wr_id matches */ if (peek_sq == sq && sq->swq[peek_sq_cons_idx].wr_id == @@ -2282,7 +2313,7 @@ static int do_wa9060(struct bnxt_qplib_qp *qp, struct bnxt_qplib_cq *cq, } dev_err(&cq->hwq.pdev->dev, "Should not have come here! cq_cons=0x%x qp=0x%x sq cons sw=0x%x hw=0x%x\n", - cq_cons, qp->id, sw_sq_cons, cqe_sq_cons); + cq_cons, qp->id, swq_last, cqe_sq_cons); rc = -EINVAL; } out: @@ -2294,11 +2325,11 @@ static int bnxt_qplib_cq_process_req(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe **pcqe, int *budget, u32 cq_cons, struct bnxt_qplib_qp **lib_qp) { + struct bnxt_qplib_swq *swq; + struct bnxt_qplib_cqe *cqe; struct bnxt_qplib_qp *qp; struct bnxt_qplib_q *sq; - struct bnxt_qplib_cqe *cqe; - u32 sw_sq_cons, cqe_sq_cons; - struct bnxt_qplib_swq *swq; + u32 cqe_sq_cons; int rc = 0; qp = (struct bnxt_qplib_qp *)((unsigned long) @@ -2310,14 +2341,7 @@ static int bnxt_qplib_cq_process_req(struct bnxt_qplib_cq *cq, } sq = &qp->sq; - cqe_sq_cons = HWQ_CMP(le16_to_cpu(hwcqe->sq_cons_idx), &sq->hwq); - if (cqe_sq_cons > sq->hwq.max_elements) { - dev_err(&cq->hwq.pdev->dev, - "FP: CQ Process req reported sq_cons_idx 0x%x which exceeded max 0x%x\n", - cqe_sq_cons, sq->hwq.max_elements); - return -EINVAL; - } - + cqe_sq_cons = le16_to_cpu(hwcqe->sq_cons_idx) % sq->max_wqe; if (qp->sq.flushed) { dev_dbg(&cq->hwq.pdev->dev, "%s: QP in Flush QP = %p\n", __func__, qp); @@ -2329,12 +2353,11 @@ static int bnxt_qplib_cq_process_req(struct bnxt_qplib_cq *cq, */ cqe = *pcqe; while (*budget) { - sw_sq_cons = HWQ_CMP(sq->hwq.cons, &sq->hwq); - if (sw_sq_cons == cqe_sq_cons) + if (sq->swq_last == cqe_sq_cons) /* Done */ break; - swq = &sq->swq[sw_sq_cons]; + swq = &sq->swq[sq->swq_last]; memset(cqe, 0, sizeof(*cqe)); cqe->opcode = CQ_BASE_CQE_TYPE_REQ; cqe->qp_handle = (u64)(unsigned long)qp; @@ -2348,12 +2371,12 @@ static int bnxt_qplib_cq_process_req(struct bnxt_qplib_cq *cq, * of the request being signaled or not, it must complete with * the hwcqe error status */ - if (HWQ_CMP((sw_sq_cons + 1), &sq->hwq) == cqe_sq_cons && + if (swq->next_idx == cqe_sq_cons && hwcqe->status != CQ_REQ_STATUS_OK) { cqe->status = hwcqe->status; dev_err(&cq->hwq.pdev->dev, "FP: CQ Processed Req wr_id[%d] = 0x%llx with status 0x%x\n", - sw_sq_cons, cqe->wr_id, cqe->status); + sq->swq_last, cqe->wr_id, cqe->status); cqe++; (*budget)--; bnxt_qplib_mark_qp_error(qp); @@ -2361,7 +2384,7 @@ static int bnxt_qplib_cq_process_req(struct bnxt_qplib_cq *cq, bnxt_qplib_add_flush_qp(qp); } else { /* Before we complete, do WA 9060 */ - if (do_wa9060(qp, cq, cq_cons, sw_sq_cons, + if (do_wa9060(qp, cq, cq_cons, sq->swq_last, cqe_sq_cons)) { *lib_qp = qp; goto out; @@ -2373,13 +2396,14 @@ static int bnxt_qplib_cq_process_req(struct bnxt_qplib_cq *cq, } } skip: - sq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&sq->hwq, swq->slots); + sq->swq_last = swq->next_idx; if (sq->single) break; } out: *pcqe = cqe; - if (HWQ_CMP(sq->hwq.cons, &sq->hwq) != cqe_sq_cons) { + if (sq->swq_last != cqe_sq_cons) { /* Out of budget */ rc = -EAGAIN; goto done; @@ -2408,10 +2432,10 @@ static int bnxt_qplib_cq_process_res_rc(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe **pcqe, int *budget) { - struct bnxt_qplib_qp *qp; - struct bnxt_qplib_q *rq; struct bnxt_qplib_srq *srq; struct bnxt_qplib_cqe *cqe; + struct bnxt_qplib_qp *qp; + struct bnxt_qplib_q *rq; u32 wr_id_idx; int rc = 0; @@ -2454,17 +2478,23 @@ static int bnxt_qplib_cq_process_res_rc(struct bnxt_qplib_cq *cq, (*budget)--; *pcqe = cqe; } else { + struct bnxt_qplib_swq *swq; + rq = &qp->rq; - if (wr_id_idx >= rq->hwq.max_elements) { + if (wr_id_idx > (rq->max_wqe - 1)) { dev_err(&cq->hwq.pdev->dev, "FP: CQ Process RC wr_id idx 0x%x exceeded RQ max 0x%x\n", - wr_id_idx, rq->hwq.max_elements); + wr_id_idx, rq->max_wqe); return -EINVAL; } - cqe->wr_id = rq->swq[wr_id_idx].wr_id; + if (wr_id_idx != rq->swq_last) + return -EINVAL; + swq = &rq->swq[rq->swq_last]; + cqe->wr_id = swq->wr_id; cqe++; (*budget)--; - rq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&rq->hwq, swq->slots); + rq->swq_last = swq->next_idx; *pcqe = cqe; if (hwcqe->status != CQ_RES_RC_STATUS_OK) { @@ -2483,10 +2513,10 @@ static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe **pcqe, int *budget) { - struct bnxt_qplib_qp *qp; - struct bnxt_qplib_q *rq; struct bnxt_qplib_srq *srq; struct bnxt_qplib_cqe *cqe; + struct bnxt_qplib_qp *qp; + struct bnxt_qplib_q *rq; u32 wr_id_idx; int rc = 0; @@ -2535,18 +2565,24 @@ static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq, (*budget)--; *pcqe = cqe; } else { + struct bnxt_qplib_swq *swq; + rq = &qp->rq; - if (wr_id_idx >= rq->hwq.max_elements) { + if (wr_id_idx > (rq->max_wqe - 1)) { dev_err(&cq->hwq.pdev->dev, "FP: CQ Process UD wr_id idx 0x%x exceeded RQ max 0x%x\n", - wr_id_idx, rq->hwq.max_elements); + wr_id_idx, rq->max_wqe); return -EINVAL; } - cqe->wr_id = rq->swq[wr_id_idx].wr_id; + if (rq->swq_last != wr_id_idx) + return -EINVAL; + swq = &rq->swq[rq->swq_last]; + cqe->wr_id = swq->wr_id; cqe++; (*budget)--; - rq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&rq->hwq, swq->slots); + rq->swq_last = swq->next_idx; *pcqe = cqe; if (hwcqe->status != CQ_RES_RC_STATUS_OK) { @@ -2561,15 +2597,13 @@ done: bool bnxt_qplib_is_cq_empty(struct bnxt_qplib_cq *cq) { - struct cq_base *hw_cqe, **hw_cqe_ptr; + struct cq_base *hw_cqe; u32 sw_cons, raw_cons; bool rc = true; raw_cons = cq->hwq.cons; sw_cons = HWQ_CMP(raw_cons, &cq->hwq); - hw_cqe_ptr = (struct cq_base **)cq->hwq.pbl_ptr; - hw_cqe = &hw_cqe_ptr[CQE_PG(sw_cons)][CQE_IDX(sw_cons)]; - + hw_cqe = bnxt_qplib_get_qe(&cq->hwq, sw_cons, NULL); /* Check for Valid bit. If the CQE is valid, return false */ rc = !CQE_CMP_VALID(hw_cqe, raw_cons, cq->hwq.max_elements); return rc; @@ -2639,17 +2673,23 @@ static int bnxt_qplib_cq_process_res_raweth_qp1(struct bnxt_qplib_cq *cq, (*budget)--; *pcqe = cqe; } else { + struct bnxt_qplib_swq *swq; + rq = &qp->rq; - if (wr_id_idx >= rq->hwq.max_elements) { + if (wr_id_idx > (rq->max_wqe - 1)) { dev_err(&cq->hwq.pdev->dev, "FP: CQ Process Raw/QP1 RQ wr_id idx 0x%x exceeded RQ max 0x%x\n", - wr_id_idx, rq->hwq.max_elements); + wr_id_idx, rq->max_wqe); return -EINVAL; } - cqe->wr_id = rq->swq[wr_id_idx].wr_id; + if (rq->swq_last != wr_id_idx) + return -EINVAL; + swq = &rq->swq[rq->swq_last]; + cqe->wr_id = swq->wr_id; cqe++; (*budget)--; - rq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&rq->hwq, swq->slots); + rq->swq_last = swq->next_idx; *pcqe = cqe; if (hwcqe->status != CQ_RES_RC_STATUS_OK) { @@ -2671,7 +2711,7 @@ static int bnxt_qplib_cq_process_terminal(struct bnxt_qplib_cq *cq, struct bnxt_qplib_qp *qp; struct bnxt_qplib_q *sq, *rq; struct bnxt_qplib_cqe *cqe; - u32 sw_cons = 0, cqe_cons; + u32 swq_last = 0, cqe_cons; int rc = 0; /* Check the Status */ @@ -2697,13 +2737,7 @@ static int bnxt_qplib_cq_process_terminal(struct bnxt_qplib_cq *cq, cqe_cons = le16_to_cpu(hwcqe->sq_cons_idx); if (cqe_cons == 0xFFFF) goto do_rq; - - if (cqe_cons > sq->hwq.max_elements) { - dev_err(&cq->hwq.pdev->dev, - "FP: CQ Process terminal reported sq_cons_idx 0x%x which exceeded max 0x%x\n", - cqe_cons, sq->hwq.max_elements); - goto do_rq; - } + cqe_cons %= sq->max_wqe; if (qp->sq.flushed) { dev_dbg(&cq->hwq.pdev->dev, @@ -2717,24 +2751,25 @@ static int bnxt_qplib_cq_process_terminal(struct bnxt_qplib_cq *cq, */ cqe = *pcqe; while (*budget) { - sw_cons = HWQ_CMP(sq->hwq.cons, &sq->hwq); - if (sw_cons == cqe_cons) + swq_last = sq->swq_last; + if (swq_last == cqe_cons) break; - if (sq->swq[sw_cons].flags & SQ_SEND_FLAGS_SIGNAL_COMP) { + if (sq->swq[swq_last].flags & SQ_SEND_FLAGS_SIGNAL_COMP) { memset(cqe, 0, sizeof(*cqe)); cqe->status = CQ_REQ_STATUS_OK; cqe->opcode = CQ_BASE_CQE_TYPE_REQ; cqe->qp_handle = (u64)(unsigned long)qp; cqe->src_qp = qp->id; - cqe->wr_id = sq->swq[sw_cons].wr_id; - cqe->type = sq->swq[sw_cons].type; + cqe->wr_id = sq->swq[swq_last].wr_id; + cqe->type = sq->swq[swq_last].type; cqe++; (*budget)--; } - sq->hwq.cons++; + bnxt_qplib_hwq_incr_cons(&sq->hwq, sq->swq[swq_last].slots); + sq->swq_last = sq->swq[swq_last].next_idx; } *pcqe = cqe; - if (!(*budget) && sw_cons != cqe_cons) { + if (!(*budget) && swq_last != cqe_cons) { /* Out of budget */ rc = -EAGAIN; goto sq_done; @@ -2746,10 +2781,10 @@ do_rq: cqe_cons = le16_to_cpu(hwcqe->rq_cons_idx); if (cqe_cons == 0xFFFF) { goto done; - } else if (cqe_cons > rq->hwq.max_elements) { + } else if (cqe_cons > rq->max_wqe - 1) { dev_err(&cq->hwq.pdev->dev, "FP: CQ Processed terminal reported rq_cons_idx 0x%x exceeds max 0x%x\n", - cqe_cons, rq->hwq.max_elements); + cqe_cons, rq->max_wqe); goto done; } @@ -2813,7 +2848,7 @@ int bnxt_qplib_process_flush_list(struct bnxt_qplib_cq *cq, int bnxt_qplib_poll_cq(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe *cqe, int num_cqes, struct bnxt_qplib_qp **lib_qp) { - struct cq_base *hw_cqe, **hw_cqe_ptr; + struct cq_base *hw_cqe; u32 sw_cons, raw_cons; int budget, rc = 0; @@ -2822,8 +2857,7 @@ int bnxt_qplib_poll_cq(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe *cqe, while (budget) { sw_cons = HWQ_CMP(raw_cons, &cq->hwq); - hw_cqe_ptr = (struct cq_base **)cq->hwq.pbl_ptr; - hw_cqe = &hw_cqe_ptr[CQE_PG(sw_cons)][CQE_IDX(sw_cons)]; + hw_cqe = bnxt_qplib_get_qe(&cq->hwq, sw_cons, NULL); /* Check for Valid bit */ if (!CQE_CMP_VALID(hw_cqe, raw_cons, cq->hwq.max_elements)) diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.h b/drivers/infiniband/hw/bnxt_re/qplib_fp.h index 7edb70b6bb16..f50784405e27 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_fp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.h @@ -39,12 +39,58 @@ #ifndef __BNXT_QPLIB_FP_H__ #define __BNXT_QPLIB_FP_H__ +/* Few helper structures temporarily defined here + * should get rid of these when roce_hsi.h is updated + * in original code base + */ +struct sq_ud_ext_hdr { + __le32 dst_qp; + __le32 avid; + __le64 rsvd; +}; + +struct sq_raw_ext_hdr { + __le32 cfa_meta; + __le32 rsvd0; + __le64 rsvd1; +}; + +struct sq_rdma_ext_hdr { + __le64 remote_va; + __le32 remote_key; + __le32 rsvd; +}; + +struct sq_atomic_ext_hdr { + __le64 swap_data; + __le64 cmp_data; +}; + +struct sq_fr_pmr_ext_hdr { + __le64 pblptr; + __le64 va; +}; + +struct sq_bind_ext_hdr { + __le64 va; + __le32 length_lo; + __le32 length_hi; +}; + +struct rq_ext_hdr { + __le64 rsvd1; + __le64 rsvd2; +}; + +/* Helper structures end */ + struct bnxt_qplib_srq { struct bnxt_qplib_pd *pd; struct bnxt_qplib_dpi *dpi; struct bnxt_qplib_db_info dbinfo; u64 srq_handle; u32 id; + u16 wqe_size; u32 max_wqe; u32 max_sge; u32 threshold; @@ -65,38 +111,7 @@ struct bnxt_qplib_sge { u32 size; }; -#define BNXT_QPLIB_MAX_SQE_ENTRY_SIZE sizeof(struct sq_send) - -#define SQE_CNT_PER_PG (PAGE_SIZE / BNXT_QPLIB_MAX_SQE_ENTRY_SIZE) -#define SQE_MAX_IDX_PER_PG (SQE_CNT_PER_PG - 1) - -static inline u32 get_sqe_pg(u32 val) -{ - return ((val & ~SQE_MAX_IDX_PER_PG) / SQE_CNT_PER_PG); -} - -static inline u32 get_sqe_idx(u32 val) -{ - return (val & SQE_MAX_IDX_PER_PG); -} - -#define BNXT_QPLIB_MAX_PSNE_ENTRY_SIZE sizeof(struct sq_psn_search) - -#define PSNE_CNT_PER_PG (PAGE_SIZE / BNXT_QPLIB_MAX_PSNE_ENTRY_SIZE) -#define PSNE_MAX_IDX_PER_PG (PSNE_CNT_PER_PG - 1) - -static inline u32 get_psne_pg(u32 val) -{ - return ((val & ~PSNE_MAX_IDX_PER_PG) / PSNE_CNT_PER_PG); -} - -static inline u32 get_psne_idx(u32 val) -{ - return (val & PSNE_MAX_IDX_PER_PG); -} - #define BNXT_QPLIB_QP_MAX_SGL 6 - struct bnxt_qplib_swq { u64 wr_id; int next_idx; @@ -104,6 +119,8 @@ struct bnxt_qplib_swq { u8 flags; u32 start_psn; u32 next_psn; + u32 slot_idx; + u8 slots; struct sq_psn_search *psn_search; struct sq_psn_search_ext *psn_ext; }; @@ -226,19 +243,13 @@ struct bnxt_qplib_swqe { }; }; -#define BNXT_QPLIB_MAX_RQE_ENTRY_SIZE sizeof(struct rq_wqe) - -#define RQE_CNT_PER_PG (PAGE_SIZE / BNXT_QPLIB_MAX_RQE_ENTRY_SIZE) -#define RQE_MAX_IDX_PER_PG (RQE_CNT_PER_PG - 1) -#define RQE_PG(x) (((x) & ~RQE_MAX_IDX_PER_PG) / RQE_CNT_PER_PG) -#define RQE_IDX(x) ((x) & RQE_MAX_IDX_PER_PG) - struct bnxt_qplib_q { struct bnxt_qplib_hwq hwq; struct bnxt_qplib_swq *swq; struct bnxt_qplib_db_info dbinfo; struct bnxt_qplib_sg_info sg_info; u32 max_wqe; + u16 wqe_size; u16 q_full_delta; u16 max_sge; u32 psn; @@ -249,6 +260,8 @@ struct bnxt_qplib_q { u32 phantom_cqe_cnt; u32 next_cq_cons; bool flushed; + u32 swq_start; + u32 swq_last; }; struct bnxt_qplib_qp { @@ -256,13 +269,14 @@ struct bnxt_qplib_qp { struct bnxt_qplib_dpi *dpi; struct bnxt_qplib_chip_ctx *cctx; u64 qp_handle; -#define BNXT_QPLIB_QP_ID_INVALID 0xFFFFFFFF +#define BNXT_QPLIB_QP_ID_INVALID 0xFFFFFFFF u32 id; u8 type; u8 sig_type; - u32 modify_flags; + u8 wqe_mode; u8 state; u8 cur_qp_state; + u64 modify_flags; u32 max_inline_data; u32 mtu; u8 path_mtu; @@ -336,11 +350,18 @@ struct bnxt_qplib_qp { (!!((hdr)->cqe_type_toggle & CQ_BASE_TOGGLE) == \ !((raw_cons) & (cp_bit))) -static inline bool bnxt_qplib_queue_full(struct bnxt_qplib_q *qplib_q) +static inline bool bnxt_qplib_queue_full(struct bnxt_qplib_q *que, + u8 slots) { - return HWQ_CMP((qplib_q->hwq.prod + qplib_q->q_full_delta), - &qplib_q->hwq) == HWQ_CMP(qplib_q->hwq.cons, - &qplib_q->hwq); + struct bnxt_qplib_hwq *hwq; + int avail; + + hwq = &que->hwq; + /* False full is possible, retrying post-send makes sense */ + avail = hwq->cons - hwq->prod; + if (hwq->cons <= hwq->prod) + avail += hwq->depth; + return avail <= slots; } struct bnxt_qplib_cqe { @@ -525,4 +546,64 @@ int bnxt_qplib_process_flush_list(struct bnxt_qplib_cq *cq, struct bnxt_qplib_cqe *cqe, int num_cqes); void bnxt_qplib_flush_cqn_wq(struct bnxt_qplib_qp *qp); + +static inline void *bnxt_qplib_get_swqe(struct bnxt_qplib_q *que, u32 *swq_idx) +{ + u32 idx; + + idx = que->swq_start; + if (swq_idx) + *swq_idx = idx; + return &que->swq[idx]; +} + +static inline void bnxt_qplib_swq_mod_start(struct bnxt_qplib_q *que, u32 idx) +{ + que->swq_start = que->swq[idx].next_idx; +} + +static inline u32 bnxt_qplib_get_depth(struct bnxt_qplib_q *que) +{ + return (que->wqe_size * que->max_wqe) / sizeof(struct sq_sge); +} + +static inline u32 bnxt_qplib_set_sq_size(struct bnxt_qplib_q *que, u8 wqe_mode) +{ + return (wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) ? + que->max_wqe : bnxt_qplib_get_depth(que); +} + +static inline u32 bnxt_qplib_set_sq_max_slot(u8 wqe_mode) +{ + return (wqe_mode == BNXT_QPLIB_WQE_MODE_STATIC) ? + sizeof(struct sq_send) / sizeof(struct sq_sge) : 1; +} + +static inline u32 bnxt_qplib_set_rq_max_slot(u32 wqe_size) +{ + return (wqe_size / sizeof(struct sq_sge)); +} + +static inline u16 __xlate_qfd(u16 delta, u16 wqe_bytes) +{ + /* For Cu/Wh delta = 128, stride = 16, wqe_bytes = 128 + * For Gen-p5 B/C mode delta = 0, stride = 16, wqe_bytes = 128. + * For Gen-p5 delta = 0, stride = 16, 32 <= wqe_bytes <= 512. + * when 8916 is disabled. + */ + return (delta * wqe_bytes) / sizeof(struct sq_sge); +} + +static inline u16 bnxt_qplib_calc_ilsize(struct bnxt_qplib_swqe *wqe, u16 max) +{ + u16 size = 0; + int indx; + + for (indx = 0; indx < wqe->num_sge; indx++) + size += wqe->sg_list[indx].size; + if (size > max) + size = max; + + return size; +} #endif /* __BNXT_QPLIB_FP_H__ */ diff --git a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c index f01e864bb611..f7736e34ac64 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c @@ -89,10 +89,9 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req, struct creq_base *resp, void *sb, u8 is_block) { struct bnxt_qplib_cmdq_ctx *cmdq = &rcfw->cmdq; - struct bnxt_qplib_cmdqe *cmdqe, **hwq_ptr; struct bnxt_qplib_hwq *hwq = &cmdq->hwq; struct bnxt_qplib_crsqe *crsqe; - u32 cmdq_depth = rcfw->cmdq_depth; + struct bnxt_qplib_cmdqe *cmdqe; u32 sw_prod, cmdq_prod; struct pci_dev *pdev; unsigned long flags; @@ -163,13 +162,11 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req, BNXT_QPLIB_CMDQE_UNITS; } - hwq_ptr = (struct bnxt_qplib_cmdqe **)hwq->pbl_ptr; preq = (u8 *)req; do { /* Locate the next cmdq slot */ sw_prod = HWQ_CMP(hwq->prod, hwq); - cmdqe = &hwq_ptr[get_cmdq_pg(sw_prod, cmdq_depth)] - [get_cmdq_idx(sw_prod, cmdq_depth)]; + cmdqe = bnxt_qplib_get_qe(hwq, sw_prod, NULL); if (!cmdqe) { dev_err(&pdev->dev, "RCFW request failed with no cmdqe!\n"); @@ -310,14 +307,15 @@ static int bnxt_qplib_process_qp_event(struct bnxt_qplib_rcfw *rcfw, __le16 mcookie; u16 cookie; int rc = 0; - u32 qp_id; + u32 qp_id, tbl_indx; pdev = rcfw->pdev; switch (qp_event->event) { case CREQ_QP_EVENT_EVENT_QP_ERROR_NOTIFICATION: err_event = (struct creq_qp_error_notification *)qp_event; qp_id = le32_to_cpu(err_event->xid); - qp = rcfw->qp_tbl[qp_id].qp_handle; + tbl_indx = map_qp_id_to_tbl_indx(qp_id, rcfw); + qp = rcfw->qp_tbl[tbl_indx].qp_handle; dev_dbg(&pdev->dev, "Received QP error notification\n"); dev_dbg(&pdev->dev, "qpid 0x%x, req_err=0x%x, resp_err=0x%x\n", @@ -378,7 +376,7 @@ static void bnxt_qplib_service_creq(unsigned long data) struct bnxt_qplib_creq_ctx *creq = &rcfw->creq; u32 type, budget = CREQ_ENTRY_POLL_BUDGET; struct bnxt_qplib_hwq *hwq = &creq->hwq; - struct creq_base *creqe, **hwq_ptr; + struct creq_base *creqe; u32 sw_cons, raw_cons; unsigned long flags; @@ -387,8 +385,7 @@ static void bnxt_qplib_service_creq(unsigned long data) raw_cons = hwq->cons; while (budget > 0) { sw_cons = HWQ_CMP(raw_cons, hwq); - hwq_ptr = (struct creq_base **)hwq->pbl_ptr; - creqe = &hwq_ptr[get_creq_pg(sw_cons)][get_creq_idx(sw_cons)]; + creqe = bnxt_qplib_get_qe(hwq, sw_cons, NULL); if (!CREQ_CMP_VALID(creqe, raw_cons, hwq->max_elements)) break; /* The valid test of the entry must be done first before @@ -434,7 +431,6 @@ static irqreturn_t bnxt_qplib_creq_irq(int irq, void *dev_instance) { struct bnxt_qplib_rcfw *rcfw = dev_instance; struct bnxt_qplib_creq_ctx *creq; - struct creq_base **creq_ptr; struct bnxt_qplib_hwq *hwq; u32 sw_cons; @@ -442,8 +438,7 @@ static irqreturn_t bnxt_qplib_creq_irq(int irq, void *dev_instance) hwq = &creq->hwq; /* Prefetch the CREQ element */ sw_cons = HWQ_CMP(hwq->cons, hwq); - creq_ptr = (struct creq_base **)creq->hwq.pbl_ptr; - prefetch(&creq_ptr[get_creq_pg(sw_cons)][get_creq_idx(sw_cons)]); + prefetch(bnxt_qplib_get_qe(hwq, sw_cons, NULL)); tasklet_schedule(&creq->creq_tasklet); @@ -468,29 +463,13 @@ int bnxt_qplib_deinit_rcfw(struct bnxt_qplib_rcfw *rcfw) return 0; } -static int __get_pbl_pg_idx(struct bnxt_qplib_pbl *pbl) -{ - return (pbl->pg_size == ROCE_PG_SIZE_4K ? - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_4K : - pbl->pg_size == ROCE_PG_SIZE_8K ? - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_8K : - pbl->pg_size == ROCE_PG_SIZE_64K ? - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_64K : - pbl->pg_size == ROCE_PG_SIZE_2M ? - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_2M : - pbl->pg_size == ROCE_PG_SIZE_8M ? - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_8M : - pbl->pg_size == ROCE_PG_SIZE_1G ? - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_1G : - CMDQ_INITIALIZE_FW_QPC_PG_SIZE_PG_4K); -} - int bnxt_qplib_init_rcfw(struct bnxt_qplib_rcfw *rcfw, struct bnxt_qplib_ctx *ctx, int is_virtfn) { - struct cmdq_initialize_fw req; struct creq_initialize_fw_resp resp; - u16 cmd_flags = 0, level; + struct cmdq_initialize_fw req; + u16 cmd_flags = 0; + u8 pgsz, lvl; int rc; RCFW_CMD_PREP(req, INITIALIZE_FW, cmd_flags); @@ -511,32 +490,30 @@ int bnxt_qplib_init_rcfw(struct bnxt_qplib_rcfw *rcfw, if (bnxt_qplib_is_chip_gen_p5(rcfw->res->cctx)) goto config_vf_res; - level = ctx->qpc_tbl.level; - req.qpc_pg_size_qpc_lvl = (level << CMDQ_INITIALIZE_FW_QPC_LVL_SFT) | - __get_pbl_pg_idx(&ctx->qpc_tbl.pbl[level]); - level = ctx->mrw_tbl.level; - req.mrw_pg_size_mrw_lvl = (level << CMDQ_INITIALIZE_FW_MRW_LVL_SFT) | - __get_pbl_pg_idx(&ctx->mrw_tbl.pbl[level]); - level = ctx->srqc_tbl.level; - req.srq_pg_size_srq_lvl = (level << CMDQ_INITIALIZE_FW_SRQ_LVL_SFT) | - __get_pbl_pg_idx(&ctx->srqc_tbl.pbl[level]); - level = ctx->cq_tbl.level; - req.cq_pg_size_cq_lvl = (level << CMDQ_INITIALIZE_FW_CQ_LVL_SFT) | - __get_pbl_pg_idx(&ctx->cq_tbl.pbl[level]); - level = ctx->srqc_tbl.level; - req.srq_pg_size_srq_lvl = (level << CMDQ_INITIALIZE_FW_SRQ_LVL_SFT) | - __get_pbl_pg_idx(&ctx->srqc_tbl.pbl[level]); - level = ctx->cq_tbl.level; - req.cq_pg_size_cq_lvl = (level << CMDQ_INITIALIZE_FW_CQ_LVL_SFT) | - __get_pbl_pg_idx(&ctx->cq_tbl.pbl[level]); - level = ctx->tim_tbl.level; - req.tim_pg_size_tim_lvl = (level << CMDQ_INITIALIZE_FW_TIM_LVL_SFT) | - __get_pbl_pg_idx(&ctx->tim_tbl.pbl[level]); - level = ctx->tqm_ctx.pde.level; - req.tqm_pg_size_tqm_lvl = - (level << CMDQ_INITIALIZE_FW_TQM_LVL_SFT) | - __get_pbl_pg_idx(&ctx->tqm_ctx.pde.pbl[level]); - + lvl = ctx->qpc_tbl.level; + pgsz = bnxt_qplib_base_pg_size(&ctx->qpc_tbl); + req.qpc_pg_size_qpc_lvl = (pgsz << CMDQ_INITIALIZE_FW_QPC_PG_SIZE_SFT) | + lvl; + lvl = ctx->mrw_tbl.level; + pgsz = bnxt_qplib_base_pg_size(&ctx->mrw_tbl); + req.mrw_pg_size_mrw_lvl = (pgsz << CMDQ_INITIALIZE_FW_QPC_PG_SIZE_SFT) | + lvl; + lvl = ctx->srqc_tbl.level; + pgsz = bnxt_qplib_base_pg_size(&ctx->srqc_tbl); + req.srq_pg_size_srq_lvl = (pgsz << CMDQ_INITIALIZE_FW_QPC_PG_SIZE_SFT) | + lvl; + lvl = ctx->cq_tbl.level; + pgsz = bnxt_qplib_base_pg_size(&ctx->cq_tbl); + req.cq_pg_size_cq_lvl = (pgsz << CMDQ_INITIALIZE_FW_QPC_PG_SIZE_SFT) | + lvl; + lvl = ctx->tim_tbl.level; + pgsz = bnxt_qplib_base_pg_size(&ctx->tim_tbl); + req.tim_pg_size_tim_lvl = (pgsz << CMDQ_INITIALIZE_FW_QPC_PG_SIZE_SFT) | + lvl; + lvl = ctx->tqm_ctx.pde.level; + pgsz = bnxt_qplib_base_pg_size(&ctx->tqm_ctx.pde); + req.tqm_pg_size_tqm_lvl = (pgsz << CMDQ_INITIALIZE_FW_QPC_PG_SIZE_SFT) | + lvl; req.qpc_page_dir = cpu_to_le64(ctx->qpc_tbl.pbl[PBL_LVL_0].pg_map_arr[0]); req.mrw_page_dir = @@ -639,8 +616,9 @@ int bnxt_qplib_alloc_rcfw_channel(struct bnxt_qplib_res *res, cmdq->bmap_size = bmap_size; - rcfw->qp_tbl_size = qp_tbl_sz; - rcfw->qp_tbl = kcalloc(qp_tbl_sz, sizeof(struct bnxt_qplib_qp_node), + /* Allocate one extra to hold the QP1 entries */ + rcfw->qp_tbl_size = qp_tbl_sz + 1; + rcfw->qp_tbl = kcalloc(rcfw->qp_tbl_size, sizeof(struct bnxt_qplib_qp_node), GFP_KERNEL); if (!rcfw->qp_tbl) goto fail; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h index 411fce3493b6..5f2f0a5a3560 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.h @@ -87,12 +87,6 @@ static inline u32 bnxt_qplib_cmdqe_page_size(u32 depth) return (bnxt_qplib_cmdqe_npages(depth) * PAGE_SIZE); } -static inline u32 bnxt_qplib_cmdqe_cnt_per_pg(u32 depth) -{ - return (bnxt_qplib_cmdqe_page_size(depth) / - BNXT_QPLIB_CMDQE_UNITS); -} - /* Set the cmd_size to a factor of CMDQE unit */ static inline void bnxt_qplib_set_cmd_slots(struct cmdq_base *req) { @@ -100,30 +94,12 @@ static inline void bnxt_qplib_set_cmd_slots(struct cmdq_base *req) BNXT_QPLIB_CMDQE_UNITS; } -#define MAX_CMDQ_IDX(depth) ((depth) - 1) - -static inline u32 bnxt_qplib_max_cmdq_idx_per_pg(u32 depth) -{ - return (bnxt_qplib_cmdqe_cnt_per_pg(depth) - 1); -} - #define RCFW_MAX_COOKIE_VALUE 0x7FFF #define RCFW_CMD_IS_BLOCKING 0x8000 #define RCFW_BLOCKED_CMD_WAIT_COUNT 0x4E20 #define HWRM_VERSION_RCFW_CMDQ_DEPTH_CHECK 0x1000900020011ULL -static inline u32 get_cmdq_pg(u32 val, u32 depth) -{ - return (val & ~(bnxt_qplib_max_cmdq_idx_per_pg(depth))) / - (bnxt_qplib_cmdqe_cnt_per_pg(depth)); -} - -static inline u32 get_cmdq_idx(u32 val, u32 depth) -{ - return val & (bnxt_qplib_max_cmdq_idx_per_pg(depth)); -} - /* Crsq buf is 1024-Byte */ struct bnxt_qplib_crsbe { u8 data[1024]; @@ -133,76 +109,9 @@ struct bnxt_qplib_crsbe { /* Allocate 1 per QP for async error notification for now */ #define BNXT_QPLIB_CREQE_MAX_CNT (64 * 1024) #define BNXT_QPLIB_CREQE_UNITS 16 /* 16-Bytes per prod unit */ -#define BNXT_QPLIB_CREQE_CNT_PER_PG (PAGE_SIZE / BNXT_QPLIB_CREQE_UNITS) - -#define MAX_CREQ_IDX (BNXT_QPLIB_CREQE_MAX_CNT - 1) -#define MAX_CREQ_IDX_PER_PG (BNXT_QPLIB_CREQE_CNT_PER_PG - 1) - -static inline u32 get_creq_pg(u32 val) -{ - return (val & ~MAX_CREQ_IDX_PER_PG) / BNXT_QPLIB_CREQE_CNT_PER_PG; -} - -static inline u32 get_creq_idx(u32 val) -{ - return val & MAX_CREQ_IDX_PER_PG; -} - -#define BNXT_QPLIB_CREQE_PER_PG (PAGE_SIZE / sizeof(struct creq_base)) - #define CREQ_CMP_VALID(hdr, raw_cons, cp_bit) \ (!!((hdr)->v & CREQ_BASE_V) == \ !((raw_cons) & (cp_bit))) - -#define CREQ_DB_KEY_CP (0x2 << CMPL_DOORBELL_KEY_SFT) -#define CREQ_DB_IDX_VALID CMPL_DOORBELL_IDX_VALID -#define CREQ_DB_IRQ_DIS CMPL_DOORBELL_MASK -#define CREQ_DB_CP_FLAGS_REARM (CREQ_DB_KEY_CP | \ - CREQ_DB_IDX_VALID) -#define CREQ_DB_CP_FLAGS (CREQ_DB_KEY_CP | \ - CREQ_DB_IDX_VALID | \ - CREQ_DB_IRQ_DIS) - -static inline void bnxt_qplib_ring_creq_db64(void __iomem *db, u32 index, - u32 xid, bool arm) -{ - u64 val = 0; - - val = xid & DBC_DBC_XID_MASK; - val |= DBC_DBC_PATH_ROCE; - val |= arm ? DBC_DBC_TYPE_NQ_ARM : DBC_DBC_TYPE_NQ; - val <<= 32; - val |= index & DBC_DBC_INDEX_MASK; - - writeq(val, db); -} - -static inline void bnxt_qplib_ring_creq_db_rearm(void __iomem *db, u32 raw_cons, - u32 max_elements, u32 xid, - bool gen_p5) -{ - u32 index = raw_cons & (max_elements - 1); - - if (gen_p5) - bnxt_qplib_ring_creq_db64(db, index, xid, true); - else - writel(CREQ_DB_CP_FLAGS_REARM | (index & DBC_DBC32_XID_MASK), - db); -} - -static inline void bnxt_qplib_ring_creq_db(void __iomem *db, u32 raw_cons, - u32 max_elements, u32 xid, - bool gen_p5) -{ - u32 index = raw_cons & (max_elements - 1); - - if (gen_p5) - bnxt_qplib_ring_creq_db64(db, index, xid, true); - else - writel(CREQ_DB_CP_FLAGS | (index & DBC_DBC32_XID_MASK), - db); -} - #define CREQ_ENTRY_POLL_BUDGET 0x100 /* HWQ */ @@ -307,4 +216,9 @@ int bnxt_qplib_deinit_rcfw(struct bnxt_qplib_rcfw *rcfw); int bnxt_qplib_init_rcfw(struct bnxt_qplib_rcfw *rcfw, struct bnxt_qplib_ctx *ctx, int is_virtfn); void bnxt_qplib_mark_qp_error(void *qp_handle); +static inline u32 map_qp_id_to_tbl_indx(u32 qid, struct bnxt_qplib_rcfw *rcfw) +{ + /* Last index of the qp_tbl is for QP1 ie. qp_tbl_size - 1*/ + return (qid == 1) ? rcfw->qp_tbl_size - 1 : qid % rcfw->qp_tbl_size - 2; +} #endif /* __BNXT_QPLIB_RCFW_H__ */ diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.c b/drivers/infiniband/hw/bnxt_re/qplib_res.c index cab1adf1fed9..7efa6e5dce62 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.c @@ -347,6 +347,7 @@ done: hwq->depth = hwq_attr->depth; hwq->max_elements = depth; hwq->element_size = stride; + hwq->qe_ppg = pg_size / stride; /* For direct access to the elements */ lvl = hwq->level; if (hwq_attr->sginfo->nopte && hwq->level) diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.h b/drivers/infiniband/hw/bnxt_re/qplib_res.h index 95b645dbbc2d..9da470d1e4a3 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.h @@ -41,6 +41,28 @@ extern const struct bnxt_qplib_gid bnxt_qplib_gid_zero; +#define CHIP_NUM_57508 0x1750 +#define CHIP_NUM_57504 0x1751 +#define CHIP_NUM_57502 0x1752 + +enum bnxt_qplib_wqe_mode { + BNXT_QPLIB_WQE_MODE_STATIC = 0x00, + BNXT_QPLIB_WQE_MODE_VARIABLE = 0x01, + BNXT_QPLIB_WQE_MODE_INVALID = 0x02 +}; + +struct bnxt_qplib_drv_modes { + u8 wqe_mode; + /* Other modes to follow here */ +}; + +struct bnxt_qplib_chip_ctx { + u16 chip_num; + u8 chip_rev; + u8 chip_metal; + struct bnxt_qplib_drv_modes modes; +}; + #define PTR_CNT_PER_PG (PAGE_SIZE / sizeof(void *)) #define PTR_MAX_IDX_PER_PG (PTR_CNT_PER_PG - 1) #define PTR_PG(x) (((x) & ~PTR_MAX_IDX_PER_PG) / PTR_CNT_PER_PG) @@ -80,6 +102,15 @@ enum bnxt_qplib_pbl_lvl { #define ROCE_PG_SIZE_8M (8 * 1024 * 1024) #define ROCE_PG_SIZE_1G (1024 * 1024 * 1024) +enum bnxt_qplib_hwrm_pg_size { + BNXT_QPLIB_HWRM_PG_SIZE_4K = 0, + BNXT_QPLIB_HWRM_PG_SIZE_8K = 1, + BNXT_QPLIB_HWRM_PG_SIZE_64K = 2, + BNXT_QPLIB_HWRM_PG_SIZE_2M = 3, + BNXT_QPLIB_HWRM_PG_SIZE_8M = 4, + BNXT_QPLIB_HWRM_PG_SIZE_1G = 5, +}; + struct bnxt_qplib_reg_desc { u8 bar_id; resource_size_t bar_base; @@ -126,11 +157,15 @@ struct bnxt_qplib_hwq { u32 max_elements; u32 depth; u16 element_size; /* Size of each entry */ + u16 qe_ppg; /* queue entry per page */ u32 prod; /* raw */ u32 cons; /* raw */ u8 cp_bit; u8 is_user; + u64 *pad_pg; + u32 pad_stride; + u32 pad_pgofft; }; struct bnxt_qplib_db_info { @@ -138,6 +173,7 @@ struct bnxt_qplib_db_info { void __iomem *priv_db; struct bnxt_qplib_hwq *hwq; u32 xid; + u32 max_slot; }; /* Tables */ @@ -220,16 +256,6 @@ struct bnxt_qplib_ctx { u64 hwrm_intf_ver; }; -struct bnxt_qplib_chip_ctx { - u16 chip_num; - u8 chip_rev; - u8 chip_metal; -}; - -#define CHIP_NUM_57508 0x1750 -#define CHIP_NUM_57504 0x1751 -#define CHIP_NUM_57502 0x1752 - struct bnxt_qplib_res { struct pci_dev *pdev; struct bnxt_qplib_chip_ctx *cctx; @@ -263,6 +289,57 @@ static inline u8 bnxt_qplib_get_ring_type(struct bnxt_qplib_chip_ctx *cctx) RING_ALLOC_REQ_RING_TYPE_ROCE_CMPL; } +static inline u8 bnxt_qplib_base_pg_size(struct bnxt_qplib_hwq *hwq) +{ + u8 pg_size = BNXT_QPLIB_HWRM_PG_SIZE_4K; + struct bnxt_qplib_pbl *pbl; + + pbl = &hwq->pbl[PBL_LVL_0]; + switch (pbl->pg_size) { + case ROCE_PG_SIZE_4K: + pg_size = BNXT_QPLIB_HWRM_PG_SIZE_4K; + break; + case ROCE_PG_SIZE_8K: + pg_size = BNXT_QPLIB_HWRM_PG_SIZE_8K; + break; + case ROCE_PG_SIZE_64K: + pg_size = BNXT_QPLIB_HWRM_PG_SIZE_64K; + break; + case ROCE_PG_SIZE_2M: + pg_size = BNXT_QPLIB_HWRM_PG_SIZE_2M; + break; + case ROCE_PG_SIZE_8M: + pg_size = BNXT_QPLIB_HWRM_PG_SIZE_8M; + break; + case ROCE_PG_SIZE_1G: + pg_size = BNXT_QPLIB_HWRM_PG_SIZE_1G; + break; + default: + break; + } + + return pg_size; +} + +static inline void *bnxt_qplib_get_qe(struct bnxt_qplib_hwq *hwq, + u32 indx, u64 *pg) +{ + u32 pg_num, pg_idx; + + pg_num = (indx / hwq->qe_ppg); + pg_idx = (indx % hwq->qe_ppg); + if (pg) + *pg = (u64)&hwq->pbl_ptr[pg_num]; + return (void *)(hwq->pbl_ptr[pg_num] + hwq->element_size * pg_idx); +} + +static inline void *bnxt_qplib_get_prod_qe(struct bnxt_qplib_hwq *hwq, u32 idx) +{ + idx += hwq->prod; + if (idx >= hwq->depth) + idx -= hwq->depth; + return bnxt_qplib_get_qe(hwq, idx, NULL); +} #define to_bnxt_qplib(ptr, type, member) \ container_of(ptr, type, member) @@ -298,6 +375,17 @@ int bnxt_qplib_alloc_ctx(struct bnxt_qplib_res *res, struct bnxt_qplib_ctx *ctx, bool virt_fn, bool is_p5); +static inline void bnxt_qplib_hwq_incr_prod(struct bnxt_qplib_hwq *hwq, u32 cnt) +{ + hwq->prod = (hwq->prod + cnt) % hwq->depth; +} + +static inline void bnxt_qplib_hwq_incr_cons(struct bnxt_qplib_hwq *hwq, + u32 cnt) +{ + hwq->cons = (hwq->cons + cnt) % hwq->depth; +} + static inline void bnxt_qplib_ring_db32(struct bnxt_qplib_db_info *info, bool arm) { @@ -330,8 +418,7 @@ static inline void bnxt_qplib_ring_prod_db(struct bnxt_qplib_db_info *info, key = (info->xid & DBC_DBC_XID_MASK) | DBC_DBC_PATH_ROCE | type; key <<= 32; - key |= (info->hwq->prod & (info->hwq->max_elements - 1)) & - DBC_DBC_INDEX_MASK; + key |= ((info->hwq->prod / info->max_slot)) & DBC_DBC_INDEX_MASK; writeq(key, info->db); } diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c index 66954ff6a2f2..64d44f51db4b 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c @@ -132,9 +132,6 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw, attr->max_raw_ethy_qp = le32_to_cpu(sb->max_raw_eth_qp); attr->max_ah = le32_to_cpu(sb->max_ah); - attr->max_fmr = le32_to_cpu(sb->max_fmr); - attr->max_map_per_fmr = sb->max_map_per_fmr; - attr->max_srq = le16_to_cpu(sb->max_srq); attr->max_srq_wqes = le32_to_cpu(sb->max_srq_wr) - 1; attr->max_srq_sges = sb->max_srq_sge; @@ -152,7 +149,7 @@ int bnxt_qplib_get_dev_attr(struct bnxt_qplib_rcfw *rcfw, attr->max_inline_data = le32_to_cpu(sb->max_inline_data); attr->l2_db_size = (sb->l2_db_space_size + 1) * (0x01 << RCFW_DBR_BASE_PAGE_SHIFT); - attr->max_sgid = le32_to_cpu(sb->max_gid); + attr->max_sgid = BNXT_QPLIB_NUM_GIDS_SUPPORTED; bnxt_qplib_query_version(rcfw, attr->fw_ver); diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h index 13d9432d5ce2..967890cd81f2 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h @@ -47,6 +47,7 @@ struct bnxt_qplib_dev_attr { #define FW_VER_ARR_LEN 4 u8 fw_ver[FW_VER_ARR_LEN]; +#define BNXT_QPLIB_NUM_GIDS_SUPPORTED 256 u16 max_sgid; u16 max_mrw; u32 max_qp; @@ -64,8 +65,6 @@ struct bnxt_qplib_dev_attr { u32 max_mw; u32 max_raw_ethy_qp; u32 max_ah; - u32 max_fmr; - u32 max_map_per_fmr; u32 max_srq; u32 max_srq_wqes; u32 max_srq_sges; diff --git a/drivers/infiniband/hw/bnxt_re/roce_hsi.h b/drivers/infiniband/hw/bnxt_re/roce_hsi.h index e4b09e7c2175..3e40e0d76efd 100644 --- a/drivers/infiniband/hw/bnxt_re/roce_hsi.h +++ b/drivers/infiniband/hw/bnxt_re/roce_hsi.h @@ -210,6 +210,20 @@ struct sq_send { __le32 data[24]; }; +/* sq_send_hdr (size:256b/32B) */ +struct sq_send_hdr { + u8 wqe_type; + u8 flags; + u8 wqe_size; + u8 reserved8_1; + __le32 inv_key_or_imm_data; + __le32 length; + __le32 q_key; + __le32 dst_qp; + __le32 avid; + __le64 reserved64; +}; + /* Send Raw Ethernet and QP1 SQ WQE (40 bytes) */ struct sq_send_raweth_qp1 { u8 wqe_type; @@ -265,6 +279,21 @@ struct sq_send_raweth_qp1 { __le32 data[24]; }; +/* sq_send_raweth_qp1_hdr (size:256b/32B) */ +struct sq_send_raweth_qp1_hdr { + u8 wqe_type; + u8 flags; + u8 wqe_size; + u8 reserved8; + __le16 lflags; + __le16 cfa_action; + __le32 length; + __le32 reserved32_1; + __le32 cfa_meta; + __le32 reserved32_2; + __le64 reserved64; +}; + /* RDMA SQ WQE (40 bytes) */ struct sq_rdma { u8 wqe_type; @@ -288,6 +317,20 @@ struct sq_rdma { __le32 data[24]; }; +/* sq_rdma_hdr (size:256b/32B) */ +struct sq_rdma_hdr { + u8 wqe_type; + u8 flags; + u8 wqe_size; + u8 reserved8; + __le32 imm_data; + __le32 length; + __le32 reserved32_1; + __le64 remote_va; + __le32 remote_key; + __le32 reserved32_2; +}; + /* Atomic SQ WQE (40 bytes) */ struct sq_atomic { u8 wqe_type; @@ -307,6 +350,17 @@ struct sq_atomic { __le32 data[24]; }; +/* sq_atomic_hdr (size:256b/32B) */ +struct sq_atomic_hdr { + u8 wqe_type; + u8 flags; + __le16 reserved16; + __le32 remote_key; + __le64 remote_va; + __le64 swap_data; + __le64 cmp_data; +}; + /* Local Invalidate SQ WQE (40 bytes) */ struct sq_localinvalidate { u8 wqe_type; @@ -324,6 +378,16 @@ struct sq_localinvalidate { __le32 data[24]; }; +/* sq_localinvalidate_hdr (size:256b/32B) */ +struct sq_localinvalidate_hdr { + u8 wqe_type; + u8 flags; + __le16 reserved16; + __le32 inv_l_key; + __le64 reserved64; + u8 reserved128[16]; +}; + /* FR-PMR SQ WQE (40 bytes) */ struct sq_fr_pmr { u8 wqe_type; @@ -380,6 +444,21 @@ struct sq_fr_pmr { __le32 data[24]; }; +/* sq_fr_pmr_hdr (size:256b/32B) */ +struct sq_fr_pmr_hdr { + u8 wqe_type; + u8 flags; + u8 access_cntl; + u8 zero_based_page_size_log; + __le32 l_key; + u8 length[5]; + u8 reserved8_1; + u8 reserved8_2; + u8 numlevels_pbl_page_size_log; + __le64 pblptr; + __le64 va; +}; + /* Bind SQ WQE (40 bytes) */ struct sq_bind { u8 wqe_type; @@ -417,6 +496,22 @@ struct sq_bind { #define SQ_BIND_DATA_SFT 0 }; +/* sq_bind_hdr (size:256b/32B) */ +struct sq_bind_hdr { + u8 wqe_type; + u8 flags; + u8 access_cntl; + u8 reserved8_1; + u8 mw_type_zero_based; + u8 reserved8_2; + __le16 reserved16; + __le32 parent_l_key; + __le32 l_key; + __le64 va; + u8 length[5]; + u8 reserved24[3]; +}; + /* RQ/SRQ WQE Structures */ /* RQ/SRQ WQE (40 bytes) */ struct rq_wqe { @@ -435,6 +530,17 @@ struct rq_wqe { __le32 data[24]; }; +/* rq_wqe_hdr (size:256b/32B) */ +struct rq_wqe_hdr { + u8 wqe_type; + u8 flags; + u8 wqe_size; + u8 reserved8; + __le32 reserved32; + __le32 wr_id[2]; + u8 reserved128[16]; +}; + /* CQ CQE Structures */ /* Base CQE (32 bytes) */ struct cq_base { @@ -1020,6 +1126,7 @@ struct cmdq_create_qp { #define CMDQ_CREATE_QP_QP_FLAGS_FORCE_COMPLETION 0x2UL #define CMDQ_CREATE_QP_QP_FLAGS_RESERVED_LKEY_ENABLE 0x4UL #define CMDQ_CREATE_QP_QP_FLAGS_FR_PMR_ENABLED 0x8UL + #define CMDQ_CREATE_QP_QP_FLAGS_VARIABLE_SIZED_WQE_ENABLED 0x10UL u8 type; #define CMDQ_CREATE_QP_TYPE_RC 0x2UL #define CMDQ_CREATE_QP_TYPE_UD 0x4UL diff --git a/drivers/infiniband/hw/cxgb4/Kconfig b/drivers/infiniband/hw/cxgb4/Kconfig index b49e8d4c3854..9e2b2c348afd 100644 --- a/drivers/infiniband/hw/cxgb4/Kconfig +++ b/drivers/infiniband/hw/cxgb4/Kconfig @@ -5,7 +5,7 @@ config INFINIBAND_CXGB4 depends on INFINIBAND_ADDR_TRANS select CHELSIO_LIB select GENERIC_ALLOCATOR - ---help--- + help This is an iWARP/RDMA driver for the Chelsio T4 and T5 1GbE, 10GbE adapters and T5 40GbE adapter. diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c index 30e08bcc9afb..1f288c73ccfc 100644 --- a/drivers/infiniband/hw/cxgb4/cm.c +++ b/drivers/infiniband/hw/cxgb4/cm.c @@ -2885,7 +2885,7 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) case MORIBUND: case CLOSING: stop_ep_timer(ep); - /*FALLTHROUGH*/ + fallthrough; case FPDU_MODE: if (ep->com.qp && ep->com.qp->srq) { srqidx = ABORT_RSS_SRQIDX_G( @@ -3282,7 +3282,7 @@ static int get_lladdr(struct net_device *dev, struct in6_addr *addr, static int pick_local_ip6addrs(struct c4iw_dev *dev, struct iw_cm_id *cm_id) { - struct in6_addr uninitialized_var(addr); + struct in6_addr addr; struct sockaddr_in6 *la6 = (struct sockaddr_in6 *)&cm_id->m_local_addr; struct sockaddr_in6 *ra6 = (struct sockaddr_in6 *)&cm_id->m_remote_addr; @@ -3759,7 +3759,7 @@ static void active_ofld_conn_reply(struct c4iw_dev *dev, struct sk_buff *skb, send_fw_act_open_req(ep, atid); return; } - /* fall through */ + fallthrough; case FW_EADDRINUSE: set_bit(ACT_RETRY_INUSE, &ep->com.history); if (ep->retry_count++ < ACT_OPEN_RETRY_COUNT) { diff --git a/drivers/infiniband/hw/cxgb4/cq.c b/drivers/infiniband/hw/cxgb4/cq.c index b1bb61c65f4f..352b8af1998a 100644 --- a/drivers/infiniband/hw/cxgb4/cq.c +++ b/drivers/infiniband/hw/cxgb4/cq.c @@ -754,7 +754,7 @@ skip_cqe: static int __c4iw_poll_cq_one(struct c4iw_cq *chp, struct c4iw_qp *qhp, struct ib_wc *wc, struct c4iw_srq *srq) { - struct t4_cqe uninitialized_var(cqe); + struct t4_cqe cqe; struct t4_wq *wq = qhp ? &qhp->wq : NULL; u32 credit = 0; u8 cqe_flushed; diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c index 599340c1f0b8..541dbcf22d0e 100644 --- a/drivers/infiniband/hw/cxgb4/device.c +++ b/drivers/infiniband/hw/cxgb4/device.c @@ -953,6 +953,7 @@ void c4iw_dealloc(struct uld_ctx *ctx) static void c4iw_remove(struct uld_ctx *ctx) { pr_debug("c4iw_dev %p\n", ctx->dev); + debugfs_remove_recursive(ctx->dev->debugfs_root); c4iw_unregister_device(ctx->dev); c4iw_dealloc(ctx); } diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h index e8e11bd95e42..2b2b009b371a 100644 --- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h +++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h @@ -980,7 +980,7 @@ int c4iw_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len); void c4iw_qp_add_ref(struct ib_qp *qp); void c4iw_qp_rem_ref(struct ib_qp *qp); struct ib_mr *c4iw_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int c4iw_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); int c4iw_dealloc_mw(struct ib_mw *mw); @@ -1053,8 +1053,9 @@ int c4iw_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_wr); struct c4iw_wr_wait *c4iw_alloc_wr_wait(gfp_t gfp); -typedef int c4iw_restrack_func(struct sk_buff *msg, - struct rdma_restrack_entry *res); -extern c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX]; +int c4iw_fill_res_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr); +int c4iw_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ibcq); +int c4iw_fill_res_qp_entry(struct sk_buff *msg, struct ib_qp *ibqp); +int c4iw_fill_res_cm_id_entry(struct sk_buff *msg, struct rdma_cm_id *cm_id); #endif diff --git a/drivers/infiniband/hw/cxgb4/mem.c b/drivers/infiniband/hw/cxgb4/mem.c index 962dc97a8ff2..73936c3341b7 100644 --- a/drivers/infiniband/hw/cxgb4/mem.c +++ b/drivers/infiniband/hw/cxgb4/mem.c @@ -399,7 +399,6 @@ static int finish_mem_reg(struct c4iw_mr *mhp, u32 stag) mmid = stag >> 8; mhp->ibmr.rkey = mhp->ibmr.lkey = stag; mhp->ibmr.length = mhp->attr.len; - mhp->ibmr.iova = mhp->attr.va_fbo; mhp->ibmr.page_size = 1U << (mhp->attr.page_size + 12); pr_debug("mmid 0x%x mhp %p\n", mmid, mhp); return xa_insert_irq(&mhp->rhp->mrs, mmid, mhp, GFP_KERNEL); @@ -691,7 +690,7 @@ int c4iw_dealloc_mw(struct ib_mw *mw) } struct ib_mr *c4iw_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct c4iw_dev *rhp; struct c4iw_pd *php; diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c index ba83d942997c..6c579d2d3997 100644 --- a/drivers/infiniband/hw/cxgb4/provider.c +++ b/drivers/infiniband/hw/cxgb4/provider.c @@ -236,14 +236,6 @@ static int c4iw_allocate_pd(struct ib_pd *pd, struct ib_udata *udata) return 0; } -static int c4iw_query_pkey(struct ib_device *ibdev, u8 port, u16 index, - u16 *pkey) -{ - pr_debug("ibdev %p\n", ibdev); - *pkey = 0; - return 0; -} - static int c4iw_query_gid(struct ib_device *ibdev, u8 port, int index, union ib_gid *gid) { @@ -317,7 +309,6 @@ static int c4iw_query_port(struct ib_device *ibdev, u8 port, IB_PORT_DEVICE_MGMT_SUP | IB_PORT_VENDOR_CLASS_SUP | IB_PORT_BOOT_MGMT_SUP; props->gid_tbl_len = 1; - props->pkey_tbl_len = 1; props->max_msg_sz = -1; return ret; @@ -439,7 +430,6 @@ static int c4iw_port_immutable(struct ib_device *ibdev, u8 port_num, if (err) return err; - immutable->pkey_tbl_len = attr.pkey_tbl_len; immutable->gid_tbl_len = attr.gid_tbl_len; return 0; @@ -458,13 +448,6 @@ static void get_dev_fw_str(struct ib_device *dev, char *str) FW_HDR_FW_VER_BUILD_G(c4iw_dev->rdev.lldi.fw_vers)); } -static int fill_res_entry(struct sk_buff *msg, struct rdma_restrack_entry *res) -{ - return (res->type < ARRAY_SIZE(c4iw_restrack_funcs) && - c4iw_restrack_funcs[res->type]) ? - c4iw_restrack_funcs[res->type](msg, res) : 0; -} - static const struct ib_device_ops c4iw_dev_ops = { .owner = THIS_MODULE, .driver_id = RDMA_DRIVER_CXGB4, @@ -485,7 +468,9 @@ static const struct ib_device_ops c4iw_dev_ops = { .destroy_cq = c4iw_destroy_cq, .destroy_qp = c4iw_destroy_qp, .destroy_srq = c4iw_destroy_srq, - .fill_res_entry = fill_res_entry, + .fill_res_cq_entry = c4iw_fill_res_cq_entry, + .fill_res_cm_id_entry = c4iw_fill_res_cm_id_entry, + .fill_res_mr_entry = c4iw_fill_res_mr_entry, .get_dev_fw_str = get_dev_fw_str, .get_dma_mr = c4iw_get_dma_mr, .get_hw_stats = c4iw_get_mib, @@ -508,7 +493,6 @@ static const struct ib_device_ops c4iw_dev_ops = { .post_srq_recv = c4iw_post_srq_recv, .query_device = c4iw_query_device, .query_gid = c4iw_query_gid, - .query_pkey = c4iw_query_pkey, .query_port = c4iw_query_port, .query_qp = c4iw_ib_query_qp, .reg_user_mr = c4iw_reg_user_mr, diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c index ac48012c992f..cbddb20c6121 100644 --- a/drivers/infiniband/hw/cxgb4/qp.c +++ b/drivers/infiniband/hw/cxgb4/qp.c @@ -1165,7 +1165,7 @@ int c4iw_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, break; } fw_flags |= FW_RI_RDMA_WRITE_WITH_IMMEDIATE; - /*FALLTHROUGH*/ + fallthrough; case IB_WR_RDMA_WRITE: fw_opcode = FW_RI_RDMA_WRITE_WR; swsqe->opcode = FW_RI_RDMA_WRITE; diff --git a/drivers/infiniband/hw/cxgb4/restrack.c b/drivers/infiniband/hw/cxgb4/restrack.c index f82d46ed969d..b32e6516d65f 100644 --- a/drivers/infiniband/hw/cxgb4/restrack.c +++ b/drivers/infiniband/hw/cxgb4/restrack.c @@ -134,10 +134,8 @@ err: return -EMSGSIZE; } -static int fill_res_qp_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +int c4iw_fill_res_qp_entry(struct sk_buff *msg, struct ib_qp *ibqp) { - struct ib_qp *ibqp = container_of(res, struct ib_qp, res); struct t4_swsqe *fsp = NULL, *lsp = NULL; struct c4iw_qp *qhp = to_c4iw_qp(ibqp); u16 first_sq_idx = 0, last_sq_idx = 0; @@ -195,10 +193,9 @@ union union_ep { struct c4iw_ep ep; }; -static int fill_res_ep_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +int c4iw_fill_res_cm_id_entry(struct sk_buff *msg, + struct rdma_cm_id *cm_id) { - struct rdma_cm_id *cm_id = rdma_res_to_id(res); struct nlattr *table_attr; struct c4iw_ep_common *epcp; struct c4iw_listen_ep *listen_ep = NULL; @@ -372,10 +369,8 @@ err: return -EMSGSIZE; } -static int fill_res_cq_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +int c4iw_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ibcq) { - struct ib_cq *ibcq = container_of(res, struct ib_cq, res); struct c4iw_cq *chp = to_c4iw_cq(ibcq); struct nlattr *table_attr; struct t4_cqe hwcqes[2]; @@ -433,10 +428,8 @@ err: return -EMSGSIZE; } -static int fill_res_mr_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +int c4iw_fill_res_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr) { - struct ib_mr *ibmr = container_of(res, struct ib_mr, res); struct c4iw_mr *mhp = to_c4iw_mr(ibmr); struct c4iw_dev *dev = mhp->rhp; u32 stag = mhp->attr.stag; @@ -492,10 +485,3 @@ err_cancel_table: err: return -EMSGSIZE; } - -c4iw_restrack_func *c4iw_restrack_funcs[RDMA_RESTRACK_MAX] = { - [RDMA_RESTRACK_QP] = fill_res_qp_entry, - [RDMA_RESTRACK_CM_ID] = fill_res_ep_entry, - [RDMA_RESTRACK_CQ] = fill_res_cq_entry, - [RDMA_RESTRACK_MR] = fill_res_mr_entry, -}; diff --git a/drivers/infiniband/hw/efa/efa.h b/drivers/infiniband/hw/efa/efa.h index aa7396a1588a..1889dd172a25 100644 --- a/drivers/infiniband/hw/efa/efa.h +++ b/drivers/infiniband/hw/efa/efa.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */ /* - * Copyright 2018-2019 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. */ #ifndef _EFA_H_ @@ -40,6 +40,7 @@ struct efa_sw_stats { atomic64_t reg_mr_err; atomic64_t alloc_ucontext_err; atomic64_t create_ah_err; + atomic64_t mmap_err; }; /* Don't use anything other than atomic64 */ @@ -153,8 +154,7 @@ int efa_mmap(struct ib_ucontext *ibucontext, struct vm_area_struct *vma); void efa_mmap_free(struct rdma_user_mmap_entry *rdma_entry); int efa_create_ah(struct ib_ah *ibah, - struct rdma_ah_attr *ah_attr, - u32 flags, + struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); void efa_destroy_ah(struct ib_ah *ibah, u32 flags); int efa_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, diff --git a/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h b/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h index 96b104ab5415..5484b08bbc5d 100644 --- a/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h +++ b/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h @@ -37,7 +37,7 @@ enum efa_admin_aq_feature_id { EFA_ADMIN_NETWORK_ATTR = 3, EFA_ADMIN_QUEUE_ATTR = 4, EFA_ADMIN_HW_HINTS = 5, - EFA_ADMIN_FEATURES_OPCODE_NUM = 8, + EFA_ADMIN_HOST_INFO = 6, }; /* QP transport type */ @@ -606,8 +606,8 @@ struct efa_admin_feature_queue_attr_desc { /* Number of sub-CQs to be created for each CQ */ u16 sub_cqs_per_cq; - /* MBZ */ - u16 reserved; + /* Minimum number of WQEs per SQ */ + u16 min_sq_depth; /* Maximum number of SGEs (buffers) allowed for a single send WQE */ u16 max_wr_send_sges; @@ -632,6 +632,17 @@ struct efa_admin_feature_queue_attr_desc { /* Maximum number of SGEs for a single RDMA read WQE */ u16 max_wr_rdma_sges; + + /* + * Maximum number of bytes that can be written to SQ between two + * consecutive doorbells (in units of 64B). Driver must ensure that only + * complete WQEs are written to queue before issuing a doorbell. + * Examples: max_tx_batch=16 and WQE size = 64B, means up to 16 WQEs can + * be written to SQ between two consecutive doorbells. max_tx_batch=11 + * and WQE size = 128B, means up to 5 WQEs can be written to SQ between + * two consecutive doorbells. Zero means unlimited. + */ + u16 max_tx_batch; }; struct efa_admin_feature_aenq_desc { @@ -799,6 +810,54 @@ struct efa_admin_mmio_req_read_less_resp { u32 reg_val; }; +enum efa_admin_os_type { + EFA_ADMIN_OS_LINUX = 0, +}; + +struct efa_admin_host_info { + /* OS distribution string format */ + u8 os_dist_str[128]; + + /* Defined in enum efa_admin_os_type */ + u32 os_type; + + /* Kernel version string format */ + u8 kernel_ver_str[32]; + + /* Kernel version numeric format */ + u32 kernel_ver; + + /* + * 7:0 : driver_module_type + * 15:8 : driver_sub_minor + * 23:16 : driver_minor + * 31:24 : driver_major + */ + u32 driver_ver; + + /* + * Device's Bus, Device and Function + * 2:0 : function + * 7:3 : device + * 15:8 : bus + */ + u16 bdf; + + /* + * Spec version + * 7:0 : spec_minor + * 15:8 : spec_major + */ + u16 spec_ver; + + /* + * 0 : intree - Intree driver + * 1 : gdr - GPUDirect RDMA supported + * 31:2 : reserved2 + */ + u32 flags; +}; + /* create_qp_cmd */ #define EFA_ADMIN_CREATE_QP_CMD_SQ_VIRT_MASK BIT(0) #define EFA_ADMIN_CREATE_QP_CMD_RQ_VIRT_MASK BIT(1) @@ -820,4 +879,17 @@ struct efa_admin_mmio_req_read_less_resp { /* feature_device_attr_desc */ #define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RDMA_READ_MASK BIT(0) +/* host_info */ +#define EFA_ADMIN_HOST_INFO_DRIVER_MODULE_TYPE_MASK GENMASK(7, 0) +#define EFA_ADMIN_HOST_INFO_DRIVER_SUB_MINOR_MASK GENMASK(15, 8) +#define EFA_ADMIN_HOST_INFO_DRIVER_MINOR_MASK GENMASK(23, 16) +#define EFA_ADMIN_HOST_INFO_DRIVER_MAJOR_MASK GENMASK(31, 24) +#define EFA_ADMIN_HOST_INFO_FUNCTION_MASK GENMASK(2, 0) +#define EFA_ADMIN_HOST_INFO_DEVICE_MASK GENMASK(7, 3) +#define EFA_ADMIN_HOST_INFO_BUS_MASK GENMASK(15, 8) +#define EFA_ADMIN_HOST_INFO_SPEC_MINOR_MASK GENMASK(7, 0) +#define EFA_ADMIN_HOST_INFO_SPEC_MAJOR_MASK GENMASK(15, 8) +#define EFA_ADMIN_HOST_INFO_INTREE_MASK BIT(0) +#define EFA_ADMIN_HOST_INFO_GDR_MASK BIT(1) + #endif /* _EFA_ADMIN_CMDS_H_ */ diff --git a/drivers/infiniband/hw/efa/efa_com.c b/drivers/infiniband/hw/efa/efa_com.c index 7fce69f5568f..336bc2c57bb1 100644 --- a/drivers/infiniband/hw/efa/efa_com.c +++ b/drivers/infiniband/hw/efa/efa_com.c @@ -631,17 +631,20 @@ int efa_com_cmd_exec(struct efa_com_admin_queue *aq, cmd->aq_common_descriptor.opcode, PTR_ERR(comp_ctx)); up(&aq->avail_cmds); + atomic64_inc(&aq->stats.cmd_err); return PTR_ERR(comp_ctx); } err = efa_com_wait_and_process_admin_cq(comp_ctx, aq); - if (err) + if (err) { ibdev_err_ratelimited( aq->efa_dev, "Failed to process command %s (opcode %u) comp_status %d err %d\n", efa_com_cmd_str(cmd->aq_common_descriptor.opcode), cmd->aq_common_descriptor.opcode, comp_ctx->comp_status, err); + atomic64_inc(&aq->stats.cmd_err); + } up(&aq->avail_cmds); diff --git a/drivers/infiniband/hw/efa/efa_com.h b/drivers/infiniband/hw/efa/efa_com.h index c67dd8109d1c..5e4c88877ddb 100644 --- a/drivers/infiniband/hw/efa/efa_com.h +++ b/drivers/infiniband/hw/efa/efa_com.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */ /* - * Copyright 2018-2019 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. */ #ifndef _EFA_COM_H_ @@ -47,6 +47,7 @@ struct efa_com_admin_sq { struct efa_com_stats_admin { atomic64_t submitted_cmd; atomic64_t completed_cmd; + atomic64_t cmd_err; atomic64_t no_completion; }; diff --git a/drivers/infiniband/hw/efa/efa_com_cmd.c b/drivers/infiniband/hw/efa/efa_com_cmd.c index eea5574a62e8..6ac23627f65a 100644 --- a/drivers/infiniband/hw/efa/efa_com_cmd.c +++ b/drivers/infiniband/hw/efa/efa_com_cmd.c @@ -351,7 +351,7 @@ int efa_com_destroy_ah(struct efa_com_dev *edev, return 0; } -static bool +bool efa_com_check_supported_feature_id(struct efa_com_dev *edev, enum efa_admin_aq_feature_id feature_id) { @@ -388,7 +388,7 @@ static int efa_com_get_feature_ex(struct efa_com_dev *edev, if (control_buff_size) EFA_SET(&get_cmd.aq_common_descriptor.flags, - EFA_ADMIN_AQ_COMMON_DESC_CTRL_DATA_INDIRECT, 1); + EFA_ADMIN_AQ_COMMON_DESC_CTRL_DATA, 1); efa_com_set_dma_addr(control_buf_dma_addr, &get_cmd.control_buffer.address.mem_addr_high, @@ -480,6 +480,8 @@ int efa_com_get_device_attr(struct efa_com_dev *edev, result->max_llq_size = resp.u.queue_attr.max_llq_size; result->sub_cqs_per_cq = resp.u.queue_attr.sub_cqs_per_cq; result->max_wr_rdma_sge = resp.u.queue_attr.max_wr_rdma_sges; + result->max_tx_batch = resp.u.queue_attr.max_tx_batch; + result->min_sq_depth = resp.u.queue_attr.min_sq_depth; err = efa_com_get_feature(edev, &resp, EFA_ADMIN_NETWORK_ATTR); if (err) { @@ -517,12 +519,12 @@ int efa_com_get_hw_hints(struct efa_com_dev *edev, return 0; } -static int efa_com_set_feature_ex(struct efa_com_dev *edev, - struct efa_admin_set_feature_resp *set_resp, - struct efa_admin_set_feature_cmd *set_cmd, - enum efa_admin_aq_feature_id feature_id, - dma_addr_t control_buf_dma_addr, - u32 control_buff_size) +int efa_com_set_feature_ex(struct efa_com_dev *edev, + struct efa_admin_set_feature_resp *set_resp, + struct efa_admin_set_feature_cmd *set_cmd, + enum efa_admin_aq_feature_id feature_id, + dma_addr_t control_buf_dma_addr, + u32 control_buff_size) { struct efa_com_admin_queue *aq; int err; @@ -540,7 +542,7 @@ static int efa_com_set_feature_ex(struct efa_com_dev *edev, if (control_buff_size) { set_cmd->aq_common_descriptor.flags = 0; EFA_SET(&set_cmd->aq_common_descriptor.flags, - EFA_ADMIN_AQ_COMMON_DESC_CTRL_DATA_INDIRECT, 1); + EFA_ADMIN_AQ_COMMON_DESC_CTRL_DATA, 1); efa_com_set_dma_addr(control_buf_dma_addr, &set_cmd->control_buffer.address.mem_addr_high, &set_cmd->control_buffer.address.mem_addr_low); diff --git a/drivers/infiniband/hw/efa/efa_com_cmd.h b/drivers/infiniband/hw/efa/efa_com_cmd.h index 31db5a0cbd5b..190bac23f585 100644 --- a/drivers/infiniband/hw/efa/efa_com_cmd.h +++ b/drivers/infiniband/hw/efa/efa_com_cmd.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */ /* - * Copyright 2018-2019 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. */ #ifndef _EFA_COM_CMD_H_ @@ -127,6 +127,8 @@ struct efa_com_get_device_attr_result { u16 max_sq_sge; u16 max_rq_sge; u16 max_wr_rdma_sge; + u16 max_tx_batch; + u16 min_sq_depth; u8 db_bar; }; @@ -270,6 +272,15 @@ int efa_com_get_device_attr(struct efa_com_dev *edev, struct efa_com_get_device_attr_result *result); int efa_com_get_hw_hints(struct efa_com_dev *edev, struct efa_com_get_hw_hints_result *result); +bool +efa_com_check_supported_feature_id(struct efa_com_dev *edev, + enum efa_admin_aq_feature_id feature_id); +int efa_com_set_feature_ex(struct efa_com_dev *edev, + struct efa_admin_set_feature_resp *set_resp, + struct efa_admin_set_feature_cmd *set_cmd, + enum efa_admin_aq_feature_id feature_id, + dma_addr_t control_buf_dma_addr, + u32 control_buff_size); int efa_com_set_aenq_config(struct efa_com_dev *edev, u32 groups); int efa_com_alloc_pd(struct efa_com_dev *edev, struct efa_com_alloc_pd_result *result); diff --git a/drivers/infiniband/hw/efa/efa_main.c b/drivers/infiniband/hw/efa/efa_main.c index faf3ff1bca2a..92d701146320 100644 --- a/drivers/infiniband/hw/efa/efa_main.c +++ b/drivers/infiniband/hw/efa/efa_main.c @@ -1,19 +1,23 @@ // SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause /* - * Copyright 2018-2019 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. */ #include <linux/module.h> #include <linux/pci.h> +#include <linux/utsname.h> +#include <linux/version.h> #include <rdma/ib_user_verbs.h> #include "efa.h" -#define PCI_DEV_ID_EFA_VF 0xefa0 +#define PCI_DEV_ID_EFA0_VF 0xefa0 +#define PCI_DEV_ID_EFA1_VF 0xefa1 static const struct pci_device_id efa_pci_tbl[] = { - { PCI_VDEVICE(AMAZON, PCI_DEV_ID_EFA_VF) }, + { PCI_VDEVICE(AMAZON, PCI_DEV_ID_EFA0_VF) }, + { PCI_VDEVICE(AMAZON, PCI_DEV_ID_EFA1_VF) }, { } }; @@ -187,6 +191,52 @@ static void efa_stats_init(struct efa_dev *dev) atomic64_set(s, 0); } +static void efa_set_host_info(struct efa_dev *dev) +{ + struct efa_admin_set_feature_resp resp = {}; + struct efa_admin_set_feature_cmd cmd = {}; + struct efa_admin_host_info *hinf; + u32 bufsz = sizeof(*hinf); + dma_addr_t hinf_dma; + + if (!efa_com_check_supported_feature_id(&dev->edev, + EFA_ADMIN_HOST_INFO)) + return; + + /* Failures in host info set shall not disturb probe */ + hinf = dma_alloc_coherent(&dev->pdev->dev, bufsz, &hinf_dma, + GFP_KERNEL); + if (!hinf) + return; + + strlcpy(hinf->os_dist_str, utsname()->release, + min(sizeof(hinf->os_dist_str), sizeof(utsname()->release))); + hinf->os_type = EFA_ADMIN_OS_LINUX; + strlcpy(hinf->kernel_ver_str, utsname()->version, + min(sizeof(hinf->kernel_ver_str), sizeof(utsname()->version))); + hinf->kernel_ver = LINUX_VERSION_CODE; + EFA_SET(&hinf->driver_ver, EFA_ADMIN_HOST_INFO_DRIVER_MAJOR, 0); + EFA_SET(&hinf->driver_ver, EFA_ADMIN_HOST_INFO_DRIVER_MINOR, 0); + EFA_SET(&hinf->driver_ver, EFA_ADMIN_HOST_INFO_DRIVER_SUB_MINOR, 0); + EFA_SET(&hinf->driver_ver, EFA_ADMIN_HOST_INFO_DRIVER_MODULE_TYPE, 0); + EFA_SET(&hinf->bdf, EFA_ADMIN_HOST_INFO_BUS, dev->pdev->bus->number); + EFA_SET(&hinf->bdf, EFA_ADMIN_HOST_INFO_DEVICE, + PCI_SLOT(dev->pdev->devfn)); + EFA_SET(&hinf->bdf, EFA_ADMIN_HOST_INFO_FUNCTION, + PCI_FUNC(dev->pdev->devfn)); + EFA_SET(&hinf->spec_ver, EFA_ADMIN_HOST_INFO_SPEC_MAJOR, + EFA_COMMON_SPEC_VERSION_MAJOR); + EFA_SET(&hinf->spec_ver, EFA_ADMIN_HOST_INFO_SPEC_MINOR, + EFA_COMMON_SPEC_VERSION_MINOR); + EFA_SET(&hinf->flags, EFA_ADMIN_HOST_INFO_INTREE, 1); + EFA_SET(&hinf->flags, EFA_ADMIN_HOST_INFO_GDR, 0); + + efa_com_set_feature_ex(&dev->edev, &resp, &cmd, EFA_ADMIN_HOST_INFO, + hinf_dma, bufsz); + + dma_free_coherent(&dev->pdev->dev, bufsz, hinf, hinf_dma); +} + static const struct ib_device_ops efa_dev_ops = { .owner = THIS_MODULE, .driver_id = RDMA_DRIVER_EFA, @@ -251,6 +301,8 @@ static int efa_ib_device_add(struct efa_dev *dev) if (err) goto err_release_doorbell_bar; + efa_set_host_info(dev); + dev->ibdev.node_type = RDMA_NODE_UNSPECIFIED; dev->ibdev.phys_port_cnt = 1; dev->ibdev.num_comp_vectors = 1; diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c index 5c57098a4aee..9e201f169289 100644 --- a/drivers/infiniband/hw/efa/efa_verbs.c +++ b/drivers/infiniband/hw/efa/efa_verbs.c @@ -37,13 +37,16 @@ struct efa_user_mmap_entry { op(EFA_RX_DROPS, "rx_drops") \ op(EFA_SUBMITTED_CMDS, "submitted_cmds") \ op(EFA_COMPLETED_CMDS, "completed_cmds") \ + op(EFA_CMDS_ERR, "cmds_err") \ op(EFA_NO_COMPLETION_CMDS, "no_completion_cmds") \ op(EFA_KEEP_ALIVE_RCVD, "keep_alive_rcvd") \ op(EFA_ALLOC_PD_ERR, "alloc_pd_err") \ op(EFA_CREATE_QP_ERR, "create_qp_err") \ + op(EFA_CREATE_CQ_ERR, "create_cq_err") \ op(EFA_REG_MR_ERR, "reg_mr_err") \ op(EFA_ALLOC_UCONTEXT_ERR, "alloc_ucontext_err") \ - op(EFA_CREATE_AH_ERR, "create_ah_err") + op(EFA_CREATE_AH_ERR, "create_ah_err") \ + op(EFA_MMAP_ERR, "mmap_err") #define EFA_STATS_ENUM(ename, name) ename, #define EFA_STATS_STR(ename, name) [ename] = name, @@ -209,6 +212,7 @@ int efa_query_device(struct ib_device *ibdev, props->max_send_sge = dev_attr->max_sq_sge; props->max_recv_sge = dev_attr->max_rq_sge; props->max_sge_rd = dev_attr->max_wr_rdma_sge; + props->max_pkeys = 1; if (udata && udata->outlen) { resp.max_sq_sge = dev_attr->max_sq_sge; @@ -1498,11 +1502,39 @@ static int efa_dealloc_uar(struct efa_dev *dev, u16 uarn) return efa_com_dealloc_uar(&dev->edev, ¶ms); } +#define EFA_CHECK_USER_COMP(_dev, _comp_mask, _attr, _mask, _attr_str) \ + (_attr_str = (!(_dev)->dev_attr._attr || ((_comp_mask) & (_mask))) ? \ + NULL : #_attr) + +static int efa_user_comp_handshake(const struct ib_ucontext *ibucontext, + const struct efa_ibv_alloc_ucontext_cmd *cmd) +{ + struct efa_dev *dev = to_edev(ibucontext->device); + char *attr_str; + + if (EFA_CHECK_USER_COMP(dev, cmd->comp_mask, max_tx_batch, + EFA_ALLOC_UCONTEXT_CMD_COMP_TX_BATCH, attr_str)) + goto err; + + if (EFA_CHECK_USER_COMP(dev, cmd->comp_mask, min_sq_depth, + EFA_ALLOC_UCONTEXT_CMD_COMP_MIN_SQ_WR, + attr_str)) + goto err; + + return 0; + +err: + ibdev_dbg(&dev->ibdev, "Userspace handshake failed for %s attribute\n", + attr_str); + return -EOPNOTSUPP; +} + int efa_alloc_ucontext(struct ib_ucontext *ibucontext, struct ib_udata *udata) { struct efa_ucontext *ucontext = to_eucontext(ibucontext); struct efa_dev *dev = to_edev(ibucontext->device); struct efa_ibv_alloc_ucontext_resp resp = {}; + struct efa_ibv_alloc_ucontext_cmd cmd = {}; struct efa_com_alloc_uar_result result; int err; @@ -1511,6 +1543,18 @@ int efa_alloc_ucontext(struct ib_ucontext *ibucontext, struct ib_udata *udata) * we will ack input fields in our response. */ + err = ib_copy_from_udata(&cmd, udata, + min(sizeof(cmd), udata->inlen)); + if (err) { + ibdev_dbg(&dev->ibdev, + "Cannot copy udata for alloc_ucontext\n"); + goto err_out; + } + + err = efa_user_comp_handshake(ibucontext, &cmd); + if (err) + goto err_out; + err = efa_com_alloc_uar(&dev->edev, &result); if (err) goto err_out; @@ -1522,6 +1566,8 @@ int efa_alloc_ucontext(struct ib_ucontext *ibucontext, struct ib_udata *udata) resp.sub_cqs_per_cq = dev->dev_attr.sub_cqs_per_cq; resp.inline_buf_size = dev->dev_attr.inline_buf_size; resp.max_llq_size = dev->dev_attr.max_llq_size; + resp.max_tx_batch = dev->dev_attr.max_tx_batch; + resp.min_sq_wr = dev->dev_attr.min_sq_depth; if (udata && udata->outlen) { err = ib_copy_to_udata(udata, &resp, @@ -1568,6 +1614,7 @@ static int __efa_mmap(struct efa_dev *dev, struct efa_ucontext *ucontext, ibdev_dbg(&dev->ibdev, "pgoff[%#lx] does not have valid entry\n", vma->vm_pgoff); + atomic64_inc(&dev->stats.sw_stats.mmap_err); return -EINVAL; } entry = to_emmap(rdma_entry); @@ -1603,12 +1650,14 @@ static int __efa_mmap(struct efa_dev *dev, struct efa_ucontext *ucontext, err = -EINVAL; } - if (err) + if (err) { ibdev_dbg( &dev->ibdev, "Couldn't mmap address[%#llx] length[%#zx] mmap_flag[%d] err[%d]\n", entry->address, rdma_entry->npages * PAGE_SIZE, entry->mmap_flag, err); + atomic64_inc(&dev->stats.sw_stats.mmap_err); + } rdma_user_mmap_entry_put(rdma_entry); return err; @@ -1639,10 +1688,10 @@ static int efa_ah_destroy(struct efa_dev *dev, struct efa_ah *ah) } int efa_create_ah(struct ib_ah *ibah, - struct rdma_ah_attr *ah_attr, - u32 flags, + struct rdma_ah_init_attr *init_attr, struct ib_udata *udata) { + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; struct efa_dev *dev = to_edev(ibah->device); struct efa_com_create_ah_params params = {}; struct efa_ibv_create_ah_resp resp = {}; @@ -1650,7 +1699,7 @@ int efa_create_ah(struct ib_ah *ibah, struct efa_ah *ah = to_eah(ibah); int err; - if (!(flags & RDMA_CREATE_AH_SLEEPABLE)) { + if (!(init_attr->flags & RDMA_CREATE_AH_SLEEPABLE)) { ibdev_dbg(&dev->ibdev, "Create address handle is not supported in atomic context\n"); err = -EOPNOTSUPP; @@ -1747,15 +1796,18 @@ int efa_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats, as = &dev->edev.aq.stats; stats->value[EFA_SUBMITTED_CMDS] = atomic64_read(&as->submitted_cmd); stats->value[EFA_COMPLETED_CMDS] = atomic64_read(&as->completed_cmd); + stats->value[EFA_CMDS_ERR] = atomic64_read(&as->cmd_err); stats->value[EFA_NO_COMPLETION_CMDS] = atomic64_read(&as->no_completion); s = &dev->stats; stats->value[EFA_KEEP_ALIVE_RCVD] = atomic64_read(&s->keep_alive_rcvd); stats->value[EFA_ALLOC_PD_ERR] = atomic64_read(&s->sw_stats.alloc_pd_err); stats->value[EFA_CREATE_QP_ERR] = atomic64_read(&s->sw_stats.create_qp_err); + stats->value[EFA_CREATE_CQ_ERR] = atomic64_read(&s->sw_stats.create_cq_err); stats->value[EFA_REG_MR_ERR] = atomic64_read(&s->sw_stats.reg_mr_err); stats->value[EFA_ALLOC_UCONTEXT_ERR] = atomic64_read(&s->sw_stats.alloc_ucontext_err); stats->value[EFA_CREATE_AH_ERR] = atomic64_read(&s->sw_stats.create_ah_err); + stats->value[EFA_MMAP_ERR] = atomic64_read(&s->sw_stats.mmap_err); return ARRAY_SIZE(efa_stats_names); } diff --git a/drivers/infiniband/hw/hfi1/Kconfig b/drivers/infiniband/hw/hfi1/Kconfig index 0653f4f7b26c..519866b30a13 100644 --- a/drivers/infiniband/hw/hfi1/Kconfig +++ b/drivers/infiniband/hw/hfi1/Kconfig @@ -5,19 +5,19 @@ config INFINIBAND_HFI1 select MMU_NOTIFIER select CRC32 select I2C_ALGOBIT - ---help--- + help This is a low-level driver for Intel OPA Gen1 adapter. config HFI1_DEBUG_SDMA_ORDER bool "HFI1 SDMA Order debug" depends on INFINIBAND_HFI1 default n - ---help--- + help This is a debug flag to test for out of order sdma completions for unit testing config SDMA_VERBOSITY bool "Config SDMA Verbosity" depends on INFINIBAND_HFI1 default n - ---help--- + help This is a configuration flag to enable verbose SDMA debug diff --git a/drivers/infiniband/hw/hfi1/Makefile b/drivers/infiniband/hw/hfi1/Makefile index 0405d26d0833..2e89ec10efed 100644 --- a/drivers/infiniband/hw/hfi1/Makefile +++ b/drivers/infiniband/hw/hfi1/Makefile @@ -22,9 +22,13 @@ hfi1-y := \ init.o \ intr.o \ iowait.o \ + ipoib_main.o \ + ipoib_rx.o \ + ipoib_tx.o \ mad.o \ mmu_rb.o \ msix.o \ + netdev_rx.o \ opfn.o \ pcie.o \ pio.o \ diff --git a/drivers/infiniband/hw/hfi1/affinity.c b/drivers/infiniband/hw/hfi1/affinity.c index 1aeea5d65c01..2a91b8d95e12 100644 --- a/drivers/infiniband/hw/hfi1/affinity.c +++ b/drivers/infiniband/hw/hfi1/affinity.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -64,6 +64,7 @@ struct hfi1_affinity_node_list node_affinity = { static const char * const irq_type_names[] = { "SDMA", "RCVCTXT", + "NETDEVCTXT", "GENERAL", "OTHER", }; @@ -915,6 +916,11 @@ static int get_irq_affinity(struct hfi1_devdata *dd, set = &entry->rcv_intr; scnprintf(extra, 64, "ctxt %u", rcd->ctxt); break; + case IRQ_NETDEVCTXT: + rcd = (struct hfi1_ctxtdata *)msix->arg; + set = &entry->def_intr; + scnprintf(extra, 64, "ctxt %u", rcd->ctxt); + break; default: dd_dev_err(dd, "Invalid IRQ type %d\n", msix->type); return -EINVAL; @@ -987,6 +993,10 @@ void hfi1_put_irq_affinity(struct hfi1_devdata *dd, if (rcd->ctxt != HFI1_CTRL_CTXT) set = &entry->rcv_intr; break; + case IRQ_NETDEVCTXT: + rcd = (struct hfi1_ctxtdata *)msix->arg; + set = &entry->def_intr; + break; default: mutex_unlock(&node_affinity.lock); return; diff --git a/drivers/infiniband/hw/hfi1/affinity.h b/drivers/infiniband/hw/hfi1/affinity.h index 6a7e6ea4e426..f94ed5d7c7a3 100644 --- a/drivers/infiniband/hw/hfi1/affinity.h +++ b/drivers/infiniband/hw/hfi1/affinity.h @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -52,6 +52,7 @@ enum irq_type { IRQ_SDMA, IRQ_RCVCTXT, + IRQ_NETDEVCTXT, IRQ_GENERAL, IRQ_OTHER }; diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c index e0b1238d31df..7eaf99538216 100644 --- a/drivers/infiniband/hw/hfi1/chip.c +++ b/drivers/infiniband/hw/hfi1/chip.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -66,10 +66,7 @@ #include "affinity.h" #include "debugfs.h" #include "fault.h" - -uint kdeth_qp; -module_param_named(kdeth_qp, kdeth_qp, uint, S_IRUGO); -MODULE_PARM_DESC(kdeth_qp, "Set the KDETH queue pair prefix"); +#include "netdev.h" uint num_vls = HFI1_MAX_VLS_SUPPORTED; module_param(num_vls, uint, S_IRUGO); @@ -128,13 +125,15 @@ struct flag_table { /* * RSM instance allocation - * 0 - Verbs - * 1 - User Fecn Handling - * 2 - Vnic + * 0 - User Fecn Handling + * 1 - Vnic + * 2 - AIP + * 3 - Verbs */ -#define RSM_INS_VERBS 0 -#define RSM_INS_FECN 1 -#define RSM_INS_VNIC 2 +#define RSM_INS_FECN 0 +#define RSM_INS_VNIC 1 +#define RSM_INS_AIP 2 +#define RSM_INS_VERBS 3 /* Bit offset into the GUID which carries HFI id information */ #define GUID_HFI_INDEX_SHIFT 39 @@ -175,6 +174,25 @@ struct flag_table { /* QPN[m+n:1] QW 1, OFFSET 1 */ #define QPN_SELECT_OFFSET ((1ull << QW_SHIFT) | (1ull)) +/* RSM fields for AIP */ +/* LRH.BTH above is reused for this rule */ + +/* BTH.DESTQP: QW 1, OFFSET 16 for match */ +#define BTH_DESTQP_QW 1ull +#define BTH_DESTQP_BIT_OFFSET 16ull +#define BTH_DESTQP_OFFSET(off) ((BTH_DESTQP_QW << QW_SHIFT) | (off)) +#define BTH_DESTQP_MATCH_OFFSET BTH_DESTQP_OFFSET(BTH_DESTQP_BIT_OFFSET) +#define BTH_DESTQP_MASK 0xFFull +#define BTH_DESTQP_VALUE 0x81ull + +/* DETH.SQPN: QW 1 Offset 56 for select */ +/* We use 8 most significant Soure QPN bits as entropy fpr AIP */ +#define DETH_AIP_SQPN_QW 3ull +#define DETH_AIP_SQPN_BIT_OFFSET 56ull +#define DETH_AIP_SQPN_OFFSET(off) ((DETH_AIP_SQPN_QW << QW_SHIFT) | (off)) +#define DETH_AIP_SQPN_SELECT_OFFSET \ + DETH_AIP_SQPN_OFFSET(DETH_AIP_SQPN_BIT_OFFSET) + /* RSM fields for Vnic */ /* L2_TYPE: QW 0, OFFSET 61 - for match */ #define L2_TYPE_QW 0ull @@ -7299,11 +7317,11 @@ static u16 link_width_to_bits(struct hfi1_devdata *dd, u16 width) case 1: return OPA_LINK_WIDTH_1X; case 2: return OPA_LINK_WIDTH_2X; case 3: return OPA_LINK_WIDTH_3X; + case 4: return OPA_LINK_WIDTH_4X; default: dd_dev_info(dd, "%s: invalid width %d, using 4\n", __func__, width); - /* fall through */ - case 4: return OPA_LINK_WIDTH_4X; + return OPA_LINK_WIDTH_4X; } } @@ -7358,12 +7376,13 @@ static void get_link_widths(struct hfi1_devdata *dd, u16 *tx_width, case 0: dd->pport[0].link_speed_active = OPA_LINK_SPEED_12_5G; break; + case 1: + dd->pport[0].link_speed_active = OPA_LINK_SPEED_25G; + break; default: dd_dev_err(dd, "%s: unexpected max rate %d, using 25Gb\n", __func__, (int)max_rate); - /* fall through */ - case 1: dd->pport[0].link_speed_active = OPA_LINK_SPEED_25G; break; } @@ -8463,6 +8482,49 @@ static void hfi1_rcd_eoi_intr(struct hfi1_ctxtdata *rcd) local_irq_restore(flags); } +/** + * hfi1_netdev_rx_napi - napi poll function to move eoi inline + * @napi - pointer to napi object + * @budget - netdev budget + */ +int hfi1_netdev_rx_napi(struct napi_struct *napi, int budget) +{ + struct hfi1_netdev_rxq *rxq = container_of(napi, + struct hfi1_netdev_rxq, napi); + struct hfi1_ctxtdata *rcd = rxq->rcd; + int work_done = 0; + + work_done = rcd->do_interrupt(rcd, budget); + + if (work_done < budget) { + napi_complete_done(napi, work_done); + hfi1_rcd_eoi_intr(rcd); + } + + return work_done; +} + +/* Receive packet napi handler for netdevs VNIC and AIP */ +irqreturn_t receive_context_interrupt_napi(int irq, void *data) +{ + struct hfi1_ctxtdata *rcd = data; + + receive_interrupt_common(rcd); + + if (likely(rcd->napi)) { + if (likely(napi_schedule_prep(rcd->napi))) + __napi_schedule_irqoff(rcd->napi); + else + __hfi1_rcd_eoi_intr(rcd); + } else { + WARN_ONCE(1, "Napi IRQ handler without napi set up ctxt=%d\n", + rcd->ctxt); + __hfi1_rcd_eoi_intr(rcd); + } + + return IRQ_HANDLED; +} + /* * Receive packet IRQ handler. This routine expects to be on its own IRQ. * This routine will try to handle packets immediately (latency), but if @@ -12817,11 +12879,6 @@ bail: static u32 chip_to_opa_lstate(struct hfi1_devdata *dd, u32 chip_lstate) { switch (chip_lstate) { - default: - dd_dev_err(dd, - "Unknown logical state 0x%x, reporting IB_PORT_DOWN\n", - chip_lstate); - /* fall through */ case LSTATE_DOWN: return IB_PORT_DOWN; case LSTATE_INIT: @@ -12830,6 +12887,11 @@ static u32 chip_to_opa_lstate(struct hfi1_devdata *dd, u32 chip_lstate) return IB_PORT_ARMED; case LSTATE_ACTIVE: return IB_PORT_ACTIVE; + default: + dd_dev_err(dd, + "Unknown logical state 0x%x, reporting IB_PORT_DOWN\n", + chip_lstate); + return IB_PORT_DOWN; } } @@ -12837,10 +12899,6 @@ u32 chip_to_opa_pstate(struct hfi1_devdata *dd, u32 chip_pstate) { /* look at the HFI meta-states only */ switch (chip_pstate & 0xf0) { - default: - dd_dev_err(dd, "Unexpected chip physical state of 0x%x\n", - chip_pstate); - /* fall through */ case PLS_DISABLED: return IB_PORTPHYSSTATE_DISABLED; case PLS_OFFLINE: @@ -12853,6 +12911,10 @@ u32 chip_to_opa_pstate(struct hfi1_devdata *dd, u32 chip_pstate) return IB_PORTPHYSSTATE_LINKUP; case PLS_PHYTEST: return IB_PORTPHYSSTATE_PHY_TEST; + default: + dd_dev_err(dd, "Unexpected chip physical state of 0x%x\n", + chip_pstate); + return IB_PORTPHYSSTATE_DISABLED; } } @@ -13330,13 +13392,12 @@ static int set_up_interrupts(struct hfi1_devdata *dd) * in array of contexts * freectxts - number of free user contexts * num_send_contexts - number of PIO send contexts being used - * num_vnic_contexts - number of contexts reserved for VNIC + * num_netdev_contexts - number of contexts reserved for netdev */ static int set_up_context_variables(struct hfi1_devdata *dd) { unsigned long num_kernel_contexts; - u16 num_vnic_contexts = HFI1_NUM_VNIC_CTXT; - int total_contexts; + u16 num_netdev_contexts; int ret; unsigned ngroups; int rmt_count; @@ -13373,13 +13434,6 @@ static int set_up_context_variables(struct hfi1_devdata *dd) num_kernel_contexts = send_contexts - num_vls - 1; } - /* Accommodate VNIC contexts if possible */ - if ((num_kernel_contexts + num_vnic_contexts) > rcv_contexts) { - dd_dev_err(dd, "No receive contexts available for VNIC\n"); - num_vnic_contexts = 0; - } - total_contexts = num_kernel_contexts + num_vnic_contexts; - /* * User contexts: * - default to 1 user context per real (non-HT) CPU core if @@ -13392,28 +13446,32 @@ static int set_up_context_variables(struct hfi1_devdata *dd) /* * Adjust the counts given a global max. */ - if (total_contexts + n_usr_ctxts > rcv_contexts) { + if (num_kernel_contexts + n_usr_ctxts > rcv_contexts) { dd_dev_err(dd, - "Reducing # user receive contexts to: %d, from %u\n", - rcv_contexts - total_contexts, + "Reducing # user receive contexts to: %u, from %u\n", + (u32)(rcv_contexts - num_kernel_contexts), n_usr_ctxts); /* recalculate */ - n_usr_ctxts = rcv_contexts - total_contexts; + n_usr_ctxts = rcv_contexts - num_kernel_contexts; } + num_netdev_contexts = + hfi1_num_netdev_contexts(dd, rcv_contexts - + (num_kernel_contexts + n_usr_ctxts), + &node_affinity.real_cpu_mask); /* * The RMT entries are currently allocated as shown below: * 1. QOS (0 to 128 entries); * 2. FECN (num_kernel_context - 1 + num_user_contexts + - * num_vnic_contexts); - * 3. VNIC (num_vnic_contexts). - * It should be noted that FECN oversubscribe num_vnic_contexts - * entries of RMT because both VNIC and PSM could allocate any receive + * num_netdev_contexts); + * 3. netdev (num_netdev_contexts). + * It should be noted that FECN oversubscribe num_netdev_contexts + * entries of RMT because both netdev and PSM could allocate any receive * context between dd->first_dyn_alloc_text and dd->num_rcv_contexts, * and PSM FECN must reserve an RMT entry for each possible PSM receive * context. */ - rmt_count = qos_rmt_entries(dd, NULL, NULL) + (num_vnic_contexts * 2); + rmt_count = qos_rmt_entries(dd, NULL, NULL) + (num_netdev_contexts * 2); if (HFI1_CAP_IS_KSET(TID_RDMA)) rmt_count += num_kernel_contexts - 1; if (rmt_count + n_usr_ctxts > NUM_MAP_ENTRIES) { @@ -13426,21 +13484,20 @@ static int set_up_context_variables(struct hfi1_devdata *dd) n_usr_ctxts = user_rmt_reduced; } - total_contexts += n_usr_ctxts; - - /* the first N are kernel contexts, the rest are user/vnic contexts */ - dd->num_rcv_contexts = total_contexts; + /* the first N are kernel contexts, the rest are user/netdev contexts */ + dd->num_rcv_contexts = + num_kernel_contexts + n_usr_ctxts + num_netdev_contexts; dd->n_krcv_queues = num_kernel_contexts; dd->first_dyn_alloc_ctxt = num_kernel_contexts; - dd->num_vnic_contexts = num_vnic_contexts; + dd->num_netdev_contexts = num_netdev_contexts; dd->num_user_contexts = n_usr_ctxts; dd->freectxts = n_usr_ctxts; dd_dev_info(dd, - "rcv contexts: chip %d, used %d (kernel %d, vnic %u, user %u)\n", + "rcv contexts: chip %d, used %d (kernel %d, netdev %u, user %u)\n", rcv_contexts, (int)dd->num_rcv_contexts, (int)dd->n_krcv_queues, - dd->num_vnic_contexts, + dd->num_netdev_contexts, dd->num_user_contexts); /* @@ -14119,21 +14176,12 @@ static void init_early_variables(struct hfi1_devdata *dd) static void init_kdeth_qp(struct hfi1_devdata *dd) { - /* user changed the KDETH_QP */ - if (kdeth_qp != 0 && kdeth_qp >= 0xff) { - /* out of range or illegal value */ - dd_dev_err(dd, "Invalid KDETH queue pair prefix, ignoring"); - kdeth_qp = 0; - } - if (kdeth_qp == 0) /* not set, or failed range check */ - kdeth_qp = DEFAULT_KDETH_QP; - write_csr(dd, SEND_BTH_QP, - (kdeth_qp & SEND_BTH_QP_KDETH_QP_MASK) << + (RVT_KDETH_QP_PREFIX & SEND_BTH_QP_KDETH_QP_MASK) << SEND_BTH_QP_KDETH_QP_SHIFT); write_csr(dd, RCV_BTH_QP, - (kdeth_qp & RCV_BTH_QP_KDETH_QP_MASK) << + (RVT_KDETH_QP_PREFIX & RCV_BTH_QP_KDETH_QP_MASK) << RCV_BTH_QP_KDETH_QP_SHIFT); } @@ -14249,6 +14297,12 @@ static void complete_rsm_map_table(struct hfi1_devdata *dd, } } +/* Is a receive side mapping rule */ +static bool has_rsm_rule(struct hfi1_devdata *dd, u8 rule_index) +{ + return read_csr(dd, RCV_RSM_CFG + (8 * rule_index)) != 0; +} + /* * Add a receive side mapping rule. */ @@ -14485,77 +14539,138 @@ static void init_fecn_handling(struct hfi1_devdata *dd, rmt->used += total_cnt; } -/* Initialize RSM for VNIC */ -void hfi1_init_vnic_rsm(struct hfi1_devdata *dd) +static inline bool hfi1_is_rmt_full(int start, int spare) +{ + return (start + spare) > NUM_MAP_ENTRIES; +} + +static bool hfi1_netdev_update_rmt(struct hfi1_devdata *dd) { u8 i, j; u8 ctx_id = 0; u64 reg; u32 regoff; - struct rsm_rule_data rrd; + int rmt_start = hfi1_netdev_get_free_rmt_idx(dd); + int ctxt_count = hfi1_netdev_ctxt_count(dd); - if (hfi1_vnic_is_rsm_full(dd, NUM_VNIC_MAP_ENTRIES)) { - dd_dev_err(dd, "Vnic RSM disabled, rmt entries used = %d\n", - dd->vnic.rmt_start); - return; + /* We already have contexts mapped in RMT */ + if (has_rsm_rule(dd, RSM_INS_VNIC) || has_rsm_rule(dd, RSM_INS_AIP)) { + dd_dev_info(dd, "Contexts are already mapped in RMT\n"); + return true; } - dev_dbg(&(dd)->pcidev->dev, "Vnic rsm start = %d, end %d\n", - dd->vnic.rmt_start, - dd->vnic.rmt_start + NUM_VNIC_MAP_ENTRIES); + if (hfi1_is_rmt_full(rmt_start, NUM_NETDEV_MAP_ENTRIES)) { + dd_dev_err(dd, "Not enough RMT entries used = %d\n", + rmt_start); + return false; + } + + dev_dbg(&(dd)->pcidev->dev, "RMT start = %d, end %d\n", + rmt_start, + rmt_start + NUM_NETDEV_MAP_ENTRIES); /* Update RSM mapping table, 32 regs, 256 entries - 1 ctx per byte */ - regoff = RCV_RSM_MAP_TABLE + (dd->vnic.rmt_start / 8) * 8; + regoff = RCV_RSM_MAP_TABLE + (rmt_start / 8) * 8; reg = read_csr(dd, regoff); - for (i = 0; i < NUM_VNIC_MAP_ENTRIES; i++) { - /* Update map register with vnic context */ - j = (dd->vnic.rmt_start + i) % 8; + for (i = 0; i < NUM_NETDEV_MAP_ENTRIES; i++) { + /* Update map register with netdev context */ + j = (rmt_start + i) % 8; reg &= ~(0xffllu << (j * 8)); - reg |= (u64)dd->vnic.ctxt[ctx_id++]->ctxt << (j * 8); - /* Wrap up vnic ctx index */ - ctx_id %= dd->vnic.num_ctxt; + reg |= (u64)hfi1_netdev_get_ctxt(dd, ctx_id++)->ctxt << (j * 8); + /* Wrap up netdev ctx index */ + ctx_id %= ctxt_count; /* Write back map register */ - if (j == 7 || ((i + 1) == NUM_VNIC_MAP_ENTRIES)) { + if (j == 7 || ((i + 1) == NUM_NETDEV_MAP_ENTRIES)) { dev_dbg(&(dd)->pcidev->dev, - "Vnic rsm map reg[%d] =0x%llx\n", + "RMT[%d] =0x%llx\n", regoff - RCV_RSM_MAP_TABLE, reg); write_csr(dd, regoff, reg); regoff += 8; - if (i < (NUM_VNIC_MAP_ENTRIES - 1)) + if (i < (NUM_NETDEV_MAP_ENTRIES - 1)) reg = read_csr(dd, regoff); } } - /* Add rule for vnic */ - rrd.offset = dd->vnic.rmt_start; - rrd.pkt_type = 4; - /* Match 16B packets */ - rrd.field1_off = L2_TYPE_MATCH_OFFSET; - rrd.mask1 = L2_TYPE_MASK; - rrd.value1 = L2_16B_VALUE; - /* Match ETH L4 packets */ - rrd.field2_off = L4_TYPE_MATCH_OFFSET; - rrd.mask2 = L4_16B_TYPE_MASK; - rrd.value2 = L4_16B_ETH_VALUE; - /* Calc context from veswid and entropy */ - rrd.index1_off = L4_16B_HDR_VESWID_OFFSET; - rrd.index1_width = ilog2(NUM_VNIC_MAP_ENTRIES); - rrd.index2_off = L2_16B_ENTROPY_OFFSET; - rrd.index2_width = ilog2(NUM_VNIC_MAP_ENTRIES); - add_rsm_rule(dd, RSM_INS_VNIC, &rrd); - - /* Enable RSM if not already enabled */ + return true; +} + +static void hfi1_enable_rsm_rule(struct hfi1_devdata *dd, + int rule, struct rsm_rule_data *rrd) +{ + if (!hfi1_netdev_update_rmt(dd)) { + dd_dev_err(dd, "Failed to update RMT for RSM%d rule\n", rule); + return; + } + + add_rsm_rule(dd, rule, rrd); add_rcvctrl(dd, RCV_CTRL_RCV_RSM_ENABLE_SMASK); } +void hfi1_init_aip_rsm(struct hfi1_devdata *dd) +{ + /* + * go through with the initialisation only if this rule actually doesn't + * exist yet + */ + if (atomic_fetch_inc(&dd->ipoib_rsm_usr_num) == 0) { + int rmt_start = hfi1_netdev_get_free_rmt_idx(dd); + struct rsm_rule_data rrd = { + .offset = rmt_start, + .pkt_type = IB_PACKET_TYPE, + .field1_off = LRH_BTH_MATCH_OFFSET, + .mask1 = LRH_BTH_MASK, + .value1 = LRH_BTH_VALUE, + .field2_off = BTH_DESTQP_MATCH_OFFSET, + .mask2 = BTH_DESTQP_MASK, + .value2 = BTH_DESTQP_VALUE, + .index1_off = DETH_AIP_SQPN_SELECT_OFFSET + + ilog2(NUM_NETDEV_MAP_ENTRIES), + .index1_width = ilog2(NUM_NETDEV_MAP_ENTRIES), + .index2_off = DETH_AIP_SQPN_SELECT_OFFSET, + .index2_width = ilog2(NUM_NETDEV_MAP_ENTRIES) + }; + + hfi1_enable_rsm_rule(dd, RSM_INS_AIP, &rrd); + } +} + +/* Initialize RSM for VNIC */ +void hfi1_init_vnic_rsm(struct hfi1_devdata *dd) +{ + int rmt_start = hfi1_netdev_get_free_rmt_idx(dd); + struct rsm_rule_data rrd = { + /* Add rule for vnic */ + .offset = rmt_start, + .pkt_type = 4, + /* Match 16B packets */ + .field1_off = L2_TYPE_MATCH_OFFSET, + .mask1 = L2_TYPE_MASK, + .value1 = L2_16B_VALUE, + /* Match ETH L4 packets */ + .field2_off = L4_TYPE_MATCH_OFFSET, + .mask2 = L4_16B_TYPE_MASK, + .value2 = L4_16B_ETH_VALUE, + /* Calc context from veswid and entropy */ + .index1_off = L4_16B_HDR_VESWID_OFFSET, + .index1_width = ilog2(NUM_NETDEV_MAP_ENTRIES), + .index2_off = L2_16B_ENTROPY_OFFSET, + .index2_width = ilog2(NUM_NETDEV_MAP_ENTRIES) + }; + + hfi1_enable_rsm_rule(dd, RSM_INS_VNIC, &rrd); +} + void hfi1_deinit_vnic_rsm(struct hfi1_devdata *dd) { clear_rsm_rule(dd, RSM_INS_VNIC); +} - /* Disable RSM if used only by vnic */ - if (dd->vnic.rmt_start == 0) - clear_rcvctrl(dd, RCV_CTRL_RCV_RSM_ENABLE_SMASK); +void hfi1_deinit_aip_rsm(struct hfi1_devdata *dd) +{ + /* only actually clear the rule if it's the last user asking to do so */ + if (atomic_fetch_add_unless(&dd->ipoib_rsm_usr_num, -1, 0) == 1) + clear_rsm_rule(dd, RSM_INS_AIP); } static int init_rxe(struct hfi1_devdata *dd) @@ -14574,8 +14689,8 @@ static int init_rxe(struct hfi1_devdata *dd) init_qos(dd, rmt); init_fecn_handling(dd, rmt); complete_rsm_map_table(dd, rmt); - /* record number of used rsm map entries for vnic */ - dd->vnic.rmt_start = rmt->used; + /* record number of used rsm map entries for netdev */ + hfi1_netdev_set_free_rmt_idx(dd, rmt->used); kfree(rmt); /* @@ -15129,6 +15244,10 @@ int hfi1_init_dd(struct hfi1_devdata *dd) (dd->revision >> CCE_REVISION_SW_SHIFT) & CCE_REVISION_SW_MASK); + /* alloc netdev data */ + if (hfi1_netdev_alloc(dd)) + goto bail_cleanup; + ret = set_up_context_variables(dd); if (ret) goto bail_cleanup; @@ -15229,6 +15348,7 @@ bail_clear_intr: hfi1_comp_vectors_clean_up(dd); msix_clean_up_interrupts(dd); bail_cleanup: + hfi1_netdev_free(dd); hfi1_pcie_ddcleanup(dd); bail_free: hfi1_free_devdata(dd); diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h index 725509261016..2c6f2de74d4d 100644 --- a/drivers/infiniband/hw/hfi1/chip.h +++ b/drivers/infiniband/hw/hfi1/chip.h @@ -1,7 +1,7 @@ #ifndef _CHIP_H #define _CHIP_H /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -1447,6 +1447,7 @@ irqreturn_t general_interrupt(int irq, void *data); irqreturn_t sdma_interrupt(int irq, void *data); irqreturn_t receive_context_interrupt(int irq, void *data); irqreturn_t receive_context_thread(int irq, void *data); +irqreturn_t receive_context_interrupt_napi(int irq, void *data); int set_intr_bits(struct hfi1_devdata *dd, u16 first, u16 last, bool set); void init_qsfp_int(struct hfi1_devdata *dd); @@ -1455,6 +1456,8 @@ void remap_intr(struct hfi1_devdata *dd, int isrc, int msix_intr); void remap_sdma_interrupts(struct hfi1_devdata *dd, int engine, int msix_intr); void reset_interrupts(struct hfi1_devdata *dd); u8 hfi1_get_qp_map(struct hfi1_devdata *dd, u8 idx); +void hfi1_init_aip_rsm(struct hfi1_devdata *dd); +void hfi1_deinit_aip_rsm(struct hfi1_devdata *dd); /* * Interrupt source table. diff --git a/drivers/infiniband/hw/hfi1/common.h b/drivers/infiniband/hw/hfi1/common.h index 40a1ff0c8a8e..ff423e546b80 100644 --- a/drivers/infiniband/hw/hfi1/common.h +++ b/drivers/infiniband/hw/hfi1/common.h @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -72,13 +72,6 @@ * compilation unit */ -/* - * If a packet's QP[23:16] bits match this value, then it is - * a PSM packet and the hardware will expect a KDETH header - * following the BTH. - */ -#define DEFAULT_KDETH_QP 0x80 - /* driver/hw feature set bitmask */ #define HFI1_CAP_USER_SHIFT 24 #define HFI1_CAP_MASK ((1UL << HFI1_CAP_USER_SHIFT) - 1) @@ -149,7 +142,8 @@ HFI1_CAP_NO_INTEGRITY | \ HFI1_CAP_PKEY_CHECK | \ HFI1_CAP_TID_RDMA | \ - HFI1_CAP_OPFN) << \ + HFI1_CAP_OPFN | \ + HFI1_CAP_AIP) << \ HFI1_CAP_USER_SHIFT) /* * Set of capabilities that need to be enabled for kernel context in @@ -166,6 +160,7 @@ HFI1_CAP_PKEY_CHECK | \ HFI1_CAP_MULTI_PKT_EGR | \ HFI1_CAP_EXTENDED_PSN | \ + HFI1_CAP_AIP | \ ((HFI1_CAP_HDRSUPP | \ HFI1_CAP_MULTI_PKT_EGR | \ HFI1_CAP_STATIC_RATE_CTRL | \ diff --git a/drivers/infiniband/hw/hfi1/debugfs.c b/drivers/infiniband/hw/hfi1/debugfs.c index 4633a0ce1a8c..2ced236e1553 100644 --- a/drivers/infiniband/hw/hfi1/debugfs.c +++ b/drivers/infiniband/hw/hfi1/debugfs.c @@ -985,15 +985,10 @@ static ssize_t qsfp2_debugfs_read(struct file *file, char __user *buf, static int __i2c_debugfs_open(struct inode *in, struct file *fp, u32 target) { struct hfi1_pportdata *ppd; - int ret; ppd = private2ppd(fp); - ret = acquire_chip_resource(ppd->dd, i2c_target(target), 0); - if (ret) /* failed - release the module */ - module_put(THIS_MODULE); - - return ret; + return acquire_chip_resource(ppd->dd, i2c_target(target), 0); } static int i2c1_debugfs_open(struct inode *in, struct file *fp) @@ -1013,7 +1008,6 @@ static int __i2c_debugfs_release(struct inode *in, struct file *fp, u32 target) ppd = private2ppd(fp); release_chip_resource(ppd->dd, i2c_target(target)); - module_put(THIS_MODULE); return 0; } @@ -1031,18 +1025,10 @@ static int i2c2_debugfs_release(struct inode *in, struct file *fp) static int __qsfp_debugfs_open(struct inode *in, struct file *fp, u32 target) { struct hfi1_pportdata *ppd; - int ret; - - if (!try_module_get(THIS_MODULE)) - return -ENODEV; ppd = private2ppd(fp); - ret = acquire_chip_resource(ppd->dd, i2c_target(target), 0); - if (ret) /* failed - release the module */ - module_put(THIS_MODULE); - - return ret; + return acquire_chip_resource(ppd->dd, i2c_target(target), 0); } static int qsfp1_debugfs_open(struct inode *in, struct file *fp) @@ -1062,7 +1048,6 @@ static int __qsfp_debugfs_release(struct inode *in, struct file *fp, u32 target) ppd = private2ppd(fp); release_chip_resource(ppd->dd, i2c_target(target)); - module_put(THIS_MODULE); return 0; } diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c index 049d15befe58..a40701a6e1b6 100644 --- a/drivers/infiniband/hw/hfi1/driver.c +++ b/drivers/infiniband/hw/hfi1/driver.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015-2018 Intel Corporation. + * Copyright(c) 2015-2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -54,6 +54,7 @@ #include <linux/module.h> #include <linux/prefetch.h> #include <rdma/ib_verbs.h> +#include <linux/etherdevice.h> #include "hfi.h" #include "trace.h" @@ -63,6 +64,9 @@ #include "vnic.h" #include "fault.h" +#include "ipoib.h" +#include "netdev.h" + #undef pr_fmt #define pr_fmt(fmt) DRIVER_NAME ": " fmt @@ -748,6 +752,39 @@ static noinline int skip_rcv_packet(struct hfi1_packet *packet, int thread) return ret; } +static void process_rcv_packet_napi(struct hfi1_packet *packet) +{ + packet->etype = rhf_rcv_type(packet->rhf); + + /* total length */ + packet->tlen = rhf_pkt_len(packet->rhf); /* in bytes */ + /* retrieve eager buffer details */ + packet->etail = rhf_egr_index(packet->rhf); + packet->ebuf = get_egrbuf(packet->rcd, packet->rhf, + &packet->updegr); + /* + * Prefetch the contents of the eager buffer. It is + * OK to send a negative length to prefetch_range(). + * The +2 is the size of the RHF. + */ + prefetch_range(packet->ebuf, + packet->tlen - ((packet->rcd->rcvhdrqentsize - + (rhf_hdrq_offset(packet->rhf) + + 2)) * 4)); + + packet->rcd->rhf_rcv_function_map[packet->etype](packet); + packet->numpkt++; + + /* Set up for the next packet */ + packet->rhqoff += packet->rsize; + if (packet->rhqoff >= packet->maxcnt) + packet->rhqoff = 0; + + packet->rhf_addr = (__le32 *)packet->rcd->rcvhdrq + packet->rhqoff + + packet->rcd->rhf_offset; + packet->rhf = rhf_to_cpu(packet->rhf_addr); +} + static inline int process_rcv_packet(struct hfi1_packet *packet, int thread) { int ret; @@ -827,6 +864,36 @@ static inline void finish_packet(struct hfi1_packet *packet) } /* + * handle_receive_interrupt_napi_fp - receive a packet + * @rcd: the context + * @budget: polling budget + * + * Called from interrupt handler for receive interrupt. + * This is the fast path interrupt handler + * when executing napi soft irq environment. + */ +int handle_receive_interrupt_napi_fp(struct hfi1_ctxtdata *rcd, int budget) +{ + struct hfi1_packet packet; + + init_packet(rcd, &packet); + if (last_rcv_seq(rcd, rhf_rcv_seq(packet.rhf))) + goto bail; + + while (packet.numpkt < budget) { + process_rcv_packet_napi(&packet); + if (hfi1_seq_incr(rcd, rhf_rcv_seq(packet.rhf))) + break; + + process_rcv_update(0, &packet); + } + hfi1_set_rcd_head(rcd, packet.rhqoff); +bail: + finish_packet(&packet); + return packet.numpkt; +} + +/* * Handle receive interrupts when using the no dma rtail option. */ int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *rcd, int thread) @@ -1074,6 +1141,63 @@ bail: } /* + * handle_receive_interrupt_napi_sp - receive a packet + * @rcd: the context + * @budget: polling budget + * + * Called from interrupt handler for errors or receive interrupt. + * This is the slow path interrupt handler + * when executing napi soft irq environment. + */ +int handle_receive_interrupt_napi_sp(struct hfi1_ctxtdata *rcd, int budget) +{ + struct hfi1_devdata *dd = rcd->dd; + int last = RCV_PKT_OK; + bool needset = true; + struct hfi1_packet packet; + + init_packet(rcd, &packet); + if (last_rcv_seq(rcd, rhf_rcv_seq(packet.rhf))) + goto bail; + + while (last != RCV_PKT_DONE && packet.numpkt < budget) { + if (hfi1_need_drop(dd)) { + /* On to the next packet */ + packet.rhqoff += packet.rsize; + packet.rhf_addr = (__le32 *)rcd->rcvhdrq + + packet.rhqoff + + rcd->rhf_offset; + packet.rhf = rhf_to_cpu(packet.rhf_addr); + + } else { + if (set_armed_to_active(&packet)) + goto bail; + process_rcv_packet_napi(&packet); + } + + if (hfi1_seq_incr(rcd, rhf_rcv_seq(packet.rhf))) + last = RCV_PKT_DONE; + + if (needset) { + needset = false; + set_all_fastpath(dd, rcd); + } + + process_rcv_update(last, &packet); + } + + hfi1_set_rcd_head(rcd, packet.rhqoff); + +bail: + /* + * Always write head at end, and setup rcv interrupt, even + * if no packets were processed. + */ + finish_packet(&packet); + return packet.numpkt; +} + +/* * We may discover in the interrupt that the hardware link state has * changed from ARMED to ACTIVE (due to the arrival of a non-SC15 packet), * and we need to update the driver's notion of the link state. We cannot @@ -1550,6 +1674,82 @@ void handle_eflags(struct hfi1_packet *packet) show_eflags_errs(packet); } +static void hfi1_ipoib_ib_rcv(struct hfi1_packet *packet) +{ + struct hfi1_ibport *ibp; + struct net_device *netdev; + struct hfi1_ctxtdata *rcd = packet->rcd; + struct napi_struct *napi = rcd->napi; + struct sk_buff *skb; + struct hfi1_netdev_rxq *rxq = container_of(napi, + struct hfi1_netdev_rxq, napi); + u32 extra_bytes; + u32 tlen, qpnum; + bool do_work, do_cnp; + struct hfi1_ipoib_dev_priv *priv; + + trace_hfi1_rcvhdr(packet); + + hfi1_setup_ib_header(packet); + + packet->ohdr = &((struct ib_header *)packet->hdr)->u.oth; + packet->grh = NULL; + + if (unlikely(rhf_err_flags(packet->rhf))) { + handle_eflags(packet); + return; + } + + qpnum = ib_bth_get_qpn(packet->ohdr); + netdev = hfi1_netdev_get_data(rcd->dd, qpnum); + if (!netdev) + goto drop_no_nd; + + trace_input_ibhdr(rcd->dd, packet, !!(rhf_dc_info(packet->rhf))); + trace_ctxt_rsm_hist(rcd->ctxt); + + /* handle congestion notifications */ + do_work = hfi1_may_ecn(packet); + if (unlikely(do_work)) { + do_cnp = (packet->opcode != IB_OPCODE_CNP); + (void)hfi1_process_ecn_slowpath(hfi1_ipoib_priv(netdev)->qp, + packet, do_cnp); + } + + /* + * We have split point after last byte of DETH + * lets strip padding and CRC and ICRC. + * tlen is whole packet len so we need to + * subtract header size as well. + */ + tlen = packet->tlen; + extra_bytes = ib_bth_get_pad(packet->ohdr) + (SIZE_OF_CRC << 2) + + packet->hlen; + if (unlikely(tlen < extra_bytes)) + goto drop; + + tlen -= extra_bytes; + + skb = hfi1_ipoib_prepare_skb(rxq, tlen, packet->ebuf); + if (unlikely(!skb)) + goto drop; + + priv = hfi1_ipoib_priv(netdev); + hfi1_ipoib_update_rx_netstats(priv, 1, skb->len); + + skb->dev = netdev; + skb->pkt_type = PACKET_HOST; + netif_receive_skb(skb); + + return; + +drop: + ++netdev->stats.rx_dropped; +drop_no_nd: + ibp = rcd_to_iport(packet->rcd); + ++ibp->rvp.n_pkt_drops; +} + /* * The following functions are called by the interrupt handler. They are type * specific handlers for each packet type. @@ -1572,28 +1772,10 @@ static void process_receive_ib(struct hfi1_packet *packet) hfi1_ib_rcv(packet); } -static inline bool hfi1_is_vnic_packet(struct hfi1_packet *packet) -{ - /* Packet received in VNIC context via RSM */ - if (packet->rcd->is_vnic) - return true; - - if ((hfi1_16B_get_l2(packet->ebuf) == OPA_16B_L2_TYPE) && - (hfi1_16B_get_l4(packet->ebuf) == OPA_16B_L4_ETHR)) - return true; - - return false; -} - static void process_receive_bypass(struct hfi1_packet *packet) { struct hfi1_devdata *dd = packet->rcd->dd; - if (hfi1_is_vnic_packet(packet)) { - hfi1_vnic_bypass_rcv(packet); - return; - } - if (hfi1_setup_bypass_packet(packet)) return; @@ -1757,3 +1939,14 @@ const rhf_rcv_function_ptr normal_rhf_rcv_functions[] = { [RHF_RCV_TYPE_INVALID6] = process_receive_invalid, [RHF_RCV_TYPE_INVALID7] = process_receive_invalid, }; + +const rhf_rcv_function_ptr netdev_rhf_rcv_functions[] = { + [RHF_RCV_TYPE_EXPECTED] = process_receive_invalid, + [RHF_RCV_TYPE_EAGER] = process_receive_invalid, + [RHF_RCV_TYPE_IB] = hfi1_ipoib_ib_rcv, + [RHF_RCV_TYPE_ERROR] = process_receive_error, + [RHF_RCV_TYPE_BYPASS] = hfi1_vnic_bypass_rcv, + [RHF_RCV_TYPE_INVALID5] = process_receive_invalid, + [RHF_RCV_TYPE_INVALID6] = process_receive_invalid, + [RHF_RCV_TYPE_INVALID7] = process_receive_invalid, +}; diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c index e7fdd70c6e78..8ca51e43cf53 100644 --- a/drivers/infiniband/hw/hfi1/file_ops.c +++ b/drivers/infiniband/hw/hfi1/file_ops.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015-2017 Intel Corporation. + * Copyright(c) 2015-2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -1264,7 +1264,7 @@ static int get_base_info(struct hfi1_filedata *fd, unsigned long arg, u32 len) memset(&binfo, 0, sizeof(binfo)); binfo.hw_version = dd->revision; binfo.sw_version = HFI1_KERN_SWVERSION; - binfo.bthqp = kdeth_qp; + binfo.bthqp = RVT_KDETH_QP_PREFIX; binfo.jkey = uctxt->jkey; /* * If more than 64 contexts are enabled the allocated credit diff --git a/drivers/infiniband/hw/hfi1/firmware.c b/drivers/infiniband/hw/hfi1/firmware.c index 2b57ba70ddd6..0e83d4b61e46 100644 --- a/drivers/infiniband/hw/hfi1/firmware.c +++ b/drivers/infiniband/hw/hfi1/firmware.c @@ -1868,11 +1868,8 @@ int parse_platform_config(struct hfi1_devdata *dd) 2; break; case PLATFORM_CONFIG_RX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_TX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_QSFP_ATTEN_TABLE: - /* fall through */ case PLATFORM_CONFIG_VARIABLE_SETTINGS_TABLE: pcfgcache->config_tables[table_type].num_table = table_length_dwords; @@ -1890,15 +1887,10 @@ int parse_platform_config(struct hfi1_devdata *dd) /* metadata table */ switch (table_type) { case PLATFORM_CONFIG_SYSTEM_TABLE: - /* fall through */ case PLATFORM_CONFIG_PORT_TABLE: - /* fall through */ case PLATFORM_CONFIG_RX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_TX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_QSFP_ATTEN_TABLE: - /* fall through */ case PLATFORM_CONFIG_VARIABLE_SETTINGS_TABLE: break; default: @@ -2027,15 +2019,10 @@ static int get_platform_fw_field_metadata(struct hfi1_devdata *dd, int table, switch (table) { case PLATFORM_CONFIG_SYSTEM_TABLE: - /* fall through */ case PLATFORM_CONFIG_PORT_TABLE: - /* fall through */ case PLATFORM_CONFIG_RX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_TX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_QSFP_ATTEN_TABLE: - /* fall through */ case PLATFORM_CONFIG_VARIABLE_SETTINGS_TABLE: if (field && field < platform_config_table_limits[table]) src_ptr = @@ -2138,11 +2125,8 @@ int get_platform_config_field(struct hfi1_devdata *dd, pcfgcache->config_tables[table_type].table; break; case PLATFORM_CONFIG_RX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_TX_PRESET_TABLE: - /* fall through */ case PLATFORM_CONFIG_QSFP_ATTEN_TABLE: - /* fall through */ case PLATFORM_CONFIG_VARIABLE_SETTINGS_TABLE: src_ptr = pcfgcache->config_tables[table_type].table; diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h index b06c2594105a..b4c6bff60a4e 100644 --- a/drivers/infiniband/hw/hfi1/hfi.h +++ b/drivers/infiniband/hw/hfi1/hfi.h @@ -1,7 +1,7 @@ #ifndef _HFI1_KERNEL_H #define _HFI1_KERNEL_H /* - * Copyright(c) 2015-2018 Intel Corporation. + * Copyright(c) 2015-2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -233,6 +233,8 @@ struct hfi1_ctxtdata { intr_handler fast_handler; /** slow handler */ intr_handler slow_handler; + /* napi pointer assiociated with netdev */ + struct napi_struct *napi; /* verbs rx_stats per rcd */ struct hfi1_opcode_stats_perctx *opstats; /* clear interrupt mask */ @@ -383,11 +385,11 @@ struct hfi1_packet { u32 rhqoff; u32 dlid; u32 slid; + int numpkt; u16 tlen; s16 etail; u16 pkey; u8 hlen; - u8 numpkt; u8 rsize; u8 updegr; u8 etype; @@ -985,7 +987,7 @@ typedef void (*hfi1_make_req)(struct rvt_qp *qp, struct hfi1_pkt_state *ps, struct rvt_swqe *wqe); extern const rhf_rcv_function_ptr normal_rhf_rcv_functions[]; - +extern const rhf_rcv_function_ptr netdev_rhf_rcv_functions[]; /* return values for the RHF receive functions */ #define RHF_RCV_CONTINUE 0 /* keep going */ @@ -1045,23 +1047,10 @@ struct hfi1_asic_data { #define NUM_MAP_ENTRIES 256 #define NUM_MAP_REGS 32 -/* - * Number of VNIC contexts used. Ensure it is less than or equal to - * max queues supported by VNIC (HFI1_VNIC_MAX_QUEUE). - */ -#define HFI1_NUM_VNIC_CTXT 8 - -/* Number of VNIC RSM entries */ -#define NUM_VNIC_MAP_ENTRIES 8 - /* Virtual NIC information */ struct hfi1_vnic_data { - struct hfi1_ctxtdata *ctxt[HFI1_NUM_VNIC_CTXT]; struct kmem_cache *txreq_cache; - struct xarray vesws; u8 num_vports; - u8 rmt_start; - u8 num_ctxt; }; struct hfi1_vnic_vport_info; @@ -1167,8 +1156,8 @@ struct hfi1_devdata { u64 z_send_schedule; u64 __percpu *send_schedule; - /* number of reserved contexts for VNIC usage */ - u16 num_vnic_contexts; + /* number of reserved contexts for netdev usage */ + u16 num_netdev_contexts; /* number of receive contexts in use by the driver */ u32 num_rcv_contexts; /* number of pio send contexts in use by the driver */ @@ -1417,12 +1406,12 @@ struct hfi1_devdata { struct hfi1_vnic_data vnic; /* Lock to protect IRQ SRC register access */ spinlock_t irq_src_lock; -}; + int vnic_num_vports; + struct net_device *dummy_netdev; -static inline bool hfi1_vnic_is_rsm_full(struct hfi1_devdata *dd, int spare) -{ - return (dd->vnic.rmt_start + spare) > NUM_MAP_ENTRIES; -} + /* Keeps track of IPoIB RSM rule users */ + atomic_t ipoib_rsm_usr_num; +}; /* 8051 firmware version helper */ #define dc8051_ver(a, b, c) ((a) << 16 | (b) << 8 | (c)) @@ -1500,6 +1489,8 @@ struct hfi1_ctxtdata *hfi1_rcd_get_by_index(struct hfi1_devdata *dd, u16 ctxt); int handle_receive_interrupt(struct hfi1_ctxtdata *rcd, int thread); int handle_receive_interrupt_nodma_rtail(struct hfi1_ctxtdata *rcd, int thread); int handle_receive_interrupt_dma_rtail(struct hfi1_ctxtdata *rcd, int thread); +int handle_receive_interrupt_napi_fp(struct hfi1_ctxtdata *rcd, int budget); +int handle_receive_interrupt_napi_sp(struct hfi1_ctxtdata *rcd, int budget); void set_all_slowpath(struct hfi1_devdata *dd); extern const struct pci_device_id hfi1_pci_tbl[]; @@ -2250,7 +2241,6 @@ extern int num_user_contexts; extern unsigned long n_krcvqs; extern uint krcvqs[]; extern int krcvqsset; -extern uint kdeth_qp; extern uint loopback; extern uint quick_linkup; extern uint rcv_intr_timeout; diff --git a/drivers/infiniband/hw/hfi1/init.c b/drivers/infiniband/hw/hfi1/init.c index 3759d9233a1c..cb7ad1288821 100644 --- a/drivers/infiniband/hw/hfi1/init.c +++ b/drivers/infiniband/hw/hfi1/init.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -69,6 +69,7 @@ #include "affinity.h" #include "vnic.h" #include "exp_rcv.h" +#include "netdev.h" #undef pr_fmt #define pr_fmt(fmt) DRIVER_NAME ": " fmt @@ -374,6 +375,7 @@ int hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, int numa, rcd->numa_id = numa; rcd->rcv_array_groups = dd->rcv_entries.ngroups; rcd->rhf_rcv_function_map = normal_rhf_rcv_functions; + rcd->msix_intr = CCE_NUM_MSIX_VECTORS; mutex_init(&rcd->exp_mutex); spin_lock_init(&rcd->exp_lock); @@ -829,6 +831,29 @@ wq_error: } /** + * destroy_workqueues - destroy per port workqueues + * @dd: the hfi1_ib device + */ +static void destroy_workqueues(struct hfi1_devdata *dd) +{ + int pidx; + struct hfi1_pportdata *ppd; + + for (pidx = 0; pidx < dd->num_pports; ++pidx) { + ppd = dd->pport + pidx; + + if (ppd->hfi1_wq) { + destroy_workqueue(ppd->hfi1_wq); + ppd->hfi1_wq = NULL; + } + if (ppd->link_wq) { + destroy_workqueue(ppd->link_wq); + ppd->link_wq = NULL; + } + } +} + +/** * enable_general_intr() - Enable the IRQs that will be handled by the * general interrupt handler. * @dd: valid devdata @@ -1101,15 +1126,10 @@ static void shutdown_device(struct hfi1_devdata *dd) * We can't count on interrupts since we are stopping. */ hfi1_quiet_serdes(ppd); - - if (ppd->hfi1_wq) { - destroy_workqueue(ppd->hfi1_wq); - ppd->hfi1_wq = NULL; - } - if (ppd->link_wq) { - destroy_workqueue(ppd->link_wq); - ppd->link_wq = NULL; - } + if (ppd->hfi1_wq) + flush_workqueue(ppd->hfi1_wq); + if (ppd->link_wq) + flush_workqueue(ppd->link_wq); } sdma_exit(dd); } @@ -1316,6 +1336,7 @@ static struct hfi1_devdata *hfi1_alloc_devdata(struct pci_dev *pdev, goto bail; } + atomic_set(&dd->ipoib_rsm_usr_num, 0); return dd; bail: @@ -1663,9 +1684,6 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent) /* do the generic initialization */ initfail = hfi1_init(dd, 0); - /* setup vnic */ - hfi1_vnic_setup(dd); - ret = hfi1_register_ib_device(dd); /* @@ -1704,7 +1722,6 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent) hfi1_device_remove(dd); if (!ret) hfi1_unregister_ib_device(dd); - hfi1_vnic_cleanup(dd); postinit_cleanup(dd); if (initfail) ret = initfail; @@ -1749,14 +1766,15 @@ static void remove_one(struct pci_dev *pdev) /* unregister from IB core */ hfi1_unregister_ib_device(dd); - /* cleanup vnic */ - hfi1_vnic_cleanup(dd); + /* free netdev data */ + hfi1_netdev_free(dd); /* * Disable the IB link, disable interrupts on the device, * clear dma engines, etc. */ shutdown_device(dd); + destroy_workqueues(dd); stop_timers(dd); diff --git a/drivers/infiniband/hw/hfi1/iowait.h b/drivers/infiniband/hw/hfi1/iowait.h index 07847cb72169..d580aa17ae37 100644 --- a/drivers/infiniband/hw/hfi1/iowait.h +++ b/drivers/infiniband/hw/hfi1/iowait.h @@ -399,7 +399,7 @@ static inline void iowait_get_priority(struct iowait *w) * @wait_head: the wait queue * * This function is called to insert an iowait struct into a - * wait queue after a resource (eg, sdma decriptor or pio + * wait queue after a resource (eg, sdma descriptor or pio * buffer) is run out. */ static inline void iowait_queue(bool pkts_sent, struct iowait *w, diff --git a/drivers/infiniband/hw/hfi1/ipoib.h b/drivers/infiniband/hw/hfi1/ipoib.h new file mode 100644 index 000000000000..b8c9d0a003fb --- /dev/null +++ b/drivers/infiniband/hw/hfi1/ipoib.h @@ -0,0 +1,177 @@ +/* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */ +/* + * Copyright(c) 2020 Intel Corporation. + * + */ + +/* + * This file contains HFI1 support for IPOIB functionality + */ + +#ifndef HFI1_IPOIB_H +#define HFI1_IPOIB_H + +#include <linux/types.h> +#include <linux/stddef.h> +#include <linux/atomic.h> +#include <linux/netdevice.h> +#include <linux/slab.h> +#include <linux/skbuff.h> +#include <linux/list.h> +#include <linux/if_infiniband.h> + +#include "hfi.h" +#include "iowait.h" +#include "netdev.h" + +#include <rdma/ib_verbs.h> + +#define HFI1_IPOIB_ENTROPY_SHIFT 24 + +#define HFI1_IPOIB_TXREQ_NAME_LEN 32 + +#define HFI1_IPOIB_PSEUDO_LEN 20 +#define HFI1_IPOIB_ENCAP_LEN 4 + +struct hfi1_ipoib_dev_priv; + +union hfi1_ipoib_flow { + u16 as_int; + struct { + u8 tx_queue; + u8 sc5; + } __attribute__((__packed__)); +}; + +/** + * struct hfi1_ipoib_circ_buf - List of items to be processed + * @items: ring of items + * @head: ring head + * @tail: ring tail + * @max_items: max items + 1 that the ring can contain + * @producer_lock: producer sync lock + * @consumer_lock: consumer sync lock + */ +struct hfi1_ipoib_circ_buf { + void **items; + unsigned long head; + unsigned long tail; + unsigned long max_items; + spinlock_t producer_lock; /* head sync lock */ + spinlock_t consumer_lock; /* tail sync lock */ +}; + +/** + * struct hfi1_ipoib_txq - IPOIB per Tx queue information + * @priv: private pointer + * @sde: sdma engine + * @tx_list: tx request list + * @sent_txreqs: count of txreqs posted to sdma + * @stops: count of stops of queue + * @ring_full: ring has been filled + * @no_desc: descriptor shortage seen + * @flow: tracks when list needs to be flushed for a flow change + * @q_idx: ipoib Tx queue index + * @pkts_sent: indicator packets have been sent from this queue + * @wait: iowait structure + * @complete_txreqs: count of txreqs completed by sdma + * @napi: pointer to tx napi interface + * @tx_ring: ring of ipoib txreqs to be reaped by napi callback + */ +struct hfi1_ipoib_txq { + struct hfi1_ipoib_dev_priv *priv; + struct sdma_engine *sde; + struct list_head tx_list; + u64 sent_txreqs; + atomic_t stops; + atomic_t ring_full; + atomic_t no_desc; + union hfi1_ipoib_flow flow; + u8 q_idx; + bool pkts_sent; + struct iowait wait; + + atomic64_t ____cacheline_aligned_in_smp complete_txreqs; + struct napi_struct *napi; + struct hfi1_ipoib_circ_buf tx_ring; +}; + +struct hfi1_ipoib_dev_priv { + struct hfi1_devdata *dd; + struct net_device *netdev; + struct ib_device *device; + struct hfi1_ipoib_txq *txqs; + struct kmem_cache *txreq_cache; + struct napi_struct *tx_napis; + u16 pkey; + u16 pkey_index; + u32 qkey; + u8 port_num; + + const struct net_device_ops *netdev_ops; + struct rvt_qp *qp; + struct pcpu_sw_netstats __percpu *netstats; +}; + +/* hfi1 ipoib rdma netdev's private data structure */ +struct hfi1_ipoib_rdma_netdev { + struct rdma_netdev rn; /* keep this first */ + /* followed by device private data */ + struct hfi1_ipoib_dev_priv dev_priv; +}; + +static inline struct hfi1_ipoib_dev_priv * +hfi1_ipoib_priv(const struct net_device *dev) +{ + return &((struct hfi1_ipoib_rdma_netdev *)netdev_priv(dev))->dev_priv; +} + +static inline void +hfi1_ipoib_update_rx_netstats(struct hfi1_ipoib_dev_priv *priv, + u64 packets, + u64 bytes) +{ + struct pcpu_sw_netstats *netstats = this_cpu_ptr(priv->netstats); + + u64_stats_update_begin(&netstats->syncp); + netstats->rx_packets += packets; + netstats->rx_bytes += bytes; + u64_stats_update_end(&netstats->syncp); +} + +static inline void +hfi1_ipoib_update_tx_netstats(struct hfi1_ipoib_dev_priv *priv, + u64 packets, + u64 bytes) +{ + struct pcpu_sw_netstats *netstats = this_cpu_ptr(priv->netstats); + + u64_stats_update_begin(&netstats->syncp); + netstats->tx_packets += packets; + netstats->tx_bytes += bytes; + u64_stats_update_end(&netstats->syncp); +} + +int hfi1_ipoib_send_dma(struct net_device *dev, + struct sk_buff *skb, + struct ib_ah *address, + u32 dqpn); + +int hfi1_ipoib_txreq_init(struct hfi1_ipoib_dev_priv *priv); +void hfi1_ipoib_txreq_deinit(struct hfi1_ipoib_dev_priv *priv); + +int hfi1_ipoib_rxq_init(struct net_device *dev); +void hfi1_ipoib_rxq_deinit(struct net_device *dev); + +void hfi1_ipoib_napi_tx_enable(struct net_device *dev); +void hfi1_ipoib_napi_tx_disable(struct net_device *dev); + +struct sk_buff *hfi1_ipoib_prepare_skb(struct hfi1_netdev_rxq *rxq, + int size, void *data); + +int hfi1_ipoib_rn_get_params(struct ib_device *device, + u8 port_num, + enum rdma_netdev_t type, + struct rdma_netdev_alloc_params *params); + +#endif /* _IPOIB_H */ diff --git a/drivers/infiniband/hw/hfi1/ipoib_main.c b/drivers/infiniband/hw/hfi1/ipoib_main.c new file mode 100644 index 000000000000..014351ebbefa --- /dev/null +++ b/drivers/infiniband/hw/hfi1/ipoib_main.c @@ -0,0 +1,309 @@ +// SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) +/* + * Copyright(c) 2020 Intel Corporation. + * + */ + +/* + * This file contains HFI1 support for ipoib functionality + */ + +#include "ipoib.h" +#include "hfi.h" + +static u32 qpn_from_mac(u8 *mac_arr) +{ + return (u32)mac_arr[1] << 16 | mac_arr[2] << 8 | mac_arr[3]; +} + +static int hfi1_ipoib_dev_init(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + int ret; + + priv->netstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats); + + ret = priv->netdev_ops->ndo_init(dev); + if (ret) + return ret; + + ret = hfi1_netdev_add_data(priv->dd, + qpn_from_mac(priv->netdev->dev_addr), + dev); + if (ret < 0) { + priv->netdev_ops->ndo_uninit(dev); + return ret; + } + + return 0; +} + +static void hfi1_ipoib_dev_uninit(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + + hfi1_netdev_remove_data(priv->dd, qpn_from_mac(priv->netdev->dev_addr)); + + priv->netdev_ops->ndo_uninit(dev); +} + +static int hfi1_ipoib_dev_open(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + int ret; + + ret = priv->netdev_ops->ndo_open(dev); + if (!ret) { + struct hfi1_ibport *ibp = to_iport(priv->device, + priv->port_num); + struct rvt_qp *qp; + u32 qpn = qpn_from_mac(priv->netdev->dev_addr); + + rcu_read_lock(); + qp = rvt_lookup_qpn(ib_to_rvt(priv->device), &ibp->rvp, qpn); + if (!qp) { + rcu_read_unlock(); + priv->netdev_ops->ndo_stop(dev); + return -EINVAL; + } + rvt_get_qp(qp); + priv->qp = qp; + rcu_read_unlock(); + + hfi1_netdev_enable_queues(priv->dd); + hfi1_ipoib_napi_tx_enable(dev); + } + + return ret; +} + +static int hfi1_ipoib_dev_stop(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + + if (!priv->qp) + return 0; + + hfi1_ipoib_napi_tx_disable(dev); + hfi1_netdev_disable_queues(priv->dd); + + rvt_put_qp(priv->qp); + priv->qp = NULL; + + return priv->netdev_ops->ndo_stop(dev); +} + +static void hfi1_ipoib_dev_get_stats64(struct net_device *dev, + struct rtnl_link_stats64 *storage) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + u64 rx_packets = 0ull; + u64 rx_bytes = 0ull; + u64 tx_packets = 0ull; + u64 tx_bytes = 0ull; + int i; + + netdev_stats_to_stats64(storage, &dev->stats); + + for_each_possible_cpu(i) { + const struct pcpu_sw_netstats *stats; + unsigned int start; + u64 trx_packets; + u64 trx_bytes; + u64 ttx_packets; + u64 ttx_bytes; + + stats = per_cpu_ptr(priv->netstats, i); + do { + start = u64_stats_fetch_begin_irq(&stats->syncp); + trx_packets = stats->rx_packets; + trx_bytes = stats->rx_bytes; + ttx_packets = stats->tx_packets; + ttx_bytes = stats->tx_bytes; + } while (u64_stats_fetch_retry_irq(&stats->syncp, start)); + + rx_packets += trx_packets; + rx_bytes += trx_bytes; + tx_packets += ttx_packets; + tx_bytes += ttx_bytes; + } + + storage->rx_packets += rx_packets; + storage->rx_bytes += rx_bytes; + storage->tx_packets += tx_packets; + storage->tx_bytes += tx_bytes; +} + +static const struct net_device_ops hfi1_ipoib_netdev_ops = { + .ndo_init = hfi1_ipoib_dev_init, + .ndo_uninit = hfi1_ipoib_dev_uninit, + .ndo_open = hfi1_ipoib_dev_open, + .ndo_stop = hfi1_ipoib_dev_stop, + .ndo_get_stats64 = hfi1_ipoib_dev_get_stats64, +}; + +static int hfi1_ipoib_send(struct net_device *dev, + struct sk_buff *skb, + struct ib_ah *address, + u32 dqpn) +{ + return hfi1_ipoib_send_dma(dev, skb, address, dqpn); +} + +static int hfi1_ipoib_mcast_attach(struct net_device *dev, + struct ib_device *device, + union ib_gid *mgid, + u16 mlid, + int set_qkey, + u32 qkey) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + u32 qpn = (u32)qpn_from_mac(priv->netdev->dev_addr); + struct hfi1_ibport *ibp = to_iport(priv->device, priv->port_num); + struct rvt_qp *qp; + int ret = -EINVAL; + + rcu_read_lock(); + + qp = rvt_lookup_qpn(ib_to_rvt(priv->device), &ibp->rvp, qpn); + if (qp) { + rvt_get_qp(qp); + rcu_read_unlock(); + if (set_qkey) + priv->qkey = qkey; + + /* attach QP to multicast group */ + ret = ib_attach_mcast(&qp->ibqp, mgid, mlid); + rvt_put_qp(qp); + } else { + rcu_read_unlock(); + } + + return ret; +} + +static int hfi1_ipoib_mcast_detach(struct net_device *dev, + struct ib_device *device, + union ib_gid *mgid, + u16 mlid) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + u32 qpn = (u32)qpn_from_mac(priv->netdev->dev_addr); + struct hfi1_ibport *ibp = to_iport(priv->device, priv->port_num); + struct rvt_qp *qp; + int ret = -EINVAL; + + rcu_read_lock(); + + qp = rvt_lookup_qpn(ib_to_rvt(priv->device), &ibp->rvp, qpn); + if (qp) { + rvt_get_qp(qp); + rcu_read_unlock(); + ret = ib_detach_mcast(&qp->ibqp, mgid, mlid); + rvt_put_qp(qp); + } else { + rcu_read_unlock(); + } + return ret; +} + +static void hfi1_ipoib_netdev_dtor(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + + hfi1_ipoib_txreq_deinit(priv); + hfi1_ipoib_rxq_deinit(priv->netdev); + + free_percpu(priv->netstats); +} + +static void hfi1_ipoib_free_rdma_netdev(struct net_device *dev) +{ + hfi1_ipoib_netdev_dtor(dev); + free_netdev(dev); +} + +static void hfi1_ipoib_set_id(struct net_device *dev, int id) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + + priv->pkey_index = (u16)id; + ib_query_pkey(priv->device, + priv->port_num, + priv->pkey_index, + &priv->pkey); +} + +static int hfi1_ipoib_setup_rn(struct ib_device *device, + u8 port_num, + struct net_device *netdev, + void *param) +{ + struct hfi1_devdata *dd = dd_from_ibdev(device); + struct rdma_netdev *rn = netdev_priv(netdev); + struct hfi1_ipoib_dev_priv *priv; + int rc; + + rn->send = hfi1_ipoib_send; + rn->attach_mcast = hfi1_ipoib_mcast_attach; + rn->detach_mcast = hfi1_ipoib_mcast_detach; + rn->set_id = hfi1_ipoib_set_id; + rn->hca = device; + rn->port_num = port_num; + rn->mtu = netdev->mtu; + + priv = hfi1_ipoib_priv(netdev); + priv->dd = dd; + priv->netdev = netdev; + priv->device = device; + priv->port_num = port_num; + priv->netdev_ops = netdev->netdev_ops; + + netdev->netdev_ops = &hfi1_ipoib_netdev_ops; + + ib_query_pkey(device, port_num, priv->pkey_index, &priv->pkey); + + rc = hfi1_ipoib_txreq_init(priv); + if (rc) { + dd_dev_err(dd, "IPoIB netdev TX init - failed(%d)\n", rc); + hfi1_ipoib_free_rdma_netdev(netdev); + return rc; + } + + rc = hfi1_ipoib_rxq_init(netdev); + if (rc) { + dd_dev_err(dd, "IPoIB netdev RX init - failed(%d)\n", rc); + hfi1_ipoib_free_rdma_netdev(netdev); + return rc; + } + + netdev->priv_destructor = hfi1_ipoib_netdev_dtor; + netdev->needs_free_netdev = true; + + return 0; +} + +int hfi1_ipoib_rn_get_params(struct ib_device *device, + u8 port_num, + enum rdma_netdev_t type, + struct rdma_netdev_alloc_params *params) +{ + struct hfi1_devdata *dd = dd_from_ibdev(device); + + if (type != RDMA_NETDEV_IPOIB) + return -EOPNOTSUPP; + + if (!HFI1_CAP_IS_KSET(AIP) || !dd->num_netdev_contexts) + return -EOPNOTSUPP; + + if (!port_num || port_num > dd->num_pports) + return -EINVAL; + + params->sizeof_priv = sizeof(struct hfi1_ipoib_rdma_netdev); + params->txqs = dd->num_sdma; + params->rxqs = dd->num_netdev_contexts; + params->param = NULL; + params->initialize_rdma_netdev = hfi1_ipoib_setup_rn; + + return 0; +} diff --git a/drivers/infiniband/hw/hfi1/ipoib_rx.c b/drivers/infiniband/hw/hfi1/ipoib_rx.c new file mode 100644 index 000000000000..3afa7545242c --- /dev/null +++ b/drivers/infiniband/hw/hfi1/ipoib_rx.c @@ -0,0 +1,95 @@ +// SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) +/* + * Copyright(c) 2020 Intel Corporation. + * + */ + +#include "netdev.h" +#include "ipoib.h" + +#define HFI1_IPOIB_SKB_PAD ((NET_SKB_PAD) + (NET_IP_ALIGN)) + +static void copy_ipoib_buf(struct sk_buff *skb, void *data, int size) +{ + void *dst_data; + + skb_checksum_none_assert(skb); + skb->protocol = *((__be16 *)data); + + dst_data = skb_put(skb, size); + memcpy(dst_data, data, size); + skb->mac_header = HFI1_IPOIB_PSEUDO_LEN; + skb_pull(skb, HFI1_IPOIB_ENCAP_LEN); +} + +static struct sk_buff *prepare_frag_skb(struct napi_struct *napi, int size) +{ + struct sk_buff *skb; + int skb_size = SKB_DATA_ALIGN(size + HFI1_IPOIB_SKB_PAD); + void *frag; + + skb_size += SKB_DATA_ALIGN(sizeof(struct skb_shared_info)); + skb_size = SKB_DATA_ALIGN(skb_size); + frag = napi_alloc_frag(skb_size); + + if (unlikely(!frag)) + return napi_alloc_skb(napi, size); + + skb = build_skb(frag, skb_size); + + if (unlikely(!skb)) { + skb_free_frag(frag); + return NULL; + } + + skb_reserve(skb, HFI1_IPOIB_SKB_PAD); + return skb; +} + +struct sk_buff *hfi1_ipoib_prepare_skb(struct hfi1_netdev_rxq *rxq, + int size, void *data) +{ + struct napi_struct *napi = &rxq->napi; + int skb_size = size + HFI1_IPOIB_ENCAP_LEN; + struct sk_buff *skb; + + /* + * For smaller(4k + skb overhead) allocations we will go using + * napi cache. Otherwise we will try to use napi frag cache. + */ + if (size <= SKB_WITH_OVERHEAD(PAGE_SIZE)) + skb = napi_alloc_skb(napi, skb_size); + else + skb = prepare_frag_skb(napi, skb_size); + + if (unlikely(!skb)) + return NULL; + + copy_ipoib_buf(skb, data, size); + + return skb; +} + +int hfi1_ipoib_rxq_init(struct net_device *netdev) +{ + struct hfi1_ipoib_dev_priv *ipoib_priv = hfi1_ipoib_priv(netdev); + struct hfi1_devdata *dd = ipoib_priv->dd; + int ret; + + ret = hfi1_netdev_rx_init(dd); + if (ret) + return ret; + + hfi1_init_aip_rsm(dd); + + return ret; +} + +void hfi1_ipoib_rxq_deinit(struct net_device *netdev) +{ + struct hfi1_ipoib_dev_priv *ipoib_priv = hfi1_ipoib_priv(netdev); + struct hfi1_devdata *dd = ipoib_priv->dd; + + hfi1_deinit_aip_rsm(dd); + hfi1_netdev_rx_destroy(dd); +} diff --git a/drivers/infiniband/hw/hfi1/ipoib_tx.c b/drivers/infiniband/hw/hfi1/ipoib_tx.c new file mode 100644 index 000000000000..9df292b51a05 --- /dev/null +++ b/drivers/infiniband/hw/hfi1/ipoib_tx.c @@ -0,0 +1,856 @@ +// SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) +/* + * Copyright(c) 2020 Intel Corporation. + * + */ + +/* + * This file contains HFI1 support for IPOIB SDMA functionality + */ + +#include <linux/log2.h> +#include <linux/circ_buf.h> + +#include "sdma.h" +#include "verbs.h" +#include "trace_ibhdrs.h" +#include "ipoib.h" + +/* Add a convenience helper */ +#define CIRC_ADD(val, add, size) (((val) + (add)) & ((size) - 1)) +#define CIRC_NEXT(val, size) CIRC_ADD(val, 1, size) +#define CIRC_PREV(val, size) CIRC_ADD(val, -1, size) + +/** + * struct ipoib_txreq - IPOIB transmit descriptor + * @txreq: sdma transmit request + * @sdma_hdr: 9b ib headers + * @sdma_status: status returned by sdma engine + * @priv: ipoib netdev private data + * @txq: txq on which skb was output + * @skb: skb to send + */ +struct ipoib_txreq { + struct sdma_txreq txreq; + struct hfi1_sdma_header sdma_hdr; + int sdma_status; + struct hfi1_ipoib_dev_priv *priv; + struct hfi1_ipoib_txq *txq; + struct sk_buff *skb; +}; + +struct ipoib_txparms { + struct hfi1_devdata *dd; + struct rdma_ah_attr *ah_attr; + struct hfi1_ibport *ibp; + struct hfi1_ipoib_txq *txq; + union hfi1_ipoib_flow flow; + u32 dqpn; + u8 hdr_dwords; + u8 entropy; +}; + +static u64 hfi1_ipoib_txreqs(const u64 sent, const u64 completed) +{ + return sent - completed; +} + +static u64 hfi1_ipoib_used(struct hfi1_ipoib_txq *txq) +{ + return hfi1_ipoib_txreqs(txq->sent_txreqs, + atomic64_read(&txq->complete_txreqs)); +} + +static void hfi1_ipoib_stop_txq(struct hfi1_ipoib_txq *txq) +{ + if (atomic_inc_return(&txq->stops) == 1) + netif_stop_subqueue(txq->priv->netdev, txq->q_idx); +} + +static void hfi1_ipoib_wake_txq(struct hfi1_ipoib_txq *txq) +{ + if (atomic_dec_and_test(&txq->stops)) + netif_wake_subqueue(txq->priv->netdev, txq->q_idx); +} + +static uint hfi1_ipoib_ring_hwat(struct hfi1_ipoib_txq *txq) +{ + return min_t(uint, txq->priv->netdev->tx_queue_len, + txq->tx_ring.max_items - 1); +} + +static uint hfi1_ipoib_ring_lwat(struct hfi1_ipoib_txq *txq) +{ + return min_t(uint, txq->priv->netdev->tx_queue_len, + txq->tx_ring.max_items) >> 1; +} + +static void hfi1_ipoib_check_queue_depth(struct hfi1_ipoib_txq *txq) +{ + ++txq->sent_txreqs; + if (hfi1_ipoib_used(txq) >= hfi1_ipoib_ring_hwat(txq) && + !atomic_xchg(&txq->ring_full, 1)) + hfi1_ipoib_stop_txq(txq); +} + +static void hfi1_ipoib_check_queue_stopped(struct hfi1_ipoib_txq *txq) +{ + struct net_device *dev = txq->priv->netdev; + + /* If shutting down just return as queue state is irrelevant */ + if (unlikely(dev->reg_state != NETREG_REGISTERED)) + return; + + /* + * When the queue has been drained to less than half full it will be + * restarted. + * The size of the txreq ring is fixed at initialization. + * The tx queue len can be adjusted upward while the interface is + * running. + * The tx queue len can be large enough to overflow the txreq_ring. + * Use the minimum of the current tx_queue_len or the rings max txreqs + * to protect against ring overflow. + */ + if (hfi1_ipoib_used(txq) < hfi1_ipoib_ring_lwat(txq) && + atomic_xchg(&txq->ring_full, 0)) + hfi1_ipoib_wake_txq(txq); +} + +static void hfi1_ipoib_free_tx(struct ipoib_txreq *tx, int budget) +{ + struct hfi1_ipoib_dev_priv *priv = tx->priv; + + if (likely(!tx->sdma_status)) { + hfi1_ipoib_update_tx_netstats(priv, 1, tx->skb->len); + } else { + ++priv->netdev->stats.tx_errors; + dd_dev_warn(priv->dd, + "%s: Status = 0x%x pbc 0x%llx txq = %d sde = %d\n", + __func__, tx->sdma_status, + le64_to_cpu(tx->sdma_hdr.pbc), tx->txq->q_idx, + tx->txq->sde->this_idx); + } + + napi_consume_skb(tx->skb, budget); + sdma_txclean(priv->dd, &tx->txreq); + kmem_cache_free(priv->txreq_cache, tx); +} + +static int hfi1_ipoib_drain_tx_ring(struct hfi1_ipoib_txq *txq, int budget) +{ + struct hfi1_ipoib_circ_buf *tx_ring = &txq->tx_ring; + unsigned long head; + unsigned long tail; + unsigned int max_tx; + int work_done; + int tx_count; + + spin_lock_bh(&tx_ring->consumer_lock); + + /* Read index before reading contents at that index. */ + head = smp_load_acquire(&tx_ring->head); + tail = tx_ring->tail; + max_tx = tx_ring->max_items; + + work_done = min_t(int, CIRC_CNT(head, tail, max_tx), budget); + + for (tx_count = work_done; tx_count; tx_count--) { + hfi1_ipoib_free_tx(tx_ring->items[tail], budget); + tail = CIRC_NEXT(tail, max_tx); + } + + atomic64_add(work_done, &txq->complete_txreqs); + + /* Finished freeing tx items so store the tail value. */ + smp_store_release(&tx_ring->tail, tail); + + spin_unlock_bh(&tx_ring->consumer_lock); + + hfi1_ipoib_check_queue_stopped(txq); + + return work_done; +} + +static int hfi1_ipoib_process_tx_ring(struct napi_struct *napi, int budget) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(napi->dev); + struct hfi1_ipoib_txq *txq = &priv->txqs[napi - priv->tx_napis]; + + int work_done = hfi1_ipoib_drain_tx_ring(txq, budget); + + if (work_done < budget) + napi_complete_done(napi, work_done); + + return work_done; +} + +static void hfi1_ipoib_add_tx(struct ipoib_txreq *tx) +{ + struct hfi1_ipoib_circ_buf *tx_ring = &tx->txq->tx_ring; + unsigned long head; + unsigned long tail; + size_t max_tx; + + spin_lock(&tx_ring->producer_lock); + + head = tx_ring->head; + tail = READ_ONCE(tx_ring->tail); + max_tx = tx_ring->max_items; + + if (likely(CIRC_SPACE(head, tail, max_tx))) { + tx_ring->items[head] = tx; + + /* Finish storing txreq before incrementing head. */ + smp_store_release(&tx_ring->head, CIRC_ADD(head, 1, max_tx)); + napi_schedule(tx->txq->napi); + } else { + struct hfi1_ipoib_txq *txq = tx->txq; + struct hfi1_ipoib_dev_priv *priv = tx->priv; + + /* Ring was full */ + hfi1_ipoib_free_tx(tx, 0); + atomic64_inc(&txq->complete_txreqs); + dd_dev_dbg(priv->dd, "txq %d full.\n", txq->q_idx); + } + + spin_unlock(&tx_ring->producer_lock); +} + +static void hfi1_ipoib_sdma_complete(struct sdma_txreq *txreq, int status) +{ + struct ipoib_txreq *tx = container_of(txreq, struct ipoib_txreq, txreq); + + tx->sdma_status = status; + + hfi1_ipoib_add_tx(tx); +} + +static int hfi1_ipoib_build_ulp_payload(struct ipoib_txreq *tx, + struct ipoib_txparms *txp) +{ + struct hfi1_devdata *dd = txp->dd; + struct sdma_txreq *txreq = &tx->txreq; + struct sk_buff *skb = tx->skb; + int ret = 0; + int i; + + if (skb_headlen(skb)) { + ret = sdma_txadd_kvaddr(dd, txreq, skb->data, skb_headlen(skb)); + if (unlikely(ret)) + return ret; + } + + for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) { + const skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; + + ret = sdma_txadd_page(dd, + txreq, + skb_frag_page(frag), + frag->bv_offset, + skb_frag_size(frag)); + if (unlikely(ret)) + break; + } + + return ret; +} + +static int hfi1_ipoib_build_tx_desc(struct ipoib_txreq *tx, + struct ipoib_txparms *txp) +{ + struct hfi1_devdata *dd = txp->dd; + struct sdma_txreq *txreq = &tx->txreq; + struct hfi1_sdma_header *sdma_hdr = &tx->sdma_hdr; + u16 pkt_bytes = + sizeof(sdma_hdr->pbc) + (txp->hdr_dwords << 2) + tx->skb->len; + int ret; + + ret = sdma_txinit(txreq, 0, pkt_bytes, hfi1_ipoib_sdma_complete); + if (unlikely(ret)) + return ret; + + /* add pbc + headers */ + ret = sdma_txadd_kvaddr(dd, + txreq, + sdma_hdr, + sizeof(sdma_hdr->pbc) + (txp->hdr_dwords << 2)); + if (unlikely(ret)) + return ret; + + /* add the ulp payload */ + return hfi1_ipoib_build_ulp_payload(tx, txp); +} + +static void hfi1_ipoib_build_ib_tx_headers(struct ipoib_txreq *tx, + struct ipoib_txparms *txp) +{ + struct hfi1_ipoib_dev_priv *priv = tx->priv; + struct hfi1_sdma_header *sdma_hdr = &tx->sdma_hdr; + struct sk_buff *skb = tx->skb; + struct hfi1_pportdata *ppd = ppd_from_ibp(txp->ibp); + struct rdma_ah_attr *ah_attr = txp->ah_attr; + struct ib_other_headers *ohdr; + struct ib_grh *grh; + u16 dwords; + u16 slid; + u16 dlid; + u16 lrh0; + u32 bth0; + u32 sqpn = (u32)(priv->netdev->dev_addr[1] << 16 | + priv->netdev->dev_addr[2] << 8 | + priv->netdev->dev_addr[3]); + u16 payload_dwords; + u8 pad_cnt; + + pad_cnt = -skb->len & 3; + + /* Includes ICRC */ + payload_dwords = ((skb->len + pad_cnt) >> 2) + SIZE_OF_CRC; + + /* header size in dwords LRH+BTH+DETH = (8+12+8)/4. */ + txp->hdr_dwords = 7; + + if (rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH) { + grh = &sdma_hdr->hdr.ibh.u.l.grh; + txp->hdr_dwords += + hfi1_make_grh(txp->ibp, + grh, + rdma_ah_read_grh(ah_attr), + txp->hdr_dwords - LRH_9B_DWORDS, + payload_dwords); + lrh0 = HFI1_LRH_GRH; + ohdr = &sdma_hdr->hdr.ibh.u.l.oth; + } else { + lrh0 = HFI1_LRH_BTH; + ohdr = &sdma_hdr->hdr.ibh.u.oth; + } + + lrh0 |= (rdma_ah_get_sl(ah_attr) & 0xf) << 4; + lrh0 |= (txp->flow.sc5 & 0xf) << 12; + + dlid = opa_get_lid(rdma_ah_get_dlid(ah_attr), 9B); + if (dlid == be16_to_cpu(IB_LID_PERMISSIVE)) { + slid = be16_to_cpu(IB_LID_PERMISSIVE); + } else { + u16 lid = (u16)ppd->lid; + + if (lid) { + lid |= rdma_ah_get_path_bits(ah_attr) & + ((1 << ppd->lmc) - 1); + slid = lid; + } else { + slid = be16_to_cpu(IB_LID_PERMISSIVE); + } + } + + /* Includes ICRC */ + dwords = txp->hdr_dwords + payload_dwords; + + /* Build the lrh */ + sdma_hdr->hdr.hdr_type = HFI1_PKT_TYPE_9B; + hfi1_make_ib_hdr(&sdma_hdr->hdr.ibh, lrh0, dwords, dlid, slid); + + /* Build the bth */ + bth0 = (IB_OPCODE_UD_SEND_ONLY << 24) | (pad_cnt << 20) | priv->pkey; + + ohdr->bth[0] = cpu_to_be32(bth0); + ohdr->bth[1] = cpu_to_be32(txp->dqpn); + ohdr->bth[2] = cpu_to_be32(mask_psn((u32)txp->txq->sent_txreqs)); + + /* Build the deth */ + ohdr->u.ud.deth[0] = cpu_to_be32(priv->qkey); + ohdr->u.ud.deth[1] = cpu_to_be32((txp->entropy << + HFI1_IPOIB_ENTROPY_SHIFT) | sqpn); + + /* Construct the pbc. */ + sdma_hdr->pbc = + cpu_to_le64(create_pbc(ppd, + ib_is_sc5(txp->flow.sc5) << + PBC_DC_INFO_SHIFT, + 0, + sc_to_vlt(priv->dd, txp->flow.sc5), + dwords - SIZE_OF_CRC + + (sizeof(sdma_hdr->pbc) >> 2))); +} + +static struct ipoib_txreq *hfi1_ipoib_send_dma_common(struct net_device *dev, + struct sk_buff *skb, + struct ipoib_txparms *txp) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + struct ipoib_txreq *tx; + int ret; + + tx = kmem_cache_alloc_node(priv->txreq_cache, + GFP_ATOMIC, + priv->dd->node); + if (unlikely(!tx)) + return ERR_PTR(-ENOMEM); + + /* so that we can test if the sdma descriptors are there */ + tx->txreq.num_desc = 0; + tx->priv = priv; + tx->txq = txp->txq; + tx->skb = skb; + INIT_LIST_HEAD(&tx->txreq.list); + + hfi1_ipoib_build_ib_tx_headers(tx, txp); + + ret = hfi1_ipoib_build_tx_desc(tx, txp); + if (likely(!ret)) { + if (txp->txq->flow.as_int != txp->flow.as_int) { + txp->txq->flow.tx_queue = txp->flow.tx_queue; + txp->txq->flow.sc5 = txp->flow.sc5; + txp->txq->sde = + sdma_select_engine_sc(priv->dd, + txp->flow.tx_queue, + txp->flow.sc5); + } + + return tx; + } + + sdma_txclean(priv->dd, &tx->txreq); + kmem_cache_free(priv->txreq_cache, tx); + + return ERR_PTR(ret); +} + +static int hfi1_ipoib_submit_tx_list(struct net_device *dev, + struct hfi1_ipoib_txq *txq) +{ + int ret; + u16 count_out; + + ret = sdma_send_txlist(txq->sde, + iowait_get_ib_work(&txq->wait), + &txq->tx_list, + &count_out); + if (likely(!ret) || ret == -EBUSY || ret == -ECOMM) + return ret; + + dd_dev_warn(txq->priv->dd, "cannot send skb tx list, err %d.\n", ret); + + return ret; +} + +static int hfi1_ipoib_flush_tx_list(struct net_device *dev, + struct hfi1_ipoib_txq *txq) +{ + int ret = 0; + + if (!list_empty(&txq->tx_list)) { + /* Flush the current list */ + ret = hfi1_ipoib_submit_tx_list(dev, txq); + + if (unlikely(ret)) + if (ret != -EBUSY) + ++dev->stats.tx_carrier_errors; + } + + return ret; +} + +static int hfi1_ipoib_submit_tx(struct hfi1_ipoib_txq *txq, + struct ipoib_txreq *tx) +{ + int ret; + + ret = sdma_send_txreq(txq->sde, + iowait_get_ib_work(&txq->wait), + &tx->txreq, + txq->pkts_sent); + if (likely(!ret)) { + txq->pkts_sent = true; + iowait_starve_clear(txq->pkts_sent, &txq->wait); + } + + return ret; +} + +static int hfi1_ipoib_send_dma_single(struct net_device *dev, + struct sk_buff *skb, + struct ipoib_txparms *txp) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + struct hfi1_ipoib_txq *txq = txp->txq; + struct ipoib_txreq *tx; + int ret; + + tx = hfi1_ipoib_send_dma_common(dev, skb, txp); + if (IS_ERR(tx)) { + int ret = PTR_ERR(tx); + + dev_kfree_skb_any(skb); + + if (ret == -ENOMEM) + ++dev->stats.tx_errors; + else + ++dev->stats.tx_carrier_errors; + + return NETDEV_TX_OK; + } + + ret = hfi1_ipoib_submit_tx(txq, tx); + if (likely(!ret)) { +tx_ok: + trace_sdma_output_ibhdr(tx->priv->dd, + &tx->sdma_hdr.hdr, + ib_is_sc5(txp->flow.sc5)); + hfi1_ipoib_check_queue_depth(txq); + return NETDEV_TX_OK; + } + + txq->pkts_sent = false; + + if (ret == -EBUSY || ret == -ECOMM) + goto tx_ok; + + sdma_txclean(priv->dd, &tx->txreq); + dev_kfree_skb_any(skb); + kmem_cache_free(priv->txreq_cache, tx); + ++dev->stats.tx_carrier_errors; + + return NETDEV_TX_OK; +} + +static int hfi1_ipoib_send_dma_list(struct net_device *dev, + struct sk_buff *skb, + struct ipoib_txparms *txp) +{ + struct hfi1_ipoib_txq *txq = txp->txq; + struct ipoib_txreq *tx; + + /* Has the flow change ? */ + if (txq->flow.as_int != txp->flow.as_int) { + int ret; + + ret = hfi1_ipoib_flush_tx_list(dev, txq); + if (unlikely(ret)) { + if (ret == -EBUSY) + ++dev->stats.tx_dropped; + dev_kfree_skb_any(skb); + return NETDEV_TX_OK; + } + } + tx = hfi1_ipoib_send_dma_common(dev, skb, txp); + if (IS_ERR(tx)) { + int ret = PTR_ERR(tx); + + dev_kfree_skb_any(skb); + + if (ret == -ENOMEM) + ++dev->stats.tx_errors; + else + ++dev->stats.tx_carrier_errors; + + return NETDEV_TX_OK; + } + + list_add_tail(&tx->txreq.list, &txq->tx_list); + + hfi1_ipoib_check_queue_depth(txq); + + trace_sdma_output_ibhdr(tx->priv->dd, + &tx->sdma_hdr.hdr, + ib_is_sc5(txp->flow.sc5)); + + if (!netdev_xmit_more()) + (void)hfi1_ipoib_flush_tx_list(dev, txq); + + return NETDEV_TX_OK; +} + +static u8 hfi1_ipoib_calc_entropy(struct sk_buff *skb) +{ + if (skb_transport_header_was_set(skb)) { + u8 *hdr = (u8 *)skb_transport_header(skb); + + return (hdr[0] ^ hdr[1] ^ hdr[2] ^ hdr[3]); + } + + return (u8)skb_get_queue_mapping(skb); +} + +int hfi1_ipoib_send_dma(struct net_device *dev, + struct sk_buff *skb, + struct ib_ah *address, + u32 dqpn) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + struct ipoib_txparms txp; + struct rdma_netdev *rn = netdev_priv(dev); + + if (unlikely(skb->len > rn->mtu + HFI1_IPOIB_ENCAP_LEN)) { + dd_dev_warn(priv->dd, "packet len %d (> %d) too long to send, dropping\n", + skb->len, + rn->mtu + HFI1_IPOIB_ENCAP_LEN); + ++dev->stats.tx_dropped; + ++dev->stats.tx_errors; + dev_kfree_skb_any(skb); + return NETDEV_TX_OK; + } + + txp.dd = priv->dd; + txp.ah_attr = &ibah_to_rvtah(address)->attr; + txp.ibp = to_iport(priv->device, priv->port_num); + txp.txq = &priv->txqs[skb_get_queue_mapping(skb)]; + txp.dqpn = dqpn; + txp.flow.sc5 = txp.ibp->sl_to_sc[rdma_ah_get_sl(txp.ah_attr)]; + txp.flow.tx_queue = (u8)skb_get_queue_mapping(skb); + txp.entropy = hfi1_ipoib_calc_entropy(skb); + + if (netdev_xmit_more() || !list_empty(&txp.txq->tx_list)) + return hfi1_ipoib_send_dma_list(dev, skb, &txp); + + return hfi1_ipoib_send_dma_single(dev, skb, &txp); +} + +/* + * hfi1_ipoib_sdma_sleep - ipoib sdma sleep function + * + * This function gets called from sdma_send_txreq() when there are not enough + * sdma descriptors available to send the packet. It adds Tx queue's wait + * structure to sdma engine's dmawait list to be woken up when descriptors + * become available. + */ +static int hfi1_ipoib_sdma_sleep(struct sdma_engine *sde, + struct iowait_work *wait, + struct sdma_txreq *txreq, + uint seq, + bool pkts_sent) +{ + struct hfi1_ipoib_txq *txq = + container_of(wait->iow, struct hfi1_ipoib_txq, wait); + + write_seqlock(&sde->waitlock); + + if (likely(txq->priv->netdev->reg_state == NETREG_REGISTERED)) { + if (sdma_progress(sde, seq, txreq)) { + write_sequnlock(&sde->waitlock); + return -EAGAIN; + } + + if (list_empty(&txreq->list)) + /* came from non-list submit */ + list_add_tail(&txreq->list, &txq->tx_list); + if (list_empty(&txq->wait.list)) { + if (!atomic_xchg(&txq->no_desc, 1)) + hfi1_ipoib_stop_txq(txq); + iowait_queue(pkts_sent, wait->iow, &sde->dmawait); + } + + write_sequnlock(&sde->waitlock); + return -EBUSY; + } + + write_sequnlock(&sde->waitlock); + return -EINVAL; +} + +/* + * hfi1_ipoib_sdma_wakeup - ipoib sdma wakeup function + * + * This function gets called when SDMA descriptors becomes available and Tx + * queue's wait structure was previously added to sdma engine's dmawait list. + */ +static void hfi1_ipoib_sdma_wakeup(struct iowait *wait, int reason) +{ + struct hfi1_ipoib_txq *txq = + container_of(wait, struct hfi1_ipoib_txq, wait); + + if (likely(txq->priv->netdev->reg_state == NETREG_REGISTERED)) + iowait_schedule(wait, system_highpri_wq, WORK_CPU_UNBOUND); +} + +static void hfi1_ipoib_flush_txq(struct work_struct *work) +{ + struct iowait_work *ioww = + container_of(work, struct iowait_work, iowork); + struct iowait *wait = iowait_ioww_to_iow(ioww); + struct hfi1_ipoib_txq *txq = + container_of(wait, struct hfi1_ipoib_txq, wait); + struct net_device *dev = txq->priv->netdev; + + if (likely(dev->reg_state == NETREG_REGISTERED) && + likely(!hfi1_ipoib_flush_tx_list(dev, txq))) + if (atomic_xchg(&txq->no_desc, 0)) + hfi1_ipoib_wake_txq(txq); +} + +int hfi1_ipoib_txreq_init(struct hfi1_ipoib_dev_priv *priv) +{ + struct net_device *dev = priv->netdev; + char buf[HFI1_IPOIB_TXREQ_NAME_LEN]; + unsigned long tx_ring_size; + int i; + + /* + * Ring holds 1 less than tx_ring_size + * Round up to next power of 2 in order to hold at least tx_queue_len + */ + tx_ring_size = roundup_pow_of_two((unsigned long)dev->tx_queue_len + 1); + + snprintf(buf, sizeof(buf), "hfi1_%u_ipoib_txreq_cache", priv->dd->unit); + priv->txreq_cache = kmem_cache_create(buf, + sizeof(struct ipoib_txreq), + 0, + 0, + NULL); + if (!priv->txreq_cache) + return -ENOMEM; + + priv->tx_napis = kcalloc_node(dev->num_tx_queues, + sizeof(struct napi_struct), + GFP_ATOMIC, + priv->dd->node); + if (!priv->tx_napis) + goto free_txreq_cache; + + priv->txqs = kcalloc_node(dev->num_tx_queues, + sizeof(struct hfi1_ipoib_txq), + GFP_ATOMIC, + priv->dd->node); + if (!priv->txqs) + goto free_tx_napis; + + for (i = 0; i < dev->num_tx_queues; i++) { + struct hfi1_ipoib_txq *txq = &priv->txqs[i]; + + iowait_init(&txq->wait, + 0, + hfi1_ipoib_flush_txq, + NULL, + hfi1_ipoib_sdma_sleep, + hfi1_ipoib_sdma_wakeup, + NULL, + NULL); + txq->priv = priv; + txq->sde = NULL; + INIT_LIST_HEAD(&txq->tx_list); + atomic64_set(&txq->complete_txreqs, 0); + atomic_set(&txq->stops, 0); + atomic_set(&txq->ring_full, 0); + atomic_set(&txq->no_desc, 0); + txq->q_idx = i; + txq->flow.tx_queue = 0xff; + txq->flow.sc5 = 0xff; + txq->pkts_sent = false; + + netdev_queue_numa_node_write(netdev_get_tx_queue(dev, i), + priv->dd->node); + + txq->tx_ring.items = + vzalloc_node(array_size(tx_ring_size, + sizeof(struct ipoib_txreq)), + priv->dd->node); + if (!txq->tx_ring.items) + goto free_txqs; + + spin_lock_init(&txq->tx_ring.producer_lock); + spin_lock_init(&txq->tx_ring.consumer_lock); + txq->tx_ring.max_items = tx_ring_size; + + txq->napi = &priv->tx_napis[i]; + netif_tx_napi_add(dev, txq->napi, + hfi1_ipoib_process_tx_ring, + NAPI_POLL_WEIGHT); + } + + return 0; + +free_txqs: + for (i--; i >= 0; i--) { + struct hfi1_ipoib_txq *txq = &priv->txqs[i]; + + netif_napi_del(txq->napi); + vfree(txq->tx_ring.items); + } + + kfree(priv->txqs); + priv->txqs = NULL; + +free_tx_napis: + kfree(priv->tx_napis); + priv->tx_napis = NULL; + +free_txreq_cache: + kmem_cache_destroy(priv->txreq_cache); + priv->txreq_cache = NULL; + return -ENOMEM; +} + +static void hfi1_ipoib_drain_tx_list(struct hfi1_ipoib_txq *txq) +{ + struct sdma_txreq *txreq; + struct sdma_txreq *txreq_tmp; + atomic64_t *complete_txreqs = &txq->complete_txreqs; + + list_for_each_entry_safe(txreq, txreq_tmp, &txq->tx_list, list) { + struct ipoib_txreq *tx = + container_of(txreq, struct ipoib_txreq, txreq); + + list_del(&txreq->list); + sdma_txclean(txq->priv->dd, &tx->txreq); + dev_kfree_skb_any(tx->skb); + kmem_cache_free(txq->priv->txreq_cache, tx); + atomic64_inc(complete_txreqs); + } + + if (hfi1_ipoib_used(txq)) + dd_dev_warn(txq->priv->dd, + "txq %d not empty found %llu requests\n", + txq->q_idx, + hfi1_ipoib_txreqs(txq->sent_txreqs, + atomic64_read(complete_txreqs))); +} + +void hfi1_ipoib_txreq_deinit(struct hfi1_ipoib_dev_priv *priv) +{ + int i; + + for (i = 0; i < priv->netdev->num_tx_queues; i++) { + struct hfi1_ipoib_txq *txq = &priv->txqs[i]; + + iowait_cancel_work(&txq->wait); + iowait_sdma_drain(&txq->wait); + hfi1_ipoib_drain_tx_list(txq); + netif_napi_del(txq->napi); + (void)hfi1_ipoib_drain_tx_ring(txq, txq->tx_ring.max_items); + vfree(txq->tx_ring.items); + } + + kfree(priv->txqs); + priv->txqs = NULL; + + kfree(priv->tx_napis); + priv->tx_napis = NULL; + + kmem_cache_destroy(priv->txreq_cache); + priv->txreq_cache = NULL; +} + +void hfi1_ipoib_napi_tx_enable(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + int i; + + for (i = 0; i < dev->num_tx_queues; i++) { + struct hfi1_ipoib_txq *txq = &priv->txqs[i]; + + napi_enable(txq->napi); + } +} + +void hfi1_ipoib_napi_tx_disable(struct net_device *dev) +{ + struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev); + int i; + + for (i = 0; i < dev->num_tx_queues; i++) { + struct hfi1_ipoib_txq *txq = &priv->txqs[i]; + + napi_disable(txq->napi); + (void)hfi1_ipoib_drain_tx_ring(txq, txq->tx_ring.max_items); + } +} diff --git a/drivers/infiniband/hw/hfi1/mad.c b/drivers/infiniband/hw/hfi1/mad.c index 7073f237a949..3222e3acb79c 100644 --- a/drivers/infiniband/hw/hfi1/mad.c +++ b/drivers/infiniband/hw/hfi1/mad.c @@ -721,7 +721,7 @@ static int check_mkey(struct hfi1_ibport *ibp, struct ib_mad_hdr *mad, /* Bad mkey not a violation below level 2 */ if (ibp->rvp.mkeyprot < 2) break; - /* fall through */ + fallthrough; case IB_MGMT_METHOD_SET: case IB_MGMT_METHOD_TRAP_REPRESS: if (ibp->rvp.mkey_violations != 0xFFFF) @@ -1272,7 +1272,7 @@ static int set_port_states(struct hfi1_pportdata *ppd, struct opa_smp *smp, case IB_PORT_NOP: if (phys_state == IB_PORTPHYSSTATE_NOP) break; - /* FALLTHROUGH */ + fallthrough; case IB_PORT_DOWN: if (phys_state == IB_PORTPHYSSTATE_NOP) { link_state = HLS_DN_DOWNDEF; @@ -2300,7 +2300,6 @@ static int __subn_set_opa_vl_arb(struct opa_smp *smp, u32 am, u8 *data, * can be changed from the default values */ case OPA_VLARB_PREEMPT_ELEMENTS: - /* FALLTHROUGH */ case OPA_VLARB_PREEMPT_MATRIX: smp->status |= IB_SMP_UNSUP_METH_ATTR; break; @@ -4170,7 +4169,7 @@ static int subn_get_opa_sma(__be16 attr_id, struct opa_smp *smp, u32 am, return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED; if (ibp->rvp.port_cap_flags & IB_PORT_SM) return IB_MAD_RESULT_SUCCESS; - /* FALLTHROUGH */ + fallthrough; default: smp->status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_mad_hdr *)smp); @@ -4240,7 +4239,7 @@ static int subn_set_opa_sma(__be16 attr_id, struct opa_smp *smp, u32 am, return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED; if (ibp->rvp.port_cap_flags & IB_PORT_SM) return IB_MAD_RESULT_SUCCESS; - /* FALLTHROUGH */ + fallthrough; default: smp->status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_mad_hdr *)smp); diff --git a/drivers/infiniband/hw/hfi1/mmu_rb.c b/drivers/infiniband/hw/hfi1/mmu_rb.c index 14d2a90964c3..24ca17b77b72 100644 --- a/drivers/infiniband/hw/hfi1/mmu_rb.c +++ b/drivers/infiniband/hw/hfi1/mmu_rb.c @@ -333,7 +333,7 @@ static void do_remove(struct mmu_rb_handler *handler, /* * Work queue function to remove all nodes that have been queued up to - * be removed. The key feature is that mm->mmap_sem is not being held + * be removed. The key feature is that mm->mmap_lock is not being held * and the remove callback can sleep while taking it, if needed. */ static void handle_remove(struct work_struct *work) diff --git a/drivers/infiniband/hw/hfi1/msix.c b/drivers/infiniband/hw/hfi1/msix.c index db82db497b2c..d61ee853d215 100644 --- a/drivers/infiniband/hw/hfi1/msix.c +++ b/drivers/infiniband/hw/hfi1/msix.c @@ -1,6 +1,6 @@ // SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) /* - * Copyright(c) 2018 Intel Corporation. + * Copyright(c) 2018 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -49,6 +49,7 @@ #include "hfi.h" #include "affinity.h" #include "sdma.h" +#include "netdev.h" /** * msix_initialize() - Calculate, request and configure MSIx IRQs @@ -69,7 +70,7 @@ int msix_initialize(struct hfi1_devdata *dd) * one for each VNIC context * ...any new IRQs should be added here. */ - total = 1 + dd->num_sdma + dd->n_krcv_queues + dd->num_vnic_contexts; + total = 1 + dd->num_sdma + dd->n_krcv_queues + dd->num_netdev_contexts; if (total >= CCE_NUM_MSIX_VECTORS) return -EINVAL; @@ -140,7 +141,7 @@ static int msix_request_irq(struct hfi1_devdata *dd, void *arg, ret = pci_request_irq(dd->pcidev, nr, handler, thread, arg, name); if (ret) { dd_dev_err(dd, - "%s: request for IRQ %d failed, MSIx %lu, err %d\n", + "%s: request for IRQ %d failed, MSIx %lx, err %d\n", name, irq, nr, ret); spin_lock(&dd->msix_info.msix_lock); __clear_bit(nr, dd->msix_info.in_use_msix); @@ -160,7 +161,7 @@ static int msix_request_irq(struct hfi1_devdata *dd, void *arg, /* This is a request, so a failure is not fatal */ ret = hfi1_get_irq_affinity(dd, me); if (ret) - dd_dev_err(dd, "unable to pin IRQ %d\n", ret); + dd_dev_err(dd, "%s: unable to pin IRQ %d\n", name, ret); return nr; } @@ -171,7 +172,8 @@ static int msix_request_rcd_irq_common(struct hfi1_ctxtdata *rcd, const char *name) { int nr = msix_request_irq(rcd->dd, rcd, handler, thread, - IRQ_RCVCTXT, name); + rcd->is_vnic ? IRQ_NETDEVCTXT : IRQ_RCVCTXT, + name); if (nr < 0) return nr; @@ -204,6 +206,21 @@ int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd) } /** + * msix_request_rcd_irq() - Helper function for RCVAVAIL IRQs + * for netdev context + * @rcd: valid netdev contexti + */ +int msix_netdev_request_rcd_irq(struct hfi1_ctxtdata *rcd) +{ + char name[MAX_NAME_SIZE]; + + snprintf(name, sizeof(name), DRIVER_NAME "_%d nd kctxt%d", + rcd->dd->unit, rcd->ctxt); + return msix_request_rcd_irq_common(rcd, receive_context_interrupt_napi, + NULL, name); +} + +/** * msix_request_smda_ira() - Helper for getting SDMA IRQ resources * @sde: valid sdma engine * @@ -355,15 +372,16 @@ void msix_clean_up_interrupts(struct hfi1_devdata *dd) } /** - * msix_vnic_syncrhonize_irq() - Vnic IRQ synchronize + * msix_netdev_syncrhonize_irq() - netdev IRQ synchronize * @dd: valid devdata */ -void msix_vnic_synchronize_irq(struct hfi1_devdata *dd) +void msix_netdev_synchronize_irq(struct hfi1_devdata *dd) { int i; + int ctxt_count = hfi1_netdev_ctxt_count(dd); - for (i = 0; i < dd->vnic.num_ctxt; i++) { - struct hfi1_ctxtdata *rcd = dd->vnic.ctxt[i]; + for (i = 0; i < ctxt_count; i++) { + struct hfi1_ctxtdata *rcd = hfi1_netdev_get_ctxt(dd, i); struct hfi1_msix_entry *me; me = &dd->msix_info.msix_entries[rcd->msix_intr]; diff --git a/drivers/infiniband/hw/hfi1/msix.h b/drivers/infiniband/hw/hfi1/msix.h index 1a02ab7971c8..e63e944bf0fc 100644 --- a/drivers/infiniband/hw/hfi1/msix.h +++ b/drivers/infiniband/hw/hfi1/msix.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */ /* - * Copyright(c) 2018 Intel Corporation. + * Copyright(c) 2018 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -59,7 +59,8 @@ int msix_request_rcd_irq(struct hfi1_ctxtdata *rcd); int msix_request_sdma_irq(struct sdma_engine *sde); void msix_free_irq(struct hfi1_devdata *dd, u8 msix_intr); -/* VNIC interface */ -void msix_vnic_synchronize_irq(struct hfi1_devdata *dd); +/* Netdev interface */ +void msix_netdev_synchronize_irq(struct hfi1_devdata *dd); +int msix_netdev_request_rcd_irq(struct hfi1_ctxtdata *rcd); #endif diff --git a/drivers/infiniband/hw/hfi1/netdev.h b/drivers/infiniband/hw/hfi1/netdev.h new file mode 100644 index 000000000000..947543a3e0c4 --- /dev/null +++ b/drivers/infiniband/hw/hfi1/netdev.h @@ -0,0 +1,118 @@ +/* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */ +/* + * Copyright(c) 2020 Intel Corporation. + * + */ + +#ifndef HFI1_NETDEV_H +#define HFI1_NETDEV_H + +#include "hfi.h" + +#include <linux/netdevice.h> +#include <linux/xarray.h> + +/** + * struct hfi1_netdev_rxq - Receive Queue for HFI + * dummy netdev. Both IPoIB and VNIC netdevices will be working on + * top of this device. + * @napi: napi object + * @priv: ptr to netdev_priv + * @rcd: ptr to receive context data + */ +struct hfi1_netdev_rxq { + struct napi_struct napi; + struct hfi1_netdev_priv *priv; + struct hfi1_ctxtdata *rcd; +}; + +/* + * Number of netdev contexts used. Ensure it is less than or equal to + * max queues supported by VNIC (HFI1_VNIC_MAX_QUEUE). + */ +#define HFI1_MAX_NETDEV_CTXTS 8 + +/* Number of NETDEV RSM entries */ +#define NUM_NETDEV_MAP_ENTRIES HFI1_MAX_NETDEV_CTXTS + +/** + * struct hfi1_netdev_priv: data required to setup and run HFI netdev. + * @dd: hfi1_devdata + * @rxq: pointer to dummy netdev receive queues. + * @num_rx_q: number of receive queues + * @rmt_index: first free index in RMT Array + * @msix_start: first free MSI-X interrupt vector. + * @dev_tbl: netdev table for unique identifier VNIC and IPoIb VLANs. + * @enabled: atomic counter of netdevs enabling receive queues. + * When 0 NAPI will be disabled. + * @netdevs: atomic counter of netdevs using dummy netdev. + * When 0 receive queues will be freed. + */ +struct hfi1_netdev_priv { + struct hfi1_devdata *dd; + struct hfi1_netdev_rxq *rxq; + int num_rx_q; + int rmt_start; + struct xarray dev_tbl; + /* count of enabled napi polls */ + atomic_t enabled; + /* count of netdevs on top */ + atomic_t netdevs; +}; + +static inline +struct hfi1_netdev_priv *hfi1_netdev_priv(struct net_device *dev) +{ + return (struct hfi1_netdev_priv *)&dev[1]; +} + +static inline +int hfi1_netdev_ctxt_count(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + return priv->num_rx_q; +} + +static inline +struct hfi1_ctxtdata *hfi1_netdev_get_ctxt(struct hfi1_devdata *dd, int ctxt) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + return priv->rxq[ctxt].rcd; +} + +static inline +int hfi1_netdev_get_free_rmt_idx(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + return priv->rmt_start; +} + +static inline +void hfi1_netdev_set_free_rmt_idx(struct hfi1_devdata *dd, int rmt_idx) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + priv->rmt_start = rmt_idx; +} + +u32 hfi1_num_netdev_contexts(struct hfi1_devdata *dd, u32 available_contexts, + struct cpumask *cpu_mask); + +void hfi1_netdev_enable_queues(struct hfi1_devdata *dd); +void hfi1_netdev_disable_queues(struct hfi1_devdata *dd); +int hfi1_netdev_rx_init(struct hfi1_devdata *dd); +int hfi1_netdev_rx_destroy(struct hfi1_devdata *dd); +int hfi1_netdev_alloc(struct hfi1_devdata *dd); +void hfi1_netdev_free(struct hfi1_devdata *dd); +int hfi1_netdev_add_data(struct hfi1_devdata *dd, int id, void *data); +void *hfi1_netdev_remove_data(struct hfi1_devdata *dd, int id); +void *hfi1_netdev_get_data(struct hfi1_devdata *dd, int id); +void *hfi1_netdev_get_first_data(struct hfi1_devdata *dd, int *start_id); + +/* chip.c */ +int hfi1_netdev_rx_napi(struct napi_struct *napi, int budget); + +#endif /* HFI1_NETDEV_H */ diff --git a/drivers/infiniband/hw/hfi1/netdev_rx.c b/drivers/infiniband/hw/hfi1/netdev_rx.c new file mode 100644 index 000000000000..6d263c9749b3 --- /dev/null +++ b/drivers/infiniband/hw/hfi1/netdev_rx.c @@ -0,0 +1,481 @@ +// SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) +/* + * Copyright(c) 2020 Intel Corporation. + * + */ + +/* + * This file contains HFI1 support for netdev RX functionality + */ + +#include "sdma.h" +#include "verbs.h" +#include "netdev.h" +#include "hfi.h" + +#include <linux/netdevice.h> +#include <linux/etherdevice.h> +#include <rdma/ib_verbs.h> + +static int hfi1_netdev_setup_ctxt(struct hfi1_netdev_priv *priv, + struct hfi1_ctxtdata *uctxt) +{ + unsigned int rcvctrl_ops; + struct hfi1_devdata *dd = priv->dd; + int ret; + + uctxt->rhf_rcv_function_map = netdev_rhf_rcv_functions; + uctxt->do_interrupt = &handle_receive_interrupt_napi_sp; + + /* Now allocate the RcvHdr queue and eager buffers. */ + ret = hfi1_create_rcvhdrq(dd, uctxt); + if (ret) + goto done; + + ret = hfi1_setup_eagerbufs(uctxt); + if (ret) + goto done; + + clear_rcvhdrtail(uctxt); + + rcvctrl_ops = HFI1_RCVCTRL_CTXT_DIS; + rcvctrl_ops |= HFI1_RCVCTRL_INTRAVAIL_DIS; + + if (!HFI1_CAP_KGET_MASK(uctxt->flags, MULTI_PKT_EGR)) + rcvctrl_ops |= HFI1_RCVCTRL_ONE_PKT_EGR_ENB; + if (HFI1_CAP_KGET_MASK(uctxt->flags, NODROP_EGR_FULL)) + rcvctrl_ops |= HFI1_RCVCTRL_NO_EGR_DROP_ENB; + if (HFI1_CAP_KGET_MASK(uctxt->flags, NODROP_RHQ_FULL)) + rcvctrl_ops |= HFI1_RCVCTRL_NO_RHQ_DROP_ENB; + if (HFI1_CAP_KGET_MASK(uctxt->flags, DMA_RTAIL)) + rcvctrl_ops |= HFI1_RCVCTRL_TAILUPD_ENB; + + hfi1_rcvctrl(uctxt->dd, rcvctrl_ops, uctxt); +done: + return ret; +} + +static int hfi1_netdev_allocate_ctxt(struct hfi1_devdata *dd, + struct hfi1_ctxtdata **ctxt) +{ + struct hfi1_ctxtdata *uctxt; + int ret; + + if (dd->flags & HFI1_FROZEN) + return -EIO; + + ret = hfi1_create_ctxtdata(dd->pport, dd->node, &uctxt); + if (ret < 0) { + dd_dev_err(dd, "Unable to create ctxtdata, failing open\n"); + return -ENOMEM; + } + + uctxt->flags = HFI1_CAP_KGET(MULTI_PKT_EGR) | + HFI1_CAP_KGET(NODROP_RHQ_FULL) | + HFI1_CAP_KGET(NODROP_EGR_FULL) | + HFI1_CAP_KGET(DMA_RTAIL); + /* Netdev contexts are always NO_RDMA_RTAIL */ + uctxt->fast_handler = handle_receive_interrupt_napi_fp; + uctxt->slow_handler = handle_receive_interrupt_napi_sp; + hfi1_set_seq_cnt(uctxt, 1); + uctxt->is_vnic = true; + + hfi1_stats.sps_ctxts++; + + dd_dev_info(dd, "created netdev context %d\n", uctxt->ctxt); + *ctxt = uctxt; + + return 0; +} + +static void hfi1_netdev_deallocate_ctxt(struct hfi1_devdata *dd, + struct hfi1_ctxtdata *uctxt) +{ + flush_wc(); + + /* + * Disable receive context and interrupt available, reset all + * RcvCtxtCtrl bits to default values. + */ + hfi1_rcvctrl(dd, HFI1_RCVCTRL_CTXT_DIS | + HFI1_RCVCTRL_TIDFLOW_DIS | + HFI1_RCVCTRL_INTRAVAIL_DIS | + HFI1_RCVCTRL_ONE_PKT_EGR_DIS | + HFI1_RCVCTRL_NO_RHQ_DROP_DIS | + HFI1_RCVCTRL_NO_EGR_DROP_DIS, uctxt); + + if (uctxt->msix_intr != CCE_NUM_MSIX_VECTORS) + msix_free_irq(dd, uctxt->msix_intr); + + uctxt->msix_intr = CCE_NUM_MSIX_VECTORS; + uctxt->event_flags = 0; + + hfi1_clear_tids(uctxt); + hfi1_clear_ctxt_pkey(dd, uctxt); + + hfi1_stats.sps_ctxts--; + + hfi1_free_ctxt(uctxt); +} + +static int hfi1_netdev_allot_ctxt(struct hfi1_netdev_priv *priv, + struct hfi1_ctxtdata **ctxt) +{ + int rc; + struct hfi1_devdata *dd = priv->dd; + + rc = hfi1_netdev_allocate_ctxt(dd, ctxt); + if (rc) { + dd_dev_err(dd, "netdev ctxt alloc failed %d\n", rc); + return rc; + } + + rc = hfi1_netdev_setup_ctxt(priv, *ctxt); + if (rc) { + dd_dev_err(dd, "netdev ctxt setup failed %d\n", rc); + hfi1_netdev_deallocate_ctxt(dd, *ctxt); + *ctxt = NULL; + } + + return rc; +} + +/** + * hfi1_num_netdev_contexts - Count of netdev recv contexts to use. + * @dd: device on which to allocate netdev contexts + * @available_contexts: count of available receive contexts + * @cpu_mask: mask of possible cpus to include for contexts + * + * Return: count of physical cores on a node or the remaining available recv + * contexts for netdev recv context usage up to the maximum of + * HFI1_MAX_NETDEV_CTXTS. + * A value of 0 can be returned when acceleration is explicitly turned off, + * a memory allocation error occurs or when there are no available contexts. + * + */ +u32 hfi1_num_netdev_contexts(struct hfi1_devdata *dd, u32 available_contexts, + struct cpumask *cpu_mask) +{ + cpumask_var_t node_cpu_mask; + unsigned int available_cpus; + + if (!HFI1_CAP_IS_KSET(AIP)) + return 0; + + /* Always give user contexts priority over netdev contexts */ + if (available_contexts == 0) { + dd_dev_info(dd, "No receive contexts available for netdevs.\n"); + return 0; + } + + if (!zalloc_cpumask_var(&node_cpu_mask, GFP_KERNEL)) { + dd_dev_err(dd, "Unable to allocate cpu_mask for netdevs.\n"); + return 0; + } + + cpumask_and(node_cpu_mask, cpu_mask, + cpumask_of_node(pcibus_to_node(dd->pcidev->bus))); + + available_cpus = cpumask_weight(node_cpu_mask); + + free_cpumask_var(node_cpu_mask); + + return min3(available_cpus, available_contexts, + (u32)HFI1_MAX_NETDEV_CTXTS); +} + +static int hfi1_netdev_rxq_init(struct net_device *dev) +{ + int i; + int rc; + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dev); + struct hfi1_devdata *dd = priv->dd; + + priv->num_rx_q = dd->num_netdev_contexts; + priv->rxq = kcalloc_node(priv->num_rx_q, sizeof(struct hfi1_netdev_rxq), + GFP_KERNEL, dd->node); + + if (!priv->rxq) { + dd_dev_err(dd, "Unable to allocate netdev queue data\n"); + return (-ENOMEM); + } + + for (i = 0; i < priv->num_rx_q; i++) { + struct hfi1_netdev_rxq *rxq = &priv->rxq[i]; + + rc = hfi1_netdev_allot_ctxt(priv, &rxq->rcd); + if (rc) + goto bail_context_irq_failure; + + hfi1_rcd_get(rxq->rcd); + rxq->priv = priv; + rxq->rcd->napi = &rxq->napi; + dd_dev_info(dd, "Setting rcv queue %d napi to context %d\n", + i, rxq->rcd->ctxt); + /* + * Disable BUSY_POLL on this NAPI as this is not supported + * right now. + */ + set_bit(NAPI_STATE_NO_BUSY_POLL, &rxq->napi.state); + netif_napi_add(dev, &rxq->napi, hfi1_netdev_rx_napi, 64); + rc = msix_netdev_request_rcd_irq(rxq->rcd); + if (rc) + goto bail_context_irq_failure; + } + + return 0; + +bail_context_irq_failure: + dd_dev_err(dd, "Unable to allot receive context\n"); + for (; i >= 0; i--) { + struct hfi1_netdev_rxq *rxq = &priv->rxq[i]; + + if (rxq->rcd) { + hfi1_netdev_deallocate_ctxt(dd, rxq->rcd); + hfi1_rcd_put(rxq->rcd); + rxq->rcd = NULL; + } + } + kfree(priv->rxq); + priv->rxq = NULL; + + return rc; +} + +static void hfi1_netdev_rxq_deinit(struct net_device *dev) +{ + int i; + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dev); + struct hfi1_devdata *dd = priv->dd; + + for (i = 0; i < priv->num_rx_q; i++) { + struct hfi1_netdev_rxq *rxq = &priv->rxq[i]; + + netif_napi_del(&rxq->napi); + hfi1_netdev_deallocate_ctxt(dd, rxq->rcd); + hfi1_rcd_put(rxq->rcd); + rxq->rcd = NULL; + } + + kfree(priv->rxq); + priv->rxq = NULL; + priv->num_rx_q = 0; +} + +static void enable_queues(struct hfi1_netdev_priv *priv) +{ + int i; + + for (i = 0; i < priv->num_rx_q; i++) { + struct hfi1_netdev_rxq *rxq = &priv->rxq[i]; + + dd_dev_info(priv->dd, "enabling queue %d on context %d\n", i, + rxq->rcd->ctxt); + napi_enable(&rxq->napi); + hfi1_rcvctrl(priv->dd, + HFI1_RCVCTRL_CTXT_ENB | HFI1_RCVCTRL_INTRAVAIL_ENB, + rxq->rcd); + } +} + +static void disable_queues(struct hfi1_netdev_priv *priv) +{ + int i; + + msix_netdev_synchronize_irq(priv->dd); + + for (i = 0; i < priv->num_rx_q; i++) { + struct hfi1_netdev_rxq *rxq = &priv->rxq[i]; + + dd_dev_info(priv->dd, "disabling queue %d on context %d\n", i, + rxq->rcd->ctxt); + + /* wait for napi if it was scheduled */ + hfi1_rcvctrl(priv->dd, + HFI1_RCVCTRL_CTXT_DIS | HFI1_RCVCTRL_INTRAVAIL_DIS, + rxq->rcd); + napi_synchronize(&rxq->napi); + napi_disable(&rxq->napi); + } +} + +/** + * hfi1_netdev_rx_init - Incrememnts netdevs counter. When called first time, + * it allocates receive queue data and calls netif_napi_add + * for each queue. + * + * @dd: hfi1 dev data + */ +int hfi1_netdev_rx_init(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + int res; + + if (atomic_fetch_inc(&priv->netdevs)) + return 0; + + mutex_lock(&hfi1_mutex); + init_dummy_netdev(dd->dummy_netdev); + res = hfi1_netdev_rxq_init(dd->dummy_netdev); + mutex_unlock(&hfi1_mutex); + return res; +} + +/** + * hfi1_netdev_rx_destroy - Decrements netdevs counter, when it reaches 0 + * napi is deleted and receive queses memory is freed. + * + * @dd: hfi1 dev data + */ +int hfi1_netdev_rx_destroy(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + /* destroy the RX queues only if it is the last netdev going away */ + if (atomic_fetch_add_unless(&priv->netdevs, -1, 0) == 1) { + mutex_lock(&hfi1_mutex); + hfi1_netdev_rxq_deinit(dd->dummy_netdev); + mutex_unlock(&hfi1_mutex); + } + + return 0; +} + +/** + * hfi1_netdev_alloc - Allocates netdev and private data. It is required + * because RMT index and MSI-X interrupt can be set only + * during driver initialization. + * + * @dd: hfi1 dev data + */ +int hfi1_netdev_alloc(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv; + const int netdev_size = sizeof(*dd->dummy_netdev) + + sizeof(struct hfi1_netdev_priv); + + dd_dev_info(dd, "allocating netdev size %d\n", netdev_size); + dd->dummy_netdev = kcalloc_node(1, netdev_size, GFP_KERNEL, dd->node); + + if (!dd->dummy_netdev) + return -ENOMEM; + + priv = hfi1_netdev_priv(dd->dummy_netdev); + priv->dd = dd; + xa_init(&priv->dev_tbl); + atomic_set(&priv->enabled, 0); + atomic_set(&priv->netdevs, 0); + + return 0; +} + +void hfi1_netdev_free(struct hfi1_devdata *dd) +{ + if (dd->dummy_netdev) { + dd_dev_info(dd, "hfi1 netdev freed\n"); + kfree(dd->dummy_netdev); + dd->dummy_netdev = NULL; + } +} + +/** + * hfi1_netdev_enable_queues - This is napi enable function. + * It enables napi objects associated with queues. + * When at least one device has called it it increments atomic counter. + * Disable function decrements counter and when it is 0, + * calls napi_disable for every queue. + * + * @dd: hfi1 dev data + */ +void hfi1_netdev_enable_queues(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv; + + if (!dd->dummy_netdev) + return; + + priv = hfi1_netdev_priv(dd->dummy_netdev); + if (atomic_fetch_inc(&priv->enabled)) + return; + + mutex_lock(&hfi1_mutex); + enable_queues(priv); + mutex_unlock(&hfi1_mutex); +} + +void hfi1_netdev_disable_queues(struct hfi1_devdata *dd) +{ + struct hfi1_netdev_priv *priv; + + if (!dd->dummy_netdev) + return; + + priv = hfi1_netdev_priv(dd->dummy_netdev); + if (atomic_dec_if_positive(&priv->enabled)) + return; + + mutex_lock(&hfi1_mutex); + disable_queues(priv); + mutex_unlock(&hfi1_mutex); +} + +/** + * hfi1_netdev_add_data - Registers data with unique identifier + * to be requested later this is needed for VNIC and IPoIB VLANs + * implementations. + * This call is protected by mutex idr_lock. + * + * @dd: hfi1 dev data + * @id: requested integer id up to INT_MAX + * @data: data to be associated with index + */ +int hfi1_netdev_add_data(struct hfi1_devdata *dd, int id, void *data) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + return xa_insert(&priv->dev_tbl, id, data, GFP_NOWAIT); +} + +/** + * hfi1_netdev_remove_data - Removes data with previously given id. + * Returns the reference to removed entry. + * + * @dd: hfi1 dev data + * @id: requested integer id up to INT_MAX + */ +void *hfi1_netdev_remove_data(struct hfi1_devdata *dd, int id) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + return xa_erase(&priv->dev_tbl, id); +} + +/** + * hfi1_netdev_get_data - Gets data with given id + * + * @dd: hfi1 dev data + * @id: requested integer id up to INT_MAX + */ +void *hfi1_netdev_get_data(struct hfi1_devdata *dd, int id) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + + return xa_load(&priv->dev_tbl, id); +} + +/** + * hfi1_netdev_get_first_dat - Gets first entry with greater or equal id. + * + * @dd: hfi1 dev data + * @id: requested integer id up to INT_MAX + */ +void *hfi1_netdev_get_first_data(struct hfi1_devdata *dd, int *start_id) +{ + struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev); + unsigned long index = *start_id; + void *ret; + + ret = xa_find(&priv->dev_tbl, &index, UINT_MAX, XA_PRESENT); + *start_id = (int)index; + return ret; +} diff --git a/drivers/infiniband/hw/hfi1/pcie.c b/drivers/infiniband/hw/hfi1/pcie.c index 1a6268d61977..18d32f053d26 100644 --- a/drivers/infiniband/hw/hfi1/pcie.c +++ b/drivers/infiniband/hw/hfi1/pcie.c @@ -306,7 +306,7 @@ int pcie_speeds(struct hfi1_devdata *dd) ret = pcie_capability_read_dword(dd->pcidev, PCI_EXP_LNKCAP, &linkcap); if (ret) { dd_dev_err(dd, "Unable to read from PCI config\n"); - return ret; + return pcibios_err_to_errno(ret); } if ((linkcap & PCI_EXP_LNKCAP_SLS) != PCI_EXP_LNKCAP_SLS_8_0GB) { @@ -334,10 +334,14 @@ int pcie_speeds(struct hfi1_devdata *dd) return 0; } -/* restore command and BARs after a reset has wiped them out */ +/** + * Restore command and BARs after a reset has wiped them out + * + * Returns 0 on success, otherwise a negative error value + */ int restore_pci_variables(struct hfi1_devdata *dd) { - int ret = 0; + int ret; ret = pci_write_config_word(dd->pcidev, PCI_COMMAND, dd->pci_command); if (ret) @@ -386,13 +390,17 @@ int restore_pci_variables(struct hfi1_devdata *dd) error: dd_dev_err(dd, "Unable to write to PCI config\n"); - return ret; + return pcibios_err_to_errno(ret); } -/* Save BARs and command to rewrite after device reset */ +/** + * Save BARs and command to rewrite after device reset + * + * Returns 0 on success, otherwise a negative error value + */ int save_pci_variables(struct hfi1_devdata *dd) { - int ret = 0; + int ret; ret = pci_read_config_dword(dd->pcidev, PCI_BASE_ADDRESS_0, &dd->pcibar0); @@ -441,7 +449,7 @@ int save_pci_variables(struct hfi1_devdata *dd) error: dd_dev_err(dd, "Unable to read from PCI config\n"); - return ret; + return pcibios_err_to_errno(ret); } /* diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c index 79126b2b14ab..ff864f6f0266 100644 --- a/drivers/infiniband/hw/hfi1/pio.c +++ b/drivers/infiniband/hw/hfi1/pio.c @@ -86,7 +86,7 @@ void pio_send_control(struct hfi1_devdata *dd, int op) switch (op) { case PSC_GLOBAL_ENABLE: reg |= SEND_CTRL_SEND_ENABLE_SMASK; - /* Fall through */ + fallthrough; case PSC_DATA_VL_ENABLE: mask = 0; for (i = 0; i < ARRAY_SIZE(dd->vld); i++) diff --git a/drivers/infiniband/hw/hfi1/pio_copy.c b/drivers/infiniband/hw/hfi1/pio_copy.c index 03024cec78dd..4a4ec2397857 100644 --- a/drivers/infiniband/hw/hfi1/pio_copy.c +++ b/drivers/infiniband/hw/hfi1/pio_copy.c @@ -191,25 +191,24 @@ static inline void jcopy(u8 *dest, const u8 *src, u32 n) switch (n) { case 7: *dest++ = *src++; - /* fall through */ + fallthrough; case 6: *dest++ = *src++; - /* fall through */ + fallthrough; case 5: *dest++ = *src++; - /* fall through */ + fallthrough; case 4: *dest++ = *src++; - /* fall through */ + fallthrough; case 3: *dest++ = *src++; - /* fall through */ + fallthrough; case 2: *dest++ = *src++; - /* fall through */ + fallthrough; case 1: *dest++ = *src++; - /* fall through */ } } diff --git a/drivers/infiniband/hw/hfi1/platform.c b/drivers/infiniband/hw/hfi1/platform.c index 36593f2efe26..4642d6ceb890 100644 --- a/drivers/infiniband/hw/hfi1/platform.c +++ b/drivers/infiniband/hw/hfi1/platform.c @@ -668,8 +668,8 @@ static u8 aoc_low_power_setting(struct hfi1_pportdata *ppd) /* active optical cables only */ switch ((cache[QSFP_MOD_TECH_OFFS] & 0xF0) >> 4) { - case 0x0 ... 0x9: /* fallthrough */ - case 0xC: /* fallthrough */ + case 0x0 ... 0x9: fallthrough; + case 0xC: fallthrough; case 0xE: /* active AOC */ power_class = get_qsfp_power_class(cache[QSFP_MOD_PWR_OFFS]); @@ -899,8 +899,8 @@ static int tune_qsfp(struct hfi1_pportdata *ppd, *ptr_tuning_method = OPA_PASSIVE_TUNING; break; - case 0x0 ... 0x9: /* fallthrough */ - case 0xC: /* fallthrough */ + case 0x0 ... 0x9: fallthrough; + case 0xC: fallthrough; case 0xE: ret = tune_active_qsfp(ppd, ptr_tx_preset, ptr_rx_preset, ptr_total_atten); @@ -909,7 +909,7 @@ static int tune_qsfp(struct hfi1_pportdata *ppd, *ptr_tuning_method = OPA_ACTIVE_TUNING; break; - case 0xD: /* fallthrough */ + case 0xD: fallthrough; case 0xF: default: dd_dev_warn(ppd->dd, "%s: Unknown/unsupported cable\n", diff --git a/drivers/infiniband/hw/hfi1/qp.c b/drivers/infiniband/hw/hfi1/qp.c index f8e733aa3bb8..356518e17fa6 100644 --- a/drivers/infiniband/hw/hfi1/qp.c +++ b/drivers/infiniband/hw/hfi1/qp.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2019 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -186,15 +186,6 @@ static void flush_iowait(struct rvt_qp *qp) write_sequnlock_irqrestore(lock, flags); } -static inline int opa_mtu_enum_to_int(int mtu) -{ - switch (mtu) { - case OPA_MTU_8192: return 8192; - case OPA_MTU_10240: return 10240; - default: return -1; - } -} - /** * This function is what we would push to the core layer if we wanted to be a * "first class citizen". Instead we hide this here and rely on Verbs ULPs @@ -202,15 +193,10 @@ static inline int opa_mtu_enum_to_int(int mtu) */ static inline int verbs_mtu_enum_to_int(struct ib_device *dev, enum ib_mtu mtu) { - int val; - /* Constraining 10KB packets to 8KB packets */ if (mtu == (enum ib_mtu)OPA_MTU_10240) - mtu = OPA_MTU_8192; - val = opa_mtu_enum_to_int((int)mtu); - if (val > 0) - return val; - return ib_mtu_enum_to_int(mtu); + mtu = (enum ib_mtu)OPA_MTU_8192; + return opa_mtu_enum_to_int((enum opa_mtu)mtu); } int hfi1_check_modify_qp(struct rvt_qp *qp, struct ib_qp_attr *attr, @@ -326,7 +312,7 @@ int hfi1_setup_wqe(struct rvt_qp *qp, struct rvt_swqe *wqe, bool *call_send) switch (qp->ibqp.qp_type) { case IB_QPT_RC: hfi1_setup_tid_rdma_wqe(qp, wqe); - /* fall through */ + fallthrough; case IB_QPT_UC: if (wqe->length > 0x80000000U) return -EINVAL; @@ -381,7 +367,10 @@ bool _hfi1_schedule_send(struct rvt_qp *qp) struct hfi1_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num); struct hfi1_pportdata *ppd = ppd_from_ibp(ibp); - struct hfi1_devdata *dd = dd_from_ibdev(qp->ibqp.device); + struct hfi1_devdata *dd = ppd->dd; + + if (dd->flags & HFI1_SHUTDOWN) + return true; return iowait_schedule(&priv->s_iowait, ppd->hfi1_wq, priv->s_sde ? diff --git a/drivers/infiniband/hw/hfi1/qp.h b/drivers/infiniband/hw/hfi1/qp.h index b670321365d3..b0d053d12129 100644 --- a/drivers/infiniband/hw/hfi1/qp.h +++ b/drivers/infiniband/hw/hfi1/qp.h @@ -113,20 +113,6 @@ static inline void clear_ahg(struct rvt_qp *qp) } /** - * hfi1_create_qp - create a queue pair for a device - * @ibpd: the protection domain who's device we create the queue pair for - * @init_attr: the attributes of the queue pair - * @udata: user data for libibverbs.so - * - * Returns the queue pair on success, otherwise returns an errno. - * - * Called by the ib_create_qp() core verbs function. - */ -struct ib_qp *hfi1_create_qp(struct ib_pd *ibpd, - struct ib_qp_init_attr *init_attr, - struct ib_udata *udata); - -/** * hfi1_qp_wakeup - wake up on the indicated event * @qp: the QP * @flag: flag the qp on which the qp is stalled diff --git a/drivers/infiniband/hw/hfi1/qsfp.c b/drivers/infiniband/hw/hfi1/qsfp.c index b5966991d647..8386c84c2d92 100644 --- a/drivers/infiniband/hw/hfi1/qsfp.c +++ b/drivers/infiniband/hw/hfi1/qsfp.c @@ -231,7 +231,7 @@ static int i2c_bus_write(struct hfi1_devdata *dd, struct hfi1_i2c_bus *i2c, break; case 2: offset_bytes[1] = (offset >> 8) & 0xff; - /* fall through */ + fallthrough; case 1: num_msgs = 2; offset_bytes[0] = offset & 0xff; @@ -279,7 +279,7 @@ static int i2c_bus_read(struct hfi1_devdata *dd, struct hfi1_i2c_bus *bus, break; case 2: offset_bytes[1] = (offset >> 8) & 0xff; - /* fall through */ + fallthrough; case 1: num_msgs = 2; offset_bytes[0] = offset & 0xff; diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c index f1734e5e9ac4..1bb5f57152d3 100644 --- a/drivers/infiniband/hw/hfi1/rc.c +++ b/drivers/infiniband/hw/hfi1/rc.c @@ -141,7 +141,7 @@ static int make_rc_ack(struct hfi1_ibdev *dev, struct rvt_qp *qp, case OP(RDMA_READ_RESPONSE_ONLY): e = &qp->s_ack_queue[qp->s_tail_ack_queue]; release_rdma_sge_mr(e); - /* FALLTHROUGH */ + fallthrough; case OP(ATOMIC_ACKNOWLEDGE): /* * We can increment the tail pointer now that the last @@ -160,7 +160,7 @@ static int make_rc_ack(struct hfi1_ibdev *dev, struct rvt_qp *qp, qp->s_acked_ack_queue = next; qp->s_tail_ack_queue = next; trace_hfi1_rsp_make_rc_ack(qp, e->psn); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_ONLY): case OP(ACKNOWLEDGE): /* Check for no next entry in the queue. */ @@ -267,7 +267,7 @@ static int make_rc_ack(struct hfi1_ibdev *dev, struct rvt_qp *qp, case OP(RDMA_READ_RESPONSE_FIRST): qp->s_ack_state = OP(RDMA_READ_RESPONSE_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_READ_RESPONSE_MIDDLE): ps->s_txreq->ss = &qp->s_ack_rdma_sge; ps->s_txreq->mr = qp->s_ack_rdma_sge.sge.mr; @@ -881,8 +881,7 @@ no_flow_control: goto bail; } qp->s_num_rd_atomic++; - - /* FALLTHROUGH */ + fallthrough; case IB_WR_OPFN: if (newreq && !(qp->s_flags & RVT_S_UNLIMITED_CREDIT)) qp->s_lsn++; @@ -946,10 +945,10 @@ no_flow_control: * See restart_rc(). */ qp->s_len = restart_sge(&qp->s_sge, wqe, qp->s_psn, pmtu); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_FIRST): qp->s_state = OP(SEND_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): bth2 = mask_psn(qp->s_psn++); ss = &qp->s_sge; @@ -991,10 +990,10 @@ no_flow_control: * See restart_rc(). */ qp->s_len = restart_sge(&qp->s_sge, wqe, qp->s_psn, pmtu); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_FIRST): qp->s_state = OP(RDMA_WRITE_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_MIDDLE): bth2 = mask_psn(qp->s_psn++); ss = &qp->s_sge; @@ -2901,7 +2900,7 @@ void hfi1_rc_rcv(struct hfi1_packet *packet) if (!ret) goto rnr_nak; qp->r_rcv_len = 0; - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): case OP(RDMA_WRITE_MIDDLE): send_middle: @@ -2941,7 +2940,7 @@ send_middle: goto no_immediate_data; if (opcode == OP(SEND_ONLY_WITH_INVALIDATE)) goto send_last_inv; - /* FALLTHROUGH -- for SEND_ONLY_WITH_IMMEDIATE */ + fallthrough; /* for SEND_ONLY_WITH_IMMEDIATE */ case OP(SEND_LAST_WITH_IMMEDIATE): send_last_imm: wc.ex.imm_data = ohdr->u.imm_data; @@ -2957,7 +2956,7 @@ send_last_inv: goto send_last; case OP(RDMA_WRITE_LAST): copy_last = rvt_is_user_qp(qp); - /* fall through */ + fallthrough; case OP(SEND_LAST): no_immediate_data: wc.wc_flags = 0; @@ -3010,7 +3009,7 @@ send_last: case OP(RDMA_WRITE_ONLY): copy_last = rvt_is_user_qp(qp); - /* fall through */ + fallthrough; case OP(RDMA_WRITE_FIRST): case OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE): if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE))) diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c index c93ea021cf49..04575c9afd61 100644 --- a/drivers/infiniband/hw/hfi1/sdma.c +++ b/drivers/infiniband/hw/hfi1/sdma.c @@ -2584,7 +2584,7 @@ static void __sdma_process_event(struct sdma_engine *sde, * 7220, e.g. */ ss->go_s99_running = 1; - /* fall through -- and start dma engine */ + fallthrough; /* and start dma engine */ case sdma_event_e10_go_hw_start: /* This reference means the state machine is started */ sdma_get(&sde->state); @@ -2726,7 +2726,6 @@ static void __sdma_process_event(struct sdma_engine *sde, case sdma_event_e70_go_idle: break; case sdma_event_e85_link_down: - /* fall through */ case sdma_event_e80_hw_freeze: sdma_set_state(sde, sdma_state_s80_hw_freeze); atomic_dec(&sde->dd->sdma_unfreeze_count); @@ -3007,7 +3006,7 @@ static void __sdma_process_event(struct sdma_engine *sde, case sdma_event_e60_hw_halted: need_progress = 1; sdma_err_progress_check_schedule(sde); - /* fall through */ + fallthrough; case sdma_event_e90_sw_halted: /* * SW initiated halt does not perform engines @@ -3021,7 +3020,7 @@ static void __sdma_process_event(struct sdma_engine *sde, break; case sdma_event_e85_link_down: ss->go_s99_running = 0; - /* fall through */ + fallthrough; case sdma_event_e80_hw_freeze: sdma_set_state(sde, sdma_state_s80_hw_freeze); atomic_dec(&sde->dd->sdma_unfreeze_count); @@ -3252,7 +3251,7 @@ void _sdma_txreq_ahgadd( tx->num_desc++; tx->descs[2].qw[0] = 0; tx->descs[2].qw[1] = 0; - /* FALLTHROUGH */ + fallthrough; case SDMA_AHG_APPLY_UPDATE2: tx->num_desc++; tx->descs[1].qw[0] = 0; diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.c b/drivers/infiniband/hw/hfi1/tid_rdma.c index 8a2e0d9351e9..73d197e21730 100644 --- a/drivers/infiniband/hw/hfi1/tid_rdma.c +++ b/drivers/infiniband/hw/hfi1/tid_rdma.c @@ -1,6 +1,6 @@ // SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) /* - * Copyright(c) 2018 Intel Corporation. + * Copyright(c) 2018 - 2020 Intel Corporation. * */ @@ -194,7 +194,7 @@ void tid_rdma_opfn_init(struct rvt_qp *qp, struct tid_rdma_params *p) { struct hfi1_qp_priv *priv = qp->priv; - p->qp = (kdeth_qp << 16) | priv->rcd->ctxt; + p->qp = (RVT_KDETH_QP_PREFIX << 16) | priv->rcd->ctxt; p->max_len = TID_RDMA_MAX_SEGMENT_SIZE; p->jkey = priv->rcd->jkey; p->max_read = TID_RDMA_MAX_READ_SEGS_PER_REQ; @@ -3215,6 +3215,7 @@ bool hfi1_tid_rdma_wqe_interlock(struct rvt_qp *qp, struct rvt_swqe *wqe) case IB_WR_ATOMIC_CMP_AND_SWP: case IB_WR_ATOMIC_FETCH_AND_ADD: case IB_WR_RDMA_WRITE: + case IB_WR_RDMA_WRITE_WITH_IMM: switch (prev->wr.opcode) { case IB_WR_TID_RDMA_WRITE: req = wqe_to_tid_req(prev); @@ -3227,7 +3228,7 @@ bool hfi1_tid_rdma_wqe_interlock(struct rvt_qp *qp, struct rvt_swqe *wqe) case IB_WR_RDMA_READ: if (prev->wr.opcode != IB_WR_TID_RDMA_WRITE) break; - /* fall through */ + fallthrough; case IB_WR_TID_RDMA_READ: switch (prev->wr.opcode) { case IB_WR_RDMA_READ: @@ -5067,7 +5068,7 @@ int hfi1_make_tid_rdma_pkt(struct rvt_qp *qp, struct hfi1_pkt_state *ps) if (priv->s_state == TID_OP(WRITE_REQ)) hfi1_tid_rdma_restart_req(qp, wqe, &bth2); priv->s_state = TID_OP(WRITE_DATA); - /* fall through */ + fallthrough; case TID_OP(WRITE_DATA): /* @@ -5406,7 +5407,10 @@ static bool _hfi1_schedule_tid_send(struct rvt_qp *qp) struct hfi1_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num); struct hfi1_pportdata *ppd = ppd_from_ibp(ibp); - struct hfi1_devdata *dd = dd_from_ibdev(qp->ibqp.device); + struct hfi1_devdata *dd = ppd->dd; + + if ((dd->flags & HFI1_SHUTDOWN)) + return true; return iowait_tid_schedule(&priv->s_iowait, ppd->hfi1_wq, priv->s_sde ? diff --git a/drivers/infiniband/hw/hfi1/trace.c b/drivers/infiniband/hw/hfi1/trace.c index 9a3d236bcc88..b219ea90fd6f 100644 --- a/drivers/infiniband/hw/hfi1/trace.c +++ b/drivers/infiniband/hw/hfi1/trace.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -47,6 +47,7 @@ #define CREATE_TRACE_POINTS #include "trace.h" #include "exp_rcv.h" +#include "ipoib.h" static u8 __get_ib_hdr_len(struct ib_header *hdr) { @@ -126,6 +127,7 @@ const char *hfi1_trace_get_packet_l2_str(u8 l2) #define RETH_PRN "reth vaddr:0x%.16llx rkey:0x%.8x dlen:0x%.8x" #define AETH_PRN "aeth syn:0x%.2x %s msn:0x%.8x" #define DETH_PRN "deth qkey:0x%.8x sqpn:0x%.6x" +#define DETH_ENTROPY_PRN "deth qkey:0x%.8x sqpn:0x%.6x entropy:0x%.2x" #define IETH_PRN "ieth rkey:0x%.8x" #define ATOMICACKETH_PRN "origdata:%llx" #define ATOMICETH_PRN "vaddr:0x%llx rkey:0x%.8x sdata:%llx cdata:%llx" @@ -444,6 +446,12 @@ const char *parse_everbs_hdrs( break; /* deth */ case OP(UD, SEND_ONLY): + trace_seq_printf(p, DETH_ENTROPY_PRN, + be32_to_cpu(eh->ud.deth[0]), + be32_to_cpu(eh->ud.deth[1]) & RVT_QPN_MASK, + be32_to_cpu(eh->ud.deth[1]) >> + HFI1_IPOIB_ENTROPY_SHIFT); + break; case OP(UD, SEND_ONLY_WITH_IMMEDIATE): trace_seq_printf(p, DETH_PRN, be32_to_cpu(eh->ud.deth[0]), @@ -512,6 +520,38 @@ u16 hfi1_trace_get_tid_idx(u32 ent) return EXP_TID_GET(ent, IDX); } +struct hfi1_ctxt_hist { + atomic_t count; + atomic_t data[255]; +}; + +struct hfi1_ctxt_hist hist = { + .count = ATOMIC_INIT(0) +}; + +const char *hfi1_trace_print_rsm_hist(struct trace_seq *p, unsigned int ctxt) +{ + int i, len = ARRAY_SIZE(hist.data); + const char *ret = trace_seq_buffer_ptr(p); + unsigned long packet_count = atomic_fetch_inc(&hist.count); + + trace_seq_printf(p, "packet[%lu]", packet_count); + for (i = 0; i < len; ++i) { + unsigned long val; + atomic_t *count = &hist.data[i]; + + if (ctxt == i) + val = atomic_fetch_inc(count); + else + val = atomic_read(count); + + if (val) + trace_seq_printf(p, "(%d:%lu)", i, val); + } + trace_seq_putc(p, 0); + return ret; +} + __hfi1_trace_fn(AFFINITY); __hfi1_trace_fn(PKT); __hfi1_trace_fn(PROC); diff --git a/drivers/infiniband/hw/hfi1/trace_ctxts.h b/drivers/infiniband/hw/hfi1/trace_ctxts.h index b5fc5c6cd52f..d8c168dc3ea8 100644 --- a/drivers/infiniband/hw/hfi1/trace_ctxts.h +++ b/drivers/infiniband/hw/hfi1/trace_ctxts.h @@ -1,5 +1,5 @@ /* -* Copyright(c) 2015, 2016 Intel Corporation. +* Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -138,6 +138,15 @@ TRACE_EVENT(hfi1_ctxt_info, ) ); +const char *hfi1_trace_print_rsm_hist(struct trace_seq *p, unsigned int ctxt); +TRACE_EVENT(ctxt_rsm_hist, + TP_PROTO(unsigned int ctxt), + TP_ARGS(ctxt), + TP_STRUCT__entry(__field(unsigned int, ctxt)), + TP_fast_assign(__entry->ctxt = ctxt;), + TP_printk("%s", hfi1_trace_print_rsm_hist(p, __entry->ctxt)) +); + #endif /* __HFI1_TRACE_CTXTS_H */ #undef TRACE_INCLUDE_PATH diff --git a/drivers/infiniband/hw/hfi1/uc.c b/drivers/infiniband/hw/hfi1/uc.c index 0c77f18120ed..1fb918399da0 100644 --- a/drivers/infiniband/hw/hfi1/uc.c +++ b/drivers/infiniband/hw/hfi1/uc.c @@ -216,7 +216,7 @@ int hfi1_make_uc_req(struct rvt_qp *qp, struct hfi1_pkt_state *ps) case OP(SEND_FIRST): qp->s_state = OP(SEND_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): len = qp->s_len; if (len > pmtu) { @@ -241,7 +241,7 @@ int hfi1_make_uc_req(struct rvt_qp *qp, struct hfi1_pkt_state *ps) case OP(RDMA_WRITE_FIRST): qp->s_state = OP(RDMA_WRITE_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_MIDDLE): len = qp->s_len; if (len > pmtu) { @@ -414,7 +414,7 @@ send_first: goto no_immediate_data; else if (opcode == OP(SEND_ONLY_WITH_IMMEDIATE)) goto send_last_imm; - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): /* Check for invalid length PMTU or posted rwqe len. */ /* @@ -515,7 +515,7 @@ rdma_first: wc.ex.imm_data = ohdr->u.rc.imm_data; goto rdma_last_imm; } - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_MIDDLE): /* Check for invalid length PMTU or posted rwqe len. */ if (unlikely(tlen != (hdrsize + pmtu + 4))) diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c index 4da03f823474..f81ca20f4b69 100644 --- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c +++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c @@ -206,13 +206,6 @@ static int pin_rcv_pages(struct hfi1_filedata *fd, struct tid_user_buf *tidbuf) return -EINVAL; } - /* Verify that access is OK for the user buffer */ - if (!access_ok((void __user *)vaddr, - npages * PAGE_SIZE)) { - dd_dev_err(dd, "Fail vaddr %p, %u pages, !access_ok\n", - (void *)vaddr, npages); - return -EFAULT; - } /* Allocate the array of struct page pointers needed for pinning */ pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL); if (!pages) diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 2f6323ad9c59..30865635b449 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015 - 2018 Intel Corporation. + * Copyright(c) 2015 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -66,6 +66,7 @@ #include "vnic.h" #include "fault.h" #include "affinity.h" +#include "ipoib.h" static unsigned int hfi1_lkey_table_size = 16; module_param_named(lkey_table_size, hfi1_lkey_table_size, uint, @@ -1342,7 +1343,7 @@ static void hfi1_fill_device_attr(struct hfi1_devdata *dd) IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN | IB_DEVICE_PORT_ACTIVE_EVENT | IB_DEVICE_SRQ_RESIZE | IB_DEVICE_MEM_MGT_EXTENSIONS | - IB_DEVICE_RDMA_NETDEV_OPA_VNIC; + IB_DEVICE_RDMA_NETDEV_OPA; rdi->dparms.props.page_size_cap = PAGE_SIZE; rdi->dparms.props.vendor_id = dd->oui1 << 16 | dd->oui2 << 8 | dd->oui3; rdi->dparms.props.vendor_part_id = dd->pcidev->device; @@ -1360,7 +1361,6 @@ static void hfi1_fill_device_attr(struct hfi1_devdata *dd) rdi->dparms.props.max_cq = hfi1_max_cqs; rdi->dparms.props.max_ah = hfi1_max_ahs; rdi->dparms.props.max_cqe = hfi1_max_cqes; - rdi->dparms.props.max_map_per_fmr = 32767; rdi->dparms.props.max_pd = hfi1_max_pds; rdi->dparms.props.max_qp_rd_atom = HFI1_MAX_RDMA_ATOMIC; rdi->dparms.props.max_qp_init_rd_atom = 255; @@ -1439,6 +1439,8 @@ static int query_port(struct rvt_dev_info *rdi, u8 port_num, 4096 : hfi1_max_mtu), IB_MTU_4096); props->active_mtu = !valid_ib_mtu(ppd->ibmtu) ? props->max_mtu : mtu_to_enum(ppd->ibmtu, IB_MTU_4096); + props->phys_mtu = HFI1_CAP_IS_KSET(AIP) ? hfi1_max_mtu : + ib_mtu_enum_to_int(props->max_mtu); return 0; } @@ -1793,6 +1795,7 @@ static const struct ib_device_ops hfi1_dev_ops = { .modify_device = modify_device, /* keep process mad in the driver */ .process_mad = hfi1_process_mad, + .rdma_netdev_get_params = hfi1_ipoib_rn_get_params, }; /** @@ -1863,9 +1866,8 @@ int hfi1_register_ib_device(struct hfi1_devdata *dd) dd->verbs_dev.rdi.dparms.qpn_start = 0; dd->verbs_dev.rdi.dparms.qpn_inc = 1; dd->verbs_dev.rdi.dparms.qos_shift = dd->qos_shift; - dd->verbs_dev.rdi.dparms.qpn_res_start = kdeth_qp << 16; - dd->verbs_dev.rdi.dparms.qpn_res_end = - dd->verbs_dev.rdi.dparms.qpn_res_start + 65535; + dd->verbs_dev.rdi.dparms.qpn_res_start = RVT_KDETH_QP_BASE; + dd->verbs_dev.rdi.dparms.qpn_res_end = RVT_AIP_QP_MAX; dd->verbs_dev.rdi.dparms.max_rdma_atomic = HFI1_MAX_RDMA_ATOMIC; dd->verbs_dev.rdi.dparms.psn_mask = PSN_MASK; dd->verbs_dev.rdi.dparms.psn_shift = PSN_SHIFT; diff --git a/drivers/infiniband/hw/hfi1/verbs_txreq.h b/drivers/infiniband/hw/hfi1/verbs_txreq.h index bfa6e081cb56..d2d526c5a756 100644 --- a/drivers/infiniband/hw/hfi1/verbs_txreq.h +++ b/drivers/infiniband/hw/hfi1/verbs_txreq.h @@ -91,7 +91,7 @@ static inline struct verbs_txreq *get_txreq(struct hfi1_ibdev *dev, tx->mr = NULL; tx->sde = priv->s_sde; tx->psc = priv->s_sendcontext; - /* so that we can test if the sdma decriptors are there */ + /* so that we can test if the sdma descriptors are there */ tx->txreq.num_desc = 0; /* Set the header type */ tx->phdr.hdr.hdr_type = priv->hdr_type; diff --git a/drivers/infiniband/hw/hfi1/vnic.h b/drivers/infiniband/hw/hfi1/vnic.h index 5ae781514e32..66150a13f374 100644 --- a/drivers/infiniband/hw/hfi1/vnic.h +++ b/drivers/infiniband/hw/hfi1/vnic.h @@ -1,7 +1,7 @@ #ifndef _HFI1_VNIC_H #define _HFI1_VNIC_H /* - * Copyright(c) 2017 Intel Corporation. + * Copyright(c) 2017 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -69,6 +69,7 @@ #define HFI1_VNIC_SC_SHIFT 4 #define HFI1_VNIC_MAX_QUEUE 16 +#define HFI1_NUM_VNIC_CTXT 8 /** * struct hfi1_vnic_sdma - VNIC per Tx ring SDMA information @@ -104,7 +105,6 @@ struct hfi1_vnic_rx_queue { struct hfi1_vnic_vport_info *vinfo; struct net_device *netdev; struct napi_struct napi; - struct sk_buff_head skbq; }; /** @@ -146,7 +146,6 @@ struct hfi1_vnic_vport_info { /* vnic hfi1 internal functions */ void hfi1_vnic_setup(struct hfi1_devdata *dd); -void hfi1_vnic_cleanup(struct hfi1_devdata *dd); int hfi1_vnic_txreq_init(struct hfi1_devdata *dd); void hfi1_vnic_txreq_deinit(struct hfi1_devdata *dd); diff --git a/drivers/infiniband/hw/hfi1/vnic_main.c b/drivers/infiniband/hw/hfi1/vnic_main.c index 6b14581b9965..a90824de0f57 100644 --- a/drivers/infiniband/hw/hfi1/vnic_main.c +++ b/drivers/infiniband/hw/hfi1/vnic_main.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2017 - 2018 Intel Corporation. + * Copyright(c) 2017 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -53,6 +53,7 @@ #include <linux/if_vlan.h> #include "vnic.h" +#include "netdev.h" #define HFI_TX_TIMEOUT_MS 1000 @@ -62,114 +63,6 @@ static DEFINE_SPINLOCK(vport_cntr_lock); -static int setup_vnic_ctxt(struct hfi1_devdata *dd, struct hfi1_ctxtdata *uctxt) -{ - unsigned int rcvctrl_ops = 0; - int ret; - - uctxt->do_interrupt = &handle_receive_interrupt; - - /* Now allocate the RcvHdr queue and eager buffers. */ - ret = hfi1_create_rcvhdrq(dd, uctxt); - if (ret) - goto done; - - ret = hfi1_setup_eagerbufs(uctxt); - if (ret) - goto done; - - if (hfi1_rcvhdrtail_kvaddr(uctxt)) - clear_rcvhdrtail(uctxt); - - rcvctrl_ops = HFI1_RCVCTRL_CTXT_ENB; - rcvctrl_ops |= HFI1_RCVCTRL_INTRAVAIL_ENB; - - if (!HFI1_CAP_KGET_MASK(uctxt->flags, MULTI_PKT_EGR)) - rcvctrl_ops |= HFI1_RCVCTRL_ONE_PKT_EGR_ENB; - if (HFI1_CAP_KGET_MASK(uctxt->flags, NODROP_EGR_FULL)) - rcvctrl_ops |= HFI1_RCVCTRL_NO_EGR_DROP_ENB; - if (HFI1_CAP_KGET_MASK(uctxt->flags, NODROP_RHQ_FULL)) - rcvctrl_ops |= HFI1_RCVCTRL_NO_RHQ_DROP_ENB; - if (HFI1_CAP_KGET_MASK(uctxt->flags, DMA_RTAIL)) - rcvctrl_ops |= HFI1_RCVCTRL_TAILUPD_ENB; - - hfi1_rcvctrl(uctxt->dd, rcvctrl_ops, uctxt); -done: - return ret; -} - -static int allocate_vnic_ctxt(struct hfi1_devdata *dd, - struct hfi1_ctxtdata **vnic_ctxt) -{ - struct hfi1_ctxtdata *uctxt; - int ret; - - if (dd->flags & HFI1_FROZEN) - return -EIO; - - ret = hfi1_create_ctxtdata(dd->pport, dd->node, &uctxt); - if (ret < 0) { - dd_dev_err(dd, "Unable to create ctxtdata, failing open\n"); - return -ENOMEM; - } - - uctxt->flags = HFI1_CAP_KGET(MULTI_PKT_EGR) | - HFI1_CAP_KGET(NODROP_RHQ_FULL) | - HFI1_CAP_KGET(NODROP_EGR_FULL) | - HFI1_CAP_KGET(DMA_RTAIL); - uctxt->seq_cnt = 1; - uctxt->is_vnic = true; - - msix_request_rcd_irq(uctxt); - - hfi1_stats.sps_ctxts++; - dd_dev_dbg(dd, "created vnic context %d\n", uctxt->ctxt); - *vnic_ctxt = uctxt; - - return 0; -} - -static void deallocate_vnic_ctxt(struct hfi1_devdata *dd, - struct hfi1_ctxtdata *uctxt) -{ - dd_dev_dbg(dd, "closing vnic context %d\n", uctxt->ctxt); - flush_wc(); - - /* - * Disable receive context and interrupt available, reset all - * RcvCtxtCtrl bits to default values. - */ - hfi1_rcvctrl(dd, HFI1_RCVCTRL_CTXT_DIS | - HFI1_RCVCTRL_TIDFLOW_DIS | - HFI1_RCVCTRL_INTRAVAIL_DIS | - HFI1_RCVCTRL_ONE_PKT_EGR_DIS | - HFI1_RCVCTRL_NO_RHQ_DROP_DIS | - HFI1_RCVCTRL_NO_EGR_DROP_DIS, uctxt); - - /* msix_intr will always be > 0, only clean up if this is true */ - if (uctxt->msix_intr) - msix_free_irq(dd, uctxt->msix_intr); - - uctxt->event_flags = 0; - - hfi1_clear_tids(uctxt); - hfi1_clear_ctxt_pkey(dd, uctxt); - - hfi1_stats.sps_ctxts--; - - hfi1_free_ctxt(uctxt); -} - -void hfi1_vnic_setup(struct hfi1_devdata *dd) -{ - xa_init(&dd->vnic.vesws); -} - -void hfi1_vnic_cleanup(struct hfi1_devdata *dd) -{ - WARN_ON(!xa_empty(&dd->vnic.vesws)); -} - #define SUM_GRP_COUNTERS(stats, qstats, x_grp) do { \ u64 *src64, *dst64; \ for (src64 = &qstats->x_grp.unicast, \ @@ -179,6 +72,9 @@ void hfi1_vnic_cleanup(struct hfi1_devdata *dd) } \ } while (0) +#define VNIC_MASK (0xFF) +#define VNIC_ID(val) ((1ull << 24) | ((val) & VNIC_MASK)) + /* hfi1_vnic_update_stats - update statistics */ static void hfi1_vnic_update_stats(struct hfi1_vnic_vport_info *vinfo, struct opa_vnic_stats *stats) @@ -454,71 +350,25 @@ static inline int hfi1_vnic_decap_skb(struct hfi1_vnic_rx_queue *rxq, return rc; } -static inline struct sk_buff *hfi1_vnic_get_skb(struct hfi1_vnic_rx_queue *rxq) +static struct hfi1_vnic_vport_info *get_vnic_port(struct hfi1_devdata *dd, + int vesw_id) { - unsigned char *pad_info; - struct sk_buff *skb; + int vnic_id = VNIC_ID(vesw_id); - skb = skb_dequeue(&rxq->skbq); - if (unlikely(!skb)) - return NULL; - - /* remove tail padding and icrc */ - pad_info = skb->data + skb->len - 1; - skb_trim(skb, (skb->len - OPA_VNIC_ICRC_TAIL_LEN - - ((*pad_info) & 0x7))); - - return skb; + return hfi1_netdev_get_data(dd, vnic_id); } -/* hfi1_vnic_handle_rx - handle skb receive */ -static void hfi1_vnic_handle_rx(struct hfi1_vnic_rx_queue *rxq, - int *work_done, int work_to_do) +static struct hfi1_vnic_vport_info *get_first_vnic_port(struct hfi1_devdata *dd) { - struct hfi1_vnic_vport_info *vinfo = rxq->vinfo; - struct sk_buff *skb; - int rc; - - while (1) { - if (*work_done >= work_to_do) - break; - - skb = hfi1_vnic_get_skb(rxq); - if (unlikely(!skb)) - break; - - rc = hfi1_vnic_decap_skb(rxq, skb); - /* update rx counters */ - hfi1_vnic_update_rx_counters(vinfo, rxq->idx, skb, rc); - if (unlikely(rc)) { - dev_kfree_skb_any(skb); - continue; - } - - skb_checksum_none_assert(skb); - skb->protocol = eth_type_trans(skb, rxq->netdev); - - napi_gro_receive(&rxq->napi, skb); - (*work_done)++; - } -} - -/* hfi1_vnic_napi - napi receive polling callback function */ -static int hfi1_vnic_napi(struct napi_struct *napi, int budget) -{ - struct hfi1_vnic_rx_queue *rxq = container_of(napi, - struct hfi1_vnic_rx_queue, napi); - struct hfi1_vnic_vport_info *vinfo = rxq->vinfo; - int work_done = 0; + struct hfi1_vnic_vport_info *vinfo; + int next_id = VNIC_ID(0); - v_dbg("napi %d budget %d\n", rxq->idx, budget); - hfi1_vnic_handle_rx(rxq, &work_done, budget); + vinfo = hfi1_netdev_get_first_data(dd, &next_id); - v_dbg("napi %d work_done %d\n", rxq->idx, work_done); - if (work_done < budget) - napi_complete(napi); + if (next_id > VNIC_ID(VNIC_MASK)) + return NULL; - return work_done; + return vinfo; } void hfi1_vnic_bypass_rcv(struct hfi1_packet *packet) @@ -527,13 +377,14 @@ void hfi1_vnic_bypass_rcv(struct hfi1_packet *packet) struct hfi1_vnic_vport_info *vinfo = NULL; struct hfi1_vnic_rx_queue *rxq; struct sk_buff *skb; - int l4_type, vesw_id = -1; + int l4_type, vesw_id = -1, rc; u8 q_idx; + unsigned char *pad_info; l4_type = hfi1_16B_get_l4(packet->ebuf); if (likely(l4_type == OPA_16B_L4_ETHR)) { vesw_id = HFI1_VNIC_GET_VESWID(packet->ebuf); - vinfo = xa_load(&dd->vnic.vesws, vesw_id); + vinfo = get_vnic_port(dd, vesw_id); /* * In case of invalid vesw id, count the error on @@ -541,10 +392,8 @@ void hfi1_vnic_bypass_rcv(struct hfi1_packet *packet) */ if (unlikely(!vinfo)) { struct hfi1_vnic_vport_info *vinfo_tmp; - unsigned long index = 0; - vinfo_tmp = xa_find(&dd->vnic.vesws, &index, ULONG_MAX, - XA_PRESENT); + vinfo_tmp = get_first_vnic_port(dd); if (vinfo_tmp) { spin_lock(&vport_cntr_lock); vinfo_tmp->stats[0].netstats.rx_nohandler++; @@ -563,12 +412,6 @@ void hfi1_vnic_bypass_rcv(struct hfi1_packet *packet) rxq = &vinfo->rxq[q_idx]; if (unlikely(!netif_oper_up(vinfo->netdev))) { vinfo->stats[q_idx].rx_drop_state++; - skb_queue_purge(&rxq->skbq); - return; - } - - if (unlikely(skb_queue_len(&rxq->skbq) > HFI1_VNIC_RCV_Q_SIZE)) { - vinfo->stats[q_idx].netstats.rx_fifo_errors++; return; } @@ -580,62 +423,65 @@ void hfi1_vnic_bypass_rcv(struct hfi1_packet *packet) memcpy(skb->data, packet->ebuf, packet->tlen); skb_put(skb, packet->tlen); - skb_queue_tail(&rxq->skbq, skb); - if (napi_schedule_prep(&rxq->napi)) { - v_dbg("napi %d scheduling\n", q_idx); - __napi_schedule(&rxq->napi); + pad_info = skb->data + skb->len - 1; + skb_trim(skb, (skb->len - OPA_VNIC_ICRC_TAIL_LEN - + ((*pad_info) & 0x7))); + + rc = hfi1_vnic_decap_skb(rxq, skb); + + /* update rx counters */ + hfi1_vnic_update_rx_counters(vinfo, rxq->idx, skb, rc); + if (unlikely(rc)) { + dev_kfree_skb_any(skb); + return; } + + skb_checksum_none_assert(skb); + skb->protocol = eth_type_trans(skb, rxq->netdev); + + napi_gro_receive(&rxq->napi, skb); } static int hfi1_vnic_up(struct hfi1_vnic_vport_info *vinfo) { struct hfi1_devdata *dd = vinfo->dd; struct net_device *netdev = vinfo->netdev; - int i, rc; + int rc; /* ensure virtual eth switch id is valid */ if (!vinfo->vesw_id) return -EINVAL; - rc = xa_insert(&dd->vnic.vesws, vinfo->vesw_id, vinfo, GFP_KERNEL); + rc = hfi1_netdev_add_data(dd, VNIC_ID(vinfo->vesw_id), vinfo); if (rc < 0) return rc; - for (i = 0; i < vinfo->num_rx_q; i++) { - struct hfi1_vnic_rx_queue *rxq = &vinfo->rxq[i]; - - skb_queue_head_init(&rxq->skbq); - napi_enable(&rxq->napi); - } + rc = hfi1_netdev_rx_init(dd); + if (rc) + goto err_remove; netif_carrier_on(netdev); netif_tx_start_all_queues(netdev); set_bit(HFI1_VNIC_UP, &vinfo->flags); return 0; + +err_remove: + hfi1_netdev_remove_data(dd, VNIC_ID(vinfo->vesw_id)); + return rc; } static void hfi1_vnic_down(struct hfi1_vnic_vport_info *vinfo) { struct hfi1_devdata *dd = vinfo->dd; - u8 i; clear_bit(HFI1_VNIC_UP, &vinfo->flags); netif_carrier_off(vinfo->netdev); netif_tx_disable(vinfo->netdev); - xa_erase(&dd->vnic.vesws, vinfo->vesw_id); - - /* ensure irqs see the change */ - msix_vnic_synchronize_irq(dd); + hfi1_netdev_remove_data(dd, VNIC_ID(vinfo->vesw_id)); - /* remove unread skbs */ - for (i = 0; i < vinfo->num_rx_q; i++) { - struct hfi1_vnic_rx_queue *rxq = &vinfo->rxq[i]; - - napi_disable(&rxq->napi); - skb_queue_purge(&rxq->skbq); - } + hfi1_netdev_rx_destroy(dd); } static int hfi1_netdev_open(struct net_device *netdev) @@ -660,70 +506,31 @@ static int hfi1_netdev_close(struct net_device *netdev) return 0; } -static int hfi1_vnic_allot_ctxt(struct hfi1_devdata *dd, - struct hfi1_ctxtdata **vnic_ctxt) -{ - int rc; - - rc = allocate_vnic_ctxt(dd, vnic_ctxt); - if (rc) { - dd_dev_err(dd, "vnic ctxt alloc failed %d\n", rc); - return rc; - } - - rc = setup_vnic_ctxt(dd, *vnic_ctxt); - if (rc) { - dd_dev_err(dd, "vnic ctxt setup failed %d\n", rc); - deallocate_vnic_ctxt(dd, *vnic_ctxt); - *vnic_ctxt = NULL; - } - - return rc; -} - static int hfi1_vnic_init(struct hfi1_vnic_vport_info *vinfo) { struct hfi1_devdata *dd = vinfo->dd; - int i, rc = 0; + int rc = 0; mutex_lock(&hfi1_mutex); - if (!dd->vnic.num_vports) { + if (!dd->vnic_num_vports) { rc = hfi1_vnic_txreq_init(dd); if (rc) goto txreq_fail; } - for (i = dd->vnic.num_ctxt; i < vinfo->num_rx_q; i++) { - rc = hfi1_vnic_allot_ctxt(dd, &dd->vnic.ctxt[i]); - if (rc) - break; - hfi1_rcd_get(dd->vnic.ctxt[i]); - dd->vnic.ctxt[i]->vnic_q_idx = i; - } - - if (i < vinfo->num_rx_q) { - /* - * If required amount of contexts is not - * allocated successfully then remaining contexts - * are released. - */ - while (i-- > dd->vnic.num_ctxt) { - deallocate_vnic_ctxt(dd, dd->vnic.ctxt[i]); - hfi1_rcd_put(dd->vnic.ctxt[i]); - dd->vnic.ctxt[i] = NULL; - } + rc = hfi1_netdev_rx_init(dd); + if (rc) { + dd_dev_err(dd, "Unable to initialize netdev contexts\n"); goto alloc_fail; } - if (dd->vnic.num_ctxt != i) { - dd->vnic.num_ctxt = i; - hfi1_init_vnic_rsm(dd); - } + hfi1_init_vnic_rsm(dd); - dd->vnic.num_vports++; + dd->vnic_num_vports++; hfi1_vnic_sdma_init(vinfo); + alloc_fail: - if (!dd->vnic.num_vports) + if (!dd->vnic_num_vports) hfi1_vnic_txreq_deinit(dd); txreq_fail: mutex_unlock(&hfi1_mutex); @@ -733,20 +540,14 @@ txreq_fail: static void hfi1_vnic_deinit(struct hfi1_vnic_vport_info *vinfo) { struct hfi1_devdata *dd = vinfo->dd; - int i; mutex_lock(&hfi1_mutex); - if (--dd->vnic.num_vports == 0) { - for (i = 0; i < dd->vnic.num_ctxt; i++) { - deallocate_vnic_ctxt(dd, dd->vnic.ctxt[i]); - hfi1_rcd_put(dd->vnic.ctxt[i]); - dd->vnic.ctxt[i] = NULL; - } + if (--dd->vnic_num_vports == 0) { hfi1_deinit_vnic_rsm(dd); - dd->vnic.num_ctxt = 0; hfi1_vnic_txreq_deinit(dd); } mutex_unlock(&hfi1_mutex); + hfi1_netdev_rx_destroy(dd); } static void hfi1_vnic_set_vesw_id(struct net_device *netdev, int id) @@ -804,7 +605,7 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, struct rdma_netdev *rn; int i, size, rc; - if (!dd->num_vnic_contexts) + if (!dd->num_netdev_contexts) return ERR_PTR(-ENOMEM); if (!port_num || (port_num > dd->num_pports)) @@ -815,15 +616,16 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, size = sizeof(struct opa_vnic_rdma_netdev) + sizeof(*vinfo); netdev = alloc_netdev_mqs(size, name, name_assign_type, setup, - dd->num_sdma, dd->num_vnic_contexts); + chip_sdma_engines(dd), + dd->num_netdev_contexts); if (!netdev) return ERR_PTR(-ENOMEM); rn = netdev_priv(netdev); vinfo = opa_vnic_dev_priv(netdev); vinfo->dd = dd; - vinfo->num_tx_q = dd->num_sdma; - vinfo->num_rx_q = dd->num_vnic_contexts; + vinfo->num_tx_q = chip_sdma_engines(dd); + vinfo->num_rx_q = dd->num_netdev_contexts; vinfo->netdev = netdev; rn->free_rdma_netdev = hfi1_vnic_free_rn; rn->set_id = hfi1_vnic_set_vesw_id; @@ -841,7 +643,6 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, rxq->idx = i; rxq->vinfo = vinfo; rxq->netdev = netdev; - netif_napi_add(netdev, &rxq->napi, hfi1_vnic_napi, 64); } rc = hfi1_vnic_init(vinfo); diff --git a/drivers/infiniband/hw/hns/Kconfig b/drivers/infiniband/hw/hns/Kconfig index 4921c1e40ccd..18d10ebf900b 100644 --- a/drivers/infiniband/hw/hns/Kconfig +++ b/drivers/infiniband/hw/hns/Kconfig @@ -4,7 +4,7 @@ config INFINIBAND_HNS depends on NET_VENDOR_HISILICON depends on ARM64 || (COMPILE_TEST && 64BIT) depends on (HNS_DSAF && HNS_ENET) || HNS3 - ---help--- + help This is a RoCE/RDMA driver for the Hisilicon RoCE engine. The engine is used in Hisilicon Hip06 and more further ICT SoC based on platform device. @@ -15,7 +15,7 @@ config INFINIBAND_HNS_HIP06 bool "Hisilicon Hip06 Family RoCE support" depends on INFINIBAND_HNS && HNS && HNS_DSAF && HNS_ENET depends on INFINIBAND_HNS=m || (HNS_DSAF=y && HNS_ENET=y) - ---help--- + help RoCE driver support for Hisilicon RoCE engine in Hisilicon Hip06 and Hip07 SoC. These RoCE engines are platform devices. @@ -26,7 +26,7 @@ config INFINIBAND_HNS_HIP08 bool "Hisilicon Hip08 Family RoCE support" depends on INFINIBAND_HNS && PCI && HNS3 depends on INFINIBAND_HNS=m || HNS3=y - ---help--- + help RoCE driver support for Hisilicon RoCE engine in Hisilicon Hip08 SoC. The RoCE engine is a PCI device. diff --git a/drivers/infiniband/hw/hns/hns_roce_ah.c b/drivers/infiniband/hw/hns/hns_roce_ah.c index 8a522e14ef62..5b2f9314edd3 100644 --- a/drivers/infiniband/hw/hns/hns_roce_ah.c +++ b/drivers/infiniband/hw/hns/hns_roce_ah.c @@ -39,13 +39,14 @@ #define HNS_ROCE_VLAN_SL_BIT_MASK 7 #define HNS_ROCE_VLAN_SL_SHIFT 13 -int hns_roce_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata) +int hns_roce_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(ibah->device); const struct ib_gid_attr *gid_attr; struct device *dev = hr_dev->dev; struct hns_roce_ah *ah = to_hr_ah(ibah); + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; const struct ib_global_route *grh = rdma_ah_read_grh(ah_attr); u16 vlan_id = 0xffff; bool vlan_en = false; diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c b/drivers/infiniband/hw/hns/hns_roce_alloc.c index da574c26e063..a522cb2d29ea 100644 --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c @@ -157,84 +157,78 @@ void hns_roce_bitmap_cleanup(struct hns_roce_bitmap *bitmap) kfree(bitmap->table); } -void hns_roce_buf_free(struct hns_roce_dev *hr_dev, u32 size, - struct hns_roce_buf *buf) +void hns_roce_buf_free(struct hns_roce_dev *hr_dev, struct hns_roce_buf *buf) { - int i; struct device *dev = hr_dev->dev; + u32 size = buf->size; + int i; + + if (size == 0) + return; - if (buf->nbufs == 1) { + buf->size = 0; + + if (hns_roce_buf_is_direct(buf)) { dma_free_coherent(dev, size, buf->direct.buf, buf->direct.map); } else { - for (i = 0; i < buf->nbufs; ++i) + for (i = 0; i < buf->npages; ++i) if (buf->page_list[i].buf) dma_free_coherent(dev, 1 << buf->page_shift, buf->page_list[i].buf, buf->page_list[i].map); kfree(buf->page_list); + buf->page_list = NULL; } } int hns_roce_buf_alloc(struct hns_roce_dev *hr_dev, u32 size, u32 max_direct, struct hns_roce_buf *buf, u32 page_shift) { - int i = 0; - dma_addr_t t; + struct hns_roce_buf_list *buf_list; struct device *dev = hr_dev->dev; - u32 page_size = 1 << page_shift; - u32 order; + u32 page_size; + int i; - /* SQ/RQ buf lease than one page, SQ + RQ = 8K */ + /* The minimum shift of the page accessed by hw is HNS_HW_PAGE_SHIFT */ + buf->page_shift = max_t(int, HNS_HW_PAGE_SHIFT, page_shift); + + page_size = 1 << buf->page_shift; + buf->npages = DIV_ROUND_UP(size, page_size); + + /* required size is not bigger than one trunk size */ if (size <= max_direct) { - buf->nbufs = 1; - /* Npages calculated by page_size */ - order = get_order(size); - if (order <= page_shift - PAGE_SHIFT) - order = 0; - else - order -= page_shift - PAGE_SHIFT; - buf->npages = 1 << order; - buf->page_shift = page_shift; - /* MTT PA must be recorded in 4k alignment, t is 4k aligned */ - buf->direct.buf = dma_alloc_coherent(dev, size, &t, + buf->page_list = NULL; + buf->direct.buf = dma_alloc_coherent(dev, size, + &buf->direct.map, GFP_KERNEL); if (!buf->direct.buf) return -ENOMEM; - - buf->direct.map = t; - - while (t & ((1 << buf->page_shift) - 1)) { - --buf->page_shift; - buf->npages *= 2; - } } else { - buf->nbufs = (size + page_size - 1) / page_size; - buf->npages = buf->nbufs; - buf->page_shift = page_shift; - buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list), - GFP_KERNEL); - - if (!buf->page_list) + buf_list = kcalloc(buf->npages, sizeof(*buf_list), GFP_KERNEL); + if (!buf_list) return -ENOMEM; - for (i = 0; i < buf->nbufs; ++i) { - buf->page_list[i].buf = dma_alloc_coherent(dev, - page_size, - &t, - GFP_KERNEL); - - if (!buf->page_list[i].buf) - goto err_free; + for (i = 0; i < buf->npages; i++) { + buf_list[i].buf = dma_alloc_coherent(dev, page_size, + &buf_list[i].map, + GFP_KERNEL); + if (!buf_list[i].buf) + break; + } - buf->page_list[i].map = t; + if (i != buf->npages && i > 0) { + while (i-- > 0) + dma_free_coherent(dev, page_size, + buf_list[i].buf, + buf_list[i].map); + kfree(buf_list); + return -ENOMEM; } + buf->page_list = buf_list; } + buf->size = size; return 0; - -err_free: - hns_roce_buf_free(hr_dev, size, buf); - return -ENOMEM; } int hns_roce_get_kmem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs, @@ -246,33 +240,30 @@ int hns_roce_get_kmem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs, end = start + buf_cnt; if (end > buf->npages) { dev_err(hr_dev->dev, - "invalid kmem region,offset %d,buf_cnt %d,total %d!\n", + "Failed to check kmem bufs, end %d + %d total %d!\n", start, buf_cnt, buf->npages); return -EINVAL; } total = 0; for (i = start; i < end; i++) - if (buf->nbufs == 1) - bufs[total++] = buf->direct.map + - ((dma_addr_t)i << buf->page_shift); - else - bufs[total++] = buf->page_list[i].map; + bufs[total++] = hns_roce_buf_page(buf, i); return total; } int hns_roce_get_umem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs, int buf_cnt, int start, struct ib_umem *umem, - int page_shift) + unsigned int page_shift) { struct ib_block_iter biter; int total = 0; int idx = 0; u64 addr; - if (page_shift < PAGE_SHIFT) { - dev_err(hr_dev->dev, "invalid page shift %d!\n", page_shift); + if (page_shift < HNS_HW_PAGE_SHIFT) { + dev_err(hr_dev->dev, "Failed to check umem page shift %d!\n", + page_shift); return -EINVAL; } @@ -292,49 +283,6 @@ done: return total; } -void hns_roce_init_buf_region(struct hns_roce_buf_region *region, int hopnum, - int offset, int buf_cnt) -{ - if (hopnum == HNS_ROCE_HOP_NUM_0) - region->hopnum = 0; - else - region->hopnum = hopnum; - - region->offset = offset; - region->count = buf_cnt; -} - -void hns_roce_free_buf_list(dma_addr_t **bufs, int region_cnt) -{ - int i; - - for (i = 0; i < region_cnt; i++) { - kfree(bufs[i]); - bufs[i] = NULL; - } -} - -int hns_roce_alloc_buf_list(struct hns_roce_buf_region *regions, - dma_addr_t **bufs, int region_cnt) -{ - struct hns_roce_buf_region *r; - int i; - - for (i = 0; i < region_cnt; i++) { - r = ®ions[i]; - bufs[i] = kcalloc(r->count, sizeof(dma_addr_t), GFP_KERNEL); - if (!bufs[i]) - goto err_alloc; - } - - return 0; - -err_alloc: - hns_roce_free_buf_list(bufs, i); - - return -ENOMEM; -} - void hns_roce_cleanup_bitmap(struct hns_roce_dev *hr_dev) { if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h index 8e95a1aa1b4f..f5669ff8cfeb 100644 --- a/drivers/infiniband/hw/hns/hns_roce_common.h +++ b/drivers/infiniband/hw/hns/hns_roce_common.h @@ -33,10 +33,6 @@ #ifndef _HNS_ROCE_COMMON_H #define _HNS_ROCE_COMMON_H -#ifndef assert -#define assert(cond) -#endif - #define roce_write(dev, reg, val) writel((val), (dev)->reg_base + (reg)) #define roce_read(dev, reg) readl((dev)->reg_base + (reg)) #define roce_raw_write(value, addr) \ diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c index 5bfb52ffd590..e87d616f7988 100644 --- a/drivers/infiniband/hw/hns/hns_roce_cq.c +++ b/drivers/infiniband/hw/hns/hns_roce_cq.c @@ -39,51 +39,40 @@ #include <rdma/hns-abi.h> #include "hns_roce_common.h" -static int hns_roce_alloc_cqc(struct hns_roce_dev *hr_dev, - struct hns_roce_cq *hr_cq) +static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) { struct hns_roce_cmd_mailbox *mailbox; - struct hns_roce_hem_table *mtt_table; struct hns_roce_cq_table *cq_table; - struct device *dev = hr_dev->dev; + struct ib_device *ibdev = &hr_dev->ib_dev; + u64 mtts[MTT_MIN_COUNT] = { 0 }; dma_addr_t dma_handle; - u64 *mtts; int ret; - cq_table = &hr_dev->cq_table; - - /* Get the physical address of cq buf */ - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - mtt_table = &hr_dev->mr_table.mtt_cqe_table; - else - mtt_table = &hr_dev->mr_table.mtt_table; - - mtts = hns_roce_table_find(hr_dev, mtt_table, hr_cq->mtt.first_seg, - &dma_handle); - - if (!mtts) { - dev_err(dev, "Failed to find mtt for CQ buf.\n"); + ret = hns_roce_mtr_find(hr_dev, &hr_cq->mtr, 0, mtts, ARRAY_SIZE(mtts), + &dma_handle); + if (ret < 1) { + ibdev_err(ibdev, "Failed to find CQ mtr\n"); return -EINVAL; } + cq_table = &hr_dev->cq_table; ret = hns_roce_bitmap_alloc(&cq_table->bitmap, &hr_cq->cqn); if (ret) { - dev_err(dev, "Num of CQ out of range.\n"); + ibdev_err(ibdev, "Failed to alloc CQ bitmap, err %d\n", ret); return ret; } /* Get CQC memory HEM(Hardware Entry Memory) table */ ret = hns_roce_table_get(hr_dev, &cq_table->table, hr_cq->cqn); if (ret) { - dev_err(dev, - "Get context mem failed(%d) when CQ(0x%lx) alloc.\n", - ret, hr_cq->cqn); + ibdev_err(ibdev, "Failed to get CQ(0x%lx) context, err %d\n", + hr_cq->cqn, ret); goto err_out; } ret = xa_err(xa_store(&cq_table->array, hr_cq->cqn, hr_cq, GFP_KERNEL)); if (ret) { - dev_err(dev, "Failed to xa_store CQ.\n"); + ibdev_err(ibdev, "Failed to xa_store CQ\n"); goto err_put; } @@ -101,9 +90,9 @@ static int hns_roce_alloc_cqc(struct hns_roce_dev *hr_dev, HNS_ROCE_CMD_CREATE_CQC, HNS_ROCE_CMD_TIMEOUT_MSECS); hns_roce_free_cmd_mailbox(hr_dev, mailbox); if (ret) { - dev_err(dev, - "Send cmd mailbox failed(%d) when CQ(0x%lx) alloc.\n", - ret, hr_cq->cqn); + ibdev_err(ibdev, + "Failed to send create cmd for CQ(0x%lx), err %d\n", + hr_cq->cqn, ret); goto err_xa; } @@ -126,7 +115,7 @@ err_out: return ret; } -void hns_roce_free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) +static void free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) { struct hns_roce_cq_table *cq_table = &hr_dev->cq_table; struct device *dev = hr_dev->dev; @@ -153,190 +142,86 @@ void hns_roce_free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) hns_roce_bitmap_free(&cq_table->bitmap, hr_cq->cqn, BITMAP_NO_RR); } -static int get_cq_umem(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, - struct hns_roce_ib_create_cq ucmd, - struct ib_udata *udata) -{ - struct hns_roce_buf *buf = &hr_cq->buf; - struct hns_roce_mtt *mtt = &hr_cq->mtt; - struct ib_umem **umem = &hr_cq->umem; - u32 npages; - int ret; - - *umem = ib_umem_get(&hr_dev->ib_dev, ucmd.buf_addr, buf->size, - IB_ACCESS_LOCAL_WRITE); - if (IS_ERR(*umem)) - return PTR_ERR(*umem); - - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - mtt->mtt_type = MTT_TYPE_CQE; - else - mtt->mtt_type = MTT_TYPE_WQE; - - npages = DIV_ROUND_UP(ib_umem_page_count(*umem), - 1 << hr_dev->caps.cqe_buf_pg_sz); - ret = hns_roce_mtt_init(hr_dev, npages, buf->page_shift, mtt); - if (ret) - goto err_buf; - - ret = hns_roce_ib_umem_write_mtt(hr_dev, mtt, *umem); - if (ret) - goto err_mtt; - - return 0; - -err_mtt: - hns_roce_mtt_cleanup(hr_dev, mtt); - -err_buf: - ib_umem_release(*umem); - return ret; -} - -static int alloc_cq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) +static int alloc_cq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, + struct ib_udata *udata, unsigned long addr) { - struct hns_roce_buf *buf = &hr_cq->buf; - struct hns_roce_mtt *mtt = &hr_cq->mtt; - int ret; - - ret = hns_roce_buf_alloc(hr_dev, buf->size, (1 << buf->page_shift) * 2, - buf, buf->page_shift); - if (ret) - goto out; - - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - mtt->mtt_type = MTT_TYPE_CQE; - else - mtt->mtt_type = MTT_TYPE_WQE; - - ret = hns_roce_mtt_init(hr_dev, buf->npages, buf->page_shift, mtt); - if (ret) - goto err_buf; - - ret = hns_roce_buf_write_mtt(hr_dev, mtt, buf); - if (ret) - goto err_mtt; - - return 0; - -err_mtt: - hns_roce_mtt_cleanup(hr_dev, mtt); - -err_buf: - hns_roce_buf_free(hr_dev, buf->size, buf); - -out: - return ret; + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_buf_attr buf_attr = {}; + int err; + + buf_attr.page_shift = hr_dev->caps.cqe_buf_pg_sz + HNS_HW_PAGE_SHIFT; + buf_attr.region[0].size = hr_cq->cq_depth * hr_dev->caps.cq_entry_sz; + buf_attr.region[0].hopnum = hr_dev->caps.cqe_hop_num; + buf_attr.region_count = 1; + buf_attr.fixed_page = true; + + err = hns_roce_mtr_create(hr_dev, &hr_cq->mtr, &buf_attr, + hr_dev->caps.cqe_ba_pg_sz + HNS_HW_PAGE_SHIFT, + udata, addr); + if (err) + ibdev_err(ibdev, "Failed to alloc CQ mtr, err %d\n", err); + + return err; } static void free_cq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) { - hns_roce_buf_free(hr_dev, hr_cq->buf.size, &hr_cq->buf); + hns_roce_mtr_destroy(hr_dev, &hr_cq->mtr); } -static int create_user_cq(struct hns_roce_dev *hr_dev, - struct hns_roce_cq *hr_cq, - struct ib_udata *udata, - struct hns_roce_ib_create_cq_resp *resp) +static int alloc_cq_db(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, + struct ib_udata *udata, unsigned long addr, + struct hns_roce_ib_create_cq_resp *resp) { - struct hns_roce_ib_create_cq ucmd; - struct device *dev = hr_dev->dev; - int ret; - struct hns_roce_ucontext *context = rdma_udata_to_drv_context( - udata, struct hns_roce_ucontext, ibucontext); - - if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) { - dev_err(dev, "Failed to copy_from_udata.\n"); - return -EFAULT; - } + bool has_db = hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB; + struct hns_roce_ucontext *uctx; + int err; - /* Get user space address, write it into mtt table */ - ret = get_cq_umem(hr_dev, hr_cq, ucmd, udata); - if (ret) { - dev_err(dev, "Failed to get_cq_umem.\n"); - return ret; - } - - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB && - udata->outlen >= offsetofend(typeof(*resp), cap_flags)) { - ret = hns_roce_db_map_user(context, udata, ucmd.db_addr, - &hr_cq->db); - if (ret) { - dev_err(dev, "cq record doorbell map failed!\n"); - goto err_mtt; + if (udata) { + if (has_db && + udata->outlen >= offsetofend(typeof(*resp), cap_flags)) { + uctx = rdma_udata_to_drv_context(udata, + struct hns_roce_ucontext, ibucontext); + err = hns_roce_db_map_user(uctx, udata, addr, + &hr_cq->db); + if (err) + return err; + hr_cq->flags |= HNS_ROCE_CQ_FLAG_RECORD_DB; + resp->cap_flags |= HNS_ROCE_CQ_FLAG_RECORD_DB; } - hr_cq->db_en = 1; - resp->cap_flags |= HNS_ROCE_SUPPORT_CQ_RECORD_DB; - } - - return 0; - -err_mtt: - hns_roce_mtt_cleanup(hr_dev, &hr_cq->mtt); - ib_umem_release(hr_cq->umem); - - return ret; -} - -static int create_kernel_cq(struct hns_roce_dev *hr_dev, - struct hns_roce_cq *hr_cq) -{ - struct device *dev = hr_dev->dev; - int ret; - - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) { - ret = hns_roce_alloc_db(hr_dev, &hr_cq->db, 1); - if (ret) - return ret; - - hr_cq->set_ci_db = hr_cq->db.db_record; - *hr_cq->set_ci_db = 0; - hr_cq->db_en = 1; - } - - /* Init mtt table and write buff address to mtt table */ - ret = alloc_cq_buf(hr_dev, hr_cq); - if (ret) { - dev_err(dev, "Failed to alloc_cq_buf.\n"); - goto err_db; + } else { + if (has_db) { + err = hns_roce_alloc_db(hr_dev, &hr_cq->db, 1); + if (err) + return err; + hr_cq->set_ci_db = hr_cq->db.db_record; + *hr_cq->set_ci_db = 0; + hr_cq->flags |= HNS_ROCE_CQ_FLAG_RECORD_DB; + } + hr_cq->cq_db_l = hr_dev->reg_base + hr_dev->odb_offset + + DB_REG_OFFSET * hr_dev->priv_uar.index; } - hr_cq->cq_db_l = hr_dev->reg_base + hr_dev->odb_offset + - DB_REG_OFFSET * hr_dev->priv_uar.index; - return 0; - -err_db: - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) - hns_roce_free_db(hr_dev, &hr_cq->db); - - return ret; } -static void destroy_user_cq(struct hns_roce_dev *hr_dev, - struct hns_roce_cq *hr_cq, - struct ib_udata *udata, - struct hns_roce_ib_create_cq_resp *resp) +static void free_cq_db(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, + struct ib_udata *udata) { - struct hns_roce_ucontext *context = rdma_udata_to_drv_context( - udata, struct hns_roce_ucontext, ibucontext); + struct hns_roce_ucontext *uctx; - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB && - udata->outlen >= offsetofend(typeof(*resp), cap_flags)) - hns_roce_db_unmap_user(context, &hr_cq->db); - - hns_roce_mtt_cleanup(hr_dev, &hr_cq->mtt); - ib_umem_release(hr_cq->umem); -} - -static void destroy_kernel_cq(struct hns_roce_dev *hr_dev, - struct hns_roce_cq *hr_cq) -{ - hns_roce_mtt_cleanup(hr_dev, &hr_cq->mtt); - free_cq_buf(hr_dev, hr_cq); + if (!(hr_cq->flags & HNS_ROCE_CQ_FLAG_RECORD_DB)) + return; - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) + hr_cq->flags &= ~HNS_ROCE_CQ_FLAG_RECORD_DB; + if (udata) { + uctx = rdma_udata_to_drv_context(udata, + struct hns_roce_ucontext, + ibucontext); + hns_roce_db_unmap_user(uctx, &hr_cq->db); + } else { hns_roce_free_db(hr_dev, &hr_cq->db); + } } int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, @@ -345,20 +230,21 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, struct hns_roce_dev *hr_dev = to_hr_dev(ib_cq->device); struct hns_roce_ib_create_cq_resp resp = {}; struct hns_roce_cq *hr_cq = to_hr_cq(ib_cq); - struct device *dev = hr_dev->dev; + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_ib_create_cq ucmd = {}; int vector = attr->comp_vector; u32 cq_entries = attr->cqe; int ret; if (cq_entries < 1 || cq_entries > hr_dev->caps.max_cqes) { - dev_err(dev, "Create CQ failed. entries=%d, max=%d\n", - cq_entries, hr_dev->caps.max_cqes); + ibdev_err(ibdev, "Failed to check CQ count %d max=%d\n", + cq_entries, hr_dev->caps.max_cqes); return -EINVAL; } if (vector >= hr_dev->caps.num_comp_vectors) { - dev_err(dev, "Create CQ failed, vector=%d, max=%d\n", - vector, hr_dev->caps.num_comp_vectors); + ibdev_err(ibdev, "Failed to check CQ vector=%d max=%d\n", + vector, hr_dev->caps.num_comp_vectors); return -EINVAL; } @@ -367,30 +253,35 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, hr_cq->ib_cq.cqe = cq_entries - 1; /* used as cqe index */ hr_cq->cq_depth = cq_entries; hr_cq->vector = vector; - hr_cq->buf.size = hr_cq->cq_depth * hr_dev->caps.cq_entry_sz; - hr_cq->buf.page_shift = PAGE_SHIFT + hr_dev->caps.cqe_buf_pg_sz; spin_lock_init(&hr_cq->lock); INIT_LIST_HEAD(&hr_cq->sq_list); INIT_LIST_HEAD(&hr_cq->rq_list); if (udata) { - ret = create_user_cq(hr_dev, hr_cq, udata, &resp); - if (ret) { - dev_err(dev, "Create cq failed in user mode!\n"); - goto err_cq; - } - } else { - ret = create_kernel_cq(hr_dev, hr_cq); + ret = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd)); if (ret) { - dev_err(dev, "Create cq failed in kernel mode!\n"); - goto err_cq; + ibdev_err(ibdev, "Failed to copy CQ udata, err %d\n", + ret); + return ret; } } - ret = hns_roce_alloc_cqc(hr_dev, hr_cq); + ret = alloc_cq_buf(hr_dev, hr_cq, udata, ucmd.buf_addr); + if (ret) { + ibdev_err(ibdev, "Failed to alloc CQ buf, err %d\n", ret); + return ret; + } + + ret = alloc_cq_db(hr_dev, hr_cq, udata, ucmd.db_addr, &resp); if (ret) { - dev_err(dev, "Alloc CQ failed(%d).\n", ret); - goto err_dbmap; + ibdev_err(ibdev, "Failed to alloc CQ db, err %d\n", ret); + goto err_cq_buf; + } + + ret = alloc_cqc(hr_dev, hr_cq); + if (ret) { + ibdev_err(ibdev, "Failed to alloc CQ context, err %d\n", ret); + goto err_cq_db; } /* @@ -412,15 +303,11 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, return 0; err_cqc: - hns_roce_free_cqc(hr_dev, hr_cq); - -err_dbmap: - if (udata) - destroy_user_cq(hr_dev, hr_cq, udata, &resp); - else - destroy_kernel_cq(hr_dev, hr_cq); - -err_cq: + free_cqc(hr_dev, hr_cq); +err_cq_db: + free_cq_db(hr_dev, hr_cq, udata); +err_cq_buf: + free_cq_buf(hr_dev, hr_cq); return ret; } @@ -429,28 +316,12 @@ void hns_roce_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata) struct hns_roce_dev *hr_dev = to_hr_dev(ib_cq->device); struct hns_roce_cq *hr_cq = to_hr_cq(ib_cq); - if (hr_dev->hw->destroy_cq) { + if (hr_dev->hw->destroy_cq) hr_dev->hw->destroy_cq(ib_cq, udata); - return; - } - - hns_roce_free_cqc(hr_dev, hr_cq); - hns_roce_mtt_cleanup(hr_dev, &hr_cq->mtt); - ib_umem_release(hr_cq->umem); - if (udata) { - if (hr_cq->db_en == 1) - hns_roce_db_unmap_user(rdma_udata_to_drv_context( - udata, - struct hns_roce_ucontext, - ibucontext), - &hr_cq->db); - } else { - /* Free the buff of stored cq */ - free_cq_buf(hr_dev, hr_cq); - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) - hns_roce_free_db(hr_dev, &hr_cq->db); - } + free_cq_buf(hr_dev, hr_cq); + free_cq_db(hr_dev, hr_cq, udata); + free_cqc(hr_dev, hr_cq); } void hns_roce_cq_completion(struct hns_roce_dev *hr_dev, u32 cqn) diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h index f6b3cf6b95d6..6edcbdcd8f43 100644 --- a/drivers/infiniband/hw/hns/hns_roce_device.h +++ b/drivers/infiniband/hw/hns/hns_roce_device.h @@ -37,9 +37,8 @@ #define DRV_NAME "hns_roce" -/* hip08 is a pci device, it includes two version according pci version id */ -#define PCI_REVISION_ID_HIP08_A 0x20 -#define PCI_REVISION_ID_HIP08_B 0x21 +/* hip08 is a pci device */ +#define PCI_REVISION_ID_HIP08 0x21 #define HNS_ROCE_HW_VER1 ('h' << 24 | 'i' << 16 | '0' << 8 | '6') @@ -131,12 +130,12 @@ enum { }; enum { - HNS_ROCE_SUPPORT_RQ_RECORD_DB = 1 << 0, - HNS_ROCE_SUPPORT_SQ_RECORD_DB = 1 << 1, + HNS_ROCE_QP_CAP_RQ_RECORD_DB = BIT(0), + HNS_ROCE_QP_CAP_SQ_RECORD_DB = BIT(1), }; -enum { - HNS_ROCE_SUPPORT_CQ_RECORD_DB = 1 << 0, +enum hns_roce_cq_flags { + HNS_ROCE_CQ_FLAG_RECORD_DB = BIT(0), }; enum hns_roce_qp_state { @@ -209,6 +208,8 @@ enum { HNS_ROCE_OPCODE_RDMA_WITH_IMM_RECEIVE = 0x07, }; +#define HNS_ROCE_CAP_FLAGS_EX_SHIFT 12 + enum { HNS_ROCE_CAP_FLAG_REREG_MR = BIT(0), HNS_ROCE_CAP_FLAG_ROCE_V1_V2 = BIT(1), @@ -222,13 +223,6 @@ enum { HNS_ROCE_CAP_FLAG_ATOMIC = BIT(10), }; -enum hns_roce_mtt_type { - MTT_TYPE_WQE, - MTT_TYPE_CQE, - MTT_TYPE_SRQWQE, - MTT_TYPE_IDX -}; - #define HNS_ROCE_DB_TYPE_COUNT 2 #define HNS_ROCE_DB_UNIT_SIZE 4 @@ -267,9 +261,12 @@ enum { #define HNS_ROCE_PORT_DOWN 0 #define HNS_ROCE_PORT_UP 1 -#define HNS_ROCE_MTT_ENTRY_PER_SEG 8 +/* The minimum page size is 4K for hardware */ +#define HNS_HW_PAGE_SHIFT 12 +#define HNS_HW_PAGE_SIZE (1 << HNS_HW_PAGE_SHIFT) -#define PAGE_ADDR_SHIFT 12 +/* The minimum page count for hardware access page directly. */ +#define HNS_HW_DIRECT_PAGE_COUNT 2 struct hns_roce_uar { u64 pfn; @@ -300,22 +297,6 @@ struct hns_roce_bitmap { unsigned long *table; }; -/* Order bitmap length -- bit num compute formula: 1 << (max_order - order) */ -/* Order = 0: bitmap is biggest, order = max bitmap is least (only a bit) */ -/* Every bit repesent to a partner free/used status in bitmap */ -/* - * Initial, bits of other bitmap are all 0 except that a bit of max_order is 1 - * Bit = 1 represent to idle and available; bit = 0: not available - */ -struct hns_roce_buddy { - /* Members point to every order level bitmap */ - unsigned long **bits; - /* Represent to avail bits of the order level bitmap */ - u32 *num_free; - int max_order; - spinlock_t lock; -}; - /* For Hardware Entry Memory */ struct hns_roce_hem_table { /* HEM type: 0 = qpc, 1 = mtt, 2 = cqc, 3 = srq, 4 = other */ @@ -336,13 +317,6 @@ struct hns_roce_hem_table { dma_addr_t *bt_l0_dma_addr; }; -struct hns_roce_mtt { - unsigned long first_seg; - int order; - int page_shift; - enum hns_roce_mtt_type mtt_type; -}; - struct hns_roce_buf_region { int offset; /* page offset */ u32 count; /* page count */ @@ -357,13 +331,36 @@ struct hns_roce_hem_list { struct list_head mid_bt[HNS_ROCE_MAX_BT_REGION][HNS_ROCE_MAX_BT_LEVEL]; struct list_head btm_bt; /* link all bottom bt in @mid_bt */ dma_addr_t root_ba; /* pointer to the root ba table */ - int bt_pg_shift; +}; + +struct hns_roce_buf_attr { + struct { + size_t size; /* region size */ + int hopnum; /* multi-hop addressing hop num */ + } region[HNS_ROCE_MAX_BT_REGION]; + int region_count; /* valid region count */ + unsigned int page_shift; /* buffer page shift */ + bool fixed_page; /* decide page shift is fixed-size or maximum size */ + int user_access; /* umem access flag */ + bool mtt_only; /* only alloc buffer-required MTT memory */ +}; + +struct hns_roce_hem_cfg { + dma_addr_t root_ba; /* root BA table's address */ + bool is_direct; /* addressing without BA table */ + unsigned int ba_pg_shift; /* BA table page shift */ + unsigned int buf_pg_shift; /* buffer page shift */ + unsigned int buf_pg_count; /* buffer page count */ + struct hns_roce_buf_region region[HNS_ROCE_MAX_BT_REGION]; + int region_count; }; /* memory translate region */ struct hns_roce_mtr { - struct hns_roce_hem_list hem_list; - int buf_pg_shift; + struct hns_roce_hem_list hem_list; /* multi-hop addressing resource */ + struct ib_umem *umem; /* user space buffer */ + struct hns_roce_buf *kmem; /* kernel space buffer */ + struct hns_roce_hem_cfg hem_cfg; /* config for hardware addressing */ }; struct hns_roce_mw { @@ -381,43 +378,22 @@ struct hns_roce_mw { struct hns_roce_mr { struct ib_mr ibmr; - struct ib_umem *umem; u64 iova; /* MR's virtual orignal addr */ u64 size; /* Address range of MR */ u32 key; /* Key of MR */ u32 pd; /* PD num of MR */ u32 access; /* Access permission of MR */ - u32 npages; int enabled; /* MR's active status */ int type; /* MR's register type */ - u64 *pbl_buf; /* MR's PBL space */ - dma_addr_t pbl_dma_addr; /* MR's PBL space PA */ - u32 pbl_size; /* PA number in the PBL */ - u64 pbl_ba; /* page table address */ - u32 l0_chunk_last_num; /* L0 last number */ - u32 l1_chunk_last_num; /* L1 last number */ - u64 **pbl_bt_l2; /* PBL BT L2 */ - u64 **pbl_bt_l1; /* PBL BT L1 */ - u64 *pbl_bt_l0; /* PBL BT L0 */ - dma_addr_t *pbl_l2_dma_addr; /* PBL BT L2 dma addr */ - dma_addr_t *pbl_l1_dma_addr; /* PBL BT L1 dma addr */ - dma_addr_t pbl_l0_dma_addr; /* PBL BT L0 dma addr */ - u32 pbl_ba_pg_sz; /* BT chunk page size */ - u32 pbl_buf_pg_sz; /* buf chunk page size */ u32 pbl_hop_num; /* multi-hop number */ + struct hns_roce_mtr pbl_mtr; + u32 npages; + dma_addr_t *page_list; }; struct hns_roce_mr_table { struct hns_roce_bitmap mtpt_bitmap; - struct hns_roce_buddy mtt_buddy; - struct hns_roce_hem_table mtt_table; struct hns_roce_hem_table mtpt_table; - struct hns_roce_buddy mtt_cqe_buddy; - struct hns_roce_hem_table mtt_cqe_table; - struct hns_roce_buddy mtt_srqwqe_buddy; - struct hns_roce_hem_table mtt_srqwqe_table; - struct hns_roce_buddy mtt_idx_buddy; - struct hns_roce_hem_table mtt_idx_table; }; struct hns_roce_wq { @@ -433,7 +409,7 @@ struct hns_roce_wq { }; struct hns_roce_sge { - int sge_cnt; /* SGE num */ + unsigned int sge_cnt; /* SGE num */ int offset; int sge_shift; /* SGE size */ }; @@ -446,10 +422,9 @@ struct hns_roce_buf_list { struct hns_roce_buf { struct hns_roce_buf_list direct; struct hns_roce_buf_list *page_list; - int nbufs; u32 npages; u32 size; - int page_shift; + unsigned int page_shift; }; struct hns_roce_db_pgdir { @@ -482,12 +457,10 @@ struct hns_roce_db { struct hns_roce_cq { struct ib_cq ib_cq; - struct hns_roce_buf buf; - struct hns_roce_mtt mtt; + struct hns_roce_mtr mtr; struct hns_roce_db db; - u8 db_en; + u32 flags; spinlock_t lock; - struct ib_umem *umem; u32 cq_depth; u32 cons_index; u32 *set_ci_db; @@ -505,11 +478,8 @@ struct hns_roce_cq { }; struct hns_roce_idx_que { - struct hns_roce_buf idx_buf; - int entry_sz; - u32 buf_size; - struct ib_umem *umem; - struct hns_roce_mtt mtt; + struct hns_roce_mtr mtr; + int entry_shift; unsigned long *bitmap; }; @@ -524,10 +494,9 @@ struct hns_roce_srq { atomic_t refcount; struct completion free; - struct hns_roce_buf buf; + struct hns_roce_mtr buf_mtr; + u64 *wrid; - struct ib_umem *umem; - struct hns_roce_mtt mtt; struct hns_roce_idx_que idx_que; spinlock_t lock; int head; @@ -656,20 +625,15 @@ struct hns_roce_work { struct hns_roce_qp { struct ib_qp ibqp; - struct hns_roce_buf hr_buf; struct hns_roce_wq rq; struct hns_roce_db rdb; struct hns_roce_db sdb; - u8 rdb_en; - u8 sdb_en; + unsigned long en_flags; u32 doorbell_qpn; u32 sq_signal_bits; struct hns_roce_wq sq; - struct ib_umem *umem; - struct hns_roce_mtt mtt; struct hns_roce_mtr mtr; - int wqe_bt_pg_shift; u32 buff_size; struct mutex mutex; @@ -769,17 +733,11 @@ struct hns_roce_eq { int over_ignore; int coalesce; int arm_st; - u64 eqe_ba; - int eqe_ba_pg_sz; - int eqe_buf_pg_sz; int hop_num; struct hns_roce_mtr mtr; - struct hns_roce_buf buf; - int eq_max_cnt; + u16 eq_max_cnt; int eq_period; int shift; - dma_addr_t cur_eqe_ba; - dma_addr_t nxt_eqe_ba; int event_type; int sub_type; }; @@ -939,13 +897,14 @@ struct hns_roce_hw { int (*set_mac)(struct hns_roce_dev *hr_dev, u8 phy_port, u8 *addr); void (*set_mtu)(struct hns_roce_dev *hr_dev, u8 phy_port, enum ib_mtu mtu); - int (*write_mtpt)(void *mb_buf, struct hns_roce_mr *mr, - unsigned long mtpt_idx); + int (*write_mtpt)(struct hns_roce_dev *hr_dev, void *mb_buf, + struct hns_roce_mr *mr, unsigned long mtpt_idx); int (*rereg_write_mtpt)(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, int flags, u32 pdn, int mr_access_flags, u64 iova, u64 size, void *mb_buf); - int (*frmr_write_mtpt)(void *mb_buf, struct hns_roce_mr *mr); + int (*frmr_write_mtpt)(struct hns_roce_dev *hr_dev, void *mb_buf, + struct hns_roce_mr *mr); int (*mw_write_mtpt)(void *mb_buf, struct hns_roce_mw *mw); void (*write_cqc)(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, void *mb_buf, u64 *mtts, @@ -1102,15 +1061,67 @@ static inline struct hns_roce_qp return xa_load(&hr_dev->qp_table_xa, qpn & (hr_dev->caps.num_qps - 1)); } +static inline bool hns_roce_buf_is_direct(struct hns_roce_buf *buf) +{ + if (buf->page_list) + return false; + + return true; +} + static inline void *hns_roce_buf_offset(struct hns_roce_buf *buf, int offset) { - u32 page_size = 1 << buf->page_shift; + if (hns_roce_buf_is_direct(buf)) + return (char *)(buf->direct.buf) + (offset & (buf->size - 1)); - if (buf->nbufs == 1) - return (char *)(buf->direct.buf) + offset; + return (char *)(buf->page_list[offset >> buf->page_shift].buf) + + (offset & ((1 << buf->page_shift) - 1)); +} + +static inline dma_addr_t hns_roce_buf_page(struct hns_roce_buf *buf, int idx) +{ + if (hns_roce_buf_is_direct(buf)) + return buf->direct.map + ((dma_addr_t)idx << buf->page_shift); else - return (char *)(buf->page_list[offset >> buf->page_shift].buf) + - (offset & (page_size - 1)); + return buf->page_list[idx].map; +} + +#define hr_hw_page_align(x) ALIGN(x, 1 << HNS_HW_PAGE_SHIFT) + +static inline u64 to_hr_hw_page_addr(u64 addr) +{ + return addr >> HNS_HW_PAGE_SHIFT; +} + +static inline u32 to_hr_hw_page_shift(u32 page_shift) +{ + return page_shift - HNS_HW_PAGE_SHIFT; +} + +static inline u32 to_hr_hem_hopnum(u32 hopnum, u32 count) +{ + if (count > 0) + return hopnum == HNS_ROCE_HOP_NUM_0 ? 0 : hopnum; + + return 0; +} + +static inline u32 to_hr_hem_entries_size(u32 count, u32 buf_shift) +{ + return hr_hw_page_align(count << buf_shift); +} + +static inline u32 to_hr_hem_entries_count(u32 count, u32 buf_shift) +{ + return hr_hw_page_align(count << buf_shift) >> buf_shift; +} + +static inline u32 to_hr_hem_entries_shift(u32 count, u32 buf_shift) +{ + if (!count) + return 0; + + return ilog2(to_hr_hem_entries_count(count, buf_shift)); } int hns_roce_init_uar_table(struct hns_roce_dev *dev); @@ -1125,25 +1136,18 @@ void hns_roce_cmd_event(struct hns_roce_dev *hr_dev, u16 token, u8 status, int hns_roce_cmd_use_events(struct hns_roce_dev *hr_dev); void hns_roce_cmd_use_polling(struct hns_roce_dev *hr_dev); -int hns_roce_mtt_init(struct hns_roce_dev *hr_dev, int npages, int page_shift, - struct hns_roce_mtt *mtt); -void hns_roce_mtt_cleanup(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt); -int hns_roce_buf_write_mtt(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, struct hns_roce_buf *buf); - -void hns_roce_mtr_init(struct hns_roce_mtr *mtr, int bt_pg_shift, - int buf_pg_shift); -int hns_roce_mtr_attach(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, - dma_addr_t **bufs, struct hns_roce_buf_region *regions, - int region_cnt); -void hns_roce_mtr_cleanup(struct hns_roce_dev *hr_dev, - struct hns_roce_mtr *mtr); - /* hns roce hw need current block and next block addr from mtt */ #define MTT_MIN_COUNT 2 int hns_roce_mtr_find(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, int offset, u64 *mtt_buf, int mtt_max, u64 *base_addr); +int hns_roce_mtr_create(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + struct hns_roce_buf_attr *buf_attr, + unsigned int page_shift, struct ib_udata *udata, + unsigned long user_addr); +void hns_roce_mtr_destroy(struct hns_roce_dev *hr_dev, + struct hns_roce_mtr *mtr); +int hns_roce_mtr_map(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + dma_addr_t *pages, int page_cnt); int hns_roce_init_pd_table(struct hns_roce_dev *hr_dev); int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev); @@ -1171,8 +1175,8 @@ void hns_roce_bitmap_free_range(struct hns_roce_bitmap *bitmap, unsigned long obj, int cnt, int rr); -int hns_roce_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata); +int hns_roce_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata); int hns_roce_query_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr); void hns_roce_destroy_ah(struct ib_ah *ah, u32 flags); @@ -1187,7 +1191,7 @@ int hns_roce_rereg_user_mr(struct ib_mr *mr, int flags, u64 start, u64 length, u64 virt_addr, int mr_access_flags, struct ib_pd *pd, struct ib_udata *udata); struct ib_mr *hns_roce_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int hns_roce_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); int hns_roce_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata); @@ -1200,25 +1204,15 @@ struct ib_mw *hns_roce_alloc_mw(struct ib_pd *pd, enum ib_mw_type, struct ib_udata *udata); int hns_roce_dealloc_mw(struct ib_mw *ibmw); -void hns_roce_buf_free(struct hns_roce_dev *hr_dev, u32 size, - struct hns_roce_buf *buf); +void hns_roce_buf_free(struct hns_roce_dev *hr_dev, struct hns_roce_buf *buf); int hns_roce_buf_alloc(struct hns_roce_dev *hr_dev, u32 size, u32 max_direct, struct hns_roce_buf *buf, u32 page_shift); -int hns_roce_ib_umem_write_mtt(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, struct ib_umem *umem); - -void hns_roce_init_buf_region(struct hns_roce_buf_region *region, int hopnum, - int offset, int buf_cnt); -int hns_roce_alloc_buf_list(struct hns_roce_buf_region *regions, - dma_addr_t **bufs, int count); -void hns_roce_free_buf_list(dma_addr_t **bufs, int count); - int hns_roce_get_kmem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs, int buf_cnt, int start, struct hns_roce_buf *buf); int hns_roce_get_umem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs, int buf_cnt, int start, struct ib_umem *umem, - int page_shift); + unsigned int page_shift); int hns_roce_create_srq(struct ib_srq *srq, struct ib_srq_init_attr *srq_init_attr, @@ -1254,8 +1248,6 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, struct ib_udata *udata); void hns_roce_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata); -void hns_roce_free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq); - int hns_roce_db_map_user(struct hns_roce_ucontext *context, struct ib_udata *udata, unsigned long virt, struct hns_roce_db *db); @@ -1274,6 +1266,6 @@ void hns_roce_handle_device_err(struct hns_roce_dev *hr_dev); int hns_roce_init(struct hns_roce_dev *hr_dev); void hns_roce_exit(struct hns_roce_dev *hr_dev); -int hns_roce_fill_res_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res); +int hns_roce_fill_res_cq_entry(struct sk_buff *msg, + struct ib_cq *ib_cq); #endif /* _HNS_ROCE_DEVICE_H */ diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c b/drivers/infiniband/hw/hns/hns_roce_hem.c index 263338b90d7a..c8db6f8ae018 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.c +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c @@ -75,18 +75,6 @@ bool hns_roce_check_whether_mhop(struct hns_roce_dev *hr_dev, u32 type) case HEM_TYPE_CQC_TIMER: hop_num = hr_dev->caps.cqc_timer_hop_num; break; - case HEM_TYPE_CQE: - hop_num = hr_dev->caps.cqe_hop_num; - break; - case HEM_TYPE_MTT: - hop_num = hr_dev->caps.mtt_hop_num; - break; - case HEM_TYPE_SRQWQE: - hop_num = hr_dev->caps.srqwqe_hop_num; - break; - case HEM_TYPE_IDX: - hop_num = hr_dev->caps.idx_hop_num; - break; default: return false; } @@ -195,38 +183,6 @@ static int get_hem_table_config(struct hns_roce_dev *hr_dev, mhop->ba_l0_num = hr_dev->caps.srqc_bt_num; mhop->hop_num = hr_dev->caps.srqc_hop_num; break; - case HEM_TYPE_MTT: - mhop->buf_chunk_size = 1 << (hr_dev->caps.mtt_buf_pg_sz - + PAGE_SHIFT); - mhop->bt_chunk_size = 1 << (hr_dev->caps.mtt_ba_pg_sz - + PAGE_SHIFT); - mhop->ba_l0_num = mhop->bt_chunk_size / BA_BYTE_LEN; - mhop->hop_num = hr_dev->caps.mtt_hop_num; - break; - case HEM_TYPE_CQE: - mhop->buf_chunk_size = 1 << (hr_dev->caps.cqe_buf_pg_sz - + PAGE_SHIFT); - mhop->bt_chunk_size = 1 << (hr_dev->caps.cqe_ba_pg_sz - + PAGE_SHIFT); - mhop->ba_l0_num = mhop->bt_chunk_size / BA_BYTE_LEN; - mhop->hop_num = hr_dev->caps.cqe_hop_num; - break; - case HEM_TYPE_SRQWQE: - mhop->buf_chunk_size = 1 << (hr_dev->caps.srqwqe_buf_pg_sz - + PAGE_SHIFT); - mhop->bt_chunk_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz - + PAGE_SHIFT); - mhop->ba_l0_num = mhop->bt_chunk_size / BA_BYTE_LEN; - mhop->hop_num = hr_dev->caps.srqwqe_hop_num; - break; - case HEM_TYPE_IDX: - mhop->buf_chunk_size = 1 << (hr_dev->caps.idx_buf_pg_sz - + PAGE_SHIFT); - mhop->bt_chunk_size = 1 << (hr_dev->caps.idx_ba_pg_sz - + PAGE_SHIFT); - mhop->ba_l0_num = mhop->bt_chunk_size / BA_BYTE_LEN; - mhop->hop_num = hr_dev->caps.idx_hop_num; - break; default: dev_err(dev, "Table %d not support multi-hop addressing!\n", type); @@ -899,57 +855,6 @@ out: return addr; } -int hns_roce_table_get_range(struct hns_roce_dev *hr_dev, - struct hns_roce_hem_table *table, - unsigned long start, unsigned long end) -{ - struct hns_roce_hem_mhop mhop; - unsigned long inc = table->table_chunk_size / table->obj_size; - unsigned long i = 0; - int ret; - - if (hns_roce_check_whether_mhop(hr_dev, table->type)) { - ret = hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop); - if (ret) - goto fail; - inc = mhop.bt_chunk_size / table->obj_size; - } - - /* Allocate MTT entry memory according to chunk(128K) */ - for (i = start; i <= end; i += inc) { - ret = hns_roce_table_get(hr_dev, table, i); - if (ret) - goto fail; - } - - return 0; - -fail: - while (i > start) { - i -= inc; - hns_roce_table_put(hr_dev, table, i); - } - return ret; -} - -void hns_roce_table_put_range(struct hns_roce_dev *hr_dev, - struct hns_roce_hem_table *table, - unsigned long start, unsigned long end) -{ - struct hns_roce_hem_mhop mhop; - unsigned long inc = table->table_chunk_size / table->obj_size; - unsigned long i; - - if (hns_roce_check_whether_mhop(hr_dev, table->type)) { - if (hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop)) - return; - inc = mhop.bt_chunk_size / table->obj_size; - } - - for (i = start; i <= end; i += inc) - hns_roce_table_put(hr_dev, table, i); -} - int hns_roce_init_hem_table(struct hns_roce_dev *hr_dev, struct hns_roce_hem_table *table, u32 type, unsigned long obj_size, unsigned long nobj, @@ -1112,12 +1017,6 @@ void hns_roce_cleanup_hem_table(struct hns_roce_dev *hr_dev, void hns_roce_cleanup_hem(struct hns_roce_dev *hr_dev) { - if ((hr_dev->caps.num_idx_segs)) - hns_roce_cleanup_hem_table(hr_dev, - &hr_dev->mr_table.mtt_idx_table); - if (hr_dev->caps.num_srqwqe_segs) - hns_roce_cleanup_hem_table(hr_dev, - &hr_dev->mr_table.mtt_srqwqe_table); if (hr_dev->caps.srqc_entry_sz) hns_roce_cleanup_hem_table(hr_dev, &hr_dev->srq_table.table); @@ -1137,10 +1036,6 @@ void hns_roce_cleanup_hem(struct hns_roce_dev *hr_dev) hns_roce_cleanup_hem_table(hr_dev, &hr_dev->qp_table.irrl_table); hns_roce_cleanup_hem_table(hr_dev, &hr_dev->qp_table.qp_table); hns_roce_cleanup_hem_table(hr_dev, &hr_dev->mr_table.mtpt_table); - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - hns_roce_cleanup_hem_table(hr_dev, - &hr_dev->mr_table.mtt_cqe_table); - hns_roce_cleanup_hem_table(hr_dev, &hr_dev->mr_table.mtt_table); } struct roce_hem_item { @@ -1505,7 +1400,7 @@ err_exit: int hns_roce_hem_list_request(struct hns_roce_dev *hr_dev, struct hns_roce_hem_list *hem_list, const struct hns_roce_buf_region *regions, - int region_cnt) + int region_cnt, unsigned int bt_pg_shift) { const struct hns_roce_buf_region *r; int ofs, end; @@ -1519,7 +1414,7 @@ int hns_roce_hem_list_request(struct hns_roce_dev *hr_dev, return -EINVAL; } - unit = (1 << hem_list->bt_pg_shift) / BA_BYTE_LEN; + unit = (1 << bt_pg_shift) / BA_BYTE_LEN; for (i = 0; i < region_cnt; i++) { r = ®ions[i]; if (!r->count) @@ -1566,8 +1461,7 @@ void hns_roce_hem_list_release(struct hns_roce_dev *hr_dev, hem_list->root_ba = 0; } -void hns_roce_hem_list_init(struct hns_roce_hem_list *hem_list, - int bt_page_order) +void hns_roce_hem_list_init(struct hns_roce_hem_list *hem_list) { int i, j; @@ -1576,8 +1470,6 @@ void hns_roce_hem_list_init(struct hns_roce_hem_list *hem_list, for (i = 0; i < HNS_ROCE_MAX_BT_REGION; i++) for (j = 0; j < HNS_ROCE_MAX_BT_LEVEL; j++) INIT_LIST_HEAD(&hem_list->mid_bt[i][j]); - - hem_list->bt_pg_shift = bt_page_order; } void *hns_roce_hem_list_find_mtt(struct hns_roce_dev *hr_dev, diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h b/drivers/infiniband/hw/hns/hns_roce_hem.h index 3bb8f78fb7b0..b34c940077bb 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.h +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h @@ -115,12 +115,6 @@ void hns_roce_table_put(struct hns_roce_dev *hr_dev, void *hns_roce_table_find(struct hns_roce_dev *hr_dev, struct hns_roce_hem_table *table, unsigned long obj, dma_addr_t *dma_handle); -int hns_roce_table_get_range(struct hns_roce_dev *hr_dev, - struct hns_roce_hem_table *table, - unsigned long start, unsigned long end); -void hns_roce_table_put_range(struct hns_roce_dev *hr_dev, - struct hns_roce_hem_table *table, - unsigned long start, unsigned long end); int hns_roce_init_hem_table(struct hns_roce_dev *hr_dev, struct hns_roce_hem_table *table, u32 type, unsigned long obj_size, unsigned long nobj, @@ -133,14 +127,13 @@ int hns_roce_calc_hem_mhop(struct hns_roce_dev *hr_dev, struct hns_roce_hem_mhop *mhop); bool hns_roce_check_whether_mhop(struct hns_roce_dev *hr_dev, u32 type); -void hns_roce_hem_list_init(struct hns_roce_hem_list *hem_list, - int bt_page_order); +void hns_roce_hem_list_init(struct hns_roce_hem_list *hem_list); int hns_roce_hem_list_calc_root_ba(const struct hns_roce_buf_region *regions, int region_cnt, int unit); int hns_roce_hem_list_request(struct hns_roce_dev *hr_dev, struct hns_roce_hem_list *hem_list, const struct hns_roce_buf_region *regions, - int region_cnt); + int region_cnt, unsigned int bt_pg_shift); void hns_roce_hem_list_release(struct hns_roce_dev *hr_dev, struct hns_roce_hem_list *hem_list); void *hns_roce_hem_list_find_mtt(struct hns_roce_dev *hr_dev, diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c index 5ff028d77be3..aeb3a6fa7d47 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c @@ -503,16 +503,13 @@ static void hns_roce_set_odb(struct hns_roce_dev *hr_dev, u32 odb_alept, static void hns_roce_set_sdb_ext(struct hns_roce_dev *hr_dev, u32 ext_sdb_alept, u32 ext_sdb_alful) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_db_table *db = &priv->db_table; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - struct hns_roce_db_table *db; dma_addr_t sdb_dma_addr; __le32 tmp; u32 val; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - db = &priv->db_table; - /* Configure extend SDB threshold */ roce_write(hr_dev, ROCEE_EXT_DB_SQ_WL_EMPTY_REG, ext_sdb_alept); roce_write(hr_dev, ROCEE_EXT_DB_SQ_WL_REG, ext_sdb_alful); @@ -538,23 +535,20 @@ static void hns_roce_set_sdb_ext(struct hns_roce_dev *hr_dev, u32 ext_sdb_alept, roce_write(hr_dev, ROCEE_EXT_DB_SQ_H_REG, val); dev_dbg(dev, "ext SDB depth: 0x%x\n", db->ext_db->esdb_dep); - dev_dbg(dev, "ext SDB threshold: epmty: 0x%x, ful: 0x%x\n", + dev_dbg(dev, "ext SDB threshold: empty: 0x%x, ful: 0x%x\n", ext_sdb_alept, ext_sdb_alful); } static void hns_roce_set_odb_ext(struct hns_roce_dev *hr_dev, u32 ext_odb_alept, u32 ext_odb_alful) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_db_table *db = &priv->db_table; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - struct hns_roce_db_table *db; dma_addr_t odb_dma_addr; __le32 tmp; u32 val; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - db = &priv->db_table; - /* Configure extend ODB threshold */ roce_write(hr_dev, ROCEE_EXT_DB_OTHERS_WL_EMPTY_REG, ext_odb_alept); roce_write(hr_dev, ROCEE_EXT_DB_OTHERS_WL_REG, ext_odb_alful); @@ -583,16 +577,13 @@ static void hns_roce_set_odb_ext(struct hns_roce_dev *hr_dev, u32 ext_odb_alept, static int hns_roce_db_ext_init(struct hns_roce_dev *hr_dev, u32 sdb_ext_mod, u32 odb_ext_mod) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_db_table *db = &priv->db_table; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - struct hns_roce_db_table *db; dma_addr_t sdb_dma_addr; dma_addr_t odb_dma_addr; int ret = 0; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - db = &priv->db_table; - db->ext_db = kmalloc(sizeof(*db->ext_db), GFP_KERNEL); if (!db->ext_db) return -ENOMEM; @@ -692,14 +683,14 @@ static struct hns_roce_qp *hns_roce_v1_create_lp_qp(struct hns_roce_dev *hr_dev, static int hns_roce_v1_rsv_lp_qp(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; struct hns_roce_caps *caps = &hr_dev->caps; + struct ib_device *ibdev = &hr_dev->ib_dev; struct device *dev = &hr_dev->pdev->dev; struct ib_cq_init_attr cq_init_attr; - struct hns_roce_free_mr *free_mr; struct ib_qp_attr attr = { 0 }; - struct hns_roce_v1_priv *priv; struct hns_roce_qp *hr_qp; - struct ib_device *ibdev; struct ib_cq *cq; struct ib_pd *pd; union ib_gid dgid; @@ -712,14 +703,10 @@ static int hns_roce_v1_rsv_lp_qp(struct hns_roce_dev *hr_dev) u8 port = 0; u8 sl; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; - /* Reserved cq for loop qp */ cq_init_attr.cqe = HNS_ROCE_MIN_WQE_NUM * 2; cq_init_attr.comp_vector = 0; - ibdev = &hr_dev->ib_dev; cq = rdma_zalloc_drv_obj(ibdev, ib_cq); if (!cq) return -ENOMEM; @@ -868,16 +855,13 @@ alloc_cq_failed: static void hns_roce_v1_release_lp_qp(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_free_mr *free_mr; - struct hns_roce_v1_priv *priv; struct hns_roce_qp *hr_qp; int ret; int i; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; - for (i = 0; i < HNS_ROCE_V1_RESV_QP; i++) { hr_qp = free_mr->mr_free_qp[i]; if (!hr_qp) @@ -897,18 +881,15 @@ static void hns_roce_v1_release_lp_qp(struct hns_roce_dev *hr_dev) static int hns_roce_db_init(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_db_table *db = &priv->db_table; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - struct hns_roce_db_table *db; u32 sdb_ext_mod; u32 odb_ext_mod; u32 sdb_evt_mod; u32 odb_evt_mod; int ret = 0; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - db = &priv->db_table; - memset(db, 0, sizeof(*db)); /* Default DB mode */ @@ -954,15 +935,12 @@ static void hns_roce_v1_recreate_lp_qp_work_fn(struct work_struct *work) static int hns_roce_v1_recreate_lp_qp(struct hns_roce_dev *hr_dev) { - struct device *dev = &hr_dev->pdev->dev; + long end = HNS_ROCE_V1_RECREATE_LP_QP_TIMEOUT_MSECS; + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; struct hns_roce_recreate_lp_qp_work *lp_qp_work; - struct hns_roce_free_mr *free_mr; - struct hns_roce_v1_priv *priv; + struct device *dev = &hr_dev->pdev->dev; struct completion comp; - long end = HNS_ROCE_V1_RECREATE_LP_QP_TIMEOUT_MSECS; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; lp_qp_work = kzalloc(sizeof(struct hns_roce_recreate_lp_qp_work), GFP_KERNEL); @@ -1021,29 +999,21 @@ static int hns_roce_v1_send_lp_wqe(struct hns_roce_qp *hr_qp) static void hns_roce_v1_mr_free_work_fn(struct work_struct *work) { - struct hns_roce_mr_free_work *mr_work; - struct ib_wc wc[HNS_ROCE_V1_RESV_QP]; - struct hns_roce_free_mr *free_mr; - struct hns_roce_cq *mr_free_cq; - struct hns_roce_v1_priv *priv; - struct hns_roce_dev *hr_dev; - struct hns_roce_mr *hr_mr; - struct hns_roce_qp *hr_qp; - struct device *dev; unsigned long end = msecs_to_jiffies(HNS_ROCE_V1_FREE_MR_TIMEOUT_MSECS) + jiffies; - int i; - int ret; + struct hns_roce_mr_free_work *mr_work = + container_of(work, struct hns_roce_mr_free_work, work); + struct hns_roce_dev *hr_dev = to_hr_dev(mr_work->ib_dev); + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; + struct hns_roce_cq *mr_free_cq = free_mr->mr_free_cq; + struct hns_roce_mr *hr_mr = mr_work->mr; + struct device *dev = &hr_dev->pdev->dev; + struct ib_wc wc[HNS_ROCE_V1_RESV_QP]; + struct hns_roce_qp *hr_qp; int ne = 0; - - mr_work = container_of(work, struct hns_roce_mr_free_work, work); - hr_mr = (struct hns_roce_mr *)mr_work->mr; - hr_dev = to_hr_dev(mr_work->ib_dev); - dev = &hr_dev->pdev->dev; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; - mr_free_cq = free_mr->mr_free_cq; + int ret; + int i; for (i = 0; i < HNS_ROCE_V1_RESV_QP; i++) { hr_qp = free_mr->mr_free_qp[i]; @@ -1092,19 +1062,15 @@ free_work: static int hns_roce_v1_dereg_mr(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, struct ib_udata *udata) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; + long end = HNS_ROCE_V1_FREE_MR_TIMEOUT_MSECS; struct device *dev = &hr_dev->pdev->dev; struct hns_roce_mr_free_work *mr_work; - struct hns_roce_free_mr *free_mr; - struct hns_roce_v1_priv *priv; - struct completion comp; - long end = HNS_ROCE_V1_FREE_MR_TIMEOUT_MSECS; unsigned long start = jiffies; - int npages; + struct completion comp; int ret = 0; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; - if (mr->enabled) { if (hns_roce_hw_destroy_mpt(hr_dev, NULL, key_to_hw_index(mr->key) & @@ -1146,17 +1112,9 @@ free_mr: dev_dbg(dev, "Free mr 0x%x use 0x%x us.\n", mr->key, jiffies_to_usecs(jiffies) - jiffies_to_usecs(start)); - if (mr->size != ~0ULL) { - npages = ib_umem_page_count(mr->umem); - dma_free_coherent(dev, npages * 8, mr->pbl_buf, - mr->pbl_dma_addr); - } - hns_roce_bitmap_free(&hr_dev->mr_table.mtpt_bitmap, key_to_hw_index(mr->key), 0); - - ib_umem_release(mr->umem); - + hns_roce_mtr_destroy(hr_dev, &mr->pbl_mtr); kfree(mr); return ret; @@ -1164,12 +1122,9 @@ free_mr: static void hns_roce_db_free(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_db_table *db = &priv->db_table; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - struct hns_roce_db_table *db; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - db = &priv->db_table; if (db->sdb_ext_mod) { dma_free_coherent(dev, HNS_ROCE_V1_EXT_SDB_SIZE, @@ -1190,17 +1145,14 @@ static void hns_roce_db_free(struct hns_roce_dev *hr_dev) static int hns_roce_raq_init(struct hns_roce_dev *hr_dev) { - int ret; - u32 val; - __le32 tmp; + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_raq_table *raq = &priv->raq_table; + struct device *dev = &hr_dev->pdev->dev; int raq_shift = 0; dma_addr_t addr; - struct hns_roce_v1_priv *priv; - struct hns_roce_raq_table *raq; - struct device *dev = &hr_dev->pdev->dev; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - raq = &priv->raq_table; + __le32 tmp; + u32 val; + int ret; raq->e_raq_buf = kzalloc(sizeof(*(raq->e_raq_buf)), GFP_KERNEL); if (!raq->e_raq_buf) @@ -1280,12 +1232,9 @@ err_dma_alloc_raq: static void hns_roce_raq_free(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_raq_table *raq = &priv->raq_table; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - struct hns_roce_raq_table *raq; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - raq = &priv->raq_table; dma_free_coherent(dev, HNS_ROCE_V1_RAQ_SIZE, raq->e_raq_buf->buf, raq->e_raq_buf->map); @@ -1319,12 +1268,10 @@ static void hns_roce_port_enable(struct hns_roce_dev *hr_dev, int enable_flag) static int hns_roce_bt_init(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; int ret; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - priv->bt_table.qpc_buf.buf = dma_alloc_coherent(dev, HNS_ROCE_BT_RSV_BUF_SIZE, &priv->bt_table.qpc_buf.map, GFP_KERNEL); @@ -1362,10 +1309,8 @@ err_failed_alloc_mtpt_buf: static void hns_roce_bt_free(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; dma_free_coherent(dev, HNS_ROCE_BT_RSV_BUF_SIZE, priv->bt_table.cqc_buf.buf, priv->bt_table.cqc_buf.map); @@ -1379,12 +1324,9 @@ static void hns_roce_bt_free(struct hns_roce_dev *hr_dev) static int hns_roce_tptr_init(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_buf_list *tptr_buf = &priv->tptr_table.tptr_buf; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_buf_list *tptr_buf; - struct hns_roce_v1_priv *priv; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - tptr_buf = &priv->tptr_table.tptr_buf; /* * This buffer will be used for CQ's tptr(tail pointer), also @@ -1405,12 +1347,9 @@ static int hns_roce_tptr_init(struct hns_roce_dev *hr_dev) static void hns_roce_tptr_free(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_buf_list *tptr_buf = &priv->tptr_table.tptr_buf; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_buf_list *tptr_buf; - struct hns_roce_v1_priv *priv; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - tptr_buf = &priv->tptr_table.tptr_buf; dma_free_coherent(dev, HNS_ROCE_V1_TPTR_BUF_SIZE, tptr_buf->buf, tptr_buf->map); @@ -1418,14 +1357,11 @@ static void hns_roce_tptr_free(struct hns_roce_dev *hr_dev) static int hns_roce_free_mr_init(struct hns_roce_dev *hr_dev) { + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_free_mr *free_mr; - struct hns_roce_v1_priv *priv; int ret = 0; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; - free_mr->free_mr_wq = create_singlethread_workqueue("hns_roce_free_mr"); if (!free_mr->free_mr_wq) { dev_err(dev, "Create free mr workqueue failed!\n"); @@ -1444,11 +1380,8 @@ static int hns_roce_free_mr_init(struct hns_roce_dev *hr_dev) static void hns_roce_free_mr_free(struct hns_roce_dev *hr_dev) { - struct hns_roce_free_mr *free_mr; - struct hns_roce_v1_priv *priv; - - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - free_mr = &priv->free_mr; + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_free_mr *free_mr = &priv->free_mr; flush_workqueue(free_mr->free_mr_wq); destroy_workqueue(free_mr->free_mr_wq); @@ -1823,12 +1756,15 @@ static void hns_roce_v1_set_mtu(struct hns_roce_dev *hr_dev, u8 phy_port, val); } -static int hns_roce_v1_write_mtpt(void *mb_buf, struct hns_roce_mr *mr, +static int hns_roce_v1_write_mtpt(struct hns_roce_dev *hr_dev, void *mb_buf, + struct hns_roce_mr *mr, unsigned long mtpt_idx) { + u64 pages[HNS_ROCE_MAX_INNER_MTPT_NUM] = { 0 }; + struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_v1_mpt_entry *mpt_entry; - struct sg_dma_page_iter sg_iter; - u64 *pages; + dma_addr_t pbl_ba; + int count; int i; /* MPT filled into mailbox buf */ @@ -1878,22 +1814,15 @@ static int hns_roce_v1_write_mtpt(void *mb_buf, struct hns_roce_mr *mr, if (mr->type == MR_TYPE_DMA) return 0; - pages = (u64 *) __get_free_page(GFP_KERNEL); - if (!pages) - return -ENOMEM; - - i = 0; - for_each_sg_dma_page(mr->umem->sg_head.sgl, &sg_iter, mr->umem->nmap, 0) { - pages[i] = ((u64)sg_page_iter_dma_address(&sg_iter)) >> 12; - - /* Directly record to MTPT table firstly 7 entry */ - if (i >= HNS_ROCE_MAX_INNER_MTPT_NUM) - break; - i++; + count = hns_roce_mtr_find(hr_dev, &mr->pbl_mtr, 0, pages, + ARRAY_SIZE(pages), &pbl_ba); + if (count < 1) { + ibdev_err(ibdev, "failed to find PBL mtr, count = %d.", count); + return -ENOBUFS; } /* Register user mr */ - for (i = 0; i < HNS_ROCE_MAX_INNER_MTPT_NUM; i++) { + for (i = 0; i < count; i++) { switch (i) { case 0: mpt_entry->pa0_l = cpu_to_le32((u32)(pages[i])); @@ -1959,20 +1888,17 @@ static int hns_roce_v1_write_mtpt(void *mb_buf, struct hns_roce_mr *mr, } } - free_page((unsigned long) pages); - - mpt_entry->pbl_addr_l = cpu_to_le32((u32)(mr->pbl_dma_addr)); - + mpt_entry->pbl_addr_l = cpu_to_le32(pbl_ba); roce_set_field(mpt_entry->mpt_byte_12, MPT_BYTE_12_PBL_ADDR_H_M, - MPT_BYTE_12_PBL_ADDR_H_S, - ((u32)(mr->pbl_dma_addr >> 32))); + MPT_BYTE_12_PBL_ADDR_H_S, upper_32_bits(pbl_ba)); return 0; } static void *get_cqe(struct hns_roce_cq *hr_cq, int n) { - return hns_roce_buf_offset(&hr_cq->buf, n * HNS_ROCE_V1_CQE_ENTRY_SIZE); + return hns_roce_buf_offset(hr_cq->mtr.kmem, + n * HNS_ROCE_V1_CQE_ENTRY_SIZE); } static void *get_sw_cqe(struct hns_roce_cq *hr_cq, int n) @@ -2066,16 +1992,12 @@ static void hns_roce_v1_write_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, void *mb_buf, u64 *mtts, dma_addr_t dma_handle) { - struct hns_roce_cq_context *cq_context = NULL; - struct hns_roce_buf_list *tptr_buf; - struct hns_roce_v1_priv *priv; + struct hns_roce_v1_priv *priv = hr_dev->priv; + struct hns_roce_buf_list *tptr_buf = &priv->tptr_table.tptr_buf; + struct hns_roce_cq_context *cq_context = mb_buf; dma_addr_t tptr_dma_addr; int offset; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - tptr_buf = &priv->tptr_table.tptr_buf; - - cq_context = mb_buf; memset(cq_context, 0, sizeof(*cq_context)); /* Get the tptr for this CQ. */ @@ -2416,16 +2338,14 @@ static int hns_roce_v1_clear_hem(struct hns_roce_dev *hr_dev, struct hns_roce_hem_table *table, int obj, int step_idx) { + struct hns_roce_v1_priv *priv = hr_dev->priv; struct device *dev = &hr_dev->pdev->dev; - struct hns_roce_v1_priv *priv; - unsigned long flags = 0; long end = HW_SYNC_TIMEOUT_MSECS; __le32 bt_cmd_val[2] = {0}; + unsigned long flags = 0; void __iomem *bt_cmd; u64 bt_ba = 0; - priv = (struct hns_roce_v1_priv *)hr_dev->priv; - switch (table->type) { case HEM_TYPE_QPC: bt_ba = priv->bt_table.qpc_buf.map >> 12; @@ -2479,7 +2399,6 @@ static int hns_roce_v1_clear_hem(struct hns_roce_dev *hr_dev, } static int hns_roce_v1_qp_modify(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, enum hns_roce_qp_state cur_state, enum hns_roce_qp_state new_state, struct hns_roce_qp_context *context, @@ -2560,6 +2479,28 @@ static int hns_roce_v1_qp_modify(struct hns_roce_dev *hr_dev, return ret; } +static int find_wqe_mtt(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, + u64 *sq_ba, u64 *rq_ba, dma_addr_t *bt_ba) +{ + struct ib_device *ibdev = &hr_dev->ib_dev; + int count; + + count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, 0, sq_ba, 1, bt_ba); + if (count < 1) { + ibdev_err(ibdev, "Failed to find SQ ba\n"); + return -ENOBUFS; + } + + count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, hr_qp->rq.offset, rq_ba, + 1, NULL); + if (!count) { + ibdev_err(ibdev, "Failed to find RQ ba\n"); + return -ENOBUFS; + } + + return 0; +} + static int hns_roce_v1_m_sqp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, enum ib_qp_state cur_state, enum ib_qp_state new_state) @@ -2567,25 +2508,20 @@ static int hns_roce_v1_m_sqp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); struct hns_roce_sqp_context *context; - struct device *dev = &hr_dev->pdev->dev; dma_addr_t dma_handle = 0; u32 __iomem *addr; - int rq_pa_start; + u64 sq_ba = 0; + u64 rq_ba = 0; __le32 tmp; u32 reg_val; - u64 *mtts; context = kzalloc(sizeof(*context), GFP_KERNEL); if (!context) return -ENOMEM; /* Search QP buf's MTTs */ - mtts = hns_roce_table_find(hr_dev, &hr_dev->mr_table.mtt_table, - hr_qp->mtt.first_seg, &dma_handle); - if (!mtts) { - dev_err(dev, "qp buf pa find failed\n"); + if (find_wqe_mtt(hr_dev, hr_qp, &sq_ba, &rq_ba, &dma_handle)) goto out; - } if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { roce_set_field(context->qp1c_bytes_4, @@ -2599,11 +2535,11 @@ static int hns_roce_v1_m_sqp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, roce_set_field(context->qp1c_bytes_4, QP1C_BYTES_4_PD_M, QP1C_BYTES_4_PD_S, to_hr_pd(ibqp->pd)->pdn); - context->sq_rq_bt_l = cpu_to_le32((u32)(dma_handle)); + context->sq_rq_bt_l = cpu_to_le32(dma_handle); roce_set_field(context->qp1c_bytes_12, QP1C_BYTES_12_SQ_RQ_BT_H_M, QP1C_BYTES_12_SQ_RQ_BT_H_S, - ((u32)(dma_handle >> 32))); + upper_32_bits(dma_handle)); roce_set_field(context->qp1c_bytes_16, QP1C_BYTES_16_RQ_HEAD_M, QP1C_BYTES_16_RQ_HEAD_S, hr_qp->rq.head); @@ -2624,14 +2560,12 @@ static int hns_roce_v1_m_sqp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, roce_set_field(context->qp1c_bytes_20, QP1C_BYTES_20_PKEY_IDX_M, QP1C_BYTES_20_PKEY_IDX_S, attr->pkey_index); - rq_pa_start = (u32)hr_qp->rq.offset / PAGE_SIZE; - context->cur_rq_wqe_ba_l = - cpu_to_le32((u32)(mtts[rq_pa_start])); + context->cur_rq_wqe_ba_l = cpu_to_le32(rq_ba); roce_set_field(context->qp1c_bytes_28, QP1C_BYTES_28_CUR_RQ_WQE_BA_H_M, QP1C_BYTES_28_CUR_RQ_WQE_BA_H_S, - (mtts[rq_pa_start]) >> 32); + upper_32_bits(rq_ba)); roce_set_field(context->qp1c_bytes_28, QP1C_BYTES_28_RQ_CUR_IDX_M, QP1C_BYTES_28_RQ_CUR_IDX_S, 0); @@ -2645,12 +2579,12 @@ static int hns_roce_v1_m_sqp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, QP1C_BYTES_32_TX_CQ_NUM_S, to_hr_cq(ibqp->send_cq)->cqn); - context->cur_sq_wqe_ba_l = cpu_to_le32((u32)mtts[0]); + context->cur_sq_wqe_ba_l = cpu_to_le32(sq_ba); roce_set_field(context->qp1c_bytes_40, QP1C_BYTES_40_CUR_SQ_WQE_BA_H_M, QP1C_BYTES_40_CUR_SQ_WQE_BA_H_S, - (mtts[0]) >> 32); + upper_32_bits(sq_ba)); roce_set_field(context->qp1c_bytes_40, QP1C_BYTES_40_SQ_CUR_IDX_M, QP1C_BYTES_40_SQ_CUR_IDX_S, 0); @@ -2704,6 +2638,28 @@ out: return -EINVAL; } +static bool check_qp_state(enum ib_qp_state cur_state, + enum ib_qp_state new_state) +{ + static const bool sm[][IB_QPS_ERR + 1] = { + [IB_QPS_RESET] = { [IB_QPS_RESET] = true, + [IB_QPS_INIT] = true }, + [IB_QPS_INIT] = { [IB_QPS_RESET] = true, + [IB_QPS_INIT] = true, + [IB_QPS_RTR] = true, + [IB_QPS_ERR] = true }, + [IB_QPS_RTR] = { [IB_QPS_RESET] = true, + [IB_QPS_RTS] = true, + [IB_QPS_ERR] = true }, + [IB_QPS_RTS] = { [IB_QPS_RESET] = true, [IB_QPS_ERR] = true }, + [IB_QPS_SQD] = {}, + [IB_QPS_SQE] = {}, + [IB_QPS_ERR] = { [IB_QPS_RESET] = true, [IB_QPS_ERR] = true } + }; + + return sm[cur_state][new_state]; +} + static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, enum ib_qp_state cur_state, enum ib_qp_state new_state) @@ -2716,26 +2672,29 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, dma_addr_t dma_handle_2 = 0; dma_addr_t dma_handle = 0; __le32 doorbell[2] = {0}; - int rq_pa_start = 0; u64 *mtts_2 = NULL; int ret = -EINVAL; - u64 *mtts = NULL; + u64 sq_ba = 0; + u64 rq_ba = 0; int port; u8 port_num; u8 *dmac; u8 *smac; + if (!check_qp_state(cur_state, new_state)) { + ibdev_err(ibqp->device, + "not support QP(%u) status from %d to %d\n", + ibqp->qp_num, cur_state, new_state); + return -EINVAL; + } + context = kzalloc(sizeof(*context), GFP_KERNEL); if (!context) return -ENOMEM; /* Search qp buf's mtts */ - mtts = hns_roce_table_find(hr_dev, &hr_dev->mr_table.mtt_table, - hr_qp->mtt.first_seg, &dma_handle); - if (mtts == NULL) { - dev_err(dev, "qp buf pa find failed\n"); + if (find_wqe_mtt(hr_dev, hr_qp, &sq_ba, &rq_ba, &dma_handle)) goto out; - } /* Search IRRL's mtts */ mtts_2 = hns_roce_table_find(hr_dev, &hr_dev->qp_table.irrl_table, @@ -2890,11 +2849,11 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, dmac = (u8 *)attr->ah_attr.roce.dmac; - context->sq_rq_bt_l = cpu_to_le32((u32)(dma_handle)); + context->sq_rq_bt_l = cpu_to_le32(dma_handle); roce_set_field(context->qpc_bytes_24, QP_CONTEXT_QPC_BYTES_24_SQ_RQ_BT_H_M, QP_CONTEXT_QPC_BYTES_24_SQ_RQ_BT_H_S, - ((u32)(dma_handle >> 32))); + upper_32_bits(dma_handle)); roce_set_bit(context->qpc_bytes_24, QP_CONTEXT_QPC_BYTE_24_REMOTE_ENABLE_E2E_CREDITS_S, 1); @@ -2993,14 +2952,12 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, QP_CONTEXT_QPC_BYTES_68_RQ_CUR_INDEX_M, QP_CONTEXT_QPC_BYTES_68_RQ_CUR_INDEX_S, 0); - rq_pa_start = (u32)hr_qp->rq.offset / PAGE_SIZE; - context->cur_rq_wqe_ba_l = - cpu_to_le32((u32)(mtts[rq_pa_start])); + context->cur_rq_wqe_ba_l = cpu_to_le32(rq_ba); roce_set_field(context->qpc_bytes_76, QP_CONTEXT_QPC_BYTES_76_CUR_RQ_WQE_BA_H_M, QP_CONTEXT_QPC_BYTES_76_CUR_RQ_WQE_BA_H_S, - mtts[rq_pa_start] >> 32); + upper_32_bits(rq_ba)); roce_set_field(context->qpc_bytes_76, QP_CONTEXT_QPC_BYTES_76_RX_REQ_MSN_M, QP_CONTEXT_QPC_BYTES_76_RX_REQ_MSN_S, 0); @@ -3062,8 +3019,7 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, QP_CONTEXT_QPC_BYTES_156_SL_S, rdma_ah_get_sl(&attr->ah_attr)); hr_qp->sl = rdma_ah_get_sl(&attr->ah_attr); - } else if (cur_state == IB_QPS_RTR && - new_state == IB_QPS_RTS) { + } else if (cur_state == IB_QPS_RTR && new_state == IB_QPS_RTS) { /* If exist optional param, return error */ if ((attr_mask & IB_QP_ALT_PATH) || (attr_mask & IB_QP_ACCESS_FLAGS) || @@ -3075,12 +3031,12 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, goto out; } - context->rx_cur_sq_wqe_ba_l = cpu_to_le32((u32)(mtts[0])); + context->rx_cur_sq_wqe_ba_l = cpu_to_le32(sq_ba); roce_set_field(context->qpc_bytes_120, QP_CONTEXT_QPC_BYTES_120_RX_CUR_SQ_WQE_BA_H_M, QP_CONTEXT_QPC_BYTES_120_RX_CUR_SQ_WQE_BA_H_S, - (mtts[0]) >> 32); + upper_32_bits(sq_ba)); roce_set_field(context->qpc_bytes_124, QP_CONTEXT_QPC_BYTES_124_RX_ACK_MSN_M, @@ -3223,28 +3179,18 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, QP_CONTEXT_QPC_BYTES_180_SQ_HEAD_M, QP_CONTEXT_QPC_BYTES_180_SQ_HEAD_S, 0); - context->tx_cur_sq_wqe_ba_l = cpu_to_le32((u32)(mtts[0])); + context->tx_cur_sq_wqe_ba_l = cpu_to_le32(sq_ba); roce_set_field(context->qpc_bytes_188, QP_CONTEXT_QPC_BYTES_188_TX_CUR_SQ_WQE_BA_H_M, QP_CONTEXT_QPC_BYTES_188_TX_CUR_SQ_WQE_BA_H_S, - (mtts[0]) >> 32); + upper_32_bits(sq_ba)); roce_set_bit(context->qpc_bytes_188, QP_CONTEXT_QPC_BYTES_188_PKT_RETRY_FLG_S, 0); roce_set_field(context->qpc_bytes_188, QP_CONTEXT_QPC_BYTES_188_TX_RETRY_CUR_INDEX_M, QP_CONTEXT_QPC_BYTES_188_TX_RETRY_CUR_INDEX_S, 0); - } else if (!((cur_state == IB_QPS_INIT && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_INIT && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_RTR && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_RTR && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_RTS && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_RTS && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_ERR && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_ERR && new_state == IB_QPS_ERR))) { - dev_err(dev, "not support this status migration\n"); - goto out; } /* Every status migrate must change state */ @@ -3253,8 +3199,7 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, QP_CONTEXT_QPC_BYTES_144_QP_STATE_S, new_state); /* SW pass context to HW */ - ret = hns_roce_v1_qp_modify(hr_dev, &hr_qp->mtt, - to_hns_roce_state(cur_state), + ret = hns_roce_v1_qp_modify(hr_dev, to_hns_roce_state(cur_state), to_hns_roce_state(new_state), context, hr_qp); if (ret) { @@ -3636,8 +3581,6 @@ static void hns_roce_v1_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata) u32 cqe_cnt_cur; int wait_time = 0; - hns_roce_free_cqc(hr_dev, hr_cq); - /* * Before freeing cq buffer, we need to ensure that the outstanding CQE * have been written by checking the CQE counter. @@ -3660,14 +3603,6 @@ static void hns_roce_v1_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata) } wait_time++; } - - hns_roce_mtt_cleanup(hr_dev, &hr_cq->mtt); - - ib_umem_release(hr_cq->umem); - if (!udata) { - /* Free the buff of stored cq */ - hns_roce_buf_free(hr_dev, hr_cq->buf.size, &hr_cq->buf); - } } static void set_eq_cons_index_v1(struct hns_roce_eq *eq, int req_not) diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c index c3316672b70e..4cda95ed1fbe 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -91,10 +91,12 @@ static u32 to_hr_opcode(u32 ib_opcode) } static void set_frmr_seg(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, - void *wqe, const struct ib_reg_wr *wr) + const struct ib_reg_wr *wr) { + struct hns_roce_wqe_frmr_seg *fseg = + (void *)rc_sq_wqe + sizeof(struct hns_roce_v2_rc_send_wqe); struct hns_roce_mr *mr = to_hr_mr(wr->mr); - struct hns_roce_wqe_frmr_seg *fseg = wqe; + u64 pbl_ba; /* use ib_access_flags */ roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_BIND_EN_S, @@ -109,31 +111,34 @@ static void set_frmr_seg(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, wr->access & IB_ACCESS_LOCAL_WRITE ? 1 : 0); /* Data structure reuse may lead to confusion */ - rc_sq_wqe->msg_len = cpu_to_le32(mr->pbl_ba & 0xffffffff); - rc_sq_wqe->inv_key = cpu_to_le32(mr->pbl_ba >> 32); + pbl_ba = mr->pbl_mtr.hem_cfg.root_ba; + rc_sq_wqe->msg_len = cpu_to_le32(lower_32_bits(pbl_ba)); + rc_sq_wqe->inv_key = cpu_to_le32(upper_32_bits(pbl_ba)); rc_sq_wqe->byte_16 = cpu_to_le32(wr->mr->length & 0xffffffff); rc_sq_wqe->byte_20 = cpu_to_le32(wr->mr->length >> 32); rc_sq_wqe->rkey = cpu_to_le32(wr->key); rc_sq_wqe->va = cpu_to_le64(wr->mr->iova); - fseg->pbl_size = cpu_to_le32(mr->pbl_size); + fseg->pbl_size = cpu_to_le32(mr->npages); roce_set_field(fseg->mode_buf_pg_sz, V2_RC_FRMR_WQE_BYTE_40_PBL_BUF_PG_SZ_M, V2_RC_FRMR_WQE_BYTE_40_PBL_BUF_PG_SZ_S, - mr->pbl_buf_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.buf_pg_shift)); roce_set_bit(fseg->mode_buf_pg_sz, V2_RC_FRMR_WQE_BYTE_40_BLK_MODE_S, 0); } -static void set_atomic_seg(const struct ib_send_wr *wr, void *wqe, +static void set_atomic_seg(const struct ib_send_wr *wr, struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, - int valid_num_sge) + unsigned int valid_num_sge) { - struct hns_roce_wqe_atomic_seg *aseg; + struct hns_roce_v2_wqe_data_seg *dseg = + (void *)rc_sq_wqe + sizeof(struct hns_roce_v2_rc_send_wqe); + struct hns_roce_wqe_atomic_seg *aseg = + (void *)dseg + sizeof(struct hns_roce_v2_wqe_data_seg); - set_data_seg_v2(wqe, wr->sg_list); - aseg = wqe + sizeof(struct hns_roce_v2_wqe_data_seg); + set_data_seg_v2(dseg, wr->sg_list); if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP) { aseg->fetchadd_swap_data = cpu_to_le64(atomic_wr(wr)->swap); @@ -141,7 +146,7 @@ static void set_atomic_seg(const struct ib_send_wr *wr, void *wqe, } else { aseg->fetchadd_swap_data = cpu_to_le64(atomic_wr(wr)->compare_add); - aseg->cmp_data = 0; + aseg->cmp_data = 0; } roce_set_field(rc_sq_wqe->byte_16, V2_RC_SEND_WQE_BYTE_16_SGE_NUM_M, @@ -149,74 +154,53 @@ static void set_atomic_seg(const struct ib_send_wr *wr, void *wqe, } static void set_extend_sge(struct hns_roce_qp *qp, const struct ib_send_wr *wr, - unsigned int *sge_ind, int valid_num_sge) + unsigned int *sge_ind, unsigned int valid_num_sge) { struct hns_roce_v2_wqe_data_seg *dseg; - struct ib_sge *sg; - int num_in_wqe = 0; - int extend_sge_num; - int fi_sge_num; - int se_sge_num; - int shift; - int i; + unsigned int cnt = valid_num_sge; + struct ib_sge *sge = wr->sg_list; + unsigned int idx = *sge_ind; - if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC) - num_in_wqe = HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE; - extend_sge_num = valid_num_sge - num_in_wqe; - sg = wr->sg_list + num_in_wqe; - shift = qp->hr_buf.page_shift; + if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC) { + cnt -= HNS_ROCE_SGE_IN_WQE; + sge += HNS_ROCE_SGE_IN_WQE; + } - /* - * Check whether wr->num_sge sges are in the same page. If not, we - * should calculate how many sges in the first page and the second - * page. - */ - dseg = hns_roce_get_extend_sge(qp, (*sge_ind) & (qp->sge.sge_cnt - 1)); - fi_sge_num = (round_up((uintptr_t)dseg, 1 << shift) - - (uintptr_t)dseg) / - sizeof(struct hns_roce_v2_wqe_data_seg); - if (extend_sge_num > fi_sge_num) { - se_sge_num = extend_sge_num - fi_sge_num; - for (i = 0; i < fi_sge_num; i++) { - set_data_seg_v2(dseg++, sg + i); - (*sge_ind)++; - } - dseg = hns_roce_get_extend_sge(qp, - (*sge_ind) & (qp->sge.sge_cnt - 1)); - for (i = 0; i < se_sge_num; i++) { - set_data_seg_v2(dseg++, sg + fi_sge_num + i); - (*sge_ind)++; - } - } else { - for (i = 0; i < extend_sge_num; i++) { - set_data_seg_v2(dseg++, sg + i); - (*sge_ind)++; - } + while (cnt > 0) { + dseg = hns_roce_get_extend_sge(qp, idx & (qp->sge.sge_cnt - 1)); + set_data_seg_v2(dseg, sge); + idx++; + sge++; + cnt--; } + + *sge_ind = idx; } static int set_rwqe_data_seg(struct ib_qp *ibqp, const struct ib_send_wr *wr, struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, - void *wqe, unsigned int *sge_ind, - int valid_num_sge) + unsigned int *sge_ind, + unsigned int valid_num_sge) { struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); - struct hns_roce_v2_wqe_data_seg *dseg = wqe; + struct hns_roce_v2_wqe_data_seg *dseg = + (void *)rc_sq_wqe + sizeof(struct hns_roce_v2_rc_send_wqe); struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_qp *qp = to_hr_qp(ibqp); + void *wqe = dseg; int j = 0; int i; if (wr->send_flags & IB_SEND_INLINE && valid_num_sge) { - if (le32_to_cpu(rc_sq_wqe->msg_len) > - hr_dev->caps.max_sq_inline) { + if (unlikely(le32_to_cpu(rc_sq_wqe->msg_len) > + hr_dev->caps.max_sq_inline)) { ibdev_err(ibdev, "inline len(1-%d)=%d, illegal", rc_sq_wqe->msg_len, hr_dev->caps.max_sq_inline); return -EINVAL; } - if (wr->opcode == IB_WR_RDMA_READ) { + if (unlikely(wr->opcode == IB_WR_RDMA_READ)) { ibdev_err(ibdev, "Not support inline data!\n"); return -EINVAL; } @@ -230,7 +214,7 @@ static int set_rwqe_data_seg(struct ib_qp *ibqp, const struct ib_send_wr *wr, roce_set_bit(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_INLINE_S, 1); } else { - if (valid_num_sge <= HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) { + if (valid_num_sge <= HNS_ROCE_SGE_IN_WQE) { for (i = 0; i < wr->num_sge; i++) { if (likely(wr->sg_list[i].length)) { set_data_seg_v2(dseg, wr->sg_list + i); @@ -243,8 +227,8 @@ static int set_rwqe_data_seg(struct ib_qp *ibqp, const struct ib_send_wr *wr, V2_RC_SEND_WQE_BYTE_20_MSG_START_SGE_IDX_S, (*sge_ind) & (qp->sge.sge_cnt - 1)); - for (i = 0; i < wr->num_sge && - j < HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE; i++) { + for (i = 0; i < wr->num_sge && j < HNS_ROCE_SGE_IN_WQE; + i++) { if (likely(wr->sg_list[i].length)) { set_data_seg_v2(dseg, wr->sg_list + i); dseg++; @@ -290,10 +274,11 @@ static int check_send_valid(struct hns_roce_dev *hr_dev, return 0; } -static inline int calc_wr_sge_num(const struct ib_send_wr *wr, u32 *sge_len) +static unsigned int calc_wr_sge_num(const struct ib_send_wr *wr, + unsigned int *sge_len) { - int valid_num = 0; - u32 len = 0; + unsigned int valid_num = 0; + unsigned int len = 0; int i; for (i = 0; i < wr->num_sge; i++) { @@ -424,7 +409,7 @@ static inline int set_rc_wqe(struct hns_roce_qp *qp, { struct hns_roce_v2_rc_send_wqe *rc_sq_wqe = wqe; unsigned int curr_idx = *sge_idx; - int valid_num_sge; + unsigned int valid_num_sge; u32 msg_len = 0; int ret = 0; @@ -458,7 +443,6 @@ static inline int set_rc_wqe(struct hns_roce_qp *qp, roce_set_bit(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_OWNER_S, owner_bit); - wqe += sizeof(struct hns_roce_v2_rc_send_wqe); switch (wr->opcode) { case IB_WR_RDMA_READ: case IB_WR_RDMA_WRITE: @@ -471,7 +455,7 @@ static inline int set_rc_wqe(struct hns_roce_qp *qp, rc_sq_wqe->inv_key = cpu_to_le32(wr->ex.invalidate_rkey); break; case IB_WR_REG_MR: - set_frmr_seg(rc_sq_wqe, wqe, reg_wr(wr)); + set_frmr_seg(rc_sq_wqe, reg_wr(wr)); break; case IB_WR_ATOMIC_CMP_AND_SWP: case IB_WR_ATOMIC_FETCH_AND_ADD: @@ -488,10 +472,10 @@ static inline int set_rc_wqe(struct hns_roce_qp *qp, if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP || wr->opcode == IB_WR_ATOMIC_FETCH_AND_ADD) - set_atomic_seg(wr, wqe, rc_sq_wqe, valid_num_sge); + set_atomic_seg(wr, rc_sq_wqe, valid_num_sge); else if (wr->opcode != IB_WR_REG_MR) ret = set_rwqe_data_seg(&qp->ibqp, wr, rc_sq_wqe, - wqe, &curr_idx, valid_num_sge); + &curr_idx, valid_num_sge); *sge_idx = curr_idx; @@ -521,8 +505,7 @@ static inline void update_sq_db(struct hns_roce_dev *hr_dev, roce_set_field(sq_db.byte_4, V2_DB_BYTE_4_CMD_M, V2_DB_BYTE_4_CMD_S, HNS_ROCE_V2_SQ_DB); roce_set_field(sq_db.parameter, V2_DB_PARAMETER_IDX_M, - V2_DB_PARAMETER_IDX_S, - qp->sq.head & ((qp->sq.wqe_cnt << 1) - 1)); + V2_DB_PARAMETER_IDX_S, qp->sq.head); roce_set_field(sq_db.parameter, V2_DB_PARAMETER_SL_M, V2_DB_PARAMETER_SL_S, qp->sl); @@ -548,7 +531,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp, spin_lock_irqsave(&qp->sq.lock, flags); ret = check_send_valid(hr_dev, qp); - if (ret) { + if (unlikely(ret)) { *bad_wr = wr; nreq = 0; goto out; @@ -584,7 +567,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp, else if (ibqp->qp_type == IB_QPT_RC) ret = set_rc_wqe(qp, wr, wqe, &sge_idx, owner_bit); - if (ret) { + if (unlikely(ret)) { *bad_wr = wr; goto out; } @@ -634,15 +617,15 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, spin_lock_irqsave(&hr_qp->rq.lock, flags); ret = check_recv_valid(hr_dev, hr_qp); - if (ret) { + if (unlikely(ret)) { *bad_wr = wr; nreq = 0; goto out; } for (nreq = 0; wr; ++nreq, wr = wr->next) { - if (hns_roce_wq_overflow(&hr_qp->rq, nreq, - hr_qp->ibqp.recv_cq)) { + if (unlikely(hns_roce_wq_overflow(&hr_qp->rq, nreq, + hr_qp->ibqp.recv_cq))) { ret = -ENOMEM; *bad_wr = wr; goto out; @@ -667,13 +650,13 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, dseg++; } - if (i < hr_qp->rq.max_gs) { + if (wr->num_sge < hr_qp->rq.max_gs) { dseg->lkey = cpu_to_le32(HNS_ROCE_INVALID_LKEY); dseg->addr = 0; } /* rq support inline data */ - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) { + if (hr_qp->rq_inl_buf.wqe_cnt) { sge_list = hr_qp->rq_inl_buf.wqe_list[wqe_idx].sg_list; hr_qp->rq_inl_buf.wqe_list[wqe_idx].sge_cnt = (u32)wr->num_sge; @@ -715,6 +698,129 @@ out: return ret; } +static void *get_srq_wqe(struct hns_roce_srq *srq, int n) +{ + return hns_roce_buf_offset(srq->buf_mtr.kmem, n << srq->wqe_shift); +} + +static void *get_idx_buf(struct hns_roce_idx_que *idx_que, int n) +{ + return hns_roce_buf_offset(idx_que->mtr.kmem, + n << idx_que->entry_shift); +} + +static void hns_roce_free_srq_wqe(struct hns_roce_srq *srq, int wqe_index) +{ + /* always called with interrupts disabled. */ + spin_lock(&srq->lock); + + bitmap_clear(srq->idx_que.bitmap, wqe_index, 1); + srq->tail++; + + spin_unlock(&srq->lock); +} + +static int find_empty_entry(struct hns_roce_idx_que *idx_que, + unsigned long size) +{ + int wqe_idx; + + if (unlikely(bitmap_full(idx_que->bitmap, size))) + return -ENOSPC; + + wqe_idx = find_first_zero_bit(idx_que->bitmap, size); + + bitmap_set(idx_que->bitmap, wqe_idx, 1); + + return wqe_idx; +} + +static int hns_roce_v2_post_srq_recv(struct ib_srq *ibsrq, + const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); + struct hns_roce_srq *srq = to_hr_srq(ibsrq); + struct hns_roce_v2_wqe_data_seg *dseg; + struct hns_roce_v2_db srq_db; + unsigned long flags; + __le32 *srq_idx; + int ret = 0; + int wqe_idx; + void *wqe; + int nreq; + int ind; + int i; + + spin_lock_irqsave(&srq->lock, flags); + + ind = srq->head & (srq->wqe_cnt - 1); + + for (nreq = 0; wr; ++nreq, wr = wr->next) { + if (unlikely(wr->num_sge >= srq->max_gs)) { + ret = -EINVAL; + *bad_wr = wr; + break; + } + + if (unlikely(srq->head == srq->tail)) { + ret = -ENOMEM; + *bad_wr = wr; + break; + } + + wqe_idx = find_empty_entry(&srq->idx_que, srq->wqe_cnt); + if (unlikely(wqe_idx < 0)) { + ret = -ENOMEM; + *bad_wr = wr; + break; + } + + wqe = get_srq_wqe(srq, wqe_idx); + dseg = (struct hns_roce_v2_wqe_data_seg *)wqe; + + for (i = 0; i < wr->num_sge; ++i) { + dseg[i].len = cpu_to_le32(wr->sg_list[i].length); + dseg[i].lkey = cpu_to_le32(wr->sg_list[i].lkey); + dseg[i].addr = cpu_to_le64(wr->sg_list[i].addr); + } + + if (wr->num_sge < srq->max_gs) { + dseg[i].len = 0; + dseg[i].lkey = cpu_to_le32(0x100); + dseg[i].addr = 0; + } + + srq_idx = get_idx_buf(&srq->idx_que, ind); + *srq_idx = cpu_to_le32(wqe_idx); + + srq->wrid[wqe_idx] = wr->wr_id; + ind = (ind + 1) & (srq->wqe_cnt - 1); + } + + if (likely(nreq)) { + srq->head += nreq; + + /* + * Make sure that descriptors are written before + * doorbell record. + */ + wmb(); + + srq_db.byte_4 = + cpu_to_le32(HNS_ROCE_V2_SRQ_DB << V2_DB_BYTE_4_CMD_S | + (srq->srqn & V2_DB_BYTE_4_TAG_M)); + srq_db.parameter = + cpu_to_le32(srq->head & V2_DB_PARAMETER_IDX_M); + + hns_roce_write64(hr_dev, (__le32 *)&srq_db, srq->db_reg_l); + } + + spin_unlock_irqrestore(&srq->lock, flags); + + return ret; +} + static int hns_roce_v2_cmd_hw_reseted(struct hns_roce_dev *hr_dev, unsigned long instance_stage, unsigned long reset_stage) @@ -742,7 +848,7 @@ static int hns_roce_v2_cmd_hw_resetting(struct hns_roce_dev *hr_dev, unsigned long instance_stage, unsigned long reset_stage) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hnae3_handle *handle = priv->handle; const struct hnae3_ae_ops *ops = handle->ae_algo->ops; @@ -768,7 +874,7 @@ static int hns_roce_v2_cmd_hw_resetting(struct hns_roce_dev *hr_dev, static int hns_roce_v2_cmd_sw_resetting(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hnae3_handle *handle = priv->handle; const struct hnae3_ae_ops *ops = handle->ae_algo->ops; @@ -785,7 +891,7 @@ static int hns_roce_v2_cmd_sw_resetting(struct hns_roce_dev *hr_dev) static int hns_roce_v2_rst_process_cmd(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hnae3_handle *handle = priv->handle; const struct hnae3_ae_ops *ops = handle->ae_algo->ops; unsigned long instance_stage; /* the current instance stage */ @@ -807,7 +913,7 @@ static int hns_roce_v2_rst_process_cmd(struct hns_roce_dev *hr_dev) instance_stage = handle->rinfo.instance_state; reset_stage = handle->rinfo.reset_state; reset_cnt = ops->ae_dev_reset_cnt(handle); - hw_resetting = ops->get_hw_reset_stat(handle); + hw_resetting = ops->get_cmdq_stat(handle); sw_resetting = ops->ae_dev_resetting(handle); if (reset_cnt != hr_dev->reset_cnt) @@ -865,7 +971,7 @@ static void hns_roce_free_cmq_desc(struct hns_roce_dev *hr_dev, static int hns_roce_init_cmq_ring(struct hns_roce_dev *hr_dev, bool ring_type) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hns_roce_v2_cmq_ring *ring = (ring_type == TYPE_CSQ) ? &priv->cmq.csq : &priv->cmq.crq; @@ -878,7 +984,7 @@ static int hns_roce_init_cmq_ring(struct hns_roce_dev *hr_dev, bool ring_type) static void hns_roce_cmq_init_regs(struct hns_roce_dev *hr_dev, bool ring_type) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hns_roce_v2_cmq_ring *ring = (ring_type == TYPE_CSQ) ? &priv->cmq.csq : &priv->cmq.crq; dma_addr_t dma = ring->desc_dma_addr; @@ -904,7 +1010,7 @@ static void hns_roce_cmq_init_regs(struct hns_roce_dev *hr_dev, bool ring_type) static int hns_roce_v2_cmq_init(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; int ret; /* Setup the queue entries for command queue */ @@ -948,7 +1054,7 @@ err_crq: static void hns_roce_v2_cmq_exit(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; hns_roce_free_cmq_desc(hr_dev, &priv->cmq.csq); hns_roce_free_cmq_desc(hr_dev, &priv->cmq.crq); @@ -970,15 +1076,15 @@ static void hns_roce_cmq_setup_basic_desc(struct hns_roce_cmq_desc *desc, static int hns_roce_cmq_csq_done(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; u32 head = roce_read(hr_dev, ROCEE_TX_CMQ_HEAD_REG); + struct hns_roce_v2_priv *priv = hr_dev->priv; return head == priv->cmq.csq.next_to_use; } static int hns_roce_cmq_csq_clean(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hns_roce_v2_cmq_ring *csq = &priv->cmq.csq; struct hns_roce_cmq_desc *desc; u16 ntc = csq->next_to_clean; @@ -1003,7 +1109,7 @@ static int hns_roce_cmq_csq_clean(struct hns_roce_dev *hr_dev) static int __hns_roce_cmq_send(struct hns_roce_dev *hr_dev, struct hns_roce_cmq_desc *desc, int num) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hns_roce_v2_cmq_ring *csq = &priv->cmq.csq; struct hns_roce_cmq_desc *desc_to_use; bool complete = false; @@ -1131,7 +1237,7 @@ static int hns_roce_cmq_query_hw_info(struct hns_roce_dev *hr_dev) static bool hns_roce_func_clr_chk_rst(struct hns_roce_dev *hr_dev) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hnae3_handle *handle = priv->handle; const struct hnae3_ae_ops *ops = handle->ae_algo->ops; unsigned long reset_cnt; @@ -1151,7 +1257,7 @@ static bool hns_roce_func_clr_chk_rst(struct hns_roce_dev *hr_dev) static void hns_roce_func_clr_rst_prc(struct hns_roce_dev *hr_dev, int retval, int flag) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hnae3_handle *handle = priv->handle; const struct hnae3_ae_ops *ops = handle->ae_algo->ops; unsigned long instance_stage; @@ -1349,34 +1455,26 @@ static int hns_roce_query_pf_resource(struct hns_roce_dev *hr_dev) static int hns_roce_query_pf_timer_resource(struct hns_roce_dev *hr_dev) { struct hns_roce_pf_timer_res_a *req_a; - struct hns_roce_cmq_desc desc[2]; - int ret, i; - - for (i = 0; i < 2; i++) { - hns_roce_cmq_setup_basic_desc(&desc[i], - HNS_ROCE_OPC_QUERY_PF_TIMER_RES, - true); + struct hns_roce_cmq_desc desc; + int ret; - if (i == 0) - desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - else - desc[i].flag &= ~cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - } + hns_roce_cmq_setup_basic_desc(&desc, HNS_ROCE_OPC_QUERY_PF_TIMER_RES, + true); - ret = hns_roce_cmq_send(hr_dev, desc, 2); + ret = hns_roce_cmq_send(hr_dev, &desc, 1); if (ret) return ret; - req_a = (struct hns_roce_pf_timer_res_a *)desc[0].data; + req_a = (struct hns_roce_pf_timer_res_a *)desc.data; hr_dev->caps.qpc_timer_bt_num = - roce_get_field(req_a->qpc_timer_bt_idx_num, - PF_RES_DATA_1_PF_QPC_TIMER_BT_NUM_M, - PF_RES_DATA_1_PF_QPC_TIMER_BT_NUM_S); + roce_get_field(req_a->qpc_timer_bt_idx_num, + PF_RES_DATA_1_PF_QPC_TIMER_BT_NUM_M, + PF_RES_DATA_1_PF_QPC_TIMER_BT_NUM_S); hr_dev->caps.cqc_timer_bt_num = - roce_get_field(req_a->cqc_timer_bt_idx_num, - PF_RES_DATA_2_PF_CQC_TIMER_BT_NUM_M, - PF_RES_DATA_2_PF_CQC_TIMER_BT_NUM_S); + roce_get_field(req_a->cqc_timer_bt_idx_num, + PF_RES_DATA_2_PF_CQC_TIMER_BT_NUM_M, + PF_RES_DATA_2_PF_CQC_TIMER_BT_NUM_S); return 0; } @@ -1415,8 +1513,6 @@ static int hns_roce_alloc_vf_resource(struct hns_roce_dev *hr_dev) req_a = (struct hns_roce_vf_res_a *)desc[0].data; req_b = (struct hns_roce_vf_res_b *)desc[1].data; - memset(req_a, 0, sizeof(*req_a)); - memset(req_b, 0, sizeof(*req_b)); for (i = 0; i < 2; i++) { hns_roce_cmq_setup_basic_desc(&desc[i], HNS_ROCE_OPC_ALLOC_VF_RES, false); @@ -1649,27 +1745,25 @@ static void set_default_caps(struct hns_roce_dev *hr_dev) caps->max_srq_wrs = HNS_ROCE_V2_MAX_SRQ_WR; caps->max_srq_sges = HNS_ROCE_V2_MAX_SRQ_SGE; - if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP08_B) { - caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC | HNS_ROCE_CAP_FLAG_MW | - HNS_ROCE_CAP_FLAG_SRQ | HNS_ROCE_CAP_FLAG_FRMR | - HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL; + caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC | HNS_ROCE_CAP_FLAG_MW | + HNS_ROCE_CAP_FLAG_SRQ | HNS_ROCE_CAP_FLAG_FRMR | + HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL; - caps->num_qpc_timer = HNS_ROCE_V2_MAX_QPC_TIMER_NUM; - caps->qpc_timer_entry_sz = HNS_ROCE_V2_QPC_TIMER_ENTRY_SZ; - caps->qpc_timer_ba_pg_sz = 0; - caps->qpc_timer_buf_pg_sz = 0; - caps->qpc_timer_hop_num = HNS_ROCE_HOP_NUM_0; - caps->num_cqc_timer = HNS_ROCE_V2_MAX_CQC_TIMER_NUM; - caps->cqc_timer_entry_sz = HNS_ROCE_V2_CQC_TIMER_ENTRY_SZ; - caps->cqc_timer_ba_pg_sz = 0; - caps->cqc_timer_buf_pg_sz = 0; - caps->cqc_timer_hop_num = HNS_ROCE_HOP_NUM_0; + caps->num_qpc_timer = HNS_ROCE_V2_MAX_QPC_TIMER_NUM; + caps->qpc_timer_entry_sz = HNS_ROCE_V2_QPC_TIMER_ENTRY_SZ; + caps->qpc_timer_ba_pg_sz = 0; + caps->qpc_timer_buf_pg_sz = 0; + caps->qpc_timer_hop_num = HNS_ROCE_HOP_NUM_0; + caps->num_cqc_timer = HNS_ROCE_V2_MAX_CQC_TIMER_NUM; + caps->cqc_timer_entry_sz = HNS_ROCE_V2_CQC_TIMER_ENTRY_SZ; + caps->cqc_timer_ba_pg_sz = 0; + caps->cqc_timer_buf_pg_sz = 0; + caps->cqc_timer_hop_num = HNS_ROCE_HOP_NUM_0; - caps->sccc_entry_sz = HNS_ROCE_V2_SCCC_ENTRY_SZ; - caps->sccc_ba_pg_sz = 0; - caps->sccc_buf_pg_sz = 0; - caps->sccc_hop_num = HNS_ROCE_SCCC_HOP_NUM; - } + caps->sccc_entry_sz = HNS_ROCE_V2_SCCC_ENTRY_SZ; + caps->sccc_ba_pg_sz = 0; + caps->sccc_buf_pg_sz = 0; + caps->sccc_hop_num = HNS_ROCE_SCCC_HOP_NUM; } static void calc_pg_sz(int obj_num, int obj_size, int hop_num, int ctx_bt_num, @@ -1786,6 +1880,9 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev) caps->flags = roce_get_field(resp_c->cap_flags_num_pds, V2_QUERY_PF_CAPS_C_CAP_FLAGS_M, V2_QUERY_PF_CAPS_C_CAP_FLAGS_S); + caps->flags |= le16_to_cpu(resp_d->cap_flags_ex) << + HNS_ROCE_CAP_FLAGS_EX_SHIFT; + caps->num_cqs = 1 << roce_get_field(resp_c->max_gid_num_cqs, V2_QUERY_PF_CAPS_C_NUM_CQS_M, V2_QUERY_PF_CAPS_C_NUM_CQS_S); @@ -1897,20 +1994,18 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev) caps->srqc_bt_num, &caps->srqc_buf_pg_sz, &caps->srqc_ba_pg_sz, HEM_TYPE_SRQC); - if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP08_B) { - caps->sccc_hop_num = ctx_hop_num; - caps->qpc_timer_hop_num = HNS_ROCE_HOP_NUM_0; - caps->cqc_timer_hop_num = HNS_ROCE_HOP_NUM_0; + caps->sccc_hop_num = ctx_hop_num; + caps->qpc_timer_hop_num = HNS_ROCE_HOP_NUM_0; + caps->cqc_timer_hop_num = HNS_ROCE_HOP_NUM_0; - calc_pg_sz(caps->num_qps, caps->sccc_entry_sz, - caps->sccc_hop_num, caps->sccc_bt_num, - &caps->sccc_buf_pg_sz, &caps->sccc_ba_pg_sz, - HEM_TYPE_SCCC); - calc_pg_sz(caps->num_cqc_timer, caps->cqc_timer_entry_sz, - caps->cqc_timer_hop_num, caps->cqc_timer_bt_num, - &caps->cqc_timer_buf_pg_sz, - &caps->cqc_timer_ba_pg_sz, HEM_TYPE_CQC_TIMER); - } + calc_pg_sz(caps->num_qps, caps->sccc_entry_sz, + caps->sccc_hop_num, caps->sccc_bt_num, + &caps->sccc_buf_pg_sz, &caps->sccc_ba_pg_sz, + HEM_TYPE_SCCC); + calc_pg_sz(caps->num_cqc_timer, caps->cqc_timer_entry_sz, + caps->cqc_timer_hop_num, caps->cqc_timer_bt_num, + &caps->cqc_timer_buf_pg_sz, + &caps->cqc_timer_ba_pg_sz, HEM_TYPE_CQC_TIMER); calc_pg_sz(caps->num_cqe_segs, caps->mtt_entry_sz, caps->cqe_hop_num, 1, &caps->cqe_buf_pg_sz, &caps->cqe_ba_pg_sz, HEM_TYPE_CQE); @@ -1957,32 +2052,24 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) return ret; } - if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP08_B) { - ret = hns_roce_query_pf_timer_resource(hr_dev); - if (ret) { - dev_err(hr_dev->dev, - "Query pf timer resource fail, ret = %d.\n", - ret); - return ret; - } + ret = hns_roce_query_pf_timer_resource(hr_dev); + if (ret) { + dev_err(hr_dev->dev, + "failed to query pf timer resource, ret = %d.\n", ret); + return ret; + } - ret = hns_roce_set_vf_switch_param(hr_dev, 0); - if (ret) { - dev_err(hr_dev->dev, - "Set function switch param fail, ret = %d.\n", - ret); - return ret; - } + ret = hns_roce_set_vf_switch_param(hr_dev, 0); + if (ret) { + dev_err(hr_dev->dev, + "failed to set function switch param, ret = %d.\n", + ret); + return ret; } hr_dev->vendor_part_id = hr_dev->pci_dev->device; hr_dev->sys_image_guid = be64_to_cpu(hr_dev->ib_dev.node_guid); - caps->num_mtt_segs = HNS_ROCE_V2_MAX_MTT_SEGS; - caps->num_cqe_segs = HNS_ROCE_V2_MAX_CQE_SEGS; - caps->num_srqwqe_segs = HNS_ROCE_V2_MAX_SRQWQE_SEGS; - caps->num_idx_segs = HNS_ROCE_V2_MAX_IDX_SEGS; - caps->pbl_ba_pg_sz = HNS_ROCE_BA_PG_SZ_SUPPORTED_16K; caps->pbl_buf_pg_sz = 0; caps->pbl_hop_num = HNS_ROCE_PBL_HOP_NUM; @@ -2040,8 +2127,6 @@ static int hns_roce_config_link_table(struct hns_roce_dev *hr_dev, page_num = link_tbl->npages; entry = link_tbl->table.buf; - memset(req_a, 0, sizeof(*req_a)); - memset(req_b, 0, sizeof(*req_b)); for (i = 0; i < 2; i++) { hns_roce_cmq_setup_basic_desc(&desc[i], opcode, false); @@ -2050,39 +2135,30 @@ static int hns_roce_config_link_table(struct hns_roce_dev *hr_dev, desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); else desc[i].flag &= ~cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - - if (i == 0) { - req_a->base_addr_l = - cpu_to_le32(link_tbl->table.map & 0xffffffff); - req_a->base_addr_h = - cpu_to_le32(link_tbl->table.map >> 32); - roce_set_field(req_a->depth_pgsz_init_en, - CFG_LLM_QUE_DEPTH_M, CFG_LLM_QUE_DEPTH_S, - link_tbl->npages); - roce_set_field(req_a->depth_pgsz_init_en, - CFG_LLM_QUE_PGSZ_M, CFG_LLM_QUE_PGSZ_S, - link_tbl->pg_sz); - req_a->head_ba_l = cpu_to_le32(entry[0].blk_ba0); - req_a->head_ba_h_nxtptr = - cpu_to_le32(entry[0].blk_ba1_nxt_ptr); - roce_set_field(req_a->head_ptr, CFG_LLM_HEAD_PTR_M, - CFG_LLM_HEAD_PTR_S, 0); - } else { - req_b->tail_ba_l = - cpu_to_le32(entry[page_num - 1].blk_ba0); - roce_set_field(req_b->tail_ba_h, CFG_LLM_TAIL_BA_H_M, - CFG_LLM_TAIL_BA_H_S, - entry[page_num - 1].blk_ba1_nxt_ptr & - HNS_ROCE_LINK_TABLE_BA1_M); - roce_set_field(req_b->tail_ptr, CFG_LLM_TAIL_PTR_M, - CFG_LLM_TAIL_PTR_S, - (entry[page_num - 2].blk_ba1_nxt_ptr & - HNS_ROCE_LINK_TABLE_NXT_PTR_M) >> - HNS_ROCE_LINK_TABLE_NXT_PTR_S); - } } + + req_a->base_addr_l = cpu_to_le32(link_tbl->table.map & 0xffffffff); + req_a->base_addr_h = cpu_to_le32(link_tbl->table.map >> 32); + roce_set_field(req_a->depth_pgsz_init_en, CFG_LLM_QUE_DEPTH_M, + CFG_LLM_QUE_DEPTH_S, link_tbl->npages); + roce_set_field(req_a->depth_pgsz_init_en, CFG_LLM_QUE_PGSZ_M, + CFG_LLM_QUE_PGSZ_S, link_tbl->pg_sz); roce_set_field(req_a->depth_pgsz_init_en, CFG_LLM_INIT_EN_M, CFG_LLM_INIT_EN_S, 1); + req_a->head_ba_l = cpu_to_le32(entry[0].blk_ba0); + req_a->head_ba_h_nxtptr = cpu_to_le32(entry[0].blk_ba1_nxt_ptr); + roce_set_field(req_a->head_ptr, CFG_LLM_HEAD_PTR_M, CFG_LLM_HEAD_PTR_S, + 0); + + req_b->tail_ba_l = cpu_to_le32(entry[page_num - 1].blk_ba0); + roce_set_field(req_b->tail_ba_h, CFG_LLM_TAIL_BA_H_M, + CFG_LLM_TAIL_BA_H_S, + entry[page_num - 1].blk_ba1_nxt_ptr & + HNS_ROCE_LINK_TABLE_BA1_M); + roce_set_field(req_b->tail_ptr, CFG_LLM_TAIL_PTR_M, CFG_LLM_TAIL_PTR_S, + (entry[page_num - 2].blk_ba1_nxt_ptr & + HNS_ROCE_LINK_TABLE_NXT_PTR_M) >> + HNS_ROCE_LINK_TABLE_NXT_PTR_S); return hns_roce_cmq_send(hr_dev, desc, 2); } @@ -2254,8 +2330,7 @@ static void hns_roce_v2_exit(struct hns_roce_dev *hr_dev) { struct hns_roce_v2_priv *priv = hr_dev->priv; - if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP08_B) - hns_roce_function_clear(hr_dev); + hns_roce_function_clear(hr_dev); hns_roce_free_link_table(hr_dev, &priv->tpq); hns_roce_free_link_table(hr_dev, &priv->tsq); @@ -2438,47 +2513,42 @@ static int hns_roce_v2_set_mac(struct hns_roce_dev *hr_dev, u8 phy_port, reg_smac_l = *(u32 *)(&addr[0]); reg_smac_h = *(u16 *)(&addr[4]); - memset(smac_tb, 0, sizeof(*smac_tb)); - roce_set_field(smac_tb->tb_idx_rsv, - CFG_SMAC_TB_IDX_M, + roce_set_field(smac_tb->tb_idx_rsv, CFG_SMAC_TB_IDX_M, CFG_SMAC_TB_IDX_S, phy_port); - roce_set_field(smac_tb->vf_smac_h_rsv, - CFG_SMAC_TB_VF_SMAC_H_M, + roce_set_field(smac_tb->vf_smac_h_rsv, CFG_SMAC_TB_VF_SMAC_H_M, CFG_SMAC_TB_VF_SMAC_H_S, reg_smac_h); smac_tb->vf_smac_l = cpu_to_le32(reg_smac_l); return hns_roce_cmq_send(hr_dev, &desc, 1); } -static int set_mtpt_pbl(struct hns_roce_v2_mpt_entry *mpt_entry, +static int set_mtpt_pbl(struct hns_roce_dev *hr_dev, + struct hns_roce_v2_mpt_entry *mpt_entry, struct hns_roce_mr *mr) { - struct sg_dma_page_iter sg_iter; - u64 page_addr; - u64 *pages; - int i; + u64 pages[HNS_ROCE_V2_MAX_INNER_MTPT_NUM] = { 0 }; + struct ib_device *ibdev = &hr_dev->ib_dev; + dma_addr_t pbl_ba; + int i, count; - mpt_entry->pbl_size = cpu_to_le32(mr->pbl_size); - mpt_entry->pbl_ba_l = cpu_to_le32(lower_32_bits(mr->pbl_ba >> 3)); - roce_set_field(mpt_entry->byte_48_mode_ba, - V2_MPT_BYTE_48_PBL_BA_H_M, V2_MPT_BYTE_48_PBL_BA_H_S, - upper_32_bits(mr->pbl_ba >> 3)); + count = hns_roce_mtr_find(hr_dev, &mr->pbl_mtr, 0, pages, + ARRAY_SIZE(pages), &pbl_ba); + if (count < 1) { + ibdev_err(ibdev, "failed to find PBL mtr, count = %d.\n", + count); + return -ENOBUFS; + } - pages = (u64 *)__get_free_page(GFP_KERNEL); - if (!pages) - return -ENOMEM; + /* Aligned to the hardware address access unit */ + for (i = 0; i < count; i++) + pages[i] >>= 6; - i = 0; - for_each_sg_dma_page(mr->umem->sg_head.sgl, &sg_iter, mr->umem->nmap, 0) { - page_addr = sg_page_iter_dma_address(&sg_iter); - pages[i] = page_addr >> 6; + mpt_entry->pbl_size = cpu_to_le32(mr->npages); + mpt_entry->pbl_ba_l = cpu_to_le32(pbl_ba >> 3); + roce_set_field(mpt_entry->byte_48_mode_ba, + V2_MPT_BYTE_48_PBL_BA_H_M, V2_MPT_BYTE_48_PBL_BA_H_S, + upper_32_bits(pbl_ba >> 3)); - /* Record the first 2 entry directly to MTPT table */ - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1) - goto found; - i++; - } -found: mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0])); roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M, V2_MPT_BYTE_56_PA0_H_S, upper_32_bits(pages[0])); @@ -2489,14 +2559,13 @@ found: roce_set_field(mpt_entry->byte_64_buf_pa1, V2_MPT_BYTE_64_PBL_BUF_PG_SZ_M, V2_MPT_BYTE_64_PBL_BUF_PG_SZ_S, - mr->pbl_buf_pg_sz + PG_SHIFT_OFFSET); - - free_page((unsigned long)pages); + to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.buf_pg_shift)); return 0; } -static int hns_roce_v2_write_mtpt(void *mb_buf, struct hns_roce_mr *mr, +static int hns_roce_v2_write_mtpt(struct hns_roce_dev *hr_dev, + void *mb_buf, struct hns_roce_mr *mr, unsigned long mtpt_idx) { struct hns_roce_v2_mpt_entry *mpt_entry; @@ -2513,7 +2582,7 @@ static int hns_roce_v2_write_mtpt(void *mb_buf, struct hns_roce_mr *mr, roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PBL_BA_PG_SZ_M, V2_MPT_BYTE_4_PBL_BA_PG_SZ_S, - mr->pbl_ba_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.ba_pg_shift)); roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M, V2_MPT_BYTE_4_PD_S, mr->pd); @@ -2545,7 +2614,7 @@ static int hns_roce_v2_write_mtpt(void *mb_buf, struct hns_roce_mr *mr, if (mr->type == MR_TYPE_DMA) return 0; - ret = set_mtpt_pbl(mpt_entry, mr); + ret = set_mtpt_pbl(hr_dev, mpt_entry, mr); return ret; } @@ -2591,19 +2660,27 @@ static int hns_roce_v2_rereg_write_mtpt(struct hns_roce_dev *hr_dev, mr->iova = iova; mr->size = size; - ret = set_mtpt_pbl(mpt_entry, mr); + ret = set_mtpt_pbl(hr_dev, mpt_entry, mr); } return ret; } -static int hns_roce_v2_frmr_write_mtpt(void *mb_buf, struct hns_roce_mr *mr) +static int hns_roce_v2_frmr_write_mtpt(struct hns_roce_dev *hr_dev, + void *mb_buf, struct hns_roce_mr *mr) { + struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_v2_mpt_entry *mpt_entry; + dma_addr_t pbl_ba = 0; mpt_entry = mb_buf; memset(mpt_entry, 0, sizeof(*mpt_entry)); + if (hns_roce_mtr_find(hr_dev, &mr->pbl_mtr, 0, NULL, 0, &pbl_ba) < 0) { + ibdev_err(ibdev, "failed to find frmr mtr.\n"); + return -ENOBUFS; + } + roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_MPT_ST_M, V2_MPT_BYTE_4_MPT_ST_S, V2_MPT_ST_FREE); roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PBL_HOP_NUM_M, @@ -2611,7 +2688,7 @@ static int hns_roce_v2_frmr_write_mtpt(void *mb_buf, struct hns_roce_mr *mr) roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PBL_BA_PG_SZ_M, V2_MPT_BYTE_4_PBL_BA_PG_SZ_S, - mr->pbl_ba_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.ba_pg_shift)); roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M, V2_MPT_BYTE_4_PD_S, mr->pd); @@ -2624,17 +2701,17 @@ static int hns_roce_v2_frmr_write_mtpt(void *mb_buf, struct hns_roce_mr *mr) roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_MR_MW_S, 0); roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_BPD_S, 1); - mpt_entry->pbl_size = cpu_to_le32(mr->pbl_size); + mpt_entry->pbl_size = cpu_to_le32(mr->npages); - mpt_entry->pbl_ba_l = cpu_to_le32(lower_32_bits(mr->pbl_ba >> 3)); + mpt_entry->pbl_ba_l = cpu_to_le32(lower_32_bits(pbl_ba >> 3)); roce_set_field(mpt_entry->byte_48_mode_ba, V2_MPT_BYTE_48_PBL_BA_H_M, V2_MPT_BYTE_48_PBL_BA_H_S, - upper_32_bits(mr->pbl_ba >> 3)); + upper_32_bits(pbl_ba >> 3)); roce_set_field(mpt_entry->byte_64_buf_pa1, V2_MPT_BYTE_64_PBL_BUF_PG_SZ_M, V2_MPT_BYTE_64_PBL_BUF_PG_SZ_S, - mr->pbl_buf_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.buf_pg_shift)); return 0; } @@ -2680,7 +2757,8 @@ static int hns_roce_v2_mw_write_mtpt(void *mb_buf, struct hns_roce_mw *mw) static void *get_cqe_v2(struct hns_roce_cq *hr_cq, int n) { - return hns_roce_buf_offset(&hr_cq->buf, n * HNS_ROCE_V2_CQE_ENTRY_SIZE); + return hns_roce_buf_offset(hr_cq->mtr.kmem, + n * HNS_ROCE_V2_CQE_ENTRY_SIZE); } static void *get_sw_cqe_v2(struct hns_roce_cq *hr_cq, int n) @@ -2692,30 +2770,9 @@ static void *get_sw_cqe_v2(struct hns_roce_cq *hr_cq, int n) !!(n & hr_cq->cq_depth)) ? cqe : NULL; } -static struct hns_roce_v2_cqe *next_cqe_sw_v2(struct hns_roce_cq *hr_cq) +static inline void hns_roce_v2_cq_set_ci(struct hns_roce_cq *hr_cq, u32 ci) { - return get_sw_cqe_v2(hr_cq, hr_cq->cons_index); -} - -static void *get_srq_wqe(struct hns_roce_srq *srq, int n) -{ - return hns_roce_buf_offset(&srq->buf, n << srq->wqe_shift); -} - -static void hns_roce_free_srq_wqe(struct hns_roce_srq *srq, int wqe_index) -{ - /* always called with interrupts disabled. */ - spin_lock(&srq->lock); - - bitmap_clear(srq->idx_que.bitmap, wqe_index, 1); - srq->tail++; - - spin_unlock(&srq->lock); -} - -static void hns_roce_v2_cq_set_ci(struct hns_roce_cq *hr_cq, u32 cons_index) -{ - *hr_cq->set_ci_db = cons_index & V2_CQ_DB_PARAMETER_CONS_IDX_M; + *hr_cq->set_ci_db = ci & V2_CQ_DB_PARAMETER_CONS_IDX_M; } static void __hns_roce_v2_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn, @@ -2801,39 +2858,39 @@ static void hns_roce_v2_write_cqc(struct hns_roce_dev *hr_dev, roce_set_field(cq_context->byte_8_cqn, V2_CQC_BYTE_8_CQN_M, V2_CQC_BYTE_8_CQN_S, hr_cq->cqn); - cq_context->cqe_cur_blk_addr = cpu_to_le32(mtts[0] >> PAGE_ADDR_SHIFT); + cq_context->cqe_cur_blk_addr = cpu_to_le32(to_hr_hw_page_addr(mtts[0])); roce_set_field(cq_context->byte_16_hop_addr, V2_CQC_BYTE_16_CQE_CUR_BLK_ADDR_M, V2_CQC_BYTE_16_CQE_CUR_BLK_ADDR_S, - mtts[0] >> (32 + PAGE_ADDR_SHIFT)); + upper_32_bits(to_hr_hw_page_addr(mtts[0]))); roce_set_field(cq_context->byte_16_hop_addr, V2_CQC_BYTE_16_CQE_HOP_NUM_M, V2_CQC_BYTE_16_CQE_HOP_NUM_S, hr_dev->caps.cqe_hop_num == HNS_ROCE_HOP_NUM_0 ? 0 : hr_dev->caps.cqe_hop_num); - cq_context->cqe_nxt_blk_addr = cpu_to_le32(mtts[1] >> PAGE_ADDR_SHIFT); + cq_context->cqe_nxt_blk_addr = cpu_to_le32(to_hr_hw_page_addr(mtts[1])); roce_set_field(cq_context->byte_24_pgsz_addr, V2_CQC_BYTE_24_CQE_NXT_BLK_ADDR_M, V2_CQC_BYTE_24_CQE_NXT_BLK_ADDR_S, - mtts[1] >> (32 + PAGE_ADDR_SHIFT)); + upper_32_bits(to_hr_hw_page_addr(mtts[1]))); roce_set_field(cq_context->byte_24_pgsz_addr, V2_CQC_BYTE_24_CQE_BA_PG_SZ_M, V2_CQC_BYTE_24_CQE_BA_PG_SZ_S, - hr_dev->caps.cqe_ba_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(hr_cq->mtr.hem_cfg.ba_pg_shift)); roce_set_field(cq_context->byte_24_pgsz_addr, V2_CQC_BYTE_24_CQE_BUF_PG_SZ_M, V2_CQC_BYTE_24_CQE_BUF_PG_SZ_S, - hr_dev->caps.cqe_buf_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(hr_cq->mtr.hem_cfg.buf_pg_shift)); cq_context->cqe_ba = cpu_to_le32(dma_handle >> 3); roce_set_field(cq_context->byte_40_cqe_ba, V2_CQC_BYTE_40_CQE_BA_M, V2_CQC_BYTE_40_CQE_BA_S, (dma_handle >> (32 + 3))); - if (hr_cq->db_en) - roce_set_bit(cq_context->byte_44_db_record, - V2_CQC_BYTE_44_DB_RECORD_EN_S, 1); + roce_set_bit(cq_context->byte_44_db_record, + V2_CQC_BYTE_44_DB_RECORD_EN_S, + (hr_cq->flags & HNS_ROCE_CQ_FLAG_RECORD_DB) ? 1 : 0); roce_set_field(cq_context->byte_44_db_record, V2_CQC_BYTE_44_DB_RECORD_ADDR_M, @@ -2873,8 +2930,7 @@ static int hns_roce_v2_req_notify_cq(struct ib_cq *ibcq, roce_set_field(doorbell[0], V2_CQ_DB_BYTE_4_CMD_M, V2_DB_BYTE_4_CMD_S, HNS_ROCE_V2_CQ_DB_NTR); roce_set_field(doorbell[1], V2_CQ_DB_PARAMETER_CONS_IDX_M, - V2_CQ_DB_PARAMETER_CONS_IDX_S, - hr_cq->cons_index & ((hr_cq->cq_depth << 1) - 1)); + V2_CQ_DB_PARAMETER_CONS_IDX_S, hr_cq->cons_index); roce_set_field(doorbell[1], V2_CQ_DB_PARAMETER_CMD_SN_M, V2_CQ_DB_PARAMETER_CMD_SN_S, hr_cq->arm_sn & 0x3); roce_set_bit(doorbell[1], V2_CQ_DB_PARAMETER_NOTIFY_S, @@ -2911,7 +2967,7 @@ static int hns_roce_handle_recv_inl_wqe(struct hns_roce_v2_cqe *cqe, wqe_buf += size; } - if (data_len) { + if (unlikely(data_len)) { wc->status = IB_WC_LOC_LEN_ERR; return -EAGAIN; } @@ -2968,6 +3024,71 @@ out: return npolled; } +static void get_cqe_status(struct hns_roce_dev *hr_dev, struct hns_roce_qp *qp, + struct hns_roce_v2_cqe *cqe, struct ib_wc *wc) +{ + static const struct { + u32 cqe_status; + enum ib_wc_status wc_status; + } map[] = { + { HNS_ROCE_CQE_V2_SUCCESS, IB_WC_SUCCESS }, + { HNS_ROCE_CQE_V2_LOCAL_LENGTH_ERR, IB_WC_LOC_LEN_ERR }, + { HNS_ROCE_CQE_V2_LOCAL_QP_OP_ERR, IB_WC_LOC_QP_OP_ERR }, + { HNS_ROCE_CQE_V2_LOCAL_PROT_ERR, IB_WC_LOC_PROT_ERR }, + { HNS_ROCE_CQE_V2_WR_FLUSH_ERR, IB_WC_WR_FLUSH_ERR }, + { HNS_ROCE_CQE_V2_MW_BIND_ERR, IB_WC_MW_BIND_ERR }, + { HNS_ROCE_CQE_V2_BAD_RESP_ERR, IB_WC_BAD_RESP_ERR }, + { HNS_ROCE_CQE_V2_LOCAL_ACCESS_ERR, IB_WC_LOC_ACCESS_ERR }, + { HNS_ROCE_CQE_V2_REMOTE_INVAL_REQ_ERR, IB_WC_REM_INV_REQ_ERR }, + { HNS_ROCE_CQE_V2_REMOTE_ACCESS_ERR, IB_WC_REM_ACCESS_ERR }, + { HNS_ROCE_CQE_V2_REMOTE_OP_ERR, IB_WC_REM_OP_ERR }, + { HNS_ROCE_CQE_V2_TRANSPORT_RETRY_EXC_ERR, + IB_WC_RETRY_EXC_ERR }, + { HNS_ROCE_CQE_V2_RNR_RETRY_EXC_ERR, IB_WC_RNR_RETRY_EXC_ERR }, + { HNS_ROCE_CQE_V2_REMOTE_ABORT_ERR, IB_WC_REM_ABORT_ERR }, + { HNS_ROCE_CQE_V2_GENERAL_ERR, IB_WC_GENERAL_ERR} + }; + + u32 cqe_status = roce_get_field(cqe->byte_4, V2_CQE_BYTE_4_STATUS_M, + V2_CQE_BYTE_4_STATUS_S); + int i; + + wc->status = IB_WC_GENERAL_ERR; + for (i = 0; i < ARRAY_SIZE(map); i++) + if (cqe_status == map[i].cqe_status) { + wc->status = map[i].wc_status; + break; + } + + if (likely(wc->status == IB_WC_SUCCESS || + wc->status == IB_WC_WR_FLUSH_ERR)) + return; + + ibdev_err(&hr_dev->ib_dev, "error cqe status 0x%x:\n", cqe_status); + print_hex_dump(KERN_ERR, "", DUMP_PREFIX_NONE, 16, 4, cqe, + sizeof(*cqe), false); + + /* + * For hns ROCEE, GENERAL_ERR is an error type that is not defined in + * the standard protocol, the driver must ignore it and needn't to set + * the QP to an error state. + */ + if (cqe_status == HNS_ROCE_CQE_V2_GENERAL_ERR) + return; + + /* + * Hip08 hardware cannot flush the WQEs in SQ/RQ if the QP state gets + * into errored mode. Hence, as a workaround to this hardware + * limitation, driver needs to assist in flushing. But the flushing + * operation uses mailbox to convey the QP state to the hardware and + * which can sleep due to the mutex protection around the mailbox calls. + * Hence, use the deferred flush for now. Once wc error detected, the + * flushing operation is needed. + */ + if (!test_and_set_bit(HNS_ROCE_FLUSH_FLAG, &qp->flush_flag)) + init_flush_work(hr_dev, qp); +} + static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, struct hns_roce_qp **cur_qp, struct ib_wc *wc) { @@ -2979,12 +3100,11 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, int is_send; u16 wqe_ctr; u32 opcode; - u32 status; int qpn; int ret; /* Find cqe according to consumer index */ - cqe = next_cqe_sw_v2(hr_cq); + cqe = get_sw_cqe_v2(hr_cq, hr_cq->cons_index); if (!cqe) return -EAGAIN; @@ -3009,7 +3129,6 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, *cur_qp = hr_qp; } - hr_qp = *cur_qp; wc->qp = &(*cur_qp)->ibqp; wc->vendor_err = 0; @@ -3044,77 +3163,8 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, ++wq->tail; } - status = roce_get_field(cqe->byte_4, V2_CQE_BYTE_4_STATUS_M, - V2_CQE_BYTE_4_STATUS_S); - switch (status & HNS_ROCE_V2_CQE_STATUS_MASK) { - case HNS_ROCE_CQE_V2_SUCCESS: - wc->status = IB_WC_SUCCESS; - break; - case HNS_ROCE_CQE_V2_LOCAL_LENGTH_ERR: - wc->status = IB_WC_LOC_LEN_ERR; - break; - case HNS_ROCE_CQE_V2_LOCAL_QP_OP_ERR: - wc->status = IB_WC_LOC_QP_OP_ERR; - break; - case HNS_ROCE_CQE_V2_LOCAL_PROT_ERR: - wc->status = IB_WC_LOC_PROT_ERR; - break; - case HNS_ROCE_CQE_V2_WR_FLUSH_ERR: - wc->status = IB_WC_WR_FLUSH_ERR; - break; - case HNS_ROCE_CQE_V2_MW_BIND_ERR: - wc->status = IB_WC_MW_BIND_ERR; - break; - case HNS_ROCE_CQE_V2_BAD_RESP_ERR: - wc->status = IB_WC_BAD_RESP_ERR; - break; - case HNS_ROCE_CQE_V2_LOCAL_ACCESS_ERR: - wc->status = IB_WC_LOC_ACCESS_ERR; - break; - case HNS_ROCE_CQE_V2_REMOTE_INVAL_REQ_ERR: - wc->status = IB_WC_REM_INV_REQ_ERR; - break; - case HNS_ROCE_CQE_V2_REMOTE_ACCESS_ERR: - wc->status = IB_WC_REM_ACCESS_ERR; - break; - case HNS_ROCE_CQE_V2_REMOTE_OP_ERR: - wc->status = IB_WC_REM_OP_ERR; - break; - case HNS_ROCE_CQE_V2_TRANSPORT_RETRY_EXC_ERR: - wc->status = IB_WC_RETRY_EXC_ERR; - break; - case HNS_ROCE_CQE_V2_RNR_RETRY_EXC_ERR: - wc->status = IB_WC_RNR_RETRY_EXC_ERR; - break; - case HNS_ROCE_CQE_V2_REMOTE_ABORT_ERR: - wc->status = IB_WC_REM_ABORT_ERR; - break; - default: - wc->status = IB_WC_GENERAL_ERR; - break; - } - - /* - * Hip08 hardware cannot flush the WQEs in SQ/RQ if the QP state gets - * into errored mode. Hence, as a workaround to this hardware - * limitation, driver needs to assist in flushing. But the flushing - * operation uses mailbox to convey the QP state to the hardware and - * which can sleep due to the mutex protection around the mailbox calls. - * Hence, use the deferred flush for now. Once wc error detected, the - * flushing operation is needed. - */ - if (wc->status != IB_WC_SUCCESS && - wc->status != IB_WC_WR_FLUSH_ERR) { - ibdev_err(&hr_dev->ib_dev, "error cqe status is: 0x%x\n", - status & HNS_ROCE_V2_CQE_STATUS_MASK); - - if (!test_and_set_bit(HNS_ROCE_FLUSH_FLAG, &hr_qp->flush_flag)) - init_flush_work(hr_dev, hr_qp); - - return 0; - } - - if (wc->status == IB_WC_WR_FLUSH_ERR) + get_cqe_status(hr_dev, *cur_qp, cqe, wc); + if (unlikely(wc->status != IB_WC_SUCCESS)) return 0; if (is_send) { @@ -3122,51 +3172,51 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, /* SQ corresponding to CQE */ switch (roce_get_field(cqe->byte_4, V2_CQE_BYTE_4_OPCODE_M, V2_CQE_BYTE_4_OPCODE_S) & 0x1f) { - case HNS_ROCE_SQ_OPCODE_SEND: + case HNS_ROCE_V2_WQE_OP_SEND: wc->opcode = IB_WC_SEND; break; - case HNS_ROCE_SQ_OPCODE_SEND_WITH_INV: + case HNS_ROCE_V2_WQE_OP_SEND_WITH_INV: wc->opcode = IB_WC_SEND; break; - case HNS_ROCE_SQ_OPCODE_SEND_WITH_IMM: + case HNS_ROCE_V2_WQE_OP_SEND_WITH_IMM: wc->opcode = IB_WC_SEND; wc->wc_flags |= IB_WC_WITH_IMM; break; - case HNS_ROCE_SQ_OPCODE_RDMA_READ: + case HNS_ROCE_V2_WQE_OP_RDMA_READ: wc->opcode = IB_WC_RDMA_READ; wc->byte_len = le32_to_cpu(cqe->byte_cnt); break; - case HNS_ROCE_SQ_OPCODE_RDMA_WRITE: + case HNS_ROCE_V2_WQE_OP_RDMA_WRITE: wc->opcode = IB_WC_RDMA_WRITE; break; - case HNS_ROCE_SQ_OPCODE_RDMA_WRITE_WITH_IMM: + case HNS_ROCE_V2_WQE_OP_RDMA_WRITE_WITH_IMM: wc->opcode = IB_WC_RDMA_WRITE; wc->wc_flags |= IB_WC_WITH_IMM; break; - case HNS_ROCE_SQ_OPCODE_LOCAL_INV: + case HNS_ROCE_V2_WQE_OP_LOCAL_INV: wc->opcode = IB_WC_LOCAL_INV; wc->wc_flags |= IB_WC_WITH_INVALIDATE; break; - case HNS_ROCE_SQ_OPCODE_ATOMIC_COMP_AND_SWAP: + case HNS_ROCE_V2_WQE_OP_ATOM_CMP_AND_SWAP: wc->opcode = IB_WC_COMP_SWAP; wc->byte_len = 8; break; - case HNS_ROCE_SQ_OPCODE_ATOMIC_FETCH_AND_ADD: + case HNS_ROCE_V2_WQE_OP_ATOM_FETCH_AND_ADD: wc->opcode = IB_WC_FETCH_ADD; wc->byte_len = 8; break; - case HNS_ROCE_SQ_OPCODE_ATOMIC_MASK_COMP_AND_SWAP: + case HNS_ROCE_V2_WQE_OP_ATOM_MSK_CMP_AND_SWAP: wc->opcode = IB_WC_MASKED_COMP_SWAP; wc->byte_len = 8; break; - case HNS_ROCE_SQ_OPCODE_ATOMIC_MASK_FETCH_AND_ADD: + case HNS_ROCE_V2_WQE_OP_ATOM_MSK_FETCH_AND_ADD: wc->opcode = IB_WC_MASKED_FETCH_ADD; wc->byte_len = 8; break; - case HNS_ROCE_SQ_OPCODE_FAST_REG_WR: + case HNS_ROCE_V2_WQE_OP_FAST_REG_PMR: wc->opcode = IB_WC_REG_MR; break; - case HNS_ROCE_SQ_OPCODE_BIND_MW: + case HNS_ROCE_V2_WQE_OP_BIND_MW: wc->opcode = IB_WC_REG_MR; break; default: @@ -3213,7 +3263,7 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, opcode == HNS_ROCE_V2_OPCODE_SEND_WITH_INV) && (roce_get_bit(cqe->byte_4, V2_CQE_BYTE_4_RQ_INLINE_S))) { ret = hns_roce_handle_recv_inl_wqe(cqe, cur_qp, wc); - if (ret) + if (unlikely(ret)) return -EAGAIN; } @@ -3326,11 +3376,33 @@ static int get_op_for_set_hem(struct hns_roce_dev *hr_dev, u32 type, return op + step_idx; } +static int set_hem_to_hw(struct hns_roce_dev *hr_dev, int obj, u64 bt_ba, + u32 hem_type, int step_idx) +{ + struct hns_roce_cmd_mailbox *mailbox; + int ret; + int op; + + op = get_op_for_set_hem(hr_dev, hem_type, step_idx); + if (op < 0) + return 0; + + mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + ret = hns_roce_cmd_mbox(hr_dev, bt_ba, mailbox->dma, obj, + 0, op, HNS_ROCE_CMD_TIMEOUT_MSECS); + + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + + return ret; +} + static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, struct hns_roce_hem_table *table, int obj, int step_idx) { - struct hns_roce_cmd_mailbox *mailbox; struct hns_roce_hem_iter iter; struct hns_roce_hem_mhop mhop; struct hns_roce_hem *hem; @@ -3342,7 +3414,6 @@ static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, u64 bt_ba = 0; u32 chunk_ba_num; u32 hop_num; - int op; if (!hns_roce_check_whether_mhop(hr_dev, table->type)) return 0; @@ -3364,14 +3435,6 @@ static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, hem_idx = i; } - op = get_op_for_set_hem(hr_dev, table->type, step_idx); - if (op == -EINVAL) - return 0; - - mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); - if (IS_ERR(mailbox)) - return PTR_ERR(mailbox); - if (table->type == HEM_TYPE_SCCC) obj = mhop.l0_idx; @@ -3380,11 +3443,8 @@ static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, for (hns_roce_hem_first(hem, &iter); !hns_roce_hem_last(&iter); hns_roce_hem_next(&iter)) { bt_ba = hns_roce_hem_addr(&iter); - - /* configure the ba, tag, and op */ - ret = hns_roce_cmd_mbox(hr_dev, bt_ba, mailbox->dma, - obj, 0, op, - HNS_ROCE_CMD_TIMEOUT_MSECS); + ret = set_hem_to_hw(hr_dev, obj, bt_ba, table->type, + step_idx); } } else { if (step_idx == 0) @@ -3392,12 +3452,9 @@ static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, else if (step_idx == 1 && hop_num == 2) bt_ba = table->bt_l1_dma_addr[l1_idx]; - /* configure the ba, tag, and op */ - ret = hns_roce_cmd_mbox(hr_dev, bt_ba, mailbox->dma, obj, - 0, op, HNS_ROCE_CMD_TIMEOUT_MSECS); + ret = set_hem_to_hw(hr_dev, obj, bt_ba, table->type, step_idx); } - hns_roce_free_cmd_mailbox(hr_dev, mailbox); return ret; } @@ -3514,29 +3571,18 @@ static void set_qpc_wqe_cnt(struct hns_roce_qp *hr_qp, struct hns_roce_v2_qp_context *context, struct hns_roce_v2_qp_context *qpc_mask) { - if (hr_qp->ibqp.qp_type == IB_QPT_GSI) - roce_set_field(context->byte_4_sqpn_tst, - V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, - ilog2((unsigned int)hr_qp->sge.sge_cnt)); - else - roce_set_field(context->byte_4_sqpn_tst, - V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, - hr_qp->sq.max_gs > - HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE ? - ilog2((unsigned int)hr_qp->sge.sge_cnt) : 0); + roce_set_field(context->byte_4_sqpn_tst, + V2_QPC_BYTE_4_SGE_SHIFT_M, V2_QPC_BYTE_4_SGE_SHIFT_S, + to_hr_hem_entries_shift(hr_qp->sge.sge_cnt, + hr_qp->sge.sge_shift)); roce_set_field(context->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, - ilog2((unsigned int)hr_qp->sq.wqe_cnt)); + ilog2(hr_qp->sq.wqe_cnt)); roce_set_field(context->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, - (hr_qp->ibqp.qp_type == IB_QPT_XRC_INI || - hr_qp->ibqp.qp_type == IB_QPT_XRC_TGT || - hr_qp->ibqp.srq) ? 0 : - ilog2((unsigned int)hr_qp->rq.wqe_cnt)); + ilog2(hr_qp->rq.wqe_cnt)); } static void modify_qp_reset_to_init(struct ib_qp *ibqp, @@ -3572,7 +3618,7 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp, roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_VLAN_ID_M, V2_QPC_BYTE_24_VLAN_ID_S, 0xfff); - if (hr_qp->rdb_en) + if (hr_qp->en_flags & HNS_ROCE_QP_CAP_RQ_RECORD_DB) roce_set_bit(context->byte_68_rq_db, V2_QPC_BYTE_68_RQ_RECORD_EN_S, 1); @@ -3708,86 +3754,24 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp, } } -static bool check_wqe_rq_mtt_count(struct hns_roce_dev *hr_dev, - struct hns_roce_qp *hr_qp, int mtt_cnt, - u32 page_size) -{ - struct ib_device *ibdev = &hr_dev->ib_dev; - - if (hr_qp->rq.wqe_cnt < 1) - return true; - - if (mtt_cnt < 1) { - ibdev_err(ibdev, "failed to find RQWQE buf ba of QP(0x%lx)\n", - hr_qp->qpn); - return false; - } - - if (mtt_cnt < MTT_MIN_COUNT && - (hr_qp->rq.offset + page_size) < hr_qp->buff_size) { - ibdev_err(ibdev, - "failed to find next RQWQE buf ba of QP(0x%lx)\n", - hr_qp->qpn); - return false; - } - - return true; -} - -static int modify_qp_init_to_rtr(struct ib_qp *ibqp, - const struct ib_qp_attr *attr, int attr_mask, - struct hns_roce_v2_qp_context *context, - struct hns_roce_v2_qp_context *qpc_mask) +static int config_qp_rq_buf(struct hns_roce_dev *hr_dev, + struct hns_roce_qp *hr_qp, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) { - const struct ib_global_route *grh = rdma_ah_read_grh(&attr->ah_attr); - struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); - struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); - struct ib_device *ibdev = &hr_dev->ib_dev; u64 mtts[MTT_MIN_COUNT] = { 0 }; - dma_addr_t dma_handle_3; - dma_addr_t dma_handle_2; u64 wqe_sge_ba; - u32 page_size; - u8 port_num; - u64 *mtts_3; - u64 *mtts_2; int count; - u8 *dmac; - u8 *smac; - int port; /* Search qp buf's mtts */ - page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT); - count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, - hr_qp->rq.offset / page_size, mtts, + count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, hr_qp->rq.offset, mtts, MTT_MIN_COUNT, &wqe_sge_ba); - if (!ibqp->srq) - if (!check_wqe_rq_mtt_count(hr_dev, hr_qp, count, page_size)) - return -EINVAL; - - /* Search IRRL's mtts */ - mtts_2 = hns_roce_table_find(hr_dev, &hr_dev->qp_table.irrl_table, - hr_qp->qpn, &dma_handle_2); - if (!mtts_2) { - ibdev_err(ibdev, "failed to find QP irrl_table\n"); - return -EINVAL; - } - - /* Search TRRL's mtts */ - mtts_3 = hns_roce_table_find(hr_dev, &hr_dev->qp_table.trrl_table, - hr_qp->qpn, &dma_handle_3); - if (!mtts_3) { - ibdev_err(ibdev, "failed to find QP trrl_table\n"); - return -EINVAL; - } - - if (attr_mask & IB_QP_ALT_PATH) { - ibdev_err(ibdev, "INIT2RTR attr_mask (0x%x) error\n", - attr_mask); + if (hr_qp->rq.wqe_cnt && count < 1) { + ibdev_err(&hr_dev->ib_dev, + "failed to find RQ WQE, QPN = 0x%lx.\n", hr_qp->qpn); return -EINVAL; } - dmac = (u8 *)attr->ah_attr.roce.dmac; context->wqe_sge_ba = cpu_to_le32(wqe_sge_ba >> 3); qpc_mask->wqe_sge_ba = 0; @@ -3804,17 +3788,16 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, roce_set_field(context->byte_12_sq_hop, V2_QPC_BYTE_12_SQ_HOP_NUM_M, V2_QPC_BYTE_12_SQ_HOP_NUM_S, - hr_dev->caps.wqe_sq_hop_num == HNS_ROCE_HOP_NUM_0 ? - 0 : hr_dev->caps.wqe_sq_hop_num); + to_hr_hem_hopnum(hr_dev->caps.wqe_sq_hop_num, + hr_qp->sq.wqe_cnt)); roce_set_field(qpc_mask->byte_12_sq_hop, V2_QPC_BYTE_12_SQ_HOP_NUM_M, V2_QPC_BYTE_12_SQ_HOP_NUM_S, 0); roce_set_field(context->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_SGE_HOP_NUM_M, V2_QPC_BYTE_20_SGE_HOP_NUM_S, - ((ibqp->qp_type == IB_QPT_GSI) || - hr_qp->sq.max_gs > HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) ? - hr_dev->caps.wqe_sge_hop_num : 0); + to_hr_hem_hopnum(hr_dev->caps.wqe_sge_hop_num, + hr_qp->sge.sge_cnt)); roce_set_field(qpc_mask->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_SGE_HOP_NUM_M, V2_QPC_BYTE_20_SGE_HOP_NUM_S, 0); @@ -3822,8 +3805,9 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, roce_set_field(context->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_HOP_NUM_M, V2_QPC_BYTE_20_RQ_HOP_NUM_S, - hr_dev->caps.wqe_rq_hop_num == HNS_ROCE_HOP_NUM_0 ? - 0 : hr_dev->caps.wqe_rq_hop_num); + to_hr_hem_hopnum(hr_dev->caps.wqe_rq_hop_num, + hr_qp->rq.wqe_cnt)); + roce_set_field(qpc_mask->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_HOP_NUM_M, V2_QPC_BYTE_20_RQ_HOP_NUM_S, 0); @@ -3831,7 +3815,7 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, roce_set_field(context->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_M, V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_S, - hr_qp->wqe_bt_pg_shift + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(hr_qp->mtr.hem_cfg.ba_pg_shift)); roce_set_field(qpc_mask->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_M, V2_QPC_BYTE_16_WQE_SGE_BA_PG_SZ_S, 0); @@ -3839,50 +3823,189 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, roce_set_field(context->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_WQE_SGE_BUF_PG_SZ_M, V2_QPC_BYTE_16_WQE_SGE_BUF_PG_SZ_S, - hr_dev->caps.mtt_buf_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(hr_qp->mtr.hem_cfg.buf_pg_shift)); roce_set_field(qpc_mask->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_WQE_SGE_BUF_PG_SZ_M, V2_QPC_BYTE_16_WQE_SGE_BUF_PG_SZ_S, 0); - context->rq_cur_blk_addr = cpu_to_le32(mtts[0] >> PAGE_ADDR_SHIFT); + context->rq_cur_blk_addr = cpu_to_le32(to_hr_hw_page_addr(mtts[0])); qpc_mask->rq_cur_blk_addr = 0; roce_set_field(context->byte_92_srq_info, V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_M, V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_S, - mtts[0] >> (32 + PAGE_ADDR_SHIFT)); + upper_32_bits(to_hr_hw_page_addr(mtts[0]))); roce_set_field(qpc_mask->byte_92_srq_info, V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_M, V2_QPC_BYTE_92_RQ_CUR_BLK_ADDR_S, 0); - context->rq_nxt_blk_addr = cpu_to_le32(mtts[1] >> PAGE_ADDR_SHIFT); + context->rq_nxt_blk_addr = cpu_to_le32(to_hr_hw_page_addr(mtts[1])); qpc_mask->rq_nxt_blk_addr = 0; roce_set_field(context->byte_104_rq_sge, V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_M, V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_S, - mtts[1] >> (32 + PAGE_ADDR_SHIFT)); + upper_32_bits(to_hr_hw_page_addr(mtts[1]))); roce_set_field(qpc_mask->byte_104_rq_sge, V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_M, V2_QPC_BYTE_104_RQ_NXT_BLK_ADDR_S, 0); + roce_set_field(context->byte_84_rq_ci_pi, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, hr_qp->rq.head); + roce_set_field(qpc_mask->byte_84_rq_ci_pi, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, 0); + + roce_set_field(qpc_mask->byte_84_rq_ci_pi, + V2_QPC_BYTE_84_RQ_CONSUMER_IDX_M, + V2_QPC_BYTE_84_RQ_CONSUMER_IDX_S, 0); + + return 0; +} + +static int config_qp_sq_buf(struct hns_roce_dev *hr_dev, + struct hns_roce_qp *hr_qp, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) +{ + struct ib_device *ibdev = &hr_dev->ib_dev; + u64 sge_cur_blk = 0; + u64 sq_cur_blk = 0; + int count; + + /* search qp buf's mtts */ + count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, 0, &sq_cur_blk, 1, NULL); + if (count < 1) { + ibdev_err(ibdev, "failed to find QP(0x%lx) SQ buf.\n", + hr_qp->qpn); + return -EINVAL; + } + if (hr_qp->sge.sge_cnt > 0) { + count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, + hr_qp->sge.offset, + &sge_cur_blk, 1, NULL); + if (count < 1) { + ibdev_err(ibdev, "failed to find QP(0x%lx) SGE buf.\n", + hr_qp->qpn); + return -EINVAL; + } + } + + /* + * In v2 engine, software pass context and context mask to hardware + * when modifying qp. If software need modify some fields in context, + * we should set all bits of the relevant fields in context mask to + * 0 at the same time, else set them to 0x1. + */ + context->sq_cur_blk_addr = cpu_to_le32(to_hr_hw_page_addr(sq_cur_blk)); + roce_set_field(context->byte_168_irrl_idx, + V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_M, + V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_S, + upper_32_bits(to_hr_hw_page_addr(sq_cur_blk))); + qpc_mask->sq_cur_blk_addr = 0; + roce_set_field(qpc_mask->byte_168_irrl_idx, + V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_M, + V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_S, 0); + + context->sq_cur_sge_blk_addr = + cpu_to_le32(to_hr_hw_page_addr(sge_cur_blk)); + roce_set_field(context->byte_184_irrl_idx, + V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_M, + V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_S, + upper_32_bits(to_hr_hw_page_addr(sge_cur_blk))); + qpc_mask->sq_cur_sge_blk_addr = 0; + roce_set_field(qpc_mask->byte_184_irrl_idx, + V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_M, + V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_S, 0); + + context->rx_sq_cur_blk_addr = + cpu_to_le32(to_hr_hw_page_addr(sq_cur_blk)); + roce_set_field(context->byte_232_irrl_sge, + V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_M, + V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_S, + upper_32_bits(to_hr_hw_page_addr(sq_cur_blk))); + qpc_mask->rx_sq_cur_blk_addr = 0; + roce_set_field(qpc_mask->byte_232_irrl_sge, + V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_M, + V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_S, 0); + + return 0; +} + +static inline enum ib_mtu get_mtu(struct ib_qp *ibqp, + const struct ib_qp_attr *attr) +{ + if (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_UD) + return IB_MTU_4096; + + return attr->path_mtu; +} + +static int modify_qp_init_to_rtr(struct ib_qp *ibqp, + const struct ib_qp_attr *attr, int attr_mask, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) +{ + const struct ib_global_route *grh = rdma_ah_read_grh(&attr->ah_attr); + struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); + struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); + struct ib_device *ibdev = &hr_dev->ib_dev; + dma_addr_t trrl_ba; + dma_addr_t irrl_ba; + enum ib_mtu mtu; + u8 port_num; + u64 *mtts; + u8 *dmac; + u8 *smac; + int port; + int ret; + + ret = config_qp_rq_buf(hr_dev, hr_qp, context, qpc_mask); + if (ret) { + ibdev_err(ibdev, "failed to config rq buf, ret = %d.\n", ret); + return ret; + } + + /* Search IRRL's mtts */ + mtts = hns_roce_table_find(hr_dev, &hr_dev->qp_table.irrl_table, + hr_qp->qpn, &irrl_ba); + if (!mtts) { + ibdev_err(ibdev, "failed to find qp irrl_table.\n"); + return -EINVAL; + } + + /* Search TRRL's mtts */ + mtts = hns_roce_table_find(hr_dev, &hr_dev->qp_table.trrl_table, + hr_qp->qpn, &trrl_ba); + if (!mtts) { + ibdev_err(ibdev, "failed to find qp trrl_table.\n"); + return -EINVAL; + } + + if (attr_mask & IB_QP_ALT_PATH) { + ibdev_err(ibdev, "INIT2RTR attr_mask (0x%x) error.\n", + attr_mask); + return -EINVAL; + } + roce_set_field(context->byte_132_trrl, V2_QPC_BYTE_132_TRRL_BA_M, - V2_QPC_BYTE_132_TRRL_BA_S, dma_handle_3 >> 4); + V2_QPC_BYTE_132_TRRL_BA_S, trrl_ba >> 4); roce_set_field(qpc_mask->byte_132_trrl, V2_QPC_BYTE_132_TRRL_BA_M, V2_QPC_BYTE_132_TRRL_BA_S, 0); - context->trrl_ba = cpu_to_le32(dma_handle_3 >> (16 + 4)); + context->trrl_ba = cpu_to_le32(trrl_ba >> (16 + 4)); qpc_mask->trrl_ba = 0; roce_set_field(context->byte_140_raq, V2_QPC_BYTE_140_TRRL_BA_M, V2_QPC_BYTE_140_TRRL_BA_S, - (u32)(dma_handle_3 >> (32 + 16 + 4))); + (u32)(trrl_ba >> (32 + 16 + 4))); roce_set_field(qpc_mask->byte_140_raq, V2_QPC_BYTE_140_TRRL_BA_M, V2_QPC_BYTE_140_TRRL_BA_S, 0); - context->irrl_ba = cpu_to_le32(dma_handle_2 >> 6); + context->irrl_ba = cpu_to_le32(irrl_ba >> 6); qpc_mask->irrl_ba = 0; roce_set_field(context->byte_208_irrl, V2_QPC_BYTE_208_IRRL_BA_M, V2_QPC_BYTE_208_IRRL_BA_S, - dma_handle_2 >> (32 + 6)); + irrl_ba >> (32 + 6)); roce_set_field(qpc_mask->byte_208_irrl, V2_QPC_BYTE_208_IRRL_BA_M, V2_QPC_BYTE_208_IRRL_BA_S, 0); @@ -3897,6 +4020,7 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, port = (attr_mask & IB_QP_PORT) ? (attr->port_num - 1) : hr_qp->port; smac = (u8 *)hr_dev->dev_addr[port]; + dmac = (u8 *)attr->ah_attr.roce.dmac; /* when dmac equals smac or loop_idc is 1, it should loopback */ if (ether_addr_equal_unaligned(dmac, smac) || hr_dev->loop_idc == 0x1) { @@ -3919,6 +4043,7 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, grh->sgid_index)); roce_set_field(qpc_mask->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S, 0); + memcpy(&(context->dmac), dmac, sizeof(u32)); roce_set_field(context->byte_52_udpspn_dmac, V2_QPC_BYTE_52_DMAC_M, V2_QPC_BYTE_52_DMAC_S, *((u16 *)(&dmac[4]))); @@ -3926,32 +4051,23 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, roce_set_field(qpc_mask->byte_52_udpspn_dmac, V2_QPC_BYTE_52_DMAC_M, V2_QPC_BYTE_52_DMAC_S, 0); - /* mtu*(2^LP_PKTN_INI) should not bigger than 1 message length 64kb */ - roce_set_field(context->byte_56_dqpn_err, V2_QPC_BYTE_56_LP_PKTN_INI_M, - V2_QPC_BYTE_56_LP_PKTN_INI_S, 0); - roce_set_field(qpc_mask->byte_56_dqpn_err, V2_QPC_BYTE_56_LP_PKTN_INI_M, - V2_QPC_BYTE_56_LP_PKTN_INI_S, 0); + mtu = get_mtu(ibqp, attr); - if (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_UD) - roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_MTU_M, - V2_QPC_BYTE_24_MTU_S, IB_MTU_4096); - else if (attr_mask & IB_QP_PATH_MTU) + if (attr_mask & IB_QP_PATH_MTU) { roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_MTU_M, - V2_QPC_BYTE_24_MTU_S, attr->path_mtu); - - roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_MTU_M, - V2_QPC_BYTE_24_MTU_S, 0); + V2_QPC_BYTE_24_MTU_S, mtu); + roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_MTU_M, + V2_QPC_BYTE_24_MTU_S, 0); + } - roce_set_field(context->byte_84_rq_ci_pi, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, hr_qp->rq.head); - roce_set_field(qpc_mask->byte_84_rq_ci_pi, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, 0); +#define MAX_LP_MSG_LEN 65536 + /* MTU*(2^LP_PKTN_INI) shouldn't be bigger than 64kb */ + roce_set_field(context->byte_56_dqpn_err, V2_QPC_BYTE_56_LP_PKTN_INI_M, + V2_QPC_BYTE_56_LP_PKTN_INI_S, + ilog2(MAX_LP_MSG_LEN / ib_mtu_enum_to_int(mtu))); + roce_set_field(qpc_mask->byte_56_dqpn_err, V2_QPC_BYTE_56_LP_PKTN_INI_M, + V2_QPC_BYTE_56_LP_PKTN_INI_S, 0); - roce_set_field(qpc_mask->byte_84_rq_ci_pi, - V2_QPC_BYTE_84_RQ_CONSUMER_IDX_M, - V2_QPC_BYTE_84_RQ_CONSUMER_IDX_S, 0); roce_set_bit(qpc_mask->byte_108_rx_reqepsn, V2_QPC_BYTE_108_RX_REQ_PSN_ERR_S, 0); roce_set_field(qpc_mask->byte_96_rx_reqmsn, V2_QPC_BYTE_96_RX_REQ_MSN_M, @@ -3987,30 +4103,7 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp, struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); struct ib_device *ibdev = &hr_dev->ib_dev; - u64 sge_cur_blk = 0; - u64 sq_cur_blk = 0; - u32 page_size; - int count; - - /* Search qp buf's mtts */ - count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, 0, &sq_cur_blk, 1, NULL); - if (count < 1) { - ibdev_err(ibdev, "failed to find buf pa of QP(0x%lx)\n", - hr_qp->qpn); - return -EINVAL; - } - - if (hr_qp->sge.offset) { - page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT); - count = hns_roce_mtr_find(hr_dev, &hr_qp->mtr, - hr_qp->sge.offset / page_size, - &sge_cur_blk, 1, NULL); - if (count < 1) { - ibdev_err(ibdev, "failed to find sge pa of QP(0x%lx)\n", - hr_qp->qpn); - return -EINVAL; - } - } + int ret; /* Not support alternate path and path migration */ if (attr_mask & (IB_QP_ALT_PATH | IB_QP_PATH_MIG_STATE)) { @@ -4018,48 +4111,11 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp, return -EINVAL; } - /* - * In v2 engine, software pass context and context mask to hardware - * when modifying qp. If software need modify some fields in context, - * we should set all bits of the relevant fields in context mask to - * 0 at the same time, else set them to 0x1. - */ - context->sq_cur_blk_addr = cpu_to_le32(sq_cur_blk >> PAGE_ADDR_SHIFT); - roce_set_field(context->byte_168_irrl_idx, - V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_M, - V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_S, - sq_cur_blk >> (32 + PAGE_ADDR_SHIFT)); - qpc_mask->sq_cur_blk_addr = 0; - roce_set_field(qpc_mask->byte_168_irrl_idx, - V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_M, - V2_QPC_BYTE_168_SQ_CUR_BLK_ADDR_S, 0); - - context->sq_cur_sge_blk_addr = ((ibqp->qp_type == IB_QPT_GSI) || - hr_qp->sq.max_gs > HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) ? - cpu_to_le32(sge_cur_blk >> - PAGE_ADDR_SHIFT) : 0; - roce_set_field(context->byte_184_irrl_idx, - V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_M, - V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_S, - ((ibqp->qp_type == IB_QPT_GSI) || hr_qp->sq.max_gs > - HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE) ? - (sge_cur_blk >> - (32 + PAGE_ADDR_SHIFT)) : 0); - qpc_mask->sq_cur_sge_blk_addr = 0; - roce_set_field(qpc_mask->byte_184_irrl_idx, - V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_M, - V2_QPC_BYTE_184_SQ_CUR_SGE_BLK_ADDR_S, 0); - - context->rx_sq_cur_blk_addr = - cpu_to_le32(sq_cur_blk >> PAGE_ADDR_SHIFT); - roce_set_field(context->byte_232_irrl_sge, - V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_M, - V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_S, - sq_cur_blk >> (32 + PAGE_ADDR_SHIFT)); - qpc_mask->rx_sq_cur_blk_addr = 0; - roce_set_field(qpc_mask->byte_232_irrl_sge, - V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_M, - V2_QPC_BYTE_232_RX_SQ_CUR_BLK_ADDR_S, 0); + ret = config_qp_sq_buf(hr_dev, hr_qp, context, qpc_mask); + if (ret) { + ibdev_err(ibdev, "failed to config sq buf, ret %d\n", ret); + return ret; + } /* * Set some fields in context to zero, Because the default values @@ -4108,21 +4164,6 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp, return 0; } -static inline bool hns_roce_v2_check_qp_stat(enum ib_qp_state cur_state, - enum ib_qp_state new_state) -{ - - if ((cur_state != IB_QPS_RESET && - (new_state == IB_QPS_ERR || new_state == IB_QPS_RESET)) || - ((cur_state == IB_QPS_RTS || cur_state == IB_QPS_SQD) && - (new_state == IB_QPS_RTS || new_state == IB_QPS_SQD)) || - (cur_state == IB_QPS_SQE && new_state == IB_QPS_RTS)) - return true; - - return false; - -} - static int hns_roce_v2_set_path(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, @@ -4203,12 +4244,13 @@ static int hns_roce_v2_set_path(struct ib_qp *ibqp, roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_HOP_LIMIT_M, V2_QPC_BYTE_24_HOP_LIMIT_S, 0); - if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP08_B && is_udp) + if (is_udp) roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, V2_QPC_BYTE_24_TC_S, grh->traffic_class >> 2); else roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, V2_QPC_BYTE_24_TC_S, grh->traffic_class); + roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, V2_QPC_BYTE_24_TC_S, 0); roce_set_field(context->byte_28_at_fl, V2_QPC_BYTE_28_FL_M, @@ -4226,6 +4268,30 @@ static int hns_roce_v2_set_path(struct ib_qp *ibqp, return 0; } +static bool check_qp_state(enum ib_qp_state cur_state, + enum ib_qp_state new_state) +{ + static const bool sm[][IB_QPS_ERR + 1] = { + [IB_QPS_RESET] = { [IB_QPS_RESET] = true, + [IB_QPS_INIT] = true }, + [IB_QPS_INIT] = { [IB_QPS_RESET] = true, + [IB_QPS_INIT] = true, + [IB_QPS_RTR] = true, + [IB_QPS_ERR] = true }, + [IB_QPS_RTR] = { [IB_QPS_RESET] = true, + [IB_QPS_RTS] = true, + [IB_QPS_ERR] = true }, + [IB_QPS_RTS] = { [IB_QPS_RESET] = true, + [IB_QPS_RTS] = true, + [IB_QPS_ERR] = true }, + [IB_QPS_SQD] = {}, + [IB_QPS_SQE] = {}, + [IB_QPS_ERR] = { [IB_QPS_RESET] = true, [IB_QPS_ERR] = true } + }; + + return sm[cur_state][new_state]; +} + static int hns_roce_v2_set_abs_fields(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, @@ -4237,6 +4303,11 @@ static int hns_roce_v2_set_abs_fields(struct ib_qp *ibqp, struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); int ret = 0; + if (!check_qp_state(cur_state, new_state)) { + ibdev_err(&hr_dev->ib_dev, "Illegal state for QP!\n"); + return -EINVAL; + } + if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { memset(qpc_mask, 0, sizeof(*qpc_mask)); modify_qp_reset_to_init(ibqp, attr, attr_mask, context, @@ -4247,23 +4318,11 @@ static int hns_roce_v2_set_abs_fields(struct ib_qp *ibqp, } else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { ret = modify_qp_init_to_rtr(ibqp, attr, attr_mask, context, qpc_mask); - if (ret) - goto out; } else if (cur_state == IB_QPS_RTR && new_state == IB_QPS_RTS) { ret = modify_qp_rtr_to_rts(ibqp, attr, attr_mask, context, qpc_mask); - if (ret) - goto out; - } else if (hns_roce_v2_check_qp_stat(cur_state, new_state)) { - /* Nothing */ - ; - } else { - ibdev_err(&hr_dev->ib_dev, "Illegal state for QP!\n"); - ret = -EINVAL; - goto out; } -out: return ret; } @@ -4554,19 +4613,20 @@ out: return ret; } -static inline enum ib_qp_state to_ib_qp_st(enum hns_roce_v2_qp_state state) +static int to_ib_qp_st(enum hns_roce_v2_qp_state state) { - switch (state) { - case HNS_ROCE_QP_ST_RST: return IB_QPS_RESET; - case HNS_ROCE_QP_ST_INIT: return IB_QPS_INIT; - case HNS_ROCE_QP_ST_RTR: return IB_QPS_RTR; - case HNS_ROCE_QP_ST_RTS: return IB_QPS_RTS; - case HNS_ROCE_QP_ST_SQ_DRAINING: - case HNS_ROCE_QP_ST_SQD: return IB_QPS_SQD; - case HNS_ROCE_QP_ST_SQER: return IB_QPS_SQE; - case HNS_ROCE_QP_ST_ERR: return IB_QPS_ERR; - default: return -1; - } + static const enum ib_qp_state map[] = { + [HNS_ROCE_QP_ST_RST] = IB_QPS_RESET, + [HNS_ROCE_QP_ST_INIT] = IB_QPS_INIT, + [HNS_ROCE_QP_ST_RTR] = IB_QPS_RTR, + [HNS_ROCE_QP_ST_RTS] = IB_QPS_RTS, + [HNS_ROCE_QP_ST_SQD] = IB_QPS_SQD, + [HNS_ROCE_QP_ST_SQER] = IB_QPS_SQE, + [HNS_ROCE_QP_ST_ERR] = IB_QPS_ERR, + [HNS_ROCE_QP_ST_SQ_DRAINING] = IB_QPS_SQD + }; + + return (state < ARRAY_SIZE(map)) ? map[state] : -1; } static int hns_roce_v2_query_qpc(struct hns_roce_dev *hr_dev, @@ -4639,7 +4699,7 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, qp_attr->path_mig_state = IB_MIG_ARMED; qp_attr->ah_attr.type = RDMA_AH_ATTR_TYPE_ROCE; if (hr_qp->ibqp.qp_type == IB_QPT_UD) - qp_attr->qkey = V2_QKEY_VAL; + qp_attr->qkey = le32_to_cpu(context.qkey_xrcd); qp_attr->rq_psn = roce_get_field(context.byte_108_rx_reqepsn, V2_QPC_BYTE_108_RX_REQ_EPSN_M, @@ -4838,6 +4898,184 @@ out: return ret; } +static void hns_roce_v2_write_srqc(struct hns_roce_dev *hr_dev, + struct hns_roce_srq *srq, u32 pdn, u16 xrcd, + u32 cqn, void *mb_buf, u64 *mtts_wqe, + u64 *mtts_idx, dma_addr_t dma_handle_wqe, + dma_addr_t dma_handle_idx) +{ + struct hns_roce_srq_context *srq_context; + + srq_context = mb_buf; + memset(srq_context, 0, sizeof(*srq_context)); + + roce_set_field(srq_context->byte_4_srqn_srqst, SRQC_BYTE_4_SRQ_ST_M, + SRQC_BYTE_4_SRQ_ST_S, 1); + + roce_set_field(srq_context->byte_4_srqn_srqst, + SRQC_BYTE_4_SRQ_WQE_HOP_NUM_M, + SRQC_BYTE_4_SRQ_WQE_HOP_NUM_S, + to_hr_hem_hopnum(hr_dev->caps.srqwqe_hop_num, + srq->wqe_cnt)); + roce_set_field(srq_context->byte_4_srqn_srqst, + SRQC_BYTE_4_SRQ_SHIFT_M, SRQC_BYTE_4_SRQ_SHIFT_S, + ilog2(srq->wqe_cnt)); + + roce_set_field(srq_context->byte_4_srqn_srqst, SRQC_BYTE_4_SRQN_M, + SRQC_BYTE_4_SRQN_S, srq->srqn); + + roce_set_field(srq_context->byte_8_limit_wl, SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S, 0); + + roce_set_field(srq_context->byte_12_xrcd, SRQC_BYTE_12_SRQ_XRCD_M, + SRQC_BYTE_12_SRQ_XRCD_S, xrcd); + + srq_context->wqe_bt_ba = cpu_to_le32((u32)(dma_handle_wqe >> 3)); + + roce_set_field(srq_context->byte_24_wqe_bt_ba, + SRQC_BYTE_24_SRQ_WQE_BT_BA_M, + SRQC_BYTE_24_SRQ_WQE_BT_BA_S, + dma_handle_wqe >> 35); + + roce_set_field(srq_context->byte_28_rqws_pd, SRQC_BYTE_28_PD_M, + SRQC_BYTE_28_PD_S, pdn); + roce_set_field(srq_context->byte_28_rqws_pd, SRQC_BYTE_28_RQWS_M, + SRQC_BYTE_28_RQWS_S, srq->max_gs <= 0 ? 0 : + fls(srq->max_gs - 1)); + + srq_context->idx_bt_ba = cpu_to_le32(dma_handle_idx >> 3); + roce_set_field(srq_context->rsv_idx_bt_ba, + SRQC_BYTE_36_SRQ_IDX_BT_BA_M, + SRQC_BYTE_36_SRQ_IDX_BT_BA_S, + dma_handle_idx >> 35); + + srq_context->idx_cur_blk_addr = + cpu_to_le32(to_hr_hw_page_addr(mtts_idx[0])); + roce_set_field(srq_context->byte_44_idxbufpgsz_addr, + SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_M, + SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_S, + upper_32_bits(to_hr_hw_page_addr(mtts_idx[0]))); + roce_set_field(srq_context->byte_44_idxbufpgsz_addr, + SRQC_BYTE_44_SRQ_IDX_HOP_NUM_M, + SRQC_BYTE_44_SRQ_IDX_HOP_NUM_S, + to_hr_hem_hopnum(hr_dev->caps.idx_hop_num, + srq->wqe_cnt)); + + roce_set_field(srq_context->byte_44_idxbufpgsz_addr, + SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_M, + SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_S, + to_hr_hw_page_shift(srq->idx_que.mtr.hem_cfg.ba_pg_shift)); + roce_set_field(srq_context->byte_44_idxbufpgsz_addr, + SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_M, + SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_S, + to_hr_hw_page_shift(srq->idx_que.mtr.hem_cfg.buf_pg_shift)); + + srq_context->idx_nxt_blk_addr = + cpu_to_le32(to_hr_hw_page_addr(mtts_idx[1])); + roce_set_field(srq_context->rsv_idxnxtblkaddr, + SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_M, + SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_S, + upper_32_bits(to_hr_hw_page_addr(mtts_idx[1]))); + roce_set_field(srq_context->byte_56_xrc_cqn, + SRQC_BYTE_56_SRQ_XRC_CQN_M, SRQC_BYTE_56_SRQ_XRC_CQN_S, + cqn); + roce_set_field(srq_context->byte_56_xrc_cqn, + SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_M, + SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_S, + to_hr_hw_page_shift(srq->buf_mtr.hem_cfg.ba_pg_shift)); + roce_set_field(srq_context->byte_56_xrc_cqn, + SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_M, + SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_S, + to_hr_hw_page_shift(srq->buf_mtr.hem_cfg.buf_pg_shift)); + + roce_set_bit(srq_context->db_record_addr_record_en, + SRQC_BYTE_60_SRQ_RECORD_EN_S, 0); +} + +static int hns_roce_v2_modify_srq(struct ib_srq *ibsrq, + struct ib_srq_attr *srq_attr, + enum ib_srq_attr_mask srq_attr_mask, + struct ib_udata *udata) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); + struct hns_roce_srq *srq = to_hr_srq(ibsrq); + struct hns_roce_srq_context *srq_context; + struct hns_roce_srq_context *srqc_mask; + struct hns_roce_cmd_mailbox *mailbox; + int ret; + + if (srq_attr_mask & IB_SRQ_LIMIT) { + if (srq_attr->srq_limit >= srq->wqe_cnt) + return -EINVAL; + + mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + srq_context = mailbox->buf; + srqc_mask = (struct hns_roce_srq_context *)mailbox->buf + 1; + + memset(srqc_mask, 0xff, sizeof(*srqc_mask)); + + roce_set_field(srq_context->byte_8_limit_wl, + SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S, srq_attr->srq_limit); + roce_set_field(srqc_mask->byte_8_limit_wl, + SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S, 0); + + ret = hns_roce_cmd_mbox(hr_dev, mailbox->dma, 0, srq->srqn, 0, + HNS_ROCE_CMD_MODIFY_SRQC, + HNS_ROCE_CMD_TIMEOUT_MSECS); + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + if (ret) { + ibdev_err(&hr_dev->ib_dev, + "failed to handle cmd of modifying SRQ, ret = %d.\n", + ret); + return ret; + } + } + + return 0; +} + +static int hns_roce_v2_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); + struct hns_roce_srq *srq = to_hr_srq(ibsrq); + struct hns_roce_srq_context *srq_context; + struct hns_roce_cmd_mailbox *mailbox; + int limit_wl; + int ret; + + mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + srq_context = mailbox->buf; + ret = hns_roce_cmd_mbox(hr_dev, 0, mailbox->dma, srq->srqn, 0, + HNS_ROCE_CMD_QUERY_SRQC, + HNS_ROCE_CMD_TIMEOUT_MSECS); + if (ret) { + ibdev_err(&hr_dev->ib_dev, + "failed to process cmd of querying SRQ, ret = %d.\n", + ret); + goto out; + } + + limit_wl = roce_get_field(srq_context->byte_8_limit_wl, + SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S); + + attr->srq_limit = limit_wl; + attr->max_wr = srq->wqe_cnt - 1; + attr->max_sge = srq->max_gs; + +out: + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + return ret; +} + static int hns_roce_v2_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period) { struct hns_roce_dev *hr_dev = to_hr_dev(cq->device); @@ -4989,24 +5227,14 @@ static void set_eq_cons_index_v2(struct hns_roce_eq *eq) hns_roce_write64(hr_dev, doorbell, eq->doorbell); } -static inline void *get_eqe_buf(struct hns_roce_eq *eq, unsigned long offset) -{ - u32 buf_chk_sz; - - buf_chk_sz = 1 << (eq->eqe_buf_pg_sz + PAGE_SHIFT); - if (eq->buf.nbufs == 1) - return eq->buf.direct.buf + offset % buf_chk_sz; - else - return eq->buf.page_list[offset / buf_chk_sz].buf + - offset % buf_chk_sz; -} - static struct hns_roce_aeqe *next_aeqe_sw_v2(struct hns_roce_eq *eq) { struct hns_roce_aeqe *aeqe; - aeqe = get_eqe_buf(eq, (eq->cons_index & (eq->entries - 1)) * - HNS_ROCE_AEQ_ENTRY_SIZE); + aeqe = hns_roce_buf_offset(eq->mtr.kmem, + (eq->cons_index & (eq->entries - 1)) * + HNS_ROCE_AEQ_ENTRY_SIZE); + return (roce_get_bit(aeqe->asyn, HNS_ROCE_V2_AEQ_AEQE_OWNER_S) ^ !!(eq->cons_index & eq->entries)) ? aeqe : NULL; } @@ -5103,8 +5331,9 @@ static struct hns_roce_ceqe *next_ceqe_sw_v2(struct hns_roce_eq *eq) { struct hns_roce_ceqe *ceqe; - ceqe = get_eqe_buf(eq, (eq->cons_index & (eq->entries - 1)) * - HNS_ROCE_CEQ_ENTRY_SIZE); + ceqe = hns_roce_buf_offset(eq->mtr.kmem, + (eq->cons_index & (eq->entries - 1)) * + HNS_ROCE_CEQ_ENTRY_SIZE); return (!!(roce_get_bit(ceqe->comp, HNS_ROCE_V2_CEQ_CEQE_OWNER_S))) ^ (!!(eq->cons_index & eq->entries)) ? ceqe : NULL; } @@ -5263,17 +5492,15 @@ static void hns_roce_v2_destroy_eqc(struct hns_roce_dev *hr_dev, int eqn) static void free_eq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq) { - if (!eq->hop_num || eq->hop_num == HNS_ROCE_HOP_NUM_0) - hns_roce_mtr_cleanup(hr_dev, &eq->mtr); - hns_roce_buf_free(hr_dev, eq->buf.size, &eq->buf); + hns_roce_mtr_destroy(hr_dev, &eq->mtr); } -static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev, - struct hns_roce_eq *eq, - void *mb_buf) +static int config_eqc(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq, + void *mb_buf) { + u64 eqe_ba[MTT_MIN_COUNT] = { 0 }; struct hns_roce_eq_context *eqc; - u64 ba[MTT_MIN_COUNT] = { 0 }; + u64 bt_ba = 0; int count; eqc = mb_buf; @@ -5281,31 +5508,18 @@ static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev, /* init eqc */ eq->doorbell = hr_dev->reg_base + ROCEE_VF_EQ_DB_CFG0_REG; - eq->hop_num = hr_dev->caps.eqe_hop_num; eq->cons_index = 0; eq->over_ignore = HNS_ROCE_V2_EQ_OVER_IGNORE_0; eq->coalesce = HNS_ROCE_V2_EQ_COALESCE_0; eq->arm_st = HNS_ROCE_V2_EQ_ALWAYS_ARMED; - eq->eqe_ba_pg_sz = hr_dev->caps.eqe_ba_pg_sz; - eq->eqe_buf_pg_sz = hr_dev->caps.eqe_buf_pg_sz; eq->shift = ilog2((unsigned int)eq->entries); - /* if not muti-hop, eqe buffer only use one trunk */ - if (!eq->hop_num || eq->hop_num == HNS_ROCE_HOP_NUM_0) { - eq->eqe_ba = eq->buf.direct.map; - eq->cur_eqe_ba = eq->eqe_ba; - if (eq->buf.npages > 1) - eq->nxt_eqe_ba = eq->eqe_ba + (1 << eq->eqe_buf_pg_sz); - else - eq->nxt_eqe_ba = eq->eqe_ba; - } else { - count = hns_roce_mtr_find(hr_dev, &eq->mtr, 0, ba, - MTT_MIN_COUNT, &eq->eqe_ba); - eq->cur_eqe_ba = ba[0]; - if (count > 1) - eq->nxt_eqe_ba = ba[1]; - else - eq->nxt_eqe_ba = ba[0]; + /* if not multi-hop, eqe buffer only use one trunk */ + count = hns_roce_mtr_find(hr_dev, &eq->mtr, 0, eqe_ba, MTT_MIN_COUNT, + &bt_ba); + if (count < 1) { + dev_err(hr_dev->dev, "failed to find EQE mtr\n"); + return -ENOBUFS; } /* set eqc state */ @@ -5339,12 +5553,12 @@ static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev, /* set eqe_ba_pg_sz */ roce_set_field(eqc->byte_8, HNS_ROCE_EQC_BA_PG_SZ_M, HNS_ROCE_EQC_BA_PG_SZ_S, - eq->eqe_ba_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(eq->mtr.hem_cfg.ba_pg_shift)); /* set eqe_buf_pg_sz */ roce_set_field(eqc->byte_8, HNS_ROCE_EQC_BUF_PG_SZ_M, HNS_ROCE_EQC_BUF_PG_SZ_S, - eq->eqe_buf_pg_sz + PG_SHIFT_OFFSET); + to_hr_hw_page_shift(eq->mtr.hem_cfg.buf_pg_shift)); /* set eq_producer_idx */ roce_set_field(eqc->byte_8, HNS_ROCE_EQC_PROD_INDX_M, @@ -5363,13 +5577,13 @@ static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev, HNS_ROCE_EQC_REPORT_TIMER_S, HNS_ROCE_EQ_INIT_REPORT_TIMER); - /* set eqe_ba [34:3] */ + /* set bt_ba [34:3] */ roce_set_field(eqc->eqe_ba0, HNS_ROCE_EQC_EQE_BA_L_M, - HNS_ROCE_EQC_EQE_BA_L_S, eq->eqe_ba >> 3); + HNS_ROCE_EQC_EQE_BA_L_S, bt_ba >> 3); - /* set eqe_ba [64:35] */ + /* set bt_ba [64:35] */ roce_set_field(eqc->eqe_ba1, HNS_ROCE_EQC_EQE_BA_H_M, - HNS_ROCE_EQC_EQE_BA_H_S, eq->eqe_ba >> 35); + HNS_ROCE_EQC_EQE_BA_H_S, bt_ba >> 35); /* set eq shift */ roce_set_field(eqc->byte_28, HNS_ROCE_EQC_SHIFT_M, HNS_ROCE_EQC_SHIFT_S, @@ -5381,15 +5595,15 @@ static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev, /* set cur_eqe_ba [27:12] */ roce_set_field(eqc->byte_28, HNS_ROCE_EQC_CUR_EQE_BA_L_M, - HNS_ROCE_EQC_CUR_EQE_BA_L_S, eq->cur_eqe_ba >> 12); + HNS_ROCE_EQC_CUR_EQE_BA_L_S, eqe_ba[0] >> 12); /* set cur_eqe_ba [59:28] */ roce_set_field(eqc->byte_32, HNS_ROCE_EQC_CUR_EQE_BA_M_M, - HNS_ROCE_EQC_CUR_EQE_BA_M_S, eq->cur_eqe_ba >> 28); + HNS_ROCE_EQC_CUR_EQE_BA_M_S, eqe_ba[0] >> 28); /* set cur_eqe_ba [63:60] */ roce_set_field(eqc->byte_36, HNS_ROCE_EQC_CUR_EQE_BA_H_M, - HNS_ROCE_EQC_CUR_EQE_BA_H_S, eq->cur_eqe_ba >> 60); + HNS_ROCE_EQC_CUR_EQE_BA_H_S, eqe_ba[0] >> 60); /* set eq consumer idx */ roce_set_field(eqc->byte_36, HNS_ROCE_EQC_CONS_INDX_M, @@ -5397,97 +5611,38 @@ static void hns_roce_config_eqc(struct hns_roce_dev *hr_dev, /* set nex_eqe_ba[43:12] */ roce_set_field(eqc->nxt_eqe_ba0, HNS_ROCE_EQC_NXT_EQE_BA_L_M, - HNS_ROCE_EQC_NXT_EQE_BA_L_S, eq->nxt_eqe_ba >> 12); + HNS_ROCE_EQC_NXT_EQE_BA_L_S, eqe_ba[1] >> 12); /* set nex_eqe_ba[63:44] */ roce_set_field(eqc->nxt_eqe_ba1, HNS_ROCE_EQC_NXT_EQE_BA_H_M, - HNS_ROCE_EQC_NXT_EQE_BA_H_S, eq->nxt_eqe_ba >> 44); -} - -static int map_eq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq, - u32 page_shift) -{ - struct hns_roce_buf_region region = {}; - dma_addr_t *buf_list = NULL; - int ba_num; - int ret; - - ba_num = DIV_ROUND_UP(PAGE_ALIGN(eq->entries * eq->eqe_size), - 1 << page_shift); - hns_roce_init_buf_region(®ion, hr_dev->caps.eqe_hop_num, 0, ba_num); - - /* alloc a tmp list for storing eq buf address */ - ret = hns_roce_alloc_buf_list(®ion, &buf_list, 1); - if (ret) { - dev_err(hr_dev->dev, "alloc eq buf_list error\n"); - return ret; - } - - ba_num = hns_roce_get_kmem_bufs(hr_dev, buf_list, region.count, - region.offset, &eq->buf); - if (ba_num != region.count) { - dev_err(hr_dev->dev, "get eqe buf err,expect %d,ret %d.\n", - region.count, ba_num); - ret = -ENOBUFS; - goto done; - } - - hns_roce_mtr_init(&eq->mtr, PAGE_SHIFT + hr_dev->caps.eqe_ba_pg_sz, - page_shift); - ret = hns_roce_mtr_attach(hr_dev, &eq->mtr, &buf_list, ®ion, 1); - if (ret) - dev_err(hr_dev->dev, "mtr attach error for eqe\n"); - - goto done; + HNS_ROCE_EQC_NXT_EQE_BA_H_S, eqe_ba[1] >> 44); - hns_roce_mtr_cleanup(hr_dev, &eq->mtr); -done: - hns_roce_free_buf_list(&buf_list, 1); - - return ret; + return 0; } static int alloc_eq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq) { - struct hns_roce_buf *buf = &eq->buf; - bool is_mhop = false; - u32 page_shift; - u32 mhop_num; - u32 max_size; - int ret; + struct hns_roce_buf_attr buf_attr = {}; + int err; - page_shift = PAGE_SHIFT + hr_dev->caps.eqe_buf_pg_sz; - mhop_num = hr_dev->caps.eqe_hop_num; - if (!mhop_num) { - max_size = 1 << page_shift; - buf->size = max_size; - } else if (mhop_num == HNS_ROCE_HOP_NUM_0) { - max_size = eq->entries * eq->eqe_size; - buf->size = max_size; - } else { - max_size = 1 << page_shift; - buf->size = PAGE_ALIGN(eq->entries * eq->eqe_size); - is_mhop = true; - } + if (hr_dev->caps.eqe_hop_num == HNS_ROCE_HOP_NUM_0) + eq->hop_num = 0; + else + eq->hop_num = hr_dev->caps.eqe_hop_num; - ret = hns_roce_buf_alloc(hr_dev, buf->size, max_size, buf, page_shift); - if (ret) { - dev_err(hr_dev->dev, "alloc eq buf error\n"); - return ret; - } + buf_attr.page_shift = hr_dev->caps.eqe_buf_pg_sz + HNS_HW_PAGE_SHIFT; + buf_attr.region[0].size = eq->entries * eq->eqe_size; + buf_attr.region[0].hopnum = eq->hop_num; + buf_attr.region_count = 1; + buf_attr.fixed_page = true; - if (is_mhop) { - ret = map_eq_buf(hr_dev, eq, page_shift); - if (ret) { - dev_err(hr_dev->dev, "map roce buf error\n"); - goto err_alloc; - } - } + err = hns_roce_mtr_create(hr_dev, &eq->mtr, &buf_attr, + hr_dev->caps.eqe_ba_pg_sz + + HNS_HW_PAGE_SHIFT, NULL, 0); + if (err) + dev_err(hr_dev->dev, "Failed to alloc EQE mtr, err %d\n", err); - return 0; -err_alloc: - hns_roce_buf_free(hr_dev, buf->size, buf); - return ret; + return err; } static int hns_roce_v2_create_eq(struct hns_roce_dev *hr_dev, @@ -5499,15 +5654,16 @@ static int hns_roce_v2_create_eq(struct hns_roce_dev *hr_dev, /* Allocate mailbox memory */ mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); - if (IS_ERR(mailbox)) - return PTR_ERR(mailbox); + if (IS_ERR_OR_NULL(mailbox)) + return -ENOMEM; ret = alloc_eq_buf(hr_dev, eq); - if (ret) { - ret = -ENOMEM; + if (ret) goto free_cmd_mbox; - } - hns_roce_config_eqc(hr_dev, eq, mailbox->buf); + + ret = config_eqc(hr_dev, eq, mailbox->buf); + if (ret) + goto err_cmd_mbox; ret = hns_roce_cmd_mbox(hr_dev, mailbox->dma, 0, eq->eqn, 0, eq_cmd, HNS_ROCE_CMD_TIMEOUT_MSECS); @@ -5731,294 +5887,6 @@ static void hns_roce_v2_cleanup_eq_table(struct hns_roce_dev *hr_dev) destroy_workqueue(hr_dev->irq_workq); } -static void hns_roce_v2_write_srqc(struct hns_roce_dev *hr_dev, - struct hns_roce_srq *srq, u32 pdn, u16 xrcd, - u32 cqn, void *mb_buf, u64 *mtts_wqe, - u64 *mtts_idx, dma_addr_t dma_handle_wqe, - dma_addr_t dma_handle_idx) -{ - struct hns_roce_srq_context *srq_context; - - srq_context = mb_buf; - memset(srq_context, 0, sizeof(*srq_context)); - - roce_set_field(srq_context->byte_4_srqn_srqst, SRQC_BYTE_4_SRQ_ST_M, - SRQC_BYTE_4_SRQ_ST_S, 1); - - roce_set_field(srq_context->byte_4_srqn_srqst, - SRQC_BYTE_4_SRQ_WQE_HOP_NUM_M, - SRQC_BYTE_4_SRQ_WQE_HOP_NUM_S, - (hr_dev->caps.srqwqe_hop_num == HNS_ROCE_HOP_NUM_0 ? 0 : - hr_dev->caps.srqwqe_hop_num)); - roce_set_field(srq_context->byte_4_srqn_srqst, - SRQC_BYTE_4_SRQ_SHIFT_M, SRQC_BYTE_4_SRQ_SHIFT_S, - ilog2(srq->wqe_cnt)); - - roce_set_field(srq_context->byte_4_srqn_srqst, SRQC_BYTE_4_SRQN_M, - SRQC_BYTE_4_SRQN_S, srq->srqn); - - roce_set_field(srq_context->byte_8_limit_wl, SRQC_BYTE_8_SRQ_LIMIT_WL_M, - SRQC_BYTE_8_SRQ_LIMIT_WL_S, 0); - - roce_set_field(srq_context->byte_12_xrcd, SRQC_BYTE_12_SRQ_XRCD_M, - SRQC_BYTE_12_SRQ_XRCD_S, xrcd); - - srq_context->wqe_bt_ba = cpu_to_le32((u32)(dma_handle_wqe >> 3)); - - roce_set_field(srq_context->byte_24_wqe_bt_ba, - SRQC_BYTE_24_SRQ_WQE_BT_BA_M, - SRQC_BYTE_24_SRQ_WQE_BT_BA_S, - dma_handle_wqe >> 35); - - roce_set_field(srq_context->byte_28_rqws_pd, SRQC_BYTE_28_PD_M, - SRQC_BYTE_28_PD_S, pdn); - roce_set_field(srq_context->byte_28_rqws_pd, SRQC_BYTE_28_RQWS_M, - SRQC_BYTE_28_RQWS_S, srq->max_gs <= 0 ? 0 : - fls(srq->max_gs - 1)); - - srq_context->idx_bt_ba = cpu_to_le32(dma_handle_idx >> 3); - roce_set_field(srq_context->rsv_idx_bt_ba, - SRQC_BYTE_36_SRQ_IDX_BT_BA_M, - SRQC_BYTE_36_SRQ_IDX_BT_BA_S, - dma_handle_idx >> 35); - - srq_context->idx_cur_blk_addr = - cpu_to_le32(mtts_idx[0] >> PAGE_ADDR_SHIFT); - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_M, - SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_S, - mtts_idx[0] >> (32 + PAGE_ADDR_SHIFT)); - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_HOP_NUM_M, - SRQC_BYTE_44_SRQ_IDX_HOP_NUM_S, - hr_dev->caps.idx_hop_num == HNS_ROCE_HOP_NUM_0 ? 0 : - hr_dev->caps.idx_hop_num); - - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_M, - SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_S, - hr_dev->caps.idx_ba_pg_sz + PG_SHIFT_OFFSET); - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_M, - SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_S, - hr_dev->caps.idx_buf_pg_sz + PG_SHIFT_OFFSET); - - srq_context->idx_nxt_blk_addr = - cpu_to_le32(mtts_idx[1] >> PAGE_ADDR_SHIFT); - roce_set_field(srq_context->rsv_idxnxtblkaddr, - SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_M, - SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_S, - mtts_idx[1] >> (32 + PAGE_ADDR_SHIFT)); - roce_set_field(srq_context->byte_56_xrc_cqn, - SRQC_BYTE_56_SRQ_XRC_CQN_M, SRQC_BYTE_56_SRQ_XRC_CQN_S, - cqn); - roce_set_field(srq_context->byte_56_xrc_cqn, - SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_M, - SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_S, - hr_dev->caps.srqwqe_ba_pg_sz + PG_SHIFT_OFFSET); - roce_set_field(srq_context->byte_56_xrc_cqn, - SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_M, - SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_S, - hr_dev->caps.srqwqe_buf_pg_sz + PG_SHIFT_OFFSET); - - roce_set_bit(srq_context->db_record_addr_record_en, - SRQC_BYTE_60_SRQ_RECORD_EN_S, 0); -} - -static int hns_roce_v2_modify_srq(struct ib_srq *ibsrq, - struct ib_srq_attr *srq_attr, - enum ib_srq_attr_mask srq_attr_mask, - struct ib_udata *udata) -{ - struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); - struct hns_roce_srq *srq = to_hr_srq(ibsrq); - struct hns_roce_srq_context *srq_context; - struct hns_roce_srq_context *srqc_mask; - struct hns_roce_cmd_mailbox *mailbox; - int ret; - - if (srq_attr_mask & IB_SRQ_LIMIT) { - if (srq_attr->srq_limit >= srq->wqe_cnt) - return -EINVAL; - - mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); - if (IS_ERR(mailbox)) - return PTR_ERR(mailbox); - - srq_context = mailbox->buf; - srqc_mask = (struct hns_roce_srq_context *)mailbox->buf + 1; - - memset(srqc_mask, 0xff, sizeof(*srqc_mask)); - - roce_set_field(srq_context->byte_8_limit_wl, - SRQC_BYTE_8_SRQ_LIMIT_WL_M, - SRQC_BYTE_8_SRQ_LIMIT_WL_S, srq_attr->srq_limit); - roce_set_field(srqc_mask->byte_8_limit_wl, - SRQC_BYTE_8_SRQ_LIMIT_WL_M, - SRQC_BYTE_8_SRQ_LIMIT_WL_S, 0); - - ret = hns_roce_cmd_mbox(hr_dev, mailbox->dma, 0, srq->srqn, 0, - HNS_ROCE_CMD_MODIFY_SRQC, - HNS_ROCE_CMD_TIMEOUT_MSECS); - hns_roce_free_cmd_mailbox(hr_dev, mailbox); - if (ret) { - ibdev_err(&hr_dev->ib_dev, - "failed to process cmd when modifying SRQ, ret = %d\n", - ret); - return ret; - } - } - - return 0; -} - -static int hns_roce_v2_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr) -{ - struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); - struct hns_roce_srq *srq = to_hr_srq(ibsrq); - struct hns_roce_srq_context *srq_context; - struct hns_roce_cmd_mailbox *mailbox; - int limit_wl; - int ret; - - mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); - if (IS_ERR(mailbox)) - return PTR_ERR(mailbox); - - srq_context = mailbox->buf; - ret = hns_roce_cmd_mbox(hr_dev, 0, mailbox->dma, srq->srqn, 0, - HNS_ROCE_CMD_QUERY_SRQC, - HNS_ROCE_CMD_TIMEOUT_MSECS); - if (ret) { - ibdev_err(&hr_dev->ib_dev, - "failed to process cmd when querying SRQ, ret = %d\n", - ret); - goto out; - } - - limit_wl = roce_get_field(srq_context->byte_8_limit_wl, - SRQC_BYTE_8_SRQ_LIMIT_WL_M, - SRQC_BYTE_8_SRQ_LIMIT_WL_S); - - attr->srq_limit = limit_wl; - attr->max_wr = srq->wqe_cnt - 1; - attr->max_sge = srq->max_gs; - - memcpy(srq_context, mailbox->buf, sizeof(*srq_context)); - -out: - hns_roce_free_cmd_mailbox(hr_dev, mailbox); - return ret; -} - -static int find_empty_entry(struct hns_roce_idx_que *idx_que, - unsigned long size) -{ - int wqe_idx; - - if (unlikely(bitmap_full(idx_que->bitmap, size))) - return -ENOSPC; - - wqe_idx = find_first_zero_bit(idx_que->bitmap, size); - - bitmap_set(idx_que->bitmap, wqe_idx, 1); - - return wqe_idx; -} - -static void fill_idx_queue(struct hns_roce_idx_que *idx_que, - int cur_idx, int wqe_idx) -{ - unsigned int *addr; - - addr = (unsigned int *)hns_roce_buf_offset(&idx_que->idx_buf, - cur_idx * idx_que->entry_sz); - *addr = wqe_idx; -} - -static int hns_roce_v2_post_srq_recv(struct ib_srq *ibsrq, - const struct ib_recv_wr *wr, - const struct ib_recv_wr **bad_wr) -{ - struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); - struct hns_roce_srq *srq = to_hr_srq(ibsrq); - struct hns_roce_v2_wqe_data_seg *dseg; - struct hns_roce_v2_db srq_db; - unsigned long flags; - int ret = 0; - int wqe_idx; - void *wqe; - int nreq; - int ind; - int i; - - spin_lock_irqsave(&srq->lock, flags); - - ind = srq->head & (srq->wqe_cnt - 1); - - for (nreq = 0; wr; ++nreq, wr = wr->next) { - if (unlikely(wr->num_sge > srq->max_gs)) { - ret = -EINVAL; - *bad_wr = wr; - break; - } - - if (unlikely(srq->head == srq->tail)) { - ret = -ENOMEM; - *bad_wr = wr; - break; - } - - wqe_idx = find_empty_entry(&srq->idx_que, srq->wqe_cnt); - if (wqe_idx < 0) { - ret = -ENOMEM; - *bad_wr = wr; - break; - } - - fill_idx_queue(&srq->idx_que, ind, wqe_idx); - wqe = get_srq_wqe(srq, wqe_idx); - dseg = (struct hns_roce_v2_wqe_data_seg *)wqe; - - for (i = 0; i < wr->num_sge; ++i) { - dseg[i].len = cpu_to_le32(wr->sg_list[i].length); - dseg[i].lkey = cpu_to_le32(wr->sg_list[i].lkey); - dseg[i].addr = cpu_to_le64(wr->sg_list[i].addr); - } - - if (i < srq->max_gs) { - dseg[i].len = 0; - dseg[i].lkey = cpu_to_le32(0x100); - dseg[i].addr = 0; - } - - srq->wrid[wqe_idx] = wr->wr_id; - ind = (ind + 1) & (srq->wqe_cnt - 1); - } - - if (likely(nreq)) { - srq->head += nreq; - - /* - * Make sure that descriptors are written before - * doorbell record. - */ - wmb(); - - srq_db.byte_4 = - cpu_to_le32(HNS_ROCE_V2_SRQ_DB << V2_DB_BYTE_4_CMD_S | - (srq->srqn & V2_DB_BYTE_4_TAG_M)); - srq_db.parameter = cpu_to_le32(srq->head); - - hns_roce_write64(hr_dev, (__le32 *)&srq_db, srq->db_reg_l); - - } - - spin_unlock_irqrestore(&srq->lock, flags); - - return ret; -} - static const struct hns_roce_dfx_hw hns_roce_dfx_hw_v2 = { .query_cqc_info = hns_roce_v2_query_cqc_info, }; @@ -6161,7 +6029,7 @@ error_failed_kzalloc: static void __hns_roce_hw_v2_uninit_instance(struct hnae3_handle *handle, bool reset) { - struct hns_roce_dev *hr_dev = (struct hns_roce_dev *)handle->priv; + struct hns_roce_dev *hr_dev = handle->priv; if (!hr_dev) return; @@ -6241,7 +6109,7 @@ static int hns_roce_hw_v2_reset_notify_down(struct hnae3_handle *handle) handle->rinfo.reset_state = HNS_ROCE_STATE_RST_DOWN; clear_bit(HNS_ROCE_RST_DIRECT_RETURN, &handle->rinfo.state); - hr_dev = (struct hns_roce_dev *)handle->priv; + hr_dev = handle->priv; if (!hr_dev) return 0; diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h index 82dd9f6f4845..ac29be43b6bd 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h @@ -177,27 +177,11 @@ enum { HNS_ROCE_V2_WQE_OP_ATOM_MSK_FETCH_AND_ADD = 0x9, HNS_ROCE_V2_WQE_OP_FAST_REG_PMR = 0xa, HNS_ROCE_V2_WQE_OP_LOCAL_INV = 0xb, - HNS_ROCE_V2_WQE_OP_BIND_MW_TYPE = 0xc, + HNS_ROCE_V2_WQE_OP_BIND_MW = 0xc, HNS_ROCE_V2_WQE_OP_MASK = 0x1f, }; enum { - HNS_ROCE_SQ_OPCODE_SEND = 0x0, - HNS_ROCE_SQ_OPCODE_SEND_WITH_INV = 0x1, - HNS_ROCE_SQ_OPCODE_SEND_WITH_IMM = 0x2, - HNS_ROCE_SQ_OPCODE_RDMA_WRITE = 0x3, - HNS_ROCE_SQ_OPCODE_RDMA_WRITE_WITH_IMM = 0x4, - HNS_ROCE_SQ_OPCODE_RDMA_READ = 0x5, - HNS_ROCE_SQ_OPCODE_ATOMIC_COMP_AND_SWAP = 0x6, - HNS_ROCE_SQ_OPCODE_ATOMIC_FETCH_AND_ADD = 0x7, - HNS_ROCE_SQ_OPCODE_ATOMIC_MASK_COMP_AND_SWAP = 0x8, - HNS_ROCE_SQ_OPCODE_ATOMIC_MASK_FETCH_AND_ADD = 0x9, - HNS_ROCE_SQ_OPCODE_FAST_REG_WR = 0xa, - HNS_ROCE_SQ_OPCODE_LOCAL_INV = 0xb, - HNS_ROCE_SQ_OPCODE_BIND_MW = 0xc, -}; - -enum { /* rq operations */ HNS_ROCE_V2_OPCODE_RDMA_WRITE_IMM = 0x0, HNS_ROCE_V2_OPCODE_SEND = 0x1, @@ -228,6 +212,7 @@ enum { HNS_ROCE_CQE_V2_TRANSPORT_RETRY_EXC_ERR = 0x15, HNS_ROCE_CQE_V2_RNR_RETRY_EXC_ERR = 0x16, HNS_ROCE_CQE_V2_REMOTE_ABORT_ERR = 0x22, + HNS_ROCE_CQE_V2_GENERAL_ERR = 0x23, HNS_ROCE_V2_CQE_STATUS_MASK = 0xff, }; @@ -1241,10 +1226,9 @@ struct hns_roce_func_clear { }; #define FUNC_CLEAR_RST_FUN_DONE_S 0 -/* Each physical function manages up to 248 virtual functionsï¼› - * it takes up to 100ms for each function to execute clearï¼› - * if an abnormal reset occurs, it is executed twice at most; - * so it takes up to 249 * 2 * 100ms. +/* Each physical function manages up to 248 virtual functions, it takes up to + * 100ms for each function to execute clear. If an abnormal reset occurs, it is + * executed twice at most, so it takes up to 249 * 2 * 100ms. */ #define HNS_ROCE_V2_FUNC_CLEAR_TIMEOUT_MSECS (249 * 2 * 100) #define HNS_ROCE_V2_READ_FUNC_CLEAR_FLAG_INTERVAL 40 @@ -1648,7 +1632,7 @@ struct hns_roce_query_pf_caps_c { struct hns_roce_query_pf_caps_d { __le32 wq_hop_num_max_srqs; __le16 srq_depth; - __le16 rsv; + __le16 cap_flags_ex; __le32 num_ceqs_ceq_depth; __le32 arm_st_aeq_depth; __le32 num_uars_rsv_pds; @@ -1978,7 +1962,7 @@ int hns_roce_v2_query_cqc_info(struct hns_roce_dev *hr_dev, u32 cqn, static inline void hns_roce_write64(struct hns_roce_dev *hr_dev, __le32 val[2], void __iomem *dest) { - struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hns_roce_v2_priv *priv = hr_dev->priv; struct hnae3_handle *handle = priv->handle; const struct hnae3_ae_ops *ops = handle->ae_algo->ops; diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c index d0031d559213..5907cfd878a6 100644 --- a/drivers/infiniband/hw/hns/hns_roce_main.c +++ b/drivers/infiniband/hw/hns/hns_roce_main.c @@ -233,7 +233,6 @@ static int hns_roce_query_port(struct ib_device *ib_dev, u8 port_num, enum ib_mtu mtu; u8 port; - assert(port_num > 0); port = port_num - 1; /* props being zeroed by the caller, avoid zeroing it here */ @@ -429,7 +428,7 @@ static const struct ib_device_ops hns_roce_dev_ops = { .destroy_ah = hns_roce_destroy_ah, .destroy_cq = hns_roce_destroy_cq, .disassociate_ucontext = hns_roce_disassociate_ucontext, - .fill_res_entry = hns_roce_fill_res_entry, + .fill_res_cq_entry = hns_roce_fill_res_cq_entry, .get_dma_mr = hns_roce_get_dma_mr, .get_link_layer = hns_roce_get_link_layer, .get_port_immutable = hns_roce_port_immutable, @@ -579,33 +578,12 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev) int ret; struct device *dev = hr_dev->dev; - ret = hns_roce_init_hem_table(hr_dev, &hr_dev->mr_table.mtt_table, - HEM_TYPE_MTT, hr_dev->caps.mtt_entry_sz, - hr_dev->caps.num_mtt_segs, 1); - if (ret) { - dev_err(dev, "Failed to init MTT context memory, aborting.\n"); - return ret; - } - - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) { - ret = hns_roce_init_hem_table(hr_dev, - &hr_dev->mr_table.mtt_cqe_table, - HEM_TYPE_CQE, - hr_dev->caps.mtt_entry_sz, - hr_dev->caps.num_cqe_segs, 1); - if (ret) { - dev_err(dev, - "Failed to init CQE context memory, aborting.\n"); - goto err_unmap_cqe; - } - } - ret = hns_roce_init_hem_table(hr_dev, &hr_dev->mr_table.mtpt_table, HEM_TYPE_MTPT, hr_dev->caps.mtpt_entry_sz, hr_dev->caps.num_mtpts, 1); if (ret) { dev_err(dev, "Failed to init MTPT context memory, aborting.\n"); - goto err_unmap_mtt; + return ret; } ret = hns_roce_init_hem_table(hr_dev, &hr_dev->qp_table.qp_table, @@ -660,32 +638,6 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev) } } - if (hr_dev->caps.num_srqwqe_segs) { - ret = hns_roce_init_hem_table(hr_dev, - &hr_dev->mr_table.mtt_srqwqe_table, - HEM_TYPE_SRQWQE, - hr_dev->caps.mtt_entry_sz, - hr_dev->caps.num_srqwqe_segs, 1); - if (ret) { - dev_err(dev, - "Failed to init MTT srqwqe memory, aborting.\n"); - goto err_unmap_srq; - } - } - - if (hr_dev->caps.num_idx_segs) { - ret = hns_roce_init_hem_table(hr_dev, - &hr_dev->mr_table.mtt_idx_table, - HEM_TYPE_IDX, - hr_dev->caps.idx_entry_sz, - hr_dev->caps.num_idx_segs, 1); - if (ret) { - dev_err(dev, - "Failed to init MTT idx memory, aborting.\n"); - goto err_unmap_srqwqe; - } - } - if (hr_dev->caps.sccc_entry_sz) { ret = hns_roce_init_hem_table(hr_dev, &hr_dev->qp_table.sccc_table, @@ -695,7 +647,7 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev) if (ret) { dev_err(dev, "Failed to init SCC context memory, aborting.\n"); - goto err_unmap_idx; + goto err_unmap_srq; } } @@ -733,17 +685,6 @@ err_unmap_ctx: if (hr_dev->caps.sccc_entry_sz) hns_roce_cleanup_hem_table(hr_dev, &hr_dev->qp_table.sccc_table); - -err_unmap_idx: - if (hr_dev->caps.num_idx_segs) - hns_roce_cleanup_hem_table(hr_dev, - &hr_dev->mr_table.mtt_idx_table); - -err_unmap_srqwqe: - if (hr_dev->caps.num_srqwqe_segs) - hns_roce_cleanup_hem_table(hr_dev, - &hr_dev->mr_table.mtt_srqwqe_table); - err_unmap_srq: if (hr_dev->caps.srqc_entry_sz) hns_roce_cleanup_hem_table(hr_dev, &hr_dev->srq_table.table); @@ -765,14 +706,6 @@ err_unmap_qp: err_unmap_dmpt: hns_roce_cleanup_hem_table(hr_dev, &hr_dev->mr_table.mtpt_table); -err_unmap_mtt: - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - hns_roce_cleanup_hem_table(hr_dev, - &hr_dev->mr_table.mtt_cqe_table); - -err_unmap_cqe: - hns_roce_cleanup_hem_table(hr_dev, &hr_dev->mr_table.mtt_table); - return ret; } diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c index 176f34692f88..e5df3884b41d 100644 --- a/drivers/infiniband/hw/hns/hns_roce_mr.c +++ b/drivers/infiniband/hw/hns/hns_roce_mr.c @@ -66,645 +66,89 @@ int hns_roce_hw_destroy_mpt(struct hns_roce_dev *hr_dev, HNS_ROCE_CMD_TIMEOUT_MSECS); } -static int hns_roce_buddy_alloc(struct hns_roce_buddy *buddy, int order, - unsigned long *seg) +static int alloc_mr_key(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, + u32 pd, u64 iova, u64 size, u32 access) { - int o; - u32 m; + struct ib_device *ibdev = &hr_dev->ib_dev; + unsigned long obj = 0; + int err; - spin_lock(&buddy->lock); - - for (o = order; o <= buddy->max_order; ++o) { - if (buddy->num_free[o]) { - m = 1 << (buddy->max_order - o); - *seg = find_first_bit(buddy->bits[o], m); - if (*seg < m) - goto found; - } - } - spin_unlock(&buddy->lock); - return -EINVAL; - - found: - clear_bit(*seg, buddy->bits[o]); - --buddy->num_free[o]; - - while (o > order) { - --o; - *seg <<= 1; - set_bit(*seg ^ 1, buddy->bits[o]); - ++buddy->num_free[o]; - } - - spin_unlock(&buddy->lock); - - *seg <<= order; - return 0; -} - -static void hns_roce_buddy_free(struct hns_roce_buddy *buddy, unsigned long seg, - int order) -{ - seg >>= order; - - spin_lock(&buddy->lock); - - while (test_bit(seg ^ 1, buddy->bits[order])) { - clear_bit(seg ^ 1, buddy->bits[order]); - --buddy->num_free[order]; - seg >>= 1; - ++order; - } - - set_bit(seg, buddy->bits[order]); - ++buddy->num_free[order]; - - spin_unlock(&buddy->lock); -} - -static int hns_roce_buddy_init(struct hns_roce_buddy *buddy, int max_order) -{ - int i, s; - - buddy->max_order = max_order; - spin_lock_init(&buddy->lock); - buddy->bits = kcalloc(buddy->max_order + 1, - sizeof(*buddy->bits), - GFP_KERNEL); - buddy->num_free = kcalloc(buddy->max_order + 1, - sizeof(*buddy->num_free), - GFP_KERNEL); - if (!buddy->bits || !buddy->num_free) - goto err_out; - - for (i = 0; i <= buddy->max_order; ++i) { - s = BITS_TO_LONGS(1 << (buddy->max_order - i)); - buddy->bits[i] = kcalloc(s, sizeof(long), GFP_KERNEL | - __GFP_NOWARN); - if (!buddy->bits[i]) { - buddy->bits[i] = vzalloc(array_size(s, sizeof(long))); - if (!buddy->bits[i]) - goto err_out_free; - } - } - - set_bit(0, buddy->bits[buddy->max_order]); - buddy->num_free[buddy->max_order] = 1; - - return 0; - -err_out_free: - for (i = 0; i <= buddy->max_order; ++i) - kvfree(buddy->bits[i]); - -err_out: - kfree(buddy->bits); - kfree(buddy->num_free); - return -ENOMEM; -} - -static void hns_roce_buddy_cleanup(struct hns_roce_buddy *buddy) -{ - int i; - - for (i = 0; i <= buddy->max_order; ++i) - kvfree(buddy->bits[i]); - - kfree(buddy->bits); - kfree(buddy->num_free); -} - -static int hns_roce_alloc_mtt_range(struct hns_roce_dev *hr_dev, int order, - unsigned long *seg, u32 mtt_type) -{ - struct hns_roce_mr_table *mr_table = &hr_dev->mr_table; - struct hns_roce_hem_table *table; - struct hns_roce_buddy *buddy; - int ret; - - switch (mtt_type) { - case MTT_TYPE_WQE: - buddy = &mr_table->mtt_buddy; - table = &mr_table->mtt_table; - break; - case MTT_TYPE_CQE: - buddy = &mr_table->mtt_cqe_buddy; - table = &mr_table->mtt_cqe_table; - break; - case MTT_TYPE_SRQWQE: - buddy = &mr_table->mtt_srqwqe_buddy; - table = &mr_table->mtt_srqwqe_table; - break; - case MTT_TYPE_IDX: - buddy = &mr_table->mtt_idx_buddy; - table = &mr_table->mtt_idx_table; - break; - default: - dev_err(hr_dev->dev, "Unsupport MTT table type: %d\n", - mtt_type); - return -EINVAL; - } - - ret = hns_roce_buddy_alloc(buddy, order, seg); - if (ret) - return ret; - - ret = hns_roce_table_get_range(hr_dev, table, *seg, - *seg + (1 << order) - 1); - if (ret) { - hns_roce_buddy_free(buddy, *seg, order); - return ret; - } - - return 0; -} - -int hns_roce_mtt_init(struct hns_roce_dev *hr_dev, int npages, int page_shift, - struct hns_roce_mtt *mtt) -{ - int ret; - int i; - - /* Page num is zero, correspond to DMA memory register */ - if (!npages) { - mtt->order = -1; - mtt->page_shift = HNS_ROCE_HEM_PAGE_SHIFT; - return 0; - } - - /* Note: if page_shift is zero, FAST memory register */ - mtt->page_shift = page_shift; - - /* Compute MTT entry necessary */ - for (mtt->order = 0, i = HNS_ROCE_MTT_ENTRY_PER_SEG; i < npages; - i <<= 1) - ++mtt->order; - - /* Allocate MTT entry */ - ret = hns_roce_alloc_mtt_range(hr_dev, mtt->order, &mtt->first_seg, - mtt->mtt_type); - if (ret) + /* Allocate a key for mr from mr_table */ + err = hns_roce_bitmap_alloc(&hr_dev->mr_table.mtpt_bitmap, &obj); + if (err) { + ibdev_err(ibdev, + "failed to alloc bitmap for MR key, ret = %d.\n", + err); return -ENOMEM; - - return 0; -} - -void hns_roce_mtt_cleanup(struct hns_roce_dev *hr_dev, struct hns_roce_mtt *mtt) -{ - struct hns_roce_mr_table *mr_table = &hr_dev->mr_table; - - if (mtt->order < 0) - return; - - switch (mtt->mtt_type) { - case MTT_TYPE_WQE: - hns_roce_buddy_free(&mr_table->mtt_buddy, mtt->first_seg, - mtt->order); - hns_roce_table_put_range(hr_dev, &mr_table->mtt_table, - mtt->first_seg, - mtt->first_seg + (1 << mtt->order) - 1); - break; - case MTT_TYPE_CQE: - hns_roce_buddy_free(&mr_table->mtt_cqe_buddy, mtt->first_seg, - mtt->order); - hns_roce_table_put_range(hr_dev, &mr_table->mtt_cqe_table, - mtt->first_seg, - mtt->first_seg + (1 << mtt->order) - 1); - break; - case MTT_TYPE_SRQWQE: - hns_roce_buddy_free(&mr_table->mtt_srqwqe_buddy, mtt->first_seg, - mtt->order); - hns_roce_table_put_range(hr_dev, &mr_table->mtt_srqwqe_table, - mtt->first_seg, - mtt->first_seg + (1 << mtt->order) - 1); - break; - case MTT_TYPE_IDX: - hns_roce_buddy_free(&mr_table->mtt_idx_buddy, mtt->first_seg, - mtt->order); - hns_roce_table_put_range(hr_dev, &mr_table->mtt_idx_table, - mtt->first_seg, - mtt->first_seg + (1 << mtt->order) - 1); - break; - default: - dev_err(hr_dev->dev, - "Unsupport mtt type %d, clean mtt failed\n", - mtt->mtt_type); - break; } -} -static void hns_roce_loop_free(struct hns_roce_dev *hr_dev, - struct hns_roce_mr *mr, int err_loop_index, - int loop_i, int loop_j) -{ - struct device *dev = hr_dev->dev; - u32 mhop_num; - u32 pbl_bt_sz; - u64 bt_idx; - int i, j; - - pbl_bt_sz = 1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT); - mhop_num = hr_dev->caps.pbl_hop_num; - - i = loop_i; - if (mhop_num == 3 && err_loop_index == 2) { - for (; i >= 0; i--) { - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l1[i], - mr->pbl_l1_dma_addr[i]); - - for (j = 0; j < pbl_bt_sz / BA_BYTE_LEN; j++) { - if (i == loop_i && j >= loop_j) - break; - - bt_idx = i * pbl_bt_sz / BA_BYTE_LEN + j; - dma_free_coherent(dev, pbl_bt_sz, - mr->pbl_bt_l2[bt_idx], - mr->pbl_l2_dma_addr[bt_idx]); - } - } - } else if (mhop_num == 3 && err_loop_index == 1) { - for (i -= 1; i >= 0; i--) { - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l1[i], - mr->pbl_l1_dma_addr[i]); - - for (j = 0; j < pbl_bt_sz / BA_BYTE_LEN; j++) { - bt_idx = i * pbl_bt_sz / BA_BYTE_LEN + j; - dma_free_coherent(dev, pbl_bt_sz, - mr->pbl_bt_l2[bt_idx], - mr->pbl_l2_dma_addr[bt_idx]); - } - } - } else if (mhop_num == 2 && err_loop_index == 1) { - for (i -= 1; i >= 0; i--) - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l1[i], - mr->pbl_l1_dma_addr[i]); - } else { - dev_warn(dev, "not support: mhop_num=%d, err_loop_index=%d.", - mhop_num, err_loop_index); - return; - } - - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l0, mr->pbl_l0_dma_addr); - mr->pbl_bt_l0 = NULL; - mr->pbl_l0_dma_addr = 0; -} -static int pbl_1hop_alloc(struct hns_roce_dev *hr_dev, int npages, - struct hns_roce_mr *mr, u32 pbl_bt_sz) -{ - struct device *dev = hr_dev->dev; + mr->iova = iova; /* MR va starting addr */ + mr->size = size; /* MR addr range */ + mr->pd = pd; /* MR num */ + mr->access = access; /* MR access permit */ + mr->enabled = 0; /* MR active status */ + mr->key = hw_index_to_key(obj); /* MR key */ - if (npages > pbl_bt_sz / 8) { - dev_err(dev, "npages %d is larger than buf_pg_sz!", - npages); - return -EINVAL; + err = hns_roce_table_get(hr_dev, &hr_dev->mr_table.mtpt_table, obj); + if (err) { + ibdev_err(ibdev, "failed to alloc mtpt, ret = %d.\n", err); + goto err_free_bitmap; } - mr->pbl_buf = dma_alloc_coherent(dev, npages * 8, - &(mr->pbl_dma_addr), - GFP_KERNEL); - if (!mr->pbl_buf) - return -ENOMEM; - mr->pbl_size = npages; - mr->pbl_ba = mr->pbl_dma_addr; - mr->pbl_hop_num = 1; - mr->pbl_ba_pg_sz = hr_dev->caps.pbl_ba_pg_sz; - mr->pbl_buf_pg_sz = hr_dev->caps.pbl_buf_pg_sz; return 0; - +err_free_bitmap: + hns_roce_bitmap_free(&hr_dev->mr_table.mtpt_bitmap, obj, BITMAP_NO_RR); + return err; } - -static int pbl_2hop_alloc(struct hns_roce_dev *hr_dev, int npages, - struct hns_roce_mr *mr, u32 pbl_bt_sz) +static void free_mr_key(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr) { - struct device *dev = hr_dev->dev; - int npages_allocated; - u64 pbl_last_bt_num; - u64 pbl_bt_cnt = 0; - u64 size; - int i; - - pbl_last_bt_num = (npages + pbl_bt_sz / 8 - 1) / (pbl_bt_sz / 8); - - /* alloc L1 BT */ - for (i = 0; i < pbl_bt_sz / 8; i++) { - if (pbl_bt_cnt + 1 < pbl_last_bt_num) { - size = pbl_bt_sz; - } else { - npages_allocated = i * (pbl_bt_sz / 8); - size = (npages - npages_allocated) * 8; - } - mr->pbl_bt_l1[i] = dma_alloc_coherent(dev, size, - &(mr->pbl_l1_dma_addr[i]), - GFP_KERNEL); - if (!mr->pbl_bt_l1[i]) { - hns_roce_loop_free(hr_dev, mr, 1, i, 0); - return -ENOMEM; - } - - *(mr->pbl_bt_l0 + i) = mr->pbl_l1_dma_addr[i]; + unsigned long obj = key_to_hw_index(mr->key); - pbl_bt_cnt++; - if (pbl_bt_cnt >= pbl_last_bt_num) - break; - } - - mr->l0_chunk_last_num = i + 1; - - return 0; + hns_roce_table_put(hr_dev, &hr_dev->mr_table.mtpt_table, obj); + hns_roce_bitmap_free(&hr_dev->mr_table.mtpt_bitmap, obj, BITMAP_NO_RR); } -static int pbl_3hop_alloc(struct hns_roce_dev *hr_dev, int npages, - struct hns_roce_mr *mr, u32 pbl_bt_sz) +static int alloc_mr_pbl(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, + size_t length, struct ib_udata *udata, u64 start, + int access) { - struct device *dev = hr_dev->dev; - int mr_alloc_done = 0; - int npages_allocated; - u64 pbl_last_bt_num; - u64 pbl_bt_cnt = 0; - u64 bt_idx; - u64 size; - int i; - int j = 0; - - pbl_last_bt_num = (npages + pbl_bt_sz / 8 - 1) / (pbl_bt_sz / 8); - - mr->pbl_l2_dma_addr = kcalloc(pbl_last_bt_num, - sizeof(*mr->pbl_l2_dma_addr), - GFP_KERNEL); - if (!mr->pbl_l2_dma_addr) - return -ENOMEM; - - mr->pbl_bt_l2 = kcalloc(pbl_last_bt_num, - sizeof(*mr->pbl_bt_l2), - GFP_KERNEL); - if (!mr->pbl_bt_l2) - goto err_kcalloc_bt_l2; - - /* alloc L1, L2 BT */ - for (i = 0; i < pbl_bt_sz / 8; i++) { - mr->pbl_bt_l1[i] = dma_alloc_coherent(dev, pbl_bt_sz, - &(mr->pbl_l1_dma_addr[i]), - GFP_KERNEL); - if (!mr->pbl_bt_l1[i]) { - hns_roce_loop_free(hr_dev, mr, 1, i, 0); - goto err_dma_alloc_l0; - } - - *(mr->pbl_bt_l0 + i) = mr->pbl_l1_dma_addr[i]; - - for (j = 0; j < pbl_bt_sz / 8; j++) { - bt_idx = i * pbl_bt_sz / 8 + j; - - if (pbl_bt_cnt + 1 < pbl_last_bt_num) { - size = pbl_bt_sz; - } else { - npages_allocated = bt_idx * - (pbl_bt_sz / 8); - size = (npages - npages_allocated) * 8; - } - mr->pbl_bt_l2[bt_idx] = dma_alloc_coherent( - dev, size, - &(mr->pbl_l2_dma_addr[bt_idx]), - GFP_KERNEL); - if (!mr->pbl_bt_l2[bt_idx]) { - hns_roce_loop_free(hr_dev, mr, 2, i, j); - goto err_dma_alloc_l0; - } - - *(mr->pbl_bt_l1[i] + j) = - mr->pbl_l2_dma_addr[bt_idx]; - - pbl_bt_cnt++; - if (pbl_bt_cnt >= pbl_last_bt_num) { - mr_alloc_done = 1; - break; - } - } - - if (mr_alloc_done) - break; - } - - mr->l0_chunk_last_num = i + 1; - mr->l1_chunk_last_num = j + 1; - - - return 0; - -err_dma_alloc_l0: - kfree(mr->pbl_bt_l2); - mr->pbl_bt_l2 = NULL; - -err_kcalloc_bt_l2: - kfree(mr->pbl_l2_dma_addr); - mr->pbl_l2_dma_addr = NULL; - - return -ENOMEM; -} - - -/* PBL multi hop addressing */ -static int hns_roce_mhop_alloc(struct hns_roce_dev *hr_dev, int npages, - struct hns_roce_mr *mr) -{ - struct device *dev = hr_dev->dev; - u32 pbl_bt_sz; - u32 mhop_num; - - mhop_num = (mr->type == MR_TYPE_FRMR ? 1 : hr_dev->caps.pbl_hop_num); - pbl_bt_sz = 1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT); - - if (mhop_num == HNS_ROCE_HOP_NUM_0) - return 0; - - if (mhop_num == 1) - return pbl_1hop_alloc(hr_dev, npages, mr, pbl_bt_sz); - - mr->pbl_l1_dma_addr = kcalloc(pbl_bt_sz / 8, - sizeof(*mr->pbl_l1_dma_addr), - GFP_KERNEL); - if (!mr->pbl_l1_dma_addr) - return -ENOMEM; - - mr->pbl_bt_l1 = kcalloc(pbl_bt_sz / 8, sizeof(*mr->pbl_bt_l1), - GFP_KERNEL); - if (!mr->pbl_bt_l1) - goto err_kcalloc_bt_l1; - - /* alloc L0 BT */ - mr->pbl_bt_l0 = dma_alloc_coherent(dev, pbl_bt_sz, - &(mr->pbl_l0_dma_addr), - GFP_KERNEL); - if (!mr->pbl_bt_l0) - goto err_kcalloc_l2_dma; - - if (mhop_num == 2) { - if (pbl_2hop_alloc(hr_dev, npages, mr, pbl_bt_sz)) - goto err_kcalloc_l2_dma; - } - - if (mhop_num == 3) { - if (pbl_3hop_alloc(hr_dev, npages, mr, pbl_bt_sz)) - goto err_kcalloc_l2_dma; - } - - - mr->pbl_size = npages; - mr->pbl_ba = mr->pbl_l0_dma_addr; - mr->pbl_hop_num = hr_dev->caps.pbl_hop_num; - mr->pbl_ba_pg_sz = hr_dev->caps.pbl_ba_pg_sz; - mr->pbl_buf_pg_sz = hr_dev->caps.pbl_buf_pg_sz; - - return 0; - -err_kcalloc_l2_dma: - kfree(mr->pbl_bt_l1); - mr->pbl_bt_l1 = NULL; - -err_kcalloc_bt_l1: - kfree(mr->pbl_l1_dma_addr); - mr->pbl_l1_dma_addr = NULL; - - return -ENOMEM; -} - -static int hns_roce_mr_alloc(struct hns_roce_dev *hr_dev, u32 pd, u64 iova, - u64 size, u32 access, int npages, - struct hns_roce_mr *mr) -{ - struct device *dev = hr_dev->dev; - unsigned long index = 0; - int ret; - - /* Allocate a key for mr from mr_table */ - ret = hns_roce_bitmap_alloc(&hr_dev->mr_table.mtpt_bitmap, &index); - if (ret) - return -ENOMEM; - - mr->iova = iova; /* MR va starting addr */ - mr->size = size; /* MR addr range */ - mr->pd = pd; /* MR num */ - mr->access = access; /* MR access permit */ - mr->enabled = 0; /* MR active status */ - mr->key = hw_index_to_key(index); /* MR key */ - - if (size == ~0ull) { - mr->pbl_buf = NULL; - mr->pbl_dma_addr = 0; - /* PBL multi-hop addressing parameters */ - mr->pbl_bt_l2 = NULL; - mr->pbl_bt_l1 = NULL; - mr->pbl_bt_l0 = NULL; - mr->pbl_l2_dma_addr = NULL; - mr->pbl_l1_dma_addr = NULL; - mr->pbl_l0_dma_addr = 0; - } else { - if (!hr_dev->caps.pbl_hop_num) { - mr->pbl_buf = dma_alloc_coherent(dev, - npages * BA_BYTE_LEN, - &(mr->pbl_dma_addr), - GFP_KERNEL); - if (!mr->pbl_buf) - return -ENOMEM; - } else { - ret = hns_roce_mhop_alloc(hr_dev, npages, mr); - } - } + struct ib_device *ibdev = &hr_dev->ib_dev; + bool is_fast = mr->type == MR_TYPE_FRMR; + struct hns_roce_buf_attr buf_attr = {}; + int err; + + mr->pbl_hop_num = is_fast ? 1 : hr_dev->caps.pbl_hop_num; + buf_attr.page_shift = is_fast ? PAGE_SHIFT : + hr_dev->caps.pbl_buf_pg_sz + PAGE_SHIFT; + buf_attr.region[0].size = length; + buf_attr.region[0].hopnum = mr->pbl_hop_num; + buf_attr.region_count = 1; + buf_attr.fixed_page = true; + buf_attr.user_access = access; + /* fast MR's buffer is alloced before mapping, not at creation */ + buf_attr.mtt_only = is_fast; + + err = hns_roce_mtr_create(hr_dev, &mr->pbl_mtr, &buf_attr, + hr_dev->caps.pbl_ba_pg_sz + HNS_HW_PAGE_SHIFT, + udata, start); + if (err) + ibdev_err(ibdev, "failed to alloc pbl mtr, ret = %d.\n", err); + else + mr->npages = mr->pbl_mtr.hem_cfg.buf_pg_count; - return ret; + return err; } -static void hns_roce_mhop_free(struct hns_roce_dev *hr_dev, - struct hns_roce_mr *mr) +static void free_mr_pbl(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr) { - struct device *dev = hr_dev->dev; - int npages_allocated; - int npages; - int i, j; - u32 pbl_bt_sz; - u32 mhop_num; - u64 bt_idx; - - npages = mr->pbl_size; - pbl_bt_sz = 1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT); - mhop_num = (mr->type == MR_TYPE_FRMR) ? 1 : hr_dev->caps.pbl_hop_num; - - if (mhop_num == HNS_ROCE_HOP_NUM_0) - return; - - if (mhop_num == 1) { - dma_free_coherent(dev, (unsigned int)(npages * BA_BYTE_LEN), - mr->pbl_buf, mr->pbl_dma_addr); - return; - } - - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l0, - mr->pbl_l0_dma_addr); - - if (mhop_num == 2) { - for (i = 0; i < mr->l0_chunk_last_num; i++) { - if (i == mr->l0_chunk_last_num - 1) { - npages_allocated = - i * (pbl_bt_sz / BA_BYTE_LEN); - - dma_free_coherent(dev, - (npages - npages_allocated) * BA_BYTE_LEN, - mr->pbl_bt_l1[i], - mr->pbl_l1_dma_addr[i]); - - break; - } - - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l1[i], - mr->pbl_l1_dma_addr[i]); - } - } else if (mhop_num == 3) { - for (i = 0; i < mr->l0_chunk_last_num; i++) { - dma_free_coherent(dev, pbl_bt_sz, mr->pbl_bt_l1[i], - mr->pbl_l1_dma_addr[i]); - - for (j = 0; j < pbl_bt_sz / BA_BYTE_LEN; j++) { - bt_idx = i * (pbl_bt_sz / BA_BYTE_LEN) + j; - - if ((i == mr->l0_chunk_last_num - 1) - && j == mr->l1_chunk_last_num - 1) { - npages_allocated = bt_idx * - (pbl_bt_sz / BA_BYTE_LEN); - - dma_free_coherent(dev, - (npages - npages_allocated) * - BA_BYTE_LEN, - mr->pbl_bt_l2[bt_idx], - mr->pbl_l2_dma_addr[bt_idx]); - - break; - } - - dma_free_coherent(dev, pbl_bt_sz, - mr->pbl_bt_l2[bt_idx], - mr->pbl_l2_dma_addr[bt_idx]); - } - } - } - - kfree(mr->pbl_bt_l1); - kfree(mr->pbl_l1_dma_addr); - mr->pbl_bt_l1 = NULL; - mr->pbl_l1_dma_addr = NULL; - if (mhop_num == 3) { - kfree(mr->pbl_bt_l2); - kfree(mr->pbl_l2_dma_addr); - mr->pbl_bt_l2 = NULL; - mr->pbl_l2_dma_addr = NULL; - } + hns_roce_mtr_destroy(hr_dev, &mr->pbl_mtr); } static void hns_roce_mr_free(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr) { - struct device *dev = hr_dev->dev; - int npages = 0; + struct ib_device *ibdev = &hr_dev->ib_dev; int ret; if (mr->enabled) { @@ -712,27 +156,12 @@ static void hns_roce_mr_free(struct hns_roce_dev *hr_dev, key_to_hw_index(mr->key) & (hr_dev->caps.num_mtpts - 1)); if (ret) - dev_warn(dev, "DESTROY_MPT failed (%d)\n", ret); - } - - if (mr->size != ~0ULL) { - if (mr->type == MR_TYPE_MR) - npages = ib_umem_page_count(mr->umem); - - if (!hr_dev->caps.pbl_hop_num) - dma_free_coherent(dev, - (unsigned int)(npages * BA_BYTE_LEN), - mr->pbl_buf, mr->pbl_dma_addr); - else - hns_roce_mhop_free(hr_dev, mr); + ibdev_warn(ibdev, "failed to destroy mpt, ret = %d.\n", + ret); } - if (mr->enabled) - hns_roce_table_put(hr_dev, &hr_dev->mr_table.mtpt_table, - key_to_hw_index(mr->key)); - - hns_roce_bitmap_free(&hr_dev->mr_table.mtpt_bitmap, - key_to_hw_index(mr->key), BITMAP_NO_RR); + free_mr_pbl(hr_dev, mr); + free_mr_key(hr_dev, mr); } static int hns_roce_mr_enable(struct hns_roce_dev *hr_dev, @@ -742,24 +171,19 @@ static int hns_roce_mr_enable(struct hns_roce_dev *hr_dev, unsigned long mtpt_idx = key_to_hw_index(mr->key); struct device *dev = hr_dev->dev; struct hns_roce_cmd_mailbox *mailbox; - struct hns_roce_mr_table *mr_table = &hr_dev->mr_table; - - /* Prepare HEM entry memory */ - ret = hns_roce_table_get(hr_dev, &mr_table->mtpt_table, mtpt_idx); - if (ret) - return ret; /* Allocate mailbox memory */ mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); if (IS_ERR(mailbox)) { ret = PTR_ERR(mailbox); - goto err_table; + return ret; } if (mr->type != MR_TYPE_FRMR) - ret = hr_dev->hw->write_mtpt(mailbox->buf, mr, mtpt_idx); + ret = hr_dev->hw->write_mtpt(hr_dev, mailbox->buf, mr, + mtpt_idx); else - ret = hr_dev->hw->frmr_write_mtpt(mailbox->buf, mr); + ret = hr_dev->hw->frmr_write_mtpt(hr_dev, mailbox->buf, mr); if (ret) { dev_err(dev, "Write mtpt fail!\n"); goto err_page; @@ -780,137 +204,6 @@ static int hns_roce_mr_enable(struct hns_roce_dev *hr_dev, err_page: hns_roce_free_cmd_mailbox(hr_dev, mailbox); -err_table: - hns_roce_table_put(hr_dev, &mr_table->mtpt_table, mtpt_idx); - return ret; -} - -static int hns_roce_write_mtt_chunk(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, u32 start_index, - u32 npages, u64 *page_list) -{ - struct hns_roce_hem_table *table; - dma_addr_t dma_handle; - __le64 *mtts; - u32 bt_page_size; - u32 i; - - switch (mtt->mtt_type) { - case MTT_TYPE_WQE: - table = &hr_dev->mr_table.mtt_table; - bt_page_size = 1 << (hr_dev->caps.mtt_ba_pg_sz + PAGE_SHIFT); - break; - case MTT_TYPE_CQE: - table = &hr_dev->mr_table.mtt_cqe_table; - bt_page_size = 1 << (hr_dev->caps.cqe_ba_pg_sz + PAGE_SHIFT); - break; - case MTT_TYPE_SRQWQE: - table = &hr_dev->mr_table.mtt_srqwqe_table; - bt_page_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz + PAGE_SHIFT); - break; - case MTT_TYPE_IDX: - table = &hr_dev->mr_table.mtt_idx_table; - bt_page_size = 1 << (hr_dev->caps.idx_ba_pg_sz + PAGE_SHIFT); - break; - default: - return -EINVAL; - } - - /* All MTTs must fit in the same page */ - if (start_index / (bt_page_size / sizeof(u64)) != - (start_index + npages - 1) / (bt_page_size / sizeof(u64))) - return -EINVAL; - - if (start_index & (HNS_ROCE_MTT_ENTRY_PER_SEG - 1)) - return -EINVAL; - - mtts = hns_roce_table_find(hr_dev, table, - mtt->first_seg + - start_index / HNS_ROCE_MTT_ENTRY_PER_SEG, - &dma_handle); - if (!mtts) - return -ENOMEM; - - /* Save page addr, low 12 bits : 0 */ - for (i = 0; i < npages; ++i) { - if (!hr_dev->caps.mtt_hop_num) - mtts[i] = cpu_to_le64(page_list[i] >> PAGE_ADDR_SHIFT); - else - mtts[i] = cpu_to_le64(page_list[i]); - } - - return 0; -} - -static int hns_roce_write_mtt(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, u32 start_index, - u32 npages, u64 *page_list) -{ - int chunk; - int ret; - u32 bt_page_size; - - if (mtt->order < 0) - return -EINVAL; - - switch (mtt->mtt_type) { - case MTT_TYPE_WQE: - bt_page_size = 1 << (hr_dev->caps.mtt_ba_pg_sz + PAGE_SHIFT); - break; - case MTT_TYPE_CQE: - bt_page_size = 1 << (hr_dev->caps.cqe_ba_pg_sz + PAGE_SHIFT); - break; - case MTT_TYPE_SRQWQE: - bt_page_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz + PAGE_SHIFT); - break; - case MTT_TYPE_IDX: - bt_page_size = 1 << (hr_dev->caps.idx_ba_pg_sz + PAGE_SHIFT); - break; - default: - dev_err(hr_dev->dev, - "Unsupport mtt type %d, write mtt failed\n", - mtt->mtt_type); - return -EINVAL; - } - - while (npages > 0) { - chunk = min_t(int, bt_page_size / sizeof(u64), npages); - - ret = hns_roce_write_mtt_chunk(hr_dev, mtt, start_index, chunk, - page_list); - if (ret) - return ret; - - npages -= chunk; - start_index += chunk; - page_list += chunk; - } - - return 0; -} - -int hns_roce_buf_write_mtt(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, struct hns_roce_buf *buf) -{ - u64 *page_list; - int ret; - u32 i; - - page_list = kmalloc_array(buf->npages, sizeof(*page_list), GFP_KERNEL); - if (!page_list) - return -ENOMEM; - - for (i = 0; i < buf->npages; ++i) { - if (buf->nbufs == 1) - page_list[i] = buf->direct.map + (i << buf->page_shift); - else - page_list[i] = buf->page_list[i].map; - - } - ret = hns_roce_write_mtt(hr_dev, mtt, 0, buf->npages, page_list); - - kfree(page_list); - return ret; } @@ -923,50 +216,6 @@ int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev) hr_dev->caps.num_mtpts, hr_dev->caps.num_mtpts - 1, hr_dev->caps.reserved_mrws, 0); - if (ret) - return ret; - - ret = hns_roce_buddy_init(&mr_table->mtt_buddy, - ilog2(hr_dev->caps.num_mtt_segs)); - if (ret) - goto err_buddy; - - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) { - ret = hns_roce_buddy_init(&mr_table->mtt_cqe_buddy, - ilog2(hr_dev->caps.num_cqe_segs)); - if (ret) - goto err_buddy_cqe; - } - - if (hr_dev->caps.num_srqwqe_segs) { - ret = hns_roce_buddy_init(&mr_table->mtt_srqwqe_buddy, - ilog2(hr_dev->caps.num_srqwqe_segs)); - if (ret) - goto err_buddy_srqwqe; - } - - if (hr_dev->caps.num_idx_segs) { - ret = hns_roce_buddy_init(&mr_table->mtt_idx_buddy, - ilog2(hr_dev->caps.num_idx_segs)); - if (ret) - goto err_buddy_idx; - } - - return 0; - -err_buddy_idx: - if (hr_dev->caps.num_srqwqe_segs) - hns_roce_buddy_cleanup(&mr_table->mtt_srqwqe_buddy); - -err_buddy_srqwqe: - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - hns_roce_buddy_cleanup(&mr_table->mtt_cqe_buddy); - -err_buddy_cqe: - hns_roce_buddy_cleanup(&mr_table->mtt_buddy); - -err_buddy: - hns_roce_bitmap_cleanup(&mr_table->mtpt_bitmap); return ret; } @@ -974,30 +223,24 @@ void hns_roce_cleanup_mr_table(struct hns_roce_dev *hr_dev) { struct hns_roce_mr_table *mr_table = &hr_dev->mr_table; - if (hr_dev->caps.num_idx_segs) - hns_roce_buddy_cleanup(&mr_table->mtt_idx_buddy); - if (hr_dev->caps.num_srqwqe_segs) - hns_roce_buddy_cleanup(&mr_table->mtt_srqwqe_buddy); - hns_roce_buddy_cleanup(&mr_table->mtt_buddy); - if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) - hns_roce_buddy_cleanup(&mr_table->mtt_cqe_buddy); hns_roce_bitmap_cleanup(&mr_table->mtpt_bitmap); } struct ib_mr *hns_roce_get_dma_mr(struct ib_pd *pd, int acc) { + struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); struct hns_roce_mr *mr; int ret; - mr = kmalloc(sizeof(*mr), GFP_KERNEL); + mr = kzalloc(sizeof(*mr), GFP_KERNEL); if (mr == NULL) return ERR_PTR(-ENOMEM); mr->type = MR_TYPE_DMA; /* Allocate memory region key */ - ret = hns_roce_mr_alloc(to_hr_dev(pd->device), to_hr_pd(pd)->pdn, 0, - ~0ULL, acc, 0, mr); + hns_roce_hem_list_init(&mr->pbl_mtr.hem_list); + ret = alloc_mr_key(hr_dev, mr, to_hr_pd(pd)->pdn, 0, 0, acc); if (ret) goto err_free; @@ -1006,203 +249,52 @@ struct ib_mr *hns_roce_get_dma_mr(struct ib_pd *pd, int acc) goto err_mr; mr->ibmr.rkey = mr->ibmr.lkey = mr->key; - mr->umem = NULL; return &mr->ibmr; - err_mr: - hns_roce_mr_free(to_hr_dev(pd->device), mr); + free_mr_key(hr_dev, mr); err_free: kfree(mr); return ERR_PTR(ret); } -int hns_roce_ib_umem_write_mtt(struct hns_roce_dev *hr_dev, - struct hns_roce_mtt *mtt, struct ib_umem *umem) -{ - struct device *dev = hr_dev->dev; - struct sg_dma_page_iter sg_iter; - unsigned int order; - int npage = 0; - int ret = 0; - int i; - u64 page_addr; - u64 *pages; - u32 bt_page_size; - u32 n; - - switch (mtt->mtt_type) { - case MTT_TYPE_WQE: - order = hr_dev->caps.mtt_ba_pg_sz; - break; - case MTT_TYPE_CQE: - order = hr_dev->caps.cqe_ba_pg_sz; - break; - case MTT_TYPE_SRQWQE: - order = hr_dev->caps.srqwqe_ba_pg_sz; - break; - case MTT_TYPE_IDX: - order = hr_dev->caps.idx_ba_pg_sz; - break; - default: - dev_err(dev, "Unsupport mtt type %d, write mtt failed\n", - mtt->mtt_type); - return -EINVAL; - } - - bt_page_size = 1 << (order + PAGE_SHIFT); - - pages = (u64 *) __get_free_pages(GFP_KERNEL, order); - if (!pages) - return -ENOMEM; - - i = n = 0; - - for_each_sg_dma_page(umem->sg_head.sgl, &sg_iter, umem->nmap, 0) { - page_addr = sg_page_iter_dma_address(&sg_iter); - if (!(npage % (1 << (mtt->page_shift - PAGE_SHIFT)))) { - if (page_addr & ((1 << mtt->page_shift) - 1)) { - dev_err(dev, - "page_addr is not page_shift %d alignment!\n", - mtt->page_shift); - ret = -EINVAL; - goto out; - } - pages[i++] = page_addr; - } - npage++; - if (i == bt_page_size / sizeof(u64)) { - ret = hns_roce_write_mtt(hr_dev, mtt, n, i, pages); - if (ret) - goto out; - n += i; - i = 0; - } - } - - if (i) - ret = hns_roce_write_mtt(hr_dev, mtt, n, i, pages); - -out: - free_pages((unsigned long) pages, order); - return ret; -} - -static int hns_roce_ib_umem_write_mr(struct hns_roce_dev *hr_dev, - struct hns_roce_mr *mr, - struct ib_umem *umem) -{ - struct sg_dma_page_iter sg_iter; - int i = 0, j = 0; - u64 page_addr; - u32 pbl_bt_sz; - - if (hr_dev->caps.pbl_hop_num == HNS_ROCE_HOP_NUM_0) - return 0; - - pbl_bt_sz = 1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT); - for_each_sg_dma_page(umem->sg_head.sgl, &sg_iter, umem->nmap, 0) { - page_addr = sg_page_iter_dma_address(&sg_iter); - if (!hr_dev->caps.pbl_hop_num) { - /* for hip06, page addr is aligned to 4K */ - mr->pbl_buf[i++] = page_addr >> 12; - } else if (hr_dev->caps.pbl_hop_num == 1) { - mr->pbl_buf[i++] = page_addr; - } else { - if (hr_dev->caps.pbl_hop_num == 2) - mr->pbl_bt_l1[i][j] = page_addr; - else if (hr_dev->caps.pbl_hop_num == 3) - mr->pbl_bt_l2[i][j] = page_addr; - - j++; - if (j >= (pbl_bt_sz / BA_BYTE_LEN)) { - i++; - j = 0; - } - } - } - - /* Memory barrier */ - mb(); - - return 0; -} - struct ib_mr *hns_roce_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, u64 virt_addr, int access_flags, struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); - struct device *dev = hr_dev->dev; struct hns_roce_mr *mr; - int bt_size; int ret; - int n; - int i; - mr = kmalloc(sizeof(*mr), GFP_KERNEL); + mr = kzalloc(sizeof(*mr), GFP_KERNEL); if (!mr) return ERR_PTR(-ENOMEM); - mr->umem = ib_umem_get(pd->device, start, length, access_flags); - if (IS_ERR(mr->umem)) { - ret = PTR_ERR(mr->umem); - goto err_free; - } - - n = ib_umem_page_count(mr->umem); - - if (!hr_dev->caps.pbl_hop_num) { - if (n > HNS_ROCE_MAX_MTPT_PBL_NUM) { - dev_err(dev, - " MR len %lld err. MR is limited to 4G at most!\n", - length); - ret = -EINVAL; - goto err_umem; - } - } else { - u64 pbl_size = 1; - - bt_size = (1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT)) / - BA_BYTE_LEN; - for (i = 0; i < hr_dev->caps.pbl_hop_num; i++) - pbl_size *= bt_size; - if (n > pbl_size) { - dev_err(dev, - " MR len %lld err. MR page num is limited to %lld!\n", - length, pbl_size); - ret = -EINVAL; - goto err_umem; - } - } - mr->type = MR_TYPE_MR; - - ret = hns_roce_mr_alloc(hr_dev, to_hr_pd(pd)->pdn, virt_addr, length, - access_flags, n, mr); + ret = alloc_mr_key(hr_dev, mr, to_hr_pd(pd)->pdn, virt_addr, length, + access_flags); if (ret) - goto err_umem; + goto err_alloc_mr; - ret = hns_roce_ib_umem_write_mr(hr_dev, mr, mr->umem); + ret = alloc_mr_pbl(hr_dev, mr, length, udata, start, access_flags); if (ret) - goto err_mr; + goto err_alloc_key; ret = hns_roce_mr_enable(hr_dev, mr); if (ret) - goto err_mr; + goto err_alloc_pbl; mr->ibmr.rkey = mr->ibmr.lkey = mr->key; + mr->ibmr.length = length; return &mr->ibmr; -err_mr: - hns_roce_mr_free(hr_dev, mr); - -err_umem: - ib_umem_release(mr->umem); - -err_free: +err_alloc_pbl: + free_mr_pbl(hr_dev, mr); +err_alloc_key: + free_mr_key(hr_dev, mr); +err_alloc_mr: kfree(mr); return ERR_PTR(ret); } @@ -1214,84 +306,36 @@ static int rereg_mr_trans(struct ib_mr *ibmr, int flags, u32 pdn, struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(ibmr->device); + struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_mr *mr = to_hr_mr(ibmr); - struct device *dev = hr_dev->dev; - int npages; int ret; - if (mr->size != ~0ULL) { - npages = ib_umem_page_count(mr->umem); - - if (hr_dev->caps.pbl_hop_num) - hns_roce_mhop_free(hr_dev, mr); - else - dma_free_coherent(dev, npages * 8, - mr->pbl_buf, mr->pbl_dma_addr); - } - ib_umem_release(mr->umem); - - mr->umem = ib_umem_get(ibmr->device, start, length, mr_access_flags); - if (IS_ERR(mr->umem)) { - ret = PTR_ERR(mr->umem); - mr->umem = NULL; - return -ENOMEM; - } - npages = ib_umem_page_count(mr->umem); - - if (hr_dev->caps.pbl_hop_num) { - ret = hns_roce_mhop_alloc(hr_dev, npages, mr); - if (ret) - goto release_umem; - } else { - mr->pbl_buf = dma_alloc_coherent(dev, npages * 8, - &(mr->pbl_dma_addr), - GFP_KERNEL); - if (!mr->pbl_buf) { - ret = -ENOMEM; - goto release_umem; - } + free_mr_pbl(hr_dev, mr); + ret = alloc_mr_pbl(hr_dev, mr, length, udata, start, mr_access_flags); + if (ret) { + ibdev_err(ibdev, "failed to create mr PBL, ret = %d.\n", ret); + return ret; } ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, pdn, mr_access_flags, virt_addr, length, mailbox->buf); - if (ret) - goto release_umem; - - - ret = hns_roce_ib_umem_write_mr(hr_dev, mr, mr->umem); if (ret) { - if (mr->size != ~0ULL) { - npages = ib_umem_page_count(mr->umem); - - if (hr_dev->caps.pbl_hop_num) - hns_roce_mhop_free(hr_dev, mr); - else - dma_free_coherent(dev, npages * 8, - mr->pbl_buf, - mr->pbl_dma_addr); - } - - goto release_umem; + ibdev_err(ibdev, "failed to write mtpt, ret = %d.\n", ret); + free_mr_pbl(hr_dev, mr); } - return 0; - -release_umem: - ib_umem_release(mr->umem); return ret; - } - int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, u64 virt_addr, int mr_access_flags, struct ib_pd *pd, struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(ibmr->device); + struct ib_device *ib_dev = &hr_dev->ib_dev; struct hns_roce_mr *mr = to_hr_mr(ibmr); struct hns_roce_cmd_mailbox *mailbox; - struct device *dev = hr_dev->dev; unsigned long mtpt_idx; u32 pdn = 0; int ret; @@ -1312,7 +356,7 @@ int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, ret = hns_roce_hw_destroy_mpt(hr_dev, NULL, mtpt_idx); if (ret) - dev_warn(dev, "DESTROY_MPT failed (%d)\n", ret); + ibdev_warn(ib_dev, "failed to destroy MPT, ret = %d.\n", ret); mr->enabled = 0; @@ -1336,8 +380,7 @@ int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, ret = hns_roce_hw_create_mpt(hr_dev, mailbox, mtpt_idx); if (ret) { - dev_err(dev, "CREATE_MPT failed (%d)\n", ret); - ib_umem_release(mr->umem); + ibdev_err(ib_dev, "failed to create MPT, ret = %d.\n", ret); goto free_cmd_mbox; } @@ -1365,8 +408,6 @@ int hns_roce_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata) ret = hr_dev->hw->dereg_mr(hr_dev, mr, udata); } else { hns_roce_mr_free(hr_dev, mr); - - ib_umem_release(mr->umem); kfree(mr); } @@ -1374,18 +415,14 @@ int hns_roce_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata) } struct ib_mr *hns_roce_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); struct device *dev = hr_dev->dev; struct hns_roce_mr *mr; u64 length; - u32 page_size; int ret; - page_size = 1 << (hr_dev->caps.pbl_buf_pg_sz + PAGE_SHIFT); - length = max_num_sg * page_size; - if (mr_type != IB_MR_TYPE_MEM_REG) return ERR_PTR(-EINVAL); @@ -1402,23 +439,28 @@ struct ib_mr *hns_roce_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, mr->type = MR_TYPE_FRMR; /* Allocate memory region key */ - ret = hns_roce_mr_alloc(hr_dev, to_hr_pd(pd)->pdn, 0, length, - 0, max_num_sg, mr); + length = max_num_sg * (1 << PAGE_SHIFT); + ret = alloc_mr_key(hr_dev, mr, to_hr_pd(pd)->pdn, 0, length, 0); if (ret) goto err_free; + ret = alloc_mr_pbl(hr_dev, mr, length, NULL, 0, 0); + if (ret) + goto err_key; + ret = hns_roce_mr_enable(hr_dev, mr); if (ret) - goto err_mr; + goto err_pbl; mr->ibmr.rkey = mr->ibmr.lkey = mr->key; - mr->umem = NULL; + mr->ibmr.length = length; return &mr->ibmr; -err_mr: - hns_roce_mr_free(to_hr_dev(pd->device), mr); - +err_key: + free_mr_key(hr_dev, mr); +err_pbl: + free_mr_pbl(hr_dev, mr); err_free: kfree(mr); return ERR_PTR(ret); @@ -1428,19 +470,54 @@ static int hns_roce_set_page(struct ib_mr *ibmr, u64 addr) { struct hns_roce_mr *mr = to_hr_mr(ibmr); - mr->pbl_buf[mr->npages++] = addr; + if (likely(mr->npages < mr->pbl_mtr.hem_cfg.buf_pg_count)) { + mr->page_list[mr->npages++] = addr; + return 0; + } - return 0; + return -ENOBUFS; } int hns_roce_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset) { + struct hns_roce_dev *hr_dev = to_hr_dev(ibmr->device); + struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_mr *mr = to_hr_mr(ibmr); + struct hns_roce_mtr *mtr = &mr->pbl_mtr; + int ret = 0; mr->npages = 0; + mr->page_list = kvcalloc(mr->pbl_mtr.hem_cfg.buf_pg_count, + sizeof(dma_addr_t), GFP_KERNEL); + if (!mr->page_list) + return ret; + + ret = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, hns_roce_set_page); + if (ret < 1) { + ibdev_err(ibdev, "failed to store sg pages %d %d, cnt = %d.\n", + mr->npages, mr->pbl_mtr.hem_cfg.buf_pg_count, ret); + goto err_page_list; + } + + mtr->hem_cfg.region[0].offset = 0; + mtr->hem_cfg.region[0].count = mr->npages; + mtr->hem_cfg.region[0].hopnum = mr->pbl_hop_num; + mtr->hem_cfg.region_count = 1; + ret = hns_roce_mtr_map(hr_dev, mtr, mr->page_list, mr->npages); + if (ret) { + ibdev_err(ibdev, "failed to map sg mtr, ret = %d.\n", ret); + ret = 0; + } else { + mr->pbl_mtr.hem_cfg.buf_pg_shift = ilog2(ibmr->page_size); + ret = mr->npages; + } + +err_page_list: + kvfree(mr->page_list); + mr->page_list = NULL; - return ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, hns_roce_set_page); + return ret; } static void hns_roce_mw_free(struct hns_roce_dev *hr_dev, @@ -1564,32 +641,23 @@ int hns_roce_dealloc_mw(struct ib_mw *ibmw) return 0; } -void hns_roce_mtr_init(struct hns_roce_mtr *mtr, int bt_pg_shift, - int buf_pg_shift) -{ - hns_roce_hem_list_init(&mtr->hem_list, bt_pg_shift); - mtr->buf_pg_shift = buf_pg_shift; -} - -void hns_roce_mtr_cleanup(struct hns_roce_dev *hr_dev, - struct hns_roce_mtr *mtr) -{ - hns_roce_hem_list_release(hr_dev, &mtr->hem_list); -} - -static int hns_roce_write_mtr(struct hns_roce_dev *hr_dev, - struct hns_roce_mtr *mtr, dma_addr_t *bufs, - struct hns_roce_buf_region *r) +static int mtr_map_region(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + dma_addr_t *pages, struct hns_roce_buf_region *region) { + __le64 *mtts; int offset; int count; int npage; - u64 *mtts; + u64 addr; int end; int i; - offset = r->offset; - end = offset + r->count; + /* if hopnum is 0, buffer cannot store BAs, so skip write mtt */ + if (!region->hopnum) + return 0; + + offset = region->offset; + end = offset + region->count; npage = 0; while (offset < end) { mtts = hns_roce_hem_list_find_mtt(hr_dev, &mtr->hem_list, @@ -1597,13 +665,13 @@ static int hns_roce_write_mtr(struct hns_roce_dev *hr_dev, if (!mtts) return -ENOBUFS; - /* Save page addr, low 12 bits : 0 */ for (i = 0; i < count; i++) { if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) - mtts[i] = bufs[npage] >> PAGE_ADDR_SHIFT; + addr = to_hr_hw_page_addr(pages[npage]); else - mtts[i] = bufs[npage]; + addr = pages[npage]; + mtts[i] = cpu_to_le64(addr); npage++; } offset += count; @@ -1612,69 +680,460 @@ static int hns_roce_write_mtr(struct hns_roce_dev *hr_dev, return 0; } -int hns_roce_mtr_attach(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, - dma_addr_t **bufs, struct hns_roce_buf_region *regions, - int region_cnt) +static inline bool mtr_has_mtt(struct hns_roce_buf_attr *attr) { - struct hns_roce_buf_region *r; - int ret; int i; - ret = hns_roce_hem_list_request(hr_dev, &mtr->hem_list, regions, - region_cnt); - if (ret) - return ret; + for (i = 0; i < attr->region_count; i++) + if (attr->region[i].hopnum != HNS_ROCE_HOP_NUM_0 && + attr->region[i].hopnum > 0) + return true; + + /* because the mtr only one root base address, when hopnum is 0 means + * root base address equals the first buffer address, thus all alloced + * memory must in a continuous space accessed by direct mode. + */ + return false; +} + +static inline size_t mtr_bufs_size(struct hns_roce_buf_attr *attr) +{ + size_t size = 0; + int i; - for (i = 0; i < region_cnt; i++) { - r = ®ions[i]; - ret = hns_roce_write_mtr(hr_dev, mtr, bufs[i], r); + for (i = 0; i < attr->region_count; i++) + size += attr->region[i].size; + + return size; +} + +static inline int mtr_umem_page_count(struct ib_umem *umem, + unsigned int page_shift) +{ + int count = ib_umem_page_count(umem); + + if (page_shift >= PAGE_SHIFT) + count >>= page_shift - PAGE_SHIFT; + else + count <<= PAGE_SHIFT - page_shift; + + return count; +} + +static inline size_t mtr_kmem_direct_size(bool is_direct, size_t alloc_size, + unsigned int page_shift) +{ + if (is_direct) + return ALIGN(alloc_size, 1 << page_shift); + else + return HNS_HW_DIRECT_PAGE_COUNT << page_shift; +} + +/* + * check the given pages in continuous address space + * Returns 0 on success, or the error page num. + */ +static inline int mtr_check_direct_pages(dma_addr_t *pages, int page_count, + unsigned int page_shift) +{ + size_t page_size = 1 << page_shift; + int i; + + for (i = 1; i < page_count; i++) + if (pages[i] - pages[i - 1] != page_size) + return i; + + return 0; +} + +static void mtr_free_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr) +{ + /* release user buffers */ + if (mtr->umem) { + ib_umem_release(mtr->umem); + mtr->umem = NULL; + } + + /* release kernel buffers */ + if (mtr->kmem) { + hns_roce_buf_free(hr_dev, mtr->kmem); + kfree(mtr->kmem); + mtr->kmem = NULL; + } +} + +static int mtr_alloc_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + struct hns_roce_buf_attr *buf_attr, bool is_direct, + struct ib_udata *udata, unsigned long user_addr) +{ + struct ib_device *ibdev = &hr_dev->ib_dev; + unsigned int max_pg_shift = buf_attr->page_shift; + unsigned int best_pg_shift = 0; + int all_pg_count = 0; + size_t direct_size; + size_t total_size; + unsigned long tmp; + int ret = 0; + + total_size = mtr_bufs_size(buf_attr); + if (total_size < 1) { + ibdev_err(ibdev, "Failed to check mtr size\n"); + return -EINVAL; + } + + if (udata) { + mtr->kmem = NULL; + mtr->umem = ib_umem_get(ibdev, user_addr, total_size, + buf_attr->user_access); + if (IS_ERR_OR_NULL(mtr->umem)) { + ibdev_err(ibdev, "Failed to get umem, ret %ld\n", + PTR_ERR(mtr->umem)); + return -ENOMEM; + } + if (buf_attr->fixed_page) { + best_pg_shift = max_pg_shift; + } else { + tmp = GENMASK(max_pg_shift, 0); + ret = ib_umem_find_best_pgsz(mtr->umem, tmp, user_addr); + best_pg_shift = (ret <= PAGE_SIZE) ? + PAGE_SHIFT : ilog2(ret); + } + all_pg_count = mtr_umem_page_count(mtr->umem, best_pg_shift); + ret = 0; + } else { + mtr->umem = NULL; + mtr->kmem = kzalloc(sizeof(*mtr->kmem), GFP_KERNEL); + if (!mtr->kmem) { + ibdev_err(ibdev, "Failed to alloc kmem\n"); + return -ENOMEM; + } + direct_size = mtr_kmem_direct_size(is_direct, total_size, + max_pg_shift); + ret = hns_roce_buf_alloc(hr_dev, total_size, direct_size, + mtr->kmem, max_pg_shift); if (ret) { - dev_err(hr_dev->dev, - "write mtr[%d/%d] err %d,offset=%d.\n", - i, region_cnt, ret, r->offset); - goto err_write; + ibdev_err(ibdev, "Failed to alloc kmem, ret %d\n", ret); + goto err_alloc_mem; + } else { + best_pg_shift = max_pg_shift; + all_pg_count = mtr->kmem->npages; } } - return 0; + /* must bigger than minimum hardware page shift */ + if (best_pg_shift < HNS_HW_PAGE_SHIFT || all_pg_count < 1) { + ret = -EINVAL; + ibdev_err(ibdev, "Failed to check mtr page shift %d count %d\n", + best_pg_shift, all_pg_count); + goto err_alloc_mem; + } -err_write: - hns_roce_hem_list_release(hr_dev, &mtr->hem_list); + mtr->hem_cfg.buf_pg_shift = best_pg_shift; + mtr->hem_cfg.buf_pg_count = all_pg_count; + return 0; +err_alloc_mem: + mtr_free_bufs(hr_dev, mtr); return ret; } +static int mtr_get_pages(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + dma_addr_t *pages, int count, unsigned int page_shift) +{ + struct ib_device *ibdev = &hr_dev->ib_dev; + int npage; + int err; + + if (mtr->umem) + npage = hns_roce_get_umem_bufs(hr_dev, pages, count, 0, + mtr->umem, page_shift); + else + npage = hns_roce_get_kmem_bufs(hr_dev, pages, count, 0, + mtr->kmem); + + if (mtr->hem_cfg.is_direct && npage > 1) { + err = mtr_check_direct_pages(pages, npage, page_shift); + if (err) { + ibdev_err(ibdev, "Failed to check %s direct page-%d\n", + mtr->umem ? "user" : "kernel", err); + npage = err; + } + } + + return npage; +} + +int hns_roce_mtr_map(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + dma_addr_t *pages, int page_cnt) +{ + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_buf_region *r; + int err; + int i; + + /* + * Only use the first page address as root ba when hopnum is 0, this + * is because the addresses of all pages are consecutive in this case. + */ + if (mtr->hem_cfg.is_direct) { + mtr->hem_cfg.root_ba = pages[0]; + return 0; + } + + for (i = 0; i < mtr->hem_cfg.region_count; i++) { + r = &mtr->hem_cfg.region[i]; + if (r->offset + r->count > page_cnt) { + err = -EINVAL; + ibdev_err(ibdev, + "Failed to check mtr%d end %d + %d, max %d\n", + i, r->offset, r->count, page_cnt); + return err; + } + + err = mtr_map_region(hr_dev, mtr, &pages[r->offset], r); + if (err) { + ibdev_err(ibdev, + "Failed to map mtr%d offset %d, err %d\n", + i, r->offset, err); + return err; + } + } + + return 0; +} + int hns_roce_mtr_find(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, int offset, u64 *mtt_buf, int mtt_max, u64 *base_addr) { - u64 *mtts = mtt_buf; + struct hns_roce_hem_cfg *cfg = &mtr->hem_cfg; + int start_index; int mtt_count; int total = 0; - u64 *addr; + __le64 *mtts; int npage; + u64 addr; int left; - if (mtts == NULL || mtt_max < 1) + if (!mtt_buf || mtt_max < 1) goto done; + /* no mtt memory in direct mode, so just return the buffer address */ + if (cfg->is_direct) { + start_index = offset >> HNS_HW_PAGE_SHIFT; + for (mtt_count = 0; mtt_count < cfg->region_count && + total < mtt_max; mtt_count++) { + npage = cfg->region[mtt_count].offset; + if (npage < start_index) + continue; + + addr = cfg->root_ba + (npage << HNS_HW_PAGE_SHIFT); + if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) + mtt_buf[total] = to_hr_hw_page_addr(addr); + else + mtt_buf[total] = addr; + + total++; + } + + goto done; + } + + start_index = offset >> cfg->buf_pg_shift; left = mtt_max; while (left > 0) { mtt_count = 0; - addr = hns_roce_hem_list_find_mtt(hr_dev, &mtr->hem_list, - offset + total, + mtts = hns_roce_hem_list_find_mtt(hr_dev, &mtr->hem_list, + start_index + total, &mtt_count, NULL); - if (!addr || !mtt_count) + if (!mtts || !mtt_count) goto done; npage = min(mtt_count, left); - memcpy(&mtts[total], addr, BA_BYTE_LEN * npage); left -= npage; - total += npage; + for (mtt_count = 0; mtt_count < npage; mtt_count++) + mtt_buf[total++] = le64_to_cpu(mtts[mtt_count]); } done: if (base_addr) - *base_addr = mtr->hem_list.root_ba; + *base_addr = cfg->root_ba; return total; } + +static int mtr_init_buf_cfg(struct hns_roce_dev *hr_dev, + struct hns_roce_buf_attr *attr, + struct hns_roce_hem_cfg *cfg, + unsigned int *buf_page_shift) +{ + struct hns_roce_buf_region *r; + unsigned int page_shift = 0; + int page_cnt = 0; + size_t buf_size; + int region_cnt; + + if (cfg->is_direct) { + buf_size = cfg->buf_pg_count << cfg->buf_pg_shift; + page_cnt = DIV_ROUND_UP(buf_size, HNS_HW_PAGE_SIZE); + /* + * When HEM buffer use level-0 addressing, the page size equals + * the buffer size, and the the page size = 4K * 2^N. + */ + cfg->buf_pg_shift = HNS_HW_PAGE_SHIFT + order_base_2(page_cnt); + if (attr->region_count > 1) { + cfg->buf_pg_count = page_cnt; + page_shift = HNS_HW_PAGE_SHIFT; + } else { + cfg->buf_pg_count = 1; + page_shift = cfg->buf_pg_shift; + if (buf_size != 1 << page_shift) { + ibdev_err(&hr_dev->ib_dev, + "failed to check direct size %zu shift %d.\n", + buf_size, page_shift); + return -EINVAL; + } + } + } else { + page_shift = cfg->buf_pg_shift; + } + + /* convert buffer size to page index and page count */ + for (page_cnt = 0, region_cnt = 0; page_cnt < cfg->buf_pg_count && + region_cnt < attr->region_count && + region_cnt < ARRAY_SIZE(cfg->region); region_cnt++) { + r = &cfg->region[region_cnt]; + r->offset = page_cnt; + buf_size = hr_hw_page_align(attr->region[region_cnt].size); + r->count = DIV_ROUND_UP(buf_size, 1 << page_shift); + page_cnt += r->count; + r->hopnum = to_hr_hem_hopnum(attr->region[region_cnt].hopnum, + r->count); + } + + if (region_cnt < 1) { + ibdev_err(&hr_dev->ib_dev, + "failed to check mtr region count, pages = %d.\n", + cfg->buf_pg_count); + return -ENOBUFS; + } + + cfg->region_count = region_cnt; + *buf_page_shift = page_shift; + + return page_cnt; +} + +/** + * hns_roce_mtr_create - Create hns memory translate region. + * + * @mtr: memory translate region + * @buf_attr: buffer attribute for creating mtr + * @ba_page_shift: page shift for multi-hop base address table + * @udata: user space context, if it's NULL, means kernel space + * @user_addr: userspace virtual address to start at + */ +int hns_roce_mtr_create(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + struct hns_roce_buf_attr *buf_attr, + unsigned int ba_page_shift, struct ib_udata *udata, + unsigned long user_addr) +{ + struct hns_roce_hem_cfg *cfg = &mtr->hem_cfg; + struct ib_device *ibdev = &hr_dev->ib_dev; + unsigned int buf_page_shift = 0; + dma_addr_t *pages = NULL; + int all_pg_cnt; + int get_pg_cnt; + int ret = 0; + + /* if disable mtt, all pages must in a continuous address range */ + cfg->is_direct = !mtr_has_mtt(buf_attr); + + /* if buffer only need mtt, just init the hem cfg */ + if (buf_attr->mtt_only) { + cfg->buf_pg_shift = buf_attr->page_shift; + cfg->buf_pg_count = mtr_bufs_size(buf_attr) >> + buf_attr->page_shift; + mtr->umem = NULL; + mtr->kmem = NULL; + } else { + ret = mtr_alloc_bufs(hr_dev, mtr, buf_attr, cfg->is_direct, + udata, user_addr); + if (ret) { + ibdev_err(ibdev, + "failed to alloc mtr bufs, ret = %d.\n", ret); + return ret; + } + } + + all_pg_cnt = mtr_init_buf_cfg(hr_dev, buf_attr, cfg, &buf_page_shift); + if (all_pg_cnt < 1) { + ret = -ENOBUFS; + ibdev_err(ibdev, "failed to init mtr buf cfg.\n"); + goto err_alloc_bufs; + } + + hns_roce_hem_list_init(&mtr->hem_list); + if (!cfg->is_direct) { + ret = hns_roce_hem_list_request(hr_dev, &mtr->hem_list, + cfg->region, cfg->region_count, + ba_page_shift); + if (ret) { + ibdev_err(ibdev, "failed to request mtr hem, ret = %d.\n", + ret); + goto err_alloc_bufs; + } + cfg->root_ba = mtr->hem_list.root_ba; + cfg->ba_pg_shift = ba_page_shift; + } else { + cfg->ba_pg_shift = cfg->buf_pg_shift; + } + + /* no buffer to map */ + if (buf_attr->mtt_only) + return 0; + + /* alloc a tmp array to store buffer's dma address */ + pages = kvcalloc(all_pg_cnt, sizeof(dma_addr_t), GFP_KERNEL); + if (!pages) { + ret = -ENOMEM; + ibdev_err(ibdev, "failed to alloc mtr page list %d.\n", + all_pg_cnt); + goto err_alloc_hem_list; + } + + get_pg_cnt = mtr_get_pages(hr_dev, mtr, pages, all_pg_cnt, + buf_page_shift); + if (get_pg_cnt != all_pg_cnt) { + ibdev_err(ibdev, "failed to get mtr page %d != %d.\n", + get_pg_cnt, all_pg_cnt); + ret = -ENOBUFS; + goto err_alloc_page_list; + } + + /* write buffer's dma address to BA table */ + ret = hns_roce_mtr_map(hr_dev, mtr, pages, all_pg_cnt); + if (ret) { + ibdev_err(ibdev, "failed to map mtr pages, ret = %d.\n", ret); + goto err_alloc_page_list; + } + + /* drop tmp array */ + kvfree(pages); + return 0; +err_alloc_page_list: + kvfree(pages); +err_alloc_hem_list: + hns_roce_hem_list_release(hr_dev, &mtr->hem_list); +err_alloc_bufs: + mtr_free_bufs(hr_dev, mtr); + return ret; +} + +void hns_roce_mtr_destroy(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr) +{ + /* release multi-hop addressing resource */ + hns_roce_hem_list_release(hr_dev, &mtr->hem_list); + + /* free buffers */ + mtr_free_bufs(hr_dev, mtr); +} diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c index 6317901c4b4f..c063c450c715 100644 --- a/drivers/infiniband/hw/hns/hns_roce_qp.c +++ b/drivers/infiniband/hw/hns/hns_roce_qp.c @@ -355,16 +355,16 @@ static void free_qpn(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp) hns_roce_bitmap_free_range(&qp_table->bitmap, hr_qp->qpn, 1, BITMAP_RR); } -static int set_rq_size(struct hns_roce_dev *hr_dev, - struct ib_qp_cap *cap, bool is_user, int has_rq, - struct hns_roce_qp *hr_qp) +static int set_rq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, + struct hns_roce_qp *hr_qp, int has_rq) { - u32 max_cnt; + u32 cnt; /* If srq exist, set zero for relative number of rq */ if (!has_rq) { hr_qp->rq.wqe_cnt = 0; hr_qp->rq.max_gs = 0; + hr_qp->rq_inl_buf.wqe_cnt = 0; cap->max_recv_wr = 0; cap->max_recv_sge = 0; @@ -379,17 +379,14 @@ static int set_rq_size(struct hns_roce_dev *hr_dev, return -EINVAL; } - max_cnt = max(cap->max_recv_wr, hr_dev->caps.min_wqes); - - hr_qp->rq.wqe_cnt = roundup_pow_of_two(max_cnt); - if ((u32)hr_qp->rq.wqe_cnt > hr_dev->caps.max_wqes) { + cnt = roundup_pow_of_two(max(cap->max_recv_wr, hr_dev->caps.min_wqes)); + if (cnt > hr_dev->caps.max_wqes) { ibdev_err(&hr_dev->ib_dev, "rq depth %u too large\n", cap->max_recv_wr); return -EINVAL; } - max_cnt = max(1U, cap->max_recv_sge); - hr_qp->rq.max_gs = roundup_pow_of_two(max_cnt); + hr_qp->rq.max_gs = roundup_pow_of_two(max(1U, cap->max_recv_sge)); if (hr_dev->caps.max_rq_sg <= HNS_ROCE_SGE_IN_WQE) hr_qp->rq.wqe_shift = ilog2(hr_dev->caps.max_rq_desc_sz); @@ -397,12 +394,51 @@ static int set_rq_size(struct hns_roce_dev *hr_dev, hr_qp->rq.wqe_shift = ilog2(hr_dev->caps.max_rq_desc_sz * hr_qp->rq.max_gs); - cap->max_recv_wr = hr_qp->rq.wqe_cnt; + hr_qp->rq.wqe_cnt = cnt; + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) + hr_qp->rq_inl_buf.wqe_cnt = cnt; + else + hr_qp->rq_inl_buf.wqe_cnt = 0; + + cap->max_recv_wr = cnt; cap->max_recv_sge = hr_qp->rq.max_gs; return 0; } +static int set_extend_sge_param(struct hns_roce_dev *hr_dev, u32 sq_wqe_cnt, + struct hns_roce_qp *hr_qp, + struct ib_qp_cap *cap) +{ + u32 cnt; + + cnt = max(1U, cap->max_send_sge); + if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) { + hr_qp->sq.max_gs = roundup_pow_of_two(cnt); + hr_qp->sge.sge_cnt = 0; + + return 0; + } + + hr_qp->sq.max_gs = cnt; + + /* UD sqwqe's sge use extend sge */ + if (hr_qp->ibqp.qp_type == IB_QPT_GSI || + hr_qp->ibqp.qp_type == IB_QPT_UD) { + cnt = roundup_pow_of_two(sq_wqe_cnt * hr_qp->sq.max_gs); + } else if (hr_qp->sq.max_gs > HNS_ROCE_SGE_IN_WQE) { + cnt = roundup_pow_of_two(sq_wqe_cnt * + (hr_qp->sq.max_gs - HNS_ROCE_SGE_IN_WQE)); + } else { + cnt = 0; + } + + hr_qp->sge.sge_shift = HNS_ROCE_SGE_SHIFT; + hr_qp->sge.sge_cnt = cnt; + + return 0; +} + static int check_sq_size_with_integrity(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, struct hns_roce_ib_create_qp *ucmd) @@ -430,174 +466,79 @@ static int set_user_sq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, struct hns_roce_qp *hr_qp, struct hns_roce_ib_create_qp *ucmd) { - u32 ex_sge_num; - u32 page_size; - u32 max_cnt; + struct ib_device *ibdev = &hr_dev->ib_dev; + u32 cnt = 0; int ret; - if (check_shl_overflow(1, ucmd->log_sq_bb_count, &hr_qp->sq.wqe_cnt) || - hr_qp->sq.wqe_cnt > hr_dev->caps.max_wqes) + if (check_shl_overflow(1, ucmd->log_sq_bb_count, &cnt) || + cnt > hr_dev->caps.max_wqes) return -EINVAL; ret = check_sq_size_with_integrity(hr_dev, cap, ucmd); if (ret) { - ibdev_err(&hr_dev->ib_dev, "Failed to check user SQ size limit\n"); + ibdev_err(ibdev, "failed to check user SQ size, ret = %d.\n", + ret); return ret; } - hr_qp->sq.wqe_shift = ucmd->log_sq_stride; - - max_cnt = max(1U, cap->max_send_sge); - if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) - hr_qp->sq.max_gs = roundup_pow_of_two(max_cnt); - else - hr_qp->sq.max_gs = max_cnt; - - if (hr_qp->sq.max_gs > HNS_ROCE_SGE_IN_WQE) - hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * - (hr_qp->sq.max_gs - 2)); - - if (hr_qp->sq.max_gs > HNS_ROCE_SGE_IN_WQE && - hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP08_A) { - if (hr_qp->sge.sge_cnt > hr_dev->caps.max_extend_sg) { - ibdev_err(&hr_dev->ib_dev, - "Failed to check extended SGE size limit %d\n", - hr_qp->sge.sge_cnt); - return -EINVAL; - } - } - - hr_qp->sge.sge_shift = 4; - ex_sge_num = hr_qp->sge.sge_cnt; + ret = set_extend_sge_param(hr_dev, cnt, hr_qp, cap); + if (ret) + return ret; - /* Get buf size, SQ and RQ are aligned to page_szie */ - if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) { - hr_qp->buff_size = round_up((hr_qp->rq.wqe_cnt << - hr_qp->rq.wqe_shift), PAGE_SIZE) + - round_up((hr_qp->sq.wqe_cnt << - hr_qp->sq.wqe_shift), PAGE_SIZE); - - hr_qp->sq.offset = 0; - hr_qp->rq.offset = round_up((hr_qp->sq.wqe_cnt << - hr_qp->sq.wqe_shift), PAGE_SIZE); - } else { - page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT); - hr_qp->sge.sge_cnt = ex_sge_num ? - max(page_size / (1 << hr_qp->sge.sge_shift), ex_sge_num) : 0; - hr_qp->buff_size = round_up((hr_qp->rq.wqe_cnt << - hr_qp->rq.wqe_shift), page_size) + - round_up((hr_qp->sge.sge_cnt << - hr_qp->sge.sge_shift), page_size) + - round_up((hr_qp->sq.wqe_cnt << - hr_qp->sq.wqe_shift), page_size); - - hr_qp->sq.offset = 0; - if (ex_sge_num) { - hr_qp->sge.offset = round_up((hr_qp->sq.wqe_cnt << - hr_qp->sq.wqe_shift), - page_size); - hr_qp->rq.offset = hr_qp->sge.offset + - round_up((hr_qp->sge.sge_cnt << - hr_qp->sge.sge_shift), - page_size); - } else { - hr_qp->rq.offset = round_up((hr_qp->sq.wqe_cnt << - hr_qp->sq.wqe_shift), - page_size); - } - } + hr_qp->sq.wqe_shift = ucmd->log_sq_stride; + hr_qp->sq.wqe_cnt = cnt; return 0; } -static int split_wqe_buf_region(struct hns_roce_dev *hr_dev, - struct hns_roce_qp *hr_qp, - struct hns_roce_buf_region *regions, - int region_max, int page_shift) +static int set_wqe_buf_attr(struct hns_roce_dev *hr_dev, + struct hns_roce_qp *hr_qp, + struct hns_roce_buf_attr *buf_attr) { - int page_size = 1 << page_shift; - bool is_extend_sge; - int region_cnt = 0; int buf_size; - int buf_cnt; - - if (hr_qp->buff_size < 1 || region_max < 1) - return region_cnt; - - if (hr_qp->sge.sge_cnt > 0) - is_extend_sge = true; - else - is_extend_sge = false; - - /* sq region */ - if (is_extend_sge) - buf_size = hr_qp->sge.offset - hr_qp->sq.offset; - else - buf_size = hr_qp->rq.offset - hr_qp->sq.offset; - - if (buf_size > 0 && region_cnt < region_max) { - buf_cnt = DIV_ROUND_UP(buf_size, page_size); - hns_roce_init_buf_region(®ions[region_cnt], - hr_dev->caps.wqe_sq_hop_num, - hr_qp->sq.offset / page_size, - buf_cnt); - region_cnt++; - } + int idx = 0; - /* sge region */ - if (is_extend_sge) { - buf_size = hr_qp->rq.offset - hr_qp->sge.offset; - if (buf_size > 0 && region_cnt < region_max) { - buf_cnt = DIV_ROUND_UP(buf_size, page_size); - hns_roce_init_buf_region(®ions[region_cnt], - hr_dev->caps.wqe_sge_hop_num, - hr_qp->sge.offset / page_size, - buf_cnt); - region_cnt++; - } - } - - /* rq region */ - buf_size = hr_qp->buff_size - hr_qp->rq.offset; - if (buf_size > 0) { - buf_cnt = DIV_ROUND_UP(buf_size, page_size); - hns_roce_init_buf_region(®ions[region_cnt], - hr_dev->caps.wqe_rq_hop_num, - hr_qp->rq.offset / page_size, - buf_cnt); - region_cnt++; - } + hr_qp->buff_size = 0; - return region_cnt; -} - -static int set_extend_sge_param(struct hns_roce_dev *hr_dev, - struct hns_roce_qp *hr_qp) -{ - struct device *dev = hr_dev->dev; - - if (hr_qp->sq.max_gs > 2) { - hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * - (hr_qp->sq.max_gs - 2)); - hr_qp->sge.sge_shift = 4; - } - - /* ud sqwqe's sge use extend sge */ - if (hr_dev->hw_rev != HNS_ROCE_HW_VER1 && - hr_qp->ibqp.qp_type == IB_QPT_GSI) { - hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * - hr_qp->sq.max_gs); - hr_qp->sge.sge_shift = 4; - } + /* SQ WQE */ + hr_qp->sq.offset = 0; + buf_size = to_hr_hem_entries_size(hr_qp->sq.wqe_cnt, + hr_qp->sq.wqe_shift); + if (buf_size > 0 && idx < ARRAY_SIZE(buf_attr->region)) { + buf_attr->region[idx].size = buf_size; + buf_attr->region[idx].hopnum = hr_dev->caps.wqe_sq_hop_num; + idx++; + hr_qp->buff_size += buf_size; + } + + /* extend SGE WQE in SQ */ + hr_qp->sge.offset = hr_qp->buff_size; + buf_size = to_hr_hem_entries_size(hr_qp->sge.sge_cnt, + hr_qp->sge.sge_shift); + if (buf_size > 0 && idx < ARRAY_SIZE(buf_attr->region)) { + buf_attr->region[idx].size = buf_size; + buf_attr->region[idx].hopnum = hr_dev->caps.wqe_sge_hop_num; + idx++; + hr_qp->buff_size += buf_size; + } + + /* RQ WQE */ + hr_qp->rq.offset = hr_qp->buff_size; + buf_size = to_hr_hem_entries_size(hr_qp->rq.wqe_cnt, + hr_qp->rq.wqe_shift); + if (buf_size > 0 && idx < ARRAY_SIZE(buf_attr->region)) { + buf_attr->region[idx].size = buf_size; + buf_attr->region[idx].hopnum = hr_dev->caps.wqe_rq_hop_num; + idx++; + hr_qp->buff_size += buf_size; + } + + if (hr_qp->buff_size < 1) + return -EINVAL; - if (hr_qp->sq.max_gs > 2 && - hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP08_A) { - if (hr_qp->sge.sge_cnt > hr_dev->caps.max_extend_sg) { - dev_err(dev, "The extended sge cnt error! sge_cnt=%d\n", - hr_qp->sge.sge_cnt); - return -EINVAL; - } - } + buf_attr->page_shift = HNS_HW_PAGE_SHIFT + hr_dev->caps.mtt_buf_pg_sz; + buf_attr->fixed_page = true; + buf_attr->region_count = idx; return 0; } @@ -605,62 +546,35 @@ static int set_extend_sge_param(struct hns_roce_dev *hr_dev, static int set_kernel_sq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, struct hns_roce_qp *hr_qp) { - u32 page_size; - u32 max_cnt; - int size; + struct ib_device *ibdev = &hr_dev->ib_dev; + u32 cnt; int ret; if (!cap->max_send_wr || cap->max_send_wr > hr_dev->caps.max_wqes || cap->max_send_sge > hr_dev->caps.max_sq_sg || cap->max_inline_data > hr_dev->caps.max_sq_inline) { - ibdev_err(&hr_dev->ib_dev, - "SQ WR or sge or inline data error!\n"); + ibdev_err(ibdev, + "failed to check SQ WR, SGE or inline num, ret = %d.\n", + -EINVAL); return -EINVAL; } - hr_qp->sq.wqe_shift = ilog2(hr_dev->caps.max_sq_desc_sz); - - max_cnt = max(cap->max_send_wr, hr_dev->caps.min_wqes); - - hr_qp->sq.wqe_cnt = roundup_pow_of_two(max_cnt); - if ((u32)hr_qp->sq.wqe_cnt > hr_dev->caps.max_wqes) { - ibdev_err(&hr_dev->ib_dev, - "while setting kernel sq size, sq.wqe_cnt too large\n"); + cnt = roundup_pow_of_two(max(cap->max_send_wr, hr_dev->caps.min_wqes)); + if (cnt > hr_dev->caps.max_wqes) { + ibdev_err(ibdev, "failed to check WQE num, WQE num = %d.\n", + cnt); return -EINVAL; } - /* Get data_seg numbers */ - max_cnt = max(1U, cap->max_send_sge); - if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) - hr_qp->sq.max_gs = roundup_pow_of_two(max_cnt); - else - hr_qp->sq.max_gs = max_cnt; + hr_qp->sq.wqe_shift = ilog2(hr_dev->caps.max_sq_desc_sz); + hr_qp->sq.wqe_cnt = cnt; - ret = set_extend_sge_param(hr_dev, hr_qp); - if (ret) { - ibdev_err(&hr_dev->ib_dev, "set extend sge parameters fail\n"); + ret = set_extend_sge_param(hr_dev, cnt, hr_qp, cap); + if (ret) return ret; - } - - /* Get buf size, SQ and RQ are aligned to PAGE_SIZE */ - page_size = 1 << (hr_dev->caps.mtt_buf_pg_sz + PAGE_SHIFT); - hr_qp->sq.offset = 0; - size = round_up(hr_qp->sq.wqe_cnt << hr_qp->sq.wqe_shift, page_size); - - if (hr_dev->hw_rev != HNS_ROCE_HW_VER1 && hr_qp->sge.sge_cnt) { - hr_qp->sge.sge_cnt = max(page_size/(1 << hr_qp->sge.sge_shift), - (u32)hr_qp->sge.sge_cnt); - hr_qp->sge.offset = size; - size += round_up(hr_qp->sge.sge_cnt << hr_qp->sge.sge_shift, - page_size); - } - hr_qp->rq.offset = size; - size += round_up((hr_qp->rq.wqe_cnt << hr_qp->rq.wqe_shift), page_size); - hr_qp->buff_size = size; - - /* Get wr and sge number which send */ - cap->max_send_wr = hr_qp->sq.wqe_cnt; + /* sync the parameters of kernel QP to user's configuration */ + cap->max_send_wr = cnt; cap->max_send_sge = hr_qp->sq.max_gs; /* We don't support inline sends for kernel QPs (yet) */ @@ -691,8 +605,8 @@ static int alloc_rq_inline_buf(struct hns_roce_qp *hr_qp, struct ib_qp_init_attr *init_attr) { u32 max_recv_sge = init_attr->cap.max_recv_sge; + u32 wqe_cnt = hr_qp->rq_inl_buf.wqe_cnt; struct hns_roce_rinl_wqe *wqe_list; - u32 wqe_cnt = hr_qp->rq.wqe_cnt; int i; /* allocate recv inline buf */ @@ -714,7 +628,6 @@ static int alloc_rq_inline_buf(struct hns_roce_qp *hr_qp, wqe_list[i].sg_list = &wqe_list[0].sg_list[i * max_recv_sge]; hr_qp->rq_inl_buf.wqe_list = wqe_list; - hr_qp->rq_inl_buf.wqe_cnt = wqe_cnt; return 0; @@ -727,140 +640,55 @@ err: static void free_rq_inline_buf(struct hns_roce_qp *hr_qp) { - kfree(hr_qp->rq_inl_buf.wqe_list[0].sg_list); + if (hr_qp->rq_inl_buf.wqe_list) + kfree(hr_qp->rq_inl_buf.wqe_list[0].sg_list); kfree(hr_qp->rq_inl_buf.wqe_list); } -static int map_wqe_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, - u32 page_shift, bool is_user) -{ -/* WQE buffer include 3 parts: SQ, extend SGE and RQ. */ -#define HNS_ROCE_WQE_REGION_MAX 3 - struct hns_roce_buf_region regions[HNS_ROCE_WQE_REGION_MAX] = {}; - dma_addr_t *buf_list[HNS_ROCE_WQE_REGION_MAX] = {}; - struct ib_device *ibdev = &hr_dev->ib_dev; - struct hns_roce_buf_region *r; - int region_count; - int buf_count; - int ret; - int i; - - region_count = split_wqe_buf_region(hr_dev, hr_qp, regions, - ARRAY_SIZE(regions), page_shift); - - /* alloc a tmp list to store WQE buffers address */ - ret = hns_roce_alloc_buf_list(regions, buf_list, region_count); - if (ret) { - ibdev_err(ibdev, "Failed to alloc WQE buffer list\n"); - return ret; - } - - for (i = 0; i < region_count; i++) { - r = ®ions[i]; - if (is_user) - buf_count = hns_roce_get_umem_bufs(hr_dev, buf_list[i], - r->count, r->offset, hr_qp->umem, - page_shift); - else - buf_count = hns_roce_get_kmem_bufs(hr_dev, buf_list[i], - r->count, r->offset, &hr_qp->hr_buf); - - if (buf_count != r->count) { - ibdev_err(ibdev, "Failed to get %s WQE buf, expect %d = %d.\n", - is_user ? "user" : "kernel", - r->count, buf_count); - ret = -ENOBUFS; - goto done; - } - } - - hr_qp->wqe_bt_pg_shift = hr_dev->caps.mtt_ba_pg_sz; - hns_roce_mtr_init(&hr_qp->mtr, PAGE_SHIFT + hr_qp->wqe_bt_pg_shift, - page_shift); - ret = hns_roce_mtr_attach(hr_dev, &hr_qp->mtr, buf_list, regions, - region_count); - if (ret) - ibdev_err(ibdev, "Failed to attach WQE's mtr\n"); - - goto done; - - hns_roce_mtr_cleanup(hr_dev, &hr_qp->mtr); -done: - hns_roce_free_buf_list(buf_list, region_count); - - return ret; -} - static int alloc_qp_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, struct ib_qp_init_attr *init_attr, struct ib_udata *udata, unsigned long addr) { - u32 page_shift = PAGE_SHIFT + hr_dev->caps.mtt_buf_pg_sz; struct ib_device *ibdev = &hr_dev->ib_dev; - bool is_rq_buf_inline; + struct hns_roce_buf_attr buf_attr = {}; int ret; - is_rq_buf_inline = (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) && - hns_roce_qp_has_rq(init_attr); - if (is_rq_buf_inline) { + if (!udata && hr_qp->rq_inl_buf.wqe_cnt) { ret = alloc_rq_inline_buf(hr_qp, init_attr); if (ret) { - ibdev_err(ibdev, "Failed to alloc inline RQ buffer\n"); + ibdev_err(ibdev, + "failed to alloc inline buf, ret = %d.\n", + ret); return ret; } - } - - if (udata) { - hr_qp->umem = ib_umem_get(ibdev, addr, hr_qp->buff_size, 0); - if (IS_ERR(hr_qp->umem)) { - ret = PTR_ERR(hr_qp->umem); - goto err_inline; - } } else { - ret = hns_roce_buf_alloc(hr_dev, hr_qp->buff_size, - (1 << page_shift) * 2, - &hr_qp->hr_buf, page_shift); - if (ret) - goto err_inline; + hr_qp->rq_inl_buf.wqe_list = NULL; } - ret = map_wqe_buf(hr_dev, hr_qp, page_shift, udata); - if (ret) - goto err_alloc; + ret = set_wqe_buf_attr(hr_dev, hr_qp, &buf_attr); + if (ret) { + ibdev_err(ibdev, "failed to split WQE buf, ret = %d.\n", ret); + goto err_inline; + } + ret = hns_roce_mtr_create(hr_dev, &hr_qp->mtr, &buf_attr, + HNS_HW_PAGE_SHIFT + hr_dev->caps.mtt_ba_pg_sz, + udata, addr); + if (ret) { + ibdev_err(ibdev, "failed to create WQE mtr, ret = %d.\n", ret); + goto err_inline; + } return 0; - err_inline: - if (is_rq_buf_inline) - free_rq_inline_buf(hr_qp); - -err_alloc: - if (udata) { - ib_umem_release(hr_qp->umem); - hr_qp->umem = NULL; - } else { - hns_roce_buf_free(hr_dev, hr_qp->buff_size, &hr_qp->hr_buf); - } - - ibdev_err(ibdev, "Failed to alloc WQE buffer, ret %d.\n", ret); + free_rq_inline_buf(hr_qp); return ret; } static void free_qp_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp) { - hns_roce_mtr_cleanup(hr_dev, &hr_qp->mtr); - if (hr_qp->umem) { - ib_umem_release(hr_qp->umem); - hr_qp->umem = NULL; - } - - if (hr_qp->hr_buf.nbufs > 0) - hns_roce_buf_free(hr_dev, hr_qp->buff_size, &hr_qp->hr_buf); - - if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) && - hr_qp->rq.wqe_cnt) - free_rq_inline_buf(hr_qp); + hns_roce_mtr_destroy(hr_dev, &hr_qp->mtr); + free_rq_inline_buf(hr_qp); } static inline bool user_qp_has_sdb(struct hns_roce_dev *hr_dev, @@ -912,8 +740,8 @@ static int alloc_qp_db(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, "Failed to map user SQ doorbell\n"); goto err_out; } - hr_qp->sdb_en = 1; - resp->cap_flags |= HNS_ROCE_SUPPORT_SQ_RECORD_DB; + hr_qp->en_flags |= HNS_ROCE_QP_CAP_SQ_RECORD_DB; + resp->cap_flags |= HNS_ROCE_QP_CAP_SQ_RECORD_DB; } if (user_qp_has_rdb(hr_dev, init_attr, udata, resp)) { @@ -924,8 +752,8 @@ static int alloc_qp_db(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, "Failed to map user RQ doorbell\n"); goto err_sdb; } - hr_qp->rdb_en = 1; - resp->cap_flags |= HNS_ROCE_SUPPORT_RQ_RECORD_DB; + hr_qp->en_flags |= HNS_ROCE_QP_CAP_RQ_RECORD_DB; + resp->cap_flags |= HNS_ROCE_QP_CAP_RQ_RECORD_DB; } } else { /* QP doorbell register address */ @@ -942,13 +770,13 @@ static int alloc_qp_db(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, goto err_out; } *hr_qp->rdb.db_record = 0; - hr_qp->rdb_en = 1; + hr_qp->en_flags |= HNS_ROCE_QP_CAP_RQ_RECORD_DB; } } return 0; err_sdb: - if (udata && hr_qp->sdb_en) + if (udata && hr_qp->en_flags & HNS_ROCE_QP_CAP_SQ_RECORD_DB) hns_roce_db_unmap_user(uctx, &hr_qp->sdb); err_out: return ret; @@ -961,12 +789,12 @@ static void free_qp_db(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, udata, struct hns_roce_ucontext, ibucontext); if (udata) { - if (hr_qp->rdb_en) + if (hr_qp->en_flags & HNS_ROCE_QP_CAP_RQ_RECORD_DB) hns_roce_db_unmap_user(uctx, &hr_qp->rdb); - if (hr_qp->sdb_en) + if (hr_qp->en_flags & HNS_ROCE_QP_CAP_SQ_RECORD_DB) hns_roce_db_unmap_user(uctx, &hr_qp->sdb); } else { - if (hr_qp->rdb_en) + if (hr_qp->en_flags & HNS_ROCE_QP_CAP_RQ_RECORD_DB) hns_roce_free_db(hr_dev, &hr_qp->rdb); } } @@ -1003,8 +831,7 @@ err_sq: return ret; } -static void free_kernel_wrid(struct hns_roce_dev *hr_dev, - struct hns_roce_qp *hr_qp) +static void free_kernel_wrid(struct hns_roce_qp *hr_qp) { kfree(hr_qp->rq.wrid); kfree(hr_qp->sq.wrid); @@ -1025,10 +852,11 @@ static int set_qp_param(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, else hr_qp->sq_signal_bits = IB_SIGNAL_REQ_WR; - ret = set_rq_size(hr_dev, &init_attr->cap, udata, - hns_roce_qp_has_rq(init_attr), hr_qp); + ret = set_rq_size(hr_dev, &init_attr->cap, hr_qp, + hns_roce_qp_has_rq(init_attr)); if (ret) { - ibdev_err(ibdev, "Failed to set user RQ size\n"); + ibdev_err(ibdev, "failed to set user RQ size, ret = %d.\n", + ret); return ret; } @@ -1156,7 +984,7 @@ err_buf: err_db: free_qp_db(hr_dev, hr_qp, udata); err_wrid: - free_kernel_wrid(hr_dev, hr_qp); + free_kernel_wrid(hr_qp); return ret; } @@ -1170,7 +998,7 @@ void hns_roce_qp_destroy(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, free_qpc(hr_dev, hr_qp); free_qpn(hr_dev, hr_qp); free_qp_buf(hr_dev, hr_qp); - free_kernel_wrid(hr_dev, hr_qp); + free_kernel_wrid(hr_qp); free_qp_db(hr_dev, hr_qp, udata); kfree(hr_qp); @@ -1339,10 +1167,10 @@ int hns_roce_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, if (ibqp->uobject && (attr_mask & IB_QP_STATE) && new_state == IB_QPS_ERR) { - if (hr_qp->sdb_en == 1) { + if (hr_qp->en_flags & HNS_ROCE_QP_CAP_SQ_RECORD_DB) { hr_qp->sq.head = *(int *)(hr_qp->sdb.virt_addr); - if (hr_qp->rdb_en == 1) + if (hr_qp->en_flags & HNS_ROCE_QP_CAP_RQ_RECORD_DB) hr_qp->rq.head = *(int *)(hr_qp->rdb.virt_addr); } else { ibdev_warn(&hr_dev->ib_dev, @@ -1431,10 +1259,9 @@ void hns_roce_unlock_cqs(struct hns_roce_cq *send_cq, } } -static void *get_wqe(struct hns_roce_qp *hr_qp, int offset) +static inline void *get_wqe(struct hns_roce_qp *hr_qp, int offset) { - - return hns_roce_buf_offset(&hr_qp->hr_buf, offset); + return hns_roce_buf_offset(hr_qp->mtr.kmem, offset); } void *hns_roce_get_recv_wqe(struct hns_roce_qp *hr_qp, int n) @@ -1449,8 +1276,7 @@ void *hns_roce_get_send_wqe(struct hns_roce_qp *hr_qp, int n) void *hns_roce_get_extend_sge(struct hns_roce_qp *hr_qp, int n) { - return hns_roce_buf_offset(&hr_qp->hr_buf, hr_qp->sge.offset + - (n << hr_qp->sge.sge_shift)); + return get_wqe(hr_qp, hr_qp->sge.offset + (n << hr_qp->sge.sge_shift)); } bool hns_roce_wq_overflow(struct hns_roce_wq *hr_wq, int nreq, diff --git a/drivers/infiniband/hw/hns/hns_roce_restrack.c b/drivers/infiniband/hw/hns/hns_roce_restrack.c index 06871731ac43..259444c0a630 100644 --- a/drivers/infiniband/hw/hns/hns_roce_restrack.c +++ b/drivers/infiniband/hw/hns/hns_roce_restrack.c @@ -76,10 +76,9 @@ err: return -EMSGSIZE; } -static int hns_roce_fill_res_cq_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +int hns_roce_fill_res_cq_entry(struct sk_buff *msg, + struct ib_cq *ib_cq) { - struct ib_cq *ib_cq = container_of(res, struct ib_cq, res); struct hns_roce_dev *hr_dev = to_hr_dev(ib_cq->device); struct hns_roce_cq *hr_cq = to_hr_cq(ib_cq); struct hns_roce_v2_cq_context *context; @@ -119,12 +118,3 @@ err: kfree(context); return ret; } - -int hns_roce_fill_res_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) -{ - if (res->type == RDMA_RESTRACK_CQ) - return hns_roce_fill_res_cq_entry(msg, res); - - return 0; -} diff --git a/drivers/infiniband/hw/hns/hns_roce_srq.c b/drivers/infiniband/hw/hns/hns_roce_srq.c index 5b3dd1a337d4..b9e2dbd372b6 100644 --- a/drivers/infiniband/hw/hns/hns_roce_srq.c +++ b/drivers/infiniband/hw/hns/hns_roce_srq.c @@ -77,56 +77,56 @@ static int hns_roce_hw_destroy_srq(struct hns_roce_dev *dev, HNS_ROCE_CMD_TIMEOUT_MSECS); } -static int hns_roce_srq_alloc(struct hns_roce_dev *hr_dev, u32 pdn, u32 cqn, - u16 xrcd, struct hns_roce_mtt *hr_mtt, - u64 db_rec_addr, struct hns_roce_srq *srq) +static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, + u32 pdn, u32 cqn, u16 xrcd, u64 db_rec_addr) { struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; + struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_cmd_mailbox *mailbox; - dma_addr_t dma_handle_wqe; - dma_addr_t dma_handle_idx; - u64 *mtts_wqe; - u64 *mtts_idx; + u64 mtts_wqe[MTT_MIN_COUNT] = { 0 }; + u64 mtts_idx[MTT_MIN_COUNT] = { 0 }; + dma_addr_t dma_handle_wqe = 0; + dma_addr_t dma_handle_idx = 0; int ret; /* Get the physical address of srq buf */ - mtts_wqe = hns_roce_table_find(hr_dev, - &hr_dev->mr_table.mtt_srqwqe_table, - srq->mtt.first_seg, - &dma_handle_wqe); - if (!mtts_wqe) { - dev_err(hr_dev->dev, "Failed to find mtt for srq buf.\n"); - return -EINVAL; + ret = hns_roce_mtr_find(hr_dev, &srq->buf_mtr, 0, mtts_wqe, + ARRAY_SIZE(mtts_wqe), &dma_handle_wqe); + if (ret < 1) { + ibdev_err(ibdev, "Failed to find mtr for SRQ WQE\n"); + return -ENOBUFS; } /* Get physical address of idx que buf */ - mtts_idx = hns_roce_table_find(hr_dev, &hr_dev->mr_table.mtt_idx_table, - srq->idx_que.mtt.first_seg, - &dma_handle_idx); - if (!mtts_idx) { - dev_err(hr_dev->dev, - "Failed to find mtt for srq idx queue buf.\n"); - return -EINVAL; + ret = hns_roce_mtr_find(hr_dev, &srq->idx_que.mtr, 0, mtts_idx, + ARRAY_SIZE(mtts_idx), &dma_handle_idx); + if (ret < 1) { + ibdev_err(ibdev, "Failed to find mtr for SRQ idx\n"); + return -ENOBUFS; } ret = hns_roce_bitmap_alloc(&srq_table->bitmap, &srq->srqn); if (ret) { - dev_err(hr_dev->dev, - "Failed to alloc a bit from srq bitmap.\n"); + ibdev_err(ibdev, "Failed to alloc SRQ number, err %d\n", ret); return -ENOMEM; } ret = hns_roce_table_get(hr_dev, &srq_table->table, srq->srqn); - if (ret) + if (ret) { + ibdev_err(ibdev, "Failed to get SRQC table, err %d\n", ret); goto err_out; + } ret = xa_err(xa_store(&srq_table->xa, srq->srqn, srq, GFP_KERNEL)); - if (ret) + if (ret) { + ibdev_err(ibdev, "Failed to store SRQC, err %d\n", ret); goto err_put; + } mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); - if (IS_ERR(mailbox)) { - ret = PTR_ERR(mailbox); + if (IS_ERR_OR_NULL(mailbox)) { + ret = -ENOMEM; + ibdev_err(ibdev, "Failed to alloc mailbox for SRQC\n"); goto err_xa; } @@ -136,8 +136,10 @@ static int hns_roce_srq_alloc(struct hns_roce_dev *hr_dev, u32 pdn, u32 cqn, ret = hns_roce_hw_create_srq(hr_dev, mailbox, srq->srqn); hns_roce_free_cmd_mailbox(hr_dev, mailbox); - if (ret) + if (ret) { + ibdev_err(ibdev, "Failed to config SRQC, err %d\n", ret); goto err_xa; + } atomic_set(&srq->refcount, 1); init_completion(&srq->free); @@ -154,8 +156,7 @@ err_out: return ret; } -static void hns_roce_srq_free(struct hns_roce_dev *hr_dev, - struct hns_roce_srq *srq) +static void free_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) { struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; int ret; @@ -175,187 +176,104 @@ static void hns_roce_srq_free(struct hns_roce_dev *hr_dev, hns_roce_bitmap_free(&srq_table->bitmap, srq->srqn, BITMAP_NO_RR); } -static int create_user_srq(struct hns_roce_srq *srq, struct ib_udata *udata, - int srq_buf_size) +static int alloc_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, + struct ib_udata *udata, unsigned long addr) { - struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device); - struct hns_roce_ib_create_srq ucmd; - struct hns_roce_buf *buf; - int ret; + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_buf_attr buf_attr = {}; + int err; + + srq->wqe_shift = ilog2(roundup_pow_of_two(max(HNS_ROCE_SGE_SIZE, + HNS_ROCE_SGE_SIZE * + srq->max_gs))); + + buf_attr.page_shift = hr_dev->caps.srqwqe_buf_pg_sz + HNS_HW_PAGE_SHIFT; + buf_attr.region[0].size = to_hr_hem_entries_size(srq->wqe_cnt, + srq->wqe_shift); + buf_attr.region[0].hopnum = hr_dev->caps.srqwqe_hop_num; + buf_attr.region_count = 1; + buf_attr.fixed_page = true; + + err = hns_roce_mtr_create(hr_dev, &srq->buf_mtr, &buf_attr, + hr_dev->caps.srqwqe_ba_pg_sz + + HNS_HW_PAGE_SHIFT, udata, addr); + if (err) + ibdev_err(ibdev, "Failed to alloc SRQ buf mtr, err %d\n", err); + + return err; +} - if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) - return -EFAULT; - - srq->umem = - ib_umem_get(srq->ibsrq.device, ucmd.buf_addr, srq_buf_size, 0); - if (IS_ERR(srq->umem)) - return PTR_ERR(srq->umem); - - buf = &srq->buf; - buf->npages = (ib_umem_page_count(srq->umem) + - (1 << hr_dev->caps.srqwqe_buf_pg_sz) - 1) / - (1 << hr_dev->caps.srqwqe_buf_pg_sz); - buf->page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; - ret = hns_roce_mtt_init(hr_dev, buf->npages, buf->page_shift, - &srq->mtt); - if (ret) - goto err_user_buf; +static void free_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) +{ + hns_roce_mtr_destroy(hr_dev, &srq->buf_mtr); +} - ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->mtt, srq->umem); - if (ret) - goto err_user_srq_mtt; - - /* config index queue BA */ - srq->idx_que.umem = ib_umem_get(srq->ibsrq.device, ucmd.que_addr, - srq->idx_que.buf_size, 0); - if (IS_ERR(srq->idx_que.umem)) { - dev_err(hr_dev->dev, "ib_umem_get error for index queue\n"); - ret = PTR_ERR(srq->idx_que.umem); - goto err_user_srq_mtt; +static int alloc_srq_idx(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, + struct ib_udata *udata, unsigned long addr) +{ + struct hns_roce_idx_que *idx_que = &srq->idx_que; + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_buf_attr buf_attr = {}; + int err; + + srq->idx_que.entry_shift = ilog2(HNS_ROCE_IDX_QUE_ENTRY_SZ); + + buf_attr.page_shift = hr_dev->caps.idx_buf_pg_sz + HNS_HW_PAGE_SHIFT; + buf_attr.region[0].size = to_hr_hem_entries_size(srq->wqe_cnt, + srq->idx_que.entry_shift); + buf_attr.region[0].hopnum = hr_dev->caps.idx_hop_num; + buf_attr.region_count = 1; + buf_attr.fixed_page = true; + + err = hns_roce_mtr_create(hr_dev, &idx_que->mtr, &buf_attr, + hr_dev->caps.idx_ba_pg_sz + HNS_HW_PAGE_SHIFT, + udata, addr); + if (err) { + ibdev_err(ibdev, "Failed to alloc SRQ idx mtr, err %d\n", err); + return err; } - buf = &srq->idx_que.idx_buf; - buf->npages = DIV_ROUND_UP(ib_umem_page_count(srq->idx_que.umem), - 1 << hr_dev->caps.idx_buf_pg_sz); - buf->page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; - ret = hns_roce_mtt_init(hr_dev, buf->npages, buf->page_shift, - &srq->idx_que.mtt); - if (ret) { - dev_err(hr_dev->dev, "hns_roce_mtt_init error for idx que\n"); - goto err_user_idx_mtt; - } + if (!udata) { + idx_que->bitmap = bitmap_zalloc(srq->wqe_cnt, GFP_KERNEL); + if (!idx_que->bitmap) { + ibdev_err(ibdev, "Failed to alloc SRQ idx bitmap\n"); + err = -ENOMEM; + goto err_idx_mtr; + } - ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->idx_que.mtt, - srq->idx_que.umem); - if (ret) { - dev_err(hr_dev->dev, - "hns_roce_ib_umem_write_mtt error for idx que\n"); - goto err_user_idx_buf; } return 0; +err_idx_mtr: + hns_roce_mtr_destroy(hr_dev, &idx_que->mtr); -err_user_idx_buf: - hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); - -err_user_idx_mtt: - ib_umem_release(srq->idx_que.umem); - -err_user_srq_mtt: - hns_roce_mtt_cleanup(hr_dev, &srq->mtt); - -err_user_buf: - ib_umem_release(srq->umem); - - return ret; + return err; } -static int hns_roce_create_idx_que(struct ib_pd *pd, struct hns_roce_srq *srq, - u32 page_shift) +static void free_srq_idx(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) { - struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); struct hns_roce_idx_que *idx_que = &srq->idx_que; - idx_que->bitmap = bitmap_zalloc(srq->wqe_cnt, GFP_KERNEL); - if (!idx_que->bitmap) - return -ENOMEM; - - idx_que->buf_size = srq->idx_que.buf_size; - - if (hns_roce_buf_alloc(hr_dev, idx_que->buf_size, (1 << page_shift) * 2, - &idx_que->idx_buf, page_shift)) { - bitmap_free(idx_que->bitmap); - return -ENOMEM; - } - - return 0; + bitmap_free(idx_que->bitmap); + idx_que->bitmap = NULL; + hns_roce_mtr_destroy(hr_dev, &idx_que->mtr); } -static int create_kernel_srq(struct hns_roce_srq *srq, int srq_buf_size) +static int alloc_srq_wrid(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) { - struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device); - u32 page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; - int ret; - - if (hns_roce_buf_alloc(hr_dev, srq_buf_size, (1 << page_shift) * 2, - &srq->buf, page_shift)) - return -ENOMEM; - srq->head = 0; srq->tail = srq->wqe_cnt - 1; - - ret = hns_roce_mtt_init(hr_dev, srq->buf.npages, srq->buf.page_shift, - &srq->mtt); - if (ret) - goto err_kernel_buf; - - ret = hns_roce_buf_write_mtt(hr_dev, &srq->mtt, &srq->buf); - if (ret) - goto err_kernel_srq_mtt; - - page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; - ret = hns_roce_create_idx_que(srq->ibsrq.pd, srq, page_shift); - if (ret) { - dev_err(hr_dev->dev, "Create idx queue fail(%d)!\n", ret); - goto err_kernel_srq_mtt; - } - - /* Init mtt table for idx_que */ - ret = hns_roce_mtt_init(hr_dev, srq->idx_que.idx_buf.npages, - srq->idx_que.idx_buf.page_shift, - &srq->idx_que.mtt); - if (ret) - goto err_kernel_create_idx; - - /* Write buffer address into the mtt table */ - ret = hns_roce_buf_write_mtt(hr_dev, &srq->idx_que.mtt, - &srq->idx_que.idx_buf); - if (ret) - goto err_kernel_idx_buf; - srq->wrid = kvmalloc_array(srq->wqe_cnt, sizeof(u64), GFP_KERNEL); - if (!srq->wrid) { - ret = -ENOMEM; - goto err_kernel_idx_buf; - } + if (!srq->wrid) + return -ENOMEM; return 0; - -err_kernel_idx_buf: - hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); - -err_kernel_create_idx: - hns_roce_buf_free(hr_dev, srq->idx_que.buf_size, - &srq->idx_que.idx_buf); - kfree(srq->idx_que.bitmap); - -err_kernel_srq_mtt: - hns_roce_mtt_cleanup(hr_dev, &srq->mtt); - -err_kernel_buf: - hns_roce_buf_free(hr_dev, srq_buf_size, &srq->buf); - - return ret; -} - -static void destroy_user_srq(struct hns_roce_dev *hr_dev, - struct hns_roce_srq *srq) -{ - hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); - ib_umem_release(srq->idx_que.umem); - hns_roce_mtt_cleanup(hr_dev, &srq->mtt); - ib_umem_release(srq->umem); } -static void destroy_kernel_srq(struct hns_roce_dev *hr_dev, - struct hns_roce_srq *srq, int srq_buf_size) +static void free_srq_wrid(struct hns_roce_srq *srq) { - kvfree(srq->wrid); - hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); - hns_roce_buf_free(hr_dev, srq->idx_que.buf_size, &srq->idx_que.idx_buf); - kfree(srq->idx_que.bitmap); - hns_roce_mtt_cleanup(hr_dev, &srq->mtt); - hns_roce_buf_free(hr_dev, srq_buf_size, &srq->buf); + kfree(srq->wrid); + srq->wrid = NULL; } int hns_roce_create_srq(struct ib_srq *ib_srq, @@ -365,8 +283,8 @@ int hns_roce_create_srq(struct ib_srq *ib_srq, struct hns_roce_dev *hr_dev = to_hr_dev(ib_srq->device); struct hns_roce_ib_create_srq_resp resp = {}; struct hns_roce_srq *srq = to_hr_srq(ib_srq); - int srq_desc_size; - int srq_buf_size; + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_ib_create_srq ucmd = {}; int ret = 0; u32 cqn; @@ -381,41 +299,45 @@ int hns_roce_create_srq(struct ib_srq *ib_srq, srq->wqe_cnt = roundup_pow_of_two(init_attr->attr.max_wr + 1); srq->max_gs = init_attr->attr.max_sge; - srq_desc_size = roundup_pow_of_two(max(HNS_ROCE_SGE_SIZE, - HNS_ROCE_SGE_SIZE * srq->max_gs)); - - srq->wqe_shift = ilog2(srq_desc_size); - - srq_buf_size = srq->wqe_cnt * srq_desc_size; - - srq->idx_que.entry_sz = HNS_ROCE_IDX_QUE_ENTRY_SZ; - srq->idx_que.buf_size = srq->wqe_cnt * srq->idx_que.entry_sz; - srq->mtt.mtt_type = MTT_TYPE_SRQWQE; - srq->idx_que.mtt.mtt_type = MTT_TYPE_IDX; - if (udata) { - ret = create_user_srq(srq, udata, srq_buf_size); + ret = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd)); if (ret) { - dev_err(hr_dev->dev, "Create user srq failed\n"); - goto err_srq; + ibdev_err(ibdev, "Failed to copy SRQ udata, err %d\n", + ret); + return ret; } - } else { - ret = create_kernel_srq(srq, srq_buf_size); + } + + ret = alloc_srq_buf(hr_dev, srq, udata, ucmd.buf_addr); + if (ret) { + ibdev_err(ibdev, "Failed to alloc SRQ buffer, err %d\n", ret); + return ret; + } + + ret = alloc_srq_idx(hr_dev, srq, udata, ucmd.que_addr); + if (ret) { + ibdev_err(ibdev, "Failed to alloc SRQ idx, err %d\n", ret); + goto err_buf_alloc; + } + + if (!udata) { + ret = alloc_srq_wrid(hr_dev, srq); if (ret) { - dev_err(hr_dev->dev, "Create kernel srq failed\n"); - goto err_srq; + ibdev_err(ibdev, "Failed to alloc SRQ wrid, err %d\n", + ret); + goto err_idx_alloc; } } cqn = ib_srq_has_cq(init_attr->srq_type) ? to_hr_cq(init_attr->ext.cq)->cqn : 0; - srq->db_reg_l = hr_dev->reg_base + SRQ_DB_REG; - ret = hns_roce_srq_alloc(hr_dev, to_hr_pd(ib_srq->pd)->pdn, cqn, 0, - &srq->mtt, 0, srq); - if (ret) - goto err_wrid; + ret = alloc_srqc(hr_dev, srq, to_hr_pd(ib_srq->pd)->pdn, cqn, 0, 0); + if (ret) { + ibdev_err(ibdev, "Failed to alloc SRQ context, err %d\n", ret); + goto err_wrid_alloc; + } srq->event = hns_roce_ib_srq_event; resp.srqn = srq->srqn; @@ -431,15 +353,13 @@ int hns_roce_create_srq(struct ib_srq *ib_srq, return 0; err_srqc_alloc: - hns_roce_srq_free(hr_dev, srq); - -err_wrid: - if (udata) - destroy_user_srq(hr_dev, srq); - else - destroy_kernel_srq(hr_dev, srq, srq_buf_size); - -err_srq: + free_srqc(hr_dev, srq); +err_wrid_alloc: + free_srq_wrid(srq); +err_idx_alloc: + free_srq_idx(hr_dev, srq); +err_buf_alloc: + free_srq_buf(hr_dev, srq); return ret; } @@ -448,18 +368,10 @@ void hns_roce_destroy_srq(struct ib_srq *ibsrq, struct ib_udata *udata) struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); struct hns_roce_srq *srq = to_hr_srq(ibsrq); - hns_roce_srq_free(hr_dev, srq); - hns_roce_mtt_cleanup(hr_dev, &srq->mtt); - - if (udata) { - hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); - } else { - kvfree(srq->wrid); - hns_roce_buf_free(hr_dev, srq->wqe_cnt << srq->wqe_shift, - &srq->buf); - } - ib_umem_release(srq->idx_que.umem); - ib_umem_release(srq->umem); + free_srqc(hr_dev, srq); + free_srq_idx(hr_dev, srq); + free_srq_wrid(srq); + free_srq_buf(hr_dev, srq); } int hns_roce_init_srq_table(struct hns_roce_dev *hr_dev) diff --git a/drivers/infiniband/hw/i40iw/Kconfig b/drivers/infiniband/hw/i40iw/Kconfig index e4b45f4cd8f8..7476f3b14c39 100644 --- a/drivers/infiniband/hw/i40iw/Kconfig +++ b/drivers/infiniband/hw/i40iw/Kconfig @@ -5,5 +5,5 @@ config INFINIBAND_I40IW depends on IPV6 || !IPV6 depends on PCI select GENERIC_ALLOCATOR - ---help--- + help Intel(R) Ethernet X722 iWARP Driver diff --git a/drivers/infiniband/hw/i40iw/Makefile b/drivers/infiniband/hw/i40iw/Makefile index 8942f8229945..34da9eba8a7c 100644 --- a/drivers/infiniband/hw/i40iw/Makefile +++ b/drivers/infiniband/hw/i40iw/Makefile @@ -1,5 +1,4 @@ # SPDX-License-Identifier: GPL-2.0 -ccflags-y := -I $(srctree)/drivers/net/ethernet/intel/i40e obj-$(CONFIG_INFINIBAND_I40IW) += i40iw.o diff --git a/drivers/infiniband/hw/i40iw/i40iw.h b/drivers/infiniband/hw/i40iw/i40iw.h index 3c62c9327a9c..25747b85a79c 100644 --- a/drivers/infiniband/hw/i40iw/i40iw.h +++ b/drivers/infiniband/hw/i40iw/i40iw.h @@ -45,6 +45,7 @@ #include <linux/slab.h> #include <linux/io.h> #include <linux/crc32c.h> +#include <linux/net/intel/i40e_client.h> #include <rdma/ib_smi.h> #include <rdma/ib_verbs.h> #include <rdma/ib_pack.h> @@ -57,7 +58,6 @@ #include "i40iw_d.h" #include "i40iw_hmc.h" -#include <i40e_client.h> #include "i40iw_type.h" #include "i40iw_p.h" #include <rdma/i40iw-abi.h> @@ -382,15 +382,6 @@ static inline struct i40iw_mr *to_iwmr(struct ib_mr *ibmr) } /** - * to_iwmr_from_ibfmr - get device memory region - * @ibfmr: ib fmr - **/ -static inline struct i40iw_mr *to_iwmr_from_ibfmr(struct ib_fmr *ibfmr) -{ - return container_of(ibfmr, struct i40iw_mr, ibfmr); -} - -/** * to_iwmw - get device memory window * @ibmw: ib memory window **/ diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c index fa7a5ff498c7..a3b95805c154 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_cm.c +++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c @@ -2443,7 +2443,7 @@ static void i40iw_handle_rst_pkt(struct i40iw_cm_node *cm_node, case I40IW_CM_STATE_FIN_WAIT1: case I40IW_CM_STATE_LAST_ACK: cm_node->cm_id->rem_ref(cm_node->cm_id); - /* fall through */ + fallthrough; case I40IW_CM_STATE_TIME_WAIT: cm_node->state = I40IW_CM_STATE_CLOSED; i40iw_rem_ref_cm_node(cm_node); diff --git a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c index 688f19667221..86d3f8aff329 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c +++ b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c @@ -1964,7 +1964,6 @@ static enum i40iw_status_code i40iw_sc_get_next_aeqe(struct i40iw_sc_aeq *aeq, info->out_rdrsp = true; break; case I40IW_AE_SOURCE_RSVD: - /* fallthrough */ default: break; } @@ -3762,14 +3761,14 @@ static enum i40iw_status_code cqp_sds_wqe_fill(struct i40iw_sc_cqp *cqp, LS_64(1, I40IW_CQPSQ_UPESD_ENTRY_VALID))); set_64bit_val(wqe, 56, info->entry[2].data); - /* fallthrough */ + fallthrough; case 2: set_64bit_val(wqe, 32, (LS_64(info->entry[1].cmd, I40IW_CQPSQ_UPESD_SDCMD) | LS_64(1, I40IW_CQPSQ_UPESD_ENTRY_VALID))); set_64bit_val(wqe, 40, info->entry[1].data); - /* fallthrough */ + fallthrough; case 1: set_64bit_val(wqe, 0, LS_64(info->entry[0].cmd, I40IW_CQPSQ_UPESD_SDCMD)); diff --git a/drivers/infiniband/hw/i40iw/i40iw_hw.c b/drivers/infiniband/hw/i40iw/i40iw_hw.c index ae8b97c30665..e1085634b8d9 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_hw.c +++ b/drivers/infiniband/hw/i40iw/i40iw_hw.c @@ -353,7 +353,6 @@ void i40iw_process_aeq(struct i40iw_device *iwdev) i40iw_cm_disconn(iwqp); break; case I40IW_AE_BAD_CLOSE: - /* fall through */ case I40IW_AE_RESET_SENT: i40iw_next_iw_state(iwqp, I40IW_QP_STATE_ERROR, 1, 0, 0); i40iw_cm_disconn(iwqp); @@ -413,7 +412,7 @@ void i40iw_process_aeq(struct i40iw_device *iwdev) case I40IW_AE_UDA_XMIT_DGRAM_TOO_LONG: case I40IW_AE_UDA_XMIT_DGRAM_TOO_SHORT: ctx_info->err_rq_idx_valid = false; - /* fall through */ + fallthrough; default: if (!info->sq && ctx_info->err_rq_idx_valid) { ctx_info->err_rq_idx = info->wqe_idx; diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c index 9c96ece5e7f3..58a433135a03 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_main.c +++ b/drivers/infiniband/hw/i40iw/i40iw_main.c @@ -1489,36 +1489,35 @@ static void i40iw_deinit_device(struct i40iw_device *iwdev) iwdev->iw_status = 0; i40iw_port_ibevent(iwdev); i40iw_destroy_rdma_device(iwdev->iwibdev); - /* fallthrough */ + fallthrough; case IP_ADDR_REGISTERED: if (!iwdev->reset) i40iw_del_macip_entry(iwdev, (u8)iwdev->mac_ip_table_idx); - /* fallthrough */ - /* fallthrough */ + fallthrough; case PBLE_CHUNK_MEM: i40iw_destroy_pble_pool(dev, iwdev->pble_rsrc); - /* fallthrough */ + fallthrough; case CEQ_CREATED: i40iw_dele_ceqs(iwdev); - /* fallthrough */ + fallthrough; case AEQ_CREATED: i40iw_destroy_aeq(iwdev); - /* fallthrough */ + fallthrough; case IEQ_CREATED: i40iw_puda_dele_resources(&iwdev->vsi, I40IW_PUDA_RSRC_TYPE_IEQ, iwdev->reset); - /* fallthrough */ + fallthrough; case ILQ_CREATED: i40iw_puda_dele_resources(&iwdev->vsi, I40IW_PUDA_RSRC_TYPE_ILQ, iwdev->reset); - /* fallthrough */ + fallthrough; case CCQ_CREATED: i40iw_destroy_ccq(iwdev); - /* fallthrough */ + fallthrough; case HMC_OBJS_CREATED: i40iw_del_hmc_objects(dev, dev->hmc_info, true, iwdev->reset); - /* fallthrough */ + fallthrough; case CQP_CREATED: i40iw_destroy_cqp(iwdev, true); - /* fallthrough */ + fallthrough; case INITIAL_STATE: i40iw_cleanup_cm_core(&iwdev->cm_core); if (iwdev->vsi.pestat) { @@ -1528,7 +1527,6 @@ static void i40iw_deinit_device(struct i40iw_device *iwdev) i40iw_del_init_mem(iwdev); break; case INVALID_STATE: - /* fallthrough */ default: i40iw_pr_err("bad init_state = %d\n", iwdev->init_state); break; diff --git a/drivers/infiniband/hw/i40iw/i40iw_puda.c b/drivers/infiniband/hw/i40iw/i40iw_puda.c index d9c7ae6a7030..924be4b03c9a 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_puda.c +++ b/drivers/infiniband/hw/i40iw/i40iw_puda.c @@ -814,13 +814,13 @@ void i40iw_puda_dele_resources(struct i40iw_sc_vsi *vsi, switch (rsrc->completion) { case PUDA_HASH_CRC_COMPLETE: i40iw_free_hash_desc(rsrc->hash_desc); - /* fall through */ + fallthrough; case PUDA_QP_CREATED: if (!reset) i40iw_puda_free_qp(rsrc); i40iw_free_dma_mem(dev->hw, &rsrc->qpmem); - /* fallthrough */ + fallthrough; case PUDA_CQ_CREATED: if (!reset) i40iw_puda_free_cq(rsrc); diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c index 016524683e17..e07fb37af086 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_utils.c +++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c @@ -190,9 +190,8 @@ int i40iw_inetaddr_event(struct notifier_block *notifier, switch (event) { case NETDEV_DOWN: action = I40IW_ARP_DELETE; - /* Fall through */ + fallthrough; case NETDEV_UP: - /* Fall through */ case NETDEV_CHANGEADDR: /* Just skip if no need to handle ARP cache */ @@ -247,9 +246,8 @@ int i40iw_inet6addr_event(struct notifier_block *notifier, switch (event) { case NETDEV_DOWN: action = I40IW_ARP_DELETE; - /* Fall through */ + fallthrough; case NETDEV_UP: - /* Fall through */ case NETDEV_CHANGEADDR: i40iw_manage_arp_cache(iwdev, netdev->dev_addr, @@ -344,7 +342,7 @@ int i40iw_netdevice_event(struct notifier_block *notifier, switch (event) { case NETDEV_DOWN: iwdev->iw_status = 0; - /* Fall through */ + fallthrough; case NETDEV_UP: i40iw_port_ibevent(iwdev); break; diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c index 1b6fb1380961..b51339328a51 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c @@ -83,7 +83,6 @@ static int i40iw_query_device(struct ib_device *ibdev, props->max_qp_rd_atom = I40IW_MAX_IRD_SIZE; props->max_qp_init_rd_atom = props->max_qp_rd_atom; props->atomic_cap = IB_ATOMIC_NONE; - props->max_map_per_fmr = 1; props->max_fast_reg_page_list_len = I40IW_MAX_PAGES_PER_FMR; return 0; } @@ -102,7 +101,6 @@ static int i40iw_query_port(struct ib_device *ibdev, props->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_REINIT_SUP | IB_PORT_VENDOR_CLASS_SUP | IB_PORT_BOOT_MGMT_SUP; props->gid_tbl_len = 1; - props->pkey_tbl_len = 1; props->active_width = IB_WIDTH_4X; props->active_speed = 1; props->max_msg_sz = I40IW_MAX_OUTBOUND_MESSAGE_SIZE; @@ -812,7 +810,7 @@ void i40iw_hw_modify_qp(struct i40iw_device *iwdev, struct i40iw_qp *iwqp, case I40IW_QP_STATE_RTS: if (iwqp->iwarp_state == I40IW_QP_STATE_IDLE) i40iw_send_reset(iwqp->cm_node); - /* fall through */ + fallthrough; case I40IW_QP_STATE_IDLE: case I40IW_QP_STATE_TERMINATE: case I40IW_QP_STATE_CLOSING: @@ -1544,10 +1542,9 @@ static int i40iw_hw_alloc_stag(struct i40iw_device *iwdev, struct i40iw_mr *iwmr * @pd: ibpd pointer * @mr_type: memory for stag registrion * @max_num_sg: man number of pages - * @udata: user data or NULL for kernel objects */ static struct ib_mr *i40iw_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct i40iw_pd *iwpd = to_iwpd(pd); struct i40iw_device *iwdev = to_iwdev(pd->device); @@ -2147,7 +2144,6 @@ static int i40iw_post_send(struct ib_qp *ibqp, switch (ib_wr->opcode) { case IB_WR_SEND: - /* fall-through */ case IB_WR_SEND_WITH_INV: if (ib_wr->opcode == IB_WR_SEND) { if (ib_wr->send_flags & IB_SEND_SOLICITED) @@ -2204,7 +2200,7 @@ static int i40iw_post_send(struct ib_qp *ibqp, break; case IB_WR_RDMA_READ_WITH_INV: inv_stag = true; - /* fall-through*/ + fallthrough; case IB_WR_RDMA_READ: if (ib_wr->num_sge > I40IW_MAX_SGE_RD) { err = -EINVAL; @@ -2461,7 +2457,6 @@ static int i40iw_port_immutable(struct ib_device *ibdev, u8 port_num, if (err) return err; - immutable->pkey_tbl_len = attr.pkey_tbl_len; immutable->gid_tbl_len = attr.gid_tbl_len; return 0; @@ -2617,22 +2612,6 @@ static int i40iw_query_gid(struct ib_device *ibdev, return 0; } -/** - * i40iw_query_pkey - Query partition key - * @ibdev: device pointer from stack - * @port: port number - * @index: index of pkey - * @pkey: pointer to store the pkey - */ -static int i40iw_query_pkey(struct ib_device *ibdev, - u8 port, - u16 index, - u16 *pkey) -{ - *pkey = 0; - return 0; -} - static const struct ib_device_ops i40iw_dev_ops = { .owner = THIS_MODULE, .driver_id = RDMA_DRIVER_I40IW, @@ -2672,7 +2651,6 @@ static const struct ib_device_ops i40iw_dev_ops = { .post_send = i40iw_post_send, .query_device = i40iw_query_device, .query_gid = i40iw_query_gid, - .query_pkey = i40iw_query_pkey, .query_port = i40iw_query_port, .query_qp = i40iw_query_qp, .reg_user_mr = i40iw_reg_user_mr, diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.h b/drivers/infiniband/hw/i40iw/i40iw_verbs.h index 3a413752ccc3..331bc21cbcc7 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.h +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.h @@ -89,7 +89,6 @@ struct i40iw_mr { union { struct ib_mr ibmr; struct ib_mw ibmw; - struct ib_fmr ibfmr; }; struct ib_umem *region; u16 type; diff --git a/drivers/infiniband/hw/mlx4/Kconfig b/drivers/infiniband/hw/mlx4/Kconfig index cc7c42fe6499..f30ce9dd080a 100644 --- a/drivers/infiniband/hw/mlx4/Kconfig +++ b/drivers/infiniband/hw/mlx4/Kconfig @@ -4,7 +4,7 @@ config MLX4_INFINIBAND depends on NETDEVICES && ETHERNET && PCI && INET select NET_VENDOR_MELLANOX select MLX4_CORE - ---help--- + help This driver provides low-level InfiniBand support for Mellanox ConnectX PCI Express host channel adapters (HCAs). This is required to use InfiniBand protocols such as diff --git a/drivers/infiniband/hw/mlx4/ah.c b/drivers/infiniband/hw/mlx4/ah.c index 02a169f8027b..5f8f8d5c0ce0 100644 --- a/drivers/infiniband/hw/mlx4/ah.c +++ b/drivers/infiniband/hw/mlx4/ah.c @@ -141,10 +141,11 @@ static int create_iboe_ah(struct ib_ah *ib_ah, struct rdma_ah_attr *ah_attr) return 0; } -int mlx4_ib_create_ah(struct ib_ah *ib_ah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata) - +int mlx4_ib_create_ah(struct ib_ah *ib_ah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; + if (ah_attr->type == RDMA_AH_ATTR_TYPE_ROCE) { if (!(rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH)) return -EINVAL; @@ -167,12 +168,14 @@ int mlx4_ib_create_ah_slave(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, int slave_sgid_index, u8 *s_mac, u16 vlan_tag) { struct rdma_ah_attr slave_attr = *ah_attr; + struct rdma_ah_init_attr init_attr = {}; struct mlx4_ib_ah *mah = to_mah(ah); int ret; slave_attr.grh.sgid_attr = NULL; slave_attr.grh.sgid_index = slave_sgid_index; - ret = mlx4_ib_create_ah(ah, &slave_attr, 0, NULL); + init_attr.ah_attr = &slave_attr; + ret = mlx4_ib_create_ah(ah, &init_attr, NULL); if (ret) return ret; diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c index f8b936b76dcd..8a3436994f80 100644 --- a/drivers/infiniband/hw/mlx4/cq.c +++ b/drivers/infiniband/hw/mlx4/cq.c @@ -765,13 +765,13 @@ repoll: switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) { case MLX4_OPCODE_RDMA_WRITE_IMM: wc->wc_flags |= IB_WC_WITH_IMM; - /* fall through */ + fallthrough; case MLX4_OPCODE_RDMA_WRITE: wc->opcode = IB_WC_RDMA_WRITE; break; case MLX4_OPCODE_SEND_IMM: wc->wc_flags |= IB_WC_WITH_IMM; - /* fall through */ + fallthrough; case MLX4_OPCODE_SEND: case MLX4_OPCODE_SEND_INVAL: wc->opcode = IB_WC_SEND; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 275722cec8c6..bd4f975e7f9a 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -558,7 +558,6 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, props->max_mcast_qp_attach = dev->dev->caps.num_qp_per_mgm; props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * props->max_mcast_grp; - props->max_map_per_fmr = dev->dev->caps.max_fmr_maps; props->hca_core_clock = dev->dev->caps.hca_core_clock * 1000UL; props->timestamp_mask = 0xFFFFFFFFFFFFULL; props->max_ah = INT_MAX; @@ -785,7 +784,8 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port, props->ip_gids = true; props->gid_tbl_len = mdev->dev->caps.gid_table_len[port]; props->max_msg_sz = mdev->dev->caps.max_msg_sz; - props->pkey_tbl_len = 1; + if (mdev->dev->caps.pkey_table_len[port]) + props->pkey_tbl_len = 1; props->max_mtu = IB_MTU_4096; props->max_vl_num = 2; props->state = IB_PORT_DOWN; @@ -1220,56 +1220,47 @@ static void mlx4_ib_dealloc_pd(struct ib_pd *pd, struct ib_udata *udata) mlx4_pd_free(to_mdev(pd->device)->dev, to_mpd(pd)->pdn); } -static struct ib_xrcd *mlx4_ib_alloc_xrcd(struct ib_device *ibdev, - struct ib_udata *udata) +static int mlx4_ib_alloc_xrcd(struct ib_xrcd *ibxrcd, struct ib_udata *udata) { - struct mlx4_ib_xrcd *xrcd; + struct mlx4_ib_dev *dev = to_mdev(ibxrcd->device); + struct mlx4_ib_xrcd *xrcd = to_mxrcd(ibxrcd); struct ib_cq_init_attr cq_attr = {}; int err; - if (!(to_mdev(ibdev)->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC)) - return ERR_PTR(-ENOSYS); - - xrcd = kmalloc(sizeof *xrcd, GFP_KERNEL); - if (!xrcd) - return ERR_PTR(-ENOMEM); + if (!(dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC)) + return -EOPNOTSUPP; - err = mlx4_xrcd_alloc(to_mdev(ibdev)->dev, &xrcd->xrcdn); + err = mlx4_xrcd_alloc(dev->dev, &xrcd->xrcdn); if (err) - goto err1; + return err; - xrcd->pd = ib_alloc_pd(ibdev, 0); + xrcd->pd = ib_alloc_pd(ibxrcd->device, 0); if (IS_ERR(xrcd->pd)) { err = PTR_ERR(xrcd->pd); goto err2; } cq_attr.cqe = 1; - xrcd->cq = ib_create_cq(ibdev, NULL, NULL, xrcd, &cq_attr); + xrcd->cq = ib_create_cq(ibxrcd->device, NULL, NULL, xrcd, &cq_attr); if (IS_ERR(xrcd->cq)) { err = PTR_ERR(xrcd->cq); goto err3; } - return &xrcd->ibxrcd; + return 0; err3: ib_dealloc_pd(xrcd->pd); err2: - mlx4_xrcd_free(to_mdev(ibdev)->dev, xrcd->xrcdn); -err1: - kfree(xrcd); - return ERR_PTR(err); + mlx4_xrcd_free(dev->dev, xrcd->xrcdn); + return err; } -static int mlx4_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata) +static void mlx4_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata) { ib_destroy_cq(to_mxrcd(xrcd)->cq); ib_dealloc_pd(to_mxrcd(xrcd)->pd); mlx4_xrcd_free(to_mdev(xrcd->device)->dev, to_mxrcd(xrcd)->xrcdn); - kfree(xrcd); - - return 0; } static int add_gid_entry(struct ib_qp *ibqp, union ib_gid *gid) @@ -2600,13 +2591,6 @@ static const struct ib_device_ops mlx4_ib_dev_wq_ops = { .modify_wq = mlx4_ib_modify_wq, }; -static const struct ib_device_ops mlx4_ib_dev_fmr_ops = { - .alloc_fmr = mlx4_ib_fmr_alloc, - .dealloc_fmr = mlx4_ib_fmr_dealloc, - .map_phys_fmr = mlx4_ib_map_phys_fmr, - .unmap_fmr = mlx4_ib_unmap_fmr, -}; - static const struct ib_device_ops mlx4_ib_dev_mw_ops = { .alloc_mw = mlx4_ib_alloc_mw, .dealloc_mw = mlx4_ib_dealloc_mw, @@ -2615,6 +2599,8 @@ static const struct ib_device_ops mlx4_ib_dev_mw_ops = { static const struct ib_device_ops mlx4_ib_dev_xrc_ops = { .alloc_xrcd = mlx4_ib_alloc_xrcd, .dealloc_xrcd = mlx4_ib_dealloc_xrcd, + + INIT_RDMA_OBJ_SIZE(ib_xrcd, mlx4_ib_xrcd, ibxrcd), }; static const struct ib_device_ops mlx4_ib_dev_fs_ops = { @@ -2724,9 +2710,6 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) ib_set_device_ops(&ibdev->ib_dev, &mlx4_ib_dev_wq_ops); } - if (!mlx4_is_slave(ibdev->dev)) - ib_set_device_ops(&ibdev->ib_dev, &mlx4_ib_dev_fmr_ops); - if (dev->caps.flags & MLX4_DEV_CAP_FLAG_MEM_WINDOW || dev->caps.bmme_flags & MLX4_BMME_FLAG_TYPE_2_WIN) { ibdev->ib_dev.uverbs_cmd_mask |= diff --git a/drivers/infiniband/hw/mlx4/mcg.c b/drivers/infiniband/hw/mlx4/mcg.c index d844831179cf..5e4ec9786081 100644 --- a/drivers/infiniband/hw/mlx4/mcg.c +++ b/drivers/infiniband/hw/mlx4/mcg.c @@ -944,7 +944,7 @@ int mlx4_ib_mcg_multiplex_handler(struct ib_device *ibdev, int port, switch (sa_mad->mad_hdr.method) { case IB_MGMT_METHOD_SET: may_create = 1; - /* fall through */ + fallthrough; case IB_SA_METHOD_DELETE: req = kzalloc(sizeof *req, GFP_KERNEL); if (!req) diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index d188573187fa..38e87a700a2a 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -146,11 +146,6 @@ struct mlx4_ib_mw { struct mlx4_mw mmw; }; -struct mlx4_ib_fmr { - struct ib_fmr ibfmr; - struct mlx4_fmr mfmr; -}; - #define MAX_REGS_PER_FLOW 2 struct mlx4_flow_reg_id { @@ -679,11 +674,6 @@ static inline struct mlx4_ib_mw *to_mmw(struct ib_mw *ibmw) return container_of(ibmw, struct mlx4_ib_mw, ibmw); } -static inline struct mlx4_ib_fmr *to_mfmr(struct ib_fmr *ibfmr) -{ - return container_of(ibfmr, struct mlx4_ib_fmr, ibfmr); -} - static inline struct mlx4_ib_flow *to_mflow(struct ib_flow *ibflow) { return container_of(ibflow, struct mlx4_ib_flow, ibflow); @@ -739,7 +729,7 @@ struct ib_mw *mlx4_ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type, struct ib_udata *udata); int mlx4_ib_dealloc_mw(struct ib_mw *mw); struct ib_mr *mlx4_ib_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int mlx4_ib_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period); @@ -752,7 +742,7 @@ int mlx4_ib_arm_cq(struct ib_cq *cq, enum ib_cq_notify_flags flags); void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq); void mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq); -int mlx4_ib_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, u32 flags, +int mlx4_ib_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); int mlx4_ib_create_ah_slave(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, int slave_sgid_index, u8 *s_mac, u16 vlan_tag); @@ -794,12 +784,6 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, int mlx4_ib_mad_init(struct mlx4_ib_dev *dev); void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev); -struct ib_fmr *mlx4_ib_fmr_alloc(struct ib_pd *pd, int mr_access_flags, - struct ib_fmr_attr *fmr_attr); -int mlx4_ib_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, int npages, - u64 iova); -int mlx4_ib_unmap_fmr(struct list_head *fmr_list); -int mlx4_ib_fmr_dealloc(struct ib_fmr *fmr); int __mlx4_ib_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props, int netw_view); int __mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index, diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c index b0121c90c561..1d5ef0de12c9 100644 --- a/drivers/infiniband/hw/mlx4/mr.c +++ b/drivers/infiniband/hw/mlx4/mr.c @@ -380,7 +380,7 @@ static struct ib_umem *mlx4_get_umem_mr(struct ib_device *device, u64 start, unsigned long untagged_start = untagged_addr(start); struct vm_area_struct *vma; - down_read(¤t->mm->mmap_sem); + mmap_read_lock(current->mm); /* * FIXME: Ideally this would iterate over all the vmas that * cover the memory, but for now it requires a single vma to @@ -395,7 +395,7 @@ static struct ib_umem *mlx4_get_umem_mr(struct ib_device *device, u64 start, access_flags |= IB_ACCESS_LOCAL_WRITE; } - up_read(¤t->mm->mmap_sem); + mmap_read_unlock(current->mm); } return ib_umem_get(device, start, length, access_flags); @@ -439,7 +439,6 @@ struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key; mr->ibmr.length = length; - mr->ibmr.iova = virt_addr; mr->ibmr.page_size = 1U << shift; return &mr->ibmr; @@ -655,7 +654,7 @@ int mlx4_ib_dealloc_mw(struct ib_mw *ibmw) } struct ib_mr *mlx4_ib_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct mlx4_ib_dev *dev = to_mdev(pd->device); struct mlx4_ib_mr *mr; @@ -698,99 +697,6 @@ err_free: return ERR_PTR(err); } -struct ib_fmr *mlx4_ib_fmr_alloc(struct ib_pd *pd, int acc, - struct ib_fmr_attr *fmr_attr) -{ - struct mlx4_ib_dev *dev = to_mdev(pd->device); - struct mlx4_ib_fmr *fmr; - int err = -ENOMEM; - - fmr = kmalloc(sizeof *fmr, GFP_KERNEL); - if (!fmr) - return ERR_PTR(-ENOMEM); - - err = mlx4_fmr_alloc(dev->dev, to_mpd(pd)->pdn, convert_access(acc), - fmr_attr->max_pages, fmr_attr->max_maps, - fmr_attr->page_shift, &fmr->mfmr); - if (err) - goto err_free; - - err = mlx4_fmr_enable(to_mdev(pd->device)->dev, &fmr->mfmr); - if (err) - goto err_mr; - - fmr->ibfmr.rkey = fmr->ibfmr.lkey = fmr->mfmr.mr.key; - - return &fmr->ibfmr; - -err_mr: - (void) mlx4_mr_free(to_mdev(pd->device)->dev, &fmr->mfmr.mr); - -err_free: - kfree(fmr); - - return ERR_PTR(err); -} - -int mlx4_ib_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int npages, u64 iova) -{ - struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr); - struct mlx4_ib_dev *dev = to_mdev(ifmr->ibfmr.device); - - return mlx4_map_phys_fmr(dev->dev, &ifmr->mfmr, page_list, npages, iova, - &ifmr->ibfmr.lkey, &ifmr->ibfmr.rkey); -} - -int mlx4_ib_unmap_fmr(struct list_head *fmr_list) -{ - struct ib_fmr *ibfmr; - int err; - struct mlx4_dev *mdev = NULL; - - list_for_each_entry(ibfmr, fmr_list, list) { - if (mdev && to_mdev(ibfmr->device)->dev != mdev) - return -EINVAL; - mdev = to_mdev(ibfmr->device)->dev; - } - - if (!mdev) - return 0; - - list_for_each_entry(ibfmr, fmr_list, list) { - struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr); - - mlx4_fmr_unmap(mdev, &ifmr->mfmr, &ifmr->ibfmr.lkey, &ifmr->ibfmr.rkey); - } - - /* - * Make sure all MPT status updates are visible before issuing - * SYNC_TPT firmware command. - */ - wmb(); - - err = mlx4_SYNC_TPT(mdev); - if (err) - pr_warn("SYNC_TPT error %d when " - "unmapping FMRs\n", err); - - return 0; -} - -int mlx4_ib_fmr_dealloc(struct ib_fmr *ibfmr) -{ - struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr); - struct mlx4_ib_dev *dev = to_mdev(ibfmr->device); - int err; - - err = mlx4_fmr_free(dev->dev, &ifmr->mfmr); - - if (!err) - kfree(ifmr); - - return err; -} - static int mlx4_set_page(struct ib_mr *ibmr, u64 addr) { struct mlx4_ib_mr *mr = to_mmr(ibmr); diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index cf51e3cbd969..2975f350b9fd 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -1578,12 +1578,12 @@ static struct ib_qp *_mlx4_ib_create_qp(struct ib_pd *pd, pd = to_mxrcd(init_attr->xrcd)->pd; xrcdn = to_mxrcd(init_attr->xrcd)->xrcdn; init_attr->send_cq = to_mxrcd(init_attr->xrcd)->cq; - /* fall through */ + fallthrough; case IB_QPT_XRC_INI: if (!(to_mdev(pd->device)->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC)) return ERR_PTR(-ENOSYS); init_attr->recv_cq = init_attr->send_cq; - /* fall through */ + fallthrough; case IB_QPT_RC: case IB_QPT_UC: case IB_QPT_RAW_PACKET: @@ -1592,7 +1592,7 @@ static struct ib_qp *_mlx4_ib_create_qp(struct ib_pd *pd, return ERR_PTR(-ENOMEM); qp->pri.vid = 0xFFFF; qp->alt.vid = 0xFFFF; - /* fall through */ + fallthrough; case IB_QPT_UD: { err = create_qp_common(pd, init_attr, udata, 0, &qp); @@ -3541,11 +3541,11 @@ static int _mlx4_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, int nreq; int err = 0; unsigned ind; - int uninitialized_var(size); - unsigned uninitialized_var(seglen); + int size; + unsigned seglen; __be32 dummy; __be32 *lso_wqe; - __be32 uninitialized_var(lso_hdr_sz); + __be32 lso_hdr_sz; __be32 blh; int i; struct mlx4_ib_dev *mdev = to_mdev(ibqp->device); diff --git a/drivers/infiniband/hw/mlx5/Kconfig b/drivers/infiniband/hw/mlx5/Kconfig index ea248def4556..ef1ff42eaec5 100644 --- a/drivers/infiniband/hw/mlx5/Kconfig +++ b/drivers/infiniband/hw/mlx5/Kconfig @@ -2,7 +2,7 @@ config MLX5_INFINIBAND tristate "Mellanox 5th generation network adapters (ConnectX series) support" depends on NETDEVICES && ETHERNET && PCI && MLX5_CORE - ---help--- + help This driver provides low-level InfiniBand support for Mellanox Connect-IB PCI Express host channel adapters (HCAs). This is required to use InfiniBand protocols such as diff --git a/drivers/infiniband/hw/mlx5/Makefile b/drivers/infiniband/hw/mlx5/Makefile index 2a334800f109..b4c009bb0db6 100644 --- a/drivers/infiniband/hw/mlx5/Makefile +++ b/drivers/infiniband/hw/mlx5/Makefile @@ -1,11 +1,28 @@ # SPDX-License-Identifier: GPL-2.0-only -obj-$(CONFIG_MLX5_INFINIBAND) += mlx5_ib.o +obj-$(CONFIG_MLX5_INFINIBAND) += mlx5_ib.o + +mlx5_ib-y := ah.o \ + cmd.o \ + cong.o \ + counters.o \ + cq.o \ + doorbell.o \ + gsi.o \ + ib_virt.o \ + mad.o \ + main.o \ + mem.o \ + mr.o \ + qp.o \ + qpc.o \ + restrack.o \ + srq.o \ + srq_cmd.o \ + wr.o -mlx5_ib-y := main.o cq.o doorbell.o qp.o mem.o srq_cmd.o \ - srq.o mr.o ah.o mad.o gsi.o ib_virt.o cmd.o \ - cong.o restrack.o mlx5_ib-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += odp.o mlx5_ib-$(CONFIG_MLX5_ESWITCH) += ib_rep.o -mlx5_ib-$(CONFIG_INFINIBAND_USER_ACCESS) += devx.o -mlx5_ib-$(CONFIG_INFINIBAND_USER_ACCESS) += flow.o -mlx5_ib-$(CONFIG_INFINIBAND_USER_ACCESS) += qos.o +mlx5_ib-$(CONFIG_INFINIBAND_USER_ACCESS) += devx.o \ + fs.o \ + qos.o \ + std_types.o diff --git a/drivers/infiniband/hw/mlx5/ah.c b/drivers/infiniband/hw/mlx5/ah.c index 80642dd359bc..59e5ec39b447 100644 --- a/drivers/infiniband/hw/mlx5/ah.c +++ b/drivers/infiniband/hw/mlx5/ah.c @@ -32,9 +32,28 @@ #include "mlx5_ib.h" +static __be16 mlx5_ah_get_udp_sport(const struct mlx5_ib_dev *dev, + const struct rdma_ah_attr *ah_attr) +{ + enum ib_gid_type gid_type = ah_attr->grh.sgid_attr->gid_type; + __be16 sport; + + if ((gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) && + (rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH) && + (ah_attr->grh.flow_label & IB_GRH_FLOWLABEL_MASK)) + sport = cpu_to_be16( + rdma_flow_label_to_udp_sport(ah_attr->grh.flow_label)); + else + sport = mlx5_get_roce_udp_sport_min(dev, + ah_attr->grh.sgid_attr); + + return sport; +} + static void create_ib_ah(struct mlx5_ib_dev *dev, struct mlx5_ib_ah *ah, - struct rdma_ah_attr *ah_attr) + struct rdma_ah_init_attr *init_attr) { + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; enum ib_gid_type gid_type; if (rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH) { @@ -51,12 +70,15 @@ static void create_ib_ah(struct mlx5_ib_dev *dev, struct mlx5_ib_ah *ah, ah->av.stat_rate_sl = (rdma_ah_get_static_rate(ah_attr) << 4); if (ah_attr->type == RDMA_AH_ATTR_TYPE_ROCE) { + if (init_attr->xmit_slave) + ah->xmit_port = + mlx5_lag_get_slave_port(dev->mdev, + init_attr->xmit_slave); gid_type = ah_attr->grh.sgid_attr->gid_type; memcpy(ah->av.rmac, ah_attr->roce.dmac, sizeof(ah_attr->roce.dmac)); - ah->av.udp_sport = - mlx5_get_roce_udp_sport(dev, ah_attr->grh.sgid_attr); + ah->av.udp_sport = mlx5_ah_get_udp_sport(dev, ah_attr); ah->av.stat_rate_sl |= (rdma_ah_get_sl(ah_attr) & 0x7) << 1; if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) #define MLX5_ECN_ENABLED BIT(1) @@ -68,10 +90,11 @@ static void create_ib_ah(struct mlx5_ib_dev *dev, struct mlx5_ib_ah *ah, } } -int mlx5_ib_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata) +int mlx5_ib_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; struct mlx5_ib_ah *ah = to_mah(ibah); struct mlx5_ib_dev *dev = to_mdev(ibah->device); enum rdma_ah_attr_type ah_type = ah_attr->type; @@ -97,7 +120,7 @@ int mlx5_ib_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, return err; } - create_ib_ah(dev, ah, ah_attr); + create_ib_ah(dev, ah, init_attr); return 0; } diff --git a/drivers/infiniband/hw/mlx5/cmd.c b/drivers/infiniband/hw/mlx5/cmd.c index 4c26492ab8a3..ebb2f108b64f 100644 --- a/drivers/infiniband/hw/mlx5/cmd.c +++ b/drivers/infiniband/hw/mlx5/cmd.c @@ -1,46 +1,19 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* - * Copyright (c) 2017, Mellanox Technologies. All rights reserved. - * - * This software is available to you under a choice of one of two - * licenses. You may choose to be licensed under the terms of the GNU - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. + * Copyright (c) 2017-2020, Mellanox Technologies inc. All rights reserved. */ #include "cmd.h" int mlx5_cmd_dump_fill_mkey(struct mlx5_core_dev *dev, u32 *mkey) { - u32 out[MLX5_ST_SZ_DW(query_special_contexts_out)] = {0}; - u32 in[MLX5_ST_SZ_DW(query_special_contexts_in)] = {0}; + u32 out[MLX5_ST_SZ_DW(query_special_contexts_out)] = {}; + u32 in[MLX5_ST_SZ_DW(query_special_contexts_in)] = {}; int err; MLX5_SET(query_special_contexts_in, in, opcode, MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS); - err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + err = mlx5_cmd_exec_inout(dev, query_special_contexts, in, out); if (!err) *mkey = MLX5_GET(query_special_contexts_out, out, dump_fill_mkey); @@ -50,12 +23,12 @@ int mlx5_cmd_dump_fill_mkey(struct mlx5_core_dev *dev, u32 *mkey) int mlx5_cmd_null_mkey(struct mlx5_core_dev *dev, u32 *null_mkey) { u32 out[MLX5_ST_SZ_DW(query_special_contexts_out)] = {}; - u32 in[MLX5_ST_SZ_DW(query_special_contexts_in)] = {}; + u32 in[MLX5_ST_SZ_DW(query_special_contexts_in)] = {}; int err; MLX5_SET(query_special_contexts_in, in, opcode, MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS); - err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + err = mlx5_cmd_exec_inout(dev, query_special_contexts, in, out); if (!err) *null_mkey = MLX5_GET(query_special_contexts_out, out, null_mkey); @@ -63,23 +36,15 @@ int mlx5_cmd_null_mkey(struct mlx5_core_dev *dev, u32 *null_mkey) } int mlx5_cmd_query_cong_params(struct mlx5_core_dev *dev, int cong_point, - void *out, int out_size) + void *out) { - u32 in[MLX5_ST_SZ_DW(query_cong_params_in)] = { }; + u32 in[MLX5_ST_SZ_DW(query_cong_params_in)] = {}; MLX5_SET(query_cong_params_in, in, opcode, MLX5_CMD_OP_QUERY_CONG_PARAMS); MLX5_SET(query_cong_params_in, in, cong_protocol, cong_point); - return mlx5_cmd_exec(dev, in, sizeof(in), out, out_size); -} - -int mlx5_cmd_modify_cong_params(struct mlx5_core_dev *dev, - void *in, int in_size) -{ - u32 out[MLX5_ST_SZ_DW(modify_cong_params_out)] = { }; - - return mlx5_cmd_exec(dev, in, in_size, out, sizeof(out)); + return mlx5_cmd_exec_inout(dev, query_cong_params, in, out); } int mlx5_cmd_alloc_memic(struct mlx5_dm *dm, phys_addr_t *addr, @@ -133,7 +98,7 @@ int mlx5_cmd_alloc_memic(struct mlx5_dm *dm, phys_addr_t *addr, MLX5_SET64(alloc_memic_in, in, range_start_addr, hw_start_addr + (page_idx * PAGE_SIZE)); - ret = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + ret = mlx5_cmd_exec_inout(dev, alloc_memic, in, out); if (ret) { spin_lock(&dm->lock); bitmap_clear(dm->memic_alloc_pages, @@ -162,8 +127,7 @@ void mlx5_cmd_dealloc_memic(struct mlx5_dm *dm, phys_addr_t addr, u64 length) struct mlx5_core_dev *dev = dm->dev; u64 hw_start_addr = MLX5_CAP64_DEV_MEM(dev, memic_bar_start_addr); u32 num_pages = DIV_ROUND_UP(length, PAGE_SIZE); - u32 out[MLX5_ST_SZ_DW(dealloc_memic_out)] = {0}; - u32 in[MLX5_ST_SZ_DW(dealloc_memic_in)] = {0}; + u32 in[MLX5_ST_SZ_DW(dealloc_memic_in)] = {}; u64 start_page_idx; int err; @@ -174,7 +138,7 @@ void mlx5_cmd_dealloc_memic(struct mlx5_dm *dm, phys_addr_t addr, u64 length) MLX5_SET64(dealloc_memic_in, in, memic_start_addr, addr); MLX5_SET(dealloc_memic_in, in, memic_size, length); - err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + err = mlx5_cmd_exec_in(dev, dealloc_memic, in); if (err) return; @@ -184,63 +148,48 @@ void mlx5_cmd_dealloc_memic(struct mlx5_dm *dm, phys_addr_t addr, u64 length) spin_unlock(&dm->lock); } -int mlx5_cmd_query_ext_ppcnt_counters(struct mlx5_core_dev *dev, void *out) -{ - u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {}; - int sz = MLX5_ST_SZ_BYTES(ppcnt_reg); - - MLX5_SET(ppcnt_reg, in, local_port, 1); - - MLX5_SET(ppcnt_reg, in, grp, MLX5_ETHERNET_EXTENDED_COUNTERS_GROUP); - return mlx5_core_access_reg(dev, in, sz, out, sz, MLX5_REG_PPCNT, - 0, 0); -} - void mlx5_cmd_destroy_tir(struct mlx5_core_dev *dev, u32 tirn, u16 uid) { - u32 in[MLX5_ST_SZ_DW(destroy_tir_in)] = {}; - u32 out[MLX5_ST_SZ_DW(destroy_tir_out)] = {}; + u32 in[MLX5_ST_SZ_DW(destroy_tir_in)] = {}; MLX5_SET(destroy_tir_in, in, opcode, MLX5_CMD_OP_DESTROY_TIR); MLX5_SET(destroy_tir_in, in, tirn, tirn); MLX5_SET(destroy_tir_in, in, uid, uid); - mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + mlx5_cmd_exec_in(dev, destroy_tir, in); } void mlx5_cmd_destroy_tis(struct mlx5_core_dev *dev, u32 tisn, u16 uid) { - u32 in[MLX5_ST_SZ_DW(destroy_tis_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(destroy_tis_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(destroy_tis_in)] = {}; MLX5_SET(destroy_tis_in, in, opcode, MLX5_CMD_OP_DESTROY_TIS); MLX5_SET(destroy_tis_in, in, tisn, tisn); MLX5_SET(destroy_tis_in, in, uid, uid); - mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + mlx5_cmd_exec_in(dev, destroy_tis, in); } void mlx5_cmd_destroy_rqt(struct mlx5_core_dev *dev, u32 rqtn, u16 uid) { - u32 in[MLX5_ST_SZ_DW(destroy_rqt_in)] = {}; - u32 out[MLX5_ST_SZ_DW(destroy_rqt_out)] = {}; + u32 in[MLX5_ST_SZ_DW(destroy_rqt_in)] = {}; MLX5_SET(destroy_rqt_in, in, opcode, MLX5_CMD_OP_DESTROY_RQT); MLX5_SET(destroy_rqt_in, in, rqtn, rqtn); MLX5_SET(destroy_rqt_in, in, uid, uid); - mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + mlx5_cmd_exec_in(dev, destroy_rqt, in); } int mlx5_cmd_alloc_transport_domain(struct mlx5_core_dev *dev, u32 *tdn, u16 uid) { - u32 in[MLX5_ST_SZ_DW(alloc_transport_domain_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(alloc_transport_domain_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(alloc_transport_domain_in)] = {}; + u32 out[MLX5_ST_SZ_DW(alloc_transport_domain_out)] = {}; int err; MLX5_SET(alloc_transport_domain_in, in, opcode, MLX5_CMD_OP_ALLOC_TRANSPORT_DOMAIN); MLX5_SET(alloc_transport_domain_in, in, uid, uid); - err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + err = mlx5_cmd_exec_inout(dev, alloc_transport_domain, in, out); if (!err) *tdn = MLX5_GET(alloc_transport_domain_out, out, transport_domain); @@ -251,32 +200,29 @@ int mlx5_cmd_alloc_transport_domain(struct mlx5_core_dev *dev, u32 *tdn, void mlx5_cmd_dealloc_transport_domain(struct mlx5_core_dev *dev, u32 tdn, u16 uid) { - u32 in[MLX5_ST_SZ_DW(dealloc_transport_domain_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(dealloc_transport_domain_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(dealloc_transport_domain_in)] = {}; MLX5_SET(dealloc_transport_domain_in, in, opcode, MLX5_CMD_OP_DEALLOC_TRANSPORT_DOMAIN); MLX5_SET(dealloc_transport_domain_in, in, uid, uid); MLX5_SET(dealloc_transport_domain_in, in, transport_domain, tdn); - mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + mlx5_cmd_exec_in(dev, dealloc_transport_domain, in); } void mlx5_cmd_dealloc_pd(struct mlx5_core_dev *dev, u32 pdn, u16 uid) { - u32 out[MLX5_ST_SZ_DW(dealloc_pd_out)] = {}; - u32 in[MLX5_ST_SZ_DW(dealloc_pd_in)] = {}; + u32 in[MLX5_ST_SZ_DW(dealloc_pd_in)] = {}; MLX5_SET(dealloc_pd_in, in, opcode, MLX5_CMD_OP_DEALLOC_PD); MLX5_SET(dealloc_pd_in, in, pd, pdn); MLX5_SET(dealloc_pd_in, in, uid, uid); - mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + mlx5_cmd_exec_in(dev, dealloc_pd, in); } int mlx5_cmd_attach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn, u16 uid) { - u32 out[MLX5_ST_SZ_DW(attach_to_mcg_out)] = {}; - u32 in[MLX5_ST_SZ_DW(attach_to_mcg_in)] = {}; + u32 in[MLX5_ST_SZ_DW(attach_to_mcg_in)] = {}; void *gid; MLX5_SET(attach_to_mcg_in, in, opcode, MLX5_CMD_OP_ATTACH_TO_MCG); @@ -284,14 +230,13 @@ int mlx5_cmd_attach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, MLX5_SET(attach_to_mcg_in, in, uid, uid); gid = MLX5_ADDR_OF(attach_to_mcg_in, in, multicast_gid); memcpy(gid, mgid, sizeof(*mgid)); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec_in(dev, attach_to_mcg, in); } int mlx5_cmd_detach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn, u16 uid) { - u32 out[MLX5_ST_SZ_DW(detach_from_mcg_out)] = {}; - u32 in[MLX5_ST_SZ_DW(detach_from_mcg_in)] = {}; + u32 in[MLX5_ST_SZ_DW(detach_from_mcg_in)] = {}; void *gid; MLX5_SET(detach_from_mcg_in, in, opcode, MLX5_CMD_OP_DETACH_FROM_MCG); @@ -299,18 +244,18 @@ int mlx5_cmd_detach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, MLX5_SET(detach_from_mcg_in, in, uid, uid); gid = MLX5_ADDR_OF(detach_from_mcg_in, in, multicast_gid); memcpy(gid, mgid, sizeof(*mgid)); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec_in(dev, detach_from_mcg, in); } int mlx5_cmd_xrcd_alloc(struct mlx5_core_dev *dev, u32 *xrcdn, u16 uid) { u32 out[MLX5_ST_SZ_DW(alloc_xrcd_out)] = {}; - u32 in[MLX5_ST_SZ_DW(alloc_xrcd_in)] = {}; + u32 in[MLX5_ST_SZ_DW(alloc_xrcd_in)] = {}; int err; MLX5_SET(alloc_xrcd_in, in, opcode, MLX5_CMD_OP_ALLOC_XRCD); MLX5_SET(alloc_xrcd_in, in, uid, uid); - err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + err = mlx5_cmd_exec_inout(dev, alloc_xrcd, in, out); if (!err) *xrcdn = MLX5_GET(alloc_xrcd_out, out, xrcd); return err; @@ -318,30 +263,12 @@ int mlx5_cmd_xrcd_alloc(struct mlx5_core_dev *dev, u32 *xrcdn, u16 uid) int mlx5_cmd_xrcd_dealloc(struct mlx5_core_dev *dev, u32 xrcdn, u16 uid) { - u32 out[MLX5_ST_SZ_DW(dealloc_xrcd_out)] = {}; - u32 in[MLX5_ST_SZ_DW(dealloc_xrcd_in)] = {}; + u32 in[MLX5_ST_SZ_DW(dealloc_xrcd_in)] = {}; MLX5_SET(dealloc_xrcd_in, in, opcode, MLX5_CMD_OP_DEALLOC_XRCD); MLX5_SET(dealloc_xrcd_in, in, xrcd, xrcdn); MLX5_SET(dealloc_xrcd_in, in, uid, uid); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); -} - -int mlx5_cmd_alloc_q_counter(struct mlx5_core_dev *dev, u16 *counter_id, - u16 uid) -{ - u32 in[MLX5_ST_SZ_DW(alloc_q_counter_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(alloc_q_counter_out)] = {0}; - int err; - - MLX5_SET(alloc_q_counter_in, in, opcode, MLX5_CMD_OP_ALLOC_Q_COUNTER); - MLX5_SET(alloc_q_counter_in, in, uid, uid); - - err = mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); - if (!err) - *counter_id = MLX5_GET(alloc_q_counter_out, out, - counter_set_id); - return err; + return mlx5_cmd_exec_in(dev, dealloc_xrcd, in); } int mlx5_cmd_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb, @@ -367,7 +294,7 @@ int mlx5_cmd_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb, data = MLX5_ADDR_OF(mad_ifc_in, in, mad); memcpy(data, inb, MLX5_FLD_SZ_BYTES(mad_ifc_in, mad)); - err = mlx5_cmd_exec(dev, in, inlen, out, outlen); + err = mlx5_cmd_exec_inout(dev, mad_ifc, in, out); if (err) goto out; diff --git a/drivers/infiniband/hw/mlx5/cmd.h b/drivers/infiniband/hw/mlx5/cmd.h index 945ebce73613..1d192a8ca87d 100644 --- a/drivers/infiniband/hw/mlx5/cmd.h +++ b/drivers/infiniband/hw/mlx5/cmd.h @@ -40,10 +40,7 @@ int mlx5_cmd_dump_fill_mkey(struct mlx5_core_dev *dev, u32 *mkey); int mlx5_cmd_null_mkey(struct mlx5_core_dev *dev, u32 *null_mkey); int mlx5_cmd_query_cong_params(struct mlx5_core_dev *dev, int cong_point, - void *out, int out_size); -int mlx5_cmd_query_ext_ppcnt_counters(struct mlx5_core_dev *dev, void *out); -int mlx5_cmd_modify_cong_params(struct mlx5_core_dev *mdev, - void *in, int in_size); + void *out); int mlx5_cmd_alloc_memic(struct mlx5_dm *dm, phys_addr_t *addr, u64 length, u32 alignment); void mlx5_cmd_dealloc_memic(struct mlx5_dm *dm, phys_addr_t addr, u64 length); @@ -61,8 +58,6 @@ int mlx5_cmd_detach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn, u16 uid); int mlx5_cmd_xrcd_alloc(struct mlx5_core_dev *dev, u32 *xrcdn, u16 uid); int mlx5_cmd_xrcd_dealloc(struct mlx5_core_dev *dev, u32 xrcdn, u16 uid); -int mlx5_cmd_alloc_q_counter(struct mlx5_core_dev *dev, u16 *counter_id, - u16 uid); int mlx5_cmd_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb, u16 opmod, u8 port); #endif /* MLX5_IB_CMD_H */ diff --git a/drivers/infiniband/hw/mlx5/cong.c b/drivers/infiniband/hw/mlx5/cong.c index de4da92b81a6..b9291e482428 100644 --- a/drivers/infiniband/hw/mlx5/cong.c +++ b/drivers/infiniband/hw/mlx5/cong.c @@ -290,7 +290,7 @@ static int mlx5_ib_get_cc_params(struct mlx5_ib_dev *dev, u8 port_num, node = mlx5_ib_param_to_node(offset); - err = mlx5_cmd_query_cong_params(mdev, node, out, outlen); + err = mlx5_cmd_query_cong_params(mdev, node, out); if (err) goto free; @@ -339,7 +339,7 @@ static int mlx5_ib_set_cc_params(struct mlx5_ib_dev *dev, u8 port_num, MLX5_SET(field_select_r_roce_rp, field, field_select_r_roce_rp, attr_mask); - err = mlx5_cmd_modify_cong_params(mdev, in, inlen); + err = mlx5_cmd_exec_in(dev->mdev, modify_cong_params, in); kvfree(in); alloc_err: mlx5_ib_put_native_port_mdev(dev, port_num + 1); diff --git a/drivers/infiniband/hw/mlx5/counters.c b/drivers/infiniband/hw/mlx5/counters.c new file mode 100644 index 000000000000..145f3cb40ccb --- /dev/null +++ b/drivers/infiniband/hw/mlx5/counters.c @@ -0,0 +1,709 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + */ + +#include "mlx5_ib.h" +#include <linux/mlx5/eswitch.h> +#include "counters.h" +#include "ib_rep.h" +#include "qp.h" + +struct mlx5_ib_counter { + const char *name; + size_t offset; +}; + +#define INIT_Q_COUNTER(_name) \ + { .name = #_name, .offset = MLX5_BYTE_OFF(query_q_counter_out, _name)} + +static const struct mlx5_ib_counter basic_q_cnts[] = { + INIT_Q_COUNTER(rx_write_requests), + INIT_Q_COUNTER(rx_read_requests), + INIT_Q_COUNTER(rx_atomic_requests), + INIT_Q_COUNTER(out_of_buffer), +}; + +static const struct mlx5_ib_counter out_of_seq_q_cnts[] = { + INIT_Q_COUNTER(out_of_sequence), +}; + +static const struct mlx5_ib_counter retrans_q_cnts[] = { + INIT_Q_COUNTER(duplicate_request), + INIT_Q_COUNTER(rnr_nak_retry_err), + INIT_Q_COUNTER(packet_seq_err), + INIT_Q_COUNTER(implied_nak_seq_err), + INIT_Q_COUNTER(local_ack_timeout_err), +}; + +#define INIT_CONG_COUNTER(_name) \ + { .name = #_name, .offset = \ + MLX5_BYTE_OFF(query_cong_statistics_out, _name ## _high)} + +static const struct mlx5_ib_counter cong_cnts[] = { + INIT_CONG_COUNTER(rp_cnp_ignored), + INIT_CONG_COUNTER(rp_cnp_handled), + INIT_CONG_COUNTER(np_ecn_marked_roce_packets), + INIT_CONG_COUNTER(np_cnp_sent), +}; + +static const struct mlx5_ib_counter extended_err_cnts[] = { + INIT_Q_COUNTER(resp_local_length_error), + INIT_Q_COUNTER(resp_cqe_error), + INIT_Q_COUNTER(req_cqe_error), + INIT_Q_COUNTER(req_remote_invalid_request), + INIT_Q_COUNTER(req_remote_access_errors), + INIT_Q_COUNTER(resp_remote_access_errors), + INIT_Q_COUNTER(resp_cqe_flush_error), + INIT_Q_COUNTER(req_cqe_flush_error), +}; + +static const struct mlx5_ib_counter roce_accl_cnts[] = { + INIT_Q_COUNTER(roce_adp_retrans), + INIT_Q_COUNTER(roce_adp_retrans_to), + INIT_Q_COUNTER(roce_slow_restart), + INIT_Q_COUNTER(roce_slow_restart_cnps), + INIT_Q_COUNTER(roce_slow_restart_trans), +}; + +#define INIT_EXT_PPCNT_COUNTER(_name) \ + { .name = #_name, .offset = \ + MLX5_BYTE_OFF(ppcnt_reg, \ + counter_set.eth_extended_cntrs_grp_data_layout._name##_high)} + +static const struct mlx5_ib_counter ext_ppcnt_cnts[] = { + INIT_EXT_PPCNT_COUNTER(rx_icrc_encapsulated), +}; + +static int mlx5_ib_read_counters(struct ib_counters *counters, + struct ib_counters_read_attr *read_attr, + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); + struct mlx5_read_counters_attr mread_attr = {}; + struct mlx5_ib_flow_counters_desc *desc; + int ret, i; + + mutex_lock(&mcounters->mcntrs_mutex); + if (mcounters->cntrs_max_index > read_attr->ncounters) { + ret = -EINVAL; + goto err_bound; + } + + mread_attr.out = kcalloc(mcounters->counters_num, sizeof(u64), + GFP_KERNEL); + if (!mread_attr.out) { + ret = -ENOMEM; + goto err_bound; + } + + mread_attr.hw_cntrs_hndl = mcounters->hw_cntrs_hndl; + mread_attr.flags = read_attr->flags; + ret = mcounters->read_counters(counters->device, &mread_attr); + if (ret) + goto err_read; + + /* do the pass over the counters data array to assign according to the + * descriptions and indexing pairs + */ + desc = mcounters->counters_data; + for (i = 0; i < mcounters->ncounters; i++) + read_attr->counters_buff[desc[i].index] += mread_attr.out[desc[i].description]; + +err_read: + kfree(mread_attr.out); +err_bound: + mutex_unlock(&mcounters->mcntrs_mutex); + return ret; +} + +static void mlx5_ib_destroy_counters(struct ib_counters *counters) +{ + struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); + + mlx5_ib_counters_clear_description(counters); + if (mcounters->hw_cntrs_hndl) + mlx5_fc_destroy(to_mdev(counters->device)->mdev, + mcounters->hw_cntrs_hndl); +} + +static int mlx5_ib_create_counters(struct ib_counters *counters, + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); + + mutex_init(&mcounters->mcntrs_mutex); + return 0; +} + + +static bool is_mdev_switchdev_mode(const struct mlx5_core_dev *mdev) +{ + return MLX5_ESWITCH_MANAGER(mdev) && + mlx5_ib_eswitch_mode(mdev->priv.eswitch) == + MLX5_ESWITCH_OFFLOADS; +} + +static const struct mlx5_ib_counters *get_counters(struct mlx5_ib_dev *dev, + u8 port_num) +{ + return is_mdev_switchdev_mode(dev->mdev) ? &dev->port[0].cnts : + &dev->port[port_num].cnts; +} + +/** + * mlx5_ib_get_counters_id - Returns counters id to use for device+port + * @dev: Pointer to mlx5 IB device + * @port_num: Zero based port number + * + * mlx5_ib_get_counters_id() Returns counters set id to use for given + * device port combination in switchdev and non switchdev mode of the + * parent device. + */ +u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num) +{ + const struct mlx5_ib_counters *cnts = get_counters(dev, port_num); + + return cnts->set_id; +} + +static struct rdma_hw_stats *mlx5_ib_alloc_hw_stats(struct ib_device *ibdev, + u8 port_num) +{ + struct mlx5_ib_dev *dev = to_mdev(ibdev); + const struct mlx5_ib_counters *cnts; + bool is_switchdev = is_mdev_switchdev_mode(dev->mdev); + + if ((is_switchdev && port_num) || (!is_switchdev && !port_num)) + return NULL; + + cnts = get_counters(dev, port_num - 1); + + return rdma_alloc_hw_stats_struct(cnts->names, + cnts->num_q_counters + + cnts->num_cong_counters + + cnts->num_ext_ppcnt_counters, + RDMA_HW_STATS_DEFAULT_LIFESPAN); +} + +static int mlx5_ib_query_q_counters(struct mlx5_core_dev *mdev, + const struct mlx5_ib_counters *cnts, + struct rdma_hw_stats *stats, + u16 set_id) +{ + u32 out[MLX5_ST_SZ_DW(query_q_counter_out)] = {}; + u32 in[MLX5_ST_SZ_DW(query_q_counter_in)] = {}; + __be32 val; + int ret, i; + + MLX5_SET(query_q_counter_in, in, opcode, MLX5_CMD_OP_QUERY_Q_COUNTER); + MLX5_SET(query_q_counter_in, in, counter_set_id, set_id); + ret = mlx5_cmd_exec_inout(mdev, query_q_counter, in, out); + if (ret) + return ret; + + for (i = 0; i < cnts->num_q_counters; i++) { + val = *(__be32 *)((void *)out + cnts->offsets[i]); + stats->value[i] = (u64)be32_to_cpu(val); + } + + return 0; +} + +static int mlx5_ib_query_ext_ppcnt_counters(struct mlx5_ib_dev *dev, + const struct mlx5_ib_counters *cnts, + struct rdma_hw_stats *stats) +{ + int offset = cnts->num_q_counters + cnts->num_cong_counters; + u32 in[MLX5_ST_SZ_DW(ppcnt_reg)] = {}; + int sz = MLX5_ST_SZ_BYTES(ppcnt_reg); + int ret, i; + void *out; + + out = kvzalloc(sz, GFP_KERNEL); + if (!out) + return -ENOMEM; + + MLX5_SET(ppcnt_reg, in, local_port, 1); + MLX5_SET(ppcnt_reg, in, grp, MLX5_ETHERNET_EXTENDED_COUNTERS_GROUP); + ret = mlx5_core_access_reg(dev->mdev, in, sz, out, sz, MLX5_REG_PPCNT, + 0, 0); + if (ret) + goto free; + + for (i = 0; i < cnts->num_ext_ppcnt_counters; i++) + stats->value[i + offset] = + be64_to_cpup((__be64 *)(out + + cnts->offsets[i + offset])); +free: + kvfree(out); + return ret; +} + +static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, + struct rdma_hw_stats *stats, + u8 port_num, int index) +{ + struct mlx5_ib_dev *dev = to_mdev(ibdev); + const struct mlx5_ib_counters *cnts = get_counters(dev, port_num - 1); + struct mlx5_core_dev *mdev; + int ret, num_counters; + u8 mdev_port_num; + + if (!stats) + return -EINVAL; + + num_counters = cnts->num_q_counters + + cnts->num_cong_counters + + cnts->num_ext_ppcnt_counters; + + /* q_counters are per IB device, query the master mdev */ + ret = mlx5_ib_query_q_counters(dev->mdev, cnts, stats, cnts->set_id); + if (ret) + return ret; + + if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { + ret = mlx5_ib_query_ext_ppcnt_counters(dev, cnts, stats); + if (ret) + return ret; + } + + if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) { + mdev = mlx5_ib_get_native_port_mdev(dev, port_num, + &mdev_port_num); + if (!mdev) { + /* If port is not affiliated yet, its in down state + * which doesn't have any counters yet, so it would be + * zero. So no need to read from the HCA. + */ + goto done; + } + ret = mlx5_lag_query_cong_counters(dev->mdev, + stats->value + + cnts->num_q_counters, + cnts->num_cong_counters, + cnts->offsets + + cnts->num_q_counters); + + mlx5_ib_put_native_port_mdev(dev, port_num); + if (ret) + return ret; + } + +done: + return num_counters; +} + +static struct rdma_hw_stats * +mlx5_ib_counter_alloc_stats(struct rdma_counter *counter) +{ + struct mlx5_ib_dev *dev = to_mdev(counter->device); + const struct mlx5_ib_counters *cnts = + get_counters(dev, counter->port - 1); + + return rdma_alloc_hw_stats_struct(cnts->names, + cnts->num_q_counters + + cnts->num_cong_counters + + cnts->num_ext_ppcnt_counters, + RDMA_HW_STATS_DEFAULT_LIFESPAN); +} + +static int mlx5_ib_counter_update_stats(struct rdma_counter *counter) +{ + struct mlx5_ib_dev *dev = to_mdev(counter->device); + const struct mlx5_ib_counters *cnts = + get_counters(dev, counter->port - 1); + + return mlx5_ib_query_q_counters(dev->mdev, cnts, + counter->stats, counter->id); +} + +static int mlx5_ib_counter_dealloc(struct rdma_counter *counter) +{ + struct mlx5_ib_dev *dev = to_mdev(counter->device); + u32 in[MLX5_ST_SZ_DW(dealloc_q_counter_in)] = {}; + + if (!counter->id) + return 0; + + MLX5_SET(dealloc_q_counter_in, in, opcode, + MLX5_CMD_OP_DEALLOC_Q_COUNTER); + MLX5_SET(dealloc_q_counter_in, in, counter_set_id, counter->id); + return mlx5_cmd_exec_in(dev->mdev, dealloc_q_counter, in); +} + +static int mlx5_ib_counter_bind_qp(struct rdma_counter *counter, + struct ib_qp *qp) +{ + struct mlx5_ib_dev *dev = to_mdev(qp->device); + int err; + + if (!counter->id) { + u32 out[MLX5_ST_SZ_DW(alloc_q_counter_out)] = {}; + u32 in[MLX5_ST_SZ_DW(alloc_q_counter_in)] = {}; + + MLX5_SET(alloc_q_counter_in, in, opcode, + MLX5_CMD_OP_ALLOC_Q_COUNTER); + MLX5_SET(alloc_q_counter_in, in, uid, MLX5_SHARED_RESOURCE_UID); + err = mlx5_cmd_exec_inout(dev->mdev, alloc_q_counter, in, out); + if (err) + return err; + counter->id = + MLX5_GET(alloc_q_counter_out, out, counter_set_id); + } + + err = mlx5_ib_qp_set_counter(qp, counter); + if (err) + goto fail_set_counter; + + return 0; + +fail_set_counter: + mlx5_ib_counter_dealloc(counter); + counter->id = 0; + + return err; +} + +static int mlx5_ib_counter_unbind_qp(struct ib_qp *qp) +{ + return mlx5_ib_qp_set_counter(qp, NULL); +} + + +static void mlx5_ib_fill_counters(struct mlx5_ib_dev *dev, + const char **names, + size_t *offsets) +{ + int i; + int j = 0; + + for (i = 0; i < ARRAY_SIZE(basic_q_cnts); i++, j++) { + names[j] = basic_q_cnts[i].name; + offsets[j] = basic_q_cnts[i].offset; + } + + if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt)) { + for (i = 0; i < ARRAY_SIZE(out_of_seq_q_cnts); i++, j++) { + names[j] = out_of_seq_q_cnts[i].name; + offsets[j] = out_of_seq_q_cnts[i].offset; + } + } + + if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) { + for (i = 0; i < ARRAY_SIZE(retrans_q_cnts); i++, j++) { + names[j] = retrans_q_cnts[i].name; + offsets[j] = retrans_q_cnts[i].offset; + } + } + + if (MLX5_CAP_GEN(dev->mdev, enhanced_error_q_counters)) { + for (i = 0; i < ARRAY_SIZE(extended_err_cnts); i++, j++) { + names[j] = extended_err_cnts[i].name; + offsets[j] = extended_err_cnts[i].offset; + } + } + + if (MLX5_CAP_GEN(dev->mdev, roce_accl)) { + for (i = 0; i < ARRAY_SIZE(roce_accl_cnts); i++, j++) { + names[j] = roce_accl_cnts[i].name; + offsets[j] = roce_accl_cnts[i].offset; + } + } + + if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) { + for (i = 0; i < ARRAY_SIZE(cong_cnts); i++, j++) { + names[j] = cong_cnts[i].name; + offsets[j] = cong_cnts[i].offset; + } + } + + if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { + for (i = 0; i < ARRAY_SIZE(ext_ppcnt_cnts); i++, j++) { + names[j] = ext_ppcnt_cnts[i].name; + offsets[j] = ext_ppcnt_cnts[i].offset; + } + } +} + + +static int __mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev, + struct mlx5_ib_counters *cnts) +{ + u32 num_counters; + + num_counters = ARRAY_SIZE(basic_q_cnts); + + if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt)) + num_counters += ARRAY_SIZE(out_of_seq_q_cnts); + + if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) + num_counters += ARRAY_SIZE(retrans_q_cnts); + + if (MLX5_CAP_GEN(dev->mdev, enhanced_error_q_counters)) + num_counters += ARRAY_SIZE(extended_err_cnts); + + if (MLX5_CAP_GEN(dev->mdev, roce_accl)) + num_counters += ARRAY_SIZE(roce_accl_cnts); + + cnts->num_q_counters = num_counters; + + if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) { + cnts->num_cong_counters = ARRAY_SIZE(cong_cnts); + num_counters += ARRAY_SIZE(cong_cnts); + } + if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { + cnts->num_ext_ppcnt_counters = ARRAY_SIZE(ext_ppcnt_cnts); + num_counters += ARRAY_SIZE(ext_ppcnt_cnts); + } + cnts->names = kcalloc(num_counters, sizeof(cnts->names), GFP_KERNEL); + if (!cnts->names) + return -ENOMEM; + + cnts->offsets = kcalloc(num_counters, + sizeof(cnts->offsets), GFP_KERNEL); + if (!cnts->offsets) + goto err_names; + + return 0; + +err_names: + kfree(cnts->names); + cnts->names = NULL; + return -ENOMEM; +} + +static void mlx5_ib_dealloc_counters(struct mlx5_ib_dev *dev) +{ + u32 in[MLX5_ST_SZ_DW(dealloc_q_counter_in)] = {}; + int num_cnt_ports; + int i; + + num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports; + + MLX5_SET(dealloc_q_counter_in, in, opcode, + MLX5_CMD_OP_DEALLOC_Q_COUNTER); + + for (i = 0; i < num_cnt_ports; i++) { + if (dev->port[i].cnts.set_id) { + MLX5_SET(dealloc_q_counter_in, in, counter_set_id, + dev->port[i].cnts.set_id); + mlx5_cmd_exec_in(dev->mdev, dealloc_q_counter, in); + } + kfree(dev->port[i].cnts.names); + kfree(dev->port[i].cnts.offsets); + } +} + +static int mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev) +{ + u32 out[MLX5_ST_SZ_DW(alloc_q_counter_out)] = {}; + u32 in[MLX5_ST_SZ_DW(alloc_q_counter_in)] = {}; + int num_cnt_ports; + int err = 0; + int i; + bool is_shared; + + MLX5_SET(alloc_q_counter_in, in, opcode, MLX5_CMD_OP_ALLOC_Q_COUNTER); + is_shared = MLX5_CAP_GEN(dev->mdev, log_max_uctx) != 0; + num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports; + + for (i = 0; i < num_cnt_ports; i++) { + err = __mlx5_ib_alloc_counters(dev, &dev->port[i].cnts); + if (err) + goto err_alloc; + + mlx5_ib_fill_counters(dev, dev->port[i].cnts.names, + dev->port[i].cnts.offsets); + + MLX5_SET(alloc_q_counter_in, in, uid, + is_shared ? MLX5_SHARED_RESOURCE_UID : 0); + + err = mlx5_cmd_exec_inout(dev->mdev, alloc_q_counter, in, out); + if (err) { + mlx5_ib_warn(dev, + "couldn't allocate queue counter for port %d, err %d\n", + i + 1, err); + goto err_alloc; + } + + dev->port[i].cnts.set_id = + MLX5_GET(alloc_q_counter_out, out, counter_set_id); + } + return 0; + +err_alloc: + mlx5_ib_dealloc_counters(dev); + return err; +} + +static int read_flow_counters(struct ib_device *ibdev, + struct mlx5_read_counters_attr *read_attr) +{ + struct mlx5_fc *fc = read_attr->hw_cntrs_hndl; + struct mlx5_ib_dev *dev = to_mdev(ibdev); + + return mlx5_fc_query(dev->mdev, fc, + &read_attr->out[IB_COUNTER_PACKETS], + &read_attr->out[IB_COUNTER_BYTES]); +} + +/* flow counters currently expose two counters packets and bytes */ +#define FLOW_COUNTERS_NUM 2 +static int counters_set_description( + struct ib_counters *counters, enum mlx5_ib_counters_type counters_type, + struct mlx5_ib_flow_counters_desc *desc_data, u32 ncounters) +{ + struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); + u32 cntrs_max_index = 0; + int i; + + if (counters_type != MLX5_IB_COUNTERS_FLOW) + return -EINVAL; + + /* init the fields for the object */ + mcounters->type = counters_type; + mcounters->read_counters = read_flow_counters; + mcounters->counters_num = FLOW_COUNTERS_NUM; + mcounters->ncounters = ncounters; + /* each counter entry have both description and index pair */ + for (i = 0; i < ncounters; i++) { + if (desc_data[i].description > IB_COUNTER_BYTES) + return -EINVAL; + + if (cntrs_max_index <= desc_data[i].index) + cntrs_max_index = desc_data[i].index + 1; + } + + mutex_lock(&mcounters->mcntrs_mutex); + mcounters->counters_data = desc_data; + mcounters->cntrs_max_index = cntrs_max_index; + mutex_unlock(&mcounters->mcntrs_mutex); + + return 0; +} + +#define MAX_COUNTERS_NUM (USHRT_MAX / (sizeof(u32) * 2)) +int mlx5_ib_flow_counters_set_data(struct ib_counters *ibcounters, + struct mlx5_ib_create_flow *ucmd) +{ + struct mlx5_ib_mcounters *mcounters = to_mcounters(ibcounters); + struct mlx5_ib_flow_counters_data *cntrs_data = NULL; + struct mlx5_ib_flow_counters_desc *desc_data = NULL; + bool hw_hndl = false; + int ret = 0; + + if (ucmd && ucmd->ncounters_data != 0) { + cntrs_data = ucmd->data; + if (cntrs_data->ncounters > MAX_COUNTERS_NUM) + return -EINVAL; + + desc_data = kcalloc(cntrs_data->ncounters, + sizeof(*desc_data), + GFP_KERNEL); + if (!desc_data) + return -ENOMEM; + + if (copy_from_user(desc_data, + u64_to_user_ptr(cntrs_data->counters_data), + sizeof(*desc_data) * cntrs_data->ncounters)) { + ret = -EFAULT; + goto free; + } + } + + if (!mcounters->hw_cntrs_hndl) { + mcounters->hw_cntrs_hndl = mlx5_fc_create( + to_mdev(ibcounters->device)->mdev, false); + if (IS_ERR(mcounters->hw_cntrs_hndl)) { + ret = PTR_ERR(mcounters->hw_cntrs_hndl); + goto free; + } + hw_hndl = true; + } + + if (desc_data) { + /* counters already bound to at least one flow */ + if (mcounters->cntrs_max_index) { + ret = -EINVAL; + goto free_hndl; + } + + ret = counters_set_description(ibcounters, + MLX5_IB_COUNTERS_FLOW, + desc_data, + cntrs_data->ncounters); + if (ret) + goto free_hndl; + + } else if (!mcounters->cntrs_max_index) { + /* counters not bound yet, must have udata passed */ + ret = -EINVAL; + goto free_hndl; + } + + return 0; + +free_hndl: + if (hw_hndl) { + mlx5_fc_destroy(to_mdev(ibcounters->device)->mdev, + mcounters->hw_cntrs_hndl); + mcounters->hw_cntrs_hndl = NULL; + } +free: + kfree(desc_data); + return ret; +} + +void mlx5_ib_counters_clear_description(struct ib_counters *counters) +{ + struct mlx5_ib_mcounters *mcounters; + + if (!counters || atomic_read(&counters->usecnt) != 1) + return; + + mcounters = to_mcounters(counters); + + mutex_lock(&mcounters->mcntrs_mutex); + kfree(mcounters->counters_data); + mcounters->counters_data = NULL; + mcounters->cntrs_max_index = 0; + mutex_unlock(&mcounters->mcntrs_mutex); +} + +static const struct ib_device_ops hw_stats_ops = { + .alloc_hw_stats = mlx5_ib_alloc_hw_stats, + .get_hw_stats = mlx5_ib_get_hw_stats, + .counter_bind_qp = mlx5_ib_counter_bind_qp, + .counter_unbind_qp = mlx5_ib_counter_unbind_qp, + .counter_dealloc = mlx5_ib_counter_dealloc, + .counter_alloc_stats = mlx5_ib_counter_alloc_stats, + .counter_update_stats = mlx5_ib_counter_update_stats, +}; + +static const struct ib_device_ops counters_ops = { + .create_counters = mlx5_ib_create_counters, + .destroy_counters = mlx5_ib_destroy_counters, + .read_counters = mlx5_ib_read_counters, + + INIT_RDMA_OBJ_SIZE(ib_counters, mlx5_ib_mcounters, ibcntrs), +}; + +int mlx5_ib_counters_init(struct mlx5_ib_dev *dev) +{ + ib_set_device_ops(&dev->ib_dev, &counters_ops); + + if (!MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) + return 0; + + ib_set_device_ops(&dev->ib_dev, &hw_stats_ops); + return mlx5_ib_alloc_counters(dev); +} + +void mlx5_ib_counters_cleanup(struct mlx5_ib_dev *dev) +{ + if (!MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) + return; + + mlx5_ib_dealloc_counters(dev); +} diff --git a/drivers/infiniband/hw/mlx5/counters.h b/drivers/infiniband/hw/mlx5/counters.h new file mode 100644 index 000000000000..1aa30c2f3f4d --- /dev/null +++ b/drivers/infiniband/hw/mlx5/counters.h @@ -0,0 +1,17 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + */ + +#ifndef _MLX5_IB_COUNTERS_H +#define _MLX5_IB_COUNTERS_H + +#include "mlx5_ib.h" + +int mlx5_ib_counters_init(struct mlx5_ib_dev *dev); +void mlx5_ib_counters_cleanup(struct mlx5_ib_dev *dev); +void mlx5_ib_counters_clear_description(struct ib_counters *counters); +int mlx5_ib_flow_counters_set_data(struct ib_counters *ibcounters, + struct mlx5_ib_create_flow *ucmd); +u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num); +#endif /* _MLX5_IB_COUNTERS_H */ diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c index 146ba2966744..dceb0eb2bed1 100644 --- a/drivers/infiniband/hw/mlx5/cq.c +++ b/drivers/infiniband/hw/mlx5/cq.c @@ -36,6 +36,7 @@ #include <rdma/ib_cache.h> #include "mlx5_ib.h" #include "srq.h" +#include "qp.h" static void mlx5_ib_cq_comp(struct mlx5_core_cq *cq, struct mlx5_eqe *eqe) { @@ -120,13 +121,13 @@ static void handle_good_req(struct ib_wc *wc, struct mlx5_cqe64 *cqe, switch (be32_to_cpu(cqe->sop_drop_qpn) >> 24) { case MLX5_OPCODE_RDMA_WRITE_IMM: wc->wc_flags |= IB_WC_WITH_IMM; - /* fall through */ + fallthrough; case MLX5_OPCODE_RDMA_WRITE: wc->opcode = IB_WC_RDMA_WRITE; break; case MLX5_OPCODE_SEND_IMM: wc->wc_flags |= IB_WC_WITH_IMM; - /* fall through */ + fallthrough; case MLX5_OPCODE_SEND: case MLX5_OPCODE_SEND_INVAL: wc->opcode = IB_WC_SEND; @@ -201,7 +202,7 @@ static void handle_responder(struct ib_wc *wc, struct mlx5_cqe64 *cqe, case MLX5_CQE_RESP_WR_IMM: wc->opcode = IB_WC_RECV_RDMA_WITH_IMM; wc->wc_flags = IB_WC_WITH_IMM; - wc->ex.imm_data = cqe->imm_inval_pkey; + wc->ex.imm_data = cqe->immediate; break; case MLX5_CQE_RESP_SEND: wc->opcode = IB_WC_RECV; @@ -213,12 +214,12 @@ static void handle_responder(struct ib_wc *wc, struct mlx5_cqe64 *cqe, case MLX5_CQE_RESP_SEND_IMM: wc->opcode = IB_WC_RECV; wc->wc_flags = IB_WC_WITH_IMM; - wc->ex.imm_data = cqe->imm_inval_pkey; + wc->ex.imm_data = cqe->immediate; break; case MLX5_CQE_RESP_SEND_INV: wc->opcode = IB_WC_RECV; wc->wc_flags = IB_WC_WITH_INVALIDATE; - wc->ex.invalidate_rkey = be32_to_cpu(cqe->imm_inval_pkey); + wc->ex.invalidate_rkey = be32_to_cpu(cqe->inval_rkey); break; } wc->src_qp = be32_to_cpu(cqe->flags_rqpn) & 0xffffff; @@ -226,7 +227,7 @@ static void handle_responder(struct ib_wc *wc, struct mlx5_cqe64 *cqe, g = (be32_to_cpu(cqe->flags_rqpn) >> 28) & 3; wc->wc_flags |= g ? IB_WC_GRH : 0; if (unlikely(is_qp1(qp->ibqp.qp_type))) { - u16 pkey = be32_to_cpu(cqe->imm_inval_pkey) & 0xffff; + u16 pkey = be32_to_cpu(cqe->pkey) & 0xffff; ib_find_cached_pkey(&dev->ib_dev, qp->port, pkey, &wc->pkey_index); @@ -484,7 +485,7 @@ repoll: * because CQs will be locked while QPs are removed * from the table. */ - mqp = __mlx5_qp_lookup(dev->mdev, qpn); + mqp = radix_tree_lookup(&dev->qp_table.tree, qpn); *cur_qp = to_mibqp(mqp); } @@ -924,8 +925,8 @@ int mlx5_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_ib_cq *cq = to_mcq(ibcq); u32 out[MLX5_ST_SZ_DW(create_cq_out)]; - int uninitialized_var(index); - int uninitialized_var(inlen); + int index; + int inlen; u32 *cqb = NULL; void *cqc; int cqe_size; @@ -1245,7 +1246,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) __be64 *pas; int page_shift; int inlen; - int uninitialized_var(cqe_size); + int cqe_size; unsigned long flags; if (!MLX5_CAP_GEN(dev->mdev, cq_resize)) { diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c index 46e1ab771f10..9e3d8b826498 100644 --- a/drivers/infiniband/hw/mlx5/devx.c +++ b/drivers/infiniband/hw/mlx5/devx.c @@ -14,6 +14,8 @@ #include <linux/mlx5/driver.h> #include <linux/mlx5/fs.h> #include "mlx5_ib.h" +#include "devx.h" +#include "qp.h" #include <linux/xarray.h> #define UVERBS_MODULE_NAME mlx5_ib @@ -88,22 +90,6 @@ struct devx_async_event_file { u8 is_destroyed:1; }; -#define MLX5_MAX_DESTROY_INBOX_SIZE_DW MLX5_ST_SZ_DW(delete_fte_in) -struct devx_obj { - struct mlx5_ib_dev *ib_dev; - u64 obj_id; - u32 dinlen; /* destroy inbox length */ - u32 dinbox[MLX5_MAX_DESTROY_INBOX_SIZE_DW]; - u32 flags; - union { - struct mlx5_ib_devx_mr devx_mr; - struct mlx5_core_dct core_dct; - struct mlx5_core_cq core_cq; - u32 flow_counter_bulk_size; - }; - struct list_head event_sub; /* holds devx_event_subscription entries */ -}; - struct devx_umem { struct mlx5_core_dev *mdev; struct ib_umem *umem; @@ -170,48 +156,6 @@ void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid) mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); } -bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, int *dest_type) -{ - struct devx_obj *devx_obj = obj; - u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, opcode); - - switch (opcode) { - case MLX5_CMD_OP_DESTROY_TIR: - *dest_type = MLX5_FLOW_DESTINATION_TYPE_TIR; - *dest_id = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, - obj_id); - return true; - - case MLX5_CMD_OP_DESTROY_FLOW_TABLE: - *dest_type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE; - *dest_id = MLX5_GET(destroy_flow_table_in, devx_obj->dinbox, - table_id); - return true; - default: - return false; - } -} - -bool mlx5_ib_devx_is_flow_counter(void *obj, u32 offset, u32 *counter_id) -{ - struct devx_obj *devx_obj = obj; - u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, opcode); - - if (opcode == MLX5_CMD_OP_DEALLOC_FLOW_COUNTER) { - - if (offset && offset >= devx_obj->flow_counter_bulk_size) - return false; - - *counter_id = MLX5_GET(dealloc_flow_counter_in, - devx_obj->dinbox, - flow_counter_id); - *counter_id += offset; - return true; - } - - return false; -} - static bool is_legacy_unaffiliated_event_num(u16 event_num) { switch (event_num) { @@ -494,6 +438,10 @@ static u64 devx_get_obj_id(const void *in) obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_QP, MLX5_GET(rst2init_qp_in, in, qpn)); break; + case MLX5_CMD_OP_INIT2INIT_QP: + obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_QP, + MLX5_GET(init2init_qp_in, in, qpn)); + break; case MLX5_CMD_OP_INIT2RTR_QP: obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_QP, MLX5_GET(init2rtr_qp_in, in, qpn)); @@ -614,7 +562,7 @@ static bool devx_is_valid_obj_id(struct uverbs_attr_bundle *attrs, enum ib_qp_type qp_type = qp->ibqp.qp_type; if (qp_type == IB_QPT_RAW_PACKET || - (qp->flags & MLX5_IB_QP_UNDERLAY)) { + (qp->flags & IB_QP_CREATE_SOURCE_QPN)) { struct mlx5_ib_raw_packet_qp *raw_packet_qp = &qp->raw_packet_qp; struct mlx5_ib_rq *rq = &raw_packet_qp->rq; @@ -819,6 +767,7 @@ static bool devx_is_obj_modify_cmd(const void *in) case MLX5_CMD_OP_SET_L2_TABLE_ENTRY: case MLX5_CMD_OP_RST2INIT_QP: case MLX5_CMD_OP_INIT2RTR_QP: + case MLX5_CMD_OP_INIT2INIT_QP: case MLX5_CMD_OP_RTR2RTS_QP: case MLX5_CMD_OP_RTS2RTS_QP: case MLX5_CMD_OP_SQERR2RTS_QP: @@ -1356,7 +1305,7 @@ static int devx_obj_cleanup(struct ib_uobject *uobject, } if (obj->flags & DEVX_OBJ_FLAGS_DCT) - ret = mlx5_core_destroy_dct(obj->ib_dev->mdev, &obj->core_dct); + ret = mlx5_core_destroy_dct(obj->ib_dev, &obj->core_dct); else if (obj->flags & DEVX_OBJ_FLAGS_CQ) ret = mlx5_core_destroy_cq(obj->ib_dev->mdev, &obj->core_cq); else @@ -1450,9 +1399,8 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)( if (opcode == MLX5_CMD_OP_CREATE_DCT) { obj->flags |= DEVX_OBJ_FLAGS_DCT; - err = mlx5_core_create_dct(dev->mdev, &obj->core_dct, - cmd_in, cmd_in_len, - cmd_out, cmd_out_len); + err = mlx5_core_create_dct(dev, &obj->core_dct, cmd_in, + cmd_in_len, cmd_out, cmd_out_len); } else if (opcode == MLX5_CMD_OP_CREATE_CQ) { obj->flags |= DEVX_OBJ_FLAGS_CQ; obj->core_cq.comp = devx_cq_comp; @@ -1499,7 +1447,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)( obj_destroy: if (obj->flags & DEVX_OBJ_FLAGS_DCT) - mlx5_core_destroy_dct(obj->ib_dev->mdev, &obj->core_dct); + mlx5_core_destroy_dct(obj->ib_dev, &obj->core_dct); else if (obj->flags & DEVX_OBJ_FLAGS_CQ) mlx5_core_destroy_cq(obj->ib_dev->mdev, &obj->core_cq); else @@ -2217,14 +2165,12 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_UMEM_REG)( obj->mdev = dev->mdev; uobj->object = obj; devx_obj_build_destroy_cmd(cmd.in, cmd.out, obj->dinbox, &obj->dinlen, &obj_id); - err = uverbs_copy_to(attrs, MLX5_IB_ATTR_DEVX_UMEM_REG_OUT_ID, &obj_id, sizeof(obj_id)); - if (err) - goto err_umem_destroy; + uverbs_finalize_uobj_create(attrs, MLX5_IB_ATTR_DEVX_UMEM_REG_HANDLE); - return 0; + err = uverbs_copy_to(attrs, MLX5_IB_ATTR_DEVX_UMEM_REG_OUT_ID, &obj_id, + sizeof(obj_id)); + return err; -err_umem_destroy: - mlx5_cmd_exec(obj->mdev, obj->dinbox, obj->dinlen, cmd.out, sizeof(cmd.out)); err_umem_release: ib_umem_release(obj->umem); err_obj_free: @@ -2416,17 +2362,24 @@ static int devx_event_notifier(struct notifier_block *nb, return NOTIFY_OK; } -void mlx5_ib_devx_init_event_table(struct mlx5_ib_dev *dev) +int mlx5_ib_devx_init(struct mlx5_ib_dev *dev) { struct mlx5_devx_event_table *table = &dev->devx_event_table; + int uid; + + uid = mlx5_ib_devx_create(dev, false); + if (uid > 0) { + dev->devx_whitelist_uid = uid; + xa_init(&table->event_xa); + mutex_init(&table->event_xa_lock); + MLX5_NB_INIT(&table->devx_nb, devx_event_notifier, NOTIFY_ANY); + mlx5_eq_notifier_register(dev->mdev, &table->devx_nb); + } - xa_init(&table->event_xa); - mutex_init(&table->event_xa_lock); - MLX5_NB_INIT(&table->devx_nb, devx_event_notifier, NOTIFY_ANY); - mlx5_eq_notifier_register(dev->mdev, &table->devx_nb); + return 0; } -void mlx5_ib_devx_cleanup_event_table(struct mlx5_ib_dev *dev) +void mlx5_ib_devx_cleanup(struct mlx5_ib_dev *dev) { struct mlx5_devx_event_table *table = &dev->devx_event_table; struct devx_event_subscription *sub, *tmp; @@ -2434,17 +2387,21 @@ void mlx5_ib_devx_cleanup_event_table(struct mlx5_ib_dev *dev) void *entry; unsigned long id; - mlx5_eq_notifier_unregister(dev->mdev, &table->devx_nb); - mutex_lock(&dev->devx_event_table.event_xa_lock); - xa_for_each(&table->event_xa, id, entry) { - event = entry; - list_for_each_entry_safe(sub, tmp, &event->unaffiliated_list, - xa_list) - devx_cleanup_subscription(dev, sub); - kfree(entry); + if (dev->devx_whitelist_uid) { + mlx5_eq_notifier_unregister(dev->mdev, &table->devx_nb); + mutex_lock(&dev->devx_event_table.event_xa_lock); + xa_for_each(&table->event_xa, id, entry) { + event = entry; + list_for_each_entry_safe( + sub, tmp, &event->unaffiliated_list, xa_list) + devx_cleanup_subscription(dev, sub); + kfree(entry); + } + mutex_unlock(&dev->devx_event_table.event_xa_lock); + xa_destroy(&table->event_xa); + + mlx5_ib_devx_destroy(dev, dev->devx_whitelist_uid); } - mutex_unlock(&dev->devx_event_table.event_xa_lock); - xa_destroy(&table->event_xa); } static ssize_t devx_async_cmd_event_read(struct file *filp, char __user *buf, @@ -2533,7 +2490,7 @@ static ssize_t devx_async_event_read(struct file *filp, char __user *buf, { struct devx_async_event_file *ev_file = filp->private_data; struct devx_event_subscription *event_sub; - struct devx_async_event_data *uninitialized_var(event); + struct devx_async_event_data *event; int ret = 0; size_t eventsz; bool omit_data; diff --git a/drivers/infiniband/hw/mlx5/devx.h b/drivers/infiniband/hw/mlx5/devx.h new file mode 100644 index 000000000000..1f69866aed16 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/devx.h @@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2019-2020, Mellanox Technologies inc. All rights reserved. + */ + +#ifndef _MLX5_IB_DEVX_H +#define _MLX5_IB_DEVX_H + +#include "mlx5_ib.h" + +#define MLX5_MAX_DESTROY_INBOX_SIZE_DW MLX5_ST_SZ_DW(delete_fte_in) +struct devx_obj { + struct mlx5_ib_dev *ib_dev; + u64 obj_id; + u32 dinlen; /* destroy inbox length */ + u32 dinbox[MLX5_MAX_DESTROY_INBOX_SIZE_DW]; + u32 flags; + union { + struct mlx5_ib_devx_mr devx_mr; + struct mlx5_core_dct core_dct; + struct mlx5_core_cq core_cq; + u32 flow_counter_bulk_size; + }; + struct list_head event_sub; /* holds devx_event_subscription entries */ +}; +#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) +int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user); +void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid); +int mlx5_ib_devx_init(struct mlx5_ib_dev *dev); +void mlx5_ib_devx_cleanup(struct mlx5_ib_dev *dev); +#else +static inline int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user) +{ + return -EOPNOTSUPP; +} +static inline void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid) {} +static inline int mlx5_ib_devx_init(struct mlx5_ib_dev *dev) +{ + return 0; +} +static inline void mlx5_ib_devx_cleanup(struct mlx5_ib_dev *dev) +{ +} +#endif +#endif /* _MLX5_IB_DEVX_H */ diff --git a/drivers/infiniband/hw/mlx5/flow.c b/drivers/infiniband/hw/mlx5/flow.c deleted file mode 100644 index 862b7bf3e646..000000000000 --- a/drivers/infiniband/hw/mlx5/flow.c +++ /dev/null @@ -1,725 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB -/* - * Copyright (c) 2018, Mellanox Technologies inc. All rights reserved. - */ - -#include <rdma/ib_user_verbs.h> -#include <rdma/ib_verbs.h> -#include <rdma/uverbs_types.h> -#include <rdma/uverbs_ioctl.h> -#include <rdma/uverbs_std_types.h> -#include <rdma/mlx5_user_ioctl_cmds.h> -#include <rdma/mlx5_user_ioctl_verbs.h> -#include <rdma/ib_umem.h> -#include <linux/mlx5/driver.h> -#include <linux/mlx5/fs.h> -#include "mlx5_ib.h" - -#define UVERBS_MODULE_NAME mlx5_ib -#include <rdma/uverbs_named_ioctl.h> - -static int -mlx5_ib_ft_type_to_namespace(enum mlx5_ib_uapi_flow_table_type table_type, - enum mlx5_flow_namespace_type *namespace) -{ - switch (table_type) { - case MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX: - *namespace = MLX5_FLOW_NAMESPACE_BYPASS; - break; - case MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX: - *namespace = MLX5_FLOW_NAMESPACE_EGRESS; - break; - case MLX5_IB_UAPI_FLOW_TABLE_TYPE_FDB: - *namespace = MLX5_FLOW_NAMESPACE_FDB; - break; - case MLX5_IB_UAPI_FLOW_TABLE_TYPE_RDMA_RX: - *namespace = MLX5_FLOW_NAMESPACE_RDMA_RX; - break; - case MLX5_IB_UAPI_FLOW_TABLE_TYPE_RDMA_TX: - *namespace = MLX5_FLOW_NAMESPACE_RDMA_TX; - break; - default: - return -EINVAL; - } - - return 0; -} - -static const struct uverbs_attr_spec mlx5_ib_flow_type[] = { - [MLX5_IB_FLOW_TYPE_NORMAL] = { - .type = UVERBS_ATTR_TYPE_PTR_IN, - .u.ptr = { - .len = sizeof(u16), /* data is priority */ - .min_len = sizeof(u16), - } - }, - [MLX5_IB_FLOW_TYPE_SNIFFER] = { - .type = UVERBS_ATTR_TYPE_PTR_IN, - UVERBS_ATTR_NO_DATA(), - }, - [MLX5_IB_FLOW_TYPE_ALL_DEFAULT] = { - .type = UVERBS_ATTR_TYPE_PTR_IN, - UVERBS_ATTR_NO_DATA(), - }, - [MLX5_IB_FLOW_TYPE_MC_DEFAULT] = { - .type = UVERBS_ATTR_TYPE_PTR_IN, - UVERBS_ATTR_NO_DATA(), - }, -}; - -#define MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS 2 -static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( - struct uverbs_attr_bundle *attrs) -{ - struct mlx5_flow_context flow_context = {.flow_tag = MLX5_FS_DEFAULT_FLOW_TAG}; - struct mlx5_ib_flow_handler *flow_handler; - struct mlx5_ib_flow_matcher *fs_matcher; - struct ib_uobject **arr_flow_actions; - struct ib_uflow_resources *uflow_res; - struct mlx5_flow_act flow_act = {}; - void *devx_obj; - int dest_id, dest_type; - void *cmd_in; - int inlen; - bool dest_devx, dest_qp; - struct ib_qp *qp = NULL; - struct ib_uobject *uobj = - uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_CREATE_FLOW_HANDLE); - struct mlx5_ib_dev *dev = mlx5_udata_to_mdev(&attrs->driver_udata); - int len, ret, i; - u32 counter_id = 0; - u32 *offset_attr; - u32 offset = 0; - - if (!capable(CAP_NET_RAW)) - return -EPERM; - - dest_devx = - uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); - dest_qp = uverbs_attr_is_valid(attrs, - MLX5_IB_ATTR_CREATE_FLOW_DEST_QP); - - fs_matcher = uverbs_attr_get_obj(attrs, - MLX5_IB_ATTR_CREATE_FLOW_MATCHER); - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS && - ((dest_devx && dest_qp) || (!dest_devx && !dest_qp))) - return -EINVAL; - - /* Allow only DEVX object as dest when inserting to FDB */ - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && !dest_devx) - return -EINVAL; - - /* Allow only DEVX object or QP as dest when inserting to RDMA_RX */ - if ((fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) && - ((!dest_devx && !dest_qp) || (dest_devx && dest_qp))) - return -EINVAL; - - if (dest_devx) { - devx_obj = uverbs_attr_get_obj( - attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); - if (IS_ERR(devx_obj)) - return PTR_ERR(devx_obj); - - /* Verify that the given DEVX object is a flow - * steering destination. - */ - if (!mlx5_ib_devx_is_flow_dest(devx_obj, &dest_id, &dest_type)) - return -EINVAL; - /* Allow only flow table as dest when inserting to FDB or RDMA_RX */ - if ((fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB || - fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) && - dest_type != MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) - return -EINVAL; - } else if (dest_qp) { - struct mlx5_ib_qp *mqp; - - qp = uverbs_attr_get_obj(attrs, - MLX5_IB_ATTR_CREATE_FLOW_DEST_QP); - if (IS_ERR(qp)) - return PTR_ERR(qp); - - if (qp->qp_type != IB_QPT_RAW_PACKET) - return -EINVAL; - - mqp = to_mqp(qp); - if (mqp->flags & MLX5_IB_QP_RSS) - dest_id = mqp->rss_qp.tirn; - else - dest_id = mqp->raw_packet_qp.rq.tirn; - dest_type = MLX5_FLOW_DESTINATION_TYPE_TIR; - } else { - dest_type = MLX5_FLOW_DESTINATION_TYPE_PORT; - } - - len = uverbs_attr_get_uobjs_arr(attrs, - MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, &arr_flow_actions); - if (len) { - devx_obj = arr_flow_actions[0]->object; - - if (uverbs_attr_is_valid(attrs, - MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET)) { - - int num_offsets = uverbs_attr_ptr_get_array_size( - attrs, - MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET, - sizeof(u32)); - - if (num_offsets != 1) - return -EINVAL; - - offset_attr = uverbs_attr_get_alloced_ptr( - attrs, - MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET); - offset = *offset_attr; - } - - if (!mlx5_ib_devx_is_flow_counter(devx_obj, offset, - &counter_id)) - return -EINVAL; - - flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT; - } - - if (dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR && - fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS) - return -EINVAL; - - cmd_in = uverbs_attr_get_alloced_ptr( - attrs, MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE); - inlen = uverbs_attr_get_len(attrs, - MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE); - - uflow_res = flow_resources_alloc(MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS); - if (!uflow_res) - return -ENOMEM; - - len = uverbs_attr_get_uobjs_arr(attrs, - MLX5_IB_ATTR_CREATE_FLOW_ARR_FLOW_ACTIONS, &arr_flow_actions); - for (i = 0; i < len; i++) { - struct mlx5_ib_flow_action *maction = - to_mflow_act(arr_flow_actions[i]->object); - - ret = parse_flow_flow_action(maction, false, &flow_act); - if (ret) - goto err_out; - flow_resources_add(uflow_res, IB_FLOW_SPEC_ACTION_HANDLE, - arr_flow_actions[i]->object); - } - - ret = uverbs_copy_from(&flow_context.flow_tag, attrs, - MLX5_IB_ATTR_CREATE_FLOW_TAG); - if (!ret) { - if (flow_context.flow_tag >= BIT(24)) { - ret = -EINVAL; - goto err_out; - } - flow_context.flags |= FLOW_CONTEXT_HAS_TAG; - } - - flow_handler = mlx5_ib_raw_fs_rule_add(dev, fs_matcher, - &flow_context, - &flow_act, - counter_id, - cmd_in, inlen, - dest_id, dest_type); - if (IS_ERR(flow_handler)) { - ret = PTR_ERR(flow_handler); - goto err_out; - } - - ib_set_flow(uobj, &flow_handler->ibflow, qp, &dev->ib_dev, uflow_res); - - return 0; -err_out: - ib_uverbs_flow_resources_free(uflow_res); - return ret; -} - -static int flow_matcher_cleanup(struct ib_uobject *uobject, - enum rdma_remove_reason why, - struct uverbs_attr_bundle *attrs) -{ - struct mlx5_ib_flow_matcher *obj = uobject->object; - int ret; - - ret = ib_destroy_usecnt(&obj->usecnt, why, uobject); - if (ret) - return ret; - - kfree(obj); - return 0; -} - -static int mlx5_ib_matcher_ns(struct uverbs_attr_bundle *attrs, - struct mlx5_ib_flow_matcher *obj) -{ - enum mlx5_ib_uapi_flow_table_type ft_type = - MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX; - u32 flags; - int err; - - /* New users should use MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE and older - * users should switch to it. We leave this to not break userspace - */ - if (uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE) && - uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS)) - return -EINVAL; - - if (uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE)) { - err = uverbs_get_const(&ft_type, attrs, - MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE); - if (err) - return err; - - err = mlx5_ib_ft_type_to_namespace(ft_type, &obj->ns_type); - if (err) - return err; - - return 0; - } - - if (uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS)) { - err = uverbs_get_flags32(&flags, attrs, - MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS, - IB_FLOW_ATTR_FLAGS_EGRESS); - if (err) - return err; - - if (flags) { - mlx5_ib_ft_type_to_namespace( - MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX, - &obj->ns_type); - return 0; - } - } - - obj->ns_type = MLX5_FLOW_NAMESPACE_BYPASS; - - return 0; -} - -static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_MATCHER_CREATE)( - struct uverbs_attr_bundle *attrs) -{ - struct ib_uobject *uobj = uverbs_attr_get_uobject( - attrs, MLX5_IB_ATTR_FLOW_MATCHER_CREATE_HANDLE); - struct mlx5_ib_dev *dev = mlx5_udata_to_mdev(&attrs->driver_udata); - struct mlx5_ib_flow_matcher *obj; - int err; - - obj = kzalloc(sizeof(struct mlx5_ib_flow_matcher), GFP_KERNEL); - if (!obj) - return -ENOMEM; - - obj->mask_len = uverbs_attr_get_len( - attrs, MLX5_IB_ATTR_FLOW_MATCHER_MATCH_MASK); - err = uverbs_copy_from(&obj->matcher_mask, - attrs, - MLX5_IB_ATTR_FLOW_MATCHER_MATCH_MASK); - if (err) - goto end; - - obj->flow_type = uverbs_attr_get_enum_id( - attrs, MLX5_IB_ATTR_FLOW_MATCHER_FLOW_TYPE); - - if (obj->flow_type == MLX5_IB_FLOW_TYPE_NORMAL) { - err = uverbs_copy_from(&obj->priority, - attrs, - MLX5_IB_ATTR_FLOW_MATCHER_FLOW_TYPE); - if (err) - goto end; - } - - err = uverbs_copy_from(&obj->match_criteria_enable, - attrs, - MLX5_IB_ATTR_FLOW_MATCHER_MATCH_CRITERIA); - if (err) - goto end; - - err = mlx5_ib_matcher_ns(attrs, obj); - if (err) - goto end; - - uobj->object = obj; - obj->mdev = dev->mdev; - atomic_set(&obj->usecnt, 0); - return 0; - -end: - kfree(obj); - return err; -} - -void mlx5_ib_destroy_flow_action_raw(struct mlx5_ib_flow_action *maction) -{ - switch (maction->flow_action_raw.sub_type) { - case MLX5_IB_FLOW_ACTION_MODIFY_HEADER: - mlx5_modify_header_dealloc(maction->flow_action_raw.dev->mdev, - maction->flow_action_raw.modify_hdr); - break; - case MLX5_IB_FLOW_ACTION_PACKET_REFORMAT: - mlx5_packet_reformat_dealloc(maction->flow_action_raw.dev->mdev, - maction->flow_action_raw.pkt_reformat); - break; - case MLX5_IB_FLOW_ACTION_DECAP: - break; - default: - break; - } -} - -static struct ib_flow_action * -mlx5_ib_create_modify_header(struct mlx5_ib_dev *dev, - enum mlx5_ib_uapi_flow_table_type ft_type, - u8 num_actions, void *in) -{ - enum mlx5_flow_namespace_type namespace; - struct mlx5_ib_flow_action *maction; - int ret; - - ret = mlx5_ib_ft_type_to_namespace(ft_type, &namespace); - if (ret) - return ERR_PTR(-EINVAL); - - maction = kzalloc(sizeof(*maction), GFP_KERNEL); - if (!maction) - return ERR_PTR(-ENOMEM); - - maction->flow_action_raw.modify_hdr = - mlx5_modify_header_alloc(dev->mdev, namespace, num_actions, in); - - if (IS_ERR(maction->flow_action_raw.modify_hdr)) { - ret = PTR_ERR(maction->flow_action_raw.modify_hdr); - kfree(maction); - return ERR_PTR(ret); - } - maction->flow_action_raw.sub_type = - MLX5_IB_FLOW_ACTION_MODIFY_HEADER; - maction->flow_action_raw.dev = dev; - - return &maction->ib_action; -} - -static bool mlx5_ib_modify_header_supported(struct mlx5_ib_dev *dev) -{ - return MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, - max_modify_header_actions) || - MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, max_modify_header_actions); -} - -static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER)( - struct uverbs_attr_bundle *attrs) -{ - struct ib_uobject *uobj = uverbs_attr_get_uobject( - attrs, MLX5_IB_ATTR_CREATE_MODIFY_HEADER_HANDLE); - struct mlx5_ib_dev *mdev = mlx5_udata_to_mdev(&attrs->driver_udata); - enum mlx5_ib_uapi_flow_table_type ft_type; - struct ib_flow_action *action; - int num_actions; - void *in; - int ret; - - if (!mlx5_ib_modify_header_supported(mdev)) - return -EOPNOTSUPP; - - in = uverbs_attr_get_alloced_ptr(attrs, - MLX5_IB_ATTR_CREATE_MODIFY_HEADER_ACTIONS_PRM); - - num_actions = uverbs_attr_ptr_get_array_size( - attrs, MLX5_IB_ATTR_CREATE_MODIFY_HEADER_ACTIONS_PRM, - MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto)); - if (num_actions < 0) - return num_actions; - - ret = uverbs_get_const(&ft_type, attrs, - MLX5_IB_ATTR_CREATE_MODIFY_HEADER_FT_TYPE); - if (ret) - return ret; - action = mlx5_ib_create_modify_header(mdev, ft_type, num_actions, in); - if (IS_ERR(action)) - return PTR_ERR(action); - - uverbs_flow_action_fill_action(action, uobj, &mdev->ib_dev, - IB_FLOW_ACTION_UNSPECIFIED); - - return 0; -} - -static bool mlx5_ib_flow_action_packet_reformat_valid(struct mlx5_ib_dev *ibdev, - u8 packet_reformat_type, - u8 ft_type) -{ - switch (packet_reformat_type) { - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L2_TUNNEL: - if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX) - return MLX5_CAP_FLOWTABLE(ibdev->mdev, - encap_general_header); - break; - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L3_TUNNEL: - if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX) - return MLX5_CAP_FLOWTABLE_NIC_TX(ibdev->mdev, - reformat_l2_to_l3_tunnel); - break; - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L3_TUNNEL_TO_L2: - if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX) - return MLX5_CAP_FLOWTABLE_NIC_RX(ibdev->mdev, - reformat_l3_tunnel_to_l2); - break; - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TUNNEL_TO_L2: - if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX) - return MLX5_CAP_FLOWTABLE_NIC_RX(ibdev->mdev, decap); - break; - default: - break; - } - - return false; -} - -static int mlx5_ib_dv_to_prm_packet_reforamt_type(u8 dv_prt, u8 *prm_prt) -{ - switch (dv_prt) { - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L2_TUNNEL: - *prm_prt = MLX5_REFORMAT_TYPE_L2_TO_L2_TUNNEL; - break; - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L3_TUNNEL_TO_L2: - *prm_prt = MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2; - break; - case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L3_TUNNEL: - *prm_prt = MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL; - break; - default: - return -EINVAL; - } - - return 0; -} - -static int mlx5_ib_flow_action_create_packet_reformat_ctx( - struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_action *maction, - u8 ft_type, u8 dv_prt, - void *in, size_t len) -{ - enum mlx5_flow_namespace_type namespace; - u8 prm_prt; - int ret; - - ret = mlx5_ib_ft_type_to_namespace(ft_type, &namespace); - if (ret) - return ret; - - ret = mlx5_ib_dv_to_prm_packet_reforamt_type(dv_prt, &prm_prt); - if (ret) - return ret; - - maction->flow_action_raw.pkt_reformat = - mlx5_packet_reformat_alloc(dev->mdev, prm_prt, len, - in, namespace); - if (IS_ERR(maction->flow_action_raw.pkt_reformat)) { - ret = PTR_ERR(maction->flow_action_raw.pkt_reformat); - return ret; - } - - maction->flow_action_raw.sub_type = - MLX5_IB_FLOW_ACTION_PACKET_REFORMAT; - maction->flow_action_raw.dev = dev; - - return 0; -} - -static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT)( - struct uverbs_attr_bundle *attrs) -{ - struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, - MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_HANDLE); - struct mlx5_ib_dev *mdev = mlx5_udata_to_mdev(&attrs->driver_udata); - enum mlx5_ib_uapi_flow_action_packet_reformat_type dv_prt; - enum mlx5_ib_uapi_flow_table_type ft_type; - struct mlx5_ib_flow_action *maction; - int ret; - - ret = uverbs_get_const(&ft_type, attrs, - MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_FT_TYPE); - if (ret) - return ret; - - ret = uverbs_get_const(&dv_prt, attrs, - MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_TYPE); - if (ret) - return ret; - - if (!mlx5_ib_flow_action_packet_reformat_valid(mdev, dv_prt, ft_type)) - return -EOPNOTSUPP; - - maction = kzalloc(sizeof(*maction), GFP_KERNEL); - if (!maction) - return -ENOMEM; - - if (dv_prt == - MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TUNNEL_TO_L2) { - maction->flow_action_raw.sub_type = - MLX5_IB_FLOW_ACTION_DECAP; - maction->flow_action_raw.dev = mdev; - } else { - void *in; - int len; - - in = uverbs_attr_get_alloced_ptr(attrs, - MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_DATA_BUF); - if (IS_ERR(in)) { - ret = PTR_ERR(in); - goto free_maction; - } - - len = uverbs_attr_get_len(attrs, - MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_DATA_BUF); - - ret = mlx5_ib_flow_action_create_packet_reformat_ctx(mdev, - maction, ft_type, dv_prt, in, len); - if (ret) - goto free_maction; - } - - uverbs_flow_action_fill_action(&maction->ib_action, uobj, &mdev->ib_dev, - IB_FLOW_ACTION_UNSPECIFIED); - return 0; - -free_maction: - kfree(maction); - return ret; -} - -DECLARE_UVERBS_NAMED_METHOD( - MLX5_IB_METHOD_CREATE_FLOW, - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_HANDLE, - UVERBS_OBJECT_FLOW, - UVERBS_ACCESS_NEW, - UA_MANDATORY), - UVERBS_ATTR_PTR_IN( - MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE, - UVERBS_ATTR_SIZE(1, sizeof(struct mlx5_ib_match_params)), - UA_MANDATORY, - UA_ALLOC_AND_COPY), - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_MATCHER, - MLX5_IB_OBJECT_FLOW_MATCHER, - UVERBS_ACCESS_READ, - UA_MANDATORY), - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_DEST_QP, - UVERBS_OBJECT_QP, - UVERBS_ACCESS_READ), - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX, - MLX5_IB_OBJECT_DEVX_OBJ, - UVERBS_ACCESS_READ), - UVERBS_ATTR_IDRS_ARR(MLX5_IB_ATTR_CREATE_FLOW_ARR_FLOW_ACTIONS, - UVERBS_OBJECT_FLOW_ACTION, - UVERBS_ACCESS_READ, 1, - MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS, - UA_OPTIONAL), - UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_TAG, - UVERBS_ATTR_TYPE(u32), - UA_OPTIONAL), - UVERBS_ATTR_IDRS_ARR(MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, - MLX5_IB_OBJECT_DEVX_OBJ, - UVERBS_ACCESS_READ, 1, 1, - UA_OPTIONAL), - UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET, - UVERBS_ATTR_MIN_SIZE(sizeof(u32)), - UA_OPTIONAL, - UA_ALLOC_AND_COPY)); - -DECLARE_UVERBS_NAMED_METHOD_DESTROY( - MLX5_IB_METHOD_DESTROY_FLOW, - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_HANDLE, - UVERBS_OBJECT_FLOW, - UVERBS_ACCESS_DESTROY, - UA_MANDATORY)); - -ADD_UVERBS_METHODS(mlx5_ib_fs, - UVERBS_OBJECT_FLOW, - &UVERBS_METHOD(MLX5_IB_METHOD_CREATE_FLOW), - &UVERBS_METHOD(MLX5_IB_METHOD_DESTROY_FLOW)); - -DECLARE_UVERBS_NAMED_METHOD( - MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER, - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_MODIFY_HEADER_HANDLE, - UVERBS_OBJECT_FLOW_ACTION, - UVERBS_ACCESS_NEW, - UA_MANDATORY), - UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_MODIFY_HEADER_ACTIONS_PRM, - UVERBS_ATTR_MIN_SIZE(MLX5_UN_SZ_BYTES( - set_action_in_add_action_in_auto)), - UA_MANDATORY, - UA_ALLOC_AND_COPY), - UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_CREATE_MODIFY_HEADER_FT_TYPE, - enum mlx5_ib_uapi_flow_table_type, - UA_MANDATORY)); - -DECLARE_UVERBS_NAMED_METHOD( - MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT, - UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_HANDLE, - UVERBS_OBJECT_FLOW_ACTION, - UVERBS_ACCESS_NEW, - UA_MANDATORY), - UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_DATA_BUF, - UVERBS_ATTR_MIN_SIZE(1), - UA_ALLOC_AND_COPY, - UA_OPTIONAL), - UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_TYPE, - enum mlx5_ib_uapi_flow_action_packet_reformat_type, - UA_MANDATORY), - UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_FT_TYPE, - enum mlx5_ib_uapi_flow_table_type, - UA_MANDATORY)); - -ADD_UVERBS_METHODS( - mlx5_ib_flow_actions, - UVERBS_OBJECT_FLOW_ACTION, - &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER), - &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT)); - -DECLARE_UVERBS_NAMED_METHOD( - MLX5_IB_METHOD_FLOW_MATCHER_CREATE, - UVERBS_ATTR_IDR(MLX5_IB_ATTR_FLOW_MATCHER_CREATE_HANDLE, - MLX5_IB_OBJECT_FLOW_MATCHER, - UVERBS_ACCESS_NEW, - UA_MANDATORY), - UVERBS_ATTR_PTR_IN( - MLX5_IB_ATTR_FLOW_MATCHER_MATCH_MASK, - UVERBS_ATTR_SIZE(1, sizeof(struct mlx5_ib_match_params)), - UA_MANDATORY), - UVERBS_ATTR_ENUM_IN(MLX5_IB_ATTR_FLOW_MATCHER_FLOW_TYPE, - mlx5_ib_flow_type, - UA_MANDATORY), - UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_FLOW_MATCHER_MATCH_CRITERIA, - UVERBS_ATTR_TYPE(u8), - UA_MANDATORY), - UVERBS_ATTR_FLAGS_IN(MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS, - enum ib_flow_flags, - UA_OPTIONAL), - UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE, - enum mlx5_ib_uapi_flow_table_type, - UA_OPTIONAL)); - -DECLARE_UVERBS_NAMED_METHOD_DESTROY( - MLX5_IB_METHOD_FLOW_MATCHER_DESTROY, - UVERBS_ATTR_IDR(MLX5_IB_ATTR_FLOW_MATCHER_DESTROY_HANDLE, - MLX5_IB_OBJECT_FLOW_MATCHER, - UVERBS_ACCESS_DESTROY, - UA_MANDATORY)); - -DECLARE_UVERBS_NAMED_OBJECT(MLX5_IB_OBJECT_FLOW_MATCHER, - UVERBS_TYPE_ALLOC_IDR(flow_matcher_cleanup), - &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_MATCHER_CREATE), - &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_MATCHER_DESTROY)); - -const struct uapi_definition mlx5_ib_flow_defs[] = { - UAPI_DEF_CHAIN_OBJ_TREE_NAMED( - MLX5_IB_OBJECT_FLOW_MATCHER), - UAPI_DEF_CHAIN_OBJ_TREE( - UVERBS_OBJECT_FLOW, - &mlx5_ib_fs), - UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION, - &mlx5_ib_flow_actions), - {}, -}; diff --git a/drivers/infiniband/hw/mlx5/fs.c b/drivers/infiniband/hw/mlx5/fs.c new file mode 100644 index 000000000000..e9cfb9a2ef41 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/fs.c @@ -0,0 +1,2516 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2018, Mellanox Technologies inc. All rights reserved. + */ + +#include <rdma/ib_user_verbs.h> +#include <rdma/ib_verbs.h> +#include <rdma/uverbs_types.h> +#include <rdma/uverbs_ioctl.h> +#include <rdma/uverbs_std_types.h> +#include <rdma/mlx5_user_ioctl_cmds.h> +#include <rdma/mlx5_user_ioctl_verbs.h> +#include <rdma/ib_umem.h> +#include <linux/mlx5/driver.h> +#include <linux/mlx5/fs.h> +#include <linux/mlx5/fs_helpers.h> +#include <linux/mlx5/accel.h> +#include <linux/mlx5/eswitch.h> +#include "mlx5_ib.h" +#include "counters.h" +#include "devx.h" +#include "fs.h" + +#define UVERBS_MODULE_NAME mlx5_ib +#include <rdma/uverbs_named_ioctl.h> + +enum { + MATCH_CRITERIA_ENABLE_OUTER_BIT, + MATCH_CRITERIA_ENABLE_MISC_BIT, + MATCH_CRITERIA_ENABLE_INNER_BIT, + MATCH_CRITERIA_ENABLE_MISC2_BIT +}; + +#define HEADER_IS_ZERO(match_criteria, headers) \ + !(memchr_inv(MLX5_ADDR_OF(fte_match_param, match_criteria, headers), \ + 0, MLX5_FLD_SZ_BYTES(fte_match_param, headers))) \ + +static u8 get_match_criteria_enable(u32 *match_criteria) +{ + u8 match_criteria_enable; + + match_criteria_enable = + (!HEADER_IS_ZERO(match_criteria, outer_headers)) << + MATCH_CRITERIA_ENABLE_OUTER_BIT; + match_criteria_enable |= + (!HEADER_IS_ZERO(match_criteria, misc_parameters)) << + MATCH_CRITERIA_ENABLE_MISC_BIT; + match_criteria_enable |= + (!HEADER_IS_ZERO(match_criteria, inner_headers)) << + MATCH_CRITERIA_ENABLE_INNER_BIT; + match_criteria_enable |= + (!HEADER_IS_ZERO(match_criteria, misc_parameters_2)) << + MATCH_CRITERIA_ENABLE_MISC2_BIT; + + return match_criteria_enable; +} + +static int set_proto(void *outer_c, void *outer_v, u8 mask, u8 val) +{ + u8 entry_mask; + u8 entry_val; + int err = 0; + + if (!mask) + goto out; + + entry_mask = MLX5_GET(fte_match_set_lyr_2_4, outer_c, + ip_protocol); + entry_val = MLX5_GET(fte_match_set_lyr_2_4, outer_v, + ip_protocol); + if (!entry_mask) { + MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_protocol, mask); + MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_protocol, val); + goto out; + } + /* Don't override existing ip protocol */ + if (mask != entry_mask || val != entry_val) + err = -EINVAL; +out: + return err; +} + +static void set_flow_label(void *misc_c, void *misc_v, u32 mask, u32 val, + bool inner) +{ + if (inner) { + MLX5_SET(fte_match_set_misc, + misc_c, inner_ipv6_flow_label, mask); + MLX5_SET(fte_match_set_misc, + misc_v, inner_ipv6_flow_label, val); + } else { + MLX5_SET(fte_match_set_misc, + misc_c, outer_ipv6_flow_label, mask); + MLX5_SET(fte_match_set_misc, + misc_v, outer_ipv6_flow_label, val); + } +} + +static void set_tos(void *outer_c, void *outer_v, u8 mask, u8 val) +{ + MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_ecn, mask); + MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_ecn, val); + MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_dscp, mask >> 2); + MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_dscp, val >> 2); +} + +static int check_mpls_supp_fields(u32 field_support, const __be32 *set_mask) +{ + if (MLX5_GET(fte_match_mpls, set_mask, mpls_label) && + !(field_support & MLX5_FIELD_SUPPORT_MPLS_LABEL)) + return -EOPNOTSUPP; + + if (MLX5_GET(fte_match_mpls, set_mask, mpls_exp) && + !(field_support & MLX5_FIELD_SUPPORT_MPLS_EXP)) + return -EOPNOTSUPP; + + if (MLX5_GET(fte_match_mpls, set_mask, mpls_s_bos) && + !(field_support & MLX5_FIELD_SUPPORT_MPLS_S_BOS)) + return -EOPNOTSUPP; + + if (MLX5_GET(fte_match_mpls, set_mask, mpls_ttl) && + !(field_support & MLX5_FIELD_SUPPORT_MPLS_TTL)) + return -EOPNOTSUPP; + + return 0; +} + +#define LAST_ETH_FIELD vlan_tag +#define LAST_IB_FIELD sl +#define LAST_IPV4_FIELD tos +#define LAST_IPV6_FIELD traffic_class +#define LAST_TCP_UDP_FIELD src_port +#define LAST_TUNNEL_FIELD tunnel_id +#define LAST_FLOW_TAG_FIELD tag_id +#define LAST_DROP_FIELD size +#define LAST_COUNTERS_FIELD counters + +/* Field is the last supported field */ +#define FIELDS_NOT_SUPPORTED(filter, field)\ + memchr_inv((void *)&filter.field +\ + sizeof(filter.field), 0,\ + sizeof(filter) -\ + offsetof(typeof(filter), field) -\ + sizeof(filter.field)) + +int parse_flow_flow_action(struct mlx5_ib_flow_action *maction, + bool is_egress, + struct mlx5_flow_act *action) +{ + + switch (maction->ib_action.type) { + case IB_FLOW_ACTION_ESP: + if (action->action & (MLX5_FLOW_CONTEXT_ACTION_ENCRYPT | + MLX5_FLOW_CONTEXT_ACTION_DECRYPT)) + return -EINVAL; + /* Currently only AES_GCM keymat is supported by the driver */ + action->esp_id = (uintptr_t)maction->esp_aes_gcm.ctx; + action->action |= is_egress ? + MLX5_FLOW_CONTEXT_ACTION_ENCRYPT : + MLX5_FLOW_CONTEXT_ACTION_DECRYPT; + return 0; + case IB_FLOW_ACTION_UNSPECIFIED: + if (maction->flow_action_raw.sub_type == + MLX5_IB_FLOW_ACTION_MODIFY_HEADER) { + if (action->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) + return -EINVAL; + action->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR; + action->modify_hdr = + maction->flow_action_raw.modify_hdr; + return 0; + } + if (maction->flow_action_raw.sub_type == + MLX5_IB_FLOW_ACTION_DECAP) { + if (action->action & MLX5_FLOW_CONTEXT_ACTION_DECAP) + return -EINVAL; + action->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP; + return 0; + } + if (maction->flow_action_raw.sub_type == + MLX5_IB_FLOW_ACTION_PACKET_REFORMAT) { + if (action->action & + MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) + return -EINVAL; + action->action |= + MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT; + action->pkt_reformat = + maction->flow_action_raw.pkt_reformat; + return 0; + } + fallthrough; + default: + return -EOPNOTSUPP; + } +} + +static int parse_flow_attr(struct mlx5_core_dev *mdev, + struct mlx5_flow_spec *spec, + const union ib_flow_spec *ib_spec, + const struct ib_flow_attr *flow_attr, + struct mlx5_flow_act *action, u32 prev_type) +{ + struct mlx5_flow_context *flow_context = &spec->flow_context; + u32 *match_c = spec->match_criteria; + u32 *match_v = spec->match_value; + void *misc_params_c = MLX5_ADDR_OF(fte_match_param, match_c, + misc_parameters); + void *misc_params_v = MLX5_ADDR_OF(fte_match_param, match_v, + misc_parameters); + void *misc_params2_c = MLX5_ADDR_OF(fte_match_param, match_c, + misc_parameters_2); + void *misc_params2_v = MLX5_ADDR_OF(fte_match_param, match_v, + misc_parameters_2); + void *headers_c; + void *headers_v; + int match_ipv; + int ret; + + if (ib_spec->type & IB_FLOW_SPEC_INNER) { + headers_c = MLX5_ADDR_OF(fte_match_param, match_c, + inner_headers); + headers_v = MLX5_ADDR_OF(fte_match_param, match_v, + inner_headers); + match_ipv = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.inner_ip_version); + } else { + headers_c = MLX5_ADDR_OF(fte_match_param, match_c, + outer_headers); + headers_v = MLX5_ADDR_OF(fte_match_param, match_v, + outer_headers); + match_ipv = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.outer_ip_version); + } + + switch (ib_spec->type & ~IB_FLOW_SPEC_INNER) { + case IB_FLOW_SPEC_ETH: + if (FIELDS_NOT_SUPPORTED(ib_spec->eth.mask, LAST_ETH_FIELD)) + return -EOPNOTSUPP; + + ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, + dmac_47_16), + ib_spec->eth.mask.dst_mac); + ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, + dmac_47_16), + ib_spec->eth.val.dst_mac); + + ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, + smac_47_16), + ib_spec->eth.mask.src_mac); + ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, + smac_47_16), + ib_spec->eth.val.src_mac); + + if (ib_spec->eth.mask.vlan_tag) { + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + cvlan_tag, 1); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + cvlan_tag, 1); + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + first_vid, ntohs(ib_spec->eth.mask.vlan_tag)); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + first_vid, ntohs(ib_spec->eth.val.vlan_tag)); + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + first_cfi, + ntohs(ib_spec->eth.mask.vlan_tag) >> 12); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + first_cfi, + ntohs(ib_spec->eth.val.vlan_tag) >> 12); + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + first_prio, + ntohs(ib_spec->eth.mask.vlan_tag) >> 13); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + first_prio, + ntohs(ib_spec->eth.val.vlan_tag) >> 13); + } + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + ethertype, ntohs(ib_spec->eth.mask.ether_type)); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + ethertype, ntohs(ib_spec->eth.val.ether_type)); + break; + case IB_FLOW_SPEC_IPV4: + if (FIELDS_NOT_SUPPORTED(ib_spec->ipv4.mask, LAST_IPV4_FIELD)) + return -EOPNOTSUPP; + + if (match_ipv) { + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + ip_version, 0xf); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + ip_version, MLX5_FS_IPV4_VERSION); + } else { + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + ethertype, 0xffff); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + ethertype, ETH_P_IP); + } + + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, + src_ipv4_src_ipv6.ipv4_layout.ipv4), + &ib_spec->ipv4.mask.src_ip, + sizeof(ib_spec->ipv4.mask.src_ip)); + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, + src_ipv4_src_ipv6.ipv4_layout.ipv4), + &ib_spec->ipv4.val.src_ip, + sizeof(ib_spec->ipv4.val.src_ip)); + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, + dst_ipv4_dst_ipv6.ipv4_layout.ipv4), + &ib_spec->ipv4.mask.dst_ip, + sizeof(ib_spec->ipv4.mask.dst_ip)); + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, + dst_ipv4_dst_ipv6.ipv4_layout.ipv4), + &ib_spec->ipv4.val.dst_ip, + sizeof(ib_spec->ipv4.val.dst_ip)); + + set_tos(headers_c, headers_v, + ib_spec->ipv4.mask.tos, ib_spec->ipv4.val.tos); + + if (set_proto(headers_c, headers_v, + ib_spec->ipv4.mask.proto, + ib_spec->ipv4.val.proto)) + return -EINVAL; + break; + case IB_FLOW_SPEC_IPV6: + if (FIELDS_NOT_SUPPORTED(ib_spec->ipv6.mask, LAST_IPV6_FIELD)) + return -EOPNOTSUPP; + + if (match_ipv) { + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + ip_version, 0xf); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + ip_version, MLX5_FS_IPV6_VERSION); + } else { + MLX5_SET(fte_match_set_lyr_2_4, headers_c, + ethertype, 0xffff); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, + ethertype, ETH_P_IPV6); + } + + memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, 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, 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, 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, headers_v, + dst_ipv4_dst_ipv6.ipv6_layout.ipv6), + &ib_spec->ipv6.val.dst_ip, + sizeof(ib_spec->ipv6.val.dst_ip)); + + set_tos(headers_c, headers_v, + ib_spec->ipv6.mask.traffic_class, + ib_spec->ipv6.val.traffic_class); + + if (set_proto(headers_c, headers_v, + ib_spec->ipv6.mask.next_hdr, + ib_spec->ipv6.val.next_hdr)) + return -EINVAL; + + set_flow_label(misc_params_c, misc_params_v, + ntohl(ib_spec->ipv6.mask.flow_label), + ntohl(ib_spec->ipv6.val.flow_label), + ib_spec->type & IB_FLOW_SPEC_INNER); + break; + case IB_FLOW_SPEC_ESP: + if (ib_spec->esp.mask.seq) + return -EOPNOTSUPP; + + MLX5_SET(fte_match_set_misc, misc_params_c, outer_esp_spi, + ntohl(ib_spec->esp.mask.spi)); + MLX5_SET(fte_match_set_misc, misc_params_v, outer_esp_spi, + ntohl(ib_spec->esp.val.spi)); + break; + case IB_FLOW_SPEC_TCP: + if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask, + LAST_TCP_UDP_FIELD)) + return -EOPNOTSUPP; + + if (set_proto(headers_c, headers_v, 0xff, IPPROTO_TCP)) + return -EINVAL; + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_sport, + ntohs(ib_spec->tcp_udp.mask.src_port)); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, tcp_sport, + ntohs(ib_spec->tcp_udp.val.src_port)); + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_dport, + ntohs(ib_spec->tcp_udp.mask.dst_port)); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, tcp_dport, + ntohs(ib_spec->tcp_udp.val.dst_port)); + break; + case IB_FLOW_SPEC_UDP: + if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask, + LAST_TCP_UDP_FIELD)) + return -EOPNOTSUPP; + + if (set_proto(headers_c, headers_v, 0xff, IPPROTO_UDP)) + return -EINVAL; + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_sport, + ntohs(ib_spec->tcp_udp.mask.src_port)); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, udp_sport, + ntohs(ib_spec->tcp_udp.val.src_port)); + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_dport, + ntohs(ib_spec->tcp_udp.mask.dst_port)); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, udp_dport, + ntohs(ib_spec->tcp_udp.val.dst_port)); + break; + case IB_FLOW_SPEC_GRE: + if (ib_spec->gre.mask.c_ks_res0_ver) + return -EOPNOTSUPP; + + if (set_proto(headers_c, headers_v, 0xff, IPPROTO_GRE)) + return -EINVAL; + + MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol, + 0xff); + MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol, + IPPROTO_GRE); + + MLX5_SET(fte_match_set_misc, misc_params_c, gre_protocol, + ntohs(ib_spec->gre.mask.protocol)); + MLX5_SET(fte_match_set_misc, misc_params_v, gre_protocol, + ntohs(ib_spec->gre.val.protocol)); + + memcpy(MLX5_ADDR_OF(fte_match_set_misc, misc_params_c, + gre_key.nvgre.hi), + &ib_spec->gre.mask.key, + sizeof(ib_spec->gre.mask.key)); + memcpy(MLX5_ADDR_OF(fte_match_set_misc, misc_params_v, + gre_key.nvgre.hi), + &ib_spec->gre.val.key, + sizeof(ib_spec->gre.val.key)); + break; + case IB_FLOW_SPEC_MPLS: + switch (prev_type) { + case IB_FLOW_SPEC_UDP: + if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.outer_first_mpls_over_udp), + &ib_spec->mpls.mask.tag)) + return -EOPNOTSUPP; + + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, + outer_first_mpls_over_udp), + &ib_spec->mpls.val.tag, + sizeof(ib_spec->mpls.val.tag)); + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, + outer_first_mpls_over_udp), + &ib_spec->mpls.mask.tag, + sizeof(ib_spec->mpls.mask.tag)); + break; + case IB_FLOW_SPEC_GRE: + if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.outer_first_mpls_over_gre), + &ib_spec->mpls.mask.tag)) + return -EOPNOTSUPP; + + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, + outer_first_mpls_over_gre), + &ib_spec->mpls.val.tag, + sizeof(ib_spec->mpls.val.tag)); + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, + outer_first_mpls_over_gre), + &ib_spec->mpls.mask.tag, + sizeof(ib_spec->mpls.mask.tag)); + break; + default: + if (ib_spec->type & IB_FLOW_SPEC_INNER) { + if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.inner_first_mpls), + &ib_spec->mpls.mask.tag)) + return -EOPNOTSUPP; + + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, + inner_first_mpls), + &ib_spec->mpls.val.tag, + sizeof(ib_spec->mpls.val.tag)); + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, + inner_first_mpls), + &ib_spec->mpls.mask.tag, + sizeof(ib_spec->mpls.mask.tag)); + } else { + if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.outer_first_mpls), + &ib_spec->mpls.mask.tag)) + return -EOPNOTSUPP; + + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, + outer_first_mpls), + &ib_spec->mpls.val.tag, + sizeof(ib_spec->mpls.val.tag)); + memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, + outer_first_mpls), + &ib_spec->mpls.mask.tag, + sizeof(ib_spec->mpls.mask.tag)); + } + } + break; + case IB_FLOW_SPEC_VXLAN_TUNNEL: + if (FIELDS_NOT_SUPPORTED(ib_spec->tunnel.mask, + LAST_TUNNEL_FIELD)) + return -EOPNOTSUPP; + + MLX5_SET(fte_match_set_misc, misc_params_c, vxlan_vni, + ntohl(ib_spec->tunnel.mask.tunnel_id)); + MLX5_SET(fte_match_set_misc, misc_params_v, vxlan_vni, + ntohl(ib_spec->tunnel.val.tunnel_id)); + break; + case IB_FLOW_SPEC_ACTION_TAG: + if (FIELDS_NOT_SUPPORTED(ib_spec->flow_tag, + LAST_FLOW_TAG_FIELD)) + return -EOPNOTSUPP; + if (ib_spec->flow_tag.tag_id >= BIT(24)) + return -EINVAL; + + flow_context->flow_tag = ib_spec->flow_tag.tag_id; + flow_context->flags |= FLOW_CONTEXT_HAS_TAG; + break; + case IB_FLOW_SPEC_ACTION_DROP: + if (FIELDS_NOT_SUPPORTED(ib_spec->drop, + LAST_DROP_FIELD)) + return -EOPNOTSUPP; + action->action |= MLX5_FLOW_CONTEXT_ACTION_DROP; + break; + case IB_FLOW_SPEC_ACTION_HANDLE: + ret = parse_flow_flow_action(to_mflow_act(ib_spec->action.act), + flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS, action); + if (ret) + return ret; + break; + case IB_FLOW_SPEC_ACTION_COUNT: + if (FIELDS_NOT_SUPPORTED(ib_spec->flow_count, + LAST_COUNTERS_FIELD)) + return -EOPNOTSUPP; + + /* for now support only one counters spec per flow */ + if (action->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) + return -EINVAL; + + action->counters = ib_spec->flow_count.counters; + action->action |= MLX5_FLOW_CONTEXT_ACTION_COUNT; + break; + default: + return -EINVAL; + } + + return 0; +} + +/* If a flow could catch both multicast and unicast packets, + * it won't fall into the multicast flow steering table and this rule + * could steal other multicast packets. + */ +static bool flow_is_multicast_only(const struct ib_flow_attr *ib_attr) +{ + union ib_flow_spec *flow_spec; + + if (ib_attr->type != IB_FLOW_ATTR_NORMAL || + ib_attr->num_of_specs < 1) + return false; + + flow_spec = (union ib_flow_spec *)(ib_attr + 1); + if (flow_spec->type == IB_FLOW_SPEC_IPV4) { + struct ib_flow_spec_ipv4 *ipv4_spec; + + ipv4_spec = (struct ib_flow_spec_ipv4 *)flow_spec; + if (ipv4_is_multicast(ipv4_spec->val.dst_ip)) + return true; + + return false; + } + + if (flow_spec->type == IB_FLOW_SPEC_ETH) { + struct ib_flow_spec_eth *eth_spec; + + eth_spec = (struct ib_flow_spec_eth *)flow_spec; + return is_multicast_ether_addr(eth_spec->mask.dst_mac) && + is_multicast_ether_addr(eth_spec->val.dst_mac); + } + + return false; +} + +enum valid_spec { + VALID_SPEC_INVALID, + VALID_SPEC_VALID, + VALID_SPEC_NA, +}; + +static enum valid_spec +is_valid_esp_aes_gcm(struct mlx5_core_dev *mdev, + const struct mlx5_flow_spec *spec, + const struct mlx5_flow_act *flow_act, + bool egress) +{ + const u32 *match_c = spec->match_criteria; + bool is_crypto = + (flow_act->action & (MLX5_FLOW_CONTEXT_ACTION_ENCRYPT | + MLX5_FLOW_CONTEXT_ACTION_DECRYPT)); + bool is_ipsec = mlx5_fs_is_ipsec_flow(match_c); + bool is_drop = flow_act->action & MLX5_FLOW_CONTEXT_ACTION_DROP; + + /* + * Currently only crypto is supported in egress, when regular egress + * rules would be supported, always return VALID_SPEC_NA. + */ + if (!is_crypto) + return VALID_SPEC_NA; + + return is_crypto && is_ipsec && + (!egress || (!is_drop && + !(spec->flow_context.flags & FLOW_CONTEXT_HAS_TAG))) ? + VALID_SPEC_VALID : VALID_SPEC_INVALID; +} + +static bool is_valid_spec(struct mlx5_core_dev *mdev, + const struct mlx5_flow_spec *spec, + const struct mlx5_flow_act *flow_act, + bool egress) +{ + /* We curretly only support ipsec egress flow */ + return is_valid_esp_aes_gcm(mdev, spec, flow_act, egress) != VALID_SPEC_INVALID; +} + +static bool is_valid_ethertype(struct mlx5_core_dev *mdev, + const struct ib_flow_attr *flow_attr, + bool check_inner) +{ + union ib_flow_spec *ib_spec = (union ib_flow_spec *)(flow_attr + 1); + int match_ipv = check_inner ? + MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.inner_ip_version) : + MLX5_CAP_FLOWTABLE_NIC_RX(mdev, + ft_field_support.outer_ip_version); + int inner_bit = check_inner ? IB_FLOW_SPEC_INNER : 0; + bool ipv4_spec_valid, ipv6_spec_valid; + unsigned int ip_spec_type = 0; + bool has_ethertype = false; + unsigned int spec_index; + bool mask_valid = true; + u16 eth_type = 0; + bool type_valid; + + /* Validate that ethertype is correct */ + for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) { + if ((ib_spec->type == (IB_FLOW_SPEC_ETH | inner_bit)) && + ib_spec->eth.mask.ether_type) { + mask_valid = (ib_spec->eth.mask.ether_type == + htons(0xffff)); + has_ethertype = true; + eth_type = ntohs(ib_spec->eth.val.ether_type); + } else if ((ib_spec->type == (IB_FLOW_SPEC_IPV4 | inner_bit)) || + (ib_spec->type == (IB_FLOW_SPEC_IPV6 | inner_bit))) { + ip_spec_type = ib_spec->type; + } + ib_spec = (void *)ib_spec + ib_spec->size; + } + + type_valid = (!has_ethertype) || (!ip_spec_type); + if (!type_valid && mask_valid) { + ipv4_spec_valid = (eth_type == ETH_P_IP) && + (ip_spec_type == (IB_FLOW_SPEC_IPV4 | inner_bit)); + ipv6_spec_valid = (eth_type == ETH_P_IPV6) && + (ip_spec_type == (IB_FLOW_SPEC_IPV6 | inner_bit)); + + type_valid = (ipv4_spec_valid) || (ipv6_spec_valid) || + (((eth_type == ETH_P_MPLS_UC) || + (eth_type == ETH_P_MPLS_MC)) && match_ipv); + } + + return type_valid; +} + +static bool is_valid_attr(struct mlx5_core_dev *mdev, + const struct ib_flow_attr *flow_attr) +{ + return is_valid_ethertype(mdev, flow_attr, false) && + is_valid_ethertype(mdev, flow_attr, true); +} + +static void put_flow_table(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_prio *prio, bool ft_added) +{ + prio->refcount -= !!ft_added; + if (!prio->refcount) { + mlx5_destroy_flow_table(prio->flow_table); + prio->flow_table = NULL; + } +} + +static int mlx5_ib_destroy_flow(struct ib_flow *flow_id) +{ + struct mlx5_ib_flow_handler *handler = container_of(flow_id, + struct mlx5_ib_flow_handler, + ibflow); + struct mlx5_ib_flow_handler *iter, *tmp; + struct mlx5_ib_dev *dev = handler->dev; + + mutex_lock(&dev->flow_db->lock); + + list_for_each_entry_safe(iter, tmp, &handler->list, list) { + mlx5_del_flow_rules(iter->rule); + put_flow_table(dev, iter->prio, true); + list_del(&iter->list); + kfree(iter); + } + + mlx5_del_flow_rules(handler->rule); + put_flow_table(dev, handler->prio, true); + mlx5_ib_counters_clear_description(handler->ibcounters); + mutex_unlock(&dev->flow_db->lock); + if (handler->flow_matcher) + atomic_dec(&handler->flow_matcher->usecnt); + kfree(handler); + + return 0; +} + +static int ib_prio_to_core_prio(unsigned int priority, bool dont_trap) +{ + priority *= 2; + if (!dont_trap) + priority++; + return priority; +} + +enum flow_table_type { + MLX5_IB_FT_RX, + MLX5_IB_FT_TX +}; + +#define MLX5_FS_MAX_TYPES 6 +#define MLX5_FS_MAX_ENTRIES BIT(16) + +static struct mlx5_ib_flow_prio *_get_prio(struct mlx5_flow_namespace *ns, + struct mlx5_ib_flow_prio *prio, + int priority, + int num_entries, int num_groups, + u32 flags) +{ + struct mlx5_flow_table_attr ft_attr = {}; + struct mlx5_flow_table *ft; + + ft_attr.prio = priority; + ft_attr.max_fte = num_entries; + ft_attr.flags = flags; + ft_attr.autogroup.max_num_groups = num_groups; + ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr); + if (IS_ERR(ft)) + return ERR_CAST(ft); + + prio->flow_table = ft; + prio->refcount = 0; + return prio; +} + +static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev, + struct ib_flow_attr *flow_attr, + enum flow_table_type ft_type) +{ + bool dont_trap = flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP; + struct mlx5_flow_namespace *ns = NULL; + struct mlx5_ib_flow_prio *prio; + struct mlx5_flow_table *ft; + int max_table_size; + int num_entries; + int num_groups; + bool esw_encap; + u32 flags = 0; + int priority; + + max_table_size = BIT(MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, + log_max_ft_size)); + esw_encap = mlx5_eswitch_get_encap_mode(dev->mdev) != + DEVLINK_ESWITCH_ENCAP_MODE_NONE; + if (flow_attr->type == IB_FLOW_ATTR_NORMAL) { + enum mlx5_flow_namespace_type fn_type; + + if (flow_is_multicast_only(flow_attr) && + !dont_trap) + priority = MLX5_IB_FLOW_MCAST_PRIO; + else + priority = ib_prio_to_core_prio(flow_attr->priority, + dont_trap); + if (ft_type == MLX5_IB_FT_RX) { + fn_type = MLX5_FLOW_NAMESPACE_BYPASS; + prio = &dev->flow_db->prios[priority]; + if (!dev->is_rep && !esw_encap && + MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, decap)) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_DECAP; + if (!dev->is_rep && !esw_encap && + MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, + reformat_l3_tunnel_to_l2)) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; + } else { + max_table_size = + BIT(MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, + log_max_ft_size)); + fn_type = MLX5_FLOW_NAMESPACE_EGRESS; + prio = &dev->flow_db->egress_prios[priority]; + if (!dev->is_rep && !esw_encap && + MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, reformat)) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; + } + ns = mlx5_get_flow_namespace(dev->mdev, fn_type); + num_entries = MLX5_FS_MAX_ENTRIES; + num_groups = MLX5_FS_MAX_TYPES; + } else if (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || + flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT) { + ns = mlx5_get_flow_namespace(dev->mdev, + MLX5_FLOW_NAMESPACE_LEFTOVERS); + build_leftovers_ft_param(&priority, + &num_entries, + &num_groups); + prio = &dev->flow_db->prios[MLX5_IB_FLOW_LEFTOVERS_PRIO]; + } else if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { + if (!MLX5_CAP_FLOWTABLE(dev->mdev, + allow_sniffer_and_nic_rx_shared_tir)) + return ERR_PTR(-EOPNOTSUPP); + + ns = mlx5_get_flow_namespace(dev->mdev, ft_type == MLX5_IB_FT_RX ? + MLX5_FLOW_NAMESPACE_SNIFFER_RX : + MLX5_FLOW_NAMESPACE_SNIFFER_TX); + + prio = &dev->flow_db->sniffer[ft_type]; + priority = 0; + num_entries = 1; + num_groups = 1; + } + + if (!ns) + return ERR_PTR(-EOPNOTSUPP); + + max_table_size = min_t(int, num_entries, max_table_size); + + ft = prio->flow_table; + if (!ft) + return _get_prio(ns, prio, priority, max_table_size, num_groups, + flags); + + return prio; +} + +static void set_underlay_qp(struct mlx5_ib_dev *dev, + struct mlx5_flow_spec *spec, + u32 underlay_qpn) +{ + void *misc_params_c = MLX5_ADDR_OF(fte_match_param, + spec->match_criteria, + misc_parameters); + void *misc_params_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, + misc_parameters); + + if (underlay_qpn && + MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, + ft_field_support.bth_dst_qp)) { + MLX5_SET(fte_match_set_misc, + misc_params_v, bth_dst_qp, underlay_qpn); + MLX5_SET(fte_match_set_misc, + misc_params_c, bth_dst_qp, 0xffffff); + } +} + +static void mlx5_ib_set_rule_source_port(struct mlx5_ib_dev *dev, + struct mlx5_flow_spec *spec, + struct mlx5_eswitch_rep *rep) +{ + struct mlx5_eswitch *esw = dev->mdev->priv.eswitch; + void *misc; + + if (mlx5_eswitch_vport_match_metadata_enabled(esw)) { + misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, + misc_parameters_2); + + MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0, + mlx5_eswitch_get_vport_metadata_for_match(esw, + rep->vport)); + misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, + misc_parameters_2); + + MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0, + mlx5_eswitch_get_vport_metadata_mask()); + } else { + misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, + misc_parameters); + + MLX5_SET(fte_match_set_misc, misc, source_port, rep->vport); + + misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, + misc_parameters); + + MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port); + } +} + +static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_prio *ft_prio, + const struct ib_flow_attr *flow_attr, + struct mlx5_flow_destination *dst, + u32 underlay_qpn, + struct mlx5_ib_create_flow *ucmd) +{ + struct mlx5_flow_table *ft = ft_prio->flow_table; + struct mlx5_ib_flow_handler *handler; + struct mlx5_flow_act flow_act = {}; + struct mlx5_flow_spec *spec; + struct mlx5_flow_destination dest_arr[2] = {}; + struct mlx5_flow_destination *rule_dst = dest_arr; + const void *ib_flow = (const void *)flow_attr + sizeof(*flow_attr); + unsigned int spec_index; + u32 prev_type = 0; + int err = 0; + int dest_num = 0; + bool is_egress = flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS; + + if (!is_valid_attr(dev->mdev, flow_attr)) + return ERR_PTR(-EINVAL); + + if (dev->is_rep && is_egress) + return ERR_PTR(-EINVAL); + + spec = kvzalloc(sizeof(*spec), GFP_KERNEL); + handler = kzalloc(sizeof(*handler), GFP_KERNEL); + if (!handler || !spec) { + err = -ENOMEM; + goto free; + } + + INIT_LIST_HEAD(&handler->list); + + for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) { + err = parse_flow_attr(dev->mdev, spec, + ib_flow, flow_attr, &flow_act, + prev_type); + if (err < 0) + goto free; + + prev_type = ((union ib_flow_spec *)ib_flow)->type; + ib_flow += ((union ib_flow_spec *)ib_flow)->size; + } + + if (dst && !(flow_act.action & MLX5_FLOW_CONTEXT_ACTION_DROP)) { + memcpy(&dest_arr[0], dst, sizeof(*dst)); + dest_num++; + } + + if (!flow_is_multicast_only(flow_attr)) + set_underlay_qp(dev, spec, underlay_qpn); + + if (dev->is_rep) { + struct mlx5_eswitch_rep *rep; + + rep = dev->port[flow_attr->port - 1].rep; + if (!rep) { + err = -EINVAL; + goto free; + } + + mlx5_ib_set_rule_source_port(dev, spec, rep); + } + + spec->match_criteria_enable = get_match_criteria_enable(spec->match_criteria); + + if (is_egress && + !is_valid_spec(dev->mdev, spec, &flow_act, is_egress)) { + err = -EINVAL; + goto free; + } + + if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) { + struct mlx5_ib_mcounters *mcounters; + + err = mlx5_ib_flow_counters_set_data(flow_act.counters, ucmd); + if (err) + goto free; + + mcounters = to_mcounters(flow_act.counters); + handler->ibcounters = flow_act.counters; + dest_arr[dest_num].type = + MLX5_FLOW_DESTINATION_TYPE_COUNTER; + dest_arr[dest_num].counter_id = + mlx5_fc_id(mcounters->hw_cntrs_hndl); + dest_num++; + } + + if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_DROP) { + if (!dest_num) + rule_dst = NULL; + } else { + if (flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP) + flow_act.action |= + MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO; + if (is_egress) + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW; + else if (dest_num) + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + } + + if ((spec->flow_context.flags & FLOW_CONTEXT_HAS_TAG) && + (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || + flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT)) { + mlx5_ib_warn(dev, "Flow tag %u and attribute type %x isn't allowed in leftovers\n", + spec->flow_context.flow_tag, flow_attr->type); + err = -EINVAL; + goto free; + } + handler->rule = mlx5_add_flow_rules(ft, spec, + &flow_act, + rule_dst, dest_num); + + if (IS_ERR(handler->rule)) { + err = PTR_ERR(handler->rule); + goto free; + } + + ft_prio->refcount++; + handler->prio = ft_prio; + handler->dev = dev; + + ft_prio->flow_table = ft; +free: + if (err && handler) { + mlx5_ib_counters_clear_description(handler->ibcounters); + kfree(handler); + } + kvfree(spec); + return err ? ERR_PTR(err) : handler; +} + +static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_prio *ft_prio, + const struct ib_flow_attr *flow_attr, + struct mlx5_flow_destination *dst) +{ + return _create_flow_rule(dev, ft_prio, flow_attr, dst, 0, NULL); +} + +enum { + LEFTOVERS_MC, + LEFTOVERS_UC, +}; + +static struct mlx5_ib_flow_handler *create_leftovers_rule(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_prio *ft_prio, + struct ib_flow_attr *flow_attr, + struct mlx5_flow_destination *dst) +{ + struct mlx5_ib_flow_handler *handler_ucast = NULL; + struct mlx5_ib_flow_handler *handler = NULL; + + static struct { + struct ib_flow_attr flow_attr; + struct ib_flow_spec_eth eth_flow; + } leftovers_specs[] = { + [LEFTOVERS_MC] = { + .flow_attr = { + .num_of_specs = 1, + .size = sizeof(leftovers_specs[0]) + }, + .eth_flow = { + .type = IB_FLOW_SPEC_ETH, + .size = sizeof(struct ib_flow_spec_eth), + .mask = {.dst_mac = {0x1} }, + .val = {.dst_mac = {0x1} } + } + }, + [LEFTOVERS_UC] = { + .flow_attr = { + .num_of_specs = 1, + .size = sizeof(leftovers_specs[0]) + }, + .eth_flow = { + .type = IB_FLOW_SPEC_ETH, + .size = sizeof(struct ib_flow_spec_eth), + .mask = {.dst_mac = {0x1} }, + .val = {.dst_mac = {} } + } + } + }; + + handler = create_flow_rule(dev, ft_prio, + &leftovers_specs[LEFTOVERS_MC].flow_attr, + dst); + if (!IS_ERR(handler) && + flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT) { + handler_ucast = create_flow_rule(dev, ft_prio, + &leftovers_specs[LEFTOVERS_UC].flow_attr, + dst); + if (IS_ERR(handler_ucast)) { + mlx5_del_flow_rules(handler->rule); + ft_prio->refcount--; + kfree(handler); + handler = handler_ucast; + } else { + list_add(&handler_ucast->list, &handler->list); + } + } + + return handler; +} + +static struct mlx5_ib_flow_handler *create_sniffer_rule(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_prio *ft_rx, + struct mlx5_ib_flow_prio *ft_tx, + struct mlx5_flow_destination *dst) +{ + struct mlx5_ib_flow_handler *handler_rx; + struct mlx5_ib_flow_handler *handler_tx; + int err; + static const struct ib_flow_attr flow_attr = { + .num_of_specs = 0, + .size = sizeof(flow_attr) + }; + + handler_rx = create_flow_rule(dev, ft_rx, &flow_attr, dst); + if (IS_ERR(handler_rx)) { + err = PTR_ERR(handler_rx); + goto err; + } + + handler_tx = create_flow_rule(dev, ft_tx, &flow_attr, dst); + if (IS_ERR(handler_tx)) { + err = PTR_ERR(handler_tx); + goto err_tx; + } + + list_add(&handler_tx->list, &handler_rx->list); + + return handler_rx; + +err_tx: + mlx5_del_flow_rules(handler_rx->rule); + ft_rx->refcount--; + kfree(handler_rx); +err: + return ERR_PTR(err); +} + + +static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp, + struct ib_flow_attr *flow_attr, + int domain, + struct ib_udata *udata) +{ + struct mlx5_ib_dev *dev = to_mdev(qp->device); + struct mlx5_ib_qp *mqp = to_mqp(qp); + struct mlx5_ib_flow_handler *handler = NULL; + struct mlx5_flow_destination *dst = NULL; + struct mlx5_ib_flow_prio *ft_prio_tx = NULL; + struct mlx5_ib_flow_prio *ft_prio; + bool is_egress = flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS; + struct mlx5_ib_create_flow *ucmd = NULL, ucmd_hdr; + size_t min_ucmd_sz, required_ucmd_sz; + int err; + int underlay_qpn; + + if (udata && udata->inlen) { + min_ucmd_sz = offsetof(typeof(ucmd_hdr), reserved) + + sizeof(ucmd_hdr.reserved); + if (udata->inlen < min_ucmd_sz) + return ERR_PTR(-EOPNOTSUPP); + + err = ib_copy_from_udata(&ucmd_hdr, udata, min_ucmd_sz); + if (err) + return ERR_PTR(err); + + /* currently supports only one counters data */ + if (ucmd_hdr.ncounters_data > 1) + return ERR_PTR(-EINVAL); + + required_ucmd_sz = min_ucmd_sz + + sizeof(struct mlx5_ib_flow_counters_data) * + ucmd_hdr.ncounters_data; + if (udata->inlen > required_ucmd_sz && + !ib_is_udata_cleared(udata, required_ucmd_sz, + udata->inlen - required_ucmd_sz)) + return ERR_PTR(-EOPNOTSUPP); + + ucmd = kzalloc(required_ucmd_sz, GFP_KERNEL); + if (!ucmd) + return ERR_PTR(-ENOMEM); + + err = ib_copy_from_udata(ucmd, udata, required_ucmd_sz); + if (err) + goto free_ucmd; + } + + if (flow_attr->priority > MLX5_IB_FLOW_LAST_PRIO) { + err = -ENOMEM; + goto free_ucmd; + } + + if (domain != IB_FLOW_DOMAIN_USER || + flow_attr->port > dev->num_ports || + (flow_attr->flags & ~(IB_FLOW_ATTR_FLAGS_DONT_TRAP | + IB_FLOW_ATTR_FLAGS_EGRESS))) { + err = -EINVAL; + goto free_ucmd; + } + + if (is_egress && + (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || + flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT)) { + err = -EINVAL; + goto free_ucmd; + } + + dst = kzalloc(sizeof(*dst), GFP_KERNEL); + if (!dst) { + err = -ENOMEM; + goto free_ucmd; + } + + mutex_lock(&dev->flow_db->lock); + + ft_prio = get_flow_table(dev, flow_attr, + is_egress ? MLX5_IB_FT_TX : MLX5_IB_FT_RX); + if (IS_ERR(ft_prio)) { + err = PTR_ERR(ft_prio); + goto unlock; + } + if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { + ft_prio_tx = get_flow_table(dev, flow_attr, MLX5_IB_FT_TX); + if (IS_ERR(ft_prio_tx)) { + err = PTR_ERR(ft_prio_tx); + ft_prio_tx = NULL; + goto destroy_ft; + } + } + + if (is_egress) { + dst->type = MLX5_FLOW_DESTINATION_TYPE_PORT; + } else { + dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR; + if (mqp->is_rss) + dst->tir_num = mqp->rss_qp.tirn; + else + dst->tir_num = mqp->raw_packet_qp.rq.tirn; + } + + if (flow_attr->type == IB_FLOW_ATTR_NORMAL) { + underlay_qpn = (mqp->flags & IB_QP_CREATE_SOURCE_QPN) ? + mqp->underlay_qpn : + 0; + handler = _create_flow_rule(dev, ft_prio, flow_attr, dst, + underlay_qpn, ucmd); + } else if (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || + flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT) { + handler = create_leftovers_rule(dev, ft_prio, flow_attr, + dst); + } else if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { + handler = create_sniffer_rule(dev, ft_prio, ft_prio_tx, dst); + } else { + err = -EINVAL; + goto destroy_ft; + } + + if (IS_ERR(handler)) { + err = PTR_ERR(handler); + handler = NULL; + goto destroy_ft; + } + + mutex_unlock(&dev->flow_db->lock); + kfree(dst); + kfree(ucmd); + + return &handler->ibflow; + +destroy_ft: + put_flow_table(dev, ft_prio, false); + if (ft_prio_tx) + put_flow_table(dev, ft_prio_tx, false); +unlock: + mutex_unlock(&dev->flow_db->lock); + kfree(dst); +free_ucmd: + kfree(ucmd); + return ERR_PTR(err); +} + +static struct mlx5_ib_flow_prio * +_get_flow_table(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_matcher *fs_matcher, + bool mcast) +{ + struct mlx5_flow_namespace *ns = NULL; + struct mlx5_ib_flow_prio *prio = NULL; + int max_table_size = 0; + bool esw_encap; + u32 flags = 0; + int priority; + + if (mcast) + priority = MLX5_IB_FLOW_MCAST_PRIO; + else + priority = ib_prio_to_core_prio(fs_matcher->priority, false); + + esw_encap = mlx5_eswitch_get_encap_mode(dev->mdev) != + DEVLINK_ESWITCH_ENCAP_MODE_NONE; + if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) { + max_table_size = BIT(MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, + log_max_ft_size)); + if (MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, decap) && !esw_encap) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_DECAP; + if (MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, + reformat_l3_tunnel_to_l2) && + !esw_encap) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; + } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS) { + max_table_size = BIT( + MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, log_max_ft_size)); + if (MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, reformat) && !esw_encap) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; + } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB) { + max_table_size = BIT( + MLX5_CAP_ESW_FLOWTABLE_FDB(dev->mdev, log_max_ft_size)); + if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev->mdev, decap) && esw_encap) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_DECAP; + if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev->mdev, reformat_l3_tunnel_to_l2) && + esw_encap) + flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; + priority = FDB_BYPASS_PATH; + } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) { + max_table_size = + BIT(MLX5_CAP_FLOWTABLE_RDMA_RX(dev->mdev, + log_max_ft_size)); + priority = fs_matcher->priority; + } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_TX) { + max_table_size = + BIT(MLX5_CAP_FLOWTABLE_RDMA_TX(dev->mdev, + log_max_ft_size)); + priority = fs_matcher->priority; + } + + max_table_size = min_t(int, max_table_size, MLX5_FS_MAX_ENTRIES); + + ns = mlx5_get_flow_namespace(dev->mdev, fs_matcher->ns_type); + if (!ns) + return ERR_PTR(-EOPNOTSUPP); + + if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) + prio = &dev->flow_db->prios[priority]; + else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS) + prio = &dev->flow_db->egress_prios[priority]; + else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB) + prio = &dev->flow_db->fdb; + else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) + prio = &dev->flow_db->rdma_rx[priority]; + else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_TX) + prio = &dev->flow_db->rdma_tx[priority]; + + if (!prio) + return ERR_PTR(-EINVAL); + + if (prio->flow_table) + return prio; + + return _get_prio(ns, prio, priority, max_table_size, + MLX5_FS_MAX_TYPES, flags); +} + +static struct mlx5_ib_flow_handler * +_create_raw_flow_rule(struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_prio *ft_prio, + struct mlx5_flow_destination *dst, + struct mlx5_ib_flow_matcher *fs_matcher, + struct mlx5_flow_context *flow_context, + struct mlx5_flow_act *flow_act, + void *cmd_in, int inlen, + int dst_num) +{ + struct mlx5_ib_flow_handler *handler; + struct mlx5_flow_spec *spec; + struct mlx5_flow_table *ft = ft_prio->flow_table; + int err = 0; + + spec = kvzalloc(sizeof(*spec), GFP_KERNEL); + handler = kzalloc(sizeof(*handler), GFP_KERNEL); + if (!handler || !spec) { + err = -ENOMEM; + goto free; + } + + INIT_LIST_HEAD(&handler->list); + + memcpy(spec->match_value, cmd_in, inlen); + memcpy(spec->match_criteria, fs_matcher->matcher_mask.match_params, + fs_matcher->mask_len); + spec->match_criteria_enable = fs_matcher->match_criteria_enable; + spec->flow_context = *flow_context; + + handler->rule = mlx5_add_flow_rules(ft, spec, + flow_act, dst, dst_num); + + if (IS_ERR(handler->rule)) { + err = PTR_ERR(handler->rule); + goto free; + } + + ft_prio->refcount++; + handler->prio = ft_prio; + handler->dev = dev; + ft_prio->flow_table = ft; + +free: + if (err) + kfree(handler); + kvfree(spec); + return err ? ERR_PTR(err) : handler; +} + +static bool raw_fs_is_multicast(struct mlx5_ib_flow_matcher *fs_matcher, + void *match_v) +{ + void *match_c; + void *match_v_set_lyr_2_4, *match_c_set_lyr_2_4; + void *dmac, *dmac_mask; + void *ipv4, *ipv4_mask; + + if (!(fs_matcher->match_criteria_enable & + (1 << MATCH_CRITERIA_ENABLE_OUTER_BIT))) + return false; + + match_c = fs_matcher->matcher_mask.match_params; + match_v_set_lyr_2_4 = MLX5_ADDR_OF(fte_match_param, match_v, + outer_headers); + match_c_set_lyr_2_4 = MLX5_ADDR_OF(fte_match_param, match_c, + outer_headers); + + dmac = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_v_set_lyr_2_4, + dmac_47_16); + dmac_mask = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_c_set_lyr_2_4, + dmac_47_16); + + if (is_multicast_ether_addr(dmac) && + is_multicast_ether_addr(dmac_mask)) + return true; + + ipv4 = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_v_set_lyr_2_4, + dst_ipv4_dst_ipv6.ipv4_layout.ipv4); + + ipv4_mask = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_c_set_lyr_2_4, + dst_ipv4_dst_ipv6.ipv4_layout.ipv4); + + if (ipv4_is_multicast(*(__be32 *)(ipv4)) && + ipv4_is_multicast(*(__be32 *)(ipv4_mask))) + return true; + + return false; +} + +static struct mlx5_ib_flow_handler *raw_fs_rule_add( + struct mlx5_ib_dev *dev, struct mlx5_ib_flow_matcher *fs_matcher, + struct mlx5_flow_context *flow_context, struct mlx5_flow_act *flow_act, + u32 counter_id, void *cmd_in, int inlen, int dest_id, int dest_type) +{ + struct mlx5_flow_destination *dst; + struct mlx5_ib_flow_prio *ft_prio; + struct mlx5_ib_flow_handler *handler; + int dst_num = 0; + bool mcast; + int err; + + if (fs_matcher->flow_type != MLX5_IB_FLOW_TYPE_NORMAL) + return ERR_PTR(-EOPNOTSUPP); + + if (fs_matcher->priority > MLX5_IB_FLOW_LAST_PRIO) + return ERR_PTR(-ENOMEM); + + dst = kcalloc(2, sizeof(*dst), GFP_KERNEL); + if (!dst) + return ERR_PTR(-ENOMEM); + + mcast = raw_fs_is_multicast(fs_matcher, cmd_in); + mutex_lock(&dev->flow_db->lock); + + ft_prio = _get_flow_table(dev, fs_matcher, mcast); + if (IS_ERR(ft_prio)) { + err = PTR_ERR(ft_prio); + goto unlock; + } + + if (dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR) { + dst[dst_num].type = dest_type; + dst[dst_num++].tir_num = dest_id; + flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + } else if (dest_type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) { + dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM; + dst[dst_num++].ft_num = dest_id; + flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + } else if (dest_type == MLX5_FLOW_DESTINATION_TYPE_PORT) { + dst[dst_num++].type = MLX5_FLOW_DESTINATION_TYPE_PORT; + flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW; + } + + + if (flow_act->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) { + dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER; + dst[dst_num].counter_id = counter_id; + dst_num++; + } + + handler = _create_raw_flow_rule(dev, ft_prio, dst, fs_matcher, + flow_context, flow_act, + cmd_in, inlen, dst_num); + + if (IS_ERR(handler)) { + err = PTR_ERR(handler); + goto destroy_ft; + } + + mutex_unlock(&dev->flow_db->lock); + atomic_inc(&fs_matcher->usecnt); + handler->flow_matcher = fs_matcher; + + kfree(dst); + + return handler; + +destroy_ft: + put_flow_table(dev, ft_prio, false); +unlock: + mutex_unlock(&dev->flow_db->lock); + kfree(dst); + + return ERR_PTR(err); +} + +static u32 mlx5_ib_flow_action_flags_to_accel_xfrm_flags(u32 mlx5_flags) +{ + u32 flags = 0; + + if (mlx5_flags & MLX5_IB_UAPI_FLOW_ACTION_FLAGS_REQUIRE_METADATA) + flags |= MLX5_ACCEL_XFRM_FLAG_REQUIRE_METADATA; + + return flags; +} + +#define MLX5_FLOW_ACTION_ESP_CREATE_LAST_SUPPORTED \ + MLX5_IB_UAPI_FLOW_ACTION_FLAGS_REQUIRE_METADATA +static struct ib_flow_action * +mlx5_ib_create_flow_action_esp(struct ib_device *device, + const struct ib_flow_action_attrs_esp *attr, + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_ib_dev *mdev = to_mdev(device); + struct ib_uverbs_flow_action_esp_keymat_aes_gcm *aes_gcm; + struct mlx5_accel_esp_xfrm_attrs accel_attrs = {}; + struct mlx5_ib_flow_action *action; + u64 action_flags; + u64 flags; + int err = 0; + + err = uverbs_get_flags64( + &action_flags, attrs, MLX5_IB_ATTR_CREATE_FLOW_ACTION_FLAGS, + ((MLX5_FLOW_ACTION_ESP_CREATE_LAST_SUPPORTED << 1) - 1)); + if (err) + return ERR_PTR(err); + + flags = mlx5_ib_flow_action_flags_to_accel_xfrm_flags(action_flags); + + /* We current only support a subset of the standard features. Only a + * keymat of type AES_GCM, with icv_len == 16, iv_algo == SEQ and esn + * (with overlap). Full offload mode isn't supported. + */ + if (!attr->keymat || attr->replay || attr->encap || + attr->spi || attr->seq || attr->tfc_pad || + attr->hard_limit_pkts || + (attr->flags & ~(IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED | + IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ENCRYPT))) + return ERR_PTR(-EOPNOTSUPP); + + if (attr->keymat->protocol != + IB_UVERBS_FLOW_ACTION_ESP_KEYMAT_AES_GCM) + return ERR_PTR(-EOPNOTSUPP); + + aes_gcm = &attr->keymat->keymat.aes_gcm; + + if (aes_gcm->icv_len != 16 || + aes_gcm->iv_algo != IB_UVERBS_FLOW_ACTION_IV_ALGO_SEQ) + return ERR_PTR(-EOPNOTSUPP); + + action = kmalloc(sizeof(*action), GFP_KERNEL); + if (!action) + return ERR_PTR(-ENOMEM); + + action->esp_aes_gcm.ib_flags = attr->flags; + memcpy(&accel_attrs.keymat.aes_gcm.aes_key, &aes_gcm->aes_key, + sizeof(accel_attrs.keymat.aes_gcm.aes_key)); + accel_attrs.keymat.aes_gcm.key_len = aes_gcm->key_len * 8; + memcpy(&accel_attrs.keymat.aes_gcm.salt, &aes_gcm->salt, + sizeof(accel_attrs.keymat.aes_gcm.salt)); + memcpy(&accel_attrs.keymat.aes_gcm.seq_iv, &aes_gcm->iv, + sizeof(accel_attrs.keymat.aes_gcm.seq_iv)); + accel_attrs.keymat.aes_gcm.icv_len = aes_gcm->icv_len * 8; + accel_attrs.keymat.aes_gcm.iv_algo = MLX5_ACCEL_ESP_AES_GCM_IV_ALGO_SEQ; + accel_attrs.keymat_type = MLX5_ACCEL_ESP_KEYMAT_AES_GCM; + + accel_attrs.esn = attr->esn; + if (attr->flags & IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED) + accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED; + if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW) + accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP; + + if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ENCRYPT) + accel_attrs.action |= MLX5_ACCEL_ESP_ACTION_ENCRYPT; + + action->esp_aes_gcm.ctx = + mlx5_accel_esp_create_xfrm(mdev->mdev, &accel_attrs, flags); + if (IS_ERR(action->esp_aes_gcm.ctx)) { + err = PTR_ERR(action->esp_aes_gcm.ctx); + goto err_parse; + } + + action->esp_aes_gcm.ib_flags = attr->flags; + + return &action->ib_action; + +err_parse: + kfree(action); + return ERR_PTR(err); +} + +static int +mlx5_ib_modify_flow_action_esp(struct ib_flow_action *action, + const struct ib_flow_action_attrs_esp *attr, + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_ib_flow_action *maction = to_mflow_act(action); + struct mlx5_accel_esp_xfrm_attrs accel_attrs; + int err = 0; + + if (attr->keymat || attr->replay || attr->encap || + attr->spi || attr->seq || attr->tfc_pad || + attr->hard_limit_pkts || + (attr->flags & ~(IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED | + IB_FLOW_ACTION_ESP_FLAGS_MOD_ESP_ATTRS | + IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW))) + return -EOPNOTSUPP; + + /* Only the ESN value or the MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP can + * be modified. + */ + if (!(maction->esp_aes_gcm.ib_flags & + IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED) && + attr->flags & (IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED | + IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW)) + return -EINVAL; + + memcpy(&accel_attrs, &maction->esp_aes_gcm.ctx->attrs, + sizeof(accel_attrs)); + + accel_attrs.esn = attr->esn; + if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW) + accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP; + else + accel_attrs.flags &= ~MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP; + + err = mlx5_accel_esp_modify_xfrm(maction->esp_aes_gcm.ctx, + &accel_attrs); + if (err) + return err; + + maction->esp_aes_gcm.ib_flags &= + ~IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW; + maction->esp_aes_gcm.ib_flags |= + attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW; + + return 0; +} + +static void destroy_flow_action_raw(struct mlx5_ib_flow_action *maction) +{ + switch (maction->flow_action_raw.sub_type) { + case MLX5_IB_FLOW_ACTION_MODIFY_HEADER: + mlx5_modify_header_dealloc(maction->flow_action_raw.dev->mdev, + maction->flow_action_raw.modify_hdr); + break; + case MLX5_IB_FLOW_ACTION_PACKET_REFORMAT: + mlx5_packet_reformat_dealloc(maction->flow_action_raw.dev->mdev, + maction->flow_action_raw.pkt_reformat); + break; + case MLX5_IB_FLOW_ACTION_DECAP: + break; + default: + break; + } +} + +static int mlx5_ib_destroy_flow_action(struct ib_flow_action *action) +{ + struct mlx5_ib_flow_action *maction = to_mflow_act(action); + + switch (action->type) { + case IB_FLOW_ACTION_ESP: + /* + * We only support aes_gcm by now, so we implicitly know this is + * the underline crypto. + */ + mlx5_accel_esp_destroy_xfrm(maction->esp_aes_gcm.ctx); + break; + case IB_FLOW_ACTION_UNSPECIFIED: + destroy_flow_action_raw(maction); + break; + default: + WARN_ON(true); + break; + } + + kfree(maction); + return 0; +} + +static int +mlx5_ib_ft_type_to_namespace(enum mlx5_ib_uapi_flow_table_type table_type, + enum mlx5_flow_namespace_type *namespace) +{ + switch (table_type) { + case MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX: + *namespace = MLX5_FLOW_NAMESPACE_BYPASS; + break; + case MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX: + *namespace = MLX5_FLOW_NAMESPACE_EGRESS; + break; + case MLX5_IB_UAPI_FLOW_TABLE_TYPE_FDB: + *namespace = MLX5_FLOW_NAMESPACE_FDB; + break; + case MLX5_IB_UAPI_FLOW_TABLE_TYPE_RDMA_RX: + *namespace = MLX5_FLOW_NAMESPACE_RDMA_RX; + break; + case MLX5_IB_UAPI_FLOW_TABLE_TYPE_RDMA_TX: + *namespace = MLX5_FLOW_NAMESPACE_RDMA_TX; + break; + default: + return -EINVAL; + } + + return 0; +} + +static const struct uverbs_attr_spec mlx5_ib_flow_type[] = { + [MLX5_IB_FLOW_TYPE_NORMAL] = { + .type = UVERBS_ATTR_TYPE_PTR_IN, + .u.ptr = { + .len = sizeof(u16), /* data is priority */ + .min_len = sizeof(u16), + } + }, + [MLX5_IB_FLOW_TYPE_SNIFFER] = { + .type = UVERBS_ATTR_TYPE_PTR_IN, + UVERBS_ATTR_NO_DATA(), + }, + [MLX5_IB_FLOW_TYPE_ALL_DEFAULT] = { + .type = UVERBS_ATTR_TYPE_PTR_IN, + UVERBS_ATTR_NO_DATA(), + }, + [MLX5_IB_FLOW_TYPE_MC_DEFAULT] = { + .type = UVERBS_ATTR_TYPE_PTR_IN, + UVERBS_ATTR_NO_DATA(), + }, +}; + +static bool is_flow_dest(void *obj, int *dest_id, int *dest_type) +{ + struct devx_obj *devx_obj = obj; + u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, opcode); + + switch (opcode) { + case MLX5_CMD_OP_DESTROY_TIR: + *dest_type = MLX5_FLOW_DESTINATION_TYPE_TIR; + *dest_id = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, + obj_id); + return true; + + case MLX5_CMD_OP_DESTROY_FLOW_TABLE: + *dest_type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE; + *dest_id = MLX5_GET(destroy_flow_table_in, devx_obj->dinbox, + table_id); + return true; + default: + return false; + } +} + +static int get_dests(struct uverbs_attr_bundle *attrs, + struct mlx5_ib_flow_matcher *fs_matcher, int *dest_id, + int *dest_type, struct ib_qp **qp, u32 *flags) +{ + bool dest_devx, dest_qp; + void *devx_obj; + int err; + + dest_devx = uverbs_attr_is_valid(attrs, + MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); + dest_qp = uverbs_attr_is_valid(attrs, + MLX5_IB_ATTR_CREATE_FLOW_DEST_QP); + + *flags = 0; + err = uverbs_get_flags32(flags, attrs, MLX5_IB_ATTR_CREATE_FLOW_FLAGS, + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS | + MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP); + if (err) + return err; + + /* Both flags are not allowed */ + if (*flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS && + *flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP) + return -EINVAL; + + if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) { + if (dest_devx && (dest_qp || *flags)) + return -EINVAL; + else if (dest_qp && *flags) + return -EINVAL; + } + + /* Allow only DEVX object, drop as dest for FDB */ + if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && !(dest_devx || + (*flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP))) + return -EINVAL; + + /* Allow only DEVX object or QP as dest when inserting to RDMA_RX */ + if ((fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) && + ((!dest_devx && !dest_qp) || (dest_devx && dest_qp))) + return -EINVAL; + + *qp = NULL; + if (dest_devx) { + devx_obj = + uverbs_attr_get_obj(attrs, + MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); + + /* Verify that the given DEVX object is a flow + * steering destination. + */ + if (!is_flow_dest(devx_obj, dest_id, dest_type)) + return -EINVAL; + /* Allow only flow table as dest when inserting to FDB or RDMA_RX */ + if ((fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB || + fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) && + *dest_type != MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) + return -EINVAL; + } else if (dest_qp) { + struct mlx5_ib_qp *mqp; + + *qp = uverbs_attr_get_obj(attrs, + MLX5_IB_ATTR_CREATE_FLOW_DEST_QP); + if (IS_ERR(*qp)) + return PTR_ERR(*qp); + + if ((*qp)->qp_type != IB_QPT_RAW_PACKET) + return -EINVAL; + + mqp = to_mqp(*qp); + if (mqp->is_rss) + *dest_id = mqp->rss_qp.tirn; + else + *dest_id = mqp->raw_packet_qp.rq.tirn; + *dest_type = MLX5_FLOW_DESTINATION_TYPE_TIR; + } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS || + fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_TX) { + *dest_type = MLX5_FLOW_DESTINATION_TYPE_PORT; + } + + if (*dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR && + (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS || + fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_TX)) + return -EINVAL; + + return 0; +} + +static bool is_flow_counter(void *obj, u32 offset, u32 *counter_id) +{ + struct devx_obj *devx_obj = obj; + u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, opcode); + + if (opcode == MLX5_CMD_OP_DEALLOC_FLOW_COUNTER) { + + if (offset && offset >= devx_obj->flow_counter_bulk_size) + return false; + + *counter_id = MLX5_GET(dealloc_flow_counter_in, + devx_obj->dinbox, + flow_counter_id); + *counter_id += offset; + return true; + } + + return false; +} + +#define MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS 2 +static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_flow_context flow_context = {.flow_tag = + MLX5_FS_DEFAULT_FLOW_TAG}; + u32 *offset_attr, offset = 0, counter_id = 0; + int dest_id, dest_type = -1, inlen, len, ret, i; + struct mlx5_ib_flow_handler *flow_handler; + struct mlx5_ib_flow_matcher *fs_matcher; + struct ib_uobject **arr_flow_actions; + struct ib_uflow_resources *uflow_res; + struct mlx5_flow_act flow_act = {}; + struct ib_qp *qp = NULL; + void *devx_obj, *cmd_in; + struct ib_uobject *uobj; + struct mlx5_ib_dev *dev; + u32 flags; + + if (!capable(CAP_NET_RAW)) + return -EPERM; + + fs_matcher = uverbs_attr_get_obj(attrs, + MLX5_IB_ATTR_CREATE_FLOW_MATCHER); + uobj = uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_CREATE_FLOW_HANDLE); + dev = mlx5_udata_to_mdev(&attrs->driver_udata); + + if (get_dests(attrs, fs_matcher, &dest_id, &dest_type, &qp, &flags)) + return -EINVAL; + + if (flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DEFAULT_MISS) + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_NS; + + if (flags & MLX5_IB_ATTR_CREATE_FLOW_FLAGS_DROP) + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_DROP; + + len = uverbs_attr_get_uobjs_arr(attrs, + MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, &arr_flow_actions); + if (len) { + devx_obj = arr_flow_actions[0]->object; + + if (uverbs_attr_is_valid(attrs, + MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET)) { + + int num_offsets = uverbs_attr_ptr_get_array_size( + attrs, + MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET, + sizeof(u32)); + + if (num_offsets != 1) + return -EINVAL; + + offset_attr = uverbs_attr_get_alloced_ptr( + attrs, + MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET); + offset = *offset_attr; + } + + if (!is_flow_counter(devx_obj, offset, &counter_id)) + return -EINVAL; + + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT; + } + + cmd_in = uverbs_attr_get_alloced_ptr( + attrs, MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE); + inlen = uverbs_attr_get_len(attrs, + MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE); + + uflow_res = flow_resources_alloc(MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS); + if (!uflow_res) + return -ENOMEM; + + len = uverbs_attr_get_uobjs_arr(attrs, + MLX5_IB_ATTR_CREATE_FLOW_ARR_FLOW_ACTIONS, &arr_flow_actions); + for (i = 0; i < len; i++) { + struct mlx5_ib_flow_action *maction = + to_mflow_act(arr_flow_actions[i]->object); + + ret = parse_flow_flow_action(maction, false, &flow_act); + if (ret) + goto err_out; + flow_resources_add(uflow_res, IB_FLOW_SPEC_ACTION_HANDLE, + arr_flow_actions[i]->object); + } + + ret = uverbs_copy_from(&flow_context.flow_tag, attrs, + MLX5_IB_ATTR_CREATE_FLOW_TAG); + if (!ret) { + if (flow_context.flow_tag >= BIT(24)) { + ret = -EINVAL; + goto err_out; + } + flow_context.flags |= FLOW_CONTEXT_HAS_TAG; + } + + flow_handler = + raw_fs_rule_add(dev, fs_matcher, &flow_context, &flow_act, + counter_id, cmd_in, inlen, dest_id, dest_type); + if (IS_ERR(flow_handler)) { + ret = PTR_ERR(flow_handler); + goto err_out; + } + + ib_set_flow(uobj, &flow_handler->ibflow, qp, &dev->ib_dev, uflow_res); + + return 0; +err_out: + ib_uverbs_flow_resources_free(uflow_res); + return ret; +} + +static int flow_matcher_cleanup(struct ib_uobject *uobject, + enum rdma_remove_reason why, + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_ib_flow_matcher *obj = uobject->object; + int ret; + + ret = ib_destroy_usecnt(&obj->usecnt, why, uobject); + if (ret) + return ret; + + kfree(obj); + return 0; +} + +static int mlx5_ib_matcher_ns(struct uverbs_attr_bundle *attrs, + struct mlx5_ib_flow_matcher *obj) +{ + enum mlx5_ib_uapi_flow_table_type ft_type = + MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX; + u32 flags; + int err; + + /* New users should use MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE and older + * users should switch to it. We leave this to not break userspace + */ + if (uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE) && + uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS)) + return -EINVAL; + + if (uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE)) { + err = uverbs_get_const(&ft_type, attrs, + MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE); + if (err) + return err; + + err = mlx5_ib_ft_type_to_namespace(ft_type, &obj->ns_type); + if (err) + return err; + + return 0; + } + + if (uverbs_attr_is_valid(attrs, MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS)) { + err = uverbs_get_flags32(&flags, attrs, + MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS, + IB_FLOW_ATTR_FLAGS_EGRESS); + if (err) + return err; + + if (flags) { + mlx5_ib_ft_type_to_namespace( + MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX, + &obj->ns_type); + return 0; + } + } + + obj->ns_type = MLX5_FLOW_NAMESPACE_BYPASS; + + return 0; +} + +static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_MATCHER_CREATE)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = uverbs_attr_get_uobject( + attrs, MLX5_IB_ATTR_FLOW_MATCHER_CREATE_HANDLE); + struct mlx5_ib_dev *dev = mlx5_udata_to_mdev(&attrs->driver_udata); + struct mlx5_ib_flow_matcher *obj; + int err; + + obj = kzalloc(sizeof(struct mlx5_ib_flow_matcher), GFP_KERNEL); + if (!obj) + return -ENOMEM; + + obj->mask_len = uverbs_attr_get_len( + attrs, MLX5_IB_ATTR_FLOW_MATCHER_MATCH_MASK); + err = uverbs_copy_from(&obj->matcher_mask, + attrs, + MLX5_IB_ATTR_FLOW_MATCHER_MATCH_MASK); + if (err) + goto end; + + obj->flow_type = uverbs_attr_get_enum_id( + attrs, MLX5_IB_ATTR_FLOW_MATCHER_FLOW_TYPE); + + if (obj->flow_type == MLX5_IB_FLOW_TYPE_NORMAL) { + err = uverbs_copy_from(&obj->priority, + attrs, + MLX5_IB_ATTR_FLOW_MATCHER_FLOW_TYPE); + if (err) + goto end; + } + + err = uverbs_copy_from(&obj->match_criteria_enable, + attrs, + MLX5_IB_ATTR_FLOW_MATCHER_MATCH_CRITERIA); + if (err) + goto end; + + err = mlx5_ib_matcher_ns(attrs, obj); + if (err) + goto end; + + uobj->object = obj; + obj->mdev = dev->mdev; + atomic_set(&obj->usecnt, 0); + return 0; + +end: + kfree(obj); + return err; +} + +static struct ib_flow_action * +mlx5_ib_create_modify_header(struct mlx5_ib_dev *dev, + enum mlx5_ib_uapi_flow_table_type ft_type, + u8 num_actions, void *in) +{ + enum mlx5_flow_namespace_type namespace; + struct mlx5_ib_flow_action *maction; + int ret; + + ret = mlx5_ib_ft_type_to_namespace(ft_type, &namespace); + if (ret) + return ERR_PTR(-EINVAL); + + maction = kzalloc(sizeof(*maction), GFP_KERNEL); + if (!maction) + return ERR_PTR(-ENOMEM); + + maction->flow_action_raw.modify_hdr = + mlx5_modify_header_alloc(dev->mdev, namespace, num_actions, in); + + if (IS_ERR(maction->flow_action_raw.modify_hdr)) { + ret = PTR_ERR(maction->flow_action_raw.modify_hdr); + kfree(maction); + return ERR_PTR(ret); + } + maction->flow_action_raw.sub_type = + MLX5_IB_FLOW_ACTION_MODIFY_HEADER; + maction->flow_action_raw.dev = dev; + + return &maction->ib_action; +} + +static bool mlx5_ib_modify_header_supported(struct mlx5_ib_dev *dev) +{ + return MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, + max_modify_header_actions) || + MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, + max_modify_header_actions) || + MLX5_CAP_FLOWTABLE_RDMA_TX(dev->mdev, + max_modify_header_actions); +} + +static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = uverbs_attr_get_uobject( + attrs, MLX5_IB_ATTR_CREATE_MODIFY_HEADER_HANDLE); + struct mlx5_ib_dev *mdev = mlx5_udata_to_mdev(&attrs->driver_udata); + enum mlx5_ib_uapi_flow_table_type ft_type; + struct ib_flow_action *action; + int num_actions; + void *in; + int ret; + + if (!mlx5_ib_modify_header_supported(mdev)) + return -EOPNOTSUPP; + + in = uverbs_attr_get_alloced_ptr(attrs, + MLX5_IB_ATTR_CREATE_MODIFY_HEADER_ACTIONS_PRM); + + num_actions = uverbs_attr_ptr_get_array_size( + attrs, MLX5_IB_ATTR_CREATE_MODIFY_HEADER_ACTIONS_PRM, + MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)); + if (num_actions < 0) + return num_actions; + + ret = uverbs_get_const(&ft_type, attrs, + MLX5_IB_ATTR_CREATE_MODIFY_HEADER_FT_TYPE); + if (ret) + return ret; + action = mlx5_ib_create_modify_header(mdev, ft_type, num_actions, in); + if (IS_ERR(action)) + return PTR_ERR(action); + + uverbs_flow_action_fill_action(action, uobj, &mdev->ib_dev, + IB_FLOW_ACTION_UNSPECIFIED); + + return 0; +} + +static bool mlx5_ib_flow_action_packet_reformat_valid(struct mlx5_ib_dev *ibdev, + u8 packet_reformat_type, + u8 ft_type) +{ + switch (packet_reformat_type) { + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L2_TUNNEL: + if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX) + return MLX5_CAP_FLOWTABLE(ibdev->mdev, + encap_general_header); + break; + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L3_TUNNEL: + if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_TX) + return MLX5_CAP_FLOWTABLE_NIC_TX(ibdev->mdev, + reformat_l2_to_l3_tunnel); + break; + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L3_TUNNEL_TO_L2: + if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX) + return MLX5_CAP_FLOWTABLE_NIC_RX(ibdev->mdev, + reformat_l3_tunnel_to_l2); + break; + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TUNNEL_TO_L2: + if (ft_type == MLX5_IB_UAPI_FLOW_TABLE_TYPE_NIC_RX) + return MLX5_CAP_FLOWTABLE_NIC_RX(ibdev->mdev, decap); + break; + default: + break; + } + + return false; +} + +static int mlx5_ib_dv_to_prm_packet_reforamt_type(u8 dv_prt, u8 *prm_prt) +{ + switch (dv_prt) { + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L2_TUNNEL: + *prm_prt = MLX5_REFORMAT_TYPE_L2_TO_L2_TUNNEL; + break; + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L3_TUNNEL_TO_L2: + *prm_prt = MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2; + break; + case MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TO_L3_TUNNEL: + *prm_prt = MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL; + break; + default: + return -EINVAL; + } + + return 0; +} + +static int mlx5_ib_flow_action_create_packet_reformat_ctx( + struct mlx5_ib_dev *dev, + struct mlx5_ib_flow_action *maction, + u8 ft_type, u8 dv_prt, + void *in, size_t len) +{ + enum mlx5_flow_namespace_type namespace; + u8 prm_prt; + int ret; + + ret = mlx5_ib_ft_type_to_namespace(ft_type, &namespace); + if (ret) + return ret; + + ret = mlx5_ib_dv_to_prm_packet_reforamt_type(dv_prt, &prm_prt); + if (ret) + return ret; + + maction->flow_action_raw.pkt_reformat = + mlx5_packet_reformat_alloc(dev->mdev, prm_prt, len, + in, namespace); + if (IS_ERR(maction->flow_action_raw.pkt_reformat)) { + ret = PTR_ERR(maction->flow_action_raw.pkt_reformat); + return ret; + } + + maction->flow_action_raw.sub_type = + MLX5_IB_FLOW_ACTION_PACKET_REFORMAT; + maction->flow_action_raw.dev = dev; + + return 0; +} + +static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, + MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_HANDLE); + struct mlx5_ib_dev *mdev = mlx5_udata_to_mdev(&attrs->driver_udata); + enum mlx5_ib_uapi_flow_action_packet_reformat_type dv_prt; + enum mlx5_ib_uapi_flow_table_type ft_type; + struct mlx5_ib_flow_action *maction; + int ret; + + ret = uverbs_get_const(&ft_type, attrs, + MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_FT_TYPE); + if (ret) + return ret; + + ret = uverbs_get_const(&dv_prt, attrs, + MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_TYPE); + if (ret) + return ret; + + if (!mlx5_ib_flow_action_packet_reformat_valid(mdev, dv_prt, ft_type)) + return -EOPNOTSUPP; + + maction = kzalloc(sizeof(*maction), GFP_KERNEL); + if (!maction) + return -ENOMEM; + + if (dv_prt == + MLX5_IB_UAPI_FLOW_ACTION_PACKET_REFORMAT_TYPE_L2_TUNNEL_TO_L2) { + maction->flow_action_raw.sub_type = + MLX5_IB_FLOW_ACTION_DECAP; + maction->flow_action_raw.dev = mdev; + } else { + void *in; + int len; + + in = uverbs_attr_get_alloced_ptr(attrs, + MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_DATA_BUF); + if (IS_ERR(in)) { + ret = PTR_ERR(in); + goto free_maction; + } + + len = uverbs_attr_get_len(attrs, + MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_DATA_BUF); + + ret = mlx5_ib_flow_action_create_packet_reformat_ctx(mdev, + maction, ft_type, dv_prt, in, len); + if (ret) + goto free_maction; + } + + uverbs_flow_action_fill_action(&maction->ib_action, uobj, &mdev->ib_dev, + IB_FLOW_ACTION_UNSPECIFIED); + return 0; + +free_maction: + kfree(maction); + return ret; +} + +DECLARE_UVERBS_NAMED_METHOD( + MLX5_IB_METHOD_CREATE_FLOW, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_HANDLE, + UVERBS_OBJECT_FLOW, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN( + MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE, + UVERBS_ATTR_SIZE(1, sizeof(struct mlx5_ib_match_params)), + UA_MANDATORY, + UA_ALLOC_AND_COPY), + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_MATCHER, + MLX5_IB_OBJECT_FLOW_MATCHER, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_DEST_QP, + UVERBS_OBJECT_QP, + UVERBS_ACCESS_READ), + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX, + MLX5_IB_OBJECT_DEVX_OBJ, + UVERBS_ACCESS_READ), + UVERBS_ATTR_IDRS_ARR(MLX5_IB_ATTR_CREATE_FLOW_ARR_FLOW_ACTIONS, + UVERBS_OBJECT_FLOW_ACTION, + UVERBS_ACCESS_READ, 1, + MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_TAG, + UVERBS_ATTR_TYPE(u32), + UA_OPTIONAL), + UVERBS_ATTR_IDRS_ARR(MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, + MLX5_IB_OBJECT_DEVX_OBJ, + UVERBS_ACCESS_READ, 1, 1, + UA_OPTIONAL), + UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX_OFFSET, + UVERBS_ATTR_MIN_SIZE(sizeof(u32)), + UA_OPTIONAL, + UA_ALLOC_AND_COPY), + UVERBS_ATTR_FLAGS_IN(MLX5_IB_ATTR_CREATE_FLOW_FLAGS, + enum mlx5_ib_create_flow_flags, + UA_OPTIONAL)); + +DECLARE_UVERBS_NAMED_METHOD_DESTROY( + MLX5_IB_METHOD_DESTROY_FLOW, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_FLOW_HANDLE, + UVERBS_OBJECT_FLOW, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY)); + +ADD_UVERBS_METHODS(mlx5_ib_fs, + UVERBS_OBJECT_FLOW, + &UVERBS_METHOD(MLX5_IB_METHOD_CREATE_FLOW), + &UVERBS_METHOD(MLX5_IB_METHOD_DESTROY_FLOW)); + +DECLARE_UVERBS_NAMED_METHOD( + MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_MODIFY_HEADER_HANDLE, + UVERBS_OBJECT_FLOW_ACTION, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_MODIFY_HEADER_ACTIONS_PRM, + UVERBS_ATTR_MIN_SIZE(MLX5_UN_SZ_BYTES( + set_add_copy_action_in_auto)), + UA_MANDATORY, + UA_ALLOC_AND_COPY), + UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_CREATE_MODIFY_HEADER_FT_TYPE, + enum mlx5_ib_uapi_flow_table_type, + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_METHOD( + MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_HANDLE, + UVERBS_OBJECT_FLOW_ACTION, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_DATA_BUF, + UVERBS_ATTR_MIN_SIZE(1), + UA_ALLOC_AND_COPY, + UA_OPTIONAL), + UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_TYPE, + enum mlx5_ib_uapi_flow_action_packet_reformat_type, + UA_MANDATORY), + UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_CREATE_PACKET_REFORMAT_FT_TYPE, + enum mlx5_ib_uapi_flow_table_type, + UA_MANDATORY)); + +ADD_UVERBS_METHODS( + mlx5_ib_flow_actions, + UVERBS_OBJECT_FLOW_ACTION, + &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER), + &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT)); + +DECLARE_UVERBS_NAMED_METHOD( + MLX5_IB_METHOD_FLOW_MATCHER_CREATE, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_FLOW_MATCHER_CREATE_HANDLE, + MLX5_IB_OBJECT_FLOW_MATCHER, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN( + MLX5_IB_ATTR_FLOW_MATCHER_MATCH_MASK, + UVERBS_ATTR_SIZE(1, sizeof(struct mlx5_ib_match_params)), + UA_MANDATORY), + UVERBS_ATTR_ENUM_IN(MLX5_IB_ATTR_FLOW_MATCHER_FLOW_TYPE, + mlx5_ib_flow_type, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_FLOW_MATCHER_MATCH_CRITERIA, + UVERBS_ATTR_TYPE(u8), + UA_MANDATORY), + UVERBS_ATTR_FLAGS_IN(MLX5_IB_ATTR_FLOW_MATCHER_FLOW_FLAGS, + enum ib_flow_flags, + UA_OPTIONAL), + UVERBS_ATTR_CONST_IN(MLX5_IB_ATTR_FLOW_MATCHER_FT_TYPE, + enum mlx5_ib_uapi_flow_table_type, + UA_OPTIONAL)); + +DECLARE_UVERBS_NAMED_METHOD_DESTROY( + MLX5_IB_METHOD_FLOW_MATCHER_DESTROY, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_FLOW_MATCHER_DESTROY_HANDLE, + MLX5_IB_OBJECT_FLOW_MATCHER, + UVERBS_ACCESS_DESTROY, + UA_MANDATORY)); + +DECLARE_UVERBS_NAMED_OBJECT(MLX5_IB_OBJECT_FLOW_MATCHER, + UVERBS_TYPE_ALLOC_IDR(flow_matcher_cleanup), + &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_MATCHER_CREATE), + &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_MATCHER_DESTROY)); + +const struct uapi_definition mlx5_ib_flow_defs[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED( + MLX5_IB_OBJECT_FLOW_MATCHER), + UAPI_DEF_CHAIN_OBJ_TREE( + UVERBS_OBJECT_FLOW, + &mlx5_ib_fs), + UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION, + &mlx5_ib_flow_actions), + {}, +}; + +static const struct ib_device_ops flow_ops = { + .create_flow = mlx5_ib_create_flow, + .destroy_flow = mlx5_ib_destroy_flow, + .destroy_flow_action = mlx5_ib_destroy_flow_action, +}; + +static const struct ib_device_ops flow_ipsec_ops = { + .create_flow_action_esp = mlx5_ib_create_flow_action_esp, + .modify_flow_action_esp = mlx5_ib_modify_flow_action_esp, +}; + +int mlx5_ib_fs_init(struct mlx5_ib_dev *dev) +{ + dev->flow_db = kzalloc(sizeof(*dev->flow_db), GFP_KERNEL); + + if (!dev->flow_db) + return -ENOMEM; + + mutex_init(&dev->flow_db->lock); + + ib_set_device_ops(&dev->ib_dev, &flow_ops); + if (mlx5_accel_ipsec_device_caps(dev->mdev) & + MLX5_ACCEL_IPSEC_CAP_DEVICE) + ib_set_device_ops(&dev->ib_dev, &flow_ipsec_ops); + + return 0; +} diff --git a/drivers/infiniband/hw/mlx5/fs.h b/drivers/infiniband/hw/mlx5/fs.h new file mode 100644 index 000000000000..ad320adaf321 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/fs.h @@ -0,0 +1,29 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + */ + +#ifndef _MLX5_IB_FS_H +#define _MLX5_IB_FS_H + +#include "mlx5_ib.h" + +#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) +int mlx5_ib_fs_init(struct mlx5_ib_dev *dev); +#else +static inline int mlx5_ib_fs_init(struct mlx5_ib_dev *dev) +{ + dev->flow_db = kzalloc(sizeof(*dev->flow_db), GFP_KERNEL); + + if (!dev->flow_db) + return -ENOMEM; + + mutex_init(&dev->flow_db->lock); + return 0; +} +#endif +static inline void mlx5_ib_fs_cleanup(struct mlx5_ib_dev *dev) +{ + kfree(dev->flow_db); +} +#endif /* _MLX5_IB_FS_H */ diff --git a/drivers/infiniband/hw/mlx5/gsi.c b/drivers/infiniband/hw/mlx5/gsi.c index 1ae6fd95acaa..40d418153891 100644 --- a/drivers/infiniband/hw/mlx5/gsi.c +++ b/drivers/infiniband/hw/mlx5/gsi.c @@ -119,17 +119,15 @@ struct ib_qp *mlx5_ib_gsi_create_qp(struct ib_pd *pd, struct mlx5_ib_gsi_qp *gsi; struct ib_qp_init_attr hw_init_attr = *init_attr; const u8 port_num = init_attr->port_num; - const int num_pkeys = pd->device->attrs.max_pkeys; - const int num_qps = mlx5_ib_deth_sqpn_cap(dev) ? num_pkeys : 0; + int num_qps = 0; int ret; - mlx5_ib_dbg(dev, "creating GSI QP\n"); - - if (port_num > ARRAY_SIZE(dev->devr.ports) || port_num < 1) { - mlx5_ib_warn(dev, - "invalid port number %d during GSI QP creation\n", - port_num); - return ERR_PTR(-EINVAL); + if (mlx5_ib_deth_sqpn_cap(dev)) { + if (MLX5_CAP_GEN(dev->mdev, + port_type) == MLX5_CAP_PORT_TYPE_IB) + num_qps = pd->device->attrs.max_pkeys; + else if (dev->lag_active) + num_qps = MLX5_MAX_PORTS; } gsi = kzalloc(sizeof(*gsi), GFP_KERNEL); @@ -270,7 +268,7 @@ static struct ib_qp *create_gsi_ud_qp(struct mlx5_ib_gsi_qp *gsi) } static int modify_to_rts(struct mlx5_ib_gsi_qp *gsi, struct ib_qp *qp, - u16 qp_index) + u16 pkey_index) { struct mlx5_ib_dev *dev = to_mdev(qp->device); struct ib_qp_attr attr; @@ -279,7 +277,7 @@ static int modify_to_rts(struct mlx5_ib_gsi_qp *gsi, struct ib_qp *qp, mask = IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_QKEY | IB_QP_PORT; attr.qp_state = IB_QPS_INIT; - attr.pkey_index = qp_index; + attr.pkey_index = pkey_index; attr.qkey = IB_QP1_QKEY; attr.port_num = gsi->port_num; ret = ib_modify_qp(qp, &attr, mask); @@ -313,12 +311,17 @@ static void setup_qp(struct mlx5_ib_gsi_qp *gsi, u16 qp_index) { struct ib_device *device = gsi->rx_qp->device; struct mlx5_ib_dev *dev = to_mdev(device); + int pkey_index = qp_index; + struct mlx5_ib_qp *mqp; struct ib_qp *qp; unsigned long flags; u16 pkey; int ret; - ret = ib_query_pkey(device, gsi->port_num, qp_index, &pkey); + if (MLX5_CAP_GEN(dev->mdev, port_type) != MLX5_CAP_PORT_TYPE_IB) + pkey_index = 0; + + ret = ib_query_pkey(device, gsi->port_num, pkey_index, &pkey); if (ret) { mlx5_ib_warn(dev, "unable to read P_Key at port %d, index %d\n", gsi->port_num, qp_index); @@ -347,7 +350,10 @@ static void setup_qp(struct mlx5_ib_gsi_qp *gsi, u16 qp_index) return; } - ret = modify_to_rts(gsi, qp, qp_index); + mqp = to_mqp(qp); + if (dev->lag_active) + mqp->gsi_lag_port = qp_index + 1; + ret = modify_to_rts(gsi, qp, pkey_index); if (ret) goto err_destroy_qp; @@ -466,11 +472,15 @@ static int mlx5_ib_gsi_silent_drop(struct mlx5_ib_gsi_qp *gsi, static struct ib_qp *get_tx_qp(struct mlx5_ib_gsi_qp *gsi, struct ib_ud_wr *wr) { struct mlx5_ib_dev *dev = to_mdev(gsi->rx_qp->device); + struct mlx5_ib_ah *ah = to_mah(wr->ah); int qp_index = wr->pkey_index; - if (!mlx5_ib_deth_sqpn_cap(dev)) + if (!gsi->num_qps) return gsi->rx_qp; + if (dev->lag_active && ah->xmit_port) + qp_index = ah->xmit_port - 1; + if (qp_index >= gsi->num_qps) return NULL; diff --git a/drivers/infiniband/hw/mlx5/ib_rep.h b/drivers/infiniband/hw/mlx5/ib_rep.h index 3b6750cba796..5b30d3fa8f8d 100644 --- a/drivers/infiniband/hw/mlx5/ib_rep.h +++ b/drivers/infiniband/hw/mlx5/ib_rep.h @@ -9,9 +9,9 @@ #include <linux/mlx5/eswitch.h> #include "mlx5_ib.h" -#ifdef CONFIG_MLX5_ESWITCH extern const struct mlx5_ib_profile raw_eth_profile; +#ifdef CONFIG_MLX5_ESWITCH u8 mlx5_ib_eswitch_mode(struct mlx5_eswitch *esw); struct mlx5_ib_dev *mlx5_ib_get_rep_ibdev(struct mlx5_eswitch *esw, u16 vport_num); diff --git a/drivers/infiniband/hw/mlx5/ib_virt.c b/drivers/infiniband/hw/mlx5/ib_virt.c index b61165359954..46b2d370fb3f 100644 --- a/drivers/infiniband/hw/mlx5/ib_virt.c +++ b/drivers/infiniband/hw/mlx5/ib_virt.c @@ -134,7 +134,7 @@ int mlx5_ib_get_vf_stats(struct ib_device *device, int vf, if (!out) return -ENOMEM; - err = mlx5_core_query_vport_counter(mdev, true, vf, port, out, out_sz); + err = mlx5_core_query_vport_counter(mdev, true, vf, port, out); if (err) goto ex; diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c index 14e0c17de6a9..9bb9bb058932 100644 --- a/drivers/infiniband/hw/mlx5/mad.c +++ b/drivers/infiniband/hw/mlx5/mad.c @@ -30,7 +30,6 @@ * SOFTWARE. */ -#include <linux/mlx5/cmd.h> #include <linux/mlx5/vport.h> #include <rdma/ib_mad.h> #include <rdma/ib_smi.h> @@ -188,8 +187,8 @@ static int process_pma_cmd(struct mlx5_ib_dev *dev, u8 port_num, goto done; } - err = mlx5_core_query_vport_counter(mdev, 0, 0, - mdev_port_num, out_cnt, sz); + err = mlx5_core_query_vport_counter(mdev, 0, 0, mdev_port_num, + out_cnt); if (!err) pma_cnt_ext_assign(pma_cnt_ext, out_cnt); } else { @@ -251,9 +250,8 @@ int mlx5_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, if (MLX5_CAP_GEN(dev->mdev, vport_counters) && method == IB_MGMT_METHOD_GET) return process_pma_cmd(dev, port_num, in, out); - /* fallthrough */ + fallthrough; case MLX5_IB_VENDOR_CLASS1: - /* fallthrough */ case MLX5_IB_VENDOR_CLASS2: case IB_MGMT_CLASS_CONG_MGMT: { if (method != IB_MGMT_METHOD_GET && diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 6679756506e6..d60d63221b14 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -1,33 +1,6 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* - * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved. - * - * This software is available to you under a choice of one of two - * licenses. You may choose to be licensed under the terms of the GNU - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. */ #include <linux/debugfs.h> @@ -53,13 +26,19 @@ #include <linux/list.h> #include <rdma/ib_smi.h> #include <rdma/ib_umem.h> +#include <rdma/lag.h> #include <linux/in.h> #include <linux/etherdevice.h> #include "mlx5_ib.h" #include "ib_rep.h" #include "cmd.h" +#include "devx.h" +#include "fs.h" #include "srq.h" -#include <linux/mlx5/fs_helpers.h> +#include "qp.h" +#include "wr.h" +#include "restrack.h" +#include "counters.h" #include <linux/mlx5/accel.h> #include <rdma/uverbs_std_types.h> #include <rdma/mlx5_user_ioctl_verbs.h> @@ -69,17 +48,10 @@ #define UVERBS_MODULE_NAME mlx5_ib #include <rdma/uverbs_named_ioctl.h> -#define DRIVER_NAME "mlx5_ib" -#define DRIVER_VERSION "5.0-0" - MODULE_AUTHOR("Eli Cohen <eli@mellanox.com>"); -MODULE_DESCRIPTION("Mellanox Connect-IB HCA IB driver"); +MODULE_DESCRIPTION("Mellanox 5th generation network adapters (ConnectX series) IB driver"); MODULE_LICENSE("Dual BSD/GPL"); -static char mlx5_version[] = - DRIVER_NAME ": Mellanox Connect-IB Infiniband driver v" - DRIVER_VERSION "\n"; - struct mlx5_ib_event_work { struct work_struct work; union { @@ -315,9 +287,6 @@ struct mlx5_core_dev *mlx5_ib_get_native_port_mdev(struct mlx5_ib_dev *ibdev, *native_port_num = 1; port = &ibdev->port[ib_port_num - 1]; - if (!port) - return NULL; - spin_lock(&port->mp.mpi_lock); mpi = ibdev->port[ib_port_num - 1].mp.mpi; if (mpi && !mpi->unaffiliate) { @@ -515,7 +484,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num, mdev_port_num); if (err) goto out; - ext = MLX5_CAP_PCAM_FEATURE(dev->mdev, ptys_extended_ethernet); + ext = !!MLX5_GET_ETH_PROTO(ptys_reg, out, true, eth_proto_capability); eth_prot_oper = MLX5_GET_ETH_PROTO(ptys_reg, out, ext, eth_proto_oper); props->active_width = IB_WIDTH_4X; @@ -627,8 +596,8 @@ static int mlx5_ib_del_gid(const struct ib_gid_attr *attr, attr->index, NULL, NULL); } -__be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, - const struct ib_gid_attr *attr) +__be16 mlx5_get_roce_udp_sport_min(const struct mlx5_ib_dev *dev, + const struct ib_gid_attr *attr) { if (attr->gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP) return 0; @@ -1003,7 +972,6 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, props->max_mcast_qp_attach = MLX5_CAP_GEN(mdev, max_qp_mcg); props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * props->max_mcast_grp; - props->max_map_per_fmr = INT_MAX; /* no limit in ConnectIB */ props->max_ah = INT_MAX; props->hca_core_clock = MLX5_CAP_GEN(mdev, device_frequency_khz); props->timestamp_mask = 0x7FFFFFFFFFFFFFFFULL; @@ -1770,6 +1738,92 @@ static void mlx5_ib_dealloc_transport_domain(struct mlx5_ib_dev *dev, u32 tdn, mlx5_ib_disable_lb(dev, true, false); } +static int set_ucontext_resp(struct ib_ucontext *uctx, + struct mlx5_ib_alloc_ucontext_resp *resp) +{ + struct ib_device *ibdev = uctx->device; + struct mlx5_ib_dev *dev = to_mdev(ibdev); + struct mlx5_ib_ucontext *context = to_mucontext(uctx); + struct mlx5_bfreg_info *bfregi = &context->bfregi; + int err; + + if (MLX5_CAP_GEN(dev->mdev, dump_fill_mkey)) { + err = mlx5_cmd_dump_fill_mkey(dev->mdev, + &resp->dump_fill_mkey); + if (err) + return err; + resp->comp_mask |= + MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_DUMP_FILL_MKEY; + } + + resp->qp_tab_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp); + if (dev->wc_support) + resp->bf_reg_size = 1 << MLX5_CAP_GEN(dev->mdev, + log_bf_reg_size); + resp->cache_line_size = cache_line_size(); + resp->max_sq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq); + resp->max_rq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq); + resp->max_send_wqebb = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz); + resp->max_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz); + resp->max_srq_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz); + resp->cqe_version = context->cqe_version; + resp->log_uar_size = MLX5_CAP_GEN(dev->mdev, uar_4k) ? + MLX5_ADAPTER_PAGE_SHIFT : PAGE_SHIFT; + resp->num_uars_per_page = MLX5_CAP_GEN(dev->mdev, uar_4k) ? + MLX5_CAP_GEN(dev->mdev, + num_of_uars_per_page) : 1; + + if (mlx5_accel_ipsec_device_caps(dev->mdev) & + MLX5_ACCEL_IPSEC_CAP_DEVICE) { + if (mlx5_get_flow_namespace(dev->mdev, + MLX5_FLOW_NAMESPACE_EGRESS)) + resp->flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM; + if (mlx5_accel_ipsec_device_caps(dev->mdev) & + MLX5_ACCEL_IPSEC_CAP_REQUIRED_METADATA) + resp->flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_REQ_METADATA; + if (MLX5_CAP_FLOWTABLE(dev->mdev, flow_table_properties_nic_receive.ft_field_support.outer_esp_spi)) + resp->flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_SPI_STEERING; + if (mlx5_accel_ipsec_device_caps(dev->mdev) & + MLX5_ACCEL_IPSEC_CAP_TX_IV_IS_ESN) + resp->flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_TX_IV_IS_ESN; + /* MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_FULL_OFFLOAD is currently always 0 */ + } + + resp->tot_bfregs = bfregi->lib_uar_dyn ? 0 : + bfregi->total_num_bfregs - bfregi->num_dyn_bfregs; + resp->num_ports = dev->num_ports; + resp->cmds_supp_uhw |= MLX5_USER_CMDS_SUPP_UHW_QUERY_DEVICE | + MLX5_USER_CMDS_SUPP_UHW_CREATE_AH; + + if (mlx5_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET) { + mlx5_query_min_inline(dev->mdev, &resp->eth_min_inline); + resp->eth_min_inline++; + } + + if (dev->mdev->clock_info) + resp->clock_info_versions = BIT(MLX5_IB_CLOCK_INFO_V1); + + /* + * 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 + * pretend we don't support reading the HCA's core clock. This is also + * forced by mmap function. + */ + if (PAGE_SIZE <= 4096) { + resp->comp_mask |= + MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_CORE_CLOCK_OFFSET; + resp->hca_core_clock_offset = + offsetof(struct mlx5_init_seg, + internal_timer_h) % PAGE_SIZE; + } + + if (MLX5_CAP_GEN(dev->mdev, ece_support)) + resp->comp_mask |= MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_ECE; + + resp->num_dyn_bfregs = bfregi->num_dyn_bfregs; + return 0; +} + static int mlx5_ib_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata) { @@ -1777,14 +1831,12 @@ static int mlx5_ib_alloc_ucontext(struct ib_ucontext *uctx, struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_ib_alloc_ucontext_req_v2 req = {}; struct mlx5_ib_alloc_ucontext_resp resp = {}; - struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_ucontext *context = to_mucontext(uctx); struct mlx5_bfreg_info *bfregi; int ver; int err; size_t min_req_v2 = offsetof(struct mlx5_ib_alloc_ucontext_req_v2, max_cqe_version); - u32 dump_fill_mkey; bool lib_uar_4k; bool lib_uar_dyn; @@ -1813,37 +1865,6 @@ static int mlx5_ib_alloc_ucontext(struct ib_ucontext *uctx, if (req.num_low_latency_bfregs > req.total_num_bfregs - 1) return -EINVAL; - resp.qp_tab_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp); - if (dev->wc_support) - resp.bf_reg_size = 1 << MLX5_CAP_GEN(dev->mdev, log_bf_reg_size); - resp.cache_line_size = cache_line_size(); - resp.max_sq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq); - resp.max_rq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq); - resp.max_send_wqebb = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz); - resp.max_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz); - resp.max_srq_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz); - resp.cqe_version = min_t(__u8, - (__u8)MLX5_CAP_GEN(dev->mdev, cqe_version), - req.max_cqe_version); - resp.log_uar_size = MLX5_CAP_GEN(dev->mdev, uar_4k) ? - MLX5_ADAPTER_PAGE_SHIFT : PAGE_SHIFT; - resp.num_uars_per_page = MLX5_CAP_GEN(dev->mdev, uar_4k) ? - MLX5_CAP_GEN(dev->mdev, num_of_uars_per_page) : 1; - resp.response_length = min(offsetof(typeof(resp), response_length) + - sizeof(resp.response_length), udata->outlen); - - if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_DEVICE) { - if (mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_EGRESS)) - resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM; - if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_REQUIRED_METADATA) - resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_REQ_METADATA; - if (MLX5_CAP_FLOWTABLE(dev->mdev, flow_table_properties_nic_receive.ft_field_support.outer_esp_spi)) - resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_SPI_STEERING; - if (mlx5_accel_ipsec_device_caps(dev->mdev) & MLX5_ACCEL_IPSEC_CAP_TX_IV_IS_ESN) - resp.flow_action_flags |= MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_TX_IV_IS_ESN; - /* MLX5_USER_ALLOC_UCONTEXT_FLOW_ACTION_FLAGS_ESP_AES_GCM_FULL_OFFLOAD is currently always 0 */ - } - lib_uar_4k = req.lib_caps & MLX5_LIB_CAP_4K_UAR; lib_uar_dyn = req.lib_caps & MLX5_LIB_CAP_DYN_UAR; bfregi = &context->bfregi; @@ -1892,88 +1913,28 @@ uar_done: if (err) goto out_devx; - if (MLX5_CAP_GEN(dev->mdev, dump_fill_mkey)) { - err = mlx5_cmd_dump_fill_mkey(dev->mdev, &dump_fill_mkey); - if (err) - goto out_mdev; - } - INIT_LIST_HEAD(&context->db_page_list); mutex_init(&context->db_page_mutex); - resp.tot_bfregs = lib_uar_dyn ? 0 : req.total_num_bfregs; - resp.num_ports = dev->num_ports; - - if (offsetofend(typeof(resp), cqe_version) <= udata->outlen) - resp.response_length += sizeof(resp.cqe_version); - - if (offsetofend(typeof(resp), cmds_supp_uhw) <= udata->outlen) { - resp.cmds_supp_uhw |= MLX5_USER_CMDS_SUPP_UHW_QUERY_DEVICE | - MLX5_USER_CMDS_SUPP_UHW_CREATE_AH; - resp.response_length += sizeof(resp.cmds_supp_uhw); - } - - if (offsetofend(typeof(resp), eth_min_inline) <= udata->outlen) { - if (mlx5_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET) { - mlx5_query_min_inline(dev->mdev, &resp.eth_min_inline); - resp.eth_min_inline++; - } - resp.response_length += sizeof(resp.eth_min_inline); - } - - if (offsetofend(typeof(resp), clock_info_versions) <= udata->outlen) { - if (mdev->clock_info) - resp.clock_info_versions = BIT(MLX5_IB_CLOCK_INFO_V1); - resp.response_length += sizeof(resp.clock_info_versions); - } - - /* - * 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 - * pretend we don't support reading the HCA's core clock. This is also - * forced by mmap function. - */ - if (offsetofend(typeof(resp), hca_core_clock_offset) <= udata->outlen) { - if (PAGE_SIZE <= 4096) { - resp.comp_mask |= - MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_CORE_CLOCK_OFFSET; - resp.hca_core_clock_offset = - offsetof(struct mlx5_init_seg, internal_timer_h) % PAGE_SIZE; - } - resp.response_length += sizeof(resp.hca_core_clock_offset); - } - - if (offsetofend(typeof(resp), log_uar_size) <= udata->outlen) - resp.response_length += sizeof(resp.log_uar_size); - - if (offsetofend(typeof(resp), num_uars_per_page) <= udata->outlen) - resp.response_length += sizeof(resp.num_uars_per_page); - - if (offsetofend(typeof(resp), num_dyn_bfregs) <= udata->outlen) { - resp.num_dyn_bfregs = bfregi->num_dyn_bfregs; - resp.response_length += sizeof(resp.num_dyn_bfregs); - } + context->cqe_version = min_t(__u8, + (__u8)MLX5_CAP_GEN(dev->mdev, cqe_version), + req.max_cqe_version); - if (offsetofend(typeof(resp), dump_fill_mkey) <= udata->outlen) { - if (MLX5_CAP_GEN(dev->mdev, dump_fill_mkey)) { - resp.dump_fill_mkey = dump_fill_mkey; - resp.comp_mask |= - MLX5_IB_ALLOC_UCONTEXT_RESP_MASK_DUMP_FILL_MKEY; - } - resp.response_length += sizeof(resp.dump_fill_mkey); - } + err = set_ucontext_resp(uctx, &resp); + if (err) + goto out_mdev; + resp.response_length = min(udata->outlen, sizeof(resp)); err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) goto out_mdev; bfregi->ver = ver; bfregi->num_low_latency_bfregs = req.num_low_latency_bfregs; - context->cqe_version = resp.cqe_version; context->lib_caps = req.lib_caps; print_lib_caps(dev, context->lib_caps); - if (dev->lag_active) { + if (mlx5_ib_lag_should_assign_affinity(dev)) { u8 port = mlx5_core_native_port_num(dev->mdev) - 1; atomic_set(&context->tx_port_affinity, @@ -2002,6 +1963,29 @@ out_ctx: return err; } +static int mlx5_ib_query_ucontext(struct ib_ucontext *ibcontext, + struct uverbs_attr_bundle *attrs) +{ + struct mlx5_ib_alloc_ucontext_resp uctx_resp = {}; + int ret; + + ret = set_ucontext_resp(ibcontext, &uctx_resp); + if (ret) + return ret; + + uctx_resp.response_length = + min_t(size_t, + uverbs_attr_get_len(attrs, + MLX5_IB_ATTR_QUERY_CONTEXT_RESP_UCTX), + sizeof(uctx_resp)); + + ret = uverbs_copy_to_struct_or_zero(attrs, + MLX5_IB_ATTR_QUERY_CONTEXT_RESP_UCTX, + &uctx_resp, + sizeof(uctx_resp)); + return ret; +} + static void mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext) { struct mlx5_ib_ucontext *context = to_mucontext(ibcontext); @@ -2443,7 +2427,7 @@ static int handle_alloc_dm_sw_icm(struct ib_ucontext *ctx, act_size = roundup_pow_of_two(act_size); dm->size = act_size; - err = mlx5_dm_sw_icm_alloc(dev, type, act_size, + err = mlx5_dm_sw_icm_alloc(dev, type, act_size, attr->alignment, to_mucontext(ctx)->devx_uid, &dm->dev_addr, &dm->icm_dm.obj_id); if (err) @@ -2560,7 +2544,7 @@ static int mlx5_ib_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) struct mlx5_ib_alloc_pd_resp resp; int err; u32 out[MLX5_ST_SZ_DW(alloc_pd_out)] = {}; - u32 in[MLX5_ST_SZ_DW(alloc_pd_in)] = {}; + u32 in[MLX5_ST_SZ_DW(alloc_pd_in)] = {}; u16 uid = 0; struct mlx5_ib_ucontext *context = rdma_udata_to_drv_context( udata, struct mlx5_ib_ucontext, ibucontext); @@ -2568,8 +2552,7 @@ static int mlx5_ib_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) uid = context ? context->devx_uid : 0; MLX5_SET(alloc_pd_in, in, opcode, MLX5_CMD_OP_ALLOC_PD); MLX5_SET(alloc_pd_in, in, uid, uid); - err = mlx5_cmd_exec(to_mdev(ibdev)->mdev, in, sizeof(in), - out, sizeof(out)); + err = mlx5_cmd_exec_inout(to_mdev(ibdev)->mdev, alloc_pd, in, out); if (err) return err; @@ -2594,1848 +2577,6 @@ static void mlx5_ib_dealloc_pd(struct ib_pd *pd, struct ib_udata *udata) mlx5_cmd_dealloc_pd(mdev->mdev, mpd->pdn, mpd->uid); } -enum { - MATCH_CRITERIA_ENABLE_OUTER_BIT, - MATCH_CRITERIA_ENABLE_MISC_BIT, - MATCH_CRITERIA_ENABLE_INNER_BIT, - MATCH_CRITERIA_ENABLE_MISC2_BIT -}; - -#define HEADER_IS_ZERO(match_criteria, headers) \ - !(memchr_inv(MLX5_ADDR_OF(fte_match_param, match_criteria, headers), \ - 0, MLX5_FLD_SZ_BYTES(fte_match_param, headers))) \ - -static u8 get_match_criteria_enable(u32 *match_criteria) -{ - u8 match_criteria_enable; - - match_criteria_enable = - (!HEADER_IS_ZERO(match_criteria, outer_headers)) << - MATCH_CRITERIA_ENABLE_OUTER_BIT; - match_criteria_enable |= - (!HEADER_IS_ZERO(match_criteria, misc_parameters)) << - MATCH_CRITERIA_ENABLE_MISC_BIT; - match_criteria_enable |= - (!HEADER_IS_ZERO(match_criteria, inner_headers)) << - MATCH_CRITERIA_ENABLE_INNER_BIT; - match_criteria_enable |= - (!HEADER_IS_ZERO(match_criteria, misc_parameters_2)) << - MATCH_CRITERIA_ENABLE_MISC2_BIT; - - return match_criteria_enable; -} - -static int set_proto(void *outer_c, void *outer_v, u8 mask, u8 val) -{ - u8 entry_mask; - u8 entry_val; - int err = 0; - - if (!mask) - goto out; - - entry_mask = MLX5_GET(fte_match_set_lyr_2_4, outer_c, - ip_protocol); - entry_val = MLX5_GET(fte_match_set_lyr_2_4, outer_v, - ip_protocol); - if (!entry_mask) { - MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_protocol, mask); - MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_protocol, val); - goto out; - } - /* Don't override existing ip protocol */ - if (mask != entry_mask || val != entry_val) - err = -EINVAL; -out: - return err; -} - -static void set_flow_label(void *misc_c, void *misc_v, u32 mask, u32 val, - bool inner) -{ - if (inner) { - MLX5_SET(fte_match_set_misc, - misc_c, inner_ipv6_flow_label, mask); - MLX5_SET(fte_match_set_misc, - misc_v, inner_ipv6_flow_label, val); - } else { - MLX5_SET(fte_match_set_misc, - misc_c, outer_ipv6_flow_label, mask); - MLX5_SET(fte_match_set_misc, - misc_v, outer_ipv6_flow_label, val); - } -} - -static void set_tos(void *outer_c, void *outer_v, u8 mask, u8 val) -{ - MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_ecn, mask); - MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_ecn, val); - MLX5_SET(fte_match_set_lyr_2_4, outer_c, ip_dscp, mask >> 2); - MLX5_SET(fte_match_set_lyr_2_4, outer_v, ip_dscp, val >> 2); -} - -static int check_mpls_supp_fields(u32 field_support, const __be32 *set_mask) -{ - if (MLX5_GET(fte_match_mpls, set_mask, mpls_label) && - !(field_support & MLX5_FIELD_SUPPORT_MPLS_LABEL)) - return -EOPNOTSUPP; - - if (MLX5_GET(fte_match_mpls, set_mask, mpls_exp) && - !(field_support & MLX5_FIELD_SUPPORT_MPLS_EXP)) - return -EOPNOTSUPP; - - if (MLX5_GET(fte_match_mpls, set_mask, mpls_s_bos) && - !(field_support & MLX5_FIELD_SUPPORT_MPLS_S_BOS)) - return -EOPNOTSUPP; - - if (MLX5_GET(fte_match_mpls, set_mask, mpls_ttl) && - !(field_support & MLX5_FIELD_SUPPORT_MPLS_TTL)) - return -EOPNOTSUPP; - - return 0; -} - -#define LAST_ETH_FIELD vlan_tag -#define LAST_IB_FIELD sl -#define LAST_IPV4_FIELD tos -#define LAST_IPV6_FIELD traffic_class -#define LAST_TCP_UDP_FIELD src_port -#define LAST_TUNNEL_FIELD tunnel_id -#define LAST_FLOW_TAG_FIELD tag_id -#define LAST_DROP_FIELD size -#define LAST_COUNTERS_FIELD counters - -/* Field is the last supported field */ -#define FIELDS_NOT_SUPPORTED(filter, field)\ - memchr_inv((void *)&filter.field +\ - sizeof(filter.field), 0,\ - sizeof(filter) -\ - offsetof(typeof(filter), field) -\ - sizeof(filter.field)) - -int parse_flow_flow_action(struct mlx5_ib_flow_action *maction, - bool is_egress, - struct mlx5_flow_act *action) -{ - - switch (maction->ib_action.type) { - case IB_FLOW_ACTION_ESP: - if (action->action & (MLX5_FLOW_CONTEXT_ACTION_ENCRYPT | - MLX5_FLOW_CONTEXT_ACTION_DECRYPT)) - return -EINVAL; - /* Currently only AES_GCM keymat is supported by the driver */ - action->esp_id = (uintptr_t)maction->esp_aes_gcm.ctx; - action->action |= is_egress ? - MLX5_FLOW_CONTEXT_ACTION_ENCRYPT : - MLX5_FLOW_CONTEXT_ACTION_DECRYPT; - return 0; - case IB_FLOW_ACTION_UNSPECIFIED: - if (maction->flow_action_raw.sub_type == - MLX5_IB_FLOW_ACTION_MODIFY_HEADER) { - if (action->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) - return -EINVAL; - action->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR; - action->modify_hdr = - maction->flow_action_raw.modify_hdr; - return 0; - } - if (maction->flow_action_raw.sub_type == - MLX5_IB_FLOW_ACTION_DECAP) { - if (action->action & MLX5_FLOW_CONTEXT_ACTION_DECAP) - return -EINVAL; - action->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP; - return 0; - } - if (maction->flow_action_raw.sub_type == - MLX5_IB_FLOW_ACTION_PACKET_REFORMAT) { - if (action->action & - MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) - return -EINVAL; - action->action |= - MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT; - action->pkt_reformat = - maction->flow_action_raw.pkt_reformat; - return 0; - } - /* fall through */ - default: - return -EOPNOTSUPP; - } -} - -static int parse_flow_attr(struct mlx5_core_dev *mdev, - struct mlx5_flow_spec *spec, - const union ib_flow_spec *ib_spec, - const struct ib_flow_attr *flow_attr, - struct mlx5_flow_act *action, u32 prev_type) -{ - struct mlx5_flow_context *flow_context = &spec->flow_context; - u32 *match_c = spec->match_criteria; - u32 *match_v = spec->match_value; - void *misc_params_c = MLX5_ADDR_OF(fte_match_param, match_c, - misc_parameters); - void *misc_params_v = MLX5_ADDR_OF(fte_match_param, match_v, - misc_parameters); - void *misc_params2_c = MLX5_ADDR_OF(fte_match_param, match_c, - misc_parameters_2); - void *misc_params2_v = MLX5_ADDR_OF(fte_match_param, match_v, - misc_parameters_2); - void *headers_c; - void *headers_v; - int match_ipv; - int ret; - - if (ib_spec->type & IB_FLOW_SPEC_INNER) { - headers_c = MLX5_ADDR_OF(fte_match_param, match_c, - inner_headers); - headers_v = MLX5_ADDR_OF(fte_match_param, match_v, - inner_headers); - match_ipv = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.inner_ip_version); - } else { - headers_c = MLX5_ADDR_OF(fte_match_param, match_c, - outer_headers); - headers_v = MLX5_ADDR_OF(fte_match_param, match_v, - outer_headers); - match_ipv = MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.outer_ip_version); - } - - switch (ib_spec->type & ~IB_FLOW_SPEC_INNER) { - case IB_FLOW_SPEC_ETH: - if (FIELDS_NOT_SUPPORTED(ib_spec->eth.mask, LAST_ETH_FIELD)) - return -EOPNOTSUPP; - - ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, - dmac_47_16), - ib_spec->eth.mask.dst_mac); - ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, - dmac_47_16), - ib_spec->eth.val.dst_mac); - - ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, - smac_47_16), - ib_spec->eth.mask.src_mac); - ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, - smac_47_16), - ib_spec->eth.val.src_mac); - - if (ib_spec->eth.mask.vlan_tag) { - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - cvlan_tag, 1); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - cvlan_tag, 1); - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - first_vid, ntohs(ib_spec->eth.mask.vlan_tag)); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - first_vid, ntohs(ib_spec->eth.val.vlan_tag)); - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - first_cfi, - ntohs(ib_spec->eth.mask.vlan_tag) >> 12); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - first_cfi, - ntohs(ib_spec->eth.val.vlan_tag) >> 12); - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - first_prio, - ntohs(ib_spec->eth.mask.vlan_tag) >> 13); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - first_prio, - ntohs(ib_spec->eth.val.vlan_tag) >> 13); - } - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - ethertype, ntohs(ib_spec->eth.mask.ether_type)); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - ethertype, ntohs(ib_spec->eth.val.ether_type)); - break; - case IB_FLOW_SPEC_IPV4: - if (FIELDS_NOT_SUPPORTED(ib_spec->ipv4.mask, LAST_IPV4_FIELD)) - return -EOPNOTSUPP; - - if (match_ipv) { - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - ip_version, 0xf); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - ip_version, MLX5_FS_IPV4_VERSION); - } else { - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - ethertype, 0xffff); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - ethertype, ETH_P_IP); - } - - memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, - src_ipv4_src_ipv6.ipv4_layout.ipv4), - &ib_spec->ipv4.mask.src_ip, - sizeof(ib_spec->ipv4.mask.src_ip)); - memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, - src_ipv4_src_ipv6.ipv4_layout.ipv4), - &ib_spec->ipv4.val.src_ip, - sizeof(ib_spec->ipv4.val.src_ip)); - memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c, - dst_ipv4_dst_ipv6.ipv4_layout.ipv4), - &ib_spec->ipv4.mask.dst_ip, - sizeof(ib_spec->ipv4.mask.dst_ip)); - memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v, - dst_ipv4_dst_ipv6.ipv4_layout.ipv4), - &ib_spec->ipv4.val.dst_ip, - sizeof(ib_spec->ipv4.val.dst_ip)); - - set_tos(headers_c, headers_v, - ib_spec->ipv4.mask.tos, ib_spec->ipv4.val.tos); - - if (set_proto(headers_c, headers_v, - ib_spec->ipv4.mask.proto, - ib_spec->ipv4.val.proto)) - return -EINVAL; - break; - case IB_FLOW_SPEC_IPV6: - if (FIELDS_NOT_SUPPORTED(ib_spec->ipv6.mask, LAST_IPV6_FIELD)) - return -EOPNOTSUPP; - - if (match_ipv) { - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - ip_version, 0xf); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - ip_version, MLX5_FS_IPV6_VERSION); - } else { - MLX5_SET(fte_match_set_lyr_2_4, headers_c, - ethertype, 0xffff); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, - ethertype, ETH_P_IPV6); - } - - memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, 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, 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, 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, headers_v, - dst_ipv4_dst_ipv6.ipv6_layout.ipv6), - &ib_spec->ipv6.val.dst_ip, - sizeof(ib_spec->ipv6.val.dst_ip)); - - set_tos(headers_c, headers_v, - ib_spec->ipv6.mask.traffic_class, - ib_spec->ipv6.val.traffic_class); - - if (set_proto(headers_c, headers_v, - ib_spec->ipv6.mask.next_hdr, - ib_spec->ipv6.val.next_hdr)) - return -EINVAL; - - set_flow_label(misc_params_c, misc_params_v, - ntohl(ib_spec->ipv6.mask.flow_label), - ntohl(ib_spec->ipv6.val.flow_label), - ib_spec->type & IB_FLOW_SPEC_INNER); - break; - case IB_FLOW_SPEC_ESP: - if (ib_spec->esp.mask.seq) - return -EOPNOTSUPP; - - MLX5_SET(fte_match_set_misc, misc_params_c, outer_esp_spi, - ntohl(ib_spec->esp.mask.spi)); - MLX5_SET(fte_match_set_misc, misc_params_v, outer_esp_spi, - ntohl(ib_spec->esp.val.spi)); - break; - case IB_FLOW_SPEC_TCP: - if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask, - LAST_TCP_UDP_FIELD)) - return -EOPNOTSUPP; - - if (set_proto(headers_c, headers_v, 0xff, IPPROTO_TCP)) - return -EINVAL; - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_sport, - ntohs(ib_spec->tcp_udp.mask.src_port)); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, tcp_sport, - ntohs(ib_spec->tcp_udp.val.src_port)); - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, tcp_dport, - ntohs(ib_spec->tcp_udp.mask.dst_port)); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, tcp_dport, - ntohs(ib_spec->tcp_udp.val.dst_port)); - break; - case IB_FLOW_SPEC_UDP: - if (FIELDS_NOT_SUPPORTED(ib_spec->tcp_udp.mask, - LAST_TCP_UDP_FIELD)) - return -EOPNOTSUPP; - - if (set_proto(headers_c, headers_v, 0xff, IPPROTO_UDP)) - return -EINVAL; - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_sport, - ntohs(ib_spec->tcp_udp.mask.src_port)); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, udp_sport, - ntohs(ib_spec->tcp_udp.val.src_port)); - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, udp_dport, - ntohs(ib_spec->tcp_udp.mask.dst_port)); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, udp_dport, - ntohs(ib_spec->tcp_udp.val.dst_port)); - break; - case IB_FLOW_SPEC_GRE: - if (ib_spec->gre.mask.c_ks_res0_ver) - return -EOPNOTSUPP; - - if (set_proto(headers_c, headers_v, 0xff, IPPROTO_GRE)) - return -EINVAL; - - MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_protocol, - 0xff); - MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_protocol, - IPPROTO_GRE); - - MLX5_SET(fte_match_set_misc, misc_params_c, gre_protocol, - ntohs(ib_spec->gre.mask.protocol)); - MLX5_SET(fte_match_set_misc, misc_params_v, gre_protocol, - ntohs(ib_spec->gre.val.protocol)); - - memcpy(MLX5_ADDR_OF(fte_match_set_misc, misc_params_c, - gre_key.nvgre.hi), - &ib_spec->gre.mask.key, - sizeof(ib_spec->gre.mask.key)); - memcpy(MLX5_ADDR_OF(fte_match_set_misc, misc_params_v, - gre_key.nvgre.hi), - &ib_spec->gre.val.key, - sizeof(ib_spec->gre.val.key)); - break; - case IB_FLOW_SPEC_MPLS: - switch (prev_type) { - case IB_FLOW_SPEC_UDP: - if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.outer_first_mpls_over_udp), - &ib_spec->mpls.mask.tag)) - return -EOPNOTSUPP; - - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, - outer_first_mpls_over_udp), - &ib_spec->mpls.val.tag, - sizeof(ib_spec->mpls.val.tag)); - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, - outer_first_mpls_over_udp), - &ib_spec->mpls.mask.tag, - sizeof(ib_spec->mpls.mask.tag)); - break; - case IB_FLOW_SPEC_GRE: - if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.outer_first_mpls_over_gre), - &ib_spec->mpls.mask.tag)) - return -EOPNOTSUPP; - - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, - outer_first_mpls_over_gre), - &ib_spec->mpls.val.tag, - sizeof(ib_spec->mpls.val.tag)); - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, - outer_first_mpls_over_gre), - &ib_spec->mpls.mask.tag, - sizeof(ib_spec->mpls.mask.tag)); - break; - default: - if (ib_spec->type & IB_FLOW_SPEC_INNER) { - if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.inner_first_mpls), - &ib_spec->mpls.mask.tag)) - return -EOPNOTSUPP; - - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, - inner_first_mpls), - &ib_spec->mpls.val.tag, - sizeof(ib_spec->mpls.val.tag)); - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, - inner_first_mpls), - &ib_spec->mpls.mask.tag, - sizeof(ib_spec->mpls.mask.tag)); - } else { - if (check_mpls_supp_fields(MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.outer_first_mpls), - &ib_spec->mpls.mask.tag)) - return -EOPNOTSUPP; - - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_v, - outer_first_mpls), - &ib_spec->mpls.val.tag, - sizeof(ib_spec->mpls.val.tag)); - memcpy(MLX5_ADDR_OF(fte_match_set_misc2, misc_params2_c, - outer_first_mpls), - &ib_spec->mpls.mask.tag, - sizeof(ib_spec->mpls.mask.tag)); - } - } - break; - case IB_FLOW_SPEC_VXLAN_TUNNEL: - if (FIELDS_NOT_SUPPORTED(ib_spec->tunnel.mask, - LAST_TUNNEL_FIELD)) - return -EOPNOTSUPP; - - MLX5_SET(fte_match_set_misc, misc_params_c, vxlan_vni, - ntohl(ib_spec->tunnel.mask.tunnel_id)); - MLX5_SET(fte_match_set_misc, misc_params_v, vxlan_vni, - ntohl(ib_spec->tunnel.val.tunnel_id)); - break; - case IB_FLOW_SPEC_ACTION_TAG: - if (FIELDS_NOT_SUPPORTED(ib_spec->flow_tag, - LAST_FLOW_TAG_FIELD)) - return -EOPNOTSUPP; - if (ib_spec->flow_tag.tag_id >= BIT(24)) - return -EINVAL; - - flow_context->flow_tag = ib_spec->flow_tag.tag_id; - flow_context->flags |= FLOW_CONTEXT_HAS_TAG; - break; - case IB_FLOW_SPEC_ACTION_DROP: - if (FIELDS_NOT_SUPPORTED(ib_spec->drop, - LAST_DROP_FIELD)) - return -EOPNOTSUPP; - action->action |= MLX5_FLOW_CONTEXT_ACTION_DROP; - break; - case IB_FLOW_SPEC_ACTION_HANDLE: - ret = parse_flow_flow_action(to_mflow_act(ib_spec->action.act), - flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS, action); - if (ret) - return ret; - break; - case IB_FLOW_SPEC_ACTION_COUNT: - if (FIELDS_NOT_SUPPORTED(ib_spec->flow_count, - LAST_COUNTERS_FIELD)) - return -EOPNOTSUPP; - - /* for now support only one counters spec per flow */ - if (action->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) - return -EINVAL; - - action->counters = ib_spec->flow_count.counters; - action->action |= MLX5_FLOW_CONTEXT_ACTION_COUNT; - break; - default: - return -EINVAL; - } - - return 0; -} - -/* If a flow could catch both multicast and unicast packets, - * it won't fall into the multicast flow steering table and this rule - * could steal other multicast packets. - */ -static bool flow_is_multicast_only(const struct ib_flow_attr *ib_attr) -{ - union ib_flow_spec *flow_spec; - - if (ib_attr->type != IB_FLOW_ATTR_NORMAL || - ib_attr->num_of_specs < 1) - return false; - - flow_spec = (union ib_flow_spec *)(ib_attr + 1); - if (flow_spec->type == IB_FLOW_SPEC_IPV4) { - struct ib_flow_spec_ipv4 *ipv4_spec; - - ipv4_spec = (struct ib_flow_spec_ipv4 *)flow_spec; - if (ipv4_is_multicast(ipv4_spec->val.dst_ip)) - return true; - - return false; - } - - if (flow_spec->type == IB_FLOW_SPEC_ETH) { - struct ib_flow_spec_eth *eth_spec; - - eth_spec = (struct ib_flow_spec_eth *)flow_spec; - return is_multicast_ether_addr(eth_spec->mask.dst_mac) && - is_multicast_ether_addr(eth_spec->val.dst_mac); - } - - return false; -} - -enum valid_spec { - VALID_SPEC_INVALID, - VALID_SPEC_VALID, - VALID_SPEC_NA, -}; - -static enum valid_spec -is_valid_esp_aes_gcm(struct mlx5_core_dev *mdev, - const struct mlx5_flow_spec *spec, - const struct mlx5_flow_act *flow_act, - bool egress) -{ - const u32 *match_c = spec->match_criteria; - bool is_crypto = - (flow_act->action & (MLX5_FLOW_CONTEXT_ACTION_ENCRYPT | - MLX5_FLOW_CONTEXT_ACTION_DECRYPT)); - bool is_ipsec = mlx5_fs_is_ipsec_flow(match_c); - bool is_drop = flow_act->action & MLX5_FLOW_CONTEXT_ACTION_DROP; - - /* - * Currently only crypto is supported in egress, when regular egress - * rules would be supported, always return VALID_SPEC_NA. - */ - if (!is_crypto) - return VALID_SPEC_NA; - - return is_crypto && is_ipsec && - (!egress || (!is_drop && - !(spec->flow_context.flags & FLOW_CONTEXT_HAS_TAG))) ? - VALID_SPEC_VALID : VALID_SPEC_INVALID; -} - -static bool is_valid_spec(struct mlx5_core_dev *mdev, - const struct mlx5_flow_spec *spec, - const struct mlx5_flow_act *flow_act, - bool egress) -{ - /* We curretly only support ipsec egress flow */ - return is_valid_esp_aes_gcm(mdev, spec, flow_act, egress) != VALID_SPEC_INVALID; -} - -static bool is_valid_ethertype(struct mlx5_core_dev *mdev, - const struct ib_flow_attr *flow_attr, - bool check_inner) -{ - union ib_flow_spec *ib_spec = (union ib_flow_spec *)(flow_attr + 1); - int match_ipv = check_inner ? - MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.inner_ip_version) : - MLX5_CAP_FLOWTABLE_NIC_RX(mdev, - ft_field_support.outer_ip_version); - int inner_bit = check_inner ? IB_FLOW_SPEC_INNER : 0; - bool ipv4_spec_valid, ipv6_spec_valid; - unsigned int ip_spec_type = 0; - bool has_ethertype = false; - unsigned int spec_index; - bool mask_valid = true; - u16 eth_type = 0; - bool type_valid; - - /* Validate that ethertype is correct */ - for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) { - if ((ib_spec->type == (IB_FLOW_SPEC_ETH | inner_bit)) && - ib_spec->eth.mask.ether_type) { - mask_valid = (ib_spec->eth.mask.ether_type == - htons(0xffff)); - has_ethertype = true; - eth_type = ntohs(ib_spec->eth.val.ether_type); - } else if ((ib_spec->type == (IB_FLOW_SPEC_IPV4 | inner_bit)) || - (ib_spec->type == (IB_FLOW_SPEC_IPV6 | inner_bit))) { - ip_spec_type = ib_spec->type; - } - ib_spec = (void *)ib_spec + ib_spec->size; - } - - type_valid = (!has_ethertype) || (!ip_spec_type); - if (!type_valid && mask_valid) { - ipv4_spec_valid = (eth_type == ETH_P_IP) && - (ip_spec_type == (IB_FLOW_SPEC_IPV4 | inner_bit)); - ipv6_spec_valid = (eth_type == ETH_P_IPV6) && - (ip_spec_type == (IB_FLOW_SPEC_IPV6 | inner_bit)); - - type_valid = (ipv4_spec_valid) || (ipv6_spec_valid) || - (((eth_type == ETH_P_MPLS_UC) || - (eth_type == ETH_P_MPLS_MC)) && match_ipv); - } - - return type_valid; -} - -static bool is_valid_attr(struct mlx5_core_dev *mdev, - const struct ib_flow_attr *flow_attr) -{ - return is_valid_ethertype(mdev, flow_attr, false) && - is_valid_ethertype(mdev, flow_attr, true); -} - -static void put_flow_table(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *prio, bool ft_added) -{ - prio->refcount -= !!ft_added; - if (!prio->refcount) { - mlx5_destroy_flow_table(prio->flow_table); - prio->flow_table = NULL; - } -} - -static void counters_clear_description(struct ib_counters *counters) -{ - struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); - - mutex_lock(&mcounters->mcntrs_mutex); - kfree(mcounters->counters_data); - mcounters->counters_data = NULL; - mcounters->cntrs_max_index = 0; - mutex_unlock(&mcounters->mcntrs_mutex); -} - -static int mlx5_ib_destroy_flow(struct ib_flow *flow_id) -{ - struct mlx5_ib_flow_handler *handler = container_of(flow_id, - struct mlx5_ib_flow_handler, - ibflow); - struct mlx5_ib_flow_handler *iter, *tmp; - struct mlx5_ib_dev *dev = handler->dev; - - mutex_lock(&dev->flow_db->lock); - - list_for_each_entry_safe(iter, tmp, &handler->list, list) { - mlx5_del_flow_rules(iter->rule); - put_flow_table(dev, iter->prio, true); - list_del(&iter->list); - kfree(iter); - } - - mlx5_del_flow_rules(handler->rule); - put_flow_table(dev, handler->prio, true); - if (handler->ibcounters && - atomic_read(&handler->ibcounters->usecnt) == 1) - counters_clear_description(handler->ibcounters); - - mutex_unlock(&dev->flow_db->lock); - if (handler->flow_matcher) - atomic_dec(&handler->flow_matcher->usecnt); - kfree(handler); - - return 0; -} - -static int ib_prio_to_core_prio(unsigned int priority, bool dont_trap) -{ - priority *= 2; - if (!dont_trap) - priority++; - return priority; -} - -enum flow_table_type { - MLX5_IB_FT_RX, - MLX5_IB_FT_TX -}; - -#define MLX5_FS_MAX_TYPES 6 -#define MLX5_FS_MAX_ENTRIES BIT(16) - -static struct mlx5_ib_flow_prio *_get_prio(struct mlx5_flow_namespace *ns, - struct mlx5_ib_flow_prio *prio, - int priority, - int num_entries, int num_groups, - u32 flags) -{ - struct mlx5_flow_table_attr ft_attr = {}; - struct mlx5_flow_table *ft; - - ft_attr.prio = priority; - ft_attr.max_fte = num_entries; - ft_attr.flags = flags; - ft_attr.autogroup.max_num_groups = num_groups; - ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr); - if (IS_ERR(ft)) - return ERR_CAST(ft); - - prio->flow_table = ft; - prio->refcount = 0; - return prio; -} - -static struct mlx5_ib_flow_prio *get_flow_table(struct mlx5_ib_dev *dev, - struct ib_flow_attr *flow_attr, - enum flow_table_type ft_type) -{ - bool dont_trap = flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP; - struct mlx5_flow_namespace *ns = NULL; - struct mlx5_ib_flow_prio *prio; - struct mlx5_flow_table *ft; - int max_table_size; - int num_entries; - int num_groups; - bool esw_encap; - u32 flags = 0; - int priority; - - max_table_size = BIT(MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, - log_max_ft_size)); - esw_encap = mlx5_eswitch_get_encap_mode(dev->mdev) != - DEVLINK_ESWITCH_ENCAP_MODE_NONE; - if (flow_attr->type == IB_FLOW_ATTR_NORMAL) { - enum mlx5_flow_namespace_type fn_type; - - if (flow_is_multicast_only(flow_attr) && - !dont_trap) - priority = MLX5_IB_FLOW_MCAST_PRIO; - else - priority = ib_prio_to_core_prio(flow_attr->priority, - dont_trap); - if (ft_type == MLX5_IB_FT_RX) { - fn_type = MLX5_FLOW_NAMESPACE_BYPASS; - prio = &dev->flow_db->prios[priority]; - if (!dev->is_rep && !esw_encap && - MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, decap)) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_DECAP; - if (!dev->is_rep && !esw_encap && - MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, - reformat_l3_tunnel_to_l2)) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; - } else { - max_table_size = - BIT(MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, - log_max_ft_size)); - fn_type = MLX5_FLOW_NAMESPACE_EGRESS; - prio = &dev->flow_db->egress_prios[priority]; - if (!dev->is_rep && !esw_encap && - MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, reformat)) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; - } - ns = mlx5_get_flow_namespace(dev->mdev, fn_type); - num_entries = MLX5_FS_MAX_ENTRIES; - num_groups = MLX5_FS_MAX_TYPES; - } else if (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || - flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT) { - ns = mlx5_get_flow_namespace(dev->mdev, - MLX5_FLOW_NAMESPACE_LEFTOVERS); - build_leftovers_ft_param(&priority, - &num_entries, - &num_groups); - prio = &dev->flow_db->prios[MLX5_IB_FLOW_LEFTOVERS_PRIO]; - } else if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { - if (!MLX5_CAP_FLOWTABLE(dev->mdev, - allow_sniffer_and_nic_rx_shared_tir)) - return ERR_PTR(-ENOTSUPP); - - ns = mlx5_get_flow_namespace(dev->mdev, ft_type == MLX5_IB_FT_RX ? - MLX5_FLOW_NAMESPACE_SNIFFER_RX : - MLX5_FLOW_NAMESPACE_SNIFFER_TX); - - prio = &dev->flow_db->sniffer[ft_type]; - priority = 0; - num_entries = 1; - num_groups = 1; - } - - if (!ns) - return ERR_PTR(-ENOTSUPP); - - max_table_size = min_t(int, num_entries, max_table_size); - - ft = prio->flow_table; - if (!ft) - return _get_prio(ns, prio, priority, max_table_size, num_groups, - flags); - - return prio; -} - -static void set_underlay_qp(struct mlx5_ib_dev *dev, - struct mlx5_flow_spec *spec, - u32 underlay_qpn) -{ - void *misc_params_c = MLX5_ADDR_OF(fte_match_param, - spec->match_criteria, - misc_parameters); - void *misc_params_v = MLX5_ADDR_OF(fte_match_param, spec->match_value, - misc_parameters); - - if (underlay_qpn && - MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, - ft_field_support.bth_dst_qp)) { - MLX5_SET(fte_match_set_misc, - misc_params_v, bth_dst_qp, underlay_qpn); - MLX5_SET(fte_match_set_misc, - misc_params_c, bth_dst_qp, 0xffffff); - } -} - -static int read_flow_counters(struct ib_device *ibdev, - struct mlx5_read_counters_attr *read_attr) -{ - struct mlx5_fc *fc = read_attr->hw_cntrs_hndl; - struct mlx5_ib_dev *dev = to_mdev(ibdev); - - return mlx5_fc_query(dev->mdev, fc, - &read_attr->out[IB_COUNTER_PACKETS], - &read_attr->out[IB_COUNTER_BYTES]); -} - -/* flow counters currently expose two counters packets and bytes */ -#define FLOW_COUNTERS_NUM 2 -static int counters_set_description(struct ib_counters *counters, - enum mlx5_ib_counters_type counters_type, - struct mlx5_ib_flow_counters_desc *desc_data, - u32 ncounters) -{ - struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); - u32 cntrs_max_index = 0; - int i; - - if (counters_type != MLX5_IB_COUNTERS_FLOW) - return -EINVAL; - - /* init the fields for the object */ - mcounters->type = counters_type; - mcounters->read_counters = read_flow_counters; - mcounters->counters_num = FLOW_COUNTERS_NUM; - mcounters->ncounters = ncounters; - /* each counter entry have both description and index pair */ - for (i = 0; i < ncounters; i++) { - if (desc_data[i].description > IB_COUNTER_BYTES) - return -EINVAL; - - if (cntrs_max_index <= desc_data[i].index) - cntrs_max_index = desc_data[i].index + 1; - } - - mutex_lock(&mcounters->mcntrs_mutex); - mcounters->counters_data = desc_data; - mcounters->cntrs_max_index = cntrs_max_index; - mutex_unlock(&mcounters->mcntrs_mutex); - - return 0; -} - -#define MAX_COUNTERS_NUM (USHRT_MAX / (sizeof(u32) * 2)) -static int flow_counters_set_data(struct ib_counters *ibcounters, - struct mlx5_ib_create_flow *ucmd) -{ - struct mlx5_ib_mcounters *mcounters = to_mcounters(ibcounters); - struct mlx5_ib_flow_counters_data *cntrs_data = NULL; - struct mlx5_ib_flow_counters_desc *desc_data = NULL; - bool hw_hndl = false; - int ret = 0; - - if (ucmd && ucmd->ncounters_data != 0) { - cntrs_data = ucmd->data; - if (cntrs_data->ncounters > MAX_COUNTERS_NUM) - return -EINVAL; - - desc_data = kcalloc(cntrs_data->ncounters, - sizeof(*desc_data), - GFP_KERNEL); - if (!desc_data) - return -ENOMEM; - - if (copy_from_user(desc_data, - u64_to_user_ptr(cntrs_data->counters_data), - sizeof(*desc_data) * cntrs_data->ncounters)) { - ret = -EFAULT; - goto free; - } - } - - if (!mcounters->hw_cntrs_hndl) { - mcounters->hw_cntrs_hndl = mlx5_fc_create( - to_mdev(ibcounters->device)->mdev, false); - if (IS_ERR(mcounters->hw_cntrs_hndl)) { - ret = PTR_ERR(mcounters->hw_cntrs_hndl); - goto free; - } - hw_hndl = true; - } - - if (desc_data) { - /* counters already bound to at least one flow */ - if (mcounters->cntrs_max_index) { - ret = -EINVAL; - goto free_hndl; - } - - ret = counters_set_description(ibcounters, - MLX5_IB_COUNTERS_FLOW, - desc_data, - cntrs_data->ncounters); - if (ret) - goto free_hndl; - - } else if (!mcounters->cntrs_max_index) { - /* counters not bound yet, must have udata passed */ - ret = -EINVAL; - goto free_hndl; - } - - return 0; - -free_hndl: - if (hw_hndl) { - mlx5_fc_destroy(to_mdev(ibcounters->device)->mdev, - mcounters->hw_cntrs_hndl); - mcounters->hw_cntrs_hndl = NULL; - } -free: - kfree(desc_data); - return ret; -} - -static void mlx5_ib_set_rule_source_port(struct mlx5_ib_dev *dev, - struct mlx5_flow_spec *spec, - struct mlx5_eswitch_rep *rep) -{ - struct mlx5_eswitch *esw = dev->mdev->priv.eswitch; - void *misc; - - if (mlx5_eswitch_vport_match_metadata_enabled(esw)) { - misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, - misc_parameters_2); - - MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0, - mlx5_eswitch_get_vport_metadata_for_match(esw, - rep->vport)); - misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, - misc_parameters_2); - - MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0, - mlx5_eswitch_get_vport_metadata_mask()); - } else { - misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, - misc_parameters); - - MLX5_SET(fte_match_set_misc, misc, source_port, rep->vport); - - misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, - misc_parameters); - - MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port); - } -} - -static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *ft_prio, - const struct ib_flow_attr *flow_attr, - struct mlx5_flow_destination *dst, - u32 underlay_qpn, - struct mlx5_ib_create_flow *ucmd) -{ - struct mlx5_flow_table *ft = ft_prio->flow_table; - struct mlx5_ib_flow_handler *handler; - struct mlx5_flow_act flow_act = {}; - struct mlx5_flow_spec *spec; - struct mlx5_flow_destination dest_arr[2] = {}; - struct mlx5_flow_destination *rule_dst = dest_arr; - const void *ib_flow = (const void *)flow_attr + sizeof(*flow_attr); - unsigned int spec_index; - u32 prev_type = 0; - int err = 0; - int dest_num = 0; - bool is_egress = flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS; - - if (!is_valid_attr(dev->mdev, flow_attr)) - return ERR_PTR(-EINVAL); - - if (dev->is_rep && is_egress) - return ERR_PTR(-EINVAL); - - spec = kvzalloc(sizeof(*spec), GFP_KERNEL); - handler = kzalloc(sizeof(*handler), GFP_KERNEL); - if (!handler || !spec) { - err = -ENOMEM; - goto free; - } - - INIT_LIST_HEAD(&handler->list); - - for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) { - err = parse_flow_attr(dev->mdev, spec, - ib_flow, flow_attr, &flow_act, - prev_type); - if (err < 0) - goto free; - - prev_type = ((union ib_flow_spec *)ib_flow)->type; - ib_flow += ((union ib_flow_spec *)ib_flow)->size; - } - - if (dst && !(flow_act.action & MLX5_FLOW_CONTEXT_ACTION_DROP)) { - memcpy(&dest_arr[0], dst, sizeof(*dst)); - dest_num++; - } - - if (!flow_is_multicast_only(flow_attr)) - set_underlay_qp(dev, spec, underlay_qpn); - - if (dev->is_rep) { - struct mlx5_eswitch_rep *rep; - - rep = dev->port[flow_attr->port - 1].rep; - if (!rep) { - err = -EINVAL; - goto free; - } - - mlx5_ib_set_rule_source_port(dev, spec, rep); - } - - spec->match_criteria_enable = get_match_criteria_enable(spec->match_criteria); - - if (is_egress && - !is_valid_spec(dev->mdev, spec, &flow_act, is_egress)) { - err = -EINVAL; - goto free; - } - - if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) { - struct mlx5_ib_mcounters *mcounters; - - err = flow_counters_set_data(flow_act.counters, ucmd); - if (err) - goto free; - - mcounters = to_mcounters(flow_act.counters); - handler->ibcounters = flow_act.counters; - dest_arr[dest_num].type = - MLX5_FLOW_DESTINATION_TYPE_COUNTER; - dest_arr[dest_num].counter_id = - mlx5_fc_id(mcounters->hw_cntrs_hndl); - dest_num++; - } - - if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_DROP) { - if (!dest_num) - rule_dst = NULL; - } else { - if (is_egress) - flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW; - else - flow_act.action |= - dest_num ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST : - MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO; - } - - if ((spec->flow_context.flags & FLOW_CONTEXT_HAS_TAG) && - (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || - flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT)) { - mlx5_ib_warn(dev, "Flow tag %u and attribute type %x isn't allowed in leftovers\n", - spec->flow_context.flow_tag, flow_attr->type); - err = -EINVAL; - goto free; - } - handler->rule = mlx5_add_flow_rules(ft, spec, - &flow_act, - rule_dst, dest_num); - - if (IS_ERR(handler->rule)) { - err = PTR_ERR(handler->rule); - goto free; - } - - ft_prio->refcount++; - handler->prio = ft_prio; - handler->dev = dev; - - ft_prio->flow_table = ft; -free: - if (err && handler) { - if (handler->ibcounters && - atomic_read(&handler->ibcounters->usecnt) == 1) - counters_clear_description(handler->ibcounters); - kfree(handler); - } - kvfree(spec); - return err ? ERR_PTR(err) : handler; -} - -static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *ft_prio, - const struct ib_flow_attr *flow_attr, - struct mlx5_flow_destination *dst) -{ - return _create_flow_rule(dev, ft_prio, flow_attr, dst, 0, NULL); -} - -static struct mlx5_ib_flow_handler *create_dont_trap_rule(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *ft_prio, - struct ib_flow_attr *flow_attr, - struct mlx5_flow_destination *dst) -{ - struct mlx5_ib_flow_handler *handler_dst = NULL; - struct mlx5_ib_flow_handler *handler = NULL; - - handler = create_flow_rule(dev, ft_prio, flow_attr, NULL); - if (!IS_ERR(handler)) { - handler_dst = create_flow_rule(dev, ft_prio, - flow_attr, dst); - if (IS_ERR(handler_dst)) { - mlx5_del_flow_rules(handler->rule); - ft_prio->refcount--; - kfree(handler); - handler = handler_dst; - } else { - list_add(&handler_dst->list, &handler->list); - } - } - - return handler; -} -enum { - LEFTOVERS_MC, - LEFTOVERS_UC, -}; - -static struct mlx5_ib_flow_handler *create_leftovers_rule(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *ft_prio, - struct ib_flow_attr *flow_attr, - struct mlx5_flow_destination *dst) -{ - struct mlx5_ib_flow_handler *handler_ucast = NULL; - struct mlx5_ib_flow_handler *handler = NULL; - - static struct { - struct ib_flow_attr flow_attr; - struct ib_flow_spec_eth eth_flow; - } leftovers_specs[] = { - [LEFTOVERS_MC] = { - .flow_attr = { - .num_of_specs = 1, - .size = sizeof(leftovers_specs[0]) - }, - .eth_flow = { - .type = IB_FLOW_SPEC_ETH, - .size = sizeof(struct ib_flow_spec_eth), - .mask = {.dst_mac = {0x1} }, - .val = {.dst_mac = {0x1} } - } - }, - [LEFTOVERS_UC] = { - .flow_attr = { - .num_of_specs = 1, - .size = sizeof(leftovers_specs[0]) - }, - .eth_flow = { - .type = IB_FLOW_SPEC_ETH, - .size = sizeof(struct ib_flow_spec_eth), - .mask = {.dst_mac = {0x1} }, - .val = {.dst_mac = {} } - } - } - }; - - handler = create_flow_rule(dev, ft_prio, - &leftovers_specs[LEFTOVERS_MC].flow_attr, - dst); - if (!IS_ERR(handler) && - flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT) { - handler_ucast = create_flow_rule(dev, ft_prio, - &leftovers_specs[LEFTOVERS_UC].flow_attr, - dst); - if (IS_ERR(handler_ucast)) { - mlx5_del_flow_rules(handler->rule); - ft_prio->refcount--; - kfree(handler); - handler = handler_ucast; - } else { - list_add(&handler_ucast->list, &handler->list); - } - } - - return handler; -} - -static struct mlx5_ib_flow_handler *create_sniffer_rule(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *ft_rx, - struct mlx5_ib_flow_prio *ft_tx, - struct mlx5_flow_destination *dst) -{ - struct mlx5_ib_flow_handler *handler_rx; - struct mlx5_ib_flow_handler *handler_tx; - int err; - static const struct ib_flow_attr flow_attr = { - .num_of_specs = 0, - .size = sizeof(flow_attr) - }; - - handler_rx = create_flow_rule(dev, ft_rx, &flow_attr, dst); - if (IS_ERR(handler_rx)) { - err = PTR_ERR(handler_rx); - goto err; - } - - handler_tx = create_flow_rule(dev, ft_tx, &flow_attr, dst); - if (IS_ERR(handler_tx)) { - err = PTR_ERR(handler_tx); - goto err_tx; - } - - list_add(&handler_tx->list, &handler_rx->list); - - return handler_rx; - -err_tx: - mlx5_del_flow_rules(handler_rx->rule); - ft_rx->refcount--; - kfree(handler_rx); -err: - return ERR_PTR(err); -} - -static struct ib_flow *mlx5_ib_create_flow(struct ib_qp *qp, - struct ib_flow_attr *flow_attr, - int domain, - struct ib_udata *udata) -{ - struct mlx5_ib_dev *dev = to_mdev(qp->device); - struct mlx5_ib_qp *mqp = to_mqp(qp); - struct mlx5_ib_flow_handler *handler = NULL; - struct mlx5_flow_destination *dst = NULL; - struct mlx5_ib_flow_prio *ft_prio_tx = NULL; - struct mlx5_ib_flow_prio *ft_prio; - bool is_egress = flow_attr->flags & IB_FLOW_ATTR_FLAGS_EGRESS; - struct mlx5_ib_create_flow *ucmd = NULL, ucmd_hdr; - size_t min_ucmd_sz, required_ucmd_sz; - int err; - int underlay_qpn; - - if (udata && udata->inlen) { - min_ucmd_sz = offsetof(typeof(ucmd_hdr), reserved) + - sizeof(ucmd_hdr.reserved); - if (udata->inlen < min_ucmd_sz) - return ERR_PTR(-EOPNOTSUPP); - - err = ib_copy_from_udata(&ucmd_hdr, udata, min_ucmd_sz); - if (err) - return ERR_PTR(err); - - /* currently supports only one counters data */ - if (ucmd_hdr.ncounters_data > 1) - return ERR_PTR(-EINVAL); - - required_ucmd_sz = min_ucmd_sz + - sizeof(struct mlx5_ib_flow_counters_data) * - ucmd_hdr.ncounters_data; - if (udata->inlen > required_ucmd_sz && - !ib_is_udata_cleared(udata, required_ucmd_sz, - udata->inlen - required_ucmd_sz)) - return ERR_PTR(-EOPNOTSUPP); - - ucmd = kzalloc(required_ucmd_sz, GFP_KERNEL); - if (!ucmd) - return ERR_PTR(-ENOMEM); - - err = ib_copy_from_udata(ucmd, udata, required_ucmd_sz); - if (err) - goto free_ucmd; - } - - if (flow_attr->priority > MLX5_IB_FLOW_LAST_PRIO) { - err = -ENOMEM; - goto free_ucmd; - } - - if (domain != IB_FLOW_DOMAIN_USER || - flow_attr->port > dev->num_ports || - (flow_attr->flags & ~(IB_FLOW_ATTR_FLAGS_DONT_TRAP | - IB_FLOW_ATTR_FLAGS_EGRESS))) { - err = -EINVAL; - goto free_ucmd; - } - - if (is_egress && - (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || - flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT)) { - err = -EINVAL; - goto free_ucmd; - } - - dst = kzalloc(sizeof(*dst), GFP_KERNEL); - if (!dst) { - err = -ENOMEM; - goto free_ucmd; - } - - mutex_lock(&dev->flow_db->lock); - - ft_prio = get_flow_table(dev, flow_attr, - is_egress ? MLX5_IB_FT_TX : MLX5_IB_FT_RX); - if (IS_ERR(ft_prio)) { - err = PTR_ERR(ft_prio); - goto unlock; - } - if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { - ft_prio_tx = get_flow_table(dev, flow_attr, MLX5_IB_FT_TX); - if (IS_ERR(ft_prio_tx)) { - err = PTR_ERR(ft_prio_tx); - ft_prio_tx = NULL; - goto destroy_ft; - } - } - - if (is_egress) { - dst->type = MLX5_FLOW_DESTINATION_TYPE_PORT; - } else { - dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR; - if (mqp->flags & MLX5_IB_QP_RSS) - dst->tir_num = mqp->rss_qp.tirn; - else - dst->tir_num = mqp->raw_packet_qp.rq.tirn; - } - - if (flow_attr->type == IB_FLOW_ATTR_NORMAL) { - if (flow_attr->flags & IB_FLOW_ATTR_FLAGS_DONT_TRAP) { - handler = create_dont_trap_rule(dev, ft_prio, - flow_attr, dst); - } else { - underlay_qpn = (mqp->flags & MLX5_IB_QP_UNDERLAY) ? - mqp->underlay_qpn : 0; - handler = _create_flow_rule(dev, ft_prio, flow_attr, - dst, underlay_qpn, ucmd); - } - } else if (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT || - flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT) { - handler = create_leftovers_rule(dev, ft_prio, flow_attr, - dst); - } else if (flow_attr->type == IB_FLOW_ATTR_SNIFFER) { - handler = create_sniffer_rule(dev, ft_prio, ft_prio_tx, dst); - } else { - err = -EINVAL; - goto destroy_ft; - } - - if (IS_ERR(handler)) { - err = PTR_ERR(handler); - handler = NULL; - goto destroy_ft; - } - - mutex_unlock(&dev->flow_db->lock); - kfree(dst); - kfree(ucmd); - - return &handler->ibflow; - -destroy_ft: - put_flow_table(dev, ft_prio, false); - if (ft_prio_tx) - put_flow_table(dev, ft_prio_tx, false); -unlock: - mutex_unlock(&dev->flow_db->lock); - kfree(dst); -free_ucmd: - kfree(ucmd); - return ERR_PTR(err); -} - -static struct mlx5_ib_flow_prio * -_get_flow_table(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_matcher *fs_matcher, - bool mcast) -{ - struct mlx5_flow_namespace *ns = NULL; - struct mlx5_ib_flow_prio *prio = NULL; - int max_table_size = 0; - bool esw_encap; - u32 flags = 0; - int priority; - - if (mcast) - priority = MLX5_IB_FLOW_MCAST_PRIO; - else - priority = ib_prio_to_core_prio(fs_matcher->priority, false); - - esw_encap = mlx5_eswitch_get_encap_mode(dev->mdev) != - DEVLINK_ESWITCH_ENCAP_MODE_NONE; - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) { - max_table_size = BIT(MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, - log_max_ft_size)); - if (MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, decap) && !esw_encap) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_DECAP; - if (MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, - reformat_l3_tunnel_to_l2) && - !esw_encap) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; - } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS) { - max_table_size = BIT( - MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, log_max_ft_size)); - if (MLX5_CAP_FLOWTABLE_NIC_TX(dev->mdev, reformat) && !esw_encap) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; - } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB) { - max_table_size = BIT( - MLX5_CAP_ESW_FLOWTABLE_FDB(dev->mdev, log_max_ft_size)); - if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev->mdev, decap) && esw_encap) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_DECAP; - if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev->mdev, reformat_l3_tunnel_to_l2) && - esw_encap) - flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; - priority = FDB_BYPASS_PATH; - } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) { - max_table_size = - BIT(MLX5_CAP_FLOWTABLE_RDMA_RX(dev->mdev, - log_max_ft_size)); - priority = fs_matcher->priority; - } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_TX) { - max_table_size = - BIT(MLX5_CAP_FLOWTABLE_RDMA_TX(dev->mdev, - log_max_ft_size)); - priority = fs_matcher->priority; - } - - max_table_size = min_t(int, max_table_size, MLX5_FS_MAX_ENTRIES); - - ns = mlx5_get_flow_namespace(dev->mdev, fs_matcher->ns_type); - if (!ns) - return ERR_PTR(-ENOTSUPP); - - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_BYPASS) - prio = &dev->flow_db->prios[priority]; - else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS) - prio = &dev->flow_db->egress_prios[priority]; - else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB) - prio = &dev->flow_db->fdb; - else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) - prio = &dev->flow_db->rdma_rx[priority]; - else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_TX) - prio = &dev->flow_db->rdma_tx[priority]; - - if (!prio) - return ERR_PTR(-EINVAL); - - if (prio->flow_table) - return prio; - - return _get_prio(ns, prio, priority, max_table_size, - MLX5_FS_MAX_TYPES, flags); -} - -static struct mlx5_ib_flow_handler * -_create_raw_flow_rule(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_prio *ft_prio, - struct mlx5_flow_destination *dst, - struct mlx5_ib_flow_matcher *fs_matcher, - struct mlx5_flow_context *flow_context, - struct mlx5_flow_act *flow_act, - void *cmd_in, int inlen, - int dst_num) -{ - struct mlx5_ib_flow_handler *handler; - struct mlx5_flow_spec *spec; - struct mlx5_flow_table *ft = ft_prio->flow_table; - int err = 0; - - spec = kvzalloc(sizeof(*spec), GFP_KERNEL); - handler = kzalloc(sizeof(*handler), GFP_KERNEL); - if (!handler || !spec) { - err = -ENOMEM; - goto free; - } - - INIT_LIST_HEAD(&handler->list); - - memcpy(spec->match_value, cmd_in, inlen); - memcpy(spec->match_criteria, fs_matcher->matcher_mask.match_params, - fs_matcher->mask_len); - spec->match_criteria_enable = fs_matcher->match_criteria_enable; - spec->flow_context = *flow_context; - - handler->rule = mlx5_add_flow_rules(ft, spec, - flow_act, dst, dst_num); - - if (IS_ERR(handler->rule)) { - err = PTR_ERR(handler->rule); - goto free; - } - - ft_prio->refcount++; - handler->prio = ft_prio; - handler->dev = dev; - ft_prio->flow_table = ft; - -free: - if (err) - kfree(handler); - kvfree(spec); - return err ? ERR_PTR(err) : handler; -} - -static bool raw_fs_is_multicast(struct mlx5_ib_flow_matcher *fs_matcher, - void *match_v) -{ - void *match_c; - void *match_v_set_lyr_2_4, *match_c_set_lyr_2_4; - void *dmac, *dmac_mask; - void *ipv4, *ipv4_mask; - - if (!(fs_matcher->match_criteria_enable & - (1 << MATCH_CRITERIA_ENABLE_OUTER_BIT))) - return false; - - match_c = fs_matcher->matcher_mask.match_params; - match_v_set_lyr_2_4 = MLX5_ADDR_OF(fte_match_param, match_v, - outer_headers); - match_c_set_lyr_2_4 = MLX5_ADDR_OF(fte_match_param, match_c, - outer_headers); - - dmac = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_v_set_lyr_2_4, - dmac_47_16); - dmac_mask = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_c_set_lyr_2_4, - dmac_47_16); - - if (is_multicast_ether_addr(dmac) && - is_multicast_ether_addr(dmac_mask)) - return true; - - ipv4 = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_v_set_lyr_2_4, - dst_ipv4_dst_ipv6.ipv4_layout.ipv4); - - ipv4_mask = MLX5_ADDR_OF(fte_match_set_lyr_2_4, match_c_set_lyr_2_4, - dst_ipv4_dst_ipv6.ipv4_layout.ipv4); - - if (ipv4_is_multicast(*(__be32 *)(ipv4)) && - ipv4_is_multicast(*(__be32 *)(ipv4_mask))) - return true; - - return false; -} - -struct mlx5_ib_flow_handler * -mlx5_ib_raw_fs_rule_add(struct mlx5_ib_dev *dev, - struct mlx5_ib_flow_matcher *fs_matcher, - struct mlx5_flow_context *flow_context, - struct mlx5_flow_act *flow_act, - u32 counter_id, - void *cmd_in, int inlen, int dest_id, - int dest_type) -{ - struct mlx5_flow_destination *dst; - struct mlx5_ib_flow_prio *ft_prio; - struct mlx5_ib_flow_handler *handler; - int dst_num = 0; - bool mcast; - int err; - - if (fs_matcher->flow_type != MLX5_IB_FLOW_TYPE_NORMAL) - return ERR_PTR(-EOPNOTSUPP); - - if (fs_matcher->priority > MLX5_IB_FLOW_LAST_PRIO) - return ERR_PTR(-ENOMEM); - - dst = kcalloc(2, sizeof(*dst), GFP_KERNEL); - if (!dst) - return ERR_PTR(-ENOMEM); - - mcast = raw_fs_is_multicast(fs_matcher, cmd_in); - mutex_lock(&dev->flow_db->lock); - - ft_prio = _get_flow_table(dev, fs_matcher, mcast); - if (IS_ERR(ft_prio)) { - err = PTR_ERR(ft_prio); - goto unlock; - } - - if (dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR) { - dst[dst_num].type = dest_type; - dst[dst_num].tir_num = dest_id; - flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; - } else if (dest_type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) { - dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM; - dst[dst_num].ft_num = dest_id; - flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; - } else { - dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_PORT; - flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW; - } - - dst_num++; - - if (flow_act->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) { - dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER; - dst[dst_num].counter_id = counter_id; - dst_num++; - } - - handler = _create_raw_flow_rule(dev, ft_prio, dst, fs_matcher, - flow_context, flow_act, - cmd_in, inlen, dst_num); - - if (IS_ERR(handler)) { - err = PTR_ERR(handler); - goto destroy_ft; - } - - mutex_unlock(&dev->flow_db->lock); - atomic_inc(&fs_matcher->usecnt); - handler->flow_matcher = fs_matcher; - - kfree(dst); - - return handler; - -destroy_ft: - put_flow_table(dev, ft_prio, false); -unlock: - mutex_unlock(&dev->flow_db->lock); - kfree(dst); - - return ERR_PTR(err); -} - -static u32 mlx5_ib_flow_action_flags_to_accel_xfrm_flags(u32 mlx5_flags) -{ - u32 flags = 0; - - if (mlx5_flags & MLX5_IB_UAPI_FLOW_ACTION_FLAGS_REQUIRE_METADATA) - flags |= MLX5_ACCEL_XFRM_FLAG_REQUIRE_METADATA; - - return flags; -} - -#define MLX5_FLOW_ACTION_ESP_CREATE_LAST_SUPPORTED MLX5_IB_UAPI_FLOW_ACTION_FLAGS_REQUIRE_METADATA -static struct ib_flow_action * -mlx5_ib_create_flow_action_esp(struct ib_device *device, - const struct ib_flow_action_attrs_esp *attr, - struct uverbs_attr_bundle *attrs) -{ - struct mlx5_ib_dev *mdev = to_mdev(device); - struct ib_uverbs_flow_action_esp_keymat_aes_gcm *aes_gcm; - struct mlx5_accel_esp_xfrm_attrs accel_attrs = {}; - struct mlx5_ib_flow_action *action; - u64 action_flags; - u64 flags; - int err = 0; - - err = uverbs_get_flags64( - &action_flags, attrs, MLX5_IB_ATTR_CREATE_FLOW_ACTION_FLAGS, - ((MLX5_FLOW_ACTION_ESP_CREATE_LAST_SUPPORTED << 1) - 1)); - if (err) - return ERR_PTR(err); - - flags = mlx5_ib_flow_action_flags_to_accel_xfrm_flags(action_flags); - - /* We current only support a subset of the standard features. Only a - * keymat of type AES_GCM, with icv_len == 16, iv_algo == SEQ and esn - * (with overlap). Full offload mode isn't supported. - */ - if (!attr->keymat || attr->replay || attr->encap || - attr->spi || attr->seq || attr->tfc_pad || - attr->hard_limit_pkts || - (attr->flags & ~(IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED | - IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ENCRYPT))) - return ERR_PTR(-EOPNOTSUPP); - - if (attr->keymat->protocol != - IB_UVERBS_FLOW_ACTION_ESP_KEYMAT_AES_GCM) - return ERR_PTR(-EOPNOTSUPP); - - aes_gcm = &attr->keymat->keymat.aes_gcm; - - if (aes_gcm->icv_len != 16 || - aes_gcm->iv_algo != IB_UVERBS_FLOW_ACTION_IV_ALGO_SEQ) - return ERR_PTR(-EOPNOTSUPP); - - action = kmalloc(sizeof(*action), GFP_KERNEL); - if (!action) - return ERR_PTR(-ENOMEM); - - action->esp_aes_gcm.ib_flags = attr->flags; - memcpy(&accel_attrs.keymat.aes_gcm.aes_key, &aes_gcm->aes_key, - sizeof(accel_attrs.keymat.aes_gcm.aes_key)); - accel_attrs.keymat.aes_gcm.key_len = aes_gcm->key_len * 8; - memcpy(&accel_attrs.keymat.aes_gcm.salt, &aes_gcm->salt, - sizeof(accel_attrs.keymat.aes_gcm.salt)); - memcpy(&accel_attrs.keymat.aes_gcm.seq_iv, &aes_gcm->iv, - sizeof(accel_attrs.keymat.aes_gcm.seq_iv)); - accel_attrs.keymat.aes_gcm.icv_len = aes_gcm->icv_len * 8; - accel_attrs.keymat.aes_gcm.iv_algo = MLX5_ACCEL_ESP_AES_GCM_IV_ALGO_SEQ; - accel_attrs.keymat_type = MLX5_ACCEL_ESP_KEYMAT_AES_GCM; - - accel_attrs.esn = attr->esn; - if (attr->flags & IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED) - accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_TRIGGERED; - if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW) - accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP; - - if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ENCRYPT) - accel_attrs.action |= MLX5_ACCEL_ESP_ACTION_ENCRYPT; - - action->esp_aes_gcm.ctx = - mlx5_accel_esp_create_xfrm(mdev->mdev, &accel_attrs, flags); - if (IS_ERR(action->esp_aes_gcm.ctx)) { - err = PTR_ERR(action->esp_aes_gcm.ctx); - goto err_parse; - } - - action->esp_aes_gcm.ib_flags = attr->flags; - - return &action->ib_action; - -err_parse: - kfree(action); - return ERR_PTR(err); -} - -static int -mlx5_ib_modify_flow_action_esp(struct ib_flow_action *action, - const struct ib_flow_action_attrs_esp *attr, - struct uverbs_attr_bundle *attrs) -{ - struct mlx5_ib_flow_action *maction = to_mflow_act(action); - struct mlx5_accel_esp_xfrm_attrs accel_attrs; - int err = 0; - - if (attr->keymat || attr->replay || attr->encap || - attr->spi || attr->seq || attr->tfc_pad || - attr->hard_limit_pkts || - (attr->flags & ~(IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED | - IB_FLOW_ACTION_ESP_FLAGS_MOD_ESP_ATTRS | - IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW))) - return -EOPNOTSUPP; - - /* Only the ESN value or the MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP can - * be modified. - */ - if (!(maction->esp_aes_gcm.ib_flags & - IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED) && - attr->flags & (IB_FLOW_ACTION_ESP_FLAGS_ESN_TRIGGERED | - IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW)) - return -EINVAL; - - memcpy(&accel_attrs, &maction->esp_aes_gcm.ctx->attrs, - sizeof(accel_attrs)); - - accel_attrs.esn = attr->esn; - if (attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW) - accel_attrs.flags |= MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP; - else - accel_attrs.flags &= ~MLX5_ACCEL_ESP_FLAGS_ESN_STATE_OVERLAP; - - err = mlx5_accel_esp_modify_xfrm(maction->esp_aes_gcm.ctx, - &accel_attrs); - if (err) - return err; - - maction->esp_aes_gcm.ib_flags &= - ~IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW; - maction->esp_aes_gcm.ib_flags |= - attr->flags & IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW; - - return 0; -} - -static int mlx5_ib_destroy_flow_action(struct ib_flow_action *action) -{ - struct mlx5_ib_flow_action *maction = to_mflow_act(action); - - switch (action->type) { - case IB_FLOW_ACTION_ESP: - /* - * We only support aes_gcm by now, so we implicitly know this is - * the underline crypto. - */ - mlx5_accel_esp_destroy_xfrm(maction->esp_aes_gcm.ctx); - break; - case IB_FLOW_ACTION_UNSPECIFIED: - mlx5_ib_destroy_flow_action_raw(maction); - break; - default: - WARN_ON(true); - break; - } - - kfree(maction); - return 0; -} - static int mlx5_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); @@ -4446,7 +2587,7 @@ static int mlx5_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) uid = ibqp->pd ? to_mpd(ibqp->pd)->uid : 0; - if (mqp->flags & MLX5_IB_QP_UNDERLAY) { + if (mqp->flags & IB_QP_CREATE_SOURCE_QPN) { mlx5_ib_dbg(dev, "Attaching a multi cast group to underlay QP is not supported\n"); return -EOPNOTSUPP; } @@ -4632,8 +2773,7 @@ static void delay_drop_handler(struct work_struct *work) atomic_inc(&delay_drop->events_cnt); mutex_lock(&delay_drop->lock); - err = mlx5_core_set_delay_drop(delay_drop->dev->mdev, - delay_drop->timeout); + err = mlx5_core_set_delay_drop(delay_drop->dev, delay_drop->timeout); if (err) { mlx5_ib_warn(delay_drop->dev, "Failed to set delay drop, timeout=%u\n", delay_drop->timeout); @@ -4732,7 +2872,7 @@ static void mlx5_ib_handle_event(struct work_struct *_work) break; case MLX5_EVENT_TYPE_GENERAL_EVENT: handle_general_event(ibdev, work->param, &ibev); - /* fall through */ + fallthrough; default: goto out; } @@ -4880,137 +3020,6 @@ static int get_port_caps(struct mlx5_ib_dev *dev, u8 port) return __get_port_caps(dev, port); } -static void destroy_umrc_res(struct mlx5_ib_dev *dev) -{ - int err; - - err = mlx5_mr_cache_cleanup(dev); - if (err) - mlx5_ib_warn(dev, "mr cache cleanup failed\n"); - - if (dev->umrc.qp) - mlx5_ib_destroy_qp(dev->umrc.qp, NULL); - if (dev->umrc.cq) - ib_free_cq(dev->umrc.cq); - if (dev->umrc.pd) - ib_dealloc_pd(dev->umrc.pd); -} - -enum { - MAX_UMR_WR = 128, -}; - -static int create_umr_res(struct mlx5_ib_dev *dev) -{ - struct ib_qp_init_attr *init_attr = NULL; - struct ib_qp_attr *attr = NULL; - struct ib_pd *pd; - struct ib_cq *cq; - struct ib_qp *qp; - int ret; - - attr = kzalloc(sizeof(*attr), GFP_KERNEL); - init_attr = kzalloc(sizeof(*init_attr), GFP_KERNEL); - if (!attr || !init_attr) { - ret = -ENOMEM; - goto error_0; - } - - pd = ib_alloc_pd(&dev->ib_dev, 0); - if (IS_ERR(pd)) { - mlx5_ib_dbg(dev, "Couldn't create PD for sync UMR QP\n"); - ret = PTR_ERR(pd); - goto error_0; - } - - cq = ib_alloc_cq(&dev->ib_dev, NULL, 128, 0, IB_POLL_SOFTIRQ); - if (IS_ERR(cq)) { - mlx5_ib_dbg(dev, "Couldn't create CQ for sync UMR QP\n"); - ret = PTR_ERR(cq); - goto error_2; - } - - init_attr->send_cq = cq; - init_attr->recv_cq = cq; - init_attr->sq_sig_type = IB_SIGNAL_ALL_WR; - init_attr->cap.max_send_wr = MAX_UMR_WR; - init_attr->cap.max_send_sge = 1; - init_attr->qp_type = MLX5_IB_QPT_REG_UMR; - init_attr->port_num = 1; - qp = mlx5_ib_create_qp(pd, init_attr, NULL); - if (IS_ERR(qp)) { - mlx5_ib_dbg(dev, "Couldn't create sync UMR QP\n"); - ret = PTR_ERR(qp); - goto error_3; - } - qp->device = &dev->ib_dev; - qp->real_qp = qp; - qp->uobject = NULL; - qp->qp_type = MLX5_IB_QPT_REG_UMR; - qp->send_cq = init_attr->send_cq; - qp->recv_cq = init_attr->recv_cq; - - attr->qp_state = IB_QPS_INIT; - attr->port_num = 1; - ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE | IB_QP_PKEY_INDEX | - IB_QP_PORT, NULL); - if (ret) { - mlx5_ib_dbg(dev, "Couldn't modify UMR QP\n"); - goto error_4; - } - - memset(attr, 0, sizeof(*attr)); - attr->qp_state = IB_QPS_RTR; - attr->path_mtu = IB_MTU_256; - - ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE, NULL); - if (ret) { - mlx5_ib_dbg(dev, "Couldn't modify umr QP to rtr\n"); - goto error_4; - } - - memset(attr, 0, sizeof(*attr)); - attr->qp_state = IB_QPS_RTS; - ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE, NULL); - if (ret) { - mlx5_ib_dbg(dev, "Couldn't modify umr QP to rts\n"); - goto error_4; - } - - dev->umrc.qp = qp; - dev->umrc.cq = cq; - dev->umrc.pd = pd; - - sema_init(&dev->umrc.sem, MAX_UMR_WR); - ret = mlx5_mr_cache_init(dev); - if (ret) { - mlx5_ib_warn(dev, "mr cache init failed %d\n", ret); - goto error_4; - } - - kfree(attr); - kfree(init_attr); - - return 0; - -error_4: - mlx5_ib_destroy_qp(qp, NULL); - dev->umrc.qp = NULL; - -error_3: - ib_free_cq(cq); - dev->umrc.cq = NULL; - -error_2: - ib_dealloc_pd(pd); - dev->umrc.pd = NULL; - -error_0: - kfree(attr); - kfree(init_attr); - return ret; -} - static u8 mlx5_get_umr_fence(u8 umr_fence_cap) { switch (umr_fence_cap) { @@ -5023,18 +3032,20 @@ static u8 mlx5_get_umr_fence(u8 umr_fence_cap) } } -static int create_dev_resources(struct mlx5_ib_resources *devr) +static int mlx5_ib_dev_res_init(struct mlx5_ib_dev *dev) { + struct mlx5_ib_resources *devr = &dev->devr; struct ib_srq_init_attr attr; - struct mlx5_ib_dev *dev; struct ib_device *ibdev; struct ib_cq_init_attr cq_attr = {.cqe = 1}; int port; int ret = 0; - dev = container_of(devr, struct mlx5_ib_dev, devr); ibdev = &dev->ib_dev; + if (!MLX5_CAP_GEN(dev->mdev, xrc)) + return -EOPNOTSUPP; + mutex_init(&devr->mutex); devr->p0 = rdma_zalloc_drv_obj(ibdev, ib_pd); @@ -5062,34 +3073,19 @@ static int create_dev_resources(struct mlx5_ib_resources *devr) if (ret) goto err_create_cq; - devr->x0 = mlx5_ib_alloc_xrcd(&dev->ib_dev, NULL); - if (IS_ERR(devr->x0)) { - ret = PTR_ERR(devr->x0); + ret = mlx5_cmd_xrcd_alloc(dev->mdev, &devr->xrcdn0, 0); + if (ret) goto error2; - } - devr->x0->device = &dev->ib_dev; - devr->x0->inode = NULL; - atomic_set(&devr->x0->usecnt, 0); - mutex_init(&devr->x0->tgt_qp_mutex); - INIT_LIST_HEAD(&devr->x0->tgt_qp_list); - - devr->x1 = mlx5_ib_alloc_xrcd(&dev->ib_dev, NULL); - if (IS_ERR(devr->x1)) { - ret = PTR_ERR(devr->x1); + + ret = mlx5_cmd_xrcd_alloc(dev->mdev, &devr->xrcdn1, 0); + if (ret) goto error3; - } - devr->x1->device = &dev->ib_dev; - devr->x1->inode = NULL; - atomic_set(&devr->x1->usecnt, 0); - mutex_init(&devr->x1->tgt_qp_mutex); - INIT_LIST_HEAD(&devr->x1->tgt_qp_list); memset(&attr, 0, sizeof(attr)); attr.attr.max_sge = 1; attr.attr.max_wr = 1; attr.srq_type = IB_SRQT_XRC; attr.ext.cq = devr->c0; - attr.ext.xrc.xrcd = devr->x0; devr->s0 = rdma_zalloc_drv_obj(ibdev, ib_srq); if (!devr->s0) { @@ -5100,13 +3096,11 @@ static int create_dev_resources(struct mlx5_ib_resources *devr) devr->s0->device = &dev->ib_dev; devr->s0->pd = devr->p0; devr->s0->srq_type = IB_SRQT_XRC; - devr->s0->ext.xrc.xrcd = devr->x0; devr->s0->ext.cq = devr->c0; ret = mlx5_ib_create_srq(devr->s0, &attr, NULL); if (ret) goto err_create; - atomic_inc(&devr->s0->ext.xrc.xrcd->usecnt); atomic_inc(&devr->s0->ext.cq->usecnt); atomic_inc(&devr->p0->usecnt); atomic_set(&devr->s0->usecnt, 0); @@ -5148,9 +3142,9 @@ error5: err_create: kfree(devr->s0); error4: - mlx5_ib_dealloc_xrcd(devr->x1, NULL); + mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn1, 0); error3: - mlx5_ib_dealloc_xrcd(devr->x0, NULL); + mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn0, 0); error2: mlx5_ib_destroy_cq(devr->c0, NULL); err_create_cq: @@ -5162,16 +3156,17 @@ error0: return ret; } -static void destroy_dev_resources(struct mlx5_ib_resources *devr) +static void mlx5_ib_dev_res_cleanup(struct mlx5_ib_dev *dev) { + struct mlx5_ib_resources *devr = &dev->devr; int port; mlx5_ib_destroy_srq(devr->s1, NULL); kfree(devr->s1); mlx5_ib_destroy_srq(devr->s0, NULL); kfree(devr->s0); - mlx5_ib_dealloc_xrcd(devr->x0, NULL); - mlx5_ib_dealloc_xrcd(devr->x1, NULL); + mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn1, 0); + mlx5_cmd_xrcd_dealloc(dev->mdev, devr->xrcdn0, 0); mlx5_ib_destroy_cq(devr->c0, NULL); kfree(devr->c0); mlx5_ib_dealloc_pd(devr->p0, NULL); @@ -5364,448 +3359,6 @@ static void mlx5_disable_eth(struct mlx5_ib_dev *dev) mlx5_nic_vport_disable_roce(dev->mdev); } -struct mlx5_ib_counter { - const char *name; - size_t offset; -}; - -#define INIT_Q_COUNTER(_name) \ - { .name = #_name, .offset = MLX5_BYTE_OFF(query_q_counter_out, _name)} - -static const struct mlx5_ib_counter basic_q_cnts[] = { - INIT_Q_COUNTER(rx_write_requests), - INIT_Q_COUNTER(rx_read_requests), - INIT_Q_COUNTER(rx_atomic_requests), - INIT_Q_COUNTER(out_of_buffer), -}; - -static const struct mlx5_ib_counter out_of_seq_q_cnts[] = { - INIT_Q_COUNTER(out_of_sequence), -}; - -static const struct mlx5_ib_counter retrans_q_cnts[] = { - INIT_Q_COUNTER(duplicate_request), - INIT_Q_COUNTER(rnr_nak_retry_err), - INIT_Q_COUNTER(packet_seq_err), - INIT_Q_COUNTER(implied_nak_seq_err), - INIT_Q_COUNTER(local_ack_timeout_err), -}; - -#define INIT_CONG_COUNTER(_name) \ - { .name = #_name, .offset = \ - MLX5_BYTE_OFF(query_cong_statistics_out, _name ## _high)} - -static const struct mlx5_ib_counter cong_cnts[] = { - INIT_CONG_COUNTER(rp_cnp_ignored), - INIT_CONG_COUNTER(rp_cnp_handled), - INIT_CONG_COUNTER(np_ecn_marked_roce_packets), - INIT_CONG_COUNTER(np_cnp_sent), -}; - -static const struct mlx5_ib_counter extended_err_cnts[] = { - INIT_Q_COUNTER(resp_local_length_error), - INIT_Q_COUNTER(resp_cqe_error), - INIT_Q_COUNTER(req_cqe_error), - INIT_Q_COUNTER(req_remote_invalid_request), - INIT_Q_COUNTER(req_remote_access_errors), - INIT_Q_COUNTER(resp_remote_access_errors), - INIT_Q_COUNTER(resp_cqe_flush_error), - INIT_Q_COUNTER(req_cqe_flush_error), -}; - -static const struct mlx5_ib_counter roce_accl_cnts[] = { - INIT_Q_COUNTER(roce_adp_retrans), - INIT_Q_COUNTER(roce_adp_retrans_to), - INIT_Q_COUNTER(roce_slow_restart), - INIT_Q_COUNTER(roce_slow_restart_cnps), - INIT_Q_COUNTER(roce_slow_restart_trans), -}; - -#define INIT_EXT_PPCNT_COUNTER(_name) \ - { .name = #_name, .offset = \ - MLX5_BYTE_OFF(ppcnt_reg, \ - counter_set.eth_extended_cntrs_grp_data_layout._name##_high)} - -static const struct mlx5_ib_counter ext_ppcnt_cnts[] = { - INIT_EXT_PPCNT_COUNTER(rx_icrc_encapsulated), -}; - -static bool is_mdev_switchdev_mode(const struct mlx5_core_dev *mdev) -{ - return MLX5_ESWITCH_MANAGER(mdev) && - mlx5_ib_eswitch_mode(mdev->priv.eswitch) == - MLX5_ESWITCH_OFFLOADS; -} - -static void mlx5_ib_dealloc_counters(struct mlx5_ib_dev *dev) -{ - int num_cnt_ports; - int i; - - num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports; - - for (i = 0; i < num_cnt_ports; i++) { - if (dev->port[i].cnts.set_id_valid) - mlx5_core_dealloc_q_counter(dev->mdev, - dev->port[i].cnts.set_id); - kfree(dev->port[i].cnts.names); - kfree(dev->port[i].cnts.offsets); - } -} - -static int __mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev, - struct mlx5_ib_counters *cnts) -{ - u32 num_counters; - - num_counters = ARRAY_SIZE(basic_q_cnts); - - if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt)) - num_counters += ARRAY_SIZE(out_of_seq_q_cnts); - - if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) - num_counters += ARRAY_SIZE(retrans_q_cnts); - - if (MLX5_CAP_GEN(dev->mdev, enhanced_error_q_counters)) - num_counters += ARRAY_SIZE(extended_err_cnts); - - if (MLX5_CAP_GEN(dev->mdev, roce_accl)) - num_counters += ARRAY_SIZE(roce_accl_cnts); - - cnts->num_q_counters = num_counters; - - if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) { - cnts->num_cong_counters = ARRAY_SIZE(cong_cnts); - num_counters += ARRAY_SIZE(cong_cnts); - } - if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { - cnts->num_ext_ppcnt_counters = ARRAY_SIZE(ext_ppcnt_cnts); - num_counters += ARRAY_SIZE(ext_ppcnt_cnts); - } - cnts->names = kcalloc(num_counters, sizeof(cnts->names), GFP_KERNEL); - if (!cnts->names) - return -ENOMEM; - - cnts->offsets = kcalloc(num_counters, - sizeof(cnts->offsets), GFP_KERNEL); - if (!cnts->offsets) - goto err_names; - - return 0; - -err_names: - kfree(cnts->names); - cnts->names = NULL; - return -ENOMEM; -} - -static void mlx5_ib_fill_counters(struct mlx5_ib_dev *dev, - const char **names, - size_t *offsets) -{ - int i; - int j = 0; - - for (i = 0; i < ARRAY_SIZE(basic_q_cnts); i++, j++) { - names[j] = basic_q_cnts[i].name; - offsets[j] = basic_q_cnts[i].offset; - } - - if (MLX5_CAP_GEN(dev->mdev, out_of_seq_cnt)) { - for (i = 0; i < ARRAY_SIZE(out_of_seq_q_cnts); i++, j++) { - names[j] = out_of_seq_q_cnts[i].name; - offsets[j] = out_of_seq_q_cnts[i].offset; - } - } - - if (MLX5_CAP_GEN(dev->mdev, retransmission_q_counters)) { - for (i = 0; i < ARRAY_SIZE(retrans_q_cnts); i++, j++) { - names[j] = retrans_q_cnts[i].name; - offsets[j] = retrans_q_cnts[i].offset; - } - } - - if (MLX5_CAP_GEN(dev->mdev, enhanced_error_q_counters)) { - for (i = 0; i < ARRAY_SIZE(extended_err_cnts); i++, j++) { - names[j] = extended_err_cnts[i].name; - offsets[j] = extended_err_cnts[i].offset; - } - } - - if (MLX5_CAP_GEN(dev->mdev, roce_accl)) { - for (i = 0; i < ARRAY_SIZE(roce_accl_cnts); i++, j++) { - names[j] = roce_accl_cnts[i].name; - offsets[j] = roce_accl_cnts[i].offset; - } - } - - if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) { - for (i = 0; i < ARRAY_SIZE(cong_cnts); i++, j++) { - names[j] = cong_cnts[i].name; - offsets[j] = cong_cnts[i].offset; - } - } - - if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { - for (i = 0; i < ARRAY_SIZE(ext_ppcnt_cnts); i++, j++) { - names[j] = ext_ppcnt_cnts[i].name; - offsets[j] = ext_ppcnt_cnts[i].offset; - } - } -} - -static int mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev) -{ - int num_cnt_ports; - int err = 0; - int i; - bool is_shared; - - is_shared = MLX5_CAP_GEN(dev->mdev, log_max_uctx) != 0; - num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports; - - for (i = 0; i < num_cnt_ports; i++) { - err = __mlx5_ib_alloc_counters(dev, &dev->port[i].cnts); - if (err) - goto err_alloc; - - mlx5_ib_fill_counters(dev, dev->port[i].cnts.names, - dev->port[i].cnts.offsets); - - err = mlx5_cmd_alloc_q_counter(dev->mdev, - &dev->port[i].cnts.set_id, - is_shared ? - MLX5_SHARED_RESOURCE_UID : 0); - if (err) { - mlx5_ib_warn(dev, - "couldn't allocate queue counter for port %d, err %d\n", - i + 1, err); - goto err_alloc; - } - dev->port[i].cnts.set_id_valid = true; - } - return 0; - -err_alloc: - mlx5_ib_dealloc_counters(dev); - return err; -} - -static const struct mlx5_ib_counters *get_counters(struct mlx5_ib_dev *dev, - u8 port_num) -{ - return is_mdev_switchdev_mode(dev->mdev) ? &dev->port[0].cnts : - &dev->port[port_num].cnts; -} - -/** - * mlx5_ib_get_counters_id - Returns counters id to use for device+port - * @dev: Pointer to mlx5 IB device - * @port_num: Zero based port number - * - * mlx5_ib_get_counters_id() Returns counters set id to use for given - * device port combination in switchdev and non switchdev mode of the - * parent device. - */ -u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num) -{ - const struct mlx5_ib_counters *cnts = get_counters(dev, port_num); - - return cnts->set_id; -} - -static struct rdma_hw_stats *mlx5_ib_alloc_hw_stats(struct ib_device *ibdev, - u8 port_num) -{ - struct mlx5_ib_dev *dev = to_mdev(ibdev); - const struct mlx5_ib_counters *cnts; - bool is_switchdev = is_mdev_switchdev_mode(dev->mdev); - - if ((is_switchdev && port_num) || (!is_switchdev && !port_num)) - return NULL; - - cnts = get_counters(dev, port_num - 1); - - return rdma_alloc_hw_stats_struct(cnts->names, - cnts->num_q_counters + - cnts->num_cong_counters + - cnts->num_ext_ppcnt_counters, - RDMA_HW_STATS_DEFAULT_LIFESPAN); -} - -static int mlx5_ib_query_q_counters(struct mlx5_core_dev *mdev, - const struct mlx5_ib_counters *cnts, - struct rdma_hw_stats *stats, - u16 set_id) -{ - int outlen = MLX5_ST_SZ_BYTES(query_q_counter_out); - void *out; - __be32 val; - int ret, i; - - out = kvzalloc(outlen, GFP_KERNEL); - if (!out) - return -ENOMEM; - - ret = mlx5_core_query_q_counter(mdev, set_id, 0, out, outlen); - if (ret) - goto free; - - for (i = 0; i < cnts->num_q_counters; i++) { - val = *(__be32 *)(out + cnts->offsets[i]); - stats->value[i] = (u64)be32_to_cpu(val); - } - -free: - kvfree(out); - return ret; -} - -static int mlx5_ib_query_ext_ppcnt_counters(struct mlx5_ib_dev *dev, - const struct mlx5_ib_counters *cnts, - struct rdma_hw_stats *stats) -{ - int offset = cnts->num_q_counters + cnts->num_cong_counters; - int sz = MLX5_ST_SZ_BYTES(ppcnt_reg); - int ret, i; - void *out; - - out = kvzalloc(sz, GFP_KERNEL); - if (!out) - return -ENOMEM; - - ret = mlx5_cmd_query_ext_ppcnt_counters(dev->mdev, out); - if (ret) - goto free; - - for (i = 0; i < cnts->num_ext_ppcnt_counters; i++) - stats->value[i + offset] = - be64_to_cpup((__be64 *)(out + - cnts->offsets[i + offset])); -free: - kvfree(out); - return ret; -} - -static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, - struct rdma_hw_stats *stats, - u8 port_num, int index) -{ - struct mlx5_ib_dev *dev = to_mdev(ibdev); - const struct mlx5_ib_counters *cnts = get_counters(dev, port_num - 1); - struct mlx5_core_dev *mdev; - int ret, num_counters; - u8 mdev_port_num; - - if (!stats) - return -EINVAL; - - num_counters = cnts->num_q_counters + - cnts->num_cong_counters + - cnts->num_ext_ppcnt_counters; - - /* q_counters are per IB device, query the master mdev */ - ret = mlx5_ib_query_q_counters(dev->mdev, cnts, stats, cnts->set_id); - if (ret) - return ret; - - if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { - ret = mlx5_ib_query_ext_ppcnt_counters(dev, cnts, stats); - if (ret) - return ret; - } - - if (MLX5_CAP_GEN(dev->mdev, cc_query_allowed)) { - mdev = mlx5_ib_get_native_port_mdev(dev, port_num, - &mdev_port_num); - if (!mdev) { - /* If port is not affiliated yet, its in down state - * which doesn't have any counters yet, so it would be - * zero. So no need to read from the HCA. - */ - goto done; - } - ret = mlx5_lag_query_cong_counters(dev->mdev, - stats->value + - cnts->num_q_counters, - cnts->num_cong_counters, - cnts->offsets + - cnts->num_q_counters); - - mlx5_ib_put_native_port_mdev(dev, port_num); - if (ret) - return ret; - } - -done: - return num_counters; -} - -static struct rdma_hw_stats * -mlx5_ib_counter_alloc_stats(struct rdma_counter *counter) -{ - struct mlx5_ib_dev *dev = to_mdev(counter->device); - const struct mlx5_ib_counters *cnts = - get_counters(dev, counter->port - 1); - - return rdma_alloc_hw_stats_struct(cnts->names, - cnts->num_q_counters + - cnts->num_cong_counters + - cnts->num_ext_ppcnt_counters, - RDMA_HW_STATS_DEFAULT_LIFESPAN); -} - -static int mlx5_ib_counter_update_stats(struct rdma_counter *counter) -{ - struct mlx5_ib_dev *dev = to_mdev(counter->device); - const struct mlx5_ib_counters *cnts = - get_counters(dev, counter->port - 1); - - return mlx5_ib_query_q_counters(dev->mdev, cnts, - counter->stats, counter->id); -} - -static int mlx5_ib_counter_bind_qp(struct rdma_counter *counter, - struct ib_qp *qp) -{ - struct mlx5_ib_dev *dev = to_mdev(qp->device); - u16 cnt_set_id = 0; - int err; - - if (!counter->id) { - err = mlx5_cmd_alloc_q_counter(dev->mdev, - &cnt_set_id, - MLX5_SHARED_RESOURCE_UID); - if (err) - return err; - counter->id = cnt_set_id; - } - - err = mlx5_ib_qp_set_counter(qp, counter); - if (err) - goto fail_set_counter; - - return 0; - -fail_set_counter: - mlx5_core_dealloc_q_counter(dev->mdev, cnt_set_id); - counter->id = 0; - - return err; -} - -static int mlx5_ib_counter_unbind_qp(struct ib_qp *qp) -{ - return mlx5_ib_qp_set_counter(qp, NULL); -} - -static int mlx5_ib_counter_dealloc(struct rdma_counter *counter) -{ - struct mlx5_ib_dev *dev = to_mdev(counter->device); - - return mlx5_core_dealloc_q_counter(dev->mdev, counter->id); -} - static int mlx5_ib_rn_get_params(struct ib_device *device, u8 port_num, enum rdma_netdev_t type, struct rdma_netdev_alloc_params *params) @@ -5816,23 +3369,6 @@ static int mlx5_ib_rn_get_params(struct ib_device *device, u8 port_num, return mlx5_rdma_rn_get_params(to_mdev(device)->mdev, device, params); } -static void delay_drop_debugfs_cleanup(struct mlx5_ib_dev *dev) -{ - if (!dev->delay_drop.dir_debugfs) - return; - debugfs_remove_recursive(dev->delay_drop.dir_debugfs); - dev->delay_drop.dir_debugfs = NULL; -} - -static void cancel_delay_drop(struct mlx5_ib_dev *dev) -{ - if (!(dev->ib_dev.attrs.raw_packet_caps & IB_RAW_PACKET_CAP_DELAY_DROP)) - return; - - cancel_work_sync(&dev->delay_drop.delay_drop_work); - delay_drop_debugfs_cleanup(dev); -} - static ssize_t delay_drop_timeout_read(struct file *filp, char __user *buf, size_t count, loff_t *pos) { @@ -5872,40 +3408,6 @@ static const struct file_operations fops_delay_drop_timeout = { .read = delay_drop_timeout_read, }; -static void delay_drop_debugfs_init(struct mlx5_ib_dev *dev) -{ - struct dentry *root; - - if (!mlx5_debugfs_root) - return; - - root = debugfs_create_dir("delay_drop", dev->mdev->priv.dbg_root); - dev->delay_drop.dir_debugfs = root; - - debugfs_create_atomic_t("num_timeout_events", 0400, root, - &dev->delay_drop.events_cnt); - debugfs_create_atomic_t("num_rqs", 0400, root, - &dev->delay_drop.rqs_cnt); - debugfs_create_file("timeout", 0600, root, &dev->delay_drop, - &fops_delay_drop_timeout); -} - -static void init_delay_drop(struct mlx5_ib_dev *dev) -{ - if (!(dev->ib_dev.attrs.raw_packet_caps & IB_RAW_PACKET_CAP_DELAY_DROP)) - return; - - mutex_init(&dev->delay_drop.lock); - dev->delay_drop.dev = dev; - dev->delay_drop.activate = false; - dev->delay_drop.timeout = MLX5_MAX_DELAY_DROP_TIMEOUT_MS * 1000; - INIT_WORK(&dev->delay_drop.delay_drop_work, delay_drop_handler); - atomic_set(&dev->delay_drop.rqs_cnt, 0); - atomic_set(&dev->delay_drop.events_cnt, 0); - - delay_drop_debugfs_init(dev); -} - static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, struct mlx5_ib_multiport_info *mpi) { @@ -6203,26 +3705,20 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_VAR_OBJ_ALLOC)( mmap_offset = mlx5_entry_to_mmap_offset(entry); length = entry->rdma_entry.npages * PAGE_SIZE; uobj->object = entry; + uverbs_finalize_uobj_create(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE); err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET, &mmap_offset, sizeof(mmap_offset)); if (err) - goto err; + return err; err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID, &entry->page_idx, sizeof(entry->page_idx)); if (err) - goto err; + return err; err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH, &length, sizeof(length)); - if (err) - goto err; - - return 0; - -err: - rdma_user_mmap_entry_remove(&entry->rdma_entry); return err; } @@ -6336,26 +3832,20 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_UAR_OBJ_ALLOC)( mmap_offset = mlx5_entry_to_mmap_offset(entry); length = entry->rdma_entry.npages * PAGE_SIZE; uobj->object = entry; + uverbs_finalize_uobj_create(attrs, MLX5_IB_ATTR_UAR_OBJ_ALLOC_HANDLE); err = uverbs_copy_to(attrs, MLX5_IB_ATTR_UAR_OBJ_ALLOC_MMAP_OFFSET, &mmap_offset, sizeof(mmap_offset)); if (err) - goto err; + return err; err = uverbs_copy_to(attrs, MLX5_IB_ATTR_UAR_OBJ_ALLOC_PAGE_ID, &entry->page_idx, sizeof(entry->page_idx)); if (err) - goto err; + return err; err = uverbs_copy_to(attrs, MLX5_IB_ATTR_UAR_OBJ_ALLOC_MMAP_LENGTH, &length, sizeof(length)); - if (err) - goto err; - - return 0; - -err: - rdma_user_mmap_entry_remove(&entry->rdma_entry); return err; } @@ -6411,90 +3901,32 @@ ADD_UVERBS_ATTRIBUTES_SIMPLE( UVERBS_ATTR_FLAGS_IN(MLX5_IB_ATTR_CREATE_FLOW_ACTION_FLAGS, enum mlx5_ib_uapi_flow_action_flags)); +ADD_UVERBS_ATTRIBUTES_SIMPLE( + mlx5_ib_query_context, + UVERBS_OBJECT_DEVICE, + UVERBS_METHOD_QUERY_CONTEXT, + UVERBS_ATTR_PTR_OUT( + MLX5_IB_ATTR_QUERY_CONTEXT_RESP_UCTX, + UVERBS_ATTR_STRUCT(struct mlx5_ib_alloc_ucontext_resp, + dump_fill_mkey), + UA_MANDATORY)); + static const struct uapi_definition mlx5_ib_defs[] = { UAPI_DEF_CHAIN(mlx5_ib_devx_defs), UAPI_DEF_CHAIN(mlx5_ib_flow_defs), UAPI_DEF_CHAIN(mlx5_ib_qos_defs), + UAPI_DEF_CHAIN(mlx5_ib_std_types_defs), UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION, &mlx5_ib_flow_action), UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_DM, &mlx5_ib_dm), + UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_DEVICE, &mlx5_ib_query_context), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(MLX5_IB_OBJECT_VAR, UAPI_DEF_IS_OBJ_SUPPORTED(var_is_supported)), UAPI_DEF_CHAIN_OBJ_TREE_NAMED(MLX5_IB_OBJECT_UAR), {} }; -static int mlx5_ib_read_counters(struct ib_counters *counters, - struct ib_counters_read_attr *read_attr, - struct uverbs_attr_bundle *attrs) -{ - struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); - struct mlx5_read_counters_attr mread_attr = {}; - struct mlx5_ib_flow_counters_desc *desc; - int ret, i; - - mutex_lock(&mcounters->mcntrs_mutex); - if (mcounters->cntrs_max_index > read_attr->ncounters) { - ret = -EINVAL; - goto err_bound; - } - - mread_attr.out = kcalloc(mcounters->counters_num, sizeof(u64), - GFP_KERNEL); - if (!mread_attr.out) { - ret = -ENOMEM; - goto err_bound; - } - - mread_attr.hw_cntrs_hndl = mcounters->hw_cntrs_hndl; - mread_attr.flags = read_attr->flags; - ret = mcounters->read_counters(counters->device, &mread_attr); - if (ret) - goto err_read; - - /* do the pass over the counters data array to assign according to the - * descriptions and indexing pairs - */ - desc = mcounters->counters_data; - for (i = 0; i < mcounters->ncounters; i++) - read_attr->counters_buff[desc[i].index] += mread_attr.out[desc[i].description]; - -err_read: - kfree(mread_attr.out); -err_bound: - mutex_unlock(&mcounters->mcntrs_mutex); - return ret; -} - -static int mlx5_ib_destroy_counters(struct ib_counters *counters) -{ - struct mlx5_ib_mcounters *mcounters = to_mcounters(counters); - - counters_clear_description(counters); - if (mcounters->hw_cntrs_hndl) - mlx5_fc_destroy(to_mdev(counters->device)->mdev, - mcounters->hw_cntrs_hndl); - - kfree(mcounters); - - return 0; -} - -static struct ib_counters *mlx5_ib_create_counters(struct ib_device *device, - struct uverbs_attr_bundle *attrs) -{ - struct mlx5_ib_mcounters *mcounters; - - mcounters = kzalloc(sizeof(*mcounters), GFP_KERNEL); - if (!mcounters) - return ERR_PTR(-ENOMEM); - - mutex_init(&mcounters->mcntrs_mutex); - - return &mcounters->ibcntrs; -} - static void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev) { mlx5_ib_cleanup_multiport_master(dev); @@ -6549,6 +3981,7 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev) dev->ib_dev.phys_port_cnt = dev->num_ports; dev->ib_dev.num_comp_vectors = mlx5_comp_vectors_count(mdev); dev->ib_dev.dev.parent = mdev->device; + dev->ib_dev.lag_flags = RDMA_LAG_FLAGS_HASH_ALL_SLAVES; mutex_init(&dev->cap_mask_mutex); INIT_LIST_HEAD(&dev->qp_list); @@ -6572,21 +4005,16 @@ err_mp: return -ENOMEM; } -static int mlx5_ib_stage_flow_db_init(struct mlx5_ib_dev *dev) +static int mlx5_ib_enable_driver(struct ib_device *dev) { - dev->flow_db = kzalloc(sizeof(*dev->flow_db), GFP_KERNEL); - - if (!dev->flow_db) - return -ENOMEM; - - mutex_init(&dev->flow_db->lock); + struct mlx5_ib_dev *mdev = to_mdev(dev); + int ret; - return 0; -} + ret = mlx5_ib_test_wc(mdev); + mlx5_ib_dbg(mdev, "Write-Combining %s", + mdev->wc_support ? "supported" : "not supported"); -static void mlx5_ib_stage_flow_db_cleanup(struct mlx5_ib_dev *dev) -{ - kfree(dev->flow_db); + return ret; } static const struct ib_device_ops mlx5_ib_dev_ops = { @@ -6602,9 +4030,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = { .attach_mcast = mlx5_ib_mcg_attach, .check_mr_status = mlx5_ib_check_mr_status, .create_ah = mlx5_ib_create_ah, - .create_counters = mlx5_ib_create_counters, .create_cq = mlx5_ib_create_cq, - .create_flow = mlx5_ib_create_flow, .create_qp = mlx5_ib_create_qp, .create_srq = mlx5_ib_create_srq, .dealloc_pd = mlx5_ib_dealloc_pd, @@ -6612,10 +4038,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = { .del_gid = mlx5_ib_del_gid, .dereg_mr = mlx5_ib_dereg_mr, .destroy_ah = mlx5_ib_destroy_ah, - .destroy_counters = mlx5_ib_destroy_counters, .destroy_cq = mlx5_ib_destroy_cq, - .destroy_flow = mlx5_ib_destroy_flow, - .destroy_flow_action = mlx5_ib_destroy_flow_action, .destroy_qp = mlx5_ib_destroy_qp, .destroy_srq = mlx5_ib_destroy_srq, .detach_mcast = mlx5_ib_mcg_detach, @@ -6623,8 +4046,6 @@ static const struct ib_device_ops mlx5_ib_dev_ops = { .drain_rq = mlx5_ib_drain_rq, .drain_sq = mlx5_ib_drain_sq, .enable_driver = mlx5_ib_enable_driver, - .fill_res_entry = mlx5_ib_fill_res_entry, - .fill_stat_entry = mlx5_ib_fill_stat_entry, .get_dev_fw_str = get_dev_fw_str, .get_dma_mr = mlx5_ib_get_dma_mr, .get_link_layer = mlx5_ib_port_link_layer, @@ -6638,8 +4059,8 @@ static const struct ib_device_ops mlx5_ib_dev_ops = { .modify_qp = mlx5_ib_modify_qp, .modify_srq = mlx5_ib_modify_srq, .poll_cq = mlx5_ib_poll_cq, - .post_recv = mlx5_ib_post_recv, - .post_send = mlx5_ib_post_send, + .post_recv = mlx5_ib_post_recv_nodrain, + .post_send = mlx5_ib_post_send_nodrain, .post_srq_recv = mlx5_ib_post_srq_recv, .process_mad = mlx5_ib_process_mad, .query_ah = mlx5_ib_query_ah, @@ -6648,24 +4069,20 @@ static const struct ib_device_ops mlx5_ib_dev_ops = { .query_pkey = mlx5_ib_query_pkey, .query_qp = mlx5_ib_query_qp, .query_srq = mlx5_ib_query_srq, - .read_counters = mlx5_ib_read_counters, + .query_ucontext = mlx5_ib_query_ucontext, .reg_user_mr = mlx5_ib_reg_user_mr, .req_notify_cq = mlx5_ib_arm_cq, .rereg_user_mr = mlx5_ib_rereg_user_mr, .resize_cq = mlx5_ib_resize_cq, INIT_RDMA_OBJ_SIZE(ib_ah, mlx5_ib_ah, ibah), + INIT_RDMA_OBJ_SIZE(ib_counters, mlx5_ib_mcounters, ibcntrs), INIT_RDMA_OBJ_SIZE(ib_cq, mlx5_ib_cq, ibcq), INIT_RDMA_OBJ_SIZE(ib_pd, mlx5_ib_pd, ibpd), INIT_RDMA_OBJ_SIZE(ib_srq, mlx5_ib_srq, ibsrq), INIT_RDMA_OBJ_SIZE(ib_ucontext, mlx5_ib_ucontext, ibucontext), }; -static const struct ib_device_ops mlx5_ib_dev_flow_ipsec_ops = { - .create_flow_action_esp = mlx5_ib_create_flow_action_esp, - .modify_flow_action_esp = mlx5_ib_modify_flow_action_esp, -}; - static const struct ib_device_ops mlx5_ib_dev_ipoib_enhanced_ops = { .rdma_netdev_get_params = mlx5_ib_rn_get_params, }; @@ -6686,6 +4103,8 @@ static const struct ib_device_ops mlx5_ib_dev_mw_ops = { static const struct ib_device_ops mlx5_ib_dev_xrc_ops = { .alloc_xrcd = mlx5_ib_alloc_xrcd, .dealloc_xrcd = mlx5_ib_dealloc_xrcd, + + INIT_RDMA_OBJ_SIZE(ib_xrcd, mlx5_ib_xrcd, ibxrcd), }; static const struct ib_device_ops mlx5_ib_dev_dm_ops = { @@ -6794,9 +4213,6 @@ static int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev) MLX5_GENERAL_OBJ_TYPES_CAP_SW_ICM) ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_dm_ops); - if (mlx5_accel_ipsec_device_caps(dev->mdev) & - MLX5_ACCEL_IPSEC_CAP_DEVICE) - ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_flow_ipsec_ops); ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_ops); if (IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)) @@ -6854,65 +4270,36 @@ static const struct ib_device_ops mlx5_ib_dev_common_roce_ops = { .modify_wq = mlx5_ib_modify_wq, }; -static int mlx5_ib_stage_common_roce_init(struct mlx5_ib_dev *dev) -{ - u8 port_num; - - dev->ib_dev.uverbs_ex_cmd_mask |= - (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); - ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_common_roce_ops); - - port_num = mlx5_core_native_port_num(dev->mdev) - 1; - - /* Register only for native ports */ - return mlx5_add_netdev_notifier(dev, port_num); -} - -static void mlx5_ib_stage_common_roce_cleanup(struct mlx5_ib_dev *dev) -{ - u8 port_num = mlx5_core_native_port_num(dev->mdev) - 1; - - mlx5_remove_netdev_notifier(dev, port_num); -} - -static int mlx5_ib_stage_raw_eth_roce_init(struct mlx5_ib_dev *dev) -{ - struct mlx5_core_dev *mdev = dev->mdev; - enum rdma_link_layer ll; - int port_type_cap; - int err = 0; - - port_type_cap = MLX5_CAP_GEN(mdev, port_type); - ll = mlx5_port_type_cap_to_rdma_ll(port_type_cap); - - if (ll == IB_LINK_LAYER_ETHERNET) - err = mlx5_ib_stage_common_roce_init(dev); - - return err; -} - -static void mlx5_ib_stage_raw_eth_roce_cleanup(struct mlx5_ib_dev *dev) -{ - mlx5_ib_stage_common_roce_cleanup(dev); -} - -static int mlx5_ib_stage_roce_init(struct mlx5_ib_dev *dev) +static int mlx5_ib_roce_init(struct mlx5_ib_dev *dev) { struct mlx5_core_dev *mdev = dev->mdev; enum rdma_link_layer ll; int port_type_cap; + u8 port_num = 0; int err; port_type_cap = MLX5_CAP_GEN(mdev, port_type); ll = mlx5_port_type_cap_to_rdma_ll(port_type_cap); if (ll == IB_LINK_LAYER_ETHERNET) { - err = mlx5_ib_stage_common_roce_init(dev); - if (err) + dev->ib_dev.uverbs_ex_cmd_mask |= + (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); + ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_common_roce_ops); + + port_num = mlx5_core_native_port_num(dev->mdev) - 1; + + /* Register only for native ports */ + err = mlx5_add_netdev_notifier(dev, port_num); + if (err || dev->is_rep || !mlx5_is_roce_enabled(mdev)) + /* + * We don't enable ETH interface for + * 1. IB representors + * 2. User disabled ROCE through devlink interface + */ return err; err = mlx5_enable_eth(dev); @@ -6922,71 +4309,27 @@ static int mlx5_ib_stage_roce_init(struct mlx5_ib_dev *dev) return 0; cleanup: - mlx5_ib_stage_common_roce_cleanup(dev); - + mlx5_remove_netdev_notifier(dev, port_num); return err; } -static void mlx5_ib_stage_roce_cleanup(struct mlx5_ib_dev *dev) +static void mlx5_ib_roce_cleanup(struct mlx5_ib_dev *dev) { struct mlx5_core_dev *mdev = dev->mdev; enum rdma_link_layer ll; int port_type_cap; + u8 port_num; port_type_cap = MLX5_CAP_GEN(mdev, port_type); ll = mlx5_port_type_cap_to_rdma_ll(port_type_cap); if (ll == IB_LINK_LAYER_ETHERNET) { - mlx5_disable_eth(dev); - mlx5_ib_stage_common_roce_cleanup(dev); - } -} - -static int mlx5_ib_stage_dev_res_init(struct mlx5_ib_dev *dev) -{ - return create_dev_resources(&dev->devr); -} + if (!dev->is_rep) + mlx5_disable_eth(dev); -static void mlx5_ib_stage_dev_res_cleanup(struct mlx5_ib_dev *dev) -{ - destroy_dev_resources(&dev->devr); -} - -static int mlx5_ib_stage_odp_init(struct mlx5_ib_dev *dev) -{ - return mlx5_ib_odp_init_one(dev); -} - -static void mlx5_ib_stage_odp_cleanup(struct mlx5_ib_dev *dev) -{ - mlx5_ib_odp_cleanup_one(dev); -} - -static const struct ib_device_ops mlx5_ib_dev_hw_stats_ops = { - .alloc_hw_stats = mlx5_ib_alloc_hw_stats, - .get_hw_stats = mlx5_ib_get_hw_stats, - .counter_bind_qp = mlx5_ib_counter_bind_qp, - .counter_unbind_qp = mlx5_ib_counter_unbind_qp, - .counter_dealloc = mlx5_ib_counter_dealloc, - .counter_alloc_stats = mlx5_ib_counter_alloc_stats, - .counter_update_stats = mlx5_ib_counter_update_stats, -}; - -static int mlx5_ib_stage_counters_init(struct mlx5_ib_dev *dev) -{ - if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) { - ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_hw_stats_ops); - - return mlx5_ib_alloc_counters(dev); + port_num = mlx5_core_native_port_num(dev->mdev) - 1; + mlx5_remove_netdev_notifier(dev, port_num); } - - return 0; -} - -static void mlx5_ib_stage_counters_cleanup(struct mlx5_ib_dev *dev) -{ - if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) - mlx5_ib_dealloc_counters(dev); } static int mlx5_ib_stage_cong_debugfs_init(struct mlx5_ib_dev *dev) @@ -7048,7 +4391,18 @@ static int mlx5_ib_stage_ib_reg_init(struct mlx5_ib_dev *dev) static void mlx5_ib_stage_pre_ib_reg_umr_cleanup(struct mlx5_ib_dev *dev) { - destroy_umrc_res(dev); + int err; + + err = mlx5_mr_cache_cleanup(dev); + if (err) + mlx5_ib_warn(dev, "mr cache cleanup failed\n"); + + if (dev->umrc.qp) + mlx5_ib_destroy_qp(dev->umrc.qp, NULL); + if (dev->umrc.cq) + ib_free_cq(dev->umrc.cq); + if (dev->umrc.pd) + ib_dealloc_pd(dev->umrc.pd); } static void mlx5_ib_stage_ib_reg_cleanup(struct mlx5_ib_dev *dev) @@ -7056,21 +4410,162 @@ static void mlx5_ib_stage_ib_reg_cleanup(struct mlx5_ib_dev *dev) ib_unregister_device(&dev->ib_dev); } +enum { + MAX_UMR_WR = 128, +}; + static int mlx5_ib_stage_post_ib_reg_umr_init(struct mlx5_ib_dev *dev) { - return create_umr_res(dev); + struct ib_qp_init_attr *init_attr = NULL; + struct ib_qp_attr *attr = NULL; + struct ib_pd *pd; + struct ib_cq *cq; + struct ib_qp *qp; + int ret; + + attr = kzalloc(sizeof(*attr), GFP_KERNEL); + init_attr = kzalloc(sizeof(*init_attr), GFP_KERNEL); + if (!attr || !init_attr) { + ret = -ENOMEM; + goto error_0; + } + + pd = ib_alloc_pd(&dev->ib_dev, 0); + if (IS_ERR(pd)) { + mlx5_ib_dbg(dev, "Couldn't create PD for sync UMR QP\n"); + ret = PTR_ERR(pd); + goto error_0; + } + + cq = ib_alloc_cq(&dev->ib_dev, NULL, 128, 0, IB_POLL_SOFTIRQ); + if (IS_ERR(cq)) { + mlx5_ib_dbg(dev, "Couldn't create CQ for sync UMR QP\n"); + ret = PTR_ERR(cq); + goto error_2; + } + + init_attr->send_cq = cq; + init_attr->recv_cq = cq; + init_attr->sq_sig_type = IB_SIGNAL_ALL_WR; + init_attr->cap.max_send_wr = MAX_UMR_WR; + init_attr->cap.max_send_sge = 1; + init_attr->qp_type = MLX5_IB_QPT_REG_UMR; + init_attr->port_num = 1; + qp = mlx5_ib_create_qp(pd, init_attr, NULL); + if (IS_ERR(qp)) { + mlx5_ib_dbg(dev, "Couldn't create sync UMR QP\n"); + ret = PTR_ERR(qp); + goto error_3; + } + qp->device = &dev->ib_dev; + qp->real_qp = qp; + qp->uobject = NULL; + qp->qp_type = MLX5_IB_QPT_REG_UMR; + qp->send_cq = init_attr->send_cq; + qp->recv_cq = init_attr->recv_cq; + + attr->qp_state = IB_QPS_INIT; + attr->port_num = 1; + ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE | IB_QP_PKEY_INDEX | + IB_QP_PORT, NULL); + if (ret) { + mlx5_ib_dbg(dev, "Couldn't modify UMR QP\n"); + goto error_4; + } + + memset(attr, 0, sizeof(*attr)); + attr->qp_state = IB_QPS_RTR; + attr->path_mtu = IB_MTU_256; + + ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE, NULL); + if (ret) { + mlx5_ib_dbg(dev, "Couldn't modify umr QP to rtr\n"); + goto error_4; + } + + memset(attr, 0, sizeof(*attr)); + attr->qp_state = IB_QPS_RTS; + ret = mlx5_ib_modify_qp(qp, attr, IB_QP_STATE, NULL); + if (ret) { + mlx5_ib_dbg(dev, "Couldn't modify umr QP to rts\n"); + goto error_4; + } + + dev->umrc.qp = qp; + dev->umrc.cq = cq; + dev->umrc.pd = pd; + + sema_init(&dev->umrc.sem, MAX_UMR_WR); + ret = mlx5_mr_cache_init(dev); + if (ret) { + mlx5_ib_warn(dev, "mr cache init failed %d\n", ret); + goto error_4; + } + + kfree(attr); + kfree(init_attr); + + return 0; + +error_4: + mlx5_ib_destroy_qp(qp, NULL); + dev->umrc.qp = NULL; + +error_3: + ib_free_cq(cq); + dev->umrc.cq = NULL; + +error_2: + ib_dealloc_pd(pd); + dev->umrc.pd = NULL; + +error_0: + kfree(attr); + kfree(init_attr); + return ret; } static int mlx5_ib_stage_delay_drop_init(struct mlx5_ib_dev *dev) { - init_delay_drop(dev); + struct dentry *root; + if (!(dev->ib_dev.attrs.raw_packet_caps & IB_RAW_PACKET_CAP_DELAY_DROP)) + return 0; + + mutex_init(&dev->delay_drop.lock); + dev->delay_drop.dev = dev; + dev->delay_drop.activate = false; + dev->delay_drop.timeout = MLX5_MAX_DELAY_DROP_TIMEOUT_MS * 1000; + INIT_WORK(&dev->delay_drop.delay_drop_work, delay_drop_handler); + atomic_set(&dev->delay_drop.rqs_cnt, 0); + atomic_set(&dev->delay_drop.events_cnt, 0); + + if (!mlx5_debugfs_root) + return 0; + + root = debugfs_create_dir("delay_drop", dev->mdev->priv.dbg_root); + dev->delay_drop.dir_debugfs = root; + + debugfs_create_atomic_t("num_timeout_events", 0400, root, + &dev->delay_drop.events_cnt); + debugfs_create_atomic_t("num_rqs", 0400, root, + &dev->delay_drop.rqs_cnt); + debugfs_create_file("timeout", 0600, root, &dev->delay_drop, + &fops_delay_drop_timeout); return 0; } static void mlx5_ib_stage_delay_drop_cleanup(struct mlx5_ib_dev *dev) { - cancel_delay_drop(dev); + if (!(dev->ib_dev.attrs.raw_packet_caps & IB_RAW_PACKET_CAP_DELAY_DROP)) + return; + + cancel_work_sync(&dev->delay_drop.delay_drop_work); + if (!dev->delay_drop.dir_debugfs) + return; + + debugfs_remove_recursive(dev->delay_drop.dir_debugfs); + dev->delay_drop.dir_debugfs = NULL; } static int mlx5_ib_stage_dev_notifier_init(struct mlx5_ib_dev *dev) @@ -7085,38 +4580,6 @@ static void mlx5_ib_stage_dev_notifier_cleanup(struct mlx5_ib_dev *dev) mlx5_notifier_unregister(dev->mdev, &dev->mdev_events); } -static int mlx5_ib_stage_devx_init(struct mlx5_ib_dev *dev) -{ - int uid; - - uid = mlx5_ib_devx_create(dev, false); - if (uid > 0) { - dev->devx_whitelist_uid = uid; - mlx5_ib_devx_init_event_table(dev); - } - - return 0; -} -static void mlx5_ib_stage_devx_cleanup(struct mlx5_ib_dev *dev) -{ - if (dev->devx_whitelist_uid) { - mlx5_ib_devx_cleanup_event_table(dev); - mlx5_ib_devx_destroy(dev, dev->devx_whitelist_uid); - } -} - -int mlx5_ib_enable_driver(struct ib_device *dev) -{ - struct mlx5_ib_dev *mdev = to_mdev(dev); - int ret; - - ret = mlx5_ib_test_wc(mdev); - mlx5_ib_dbg(mdev, "Write-Combining %s", - mdev->wc_support ? "supported" : "not supported"); - - return ret; -} - void __mlx5_ib_remove(struct mlx5_ib_dev *dev, const struct mlx5_ib_profile *profile, int stage) @@ -7140,6 +4603,8 @@ void *__mlx5_ib_add(struct mlx5_ib_dev *dev, int err; int i; + dev->profile = profile; + for (i = 0; i < MLX5_IB_STAGE_MAX; i++) { if (profile->stage[i].init) { err = profile->stage[i].init(dev); @@ -7148,7 +4613,6 @@ void *__mlx5_ib_add(struct mlx5_ib_dev *dev, } } - dev->profile = profile; dev->ib_active = true; return dev; @@ -7163,9 +4627,9 @@ static const struct mlx5_ib_profile pf_profile = { STAGE_CREATE(MLX5_IB_STAGE_INIT, mlx5_ib_stage_init_init, mlx5_ib_stage_init_cleanup), - STAGE_CREATE(MLX5_IB_STAGE_FLOW_DB, - mlx5_ib_stage_flow_db_init, - mlx5_ib_stage_flow_db_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_FS, + mlx5_ib_fs_init, + mlx5_ib_fs_cleanup), STAGE_CREATE(MLX5_IB_STAGE_CAPS, mlx5_ib_stage_caps_init, mlx5_ib_stage_caps_cleanup), @@ -7173,23 +4637,26 @@ static const struct mlx5_ib_profile pf_profile = { mlx5_ib_stage_non_default_cb, NULL), STAGE_CREATE(MLX5_IB_STAGE_ROCE, - mlx5_ib_stage_roce_init, - mlx5_ib_stage_roce_cleanup), + mlx5_ib_roce_init, + mlx5_ib_roce_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_QP, + mlx5_init_qp_table, + mlx5_cleanup_qp_table), STAGE_CREATE(MLX5_IB_STAGE_SRQ, mlx5_init_srq_table, mlx5_cleanup_srq_table), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_RESOURCES, - mlx5_ib_stage_dev_res_init, - mlx5_ib_stage_dev_res_cleanup), + mlx5_ib_dev_res_init, + mlx5_ib_dev_res_cleanup), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_NOTIFIER, mlx5_ib_stage_dev_notifier_init, mlx5_ib_stage_dev_notifier_cleanup), STAGE_CREATE(MLX5_IB_STAGE_ODP, - mlx5_ib_stage_odp_init, - mlx5_ib_stage_odp_cleanup), + mlx5_ib_odp_init_one, + mlx5_ib_odp_cleanup_one), STAGE_CREATE(MLX5_IB_STAGE_COUNTERS, - mlx5_ib_stage_counters_init, - mlx5_ib_stage_counters_cleanup), + mlx5_ib_counters_init, + mlx5_ib_counters_cleanup), STAGE_CREATE(MLX5_IB_STAGE_CONG_DEBUGFS, mlx5_ib_stage_cong_debugfs_init, mlx5_ib_stage_cong_debugfs_cleanup), @@ -7203,8 +4670,8 @@ static const struct mlx5_ib_profile pf_profile = { NULL, mlx5_ib_stage_pre_ib_reg_umr_cleanup), STAGE_CREATE(MLX5_IB_STAGE_WHITELIST_UID, - mlx5_ib_stage_devx_init, - mlx5_ib_stage_devx_cleanup), + mlx5_ib_devx_init, + mlx5_ib_devx_cleanup), STAGE_CREATE(MLX5_IB_STAGE_IB_REG, mlx5_ib_stage_ib_reg_init, mlx5_ib_stage_ib_reg_cleanup), @@ -7214,15 +4681,18 @@ static const struct mlx5_ib_profile pf_profile = { STAGE_CREATE(MLX5_IB_STAGE_DELAY_DROP, mlx5_ib_stage_delay_drop_init, mlx5_ib_stage_delay_drop_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_RESTRACK, + mlx5_ib_restrack_init, + NULL), }; const struct mlx5_ib_profile raw_eth_profile = { STAGE_CREATE(MLX5_IB_STAGE_INIT, mlx5_ib_stage_init_init, mlx5_ib_stage_init_cleanup), - STAGE_CREATE(MLX5_IB_STAGE_FLOW_DB, - mlx5_ib_stage_flow_db_init, - mlx5_ib_stage_flow_db_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_FS, + mlx5_ib_fs_init, + mlx5_ib_fs_cleanup), STAGE_CREATE(MLX5_IB_STAGE_CAPS, mlx5_ib_stage_caps_init, mlx5_ib_stage_caps_cleanup), @@ -7230,20 +4700,23 @@ const struct mlx5_ib_profile raw_eth_profile = { mlx5_ib_stage_raw_eth_non_default_cb, NULL), STAGE_CREATE(MLX5_IB_STAGE_ROCE, - mlx5_ib_stage_raw_eth_roce_init, - mlx5_ib_stage_raw_eth_roce_cleanup), + mlx5_ib_roce_init, + mlx5_ib_roce_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_QP, + mlx5_init_qp_table, + mlx5_cleanup_qp_table), STAGE_CREATE(MLX5_IB_STAGE_SRQ, mlx5_init_srq_table, mlx5_cleanup_srq_table), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_RESOURCES, - mlx5_ib_stage_dev_res_init, - mlx5_ib_stage_dev_res_cleanup), + mlx5_ib_dev_res_init, + mlx5_ib_dev_res_cleanup), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_NOTIFIER, mlx5_ib_stage_dev_notifier_init, mlx5_ib_stage_dev_notifier_cleanup), STAGE_CREATE(MLX5_IB_STAGE_COUNTERS, - mlx5_ib_stage_counters_init, - mlx5_ib_stage_counters_cleanup), + mlx5_ib_counters_init, + mlx5_ib_counters_cleanup), STAGE_CREATE(MLX5_IB_STAGE_CONG_DEBUGFS, mlx5_ib_stage_cong_debugfs_init, mlx5_ib_stage_cong_debugfs_cleanup), @@ -7257,14 +4730,17 @@ const struct mlx5_ib_profile raw_eth_profile = { NULL, mlx5_ib_stage_pre_ib_reg_umr_cleanup), STAGE_CREATE(MLX5_IB_STAGE_WHITELIST_UID, - mlx5_ib_stage_devx_init, - mlx5_ib_stage_devx_cleanup), + mlx5_ib_devx_init, + mlx5_ib_devx_cleanup), STAGE_CREATE(MLX5_IB_STAGE_IB_REG, mlx5_ib_stage_ib_reg_init, mlx5_ib_stage_ib_reg_cleanup), STAGE_CREATE(MLX5_IB_STAGE_POST_IB_REG_UMR, mlx5_ib_stage_post_ib_reg_umr_init, NULL), + STAGE_CREATE(MLX5_IB_STAGE_RESTRACK, + mlx5_ib_restrack_init, + NULL), }; static void *mlx5_ib_add_slave_port(struct mlx5_core_dev *mdev) @@ -7316,8 +4792,6 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) int port_type_cap; int num_ports; - printk_once(KERN_INFO "%s", mlx5_version); - if (MLX5_ESWITCH_MANAGER(mdev) && mlx5_ib_eswitch_mode(mdev->priv.eswitch) == MLX5_ESWITCH_OFFLOADS) { if (!mlx5_core_mp_enabled(mdev)) diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index a4e522385de0..5287fc868662 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -1,33 +1,6 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ /* - * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved. - * - * This software is available to you under a choice of one of two - * licenses. You may choose to be licensed under the terms of the GNU - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. */ #ifndef MLX5_IB_H @@ -337,7 +310,6 @@ struct mlx5_ib_rwq { 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; @@ -346,17 +318,6 @@ struct mlx5_ib_rwq { u32 create_flags; /* Use enum mlx5_ib_wq_flags */ }; -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; @@ -443,34 +404,37 @@ struct mlx5_ib_qp { /* serialize qp state modifications */ struct mutex mutex; + /* cached variant of create_flags from struct ib_qp_init_attr */ u32 flags; u8 port; u8 state; - int wq_sig; - int scat_cqe; int max_inline_data; struct mlx5_bf bf; - int has_rq; + u8 has_rq:1; + u8 is_rss:1; /* only for user space QPs. For kernel * we have it from the bf object */ int bfregn; - int create_type; - struct list_head qps_list; struct list_head cq_recv_list; struct list_head cq_send_list; struct mlx5_rate_limit rl; u32 underlay_qpn; u32 flags_en; - /* storage for qp sub type when core qp type is IB_QPT_DRIVER */ - enum ib_qp_type qp_sub_type; + /* + * IB/core doesn't store low-level QP types, so + * store both MLX and IBTA types in the field below. + * IB_QPT_DRIVER will be break to DCI/DCT subtypes. + */ + enum ib_qp_type type; /* A flag to indicate if there's a new counter is configured * but not take effective */ u32 counter_pending; + u16 gsi_lag_port; }; struct mlx5_ib_cq_buf { @@ -481,24 +445,6 @@ struct mlx5_ib_cq_buf { int nent; }; -enum mlx5_ib_qp_flags { - MLX5_IB_QP_LSO = IB_QP_CREATE_IPOIB_UD_LSO, - MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK = IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK, - MLX5_IB_QP_CROSS_CHANNEL = IB_QP_CREATE_CROSS_CHANNEL, - MLX5_IB_QP_MANAGED_SEND = IB_QP_CREATE_MANAGED_SEND, - MLX5_IB_QP_MANAGED_RECV = IB_QP_CREATE_MANAGED_RECV, - MLX5_IB_QP_SIGNATURE_HANDLING = 1 << 5, - /* QP uses 1 as its source QP number */ - MLX5_IB_QP_SQPN_QP1 = 1 << 6, - MLX5_IB_QP_CAP_SCATTER_FCS = 1 << 7, - MLX5_IB_QP_RSS = 1 << 8, - MLX5_IB_QP_CVLAN_STRIPPING = 1 << 9, - MLX5_IB_QP_UNDERLAY = 1 << 10, - MLX5_IB_QP_PCI_WRITE_END_PADDING = 1 << 11, - MLX5_IB_QP_TUNNEL_OFFLOAD = 1 << 12, - MLX5_IB_QP_PACKET_BASED_CREDIT = 1 << 13, -}; - struct mlx5_umr_wr { struct ib_send_wr wr; u64 virt_addr; @@ -702,12 +648,6 @@ struct umr_common { struct semaphore sem; }; -enum { - MLX5_FMR_INVALID, - MLX5_FMR_VALID, - MLX5_FMR_BUSY, -}; - struct mlx5_cache_ent { struct list_head head; /* sync access to the cahce entry @@ -763,8 +703,8 @@ struct mlx5_ib_port_resources { struct mlx5_ib_resources { struct ib_cq *c0; - struct ib_xrcd *x0; - struct ib_xrcd *x1; + u32 xrcdn0; + u32 xrcdn1; struct ib_pd *p0; struct ib_srq *s0; struct ib_srq *s1; @@ -780,7 +720,6 @@ struct mlx5_ib_counters { u32 num_cong_counters; u32 num_ext_ppcnt_counters; u16 set_id; - bool set_id_valid; }; struct mlx5_ib_multiport_info; @@ -866,10 +805,11 @@ struct mlx5_ib_delay_drop { enum mlx5_ib_stages { MLX5_IB_STAGE_INIT, - MLX5_IB_STAGE_FLOW_DB, + MLX5_IB_STAGE_FS, MLX5_IB_STAGE_CAPS, MLX5_IB_STAGE_NON_DEFAULT_CB, MLX5_IB_STAGE_ROCE, + MLX5_IB_STAGE_QP, MLX5_IB_STAGE_SRQ, MLX5_IB_STAGE_DEVICE_RESOURCES, MLX5_IB_STAGE_DEVICE_NOTIFIER, @@ -883,7 +823,7 @@ enum mlx5_ib_stages { MLX5_IB_STAGE_IB_REG, MLX5_IB_STAGE_POST_IB_REG_UMR, MLX5_IB_STAGE_DELAY_DROP, - MLX5_IB_STAGE_CLASS_ATTR, + MLX5_IB_STAGE_RESTRACK, MLX5_IB_STAGE_MAX, }; @@ -1065,6 +1005,7 @@ struct mlx5_ib_dev { struct mlx5_dm dm; u16 devx_whitelist_uid; struct mlx5_srq_table srq_table; + struct mlx5_qp_table qp_table; struct mlx5_async_ctx async_ctx; struct mlx5_devx_event_table devx_event_table; struct mlx5_var_table var_table; @@ -1110,11 +1051,6 @@ static inline struct mlx5_ib_rwq *to_mibrwq(struct mlx5_core_qp *core_qp) return container_of(core_qp, struct mlx5_ib_rwq, core_qp); } -static inline struct mlx5_ib_mr *to_mibmr(struct mlx5_core_mkey *mmkey) -{ - return container_of(mmkey, struct mlx5_ib_mr, mmkey); -} - static inline struct mlx5_ib_pd *to_mpd(struct ib_pd *ibpd) { return container_of(ibpd, struct mlx5_ib_pd, ibpd); @@ -1180,7 +1116,7 @@ void mlx5_ib_db_unmap_user(struct mlx5_ib_ucontext *context, struct mlx5_db *db) void __mlx5_ib_cq_clean(struct mlx5_ib_cq *cq, u32 qpn, struct mlx5_ib_srq *srq); void mlx5_ib_cq_clean(struct mlx5_ib_cq *cq, u32 qpn, struct mlx5_ib_srq *srq); void mlx5_ib_free_srq_wqe(struct mlx5_ib_srq *srq, int wqe_index); -int mlx5_ib_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, u32 flags, +int mlx5_ib_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); int mlx5_ib_query_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr); void mlx5_ib_destroy_ah(struct ib_ah *ah, u32 flags); @@ -1204,10 +1140,6 @@ int mlx5_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr int mlx5_ib_destroy_qp(struct ib_qp *qp, struct ib_udata *udata); void mlx5_ib_drain_sq(struct ib_qp *qp); void mlx5_ib_drain_rq(struct ib_qp *qp); -int mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, - const struct ib_send_wr **bad_wr); -int mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, - const struct ib_recv_wr **bad_wr); int mlx5_ib_read_wqe_sq(struct mlx5_ib_qp *qp, int wqe_index, void *buffer, size_t buflen, size_t *bc); int mlx5_ib_read_wqe_rq(struct mlx5_ib_qp *qp, int wqe_index, void *buffer, @@ -1246,7 +1178,7 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start, struct ib_pd *pd, struct ib_udata *udata); int mlx5_ib_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata); struct ib_mr *mlx5_ib_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); struct ib_mr *mlx5_ib_alloc_mr_integrity(struct ib_pd *pd, u32 max_num_sg, u32 max_num_meta_sg); @@ -1260,9 +1192,8 @@ int mlx5_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, const struct ib_wc *in_wc, const struct ib_grh *in_grh, const struct ib_mad *in, struct ib_mad *out, size_t *out_mad_size, u16 *out_mad_pkey_index); -struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, - struct ib_udata *udata); -int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata); +int mlx5_ib_alloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata); +void mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata); int mlx5_ib_get_buf_offset(u64 addr, int page_shift, u32 *offset); int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port); int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev, @@ -1283,8 +1214,6 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props); int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props); -int mlx5_ib_init_fmr(struct mlx5_ib_dev *dev); -void mlx5_ib_cleanup_fmr(struct mlx5_ib_dev *dev); void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr, unsigned long max_page_shift, int *count, int *shift, @@ -1382,8 +1311,8 @@ int mlx5_ib_get_vf_guid(struct ib_device *device, int vf, u8 port, int mlx5_ib_set_vf_guid(struct ib_device *device, int vf, u8 port, u64 guid, int type); -__be16 mlx5_get_roce_udp_sport(struct mlx5_ib_dev *dev, - const struct ib_gid_attr *attr); +__be16 mlx5_get_roce_udp_sport_min(const struct mlx5_ib_dev *dev, + const struct ib_gid_attr *attr); void mlx5_ib_cleanup_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num); void mlx5_ib_init_cong_debugfs(struct mlx5_ib_dev *dev, u8 port_num); @@ -1413,46 +1342,12 @@ struct mlx5_core_dev *mlx5_ib_get_native_port_mdev(struct mlx5_ib_dev *dev, u8 *native_port_num); void mlx5_ib_put_native_port_mdev(struct mlx5_ib_dev *dev, u8 port_num); -int mlx5_ib_fill_res_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res); -int mlx5_ib_fill_stat_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res); extern const struct uapi_definition mlx5_ib_devx_defs[]; extern const struct uapi_definition mlx5_ib_flow_defs[]; extern const struct uapi_definition mlx5_ib_qos_defs[]; +extern const struct uapi_definition mlx5_ib_std_types_defs[]; -#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) -int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user); -void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid); -void mlx5_ib_devx_init_event_table(struct mlx5_ib_dev *dev); -void mlx5_ib_devx_cleanup_event_table(struct mlx5_ib_dev *dev); -struct mlx5_ib_flow_handler *mlx5_ib_raw_fs_rule_add( - struct mlx5_ib_dev *dev, struct mlx5_ib_flow_matcher *fs_matcher, - struct mlx5_flow_context *flow_context, - struct mlx5_flow_act *flow_act, u32 counter_id, - void *cmd_in, int inlen, int dest_id, int dest_type); -bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, int *dest_type); -bool mlx5_ib_devx_is_flow_counter(void *obj, u32 offset, u32 *counter_id); -void mlx5_ib_destroy_flow_action_raw(struct mlx5_ib_flow_action *maction); -#else -static inline int -mlx5_ib_devx_create(struct mlx5_ib_dev *dev, - bool is_user) { return -EOPNOTSUPP; } -static inline void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid) {} -static inline void mlx5_ib_devx_init_event_table(struct mlx5_ib_dev *dev) {} -static inline void mlx5_ib_devx_cleanup_event_table(struct mlx5_ib_dev *dev) {} -static inline bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, - int *dest_type) -{ - return false; -} -static inline void -mlx5_ib_destroy_flow_action_raw(struct mlx5_ib_flow_action *maction) -{ - return; -}; -#endif static inline void init_query_mad(struct ib_smp *mad) { mad->base_version = 1; @@ -1461,15 +1356,6 @@ static inline void init_query_mad(struct ib_smp *mad) mad->method = IB_MGMT_METHOD_GET; } -static inline u8 convert_access(int acc) -{ - return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX5_PERM_ATOMIC : 0) | - (acc & IB_ACCESS_REMOTE_WRITE ? MLX5_PERM_REMOTE_WRITE : 0) | - (acc & IB_ACCESS_REMOTE_READ ? MLX5_PERM_REMOTE_READ : 0) | - (acc & IB_ACCESS_LOCAL_WRITE ? MLX5_PERM_LOCAL_WRITE : 0) | - MLX5_PERM_LOCAL_READ; -} - static inline int is_qp1(enum ib_qp_type qp_type) { return qp_type == MLX5_IB_QPT_HW_GSI; @@ -1556,9 +1442,6 @@ int bfregn_to_uar_index(struct mlx5_ib_dev *dev, struct mlx5_bfreg_info *bfregi, u32 bfregn, bool dyn_bfreg); -int mlx5_ib_qp_set_counter(struct ib_qp *qp, struct rdma_counter *counter); -u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num); - static inline bool mlx5_ib_can_use_umr(struct mlx5_ib_dev *dev, bool do_modify_atomic, int access_flags) { @@ -1571,13 +1454,24 @@ static inline bool mlx5_ib_can_use_umr(struct mlx5_ib_dev *dev, return false; if (access_flags & IB_ACCESS_RELAXED_ORDERING && - (MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write) || - MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read))) + MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write) && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write_umr)) + return false; + + if (access_flags & IB_ACCESS_RELAXED_ORDERING && + MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read) && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr)) return false; return true; } -int mlx5_ib_enable_driver(struct ib_device *dev); int mlx5_ib_test_wc(struct mlx5_ib_dev *dev); + +static inline bool mlx5_ib_lag_should_assign_affinity(struct mlx5_ib_dev *dev) +{ + return dev->lag_active || + (MLX5_CAP_GEN(dev->mdev, num_lag_ports) > 1 && + MLX5_CAP_GEN(dev->mdev, lag_tx_port_affinity)); +} #endif /* MLX5_IB_H */ diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index 44683073be0c..3e6f2f9c6655 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -1961,7 +1961,7 @@ err_free: } struct ib_mr *mlx5_ib_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { return __mlx5_ib_alloc_mr(pd, mr_type, max_num_sg, 0); } diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 3de7606d4a1a..cfd7efab114e 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -36,6 +36,7 @@ #include "mlx5_ib.h" #include "cmd.h" +#include "qp.h" #include <linux/mlx5/eq.h> @@ -446,8 +447,7 @@ static void mlx5_ib_page_fault_resume(struct mlx5_ib_dev *dev, { int wq_num = pfault->event_subtype == MLX5_PFAULT_SUBTYPE_WQE ? pfault->wqe.wq_num : pfault->token; - u32 out[MLX5_ST_SZ_DW(page_fault_resume_out)] = { }; - u32 in[MLX5_ST_SZ_DW(page_fault_resume_in)] = { }; + u32 in[MLX5_ST_SZ_DW(page_fault_resume_in)] = {}; int err; MLX5_SET(page_fault_resume_in, in, opcode, MLX5_CMD_OP_PAGE_FAULT_RESUME); @@ -456,7 +456,7 @@ static void mlx5_ib_page_fault_resume(struct mlx5_ib_dev *dev, MLX5_SET(page_fault_resume_in, in, wq_number, wq_num); MLX5_SET(page_fault_resume_in, in, error, !!error); - err = mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); + err = mlx5_cmd_exec_in(dev->mdev, page_fault_resume, in); if (err) mlx5_ib_err(dev, "Failed to resolve the page fault on WQ 0x%x err %d\n", wq_num, err); @@ -601,6 +601,23 @@ void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr) */ synchronize_srcu(&dev->odp_srcu); + /* + * All work on the prefetch list must be completed, xa_erase() prevented + * new work from being created. + */ + wait_event(imr->q_deferred_work, !atomic_read(&imr->num_deferred_work)); + + /* + * At this point it is forbidden for any other thread to enter + * pagefault_mr() on this imr. It is already forbidden to call + * pagefault_mr() on an implicit child. Due to this additions to + * implicit_children are prevented. + */ + + /* + * Block destroy_unused_implicit_child_mr() from incrementing + * num_deferred_work. + */ xa_lock(&imr->implicit_children); xa_for_each (&imr->implicit_children, idx, mtt) { __xa_erase(&imr->implicit_children, idx); @@ -609,9 +626,8 @@ void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr) xa_unlock(&imr->implicit_children); /* - * num_deferred_work can only be incremented inside the odp_srcu, or - * under xa_lock while the child is in the xarray. Thus at this point - * it is only decreasing, and all work holding it is now on the wq. + * Wait for any concurrent destroy_unused_implicit_child_mr() to + * complete. */ wait_event(imr->q_deferred_work, !atomic_read(&imr->num_deferred_work)); @@ -800,6 +816,7 @@ static int pagefault_mr(struct mlx5_ib_mr *mr, u64 io_virt, size_t bcnt, { struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem); + lockdep_assert_held(&mr->dev->odp_srcu); if (unlikely(io_virt < mr->mmkey.iova)) return -EFAULT; @@ -913,11 +930,6 @@ next_mr: if (ret < 0) goto srcu_unlock; - /* - * When prefetching a page, page fault is generated - * in order to bring the page to the main memory. - * In the current flow, page faults are being counted. - */ mlx5_update_odp_stats(mr, faults, ret); npages += ret; @@ -1135,8 +1147,7 @@ static int mlx5_ib_mr_initiator_pfault_handler( if (qp->ibqp.qp_type == IB_QPT_XRC_INI) *wqe += sizeof(struct mlx5_wqe_xrc_seg); - if (qp->ibqp.qp_type == IB_QPT_UD || - qp->qp_sub_type == MLX5_IB_QPT_DCI) { + if (qp->type == IB_QPT_UD || qp->type == MLX5_IB_QPT_DCI) { av = *wqe; if (av->dqp_dct & cpu_to_be32(MLX5_EXTENDED_UD_AV)) *wqe += sizeof(struct mlx5_av); @@ -1189,7 +1200,7 @@ static int mlx5_ib_mr_responder_pfault_handler_rq(struct mlx5_ib_dev *dev, struct mlx5_ib_wq *wq = &qp->rq; int wqe_size = 1 << wq->wqe_shift; - if (qp->wq_sig) { + if (qp->flags_en & MLX5_QP_FLAG_SIGNATURE) { mlx5_ib_err(dev, "ODP fault with WQE signatures is not supported\n"); return -EFAULT; } @@ -1219,7 +1230,7 @@ static inline struct mlx5_core_rsc_common *odp_get_rsc(struct mlx5_ib_dev *dev, case MLX5_WQE_PF_TYPE_REQ_SEND_OR_WRITE: case MLX5_WQE_PF_TYPE_RESP: case MLX5_WQE_PF_TYPE_REQ_READ_OR_ATOMIC: - common = mlx5_core_res_hold(dev->mdev, wq_num, MLX5_RES_QP); + common = mlx5_core_res_hold(dev, wq_num, MLX5_RES_QP); break; default: break; @@ -1755,13 +1766,26 @@ static void mlx5_ib_prefetch_mr_work(struct work_struct *w) { struct prefetch_mr_work *work = container_of(w, struct prefetch_mr_work, work); + struct mlx5_ib_dev *dev; u32 bytes_mapped = 0; + int srcu_key; + int ret; u32 i; - for (i = 0; i < work->num_sge; ++i) - pagefault_mr(work->frags[i].mr, work->frags[i].io_virt, - work->frags[i].length, &bytes_mapped, - work->pf_flags); + /* We rely on IB/core that work is executed if we have num_sge != 0 only. */ + WARN_ON(!work->num_sge); + dev = work->frags[0].mr->dev; + /* SRCU should be held when calling to mlx5_odp_populate_xlt() */ + srcu_key = srcu_read_lock(&dev->odp_srcu); + for (i = 0; i < work->num_sge; ++i) { + ret = pagefault_mr(work->frags[i].mr, work->frags[i].io_virt, + work->frags[i].length, &bytes_mapped, + work->pf_flags); + if (ret <= 0) + continue; + mlx5_update_odp_stats(work->frags[i].mr, prefetch, ret); + } + srcu_read_unlock(&dev->odp_srcu, srcu_key); destroy_prefetch_work(work); } @@ -1782,9 +1806,7 @@ static bool init_prefetch_work(struct ib_pd *pd, work->frags[i].mr = get_prefetchable_mr(pd, advice, sg_list[i].lkey); if (!work->frags[i].mr) { - work->num_sge = i - 1; - if (i) - destroy_prefetch_work(work); + work->num_sge = i; return false; } @@ -1819,6 +1841,7 @@ static int mlx5_ib_prefetch_sg_list(struct ib_pd *pd, &bytes_mapped, pf_flags); if (ret < 0) goto out; + mlx5_update_odp_stats(mr, prefetch, ret); } ret = 0; @@ -1850,6 +1873,7 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, srcu_key = srcu_read_lock(&dev->odp_srcu); if (!init_prefetch_work(pd, advice, pf_flags, work, sg_list, num_sge)) { srcu_read_unlock(&dev->odp_srcu, srcu_key); + destroy_prefetch_work(work); return -EINVAL; } queue_work(system_unbound_wq, &work->work); diff --git a/drivers/infiniband/hw/mlx5/qos.c b/drivers/infiniband/hw/mlx5/qos.c index cac878a70edb..dce92554142a 100644 --- a/drivers/infiniband/hw/mlx5/qos.c +++ b/drivers/infiniband/hw/mlx5/qos.c @@ -69,17 +69,14 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_PP_OBJ_ALLOC)( if (err) goto err; - err = uverbs_copy_to(attrs, MLX5_IB_ATTR_PP_OBJ_ALLOC_INDEX, - &pp_entry->index, sizeof(pp_entry->index)); - if (err) - goto clean; - pp_entry->mdev = dev->mdev; uobj->object = pp_entry; - return 0; + uverbs_finalize_uobj_create(attrs, MLX5_IB_ATTR_PP_OBJ_ALLOC_HANDLE); + + err = uverbs_copy_to(attrs, MLX5_IB_ATTR_PP_OBJ_ALLOC_INDEX, + &pp_entry->index, sizeof(pp_entry->index)); + return err; -clean: - mlx5_rl_remove_rate_raw(dev->mdev, pp_entry->index); err: kfree(pp_entry); return err; diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 2210759843ba..5758dbe64045 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -38,10 +38,10 @@ #include <linux/mlx5/fs.h> #include "mlx5_ib.h" #include "ib_rep.h" +#include "counters.h" #include "cmd.h" - -/* not supported currently */ -static int wq_signature; +#include "qp.h" +#include "wr.h" enum { MLX5_IB_ACK_REQ_FREQ = 8, @@ -54,32 +54,6 @@ enum { MLX5_IB_LINK_TYPE_ETH = 1 }; -enum { - MLX5_IB_SQ_STRIDE = 6, - MLX5_IB_SQ_UMR_INLINE_THRESHOLD = 64, -}; - -static const u32 mlx5_ib_opcode[] = { - [IB_WR_SEND] = MLX5_OPCODE_SEND, - [IB_WR_LSO] = MLX5_OPCODE_LSO, - [IB_WR_SEND_WITH_IMM] = MLX5_OPCODE_SEND_IMM, - [IB_WR_RDMA_WRITE] = MLX5_OPCODE_RDMA_WRITE, - [IB_WR_RDMA_WRITE_WITH_IMM] = MLX5_OPCODE_RDMA_WRITE_IMM, - [IB_WR_RDMA_READ] = MLX5_OPCODE_RDMA_READ, - [IB_WR_ATOMIC_CMP_AND_SWP] = MLX5_OPCODE_ATOMIC_CS, - [IB_WR_ATOMIC_FETCH_AND_ADD] = MLX5_OPCODE_ATOMIC_FA, - [IB_WR_SEND_WITH_INV] = MLX5_OPCODE_SEND_INVAL, - [IB_WR_LOCAL_INV] = MLX5_OPCODE_UMR, - [IB_WR_REG_MR] = MLX5_OPCODE_UMR, - [IB_WR_MASKED_ATOMIC_CMP_AND_SWP] = MLX5_OPCODE_ATOMIC_MASKED_CS, - [IB_WR_MASKED_ATOMIC_FETCH_AND_ADD] = MLX5_OPCODE_ATOMIC_MASKED_FA, - [MLX5_IB_WR_UMR] = MLX5_OPCODE_UMR, -}; - -struct mlx5_wqe_eth_pad { - u8 rsvd0[16]; -}; - enum raw_qp_set_mask_map { MLX5_RAW_QP_MOD_SET_RQ_Q_CTR_ID = 1UL << 0, MLX5_RAW_QP_RATE_LIMIT = 1UL << 1, @@ -391,17 +365,26 @@ static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap, cap->max_recv_wr = 0; cap->max_recv_sge = 0; } else { + int wq_sig = !!(qp->flags_en & MLX5_QP_FLAG_SIGNATURE); + if (ucmd) { qp->rq.wqe_cnt = ucmd->rq_wqe_count; if (ucmd->rq_wqe_shift > BITS_PER_BYTE * sizeof(ucmd->rq_wqe_shift)) return -EINVAL; qp->rq.wqe_shift = ucmd->rq_wqe_shift; - if ((1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) < qp->wq_sig) + if ((1 << qp->rq.wqe_shift) / + sizeof(struct mlx5_wqe_data_seg) < + wq_sig) return -EINVAL; - qp->rq.max_gs = (1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) - qp->wq_sig; + qp->rq.max_gs = + (1 << qp->rq.wqe_shift) / + sizeof(struct mlx5_wqe_data_seg) - + wq_sig; qp->rq.max_post = qp->rq.wqe_cnt; } else { - wqe_size = qp->wq_sig ? sizeof(struct mlx5_wqe_signature_seg) : 0; + wqe_size = + wq_sig ? sizeof(struct mlx5_wqe_signature_seg) : + 0; wqe_size += cap->max_recv_sge * sizeof(struct mlx5_wqe_data_seg); wqe_size = roundup_pow_of_two(wqe_size); wq_size = roundup_pow_of_two(cap->max_recv_wr) * wqe_size; @@ -415,7 +398,10 @@ static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap, return -EINVAL; } qp->rq.wqe_shift = ilog2(wqe_size); - qp->rq.max_gs = (1 << qp->rq.wqe_shift) / sizeof(struct mlx5_wqe_data_seg) - qp->wq_sig; + qp->rq.max_gs = + (1 << qp->rq.wqe_shift) / + sizeof(struct mlx5_wqe_data_seg) - + wq_sig; qp->rq.max_post = qp->rq.wqe_cnt; } } @@ -430,7 +416,7 @@ static int sq_overhead(struct ib_qp_init_attr *attr) switch (attr->qp_type) { case IB_QPT_XRC_INI: size += sizeof(struct mlx5_wqe_xrc_seg); - /* fall through */ + fallthrough; case IB_QPT_RC: size += sizeof(struct mlx5_wqe_ctrl_seg) + max(sizeof(struct mlx5_wqe_atomic_seg) + @@ -455,7 +441,7 @@ static int sq_overhead(struct ib_qp_init_attr *attr) if (attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO) size += sizeof(struct mlx5_wqe_eth_pad) + sizeof(struct mlx5_wqe_eth_seg); - /* fall through */ + fallthrough; case IB_QPT_SMI: case MLX5_IB_QPT_HW_GSI: size += sizeof(struct mlx5_wqe_ctrl_seg) + @@ -595,7 +581,7 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev, } if (attr->qp_type == IB_QPT_RAW_PACKET || - qp->flags & MLX5_IB_QP_UNDERLAY) { + qp->flags & IB_QP_CREATE_SOURCE_QPN) { base->ubuffer.buf_size = qp->rq.wqe_cnt << qp->rq.wqe_shift; qp->raw_packet_qp.sq.ubuffer.buf_size = qp->sq.wqe_cnt << 6; } else { @@ -750,10 +736,7 @@ static int to_mlx5_st(enum ib_qp_type type) case IB_QPT_SMI: return MLX5_QP_ST_QP0; case MLX5_IB_QPT_HW_GSI: return MLX5_QP_ST_QP1; case MLX5_IB_QPT_DCI: return MLX5_QP_ST_DCI; - case IB_QPT_RAW_IPV6: return MLX5_QP_ST_RAW_IPV6; - case IB_QPT_RAW_PACKET: - case IB_QPT_RAW_ETHERTYPE: return MLX5_QP_ST_RAW_ETHERTYPE; - case IB_QPT_MAX: + case IB_QPT_RAW_PACKET: return MLX5_QP_ST_RAW_ETHERTYPE; default: return -EINVAL; } } @@ -890,7 +873,6 @@ static int create_user_rq(struct mlx5_ib_dev *dev, struct ib_pd *pd, goto err_umem; } - rwq->create_type = MLX5_WQ_USER; return 0; err_umem: @@ -905,15 +887,14 @@ static int adjust_bfregn(struct mlx5_ib_dev *dev, bfregn % MLX5_NON_FP_BFREGS_PER_UAR; } -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, - u32 **in, - struct mlx5_ib_create_qp_resp *resp, int *inlen, - struct mlx5_ib_qp_base *base) +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, u32 **in, + struct mlx5_ib_create_qp_resp *resp, int *inlen, + struct mlx5_ib_qp_base *base, + struct mlx5_ib_create_qp *ucmd) { struct mlx5_ib_ucontext *context; - struct mlx5_ib_create_qp ucmd; struct mlx5_ib_ubuffer *ubuffer = &base->ubuffer; int page_shift = 0; int uar_index = 0; @@ -927,30 +908,24 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, u16 uid; u32 uar_flags; - err = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd)); - if (err) { - mlx5_ib_dbg(dev, "copy failed\n"); - return err; - } - context = rdma_udata_to_drv_context(udata, struct mlx5_ib_ucontext, ibucontext); - uar_flags = ucmd.flags & (MLX5_QP_FLAG_UAR_PAGE_INDEX | - MLX5_QP_FLAG_BFREG_INDEX); + uar_flags = qp->flags_en & + (MLX5_QP_FLAG_UAR_PAGE_INDEX | MLX5_QP_FLAG_BFREG_INDEX); switch (uar_flags) { case MLX5_QP_FLAG_UAR_PAGE_INDEX: - uar_index = ucmd.bfreg_index; + uar_index = ucmd->bfreg_index; bfregn = MLX5_IB_INVALID_BFREG; break; case MLX5_QP_FLAG_BFREG_INDEX: uar_index = bfregn_to_uar_index(dev, &context->bfregi, - ucmd.bfreg_index, true); + ucmd->bfreg_index, true); if (uar_index < 0) return uar_index; bfregn = MLX5_IB_INVALID_BFREG; break; case 0: - if (qp->flags & MLX5_IB_QP_CROSS_CHANNEL) + if (qp->flags & IB_QP_CREATE_CROSS_CHANNEL) return -EINVAL; bfregn = alloc_bfreg(dev, &context->bfregi); if (bfregn < 0) @@ -969,12 +944,12 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB); qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift; - err = set_user_buf_size(dev, qp, &ucmd, base, attr); + err = set_user_buf_size(dev, qp, ucmd, base, attr); if (err) goto err_bfreg; - if (ucmd.buf_addr && ubuffer->buf_size) { - ubuffer->buf_addr = ucmd.buf_addr; + if (ucmd->buf_addr && ubuffer->buf_size) { + ubuffer->buf_addr = ucmd->buf_addr; err = mlx5_ib_umem_get(dev, udata, ubuffer->buf_addr, ubuffer->buf_size, &ubuffer->umem, &npages, &page_shift, &ncont, &offset); @@ -992,8 +967,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, goto err_umem; } - uid = (attr->qp_type != IB_QPT_XRC_TGT && - attr->qp_type != IB_QPT_XRC_INI) ? to_mpd(pd)->uid : 0; + uid = (attr->qp_type != IB_QPT_XRC_INI) ? to_mpd(pd)->uid : 0; MLX5_SET(create_qp_in, *in, uid, uid); pas = (__be64 *)MLX5_ADDR_OF(create_qp_in, *in, pas); if (ubuffer->umem) @@ -1011,24 +985,14 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, resp->bfreg_index = MLX5_IB_INVALID_BFREG; qp->bfregn = bfregn; - err = mlx5_ib_db_map_user(context, udata, ucmd.db_addr, &qp->db); + err = mlx5_ib_db_map_user(context, udata, ucmd->db_addr, &qp->db); if (err) { mlx5_ib_dbg(dev, "map failed\n"); goto err_free; } - err = ib_copy_to_udata(udata, resp, min(udata->outlen, sizeof(*resp))); - if (err) { - mlx5_ib_dbg(dev, "copy failed\n"); - goto err_unmap; - } - qp->create_type = MLX5_QP_USER; - return 0; -err_unmap: - mlx5_ib_db_unmap_user(context, &qp->db); - err_free: kvfree(*in); @@ -1041,72 +1005,50 @@ err_bfreg: return err; } -static void destroy_qp_user(struct mlx5_ib_dev *dev, struct ib_pd *pd, - struct mlx5_ib_qp *qp, struct mlx5_ib_qp_base *base, - struct ib_udata *udata) +static void destroy_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + struct mlx5_ib_qp_base *base, struct ib_udata *udata) { - struct mlx5_ib_ucontext *context = - rdma_udata_to_drv_context( - udata, - struct mlx5_ib_ucontext, - ibucontext); - - mlx5_ib_db_unmap_user(context, &qp->db); - ib_umem_release(base->ubuffer.umem); - - /* - * Free only the BFREGs which are handled by the kernel. - * BFREGs of UARs allocated dynamically are handled by user. - */ - if (qp->bfregn != MLX5_IB_INVALID_BFREG) - mlx5_ib_free_bfreg(dev, &context->bfregi, qp->bfregn); -} + struct mlx5_ib_ucontext *context = rdma_udata_to_drv_context( + udata, struct mlx5_ib_ucontext, ibucontext); -/* get_sq_edge - Get the next nearby edge. - * - * An 'edge' is defined as the first following address after the end - * of the fragment or the SQ. Accordingly, during the WQE construction - * which repetitively increases the pointer to write the next data, it - * simply should check if it gets to an edge. - * - * @sq - SQ buffer. - * @idx - Stride index in the SQ buffer. - * - * Return: - * The new edge. - */ -static void *get_sq_edge(struct mlx5_ib_wq *sq, u32 idx) -{ - void *fragment_end; + if (udata) { + /* User QP */ + mlx5_ib_db_unmap_user(context, &qp->db); + ib_umem_release(base->ubuffer.umem); - fragment_end = mlx5_frag_buf_get_wqe - (&sq->fbc, - mlx5_frag_buf_get_idx_last_contig_stride(&sq->fbc, idx)); + /* + * Free only the BFREGs which are handled by the kernel. + * BFREGs of UARs allocated dynamically are handled by user. + */ + if (qp->bfregn != MLX5_IB_INVALID_BFREG) + mlx5_ib_free_bfreg(dev, &context->bfregi, qp->bfregn); + return; + } - return fragment_end + MLX5_SEND_WQE_BB; + /* Kernel QP */ + kvfree(qp->sq.wqe_head); + kvfree(qp->sq.w_list); + kvfree(qp->sq.wrid); + kvfree(qp->sq.wr_data); + kvfree(qp->rq.wrid); + if (qp->db.db) + mlx5_db_free(dev->mdev, &qp->db); + if (qp->buf.frags) + mlx5_frag_buf_free(dev->mdev, &qp->buf); } -static int create_kernel_qp(struct mlx5_ib_dev *dev, - struct ib_qp_init_attr *init_attr, - struct mlx5_ib_qp *qp, - u32 **in, int *inlen, - struct mlx5_ib_qp_base *base) +static int _create_kernel_qp(struct mlx5_ib_dev *dev, + struct ib_qp_init_attr *init_attr, + struct mlx5_ib_qp *qp, u32 **in, int *inlen, + struct mlx5_ib_qp_base *base) { int uar_index; void *qpc; int err; - if (init_attr->create_flags & ~(IB_QP_CREATE_INTEGRITY_EN | - IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK | - IB_QP_CREATE_IPOIB_UD_LSO | - IB_QP_CREATE_NETIF_QP | - MLX5_IB_QP_CREATE_SQPN_QP1 | - MLX5_IB_QP_CREATE_WC_TEST)) - return -EINVAL; - if (init_attr->qp_type == MLX5_IB_QPT_REG_UMR) qp->bf.bfreg = &dev->fp_bfreg; - else if (init_attr->create_flags & MLX5_IB_QP_CREATE_WC_TEST) + else if (qp->flags & MLX5_IB_QP_CREATE_WC_TEST) qp->bf.bfreg = &dev->wc_bfreg; else qp->bf.bfreg = &dev->bfreg; @@ -1166,10 +1108,8 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, MLX5_SET(qpc, qpc, fre, 1); MLX5_SET(qpc, qpc, rlky, 1); - if (init_attr->create_flags & MLX5_IB_QP_CREATE_SQPN_QP1) { + if (qp->flags & MLX5_IB_QP_CREATE_SQPN_QP1) MLX5_SET(qpc, qpc, deth_sqpn, 1); - qp->flags |= MLX5_IB_QP_SQPN_QP1; - } mlx5_fill_page_frag_array(&qp->buf, (__be64 *)MLX5_ADDR_OF(create_qp_in, @@ -1197,7 +1137,6 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, err = -ENOMEM; goto err_wrid; } - qp->create_type = MLX5_QP_KERNEL; return 0; @@ -1217,36 +1156,15 @@ err_buf: return err; } -static void destroy_qp_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) -{ - kvfree(qp->sq.wqe_head); - kvfree(qp->sq.w_list); - kvfree(qp->sq.wrid); - kvfree(qp->sq.wr_data); - kvfree(qp->rq.wrid); - mlx5_db_free(dev->mdev, &qp->db); - mlx5_frag_buf_free(dev->mdev, &qp->buf); -} - static u32 get_rx_type(struct mlx5_ib_qp *qp, struct ib_qp_init_attr *attr) { - if (attr->srq || (attr->qp_type == IB_QPT_XRC_TGT) || - (attr->qp_type == MLX5_IB_QPT_DCI) || - (attr->qp_type == IB_QPT_XRC_INI)) + if (attr->srq || (qp->type == IB_QPT_XRC_TGT) || + (qp->type == MLX5_IB_QPT_DCI) || (qp->type == IB_QPT_XRC_INI)) return MLX5_SRQ_RQ; else if (!qp->has_rq) return MLX5_ZERO_LEN_RQ; - else - return MLX5_NON_ZERO_RQ; -} - -static int is_connected(enum ib_qp_type qp_type) -{ - if (qp_type == IB_QPT_RC || qp_type == IB_QPT_UC || - qp_type == MLX5_IB_QPT_DCI) - return 1; - return 0; + return MLX5_NON_ZERO_RQ; } static int create_raw_packet_qp_tis(struct mlx5_ib_dev *dev, @@ -1254,15 +1172,15 @@ static int create_raw_packet_qp_tis(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq, u32 tdn, struct ib_pd *pd) { - u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {0}; + u32 in[MLX5_ST_SZ_DW(create_tis_in)] = {}; void *tisc = MLX5_ADDR_OF(create_tis_in, in, ctx); MLX5_SET(create_tis_in, in, uid, to_mpd(pd)->uid); MLX5_SET(tisc, tisc, transport_domain, tdn); - if (qp->flags & MLX5_IB_QP_UNDERLAY) + if (qp->flags & IB_QP_CREATE_SOURCE_QPN) MLX5_SET(tisc, tisc, underlay_qpn, qp->underlay_qpn); - return mlx5_core_create_tis(dev->mdev, in, sizeof(in), &sq->tisn); + return mlx5_core_create_tis(dev->mdev, in, &sq->tisn); } static void destroy_raw_packet_qp_tis(struct mlx5_ib_dev *dev, @@ -1336,7 +1254,7 @@ static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev, pas = (__be64 *)MLX5_ADDR_OF(wq, wq, pas); mlx5_ib_populate_pas(dev, sq->ubuffer.umem, page_shift, pas, 0); - err = mlx5_core_create_sq_tracked(dev->mdev, in, inlen, &sq->base.mqp); + err = mlx5_core_create_sq_tracked(dev, in, inlen, &sq->base.mqp); kvfree(in); @@ -1356,7 +1274,7 @@ static void destroy_raw_packet_qp_sq(struct mlx5_ib_dev *dev, struct mlx5_ib_sq *sq) { destroy_flow_rule_vport_sq(sq); - mlx5_core_destroy_sq_tracked(dev->mdev, &sq->base.mqp); + mlx5_core_destroy_sq_tracked(dev, &sq->base.mqp); ib_umem_release(sq->ubuffer.umem); } @@ -1408,7 +1326,7 @@ static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, MLX5_SET(rqc, rqc, user_index, MLX5_GET(qpc, qpc, user_index)); MLX5_SET(rqc, rqc, cqn, MLX5_GET(qpc, qpc, cqn_rcv)); - if (mqp->flags & MLX5_IB_QP_CAP_SCATTER_FCS) + if (mqp->flags & IB_QP_CREATE_SCATTER_FCS) MLX5_SET(rqc, rqc, scatter_fcs, 1); wq = MLX5_ADDR_OF(rqc, rqc, wq); @@ -1426,7 +1344,7 @@ static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, qp_pas = (__be64 *)MLX5_ADDR_OF(create_qp_in, qpin, pas); memcpy(pas, qp_pas, rq_pas_size); - err = mlx5_core_create_rq_tracked(dev->mdev, in, inlen, &rq->base.mqp); + err = mlx5_core_create_rq_tracked(dev, in, inlen, &rq->base.mqp); kvfree(in); @@ -1436,14 +1354,7 @@ static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, static void destroy_raw_packet_qp_rq(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq) { - mlx5_core_destroy_rq_tracked(dev->mdev, &rq->base.mqp); -} - -static bool tunnel_offload_supported(struct mlx5_core_dev *dev) -{ - return (MLX5_CAP_ETH(dev, tunnel_stateless_vxlan) || - MLX5_CAP_ETH(dev, tunnel_stateless_gre) || - MLX5_CAP_ETH(dev, tunnel_stateless_geneve_rx)); + mlx5_core_destroy_rq_tracked(dev, &rq->base.mqp); } static void destroy_raw_packet_qp_tir(struct mlx5_ib_dev *dev, @@ -1459,9 +1370,8 @@ static void destroy_raw_packet_qp_tir(struct mlx5_ib_dev *dev, static int create_raw_packet_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq, u32 tdn, - u32 *qp_flags_en, - struct ib_pd *pd, - u32 *out, int outlen) + u32 *qp_flags_en, struct ib_pd *pd, + u32 *out) { u8 lb_flag = 0; u32 *in; @@ -1494,9 +1404,8 @@ static int create_raw_packet_qp_tir(struct mlx5_ib_dev *dev, } MLX5_SET(tirc, tirc, self_lb_block, lb_flag); - - err = mlx5_core_create_tir_out(dev->mdev, in, inlen, out, outlen); - + MLX5_SET(create_tir_in, in, opcode, MLX5_CMD_OP_CREATE_TIR); + err = mlx5_cmd_exec_inout(dev->mdev, create_tir, in, out); rq->tirn = MLX5_GET(create_tir_out, out, tirn); if (!err && MLX5_GET(tirc, tirc, self_lb_block)) { err = mlx5_ib_enable_lb(dev, false, true); @@ -1525,6 +1434,8 @@ static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, u16 uid = to_mpd(pd)->uid; u32 out[MLX5_ST_SZ_DW(create_tir_out)] = {}; + if (!qp->sq.wqe_cnt && !qp->rq.wqe_cnt) + return -EINVAL; if (qp->sq.wqe_cnt) { err = create_raw_packet_qp_tis(dev, qp, sq, tdn, pd); if (err) @@ -1548,17 +1459,16 @@ static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, if (qp->rq.wqe_cnt) { rq->base.container_mibqp = qp; - if (qp->flags & MLX5_IB_QP_CVLAN_STRIPPING) + if (qp->flags & IB_QP_CREATE_CVLAN_STRIPPING) rq->flags |= MLX5_IB_RQ_CVLAN_STRIPPING; - if (qp->flags & MLX5_IB_QP_PCI_WRITE_END_PADDING) + if (qp->flags & IB_QP_CREATE_PCI_WRITE_END_PADDING) rq->flags |= MLX5_IB_RQ_PCI_WRITE_END_PADDING; err = create_raw_packet_qp_rq(dev, rq, in, inlen, pd); if (err) goto err_destroy_sq; - err = create_raw_packet_qp_tir( - dev, rq, tdn, &qp->flags_en, pd, out, - MLX5_ST_SZ_BYTES(create_tir_out)); + err = create_raw_packet_qp_tir(dev, rq, tdn, &qp->flags_en, pd, + out); if (err) goto err_destroy_rq; @@ -1586,14 +1496,8 @@ static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, qp->trans_qp.base.mqp.qpn = qp->sq.wqe_cnt ? sq->base.mqp.qpn : rq->base.mqp.qpn; - err = ib_copy_to_udata(udata, resp, min(udata->outlen, sizeof(*resp))); - if (err) - goto err_destroy_tir; - return 0; -err_destroy_tir: - destroy_raw_packet_qp_tir(dev, rq, qp->flags_en, pd); err_destroy_rq: destroy_raw_packet_qp_rq(dev, rq); err_destroy_sq: @@ -1645,14 +1549,27 @@ static void destroy_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *q to_mpd(qp->ibqp.pd)->uid); } -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 mlx5_create_qp_params { + struct ib_udata *udata; + size_t inlen; + size_t outlen; + size_t ucmd_size; + void *ucmd; + u8 is_rss_raw : 1; + struct ib_qp_init_attr *attr; + u32 uidx; + struct mlx5_ib_create_qp_resp resp; +}; + +static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) { + struct ib_qp_init_attr *init_attr = params->attr; + struct mlx5_ib_create_qp_rss *ucmd = params->ucmd; + struct ib_udata *udata = params->udata; struct mlx5_ib_ucontext *mucontext = rdma_udata_to_drv_context( udata, struct mlx5_ib_ucontext, ibucontext); - struct mlx5_ib_create_qp_resp resp = {}; int inlen; int outlen; int err; @@ -1662,79 +1579,28 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, void *hfso; u32 selected_fields = 0; u32 outer_l4; - size_t min_resp_len; u32 tdn = mucontext->tdn; - struct mlx5_ib_create_qp_rss ucmd = {}; - size_t required_cmd_sz; u8 lb_flag = 0; - 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), bfreg_index) + sizeof(resp.bfreg_index); - if (udata->outlen < min_resp_len) - return -EINVAL; - - required_cmd_sz = offsetof(typeof(ucmd), flags) + sizeof(ucmd.flags); - 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) { + if (ucmd->comp_mask) { mlx5_ib_dbg(dev, "invalid comp mask\n"); return -EOPNOTSUPP; } - if (ucmd.flags & ~(MLX5_QP_FLAG_TUNNEL_OFFLOADS | - MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC | - MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC)) { - mlx5_ib_dbg(dev, "invalid flags\n"); - return -EOPNOTSUPP; - } - - if (ucmd.flags & MLX5_QP_FLAG_TUNNEL_OFFLOADS && - !tunnel_offload_supported(dev->mdev)) { - mlx5_ib_dbg(dev, "tunnel offloads isn't supported\n"); - return -EOPNOTSUPP; - } - - if (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_INNER && - !(ucmd.flags & MLX5_QP_FLAG_TUNNEL_OFFLOADS)) { + if (ucmd->rx_hash_fields_mask & MLX5_RX_HASH_INNER && + !(ucmd->flags & MLX5_QP_FLAG_TUNNEL_OFFLOADS)) { mlx5_ib_dbg(dev, "Tunnel offloads must be set for inner RSS\n"); return -EOPNOTSUPP; } - if (ucmd.flags & MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC || dev->is_rep) { - lb_flag |= MLX5_TIRC_SELF_LB_BLOCK_BLOCK_UNICAST; + if (dev->is_rep) qp->flags_en |= MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC; - } - if (ucmd.flags & MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC) { - lb_flag |= MLX5_TIRC_SELF_LB_BLOCK_BLOCK_MULTICAST; - qp->flags_en |= MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC; - } + if (qp->flags_en & MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC) + lb_flag |= MLX5_TIRC_SELF_LB_BLOCK_BLOCK_UNICAST; - err = ib_copy_to_udata(udata, &resp, min(udata->outlen, sizeof(resp))); - if (err) { - mlx5_ib_dbg(dev, "copy failed\n"); - return -EINVAL; - } + if (qp->flags_en & MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC) + lb_flag |= MLX5_TIRC_SELF_LB_BLOCK_BLOCK_MULTICAST; inlen = MLX5_ST_SZ_BYTES(create_tir_in); outlen = MLX5_ST_SZ_BYTES(create_tir_out); @@ -1753,29 +1619,29 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_outer); - if (ucmd.flags & MLX5_QP_FLAG_TUNNEL_OFFLOADS) + if (ucmd->flags & MLX5_QP_FLAG_TUNNEL_OFFLOADS) MLX5_SET(tirc, tirc, tunneled_offload_en, 1); MLX5_SET(tirc, tirc, self_lb_block, lb_flag); - if (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_INNER) + if (ucmd->rx_hash_fields_mask & MLX5_RX_HASH_INNER) hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_inner); else hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_outer); - switch (ucmd.rx_hash_function) { + 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) { + if (len != ucmd->rx_key_len) { err = -EINVAL; goto err; } MLX5_SET(tirc, tirc, rx_hash_fn, MLX5_RX_HASH_FN_TOEPLITZ); - memcpy(rss_key, ucmd.rx_hash_key, len); + memcpy(rss_key, ucmd->rx_hash_key, len); break; } default: @@ -1783,7 +1649,7 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, goto err; } - if (!ucmd.rx_hash_fields_mask) { + 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; @@ -1791,29 +1657,31 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, 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))) { + 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)) + 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)) + 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); - outer_l4 = ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || - (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) << 0 | - ((ucmd.rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) || - (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) << 1 | - (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_IPSEC_SPI) << 2; + outer_l4 = ((ucmd->rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_TCP) || + (ucmd->rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_TCP)) + << 0 | + ((ucmd->rx_hash_fields_mask & MLX5_RX_HASH_SRC_PORT_UDP) || + (ucmd->rx_hash_fields_mask & MLX5_RX_HASH_DST_PORT_UDP)) + << 1 | + (ucmd->rx_hash_fields_mask & MLX5_RX_HASH_IPSEC_SPI) << 2; /* Check that only one l4 protocol is set */ if (outer_l4 & (outer_l4 - 1)) { @@ -1822,38 +1690,39 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, } /* 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)) + 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)) + 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)) + 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)) + 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)) + 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)) + 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; - if (ucmd.rx_hash_fields_mask & MLX5_RX_HASH_IPSEC_SPI) + if (ucmd->rx_hash_fields_mask & MLX5_RX_HASH_IPSEC_SPI) selected_fields |= MLX5_HASH_FIELD_SEL_IPSEC_SPI; MLX5_SET(rx_hash_field_select, hfso, selected_fields, selected_fields); create_tir: - err = mlx5_core_create_tir_out(dev->mdev, in, inlen, out, outlen); + MLX5_SET(create_tir_in, in, opcode, MLX5_CMD_OP_CREATE_TIR); + err = mlx5_cmd_exec_inout(dev->mdev, create_tir, in, out); qp->rss_qp.tirn = MLX5_GET(create_tir_out, out, tirn); if (!err && MLX5_GET(tirc, tirc, self_lb_block)) { @@ -1868,75 +1737,44 @@ create_tir: goto err; if (mucontext->devx_uid) { - resp.comp_mask |= MLX5_IB_CREATE_QP_RESP_MASK_TIRN; - resp.tirn = qp->rss_qp.tirn; + params->resp.comp_mask |= MLX5_IB_CREATE_QP_RESP_MASK_TIRN; + params->resp.tirn = qp->rss_qp.tirn; if (MLX5_CAP_FLOWTABLE_NIC_RX(dev->mdev, sw_owner)) { - resp.tir_icm_addr = + params->resp.tir_icm_addr = MLX5_GET(create_tir_out, out, icm_address_31_0); - resp.tir_icm_addr |= (u64)MLX5_GET(create_tir_out, out, - icm_address_39_32) - << 32; - resp.tir_icm_addr |= (u64)MLX5_GET(create_tir_out, out, - icm_address_63_40) - << 40; - resp.comp_mask |= + params->resp.tir_icm_addr |= + (u64)MLX5_GET(create_tir_out, out, + icm_address_39_32) + << 32; + params->resp.tir_icm_addr |= + (u64)MLX5_GET(create_tir_out, out, + icm_address_63_40) + << 40; + params->resp.comp_mask |= MLX5_IB_CREATE_QP_RESP_MASK_TIR_ICM_ADDR; } } - err = ib_copy_to_udata(udata, &resp, min(udata->outlen, sizeof(resp))); - if (err) - goto err_copy; - kvfree(in); /* qpn is reserved for that QP */ qp->trans_qp.base.mqp.qpn = 0; - qp->flags |= MLX5_IB_QP_RSS; + qp->is_rss = true; return 0; -err_copy: - mlx5_cmd_destroy_tir(dev->mdev, qp->rss_qp.tirn, mucontext->devx_uid); err: kvfree(in); return err; } -static void configure_responder_scat_cqe(struct ib_qp_init_attr *init_attr, - void *qpc) -{ - int rcqe_sz; - - if (init_attr->qp_type == MLX5_IB_QPT_DCI) - return; - - rcqe_sz = mlx5_ib_get_cqe_size(init_attr->recv_cq); - - if (init_attr->qp_type == MLX5_IB_QPT_DCT) { - if (rcqe_sz == 128) - MLX5_SET(dctc, qpc, cs_res, MLX5_RES_SCAT_DATA64_CQE); - - return; - } - - MLX5_SET(qpc, qpc, cs_res, - rcqe_sz == 128 ? MLX5_RES_SCAT_DATA64_CQE : - MLX5_RES_SCAT_DATA32_CQE); -} - static void configure_requester_scat_cqe(struct mlx5_ib_dev *dev, + struct mlx5_ib_qp *qp, struct ib_qp_init_attr *init_attr, - struct mlx5_ib_create_qp *ucmd, void *qpc) { - enum ib_qp_type qpt = init_attr->qp_type; int scqe_sz; bool allow_scat_cqe = false; - if (qpt == IB_QPT_UC || qpt == IB_QPT_UD) - return; - - if (ucmd) - allow_scat_cqe = ucmd->flags & MLX5_QP_FLAG_ALLOW_SCATTER_CQE; + allow_scat_cqe = qp->flags_en & MLX5_QP_FLAG_ALLOW_SCATTER_CQE; if (!allow_scat_cqe && init_attr->sq_sig_type != IB_SIGNAL_ALL_WR) return; @@ -1999,269 +1837,179 @@ static int get_atomic_mode(struct mlx5_ib_dev *dev, return atomic_mode; } -static inline bool check_flags_mask(uint64_t input, uint64_t supported) -{ - return (input & ~supported) == 0; -} - -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) +static int create_xrc_tgt_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) { + struct mlx5_ib_create_qp *ucmd = params->ucmd; + struct ib_qp_init_attr *attr = params->attr; + u32 uidx = params->uidx; struct mlx5_ib_resources *devr = &dev->devr; + u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; int inlen = MLX5_ST_SZ_BYTES(create_qp_in); struct mlx5_core_dev *mdev = dev->mdev; - struct mlx5_ib_create_qp_resp resp = {}; - struct mlx5_ib_ucontext *ucontext = rdma_udata_to_drv_context( - udata, struct mlx5_ib_ucontext, ibucontext); - struct mlx5_ib_cq *send_cq; - struct mlx5_ib_cq *recv_cq; - unsigned long flags; - u32 uidx = MLX5_IB_DEFAULT_UIDX; - struct mlx5_ib_create_qp ucmd; struct mlx5_ib_qp_base *base; - int mlx5_st; + unsigned long flags; void *qpc; u32 *in; int err; - mutex_init(&qp->mutex); - spin_lock_init(&qp->sq.lock); - spin_lock_init(&qp->rq.lock); - - mlx5_st = to_mlx5_st(init_attr->qp_type); - if (mlx5_st < 0) - return -EINVAL; + if (attr->sq_sig_type == IB_SIGNAL_ALL_WR) + qp->sq_signal_bits = MLX5_WQE_CTRL_CQ_UPDATE; - if (init_attr->rwq_ind_tbl) { - if (!udata) - return -ENOSYS; + in = kvzalloc(inlen, GFP_KERNEL); + if (!in) + return -ENOMEM; - err = create_rss_raw_qp_tir(dev, qp, pd, init_attr, udata); - return err; - } + if (MLX5_CAP_GEN(mdev, ece_support) && ucmd) + MLX5_SET(create_qp_in, in, ece, ucmd->ece_options); + qpc = MLX5_ADDR_OF(create_qp_in, in, qpc); - 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"); - return -EINVAL; - } else { - qp->flags |= MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK; - } - } + MLX5_SET(qpc, qpc, st, MLX5_QP_ST_XRC); + MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED); + MLX5_SET(qpc, qpc, pd, to_mpd(devr->p0)->pdn); - if (init_attr->create_flags & - (IB_QP_CREATE_CROSS_CHANNEL | - IB_QP_CREATE_MANAGED_SEND | - IB_QP_CREATE_MANAGED_RECV)) { - if (!MLX5_CAP_GEN(mdev, cd)) { - mlx5_ib_dbg(dev, "cross-channel isn't supported\n"); - return -EINVAL; - } - if (init_attr->create_flags & IB_QP_CREATE_CROSS_CHANNEL) - qp->flags |= MLX5_IB_QP_CROSS_CHANNEL; - if (init_attr->create_flags & IB_QP_CREATE_MANAGED_SEND) - qp->flags |= MLX5_IB_QP_MANAGED_SEND; - if (init_attr->create_flags & IB_QP_CREATE_MANAGED_RECV) - qp->flags |= MLX5_IB_QP_MANAGED_RECV; - } - - if (init_attr->qp_type == IB_QPT_UD && - (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO)) - if (!MLX5_CAP_GEN(mdev, ipoib_basic_offloads)) { - mlx5_ib_dbg(dev, "ipoib UD lso qp isn't supported\n"); - return -EOPNOTSUPP; - } + if (qp->flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) + MLX5_SET(qpc, qpc, block_lb_mc, 1); + if (qp->flags & IB_QP_CREATE_CROSS_CHANNEL) + MLX5_SET(qpc, qpc, cd_master, 1); + if (qp->flags & IB_QP_CREATE_MANAGED_SEND) + MLX5_SET(qpc, qpc, cd_slave_send, 1); + if (qp->flags & IB_QP_CREATE_MANAGED_RECV) + MLX5_SET(qpc, qpc, cd_slave_receive, 1); - if (init_attr->create_flags & IB_QP_CREATE_SCATTER_FCS) { - if (init_attr->qp_type != IB_QPT_RAW_PACKET) { - mlx5_ib_dbg(dev, "Scatter FCS is supported only for Raw Packet QPs"); - return -EOPNOTSUPP; - } - if (!MLX5_CAP_GEN(dev->mdev, eth_net_offloads) || - !MLX5_CAP_ETH(dev->mdev, scatter_fcs)) { - mlx5_ib_dbg(dev, "Scatter FCS isn't supported\n"); - return -EOPNOTSUPP; - } - qp->flags |= MLX5_IB_QP_CAP_SCATTER_FCS; - } + MLX5_SET(qpc, qpc, rq_type, MLX5_SRQ_RQ); + MLX5_SET(qpc, qpc, no_sq, 1); + MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(devr->c0)->mcq.cqn); + MLX5_SET(qpc, qpc, cqn_snd, to_mcq(devr->c0)->mcq.cqn); + MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, to_msrq(devr->s0)->msrq.srqn); + MLX5_SET(qpc, qpc, xrcd, to_mxrcd(attr->xrcd)->xrcdn); + MLX5_SET64(qpc, qpc, dbr_addr, qp->db.dma); - if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) - qp->sq_signal_bits = MLX5_WQE_CTRL_CQ_UPDATE; + /* 0xffffff means we ask to work with cqe version 0 */ + if (MLX5_CAP_GEN(mdev, cqe_version) == MLX5_CQE_VERSION_V1) + MLX5_SET(qpc, qpc, user_index, uidx); - if (init_attr->create_flags & IB_QP_CREATE_CVLAN_STRIPPING) { - if (!(MLX5_CAP_GEN(dev->mdev, eth_net_offloads) && - MLX5_CAP_ETH(dev->mdev, vlan_cap)) || - (init_attr->qp_type != IB_QPT_RAW_PACKET)) - return -EOPNOTSUPP; - qp->flags |= MLX5_IB_QP_CVLAN_STRIPPING; + if (qp->flags & IB_QP_CREATE_PCI_WRITE_END_PADDING) { + MLX5_SET(qpc, qpc, end_padding_mode, + MLX5_WQ_END_PAD_MODE_ALIGN); + /* Special case to clean flag */ + qp->flags &= ~IB_QP_CREATE_PCI_WRITE_END_PADDING; } - if (udata) { - if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) { - mlx5_ib_dbg(dev, "copy failed\n"); - return -EFAULT; - } + base = &qp->trans_qp.base; + err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); + kvfree(in); + if (err) + return err; - if (!check_flags_mask(ucmd.flags, - MLX5_QP_FLAG_ALLOW_SCATTER_CQE | - MLX5_QP_FLAG_BFREG_INDEX | - MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE | - MLX5_QP_FLAG_SCATTER_CQE | - MLX5_QP_FLAG_SIGNATURE | - MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC | - MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC | - MLX5_QP_FLAG_TUNNEL_OFFLOADS | - MLX5_QP_FLAG_UAR_PAGE_INDEX | - MLX5_QP_FLAG_TYPE_DCI | - MLX5_QP_FLAG_TYPE_DCT)) - return -EINVAL; + base->container_mibqp = qp; + base->mqp.event = mlx5_ib_qp_event; + if (MLX5_CAP_GEN(mdev, ece_support)) + params->resp.ece_options = MLX5_GET(create_qp_out, out, ece); - err = get_qp_user_index(ucontext, &ucmd, udata->inlen, &uidx); - if (err) - return err; + spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); + list_add_tail(&qp->qps_list, &dev->qp_list); + spin_unlock_irqrestore(&dev->reset_flow_resource_lock, flags); - qp->wq_sig = !!(ucmd.flags & MLX5_QP_FLAG_SIGNATURE); - if (MLX5_CAP_GEN(dev->mdev, sctr_data_cqe)) - qp->scat_cqe = !!(ucmd.flags & MLX5_QP_FLAG_SCATTER_CQE); - if (ucmd.flags & MLX5_QP_FLAG_TUNNEL_OFFLOADS) { - if (init_attr->qp_type != IB_QPT_RAW_PACKET || - !tunnel_offload_supported(mdev)) { - mlx5_ib_dbg(dev, "Tunnel offload isn't supported\n"); - return -EOPNOTSUPP; - } - qp->flags_en |= MLX5_QP_FLAG_TUNNEL_OFFLOADS; - } + qp->trans_qp.xrcdn = to_mxrcd(attr->xrcd)->xrcdn; + return 0; +} - if (ucmd.flags & MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC) { - if (init_attr->qp_type != IB_QPT_RAW_PACKET) { - mlx5_ib_dbg(dev, "Self-LB UC isn't supported\n"); - return -EOPNOTSUPP; - } - qp->flags_en |= MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC; - } +static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) +{ + struct ib_qp_init_attr *init_attr = params->attr; + struct mlx5_ib_create_qp *ucmd = params->ucmd; + u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; + struct ib_udata *udata = params->udata; + u32 uidx = params->uidx; + struct mlx5_ib_resources *devr = &dev->devr; + int inlen = MLX5_ST_SZ_BYTES(create_qp_in); + struct mlx5_core_dev *mdev = dev->mdev; + struct mlx5_ib_cq *send_cq; + struct mlx5_ib_cq *recv_cq; + unsigned long flags; + struct mlx5_ib_qp_base *base; + int mlx5_st; + void *qpc; + u32 *in; + int err; - if (ucmd.flags & MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC) { - if (init_attr->qp_type != IB_QPT_RAW_PACKET) { - mlx5_ib_dbg(dev, "Self-LB UM isn't supported\n"); - return -EOPNOTSUPP; - } - qp->flags_en |= MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC; - } + spin_lock_init(&qp->sq.lock); + spin_lock_init(&qp->rq.lock); - if (ucmd.flags & MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE) { - if (init_attr->qp_type != IB_QPT_RC || - !MLX5_CAP_GEN(dev->mdev, qp_packet_based)) { - mlx5_ib_dbg(dev, "packet based credit mode isn't supported\n"); - return -EOPNOTSUPP; - } - qp->flags |= MLX5_IB_QP_PACKET_BASED_CREDIT; - } + mlx5_st = to_mlx5_st(qp->type); + if (mlx5_st < 0) + return -EINVAL; - if (init_attr->create_flags & IB_QP_CREATE_SOURCE_QPN) { - if (init_attr->qp_type != IB_QPT_UD || - (MLX5_CAP_GEN(dev->mdev, port_type) != - MLX5_CAP_PORT_TYPE_IB) || - !mlx5_get_flow_namespace(dev->mdev, MLX5_FLOW_NAMESPACE_BYPASS)) { - mlx5_ib_dbg(dev, "Source QP option isn't supported\n"); - return -EOPNOTSUPP; - } + if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) + qp->sq_signal_bits = MLX5_WQE_CTRL_CQ_UPDATE; - qp->flags |= MLX5_IB_QP_UNDERLAY; - qp->underlay_qpn = init_attr->source_qpn; - } - } else { - qp->wq_sig = !!wq_signature; - } + if (qp->flags & IB_QP_CREATE_SOURCE_QPN) + qp->underlay_qpn = init_attr->source_qpn; base = (init_attr->qp_type == IB_QPT_RAW_PACKET || - qp->flags & MLX5_IB_QP_UNDERLAY) ? + qp->flags & IB_QP_CREATE_SOURCE_QPN) ? &qp->raw_packet_qp.rq.base : &qp->trans_qp.base; qp->has_rq = qp_has_rq(init_attr); - err = set_rq_size(dev, &init_attr->cap, qp->has_rq, - qp, udata ? &ucmd : NULL); + err = set_rq_size(dev, &init_attr->cap, qp->has_rq, qp, ucmd); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); return err; } - if (pd) { - if (udata) { - __u32 max_wqes = - 1 << MLX5_CAP_GEN(mdev, log_max_qp_sz); - mlx5_ib_dbg(dev, "requested sq_wqe_count (%d)\n", ucmd.sq_wqe_count); - if (ucmd.rq_wqe_shift != qp->rq.wqe_shift || - ucmd.rq_wqe_count != qp->rq.wqe_cnt) { - mlx5_ib_dbg(dev, "invalid rq params\n"); - return -EINVAL; - } - if (ucmd.sq_wqe_count > max_wqes) { - mlx5_ib_dbg(dev, "requested sq_wqe_count (%d) > max allowed (%d)\n", - ucmd.sq_wqe_count, max_wqes); - return -EINVAL; - } - if (init_attr->create_flags & - MLX5_IB_QP_CREATE_SQPN_QP1) { - mlx5_ib_dbg(dev, "user-space is not allowed to create UD QPs spoofing as QP1\n"); - return -EINVAL; - } - err = create_user_qp(dev, pd, qp, udata, init_attr, &in, - &resp, &inlen, base); - if (err) - mlx5_ib_dbg(dev, "err %d\n", err); - } else { - err = create_kernel_qp(dev, init_attr, qp, &in, &inlen, - base); - if (err) - mlx5_ib_dbg(dev, "err %d\n", err); - } + if (ucmd->rq_wqe_shift != qp->rq.wqe_shift || + ucmd->rq_wqe_count != qp->rq.wqe_cnt) + return -EINVAL; - if (err) - return err; - } else { - in = kvzalloc(inlen, GFP_KERNEL); - if (!in) - return -ENOMEM; + if (ucmd->sq_wqe_count > (1 << MLX5_CAP_GEN(mdev, log_max_qp_sz))) + return -EINVAL; - qp->create_type = MLX5_QP_EMPTY; - } + err = _create_user_qp(dev, pd, qp, udata, init_attr, &in, ¶ms->resp, + &inlen, base, ucmd); + if (err) + return err; if (is_sqp(init_attr->qp_type)) qp->port = init_attr->port_num; + if (MLX5_CAP_GEN(mdev, ece_support)) + MLX5_SET(create_qp_in, in, ece, ucmd->ece_options); qpc = MLX5_ADDR_OF(create_qp_in, in, qpc); MLX5_SET(qpc, qpc, st, mlx5_st); MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED); + MLX5_SET(qpc, qpc, pd, to_mpd(pd)->pdn); - if (init_attr->qp_type != MLX5_IB_QPT_REG_UMR) - MLX5_SET(qpc, qpc, pd, to_mpd(pd ? pd : devr->p0)->pdn); - else - MLX5_SET(qpc, qpc, latency_sensitive, 1); - - - if (qp->wq_sig) + if (qp->flags_en & MLX5_QP_FLAG_SIGNATURE) MLX5_SET(qpc, qpc, wq_signature, 1); - if (qp->flags & MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK) + if (qp->flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) MLX5_SET(qpc, qpc, block_lb_mc, 1); - if (qp->flags & MLX5_IB_QP_CROSS_CHANNEL) + if (qp->flags & IB_QP_CREATE_CROSS_CHANNEL) MLX5_SET(qpc, qpc, cd_master, 1); - if (qp->flags & MLX5_IB_QP_MANAGED_SEND) + if (qp->flags & IB_QP_CREATE_MANAGED_SEND) MLX5_SET(qpc, qpc, cd_slave_send, 1); - if (qp->flags & MLX5_IB_QP_MANAGED_RECV) + if (qp->flags & IB_QP_CREATE_MANAGED_RECV) MLX5_SET(qpc, qpc, cd_slave_receive, 1); - if (qp->flags & MLX5_IB_QP_PACKET_BASED_CREDIT) + if (qp->flags_en & MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE) MLX5_SET(qpc, qpc, req_e2e_credit_mode, 1); - if (qp->scat_cqe && is_connected(init_attr->qp_type)) { - configure_responder_scat_cqe(init_attr, qpc); - configure_requester_scat_cqe(dev, init_attr, - udata ? &ucmd : NULL, - qpc); + if ((qp->flags_en & MLX5_QP_FLAG_SCATTER_CQE) && + (init_attr->qp_type == IB_QPT_RC || + init_attr->qp_type == IB_QPT_UC)) { + int rcqe_sz = mlx5_ib_get_cqe_size(init_attr->recv_cq); + + MLX5_SET(qpc, qpc, cs_res, + rcqe_sz == 128 ? MLX5_RES_SCAT_DATA64_CQE : + MLX5_RES_SCAT_DATA32_CQE); } + if ((qp->flags_en & MLX5_QP_FLAG_SCATTER_CQE) && + (qp->type == MLX5_IB_QPT_DCI || qp->type == IB_QPT_RC)) + configure_requester_scat_cqe(dev, qp, init_attr, qpc); if (qp->rq.wqe_cnt) { MLX5_SET(qpc, qpc, log_rq_stride, qp->rq.wqe_shift - 4); @@ -2282,23 +2030,17 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, /* Set default resources */ switch (init_attr->qp_type) { - case IB_QPT_XRC_TGT: - MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(devr->c0)->mcq.cqn); - MLX5_SET(qpc, qpc, cqn_snd, to_mcq(devr->c0)->mcq.cqn); - MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, to_msrq(devr->s0)->msrq.srqn); - MLX5_SET(qpc, qpc, xrcd, to_mxrcd(init_attr->xrcd)->xrcdn); - break; case IB_QPT_XRC_INI: MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(devr->c0)->mcq.cqn); - MLX5_SET(qpc, qpc, xrcd, to_mxrcd(devr->x1)->xrcdn); + MLX5_SET(qpc, qpc, xrcd, devr->xrcdn1); MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, to_msrq(devr->s0)->msrq.srqn); break; default: if (init_attr->srq) { - MLX5_SET(qpc, qpc, xrcd, to_mxrcd(devr->x0)->xrcdn); + MLX5_SET(qpc, qpc, xrcd, devr->xrcdn0); MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, to_msrq(init_attr->srq)->msrq.srqn); } else { - MLX5_SET(qpc, qpc, xrcd, to_mxrcd(devr->x1)->xrcdn); + MLX5_SET(qpc, qpc, xrcd, devr->xrcdn1); MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, to_msrq(devr->s1)->msrq.srqn); } } @@ -2315,52 +2057,162 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, if (MLX5_CAP_GEN(mdev, cqe_version) == MLX5_CQE_VERSION_V1) MLX5_SET(qpc, qpc, user_index, uidx); - /* we use IB_QP_CREATE_IPOIB_UD_LSO to indicates ipoib qp */ - if (init_attr->qp_type == IB_QPT_UD && - (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO)) { - MLX5_SET(qpc, qpc, ulp_stateless_offload_mode, 1); - qp->flags |= MLX5_IB_QP_LSO; + if (qp->flags & IB_QP_CREATE_PCI_WRITE_END_PADDING && + init_attr->qp_type != IB_QPT_RAW_PACKET) { + MLX5_SET(qpc, qpc, end_padding_mode, + MLX5_WQ_END_PAD_MODE_ALIGN); + /* Special case to clean flag */ + qp->flags &= ~IB_QP_CREATE_PCI_WRITE_END_PADDING; } - if (init_attr->create_flags & IB_QP_CREATE_PCI_WRITE_END_PADDING) { - if (!MLX5_CAP_GEN(dev->mdev, end_pad)) { - mlx5_ib_dbg(dev, "scatter end padding is not supported\n"); - err = -EOPNOTSUPP; - goto err; - } else if (init_attr->qp_type != IB_QPT_RAW_PACKET) { - MLX5_SET(qpc, qpc, end_padding_mode, - MLX5_WQ_END_PAD_MODE_ALIGN); - } else { - qp->flags |= MLX5_IB_QP_PCI_WRITE_END_PADDING; - } + if (init_attr->qp_type == IB_QPT_RAW_PACKET || + qp->flags & IB_QP_CREATE_SOURCE_QPN) { + qp->raw_packet_qp.sq.ubuffer.buf_addr = ucmd->sq_buf_addr; + raw_packet_qp_copy_info(qp, &qp->raw_packet_qp); + err = create_raw_packet_qp(dev, qp, in, inlen, pd, udata, + ¶ms->resp); + } else + err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); + + kvfree(in); + if (err) + goto err_create; + + base->container_mibqp = qp; + base->mqp.event = mlx5_ib_qp_event; + if (MLX5_CAP_GEN(mdev, ece_support)) + params->resp.ece_options = MLX5_GET(create_qp_out, out, ece); + + get_cqs(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: + destroy_qp(dev, qp, base, udata); + return err; +} + +static int create_kernel_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) +{ + struct ib_qp_init_attr *attr = params->attr; + u32 uidx = params->uidx; + struct mlx5_ib_resources *devr = &dev->devr; + u32 out[MLX5_ST_SZ_DW(create_qp_out)] = {}; + int inlen = MLX5_ST_SZ_BYTES(create_qp_in); + struct mlx5_core_dev *mdev = dev->mdev; + struct mlx5_ib_cq *send_cq; + struct mlx5_ib_cq *recv_cq; + unsigned long flags; + struct mlx5_ib_qp_base *base; + int mlx5_st; + void *qpc; + u32 *in; + int err; + + spin_lock_init(&qp->sq.lock); + spin_lock_init(&qp->rq.lock); + + mlx5_st = to_mlx5_st(qp->type); + if (mlx5_st < 0) + return -EINVAL; + + if (attr->sq_sig_type == IB_SIGNAL_ALL_WR) + qp->sq_signal_bits = MLX5_WQE_CTRL_CQ_UPDATE; + + base = &qp->trans_qp.base; + + qp->has_rq = qp_has_rq(attr); + err = set_rq_size(dev, &attr->cap, qp->has_rq, qp, NULL); + if (err) { + mlx5_ib_dbg(dev, "err %d\n", err); + return err; } - if (inlen < 0) { - err = -EINVAL; - goto err; + err = _create_kernel_qp(dev, attr, qp, &in, &inlen, base); + if (err) + return err; + + if (is_sqp(attr->qp_type)) + qp->port = attr->port_num; + + qpc = MLX5_ADDR_OF(create_qp_in, in, qpc); + + MLX5_SET(qpc, qpc, st, mlx5_st); + MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED); + + if (attr->qp_type != MLX5_IB_QPT_REG_UMR) + MLX5_SET(qpc, qpc, pd, to_mpd(pd ? pd : devr->p0)->pdn); + else + MLX5_SET(qpc, qpc, latency_sensitive, 1); + + + if (qp->flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) + MLX5_SET(qpc, qpc, block_lb_mc, 1); + + if (qp->rq.wqe_cnt) { + MLX5_SET(qpc, qpc, log_rq_stride, qp->rq.wqe_shift - 4); + MLX5_SET(qpc, qpc, log_rq_size, ilog2(qp->rq.wqe_cnt)); } - if (init_attr->qp_type == IB_QPT_RAW_PACKET || - qp->flags & MLX5_IB_QP_UNDERLAY) { - qp->raw_packet_qp.sq.ubuffer.buf_addr = ucmd.sq_buf_addr; - raw_packet_qp_copy_info(qp, &qp->raw_packet_qp); - err = create_raw_packet_qp(dev, qp, in, inlen, pd, udata, - &resp); + MLX5_SET(qpc, qpc, rq_type, get_rx_type(qp, attr)); + + if (qp->sq.wqe_cnt) + MLX5_SET(qpc, qpc, log_sq_size, ilog2(qp->sq.wqe_cnt)); + else + MLX5_SET(qpc, qpc, no_sq, 1); + + if (attr->srq) { + MLX5_SET(qpc, qpc, xrcd, devr->xrcdn0); + MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, + to_msrq(attr->srq)->msrq.srqn); } else { - err = mlx5_core_create_qp(dev->mdev, &base->mqp, in, inlen); + MLX5_SET(qpc, qpc, xrcd, devr->xrcdn1); + MLX5_SET(qpc, qpc, srqn_rmpn_xrqn, + to_msrq(devr->s1)->msrq.srqn); } - if (err) { - mlx5_ib_dbg(dev, "create qp failed\n"); - goto err_create; - } + if (attr->send_cq) + MLX5_SET(qpc, qpc, cqn_snd, to_mcq(attr->send_cq)->mcq.cqn); + if (attr->recv_cq) + MLX5_SET(qpc, qpc, cqn_rcv, to_mcq(attr->recv_cq)->mcq.cqn); + + MLX5_SET64(qpc, qpc, dbr_addr, qp->db.dma); + + /* 0xffffff means we ask to work with cqe version 0 */ + if (MLX5_CAP_GEN(mdev, cqe_version) == MLX5_CQE_VERSION_V1) + MLX5_SET(qpc, qpc, user_index, uidx); + + /* we use IB_QP_CREATE_IPOIB_UD_LSO to indicates ipoib qp */ + if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO) + MLX5_SET(qpc, qpc, ulp_stateless_offload_mode, 1); + + err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); kvfree(in); + if (err) + goto err_create; 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, + get_cqs(qp->type, attr->send_cq, attr->recv_cq, &send_cq, &recv_cq); spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); mlx5_ib_lock_cqs(send_cq, recv_cq); @@ -2380,13 +2232,7 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, return 0; err_create: - if (qp->create_type == MLX5_QP_USER) - destroy_qp_user(dev, pd, qp, base, udata); - else if (qp->create_type == MLX5_QP_KERNEL) - destroy_qp_kernel(dev, qp); - -err: - kvfree(in); + destroy_qp(dev, qp, base, NULL); return err; } @@ -2448,11 +2294,6 @@ static void mlx5_ib_unlock_cqs(struct mlx5_ib_cq *send_cq, struct mlx5_ib_cq *re } } -static struct mlx5_ib_pd *get_pd(struct mlx5_ib_qp *qp) -{ - return to_mpd(qp->ibqp.pd); -} - 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) @@ -2473,14 +2314,10 @@ static void get_cqs(enum ib_qp_type qp_type, case IB_QPT_RC: case IB_QPT_UC: case IB_QPT_UD: - case IB_QPT_RAW_IPV6: - case IB_QPT_RAW_ETHERTYPE: case IB_QPT_RAW_PACKET: *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: default: *send_cq = NULL; *recv_cq = NULL; @@ -2500,22 +2337,21 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, unsigned long flags; int err; - if (qp->ibqp.rwq_ind_tbl) { + if (qp->is_rss) { destroy_rss_raw_qp_tir(dev, qp); return; } - base = (qp->ibqp.qp_type == IB_QPT_RAW_PACKET || - qp->flags & MLX5_IB_QP_UNDERLAY) ? - &qp->raw_packet_qp.rq.base : - &qp->trans_qp.base; + base = (qp->type == IB_QPT_RAW_PACKET || + qp->flags & IB_QP_CREATE_SOURCE_QPN) ? + &qp->raw_packet_qp.rq.base : + &qp->trans_qp.base; if (qp->state != IB_QPS_RESET) { - if (qp->ibqp.qp_type != IB_QPT_RAW_PACKET && - !(qp->flags & MLX5_IB_QP_UNDERLAY)) { - err = mlx5_core_qp_modify(dev->mdev, - MLX5_CMD_OP_2RST_QP, 0, - NULL, &base->mqp); + if (qp->type != IB_QPT_RAW_PACKET && + !(qp->flags & IB_QP_CREATE_SOURCE_QPN)) { + err = mlx5_core_qp_modify(dev, MLX5_CMD_OP_2RST_QP, 0, + NULL, &base->mqp, NULL); } else { struct mlx5_modify_raw_qp_param raw_qp_param = { .operation = MLX5_CMD_OP_2RST_QP @@ -2528,8 +2364,8 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, base->mqp.qpn); } - get_cqs(qp->ibqp.qp_type, qp->ibqp.send_cq, qp->ibqp.recv_cq, - &send_cq, &recv_cq); + get_cqs(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); @@ -2541,7 +2377,7 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, if (recv_cq) list_del(&qp->cq_recv_list); - if (qp->create_type == MLX5_QP_KERNEL) { + if (!udata) { __mlx5_ib_cq_clean(recv_cq, base->mqp.qpn, qp->ibqp.srq ? to_msrq(qp->ibqp.srq) : NULL); if (send_cq != recv_cq) @@ -2551,264 +2387,478 @@ static void destroy_qp_common(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, 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 || - qp->flags & MLX5_IB_QP_UNDERLAY) { + if (qp->type == IB_QPT_RAW_PACKET || + qp->flags & IB_QP_CREATE_SOURCE_QPN) { destroy_raw_packet_qp(dev, qp); } else { - err = mlx5_core_destroy_qp(dev->mdev, &base->mqp); + err = mlx5_core_destroy_qp(dev, &base->mqp); if (err) mlx5_ib_warn(dev, "failed to destroy QP 0x%x\n", base->mqp.qpn); } - if (qp->create_type == MLX5_QP_KERNEL) - destroy_qp_kernel(dev, qp); - else if (qp->create_type == MLX5_QP_USER) - destroy_qp_user(dev, &get_pd(qp)->ibpd, qp, base, udata); + destroy_qp(dev, qp, base, udata); } -static const char *ib_qp_type_str(enum ib_qp_type type) +static int create_dct(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) { - switch (type) { - case IB_QPT_SMI: - return "IB_QPT_SMI"; - case IB_QPT_GSI: - return "IB_QPT_GSI"; + struct ib_qp_init_attr *attr = params->attr; + struct mlx5_ib_create_qp *ucmd = params->ucmd; + u32 uidx = params->uidx; + void *dctc; + + qp->dct.in = kzalloc(MLX5_ST_SZ_BYTES(create_dct_in), GFP_KERNEL); + if (!qp->dct.in) + return -ENOMEM; + + MLX5_SET(create_dct_in, qp->dct.in, uid, to_mpd(pd)->uid); + dctc = MLX5_ADDR_OF(create_dct_in, qp->dct.in, dct_context_entry); + MLX5_SET(dctc, dctc, pd, to_mpd(pd)->pdn); + MLX5_SET(dctc, dctc, srqn_xrqn, to_msrq(attr->srq)->msrq.srqn); + MLX5_SET(dctc, dctc, cqn, to_mcq(attr->recv_cq)->mcq.cqn); + MLX5_SET64(dctc, dctc, dc_access_key, ucmd->access_key); + MLX5_SET(dctc, dctc, user_index, uidx); + if (MLX5_CAP_GEN(dev->mdev, ece_support)) + MLX5_SET(dctc, dctc, ece, ucmd->ece_options); + + if (qp->flags_en & MLX5_QP_FLAG_SCATTER_CQE) { + int rcqe_sz = mlx5_ib_get_cqe_size(attr->recv_cq); + + if (rcqe_sz == 128) + MLX5_SET(dctc, dctc, cs_res, MLX5_RES_SCAT_DATA64_CQE); + } + + qp->state = IB_QPS_RESET; + + return 0; +} + +static int check_qp_type(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, + enum ib_qp_type *type) +{ + if (attr->qp_type == IB_QPT_DRIVER && !MLX5_CAP_GEN(dev->mdev, dct)) + goto out; + + switch (attr->qp_type) { + case IB_QPT_XRC_TGT: + case IB_QPT_XRC_INI: + if (!MLX5_CAP_GEN(dev->mdev, xrc)) + goto out; + fallthrough; case IB_QPT_RC: - return "IB_QPT_RC"; case IB_QPT_UC: - return "IB_QPT_UC"; - case IB_QPT_UD: - return "IB_QPT_UD"; - case IB_QPT_RAW_IPV6: - return "IB_QPT_RAW_IPV6"; - case IB_QPT_RAW_ETHERTYPE: - return "IB_QPT_RAW_ETHERTYPE"; - case IB_QPT_XRC_INI: - return "IB_QPT_XRC_INI"; - case IB_QPT_XRC_TGT: - return "IB_QPT_XRC_TGT"; + case IB_QPT_SMI: + case MLX5_IB_QPT_HW_GSI: + case IB_QPT_DRIVER: + case IB_QPT_GSI: + if (dev->profile == &raw_eth_profile) + goto out; case IB_QPT_RAW_PACKET: - return "IB_QPT_RAW_PACKET"; + case IB_QPT_UD: case MLX5_IB_QPT_REG_UMR: - return "MLX5_IB_QPT_REG_UMR"; - case IB_QPT_DRIVER: - return "IB_QPT_DRIVER"; - case IB_QPT_MAX: + break; default: - return "Invalid QP type"; + goto out; } + + *type = attr->qp_type; + return 0; + +out: + mlx5_ib_dbg(dev, "Unsupported QP type %d\n", attr->qp_type); + return -EOPNOTSUPP; } -static struct ib_qp *mlx5_ib_create_dct(struct ib_pd *pd, - struct ib_qp_init_attr *attr, - struct mlx5_ib_create_qp *ucmd, - struct ib_udata *udata) +static int check_valid_flow(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct ib_qp_init_attr *attr, + struct ib_udata *udata) { struct mlx5_ib_ucontext *ucontext = rdma_udata_to_drv_context( udata, struct mlx5_ib_ucontext, ibucontext); - struct mlx5_ib_qp *qp; - int err = 0; - u32 uidx = MLX5_IB_DEFAULT_UIDX; - void *dctc; - if (!attr->srq || !attr->recv_cq) - return ERR_PTR(-EINVAL); + if (!udata) { + /* Kernel create_qp callers */ + if (attr->rwq_ind_tbl) + return -EOPNOTSUPP; - err = get_qp_user_index(ucontext, ucmd, sizeof(*ucmd), &uidx); - if (err) - return ERR_PTR(err); + switch (attr->qp_type) { + case IB_QPT_RAW_PACKET: + case IB_QPT_DRIVER: + return -EOPNOTSUPP; + default: + return 0; + } + } - qp = kzalloc(sizeof(*qp), GFP_KERNEL); - if (!qp) - return ERR_PTR(-ENOMEM); + /* Userspace create_qp callers */ + if (attr->qp_type == IB_QPT_RAW_PACKET && !ucontext->cqe_version) { + mlx5_ib_dbg(dev, + "Raw Packet QP is only supported for CQE version > 0\n"); + return -EINVAL; + } - qp->dct.in = kzalloc(MLX5_ST_SZ_BYTES(create_dct_in), GFP_KERNEL); - if (!qp->dct.in) { - err = -ENOMEM; - goto err_free; + if (attr->qp_type != IB_QPT_RAW_PACKET && attr->rwq_ind_tbl) { + mlx5_ib_dbg(dev, + "Wrong QP type %d for the RWQ indirect table\n", + attr->qp_type); + return -EINVAL; } - MLX5_SET(create_dct_in, qp->dct.in, uid, to_mpd(pd)->uid); - dctc = MLX5_ADDR_OF(create_dct_in, qp->dct.in, dct_context_entry); - qp->qp_sub_type = MLX5_IB_QPT_DCT; - MLX5_SET(dctc, dctc, pd, to_mpd(pd)->pdn); - MLX5_SET(dctc, dctc, srqn_xrqn, to_msrq(attr->srq)->msrq.srqn); - MLX5_SET(dctc, dctc, cqn, to_mcq(attr->recv_cq)->mcq.cqn); - MLX5_SET64(dctc, dctc, dc_access_key, ucmd->access_key); - MLX5_SET(dctc, dctc, user_index, uidx); + switch (attr->qp_type) { + case IB_QPT_SMI: + case MLX5_IB_QPT_HW_GSI: + case MLX5_IB_QPT_REG_UMR: + case IB_QPT_GSI: + mlx5_ib_dbg(dev, "Kernel doesn't support QP type %d\n", + attr->qp_type); + return -EINVAL; + default: + break; + } - if (ucmd->flags & MLX5_QP_FLAG_SCATTER_CQE) - configure_responder_scat_cqe(attr, dctc); + /* + * We don't need to see this warning, it means that kernel code + * missing ib_pd. Placed here to catch developer's mistakes. + */ + WARN_ONCE(!pd && attr->qp_type != IB_QPT_XRC_TGT, + "There is a missing PD pointer assignment\n"); + return 0; +} - qp->state = IB_QPS_RESET; +static void process_vendor_flag(struct mlx5_ib_dev *dev, int *flags, int flag, + bool cond, struct mlx5_ib_qp *qp) +{ + if (!(*flags & flag)) + return; - return &qp->ibqp; -err_free: - kfree(qp); - return ERR_PTR(err); + if (cond) { + qp->flags_en |= flag; + *flags &= ~flag; + return; + } + + switch (flag) { + case MLX5_QP_FLAG_SCATTER_CQE: + case MLX5_QP_FLAG_ALLOW_SCATTER_CQE: + /* + * We don't return error if these flags were provided, + * and mlx5 doesn't have right capability. + */ + *flags &= ~(MLX5_QP_FLAG_SCATTER_CQE | + MLX5_QP_FLAG_ALLOW_SCATTER_CQE); + return; + default: + break; + } + mlx5_ib_dbg(dev, "Vendor create QP flag 0x%X is not supported\n", flag); } -static int set_mlx_qp_type(struct mlx5_ib_dev *dev, - struct ib_qp_init_attr *init_attr, - struct mlx5_ib_create_qp *ucmd, - struct ib_udata *udata) +static int process_vendor_flags(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + void *ucmd, struct ib_qp_init_attr *attr) { - enum { MLX_QP_FLAGS = MLX5_QP_FLAG_TYPE_DCT | MLX5_QP_FLAG_TYPE_DCI }; - int err; + struct mlx5_core_dev *mdev = dev->mdev; + bool cond; + int flags; - if (!udata) + if (attr->rwq_ind_tbl) + flags = ((struct mlx5_ib_create_qp_rss *)ucmd)->flags; + else + flags = ((struct mlx5_ib_create_qp *)ucmd)->flags; + + switch (flags & (MLX5_QP_FLAG_TYPE_DCT | MLX5_QP_FLAG_TYPE_DCI)) { + case MLX5_QP_FLAG_TYPE_DCI: + qp->type = MLX5_IB_QPT_DCI; + break; + case MLX5_QP_FLAG_TYPE_DCT: + qp->type = MLX5_IB_QPT_DCT; + break; + default: + if (qp->type != IB_QPT_DRIVER) + break; + /* + * It is IB_QPT_DRIVER and or no subtype or + * wrong subtype were provided. + */ return -EINVAL; + } - if (udata->inlen < sizeof(*ucmd)) { - mlx5_ib_dbg(dev, "create_qp user command is smaller than expected\n"); + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_TYPE_DCI, true, qp); + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_TYPE_DCT, true, qp); + + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_SIGNATURE, true, qp); + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_SCATTER_CQE, + MLX5_CAP_GEN(mdev, sctr_data_cqe), qp); + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_ALLOW_SCATTER_CQE, + MLX5_CAP_GEN(mdev, sctr_data_cqe), qp); + + if (qp->type == IB_QPT_RAW_PACKET) { + cond = MLX5_CAP_ETH(mdev, tunnel_stateless_vxlan) || + MLX5_CAP_ETH(mdev, tunnel_stateless_gre) || + MLX5_CAP_ETH(mdev, tunnel_stateless_geneve_rx); + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_TUNNEL_OFFLOADS, + cond, qp); + process_vendor_flag(dev, &flags, + MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC, true, + qp); + process_vendor_flag(dev, &flags, + MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC, true, + qp); + } + + if (qp->type == IB_QPT_RC) + process_vendor_flag(dev, &flags, + MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE, + MLX5_CAP_GEN(mdev, qp_packet_based), qp); + + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_BFREG_INDEX, true, qp); + process_vendor_flag(dev, &flags, MLX5_QP_FLAG_UAR_PAGE_INDEX, true, qp); + + cond = qp->flags_en & ~(MLX5_QP_FLAG_TUNNEL_OFFLOADS | + MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC | + MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC); + if (attr->rwq_ind_tbl && cond) { + mlx5_ib_dbg(dev, "RSS RAW QP has unsupported flags 0x%X\n", + cond); return -EINVAL; } - err = ib_copy_from_udata(ucmd, udata, sizeof(*ucmd)); - if (err) - return err; - if ((ucmd->flags & MLX_QP_FLAGS) == MLX5_QP_FLAG_TYPE_DCI) { - init_attr->qp_type = MLX5_IB_QPT_DCI; - } else { - if ((ucmd->flags & MLX_QP_FLAGS) == MLX5_QP_FLAG_TYPE_DCT) { - init_attr->qp_type = MLX5_IB_QPT_DCT; - } else { - mlx5_ib_dbg(dev, "Invalid QP flags\n"); - return -EINVAL; - } + if (flags) + mlx5_ib_dbg(dev, "udata has unsupported flags 0x%X\n", flags); + + return (flags) ? -EINVAL : 0; } - if (!MLX5_CAP_GEN(dev->mdev, dct)) { - mlx5_ib_dbg(dev, "DC transport is not supported\n"); - return -EOPNOTSUPP; +static void process_create_flag(struct mlx5_ib_dev *dev, int *flags, int flag, + bool cond, struct mlx5_ib_qp *qp) +{ + if (!(*flags & flag)) + return; + + if (cond) { + qp->flags |= flag; + *flags &= ~flag; + return; } - return 0; + if (flag == MLX5_IB_QP_CREATE_WC_TEST) { + /* + * Special case, if condition didn't meet, it won't be error, + * just different in-kernel flow. + */ + *flags &= ~MLX5_IB_QP_CREATE_WC_TEST; + return; + } + mlx5_ib_dbg(dev, "Verbs create QP flag 0x%X is not supported\n", flag); } -struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, - struct ib_qp_init_attr *verbs_init_attr, - struct ib_udata *udata) +static int process_create_flags(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + struct ib_qp_init_attr *attr) { - struct mlx5_ib_dev *dev; - struct mlx5_ib_qp *qp; - u16 xrcdn = 0; - int err; - struct ib_qp_init_attr mlx_init_attr; - struct ib_qp_init_attr *init_attr = verbs_init_attr; - struct mlx5_ib_ucontext *ucontext = rdma_udata_to_drv_context( - udata, struct mlx5_ib_ucontext, ibucontext); + enum ib_qp_type qp_type = qp->type; + struct mlx5_core_dev *mdev = dev->mdev; + int create_flags = attr->create_flags; + bool cond; - if (pd) { - dev = to_mdev(pd->device); + if (qp->type == IB_QPT_UD && dev->profile == &raw_eth_profile) + if (create_flags & ~MLX5_IB_QP_CREATE_WC_TEST) + return -EINVAL; - if (init_attr->qp_type == IB_QPT_RAW_PACKET) { - if (!ucontext) { - mlx5_ib_dbg(dev, "Raw Packet QP is not supported for kernel consumers\n"); - return ERR_PTR(-EINVAL); - } else if (!ucontext->cqe_version) { - mlx5_ib_dbg(dev, "Raw Packet QP is only supported for CQE version > 0\n"); - return ERR_PTR(-EINVAL); - } - } - } else { - /* being cautious here */ - if (init_attr->qp_type != IB_QPT_XRC_TGT && - init_attr->qp_type != MLX5_IB_QPT_REG_UMR) { - pr_warn("%s: no PD for transport %s\n", __func__, - ib_qp_type_str(init_attr->qp_type)); - return ERR_PTR(-EINVAL); - } - dev = to_mdev(to_mxrcd(init_attr->xrcd)->ibxrcd.device); + if (qp_type == MLX5_IB_QPT_DCT) + return (create_flags) ? -EINVAL : 0; + + if (qp_type == IB_QPT_RAW_PACKET && attr->rwq_ind_tbl) + return (create_flags) ? -EINVAL : 0; + + process_create_flag(dev, &create_flags, IB_QP_CREATE_NETIF_QP, + mlx5_get_flow_namespace(dev->mdev, + MLX5_FLOW_NAMESPACE_BYPASS), + qp); + process_create_flag(dev, &create_flags, + IB_QP_CREATE_INTEGRITY_EN, + MLX5_CAP_GEN(mdev, sho), qp); + process_create_flag(dev, &create_flags, + IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK, + MLX5_CAP_GEN(mdev, block_lb_mc), qp); + process_create_flag(dev, &create_flags, IB_QP_CREATE_CROSS_CHANNEL, + MLX5_CAP_GEN(mdev, cd), qp); + process_create_flag(dev, &create_flags, IB_QP_CREATE_MANAGED_SEND, + MLX5_CAP_GEN(mdev, cd), qp); + process_create_flag(dev, &create_flags, IB_QP_CREATE_MANAGED_RECV, + MLX5_CAP_GEN(mdev, cd), qp); + + if (qp_type == IB_QPT_UD) { + process_create_flag(dev, &create_flags, + IB_QP_CREATE_IPOIB_UD_LSO, + MLX5_CAP_GEN(mdev, ipoib_basic_offloads), + qp); + cond = MLX5_CAP_GEN(mdev, port_type) == MLX5_CAP_PORT_TYPE_IB; + process_create_flag(dev, &create_flags, IB_QP_CREATE_SOURCE_QPN, + cond, qp); + } + + if (qp_type == IB_QPT_RAW_PACKET) { + cond = MLX5_CAP_GEN(mdev, eth_net_offloads) && + MLX5_CAP_ETH(mdev, scatter_fcs); + process_create_flag(dev, &create_flags, + IB_QP_CREATE_SCATTER_FCS, cond, qp); + + cond = MLX5_CAP_GEN(mdev, eth_net_offloads) && + MLX5_CAP_ETH(mdev, vlan_cap); + process_create_flag(dev, &create_flags, + IB_QP_CREATE_CVLAN_STRIPPING, cond, qp); + } + + process_create_flag(dev, &create_flags, + IB_QP_CREATE_PCI_WRITE_END_PADDING, + MLX5_CAP_GEN(mdev, end_pad), qp); + + process_create_flag(dev, &create_flags, MLX5_IB_QP_CREATE_WC_TEST, + qp_type != MLX5_IB_QPT_REG_UMR, qp); + process_create_flag(dev, &create_flags, MLX5_IB_QP_CREATE_SQPN_QP1, + true, qp); + + if (create_flags) + mlx5_ib_dbg(dev, "Create QP has unsupported flags 0x%X\n", + create_flags); + + return (create_flags) ? -EINVAL : 0; +} + +static int process_udata_size(struct mlx5_ib_dev *dev, + struct mlx5_create_qp_params *params) +{ + size_t ucmd = sizeof(struct mlx5_ib_create_qp); + struct ib_udata *udata = params->udata; + size_t outlen = udata->outlen; + size_t inlen = udata->inlen; + + params->outlen = min(outlen, sizeof(struct mlx5_ib_create_qp_resp)); + params->ucmd_size = ucmd; + if (!params->is_rss_raw) { + /* User has old rdma-core, which doesn't support ECE */ + size_t min_inlen = + offsetof(struct mlx5_ib_create_qp, ece_options); + + /* + * We will check in check_ucmd_data() that user + * cleared everything after inlen. + */ + params->inlen = (inlen < min_inlen) ? 0 : min(inlen, ucmd); + goto out; } - if (init_attr->qp_type == IB_QPT_DRIVER) { - struct mlx5_ib_create_qp ucmd; + /* RSS RAW QP */ + if (inlen < offsetofend(struct mlx5_ib_create_qp_rss, flags)) + return -EINVAL; - init_attr = &mlx_init_attr; - memcpy(init_attr, verbs_init_attr, sizeof(*verbs_init_attr)); - err = set_mlx_qp_type(dev, init_attr, &ucmd, udata); - if (err) - return ERR_PTR(err); + if (outlen < offsetofend(struct mlx5_ib_create_qp_resp, bfreg_index)) + return -EINVAL; - if (init_attr->qp_type == MLX5_IB_QPT_DCI) { - if (init_attr->cap.max_recv_wr || - init_attr->cap.max_recv_sge) { - mlx5_ib_dbg(dev, "DCI QP requires zero size receive queue\n"); - return ERR_PTR(-EINVAL); - } - } else { - return mlx5_ib_create_dct(pd, init_attr, &ucmd, udata); - } + ucmd = sizeof(struct mlx5_ib_create_qp_rss); + params->ucmd_size = ucmd; + if (inlen > ucmd && !ib_is_udata_cleared(udata, ucmd, inlen - ucmd)) + return -EINVAL; + + params->inlen = min(ucmd, inlen); +out: + if (!params->inlen) + mlx5_ib_dbg(dev, "udata is too small\n"); + + return (params->inlen) ? 0 : -EINVAL; +} + +static int create_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, + struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) +{ + int err; + + if (params->is_rss_raw) { + err = create_rss_raw_qp_tir(dev, pd, qp, params); + goto out; } - switch (init_attr->qp_type) { - case IB_QPT_XRC_TGT: - case IB_QPT_XRC_INI: - if (!MLX5_CAP_GEN(dev->mdev, xrc)) { - mlx5_ib_dbg(dev, "XRC not supported\n"); - return ERR_PTR(-ENOSYS); - } - init_attr->recv_cq = NULL; - if (init_attr->qp_type == IB_QPT_XRC_TGT) { - xrcdn = to_mxrcd(init_attr->xrcd)->xrcdn; - init_attr->send_cq = NULL; - } + if (qp->type == MLX5_IB_QPT_DCT) { + err = create_dct(dev, pd, qp, params); + goto out; + } - /* fall through */ - case IB_QPT_RAW_PACKET: - case IB_QPT_RC: - case IB_QPT_UC: - case IB_QPT_UD: - case IB_QPT_SMI: - case MLX5_IB_QPT_HW_GSI: - case MLX5_IB_QPT_REG_UMR: - case MLX5_IB_QPT_DCI: - qp = kzalloc(sizeof(*qp), GFP_KERNEL); - if (!qp) - return ERR_PTR(-ENOMEM); + if (qp->type == IB_QPT_XRC_TGT) { + err = create_xrc_tgt_qp(dev, qp, params); + goto out; + } - err = create_qp_common(dev, pd, init_attr, udata, qp); - if (err) { - mlx5_ib_dbg(dev, "create_qp_common failed\n"); - kfree(qp); - return ERR_PTR(err); - } + if (params->udata) + err = create_user_qp(dev, pd, qp, params); + else + err = create_kernel_qp(dev, pd, qp, params); - if (is_qp0(init_attr->qp_type)) - qp->ibqp.qp_num = 0; - else if (is_qp1(init_attr->qp_type)) - qp->ibqp.qp_num = 1; - else - qp->ibqp.qp_num = qp->trans_qp.base.mqp.qpn; +out: + if (err) { + mlx5_ib_err(dev, "Create QP type %d failed\n", qp->type); + return err; + } - mlx5_ib_dbg(dev, "ib qpnum 0x%x, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x\n", - qp->ibqp.qp_num, qp->trans_qp.base.mqp.qpn, - init_attr->recv_cq ? to_mcq(init_attr->recv_cq)->mcq.cqn : -1, - init_attr->send_cq ? to_mcq(init_attr->send_cq)->mcq.cqn : -1); + if (is_qp0(qp->type)) + qp->ibqp.qp_num = 0; + else if (is_qp1(qp->type)) + qp->ibqp.qp_num = 1; + else + qp->ibqp.qp_num = qp->trans_qp.base.mqp.qpn; - qp->trans_qp.xrcdn = xrcdn; + mlx5_ib_dbg(dev, + "QP type %d, ib qpn 0x%X, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x, ece 0x%x\n", + qp->type, qp->ibqp.qp_num, qp->trans_qp.base.mqp.qpn, + params->attr->recv_cq ? to_mcq(params->attr->recv_cq)->mcq.cqn : + -1, + params->attr->send_cq ? to_mcq(params->attr->send_cq)->mcq.cqn : + -1, + params->resp.ece_options); - break; + return 0; +} - case IB_QPT_GSI: - return mlx5_ib_gsi_create_qp(pd, init_attr); +static int check_qp_attr(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + struct ib_qp_init_attr *attr) +{ + int ret = 0; - case IB_QPT_RAW_IPV6: - case IB_QPT_RAW_ETHERTYPE: - case IB_QPT_MAX: + switch (qp->type) { + case MLX5_IB_QPT_DCT: + ret = (!attr->srq || !attr->recv_cq) ? -EINVAL : 0; + break; + case MLX5_IB_QPT_DCI: + ret = (attr->cap.max_recv_wr || attr->cap.max_recv_sge) ? + -EINVAL : + 0; + break; + case IB_QPT_RAW_PACKET: + ret = (attr->rwq_ind_tbl && attr->send_cq) ? -EINVAL : 0; + break; default: - mlx5_ib_dbg(dev, "unsupported qp type %d\n", - init_attr->qp_type); - /* Don't support raw QPs */ - return ERR_PTR(-EOPNOTSUPP); + break; } - if (verbs_init_attr->qp_type == IB_QPT_DRIVER) - qp->qp_sub_type = init_attr->qp_type; + if (ret) + mlx5_ib_dbg(dev, "QP type %d has wrong attributes\n", qp->type); - return &qp->ibqp; + return ret; +} + +static int get_qp_uidx(struct mlx5_ib_qp *qp, + struct mlx5_create_qp_params *params) +{ + struct mlx5_ib_create_qp *ucmd = params->ucmd; + struct ib_udata *udata = params->udata; + struct mlx5_ib_ucontext *ucontext = rdma_udata_to_drv_context( + udata, struct mlx5_ib_ucontext, ibucontext); + + if (params->is_rss_raw) + return 0; + + return get_qp_user_index(ucontext, ucmd, sizeof(*ucmd), ¶ms->uidx); } static int mlx5_ib_destroy_dct(struct mlx5_ib_qp *mqp) @@ -2818,7 +2868,7 @@ static int mlx5_ib_destroy_dct(struct mlx5_ib_qp *mqp) if (mqp->state == IB_QPS_RTR) { int err; - err = mlx5_core_destroy_dct(dev->mdev, &mqp->dct.mdct); + err = mlx5_core_destroy_dct(dev, &mqp->dct.mdct); if (err) { mlx5_ib_warn(dev, "failed to destroy DCT %d\n", err); return err; @@ -2830,6 +2880,152 @@ static int mlx5_ib_destroy_dct(struct mlx5_ib_qp *mqp) return 0; } +static int check_ucmd_data(struct mlx5_ib_dev *dev, + struct mlx5_create_qp_params *params) +{ + struct ib_udata *udata = params->udata; + size_t size, last; + int ret; + + if (params->is_rss_raw) + /* + * These QPs don't have "reserved" field in their + * create_qp input struct, so their data is always valid. + */ + last = sizeof(struct mlx5_ib_create_qp_rss); + else + last = offsetof(struct mlx5_ib_create_qp, reserved); + + if (udata->inlen <= last) + return 0; + + /* + * User provides different create_qp structures based on the + * flow and we need to know if he cleared memory after our + * struct create_qp ends. + */ + size = udata->inlen - last; + ret = ib_is_udata_cleared(params->udata, last, size); + if (!ret) + mlx5_ib_dbg( + dev, + "udata is not cleared, inlen = %zu, ucmd = %zu, last = %zu, size = %zu\n", + udata->inlen, params->ucmd_size, last, size); + return ret ? 0 : -EINVAL; +} + +struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attr, + struct ib_udata *udata) +{ + struct mlx5_create_qp_params params = {}; + struct mlx5_ib_dev *dev; + struct mlx5_ib_qp *qp; + enum ib_qp_type type; + int err; + + dev = pd ? to_mdev(pd->device) : + to_mdev(to_mxrcd(attr->xrcd)->ibxrcd.device); + + err = check_qp_type(dev, attr, &type); + if (err) + return ERR_PTR(err); + + err = check_valid_flow(dev, pd, attr, udata); + if (err) + return ERR_PTR(err); + + if (attr->qp_type == IB_QPT_GSI) + return mlx5_ib_gsi_create_qp(pd, attr); + + params.udata = udata; + params.uidx = MLX5_IB_DEFAULT_UIDX; + params.attr = attr; + params.is_rss_raw = !!attr->rwq_ind_tbl; + + if (udata) { + err = process_udata_size(dev, ¶ms); + if (err) + return ERR_PTR(err); + + err = check_ucmd_data(dev, ¶ms); + if (err) + return ERR_PTR(err); + + params.ucmd = kzalloc(params.ucmd_size, GFP_KERNEL); + if (!params.ucmd) + return ERR_PTR(-ENOMEM); + + err = ib_copy_from_udata(params.ucmd, udata, params.inlen); + if (err) + goto free_ucmd; + } + + qp = kzalloc(sizeof(*qp), GFP_KERNEL); + if (!qp) { + err = -ENOMEM; + goto free_ucmd; + } + + mutex_init(&qp->mutex); + qp->type = type; + if (udata) { + err = process_vendor_flags(dev, qp, params.ucmd, attr); + if (err) + goto free_qp; + + err = get_qp_uidx(qp, ¶ms); + if (err) + goto free_qp; + } + err = process_create_flags(dev, qp, attr); + if (err) + goto free_qp; + + err = check_qp_attr(dev, qp, attr); + if (err) + goto free_qp; + + err = create_qp(dev, pd, qp, ¶ms); + if (err) + goto free_qp; + + kfree(params.ucmd); + params.ucmd = NULL; + + if (udata) + /* + * It is safe to copy response for all user create QP flows, + * including MLX5_IB_QPT_DCT, which doesn't need it. + * In that case, resp will be filled with zeros. + */ + err = ib_copy_to_udata(udata, ¶ms.resp, params.outlen); + if (err) + goto destroy_qp; + + return &qp->ibqp; + +destroy_qp: + if (qp->type == MLX5_IB_QPT_DCT) { + mlx5_ib_destroy_dct(qp); + } else { + /* + * These lines below are temp solution till QP allocation + * will be moved to be under IB/core responsiblity. + */ + qp->ibqp.send_cq = attr->send_cq; + qp->ibqp.recv_cq = attr->recv_cq; + qp->ibqp.pd = pd; + destroy_qp_common(dev, qp, udata); + } + + qp = NULL; +free_qp: + kfree(qp); +free_ucmd: + kfree(params.ucmd); + return ERR_PTR(err); +} + int mlx5_ib_destroy_qp(struct ib_qp *qp, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(qp->device); @@ -2838,7 +3034,7 @@ int mlx5_ib_destroy_qp(struct ib_qp *qp, struct ib_udata *udata) if (unlikely(qp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_destroy_qp(qp); - if (mqp->qp_sub_type == MLX5_IB_QPT_DCT) + if (mqp->type == MLX5_IB_QPT_DCT) return mlx5_ib_destroy_dct(mqp); destroy_qp_common(dev, mqp, udata); @@ -2848,14 +3044,13 @@ int mlx5_ib_destroy_qp(struct ib_qp *qp, struct ib_udata *udata) return 0; } -static int to_mlx5_access_flags(struct mlx5_ib_qp *qp, - const struct ib_qp_attr *attr, - int attr_mask, __be32 *hw_access_flags_be) +static int set_qpc_atomic_flags(struct mlx5_ib_qp *qp, + const struct ib_qp_attr *attr, int attr_mask, + void *qpc) { - u8 dest_rd_atomic; - u32 access_flags, hw_access_flags = 0; - struct mlx5_ib_dev *dev = to_mdev(qp->ibqp.device); + u8 dest_rd_atomic; + u32 access_flags; if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) dest_rd_atomic = attr->max_dest_rd_atomic; @@ -2870,8 +3065,8 @@ static int to_mlx5_access_flags(struct mlx5_ib_qp *qp, if (!dest_rd_atomic) access_flags &= IB_ACCESS_REMOTE_WRITE; - if (access_flags & IB_ACCESS_REMOTE_READ) - hw_access_flags |= MLX5_QP_BIT_RRE; + MLX5_SET(qpc, qpc, rre, !!(access_flags & IB_ACCESS_REMOTE_READ)); + if (access_flags & IB_ACCESS_REMOTE_ATOMIC) { int atomic_mode; @@ -2879,15 +3074,11 @@ static int to_mlx5_access_flags(struct mlx5_ib_qp *qp, if (atomic_mode < 0) return -EOPNOTSUPP; - hw_access_flags |= MLX5_QP_BIT_RAE; - hw_access_flags |= atomic_mode << MLX5_ATOMIC_MODE_OFFSET; + MLX5_SET(qpc, qpc, rae, 1); + MLX5_SET(qpc, qpc, atomic_mode, atomic_mode); } - if (access_flags & IB_ACCESS_REMOTE_WRITE) - hw_access_flags |= MLX5_QP_BIT_RWE; - - *hw_access_flags_be = cpu_to_be32(hw_access_flags); - + MLX5_SET(qpc, qpc, rwe, !!(access_flags & IB_ACCESS_REMOTE_WRITE)); return 0; } @@ -2933,7 +3124,7 @@ static int modify_raw_packet_eth_prio(struct mlx5_core_dev *dev, tisc = MLX5_ADDR_OF(modify_tis_in, in, ctx); MLX5_SET(tisc, tisc, prio, ((sl & 0x7) << 1)); - err = mlx5_core_modify_tis(dev, sq->tisn, in, inlen); + err = mlx5_core_modify_tis(dev, sq->tisn, in); kvfree(in); @@ -2960,18 +3151,29 @@ static int modify_raw_packet_tx_affinity(struct mlx5_core_dev *dev, tisc = MLX5_ADDR_OF(modify_tis_in, in, ctx); MLX5_SET(tisc, tisc, lag_tx_port_affinity, tx_affinity); - err = mlx5_core_modify_tis(dev, sq->tisn, in, inlen); + err = mlx5_core_modify_tis(dev, sq->tisn, in); kvfree(in); return err; } +static void mlx5_set_path_udp_sport(void *path, const struct rdma_ah_attr *ah, + u32 lqpn, u32 rqpn) + +{ + u32 fl = ah->grh.flow_label; + + if (!fl) + fl = rdma_calc_flow_label(lqpn, rqpn); + + MLX5_SET(ads, path, udp_sport, rdma_flow_label_to_udp_sport(fl)); +} + static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, - const struct rdma_ah_attr *ah, - struct mlx5_qp_path *path, u8 port, int attr_mask, - u32 path_flags, const struct ib_qp_attr *attr, - bool alt) + const struct rdma_ah_attr *ah, void *path, u8 port, + int attr_mask, u32 path_flags, + const struct ib_qp_attr *attr, bool alt) { const struct ib_global_route *grh = rdma_ah_read_grh(ah); int err; @@ -2980,8 +3182,8 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, u8 sl = rdma_ah_get_sl(ah); if (attr_mask & IB_QP_PKEY_INDEX) - path->pkey_index = cpu_to_be16(alt ? attr->alt_pkey_index : - attr->pkey_index); + MLX5_SET(ads, path, pkey_index, + alt ? attr->alt_pkey_index : attr->pkey_index); if (ah_flags & IB_AH_GRH) { if (grh->sgid_index >= @@ -2997,45 +3199,49 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, if (!(ah_flags & IB_AH_GRH)) return -EINVAL; - memcpy(path->rmac, ah->roce.dmac, sizeof(ah->roce.dmac)); - if (qp->ibqp.qp_type == IB_QPT_RC || - qp->ibqp.qp_type == IB_QPT_UC || - qp->ibqp.qp_type == IB_QPT_XRC_INI || - qp->ibqp.qp_type == IB_QPT_XRC_TGT) - path->udp_sport = - mlx5_get_roce_udp_sport(dev, ah->grh.sgid_attr); - path->dci_cfi_prio_sl = (sl & 0x7) << 4; + ether_addr_copy(MLX5_ADDR_OF(ads, path, rmac_47_32), + ah->roce.dmac); + if ((qp->ibqp.qp_type == IB_QPT_RC || + qp->ibqp.qp_type == IB_QPT_UC || + qp->ibqp.qp_type == IB_QPT_XRC_INI || + qp->ibqp.qp_type == IB_QPT_XRC_TGT) && + (grh->sgid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) && + (attr_mask & IB_QP_DEST_QPN)) + mlx5_set_path_udp_sport(path, ah, + qp->ibqp.qp_num, + attr->dest_qp_num); + MLX5_SET(ads, path, eth_prio, sl & 0x7); gid_type = ah->grh.sgid_attr->gid_type; if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) - path->ecn_dscp = (grh->traffic_class >> 2) & 0x3f; + MLX5_SET(ads, path, dscp, grh->traffic_class >> 2); } else { - path->fl_free_ar = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0; - path->fl_free_ar |= - (path_flags & MLX5_PATH_FLAG_FREE_AR) ? 0x40 : 0; - path->rlid = cpu_to_be16(rdma_ah_get_dlid(ah)); - path->grh_mlid = rdma_ah_get_path_bits(ah) & 0x7f; - if (ah_flags & IB_AH_GRH) - path->grh_mlid |= 1 << 7; - path->dci_cfi_prio_sl = sl & 0xf; + MLX5_SET(ads, path, fl, !!(path_flags & MLX5_PATH_FLAG_FL)); + MLX5_SET(ads, path, free_ar, + !!(path_flags & MLX5_PATH_FLAG_FREE_AR)); + MLX5_SET(ads, path, rlid, rdma_ah_get_dlid(ah)); + MLX5_SET(ads, path, mlid, rdma_ah_get_path_bits(ah)); + MLX5_SET(ads, path, grh, !!(ah_flags & IB_AH_GRH)); + MLX5_SET(ads, path, sl, sl); } if (ah_flags & IB_AH_GRH) { - path->mgid_index = grh->sgid_index; - path->hop_limit = grh->hop_limit; - path->tclass_flowlabel = - cpu_to_be32((grh->traffic_class << 20) | - (grh->flow_label)); - memcpy(path->rgid, grh->dgid.raw, 16); + MLX5_SET(ads, path, src_addr_index, grh->sgid_index); + MLX5_SET(ads, path, hop_limit, grh->hop_limit); + MLX5_SET(ads, path, tclass, grh->traffic_class); + MLX5_SET(ads, path, flow_label, grh->flow_label); + memcpy(MLX5_ADDR_OF(ads, path, rgid_rip), grh->dgid.raw, + sizeof(grh->dgid.raw)); } err = ib_rate_to_mlx5(dev, rdma_ah_get_static_rate(ah)); if (err < 0) return err; - path->static_rate = err; - path->port = port; + MLX5_SET(ads, path, stat_rate, err); + MLX5_SET(ads, path, vhca_port_num, port); if (attr_mask & IB_QP_TIMEOUT) - path->ackto_lt = (alt ? attr->alt_timeout : attr->timeout) << 3; + MLX5_SET(ads, path, ack_timeout, + alt ? attr->alt_timeout : attr->timeout); if ((qp->ibqp.qp_type == IB_QPT_RAW_PACKET) && qp->sq.wqe_cnt) return modify_raw_packet_eth_prio(dev->mdev, @@ -3052,10 +3258,12 @@ static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_Q MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX | - MLX5_QP_OPTPAR_PRI_PORT, + MLX5_QP_OPTPAR_PRI_PORT | + MLX5_QP_OPTPAR_LAG_TX_AFF, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX | - MLX5_QP_OPTPAR_PRI_PORT, + MLX5_QP_OPTPAR_PRI_PORT | + MLX5_QP_OPTPAR_LAG_TX_AFF, [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_Q_KEY | MLX5_QP_OPTPAR_PRI_PORT, @@ -3063,17 +3271,20 @@ static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_Q MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | MLX5_QP_OPTPAR_PKEY_INDEX | - MLX5_QP_OPTPAR_PRI_PORT, + MLX5_QP_OPTPAR_PRI_PORT | + MLX5_QP_OPTPAR_LAG_TX_AFF, }, [MLX5_QP_STATE_RTR] = { [MLX5_QP_ST_RC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | - MLX5_QP_OPTPAR_PKEY_INDEX, + MLX5_QP_OPTPAR_PKEY_INDEX | + MLX5_QP_OPTPAR_LAG_TX_AFF, [MLX5_QP_ST_UC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH | MLX5_QP_OPTPAR_RWE | - MLX5_QP_OPTPAR_PKEY_INDEX, + MLX5_QP_OPTPAR_PKEY_INDEX | + MLX5_QP_OPTPAR_LAG_TX_AFF, [MLX5_QP_ST_UD] = MLX5_QP_OPTPAR_PKEY_INDEX | MLX5_QP_OPTPAR_Q_KEY, [MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_PKEY_INDEX | @@ -3082,7 +3293,8 @@ static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_Q MLX5_QP_OPTPAR_RRE | MLX5_QP_OPTPAR_RAE | MLX5_QP_OPTPAR_RWE | - MLX5_QP_OPTPAR_PKEY_INDEX, + MLX5_QP_OPTPAR_PKEY_INDEX | + MLX5_QP_OPTPAR_LAG_TX_AFF, }, }, [MLX5_QP_STATE_RTR] = { @@ -3240,7 +3452,7 @@ static int modify_raw_packet_qp_rq( "RAW PACKET QP counters are not supported on current FW\n"); } - err = mlx5_core_modify_rq(dev->mdev, rq->base.mqp.qpn, in, inlen); + err = mlx5_core_modify_rq(dev->mdev, rq->base.mqp.qpn, in); if (err) goto out; @@ -3303,7 +3515,7 @@ static int modify_raw_packet_qp_sq( MLX5_SET(sqc, sqc, packet_pacing_rate_limit_index, rl_index); } - err = mlx5_core_modify_sq(dev, sq->base.mqp.qpn, in, inlen); + err = mlx5_core_modify_sq(dev, sq->base.mqp.qpn, in); if (err) { /* Remove new rate from table if failed */ if (new_rate_added) @@ -3343,7 +3555,7 @@ static int modify_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, switch (raw_qp_param->operation) { case MLX5_CMD_OP_RST2INIT_QP: rq_state = MLX5_RQC_STATE_RDY; - sq_state = MLX5_SQC_STATE_RDY; + sq_state = MLX5_SQC_STATE_RST; break; case MLX5_CMD_OP_2ERR_QP: rq_state = MLX5_RQC_STATE_ERR; @@ -3355,13 +3567,11 @@ static int modify_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, break; case MLX5_CMD_OP_RTR2RTS_QP: case MLX5_CMD_OP_RTS2RTS_QP: - if (raw_qp_param->set_mask == - MLX5_RAW_QP_RATE_LIMIT) { - modify_rq = 0; - sq_state = sq->state; - } else { - return raw_qp_param->set_mask ? -EINVAL : 0; - } + if (raw_qp_param->set_mask & ~MLX5_RAW_QP_RATE_LIMIT) + return -EINVAL; + + modify_rq = 0; + sq_state = MLX5_SQC_STATE_RDY; break; case MLX5_CMD_OP_INIT2INIT_QP: case MLX5_CMD_OP_INIT2RTR_QP: @@ -3416,43 +3626,80 @@ static int modify_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, return 0; } -static unsigned int get_tx_affinity(struct mlx5_ib_dev *dev, - struct mlx5_ib_pd *pd, - struct mlx5_ib_qp_base *qp_base, - u8 port_num, struct ib_udata *udata) +static unsigned int get_tx_affinity_rr(struct mlx5_ib_dev *dev, + struct ib_udata *udata) +{ + struct mlx5_ib_ucontext *ucontext = rdma_udata_to_drv_context( + udata, struct mlx5_ib_ucontext, ibucontext); + u8 port_num = mlx5_core_native_port_num(dev->mdev) - 1; + atomic_t *tx_port_affinity; + + if (ucontext) + tx_port_affinity = &ucontext->tx_port_affinity; + else + tx_port_affinity = &dev->port[port_num].roce.tx_port_affinity; + + return (unsigned int)atomic_add_return(1, tx_port_affinity) % + MLX5_MAX_PORTS + 1; +} + +static bool qp_supports_affinity(struct ib_qp *qp) +{ + if ((qp->qp_type == IB_QPT_RC) || + (qp->qp_type == IB_QPT_UD) || + (qp->qp_type == IB_QPT_UC) || + (qp->qp_type == IB_QPT_RAW_PACKET) || + (qp->qp_type == IB_QPT_XRC_INI) || + (qp->qp_type == IB_QPT_XRC_TGT)) + return true; + return false; +} + +static unsigned int get_tx_affinity(struct ib_qp *qp, + const struct ib_qp_attr *attr, + int attr_mask, u8 init, + struct ib_udata *udata) { struct mlx5_ib_ucontext *ucontext = rdma_udata_to_drv_context( udata, struct mlx5_ib_ucontext, ibucontext); - unsigned int tx_port_affinity; + struct mlx5_ib_dev *dev = to_mdev(qp->device); + struct mlx5_ib_qp *mqp = to_mqp(qp); + struct mlx5_ib_qp_base *qp_base; + unsigned int tx_affinity; + + if (!(mlx5_ib_lag_should_assign_affinity(dev) && + qp_supports_affinity(qp))) + return 0; - if (ucontext) { - tx_port_affinity = (unsigned int)atomic_add_return( - 1, &ucontext->tx_port_affinity) % - MLX5_MAX_PORTS + - 1; + if (mqp->flags & MLX5_IB_QP_CREATE_SQPN_QP1) + tx_affinity = mqp->gsi_lag_port; + else if (init) + tx_affinity = get_tx_affinity_rr(dev, udata); + else if ((attr_mask & IB_QP_AV) && attr->xmit_slave) + tx_affinity = + mlx5_lag_get_slave_port(dev->mdev, attr->xmit_slave); + else + return 0; + + qp_base = &mqp->trans_qp.base; + if (ucontext) mlx5_ib_dbg(dev, "Set tx affinity 0x%x to qpn 0x%x ucontext %p\n", - tx_port_affinity, qp_base->mqp.qpn, ucontext); - } else { - tx_port_affinity = - (unsigned int)atomic_add_return( - 1, &dev->port[port_num].roce.tx_port_affinity) % - MLX5_MAX_PORTS + - 1; + tx_affinity, qp_base->mqp.qpn, ucontext); + else mlx5_ib_dbg(dev, "Set tx affinity 0x%x to qpn 0x%x\n", - tx_port_affinity, qp_base->mqp.qpn); - } - - return tx_port_affinity; + tx_affinity, qp_base->mqp.qpn); + return tx_affinity; } static int __mlx5_ib_qp_set_counter(struct ib_qp *qp, struct rdma_counter *counter) { struct mlx5_ib_dev *dev = to_mdev(qp->device); + u32 in[MLX5_ST_SZ_DW(rts2rts_qp_in)] = {}; struct mlx5_ib_qp *mqp = to_mqp(qp); - struct mlx5_qp_context context = {}; struct mlx5_ib_qp_base *base; u32 set_id; + u32 *qpc; if (counter) set_id = counter->id; @@ -3460,12 +3707,15 @@ static int __mlx5_ib_qp_set_counter(struct ib_qp *qp, set_id = mlx5_ib_get_counters_id(dev, mqp->port - 1); base = &mqp->trans_qp.base; - context.qp_counter_set_usr_page &= cpu_to_be32(0xffffff); - context.qp_counter_set_usr_page |= cpu_to_be32(set_id << 24); - return mlx5_core_qp_modify(dev->mdev, - MLX5_CMD_OP_RTS2RTS_QP, - MLX5_QP_OPTPAR_COUNTER_SET_ID, - &context, &base->mqp); + MLX5_SET(rts2rts_qp_in, in, opcode, MLX5_CMD_OP_RTS2RTS_QP); + MLX5_SET(rts2rts_qp_in, in, qpn, base->mqp.qpn); + MLX5_SET(rts2rts_qp_in, in, uid, base->mqp.uid); + MLX5_SET(rts2rts_qp_in, in, opt_param_mask, + MLX5_QP_OPTPAR_COUNTER_SET_ID); + + qpc = MLX5_ADDR_OF(rts2rts_qp_in, in, qpc); + MLX5_SET(qpc, qpc, counter_set_id, set_id); + return mlx5_cmd_exec_in(dev->mdev, rts2rts_qp, in); } static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, @@ -3473,6 +3723,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, enum ib_qp_state cur_state, enum ib_qp_state new_state, const struct mlx5_ib_modify_qp *ucmd, + struct mlx5_ib_modify_qp_resp *resp, struct ib_udata *udata) { static const u16 optab[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE] = { @@ -3516,67 +3767,60 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_qp_base *base = &qp->trans_qp.base; struct mlx5_ib_cq *send_cq, *recv_cq; - struct mlx5_qp_context *context; struct mlx5_ib_pd *pd; enum mlx5_qp_state mlx5_cur, mlx5_new; - enum mlx5_qp_optpar optpar; + void *qpc, *pri_path, *alt_path; + enum mlx5_qp_optpar optpar = 0; u32 set_id = 0; int mlx5_st; int err; u16 op; u8 tx_affinity = 0; - mlx5_st = to_mlx5_st(ibqp->qp_type == IB_QPT_DRIVER ? - qp->qp_sub_type : ibqp->qp_type); + mlx5_st = to_mlx5_st(qp->type); if (mlx5_st < 0) return -EINVAL; - context = kzalloc(sizeof(*context), GFP_KERNEL); - if (!context) + qpc = kzalloc(MLX5_ST_SZ_BYTES(qpc), GFP_KERNEL); + if (!qpc) return -ENOMEM; - pd = get_pd(qp); - context->flags = cpu_to_be32(mlx5_st << 16); + pd = to_mpd(qp->ibqp.pd); + MLX5_SET(qpc, qpc, st, mlx5_st); if (!(attr_mask & IB_QP_PATH_MIG_STATE)) { - context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11); + MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED); } else { switch (attr->path_mig_state) { case IB_MIG_MIGRATED: - context->flags |= cpu_to_be32(MLX5_QP_PM_MIGRATED << 11); + MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_MIGRATED); break; case IB_MIG_REARM: - context->flags |= cpu_to_be32(MLX5_QP_PM_REARM << 11); + MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_REARM); break; case IB_MIG_ARMED: - context->flags |= cpu_to_be32(MLX5_QP_PM_ARMED << 11); + MLX5_SET(qpc, qpc, pm_state, MLX5_QP_PM_ARMED); break; } } - if ((cur_state == IB_QPS_RESET) && (new_state == IB_QPS_INIT)) { - if ((ibqp->qp_type == IB_QPT_RC) || - (ibqp->qp_type == IB_QPT_UD && - !(qp->flags & MLX5_IB_QP_SQPN_QP1)) || - (ibqp->qp_type == IB_QPT_UC) || - (ibqp->qp_type == IB_QPT_RAW_PACKET) || - (ibqp->qp_type == IB_QPT_XRC_INI) || - (ibqp->qp_type == IB_QPT_XRC_TGT)) { - if (dev->lag_active) { - u8 p = mlx5_core_native_port_num(dev->mdev) - 1; - tx_affinity = get_tx_affinity(dev, pd, base, p, - udata); - context->flags |= cpu_to_be32(tx_affinity << 24); - } - } - } + tx_affinity = get_tx_affinity(ibqp, attr, attr_mask, + cur_state == IB_QPS_RESET && + new_state == IB_QPS_INIT, udata); + + MLX5_SET(qpc, qpc, lag_tx_port_affinity, tx_affinity); + if (tx_affinity && new_state == IB_QPS_RTR && + MLX5_CAP_GEN(dev->mdev, init2_lag_tx_port_affinity)) + optpar |= MLX5_QP_OPTPAR_LAG_TX_AFF; if (is_sqp(ibqp->qp_type)) { - context->mtu_msgmax = (IB_MTU_256 << 5) | 8; + MLX5_SET(qpc, qpc, mtu, IB_MTU_256); + MLX5_SET(qpc, qpc, log_msg_max, 8); } else if ((ibqp->qp_type == IB_QPT_UD && - !(qp->flags & MLX5_IB_QP_UNDERLAY)) || + !(qp->flags & IB_QP_CREATE_SOURCE_QPN)) || ibqp->qp_type == MLX5_IB_QPT_REG_UMR) { - context->mtu_msgmax = (IB_MTU_4096 << 5) | 12; + MLX5_SET(qpc, qpc, mtu, IB_MTU_4096); + MLX5_SET(qpc, qpc, log_msg_max, 12); } else if (attr_mask & IB_QP_PATH_MTU) { if (attr->path_mtu < IB_MTU_256 || attr->path_mtu > IB_MTU_4096) { @@ -3584,40 +3828,45 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, err = -EINVAL; goto out; } - context->mtu_msgmax = (attr->path_mtu << 5) | - (u8)MLX5_CAP_GEN(dev->mdev, log_max_msg); + MLX5_SET(qpc, qpc, mtu, attr->path_mtu); + MLX5_SET(qpc, qpc, log_msg_max, + MLX5_CAP_GEN(dev->mdev, log_max_msg)); } if (attr_mask & IB_QP_DEST_QPN) - context->log_pg_sz_remote_qpn = cpu_to_be32(attr->dest_qp_num); + MLX5_SET(qpc, qpc, remote_qpn, attr->dest_qp_num); + + pri_path = MLX5_ADDR_OF(qpc, qpc, primary_address_path); + alt_path = MLX5_ADDR_OF(qpc, qpc, secondary_address_path); if (attr_mask & IB_QP_PKEY_INDEX) - context->pri_path.pkey_index = cpu_to_be16(attr->pkey_index); + MLX5_SET(ads, pri_path, pkey_index, attr->pkey_index); /* todo implement counter_index functionality */ if (is_sqp(ibqp->qp_type)) - context->pri_path.port = qp->port; + MLX5_SET(ads, pri_path, vhca_port_num, qp->port); if (attr_mask & IB_QP_PORT) - context->pri_path.port = attr->port_num; + MLX5_SET(ads, pri_path, vhca_port_num, attr->port_num); if (attr_mask & IB_QP_AV) { - err = mlx5_set_path(dev, qp, &attr->ah_attr, &context->pri_path, - attr_mask & IB_QP_PORT ? attr->port_num : qp->port, + err = mlx5_set_path(dev, qp, &attr->ah_attr, pri_path, + attr_mask & IB_QP_PORT ? attr->port_num : + qp->port, attr_mask, 0, attr, false); if (err) goto out; } if (attr_mask & IB_QP_TIMEOUT) - context->pri_path.ackto_lt |= attr->timeout << 3; + MLX5_SET(ads, pri_path, ack_timeout, attr->timeout); if (attr_mask & IB_QP_ALT_PATH) { - err = mlx5_set_path(dev, qp, &attr->alt_ah_attr, - &context->alt_path, + err = mlx5_set_path(dev, qp, &attr->alt_ah_attr, alt_path, attr->alt_port_num, - attr_mask | IB_QP_PKEY_INDEX | IB_QP_TIMEOUT, + attr_mask | IB_QP_PKEY_INDEX | + IB_QP_TIMEOUT, 0, attr, true); if (err) goto out; @@ -3626,75 +3875,68 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, 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; - context->cqn_recv = recv_cq ? cpu_to_be32(recv_cq->mcq.cqn) : 0; - context->params1 = cpu_to_be32(MLX5_IB_ACK_REQ_FREQ << 28); + MLX5_SET(qpc, qpc, pd, pd ? pd->pdn : to_mpd(dev->devr.p0)->pdn); + if (send_cq) + MLX5_SET(qpc, qpc, cqn_snd, send_cq->mcq.cqn); + if (recv_cq) + MLX5_SET(qpc, qpc, cqn_rcv, recv_cq->mcq.cqn); + + MLX5_SET(qpc, qpc, log_ack_req_freq, MLX5_IB_ACK_REQ_FREQ); if (attr_mask & IB_QP_RNR_RETRY) - context->params1 |= cpu_to_be32(attr->rnr_retry << 13); + MLX5_SET(qpc, qpc, rnr_retry, attr->rnr_retry); if (attr_mask & IB_QP_RETRY_CNT) - context->params1 |= cpu_to_be32(attr->retry_cnt << 16); + MLX5_SET(qpc, qpc, retry_count, attr->retry_cnt); - if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC) { - if (attr->max_rd_atomic) - context->params1 |= - cpu_to_be32(fls(attr->max_rd_atomic - 1) << 21); - } + if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && attr->max_rd_atomic) + MLX5_SET(qpc, qpc, log_sra_max, ilog2(attr->max_rd_atomic)); if (attr_mask & IB_QP_SQ_PSN) - context->next_send_psn = cpu_to_be32(attr->sq_psn); + MLX5_SET(qpc, qpc, next_send_psn, attr->sq_psn); - if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) { - if (attr->max_dest_rd_atomic) - context->params2 |= - cpu_to_be32(fls(attr->max_dest_rd_atomic - 1) << 21); - } + if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && attr->max_dest_rd_atomic) + MLX5_SET(qpc, qpc, log_rra_max, + ilog2(attr->max_dest_rd_atomic)); if (attr_mask & (IB_QP_ACCESS_FLAGS | IB_QP_MAX_DEST_RD_ATOMIC)) { - __be32 access_flags; - - err = to_mlx5_access_flags(qp, attr, attr_mask, &access_flags); + err = set_qpc_atomic_flags(qp, attr, attr_mask, qpc); if (err) goto out; - - context->params2 |= access_flags; } if (attr_mask & IB_QP_MIN_RNR_TIMER) - context->rnr_nextrecvpsn |= cpu_to_be32(attr->min_rnr_timer << 24); + MLX5_SET(qpc, qpc, min_rnr_nak, attr->min_rnr_timer); if (attr_mask & IB_QP_RQ_PSN) - context->rnr_nextrecvpsn |= cpu_to_be32(attr->rq_psn); + MLX5_SET(qpc, qpc, next_rcv_psn, attr->rq_psn); if (attr_mask & IB_QP_QKEY) - context->qkey = cpu_to_be32(attr->qkey); + MLX5_SET(qpc, qpc, q_key, attr->qkey); if (qp->rq.wqe_cnt && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) - context->db_rec_addr = cpu_to_be64(qp->db.dma); + MLX5_SET64(qpc, qpc, dbr_addr, qp->db.dma); 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; /* Underlay port should be used - index 0 function per port */ - if (qp->flags & MLX5_IB_QP_UNDERLAY) + if (qp->flags & IB_QP_CREATE_SOURCE_QPN) port_num = 0; if (ibqp->counter) set_id = ibqp->counter->id; else set_id = mlx5_ib_get_counters_id(dev, port_num); - context->qp_counter_set_usr_page |= - cpu_to_be32(set_id << 24); + MLX5_SET(qpc, qpc, counter_set_id, set_id); } if (!ibqp->uobject && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) - context->sq_crq_size |= cpu_to_be16(1 << 4); + MLX5_SET(qpc, qpc, rlky, 1); - if (qp->flags & MLX5_IB_QP_SQPN_QP1) - context->deth_sqpn = cpu_to_be32(1); + if (qp->flags & MLX5_IB_QP_CREATE_SQPN_QP1) + MLX5_SET(qpc, qpc, deth_sqpn, 1); mlx5_cur = to_mlx5_state(cur_state); mlx5_new = to_mlx5_state(new_state); @@ -3706,11 +3948,11 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, } op = optab[mlx5_cur][mlx5_new]; - optpar = ib_mask_to_mlx5_opt(attr_mask); + optpar |= ib_mask_to_mlx5_opt(attr_mask); optpar &= opt_mask[mlx5_cur][mlx5_new][mlx5_st]; if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET || - qp->flags & MLX5_IB_QP_UNDERLAY) { + qp->flags & IB_QP_CREATE_SOURCE_QPN) { struct mlx5_modify_raw_qp_param raw_qp_param = {}; raw_qp_param.operation = op; @@ -3752,8 +3994,15 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, err = modify_raw_packet_qp(dev, qp, &raw_qp_param, tx_affinity); } else { - err = mlx5_core_qp_modify(dev->mdev, op, optpar, context, - &base->mqp); + if (udata) { + /* For the kernel flows, the resp will stay zero */ + resp->ece_options = + MLX5_CAP_GEN(dev->mdev, ece_support) ? + ucmd->ece_options : 0; + resp->response_length = sizeof(*resp); + } + err = mlx5_core_qp_modify(dev, op, optpar, qpc, &base->mqp, + &resp->ece_options); } if (err) @@ -3800,7 +4049,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, } out: - kfree(context); + kfree(qpc); return err; } @@ -3858,14 +4107,15 @@ static bool modify_dci_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state new * Other transitions and attributes are illegal */ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, - int attr_mask, struct ib_udata *udata) + int attr_mask, struct mlx5_ib_modify_qp *ucmd, + struct ib_udata *udata) { struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_dev *dev = to_mdev(ibqp->device); enum ib_qp_state cur_state, new_state; - int err = 0; int required = IB_QP_STATE; void *dctc; + int err; if (!(attr_mask & IB_QP_STATE)) return -EINVAL; @@ -3874,6 +4124,15 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, new_state = attr->qp_state; dctc = MLX5_ADDR_OF(create_dct_in, qp->dct.in, dct_context_entry); + if (MLX5_CAP_GEN(dev->mdev, ece_support) && ucmd->ece_options) + /* + * DCT doesn't initialize QP till modify command is executed, + * so we need to overwrite previously set ECE field if user + * provided any value except zero, which means not set/not + * valid. + */ + MLX5_SET(dctc, dctc, ece, ucmd->ece_options); + if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { u16 set_id; @@ -3906,16 +4165,20 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, set_id = mlx5_ib_get_counters_id(dev, attr->port_num - 1); MLX5_SET(dctc, dctc, counter_set_id, set_id); - } else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { struct mlx5_ib_modify_qp_resp resp = {}; - u32 out[MLX5_ST_SZ_DW(create_dct_out)] = {0}; - u32 min_resp_len = offsetof(typeof(resp), dctn) + - sizeof(resp.dctn); + u32 out[MLX5_ST_SZ_DW(create_dct_out)] = {}; + u32 min_resp_len = offsetofend(typeof(resp), dctn); if (udata->outlen < min_resp_len) return -EINVAL; - resp.response_length = min_resp_len; + /* + * If we don't have enough space for the ECE options, + * simply indicate it with resp.response_length. + */ + resp.response_length = (udata->outlen < sizeof(resp)) ? + min_resp_len : + sizeof(resp); required |= IB_QP_MIN_RNR_TIMER | IB_QP_AV | IB_QP_PATH_MTU; if (!is_valid_mask(attr_mask, required, 0)) @@ -3927,37 +4190,37 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, MLX5_SET(dctc, dctc, my_addr_index, attr->ah_attr.grh.sgid_index); MLX5_SET(dctc, dctc, hop_limit, attr->ah_attr.grh.hop_limit); - err = mlx5_core_create_dct(dev->mdev, &qp->dct.mdct, qp->dct.in, + err = mlx5_core_create_dct(dev, &qp->dct.mdct, qp->dct.in, MLX5_ST_SZ_BYTES(create_dct_in), out, sizeof(out)); if (err) return err; resp.dctn = qp->dct.mdct.mqp.qpn; + if (MLX5_CAP_GEN(dev->mdev, ece_support)) + resp.ece_options = MLX5_GET(create_dct_out, out, ece); err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) { - mlx5_core_destroy_dct(dev->mdev, &qp->dct.mdct); + mlx5_core_destroy_dct(dev, &qp->dct.mdct); return err; } } else { mlx5_ib_warn(dev, "Modify DCT: Invalid transition from %d to %d\n", cur_state, new_state); return -EINVAL; } - if (err) - qp->state = IB_QPS_ERR; - else - qp->state = new_state; - return err; + + qp->state = new_state; + return 0; } int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibqp->device); + struct mlx5_ib_modify_qp_resp resp = {}; struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_modify_qp ucmd = {}; enum ib_qp_type qp_type; enum ib_qp_state cur_state, new_state; - size_t required_cmd_sz; int err = -EINVAL; int port; @@ -3965,9 +4228,7 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, return -ENOSYS; if (udata && udata->inlen) { - required_cmd_sz = offsetof(typeof(ucmd), reserved) + - sizeof(ucmd.reserved); - if (udata->inlen < required_cmd_sz) + if (udata->inlen < offsetofend(typeof(ucmd), ece_options)) return -EINVAL; if (udata->inlen > sizeof(ucmd) && @@ -3980,23 +4241,20 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, return -EFAULT; if (ucmd.comp_mask || - memchr_inv(&ucmd.reserved, 0, sizeof(ucmd.reserved)) || memchr_inv(&ucmd.burst_info.reserved, 0, sizeof(ucmd.burst_info.reserved))) return -EOPNOTSUPP; + } if (unlikely(ibqp->qp_type == IB_QPT_GSI)) return mlx5_ib_gsi_modify_qp(ibqp, attr, attr_mask); - if (ibqp->qp_type == IB_QPT_DRIVER) - qp_type = qp->qp_sub_type; - else - qp_type = (unlikely(ibqp->qp_type == MLX5_IB_QPT_HW_GSI)) ? - IB_QPT_GSI : ibqp->qp_type; + qp_type = (unlikely(ibqp->qp_type == MLX5_IB_QPT_HW_GSI)) ? IB_QPT_GSI : + qp->type; if (qp_type == MLX5_IB_QPT_DCT) - return mlx5_ib_modify_dct(ibqp, attr, attr_mask, udata); + return mlx5_ib_modify_dct(ibqp, attr, attr_mask, &ucmd, udata); mutex_lock(&qp->mutex); @@ -4007,7 +4265,7 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port; } - if (qp->flags & MLX5_IB_QP_UNDERLAY) { + if (qp->flags & IB_QP_CREATE_SOURCE_QPN) { if (attr_mask & ~(IB_QP_STATE | IB_QP_CUR_STATE)) { mlx5_ib_dbg(dev, "invalid attr_mask 0x%x when underlay QP is used\n", attr_mask); @@ -4067,1439 +4325,19 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, } err = __mlx5_ib_modify_qp(ibqp, attr, attr_mask, cur_state, - new_state, &ucmd, udata); - -out: - mutex_unlock(&qp->mutex); - return err; -} - -static void _handle_post_send_edge(struct mlx5_ib_wq *sq, void **seg, - u32 wqe_sz, void **cur_edge) -{ - u32 idx; - - idx = (sq->cur_post + (wqe_sz >> 2)) & (sq->wqe_cnt - 1); - *cur_edge = get_sq_edge(sq, idx); - - *seg = mlx5_frag_buf_get_wqe(&sq->fbc, idx); -} - -/* handle_post_send_edge - Check if we get to SQ edge. If yes, update to the - * next nearby edge and get new address translation for current WQE position. - * @sq - SQ buffer. - * @seg: Current WQE position (16B aligned). - * @wqe_sz: Total current WQE size [16B]. - * @cur_edge: Updated current edge. - */ -static inline void handle_post_send_edge(struct mlx5_ib_wq *sq, void **seg, - u32 wqe_sz, void **cur_edge) -{ - if (likely(*seg != *cur_edge)) - return; - - _handle_post_send_edge(sq, seg, wqe_sz, cur_edge); -} - -/* memcpy_send_wqe - copy data from src to WQE and update the relevant WQ's - * pointers. At the end @seg is aligned to 16B regardless the copied size. - * @sq - SQ buffer. - * @cur_edge: Updated current edge. - * @seg: Current WQE position (16B aligned). - * @wqe_sz: Total current WQE size [16B]. - * @src: Pointer to copy from. - * @n: Number of bytes to copy. - */ -static inline void memcpy_send_wqe(struct mlx5_ib_wq *sq, void **cur_edge, - void **seg, u32 *wqe_sz, const void *src, - size_t n) -{ - while (likely(n)) { - size_t leftlen = *cur_edge - *seg; - size_t copysz = min_t(size_t, leftlen, n); - size_t stride; - - memcpy(*seg, src, copysz); - - n -= copysz; - src += copysz; - stride = !n ? ALIGN(copysz, 16) : copysz; - *seg += stride; - *wqe_sz += stride >> 4; - handle_post_send_edge(sq, seg, *wqe_sz, cur_edge); - } -} - -static int mlx5_wq_overflow(struct mlx5_ib_wq *wq, int nreq, struct ib_cq *ib_cq) -{ - struct mlx5_ib_cq *cq; - unsigned cur; - - cur = wq->head - wq->tail; - if (likely(cur + nreq < wq->max_post)) - return 0; - - cq = to_mcq(ib_cq); - spin_lock(&cq->lock); - cur = wq->head - wq->tail; - spin_unlock(&cq->lock); - - return cur + nreq >= wq->max_post; -} - -static __always_inline void set_raddr_seg(struct mlx5_wqe_raddr_seg *rseg, - u64 remote_addr, u32 rkey) -{ - rseg->raddr = cpu_to_be64(remote_addr); - rseg->rkey = cpu_to_be32(rkey); - rseg->reserved = 0; -} - -static void set_eth_seg(const struct ib_send_wr *wr, struct mlx5_ib_qp *qp, - void **seg, int *size, void **cur_edge) -{ - struct mlx5_wqe_eth_seg *eseg = *seg; - - memset(eseg, 0, sizeof(struct mlx5_wqe_eth_seg)); - - if (wr->send_flags & IB_SEND_IP_CSUM) - eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM | - MLX5_ETH_WQE_L4_CSUM; - - if (wr->opcode == IB_WR_LSO) { - struct ib_ud_wr *ud_wr = container_of(wr, struct ib_ud_wr, wr); - size_t left, copysz; - void *pdata = ud_wr->header; - size_t stride; - - left = ud_wr->hlen; - eseg->mss = cpu_to_be16(ud_wr->mss); - eseg->inline_hdr.sz = cpu_to_be16(left); - - /* memcpy_send_wqe should get a 16B align address. Hence, we - * first copy up to the current edge and then, if needed, - * fall-through to memcpy_send_wqe. - */ - copysz = min_t(u64, *cur_edge - (void *)eseg->inline_hdr.start, - left); - memcpy(eseg->inline_hdr.start, pdata, copysz); - stride = ALIGN(sizeof(struct mlx5_wqe_eth_seg) - - sizeof(eseg->inline_hdr.start) + copysz, 16); - *size += stride / 16; - *seg += stride; - - if (copysz < left) { - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - left -= copysz; - pdata += copysz; - memcpy_send_wqe(&qp->sq, cur_edge, seg, size, pdata, - left); - } - - return; - } - - *seg += sizeof(struct mlx5_wqe_eth_seg); - *size += sizeof(struct mlx5_wqe_eth_seg) / 16; -} - -static void set_datagram_seg(struct mlx5_wqe_datagram_seg *dseg, - const struct ib_send_wr *wr) -{ - memcpy(&dseg->av, &to_mah(ud_wr(wr)->ah)->av, sizeof(struct mlx5_av)); - dseg->av.dqp_dct = cpu_to_be32(ud_wr(wr)->remote_qpn | MLX5_EXTENDED_UD_AV); - dseg->av.key.qkey.qkey = cpu_to_be32(ud_wr(wr)->remote_qkey); -} - -static void set_data_ptr_seg(struct mlx5_wqe_data_seg *dseg, struct ib_sge *sg) -{ - dseg->byte_count = cpu_to_be32(sg->length); - dseg->lkey = cpu_to_be32(sg->lkey); - dseg->addr = cpu_to_be64(sg->addr); -} - -static u64 get_xlt_octo(u64 bytes) -{ - return ALIGN(bytes, MLX5_IB_UMR_XLT_ALIGNMENT) / - MLX5_IB_UMR_OCTOWORD; -} - -static __be64 frwr_mkey_mask(bool atomic) -{ - u64 result; - - result = MLX5_MKEY_MASK_LEN | - MLX5_MKEY_MASK_PAGE_SIZE | - MLX5_MKEY_MASK_START_ADDR | - MLX5_MKEY_MASK_EN_RINVAL | - MLX5_MKEY_MASK_KEY | - MLX5_MKEY_MASK_LR | - MLX5_MKEY_MASK_LW | - MLX5_MKEY_MASK_RR | - MLX5_MKEY_MASK_RW | - MLX5_MKEY_MASK_SMALL_FENCE | - MLX5_MKEY_MASK_FREE; - - if (atomic) - result |= MLX5_MKEY_MASK_A; - - return cpu_to_be64(result); -} - -static __be64 sig_mkey_mask(void) -{ - u64 result; - - result = MLX5_MKEY_MASK_LEN | - MLX5_MKEY_MASK_PAGE_SIZE | - MLX5_MKEY_MASK_START_ADDR | - MLX5_MKEY_MASK_EN_SIGERR | - MLX5_MKEY_MASK_EN_RINVAL | - MLX5_MKEY_MASK_KEY | - MLX5_MKEY_MASK_LR | - MLX5_MKEY_MASK_LW | - MLX5_MKEY_MASK_RR | - MLX5_MKEY_MASK_RW | - MLX5_MKEY_MASK_SMALL_FENCE | - MLX5_MKEY_MASK_FREE | - MLX5_MKEY_MASK_BSF_EN; - - return cpu_to_be64(result); -} - -static void set_reg_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr, - struct mlx5_ib_mr *mr, u8 flags, bool atomic) -{ - int size = (mr->ndescs + mr->meta_ndescs) * mr->desc_size; - - memset(umr, 0, sizeof(*umr)); - - umr->flags = flags; - umr->xlt_octowords = cpu_to_be16(get_xlt_octo(size)); - umr->mkey_mask = frwr_mkey_mask(atomic); -} - -static void set_linv_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr) -{ - memset(umr, 0, sizeof(*umr)); - umr->mkey_mask = cpu_to_be64(MLX5_MKEY_MASK_FREE); - umr->flags = MLX5_UMR_INLINE; -} - -static __be64 get_umr_enable_mr_mask(void) -{ - u64 result; - - result = MLX5_MKEY_MASK_KEY | - MLX5_MKEY_MASK_FREE; - - return cpu_to_be64(result); -} - -static __be64 get_umr_disable_mr_mask(void) -{ - u64 result; - - result = MLX5_MKEY_MASK_FREE; - - return cpu_to_be64(result); -} - -static __be64 get_umr_update_translation_mask(void) -{ - u64 result; - - result = MLX5_MKEY_MASK_LEN | - MLX5_MKEY_MASK_PAGE_SIZE | - MLX5_MKEY_MASK_START_ADDR; - - return cpu_to_be64(result); -} - -static __be64 get_umr_update_access_mask(int atomic) -{ - u64 result; - - result = MLX5_MKEY_MASK_LR | - MLX5_MKEY_MASK_LW | - MLX5_MKEY_MASK_RR | - MLX5_MKEY_MASK_RW; - - if (atomic) - result |= MLX5_MKEY_MASK_A; - - return cpu_to_be64(result); -} - -static __be64 get_umr_update_pd_mask(void) -{ - u64 result; - - result = MLX5_MKEY_MASK_PD; - - return cpu_to_be64(result); -} - -static int umr_check_mkey_mask(struct mlx5_ib_dev *dev, u64 mask) -{ - if ((mask & MLX5_MKEY_MASK_PAGE_SIZE && - MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled)) || - (mask & MLX5_MKEY_MASK_A && - MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled))) - return -EPERM; - return 0; -} - -static int set_reg_umr_segment(struct mlx5_ib_dev *dev, - struct mlx5_wqe_umr_ctrl_seg *umr, - const struct ib_send_wr *wr, int atomic) -{ - const struct mlx5_umr_wr *umrwr = umr_wr(wr); - - memset(umr, 0, sizeof(*umr)); - - if (!umrwr->ignore_free_state) { - if (wr->send_flags & MLX5_IB_SEND_UMR_FAIL_IF_FREE) - /* fail if free */ - umr->flags = MLX5_UMR_CHECK_FREE; - else - /* fail if not free */ - umr->flags = MLX5_UMR_CHECK_NOT_FREE; - } - - umr->xlt_octowords = cpu_to_be16(get_xlt_octo(umrwr->xlt_size)); - if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_XLT) { - u64 offset = get_xlt_octo(umrwr->offset); - - umr->xlt_offset = cpu_to_be16(offset & 0xffff); - umr->xlt_offset_47_16 = cpu_to_be32(offset >> 16); - umr->flags |= MLX5_UMR_TRANSLATION_OFFSET_EN; - } - if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION) - umr->mkey_mask |= get_umr_update_translation_mask(); - if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_PD_ACCESS) { - umr->mkey_mask |= get_umr_update_access_mask(atomic); - umr->mkey_mask |= get_umr_update_pd_mask(); - } - if (wr->send_flags & MLX5_IB_SEND_UMR_ENABLE_MR) - umr->mkey_mask |= get_umr_enable_mr_mask(); - if (wr->send_flags & MLX5_IB_SEND_UMR_DISABLE_MR) - umr->mkey_mask |= get_umr_disable_mr_mask(); - - if (!wr->num_sge) - umr->flags |= MLX5_UMR_INLINE; - - return umr_check_mkey_mask(dev, be64_to_cpu(umr->mkey_mask)); -} - -static u8 get_umr_flags(int acc) -{ - return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX5_PERM_ATOMIC : 0) | - (acc & IB_ACCESS_REMOTE_WRITE ? MLX5_PERM_REMOTE_WRITE : 0) | - (acc & IB_ACCESS_REMOTE_READ ? MLX5_PERM_REMOTE_READ : 0) | - (acc & IB_ACCESS_LOCAL_WRITE ? MLX5_PERM_LOCAL_WRITE : 0) | - MLX5_PERM_LOCAL_READ | MLX5_PERM_UMR_EN; -} - -static void set_reg_mkey_seg(struct mlx5_mkey_seg *seg, - struct mlx5_ib_mr *mr, - u32 key, int access) -{ - int ndescs = ALIGN(mr->ndescs + mr->meta_ndescs, 8) >> 1; - - memset(seg, 0, sizeof(*seg)); - - if (mr->access_mode == MLX5_MKC_ACCESS_MODE_MTT) - seg->log2_page_size = ilog2(mr->ibmr.page_size); - else if (mr->access_mode == MLX5_MKC_ACCESS_MODE_KLMS) - /* KLMs take twice the size of MTTs */ - ndescs *= 2; - - seg->flags = get_umr_flags(access) | mr->access_mode; - seg->qpn_mkey7_0 = cpu_to_be32((key & 0xff) | 0xffffff00); - seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL); - seg->start_addr = cpu_to_be64(mr->ibmr.iova); - seg->len = cpu_to_be64(mr->ibmr.length); - seg->xlt_oct_size = cpu_to_be32(ndescs); -} - -static void set_linv_mkey_seg(struct mlx5_mkey_seg *seg) -{ - memset(seg, 0, sizeof(*seg)); - seg->status = MLX5_MKEY_STATUS_FREE; -} - -static void set_reg_mkey_segment(struct mlx5_mkey_seg *seg, - const struct ib_send_wr *wr) -{ - const struct mlx5_umr_wr *umrwr = umr_wr(wr); - - memset(seg, 0, sizeof(*seg)); - if (wr->send_flags & MLX5_IB_SEND_UMR_DISABLE_MR) - seg->status = MLX5_MKEY_STATUS_FREE; - - seg->flags = convert_access(umrwr->access_flags); - if (umrwr->pd) - seg->flags_pd = cpu_to_be32(to_mpd(umrwr->pd)->pdn); - if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION && - !umrwr->length) - seg->flags_pd |= cpu_to_be32(MLX5_MKEY_LEN64); - - seg->start_addr = cpu_to_be64(umrwr->virt_addr); - seg->len = cpu_to_be64(umrwr->length); - seg->log2_page_size = umrwr->page_shift; - seg->qpn_mkey7_0 = cpu_to_be32(0xffffff00 | - mlx5_mkey_variant(umrwr->mkey)); -} - -static void set_reg_data_seg(struct mlx5_wqe_data_seg *dseg, - struct mlx5_ib_mr *mr, - struct mlx5_ib_pd *pd) -{ - int bcount = mr->desc_size * (mr->ndescs + mr->meta_ndescs); - - dseg->addr = cpu_to_be64(mr->desc_map); - dseg->byte_count = cpu_to_be32(ALIGN(bcount, 64)); - dseg->lkey = cpu_to_be32(pd->ibpd.local_dma_lkey); -} - -static __be32 send_ieth(const struct ib_send_wr *wr) -{ - switch (wr->opcode) { - case IB_WR_SEND_WITH_IMM: - case IB_WR_RDMA_WRITE_WITH_IMM: - return wr->ex.imm_data; - - case IB_WR_SEND_WITH_INV: - return cpu_to_be32(wr->ex.invalidate_rkey); - - default: - return 0; - } -} - -static u8 calc_sig(void *wqe, int size) -{ - u8 *p = wqe; - u8 res = 0; - int i; - - for (i = 0; i < size; i++) - res ^= p[i]; - - return ~res; -} - -static u8 wq_sig(void *wqe) -{ - return calc_sig(wqe, (*((u8 *)wqe + 8) & 0x3f) << 4); -} - -static int set_data_inl_seg(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, - void **wqe, int *wqe_sz, void **cur_edge) -{ - struct mlx5_wqe_inline_seg *seg; - size_t offset; - int inl = 0; - int i; - - seg = *wqe; - *wqe += sizeof(*seg); - offset = sizeof(*seg); - - for (i = 0; i < wr->num_sge; i++) { - size_t len = wr->sg_list[i].length; - void *addr = (void *)(unsigned long)(wr->sg_list[i].addr); - - inl += len; - - if (unlikely(inl > qp->max_inline_data)) - return -ENOMEM; - - while (likely(len)) { - size_t leftlen; - size_t copysz; - - handle_post_send_edge(&qp->sq, wqe, - *wqe_sz + (offset >> 4), - cur_edge); - - leftlen = *cur_edge - *wqe; - copysz = min_t(size_t, leftlen, len); - - memcpy(*wqe, addr, copysz); - len -= copysz; - addr += copysz; - *wqe += copysz; - offset += copysz; - } - } - - seg->byte_count = cpu_to_be32(inl | MLX5_INLINE_SEG); - - *wqe_sz += ALIGN(inl + sizeof(seg->byte_count), 16) / 16; - - return 0; -} - -static u16 prot_field_size(enum ib_signature_type type) -{ - switch (type) { - case IB_SIG_TYPE_T10_DIF: - return MLX5_DIF_SIZE; - default: - return 0; - } -} - -static u8 bs_selector(int block_size) -{ - switch (block_size) { - case 512: return 0x1; - case 520: return 0x2; - case 4096: return 0x3; - case 4160: return 0x4; - case 1073741824: return 0x5; - default: return 0; - } -} - -static void mlx5_fill_inl_bsf(struct ib_sig_domain *domain, - struct mlx5_bsf_inl *inl) -{ - /* Valid inline section and allow BSF refresh */ - inl->vld_refresh = cpu_to_be16(MLX5_BSF_INL_VALID | - MLX5_BSF_REFRESH_DIF); - inl->dif_apptag = cpu_to_be16(domain->sig.dif.app_tag); - inl->dif_reftag = cpu_to_be32(domain->sig.dif.ref_tag); - /* repeating block */ - inl->rp_inv_seed = MLX5_BSF_REPEAT_BLOCK; - inl->sig_type = domain->sig.dif.bg_type == IB_T10DIF_CRC ? - MLX5_DIF_CRC : MLX5_DIF_IPCS; - - if (domain->sig.dif.ref_remap) - inl->dif_inc_ref_guard_check |= MLX5_BSF_INC_REFTAG; - - if (domain->sig.dif.app_escape) { - if (domain->sig.dif.ref_escape) - inl->dif_inc_ref_guard_check |= MLX5_BSF_APPREF_ESCAPE; - else - inl->dif_inc_ref_guard_check |= MLX5_BSF_APPTAG_ESCAPE; - } - - inl->dif_app_bitmask_check = - cpu_to_be16(domain->sig.dif.apptag_check_mask); -} - -static int mlx5_set_bsf(struct ib_mr *sig_mr, - struct ib_sig_attrs *sig_attrs, - struct mlx5_bsf *bsf, u32 data_size) -{ - struct mlx5_core_sig_ctx *msig = to_mmr(sig_mr)->sig; - struct mlx5_bsf_basic *basic = &bsf->basic; - struct ib_sig_domain *mem = &sig_attrs->mem; - struct ib_sig_domain *wire = &sig_attrs->wire; - - memset(bsf, 0, sizeof(*bsf)); - - /* Basic + Extended + Inline */ - basic->bsf_size_sbs = 1 << 7; - /* Input domain check byte mask */ - basic->check_byte_mask = sig_attrs->check_mask; - basic->raw_data_size = cpu_to_be32(data_size); - - /* Memory domain */ - switch (sig_attrs->mem.sig_type) { - case IB_SIG_TYPE_NONE: - break; - case IB_SIG_TYPE_T10_DIF: - basic->mem.bs_selector = bs_selector(mem->sig.dif.pi_interval); - basic->m_bfs_psv = cpu_to_be32(msig->psv_memory.psv_idx); - mlx5_fill_inl_bsf(mem, &bsf->m_inl); - break; - default: - return -EINVAL; - } - - /* Wire domain */ - switch (sig_attrs->wire.sig_type) { - case IB_SIG_TYPE_NONE: - break; - case IB_SIG_TYPE_T10_DIF: - if (mem->sig.dif.pi_interval == wire->sig.dif.pi_interval && - mem->sig_type == wire->sig_type) { - /* Same block structure */ - basic->bsf_size_sbs |= 1 << 4; - if (mem->sig.dif.bg_type == wire->sig.dif.bg_type) - basic->wire.copy_byte_mask |= MLX5_CPY_GRD_MASK; - if (mem->sig.dif.app_tag == wire->sig.dif.app_tag) - basic->wire.copy_byte_mask |= MLX5_CPY_APP_MASK; - if (mem->sig.dif.ref_tag == wire->sig.dif.ref_tag) - basic->wire.copy_byte_mask |= MLX5_CPY_REF_MASK; - } else - basic->wire.bs_selector = bs_selector(wire->sig.dif.pi_interval); - - basic->w_bfs_psv = cpu_to_be32(msig->psv_wire.psv_idx); - mlx5_fill_inl_bsf(wire, &bsf->w_inl); - break; - default: - return -EINVAL; - } - - return 0; -} - -static int set_sig_data_segment(const struct ib_send_wr *send_wr, - struct ib_mr *sig_mr, - struct ib_sig_attrs *sig_attrs, - struct mlx5_ib_qp *qp, void **seg, int *size, - void **cur_edge) -{ - struct mlx5_bsf *bsf; - u32 data_len; - u32 data_key; - u64 data_va; - u32 prot_len = 0; - u32 prot_key = 0; - u64 prot_va = 0; - bool prot = false; - int ret; - int wqe_size; - struct mlx5_ib_mr *mr = to_mmr(sig_mr); - struct mlx5_ib_mr *pi_mr = mr->pi_mr; - - data_len = pi_mr->data_length; - data_key = pi_mr->ibmr.lkey; - data_va = pi_mr->data_iova; - if (pi_mr->meta_ndescs) { - prot_len = pi_mr->meta_length; - prot_key = pi_mr->ibmr.lkey; - prot_va = pi_mr->pi_iova; - prot = true; - } - - if (!prot || (data_key == prot_key && data_va == prot_va && - data_len == prot_len)) { - /** - * Source domain doesn't contain signature information - * or data and protection are interleaved in memory. - * So need construct: - * ------------------ - * | data_klm | - * ------------------ - * | BSF | - * ------------------ - **/ - struct mlx5_klm *data_klm = *seg; - - data_klm->bcount = cpu_to_be32(data_len); - data_klm->key = cpu_to_be32(data_key); - data_klm->va = cpu_to_be64(data_va); - wqe_size = ALIGN(sizeof(*data_klm), 64); - } else { - /** - * Source domain contains signature information - * So need construct a strided block format: - * --------------------------- - * | stride_block_ctrl | - * --------------------------- - * | data_klm | - * --------------------------- - * | prot_klm | - * --------------------------- - * | BSF | - * --------------------------- - **/ - struct mlx5_stride_block_ctrl_seg *sblock_ctrl; - struct mlx5_stride_block_entry *data_sentry; - struct mlx5_stride_block_entry *prot_sentry; - u16 block_size = sig_attrs->mem.sig.dif.pi_interval; - int prot_size; - - sblock_ctrl = *seg; - data_sentry = (void *)sblock_ctrl + sizeof(*sblock_ctrl); - prot_sentry = (void *)data_sentry + sizeof(*data_sentry); - - prot_size = prot_field_size(sig_attrs->mem.sig_type); - if (!prot_size) { - pr_err("Bad block size given: %u\n", block_size); - return -EINVAL; - } - sblock_ctrl->bcount_per_cycle = cpu_to_be32(block_size + - prot_size); - sblock_ctrl->op = cpu_to_be32(MLX5_STRIDE_BLOCK_OP); - sblock_ctrl->repeat_count = cpu_to_be32(data_len / block_size); - sblock_ctrl->num_entries = cpu_to_be16(2); - - data_sentry->bcount = cpu_to_be16(block_size); - data_sentry->key = cpu_to_be32(data_key); - data_sentry->va = cpu_to_be64(data_va); - data_sentry->stride = cpu_to_be16(block_size); - - prot_sentry->bcount = cpu_to_be16(prot_size); - prot_sentry->key = cpu_to_be32(prot_key); - prot_sentry->va = cpu_to_be64(prot_va); - prot_sentry->stride = cpu_to_be16(prot_size); - - wqe_size = ALIGN(sizeof(*sblock_ctrl) + sizeof(*data_sentry) + - sizeof(*prot_sentry), 64); - } - - *seg += wqe_size; - *size += wqe_size / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - - bsf = *seg; - ret = mlx5_set_bsf(sig_mr, sig_attrs, bsf, data_len); - if (ret) - return -EINVAL; - - *seg += sizeof(*bsf); - *size += sizeof(*bsf) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - - return 0; -} - -static void set_sig_mkey_segment(struct mlx5_mkey_seg *seg, - struct ib_mr *sig_mr, int access_flags, - u32 size, u32 length, u32 pdn) -{ - u32 sig_key = sig_mr->rkey; - u8 sigerr = to_mmr(sig_mr)->sig->sigerr_count & 1; - - memset(seg, 0, sizeof(*seg)); - - seg->flags = get_umr_flags(access_flags) | MLX5_MKC_ACCESS_MODE_KLMS; - seg->qpn_mkey7_0 = cpu_to_be32((sig_key & 0xff) | 0xffffff00); - seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL | sigerr << 26 | - MLX5_MKEY_BSF_EN | pdn); - seg->len = cpu_to_be64(length); - seg->xlt_oct_size = cpu_to_be32(get_xlt_octo(size)); - seg->bsfs_octo_size = cpu_to_be32(MLX5_MKEY_BSF_OCTO_SIZE); -} - -static void set_sig_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr, - u32 size) -{ - memset(umr, 0, sizeof(*umr)); - - umr->flags = MLX5_FLAGS_INLINE | MLX5_FLAGS_CHECK_FREE; - umr->xlt_octowords = cpu_to_be16(get_xlt_octo(size)); - umr->bsf_octowords = cpu_to_be16(MLX5_MKEY_BSF_OCTO_SIZE); - umr->mkey_mask = sig_mkey_mask(); -} - -static int set_pi_umr_wr(const struct ib_send_wr *send_wr, - struct mlx5_ib_qp *qp, void **seg, int *size, - void **cur_edge) -{ - const struct ib_reg_wr *wr = reg_wr(send_wr); - struct mlx5_ib_mr *sig_mr = to_mmr(wr->mr); - struct mlx5_ib_mr *pi_mr = sig_mr->pi_mr; - struct ib_sig_attrs *sig_attrs = sig_mr->ibmr.sig_attrs; - u32 pdn = get_pd(qp)->pdn; - u32 xlt_size; - int region_len, ret; - - if (unlikely(send_wr->num_sge != 0) || - unlikely(wr->access & IB_ACCESS_REMOTE_ATOMIC) || - unlikely(!sig_mr->sig) || unlikely(!qp->ibqp.integrity_en) || - unlikely(!sig_mr->sig->sig_status_checked)) - return -EINVAL; - - /* length of the protected region, data + protection */ - region_len = pi_mr->ibmr.length; - - /** - * KLM octoword size - if protection was provided - * then we use strided block format (3 octowords), - * else we use single KLM (1 octoword) - **/ - if (sig_attrs->mem.sig_type != IB_SIG_TYPE_NONE) - xlt_size = 0x30; - else - xlt_size = sizeof(struct mlx5_klm); - - set_sig_umr_segment(*seg, xlt_size); - *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); - *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - - set_sig_mkey_segment(*seg, wr->mr, wr->access, xlt_size, region_len, - pdn); - *seg += sizeof(struct mlx5_mkey_seg); - *size += sizeof(struct mlx5_mkey_seg) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - - ret = set_sig_data_segment(send_wr, wr->mr, sig_attrs, qp, seg, size, - cur_edge); - if (ret) - return ret; - - sig_mr->sig->sig_status_checked = false; - return 0; -} - -static int set_psv_wr(struct ib_sig_domain *domain, - u32 psv_idx, void **seg, int *size) -{ - struct mlx5_seg_set_psv *psv_seg = *seg; - - memset(psv_seg, 0, sizeof(*psv_seg)); - psv_seg->psv_num = cpu_to_be32(psv_idx); - switch (domain->sig_type) { - case IB_SIG_TYPE_NONE: - break; - case IB_SIG_TYPE_T10_DIF: - psv_seg->transient_sig = cpu_to_be32(domain->sig.dif.bg << 16 | - domain->sig.dif.app_tag); - psv_seg->ref_tag = cpu_to_be32(domain->sig.dif.ref_tag); - break; - default: - pr_err("Bad signature type (%d) is given.\n", - domain->sig_type); - return -EINVAL; - } - - *seg += sizeof(*psv_seg); - *size += sizeof(*psv_seg) / 16; - - return 0; -} - -static int set_reg_wr(struct mlx5_ib_qp *qp, - const struct ib_reg_wr *wr, - void **seg, int *size, void **cur_edge, - bool check_not_free) -{ - struct mlx5_ib_mr *mr = to_mmr(wr->mr); - struct mlx5_ib_pd *pd = to_mpd(qp->ibqp.pd); - struct mlx5_ib_dev *dev = to_mdev(pd->ibpd.device); - int mr_list_size = (mr->ndescs + mr->meta_ndescs) * mr->desc_size; - bool umr_inline = mr_list_size <= MLX5_IB_SQ_UMR_INLINE_THRESHOLD; - bool atomic = wr->access & IB_ACCESS_REMOTE_ATOMIC; - u8 flags = 0; - - if (!mlx5_ib_can_use_umr(dev, atomic, wr->access)) { - mlx5_ib_warn(to_mdev(qp->ibqp.device), - "Fast update of %s for MR is disabled\n", - (MLX5_CAP_GEN(dev->mdev, - umr_modify_entity_size_disabled)) ? - "entity size" : - "atomic access"); - return -EINVAL; - } - - if (unlikely(wr->wr.send_flags & IB_SEND_INLINE)) { - mlx5_ib_warn(to_mdev(qp->ibqp.device), - "Invalid IB_SEND_INLINE send flag\n"); - return -EINVAL; - } - - if (check_not_free) - flags |= MLX5_UMR_CHECK_NOT_FREE; - if (umr_inline) - flags |= MLX5_UMR_INLINE; - - set_reg_umr_seg(*seg, mr, flags, atomic); - *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); - *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - - set_reg_mkey_seg(*seg, mr, wr->key, wr->access); - *seg += sizeof(struct mlx5_mkey_seg); - *size += sizeof(struct mlx5_mkey_seg) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - - if (umr_inline) { - memcpy_send_wqe(&qp->sq, cur_edge, seg, size, mr->descs, - mr_list_size); - *size = ALIGN(*size, MLX5_SEND_WQE_BB >> 4); - } else { - set_reg_data_seg(*seg, mr, pd); - *seg += sizeof(struct mlx5_wqe_data_seg); - *size += (sizeof(struct mlx5_wqe_data_seg) / 16); - } - return 0; -} - -static void set_linv_wr(struct mlx5_ib_qp *qp, void **seg, int *size, - void **cur_edge) -{ - set_linv_umr_seg(*seg); - *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); - *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - set_linv_mkey_seg(*seg); - *seg += sizeof(struct mlx5_mkey_seg); - *size += sizeof(struct mlx5_mkey_seg) / 16; - handle_post_send_edge(&qp->sq, seg, *size, cur_edge); -} - -static void dump_wqe(struct mlx5_ib_qp *qp, u32 idx, int size_16) -{ - __be32 *p = NULL; - int i, j; - - pr_debug("dump WQE index %u:\n", idx); - for (i = 0, j = 0; i < size_16 * 4; i += 4, j += 4) { - if ((i & 0xf) == 0) { - p = mlx5_frag_buf_get_wqe(&qp->sq.fbc, idx); - pr_debug("WQBB at %p:\n", (void *)p); - j = 0; - idx = (idx + 1) & (qp->sq.wqe_cnt - 1); - } - pr_debug("%08x %08x %08x %08x\n", be32_to_cpu(p[j]), - be32_to_cpu(p[j + 1]), be32_to_cpu(p[j + 2]), - be32_to_cpu(p[j + 3])); - } -} - -static int __begin_wqe(struct mlx5_ib_qp *qp, void **seg, - struct mlx5_wqe_ctrl_seg **ctrl, - const struct ib_send_wr *wr, unsigned int *idx, - int *size, void **cur_edge, int nreq, - bool send_signaled, bool solicited) -{ - if (unlikely(mlx5_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq))) - return -ENOMEM; - - *idx = qp->sq.cur_post & (qp->sq.wqe_cnt - 1); - *seg = mlx5_frag_buf_get_wqe(&qp->sq.fbc, *idx); - *ctrl = *seg; - *(uint32_t *)(*seg + 8) = 0; - (*ctrl)->imm = send_ieth(wr); - (*ctrl)->fm_ce_se = qp->sq_signal_bits | - (send_signaled ? MLX5_WQE_CTRL_CQ_UPDATE : 0) | - (solicited ? MLX5_WQE_CTRL_SOLICITED : 0); - - *seg += sizeof(**ctrl); - *size = sizeof(**ctrl) / 16; - *cur_edge = qp->sq.cur_edge; - - return 0; -} - -static int begin_wqe(struct mlx5_ib_qp *qp, void **seg, - struct mlx5_wqe_ctrl_seg **ctrl, - const struct ib_send_wr *wr, unsigned *idx, - int *size, void **cur_edge, int nreq) -{ - return __begin_wqe(qp, seg, ctrl, wr, idx, size, cur_edge, nreq, - wr->send_flags & IB_SEND_SIGNALED, - wr->send_flags & IB_SEND_SOLICITED); -} - -static void finish_wqe(struct mlx5_ib_qp *qp, - struct mlx5_wqe_ctrl_seg *ctrl, - void *seg, u8 size, void *cur_edge, - unsigned int idx, u64 wr_id, int nreq, u8 fence, - u32 mlx5_opcode) -{ - u8 opmod = 0; - - ctrl->opmod_idx_opcode = cpu_to_be32(((u32)(qp->sq.cur_post) << 8) | - mlx5_opcode | ((u32)opmod << 24)); - ctrl->qpn_ds = cpu_to_be32(size | (qp->trans_qp.base.mqp.qpn << 8)); - ctrl->fm_ce_se |= fence; - if (unlikely(qp->wq_sig)) - ctrl->signature = wq_sig(ctrl); - - qp->sq.wrid[idx] = wr_id; - qp->sq.w_list[idx].opcode = mlx5_opcode; - qp->sq.wqe_head[idx] = qp->sq.head + nreq; - qp->sq.cur_post += DIV_ROUND_UP(size * 16, MLX5_SEND_WQE_BB); - qp->sq.w_list[idx].next = qp->sq.cur_post; - - /* We save the edge which was possibly updated during the WQE - * construction, into SQ's cache. - */ - seg = PTR_ALIGN(seg, MLX5_SEND_WQE_BB); - qp->sq.cur_edge = (unlikely(seg == cur_edge)) ? - get_sq_edge(&qp->sq, qp->sq.cur_post & - (qp->sq.wqe_cnt - 1)) : - cur_edge; -} - -static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, - const struct ib_send_wr **bad_wr, bool drain) -{ - 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 ib_reg_wr reg_pi_wr; - struct mlx5_ib_qp *qp; - struct mlx5_ib_mr *mr; - struct mlx5_ib_mr *pi_mr; - struct mlx5_ib_mr pa_pi_mr; - struct ib_sig_attrs *sig_attrs; - struct mlx5_wqe_xrc_seg *xrc; - struct mlx5_bf *bf; - void *cur_edge; - int uninitialized_var(size); - unsigned long flags; - unsigned idx; - int err = 0; - int num_sge; - void *seg; - int nreq; - int i; - u8 next_fence = 0; - u8 fence; - - if (unlikely(mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR && - !drain)) { - *bad_wr = wr; - return -EIO; - } - - if (unlikely(ibqp->qp_type == IB_QPT_GSI)) - return mlx5_ib_gsi_post_send(ibqp, wr, bad_wr); - - qp = to_mqp(ibqp); - bf = &qp->bf; - - spin_lock_irqsave(&qp->sq.lock, flags); - - for (nreq = 0; wr; nreq++, wr = wr->next) { - if (unlikely(wr->opcode >= ARRAY_SIZE(mlx5_ib_opcode))) { - mlx5_ib_warn(dev, "\n"); - err = -EINVAL; - *bad_wr = wr; - goto out; - } - - num_sge = wr->num_sge; - if (unlikely(num_sge > qp->sq.max_gs)) { - mlx5_ib_warn(dev, "\n"); - err = -EINVAL; - *bad_wr = wr; - goto out; - } + new_state, &ucmd, &resp, udata); - err = begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, &cur_edge, - nreq); - if (err) { - mlx5_ib_warn(dev, "\n"); - err = -ENOMEM; - *bad_wr = wr; - goto out; - } - - if (wr->opcode == IB_WR_REG_MR || - wr->opcode == IB_WR_REG_MR_INTEGRITY) { - fence = dev->umr_fence; - next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; - } else { - if (wr->send_flags & IB_SEND_FENCE) { - if (qp->next_fence) - fence = MLX5_FENCE_MODE_SMALL_AND_FENCE; - else - fence = MLX5_FENCE_MODE_FENCE; - } else { - fence = qp->next_fence; - } - } - - switch (ibqp->qp_type) { - case IB_QPT_XRC_INI: - xrc = seg; - seg += sizeof(*xrc); - size += sizeof(*xrc) / 16; - /* fall through */ - case IB_QPT_RC: - switch (wr->opcode) { - case IB_WR_RDMA_READ: - case IB_WR_RDMA_WRITE: - case IB_WR_RDMA_WRITE_WITH_IMM: - set_raddr_seg(seg, rdma_wr(wr)->remote_addr, - rdma_wr(wr)->rkey); - seg += sizeof(struct mlx5_wqe_raddr_seg); - size += sizeof(struct mlx5_wqe_raddr_seg) / 16; - break; - - case IB_WR_ATOMIC_CMP_AND_SWP: - case IB_WR_ATOMIC_FETCH_AND_ADD: - case IB_WR_MASKED_ATOMIC_CMP_AND_SWP: - mlx5_ib_warn(dev, "Atomic operations are not supported yet\n"); - err = -ENOSYS; - *bad_wr = wr; - goto out; - - case IB_WR_LOCAL_INV: - qp->sq.wr_data[idx] = IB_WR_LOCAL_INV; - ctrl->imm = cpu_to_be32(wr->ex.invalidate_rkey); - set_linv_wr(qp, &seg, &size, &cur_edge); - num_sge = 0; - break; - - case IB_WR_REG_MR: - qp->sq.wr_data[idx] = IB_WR_REG_MR; - ctrl->imm = cpu_to_be32(reg_wr(wr)->key); - err = set_reg_wr(qp, reg_wr(wr), &seg, &size, - &cur_edge, true); - if (err) { - *bad_wr = wr; - goto out; - } - num_sge = 0; - break; - - case IB_WR_REG_MR_INTEGRITY: - qp->sq.wr_data[idx] = IB_WR_REG_MR_INTEGRITY; - - mr = to_mmr(reg_wr(wr)->mr); - pi_mr = mr->pi_mr; - - if (pi_mr) { - memset(®_pi_wr, 0, - sizeof(struct ib_reg_wr)); - - reg_pi_wr.mr = &pi_mr->ibmr; - reg_pi_wr.access = reg_wr(wr)->access; - reg_pi_wr.key = pi_mr->ibmr.rkey; - - ctrl->imm = cpu_to_be32(reg_pi_wr.key); - /* UMR for data + prot registration */ - err = set_reg_wr(qp, ®_pi_wr, &seg, - &size, &cur_edge, - false); - if (err) { - *bad_wr = wr; - goto out; - } - finish_wqe(qp, ctrl, seg, size, - cur_edge, idx, wr->wr_id, - nreq, fence, - MLX5_OPCODE_UMR); - - err = begin_wqe(qp, &seg, &ctrl, wr, - &idx, &size, &cur_edge, - nreq); - if (err) { - mlx5_ib_warn(dev, "\n"); - err = -ENOMEM; - *bad_wr = wr; - goto out; - } - } else { - memset(&pa_pi_mr, 0, - sizeof(struct mlx5_ib_mr)); - /* No UMR, use local_dma_lkey */ - pa_pi_mr.ibmr.lkey = - mr->ibmr.pd->local_dma_lkey; - - pa_pi_mr.ndescs = mr->ndescs; - pa_pi_mr.data_length = mr->data_length; - pa_pi_mr.data_iova = mr->data_iova; - if (mr->meta_ndescs) { - pa_pi_mr.meta_ndescs = - mr->meta_ndescs; - pa_pi_mr.meta_length = - mr->meta_length; - pa_pi_mr.pi_iova = mr->pi_iova; - } - - pa_pi_mr.ibmr.length = mr->ibmr.length; - mr->pi_mr = &pa_pi_mr; - } - ctrl->imm = cpu_to_be32(mr->ibmr.rkey); - /* UMR for sig MR */ - err = set_pi_umr_wr(wr, qp, &seg, &size, - &cur_edge); - if (err) { - mlx5_ib_warn(dev, "\n"); - *bad_wr = wr; - goto out; - } - finish_wqe(qp, ctrl, seg, size, cur_edge, idx, - wr->wr_id, nreq, fence, - MLX5_OPCODE_UMR); - - /* - * SET_PSV WQEs are not signaled and solicited - * on error - */ - sig_attrs = mr->ibmr.sig_attrs; - err = __begin_wqe(qp, &seg, &ctrl, wr, &idx, - &size, &cur_edge, nreq, false, - true); - if (err) { - mlx5_ib_warn(dev, "\n"); - err = -ENOMEM; - *bad_wr = wr; - goto out; - } - err = set_psv_wr(&sig_attrs->mem, - mr->sig->psv_memory.psv_idx, - &seg, &size); - if (err) { - mlx5_ib_warn(dev, "\n"); - *bad_wr = wr; - goto out; - } - finish_wqe(qp, ctrl, seg, size, cur_edge, idx, - wr->wr_id, nreq, next_fence, - MLX5_OPCODE_SET_PSV); - - err = __begin_wqe(qp, &seg, &ctrl, wr, &idx, - &size, &cur_edge, nreq, false, - true); - if (err) { - mlx5_ib_warn(dev, "\n"); - err = -ENOMEM; - *bad_wr = wr; - goto out; - } - err = set_psv_wr(&sig_attrs->wire, - mr->sig->psv_wire.psv_idx, - &seg, &size); - if (err) { - mlx5_ib_warn(dev, "\n"); - *bad_wr = wr; - goto out; - } - finish_wqe(qp, ctrl, seg, size, cur_edge, idx, - wr->wr_id, nreq, next_fence, - MLX5_OPCODE_SET_PSV); - - qp->next_fence = - MLX5_FENCE_MODE_INITIATOR_SMALL; - num_sge = 0; - goto skip_psv; - - default: - break; - } - break; - - case IB_QPT_UC: - switch (wr->opcode) { - case IB_WR_RDMA_WRITE: - case IB_WR_RDMA_WRITE_WITH_IMM: - set_raddr_seg(seg, rdma_wr(wr)->remote_addr, - rdma_wr(wr)->rkey); - seg += sizeof(struct mlx5_wqe_raddr_seg); - size += sizeof(struct mlx5_wqe_raddr_seg) / 16; - break; - - default: - break; - } - break; - - case IB_QPT_SMI: - if (unlikely(!mdev->port_caps[qp->port - 1].has_smi)) { - mlx5_ib_warn(dev, "Send SMP MADs is not allowed\n"); - err = -EPERM; - *bad_wr = wr; - goto out; - } - /* fall through */ - case MLX5_IB_QPT_HW_GSI: - set_datagram_seg(seg, wr); - seg += sizeof(struct mlx5_wqe_datagram_seg); - size += sizeof(struct mlx5_wqe_datagram_seg) / 16; - handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); - - break; - case IB_QPT_UD: - set_datagram_seg(seg, wr); - seg += sizeof(struct mlx5_wqe_datagram_seg); - size += sizeof(struct mlx5_wqe_datagram_seg) / 16; - handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); - - /* handle qp that supports ud offload */ - if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO) { - struct mlx5_wqe_eth_pad *pad; - - pad = seg; - memset(pad, 0, sizeof(struct mlx5_wqe_eth_pad)); - seg += sizeof(struct mlx5_wqe_eth_pad); - size += sizeof(struct mlx5_wqe_eth_pad) / 16; - set_eth_seg(wr, qp, &seg, &size, &cur_edge); - handle_post_send_edge(&qp->sq, &seg, size, - &cur_edge); - } - break; - case MLX5_IB_QPT_REG_UMR: - if (wr->opcode != MLX5_IB_WR_UMR) { - err = -EINVAL; - mlx5_ib_warn(dev, "bad opcode\n"); - goto out; - } - qp->sq.wr_data[idx] = MLX5_IB_WR_UMR; - ctrl->imm = cpu_to_be32(umr_wr(wr)->mkey); - err = set_reg_umr_segment(dev, seg, wr, !!(MLX5_CAP_GEN(mdev, atomic))); - if (unlikely(err)) - goto out; - seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); - size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); - set_reg_mkey_segment(seg, wr); - seg += sizeof(struct mlx5_mkey_seg); - size += sizeof(struct mlx5_mkey_seg) / 16; - handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); - break; - - default: - break; - } - - if (wr->send_flags & IB_SEND_INLINE && num_sge) { - err = set_data_inl_seg(qp, wr, &seg, &size, &cur_edge); - if (unlikely(err)) { - mlx5_ib_warn(dev, "\n"); - *bad_wr = wr; - goto out; - } - } else { - for (i = 0; i < num_sge; i++) { - handle_post_send_edge(&qp->sq, &seg, size, - &cur_edge); - if (likely(wr->sg_list[i].length)) { - set_data_ptr_seg - ((struct mlx5_wqe_data_seg *)seg, - wr->sg_list + i); - size += sizeof(struct mlx5_wqe_data_seg) / 16; - seg += sizeof(struct mlx5_wqe_data_seg); - } - } - } - - qp->next_fence = next_fence; - finish_wqe(qp, ctrl, seg, size, cur_edge, idx, wr->wr_id, nreq, - fence, mlx5_ib_opcode[wr->opcode]); -skip_psv: - if (0) - dump_wqe(qp, idx, size); - } - -out: - if (likely(nreq)) { - qp->sq.head += nreq; - - /* Make sure that descriptors are written before - * updating doorbell record and ringing the doorbell - */ - wmb(); - - qp->db.db[MLX5_SND_DBR] = cpu_to_be32(qp->sq.cur_post); - - /* Make sure doorbell record is visible to the HCA before - * we hit doorbell */ - wmb(); - - mlx5_write64((__be32 *)ctrl, bf->bfreg->map + bf->offset); - /* Make sure doorbells don't leak out of SQ spinlock - * and reach the HCA out of order. - */ - bf->offset ^= bf->buf_size; - } - - spin_unlock_irqrestore(&qp->sq.lock, flags); - - return err; -} - -int mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, - const struct ib_send_wr **bad_wr) -{ - return _mlx5_ib_post_send(ibqp, wr, bad_wr, false); -} - -static void set_sig_seg(struct mlx5_rwqe_sig *sig, int size) -{ - sig->signature = calc_sig(sig, size); -} - -static int _mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, - const struct ib_recv_wr **bad_wr, bool drain) -{ - 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; - int ind; - int i; - - if (unlikely(mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR && - !drain)) { - *bad_wr = wr; - return -EIO; - } - - if (unlikely(ibqp->qp_type == IB_QPT_GSI)) - return mlx5_ib_gsi_post_recv(ibqp, wr, bad_wr); - - spin_lock_irqsave(&qp->rq.lock, flags); - - ind = qp->rq.head & (qp->rq.wqe_cnt - 1); - - for (nreq = 0; wr; nreq++, wr = wr->next) { - if (mlx5_wq_overflow(&qp->rq, nreq, qp->ibqp.recv_cq)) { - err = -ENOMEM; - *bad_wr = wr; - goto out; - } - - if (unlikely(wr->num_sge > qp->rq.max_gs)) { - err = -EINVAL; - *bad_wr = wr; - goto out; - } - - scat = mlx5_frag_buf_get_wqe(&qp->rq.fbc, ind); - if (qp->wq_sig) - scat++; - - for (i = 0; i < wr->num_sge; i++) - set_data_ptr_seg(scat + i, wr->sg_list + i); - - if (i < qp->rq.max_gs) { - scat[i].byte_count = 0; - scat[i].lkey = cpu_to_be32(MLX5_INVALID_LKEY); - scat[i].addr = 0; - } - - if (qp->wq_sig) { - sig = (struct mlx5_rwqe_sig *)scat; - set_sig_seg(sig, (qp->rq.max_gs + 1) << 2); - } - - qp->rq.wrid[ind] = wr->wr_id; - - ind = (ind + 1) & (qp->rq.wqe_cnt - 1); - } + /* resp.response_length is set in ECE supported flows only */ + if (!err && resp.response_length && + udata->outlen >= resp.response_length) + /* Return -EFAULT to the user and expect him to destroy QP. */ + err = ib_copy_to_udata(udata, &resp, resp.response_length); out: - if (likely(nreq)) { - qp->rq.head += nreq; - - /* Make sure that descriptors are written before - * doorbell record. - */ - wmb(); - - *qp->db.db = cpu_to_be32(qp->rq.head & 0xffff); - } - - spin_unlock_irqrestore(&qp->rq.lock, flags); - + mutex_unlock(&qp->mutex); return err; } -int mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, - const struct ib_recv_wr **bad_wr) -{ - return _mlx5_ib_post_recv(ibqp, wr, bad_wr, false); -} - static inline enum ib_qp_state to_ib_qp_state(enum mlx5_qp_state mlx5_state) { switch (mlx5_state) { @@ -5525,50 +4363,34 @@ static inline enum ib_mig_state to_ib_mig_state(int mlx5_mig_state) } } -static int to_ib_qp_access_flags(int mlx5_flags) -{ - int ib_flags = 0; - - if (mlx5_flags & MLX5_QP_BIT_RRE) - ib_flags |= IB_ACCESS_REMOTE_READ; - if (mlx5_flags & MLX5_QP_BIT_RWE) - ib_flags |= IB_ACCESS_REMOTE_WRITE; - if (mlx5_flags & MLX5_QP_BIT_RAE) - ib_flags |= IB_ACCESS_REMOTE_ATOMIC; - - return ib_flags; -} - static void to_rdma_ah_attr(struct mlx5_ib_dev *ibdev, - struct rdma_ah_attr *ah_attr, - struct mlx5_qp_path *path) + struct rdma_ah_attr *ah_attr, void *path) { + int port = MLX5_GET(ads, path, vhca_port_num); + int static_rate; memset(ah_attr, 0, sizeof(*ah_attr)); - if (!path->port || path->port > ibdev->num_ports) + if (!port || port > ibdev->num_ports) return; - ah_attr->type = rdma_ah_find_type(&ibdev->ib_dev, path->port); + ah_attr->type = rdma_ah_find_type(&ibdev->ib_dev, port); - rdma_ah_set_port_num(ah_attr, path->port); - rdma_ah_set_sl(ah_attr, path->dci_cfi_prio_sl & 0xf); + rdma_ah_set_port_num(ah_attr, port); + rdma_ah_set_sl(ah_attr, MLX5_GET(ads, path, sl)); - rdma_ah_set_dlid(ah_attr, be16_to_cpu(path->rlid)); - rdma_ah_set_path_bits(ah_attr, path->grh_mlid & 0x7f); - rdma_ah_set_static_rate(ah_attr, - path->static_rate ? path->static_rate - 5 : 0); + rdma_ah_set_dlid(ah_attr, MLX5_GET(ads, path, rlid)); + rdma_ah_set_path_bits(ah_attr, MLX5_GET(ads, path, mlid)); - if (path->grh_mlid & (1 << 7) || + static_rate = MLX5_GET(ads, path, stat_rate); + rdma_ah_set_static_rate(ah_attr, static_rate ? static_rate - 5 : 0); + if (MLX5_GET(ads, path, grh) || ah_attr->type == RDMA_AH_ATTR_TYPE_ROCE) { - u32 tc_fl = be32_to_cpu(path->tclass_flowlabel); - - rdma_ah_set_grh(ah_attr, NULL, - tc_fl & 0xfffff, - path->mgid_index, - path->hop_limit, - (tc_fl >> 20) & 0xff); - rdma_ah_set_dgid_raw(ah_attr, path->rgid); + rdma_ah_set_grh(ah_attr, NULL, MLX5_GET(ads, path, flow_label), + MLX5_GET(ads, path, src_addr_index), + MLX5_GET(ads, path, hop_limit), + MLX5_GET(ads, path, tclass)); + rdma_ah_set_dgid_raw(ah_attr, MLX5_ADDR_OF(ads, path, rgid_rip)); } } @@ -5625,7 +4447,7 @@ static int sqrq_state_to_qp_state(u8 sq_state, u8 rq_state, [MLX5_SQ_STATE_NA] = IB_QPS_RESET, }, [MLX5_RQC_STATE_RDY] = { - [MLX5_SQC_STATE_RST] = MLX5_QP_STATE_BAD, + [MLX5_SQC_STATE_RST] = MLX5_QP_STATE, [MLX5_SQC_STATE_RDY] = MLX5_QP_STATE, [MLX5_SQC_STATE_ERR] = IB_QPS_SQE, [MLX5_SQ_STATE_NA] = MLX5_QP_STATE, @@ -5637,7 +4459,7 @@ static int sqrq_state_to_qp_state(u8 sq_state, u8 rq_state, [MLX5_SQ_STATE_NA] = IB_QPS_ERR, }, [MLX5_RQ_STATE_NA] = { - [MLX5_SQC_STATE_RST] = IB_QPS_RESET, + [MLX5_SQC_STATE_RST] = MLX5_QP_STATE, [MLX5_SQC_STATE_RDY] = MLX5_QP_STATE, [MLX5_SQC_STATE_ERR] = MLX5_QP_STATE, [MLX5_SQ_STATE_NA] = MLX5_QP_STATE_BAD, @@ -5690,61 +4512,58 @@ static int query_qp_attr(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct ib_qp_attr *qp_attr) { int outlen = MLX5_ST_SZ_BYTES(query_qp_out); - struct mlx5_qp_context *context; - int mlx5_state; + void *qpc, *pri_path, *alt_path; u32 *outb; - int err = 0; + int err; outb = kzalloc(outlen, GFP_KERNEL); if (!outb) return -ENOMEM; - err = mlx5_core_qp_query(dev->mdev, &qp->trans_qp.base.mqp, outb, - outlen); + err = mlx5_core_qp_query(dev, &qp->trans_qp.base.mqp, outb, outlen); if (err) goto out; - /* FIXME: use MLX5_GET rather than mlx5_qp_context manual struct */ - context = (struct mlx5_qp_context *)MLX5_ADDR_OF(query_qp_out, outb, qpc); + qpc = MLX5_ADDR_OF(query_qp_out, outb, qpc); - mlx5_state = be32_to_cpu(context->flags) >> 28; + qp->state = to_ib_qp_state(MLX5_GET(qpc, qpc, state)); + if (MLX5_GET(qpc, qpc, state) == MLX5_QP_STATE_SQ_DRAINING) + qp_attr->sq_draining = 1; - qp->state = to_ib_qp_state(mlx5_state); - qp_attr->path_mtu = context->mtu_msgmax >> 5; - qp_attr->path_mig_state = - to_ib_mig_state((be32_to_cpu(context->flags) >> 11) & 0x3); - qp_attr->qkey = be32_to_cpu(context->qkey); - qp_attr->rq_psn = be32_to_cpu(context->rnr_nextrecvpsn) & 0xffffff; - qp_attr->sq_psn = be32_to_cpu(context->next_send_psn) & 0xffffff; - qp_attr->dest_qp_num = be32_to_cpu(context->log_pg_sz_remote_qpn) & 0xffffff; - qp_attr->qp_access_flags = - to_ib_qp_access_flags(be32_to_cpu(context->params2)); + qp_attr->path_mtu = MLX5_GET(qpc, qpc, mtu); + qp_attr->path_mig_state = to_ib_mig_state(MLX5_GET(qpc, qpc, pm_state)); + qp_attr->qkey = MLX5_GET(qpc, qpc, q_key); + qp_attr->rq_psn = MLX5_GET(qpc, qpc, next_rcv_psn); + qp_attr->sq_psn = MLX5_GET(qpc, qpc, next_send_psn); + qp_attr->dest_qp_num = MLX5_GET(qpc, qpc, remote_qpn); - if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC) { - to_rdma_ah_attr(dev, &qp_attr->ah_attr, &context->pri_path); - to_rdma_ah_attr(dev, &qp_attr->alt_ah_attr, &context->alt_path); - qp_attr->alt_pkey_index = - be16_to_cpu(context->alt_path.pkey_index); - qp_attr->alt_port_num = - rdma_ah_get_port_num(&qp_attr->alt_ah_attr); - } + if (MLX5_GET(qpc, qpc, rre)) + qp_attr->qp_access_flags |= IB_ACCESS_REMOTE_READ; + if (MLX5_GET(qpc, qpc, rwe)) + qp_attr->qp_access_flags |= IB_ACCESS_REMOTE_WRITE; + if (MLX5_GET(qpc, qpc, rae)) + qp_attr->qp_access_flags |= IB_ACCESS_REMOTE_ATOMIC; - qp_attr->pkey_index = be16_to_cpu(context->pri_path.pkey_index); - qp_attr->port_num = context->pri_path.port; + qp_attr->max_rd_atomic = 1 << MLX5_GET(qpc, qpc, log_sra_max); + qp_attr->max_dest_rd_atomic = 1 << MLX5_GET(qpc, qpc, log_rra_max); + qp_attr->min_rnr_timer = MLX5_GET(qpc, qpc, min_rnr_nak); + qp_attr->retry_cnt = MLX5_GET(qpc, qpc, retry_count); + qp_attr->rnr_retry = MLX5_GET(qpc, qpc, rnr_retry); - /* qp_attr->en_sqd_async_notify is only applicable in modify qp */ - qp_attr->sq_draining = mlx5_state == MLX5_QP_STATE_SQ_DRAINING; + pri_path = MLX5_ADDR_OF(qpc, qpc, primary_address_path); + alt_path = MLX5_ADDR_OF(qpc, qpc, secondary_address_path); - qp_attr->max_rd_atomic = 1 << ((be32_to_cpu(context->params1) >> 21) & 0x7); + if (qp->ibqp.qp_type == IB_QPT_RC || qp->ibqp.qp_type == IB_QPT_UC) { + to_rdma_ah_attr(dev, &qp_attr->ah_attr, pri_path); + to_rdma_ah_attr(dev, &qp_attr->alt_ah_attr, alt_path); + qp_attr->alt_pkey_index = MLX5_GET(ads, alt_path, pkey_index); + qp_attr->alt_port_num = MLX5_GET(ads, alt_path, vhca_port_num); + } - qp_attr->max_dest_rd_atomic = - 1 << ((be32_to_cpu(context->params2) >> 21) & 0x7); - qp_attr->min_rnr_timer = - (be32_to_cpu(context->rnr_nextrecvpsn) >> 24) & 0x1f; - qp_attr->timeout = context->pri_path.ackto_lt >> 3; - qp_attr->retry_cnt = (be32_to_cpu(context->params1) >> 16) & 0x7; - qp_attr->rnr_retry = (be32_to_cpu(context->params1) >> 13) & 0x7; - qp_attr->alt_timeout = context->alt_path.ackto_lt >> 3; + qp_attr->pkey_index = MLX5_GET(ads, pri_path, pkey_index); + qp_attr->port_num = MLX5_GET(ads, pri_path, vhca_port_num); + qp_attr->timeout = MLX5_GET(ads, pri_path, ack_timeout); + qp_attr->alt_timeout = MLX5_GET(ads, alt_path, ack_timeout); out: kfree(outb); @@ -5778,7 +4597,7 @@ static int mlx5_ib_dct_query_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *mqp, if (!out) return -ENOMEM; - err = mlx5_core_dct_query(dev->mdev, dct, out, outlen); + err = mlx5_core_dct_query(dev, dct, out, outlen); if (err) goto out; @@ -5835,14 +4654,14 @@ int mlx5_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, memset(qp_init_attr, 0, sizeof(*qp_init_attr)); memset(qp_attr, 0, sizeof(*qp_attr)); - if (unlikely(qp->qp_sub_type == MLX5_IB_QPT_DCT)) + if (unlikely(qp->type == MLX5_IB_QPT_DCT)) return mlx5_ib_dct_query_qp(dev, qp, qp_attr, qp_attr_mask, qp_init_attr); mutex_lock(&qp->mutex); if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET || - qp->flags & MLX5_IB_QP_UNDERLAY) { + qp->flags & IB_QP_CREATE_SOURCE_QPN) { err = query_raw_packet_qp_state(dev, qp, &raw_packet_qp_state); if (err) goto out; @@ -5876,18 +4695,7 @@ int mlx5_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, qp_init_attr->cap = qp_attr->cap; - qp_init_attr->create_flags = 0; - if (qp->flags & MLX5_IB_QP_BLOCK_MULTICAST_LOOPBACK) - qp_init_attr->create_flags |= IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK; - - if (qp->flags & MLX5_IB_QP_CROSS_CHANNEL) - qp_init_attr->create_flags |= IB_QP_CREATE_CROSS_CHANNEL; - if (qp->flags & MLX5_IB_QP_MANAGED_SEND) - qp_init_attr->create_flags |= IB_QP_CREATE_MANAGED_SEND; - if (qp->flags & MLX5_IB_QP_MANAGED_RECV) - qp_init_attr->create_flags |= IB_QP_CREATE_MANAGED_RECV; - if (qp->flags & MLX5_IB_QP_SQPN_QP1) - qp_init_attr->create_flags |= MLX5_IB_QP_CREATE_SQPN_QP1; + qp_init_attr->create_flags = qp->flags; qp_init_attr->sq_sig_type = qp->sq_signal_bits & MLX5_WQE_CTRL_CQ_UPDATE ? IB_SIGNAL_ALL_WR : IB_SIGNAL_REQ_WR; @@ -5897,41 +4705,23 @@ out: return err; } -struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, - struct ib_udata *udata) +int mlx5_ib_alloc_xrcd(struct ib_xrcd *ibxrcd, struct ib_udata *udata) { - struct mlx5_ib_dev *dev = to_mdev(ibdev); - struct mlx5_ib_xrcd *xrcd; - int err; + struct mlx5_ib_dev *dev = to_mdev(ibxrcd->device); + struct mlx5_ib_xrcd *xrcd = to_mxrcd(ibxrcd); if (!MLX5_CAP_GEN(dev->mdev, xrc)) - return ERR_PTR(-ENOSYS); - - xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL); - if (!xrcd) - return ERR_PTR(-ENOMEM); - - err = mlx5_cmd_xrcd_alloc(dev->mdev, &xrcd->xrcdn, 0); - if (err) { - kfree(xrcd); - return ERR_PTR(-ENOMEM); - } + return -EOPNOTSUPP; - return &xrcd->ibxrcd; + return mlx5_cmd_xrcd_alloc(dev->mdev, &xrcd->xrcdn, 0); } -int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata) +void mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(xrcd->device); u32 xrcdn = to_mxrcd(xrcd)->xrcdn; - int err; - - err = mlx5_cmd_xrcd_dealloc(dev->mdev, xrcdn, 0); - if (err) - mlx5_ib_warn(dev, "failed to dealloc xrcdn 0x%x\n", xrcdn); - kfree(xrcd); - return 0; + mlx5_cmd_xrcd_dealloc(dev->mdev, xrcdn, 0); } static void mlx5_ib_wq_event(struct mlx5_core_qp *core_qp, int type) @@ -5964,7 +4754,7 @@ static int set_delay_drop(struct mlx5_ib_dev *dev) if (dev->delay_drop.activate) goto out; - err = mlx5_core_set_delay_drop(dev->mdev, dev->delay_drop.timeout); + err = mlx5_core_set_delay_drop(dev, dev->delay_drop.timeout); if (err) goto out; @@ -6070,13 +4860,13 @@ static int create_rq(struct mlx5_ib_rwq *rwq, struct ib_pd *pd, } 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_tracked(dev->mdev, in, inlen, &rwq->core_qp); + err = mlx5_core_create_rq_tracked(dev, in, inlen, &rwq->core_qp); if (!err && init_attr->create_flags & IB_WQ_FLAGS_DELAY_DROP) { err = set_delay_drop(dev); if (err) { mlx5_ib_warn(dev, "Failed to enable delay drop err=%d\n", err); - mlx5_core_destroy_rq_tracked(dev->mdev, &rwq->core_qp); + mlx5_core_destroy_rq_tracked(dev, &rwq->core_qp); } else { rwq->create_flags |= MLX5_IB_WQ_FLAGS_DELAY_DROP; } @@ -6258,7 +5048,7 @@ struct ib_wq *mlx5_ib_create_wq(struct ib_pd *pd, return &rwq->ibwq; err_copy: - mlx5_core_destroy_rq_tracked(dev->mdev, &rwq->core_qp); + mlx5_core_destroy_rq_tracked(dev, &rwq->core_qp); err_user_rq: destroy_user_rq(dev, pd, rwq, udata); err: @@ -6271,7 +5061,7 @@ void mlx5_ib_destroy_wq(struct ib_wq *wq, struct ib_udata *udata) struct mlx5_ib_dev *dev = to_mdev(wq->device); struct mlx5_ib_rwq *rwq = to_mrwq(wq); - mlx5_core_destroy_rq_tracked(dev->mdev, &rwq->core_qp); + mlx5_core_destroy_rq_tracked(dev, &rwq->core_qp); destroy_user_rq(dev, wq->pd, rwq, udata); kfree(rwq); } @@ -6449,7 +5239,7 @@ int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr, "Receive WQ counters are not supported on current FW\n"); } - err = mlx5_core_modify_rq(dev->mdev, rwq->core_qp.qpn, in, inlen); + err = mlx5_core_modify_rq(dev->mdev, rwq->core_qp.qpn, in); if (!err) rwq->ibwq.state = (wq_state == MLX5_RQC_STATE_ERR) ? IB_WQS_ERR : wq_state; @@ -6548,7 +5338,7 @@ void mlx5_ib_drain_sq(struct ib_qp *qp) sdrain.cqe.done = mlx5_ib_drain_qp_done; init_completion(&sdrain.done); - ret = _mlx5_ib_post_send(qp, &swr.wr, &bad_swr, true); + ret = mlx5_ib_post_send_drain(qp, &swr.wr, &bad_swr); if (ret) { WARN_ONCE(ret, "failed to drain send queue: %d\n", ret); return; @@ -6578,7 +5368,7 @@ void mlx5_ib_drain_rq(struct ib_qp *qp) rdrain.cqe.done = mlx5_ib_drain_qp_done; init_completion(&rdrain.done); - ret = _mlx5_ib_post_recv(qp, &rwr, &bad_rwr, true); + ret = mlx5_ib_post_recv_drain(qp, &rwr, &bad_rwr); if (ret) { WARN_ONCE(ret, "failed to drain recv queue: %d\n", ret); return; diff --git a/drivers/infiniband/hw/mlx5/qp.h b/drivers/infiniband/hw/mlx5/qp.h new file mode 100644 index 000000000000..ba899df44c5b --- /dev/null +++ b/drivers/infiniband/hw/mlx5/qp.h @@ -0,0 +1,47 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + */ + +#ifndef _MLX5_IB_QP_H +#define _MLX5_IB_QP_H + +#include "mlx5_ib.h" + +int mlx5_init_qp_table(struct mlx5_ib_dev *dev); +void mlx5_cleanup_qp_table(struct mlx5_ib_dev *dev); + +int mlx5_core_create_dct(struct mlx5_ib_dev *dev, struct mlx5_core_dct *qp, + u32 *in, int inlen, u32 *out, int outlen); +int mlx5_qpc_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, + u32 *in, int inlen, u32 *out); +int mlx5_core_qp_modify(struct mlx5_ib_dev *dev, u16 opcode, u32 opt_param_mask, + void *qpc, struct mlx5_core_qp *qp, u32 *ece); +int mlx5_core_destroy_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp); +int mlx5_core_destroy_dct(struct mlx5_ib_dev *dev, struct mlx5_core_dct *dct); +int mlx5_core_qp_query(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, + u32 *out, int outlen); +int mlx5_core_dct_query(struct mlx5_ib_dev *dev, struct mlx5_core_dct *dct, + u32 *out, int outlen); + +int mlx5_core_set_delay_drop(struct mlx5_ib_dev *dev, u32 timeout_usec); + +void mlx5_core_destroy_rq_tracked(struct mlx5_ib_dev *dev, + struct mlx5_core_qp *rq); +int mlx5_core_create_sq_tracked(struct mlx5_ib_dev *dev, u32 *in, int inlen, + struct mlx5_core_qp *sq); +void mlx5_core_destroy_sq_tracked(struct mlx5_ib_dev *dev, + struct mlx5_core_qp *sq); + +int mlx5_core_create_rq_tracked(struct mlx5_ib_dev *dev, u32 *in, int inlen, + struct mlx5_core_qp *rq); + +struct mlx5_core_rsc_common *mlx5_core_res_hold(struct mlx5_ib_dev *dev, + int res_num, + enum mlx5_res_type res_type); +void mlx5_core_res_put(struct mlx5_core_rsc_common *res); + +int mlx5_core_xrcd_alloc(struct mlx5_ib_dev *dev, u32 *xrcdn); +int mlx5_core_xrcd_dealloc(struct mlx5_ib_dev *dev, u32 xrcdn); +int mlx5_ib_qp_set_counter(struct ib_qp *qp, struct rdma_counter *counter); +#endif /* _MLX5_IB_QP_H */ diff --git a/drivers/infiniband/hw/mlx5/qpc.c b/drivers/infiniband/hw/mlx5/qpc.c new file mode 100644 index 000000000000..7c3968ef9cd1 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/qpc.c @@ -0,0 +1,641 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + */ + +#include <linux/gfp.h> +#include <linux/mlx5/qp.h> +#include <linux/mlx5/driver.h> +#include "mlx5_ib.h" +#include "qp.h" + +static int mlx5_core_drain_dct(struct mlx5_ib_dev *dev, + struct mlx5_core_dct *dct); + +static struct mlx5_core_rsc_common * +mlx5_get_rsc(struct mlx5_qp_table *table, u32 rsn) +{ + struct mlx5_core_rsc_common *common; + unsigned long flags; + + spin_lock_irqsave(&table->lock, flags); + + common = radix_tree_lookup(&table->tree, rsn); + if (common) + refcount_inc(&common->refcount); + + spin_unlock_irqrestore(&table->lock, flags); + + return common; +} + +void mlx5_core_put_rsc(struct mlx5_core_rsc_common *common) +{ + if (refcount_dec_and_test(&common->refcount)) + complete(&common->free); +} + +static u64 qp_allowed_event_types(void) +{ + u64 mask; + + mask = BIT(MLX5_EVENT_TYPE_PATH_MIG) | + BIT(MLX5_EVENT_TYPE_COMM_EST) | + BIT(MLX5_EVENT_TYPE_SQ_DRAINED) | + BIT(MLX5_EVENT_TYPE_SRQ_LAST_WQE) | + BIT(MLX5_EVENT_TYPE_WQ_CATAS_ERROR) | + BIT(MLX5_EVENT_TYPE_PATH_MIG_FAILED) | + BIT(MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR) | + BIT(MLX5_EVENT_TYPE_WQ_ACCESS_ERROR); + + return mask; +} + +static u64 rq_allowed_event_types(void) +{ + u64 mask; + + mask = BIT(MLX5_EVENT_TYPE_SRQ_LAST_WQE) | + BIT(MLX5_EVENT_TYPE_WQ_CATAS_ERROR); + + return mask; +} + +static u64 sq_allowed_event_types(void) +{ + return BIT(MLX5_EVENT_TYPE_WQ_CATAS_ERROR); +} + +static u64 dct_allowed_event_types(void) +{ + return BIT(MLX5_EVENT_TYPE_DCT_DRAINED); +} + +static bool is_event_type_allowed(int rsc_type, int event_type) +{ + switch (rsc_type) { + case MLX5_EVENT_QUEUE_TYPE_QP: + return BIT(event_type) & qp_allowed_event_types(); + case MLX5_EVENT_QUEUE_TYPE_RQ: + return BIT(event_type) & rq_allowed_event_types(); + case MLX5_EVENT_QUEUE_TYPE_SQ: + return BIT(event_type) & sq_allowed_event_types(); + case MLX5_EVENT_QUEUE_TYPE_DCT: + return BIT(event_type) & dct_allowed_event_types(); + default: + WARN(1, "Event arrived for unknown resource type"); + return false; + } +} + +static int rsc_event_notifier(struct notifier_block *nb, + unsigned long type, void *data) +{ + struct mlx5_core_rsc_common *common; + struct mlx5_qp_table *table; + struct mlx5_core_dct *dct; + u8 event_type = (u8)type; + struct mlx5_core_qp *qp; + struct mlx5_eqe *eqe; + u32 rsn; + + switch (event_type) { + case MLX5_EVENT_TYPE_DCT_DRAINED: + eqe = data; + rsn = be32_to_cpu(eqe->data.dct.dctn) & 0xffffff; + rsn |= (MLX5_RES_DCT << MLX5_USER_INDEX_LEN); + break; + case MLX5_EVENT_TYPE_PATH_MIG: + case MLX5_EVENT_TYPE_COMM_EST: + case MLX5_EVENT_TYPE_SQ_DRAINED: + case MLX5_EVENT_TYPE_SRQ_LAST_WQE: + case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: + case MLX5_EVENT_TYPE_PATH_MIG_FAILED: + case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR: + case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: + eqe = data; + rsn = be32_to_cpu(eqe->data.qp_srq.qp_srq_n) & 0xffffff; + rsn |= (eqe->data.qp_srq.type << MLX5_USER_INDEX_LEN); + break; + default: + return NOTIFY_DONE; + } + + table = container_of(nb, struct mlx5_qp_table, nb); + common = mlx5_get_rsc(table, rsn); + if (!common) + return NOTIFY_OK; + + if (!is_event_type_allowed((rsn >> MLX5_USER_INDEX_LEN), event_type)) + goto out; + + switch (common->res) { + case MLX5_RES_QP: + case MLX5_RES_RQ: + case MLX5_RES_SQ: + qp = (struct mlx5_core_qp *)common; + qp->event(qp, event_type); + break; + case MLX5_RES_DCT: + dct = (struct mlx5_core_dct *)common; + if (event_type == MLX5_EVENT_TYPE_DCT_DRAINED) + complete(&dct->drained); + break; + default: + break; + } +out: + mlx5_core_put_rsc(common); + + return NOTIFY_OK; +} + +static int create_resource_common(struct mlx5_ib_dev *dev, + struct mlx5_core_qp *qp, int rsc_type) +{ + struct mlx5_qp_table *table = &dev->qp_table; + int err; + + qp->common.res = rsc_type; + spin_lock_irq(&table->lock); + err = radix_tree_insert(&table->tree, + qp->qpn | (rsc_type << MLX5_USER_INDEX_LEN), + qp); + spin_unlock_irq(&table->lock); + if (err) + return err; + + refcount_set(&qp->common.refcount, 1); + init_completion(&qp->common.free); + qp->pid = current->pid; + + return 0; +} + +static void destroy_resource_common(struct mlx5_ib_dev *dev, + struct mlx5_core_qp *qp) +{ + struct mlx5_qp_table *table = &dev->qp_table; + unsigned long flags; + + spin_lock_irqsave(&table->lock, flags); + radix_tree_delete(&table->tree, + qp->qpn | (qp->common.res << MLX5_USER_INDEX_LEN)); + spin_unlock_irqrestore(&table->lock, flags); + mlx5_core_put_rsc((struct mlx5_core_rsc_common *)qp); + wait_for_completion(&qp->common.free); +} + +static int _mlx5_core_destroy_dct(struct mlx5_ib_dev *dev, + struct mlx5_core_dct *dct, bool need_cleanup) +{ + u32 in[MLX5_ST_SZ_DW(destroy_dct_in)] = {}; + struct mlx5_core_qp *qp = &dct->mqp; + int err; + + err = mlx5_core_drain_dct(dev, dct); + if (err) { + if (dev->mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) + goto destroy; + + return err; + } + wait_for_completion(&dct->drained); +destroy: + if (need_cleanup) + destroy_resource_common(dev, &dct->mqp); + MLX5_SET(destroy_dct_in, in, opcode, MLX5_CMD_OP_DESTROY_DCT); + MLX5_SET(destroy_dct_in, in, dctn, qp->qpn); + MLX5_SET(destroy_dct_in, in, uid, qp->uid); + err = mlx5_cmd_exec_in(dev->mdev, destroy_dct, in); + return err; +} + +int mlx5_core_create_dct(struct mlx5_ib_dev *dev, struct mlx5_core_dct *dct, + u32 *in, int inlen, u32 *out, int outlen) +{ + struct mlx5_core_qp *qp = &dct->mqp; + int err; + + init_completion(&dct->drained); + MLX5_SET(create_dct_in, in, opcode, MLX5_CMD_OP_CREATE_DCT); + + err = mlx5_cmd_exec(dev->mdev, in, inlen, out, outlen); + if (err) + return err; + + qp->qpn = MLX5_GET(create_dct_out, out, dctn); + qp->uid = MLX5_GET(create_dct_in, in, uid); + err = create_resource_common(dev, qp, MLX5_RES_DCT); + if (err) + goto err_cmd; + + return 0; +err_cmd: + _mlx5_core_destroy_dct(dev, dct, false); + return err; +} + +int mlx5_qpc_create_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, + u32 *in, int inlen, u32 *out) +{ + u32 din[MLX5_ST_SZ_DW(destroy_qp_in)] = {}; + int err; + + MLX5_SET(create_qp_in, in, opcode, MLX5_CMD_OP_CREATE_QP); + + err = mlx5_cmd_exec(dev->mdev, in, inlen, out, + MLX5_ST_SZ_BYTES(create_qp_out)); + if (err) + return err; + + qp->uid = MLX5_GET(create_qp_in, in, uid); + qp->qpn = MLX5_GET(create_qp_out, out, qpn); + + err = create_resource_common(dev, qp, MLX5_RES_QP); + if (err) + goto err_cmd; + + mlx5_debug_qp_add(dev->mdev, qp); + + return 0; + +err_cmd: + MLX5_SET(destroy_qp_in, din, opcode, MLX5_CMD_OP_DESTROY_QP); + MLX5_SET(destroy_qp_in, din, qpn, qp->qpn); + MLX5_SET(destroy_qp_in, din, uid, qp->uid); + mlx5_cmd_exec_in(dev->mdev, destroy_qp, din); + return err; +} + +static int mlx5_core_drain_dct(struct mlx5_ib_dev *dev, + struct mlx5_core_dct *dct) +{ + u32 in[MLX5_ST_SZ_DW(drain_dct_in)] = {}; + struct mlx5_core_qp *qp = &dct->mqp; + + MLX5_SET(drain_dct_in, in, opcode, MLX5_CMD_OP_DRAIN_DCT); + MLX5_SET(drain_dct_in, in, dctn, qp->qpn); + MLX5_SET(drain_dct_in, in, uid, qp->uid); + return mlx5_cmd_exec_in(dev->mdev, drain_dct, in); +} + +int mlx5_core_destroy_dct(struct mlx5_ib_dev *dev, + struct mlx5_core_dct *dct) +{ + return _mlx5_core_destroy_dct(dev, dct, true); +} + +int mlx5_core_destroy_qp(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp) +{ + u32 in[MLX5_ST_SZ_DW(destroy_qp_in)] = {}; + + mlx5_debug_qp_remove(dev->mdev, qp); + + destroy_resource_common(dev, qp); + + MLX5_SET(destroy_qp_in, in, opcode, MLX5_CMD_OP_DESTROY_QP); + MLX5_SET(destroy_qp_in, in, qpn, qp->qpn); + MLX5_SET(destroy_qp_in, in, uid, qp->uid); + mlx5_cmd_exec_in(dev->mdev, destroy_qp, in); + return 0; +} + +int mlx5_core_set_delay_drop(struct mlx5_ib_dev *dev, + u32 timeout_usec) +{ + u32 in[MLX5_ST_SZ_DW(set_delay_drop_params_in)] = {}; + + MLX5_SET(set_delay_drop_params_in, in, opcode, + MLX5_CMD_OP_SET_DELAY_DROP_PARAMS); + MLX5_SET(set_delay_drop_params_in, in, delay_drop_timeout, + timeout_usec / 100); + return mlx5_cmd_exec_in(dev->mdev, set_delay_drop_params, in); +} + +struct mbox_info { + u32 *in; + u32 *out; + int inlen; + int outlen; +}; + +static int mbox_alloc(struct mbox_info *mbox, int inlen, int outlen) +{ + mbox->inlen = inlen; + mbox->outlen = outlen; + mbox->in = kzalloc(mbox->inlen, GFP_KERNEL); + mbox->out = kzalloc(mbox->outlen, GFP_KERNEL); + if (!mbox->in || !mbox->out) { + kfree(mbox->in); + kfree(mbox->out); + return -ENOMEM; + } + + return 0; +} + +static void mbox_free(struct mbox_info *mbox) +{ + kfree(mbox->in); + kfree(mbox->out); +} + +static int get_ece_from_mbox(void *out, u16 opcode) +{ + int ece = 0; + + switch (opcode) { + case MLX5_CMD_OP_INIT2INIT_QP: + ece = MLX5_GET(init2init_qp_out, out, ece); + break; + case MLX5_CMD_OP_INIT2RTR_QP: + ece = MLX5_GET(init2rtr_qp_out, out, ece); + break; + case MLX5_CMD_OP_RTR2RTS_QP: + ece = MLX5_GET(rtr2rts_qp_out, out, ece); + break; + case MLX5_CMD_OP_RTS2RTS_QP: + ece = MLX5_GET(rts2rts_qp_out, out, ece); + break; + case MLX5_CMD_OP_RST2INIT_QP: + ece = MLX5_GET(rst2init_qp_out, out, ece); + break; + default: + break; + } + + return ece; +} + +static int modify_qp_mbox_alloc(struct mlx5_core_dev *dev, u16 opcode, int qpn, + u32 opt_param_mask, void *qpc, + struct mbox_info *mbox, u16 uid, u32 ece) +{ + mbox->out = NULL; + mbox->in = NULL; + +#define MBOX_ALLOC(mbox, typ) \ + mbox_alloc(mbox, MLX5_ST_SZ_BYTES(typ##_in), MLX5_ST_SZ_BYTES(typ##_out)) + +#define MOD_QP_IN_SET(typ, in, _opcode, _qpn, _uid) \ + do { \ + MLX5_SET(typ##_in, in, opcode, _opcode); \ + MLX5_SET(typ##_in, in, qpn, _qpn); \ + MLX5_SET(typ##_in, in, uid, _uid); \ + } while (0) + +#define MOD_QP_IN_SET_QPC(typ, in, _opcode, _qpn, _opt_p, _qpc, _uid) \ + do { \ + MOD_QP_IN_SET(typ, in, _opcode, _qpn, _uid); \ + MLX5_SET(typ##_in, in, opt_param_mask, _opt_p); \ + memcpy(MLX5_ADDR_OF(typ##_in, in, qpc), _qpc, \ + MLX5_ST_SZ_BYTES(qpc)); \ + } while (0) + + switch (opcode) { + /* 2RST & 2ERR */ + case MLX5_CMD_OP_2RST_QP: + if (MBOX_ALLOC(mbox, qp_2rst)) + return -ENOMEM; + MOD_QP_IN_SET(qp_2rst, mbox->in, opcode, qpn, uid); + break; + case MLX5_CMD_OP_2ERR_QP: + if (MBOX_ALLOC(mbox, qp_2err)) + return -ENOMEM; + MOD_QP_IN_SET(qp_2err, mbox->in, opcode, qpn, uid); + break; + + /* MODIFY with QPC */ + case MLX5_CMD_OP_RST2INIT_QP: + if (MBOX_ALLOC(mbox, rst2init_qp)) + return -ENOMEM; + MOD_QP_IN_SET_QPC(rst2init_qp, mbox->in, opcode, qpn, + opt_param_mask, qpc, uid); + MLX5_SET(rst2init_qp_in, mbox->in, ece, ece); + break; + case MLX5_CMD_OP_INIT2RTR_QP: + if (MBOX_ALLOC(mbox, init2rtr_qp)) + return -ENOMEM; + MOD_QP_IN_SET_QPC(init2rtr_qp, mbox->in, opcode, qpn, + opt_param_mask, qpc, uid); + MLX5_SET(init2rtr_qp_in, mbox->in, ece, ece); + break; + case MLX5_CMD_OP_RTR2RTS_QP: + if (MBOX_ALLOC(mbox, rtr2rts_qp)) + return -ENOMEM; + MOD_QP_IN_SET_QPC(rtr2rts_qp, mbox->in, opcode, qpn, + opt_param_mask, qpc, uid); + MLX5_SET(rtr2rts_qp_in, mbox->in, ece, ece); + break; + case MLX5_CMD_OP_RTS2RTS_QP: + if (MBOX_ALLOC(mbox, rts2rts_qp)) + return -ENOMEM; + MOD_QP_IN_SET_QPC(rts2rts_qp, mbox->in, opcode, qpn, + opt_param_mask, qpc, uid); + MLX5_SET(rts2rts_qp_in, mbox->in, ece, ece); + break; + case MLX5_CMD_OP_SQERR2RTS_QP: + if (MBOX_ALLOC(mbox, sqerr2rts_qp)) + return -ENOMEM; + MOD_QP_IN_SET_QPC(sqerr2rts_qp, mbox->in, opcode, qpn, + opt_param_mask, qpc, uid); + break; + case MLX5_CMD_OP_INIT2INIT_QP: + if (MBOX_ALLOC(mbox, init2init_qp)) + return -ENOMEM; + MOD_QP_IN_SET_QPC(init2init_qp, mbox->in, opcode, qpn, + opt_param_mask, qpc, uid); + MLX5_SET(init2init_qp_in, mbox->in, ece, ece); + break; + default: + return -EINVAL; + } + return 0; +} + +int mlx5_core_qp_modify(struct mlx5_ib_dev *dev, u16 opcode, u32 opt_param_mask, + void *qpc, struct mlx5_core_qp *qp, u32 *ece) +{ + struct mbox_info mbox; + int err; + + err = modify_qp_mbox_alloc(dev->mdev, opcode, qp->qpn, opt_param_mask, + qpc, &mbox, qp->uid, (ece) ? *ece : 0); + if (err) + return err; + + err = mlx5_cmd_exec(dev->mdev, mbox.in, mbox.inlen, mbox.out, + mbox.outlen); + + if (ece) + *ece = get_ece_from_mbox(mbox.out, opcode); + + mbox_free(&mbox); + return err; +} + +int mlx5_init_qp_table(struct mlx5_ib_dev *dev) +{ + struct mlx5_qp_table *table = &dev->qp_table; + + spin_lock_init(&table->lock); + INIT_RADIX_TREE(&table->tree, GFP_ATOMIC); + mlx5_qp_debugfs_init(dev->mdev); + + table->nb.notifier_call = rsc_event_notifier; + mlx5_notifier_register(dev->mdev, &table->nb); + + return 0; +} + +void mlx5_cleanup_qp_table(struct mlx5_ib_dev *dev) +{ + struct mlx5_qp_table *table = &dev->qp_table; + + mlx5_notifier_unregister(dev->mdev, &table->nb); + mlx5_qp_debugfs_cleanup(dev->mdev); +} + +int mlx5_core_qp_query(struct mlx5_ib_dev *dev, struct mlx5_core_qp *qp, + u32 *out, int outlen) +{ + u32 in[MLX5_ST_SZ_DW(query_qp_in)] = {}; + + MLX5_SET(query_qp_in, in, opcode, MLX5_CMD_OP_QUERY_QP); + MLX5_SET(query_qp_in, in, qpn, qp->qpn); + return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, outlen); +} + +int mlx5_core_dct_query(struct mlx5_ib_dev *dev, struct mlx5_core_dct *dct, + u32 *out, int outlen) +{ + u32 in[MLX5_ST_SZ_DW(query_dct_in)] = {}; + struct mlx5_core_qp *qp = &dct->mqp; + + MLX5_SET(query_dct_in, in, opcode, MLX5_CMD_OP_QUERY_DCT); + MLX5_SET(query_dct_in, in, dctn, qp->qpn); + + return mlx5_cmd_exec(dev->mdev, (void *)&in, sizeof(in), (void *)out, + outlen); +} + +int mlx5_core_xrcd_alloc(struct mlx5_ib_dev *dev, u32 *xrcdn) +{ + u32 out[MLX5_ST_SZ_DW(alloc_xrcd_out)] = {}; + u32 in[MLX5_ST_SZ_DW(alloc_xrcd_in)] = {}; + int err; + + MLX5_SET(alloc_xrcd_in, in, opcode, MLX5_CMD_OP_ALLOC_XRCD); + err = mlx5_cmd_exec_inout(dev->mdev, alloc_xrcd, in, out); + if (!err) + *xrcdn = MLX5_GET(alloc_xrcd_out, out, xrcd); + return err; +} + +int mlx5_core_xrcd_dealloc(struct mlx5_ib_dev *dev, u32 xrcdn) +{ + u32 in[MLX5_ST_SZ_DW(dealloc_xrcd_in)] = {}; + + MLX5_SET(dealloc_xrcd_in, in, opcode, MLX5_CMD_OP_DEALLOC_XRCD); + MLX5_SET(dealloc_xrcd_in, in, xrcd, xrcdn); + return mlx5_cmd_exec_in(dev->mdev, dealloc_xrcd, in); +} + +static void destroy_rq_tracked(struct mlx5_ib_dev *dev, u32 rqn, u16 uid) +{ + u32 in[MLX5_ST_SZ_DW(destroy_rq_in)] = {}; + + MLX5_SET(destroy_rq_in, in, opcode, MLX5_CMD_OP_DESTROY_RQ); + MLX5_SET(destroy_rq_in, in, rqn, rqn); + MLX5_SET(destroy_rq_in, in, uid, uid); + mlx5_cmd_exec_in(dev->mdev, destroy_rq, in); +} + +int mlx5_core_create_rq_tracked(struct mlx5_ib_dev *dev, u32 *in, int inlen, + struct mlx5_core_qp *rq) +{ + int err; + u32 rqn; + + err = mlx5_core_create_rq(dev->mdev, in, inlen, &rqn); + if (err) + return err; + + rq->uid = MLX5_GET(create_rq_in, in, uid); + rq->qpn = rqn; + err = create_resource_common(dev, rq, MLX5_RES_RQ); + if (err) + goto err_destroy_rq; + + return 0; + +err_destroy_rq: + destroy_rq_tracked(dev, rq->qpn, rq->uid); + + return err; +} + +void mlx5_core_destroy_rq_tracked(struct mlx5_ib_dev *dev, + struct mlx5_core_qp *rq) +{ + destroy_resource_common(dev, rq); + destroy_rq_tracked(dev, rq->qpn, rq->uid); +} + +static void destroy_sq_tracked(struct mlx5_ib_dev *dev, u32 sqn, u16 uid) +{ + u32 in[MLX5_ST_SZ_DW(destroy_sq_in)] = {}; + + MLX5_SET(destroy_sq_in, in, opcode, MLX5_CMD_OP_DESTROY_SQ); + MLX5_SET(destroy_sq_in, in, sqn, sqn); + MLX5_SET(destroy_sq_in, in, uid, uid); + mlx5_cmd_exec_in(dev->mdev, destroy_sq, in); +} + +int mlx5_core_create_sq_tracked(struct mlx5_ib_dev *dev, u32 *in, int inlen, + struct mlx5_core_qp *sq) +{ + u32 out[MLX5_ST_SZ_DW(create_sq_out)] = {}; + int err; + + MLX5_SET(create_sq_in, in, opcode, MLX5_CMD_OP_CREATE_SQ); + err = mlx5_cmd_exec(dev->mdev, in, inlen, out, sizeof(out)); + if (err) + return err; + + sq->qpn = MLX5_GET(create_sq_out, out, sqn); + sq->uid = MLX5_GET(create_sq_in, in, uid); + err = create_resource_common(dev, sq, MLX5_RES_SQ); + if (err) + goto err_destroy_sq; + + return 0; + +err_destroy_sq: + destroy_sq_tracked(dev, sq->qpn, sq->uid); + + return err; +} + +void mlx5_core_destroy_sq_tracked(struct mlx5_ib_dev *dev, + struct mlx5_core_qp *sq) +{ + destroy_resource_common(dev, sq); + destroy_sq_tracked(dev, sq->qpn, sq->uid); +} + +struct mlx5_core_rsc_common *mlx5_core_res_hold(struct mlx5_ib_dev *dev, + int res_num, + enum mlx5_res_type res_type) +{ + u32 rsn = res_num | (res_type << MLX5_USER_INDEX_LEN); + struct mlx5_qp_table *table = &dev->qp_table; + + return mlx5_get_rsc(table, rsn); +} + +void mlx5_core_res_put(struct mlx5_core_rsc_common *res) +{ + mlx5_core_put_rsc(res); +} diff --git a/drivers/infiniband/hw/mlx5/restrack.c b/drivers/infiniband/hw/mlx5/restrack.c index 8f6c04f12531..887270dd3ce2 100644 --- a/drivers/infiniband/hw/mlx5/restrack.c +++ b/drivers/infiniband/hw/mlx5/restrack.c @@ -1,17 +1,85 @@ // SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* - * Copyright (c) 2019, Mellanox Technologies inc. All rights reserved. + * Copyright (c) 2019-2020, Mellanox Technologies Ltd. All rights reserved. */ #include <uapi/rdma/rdma_netlink.h> +#include <linux/mlx5/rsc_dump.h> #include <rdma/ib_umem_odp.h> #include <rdma/restrack.h> #include "mlx5_ib.h" +#include "restrack.h" -static int fill_stat_mr_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +#define MAX_DUMP_SIZE 1024 + +static int dump_rsc(struct mlx5_core_dev *dev, enum mlx5_sgmt_type type, + int index, void *data, int *data_len) +{ + struct mlx5_core_dev *mdev = dev; + struct mlx5_rsc_dump_cmd *cmd; + struct mlx5_rsc_key key = {}; + struct page *page; + int offset = 0; + int err = 0; + int cmd_err; + int size; + + page = alloc_page(GFP_KERNEL); + if (!page) + return -ENOMEM; + + key.size = PAGE_SIZE; + key.rsc = type; + key.index1 = index; + key.num_of_obj1 = 1; + + cmd = mlx5_rsc_dump_cmd_create(mdev, &key); + if (IS_ERR(cmd)) { + err = PTR_ERR(cmd); + goto free_page; + } + + do { + cmd_err = mlx5_rsc_dump_next(mdev, cmd, page, &size); + if (cmd_err < 0 || size + offset > MAX_DUMP_SIZE) { + err = cmd_err; + goto destroy_cmd; + } + memcpy(data + offset, page_address(page), size); + offset += size; + } while (cmd_err > 0); + *data_len = offset; + +destroy_cmd: + mlx5_rsc_dump_cmd_destroy(cmd); +free_page: + __free_page(page); + return err; +} + +static int fill_res_raw(struct sk_buff *msg, struct mlx5_ib_dev *dev, + enum mlx5_sgmt_type type, u32 key) +{ + int len = 0; + void *data; + int err; + + data = kzalloc(MAX_DUMP_SIZE, GFP_KERNEL); + if (!data) + return -ENOMEM; + + err = dump_rsc(dev->mdev, type, key, data, &len); + if (err) + goto out; + + err = nla_put(msg, RDMA_NLDEV_ATTR_RES_RAW, len, data); +out: + kfree(data); + return err; +} + +static int fill_stat_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr) { - struct ib_mr *ibmr = container_of(res, struct ib_mr, res); struct mlx5_ib_mr *mr = to_mmr(ibmr); struct nlattr *table_attr; @@ -31,6 +99,9 @@ static int fill_stat_mr_entry(struct sk_buff *msg, msg, "page_invalidations", atomic64_read(&mr->odp_stats.invalidations))) goto err_table; + if (rdma_nl_stat_hwcounter_entry(msg, "page_prefetch", + atomic64_read(&mr->odp_stats.prefetch))) + goto err_table; nla_nest_end(msg, table_attr); return 0; @@ -41,10 +112,16 @@ err: return -EMSGSIZE; } -static int fill_res_mr_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +static int fill_res_mr_entry_raw(struct sk_buff *msg, struct ib_mr *ibmr) +{ + struct mlx5_ib_mr *mr = to_mmr(ibmr); + + return fill_res_raw(msg, mr->dev, MLX5_SGMT_TYPE_PRM_QUERY_MKEY, + mlx5_mkey_to_idx(mr->mmkey.key)); +} + +static int fill_res_mr_entry(struct sk_buff *msg, struct ib_mr *ibmr) { - struct ib_mr *ibmr = container_of(res, struct ib_mr, res); struct mlx5_ib_mr *mr = to_mmr(ibmr); struct nlattr *table_attr; @@ -71,20 +148,32 @@ err: return -EMSGSIZE; } -int mlx5_ib_fill_res_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +static int fill_res_cq_entry_raw(struct sk_buff *msg, struct ib_cq *ibcq) { - if (res->type == RDMA_RESTRACK_MR) - return fill_res_mr_entry(msg, res); + struct mlx5_ib_dev *dev = to_mdev(ibcq->device); + struct mlx5_ib_cq *cq = to_mcq(ibcq); - return 0; + return fill_res_raw(msg, dev, MLX5_SGMT_TYPE_PRM_QUERY_CQ, cq->mcq.cqn); } -int mlx5_ib_fill_stat_entry(struct sk_buff *msg, - struct rdma_restrack_entry *res) +static int fill_res_qp_entry_raw(struct sk_buff *msg, struct ib_qp *ibqp) { - if (res->type == RDMA_RESTRACK_MR) - return fill_stat_mr_entry(msg, res); + struct mlx5_ib_dev *dev = to_mdev(ibqp->device); + + return fill_res_raw(msg, dev, MLX5_SGMT_TYPE_PRM_QUERY_QP, + ibqp->qp_num); +} +static const struct ib_device_ops restrack_ops = { + .fill_res_cq_entry_raw = fill_res_cq_entry_raw, + .fill_res_mr_entry = fill_res_mr_entry, + .fill_res_mr_entry_raw = fill_res_mr_entry_raw, + .fill_res_qp_entry_raw = fill_res_qp_entry_raw, + .fill_stat_mr_entry = fill_stat_mr_entry, +}; + +int mlx5_ib_restrack_init(struct mlx5_ib_dev *dev) +{ + ib_set_device_ops(&dev->ib_dev, &restrack_ops); return 0; } diff --git a/drivers/infiniband/hw/mlx5/restrack.h b/drivers/infiniband/hw/mlx5/restrack.h new file mode 100644 index 000000000000..e8d81270f1b6 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/restrack.h @@ -0,0 +1,13 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2013-2020, Mellanox Technologies Ltd. All rights reserved. + */ + +#ifndef _MLX5_IB_RESTRACK_H +#define _MLX5_IB_RESTRACK_H + +#include "mlx5_ib.h" + +int mlx5_ib_restrack_init(struct mlx5_ib_dev *dev); + +#endif /* _MLX5_IB_RESTRACK_H */ diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c index b1a8a9175040..7e10cbcb6d5c 100644 --- a/drivers/infiniband/hw/mlx5/srq.c +++ b/drivers/infiniband/hw/mlx5/srq.c @@ -274,10 +274,10 @@ int mlx5_ib_create_srq(struct ib_srq *ib_srq, if (srq->wq_sig) in.flags |= MLX5_SRQ_FLAG_WQ_SIG; - if (init_attr->srq_type == IB_SRQT_XRC) + if (init_attr->srq_type == IB_SRQT_XRC && init_attr->ext.xrc.xrcd) in.xrcd = to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn; else - in.xrcd = to_mxrcd(dev->devr.x0)->xrcdn; + in.xrcd = dev->devr.xrcdn0; if (init_attr->srq_type == IB_SRQT_TM) { in.tm_log_list_size = @@ -310,12 +310,18 @@ int mlx5_ib_create_srq(struct ib_srq *ib_srq, srq->msrq.event = mlx5_ib_srq_event; srq->ibsrq.ext.xrc.srq_num = srq->msrq.srqn; - if (udata) - if (ib_copy_to_udata(udata, &srq->msrq.srqn, sizeof(__u32))) { + if (udata) { + struct mlx5_ib_create_srq_resp resp = { + .srqn = srq->msrq.srqn, + }; + + if (ib_copy_to_udata(udata, &resp, min(udata->outlen, + sizeof(resp)))) { mlx5_ib_dbg(dev, "copy to user failed\n"); err = -EFAULT; goto err_core; } + } init_attr->attr.max_wr = srq->msrq.max - 1; diff --git a/drivers/infiniband/hw/mlx5/srq_cmd.c b/drivers/infiniband/hw/mlx5/srq_cmd.c index 8fc3630a9d4c..37aaacebd3f2 100644 --- a/drivers/infiniband/hw/mlx5/srq_cmd.c +++ b/drivers/infiniband/hw/mlx5/srq_cmd.c @@ -5,9 +5,9 @@ #include <linux/kernel.h> #include <linux/mlx5/driver.h> -#include <linux/mlx5/cmd.h> #include "mlx5_ib.h" #include "srq.h" +#include "qp.h" static int get_pas_size(struct mlx5_srq_attr *in) { @@ -83,11 +83,11 @@ struct mlx5_core_srq *mlx5_cmd_get_srq(struct mlx5_ib_dev *dev, u32 srqn) struct mlx5_srq_table *table = &dev->srq_table; struct mlx5_core_srq *srq; - xa_lock(&table->array); + xa_lock_irq(&table->array); srq = xa_load(&table->array, srqn); if (srq) refcount_inc(&srq->common.refcount); - xa_unlock(&table->array); + xa_unlock_irq(&table->array); return srq; } @@ -132,38 +132,33 @@ static int create_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, static int destroy_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { - u32 srq_in[MLX5_ST_SZ_DW(destroy_srq_in)] = {0}; - u32 srq_out[MLX5_ST_SZ_DW(destroy_srq_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(destroy_srq_in)] = {}; - MLX5_SET(destroy_srq_in, srq_in, opcode, - MLX5_CMD_OP_DESTROY_SRQ); - MLX5_SET(destroy_srq_in, srq_in, srqn, srq->srqn); - MLX5_SET(destroy_srq_in, srq_in, uid, srq->uid); + MLX5_SET(destroy_srq_in, in, opcode, MLX5_CMD_OP_DESTROY_SRQ); + MLX5_SET(destroy_srq_in, in, srqn, srq->srqn); + MLX5_SET(destroy_srq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, srq_in, sizeof(srq_in), srq_out, - sizeof(srq_out)); + return mlx5_cmd_exec_in(dev->mdev, destroy_srq, in); } static int arm_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, u16 lwm, int is_srq) { - u32 srq_in[MLX5_ST_SZ_DW(arm_rq_in)] = {0}; - u32 srq_out[MLX5_ST_SZ_DW(arm_rq_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(arm_rq_in)] = {}; - MLX5_SET(arm_rq_in, srq_in, opcode, MLX5_CMD_OP_ARM_RQ); - MLX5_SET(arm_rq_in, srq_in, op_mod, MLX5_ARM_RQ_IN_OP_MOD_SRQ); - MLX5_SET(arm_rq_in, srq_in, srq_number, srq->srqn); - MLX5_SET(arm_rq_in, srq_in, lwm, lwm); - MLX5_SET(arm_rq_in, srq_in, uid, srq->uid); + MLX5_SET(arm_rq_in, in, opcode, MLX5_CMD_OP_ARM_RQ); + MLX5_SET(arm_rq_in, in, op_mod, MLX5_ARM_RQ_IN_OP_MOD_SRQ); + MLX5_SET(arm_rq_in, in, srq_number, srq->srqn); + MLX5_SET(arm_rq_in, in, lwm, lwm); + MLX5_SET(arm_rq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, srq_in, sizeof(srq_in), srq_out, - sizeof(srq_out)); + return mlx5_cmd_exec_in(dev->mdev, arm_rq, in); } static int query_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { - u32 srq_in[MLX5_ST_SZ_DW(query_srq_in)] = {0}; + u32 in[MLX5_ST_SZ_DW(query_srq_in)] = {}; u32 *srq_out; void *srqc; int err; @@ -172,11 +167,9 @@ static int query_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, if (!srq_out) return -ENOMEM; - MLX5_SET(query_srq_in, srq_in, opcode, - MLX5_CMD_OP_QUERY_SRQ); - MLX5_SET(query_srq_in, srq_in, srqn, srq->srqn); - err = mlx5_cmd_exec(dev->mdev, srq_in, sizeof(srq_in), srq_out, - MLX5_ST_SZ_BYTES(query_srq_out)); + MLX5_SET(query_srq_in, in, opcode, MLX5_CMD_OP_QUERY_SRQ); + MLX5_SET(query_srq_in, in, srqn, srq->srqn); + err = mlx5_cmd_exec_inout(dev->mdev, query_srq, in, srq_out); if (err) goto out; @@ -234,39 +227,35 @@ out: static int destroy_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { - u32 xrcsrq_in[MLX5_ST_SZ_DW(destroy_xrc_srq_in)] = {0}; - u32 xrcsrq_out[MLX5_ST_SZ_DW(destroy_xrc_srq_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(destroy_xrc_srq_in)] = {}; - MLX5_SET(destroy_xrc_srq_in, xrcsrq_in, opcode, - MLX5_CMD_OP_DESTROY_XRC_SRQ); - MLX5_SET(destroy_xrc_srq_in, xrcsrq_in, xrc_srqn, srq->srqn); - MLX5_SET(destroy_xrc_srq_in, xrcsrq_in, uid, srq->uid); + MLX5_SET(destroy_xrc_srq_in, in, opcode, MLX5_CMD_OP_DESTROY_XRC_SRQ); + MLX5_SET(destroy_xrc_srq_in, in, xrc_srqn, srq->srqn); + MLX5_SET(destroy_xrc_srq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, xrcsrq_in, sizeof(xrcsrq_in), - xrcsrq_out, sizeof(xrcsrq_out)); + return mlx5_cmd_exec_in(dev->mdev, destroy_xrc_srq, in); } static int arm_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, u16 lwm) { - u32 xrcsrq_in[MLX5_ST_SZ_DW(arm_xrc_srq_in)] = {0}; - u32 xrcsrq_out[MLX5_ST_SZ_DW(arm_xrc_srq_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(arm_xrc_srq_in)] = {}; - MLX5_SET(arm_xrc_srq_in, xrcsrq_in, opcode, MLX5_CMD_OP_ARM_XRC_SRQ); - MLX5_SET(arm_xrc_srq_in, xrcsrq_in, op_mod, MLX5_ARM_XRC_SRQ_IN_OP_MOD_XRC_SRQ); - MLX5_SET(arm_xrc_srq_in, xrcsrq_in, xrc_srqn, srq->srqn); - MLX5_SET(arm_xrc_srq_in, xrcsrq_in, lwm, lwm); - MLX5_SET(arm_xrc_srq_in, xrcsrq_in, uid, srq->uid); + MLX5_SET(arm_xrc_srq_in, in, opcode, MLX5_CMD_OP_ARM_XRC_SRQ); + MLX5_SET(arm_xrc_srq_in, in, op_mod, + MLX5_ARM_XRC_SRQ_IN_OP_MOD_XRC_SRQ); + MLX5_SET(arm_xrc_srq_in, in, xrc_srqn, srq->srqn); + MLX5_SET(arm_xrc_srq_in, in, lwm, lwm); + MLX5_SET(arm_xrc_srq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, xrcsrq_in, sizeof(xrcsrq_in), - xrcsrq_out, sizeof(xrcsrq_out)); + return mlx5_cmd_exec_in(dev->mdev, arm_xrc_srq, in); } static int query_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { - u32 xrcsrq_in[MLX5_ST_SZ_DW(query_xrc_srq_in)]; + u32 in[MLX5_ST_SZ_DW(query_xrc_srq_in)] = {}; u32 *xrcsrq_out; void *xrc_srqc; int err; @@ -274,14 +263,11 @@ static int query_xrc_srq_cmd(struct mlx5_ib_dev *dev, xrcsrq_out = kvzalloc(MLX5_ST_SZ_BYTES(query_xrc_srq_out), GFP_KERNEL); if (!xrcsrq_out) return -ENOMEM; - memset(xrcsrq_in, 0, sizeof(xrcsrq_in)); - MLX5_SET(query_xrc_srq_in, xrcsrq_in, opcode, - MLX5_CMD_OP_QUERY_XRC_SRQ); - MLX5_SET(query_xrc_srq_in, xrcsrq_in, xrc_srqn, srq->srqn); + MLX5_SET(query_xrc_srq_in, in, opcode, MLX5_CMD_OP_QUERY_XRC_SRQ); + MLX5_SET(query_xrc_srq_in, in, xrc_srqn, srq->srqn); - err = mlx5_cmd_exec(dev->mdev, xrcsrq_in, sizeof(xrcsrq_in), - xrcsrq_out, MLX5_ST_SZ_BYTES(query_xrc_srq_out)); + err = mlx5_cmd_exec_inout(dev->mdev, query_xrc_srq, in, xrcsrq_out); if (err) goto out; @@ -341,13 +327,12 @@ out: static int destroy_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { - u32 in[MLX5_ST_SZ_DW(destroy_rmp_in)] = {}; - u32 out[MLX5_ST_SZ_DW(destroy_rmp_out)] = {}; + u32 in[MLX5_ST_SZ_DW(destroy_rmp_in)] = {}; MLX5_SET(destroy_rmp_in, in, opcode, MLX5_CMD_OP_DESTROY_RMP); MLX5_SET(destroy_rmp_in, in, rmpn, srq->srqn); MLX5_SET(destroy_rmp_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec_in(dev->mdev, destroy_rmp, in); } static int arm_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, @@ -384,7 +369,7 @@ static int arm_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(rmpc, rmpc, state, MLX5_RMPC_STATE_RDY); MLX5_SET(modify_rmp_in, in, opcode, MLX5_CMD_OP_MODIFY_RMP); - err = mlx5_cmd_exec(dev->mdev, in, inlen, out, outlen); + err = mlx5_cmd_exec_inout(dev->mdev, modify_rmp, in, out); out: kvfree(in); @@ -414,7 +399,7 @@ static int query_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(query_rmp_in, rmp_in, opcode, MLX5_CMD_OP_QUERY_RMP); MLX5_SET(query_rmp_in, rmp_in, rmpn, srq->srqn); - err = mlx5_cmd_exec(dev->mdev, rmp_in, inlen, rmp_out, outlen); + err = mlx5_cmd_exec_inout(dev->mdev, query_rmp, rmp_in, rmp_out); if (err) goto out; @@ -477,36 +462,34 @@ static int create_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, static int destroy_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { - u32 in[MLX5_ST_SZ_DW(destroy_xrq_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(destroy_xrq_out)] = {0}; + u32 in[MLX5_ST_SZ_DW(destroy_xrq_in)] = {}; MLX5_SET(destroy_xrq_in, in, opcode, MLX5_CMD_OP_DESTROY_XRQ); - MLX5_SET(destroy_xrq_in, in, xrqn, srq->srqn); + MLX5_SET(destroy_xrq_in, in, xrqn, srq->srqn); MLX5_SET(destroy_xrq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec_in(dev->mdev, destroy_xrq, in); } static int arm_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, u16 lwm) { - u32 out[MLX5_ST_SZ_DW(arm_rq_out)] = {0}; - u32 in[MLX5_ST_SZ_DW(arm_rq_in)] = {0}; + u32 in[MLX5_ST_SZ_DW(arm_rq_in)] = {}; - MLX5_SET(arm_rq_in, in, opcode, MLX5_CMD_OP_ARM_RQ); - MLX5_SET(arm_rq_in, in, op_mod, MLX5_ARM_RQ_IN_OP_MOD_XRQ); + MLX5_SET(arm_rq_in, in, opcode, MLX5_CMD_OP_ARM_RQ); + MLX5_SET(arm_rq_in, in, op_mod, MLX5_ARM_RQ_IN_OP_MOD_XRQ); MLX5_SET(arm_rq_in, in, srq_number, srq->srqn); - MLX5_SET(arm_rq_in, in, lwm, lwm); + MLX5_SET(arm_rq_in, in, lwm, lwm); MLX5_SET(arm_rq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec_in(dev->mdev, arm_rq, in); } static int query_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { - u32 in[MLX5_ST_SZ_DW(query_xrq_in)] = {0}; + u32 in[MLX5_ST_SZ_DW(query_xrq_in)] = {}; u32 *xrq_out; int outlen = MLX5_ST_SZ_BYTES(query_xrq_out); void *xrqc; @@ -519,7 +502,7 @@ static int query_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(query_xrq_in, in, opcode, MLX5_CMD_OP_QUERY_XRQ); MLX5_SET(query_xrq_in, in, xrqn, srq->srqn); - err = mlx5_cmd_exec(dev->mdev, in, sizeof(in), xrq_out, outlen); + err = mlx5_cmd_exec_inout(dev->mdev, query_xrq, in, xrq_out); if (err) goto out; diff --git a/drivers/infiniband/hw/mlx5/std_types.c b/drivers/infiniband/hw/mlx5/std_types.c new file mode 100644 index 000000000000..16145fda68d0 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/std_types.c @@ -0,0 +1,45 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include <rdma/uverbs_ioctl.h> +#include <rdma/mlx5_user_ioctl_cmds.h> +#include <rdma/mlx5_user_ioctl_verbs.h> +#include <linux/mlx5/driver.h> +#include "mlx5_ib.h" + +#define UVERBS_MODULE_NAME mlx5_ib +#include <rdma/uverbs_named_ioctl.h> + +static int UVERBS_HANDLER(MLX5_IB_METHOD_PD_QUERY)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_pd *pd = + uverbs_attr_get_obj(attrs, MLX5_IB_ATTR_QUERY_PD_HANDLE); + struct mlx5_ib_pd *mpd = to_mpd(pd); + + return uverbs_copy_to(attrs, MLX5_IB_ATTR_QUERY_PD_RESP_PDN, + &mpd->pdn, sizeof(mpd->pdn)); +} + +DECLARE_UVERBS_NAMED_METHOD( + MLX5_IB_METHOD_PD_QUERY, + UVERBS_ATTR_IDR(MLX5_IB_ATTR_QUERY_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_QUERY_PD_RESP_PDN, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY)); + +ADD_UVERBS_METHODS(mlx5_ib_pd, + UVERBS_OBJECT_PD, + &UVERBS_METHOD(MLX5_IB_METHOD_PD_QUERY)); + +const struct uapi_definition mlx5_ib_std_types_defs[] = { + UAPI_DEF_CHAIN_OBJ_TREE( + UVERBS_OBJECT_PD, + &mlx5_ib_pd), + {}, +}; diff --git a/drivers/infiniband/hw/mlx5/wr.c b/drivers/infiniband/hw/mlx5/wr.c new file mode 100644 index 000000000000..43880973a512 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/wr.c @@ -0,0 +1,1536 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#include <linux/gfp.h> +#include <linux/mlx5/qp.h> +#include <linux/mlx5/driver.h> +#include "wr.h" + +static const u32 mlx5_ib_opcode[] = { + [IB_WR_SEND] = MLX5_OPCODE_SEND, + [IB_WR_LSO] = MLX5_OPCODE_LSO, + [IB_WR_SEND_WITH_IMM] = MLX5_OPCODE_SEND_IMM, + [IB_WR_RDMA_WRITE] = MLX5_OPCODE_RDMA_WRITE, + [IB_WR_RDMA_WRITE_WITH_IMM] = MLX5_OPCODE_RDMA_WRITE_IMM, + [IB_WR_RDMA_READ] = MLX5_OPCODE_RDMA_READ, + [IB_WR_ATOMIC_CMP_AND_SWP] = MLX5_OPCODE_ATOMIC_CS, + [IB_WR_ATOMIC_FETCH_AND_ADD] = MLX5_OPCODE_ATOMIC_FA, + [IB_WR_SEND_WITH_INV] = MLX5_OPCODE_SEND_INVAL, + [IB_WR_LOCAL_INV] = MLX5_OPCODE_UMR, + [IB_WR_REG_MR] = MLX5_OPCODE_UMR, + [IB_WR_MASKED_ATOMIC_CMP_AND_SWP] = MLX5_OPCODE_ATOMIC_MASKED_CS, + [IB_WR_MASKED_ATOMIC_FETCH_AND_ADD] = MLX5_OPCODE_ATOMIC_MASKED_FA, + [MLX5_IB_WR_UMR] = MLX5_OPCODE_UMR, +}; + +/* handle_post_send_edge - Check if we get to SQ edge. If yes, update to the + * next nearby edge and get new address translation for current WQE position. + * @sq - SQ buffer. + * @seg: Current WQE position (16B aligned). + * @wqe_sz: Total current WQE size [16B]. + * @cur_edge: Updated current edge. + */ +static inline void handle_post_send_edge(struct mlx5_ib_wq *sq, void **seg, + u32 wqe_sz, void **cur_edge) +{ + u32 idx; + + if (likely(*seg != *cur_edge)) + return; + + idx = (sq->cur_post + (wqe_sz >> 2)) & (sq->wqe_cnt - 1); + *cur_edge = get_sq_edge(sq, idx); + + *seg = mlx5_frag_buf_get_wqe(&sq->fbc, idx); +} + +/* memcpy_send_wqe - copy data from src to WQE and update the relevant WQ's + * pointers. At the end @seg is aligned to 16B regardless the copied size. + * @sq - SQ buffer. + * @cur_edge: Updated current edge. + * @seg: Current WQE position (16B aligned). + * @wqe_sz: Total current WQE size [16B]. + * @src: Pointer to copy from. + * @n: Number of bytes to copy. + */ +static inline void memcpy_send_wqe(struct mlx5_ib_wq *sq, void **cur_edge, + void **seg, u32 *wqe_sz, const void *src, + size_t n) +{ + while (likely(n)) { + size_t leftlen = *cur_edge - *seg; + size_t copysz = min_t(size_t, leftlen, n); + size_t stride; + + memcpy(*seg, src, copysz); + + n -= copysz; + src += copysz; + stride = !n ? ALIGN(copysz, 16) : copysz; + *seg += stride; + *wqe_sz += stride >> 4; + handle_post_send_edge(sq, seg, *wqe_sz, cur_edge); + } +} + +static int mlx5_wq_overflow(struct mlx5_ib_wq *wq, int nreq, + struct ib_cq *ib_cq) +{ + struct mlx5_ib_cq *cq; + unsigned int cur; + + cur = wq->head - wq->tail; + if (likely(cur + nreq < wq->max_post)) + return 0; + + cq = to_mcq(ib_cq); + spin_lock(&cq->lock); + cur = wq->head - wq->tail; + spin_unlock(&cq->lock); + + return cur + nreq >= wq->max_post; +} + +static __always_inline void set_raddr_seg(struct mlx5_wqe_raddr_seg *rseg, + u64 remote_addr, u32 rkey) +{ + rseg->raddr = cpu_to_be64(remote_addr); + rseg->rkey = cpu_to_be32(rkey); + rseg->reserved = 0; +} + +static void set_eth_seg(const struct ib_send_wr *wr, struct mlx5_ib_qp *qp, + void **seg, int *size, void **cur_edge) +{ + struct mlx5_wqe_eth_seg *eseg = *seg; + + memset(eseg, 0, sizeof(struct mlx5_wqe_eth_seg)); + + if (wr->send_flags & IB_SEND_IP_CSUM) + eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM | + MLX5_ETH_WQE_L4_CSUM; + + if (wr->opcode == IB_WR_LSO) { + struct ib_ud_wr *ud_wr = container_of(wr, struct ib_ud_wr, wr); + size_t left, copysz; + void *pdata = ud_wr->header; + size_t stride; + + left = ud_wr->hlen; + eseg->mss = cpu_to_be16(ud_wr->mss); + eseg->inline_hdr.sz = cpu_to_be16(left); + + /* memcpy_send_wqe should get a 16B align address. Hence, we + * first copy up to the current edge and then, if needed, + * continue to memcpy_send_wqe. + */ + copysz = min_t(u64, *cur_edge - (void *)eseg->inline_hdr.start, + left); + memcpy(eseg->inline_hdr.start, pdata, copysz); + stride = ALIGN(sizeof(struct mlx5_wqe_eth_seg) - + sizeof(eseg->inline_hdr.start) + copysz, 16); + *size += stride / 16; + *seg += stride; + + if (copysz < left) { + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + left -= copysz; + pdata += copysz; + memcpy_send_wqe(&qp->sq, cur_edge, seg, size, pdata, + left); + } + + return; + } + + *seg += sizeof(struct mlx5_wqe_eth_seg); + *size += sizeof(struct mlx5_wqe_eth_seg) / 16; +} + +static void set_datagram_seg(struct mlx5_wqe_datagram_seg *dseg, + const struct ib_send_wr *wr) +{ + memcpy(&dseg->av, &to_mah(ud_wr(wr)->ah)->av, sizeof(struct mlx5_av)); + dseg->av.dqp_dct = + cpu_to_be32(ud_wr(wr)->remote_qpn | MLX5_EXTENDED_UD_AV); + dseg->av.key.qkey.qkey = cpu_to_be32(ud_wr(wr)->remote_qkey); +} + +static void set_data_ptr_seg(struct mlx5_wqe_data_seg *dseg, struct ib_sge *sg) +{ + dseg->byte_count = cpu_to_be32(sg->length); + dseg->lkey = cpu_to_be32(sg->lkey); + dseg->addr = cpu_to_be64(sg->addr); +} + +static u64 get_xlt_octo(u64 bytes) +{ + return ALIGN(bytes, MLX5_IB_UMR_XLT_ALIGNMENT) / + MLX5_IB_UMR_OCTOWORD; +} + +static __be64 frwr_mkey_mask(bool atomic) +{ + u64 result; + + result = MLX5_MKEY_MASK_LEN | + MLX5_MKEY_MASK_PAGE_SIZE | + MLX5_MKEY_MASK_START_ADDR | + MLX5_MKEY_MASK_EN_RINVAL | + MLX5_MKEY_MASK_KEY | + MLX5_MKEY_MASK_LR | + MLX5_MKEY_MASK_LW | + MLX5_MKEY_MASK_RR | + MLX5_MKEY_MASK_RW | + MLX5_MKEY_MASK_SMALL_FENCE | + MLX5_MKEY_MASK_FREE; + + if (atomic) + result |= MLX5_MKEY_MASK_A; + + return cpu_to_be64(result); +} + +static __be64 sig_mkey_mask(void) +{ + u64 result; + + result = MLX5_MKEY_MASK_LEN | + MLX5_MKEY_MASK_PAGE_SIZE | + MLX5_MKEY_MASK_START_ADDR | + MLX5_MKEY_MASK_EN_SIGERR | + MLX5_MKEY_MASK_EN_RINVAL | + MLX5_MKEY_MASK_KEY | + MLX5_MKEY_MASK_LR | + MLX5_MKEY_MASK_LW | + MLX5_MKEY_MASK_RR | + MLX5_MKEY_MASK_RW | + MLX5_MKEY_MASK_SMALL_FENCE | + MLX5_MKEY_MASK_FREE | + MLX5_MKEY_MASK_BSF_EN; + + return cpu_to_be64(result); +} + +static void set_reg_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr, + struct mlx5_ib_mr *mr, u8 flags, bool atomic) +{ + int size = (mr->ndescs + mr->meta_ndescs) * mr->desc_size; + + memset(umr, 0, sizeof(*umr)); + + umr->flags = flags; + umr->xlt_octowords = cpu_to_be16(get_xlt_octo(size)); + umr->mkey_mask = frwr_mkey_mask(atomic); +} + +static void set_linv_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr) +{ + memset(umr, 0, sizeof(*umr)); + umr->mkey_mask = cpu_to_be64(MLX5_MKEY_MASK_FREE); + umr->flags = MLX5_UMR_INLINE; +} + +static __be64 get_umr_enable_mr_mask(void) +{ + u64 result; + + result = MLX5_MKEY_MASK_KEY | + MLX5_MKEY_MASK_FREE; + + return cpu_to_be64(result); +} + +static __be64 get_umr_disable_mr_mask(void) +{ + u64 result; + + result = MLX5_MKEY_MASK_FREE; + + return cpu_to_be64(result); +} + +static __be64 get_umr_update_translation_mask(void) +{ + u64 result; + + result = MLX5_MKEY_MASK_LEN | + MLX5_MKEY_MASK_PAGE_SIZE | + MLX5_MKEY_MASK_START_ADDR; + + return cpu_to_be64(result); +} + +static __be64 get_umr_update_access_mask(int atomic, + int relaxed_ordering_write, + int relaxed_ordering_read) +{ + u64 result; + + result = MLX5_MKEY_MASK_LR | + MLX5_MKEY_MASK_LW | + MLX5_MKEY_MASK_RR | + MLX5_MKEY_MASK_RW; + + if (atomic) + result |= MLX5_MKEY_MASK_A; + + if (relaxed_ordering_write) + result |= MLX5_MKEY_MASK_RELAXED_ORDERING_WRITE; + + if (relaxed_ordering_read) + result |= MLX5_MKEY_MASK_RELAXED_ORDERING_READ; + + return cpu_to_be64(result); +} + +static __be64 get_umr_update_pd_mask(void) +{ + u64 result; + + result = MLX5_MKEY_MASK_PD; + + return cpu_to_be64(result); +} + +static int umr_check_mkey_mask(struct mlx5_ib_dev *dev, u64 mask) +{ + if (mask & MLX5_MKEY_MASK_PAGE_SIZE && + MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled)) + return -EPERM; + + if (mask & MLX5_MKEY_MASK_A && + MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled)) + return -EPERM; + + if (mask & MLX5_MKEY_MASK_RELAXED_ORDERING_WRITE && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write_umr)) + return -EPERM; + + if (mask & MLX5_MKEY_MASK_RELAXED_ORDERING_READ && + !MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr)) + return -EPERM; + + return 0; +} + +static int set_reg_umr_segment(struct mlx5_ib_dev *dev, + struct mlx5_wqe_umr_ctrl_seg *umr, + const struct ib_send_wr *wr) +{ + const struct mlx5_umr_wr *umrwr = umr_wr(wr); + + memset(umr, 0, sizeof(*umr)); + + if (!umrwr->ignore_free_state) { + if (wr->send_flags & MLX5_IB_SEND_UMR_FAIL_IF_FREE) + /* fail if free */ + umr->flags = MLX5_UMR_CHECK_FREE; + else + /* fail if not free */ + umr->flags = MLX5_UMR_CHECK_NOT_FREE; + } + + umr->xlt_octowords = cpu_to_be16(get_xlt_octo(umrwr->xlt_size)); + if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_XLT) { + u64 offset = get_xlt_octo(umrwr->offset); + + umr->xlt_offset = cpu_to_be16(offset & 0xffff); + umr->xlt_offset_47_16 = cpu_to_be32(offset >> 16); + umr->flags |= MLX5_UMR_TRANSLATION_OFFSET_EN; + } + if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION) + umr->mkey_mask |= get_umr_update_translation_mask(); + if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_PD_ACCESS) { + umr->mkey_mask |= get_umr_update_access_mask( + !!(MLX5_CAP_GEN(dev->mdev, atomic)), + !!(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_write_umr)), + !!(MLX5_CAP_GEN(dev->mdev, relaxed_ordering_read_umr))); + umr->mkey_mask |= get_umr_update_pd_mask(); + } + if (wr->send_flags & MLX5_IB_SEND_UMR_ENABLE_MR) + umr->mkey_mask |= get_umr_enable_mr_mask(); + if (wr->send_flags & MLX5_IB_SEND_UMR_DISABLE_MR) + umr->mkey_mask |= get_umr_disable_mr_mask(); + + if (!wr->num_sge) + umr->flags |= MLX5_UMR_INLINE; + + return umr_check_mkey_mask(dev, be64_to_cpu(umr->mkey_mask)); +} + +static u8 get_umr_flags(int acc) +{ + return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX5_PERM_ATOMIC : 0) | + (acc & IB_ACCESS_REMOTE_WRITE ? MLX5_PERM_REMOTE_WRITE : 0) | + (acc & IB_ACCESS_REMOTE_READ ? MLX5_PERM_REMOTE_READ : 0) | + (acc & IB_ACCESS_LOCAL_WRITE ? MLX5_PERM_LOCAL_WRITE : 0) | + MLX5_PERM_LOCAL_READ | MLX5_PERM_UMR_EN; +} + +static void set_reg_mkey_seg(struct mlx5_mkey_seg *seg, + struct mlx5_ib_mr *mr, + u32 key, int access) +{ + int ndescs = ALIGN(mr->ndescs + mr->meta_ndescs, 8) >> 1; + + memset(seg, 0, sizeof(*seg)); + + if (mr->access_mode == MLX5_MKC_ACCESS_MODE_MTT) + seg->log2_page_size = ilog2(mr->ibmr.page_size); + else if (mr->access_mode == MLX5_MKC_ACCESS_MODE_KLMS) + /* KLMs take twice the size of MTTs */ + ndescs *= 2; + + seg->flags = get_umr_flags(access) | mr->access_mode; + seg->qpn_mkey7_0 = cpu_to_be32((key & 0xff) | 0xffffff00); + seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL); + seg->start_addr = cpu_to_be64(mr->ibmr.iova); + seg->len = cpu_to_be64(mr->ibmr.length); + seg->xlt_oct_size = cpu_to_be32(ndescs); +} + +static void set_linv_mkey_seg(struct mlx5_mkey_seg *seg) +{ + memset(seg, 0, sizeof(*seg)); + seg->status = MLX5_MKEY_STATUS_FREE; +} + +static void set_reg_mkey_segment(struct mlx5_mkey_seg *seg, + const struct ib_send_wr *wr) +{ + const struct mlx5_umr_wr *umrwr = umr_wr(wr); + + memset(seg, 0, sizeof(*seg)); + if (wr->send_flags & MLX5_IB_SEND_UMR_DISABLE_MR) + MLX5_SET(mkc, seg, free, 1); + + MLX5_SET(mkc, seg, a, + !!(umrwr->access_flags & IB_ACCESS_REMOTE_ATOMIC)); + MLX5_SET(mkc, seg, rw, + !!(umrwr->access_flags & IB_ACCESS_REMOTE_WRITE)); + MLX5_SET(mkc, seg, rr, !!(umrwr->access_flags & IB_ACCESS_REMOTE_READ)); + MLX5_SET(mkc, seg, lw, !!(umrwr->access_flags & IB_ACCESS_LOCAL_WRITE)); + MLX5_SET(mkc, seg, lr, 1); + MLX5_SET(mkc, seg, relaxed_ordering_write, + !!(umrwr->access_flags & IB_ACCESS_RELAXED_ORDERING)); + MLX5_SET(mkc, seg, relaxed_ordering_read, + !!(umrwr->access_flags & IB_ACCESS_RELAXED_ORDERING)); + + if (umrwr->pd) + MLX5_SET(mkc, seg, pd, to_mpd(umrwr->pd)->pdn); + if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_TRANSLATION && + !umrwr->length) + MLX5_SET(mkc, seg, length64, 1); + + MLX5_SET64(mkc, seg, start_addr, umrwr->virt_addr); + MLX5_SET64(mkc, seg, len, umrwr->length); + MLX5_SET(mkc, seg, log_page_size, umrwr->page_shift); + MLX5_SET(mkc, seg, qpn, 0xffffff); + MLX5_SET(mkc, seg, mkey_7_0, mlx5_mkey_variant(umrwr->mkey)); +} + +static void set_reg_data_seg(struct mlx5_wqe_data_seg *dseg, + struct mlx5_ib_mr *mr, + struct mlx5_ib_pd *pd) +{ + int bcount = mr->desc_size * (mr->ndescs + mr->meta_ndescs); + + dseg->addr = cpu_to_be64(mr->desc_map); + dseg->byte_count = cpu_to_be32(ALIGN(bcount, 64)); + dseg->lkey = cpu_to_be32(pd->ibpd.local_dma_lkey); +} + +static __be32 send_ieth(const struct ib_send_wr *wr) +{ + switch (wr->opcode) { + case IB_WR_SEND_WITH_IMM: + case IB_WR_RDMA_WRITE_WITH_IMM: + return wr->ex.imm_data; + + case IB_WR_SEND_WITH_INV: + return cpu_to_be32(wr->ex.invalidate_rkey); + + default: + return 0; + } +} + +static u8 calc_sig(void *wqe, int size) +{ + u8 *p = wqe; + u8 res = 0; + int i; + + for (i = 0; i < size; i++) + res ^= p[i]; + + return ~res; +} + +static u8 wq_sig(void *wqe) +{ + return calc_sig(wqe, (*((u8 *)wqe + 8) & 0x3f) << 4); +} + +static int set_data_inl_seg(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, + void **wqe, int *wqe_sz, void **cur_edge) +{ + struct mlx5_wqe_inline_seg *seg; + size_t offset; + int inl = 0; + int i; + + seg = *wqe; + *wqe += sizeof(*seg); + offset = sizeof(*seg); + + for (i = 0; i < wr->num_sge; i++) { + size_t len = wr->sg_list[i].length; + void *addr = (void *)(unsigned long)(wr->sg_list[i].addr); + + inl += len; + + if (unlikely(inl > qp->max_inline_data)) + return -ENOMEM; + + while (likely(len)) { + size_t leftlen; + size_t copysz; + + handle_post_send_edge(&qp->sq, wqe, + *wqe_sz + (offset >> 4), + cur_edge); + + leftlen = *cur_edge - *wqe; + copysz = min_t(size_t, leftlen, len); + + memcpy(*wqe, addr, copysz); + len -= copysz; + addr += copysz; + *wqe += copysz; + offset += copysz; + } + } + + seg->byte_count = cpu_to_be32(inl | MLX5_INLINE_SEG); + + *wqe_sz += ALIGN(inl + sizeof(seg->byte_count), 16) / 16; + + return 0; +} + +static u16 prot_field_size(enum ib_signature_type type) +{ + switch (type) { + case IB_SIG_TYPE_T10_DIF: + return MLX5_DIF_SIZE; + default: + return 0; + } +} + +static u8 bs_selector(int block_size) +{ + switch (block_size) { + case 512: return 0x1; + case 520: return 0x2; + case 4096: return 0x3; + case 4160: return 0x4; + case 1073741824: return 0x5; + default: return 0; + } +} + +static void mlx5_fill_inl_bsf(struct ib_sig_domain *domain, + struct mlx5_bsf_inl *inl) +{ + /* Valid inline section and allow BSF refresh */ + inl->vld_refresh = cpu_to_be16(MLX5_BSF_INL_VALID | + MLX5_BSF_REFRESH_DIF); + inl->dif_apptag = cpu_to_be16(domain->sig.dif.app_tag); + inl->dif_reftag = cpu_to_be32(domain->sig.dif.ref_tag); + /* repeating block */ + inl->rp_inv_seed = MLX5_BSF_REPEAT_BLOCK; + inl->sig_type = domain->sig.dif.bg_type == IB_T10DIF_CRC ? + MLX5_DIF_CRC : MLX5_DIF_IPCS; + + if (domain->sig.dif.ref_remap) + inl->dif_inc_ref_guard_check |= MLX5_BSF_INC_REFTAG; + + if (domain->sig.dif.app_escape) { + if (domain->sig.dif.ref_escape) + inl->dif_inc_ref_guard_check |= MLX5_BSF_APPREF_ESCAPE; + else + inl->dif_inc_ref_guard_check |= MLX5_BSF_APPTAG_ESCAPE; + } + + inl->dif_app_bitmask_check = + cpu_to_be16(domain->sig.dif.apptag_check_mask); +} + +static int mlx5_set_bsf(struct ib_mr *sig_mr, + struct ib_sig_attrs *sig_attrs, + struct mlx5_bsf *bsf, u32 data_size) +{ + struct mlx5_core_sig_ctx *msig = to_mmr(sig_mr)->sig; + struct mlx5_bsf_basic *basic = &bsf->basic; + struct ib_sig_domain *mem = &sig_attrs->mem; + struct ib_sig_domain *wire = &sig_attrs->wire; + + memset(bsf, 0, sizeof(*bsf)); + + /* Basic + Extended + Inline */ + basic->bsf_size_sbs = 1 << 7; + /* Input domain check byte mask */ + basic->check_byte_mask = sig_attrs->check_mask; + basic->raw_data_size = cpu_to_be32(data_size); + + /* Memory domain */ + switch (sig_attrs->mem.sig_type) { + case IB_SIG_TYPE_NONE: + break; + case IB_SIG_TYPE_T10_DIF: + basic->mem.bs_selector = bs_selector(mem->sig.dif.pi_interval); + basic->m_bfs_psv = cpu_to_be32(msig->psv_memory.psv_idx); + mlx5_fill_inl_bsf(mem, &bsf->m_inl); + break; + default: + return -EINVAL; + } + + /* Wire domain */ + switch (sig_attrs->wire.sig_type) { + case IB_SIG_TYPE_NONE: + break; + case IB_SIG_TYPE_T10_DIF: + if (mem->sig.dif.pi_interval == wire->sig.dif.pi_interval && + mem->sig_type == wire->sig_type) { + /* Same block structure */ + basic->bsf_size_sbs |= 1 << 4; + if (mem->sig.dif.bg_type == wire->sig.dif.bg_type) + basic->wire.copy_byte_mask |= MLX5_CPY_GRD_MASK; + if (mem->sig.dif.app_tag == wire->sig.dif.app_tag) + basic->wire.copy_byte_mask |= MLX5_CPY_APP_MASK; + if (mem->sig.dif.ref_tag == wire->sig.dif.ref_tag) + basic->wire.copy_byte_mask |= MLX5_CPY_REF_MASK; + } else + basic->wire.bs_selector = + bs_selector(wire->sig.dif.pi_interval); + + basic->w_bfs_psv = cpu_to_be32(msig->psv_wire.psv_idx); + mlx5_fill_inl_bsf(wire, &bsf->w_inl); + break; + default: + return -EINVAL; + } + + return 0; +} + + +static int set_sig_data_segment(const struct ib_send_wr *send_wr, + struct ib_mr *sig_mr, + struct ib_sig_attrs *sig_attrs, + struct mlx5_ib_qp *qp, void **seg, int *size, + void **cur_edge) +{ + struct mlx5_bsf *bsf; + u32 data_len; + u32 data_key; + u64 data_va; + u32 prot_len = 0; + u32 prot_key = 0; + u64 prot_va = 0; + bool prot = false; + int ret; + int wqe_size; + struct mlx5_ib_mr *mr = to_mmr(sig_mr); + struct mlx5_ib_mr *pi_mr = mr->pi_mr; + + data_len = pi_mr->data_length; + data_key = pi_mr->ibmr.lkey; + data_va = pi_mr->data_iova; + if (pi_mr->meta_ndescs) { + prot_len = pi_mr->meta_length; + prot_key = pi_mr->ibmr.lkey; + prot_va = pi_mr->pi_iova; + prot = true; + } + + if (!prot || (data_key == prot_key && data_va == prot_va && + data_len == prot_len)) { + /** + * Source domain doesn't contain signature information + * or data and protection are interleaved in memory. + * So need construct: + * ------------------ + * | data_klm | + * ------------------ + * | BSF | + * ------------------ + **/ + struct mlx5_klm *data_klm = *seg; + + data_klm->bcount = cpu_to_be32(data_len); + data_klm->key = cpu_to_be32(data_key); + data_klm->va = cpu_to_be64(data_va); + wqe_size = ALIGN(sizeof(*data_klm), 64); + } else { + /** + * Source domain contains signature information + * So need construct a strided block format: + * --------------------------- + * | stride_block_ctrl | + * --------------------------- + * | data_klm | + * --------------------------- + * | prot_klm | + * --------------------------- + * | BSF | + * --------------------------- + **/ + struct mlx5_stride_block_ctrl_seg *sblock_ctrl; + struct mlx5_stride_block_entry *data_sentry; + struct mlx5_stride_block_entry *prot_sentry; + u16 block_size = sig_attrs->mem.sig.dif.pi_interval; + int prot_size; + + sblock_ctrl = *seg; + data_sentry = (void *)sblock_ctrl + sizeof(*sblock_ctrl); + prot_sentry = (void *)data_sentry + sizeof(*data_sentry); + + prot_size = prot_field_size(sig_attrs->mem.sig_type); + if (!prot_size) { + pr_err("Bad block size given: %u\n", block_size); + return -EINVAL; + } + sblock_ctrl->bcount_per_cycle = cpu_to_be32(block_size + + prot_size); + sblock_ctrl->op = cpu_to_be32(MLX5_STRIDE_BLOCK_OP); + sblock_ctrl->repeat_count = cpu_to_be32(data_len / block_size); + sblock_ctrl->num_entries = cpu_to_be16(2); + + data_sentry->bcount = cpu_to_be16(block_size); + data_sentry->key = cpu_to_be32(data_key); + data_sentry->va = cpu_to_be64(data_va); + data_sentry->stride = cpu_to_be16(block_size); + + prot_sentry->bcount = cpu_to_be16(prot_size); + prot_sentry->key = cpu_to_be32(prot_key); + prot_sentry->va = cpu_to_be64(prot_va); + prot_sentry->stride = cpu_to_be16(prot_size); + + wqe_size = ALIGN(sizeof(*sblock_ctrl) + sizeof(*data_sentry) + + sizeof(*prot_sentry), 64); + } + + *seg += wqe_size; + *size += wqe_size / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + bsf = *seg; + ret = mlx5_set_bsf(sig_mr, sig_attrs, bsf, data_len); + if (ret) + return -EINVAL; + + *seg += sizeof(*bsf); + *size += sizeof(*bsf) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + return 0; +} + +static void set_sig_mkey_segment(struct mlx5_mkey_seg *seg, + struct ib_mr *sig_mr, int access_flags, + u32 size, u32 length, u32 pdn) +{ + u32 sig_key = sig_mr->rkey; + u8 sigerr = to_mmr(sig_mr)->sig->sigerr_count & 1; + + memset(seg, 0, sizeof(*seg)); + + seg->flags = get_umr_flags(access_flags) | MLX5_MKC_ACCESS_MODE_KLMS; + seg->qpn_mkey7_0 = cpu_to_be32((sig_key & 0xff) | 0xffffff00); + seg->flags_pd = cpu_to_be32(MLX5_MKEY_REMOTE_INVAL | sigerr << 26 | + MLX5_MKEY_BSF_EN | pdn); + seg->len = cpu_to_be64(length); + seg->xlt_oct_size = cpu_to_be32(get_xlt_octo(size)); + seg->bsfs_octo_size = cpu_to_be32(MLX5_MKEY_BSF_OCTO_SIZE); +} + +static void set_sig_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr, + u32 size) +{ + memset(umr, 0, sizeof(*umr)); + + umr->flags = MLX5_FLAGS_INLINE | MLX5_FLAGS_CHECK_FREE; + umr->xlt_octowords = cpu_to_be16(get_xlt_octo(size)); + umr->bsf_octowords = cpu_to_be16(MLX5_MKEY_BSF_OCTO_SIZE); + umr->mkey_mask = sig_mkey_mask(); +} + +static int set_pi_umr_wr(const struct ib_send_wr *send_wr, + struct mlx5_ib_qp *qp, void **seg, int *size, + void **cur_edge) +{ + const struct ib_reg_wr *wr = reg_wr(send_wr); + struct mlx5_ib_mr *sig_mr = to_mmr(wr->mr); + struct mlx5_ib_mr *pi_mr = sig_mr->pi_mr; + struct ib_sig_attrs *sig_attrs = sig_mr->ibmr.sig_attrs; + u32 pdn = to_mpd(qp->ibqp.pd)->pdn; + u32 xlt_size; + int region_len, ret; + + if (unlikely(send_wr->num_sge != 0) || + unlikely(wr->access & IB_ACCESS_REMOTE_ATOMIC) || + unlikely(!sig_mr->sig) || unlikely(!qp->ibqp.integrity_en) || + unlikely(!sig_mr->sig->sig_status_checked)) + return -EINVAL; + + /* length of the protected region, data + protection */ + region_len = pi_mr->ibmr.length; + + /** + * KLM octoword size - if protection was provided + * then we use strided block format (3 octowords), + * else we use single KLM (1 octoword) + **/ + if (sig_attrs->mem.sig_type != IB_SIG_TYPE_NONE) + xlt_size = 0x30; + else + xlt_size = sizeof(struct mlx5_klm); + + set_sig_umr_segment(*seg, xlt_size); + *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); + *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + set_sig_mkey_segment(*seg, wr->mr, wr->access, xlt_size, region_len, + pdn); + *seg += sizeof(struct mlx5_mkey_seg); + *size += sizeof(struct mlx5_mkey_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + ret = set_sig_data_segment(send_wr, wr->mr, sig_attrs, qp, seg, size, + cur_edge); + if (ret) + return ret; + + sig_mr->sig->sig_status_checked = false; + return 0; +} + +static int set_psv_wr(struct ib_sig_domain *domain, + u32 psv_idx, void **seg, int *size) +{ + struct mlx5_seg_set_psv *psv_seg = *seg; + + memset(psv_seg, 0, sizeof(*psv_seg)); + psv_seg->psv_num = cpu_to_be32(psv_idx); + switch (domain->sig_type) { + case IB_SIG_TYPE_NONE: + break; + case IB_SIG_TYPE_T10_DIF: + psv_seg->transient_sig = cpu_to_be32(domain->sig.dif.bg << 16 | + domain->sig.dif.app_tag); + psv_seg->ref_tag = cpu_to_be32(domain->sig.dif.ref_tag); + break; + default: + pr_err("Bad signature type (%d) is given.\n", + domain->sig_type); + return -EINVAL; + } + + *seg += sizeof(*psv_seg); + *size += sizeof(*psv_seg) / 16; + + return 0; +} + +static int set_reg_wr(struct mlx5_ib_qp *qp, + const struct ib_reg_wr *wr, + void **seg, int *size, void **cur_edge, + bool check_not_free) +{ + struct mlx5_ib_mr *mr = to_mmr(wr->mr); + struct mlx5_ib_pd *pd = to_mpd(qp->ibqp.pd); + struct mlx5_ib_dev *dev = to_mdev(pd->ibpd.device); + int mr_list_size = (mr->ndescs + mr->meta_ndescs) * mr->desc_size; + bool umr_inline = mr_list_size <= MLX5_IB_SQ_UMR_INLINE_THRESHOLD; + bool atomic = wr->access & IB_ACCESS_REMOTE_ATOMIC; + u8 flags = 0; + + if (!mlx5_ib_can_use_umr(dev, atomic, wr->access)) { + mlx5_ib_warn(to_mdev(qp->ibqp.device), + "Fast update of %s for MR is disabled\n", + (MLX5_CAP_GEN(dev->mdev, + umr_modify_entity_size_disabled)) ? + "entity size" : + "atomic access"); + return -EINVAL; + } + + if (unlikely(wr->wr.send_flags & IB_SEND_INLINE)) { + mlx5_ib_warn(to_mdev(qp->ibqp.device), + "Invalid IB_SEND_INLINE send flag\n"); + return -EINVAL; + } + + if (check_not_free) + flags |= MLX5_UMR_CHECK_NOT_FREE; + if (umr_inline) + flags |= MLX5_UMR_INLINE; + + set_reg_umr_seg(*seg, mr, flags, atomic); + *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); + *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + set_reg_mkey_seg(*seg, mr, wr->key, wr->access); + *seg += sizeof(struct mlx5_mkey_seg); + *size += sizeof(struct mlx5_mkey_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + if (umr_inline) { + memcpy_send_wqe(&qp->sq, cur_edge, seg, size, mr->descs, + mr_list_size); + *size = ALIGN(*size, MLX5_SEND_WQE_BB >> 4); + } else { + set_reg_data_seg(*seg, mr, pd); + *seg += sizeof(struct mlx5_wqe_data_seg); + *size += (sizeof(struct mlx5_wqe_data_seg) / 16); + } + return 0; +} + +static void set_linv_wr(struct mlx5_ib_qp *qp, void **seg, int *size, + void **cur_edge) +{ + set_linv_umr_seg(*seg); + *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); + *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + set_linv_mkey_seg(*seg); + *seg += sizeof(struct mlx5_mkey_seg); + *size += sizeof(struct mlx5_mkey_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); +} + +static void dump_wqe(struct mlx5_ib_qp *qp, u32 idx, int size_16) +{ + __be32 *p = NULL; + int i, j; + + pr_debug("dump WQE index %u:\n", idx); + for (i = 0, j = 0; i < size_16 * 4; i += 4, j += 4) { + if ((i & 0xf) == 0) { + p = mlx5_frag_buf_get_wqe(&qp->sq.fbc, idx); + pr_debug("WQBB at %p:\n", (void *)p); + j = 0; + idx = (idx + 1) & (qp->sq.wqe_cnt - 1); + } + pr_debug("%08x %08x %08x %08x\n", be32_to_cpu(p[j]), + be32_to_cpu(p[j + 1]), be32_to_cpu(p[j + 2]), + be32_to_cpu(p[j + 3])); + } +} + +static int __begin_wqe(struct mlx5_ib_qp *qp, void **seg, + struct mlx5_wqe_ctrl_seg **ctrl, + const struct ib_send_wr *wr, unsigned int *idx, + int *size, void **cur_edge, int nreq, + bool send_signaled, bool solicited) +{ + if (unlikely(mlx5_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq))) + return -ENOMEM; + + *idx = qp->sq.cur_post & (qp->sq.wqe_cnt - 1); + *seg = mlx5_frag_buf_get_wqe(&qp->sq.fbc, *idx); + *ctrl = *seg; + *(uint32_t *)(*seg + 8) = 0; + (*ctrl)->imm = send_ieth(wr); + (*ctrl)->fm_ce_se = qp->sq_signal_bits | + (send_signaled ? MLX5_WQE_CTRL_CQ_UPDATE : 0) | + (solicited ? MLX5_WQE_CTRL_SOLICITED : 0); + + *seg += sizeof(**ctrl); + *size = sizeof(**ctrl) / 16; + *cur_edge = qp->sq.cur_edge; + + return 0; +} + +static int begin_wqe(struct mlx5_ib_qp *qp, void **seg, + struct mlx5_wqe_ctrl_seg **ctrl, + const struct ib_send_wr *wr, unsigned int *idx, int *size, + void **cur_edge, int nreq) +{ + return __begin_wqe(qp, seg, ctrl, wr, idx, size, cur_edge, nreq, + wr->send_flags & IB_SEND_SIGNALED, + wr->send_flags & IB_SEND_SOLICITED); +} + +static void finish_wqe(struct mlx5_ib_qp *qp, + struct mlx5_wqe_ctrl_seg *ctrl, + void *seg, u8 size, void *cur_edge, + unsigned int idx, u64 wr_id, int nreq, u8 fence, + u32 mlx5_opcode) +{ + u8 opmod = 0; + + ctrl->opmod_idx_opcode = cpu_to_be32(((u32)(qp->sq.cur_post) << 8) | + mlx5_opcode | ((u32)opmod << 24)); + ctrl->qpn_ds = cpu_to_be32(size | (qp->trans_qp.base.mqp.qpn << 8)); + ctrl->fm_ce_se |= fence; + if (unlikely(qp->flags_en & MLX5_QP_FLAG_SIGNATURE)) + ctrl->signature = wq_sig(ctrl); + + qp->sq.wrid[idx] = wr_id; + qp->sq.w_list[idx].opcode = mlx5_opcode; + qp->sq.wqe_head[idx] = qp->sq.head + nreq; + qp->sq.cur_post += DIV_ROUND_UP(size * 16, MLX5_SEND_WQE_BB); + qp->sq.w_list[idx].next = qp->sq.cur_post; + + /* We save the edge which was possibly updated during the WQE + * construction, into SQ's cache. + */ + seg = PTR_ALIGN(seg, MLX5_SEND_WQE_BB); + qp->sq.cur_edge = (unlikely(seg == cur_edge)) ? + get_sq_edge(&qp->sq, qp->sq.cur_post & + (qp->sq.wqe_cnt - 1)) : + cur_edge; +} + +static void handle_rdma_op(const struct ib_send_wr *wr, void **seg, int *size) +{ + set_raddr_seg(*seg, rdma_wr(wr)->remote_addr, rdma_wr(wr)->rkey); + *seg += sizeof(struct mlx5_wqe_raddr_seg); + *size += sizeof(struct mlx5_wqe_raddr_seg) / 16; +} + +static void handle_local_inv(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, + struct mlx5_wqe_ctrl_seg **ctrl, void **seg, + int *size, void **cur_edge, unsigned int idx) +{ + qp->sq.wr_data[idx] = IB_WR_LOCAL_INV; + (*ctrl)->imm = cpu_to_be32(wr->ex.invalidate_rkey); + set_linv_wr(qp, seg, size, cur_edge); +} + +static int handle_reg_mr(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, + struct mlx5_wqe_ctrl_seg **ctrl, void **seg, int *size, + void **cur_edge, unsigned int idx) +{ + qp->sq.wr_data[idx] = IB_WR_REG_MR; + (*ctrl)->imm = cpu_to_be32(reg_wr(wr)->key); + return set_reg_wr(qp, reg_wr(wr), seg, size, cur_edge, true); +} + +static int handle_psv(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + const struct ib_send_wr *wr, + struct mlx5_wqe_ctrl_seg **ctrl, void **seg, int *size, + void **cur_edge, unsigned int *idx, int nreq, + struct ib_sig_domain *domain, u32 psv_index, + u8 next_fence) +{ + int err; + + /* + * SET_PSV WQEs are not signaled and solicited on error. + */ + err = __begin_wqe(qp, seg, ctrl, wr, idx, size, cur_edge, nreq, + false, true); + if (unlikely(err)) { + mlx5_ib_warn(dev, "\n"); + err = -ENOMEM; + goto out; + } + err = set_psv_wr(domain, psv_index, seg, size); + if (unlikely(err)) { + mlx5_ib_warn(dev, "\n"); + goto out; + } + finish_wqe(qp, *ctrl, *seg, *size, *cur_edge, *idx, wr->wr_id, nreq, + next_fence, MLX5_OPCODE_SET_PSV); + +out: + return err; +} + +static int handle_reg_mr_integrity(struct mlx5_ib_dev *dev, + struct mlx5_ib_qp *qp, + const struct ib_send_wr *wr, + struct mlx5_wqe_ctrl_seg **ctrl, void **seg, + int *size, void **cur_edge, + unsigned int *idx, int nreq, u8 fence, + u8 next_fence) +{ + struct mlx5_ib_mr *mr; + struct mlx5_ib_mr *pi_mr; + struct mlx5_ib_mr pa_pi_mr; + struct ib_sig_attrs *sig_attrs; + struct ib_reg_wr reg_pi_wr; + int err; + + qp->sq.wr_data[*idx] = IB_WR_REG_MR_INTEGRITY; + + mr = to_mmr(reg_wr(wr)->mr); + pi_mr = mr->pi_mr; + + if (pi_mr) { + memset(®_pi_wr, 0, + sizeof(struct ib_reg_wr)); + + reg_pi_wr.mr = &pi_mr->ibmr; + reg_pi_wr.access = reg_wr(wr)->access; + reg_pi_wr.key = pi_mr->ibmr.rkey; + + (*ctrl)->imm = cpu_to_be32(reg_pi_wr.key); + /* UMR for data + prot registration */ + err = set_reg_wr(qp, ®_pi_wr, seg, size, cur_edge, false); + if (unlikely(err)) + goto out; + + finish_wqe(qp, *ctrl, *seg, *size, *cur_edge, *idx, wr->wr_id, + nreq, fence, MLX5_OPCODE_UMR); + + err = begin_wqe(qp, seg, ctrl, wr, idx, size, cur_edge, nreq); + if (unlikely(err)) { + mlx5_ib_warn(dev, "\n"); + err = -ENOMEM; + goto out; + } + } else { + memset(&pa_pi_mr, 0, sizeof(struct mlx5_ib_mr)); + /* No UMR, use local_dma_lkey */ + pa_pi_mr.ibmr.lkey = mr->ibmr.pd->local_dma_lkey; + pa_pi_mr.ndescs = mr->ndescs; + pa_pi_mr.data_length = mr->data_length; + pa_pi_mr.data_iova = mr->data_iova; + if (mr->meta_ndescs) { + pa_pi_mr.meta_ndescs = mr->meta_ndescs; + pa_pi_mr.meta_length = mr->meta_length; + pa_pi_mr.pi_iova = mr->pi_iova; + } + + pa_pi_mr.ibmr.length = mr->ibmr.length; + mr->pi_mr = &pa_pi_mr; + } + (*ctrl)->imm = cpu_to_be32(mr->ibmr.rkey); + /* UMR for sig MR */ + err = set_pi_umr_wr(wr, qp, seg, size, cur_edge); + if (unlikely(err)) { + mlx5_ib_warn(dev, "\n"); + goto out; + } + finish_wqe(qp, *ctrl, *seg, *size, *cur_edge, *idx, wr->wr_id, nreq, + fence, MLX5_OPCODE_UMR); + + sig_attrs = mr->ibmr.sig_attrs; + err = handle_psv(dev, qp, wr, ctrl, seg, size, cur_edge, idx, nreq, + &sig_attrs->mem, mr->sig->psv_memory.psv_idx, + next_fence); + if (unlikely(err)) + goto out; + + err = handle_psv(dev, qp, wr, ctrl, seg, size, cur_edge, idx, nreq, + &sig_attrs->wire, mr->sig->psv_wire.psv_idx, + next_fence); + if (unlikely(err)) + goto out; + + qp->next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; + +out: + return err; +} + +static int handle_qpt_rc(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + const struct ib_send_wr *wr, + struct mlx5_wqe_ctrl_seg **ctrl, void **seg, int *size, + void **cur_edge, unsigned int *idx, int nreq, u8 fence, + u8 next_fence, int *num_sge) +{ + int err = 0; + + switch (wr->opcode) { + case IB_WR_RDMA_READ: + case IB_WR_RDMA_WRITE: + case IB_WR_RDMA_WRITE_WITH_IMM: + handle_rdma_op(wr, seg, size); + break; + + case IB_WR_ATOMIC_CMP_AND_SWP: + case IB_WR_ATOMIC_FETCH_AND_ADD: + case IB_WR_MASKED_ATOMIC_CMP_AND_SWP: + mlx5_ib_warn(dev, "Atomic operations are not supported yet\n"); + err = -EOPNOTSUPP; + goto out; + + case IB_WR_LOCAL_INV: + handle_local_inv(qp, wr, ctrl, seg, size, cur_edge, *idx); + *num_sge = 0; + break; + + case IB_WR_REG_MR: + err = handle_reg_mr(qp, wr, ctrl, seg, size, cur_edge, *idx); + if (unlikely(err)) + goto out; + *num_sge = 0; + break; + + case IB_WR_REG_MR_INTEGRITY: + err = handle_reg_mr_integrity(dev, qp, wr, ctrl, seg, size, + cur_edge, idx, nreq, fence, + next_fence); + if (unlikely(err)) + goto out; + *num_sge = 0; + break; + + default: + break; + } + +out: + return err; +} + +static void handle_qpt_uc(const struct ib_send_wr *wr, void **seg, int *size) +{ + switch (wr->opcode) { + case IB_WR_RDMA_WRITE: + case IB_WR_RDMA_WRITE_WITH_IMM: + handle_rdma_op(wr, seg, size); + break; + default: + break; + } +} + +static void handle_qpt_hw_gsi(struct mlx5_ib_qp *qp, + const struct ib_send_wr *wr, void **seg, + int *size, void **cur_edge) +{ + set_datagram_seg(*seg, wr); + *seg += sizeof(struct mlx5_wqe_datagram_seg); + *size += sizeof(struct mlx5_wqe_datagram_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); +} + +static void handle_qpt_ud(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, + void **seg, int *size, void **cur_edge) +{ + set_datagram_seg(*seg, wr); + *seg += sizeof(struct mlx5_wqe_datagram_seg); + *size += sizeof(struct mlx5_wqe_datagram_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + + /* handle qp that supports ud offload */ + if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO) { + struct mlx5_wqe_eth_pad *pad; + + pad = *seg; + memset(pad, 0, sizeof(struct mlx5_wqe_eth_pad)); + *seg += sizeof(struct mlx5_wqe_eth_pad); + *size += sizeof(struct mlx5_wqe_eth_pad) / 16; + set_eth_seg(wr, qp, seg, size, cur_edge); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + } +} + +static int handle_qpt_reg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, + const struct ib_send_wr *wr, + struct mlx5_wqe_ctrl_seg **ctrl, void **seg, + int *size, void **cur_edge, unsigned int idx) +{ + int err = 0; + + if (unlikely(wr->opcode != MLX5_IB_WR_UMR)) { + err = -EINVAL; + mlx5_ib_warn(dev, "bad opcode %d\n", wr->opcode); + goto out; + } + + qp->sq.wr_data[idx] = MLX5_IB_WR_UMR; + (*ctrl)->imm = cpu_to_be32(umr_wr(wr)->mkey); + err = set_reg_umr_segment(dev, *seg, wr); + if (unlikely(err)) + goto out; + *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); + *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); + set_reg_mkey_segment(*seg, wr); + *seg += sizeof(struct mlx5_mkey_seg); + *size += sizeof(struct mlx5_mkey_seg) / 16; + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); +out: + return err; +} + +int mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, + const struct ib_send_wr **bad_wr, bool drain) +{ + 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_wqe_xrc_seg *xrc; + struct mlx5_bf *bf; + void *cur_edge; + int size; + unsigned long flags; + unsigned int idx; + int err = 0; + int num_sge; + void *seg; + int nreq; + int i; + u8 next_fence = 0; + u8 fence; + + if (unlikely(mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR && + !drain)) { + *bad_wr = wr; + return -EIO; + } + + if (unlikely(ibqp->qp_type == IB_QPT_GSI)) + return mlx5_ib_gsi_post_send(ibqp, wr, bad_wr); + + qp = to_mqp(ibqp); + bf = &qp->bf; + + spin_lock_irqsave(&qp->sq.lock, flags); + + for (nreq = 0; wr; nreq++, wr = wr->next) { + if (unlikely(wr->opcode >= ARRAY_SIZE(mlx5_ib_opcode))) { + mlx5_ib_warn(dev, "\n"); + err = -EINVAL; + *bad_wr = wr; + goto out; + } + + num_sge = wr->num_sge; + if (unlikely(num_sge > qp->sq.max_gs)) { + mlx5_ib_warn(dev, "\n"); + err = -EINVAL; + *bad_wr = wr; + goto out; + } + + err = begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, &cur_edge, + nreq); + if (err) { + mlx5_ib_warn(dev, "\n"); + err = -ENOMEM; + *bad_wr = wr; + goto out; + } + + if (wr->opcode == IB_WR_REG_MR || + wr->opcode == IB_WR_REG_MR_INTEGRITY) { + fence = dev->umr_fence; + next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; + } else { + if (wr->send_flags & IB_SEND_FENCE) { + if (qp->next_fence) + fence = MLX5_FENCE_MODE_SMALL_AND_FENCE; + else + fence = MLX5_FENCE_MODE_FENCE; + } else { + fence = qp->next_fence; + } + } + + switch (ibqp->qp_type) { + case IB_QPT_XRC_INI: + xrc = seg; + seg += sizeof(*xrc); + size += sizeof(*xrc) / 16; + fallthrough; + case IB_QPT_RC: + err = handle_qpt_rc(dev, qp, wr, &ctrl, &seg, &size, + &cur_edge, &idx, nreq, fence, + next_fence, &num_sge); + if (unlikely(err)) { + *bad_wr = wr; + goto out; + } else if (wr->opcode == IB_WR_REG_MR_INTEGRITY) { + goto skip_psv; + } + break; + + case IB_QPT_UC: + handle_qpt_uc(wr, &seg, &size); + break; + case IB_QPT_SMI: + if (unlikely(!mdev->port_caps[qp->port - 1].has_smi)) { + mlx5_ib_warn(dev, "Send SMP MADs is not allowed\n"); + err = -EPERM; + *bad_wr = wr; + goto out; + } + fallthrough; + case MLX5_IB_QPT_HW_GSI: + handle_qpt_hw_gsi(qp, wr, &seg, &size, &cur_edge); + break; + case IB_QPT_UD: + handle_qpt_ud(qp, wr, &seg, &size, &cur_edge); + break; + case MLX5_IB_QPT_REG_UMR: + err = handle_qpt_reg_umr(dev, qp, wr, &ctrl, &seg, + &size, &cur_edge, idx); + if (unlikely(err)) + goto out; + break; + + default: + break; + } + + if (wr->send_flags & IB_SEND_INLINE && num_sge) { + err = set_data_inl_seg(qp, wr, &seg, &size, &cur_edge); + if (unlikely(err)) { + mlx5_ib_warn(dev, "\n"); + *bad_wr = wr; + goto out; + } + } else { + for (i = 0; i < num_sge; i++) { + handle_post_send_edge(&qp->sq, &seg, size, + &cur_edge); + if (unlikely(!wr->sg_list[i].length)) + continue; + + set_data_ptr_seg( + (struct mlx5_wqe_data_seg *)seg, + wr->sg_list + i); + size += sizeof(struct mlx5_wqe_data_seg) / 16; + seg += sizeof(struct mlx5_wqe_data_seg); + } + } + + qp->next_fence = next_fence; + finish_wqe(qp, ctrl, seg, size, cur_edge, idx, wr->wr_id, nreq, + fence, mlx5_ib_opcode[wr->opcode]); +skip_psv: + if (0) + dump_wqe(qp, idx, size); + } + +out: + if (likely(nreq)) { + qp->sq.head += nreq; + + /* Make sure that descriptors are written before + * updating doorbell record and ringing the doorbell + */ + wmb(); + + qp->db.db[MLX5_SND_DBR] = cpu_to_be32(qp->sq.cur_post); + + /* Make sure doorbell record is visible to the HCA before + * we hit doorbell. + */ + wmb(); + + mlx5_write64((__be32 *)ctrl, bf->bfreg->map + bf->offset); + /* Make sure doorbells don't leak out of SQ spinlock + * and reach the HCA out of order. + */ + bf->offset ^= bf->buf_size; + } + + spin_unlock_irqrestore(&qp->sq.lock, flags); + + return err; +} + +static void set_sig_seg(struct mlx5_rwqe_sig *sig, int max_gs) +{ + sig->signature = calc_sig(sig, (max_gs + 1) << 2); +} + +int mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr, bool drain) +{ + 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; + int ind; + int i; + + if (unlikely(mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR && + !drain)) { + *bad_wr = wr; + return -EIO; + } + + if (unlikely(ibqp->qp_type == IB_QPT_GSI)) + return mlx5_ib_gsi_post_recv(ibqp, wr, bad_wr); + + spin_lock_irqsave(&qp->rq.lock, flags); + + ind = qp->rq.head & (qp->rq.wqe_cnt - 1); + + for (nreq = 0; wr; nreq++, wr = wr->next) { + if (mlx5_wq_overflow(&qp->rq, nreq, qp->ibqp.recv_cq)) { + err = -ENOMEM; + *bad_wr = wr; + goto out; + } + + if (unlikely(wr->num_sge > qp->rq.max_gs)) { + err = -EINVAL; + *bad_wr = wr; + goto out; + } + + scat = mlx5_frag_buf_get_wqe(&qp->rq.fbc, ind); + if (qp->flags_en & MLX5_QP_FLAG_SIGNATURE) + scat++; + + for (i = 0; i < wr->num_sge; i++) + set_data_ptr_seg(scat + i, wr->sg_list + i); + + if (i < qp->rq.max_gs) { + scat[i].byte_count = 0; + scat[i].lkey = cpu_to_be32(MLX5_INVALID_LKEY); + scat[i].addr = 0; + } + + if (qp->flags_en & MLX5_QP_FLAG_SIGNATURE) { + sig = (struct mlx5_rwqe_sig *)scat; + set_sig_seg(sig, qp->rq.max_gs); + } + + qp->rq.wrid[ind] = wr->wr_id; + + ind = (ind + 1) & (qp->rq.wqe_cnt - 1); + } + +out: + if (likely(nreq)) { + qp->rq.head += nreq; + + /* Make sure that descriptors are written before + * doorbell record. + */ + wmb(); + + *qp->db.db = cpu_to_be32(qp->rq.head & 0xffff); + } + + spin_unlock_irqrestore(&qp->rq.lock, flags); + + return err; +} diff --git a/drivers/infiniband/hw/mlx5/wr.h b/drivers/infiniband/hw/mlx5/wr.h new file mode 100644 index 000000000000..4f0057516402 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/wr.h @@ -0,0 +1,76 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2020, Mellanox Technologies inc. All rights reserved. + */ + +#ifndef _MLX5_IB_WR_H +#define _MLX5_IB_WR_H + +#include "mlx5_ib.h" + +enum { + MLX5_IB_SQ_UMR_INLINE_THRESHOLD = 64, +}; + +struct mlx5_wqe_eth_pad { + u8 rsvd0[16]; +}; + + +/* get_sq_edge - Get the next nearby edge. + * + * An 'edge' is defined as the first following address after the end + * of the fragment or the SQ. Accordingly, during the WQE construction + * which repetitively increases the pointer to write the next data, it + * simply should check if it gets to an edge. + * + * @sq - SQ buffer. + * @idx - Stride index in the SQ buffer. + * + * Return: + * The new edge. + */ +static inline void *get_sq_edge(struct mlx5_ib_wq *sq, u32 idx) +{ + void *fragment_end; + + fragment_end = mlx5_frag_buf_get_wqe + (&sq->fbc, + mlx5_frag_buf_get_idx_last_contig_stride(&sq->fbc, idx)); + + return fragment_end + MLX5_SEND_WQE_BB; +} + +int mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, + const struct ib_send_wr **bad_wr, bool drain); +int mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr, bool drain); + +static inline int mlx5_ib_post_send_nodrain(struct ib_qp *ibqp, + const struct ib_send_wr *wr, + const struct ib_send_wr **bad_wr) +{ + return mlx5_ib_post_send(ibqp, wr, bad_wr, false); +} + +static inline int mlx5_ib_post_send_drain(struct ib_qp *ibqp, + const struct ib_send_wr *wr, + const struct ib_send_wr **bad_wr) +{ + return mlx5_ib_post_send(ibqp, wr, bad_wr, true); +} + +static inline int mlx5_ib_post_recv_nodrain(struct ib_qp *ibqp, + const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr) +{ + return mlx5_ib_post_recv(ibqp, wr, bad_wr, false); +} + +static inline int mlx5_ib_post_recv_drain(struct ib_qp *ibqp, + const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr) +{ + return mlx5_ib_post_recv(ibqp, wr, bad_wr, true); +} +#endif /* _MLX5_IB_WR_H */ diff --git a/drivers/infiniband/hw/mthca/Kconfig b/drivers/infiniband/hw/mthca/Kconfig index 66ff527f5928..faa7381f7d63 100644 --- a/drivers/infiniband/hw/mthca/Kconfig +++ b/drivers/infiniband/hw/mthca/Kconfig @@ -2,7 +2,7 @@ config INFINIBAND_MTHCA tristate "Mellanox HCA support" depends on PCI - ---help--- + help This is a low-level driver for Mellanox InfiniHost host channel adapters (HCAs), including the MT23108 PCI-X HCA ("Tavor") and the MT25208 PCI Express HCA ("Arbel"). @@ -11,7 +11,7 @@ config INFINIBAND_MTHCA_DEBUG bool "Verbose debugging output" if EXPERT depends on INFINIBAND_MTHCA default y - ---help--- + help This option causes debugging code to be compiled into the mthca driver. The output can be turned on via the debug_level module parameter (which can also be set after diff --git a/drivers/infiniband/hw/mthca/mthca_av.c b/drivers/infiniband/hw/mthca/mthca_av.c index 0823c0bc7e73..f051f4e06b53 100644 --- a/drivers/infiniband/hw/mthca/mthca_av.c +++ b/drivers/infiniband/hw/mthca/mthca_av.c @@ -115,7 +115,7 @@ static u8 ib_rate_to_memfree(u8 req_rate, u8 cur_rate) switch ((cur_rate - 1) / req_rate) { case 0: return MTHCA_RATE_MEMFREE_FULL; case 1: return MTHCA_RATE_MEMFREE_HALF; - case 2: /* fall through */ + case 2: case 3: return MTHCA_RATE_MEMFREE_QUARTER; default: return MTHCA_RATE_MEMFREE_EIGHTH; } diff --git a/drivers/infiniband/hw/mthca/mthca_dev.h b/drivers/infiniband/hw/mthca/mthca_dev.h index 599794c5a78f..7550e9d03dec 100644 --- a/drivers/infiniband/hw/mthca/mthca_dev.h +++ b/drivers/infiniband/hw/mthca/mthca_dev.h @@ -478,16 +478,6 @@ int mthca_mr_alloc_phys(struct mthca_dev *dev, u32 pd, u32 access, struct mthca_mr *mr); void mthca_free_mr(struct mthca_dev *dev, struct mthca_mr *mr); -int mthca_fmr_alloc(struct mthca_dev *dev, u32 pd, - u32 access, struct mthca_fmr *fmr); -int mthca_tavor_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int list_len, u64 iova); -void mthca_tavor_fmr_unmap(struct mthca_dev *dev, struct mthca_fmr *fmr); -int mthca_arbel_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int list_len, u64 iova); -void mthca_arbel_fmr_unmap(struct mthca_dev *dev, struct mthca_fmr *fmr); -int mthca_free_fmr(struct mthca_dev *dev, struct mthca_fmr *fmr); - int mthca_map_eq_icm(struct mthca_dev *dev, u64 icm_virt); void mthca_unmap_eq_icm(struct mthca_dev *dev); diff --git a/drivers/infiniband/hw/mthca/mthca_mr.c b/drivers/infiniband/hw/mthca/mthca_mr.c index 4250b2c18c64..ce0e0867e488 100644 --- a/drivers/infiniband/hw/mthca/mthca_mr.c +++ b/drivers/infiniband/hw/mthca/mthca_mr.c @@ -541,7 +541,7 @@ int mthca_mr_alloc_phys(struct mthca_dev *dev, u32 pd, return err; } -/* Free mr or fmr */ +/* Free mr */ static void mthca_free_region(struct mthca_dev *dev, u32 lkey) { mthca_table_put(dev, dev->mr_table.mpt_table, @@ -564,266 +564,6 @@ void mthca_free_mr(struct mthca_dev *dev, struct mthca_mr *mr) mthca_free_mtt(dev, mr->mtt); } -int mthca_fmr_alloc(struct mthca_dev *dev, u32 pd, - u32 access, struct mthca_fmr *mr) -{ - struct mthca_mpt_entry *mpt_entry; - struct mthca_mailbox *mailbox; - u64 mtt_seg; - u32 key, idx; - int list_len = mr->attr.max_pages; - int err = -ENOMEM; - int i; - - if (mr->attr.page_shift < 12 || mr->attr.page_shift >= 32) - return -EINVAL; - - /* For Arbel, all MTTs must fit in the same page. */ - if (mthca_is_memfree(dev) && - mr->attr.max_pages * sizeof *mr->mem.arbel.mtts > PAGE_SIZE) - return -EINVAL; - - mr->maps = 0; - - key = mthca_alloc(&dev->mr_table.mpt_alloc); - if (key == -1) - return -ENOMEM; - key = adjust_key(dev, key); - - idx = key & (dev->limits.num_mpts - 1); - mr->ibmr.rkey = mr->ibmr.lkey = hw_index_to_key(dev, key); - - if (mthca_is_memfree(dev)) { - err = mthca_table_get(dev, dev->mr_table.mpt_table, key); - if (err) - goto err_out_mpt_free; - - mr->mem.arbel.mpt = mthca_table_find(dev->mr_table.mpt_table, key, NULL); - BUG_ON(!mr->mem.arbel.mpt); - } else - mr->mem.tavor.mpt = dev->mr_table.tavor_fmr.mpt_base + - sizeof *(mr->mem.tavor.mpt) * idx; - - mr->mtt = __mthca_alloc_mtt(dev, list_len, dev->mr_table.fmr_mtt_buddy); - if (IS_ERR(mr->mtt)) { - err = PTR_ERR(mr->mtt); - goto err_out_table; - } - - mtt_seg = mr->mtt->first_seg * dev->limits.mtt_seg_size; - - if (mthca_is_memfree(dev)) { - mr->mem.arbel.mtts = mthca_table_find(dev->mr_table.mtt_table, - mr->mtt->first_seg, - &mr->mem.arbel.dma_handle); - BUG_ON(!mr->mem.arbel.mtts); - } else - mr->mem.tavor.mtts = dev->mr_table.tavor_fmr.mtt_base + mtt_seg; - - mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); - if (IS_ERR(mailbox)) { - err = PTR_ERR(mailbox); - goto err_out_free_mtt; - } - - mpt_entry = mailbox->buf; - - mpt_entry->flags = cpu_to_be32(MTHCA_MPT_FLAG_SW_OWNS | - MTHCA_MPT_FLAG_MIO | - MTHCA_MPT_FLAG_REGION | - access); - - mpt_entry->page_size = cpu_to_be32(mr->attr.page_shift - 12); - mpt_entry->key = cpu_to_be32(key); - mpt_entry->pd = cpu_to_be32(pd); - memset(&mpt_entry->start, 0, - sizeof *mpt_entry - offsetof(struct mthca_mpt_entry, start)); - mpt_entry->mtt_seg = cpu_to_be64(dev->mr_table.mtt_base + mtt_seg); - - if (0) { - mthca_dbg(dev, "Dumping MPT entry %08x:\n", mr->ibmr.lkey); - for (i = 0; i < sizeof (struct mthca_mpt_entry) / 4; ++i) { - if (i % 4 == 0) - printk("[%02x] ", i * 4); - printk(" %08x", be32_to_cpu(((__be32 *) mpt_entry)[i])); - if ((i + 1) % 4 == 0) - printk("\n"); - } - } - - err = mthca_SW2HW_MPT(dev, mailbox, - key & (dev->limits.num_mpts - 1)); - if (err) { - mthca_warn(dev, "SW2HW_MPT failed (%d)\n", err); - goto err_out_mailbox_free; - } - - mthca_free_mailbox(dev, mailbox); - return 0; - -err_out_mailbox_free: - mthca_free_mailbox(dev, mailbox); - -err_out_free_mtt: - mthca_free_mtt(dev, mr->mtt); - -err_out_table: - mthca_table_put(dev, dev->mr_table.mpt_table, key); - -err_out_mpt_free: - mthca_free(&dev->mr_table.mpt_alloc, key); - return err; -} - -int mthca_free_fmr(struct mthca_dev *dev, struct mthca_fmr *fmr) -{ - if (fmr->maps) - return -EBUSY; - - mthca_free_region(dev, fmr->ibmr.lkey); - mthca_free_mtt(dev, fmr->mtt); - - return 0; -} - -static inline int mthca_check_fmr(struct mthca_fmr *fmr, u64 *page_list, - int list_len, u64 iova) -{ - int i, page_mask; - - if (list_len > fmr->attr.max_pages) - return -EINVAL; - - page_mask = (1 << fmr->attr.page_shift) - 1; - - /* We are getting page lists, so va must be page aligned. */ - if (iova & page_mask) - return -EINVAL; - - /* Trust the user not to pass misaligned data in page_list */ - if (0) - for (i = 0; i < list_len; ++i) { - if (page_list[i] & ~page_mask) - return -EINVAL; - } - - if (fmr->maps >= fmr->attr.max_maps) - return -EINVAL; - - return 0; -} - - -int mthca_tavor_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int list_len, u64 iova) -{ - struct mthca_fmr *fmr = to_mfmr(ibfmr); - struct mthca_dev *dev = to_mdev(ibfmr->device); - struct mthca_mpt_entry mpt_entry; - u32 key; - int i, err; - - err = mthca_check_fmr(fmr, page_list, list_len, iova); - if (err) - return err; - - ++fmr->maps; - - key = tavor_key_to_hw_index(fmr->ibmr.lkey); - key += dev->limits.num_mpts; - fmr->ibmr.lkey = fmr->ibmr.rkey = tavor_hw_index_to_key(key); - - writeb(MTHCA_MPT_STATUS_SW, fmr->mem.tavor.mpt); - - for (i = 0; i < list_len; ++i) { - __be64 mtt_entry = cpu_to_be64(page_list[i] | - MTHCA_MTT_FLAG_PRESENT); - mthca_write64_raw(mtt_entry, fmr->mem.tavor.mtts + i); - } - - mpt_entry.lkey = cpu_to_be32(key); - mpt_entry.length = cpu_to_be64(list_len * (1ull << fmr->attr.page_shift)); - mpt_entry.start = cpu_to_be64(iova); - - __raw_writel((__force u32) mpt_entry.lkey, &fmr->mem.tavor.mpt->key); - memcpy_toio(&fmr->mem.tavor.mpt->start, &mpt_entry.start, - offsetof(struct mthca_mpt_entry, window_count) - - offsetof(struct mthca_mpt_entry, start)); - - writeb(MTHCA_MPT_STATUS_HW, fmr->mem.tavor.mpt); - - return 0; -} - -int mthca_arbel_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int list_len, u64 iova) -{ - struct mthca_fmr *fmr = to_mfmr(ibfmr); - struct mthca_dev *dev = to_mdev(ibfmr->device); - u32 key; - int i, err; - - err = mthca_check_fmr(fmr, page_list, list_len, iova); - if (err) - return err; - - ++fmr->maps; - - key = arbel_key_to_hw_index(fmr->ibmr.lkey); - if (dev->mthca_flags & MTHCA_FLAG_SINAI_OPT) - key += SINAI_FMR_KEY_INC; - else - key += dev->limits.num_mpts; - fmr->ibmr.lkey = fmr->ibmr.rkey = arbel_hw_index_to_key(key); - - *(u8 *) fmr->mem.arbel.mpt = MTHCA_MPT_STATUS_SW; - - wmb(); - - dma_sync_single_for_cpu(&dev->pdev->dev, fmr->mem.arbel.dma_handle, - list_len * sizeof(u64), DMA_TO_DEVICE); - - for (i = 0; i < list_len; ++i) - fmr->mem.arbel.mtts[i] = cpu_to_be64(page_list[i] | - MTHCA_MTT_FLAG_PRESENT); - - dma_sync_single_for_device(&dev->pdev->dev, fmr->mem.arbel.dma_handle, - list_len * sizeof(u64), DMA_TO_DEVICE); - - fmr->mem.arbel.mpt->key = cpu_to_be32(key); - fmr->mem.arbel.mpt->lkey = cpu_to_be32(key); - fmr->mem.arbel.mpt->length = cpu_to_be64(list_len * (1ull << fmr->attr.page_shift)); - fmr->mem.arbel.mpt->start = cpu_to_be64(iova); - - wmb(); - - *(u8 *) fmr->mem.arbel.mpt = MTHCA_MPT_STATUS_HW; - - wmb(); - - return 0; -} - -void mthca_tavor_fmr_unmap(struct mthca_dev *dev, struct mthca_fmr *fmr) -{ - if (!fmr->maps) - return; - - fmr->maps = 0; - - writeb(MTHCA_MPT_STATUS_SW, fmr->mem.tavor.mpt); -} - -void mthca_arbel_fmr_unmap(struct mthca_dev *dev, struct mthca_fmr *fmr) -{ - if (!fmr->maps) - return; - - fmr->maps = 0; - - *(u8 *) fmr->mem.arbel.mpt = MTHCA_MPT_STATUS_SW; -} - int mthca_init_mr_table(struct mthca_dev *dev) { phys_addr_t addr; diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c index 69a3e4f62fb1..9fa2f9164a47 100644 --- a/drivers/infiniband/hw/mthca/mthca_provider.c +++ b/drivers/infiniband/hw/mthca/mthca_provider.c @@ -118,16 +118,6 @@ static int mthca_query_device(struct ib_device *ibdev, struct ib_device_attr *pr props->max_mcast_qp_attach = MTHCA_QP_PER_MGM; props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * props->max_mcast_grp; - /* - * If Sinai memory key optimization is being used, then only - * the 8-bit key portion will change. For other HCAs, the - * unused index bits will also be used for FMR remapping. - */ - if (mdev->mthca_flags & MTHCA_FLAG_SINAI_OPT) - props->max_map_per_fmr = 255; - else - props->max_map_per_fmr = - (1 << (32 - ilog2(mdev->limits.num_mpts))) - 1; err = 0; out: @@ -388,14 +378,15 @@ static void mthca_dealloc_pd(struct ib_pd *pd, struct ib_udata *udata) mthca_pd_free(to_mdev(pd->device), to_mpd(pd)); } -static int mthca_ah_create(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata) +static int mthca_ah_create(struct ib_ah *ibah, + struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { struct mthca_ah *ah = to_mah(ibah); - return mthca_create_ah(to_mdev(ibah->device), to_mpd(ibah->pd), ah_attr, - ah); + return mthca_create_ah(to_mdev(ibah->device), to_mpd(ibah->pd), + init_attr->ah_attr, ah); } static void mthca_ah_destroy(struct ib_ah *ah, u32 flags) @@ -957,69 +948,6 @@ static int mthca_dereg_mr(struct ib_mr *mr, struct ib_udata *udata) return 0; } -static struct ib_fmr *mthca_alloc_fmr(struct ib_pd *pd, int mr_access_flags, - struct ib_fmr_attr *fmr_attr) -{ - struct mthca_fmr *fmr; - int err; - - fmr = kmalloc(sizeof *fmr, GFP_KERNEL); - if (!fmr) - return ERR_PTR(-ENOMEM); - - memcpy(&fmr->attr, fmr_attr, sizeof *fmr_attr); - err = mthca_fmr_alloc(to_mdev(pd->device), to_mpd(pd)->pd_num, - convert_access(mr_access_flags), fmr); - - if (err) { - kfree(fmr); - return ERR_PTR(err); - } - - return &fmr->ibmr; -} - -static int mthca_dealloc_fmr(struct ib_fmr *fmr) -{ - struct mthca_fmr *mfmr = to_mfmr(fmr); - int err; - - err = mthca_free_fmr(to_mdev(fmr->device), mfmr); - if (err) - return err; - - kfree(mfmr); - return 0; -} - -static int mthca_unmap_fmr(struct list_head *fmr_list) -{ - struct ib_fmr *fmr; - int err; - struct mthca_dev *mdev = NULL; - - list_for_each_entry(fmr, fmr_list, list) { - if (mdev && to_mdev(fmr->device) != mdev) - return -EINVAL; - mdev = to_mdev(fmr->device); - } - - if (!mdev) - return 0; - - if (mthca_is_memfree(mdev)) { - list_for_each_entry(fmr, fmr_list, list) - mthca_arbel_fmr_unmap(mdev, to_mfmr(fmr)); - - wmb(); - } else - list_for_each_entry(fmr, fmr_list, list) - mthca_tavor_fmr_unmap(mdev, to_mfmr(fmr)); - - err = mthca_SYNC_TPT(mdev); - return err; -} - static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr, char *buf) { @@ -1203,20 +1131,6 @@ static const struct ib_device_ops mthca_dev_tavor_srq_ops = { INIT_RDMA_OBJ_SIZE(ib_srq, mthca_srq, ibsrq), }; -static const struct ib_device_ops mthca_dev_arbel_fmr_ops = { - .alloc_fmr = mthca_alloc_fmr, - .dealloc_fmr = mthca_dealloc_fmr, - .map_phys_fmr = mthca_arbel_map_phys_fmr, - .unmap_fmr = mthca_unmap_fmr, -}; - -static const struct ib_device_ops mthca_dev_tavor_fmr_ops = { - .alloc_fmr = mthca_alloc_fmr, - .dealloc_fmr = mthca_dealloc_fmr, - .map_phys_fmr = mthca_tavor_map_phys_fmr, - .unmap_fmr = mthca_unmap_fmr, -}; - static const struct ib_device_ops mthca_dev_arbel_ops = { .post_recv = mthca_arbel_post_receive, .post_send = mthca_arbel_post_send, @@ -1275,15 +1189,6 @@ int mthca_register_device(struct mthca_dev *dev) &mthca_dev_tavor_srq_ops); } - if (dev->mthca_flags & MTHCA_FLAG_FMR) { - if (mthca_is_memfree(dev)) - ib_set_device_ops(&dev->ib_dev, - &mthca_dev_arbel_fmr_ops); - else - ib_set_device_ops(&dev->ib_dev, - &mthca_dev_tavor_fmr_ops); - } - ib_set_device_ops(&dev->ib_dev, &mthca_dev_ops); if (mthca_is_memfree(dev)) diff --git a/drivers/infiniband/hw/mthca/mthca_provider.h b/drivers/infiniband/hw/mthca/mthca_provider.h index 596acc45569b..84c64bff0d92 100644 --- a/drivers/infiniband/hw/mthca/mthca_provider.h +++ b/drivers/infiniband/hw/mthca/mthca_provider.h @@ -76,24 +76,6 @@ struct mthca_mr { struct mthca_mtt *mtt; }; -struct mthca_fmr { - struct ib_fmr ibmr; - struct ib_fmr_attr attr; - struct mthca_mtt *mtt; - int maps; - union { - struct { - struct mthca_mpt_entry __iomem *mpt; - u64 __iomem *mtts; - } tavor; - struct { - struct mthca_mpt_entry *mpt; - __be64 *mtts; - dma_addr_t dma_handle; - } arbel; - } mem; -}; - struct mthca_pd { struct ib_pd ibpd; u32 pd_num; @@ -301,11 +283,6 @@ static inline struct mthca_ucontext *to_mucontext(struct ib_ucontext *ibucontext return container_of(ibucontext, struct mthca_ucontext, ibucontext); } -static inline struct mthca_fmr *to_mfmr(struct ib_fmr *ibmr) -{ - return container_of(ibmr, struct mthca_fmr, ibmr); -} - static inline struct mthca_mr *to_mmr(struct ib_mr *ibmr) { return container_of(ibmr, struct mthca_mr, ibmr); diff --git a/drivers/infiniband/hw/mthca/mthca_qp.c b/drivers/infiniband/hw/mthca/mthca_qp.c index d04c245359eb..c6e95d0d760a 100644 --- a/drivers/infiniband/hw/mthca/mthca_qp.c +++ b/drivers/infiniband/hw/mthca/mthca_qp.c @@ -1639,8 +1639,8 @@ int mthca_tavor_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, * without initializing f0 and size0, and they are in fact * never used uninitialized. */ - int uninitialized_var(size0); - u32 uninitialized_var(f0); + int size0; + u32 f0; int ind; u8 op0 = 0; @@ -1835,7 +1835,7 @@ int mthca_tavor_post_receive(struct ib_qp *ibqp, const struct ib_recv_wr *wr, * without initializing size0, and it is in fact never used * uninitialized. */ - int uninitialized_var(size0); + int size0; int ind; void *wqe; void *prev_wqe; @@ -1943,8 +1943,8 @@ int mthca_arbel_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, * without initializing f0 and size0, and they are in fact * never used uninitialized. */ - int uninitialized_var(size0); - u32 uninitialized_var(f0); + int size0; + u32 f0; int ind; u8 op0 = 0; diff --git a/drivers/infiniband/hw/ocrdma/Kconfig b/drivers/infiniband/hw/ocrdma/Kconfig index dd4ec388208c..54bd70bc4d1a 100644 --- a/drivers/infiniband/hw/ocrdma/Kconfig +++ b/drivers/infiniband/hw/ocrdma/Kconfig @@ -4,6 +4,6 @@ config INFINIBAND_OCRDMA depends on ETHERNET && NETDEVICES && PCI && INET && (IPV6 || IPV6=n) select NET_VENDOR_EMULEX select BE2NET - ---help--- + help This driver provides low-level InfiniBand over Ethernet support for Emulex One Connect host channel adapters (HCAs). diff --git a/drivers/infiniband/hw/ocrdma/ocrdma.h b/drivers/infiniband/hw/ocrdma/ocrdma.h index 7baedc74e39d..fcfe0e82197a 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma.h +++ b/drivers/infiniband/hw/ocrdma/ocrdma.h @@ -98,7 +98,6 @@ struct ocrdma_dev_attr { u64 max_mr_size; u32 max_num_mr_pbl; int max_mw; - int max_fmr; int max_map_per_fmr; int max_pages_per_frmr; u16 max_ord_per_qp; diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c index 2b7f00ac41b0..6eea02b18968 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c @@ -155,7 +155,7 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah, return status; } -int ocrdma_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, u32 flags, +int ocrdma_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata) { u32 *ahid_addr; @@ -165,6 +165,7 @@ int ocrdma_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, u32 flags, u16 vlan_tag = 0xffff; const struct ib_gid_attr *sgid_attr; struct ocrdma_pd *pd = get_ocrdma_pd(ibah->pd); + struct rdma_ah_attr *attr = init_attr->ah_attr; struct ocrdma_dev *dev = get_ocrdma_dev(ibah->device); if ((attr->type != RDMA_AH_ATTR_TYPE_ROCE) || diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.h b/drivers/infiniband/hw/ocrdma/ocrdma_ah.h index 9780afcde780..8b73b3489f3a 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.h +++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.h @@ -51,7 +51,7 @@ enum { OCRDMA_AH_L3_TYPE_SHIFT = 0x1D /* 29 bits */ }; -int ocrdma_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, u32 flags, +int ocrdma_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); void ocrdma_destroy_ah(struct ib_ah *ah, u32 flags); int ocrdma_query_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr); diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c index d82d3ec3649e..e07bf0b2209a 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c @@ -1190,7 +1190,6 @@ static void ocrdma_get_attr(struct ocrdma_dev *dev, attr->max_mr = rsp->max_mr; attr->max_mr_size = ((u64)rsp->max_mr_size_hi << 32) | rsp->max_mr_size_lo; - attr->max_fmr = 0; attr->max_pages_per_frmr = rsp->max_pages_per_frmr; attr->max_num_mr_pbl = rsp->max_num_mr_pbl; attr->max_cqe = rsp->max_cq_cqes_per_cq & diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c index 10e343894595..c1751c9a0f62 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c @@ -99,8 +99,6 @@ int ocrdma_query_device(struct ib_device *ibdev, struct ib_device_attr *attr, attr->max_mw = dev->attr.max_mw; attr->max_pd = dev->attr.max_pd; attr->atomic_cap = 0; - attr->max_fmr = 0; - attr->max_map_per_fmr = 0; attr->max_qp_rd_atom = min(dev->attr.max_ord_per_qp, dev->attr.max_ird_per_qp); attr->max_qp_init_rd_atom = dev->attr.max_ord_per_qp; @@ -2136,7 +2134,7 @@ int ocrdma_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, case IB_WR_SEND_WITH_IMM: hdr->cw |= (OCRDMA_FLAG_IMM << OCRDMA_WQE_FLAGS_SHIFT); hdr->immdt = ntohl(wr->ex.imm_data); - /* fall through */ + fallthrough; case IB_WR_SEND: hdr->cw |= (OCRDMA_SEND << OCRDMA_WQE_OPCODE_SHIFT); ocrdma_build_send(qp, hdr, wr); @@ -2150,7 +2148,7 @@ int ocrdma_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, case IB_WR_RDMA_WRITE_WITH_IMM: hdr->cw |= (OCRDMA_FLAG_IMM << OCRDMA_WQE_FLAGS_SHIFT); hdr->immdt = ntohl(wr->ex.imm_data); - /* fall through */ + fallthrough; case IB_WR_RDMA_WRITE: hdr->cw |= (OCRDMA_WRITE << OCRDMA_WQE_OPCODE_SHIFT); status = ocrdma_build_write(qp, hdr, wr); @@ -2903,7 +2901,7 @@ int ocrdma_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags cq_flags) } struct ib_mr *ocrdma_alloc_mr(struct ib_pd *ibpd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { int status; struct ocrdma_mr *mr; diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h index 3a5010881be5..df8e3b923a44 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.h @@ -101,7 +101,7 @@ struct ib_mr *ocrdma_get_dma_mr(struct ib_pd *, int acc); struct ib_mr *ocrdma_reg_user_mr(struct ib_pd *, u64 start, u64 length, u64 virt, int acc, struct ib_udata *); struct ib_mr *ocrdma_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int ocrdma_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); diff --git a/drivers/infiniband/hw/qedr/Kconfig b/drivers/infiniband/hw/qedr/Kconfig index 9c30325e1414..9e31d13b03c3 100644 --- a/drivers/infiniband/hw/qedr/Kconfig +++ b/drivers/infiniband/hw/qedr/Kconfig @@ -6,6 +6,6 @@ config INFINIBAND_QEDR select QED_LL2 select QED_OOO select QED_RDMA - ---help--- + help This driver provides low-level InfiniBand over Ethernet support for QLogic QED host channel adapters (HCAs). diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c index dcdc85a1ab25..d85f992bac29 100644 --- a/drivers/infiniband/hw/qedr/main.c +++ b/drivers/infiniband/hw/qedr/main.c @@ -110,7 +110,6 @@ static int qedr_iw_port_immutable(struct ib_device *ibdev, u8 port_num, if (err) return err; - immutable->pkey_tbl_len = 1; immutable->gid_tbl_len = 1; immutable->core_cap_flags = RDMA_CORE_PORT_IWARP; immutable->max_mad_size = 0; @@ -179,6 +178,7 @@ static int qedr_iw_register_device(struct qedr_dev *dev) static const struct ib_device_ops qedr_roce_dev_ops = { .get_port_immutable = qedr_roce_port_immutable, + .query_pkey = qedr_query_pkey, }; static void qedr_roce_register_device(struct qedr_dev *dev) @@ -221,7 +221,6 @@ static const struct ib_device_ops qedr_dev_ops = { .post_srq_recv = qedr_post_srq_recv, .process_mad = qedr_process_mad, .query_device = qedr_query_device, - .query_pkey = qedr_query_pkey, .query_port = qedr_query_port, .query_qp = qedr_query_qp, .query_srq = qedr_query_srq, @@ -346,9 +345,14 @@ static void qedr_free_resources(struct qedr_dev *dev) static int qedr_alloc_resources(struct qedr_dev *dev) { + struct qed_chain_init_params params = { + .mode = QED_CHAIN_MODE_PBL, + .intended_use = QED_CHAIN_USE_TO_CONSUME, + .cnt_type = QED_CHAIN_CNT_TYPE_U16, + .elem_size = sizeof(struct regpair *), + }; struct qedr_cnq *cnq; __le16 *cons_pi; - u16 n_entries; int i, rc; dev->sgid_tbl = kcalloc(QEDR_MAX_SGID, sizeof(union ib_gid), @@ -382,7 +386,9 @@ static int qedr_alloc_resources(struct qedr_dev *dev) dev->sb_start = dev->ops->rdma_get_start_sb(dev->cdev); /* Allocate CNQ PBLs */ - n_entries = min_t(u32, QED_RDMA_MAX_CNQ_SIZE, QEDR_ROCE_MAX_CNQ_SIZE); + params.num_elems = min_t(u32, QED_RDMA_MAX_CNQ_SIZE, + QEDR_ROCE_MAX_CNQ_SIZE); + for (i = 0; i < dev->num_cnq; i++) { cnq = &dev->cnq_array[i]; @@ -391,13 +397,8 @@ static int qedr_alloc_resources(struct qedr_dev *dev) if (rc) goto err3; - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_CONSUME, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U16, - n_entries, - sizeof(struct regpair *), - &cnq->pbl, NULL); + rc = dev->ops->common->chain_alloc(dev->cdev, &cnq->pbl, + ¶ms); if (rc) goto err4; @@ -632,7 +633,6 @@ static int qedr_set_device_attr(struct qedr_dev *dev) attr->max_mr_size = qed_attr->max_mr_size; attr->max_cqe = min_t(u64, qed_attr->max_cqe, QEDR_MAX_CQES); attr->max_mw = qed_attr->max_mw; - attr->max_fmr = qed_attr->max_fmr; attr->max_mr_mw_fmr_pbl = qed_attr->max_mr_mw_fmr_pbl; attr->max_mr_mw_fmr_size = qed_attr->max_mr_mw_fmr_size; attr->max_pd = qed_attr->max_pd; diff --git a/drivers/infiniband/hw/qedr/qedr.h b/drivers/infiniband/hw/qedr/qedr.h index 5488dbd59d3c..460292179b32 100644 --- a/drivers/infiniband/hw/qedr/qedr.h +++ b/drivers/infiniband/hw/qedr/qedr.h @@ -103,7 +103,6 @@ struct qedr_device_attr { u64 max_mr_size; u32 max_cqe; u32 max_mw; - u32 max_fmr; u32 max_mr_mw_fmr_pbl; u64 max_mr_mw_fmr_size; u32 max_pd; @@ -236,6 +235,7 @@ struct qedr_ucontext { u32 dpi_size; u16 dpi; bool db_rec; + u8 edpm_mode; }; union db_prod32 { @@ -345,10 +345,10 @@ struct qedr_srq_hwq_info { u32 wqe_prod; u32 sge_prod; u32 wr_prod_cnt; - u32 wr_cons_cnt; + atomic_t wr_cons_cnt; u32 num_elems; - u32 *virt_prod_pair_addr; + struct rdma_srq_producers *virt_prod_pair_addr; dma_addr_t phy_prod_pair_addr; }; diff --git a/drivers/infiniband/hw/qedr/qedr_iw_cm.c b/drivers/infiniband/hw/qedr/qedr_iw_cm.c index 792eecd206b6..97fc7dd353b0 100644 --- a/drivers/infiniband/hw/qedr/qedr_iw_cm.c +++ b/drivers/infiniband/hw/qedr/qedr_iw_cm.c @@ -150,8 +150,17 @@ qedr_iw_issue_event(void *context, if (params->cm_info) { event.ird = params->cm_info->ird; event.ord = params->cm_info->ord; - event.private_data_len = params->cm_info->private_data_len; - event.private_data = (void *)params->cm_info->private_data; + /* Only connect_request and reply have valid private data + * the rest of the events this may be left overs from + * connection establishment. CONNECT_REQUEST is issued via + * qedr_iw_mpa_request + */ + if (event_type == IW_CM_EVENT_CONNECT_REPLY) { + event.private_data_len = + params->cm_info->private_data_len; + event.private_data = + (void *)params->cm_info->private_data; + } } if (ep->cm_id) diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c index a5bd3adaf90a..b49bef94637e 100644 --- a/drivers/infiniband/hw/qedr/verbs.c +++ b/drivers/infiniband/hw/qedr/verbs.c @@ -145,8 +145,6 @@ int qedr_query_device(struct ib_device *ibdev, attr->max_mw = qattr->max_mw; attr->max_pd = qattr->max_pd; attr->atomic_cap = dev->atomic_cap; - attr->max_fmr = qattr->max_fmr; - attr->max_map_per_fmr = 16; attr->max_qp_init_rd_atom = 1 << (fls(qattr->max_qp_req_rd_atomic_resc) - 1); attr->max_qp_rd_atom = @@ -241,7 +239,6 @@ int qedr_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *attr) attr->ip_gids = true; if (rdma_protocol_iwarp(&dev->ibdev, 1)) { attr->gid_tbl_len = 1; - attr->pkey_tbl_len = 1; } else { attr->gid_tbl_len = QEDR_MAX_SGID; attr->pkey_tbl_len = QEDR_ROCE_PKEY_TABLE_LEN; @@ -277,7 +274,8 @@ int qedr_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata) DP_ERR(dev, "Problem copying data from user space\n"); return -EFAULT; } - + ctx->edpm_mode = !!(ureq.context_flags & + QEDR_ALLOC_UCTX_EDPM_MODE); ctx->db_rec = !!(ureq.context_flags & QEDR_ALLOC_UCTX_DB_REC); } @@ -318,11 +316,15 @@ int qedr_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata) uresp.dpm_flags = QEDR_DPM_TYPE_IWARP_LEGACY; else uresp.dpm_flags = QEDR_DPM_TYPE_ROCE_ENHANCED | - QEDR_DPM_TYPE_ROCE_LEGACY; + QEDR_DPM_TYPE_ROCE_LEGACY | + QEDR_DPM_TYPE_ROCE_EDPM_MODE; - uresp.dpm_flags |= QEDR_DPM_SIZES_SET; - uresp.ldpm_limit_size = QEDR_LDPM_MAX_SIZE; - uresp.edpm_trans_size = QEDR_EDPM_TRANS_SIZE; + if (ureq.context_flags & QEDR_SUPPORT_DPM_SIZES) { + uresp.dpm_flags |= QEDR_DPM_SIZES_SET; + uresp.ldpm_limit_size = QEDR_LDPM_MAX_SIZE; + uresp.edpm_trans_size = QEDR_EDPM_TRANS_SIZE; + uresp.edpm_limit_size = QEDR_EDPM_MAX_SIZE; + } uresp.wids_enabled = 1; uresp.wid_count = oparams.wid_count; @@ -893,6 +895,12 @@ int qedr_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, udata, struct qedr_ucontext, ibucontext); struct qed_rdma_destroy_cq_out_params destroy_oparams; struct qed_rdma_destroy_cq_in_params destroy_iparams; + struct qed_chain_init_params chain_params = { + .mode = QED_CHAIN_MODE_PBL, + .intended_use = QED_CHAIN_USE_TO_CONSUME, + .cnt_type = QED_CHAIN_CNT_TYPE_U32, + .elem_size = sizeof(union rdma_cqe), + }; struct qedr_dev *dev = get_qedr_dev(ibdev); struct qed_rdma_create_cq_in_params params; struct qedr_create_cq_ureq ureq = {}; @@ -919,6 +927,7 @@ int qedr_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, chain_entries = qedr_align_cq_entries(entries); chain_entries = min_t(int, chain_entries, QEDR_MAX_CQES); + chain_params.num_elems = chain_entries; /* calc db offset. user will add DPI base, kernel will add db addr */ db_offset = DB_ADDR_SHIFT(DQ_PWM_OFFSET_UCM_RDMA_CQ_CONS_32BIT); @@ -953,13 +962,8 @@ int qedr_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, } else { cq->cq_type = QEDR_CQ_TYPE_KERNEL; - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_CONSUME, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U32, - chain_entries, - sizeof(union rdma_cqe), - &cq->pbl, NULL); + rc = dev->ops->common->chain_alloc(dev->cdev, &cq->pbl, + &chain_params); if (rc) goto err0; @@ -1448,6 +1452,12 @@ static int qedr_alloc_srq_kernel_params(struct qedr_srq *srq, struct ib_srq_init_attr *init_attr) { struct qedr_srq_hwq_info *hw_srq = &srq->hw_srq; + struct qed_chain_init_params params = { + .mode = QED_CHAIN_MODE_PBL, + .intended_use = QED_CHAIN_USE_TO_CONSUME_PRODUCE, + .cnt_type = QED_CHAIN_CNT_TYPE_U32, + .elem_size = QEDR_SRQ_WQE_ELEM_SIZE, + }; dma_addr_t phy_prod_pair_addr; u32 num_elems; void *va; @@ -1466,13 +1476,9 @@ static int qedr_alloc_srq_kernel_params(struct qedr_srq *srq, hw_srq->virt_prod_pair_addr = va; num_elems = init_attr->attr.max_wr * RDMA_MAX_SRQ_WQE_SIZE; - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_CONSUME_PRODUCE, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U32, - num_elems, - QEDR_SRQ_WQE_ELEM_SIZE, - &hw_srq->pbl, NULL); + params.num_elems = num_elems; + + rc = dev->ops->common->chain_alloc(dev->cdev, &hw_srq->pbl, ¶ms); if (rc) goto err0; @@ -1752,7 +1758,7 @@ static int qedr_create_user_qp(struct qedr_dev *dev, struct qed_rdma_create_qp_out_params out_params; struct qedr_pd *pd = get_qedr_pd(ibpd); struct qedr_create_qp_uresp uresp; - struct qedr_ucontext *ctx = NULL; + struct qedr_ucontext *ctx = pd ? pd->uctx : NULL; struct qedr_create_qp_ureq ureq; int alloc_and_init = rdma_protocol_roce(&dev->ibdev, 1); int rc = -EINVAL; @@ -1790,6 +1796,9 @@ static int qedr_create_user_qp(struct qedr_dev *dev, in_params.rq_pbl_ptr = qp->urq.pbl_tbl->pa; } + if (ctx) + SET_FIELD(in_params.flags, QED_ROCE_EDPM_MODE, ctx->edpm_mode); + qp->qed_qp = dev->ops->rdma_create_qp(dev->rdma_ctx, &in_params, &out_params); @@ -1903,29 +1912,28 @@ qedr_roce_create_kernel_qp(struct qedr_dev *dev, u32 n_sq_elems, u32 n_rq_elems) { struct qed_rdma_create_qp_out_params out_params; + struct qed_chain_init_params params = { + .mode = QED_CHAIN_MODE_PBL, + .cnt_type = QED_CHAIN_CNT_TYPE_U32, + }; int rc; - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_PRODUCE, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U32, - n_sq_elems, - QEDR_SQE_ELEMENT_SIZE, - &qp->sq.pbl, NULL); + params.intended_use = QED_CHAIN_USE_TO_PRODUCE; + params.num_elems = n_sq_elems; + params.elem_size = QEDR_SQE_ELEMENT_SIZE; + rc = dev->ops->common->chain_alloc(dev->cdev, &qp->sq.pbl, ¶ms); if (rc) return rc; in_params->sq_num_pages = qed_chain_get_page_cnt(&qp->sq.pbl); in_params->sq_pbl_ptr = qed_chain_get_pbl_phys(&qp->sq.pbl); - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_CONSUME_PRODUCE, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U32, - n_rq_elems, - QEDR_RQE_ELEMENT_SIZE, - &qp->rq.pbl, NULL); + params.intended_use = QED_CHAIN_USE_TO_CONSUME_PRODUCE; + params.num_elems = n_rq_elems; + params.elem_size = QEDR_RQE_ELEMENT_SIZE; + + rc = dev->ops->common->chain_alloc(dev->cdev, &qp->rq.pbl, ¶ms); if (rc) return rc; @@ -1951,14 +1959,19 @@ qedr_iwarp_create_kernel_qp(struct qedr_dev *dev, u32 n_sq_elems, u32 n_rq_elems) { struct qed_rdma_create_qp_out_params out_params; - struct qed_chain_ext_pbl ext_pbl; + struct qed_chain_init_params params = { + .mode = QED_CHAIN_MODE_PBL, + .cnt_type = QED_CHAIN_CNT_TYPE_U32, + }; int rc; in_params->sq_num_pages = QED_CHAIN_PAGE_CNT(n_sq_elems, QEDR_SQE_ELEMENT_SIZE, + QED_CHAIN_PAGE_SIZE, QED_CHAIN_MODE_PBL); in_params->rq_num_pages = QED_CHAIN_PAGE_CNT(n_rq_elems, QEDR_RQE_ELEMENT_SIZE, + QED_CHAIN_PAGE_SIZE, QED_CHAIN_MODE_PBL); qp->qed_qp = dev->ops->rdma_create_qp(dev->rdma_ctx, @@ -1968,31 +1981,24 @@ qedr_iwarp_create_kernel_qp(struct qedr_dev *dev, return -EINVAL; /* Now we allocate the chain */ - ext_pbl.p_pbl_virt = out_params.sq_pbl_virt; - ext_pbl.p_pbl_phys = out_params.sq_pbl_phys; - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_PRODUCE, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U32, - n_sq_elems, - QEDR_SQE_ELEMENT_SIZE, - &qp->sq.pbl, &ext_pbl); + params.intended_use = QED_CHAIN_USE_TO_PRODUCE; + params.num_elems = n_sq_elems; + params.elem_size = QEDR_SQE_ELEMENT_SIZE; + params.ext_pbl_virt = out_params.sq_pbl_virt; + params.ext_pbl_phys = out_params.sq_pbl_phys; + rc = dev->ops->common->chain_alloc(dev->cdev, &qp->sq.pbl, ¶ms); if (rc) goto err; - ext_pbl.p_pbl_virt = out_params.rq_pbl_virt; - ext_pbl.p_pbl_phys = out_params.rq_pbl_phys; - - rc = dev->ops->common->chain_alloc(dev->cdev, - QED_CHAIN_USE_TO_CONSUME_PRODUCE, - QED_CHAIN_MODE_PBL, - QED_CHAIN_CNT_TYPE_U32, - n_rq_elems, - QEDR_RQE_ELEMENT_SIZE, - &qp->rq.pbl, &ext_pbl); + params.intended_use = QED_CHAIN_USE_TO_CONSUME_PRODUCE; + params.num_elems = n_rq_elems; + params.elem_size = QEDR_RQE_ELEMENT_SIZE; + params.ext_pbl_virt = out_params.rq_pbl_virt; + params.ext_pbl_phys = out_params.rq_pbl_phys; + rc = dev->ops->common->chain_alloc(dev->cdev, &qp->rq.pbl, ¶ms); if (rc) goto err; @@ -2750,12 +2756,12 @@ int qedr_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) return 0; } -int qedr_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, u32 flags, +int qedr_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata) { struct qedr_ah *ah = get_qedr_ah(ibah); - rdma_copy_ah_attr(&ah->attr, attr); + rdma_copy_ah_attr(&ah->attr, init_attr->ah_attr); return 0; } @@ -3005,7 +3011,7 @@ err0: } struct ib_mr *qedr_alloc_mr(struct ib_pd *ibpd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct qedr_mr *mr; @@ -3522,7 +3528,7 @@ static int __qedr_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, break; case IB_WR_RDMA_READ_WITH_INV: SET_FIELD2(wqe->flags, RDMA_SQ_RDMA_WQE_1ST_READ_INV_FLG, 1); - /* fallthrough -- same is identical to RDMA READ */ + fallthrough; /* same is identical to RDMA READ */ case IB_WR_RDMA_READ: wqe->req_type = RDMA_SQ_REQ_TYPE_RDMA_RD; @@ -3688,7 +3694,7 @@ static u32 qedr_srq_elem_left(struct qedr_srq_hwq_info *hw_srq) * count and consumer count and subtract it from max * work request supported so that we get elements left. */ - used = hw_srq->wr_prod_cnt - hw_srq->wr_cons_cnt; + used = hw_srq->wr_prod_cnt - (u32)atomic_read(&hw_srq->wr_cons_cnt); return hw_srq->max_wr - used; } @@ -3703,7 +3709,6 @@ int qedr_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, unsigned long flags; int status = 0; u32 num_sge; - u32 offset; spin_lock_irqsave(&srq->lock, flags); @@ -3716,7 +3721,8 @@ int qedr_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, if (!qedr_srq_elem_left(hw_srq) || wr->num_sge > srq->hw_srq.max_sges) { DP_ERR(dev, "Can't post WR (%d,%d) || (%d > %d)\n", - hw_srq->wr_prod_cnt, hw_srq->wr_cons_cnt, + hw_srq->wr_prod_cnt, + atomic_read(&hw_srq->wr_cons_cnt), wr->num_sge, srq->hw_srq.max_sges); status = -ENOMEM; *bad_wr = wr; @@ -3750,22 +3756,20 @@ int qedr_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, hw_srq->sge_prod++; } - /* Flush WQE and SGE information before + /* Update WQE and SGE information before * updating producer. */ - wmb(); + dma_wmb(); /* SRQ producer is 8 bytes. Need to update SGE producer index * in first 4 bytes and need to update WQE producer in * next 4 bytes. */ - *srq->hw_srq.virt_prod_pair_addr = hw_srq->sge_prod; - offset = offsetof(struct rdma_srq_producers, wqe_prod); - *((u8 *)srq->hw_srq.virt_prod_pair_addr + offset) = - hw_srq->wqe_prod; + srq->hw_srq.virt_prod_pair_addr->sge_prod = hw_srq->sge_prod; + /* Make sure sge producer is updated first */ + dma_wmb(); + srq->hw_srq.virt_prod_pair_addr->wqe_prod = hw_srq->wqe_prod; - /* Flush producer after updating it. */ - wmb(); wr = wr->next; } @@ -4184,7 +4188,7 @@ static int process_resp_one_srq(struct qedr_dev *dev, struct qedr_qp *qp, } else { __process_resp_one(dev, qp, cq, wc, resp, wr_id); } - srq->hw_srq.wr_cons_cnt++; + atomic_inc(&srq->hw_srq.wr_cons_cnt); return 1; } diff --git a/drivers/infiniband/hw/qedr/verbs.h b/drivers/infiniband/hw/qedr/verbs.h index 18027844eb87..39dd6286ba39 100644 --- a/drivers/infiniband/hw/qedr/verbs.h +++ b/drivers/infiniband/hw/qedr/verbs.h @@ -70,7 +70,7 @@ int qedr_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr); void qedr_destroy_srq(struct ib_srq *ibsrq, struct ib_udata *udata); int qedr_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_recv_wr); -int qedr_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, u32 flags, +int qedr_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); void qedr_destroy_ah(struct ib_ah *ibah, u32 flags); @@ -84,7 +84,7 @@ int qedr_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); struct ib_mr *qedr_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int qedr_poll_cq(struct ib_cq *, int num_entries, struct ib_wc *wc); int qedr_post_send(struct ib_qp *, const struct ib_send_wr *, const struct ib_send_wr **bad_wr); diff --git a/drivers/infiniband/hw/qib/Kconfig b/drivers/infiniband/hw/qib/Kconfig index 376d19f29346..6c4895777042 100644 --- a/drivers/infiniband/hw/qib/Kconfig +++ b/drivers/infiniband/hw/qib/Kconfig @@ -3,7 +3,7 @@ config INFINIBAND_QIB tristate "Intel PCIe HCA support" depends on 64BIT && INFINIBAND_RDMAVT depends on PCI - ---help--- + help This is a low-level driver for Intel PCIe QLE InfiniBand host channel adapters. This driver does not support the Intel HyperTransport card (model QHT7140). @@ -12,6 +12,6 @@ config INFINIBAND_QIB_DCA bool "QIB DCA support" depends on INFINIBAND_QIB && DCA && SMP && !(INFINIBAND_QIB=y && DCA=m) default y - ---help--- + help Setting this enables DCA support on some Intel chip sets with the iba7322 HCA. diff --git a/drivers/infiniband/hw/qib/qib_file_ops.c b/drivers/infiniband/hw/qib/qib_file_ops.c index b0144229cf3b..ff87a67dd7b7 100644 --- a/drivers/infiniband/hw/qib/qib_file_ops.c +++ b/drivers/infiniband/hw/qib/qib_file_ops.c @@ -40,10 +40,10 @@ #include <linux/highmem.h> #include <linux/io.h> #include <linux/jiffies.h> -#include <asm/pgtable.h> #include <linux/delay.h> #include <linux/export.h> #include <linux/uio.h> +#include <linux/pgtable.h> #include <rdma/ib.h> diff --git a/drivers/infiniband/hw/qib/qib_iba6120.c b/drivers/infiniband/hw/qib/qib_iba6120.c index ca5ea734e3d0..44150be215bf 100644 --- a/drivers/infiniband/hw/qib/qib_iba6120.c +++ b/drivers/infiniband/hw/qib/qib_iba6120.c @@ -2973,11 +2973,11 @@ static u32 qib_6120_iblink_state(u64 ibcs) state = IB_PORT_ARMED; break; case IB_6120_L_STATE_ACTIVE: - /* fall through */ case IB_6120_L_STATE_ACT_DEFER: state = IB_PORT_ACTIVE; break; - default: /* fall through */ + default: + fallthrough; case IB_6120_L_STATE_DOWN: state = IB_PORT_DOWN; break; diff --git a/drivers/infiniband/hw/qib/qib_iba7220.c b/drivers/infiniband/hw/qib/qib_iba7220.c index ea3ddb05cbad..0a6f26d4cb31 100644 --- a/drivers/infiniband/hw/qib/qib_iba7220.c +++ b/drivers/infiniband/hw/qib/qib_iba7220.c @@ -3586,11 +3586,11 @@ static u32 qib_7220_iblink_state(u64 ibcs) state = IB_PORT_ARMED; break; case IB_7220_L_STATE_ACTIVE: - /* fall through */ case IB_7220_L_STATE_ACT_DEFER: state = IB_PORT_ACTIVE; break; - default: /* fall through */ + default: + fallthrough; case IB_7220_L_STATE_DOWN: state = IB_PORT_DOWN; break; diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c index 91d64dd71a8a..a10eab89aee4 100644 --- a/drivers/infiniband/hw/qib/qib_iba7322.c +++ b/drivers/infiniband/hw/qib/qib_iba7322.c @@ -2375,7 +2375,6 @@ static int qib_7322_bringup_serdes(struct qib_pportdata *ppd) struct qib_devdata *dd = ppd->dd; u64 val, guid, ibc; unsigned long flags; - int ret = 0; /* * SerDes model not in Pd, but still need to @@ -2510,7 +2509,7 @@ static int qib_7322_bringup_serdes(struct qib_pportdata *ppd) val | ERR_MASK_N(IBStatusChanged)); /* Always zero until we start messing with SerDes for real */ - return ret; + return 0; } /** @@ -5509,11 +5508,11 @@ static u32 qib_7322_iblink_state(u64 ibcs) state = IB_PORT_ARMED; break; case IB_7322_L_STATE_ACTIVE: - /* fall through */ case IB_7322_L_STATE_ACT_DEFER: state = IB_PORT_ACTIVE; break; - default: /* fall through */ + default: + fallthrough; case IB_7322_L_STATE_DOWN: state = IB_PORT_DOWN; break; @@ -6534,7 +6533,7 @@ static int qib_init_7322_variables(struct qib_devdata *dd) "Invalid num_vls %u, using 4 VLs\n", qib_num_cfg_vls); qib_num_cfg_vls = 4; - /* fall through */ + fallthrough; case 4: ppd->vls_supported = IB_VL_VL0_3; break; @@ -6875,7 +6874,7 @@ static int init_sdma_7322_regs(struct qib_pportdata *ppd) struct qib_devdata *dd = ppd->dd; unsigned lastbuf, erstbuf; u64 senddmabufmask[3] = { 0 }; - int n, ret = 0; + int n; qib_write_kreg_port(ppd, krp_senddmabase, ppd->sdma_descq_phys); qib_sdma_7322_setlengen(ppd); @@ -6904,7 +6903,7 @@ static int init_sdma_7322_regs(struct qib_pportdata *ppd) qib_write_kreg_port(ppd, krp_senddmabufmask0, senddmabufmask[0]); qib_write_kreg_port(ppd, krp_senddmabufmask1, senddmabufmask[1]); qib_write_kreg_port(ppd, krp_senddmabufmask2, senddmabufmask[2]); - return ret; + return 0; } /* sdma_lock must be held */ diff --git a/drivers/infiniband/hw/qib/qib_mad.c b/drivers/infiniband/hw/qib/qib_mad.c index 79bb83222e8d..e7789e724f56 100644 --- a/drivers/infiniband/hw/qib/qib_mad.c +++ b/drivers/infiniband/hw/qib/qib_mad.c @@ -433,7 +433,7 @@ static int check_mkey(struct qib_ibport *ibp, struct ib_smp *smp, int mad_flags) /* Bad mkey not a violation below level 2 */ if (ibp->rvp.mkeyprot < 2) break; - /* fall through */ + fallthrough; case IB_MGMT_METHOD_SET: case IB_MGMT_METHOD_TRAP_REPRESS: if (ibp->rvp.mkey_violations != 0xFFFF) @@ -828,7 +828,7 @@ static int subn_set_portinfo(struct ib_smp *smp, struct ib_device *ibdev, case IB_PORT_NOP: if (lstate == 0) break; - /* FALLTHROUGH */ + fallthrough; case IB_PORT_DOWN: if (lstate == 0) lstate = QIB_IB_LINKDOWN_ONLY; @@ -1928,7 +1928,7 @@ static int process_subn(struct ib_device *ibdev, int mad_flags, ret = IB_MAD_RESULT_SUCCESS; goto bail; } - /* FALLTHROUGH */ + fallthrough; default: smp->status |= IB_SMP_UNSUP_METH_ATTR; ret = reply(smp); @@ -1962,7 +1962,7 @@ static int process_subn(struct ib_device *ibdev, int mad_flags, ret = IB_MAD_RESULT_SUCCESS; goto bail; } - /* FALLTHROUGH */ + fallthrough; default: smp->status |= IB_SMP_UNSUP_METH_ATTR; ret = reply(smp); @@ -2322,7 +2322,7 @@ static int process_cc(struct ib_device *ibdev, int mad_flags, ret = cc_get_congestion_control_table(ccp, ibdev, port); goto bail; - /* FALLTHROUGH */ + fallthrough; default: ccp->status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_smp *) ccp); @@ -2339,7 +2339,7 @@ static int process_cc(struct ib_device *ibdev, int mad_flags, ret = cc_set_congestion_control_table(ccp, ibdev, port); goto bail; - /* FALLTHROUGH */ + fallthrough; default: ccp->status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_smp *) ccp); diff --git a/drivers/infiniband/hw/qib/qib_rc.c b/drivers/infiniband/hw/qib/qib_rc.c index aaf7438258fa..3915e5b4a9bc 100644 --- a/drivers/infiniband/hw/qib/qib_rc.c +++ b/drivers/infiniband/hw/qib/qib_rc.c @@ -83,7 +83,7 @@ static int qib_make_rc_ack(struct qib_ibdev *dev, struct rvt_qp *qp, rvt_put_mr(e->rdma_sge.mr); e->rdma_sge.mr = NULL; } - /* FALLTHROUGH */ + fallthrough; case OP(ATOMIC_ACKNOWLEDGE): /* * We can increment the tail pointer now that the last @@ -92,7 +92,7 @@ static int qib_make_rc_ack(struct qib_ibdev *dev, struct rvt_qp *qp, */ if (++qp->s_tail_ack_queue > QIB_MAX_RDMA_ATOMIC) qp->s_tail_ack_queue = 0; - /* FALLTHROUGH */ + fallthrough; case OP(SEND_ONLY): case OP(ACKNOWLEDGE): /* Check for no next entry in the queue. */ @@ -149,7 +149,7 @@ static int qib_make_rc_ack(struct qib_ibdev *dev, struct rvt_qp *qp, case OP(RDMA_READ_RESPONSE_FIRST): qp->s_ack_state = OP(RDMA_READ_RESPONSE_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_READ_RESPONSE_MIDDLE): qp->s_cur_sge = &qp->s_ack_rdma_sge; qp->s_rdma_mr = qp->s_ack_rdma_sge.sge.mr; @@ -471,10 +471,10 @@ no_flow_control: * See qib_restart_rc(). */ qp->s_len = restart_sge(&qp->s_sge, wqe, qp->s_psn, pmtu); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_FIRST): qp->s_state = OP(SEND_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): bth2 = qp->s_psn++ & QIB_PSN_MASK; ss = &qp->s_sge; @@ -510,10 +510,10 @@ no_flow_control: * See qib_restart_rc(). */ qp->s_len = restart_sge(&qp->s_sge, wqe, qp->s_psn, pmtu); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_FIRST): qp->s_state = OP(RDMA_WRITE_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_MIDDLE): bth2 = qp->s_psn++ & QIB_PSN_MASK; ss = &qp->s_sge; @@ -1807,7 +1807,7 @@ void qib_rc_rcv(struct qib_ctxtdata *rcd, struct ib_header *hdr, if (!ret) goto rnr_nak; qp->r_rcv_len = 0; - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): case OP(RDMA_WRITE_MIDDLE): send_middle: @@ -1839,7 +1839,7 @@ send_middle: qp->r_rcv_len = 0; if (opcode == OP(SEND_ONLY)) goto no_immediate_data; - /* fall through -- for SEND_ONLY_WITH_IMMEDIATE */ + fallthrough; /* for SEND_ONLY_WITH_IMMEDIATE */ case OP(SEND_LAST_WITH_IMMEDIATE): send_last_imm: wc.ex.imm_data = ohdr->u.imm_data; diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c index 99e11c347130..8f8d61736656 100644 --- a/drivers/infiniband/hw/qib/qib_sdma.c +++ b/drivers/infiniband/hw/qib/qib_sdma.c @@ -763,7 +763,7 @@ void __qib_sdma_process_event(struct qib_pportdata *ppd, * bringing the link up with traffic active on * 7220, e.g. */ ss->go_s99_running = 1; - /* fall through -- and start dma engine */ + fallthrough; /* and start dma engine */ case qib_sdma_event_e10_go_hw_start: /* This reference means the state machine is started */ sdma_get(&ppd->sdma_state); diff --git a/drivers/infiniband/hw/qib/qib_uc.c b/drivers/infiniband/hw/qib/qib_uc.c index e17b91e2c22a..554af4273a13 100644 --- a/drivers/infiniband/hw/qib/qib_uc.c +++ b/drivers/infiniband/hw/qib/qib_uc.c @@ -161,7 +161,7 @@ int qib_make_uc_req(struct rvt_qp *qp, unsigned long *flags) case OP(SEND_FIRST): qp->s_state = OP(SEND_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): len = qp->s_len; if (len > pmtu) { @@ -185,7 +185,7 @@ int qib_make_uc_req(struct rvt_qp *qp, unsigned long *flags) case OP(RDMA_WRITE_FIRST): qp->s_state = OP(RDMA_WRITE_MIDDLE); - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_MIDDLE): len = qp->s_len; if (len > pmtu) { @@ -351,7 +351,7 @@ send_first: goto no_immediate_data; else if (opcode == OP(SEND_ONLY_WITH_IMMEDIATE)) goto send_last_imm; - /* FALLTHROUGH */ + fallthrough; case OP(SEND_MIDDLE): /* Check for invalid length PMTU or posted rwqe len. */ if (unlikely(tlen != (hdrsize + pmtu + 4))) @@ -440,7 +440,7 @@ rdma_first: wc.ex.imm_data = ohdr->u.rc.imm_data; goto rdma_last_imm; } - /* FALLTHROUGH */ + fallthrough; case OP(RDMA_WRITE_MIDDLE): /* Check for invalid length PMTU or posted rwqe len. */ if (unlikely(tlen != (hdrsize + pmtu + 4))) diff --git a/drivers/infiniband/hw/qib/qib_user_pages.c b/drivers/infiniband/hw/qib/qib_user_pages.c index 342e3172ca40..4c24e83f3175 100644 --- a/drivers/infiniband/hw/qib/qib_user_pages.c +++ b/drivers/infiniband/hw/qib/qib_user_pages.c @@ -106,18 +106,18 @@ int qib_get_user_pages(unsigned long start_page, size_t num_pages, goto bail; } - down_read(¤t->mm->mmap_sem); + mmap_read_lock(current->mm); for (got = 0; got < num_pages; got += ret) { ret = pin_user_pages(start_page + got * PAGE_SIZE, num_pages - got, FOLL_LONGTERM | FOLL_WRITE | FOLL_FORCE, p + got, NULL); if (ret < 0) { - up_read(¤t->mm->mmap_sem); + mmap_read_unlock(current->mm); goto bail_release; } } - up_read(¤t->mm->mmap_sem); + mmap_read_unlock(current->mm); return 0; bail_release: diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c index 7508abb6a0fa..f6c01bad5a74 100644 --- a/drivers/infiniband/hw/qib/qib_verbs.c +++ b/drivers/infiniband/hw/qib/qib_verbs.c @@ -237,7 +237,7 @@ static void qib_qp_rcv(struct qib_ctxtdata *rcd, struct ib_header *hdr, case IB_QPT_GSI: if (ib_qib_disable_sma) break; - /* FALLTHROUGH */ + fallthrough; case IB_QPT_UD: qib_ud_rcv(ibp, hdr, has_grh, data, tlen, qp); break; @@ -1460,7 +1460,6 @@ static void qib_fill_device_attr(struct qib_devdata *dd) rdi->dparms.props.max_cq = ib_qib_max_cqs; rdi->dparms.props.max_cqe = ib_qib_max_cqes; rdi->dparms.props.max_ah = ib_qib_max_ahs; - rdi->dparms.props.max_map_per_fmr = 32767; rdi->dparms.props.max_qp_rd_atom = QIB_MAX_RDMA_ATOMIC; rdi->dparms.props.max_qp_init_rd_atom = 255; rdi->dparms.props.max_srq = ib_qib_max_srqs; diff --git a/drivers/infiniband/hw/usnic/Kconfig b/drivers/infiniband/hw/usnic/Kconfig index c0847d9e2539..9019599c173b 100644 --- a/drivers/infiniband/hw/usnic/Kconfig +++ b/drivers/infiniband/hw/usnic/Kconfig @@ -6,6 +6,6 @@ config INFINIBAND_USNIC select ENIC select NET_VENDOR_CISCO select PCI_IOV - ---help--- + help This is a low-level driver for Cisco's Virtual Interface Cards (VICs), including the VIC 1240 and 1280 cards. diff --git a/drivers/infiniband/hw/usnic/usnic_fwd.c b/drivers/infiniband/hw/usnic/usnic_fwd.c index 7875883621f4..398c4c00b932 100644 --- a/drivers/infiniband/hw/usnic/usnic_fwd.c +++ b/drivers/infiniband/hw/usnic/usnic_fwd.c @@ -214,7 +214,7 @@ usnic_fwd_alloc_flow(struct usnic_fwd_dev *ufdev, struct filter *filter, if (!flow) return ERR_PTR(-ENOMEM); - tlv = pci_alloc_consistent(pdev, tlv_size, &tlv_pa); + tlv = dma_alloc_coherent(&pdev->dev, tlv_size, &tlv_pa, GFP_ATOMIC); if (!tlv) { usnic_err("Failed to allocate memory\n"); status = -ENOMEM; @@ -258,7 +258,7 @@ usnic_fwd_alloc_flow(struct usnic_fwd_dev *ufdev, struct filter *filter, out_free_tlv: spin_unlock(&ufdev->lock); - pci_free_consistent(pdev, tlv_size, tlv, tlv_pa); + dma_free_coherent(&pdev->dev, tlv_size, tlv, tlv_pa); if (!status) return flow; out_free_flow: diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c index c9abe1c01e4e..662e7fc7f628 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_main.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c @@ -120,7 +120,7 @@ static void usnic_ib_qp_grp_modify_active_to_err(struct usnic_ib_dev *us_ibdev) IB_QPS_ERR, NULL); if (status) { - usnic_err("Failed to transistion qp grp %u from %s to %s\n", + usnic_err("Failed to transition qp grp %u from %s to %s\n", qp_grp->grp_id, usnic_ib_qp_grp_state_to_string (cur_state), diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c index 71f82339446c..b8a77ce11590 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c @@ -322,7 +322,6 @@ int usnic_ib_query_device(struct ib_device *ibdev, props->max_mcast_grp = 0; props->max_mcast_qp_attach = 0; props->max_total_mcast_qp_attach = 0; - props->max_map_per_fmr = 0; /* Owned by Userspace * max_qp_wr, max_sge, max_sge_rd, max_cqe */ mutex_unlock(&us_ibdev->usdev_lock); diff --git a/drivers/infiniband/hw/usnic/usnic_uiom.c b/drivers/infiniband/hw/usnic/usnic_uiom.c index bd9f944b68fc..760b254ba42d 100644 --- a/drivers/infiniband/hw/usnic/usnic_uiom.c +++ b/drivers/infiniband/hw/usnic/usnic_uiom.c @@ -123,7 +123,7 @@ static int usnic_uiom_get_pages(unsigned long addr, size_t size, int writable, npages = PAGE_ALIGN(size + (addr & ~PAGE_MASK)) >> PAGE_SHIFT; uiomr->owning_mm = mm = current->mm; - down_read(&mm->mmap_sem); + mmap_read_lock(mm); locked = atomic64_add_return(npages, ¤t->mm->pinned_vm); lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT; @@ -187,7 +187,7 @@ out: } else mmgrab(uiomr->owning_mm); - up_read(&mm->mmap_sem); + mmap_read_unlock(mm); free_page((unsigned long) page_list); return ret; } diff --git a/drivers/infiniband/hw/vmw_pvrdma/Kconfig b/drivers/infiniband/hw/vmw_pvrdma/Kconfig index b99c9f0fc06a..a11e892e5e77 100644 --- a/drivers/infiniband/hw/vmw_pvrdma/Kconfig +++ b/drivers/infiniband/hw/vmw_pvrdma/Kconfig @@ -2,7 +2,7 @@ config INFINIBAND_VMWARE_PVRDMA tristate "VMware Paravirtualized RDMA Driver" depends on NETDEVICES && ETHERNET && PCI && INET && VMXNET3 - ---help--- + help This driver provides low-level support for VMware Paravirtual RDMA adapter. It interacts with the VMXNet3 driver to provide Ethernet capabilities. diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_mr.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_mr.c index b039f1f00e05..77a010e68208 100644 --- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_mr.c +++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_mr.c @@ -202,7 +202,7 @@ err_umem: * @return: ib_mr pointer on success, otherwise returns an errno. */ struct ib_mr *pvrdma_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct pvrdma_dev *dev = to_vdev(pd->device); struct pvrdma_user_mr *mr; diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c index afcc2abcf55c..9a8f2a9507be 100644 --- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c +++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c @@ -238,7 +238,7 @@ struct ib_qp *pvrdma_create_qp(struct ib_pd *pd, ret = -EINVAL; goto err_qp; } - /* fall through */ + fallthrough; case IB_QPT_RC: case IB_QPT_UD: qp = kzalloc(sizeof(*qp), GFP_KERNEL); diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c index faf7ecd7b3fa..ccbded2d26ce 100644 --- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c +++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c @@ -509,9 +509,10 @@ void pvrdma_dealloc_pd(struct ib_pd *pd, struct ib_udata *udata) * * @return: 0 on success, otherwise errno. */ -int pvrdma_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, - u32 flags, struct ib_udata *udata) +int pvrdma_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { + struct rdma_ah_attr *ah_attr = init_attr->ah_attr; struct pvrdma_dev *dev = to_vdev(ibah->device); struct pvrdma_ah *ah = to_vah(ibah); const struct ib_global_route *grh; diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h index e4a48f5c0c85..699b20849a7e 100644 --- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h +++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.h @@ -406,7 +406,7 @@ struct ib_mr *pvrdma_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, struct ib_udata *udata); int pvrdma_dereg_mr(struct ib_mr *mr, struct ib_udata *udata); struct ib_mr *pvrdma_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int pvrdma_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); int pvrdma_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, @@ -414,7 +414,7 @@ int pvrdma_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, void pvrdma_destroy_cq(struct ib_cq *cq, struct ib_udata *udata); int pvrdma_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc); int pvrdma_req_notify_cq(struct ib_cq *cq, enum ib_cq_notify_flags flags); -int pvrdma_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, u32 flags, +int pvrdma_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, struct ib_udata *udata); void pvrdma_destroy_ah(struct ib_ah *ah, u32 flags); diff --git a/drivers/infiniband/sw/rdmavt/Kconfig b/drivers/infiniband/sw/rdmavt/Kconfig index 1f2759c72108..9ef5f5ce1ff6 100644 --- a/drivers/infiniband/sw/rdmavt/Kconfig +++ b/drivers/infiniband/sw/rdmavt/Kconfig @@ -4,5 +4,5 @@ config INFINIBAND_RDMAVT depends on X86_64 && ARCH_DMA_ADDR_T_64BIT depends on PCI select DMA_VIRT_OPS - ---help--- + help This is a common software verbs provider for RDMA networks. diff --git a/drivers/infiniband/sw/rdmavt/ah.c b/drivers/infiniband/sw/rdmavt/ah.c index ee02c6176007..75a04b1497c4 100644 --- a/drivers/infiniband/sw/rdmavt/ah.c +++ b/drivers/infiniband/sw/rdmavt/ah.c @@ -90,22 +90,21 @@ EXPORT_SYMBOL(rvt_check_ah); /** * rvt_create_ah - create an address handle * @ibah: the IB address handle - * @ah_attr: the attributes of the AH - * @create_flags: create address handle flags (see enum rdma_create_ah_flags) + * @init_attr: the attributes of the AH * @udata: pointer to user's input output buffer information. * * This may be called from interrupt context. * * Return: 0 on success */ -int rvt_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, - u32 create_flags, struct ib_udata *udata) +int rvt_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { struct rvt_ah *ah = ibah_to_rvtah(ibah); struct rvt_dev_info *dev = ib_to_rvt(ibah->device); unsigned long flags; - if (rvt_check_ah(ibah->device, ah_attr)) + if (rvt_check_ah(ibah->device, init_attr->ah_attr)) return -EINVAL; spin_lock_irqsave(&dev->n_ahs_lock, flags); @@ -117,10 +116,11 @@ int rvt_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, dev->n_ahs_allocated++; spin_unlock_irqrestore(&dev->n_ahs_lock, flags); - rdma_copy_ah_attr(&ah->attr, ah_attr); + rdma_copy_ah_attr(&ah->attr, init_attr->ah_attr); if (dev->driver_f.notify_new_ah) - dev->driver_f.notify_new_ah(ibah->device, ah_attr, ah); + dev->driver_f.notify_new_ah(ibah->device, + init_attr->ah_attr, ah); return 0; } diff --git a/drivers/infiniband/sw/rdmavt/ah.h b/drivers/infiniband/sw/rdmavt/ah.h index bbb4d3bdec4e..40b7123fec76 100644 --- a/drivers/infiniband/sw/rdmavt/ah.h +++ b/drivers/infiniband/sw/rdmavt/ah.h @@ -50,8 +50,8 @@ #include <rdma/rdma_vt.h> -int rvt_create_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr, - u32 create_flags, struct ib_udata *udata); +int rvt_create_ah(struct ib_ah *ah, struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata); void rvt_destroy_ah(struct ib_ah *ibah, u32 destroy_flags); int rvt_modify_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr); int rvt_query_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr); diff --git a/drivers/infiniband/sw/rdmavt/mmap.c b/drivers/infiniband/sw/rdmavt/mmap.c index 37853aa3bcf7..f5d0e33cf3d7 100644 --- a/drivers/infiniband/sw/rdmavt/mmap.c +++ b/drivers/infiniband/sw/rdmavt/mmap.c @@ -48,7 +48,6 @@ #include <linux/slab.h> #include <linux/vmalloc.h> #include <linux/mm.h> -#include <asm/pgtable.h> #include <rdma/uverbs_ioctl.h> #include "mmap.h" diff --git a/drivers/infiniband/sw/rdmavt/mr.c b/drivers/infiniband/sw/rdmavt/mr.c index 72f6534fbb52..2f7c25fea44a 100644 --- a/drivers/infiniband/sw/rdmavt/mr.c +++ b/drivers/infiniband/sw/rdmavt/mr.c @@ -97,7 +97,6 @@ int rvt_driver_mr_init(struct rvt_dev_info *rdi) RCU_INIT_POINTER(rdi->lkey_table.table[i], NULL); rdi->dparms.props.max_mr = rdi->lkey_table.max; - rdi->dparms.props.max_fmr = rdi->lkey_table.max; return 0; } @@ -577,7 +576,7 @@ out: * Return: the memory region on success, otherwise return an errno. */ struct ib_mr *rvt_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct rvt_mr *mr; @@ -714,160 +713,6 @@ bail: EXPORT_SYMBOL(rvt_invalidate_rkey); /** - * rvt_alloc_fmr - allocate a fast memory region - * @pd: the protection domain for this memory region - * @mr_access_flags: access flags for this memory region - * @fmr_attr: fast memory region attributes - * - * Return: the memory region on success, otherwise returns an errno. - */ -struct ib_fmr *rvt_alloc_fmr(struct ib_pd *pd, int mr_access_flags, - struct ib_fmr_attr *fmr_attr) -{ - struct rvt_fmr *fmr; - int m; - struct ib_fmr *ret; - int rval = -ENOMEM; - - /* Allocate struct plus pointers to first level page tables. */ - m = (fmr_attr->max_pages + RVT_SEGSZ - 1) / RVT_SEGSZ; - fmr = kzalloc(struct_size(fmr, mr.map, m), GFP_KERNEL); - if (!fmr) - goto bail; - - rval = rvt_init_mregion(&fmr->mr, pd, fmr_attr->max_pages, - PERCPU_REF_INIT_ATOMIC); - if (rval) - goto bail; - - /* - * ib_alloc_fmr() will initialize fmr->ibfmr except for lkey & - * rkey. - */ - rval = rvt_alloc_lkey(&fmr->mr, 0); - if (rval) - goto bail_mregion; - fmr->ibfmr.rkey = fmr->mr.lkey; - fmr->ibfmr.lkey = fmr->mr.lkey; - /* - * Resources are allocated but no valid mapping (RKEY can't be - * used). - */ - fmr->mr.access_flags = mr_access_flags; - fmr->mr.max_segs = fmr_attr->max_pages; - fmr->mr.page_shift = fmr_attr->page_shift; - - ret = &fmr->ibfmr; -done: - return ret; - -bail_mregion: - rvt_deinit_mregion(&fmr->mr); -bail: - kfree(fmr); - ret = ERR_PTR(rval); - goto done; -} - -/** - * rvt_map_phys_fmr - set up a fast memory region - * @ibfmr: the fast memory region to set up - * @page_list: the list of pages to associate with the fast memory region - * @list_len: the number of pages to associate with the fast memory region - * @iova: the virtual address of the start of the fast memory region - * - * This may be called from interrupt context. - * - * Return: 0 on success - */ - -int rvt_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int list_len, u64 iova) -{ - struct rvt_fmr *fmr = to_ifmr(ibfmr); - struct rvt_lkey_table *rkt; - unsigned long flags; - int m, n; - unsigned long i; - u32 ps; - struct rvt_dev_info *rdi = ib_to_rvt(ibfmr->device); - - i = atomic_long_read(&fmr->mr.refcount.count); - if (i > 2) - return -EBUSY; - - if (list_len > fmr->mr.max_segs) - return -EINVAL; - - rkt = &rdi->lkey_table; - spin_lock_irqsave(&rkt->lock, flags); - fmr->mr.user_base = iova; - fmr->mr.iova = iova; - ps = 1 << fmr->mr.page_shift; - fmr->mr.length = list_len * ps; - m = 0; - n = 0; - for (i = 0; i < list_len; i++) { - fmr->mr.map[m]->segs[n].vaddr = (void *)page_list[i]; - fmr->mr.map[m]->segs[n].length = ps; - trace_rvt_mr_fmr_seg(&fmr->mr, m, n, (void *)page_list[i], ps); - if (++n == RVT_SEGSZ) { - m++; - n = 0; - } - } - spin_unlock_irqrestore(&rkt->lock, flags); - return 0; -} - -/** - * rvt_unmap_fmr - unmap fast memory regions - * @fmr_list: the list of fast memory regions to unmap - * - * Return: 0 on success. - */ -int rvt_unmap_fmr(struct list_head *fmr_list) -{ - struct rvt_fmr *fmr; - struct rvt_lkey_table *rkt; - unsigned long flags; - struct rvt_dev_info *rdi; - - list_for_each_entry(fmr, fmr_list, ibfmr.list) { - rdi = ib_to_rvt(fmr->ibfmr.device); - rkt = &rdi->lkey_table; - spin_lock_irqsave(&rkt->lock, flags); - fmr->mr.user_base = 0; - fmr->mr.iova = 0; - fmr->mr.length = 0; - spin_unlock_irqrestore(&rkt->lock, flags); - } - return 0; -} - -/** - * rvt_dealloc_fmr - deallocate a fast memory region - * @ibfmr: the fast memory region to deallocate - * - * Return: 0 on success. - */ -int rvt_dealloc_fmr(struct ib_fmr *ibfmr) -{ - struct rvt_fmr *fmr = to_ifmr(ibfmr); - int ret = 0; - - rvt_free_lkey(&fmr->mr); - rvt_put_mr(&fmr->mr); /* will set completion if last */ - ret = rvt_check_refs(&fmr->mr, __func__); - if (ret) - goto out; - rvt_deinit_mregion(&fmr->mr); - kfree(fmr); -out: - return ret; -} - -/** * rvt_sge_adjacent - is isge compressible * @last_sge: last outgoing SGE written * @sge: SGE to check diff --git a/drivers/infiniband/sw/rdmavt/mr.h b/drivers/infiniband/sw/rdmavt/mr.h index 2c8d0752e8e3..b3aba359401b 100644 --- a/drivers/infiniband/sw/rdmavt/mr.h +++ b/drivers/infiniband/sw/rdmavt/mr.h @@ -49,10 +49,6 @@ */ #include <rdma/rdma_vt.h> -struct rvt_fmr { - struct ib_fmr ibfmr; - struct rvt_mregion mr; /* must be last */ -}; struct rvt_mr { struct ib_mr ibmr; @@ -60,11 +56,6 @@ struct rvt_mr { struct rvt_mregion mr; /* must be last */ }; -static inline struct rvt_fmr *to_ifmr(struct ib_fmr *ibfmr) -{ - return container_of(ibfmr, struct rvt_fmr, ibfmr); -} - static inline struct rvt_mr *to_imr(struct ib_mr *ibmr) { return container_of(ibmr, struct rvt_mr, ibmr); @@ -80,14 +71,8 @@ struct ib_mr *rvt_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, struct ib_udata *udata); int rvt_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata); struct ib_mr *rvt_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata); + u32 max_num_sg); int rvt_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset); -struct ib_fmr *rvt_alloc_fmr(struct ib_pd *pd, int mr_access_flags, - struct ib_fmr_attr *fmr_attr); -int rvt_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, - int list_len, u64 iova); -int rvt_unmap_fmr(struct list_head *fmr_list); -int rvt_dealloc_fmr(struct ib_fmr *ibfmr); #endif /* DEF_RVTMR_H */ diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c index 500a7ee04c44..ee48befc8978 100644 --- a/drivers/infiniband/sw/rdmavt/qp.c +++ b/drivers/infiniband/sw/rdmavt/qp.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2016 - 2019 Intel Corporation. + * Copyright(c) 2016 - 2020 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -525,15 +525,18 @@ static inline unsigned mk_qpn(struct rvt_qpn_table *qpt, * @rdi: rvt device info structure * @qpt: queue pair number table pointer * @port_num: IB port number, 1 based, comes from core + * @exclude_prefix: prefix of special queue pair number being allocated * * Return: The queue pair number */ static int alloc_qpn(struct rvt_dev_info *rdi, struct rvt_qpn_table *qpt, - enum ib_qp_type type, u8 port_num) + enum ib_qp_type type, u8 port_num, u8 exclude_prefix) { u32 i, offset, max_scan, qpn; struct rvt_qpn_map *map; u32 ret; + u32 max_qpn = exclude_prefix == RVT_AIP_QP_PREFIX ? + RVT_AIP_QPN_MAX : RVT_QPN_MAX; if (rdi->driver_f.alloc_qpn) return rdi->driver_f.alloc_qpn(rdi, qpt, type, port_num); @@ -553,7 +556,7 @@ static int alloc_qpn(struct rvt_dev_info *rdi, struct rvt_qpn_table *qpt, } qpn = qpt->last + qpt->incr; - if (qpn >= RVT_QPN_MAX) + if (qpn >= max_qpn) qpn = qpt->incr | ((qpt->last & 1) ^ 1); /* offset carries bit 0 */ offset = qpn & RVT_BITS_PER_PAGE_MASK; @@ -898,8 +901,6 @@ static void rvt_init_qp(struct rvt_dev_info *rdi, struct rvt_qp *qp, qp->s_tail_ack_queue = 0; qp->s_acked_ack_queue = 0; qp->s_num_rd_atomic = 0; - if (qp->r_rq.kwq) - qp->r_rq.kwq->count = qp->r_rq.size; qp->r_sge.num_sge = 0; atomic_set(&qp->s_reserved_used, 0); } @@ -987,6 +988,9 @@ static void rvt_free_qpn(struct rvt_qpn_table *qpt, u32 qpn) { struct rvt_qpn_map *map; + if ((qpn & RVT_AIP_QP_PREFIX_MASK) == RVT_AIP_QP_BASE) + qpn &= RVT_AIP_QP_SUFFIX; + map = qpt->map + (qpn & RVT_QPN_MASK) / RVT_BITS_PER_PAGE; if (map->page) clear_bit(qpn & RVT_BITS_PER_PAGE_MASK, map->page); @@ -1074,13 +1078,15 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd, struct rvt_dev_info *rdi = ib_to_rvt(ibpd->device); void *priv = NULL; size_t sqsize; + u8 exclude_prefix = 0; if (!rdi) return ERR_PTR(-EINVAL); if (init_attr->cap.max_send_sge > rdi->dparms.props.max_send_sge || init_attr->cap.max_send_wr > rdi->dparms.props.max_qp_wr || - init_attr->create_flags) + (init_attr->create_flags && + init_attr->create_flags != IB_QP_CREATE_NETDEV_USE)) return ERR_PTR(-EINVAL); /* Check receive queue parameters if no SRQ is specified. */ @@ -1105,7 +1111,7 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd, if (init_attr->port_num == 0 || init_attr->port_num > ibpd->device->phys_port_cnt) return ERR_PTR(-EINVAL); - /* fall through */ + fallthrough; case IB_QPT_UC: case IB_QPT_RC: case IB_QPT_UD: @@ -1196,17 +1202,23 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd, err = alloc_ud_wq_attr(qp, rdi->dparms.node); if (err) { ret = (ERR_PTR(err)); - goto bail_driver_priv; + goto bail_rq_rvt; } + if (init_attr->create_flags & IB_QP_CREATE_NETDEV_USE) + exclude_prefix = RVT_AIP_QP_PREFIX; + err = alloc_qpn(rdi, &rdi->qp_dev->qpn_table, init_attr->qp_type, - init_attr->port_num); + init_attr->port_num, + exclude_prefix); if (err < 0) { ret = ERR_PTR(err); goto bail_rq_wq; } qp->ibqp.qp_num = err; + if (init_attr->create_flags & IB_QP_CREATE_NETDEV_USE) + qp->ibqp.qp_num |= RVT_AIP_QP_BASE; qp->port_num = init_attr->port_num; rvt_init_qp(rdi, qp, init_attr->qp_type); if (rdi->driver_f.qp_priv_init) { @@ -1300,9 +1312,11 @@ bail_qpn: rvt_free_qpn(&rdi->qp_dev->qpn_table, qp->ibqp.qp_num); bail_rq_wq: - rvt_free_rq(&qp->r_rq); free_ud_wq_attr(qp); +bail_rq_rvt: + rvt_free_rq(&qp->r_rq); + bail_driver_priv: rdi->driver_f.qp_priv_free(rdi, qp); @@ -2351,31 +2365,6 @@ bad_lkey: } /** - * get_count - count numbers of request work queue entries - * in circular buffer - * @rq: data structure for request queue entry - * @tail: tail indices of the circular buffer - * @head: head indices of the circular buffer - * - * Return - total number of entries in the circular buffer - */ -static u32 get_count(struct rvt_rq *rq, u32 tail, u32 head) -{ - u32 count; - - count = head; - - if (count >= rq->size) - count = 0; - if (count < tail) - count += rq->size - tail; - else - count -= tail; - - return count; -} - -/** * get_rvt_head - get head indices of the circular buffer * @rq: data structure for request queue entry * @ip: the QP @@ -2449,7 +2438,7 @@ int rvt_get_rwqe(struct rvt_qp *qp, bool wr_id_only) if (kwq->count < RVT_RWQ_COUNT_THRESHOLD) { head = get_rvt_head(rq, ip); - kwq->count = get_count(rq, tail, head); + kwq->count = rvt_get_rq_count(rq, head, tail); } if (unlikely(kwq->count == 0)) { ret = 0; @@ -2484,7 +2473,9 @@ int rvt_get_rwqe(struct rvt_qp *qp, bool wr_id_only) * the number of remaining WQEs. */ if (kwq->count < srq->limit) { - kwq->count = get_count(rq, tail, get_rvt_head(rq, ip)); + kwq->count = + rvt_get_rq_count(rq, + get_rvt_head(rq, ip), tail); if (kwq->count < srq->limit) { struct ib_event ev; diff --git a/drivers/infiniband/sw/rdmavt/rc.c b/drivers/infiniband/sw/rdmavt/rc.c index 977906cc0d11..c58735f4c94a 100644 --- a/drivers/infiniband/sw/rdmavt/rc.c +++ b/drivers/infiniband/sw/rdmavt/rc.c @@ -127,9 +127,7 @@ __be32 rvt_compute_aeth(struct rvt_qp *qp) * not atomic, which is OK, since the fuzziness is * resolved as further ACKs go out. */ - credits = head - tail; - if ((int)credits < 0) - credits += qp->r_rq.size; + credits = rvt_get_rq_count(&qp->r_rq, head, tail); } /* * Binary search the credit table to find the code to diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c index 72b031ab7092..f904bb34477a 100644 --- a/drivers/infiniband/sw/rdmavt/vt.c +++ b/drivers/infiniband/sw/rdmavt/vt.c @@ -378,7 +378,6 @@ enum { static const struct ib_device_ops rvt_dev_ops = { .uverbs_abi_ver = RVT_UVERBS_ABI_VERSION, - .alloc_fmr = rvt_alloc_fmr, .alloc_mr = rvt_alloc_mr, .alloc_pd = rvt_alloc_pd, .alloc_ucontext = rvt_alloc_ucontext, @@ -387,7 +386,6 @@ static const struct ib_device_ops rvt_dev_ops = { .create_cq = rvt_create_cq, .create_qp = rvt_create_qp, .create_srq = rvt_create_srq, - .dealloc_fmr = rvt_dealloc_fmr, .dealloc_pd = rvt_dealloc_pd, .dealloc_ucontext = rvt_dealloc_ucontext, .dereg_mr = rvt_dereg_mr, @@ -399,7 +397,6 @@ static const struct ib_device_ops rvt_dev_ops = { .get_dma_mr = rvt_get_dma_mr, .get_port_immutable = rvt_get_port_immutable, .map_mr_sg = rvt_map_mr_sg, - .map_phys_fmr = rvt_map_phys_fmr, .mmap = rvt_mmap, .modify_ah = rvt_modify_ah, .modify_device = rvt_modify_device, @@ -420,7 +417,6 @@ static const struct ib_device_ops rvt_dev_ops = { .reg_user_mr = rvt_reg_user_mr, .req_notify_cq = rvt_req_notify_cq, .resize_cq = rvt_resize_cq, - .unmap_fmr = rvt_unmap_fmr, INIT_RDMA_OBJ_SIZE(ib_ah, rvt_ah, ibah), INIT_RDMA_OBJ_SIZE(ib_cq, rvt_cq, ibcq), diff --git a/drivers/infiniband/sw/rxe/Kconfig b/drivers/infiniband/sw/rxe/Kconfig index d9bcfe740588..a0c6c7dfc181 100644 --- a/drivers/infiniband/sw/rxe/Kconfig +++ b/drivers/infiniband/sw/rxe/Kconfig @@ -6,7 +6,7 @@ config RDMA_RXE select NET_UDP_TUNNEL select CRYPTO_CRC32 select DMA_VIRT_OPS - ---help--- + help This driver implements the InfiniBand RDMA transport over the Linux network stack. It enables a system with a standard Ethernet adapter to interoperate with a RoCE diff --git a/drivers/infiniband/sw/rxe/rxe.c b/drivers/infiniband/sw/rxe/rxe.c index 4afdd2e20883..77f2c7cd1216 100644 --- a/drivers/infiniband/sw/rxe/rxe.c +++ b/drivers/infiniband/sw/rxe/rxe.c @@ -40,13 +40,7 @@ MODULE_AUTHOR("Bob Pearson, Frank Zago, John Groves, Kamal Heib"); MODULE_DESCRIPTION("Soft RDMA transport"); MODULE_LICENSE("Dual BSD/GPL"); -/* free resources for all ports on a device */ -static void rxe_cleanup_ports(struct rxe_dev *rxe) -{ - kfree(rxe->port.pkey_tbl); - rxe->port.pkey_tbl = NULL; - -} +bool rxe_initialized; /* free resources for a rxe device all objects created for this device must * have been destroyed @@ -66,8 +60,6 @@ void rxe_dealloc(struct ib_device *ib_dev) rxe_pool_cleanup(&rxe->mc_grp_pool); rxe_pool_cleanup(&rxe->mc_elem_pool); - rxe_cleanup_ports(rxe); - if (rxe->tfm) crypto_free_shash(rxe->tfm); } @@ -77,6 +69,7 @@ static void rxe_init_device_param(struct rxe_dev *rxe) { rxe->max_inline_data = RXE_MAX_INLINE_DATA; + rxe->attr.vendor_id = RXE_VENDOR_ID; rxe->attr.max_mr_size = RXE_MAX_MR_SIZE; rxe->attr.page_size_cap = RXE_PAGE_SIZE_CAP; rxe->attr.max_qp = RXE_MAX_QP; @@ -110,7 +103,7 @@ static void rxe_init_device_param(struct rxe_dev *rxe) } /* initialize port attributes */ -static int rxe_init_port_param(struct rxe_port *port) +static void rxe_init_port_param(struct rxe_port *port) { port->attr.state = IB_PORT_DOWN; port->attr.max_mtu = IB_MTU_4096; @@ -133,35 +126,19 @@ static int rxe_init_port_param(struct rxe_port *port) port->attr.phys_state = RXE_PORT_PHYS_STATE; port->mtu_cap = ib_mtu_enum_to_int(IB_MTU_256); port->subnet_prefix = cpu_to_be64(RXE_PORT_SUBNET_PREFIX); - - return 0; } /* initialize port state, note IB convention that HCA ports are always * numbered from 1 */ -static int rxe_init_ports(struct rxe_dev *rxe) +static void rxe_init_ports(struct rxe_dev *rxe) { struct rxe_port *port = &rxe->port; rxe_init_port_param(port); - - if (!port->attr.pkey_tbl_len || !port->attr.gid_tbl_len) - return -EINVAL; - - port->pkey_tbl = kcalloc(port->attr.pkey_tbl_len, - sizeof(*port->pkey_tbl), GFP_KERNEL); - - if (!port->pkey_tbl) - return -ENOMEM; - - port->pkey_tbl[0] = 0xffff; addrconf_addr_eui48((unsigned char *)&port->port_guid, rxe->ndev->dev_addr); - spin_lock_init(&port->port_lock); - - return 0; } /* init pools of managed objects */ @@ -251,13 +228,11 @@ static int rxe_init(struct rxe_dev *rxe) /* init default device parameters */ rxe_init_device_param(rxe); - err = rxe_init_ports(rxe); - if (err) - goto err1; + rxe_init_ports(rxe); err = rxe_init_pools(rxe); if (err) - goto err2; + return err; /* init pending mmap list */ spin_lock_init(&rxe->mmap_offset_lock); @@ -267,11 +242,6 @@ static int rxe_init(struct rxe_dev *rxe) mutex_init(&rxe->usdev_lock); return 0; - -err2: - rxe_cleanup_ports(rxe); -err1: - return err; } void rxe_set_mtu(struct rxe_dev *rxe, unsigned int ndev_mtu) @@ -347,6 +317,7 @@ static int __init rxe_module_init(void) return err; rdma_link_register(&rxe_link_ops); + rxe_initialized = true; pr_info("loaded\n"); return 0; } @@ -358,6 +329,7 @@ static void __exit rxe_module_exit(void) rxe_net_exit(); rxe_cache_exit(); + rxe_initialized = false; pr_info("unloaded\n"); } diff --git a/drivers/infiniband/sw/rxe/rxe.h b/drivers/infiniband/sw/rxe/rxe.h index fb07eed9e402..cae1b0a24c85 100644 --- a/drivers/infiniband/sw/rxe/rxe.h +++ b/drivers/infiniband/sw/rxe/rxe.h @@ -67,6 +67,8 @@ #define RXE_ROCE_V2_SPORT (0xc000) +extern bool rxe_initialized; + static inline u32 rxe_crc32(struct rxe_dev *rxe, u32 crc, void *next, size_t len) { diff --git a/drivers/infiniband/sw/rxe/rxe_comp.c b/drivers/infiniband/sw/rxe/rxe_comp.c index 4bc88708b355..7b4df0028388 100644 --- a/drivers/infiniband/sw/rxe/rxe_comp.c +++ b/drivers/infiniband/sw/rxe/rxe_comp.c @@ -282,7 +282,7 @@ static inline enum comp_state check_ack(struct rxe_qp *qp, if ((syn & AETH_TYPE_MASK) != AETH_ACK) return COMPST_ERROR; - /* fall through */ + fallthrough; /* (IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE doesn't have an AETH) */ case IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE: diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h index 775c23becaec..39dc3bfa5d5d 100644 --- a/drivers/infiniband/sw/rxe/rxe_loc.h +++ b/drivers/infiniband/sw/rxe/rxe_loc.h @@ -103,8 +103,8 @@ enum copy_direction { from_mem_obj, }; -int rxe_mem_init_dma(struct rxe_pd *pd, - int access, struct rxe_mem *mem); +void rxe_mem_init_dma(struct rxe_pd *pd, + int access, struct rxe_mem *mem); int rxe_mem_init_user(struct rxe_pd *pd, u64 start, u64 length, u64 iova, int access, struct ib_udata *udata, @@ -132,9 +132,6 @@ struct rxe_mem *lookup_mem(struct rxe_pd *pd, int access, u32 key, int mem_check_range(struct rxe_mem *mem, u64 iova, size_t length); -int rxe_mem_map_pages(struct rxe_dev *rxe, struct rxe_mem *mem, - u64 *page, int num_pages, u64 iova); - void rxe_mem_cleanup(struct rxe_pool_entry *arg); int advance_dma_data(struct rxe_dma_info *dma, unsigned int length); @@ -145,7 +142,6 @@ int rxe_send(struct rxe_pkt_info *pkt, struct sk_buff *skb); struct sk_buff *rxe_init_packet(struct rxe_dev *rxe, struct rxe_av *av, int paylen, struct rxe_pkt_info *pkt); int rxe_prepare(struct rxe_pkt_info *pkt, struct sk_buff *skb, u32 *crc); -enum rdma_link_layer rxe_link_layer(struct rxe_dev *rxe, unsigned int port_num); const char *rxe_parent_name(struct rxe_dev *rxe, unsigned int port_num); struct device *rxe_dma_device(struct rxe_dev *rxe); int rxe_mcast_add(struct rxe_dev *rxe, union ib_gid *mgid); diff --git a/drivers/infiniband/sw/rxe/rxe_mmap.c b/drivers/infiniband/sw/rxe/rxe_mmap.c index 6a413d73b95d..7887f623f62c 100644 --- a/drivers/infiniband/sw/rxe/rxe_mmap.c +++ b/drivers/infiniband/sw/rxe/rxe_mmap.c @@ -35,7 +35,6 @@ #include <linux/vmalloc.h> #include <linux/mm.h> #include <linux/errno.h> -#include <asm/pgtable.h> #include <rdma/uverbs_ioctl.h> #include "rxe.h" diff --git a/drivers/infiniband/sw/rxe/rxe_mr.c b/drivers/infiniband/sw/rxe/rxe_mr.c index e83c7b518bfa..ce24144de16a 100644 --- a/drivers/infiniband/sw/rxe/rxe_mr.c +++ b/drivers/infiniband/sw/rxe/rxe_mr.c @@ -144,8 +144,8 @@ err1: return -ENOMEM; } -int rxe_mem_init_dma(struct rxe_pd *pd, - int access, struct rxe_mem *mem) +void rxe_mem_init_dma(struct rxe_pd *pd, + int access, struct rxe_mem *mem) { rxe_mem_init(access, mem); @@ -153,8 +153,6 @@ int rxe_mem_init_dma(struct rxe_pd *pd, mem->access = access; mem->state = RXE_MEM_STATE_VALID; mem->type = RXE_MEM_TYPE_DMA; - - return 0; } int rxe_mem_init_user(struct rxe_pd *pd, u64 start, @@ -207,6 +205,7 @@ int rxe_mem_init_user(struct rxe_pd *pd, u64 start, vaddr = page_address(sg_page_iter_page(&sg_iter)); if (!vaddr) { pr_warn("null vaddr\n"); + ib_umem_release(umem); err = -ENOMEM; goto err1; } @@ -587,47 +586,3 @@ struct rxe_mem *lookup_mem(struct rxe_pd *pd, int access, u32 key, return mem; } - -int rxe_mem_map_pages(struct rxe_dev *rxe, struct rxe_mem *mem, - u64 *page, int num_pages, u64 iova) -{ - int i; - int num_buf; - int err; - struct rxe_map **map; - struct rxe_phys_buf *buf; - int page_size; - - if (num_pages > mem->max_buf) { - err = -EINVAL; - goto err1; - } - - num_buf = 0; - page_size = 1 << mem->page_shift; - map = mem->map; - buf = map[0]->buf; - - for (i = 0; i < num_pages; i++) { - buf->addr = *page++; - buf->size = page_size; - buf++; - num_buf++; - - if (num_buf == RXE_BUF_PER_MAP) { - map++; - buf = map[0]->buf; - num_buf = 0; - } - } - - mem->iova = iova; - mem->va = iova; - mem->length = num_pages << mem->page_shift; - mem->state = RXE_MEM_STATE_VALID; - - return 0; - -err1: - return err; -} diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c index 312c2fc961c0..0c3808611f95 100644 --- a/drivers/infiniband/sw/rxe/rxe_net.c +++ b/drivers/infiniband/sw/rxe/rxe_net.c @@ -520,11 +520,6 @@ const char *rxe_parent_name(struct rxe_dev *rxe, unsigned int port_num) return rxe->ndev->name; } -enum rdma_link_layer rxe_link_layer(struct rxe_dev *rxe, unsigned int port_num) -{ - return IB_LINK_LAYER_ETHERNET; -} - int rxe_net_add(const char *ibdev_name, struct net_device *ndev) { int err; diff --git a/drivers/infiniband/sw/rxe/rxe_param.h b/drivers/infiniband/sw/rxe/rxe_param.h index f59616b02477..2f381aeafcb5 100644 --- a/drivers/infiniband/sw/rxe/rxe_param.h +++ b/drivers/infiniband/sw/rxe/rxe_param.h @@ -100,7 +100,7 @@ enum rxe_device_param { RXE_MAX_SRQ_SGE = 27, RXE_MIN_SRQ_SGE = 1, RXE_MAX_FMR_PAGE_LIST_LEN = 512, - RXE_MAX_PKEYS = 64, + RXE_MAX_PKEYS = 1, RXE_LOCAL_CA_ACK_DELAY = 15, RXE_MAX_UCONTEXT = 512, @@ -127,6 +127,9 @@ enum rxe_device_param { /* Delay before calling arbiter timer */ RXE_NSEC_ARB_TIMER_DELAY = 200, + + /* IBTA v1.4 A3.3.1 VENDOR INFORMATION section */ + RXE_VENDOR_ID = 0XFFFFFF, }; /* default/initial rxe port parameters */ @@ -145,7 +148,7 @@ enum rxe_port_param { RXE_PORT_INIT_TYPE_REPLY = 0, RXE_PORT_ACTIVE_WIDTH = IB_WIDTH_1X, RXE_PORT_ACTIVE_SPEED = 1, - RXE_PORT_PKEY_TBL_LEN = 64, + RXE_PORT_PKEY_TBL_LEN = 1, RXE_PORT_PHYS_STATE = IB_PORT_PHYS_STATE_POLLING, RXE_PORT_SUBNET_PREFIX = 0xfe80000000000000ULL, }; diff --git a/drivers/infiniband/sw/rxe/rxe_recv.c b/drivers/infiniband/sw/rxe/rxe_recv.c index 831ad578a7b2..7e123d3c4d09 100644 --- a/drivers/infiniband/sw/rxe/rxe_recv.c +++ b/drivers/infiniband/sw/rxe/rxe_recv.c @@ -101,36 +101,15 @@ static void set_qkey_viol_cntr(struct rxe_port *port) static int check_keys(struct rxe_dev *rxe, struct rxe_pkt_info *pkt, u32 qpn, struct rxe_qp *qp) { - int i; - int found_pkey = 0; struct rxe_port *port = &rxe->port; u16 pkey = bth_pkey(pkt); pkt->pkey_index = 0; - if (qpn == 1) { - for (i = 0; i < port->attr.pkey_tbl_len; i++) { - if (pkey_match(pkey, port->pkey_tbl[i])) { - pkt->pkey_index = i; - found_pkey = 1; - break; - } - } - - if (!found_pkey) { - pr_warn_ratelimited("bad pkey = 0x%x\n", pkey); - set_bad_pkey_cntr(port); - goto err1; - } - } else { - if (unlikely(!pkey_match(pkey, - port->pkey_tbl[qp->attr.pkey_index] - ))) { - pr_warn_ratelimited("bad pkey = 0x%0x\n", pkey); - set_bad_pkey_cntr(port); - goto err1; - } - pkt->pkey_index = qp->attr.pkey_index; + if (!pkey_match(pkey, IB_DEFAULT_PKEY_FULL)) { + pr_warn_ratelimited("bad pkey = 0x%x\n", pkey); + set_bad_pkey_cntr(port); + goto err1; } if ((qp_type(qp) == IB_QPT_UD || qp_type(qp) == IB_QPT_GSI) && @@ -330,10 +309,14 @@ err1: static int rxe_match_dgid(struct rxe_dev *rxe, struct sk_buff *skb) { + struct rxe_pkt_info *pkt = SKB_TO_PKT(skb); const struct ib_gid_attr *gid_attr; union ib_gid dgid; union ib_gid *pdgid; + if (pkt->mask & RXE_LOOPBACK_MASK) + return 0; + if (skb->protocol == htons(ETH_P_IP)) { ipv6_addr_set_v4mapped(ip_hdr(skb)->daddr, (struct in6_addr *)&dgid); @@ -366,7 +349,7 @@ void rxe_rcv(struct sk_buff *skb) if (unlikely(skb->len < pkt->offset + RXE_BTH_BYTES)) goto drop; - if (unlikely(rxe_match_dgid(rxe, skb) < 0)) { + if (rxe_match_dgid(rxe, skb) < 0) { pr_warn_ratelimited("failed matching dgid\n"); goto drop; } diff --git a/drivers/infiniband/sw/rxe/rxe_req.c b/drivers/infiniband/sw/rxe/rxe_req.c index e5031172c019..34df2b55e650 100644 --- a/drivers/infiniband/sw/rxe/rxe_req.c +++ b/drivers/infiniband/sw/rxe/rxe_req.c @@ -381,7 +381,6 @@ static struct sk_buff *init_req_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt) { struct rxe_dev *rxe = to_rdev(qp->ibqp.device); - struct rxe_port *port = &rxe->port; struct sk_buff *skb; struct rxe_send_wr *ibwr = &wqe->wr; struct rxe_av *av; @@ -419,9 +418,7 @@ static struct sk_buff *init_req_packet(struct rxe_qp *qp, (pkt->mask & (RXE_WRITE_MASK | RXE_IMMDT_MASK)) == (RXE_WRITE_MASK | RXE_IMMDT_MASK)); - pkey = (qp_type(qp) == IB_QPT_GSI) ? - port->pkey_tbl[ibwr->wr.ud.pkey_index] : - port->pkey_tbl[qp->attr.pkey_index]; + pkey = IB_DEFAULT_PKEY_FULL; qp_num = (pkt->mask & RXE_DETH_MASK) ? ibwr->wr.ud.remote_qpn : qp->attr.dest_qp_num; diff --git a/drivers/infiniband/sw/rxe/rxe_sysfs.c b/drivers/infiniband/sw/rxe/rxe_sysfs.c index ccda5f5a3bc0..2af31d421bfc 100644 --- a/drivers/infiniband/sw/rxe/rxe_sysfs.c +++ b/drivers/infiniband/sw/rxe/rxe_sysfs.c @@ -61,6 +61,11 @@ static int rxe_param_set_add(const char *val, const struct kernel_param *kp) struct net_device *ndev; struct rxe_dev *exists; + if (!rxe_initialized) { + pr_err("Module parameters are not supported, use rdma link add or rxe_cfg\n"); + return -EAGAIN; + } + len = sanitize_arg(val, intf, sizeof(intf)); if (!len) { pr_err("add: invalid interface name\n"); diff --git a/drivers/infiniband/sw/rxe/rxe_task.c b/drivers/infiniband/sw/rxe/rxe_task.c index 08f05ac5f5d5..ecdac3f8fcc9 100644 --- a/drivers/infiniband/sw/rxe/rxe_task.c +++ b/drivers/infiniband/sw/rxe/rxe_task.c @@ -71,7 +71,7 @@ void rxe_do_task(unsigned long data) case TASK_STATE_BUSY: task->state = TASK_STATE_ARMED; - /* fall through */ + fallthrough; case TASK_STATE_ARMED: spin_unlock_irqrestore(&task->state_lock, flags); return; diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c index 9dd4bd7aea92..8522e9a3e914 100644 --- a/drivers/infiniband/sw/rxe/rxe_verbs.c +++ b/drivers/infiniband/sw/rxe/rxe_verbs.c @@ -83,22 +83,11 @@ static int rxe_query_port(struct ib_device *dev, static int rxe_query_pkey(struct ib_device *device, u8 port_num, u16 index, u16 *pkey) { - struct rxe_dev *rxe = to_rdev(device); - struct rxe_port *port; - - port = &rxe->port; - - if (unlikely(index >= port->attr.pkey_tbl_len)) { - dev_warn(device->dev.parent, "invalid index = %d\n", - index); - goto err1; - } + if (index > 0) + return -EINVAL; - *pkey = port->pkey_tbl[index]; + *pkey = IB_DEFAULT_PKEY_FULL; return 0; - -err1: - return -EINVAL; } static int rxe_modify_device(struct ib_device *dev, @@ -141,9 +130,7 @@ static int rxe_modify_port(struct ib_device *dev, static enum rdma_link_layer rxe_get_link_layer(struct ib_device *dev, u8 port_num) { - struct rxe_dev *rxe = to_rdev(dev); - - return rxe_link_layer(rxe, port_num); + return IB_LINK_LAYER_ETHERNET; } static int rxe_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata) @@ -195,15 +182,16 @@ static void rxe_dealloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) rxe_drop_ref(pd); } -static int rxe_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, - u32 flags, struct ib_udata *udata) +static int rxe_create_ah(struct ib_ah *ibah, + struct rdma_ah_init_attr *init_attr, + struct ib_udata *udata) { int err; struct rxe_dev *rxe = to_rdev(ibah->device); struct rxe_ah *ah = to_rah(ibah); - err = rxe_av_chk_attr(rxe, attr); + err = rxe_av_chk_attr(rxe, init_attr->ah_attr); if (err) return err; @@ -211,7 +199,7 @@ static int rxe_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, if (err) return err; - rxe_init_av(attr, &ah->av); + rxe_init_av(init_attr->ah_attr, &ah->av); return 0; } @@ -552,7 +540,7 @@ static void init_send_wr(struct rxe_qp *qp, struct rxe_send_wr *wr, switch (wr->opcode) { case IB_WR_RDMA_WRITE_WITH_IMM: wr->ex.imm_data = ibwr->ex.imm_data; - /* fall through */ + fallthrough; case IB_WR_RDMA_READ: case IB_WR_RDMA_WRITE: wr->wr.rdma.remote_addr = rdma_wr(ibwr)->remote_addr; @@ -683,6 +671,7 @@ static int rxe_post_send_kernel(struct rxe_qp *qp, const struct ib_send_wr *wr, unsigned int mask; unsigned int length = 0; int i; + struct ib_send_wr *next; while (wr) { mask = wr_opcode_mask(wr->opcode, qp); @@ -699,6 +688,8 @@ static int rxe_post_send_kernel(struct rxe_qp *qp, const struct ib_send_wr *wr, break; } + next = wr->next; + length = 0; for (i = 0; i < wr->num_sge; i++) length += wr->sg_list[i].length; @@ -709,7 +700,7 @@ static int rxe_post_send_kernel(struct rxe_qp *qp, const struct ib_send_wr *wr, *bad_wr = wr; break; } - wr = wr->next; + wr = next; } rxe_run_task(&qp->req.task, 1); @@ -900,30 +891,16 @@ static struct ib_mr *rxe_get_dma_mr(struct ib_pd *ibpd, int access) struct rxe_dev *rxe = to_rdev(ibpd->device); struct rxe_pd *pd = to_rpd(ibpd); struct rxe_mem *mr; - int err; mr = rxe_alloc(&rxe->mr_pool); - if (!mr) { - err = -ENOMEM; - goto err1; - } + if (!mr) + return ERR_PTR(-ENOMEM); rxe_add_index(mr); - rxe_add_ref(pd); - - err = rxe_mem_init_dma(pd, access, mr); - if (err) - goto err2; + rxe_mem_init_dma(pd, access, mr); return &mr->ibmr; - -err2: - rxe_drop_ref(pd); - rxe_drop_index(mr); - rxe_drop_ref(mr); -err1: - return ERR_PTR(err); } static struct ib_mr *rxe_reg_user_mr(struct ib_pd *ibpd, @@ -974,7 +951,7 @@ static int rxe_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata) } static struct ib_mr *rxe_alloc_mr(struct ib_pd *ibpd, enum ib_mr_type mr_type, - u32 max_num_sg, struct ib_udata *udata) + u32 max_num_sg) { struct rxe_dev *rxe = to_rdev(ibpd->device); struct rxe_pd *pd = to_rpd(ibpd); @@ -1079,7 +1056,7 @@ static ssize_t parent_show(struct device *device, struct rxe_dev *rxe = rdma_device_to_drv_device(device, struct rxe_dev, ib_dev); - return snprintf(buf, 16, "%s\n", rxe_parent_name(rxe, 1)); + return scnprintf(buf, PAGE_SIZE, "%s\n", rxe_parent_name(rxe, 1)); } static DEVICE_ATTR_RO(parent); diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h index 92de39c4a7c1..c664c7f36ab5 100644 --- a/drivers/infiniband/sw/rxe/rxe_verbs.h +++ b/drivers/infiniband/sw/rxe/rxe_verbs.h @@ -371,7 +371,6 @@ struct rxe_mc_elem { struct rxe_port { struct ib_port_attr attr; - u16 *pkey_tbl; __be64 port_guid; __be64 subnet_prefix; spinlock_t port_lock; /* guard port */ diff --git a/drivers/infiniband/sw/siw/siw.h b/drivers/infiniband/sw/siw/siw.h index af5e9f8c0fcd..e9753831ac3f 100644 --- a/drivers/infiniband/sw/siw/siw.h +++ b/drivers/infiniband/sw/siw/siw.h @@ -30,7 +30,6 @@ #define SIW_MAX_MR (SIW_MAX_QP * 10) #define SIW_MAX_PD SIW_MAX_QP #define SIW_MAX_MW 0 /* to be set if MW's are supported */ -#define SIW_MAX_FMR SIW_MAX_MR #define SIW_MAX_SRQ SIW_MAX_QP #define SIW_MAX_SRQ_WR (SIW_MAX_QP_WR * 10) #define SIW_MAX_CONTEXT SIW_MAX_PD @@ -59,7 +58,6 @@ struct siw_dev_cap { int max_mr; int max_pd; int max_mw; - int max_fmr; int max_srq; int max_srq_wr; int max_srq_sge; @@ -139,7 +137,7 @@ struct siw_pble { struct siw_pbl { unsigned int num_buf; unsigned int max_buf; - struct siw_pble pbe[1]; + struct siw_pble pbe[]; }; /* diff --git a/drivers/infiniband/sw/siw/siw_cm.c b/drivers/infiniband/sw/siw/siw_cm.c index 559e5fd3bad8..66764f7ef072 100644 --- a/drivers/infiniband/sw/siw/siw_cm.c +++ b/drivers/infiniband/sw/siw/siw_cm.c @@ -947,16 +947,8 @@ static void siw_accept_newconn(struct siw_cep *cep) siw_cep_get(new_cep); new_s->sk->sk_user_data = new_cep; - if (siw_tcp_nagle == false) { - int val = 1; - - rv = kernel_setsockopt(new_s, SOL_TCP, TCP_NODELAY, - (char *)&val, sizeof(val)); - if (rv) { - siw_dbg_cep(cep, "setsockopt NODELAY error: %d\n", rv); - goto error; - } - } + if (siw_tcp_nagle == false) + tcp_sock_set_nodelay(new_s->sk); new_cep->state = SIW_EPSTATE_AWAIT_MPAREQ; rv = siw_cm_queue_work(new_cep, SIW_CM_WORK_MPATIMEOUT); @@ -1232,12 +1224,10 @@ static void siw_cm_llp_data_ready(struct sock *sk) switch (cep->state) { case SIW_EPSTATE_RDMA_MODE: - /* fall through */ case SIW_EPSTATE_LISTENING: break; case SIW_EPSTATE_AWAIT_MPAREQ: - /* fall through */ case SIW_EPSTATE_AWAIT_MPAREP: siw_cm_queue_work(cep, SIW_CM_WORK_READ_MPAHDR); break; @@ -1312,17 +1302,14 @@ static void siw_cm_llp_state_change(struct sock *sk) static int kernel_bindconnect(struct socket *s, struct sockaddr *laddr, struct sockaddr *raddr) { - int rv, flags = 0, s_val = 1; + int rv, flags = 0; size_t size = laddr->sa_family == AF_INET ? sizeof(struct sockaddr_in) : sizeof(struct sockaddr_in6); /* * Make address available again asap. */ - rv = kernel_setsockopt(s, SOL_SOCKET, SO_REUSEADDR, (char *)&s_val, - sizeof(s_val)); - if (rv < 0) - return rv; + sock_set_reuseaddr(s->sk); rv = s->ops->bind(s, laddr, size); if (rv < 0) @@ -1389,16 +1376,8 @@ int siw_connect(struct iw_cm_id *id, struct iw_cm_conn_param *params) siw_dbg_qp(qp, "kernel_bindconnect: error %d\n", rv); goto error; } - if (siw_tcp_nagle == false) { - int val = 1; - - rv = kernel_setsockopt(s, SOL_TCP, TCP_NODELAY, (char *)&val, - sizeof(val)); - if (rv) { - siw_dbg_qp(qp, "setsockopt NODELAY error: %d\n", rv); - goto error; - } - } + if (siw_tcp_nagle == false) + tcp_sock_set_nodelay(s->sk); cep = siw_cep_alloc(sdev); if (!cep) { rv = -ENOMEM; @@ -1781,7 +1760,7 @@ int siw_create_listen(struct iw_cm_id *id, int backlog) struct siw_cep *cep = NULL; struct siw_device *sdev = to_siw_dev(id->device); int addr_family = id->local_addr.ss_family; - int rv = 0, s_val; + int rv = 0; if (addr_family != AF_INET && addr_family != AF_INET6) return -EAFNOSUPPORT; @@ -1793,13 +1772,8 @@ int siw_create_listen(struct iw_cm_id *id, int backlog) /* * Allow binding local port when still in TIME_WAIT from last close. */ - s_val = 1; - rv = kernel_setsockopt(s, SOL_SOCKET, SO_REUSEADDR, (char *)&s_val, - sizeof(s_val)); - if (rv) { - siw_dbg(id->device, "setsockopt error: %d\n", rv); - goto error; - } + sock_set_reuseaddr(s->sk); + if (addr_family == AF_INET) { struct sockaddr_in *laddr = &to_sockaddr_in(id->local_addr); diff --git a/drivers/infiniband/sw/siw/siw_main.c b/drivers/infiniband/sw/siw/siw_main.c index 5cd40fb9e20c..d862bec84376 100644 --- a/drivers/infiniband/sw/siw/siw_main.c +++ b/drivers/infiniband/sw/siw/siw_main.c @@ -67,12 +67,13 @@ static int siw_device_register(struct siw_device *sdev, const char *name) static int dev_id = 1; int rv; + sdev->vendor_part_id = dev_id++; + rv = ib_register_device(base_dev, name); if (rv) { pr_warn("siw: device registration error %d\n", rv); return rv; } - sdev->vendor_part_id = dev_id++; siw_dbg(base_dev, "HWaddr=%pM\n", sdev->netdev->dev_addr); @@ -288,7 +289,6 @@ static const struct ib_device_ops siw_device_ops = { .post_srq_recv = siw_post_srq_recv, .query_device = siw_query_device, .query_gid = siw_query_gid, - .query_pkey = siw_query_pkey, .query_port = siw_query_port, .query_qp = siw_query_qp, .query_srq = siw_query_srq, @@ -413,7 +413,6 @@ static struct siw_device *siw_device_create(struct net_device *netdev) sdev->attrs.max_mr = SIW_MAX_MR; sdev->attrs.max_pd = SIW_MAX_PD; sdev->attrs.max_mw = SIW_MAX_MW; - sdev->attrs.max_fmr = SIW_MAX_FMR; sdev->attrs.max_srq = SIW_MAX_SRQ; sdev->attrs.max_srq_wr = SIW_MAX_SRQ_WR; sdev->attrs.max_srq_sge = SIW_MAX_SGE; diff --git a/drivers/infiniband/sw/siw/siw_mem.c b/drivers/infiniband/sw/siw/siw_mem.c index e2061dc0b043..34a910cf0edb 100644 --- a/drivers/infiniband/sw/siw/siw_mem.c +++ b/drivers/infiniband/sw/siw/siw_mem.c @@ -349,14 +349,11 @@ dma_addr_t siw_pbl_get_buffer(struct siw_pbl *pbl, u64 off, int *len, int *idx) struct siw_pbl *siw_pbl_alloc(u32 num_buf) { struct siw_pbl *pbl; - int buf_size = sizeof(*pbl); if (num_buf == 0) return ERR_PTR(-EINVAL); - buf_size += ((num_buf - 1) * sizeof(struct siw_pble)); - - pbl = kzalloc(buf_size, GFP_KERNEL); + pbl = kzalloc(struct_size(pbl, pbe, num_buf), GFP_KERNEL); if (!pbl) return ERR_PTR(-ENOMEM); @@ -397,7 +394,7 @@ struct siw_umem *siw_umem_get(u64 start, u64 len, bool writable) if (!writable) foll_flags |= FOLL_FORCE; - down_read(&mm_s->mmap_sem); + mmap_read_lock(mm_s); mlock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT; @@ -441,7 +438,7 @@ struct siw_umem *siw_umem_get(u64 start, u64 len, bool writable) num_pages -= got; } out_sem_up: - up_read(&mm_s->mmap_sem); + mmap_read_unlock(mm_s); if (rv > 0) return umem; diff --git a/drivers/infiniband/sw/siw/siw_qp_rx.c b/drivers/infiniband/sw/siw/siw_qp_rx.c index 650520244ed0..4bd1f1f84057 100644 --- a/drivers/infiniband/sw/siw/siw_qp_rx.c +++ b/drivers/infiniband/sw/siw/siw_qp_rx.c @@ -139,7 +139,8 @@ static int siw_rx_pbl(struct siw_rx_stream *srx, int *pbl_idx, break; bytes = min(bytes, len); - if (siw_rx_kva(srx, (void *)buf_addr, bytes) == bytes) { + if (siw_rx_kva(srx, (void *)(uintptr_t)buf_addr, bytes) == + bytes) { copied += bytes; offset += bytes; len -= bytes; @@ -332,7 +333,7 @@ static struct siw_wqe *siw_rqe_get(struct siw_qp *qp) struct siw_srq *srq; struct siw_wqe *wqe = NULL; bool srq_event = false; - unsigned long uninitialized_var(flags); + unsigned long flags; srq = qp->srq; if (srq) { @@ -1214,7 +1215,7 @@ static int siw_rdmap_complete(struct siw_qp *qp, int error) case RDMAP_SEND_SE: case RDMAP_SEND_SE_INVAL: wqe->rqe.flags |= SIW_WQE_SOLICITED; - /* Fall through */ + fallthrough; case RDMAP_SEND: case RDMAP_SEND_INVAL: @@ -1385,7 +1386,7 @@ int siw_tcp_rx_data(read_descriptor_t *rd_desc, struct sk_buff *skb, * DDP segment. */ qp->rx_fpdu->first_ddp_seg = 0; - /* Fall through */ + fallthrough; case SIW_GET_DATA_START: /* diff --git a/drivers/infiniband/sw/siw/siw_qp_tx.c b/drivers/infiniband/sw/siw/siw_qp_tx.c index 9f53aa4feb87..d19d8325588b 100644 --- a/drivers/infiniband/sw/siw/siw_qp_tx.c +++ b/drivers/infiniband/sw/siw/siw_qp_tx.c @@ -1042,7 +1042,7 @@ next_wqe: case SIW_OP_SEND_REMOTE_INV: case SIW_OP_WRITE: siw_wqe_put_mem(wqe, tx_type); - /* Fall through */ + fallthrough; case SIW_OP_INVAL_STAG: case SIW_OP_REG_MR: @@ -1128,7 +1128,7 @@ next_wqe: case SIW_OP_READ: case SIW_OP_READ_LOCAL_INV: siw_wqe_put_mem(wqe, tx_type); - /* Fall through */ + fallthrough; case SIW_OP_INVAL_STAG: case SIW_OP_REG_MR: diff --git a/drivers/infiniband/sw/siw/siw_verbs.c b/drivers/infiniband/sw/siw/siw_verbs.c index aeb842bc7a1e..adafa1b8bebe 100644 --- a/drivers/infiniband/sw/siw/siw_verbs.c +++ b/drivers/infiniband/sw/siw/siw_verbs.c @@ -136,7 +136,6 @@ int siw_query_device(struct ib_device *base_dev, struct ib_device_attr *attr, attr->max_cq = sdev->attrs.max_cq; attr->max_cqe = sdev->attrs.max_cqe; attr->max_fast_reg_page_list_len = SIW_MAX_SGE_PBL; - attr->max_fmr = sdev->attrs.max_fmr; attr->max_mr = sdev->attrs.max_mr; attr->max_mw = sdev->attrs.max_mw; attr->max_mr_size = ~0ull; @@ -177,7 +176,6 @@ int siw_query_port(struct ib_device *base_dev, u8 port, attr->active_mtu = ib_mtu_int_to_enum(sdev->netdev->mtu); attr->phys_state = sdev->state == IB_PORT_ACTIVE ? IB_PORT_PHYS_STATE_LINK_UP : IB_PORT_PHYS_STATE_DISABLED; - attr->pkey_tbl_len = 1; attr->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_DEVICE_MGMT_SUP; attr->state = sdev->state; /* @@ -205,20 +203,12 @@ int siw_get_port_immutable(struct ib_device *base_dev, u8 port, if (rv) return rv; - port_immutable->pkey_tbl_len = attr.pkey_tbl_len; port_immutable->gid_tbl_len = attr.gid_tbl_len; port_immutable->core_cap_flags = RDMA_CORE_PORT_IWARP; return 0; } -int siw_query_pkey(struct ib_device *base_dev, u8 port, u16 idx, u16 *pkey) -{ - /* Report the default pkey */ - *pkey = 0xffff; - return 0; -} - int siw_query_gid(struct ib_device *base_dev, u8 port, int idx, union ib_gid *gid) { @@ -1374,7 +1364,7 @@ err_out: } struct ib_mr *siw_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, - u32 max_sge, struct ib_udata *udata) + u32 max_sge) { struct siw_device *sdev = to_siw_dev(pd->device); struct siw_mr *mr = NULL; diff --git a/drivers/infiniband/sw/siw/siw_verbs.h b/drivers/infiniband/sw/siw/siw_verbs.h index 1a731989fad6..d9572275a6b6 100644 --- a/drivers/infiniband/sw/siw/siw_verbs.h +++ b/drivers/infiniband/sw/siw/siw_verbs.h @@ -46,7 +46,6 @@ int siw_create_cq(struct ib_cq *base_cq, const struct ib_cq_init_attr *attr, struct ib_udata *udata); int siw_query_port(struct ib_device *base_dev, u8 port, struct ib_port_attr *attr); -int siw_query_pkey(struct ib_device *base_dev, u8 port, u16 idx, u16 *pkey); int siw_query_gid(struct ib_device *base_dev, u8 port, int idx, union ib_gid *gid); int siw_alloc_pd(struct ib_pd *base_pd, struct ib_udata *udata); @@ -69,7 +68,7 @@ int siw_req_notify_cq(struct ib_cq *base_cq, enum ib_cq_notify_flags flags); struct ib_mr *siw_reg_user_mr(struct ib_pd *base_pd, u64 start, u64 len, u64 rnic_va, int rights, struct ib_udata *udata); struct ib_mr *siw_alloc_mr(struct ib_pd *base_pd, enum ib_mr_type mr_type, - u32 max_sge, struct ib_udata *udata); + u32 max_sge); struct ib_mr *siw_get_dma_mr(struct ib_pd *base_pd, int rights); int siw_map_mr_sg(struct ib_mr *base_mr, struct scatterlist *sl, int num_sle, unsigned int *sg_off); diff --git a/drivers/infiniband/ulp/Makefile b/drivers/infiniband/ulp/Makefile index 437813c7b481..4d0004b58377 100644 --- a/drivers/infiniband/ulp/Makefile +++ b/drivers/infiniband/ulp/Makefile @@ -5,3 +5,4 @@ obj-$(CONFIG_INFINIBAND_SRPT) += srpt/ obj-$(CONFIG_INFINIBAND_ISER) += iser/ obj-$(CONFIG_INFINIBAND_ISERT) += isert/ obj-$(CONFIG_INFINIBAND_OPA_VNIC) += opa_vnic/ +obj-$(CONFIG_INFINIBAND_RTRS) += rtrs/ diff --git a/drivers/infiniband/ulp/ipoib/Kconfig b/drivers/infiniband/ulp/ipoib/Kconfig index 7af68604af77..254e31a90a66 100644 --- a/drivers/infiniband/ulp/ipoib/Kconfig +++ b/drivers/infiniband/ulp/ipoib/Kconfig @@ -2,7 +2,7 @@ config INFINIBAND_IPOIB tristate "IP-over-InfiniBand" depends on NETDEVICES && INET && (IPV6 || IPV6=n) - ---help--- + help Support for the IP-over-InfiniBand protocol (IPoIB). This transports IP packets over InfiniBand so you can use your IB device as a fancy NIC. @@ -13,7 +13,7 @@ config INFINIBAND_IPOIB_CM bool "IP-over-InfiniBand Connected Mode support" depends on INFINIBAND_IPOIB default n - ---help--- + help This option enables support for IPoIB connected mode. After enabling this option, you need to switch to connected mode through /sys/class/net/ibXXX/mode to actually create @@ -28,7 +28,7 @@ config INFINIBAND_IPOIB_DEBUG bool "IP-over-InfiniBand debugging" if EXPERT depends on INFINIBAND_IPOIB default y - ---help--- + help This option causes debugging code to be compiled into the IPoIB driver. The output can be turned on via the debug_level and mcast_debug_level module parameters (which @@ -42,7 +42,7 @@ config INFINIBAND_IPOIB_DEBUG config INFINIBAND_IPOIB_DEBUG_DATA bool "IP-over-InfiniBand data path debugging" depends on INFINIBAND_IPOIB_DEBUG - ---help--- + help This option compiles debugging code into the data path of the IPoIB driver. The output can be turned on via the data_debug_level module parameter; however, even with output diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index 9a3379c49541..3440dc48d02c 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -515,7 +515,7 @@ void ipoib_ib_dev_cleanup(struct net_device *dev); int ipoib_ib_dev_open_default(struct net_device *dev); int ipoib_ib_dev_open(struct net_device *dev); -int ipoib_ib_dev_stop(struct net_device *dev); +void ipoib_ib_dev_stop(struct net_device *dev); void ipoib_ib_dev_up(struct net_device *dev); void ipoib_ib_dev_down(struct net_device *dev); int ipoib_ib_dev_stop_default(struct net_device *dev); @@ -527,7 +527,7 @@ void ipoib_mcast_send(struct net_device *dev, u8 *daddr, struct sk_buff *skb); void ipoib_mcast_restart_task(struct work_struct *work); void ipoib_mcast_start_thread(struct net_device *dev); -int ipoib_mcast_stop_thread(struct net_device *dev); +void ipoib_mcast_stop_thread(struct net_device *dev); void ipoib_mcast_dev_down(struct net_device *dev); void ipoib_mcast_dev_flush(struct net_device *dev); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c index 9bf0fa30df28..7c41fb040f7c 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c @@ -512,13 +512,13 @@ static int ipoib_cm_rx_handler(struct ib_cm_id *cm_id, return ipoib_cm_req_handler(cm_id, event); case IB_CM_DREQ_RECEIVED: ib_send_cm_drep(cm_id, NULL, 0); - /* Fall through */ + fallthrough; case IB_CM_REJ_RECEIVED: p = cm_id->context; priv = ipoib_priv(p->dev); if (ib_modify_qp(p->qp, &ipoib_cm_err_attr, IB_QP_STATE)) ipoib_warn(priv, "unable to move qp to error state\n"); - /* Fall through */ + fallthrough; default: return 0; } diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c index da3c5315bbb5..494f413dc3c6 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c @@ -670,13 +670,12 @@ int ipoib_send(struct net_device *dev, struct sk_buff *skb, return rc; } -static void __ipoib_reap_ah(struct net_device *dev) +static void ipoib_reap_dead_ahs(struct ipoib_dev_priv *priv) { - struct ipoib_dev_priv *priv = ipoib_priv(dev); struct ipoib_ah *ah, *tah; unsigned long flags; - netif_tx_lock_bh(dev); + netif_tx_lock_bh(priv->dev); spin_lock_irqsave(&priv->lock, flags); list_for_each_entry_safe(ah, tah, &priv->dead_ahs, list) @@ -687,37 +686,37 @@ static void __ipoib_reap_ah(struct net_device *dev) } spin_unlock_irqrestore(&priv->lock, flags); - netif_tx_unlock_bh(dev); + netif_tx_unlock_bh(priv->dev); } void ipoib_reap_ah(struct work_struct *work) { struct ipoib_dev_priv *priv = container_of(work, struct ipoib_dev_priv, ah_reap_task.work); - struct net_device *dev = priv->dev; - __ipoib_reap_ah(dev); + ipoib_reap_dead_ahs(priv); if (!test_bit(IPOIB_STOP_REAPER, &priv->flags)) queue_delayed_work(priv->wq, &priv->ah_reap_task, round_jiffies_relative(HZ)); } -static void ipoib_flush_ah(struct net_device *dev) +static void ipoib_start_ah_reaper(struct ipoib_dev_priv *priv) { - struct ipoib_dev_priv *priv = ipoib_priv(dev); - - cancel_delayed_work(&priv->ah_reap_task); - flush_workqueue(priv->wq); - ipoib_reap_ah(&priv->ah_reap_task.work); + clear_bit(IPOIB_STOP_REAPER, &priv->flags); + queue_delayed_work(priv->wq, &priv->ah_reap_task, + round_jiffies_relative(HZ)); } -static void ipoib_stop_ah(struct net_device *dev) +static void ipoib_stop_ah_reaper(struct ipoib_dev_priv *priv) { - struct ipoib_dev_priv *priv = ipoib_priv(dev); - set_bit(IPOIB_STOP_REAPER, &priv->flags); - ipoib_flush_ah(dev); + cancel_delayed_work(&priv->ah_reap_task); + /* + * After ipoib_stop_ah_reaper() we always go through + * ipoib_reap_dead_ahs() which ensures the work is really stopped and + * does a final flush out of the dead_ah's list + */ } static int recvs_pending(struct net_device *dev) @@ -846,18 +845,6 @@ timeout: return 0; } -int ipoib_ib_dev_stop(struct net_device *dev) -{ - struct ipoib_dev_priv *priv = ipoib_priv(dev); - - priv->rn_ops->ndo_stop(dev); - - clear_bit(IPOIB_FLAG_INITIALIZED, &priv->flags); - ipoib_flush_ah(dev); - - return 0; -} - int ipoib_ib_dev_open_default(struct net_device *dev) { struct ipoib_dev_priv *priv = ipoib_priv(dev); @@ -901,10 +888,7 @@ int ipoib_ib_dev_open(struct net_device *dev) return -1; } - clear_bit(IPOIB_STOP_REAPER, &priv->flags); - queue_delayed_work(priv->wq, &priv->ah_reap_task, - round_jiffies_relative(HZ)); - + ipoib_start_ah_reaper(priv); if (priv->rn_ops->ndo_open(dev)) { pr_warn("%s: Failed to open dev\n", dev->name); goto dev_stop; @@ -915,13 +899,20 @@ int ipoib_ib_dev_open(struct net_device *dev) return 0; dev_stop: - set_bit(IPOIB_STOP_REAPER, &priv->flags); - cancel_delayed_work(&priv->ah_reap_task); - set_bit(IPOIB_FLAG_INITIALIZED, &priv->flags); - ipoib_ib_dev_stop(dev); + ipoib_stop_ah_reaper(priv); return -1; } +void ipoib_ib_dev_stop(struct net_device *dev) +{ + struct ipoib_dev_priv *priv = ipoib_priv(dev); + + priv->rn_ops->ndo_stop(dev); + + clear_bit(IPOIB_FLAG_INITIALIZED, &priv->flags); + ipoib_stop_ah_reaper(priv); +} + void ipoib_pkey_dev_check_presence(struct net_device *dev) { struct ipoib_dev_priv *priv = ipoib_priv(dev); @@ -1232,7 +1223,7 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv, ipoib_mcast_dev_flush(dev); if (oper_up) set_bit(IPOIB_FLAG_OPER_UP, &priv->flags); - ipoib_flush_ah(dev); + ipoib_reap_dead_ahs(priv); } if (level >= IPOIB_FLUSH_NORMAL) @@ -1307,7 +1298,7 @@ void ipoib_ib_dev_cleanup(struct net_device *dev) * the neighbor garbage collection is stopped and reaped. * That should all be done now, so make a final ah flush. */ - ipoib_stop_ah(dev); + ipoib_reap_dead_ahs(priv); clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c index ceec24d45185..f772fe8c5b66 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c @@ -86,7 +86,7 @@ struct workqueue_struct *ipoib_workqueue; struct ib_sa_client ipoib_sa_client; -static void ipoib_add_one(struct ib_device *device); +static int ipoib_add_one(struct ib_device *device); static void ipoib_remove_one(struct ib_device *device, void *client_data); static void ipoib_neigh_reclaim(struct rcu_head *rp); static struct net_device *ipoib_get_net_dev_by_params( @@ -342,9 +342,10 @@ struct ipoib_walk_data { struct net_device *result; }; -static int ipoib_upper_walk(struct net_device *upper, void *_data) +static int ipoib_upper_walk(struct net_device *upper, + struct netdev_nested_priv *priv) { - struct ipoib_walk_data *data = _data; + struct ipoib_walk_data *data = (struct ipoib_walk_data *)priv->data; int ret = 0; if (ipoib_is_dev_match_addr_rcu(data->addr, upper)) { @@ -368,10 +369,12 @@ static int ipoib_upper_walk(struct net_device *upper, void *_data) static struct net_device *ipoib_get_net_dev_match_addr( const struct sockaddr *addr, struct net_device *dev) { + struct netdev_nested_priv priv; struct ipoib_walk_data data = { .addr = addr, }; + priv.data = (void *)&data; rcu_read_lock(); if (ipoib_is_dev_match_addr_rcu(addr, dev)) { dev_hold(dev); @@ -379,7 +382,7 @@ static struct net_device *ipoib_get_net_dev_match_addr( goto out; } - netdev_walk_all_upper_dev_rcu(dev, ipoib_upper_walk, &data); + netdev_walk_all_upper_dev_rcu(dev, ipoib_upper_walk, &priv); out: rcu_read_unlock(); return data.result; @@ -479,9 +482,6 @@ static struct net_device *ipoib_get_net_dev_by_params( if (ret) return NULL; - if (!dev_list) - return NULL; - /* See if we can find a unique device matching the L2 parameters */ matches = __ipoib_get_net_dev_by_params(dev_list, port, pkey_index, gid, NULL, &net_dev); @@ -505,7 +505,7 @@ static struct net_device *ipoib_get_net_dev_by_params( default: dev_warn_ratelimited(&dev->dev, "duplicate IP address detected\n"); - /* Fall through */ + fallthrough; case 1: return net_dev; } @@ -529,6 +529,7 @@ int ipoib_set_mode(struct net_device *dev, const char *buf) "will cause multicast packet drops\n"); netdev_update_features(dev); dev_set_mtu(dev, ipoib_cm_max_mtu(dev)); + netif_set_real_num_tx_queues(dev, 1); rtnl_unlock(); priv->tx_wr.wr.send_flags &= ~IB_SEND_IP_CSUM; @@ -540,6 +541,7 @@ int ipoib_set_mode(struct net_device *dev, const char *buf) clear_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags); netdev_update_features(dev); dev_set_mtu(dev, min(priv->mcast_mtu, dev->mtu)); + netif_set_real_num_tx_queues(dev, dev->num_tx_queues); rtnl_unlock(); ipoib_flush_paths(dev); return (!rtnl_trylock()) ? -EBUSY : 0; @@ -1860,7 +1862,7 @@ static int ipoib_parent_init(struct net_device *ndev) priv->port); return result; } - priv->max_ib_mtu = ib_mtu_enum_to_int(attr.max_mtu); + priv->max_ib_mtu = rdma_mtu_from_attr(priv->ca, priv->port, &attr); result = ib_query_pkey(priv->ca, priv->port, 0, &priv->pkey); if (result) { @@ -1893,14 +1895,22 @@ static void ipoib_child_init(struct net_device *ndev) priv->max_ib_mtu = ppriv->max_ib_mtu; set_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags); - memcpy(priv->dev->dev_addr, ppriv->dev->dev_addr, INFINIBAND_ALEN); - memcpy(&priv->local_gid, &ppriv->local_gid, sizeof(priv->local_gid)); + if (memchr_inv(priv->dev->dev_addr, 0, INFINIBAND_ALEN)) + memcpy(&priv->local_gid, priv->dev->dev_addr + 4, + sizeof(priv->local_gid)); + else { + memcpy(priv->dev->dev_addr, ppriv->dev->dev_addr, + INFINIBAND_ALEN); + memcpy(&priv->local_gid, &ppriv->local_gid, + sizeof(priv->local_gid)); + } } static int ipoib_ndo_init(struct net_device *ndev) { struct ipoib_dev_priv *priv = ipoib_priv(ndev); int rc; + struct rdma_netdev *rn = netdev_priv(ndev); if (priv->parent) { ipoib_child_init(ndev); @@ -1913,6 +1923,7 @@ static int ipoib_ndo_init(struct net_device *ndev) /* MTU will be reset when mcast join happens */ ndev->mtu = IPOIB_UD_MTU(priv->max_ib_mtu); priv->mcast_mtu = priv->admin_mtu = ndev->mtu; + rn->mtu = priv->mcast_mtu; ndev->max_mtu = IPOIB_CM_MTU; ndev->neigh_priv_len = sizeof(struct ipoib_neigh); @@ -1975,6 +1986,8 @@ static void ipoib_ndo_uninit(struct net_device *dev) /* no more works over the priv->wq */ if (priv->wq) { + /* See ipoib_mcast_carrier_on_task() */ + WARN_ON(test_bit(IPOIB_FLAG_OPER_UP, &priv->flags)); flush_workqueue(priv->wq); destroy_workqueue(priv->wq); priv->wq = NULL; @@ -2074,9 +2087,17 @@ static const struct net_device_ops ipoib_netdev_ops_vf = { .ndo_do_ioctl = ipoib_ioctl, }; +static const struct net_device_ops ipoib_netdev_default_pf = { + .ndo_init = ipoib_dev_init_default, + .ndo_uninit = ipoib_dev_uninit_default, + .ndo_open = ipoib_ib_dev_open_default, + .ndo_stop = ipoib_ib_dev_stop_default, +}; + void ipoib_setup_common(struct net_device *dev) { dev->header_ops = &ipoib_header_ops; + dev->netdev_ops = &ipoib_netdev_default_pf; ipoib_set_ethtool_ops(dev); @@ -2126,13 +2147,6 @@ static void ipoib_build_priv(struct net_device *dev) INIT_DELAYED_WORK(&priv->neigh_reap_task, ipoib_reap_neigh); } -static const struct net_device_ops ipoib_netdev_default_pf = { - .ndo_init = ipoib_dev_init_default, - .ndo_uninit = ipoib_dev_uninit_default, - .ndo_open = ipoib_ib_dev_open_default, - .ndo_stop = ipoib_ib_dev_stop_default, -}; - static struct net_device *ipoib_alloc_netdev(struct ib_device *hca, u8 port, const char *name) { @@ -2170,7 +2184,6 @@ int ipoib_intf_init(struct ib_device *hca, u8 port, const char *name, if (rc != -EOPNOTSUPP) goto out; - dev->netdev_ops = &ipoib_netdev_default_pf; rn->send = ipoib_send; rn->attach_mcast = ipoib_mcast_attach; rn->detach_mcast = ipoib_mcast_detach; @@ -2516,7 +2529,7 @@ sysfs_failed: return ERR_PTR(-ENOMEM); } -static void ipoib_add_one(struct ib_device *device) +static int ipoib_add_one(struct ib_device *device) { struct list_head *dev_list; struct net_device *dev; @@ -2526,7 +2539,7 @@ static void ipoib_add_one(struct ib_device *device) dev_list = kmalloc(sizeof(*dev_list), GFP_KERNEL); if (!dev_list) - return; + return -ENOMEM; INIT_LIST_HEAD(dev_list); @@ -2543,10 +2556,11 @@ static void ipoib_add_one(struct ib_device *device) if (!count) { kfree(dev_list); - return; + return -EOPNOTSUPP; } ib_set_client_data(device, &ipoib_client, dev_list); + return 0; } static void ipoib_remove_one(struct ib_device *device, void *client_data) @@ -2554,9 +2568,6 @@ static void ipoib_remove_one(struct ib_device *device, void *client_data) struct ipoib_dev_priv *priv, *tmp, *cpriv, *tcpriv; struct list_head *dev_list = client_data; - if (!dev_list) - return; - list_for_each_entry_safe(priv, tmp, dev_list, list) { LIST_HEAD(head); ipoib_parent_unregister_pre(priv->dev); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index b9e9562f5034..86e4ed64e4e2 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -135,12 +135,11 @@ static void ipoib_mcast_free(struct ipoib_mcast *mcast) kfree(mcast); } -static struct ipoib_mcast *ipoib_mcast_alloc(struct net_device *dev, - int can_sleep) +static struct ipoib_mcast *ipoib_mcast_alloc(struct net_device *dev) { struct ipoib_mcast *mcast; - mcast = kzalloc(sizeof(*mcast), can_sleep ? GFP_KERNEL : GFP_ATOMIC); + mcast = kzalloc(sizeof(*mcast), GFP_ATOMIC); if (!mcast) return NULL; @@ -218,6 +217,7 @@ static int ipoib_mcast_join_finish(struct ipoib_mcast *mcast, struct rdma_ah_attr av; int ret; int set_qkey = 0; + int mtu; mcast->mcmember = *mcmember; @@ -240,13 +240,12 @@ static int ipoib_mcast_join_finish(struct ipoib_mcast *mcast, priv->broadcast->mcmember.flow_label = mcmember->flow_label; priv->broadcast->mcmember.hop_limit = mcmember->hop_limit; /* assume if the admin and the mcast are the same both can be changed */ + mtu = rdma_mtu_enum_to_int(priv->ca, priv->port, + priv->broadcast->mcmember.mtu); if (priv->mcast_mtu == priv->admin_mtu) - priv->admin_mtu = - priv->mcast_mtu = - IPOIB_UD_MTU(ib_mtu_enum_to_int(priv->broadcast->mcmember.mtu)); - else - priv->mcast_mtu = - IPOIB_UD_MTU(ib_mtu_enum_to_int(priv->broadcast->mcmember.mtu)); + priv->admin_mtu = IPOIB_UD_MTU(mtu); + priv->mcast_mtu = IPOIB_UD_MTU(mtu); + rn->mtu = priv->mcast_mtu; priv->qkey = be32_to_cpu(priv->broadcast->mcmember.qkey); spin_unlock_irq(&priv->lock); @@ -599,7 +598,7 @@ void ipoib_mcast_join_task(struct work_struct *work) if (!priv->broadcast) { struct ipoib_mcast *broadcast; - broadcast = ipoib_mcast_alloc(dev, 0); + broadcast = ipoib_mcast_alloc(dev); if (!broadcast) { ipoib_warn(priv, "failed to allocate broadcast group\n"); /* @@ -681,15 +680,13 @@ void ipoib_mcast_start_thread(struct net_device *dev) spin_unlock_irqrestore(&priv->lock, flags); } -int ipoib_mcast_stop_thread(struct net_device *dev) +void ipoib_mcast_stop_thread(struct net_device *dev) { struct ipoib_dev_priv *priv = ipoib_priv(dev); ipoib_dbg_mcast(priv, "stopping multicast thread\n"); cancel_delayed_work_sync(&priv->mcast_task); - - return 0; } static int ipoib_mcast_leave(struct net_device *dev, struct ipoib_mcast *mcast) @@ -782,7 +779,7 @@ void ipoib_mcast_send(struct net_device *dev, u8 *daddr, struct sk_buff *skb) ipoib_dbg_mcast(priv, "setting up send only multicast group for %pI6\n", mgid); - mcast = ipoib_mcast_alloc(dev, 0); + mcast = ipoib_mcast_alloc(dev); if (!mcast) { ipoib_warn(priv, "unable to allocate memory " "for multicast structure\n"); @@ -936,7 +933,7 @@ void ipoib_mcast_restart_task(struct work_struct *work) ipoib_dbg_mcast(priv, "adding multicast entry for mgid %pI6\n", mgid.raw); - nmcast = ipoib_mcast_alloc(dev, 0); + nmcast = ipoib_mcast_alloc(dev); if (!nmcast) { ipoib_warn(priv, "unable to allocate memory for multicast structure\n"); continue; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c index b69304d28f06..587252fd6f57 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c @@ -206,6 +206,9 @@ int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca) if (priv->hca_caps & IB_DEVICE_MANAGED_FLOW_STEERING) init_attr.create_flags |= IB_QP_CREATE_NETIF_QP; + if (priv->hca_caps & IB_DEVICE_RDMA_NETDEV_OPA) + init_attr.create_flags |= IB_QP_CREATE_NETDEV_USE; + priv->qp = ib_create_qp(priv->pd, &init_attr); if (IS_ERR(priv->qp)) { pr_warn("%s: failed to create QP\n", ca->name); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_vlan.c b/drivers/infiniband/ulp/ipoib/ipoib_vlan.c index 8ac8e18fbe0c..30865605e098 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_vlan.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_vlan.c @@ -97,6 +97,7 @@ int __ipoib_vlan_add(struct ipoib_dev_priv *ppriv, struct ipoib_dev_priv *priv, { struct net_device *ndev = priv->dev; int result; + struct rdma_netdev *rn = netdev_priv(ndev); ASSERT_RTNL(); @@ -117,6 +118,8 @@ int __ipoib_vlan_add(struct ipoib_dev_priv *ppriv, struct ipoib_dev_priv *priv, goto out_early; } + rn->mtu = priv->mcast_mtu; + priv->parent = ppriv->dev; priv->pkey = pkey; priv->child_type = type; diff --git a/drivers/infiniband/ulp/iser/Kconfig b/drivers/infiniband/ulp/iser/Kconfig index 1d29dffeaff9..3016a0c9a9f0 100644 --- a/drivers/infiniband/ulp/iser/Kconfig +++ b/drivers/infiniband/ulp/iser/Kconfig @@ -3,7 +3,7 @@ config INFINIBAND_ISER tristate "iSCSI Extensions for RDMA (iSER)" depends on SCSI && INET && INFINIBAND_ADDR_TRANS select SCSI_ISCSI_ATTRS - ---help--- + help Support for the iSCSI Extensions for RDMA (iSER) Protocol over InfiniBand. This allows you to access storage devices that speak iSCSI over iSER over InfiniBand. diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h index 029c00163442..78ee9445f801 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.h +++ b/drivers/infiniband/ulp/iser/iscsi_iser.h @@ -65,7 +65,6 @@ #include <linux/in6.h> #include <rdma/ib_verbs.h> -#include <rdma/ib_fmr_pool.h> #include <rdma/rdma_cm.h> #define DRV_NAME "iser" @@ -301,45 +300,6 @@ struct ib_conn; struct iscsi_iser_task; /** - * struct iser_comp - iSER completion context - * - * @cq: completion queue - * @active_qps: Number of active QPs attached - * to completion context - */ -struct iser_comp { - struct ib_cq *cq; - int active_qps; -}; - -/** - * struct iser_reg_ops - Memory registration operations - * per-device registration schemes - * - * @alloc_reg_res: Allocate registration resources - * @free_reg_res: Free registration resources - * @reg_mem: Register memory buffers - * @unreg_mem: Un-register memory buffers - * @reg_desc_get: Get a registration descriptor for pool - * @reg_desc_put: Get a registration descriptor to pool - */ -struct iser_reg_ops { - int (*alloc_reg_res)(struct ib_conn *ib_conn, - unsigned cmds_max, - unsigned int size); - void (*free_reg_res)(struct ib_conn *ib_conn); - int (*reg_mem)(struct iscsi_iser_task *iser_task, - struct iser_data_buf *mem, - struct iser_reg_resources *rsc, - struct iser_mem_reg *reg); - void (*unreg_mem)(struct iscsi_iser_task *iser_task, - enum iser_data_dir cmd_dir); - struct iser_fr_desc * (*reg_desc_get)(struct ib_conn *ib_conn); - void (*reg_desc_put)(struct ib_conn *ib_conn, - struct iser_fr_desc *desc); -}; - -/** * struct iser_device - iSER device handle * * @ib_device: RDMA device @@ -348,11 +308,6 @@ struct iser_reg_ops { * @event_handler: IB events handle routine * @ig_list: entry in devices list * @refcount: Reference counter, dominated by open iser connections - * @comps_used: Number of completion contexts used, Min between online - * cpus and device max completion vectors - * @comps: Dinamically allocated array of completion handlers - * @reg_ops: Registration ops - * @remote_inv_sup: Remote invalidate is supported on this device */ struct iser_device { struct ib_device *ib_device; @@ -360,28 +315,18 @@ struct iser_device { struct ib_event_handler event_handler; struct list_head ig_list; int refcount; - int comps_used; - struct iser_comp *comps; - const struct iser_reg_ops *reg_ops; - bool remote_inv_sup; }; /** * struct iser_reg_resources - Fast registration resources * * @mr: memory region - * @fmr_pool: pool of fmrs * @sig_mr: signature memory region - * @page_vec: fast reg page list used by fmr pool * @mr_valid: is mr valid indicator */ struct iser_reg_resources { - union { - struct ib_mr *mr; - struct ib_fmr_pool *fmr_pool; - }; + struct ib_mr *mr; struct ib_mr *sig_mr; - struct iser_page_vec *page_vec; u8 mr_valid:1; }; @@ -391,6 +336,7 @@ struct iser_reg_resources { * @list: entry in connection fastreg pool * @rsc: data buffer registration resources * @sig_protected: is region protected indicator + * @all_list: first and last list members */ struct iser_fr_desc { struct list_head list; @@ -403,8 +349,9 @@ struct iser_fr_desc { * struct iser_fr_pool - connection fast registration pool * * @list: list of fastreg descriptors - * @lock: protects fmr/fastreg pool + * @lock: protects fastreg pool * @size: size of the pool + * @all_list: first and last list members */ struct iser_fr_pool { struct list_head list; @@ -418,11 +365,12 @@ struct iser_fr_pool { * * @cma_id: rdma_cm connection maneger handle * @qp: Connection Queue-pair + * @cq: Connection completion queue + * @cq_size: The number of max outstanding completions * @post_recv_buf_count: post receive counter * @sig_count: send work request signal count * @rx_wr: receive work request for batch posts * @device: reference to iser device - * @comp: iser completion context * @fr_pool: connection fast registration poool * @pi_support: Indicate device T10-PI support * @reg_cqe: completion handler @@ -430,11 +378,12 @@ struct iser_fr_pool { struct ib_conn { struct rdma_cm_id *cma_id; struct ib_qp *qp; + struct ib_cq *cq; + u32 cq_size; int post_recv_buf_count; u8 sig_count; struct ib_recv_wr rx_wr[ISER_MIN_POSTED_RX]; struct iser_device *device; - struct iser_comp *comp; struct iser_fr_pool fr_pool; bool pi_support; struct ib_cqe reg_cqe; @@ -518,12 +467,6 @@ struct iscsi_iser_task { struct iser_data_buf prot[ISER_DIRS_NUM]; }; -struct iser_page_vec { - u64 *pages; - int npages; - struct ib_mr fake_mr; -}; - /** * struct iser_global - iSER global context * @@ -548,8 +491,6 @@ extern int iser_pi_guard; extern unsigned int iser_max_sectors; extern bool iser_always_reg; -int iser_assign_reg_ops(struct iser_device *device); - int iser_send_control(struct iscsi_conn *conn, struct iscsi_task *task); @@ -591,22 +532,17 @@ void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task, struct iser_data_buf *mem, enum iser_data_dir cmd_dir); -int iser_reg_rdma_mem(struct iscsi_iser_task *task, - enum iser_data_dir dir, - bool all_imm); -void iser_unreg_rdma_mem(struct iscsi_iser_task *task, - enum iser_data_dir dir); +int iser_reg_mem_fastreg(struct iscsi_iser_task *task, + enum iser_data_dir dir, + bool all_imm); +void iser_unreg_mem_fastreg(struct iscsi_iser_task *task, + enum iser_data_dir dir); int iser_connect(struct iser_conn *iser_conn, struct sockaddr *src_addr, struct sockaddr *dst_addr, int non_blocking); -void iser_unreg_mem_fmr(struct iscsi_iser_task *iser_task, - enum iser_data_dir cmd_dir); -void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task, - enum iser_data_dir cmd_dir); - int iser_post_recvl(struct iser_conn *iser_conn); int iser_post_recvm(struct iser_conn *iser_conn, int count); int iser_post_send(struct ib_conn *ib_conn, struct iser_tx_desc *tx_desc, @@ -625,26 +561,12 @@ int iser_initialize_task_headers(struct iscsi_task *task, struct iser_tx_desc *tx_desc); int iser_alloc_rx_descriptors(struct iser_conn *iser_conn, struct iscsi_session *session); -int iser_alloc_fmr_pool(struct ib_conn *ib_conn, - unsigned cmds_max, - unsigned int size); -void iser_free_fmr_pool(struct ib_conn *ib_conn); int iser_alloc_fastreg_pool(struct ib_conn *ib_conn, unsigned cmds_max, unsigned int size); void iser_free_fastreg_pool(struct ib_conn *ib_conn); u8 iser_check_task_pi_status(struct iscsi_iser_task *iser_task, enum iser_data_dir cmd_dir, sector_t *sector); -struct iser_fr_desc * -iser_reg_desc_get_fr(struct ib_conn *ib_conn); -void -iser_reg_desc_put_fr(struct ib_conn *ib_conn, - struct iser_fr_desc *desc); -struct iser_fr_desc * -iser_reg_desc_get_fmr(struct ib_conn *ib_conn); -void -iser_reg_desc_put_fmr(struct ib_conn *ib_conn, - struct iser_fr_desc *desc); static inline struct iser_conn * to_iser_conn(struct ib_conn *ib_conn) diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c index 4a7045bb0831..27a6f75a9912 100644 --- a/drivers/infiniband/ulp/iser/iser_initiator.c +++ b/drivers/infiniband/ulp/iser/iser_initiator.c @@ -72,7 +72,7 @@ static int iser_prepare_read_cmd(struct iscsi_task *task) return err; } - err = iser_reg_rdma_mem(iser_task, ISER_DIR_IN, false); + err = iser_reg_mem_fastreg(iser_task, ISER_DIR_IN, false); if (err) { iser_err("Failed to set up Data-IN RDMA\n"); return err; @@ -126,8 +126,8 @@ iser_prepare_write_cmd(struct iscsi_task *task, return err; } - err = iser_reg_rdma_mem(iser_task, ISER_DIR_OUT, - buf_out->data_len == imm_sz); + err = iser_reg_mem_fastreg(iser_task, ISER_DIR_OUT, + buf_out->data_len == imm_sz); if (err != 0) { iser_err("Failed to register write cmd RDMA mem\n"); return err; @@ -250,8 +250,8 @@ int iser_alloc_rx_descriptors(struct iser_conn *iser_conn, iser_conn->qp_max_recv_dtos_mask = session->cmds_max - 1; /* cmds_max is 2^N */ iser_conn->min_posted_rx = iser_conn->qp_max_recv_dtos >> 2; - if (device->reg_ops->alloc_reg_res(ib_conn, session->scsi_cmds_max, - iser_conn->pages_per_mr)) + if (iser_alloc_fastreg_pool(ib_conn, session->scsi_cmds_max, + iser_conn->pages_per_mr)) goto create_rdma_reg_res_failed; if (iser_alloc_login_buf(iser_conn)) @@ -293,7 +293,7 @@ rx_desc_dma_map_failed: rx_desc_alloc_fail: iser_free_login_buf(iser_conn); alloc_login_buf_fail: - device->reg_ops->free_reg_res(ib_conn); + iser_free_fastreg_pool(ib_conn); create_rdma_reg_res_failed: iser_err("failed allocating rx descriptors / data buffers\n"); return -ENOMEM; @@ -306,8 +306,7 @@ void iser_free_rx_descriptors(struct iser_conn *iser_conn) struct ib_conn *ib_conn = &iser_conn->ib_conn; struct iser_device *device = ib_conn->device; - if (device->reg_ops->free_reg_res) - device->reg_ops->free_reg_res(ib_conn); + iser_free_fastreg_pool(ib_conn); rx_desc = iser_conn->rx_descs; for (i = 0; i < iser_conn->qp_max_recv_dtos; i++, rx_desc++) @@ -768,7 +767,7 @@ void iser_task_rdma_finalize(struct iscsi_iser_task *iser_task) int prot_count = scsi_prot_sg_count(iser_task->sc); if (iser_task->dir[ISER_DIR_IN]) { - iser_unreg_rdma_mem(iser_task, ISER_DIR_IN); + iser_unreg_mem_fastreg(iser_task, ISER_DIR_IN); iser_dma_unmap_task_data(iser_task, &iser_task->data[ISER_DIR_IN], DMA_FROM_DEVICE); @@ -779,7 +778,7 @@ void iser_task_rdma_finalize(struct iscsi_iser_task *iser_task) } if (iser_task->dir[ISER_DIR_OUT]) { - iser_unreg_rdma_mem(iser_task, ISER_DIR_OUT); + iser_unreg_mem_fastreg(iser_task, ISER_DIR_OUT); iser_dma_unmap_task_data(iser_task, &iser_task->data[ISER_DIR_OUT], DMA_TO_DEVICE); diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c index 999ef7cdd05e..d4e057fac219 100644 --- a/drivers/infiniband/ulp/iser/iser_memory.c +++ b/drivers/infiniband/ulp/iser/iser_memory.c @@ -38,62 +38,13 @@ #include <linux/scatterlist.h> #include "iscsi_iser.h" -static -int iser_fast_reg_fmr(struct iscsi_iser_task *iser_task, - struct iser_data_buf *mem, - struct iser_reg_resources *rsc, - struct iser_mem_reg *mem_reg); -static -int iser_fast_reg_mr(struct iscsi_iser_task *iser_task, - struct iser_data_buf *mem, - struct iser_reg_resources *rsc, - struct iser_mem_reg *mem_reg); - -static const struct iser_reg_ops fastreg_ops = { - .alloc_reg_res = iser_alloc_fastreg_pool, - .free_reg_res = iser_free_fastreg_pool, - .reg_mem = iser_fast_reg_mr, - .unreg_mem = iser_unreg_mem_fastreg, - .reg_desc_get = iser_reg_desc_get_fr, - .reg_desc_put = iser_reg_desc_put_fr, -}; - -static const struct iser_reg_ops fmr_ops = { - .alloc_reg_res = iser_alloc_fmr_pool, - .free_reg_res = iser_free_fmr_pool, - .reg_mem = iser_fast_reg_fmr, - .unreg_mem = iser_unreg_mem_fmr, - .reg_desc_get = iser_reg_desc_get_fmr, - .reg_desc_put = iser_reg_desc_put_fmr, -}; void iser_reg_comp(struct ib_cq *cq, struct ib_wc *wc) { iser_err_comp(wc, "memreg"); } -int iser_assign_reg_ops(struct iser_device *device) -{ - struct ib_device *ib_dev = device->ib_device; - - /* Assign function handles - based on FMR support */ - if (ib_dev->ops.alloc_fmr && ib_dev->ops.dealloc_fmr && - ib_dev->ops.map_phys_fmr && ib_dev->ops.unmap_fmr) { - iser_info("FMR supported, using FMR for registration\n"); - device->reg_ops = &fmr_ops; - } else if (ib_dev->attrs.device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) { - iser_info("FastReg supported, using FastReg for registration\n"); - device->reg_ops = &fastreg_ops; - device->remote_inv_sup = iser_always_reg; - } else { - iser_err("IB device does not support FMRs nor FastRegs, can't register memory\n"); - return -1; - } - - return 0; -} - -struct iser_fr_desc * +static struct iser_fr_desc * iser_reg_desc_get_fr(struct ib_conn *ib_conn) { struct iser_fr_pool *fr_pool = &ib_conn->fr_pool; @@ -109,7 +60,7 @@ iser_reg_desc_get_fr(struct ib_conn *ib_conn) return desc; } -void +static void iser_reg_desc_put_fr(struct ib_conn *ib_conn, struct iser_fr_desc *desc) { @@ -121,44 +72,6 @@ iser_reg_desc_put_fr(struct ib_conn *ib_conn, spin_unlock_irqrestore(&fr_pool->lock, flags); } -struct iser_fr_desc * -iser_reg_desc_get_fmr(struct ib_conn *ib_conn) -{ - struct iser_fr_pool *fr_pool = &ib_conn->fr_pool; - - return list_first_entry(&fr_pool->list, - struct iser_fr_desc, list); -} - -void -iser_reg_desc_put_fmr(struct ib_conn *ib_conn, - struct iser_fr_desc *desc) -{ -} - -static void iser_data_buf_dump(struct iser_data_buf *data, - struct ib_device *ibdev) -{ - struct scatterlist *sg; - int i; - - for_each_sg(data->sg, sg, data->dma_nents, i) - iser_dbg("sg[%d] dma_addr:0x%lX page:0x%p " - "off:0x%x sz:0x%x dma_len:0x%x\n", - i, (unsigned long)sg_dma_address(sg), - sg_page(sg), sg->offset, sg->length, sg_dma_len(sg)); -} - -static void iser_dump_page_vec(struct iser_page_vec *page_vec) -{ - int i; - - iser_err("page vec npages %d data length %lld\n", - page_vec->npages, page_vec->fake_mr.length); - for (i = 0; i < page_vec->npages; i++) - iser_err("vec[%d]: %llx\n", i, page_vec->pages[i]); -} - int iser_dma_map_task_data(struct iscsi_iser_task *iser_task, struct iser_data_buf *data, enum iser_data_dir iser_dir, @@ -213,84 +126,9 @@ iser_reg_dma(struct iser_device *device, struct iser_data_buf *mem, return 0; } -static int iser_set_page(struct ib_mr *mr, u64 addr) -{ - struct iser_page_vec *page_vec = - container_of(mr, struct iser_page_vec, fake_mr); - - page_vec->pages[page_vec->npages++] = addr; - - return 0; -} - -static -int iser_fast_reg_fmr(struct iscsi_iser_task *iser_task, - struct iser_data_buf *mem, - struct iser_reg_resources *rsc, - struct iser_mem_reg *reg) -{ - struct ib_conn *ib_conn = &iser_task->iser_conn->ib_conn; - struct iser_device *device = ib_conn->device; - struct iser_page_vec *page_vec = rsc->page_vec; - struct ib_fmr_pool *fmr_pool = rsc->fmr_pool; - struct ib_pool_fmr *fmr; - int ret, plen; - - page_vec->npages = 0; - page_vec->fake_mr.page_size = SZ_4K; - plen = ib_sg_to_pages(&page_vec->fake_mr, mem->sg, - mem->dma_nents, NULL, iser_set_page); - if (unlikely(plen < mem->dma_nents)) { - iser_err("page vec too short to hold this SG\n"); - iser_data_buf_dump(mem, device->ib_device); - iser_dump_page_vec(page_vec); - return -EINVAL; - } - - fmr = ib_fmr_pool_map_phys(fmr_pool, page_vec->pages, - page_vec->npages, page_vec->pages[0]); - if (IS_ERR(fmr)) { - ret = PTR_ERR(fmr); - iser_err("ib_fmr_pool_map_phys failed: %d\n", ret); - return ret; - } - - reg->sge.lkey = fmr->fmr->lkey; - reg->rkey = fmr->fmr->rkey; - reg->sge.addr = page_vec->fake_mr.iova; - reg->sge.length = page_vec->fake_mr.length; - reg->mem_h = fmr; - - iser_dbg("fmr reg: lkey=0x%x, rkey=0x%x, addr=0x%llx," - " length=0x%x\n", reg->sge.lkey, reg->rkey, - reg->sge.addr, reg->sge.length); - - return 0; -} - -/** - * Unregister (previosuly registered using FMR) memory. - * If memory is non-FMR does nothing. - */ -void iser_unreg_mem_fmr(struct iscsi_iser_task *iser_task, - enum iser_data_dir cmd_dir) -{ - struct iser_mem_reg *reg = &iser_task->rdma_reg[cmd_dir]; - - if (!reg->mem_h) - return; - - iser_dbg("PHYSICAL Mem.Unregister mem_h %p\n", reg->mem_h); - - ib_fmr_pool_unmap((struct ib_pool_fmr *)reg->mem_h); - - reg->mem_h = NULL; -} - void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task, enum iser_data_dir cmd_dir) { - struct iser_device *device = iser_task->iser_conn->ib_conn.device; struct iser_mem_reg *reg = &iser_task->rdma_reg[cmd_dir]; struct iser_fr_desc *desc; struct ib_mr_status mr_status; @@ -312,7 +150,7 @@ void iser_unreg_mem_fastreg(struct iscsi_iser_task *iser_task, ib_check_mr_status(desc->rsc.sig_mr, IB_MR_CHECK_SIG_STATUS, &mr_status); } - device->reg_ops->reg_desc_put(&iser_task->iser_conn->ib_conn, desc); + iser_reg_desc_put_fr(&iser_task->iser_conn->ib_conn, reg->mem_h); reg->mem_h = NULL; } @@ -509,15 +347,14 @@ iser_reg_data_sg(struct iscsi_iser_task *task, if (use_dma_key) return iser_reg_dma(device, mem, reg); - return device->reg_ops->reg_mem(task, mem, &desc->rsc, reg); + return iser_fast_reg_mr(task, mem, &desc->rsc, reg); } -int iser_reg_rdma_mem(struct iscsi_iser_task *task, - enum iser_data_dir dir, - bool all_imm) +int iser_reg_mem_fastreg(struct iscsi_iser_task *task, + enum iser_data_dir dir, + bool all_imm) { struct ib_conn *ib_conn = &task->iser_conn->ib_conn; - struct iser_device *device = ib_conn->device; struct iser_data_buf *mem = &task->data[dir]; struct iser_mem_reg *reg = &task->rdma_reg[dir]; struct iser_fr_desc *desc = NULL; @@ -528,7 +365,7 @@ int iser_reg_rdma_mem(struct iscsi_iser_task *task, scsi_get_prot_op(task->sc) == SCSI_PROT_NORMAL; if (!use_dma_key) { - desc = device->reg_ops->reg_desc_get(ib_conn); + desc = iser_reg_desc_get_fr(ib_conn); reg->mem_h = desc; } @@ -549,15 +386,8 @@ int iser_reg_rdma_mem(struct iscsi_iser_task *task, err_reg: if (desc) - device->reg_ops->reg_desc_put(ib_conn, desc); + iser_reg_desc_put_fr(ib_conn, desc); return err; } -void iser_unreg_rdma_mem(struct iscsi_iser_task *task, - enum iser_data_dir dir) -{ - struct iser_device *device = task->iser_conn->ib_conn.device; - - device->reg_ops->unreg_mem(task, dir); -} diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index 127887c6c03f..2f3ebc0a75d9 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -68,58 +68,23 @@ static void iser_event_handler(struct ib_event_handler *handler, static int iser_create_device_ib_res(struct iser_device *device) { struct ib_device *ib_dev = device->ib_device; - int ret, i, max_cqe; - ret = iser_assign_reg_ops(device); - if (ret) - return ret; - - device->comps_used = min_t(int, num_online_cpus(), - ib_dev->num_comp_vectors); - - device->comps = kcalloc(device->comps_used, sizeof(*device->comps), - GFP_KERNEL); - if (!device->comps) - goto comps_err; - - max_cqe = min(ISER_MAX_CQ_LEN, ib_dev->attrs.max_cqe); - - iser_info("using %d CQs, device %s supports %d vectors max_cqe %d\n", - device->comps_used, dev_name(&ib_dev->dev), - ib_dev->num_comp_vectors, max_cqe); + if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS)) { + iser_err("IB device does not support memory registrations\n"); + return -1; + } device->pd = ib_alloc_pd(ib_dev, iser_always_reg ? 0 : IB_PD_UNSAFE_GLOBAL_RKEY); if (IS_ERR(device->pd)) goto pd_err; - for (i = 0; i < device->comps_used; i++) { - struct iser_comp *comp = &device->comps[i]; - - comp->cq = ib_alloc_cq(ib_dev, comp, max_cqe, i, - IB_POLL_SOFTIRQ); - if (IS_ERR(comp->cq)) { - comp->cq = NULL; - goto cq_err; - } - } - INIT_IB_EVENT_HANDLER(&device->event_handler, ib_dev, iser_event_handler); ib_register_event_handler(&device->event_handler); return 0; -cq_err: - for (i = 0; i < device->comps_used; i++) { - struct iser_comp *comp = &device->comps[i]; - - if (comp->cq) - ib_free_cq(comp->cq); - } - ib_dealloc_pd(device->pd); pd_err: - kfree(device->comps); -comps_err: iser_err("failed to allocate an IB resource\n"); return -1; } @@ -130,113 +95,12 @@ comps_err: */ static void iser_free_device_ib_res(struct iser_device *device) { - int i; - - for (i = 0; i < device->comps_used; i++) { - struct iser_comp *comp = &device->comps[i]; - - ib_free_cq(comp->cq); - comp->cq = NULL; - } - ib_unregister_event_handler(&device->event_handler); ib_dealloc_pd(device->pd); - kfree(device->comps); - device->comps = NULL; device->pd = NULL; } -/** - * iser_alloc_fmr_pool - Creates FMR pool and page_vector - * @ib_conn: connection RDMA resources - * @cmds_max: max number of SCSI commands for this connection - * @size: max number of pages per map request - * - * Return: 0 on success, or errno code on failure - */ -int iser_alloc_fmr_pool(struct ib_conn *ib_conn, - unsigned cmds_max, - unsigned int size) -{ - struct iser_device *device = ib_conn->device; - struct iser_fr_pool *fr_pool = &ib_conn->fr_pool; - struct iser_page_vec *page_vec; - struct iser_fr_desc *desc; - struct ib_fmr_pool *fmr_pool; - struct ib_fmr_pool_param params; - int ret; - - INIT_LIST_HEAD(&fr_pool->list); - spin_lock_init(&fr_pool->lock); - - desc = kzalloc(sizeof(*desc), GFP_KERNEL); - if (!desc) - return -ENOMEM; - - page_vec = kmalloc(sizeof(*page_vec) + (sizeof(u64) * size), - GFP_KERNEL); - if (!page_vec) { - ret = -ENOMEM; - goto err_frpl; - } - - page_vec->pages = (u64 *)(page_vec + 1); - - params.page_shift = ilog2(SZ_4K); - params.max_pages_per_fmr = size; - /* make the pool size twice the max number of SCSI commands * - * the ML is expected to queue, watermark for unmap at 50% */ - params.pool_size = cmds_max * 2; - params.dirty_watermark = cmds_max; - params.cache = 0; - params.flush_function = NULL; - params.access = (IB_ACCESS_LOCAL_WRITE | - IB_ACCESS_REMOTE_WRITE | - IB_ACCESS_REMOTE_READ); - - fmr_pool = ib_create_fmr_pool(device->pd, ¶ms); - if (IS_ERR(fmr_pool)) { - ret = PTR_ERR(fmr_pool); - iser_err("FMR allocation failed, err %d\n", ret); - goto err_fmr; - } - - desc->rsc.page_vec = page_vec; - desc->rsc.fmr_pool = fmr_pool; - list_add(&desc->list, &fr_pool->list); - - return 0; - -err_fmr: - kfree(page_vec); -err_frpl: - kfree(desc); - - return ret; -} - -/** - * iser_free_fmr_pool - releases the FMR pool and page vec - * @ib_conn: connection RDMA resources - */ -void iser_free_fmr_pool(struct ib_conn *ib_conn) -{ - struct iser_fr_pool *fr_pool = &ib_conn->fr_pool; - struct iser_fr_desc *desc; - - desc = list_first_entry(&fr_pool->list, - struct iser_fr_desc, list); - list_del(&desc->list); - - iser_info("freeing conn %p fmr pool %p\n", - ib_conn, desc->rsc.fmr_pool); - - ib_destroy_fmr_pool(desc->rsc.fmr_pool); - kfree(desc->rsc.page_vec); - kfree(desc); -} - static struct iser_fr_desc * iser_create_fastreg_desc(struct iser_device *device, struct ib_pd *pd, @@ -376,70 +240,57 @@ static int iser_create_ib_conn_res(struct ib_conn *ib_conn) struct ib_device *ib_dev; struct ib_qp_init_attr init_attr; int ret = -ENOMEM; - int index, min_index = 0; + unsigned int max_send_wr, cq_size; BUG_ON(ib_conn->device == NULL); device = ib_conn->device; ib_dev = device->ib_device; - memset(&init_attr, 0, sizeof init_attr); + if (ib_conn->pi_support) + max_send_wr = ISER_QP_SIG_MAX_REQ_DTOS + 1; + else + max_send_wr = ISER_QP_MAX_REQ_DTOS + 1; + max_send_wr = min_t(unsigned int, max_send_wr, + (unsigned int)ib_dev->attrs.max_qp_wr); - mutex_lock(&ig.connlist_mutex); - /* select the CQ with the minimal number of usages */ - for (index = 0; index < device->comps_used; index++) { - if (device->comps[index].active_qps < - device->comps[min_index].active_qps) - min_index = index; + cq_size = max_send_wr + ISER_QP_MAX_RECV_DTOS; + ib_conn->cq = ib_cq_pool_get(ib_dev, cq_size, -1, IB_POLL_SOFTIRQ); + if (IS_ERR(ib_conn->cq)) { + ret = PTR_ERR(ib_conn->cq); + goto cq_err; } - ib_conn->comp = &device->comps[min_index]; - ib_conn->comp->active_qps++; - mutex_unlock(&ig.connlist_mutex); - iser_info("cq index %d used for ib_conn %p\n", min_index, ib_conn); + ib_conn->cq_size = cq_size; + + memset(&init_attr, 0, sizeof(init_attr)); init_attr.event_handler = iser_qp_event_callback; init_attr.qp_context = (void *)ib_conn; - init_attr.send_cq = ib_conn->comp->cq; - init_attr.recv_cq = ib_conn->comp->cq; + init_attr.send_cq = ib_conn->cq; + init_attr.recv_cq = ib_conn->cq; init_attr.cap.max_recv_wr = ISER_QP_MAX_RECV_DTOS; init_attr.cap.max_send_sge = 2; init_attr.cap.max_recv_sge = 1; init_attr.sq_sig_type = IB_SIGNAL_REQ_WR; init_attr.qp_type = IB_QPT_RC; - if (ib_conn->pi_support) { - init_attr.cap.max_send_wr = ISER_QP_SIG_MAX_REQ_DTOS + 1; + init_attr.cap.max_send_wr = max_send_wr; + if (ib_conn->pi_support) init_attr.create_flags |= IB_QP_CREATE_INTEGRITY_EN; - iser_conn->max_cmds = - ISER_GET_MAX_XMIT_CMDS(ISER_QP_SIG_MAX_REQ_DTOS); - } else { - if (ib_dev->attrs.max_qp_wr > ISER_QP_MAX_REQ_DTOS) { - init_attr.cap.max_send_wr = ISER_QP_MAX_REQ_DTOS + 1; - iser_conn->max_cmds = - ISER_GET_MAX_XMIT_CMDS(ISER_QP_MAX_REQ_DTOS); - } else { - init_attr.cap.max_send_wr = ib_dev->attrs.max_qp_wr; - iser_conn->max_cmds = - ISER_GET_MAX_XMIT_CMDS(ib_dev->attrs.max_qp_wr); - iser_dbg("device %s supports max_send_wr %d\n", - dev_name(&device->ib_device->dev), - ib_dev->attrs.max_qp_wr); - } - } + iser_conn->max_cmds = ISER_GET_MAX_XMIT_CMDS(max_send_wr - 1); ret = rdma_create_qp(ib_conn->cma_id, device->pd, &init_attr); if (ret) goto out_err; ib_conn->qp = ib_conn->cma_id->qp; - iser_info("setting conn %p cma_id %p qp %p\n", + iser_info("setting conn %p cma_id %p qp %p max_send_wr %d\n", ib_conn, ib_conn->cma_id, - ib_conn->cma_id->qp); + ib_conn->cma_id->qp, max_send_wr); return ret; out_err: - mutex_lock(&ig.connlist_mutex); - ib_conn->comp->active_qps--; - mutex_unlock(&ig.connlist_mutex); + ib_cq_pool_put(ib_conn->cq, ib_conn->cq_size); +cq_err: iser_err("unable to alloc mem or create resource, err %d\n", ret); return ret; @@ -551,10 +402,8 @@ static void iser_free_ib_conn_res(struct iser_conn *iser_conn, iser_conn, ib_conn->cma_id, ib_conn->qp); if (ib_conn->qp != NULL) { - mutex_lock(&ig.connlist_mutex); - ib_conn->comp->active_qps--; - mutex_unlock(&ig.connlist_mutex); rdma_destroy_qp(ib_conn->cma_id); + ib_cq_pool_put(ib_conn->cq, ib_conn->cq_size); ib_conn->qp = NULL; } @@ -667,13 +516,12 @@ iser_calc_scsi_params(struct iser_conn *iser_conn, u32 max_num_sg; /* - * FRs without SG_GAPS or FMRs can only map up to a (device) page per - * entry, but if the first entry is misaligned we'll end up using two - * entries (head and tail) for a single page worth data, so one - * additional entry is required. + * FRs without SG_GAPS can only map up to a (device) page per entry, + * but if the first entry is misaligned we'll end up using two entries + * (head and tail) for a single page worth data, so one additional + * entry is required. */ - if ((attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) && - (attr->device_cap_flags & IB_DEVICE_SG_GAPS_REG)) + if (attr->device_cap_flags & IB_DEVICE_SG_GAPS_REG) reserved_mr_pages = 0; else reserved_mr_pages = 1; @@ -684,14 +532,8 @@ iser_calc_scsi_params(struct iser_conn *iser_conn, max_num_sg = attr->max_fast_reg_page_list_len; sg_tablesize = DIV_ROUND_UP(max_sectors * SECTOR_SIZE, SZ_4K); - if (attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) - sup_sg_tablesize = - min_t( - uint, ISCSI_ISER_MAX_SG_TABLESIZE, - max_num_sg - reserved_mr_pages); - else - sup_sg_tablesize = ISCSI_ISER_MAX_SG_TABLESIZE; - + sup_sg_tablesize = min_t(uint, ISCSI_ISER_MAX_SG_TABLESIZE, + max_num_sg - reserved_mr_pages); iser_conn->scsi_sg_tablesize = min(sg_tablesize, sup_sg_tablesize); iser_conn->pages_per_mr = iser_conn->scsi_sg_tablesize + reserved_mr_pages; @@ -755,7 +597,7 @@ static void iser_route_handler(struct rdma_cm_id *cma_id) struct iser_cm_hdr req_hdr; struct iser_conn *iser_conn = (struct iser_conn *)cma_id->context; struct ib_conn *ib_conn = &iser_conn->ib_conn; - struct iser_device *device = ib_conn->device; + struct ib_device *ib_dev = ib_conn->device->ib_device; if (iser_conn->state != ISER_CONN_PENDING) /* bailout */ @@ -766,14 +608,14 @@ static void iser_route_handler(struct rdma_cm_id *cma_id) goto failure; memset(&conn_param, 0, sizeof conn_param); - conn_param.responder_resources = device->ib_device->attrs.max_qp_rd_atom; + conn_param.responder_resources = ib_dev->attrs.max_qp_rd_atom; conn_param.initiator_depth = 1; conn_param.retry_count = 7; conn_param.rnr_retry_count = 6; memset(&req_hdr, 0, sizeof(req_hdr)); req_hdr.flags = ISER_ZBVA_NOT_SUP; - if (!device->remote_inv_sup) + if (!iser_always_reg) req_hdr.flags |= ISER_SEND_W_INV_NOT_SUP; conn_param.private_data = (void *)&req_hdr; conn_param.private_data_len = sizeof(struct iser_cm_hdr); @@ -869,7 +711,7 @@ static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *eve case RDMA_CM_EVENT_REJECTED: iser_info("Connection rejected: %s\n", rdma_reject_msg(cma_id, event->status)); - /* FALLTHROUGH */ + fallthrough; case RDMA_CM_EVENT_ADDR_ERROR: case RDMA_CM_EVENT_ROUTE_ERROR: case RDMA_CM_EVENT_CONNECT_ERROR: diff --git a/drivers/infiniband/ulp/isert/Kconfig b/drivers/infiniband/ulp/isert/Kconfig index 1a3f5ca8354c..798147adbef1 100644 --- a/drivers/infiniband/ulp/isert/Kconfig +++ b/drivers/infiniband/ulp/isert/Kconfig @@ -2,5 +2,5 @@ config INFINIBAND_ISERT tristate "iSCSI Extensions for RDMA (iSER) target support" depends on INET && INFINIBAND_ADDR_TRANS && TARGET_CORE && ISCSI_TARGET - ---help--- + help Support for iSCSI Extensions for RDMA (iSER) Target on Infiniband fabrics. diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index a1a035270cab..695f701dc43d 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -15,6 +15,7 @@ #include <linux/in.h> #include <linux/in6.h> #include <rdma/ib_verbs.h> +#include <rdma/ib_cm.h> #include <rdma/rdma_cm.h> #include <target/target_core_base.h> #include <target/target_core_fabric.h> @@ -23,13 +24,6 @@ #include "ib_isert.h" -#define ISERT_MAX_CONN 8 -#define ISER_MAX_RX_CQ_LEN (ISERT_QP_MAX_RECV_DTOS * ISERT_MAX_CONN) -#define ISER_MAX_TX_CQ_LEN \ - ((ISERT_QP_MAX_REQ_DTOS + ISCSI_DEF_XMIT_CMDS_MAX) * ISERT_MAX_CONN) -#define ISER_MAX_CQ_LEN (ISER_MAX_RX_CQ_LEN + ISER_MAX_TX_CQ_LEN + \ - ISERT_MAX_CONN) - static int isert_debug_level; module_param_named(debug_level, isert_debug_level, int, 0644); MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0 (default:0)"); @@ -81,53 +75,34 @@ isert_qp_event_callback(struct ib_event *e, void *context) } } -static struct isert_comp * -isert_comp_get(struct isert_conn *isert_conn) -{ - struct isert_device *device = isert_conn->device; - struct isert_comp *comp; - int i, min = 0; - - mutex_lock(&device_list_mutex); - for (i = 0; i < device->comps_used; i++) - if (device->comps[i].active_qps < - device->comps[min].active_qps) - min = i; - comp = &device->comps[min]; - comp->active_qps++; - mutex_unlock(&device_list_mutex); - - isert_info("conn %p, using comp %p min_index: %d\n", - isert_conn, comp, min); - - return comp; -} - -static void -isert_comp_put(struct isert_comp *comp) -{ - mutex_lock(&device_list_mutex); - comp->active_qps--; - mutex_unlock(&device_list_mutex); -} - static struct ib_qp * isert_create_qp(struct isert_conn *isert_conn, - struct isert_comp *comp, struct rdma_cm_id *cma_id) { + u32 cq_size = ISERT_QP_MAX_REQ_DTOS + ISERT_QP_MAX_RECV_DTOS + 2; struct isert_device *device = isert_conn->device; + struct ib_device *ib_dev = device->ib_device; struct ib_qp_init_attr attr; - int ret; + int ret, factor; + + isert_conn->cq = ib_cq_pool_get(ib_dev, cq_size, -1, IB_POLL_WORKQUEUE); + if (IS_ERR(isert_conn->cq)) { + isert_err("Unable to allocate cq\n"); + ret = PTR_ERR(isert_conn->cq); + return ERR_PTR(ret); + } + isert_conn->cq_size = cq_size; memset(&attr, 0, sizeof(struct ib_qp_init_attr)); attr.event_handler = isert_qp_event_callback; attr.qp_context = isert_conn; - attr.send_cq = comp->cq; - attr.recv_cq = comp->cq; + attr.send_cq = isert_conn->cq; + attr.recv_cq = isert_conn->cq; attr.cap.max_send_wr = ISERT_QP_MAX_REQ_DTOS + 1; attr.cap.max_recv_wr = ISERT_QP_MAX_RECV_DTOS + 1; - attr.cap.max_rdma_ctxs = ISCSI_DEF_XMIT_CMDS_MAX; + factor = rdma_rw_mr_factor(device->ib_device, cma_id->port_num, + ISCSI_ISER_MAX_SG_TABLESIZE); + attr.cap.max_rdma_ctxs = ISCSI_DEF_XMIT_CMDS_MAX * factor; attr.cap.max_send_sge = device->ib_device->attrs.max_send_sge; attr.cap.max_recv_sge = 1; attr.sq_sig_type = IB_SIGNAL_REQ_WR; @@ -138,6 +113,8 @@ isert_create_qp(struct isert_conn *isert_conn, ret = rdma_create_qp(cma_id, device->pd, &attr); if (ret) { isert_err("rdma_create_qp failed for cma_id %d\n", ret); + ib_cq_pool_put(isert_conn->cq, isert_conn->cq_size); + return ERR_PTR(ret); } @@ -145,25 +122,6 @@ isert_create_qp(struct isert_conn *isert_conn, } static int -isert_conn_setup_qp(struct isert_conn *isert_conn, struct rdma_cm_id *cma_id) -{ - struct isert_comp *comp; - int ret; - - comp = isert_comp_get(isert_conn); - isert_conn->qp = isert_create_qp(isert_conn, comp, cma_id); - if (IS_ERR(isert_conn->qp)) { - ret = PTR_ERR(isert_conn->qp); - goto err; - } - - return 0; -err: - isert_comp_put(comp); - return ret; -} - -static int isert_alloc_rx_descriptors(struct isert_conn *isert_conn) { struct isert_device *device = isert_conn->device; @@ -182,15 +140,15 @@ isert_alloc_rx_descriptors(struct isert_conn *isert_conn) rx_desc = isert_conn->rx_descs; for (i = 0; i < ISERT_QP_MAX_RECV_DTOS; i++, rx_desc++) { - dma_addr = ib_dma_map_single(ib_dev, (void *)rx_desc, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + dma_addr = ib_dma_map_single(ib_dev, rx_desc->buf, + ISER_RX_SIZE, DMA_FROM_DEVICE); if (ib_dma_mapping_error(ib_dev, dma_addr)) goto dma_map_fail; rx_desc->dma_addr = dma_addr; rx_sg = &rx_desc->rx_sg; - rx_sg->addr = rx_desc->dma_addr; + rx_sg->addr = rx_desc->dma_addr + isert_get_hdr_offset(rx_desc); rx_sg->length = ISER_RX_PAYLOAD_SIZE; rx_sg->lkey = device->pd->local_dma_lkey; rx_desc->rx_cqe.done = isert_recv_done; @@ -202,7 +160,7 @@ dma_map_fail: rx_desc = isert_conn->rx_descs; for (j = 0; j < i; j++, rx_desc++) { ib_dma_unmap_single(ib_dev, rx_desc->dma_addr, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + ISER_RX_SIZE, DMA_FROM_DEVICE); } kfree(isert_conn->rx_descs); isert_conn->rx_descs = NULL; @@ -223,68 +181,13 @@ isert_free_rx_descriptors(struct isert_conn *isert_conn) rx_desc = isert_conn->rx_descs; for (i = 0; i < ISERT_QP_MAX_RECV_DTOS; i++, rx_desc++) { ib_dma_unmap_single(ib_dev, rx_desc->dma_addr, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + ISER_RX_SIZE, DMA_FROM_DEVICE); } kfree(isert_conn->rx_descs); isert_conn->rx_descs = NULL; } -static void -isert_free_comps(struct isert_device *device) -{ - int i; - - for (i = 0; i < device->comps_used; i++) { - struct isert_comp *comp = &device->comps[i]; - - if (comp->cq) - ib_free_cq(comp->cq); - } - kfree(device->comps); -} - -static int -isert_alloc_comps(struct isert_device *device) -{ - int i, max_cqe, ret = 0; - - device->comps_used = min(ISERT_MAX_CQ, min_t(int, num_online_cpus(), - device->ib_device->num_comp_vectors)); - - isert_info("Using %d CQs, %s supports %d vectors support " - "pi_capable %d\n", - device->comps_used, dev_name(&device->ib_device->dev), - device->ib_device->num_comp_vectors, - device->pi_capable); - - device->comps = kcalloc(device->comps_used, sizeof(struct isert_comp), - GFP_KERNEL); - if (!device->comps) - return -ENOMEM; - - max_cqe = min(ISER_MAX_CQ_LEN, device->ib_device->attrs.max_cqe); - - for (i = 0; i < device->comps_used; i++) { - struct isert_comp *comp = &device->comps[i]; - - comp->device = device; - comp->cq = ib_alloc_cq(device->ib_device, comp, max_cqe, i, - IB_POLL_WORKQUEUE); - if (IS_ERR(comp->cq)) { - isert_err("Unable to allocate cq\n"); - ret = PTR_ERR(comp->cq); - comp->cq = NULL; - goto out_cq; - } - } - - return 0; -out_cq: - isert_free_comps(device); - return ret; -} - static int isert_create_device_ib_res(struct isert_device *device) { @@ -295,16 +198,12 @@ isert_create_device_ib_res(struct isert_device *device) ib_dev->attrs.max_send_sge, ib_dev->attrs.max_recv_sge); isert_dbg("devattr->max_sge_rd: %d\n", ib_dev->attrs.max_sge_rd); - ret = isert_alloc_comps(device); - if (ret) - goto out; - device->pd = ib_alloc_pd(ib_dev, 0); if (IS_ERR(device->pd)) { ret = PTR_ERR(device->pd); isert_err("failed to allocate pd, device %p, ret=%d\n", device, ret); - goto out_cq; + return ret; } /* Check signature cap */ @@ -312,13 +211,6 @@ isert_create_device_ib_res(struct isert_device *device) IB_DEVICE_INTEGRITY_HANDOVER ? true : false; return 0; - -out_cq: - isert_free_comps(device); -out: - if (ret > 0) - ret = -EINVAL; - return ret; } static void @@ -327,7 +219,6 @@ isert_free_device_ib_res(struct isert_device *device) isert_info("device %p\n", device); ib_dealloc_pd(device->pd); - isert_free_comps(device); } static void @@ -408,10 +299,9 @@ isert_free_login_buf(struct isert_conn *isert_conn) ISER_RX_PAYLOAD_SIZE, DMA_TO_DEVICE); kfree(isert_conn->login_rsp_buf); - ib_dma_unmap_single(ib_dev, isert_conn->login_req_dma, - ISER_RX_PAYLOAD_SIZE, - DMA_FROM_DEVICE); - kfree(isert_conn->login_req_buf); + ib_dma_unmap_single(ib_dev, isert_conn->login_desc->dma_addr, + ISER_RX_SIZE, DMA_FROM_DEVICE); + kfree(isert_conn->login_desc); } static int @@ -420,25 +310,25 @@ isert_alloc_login_buf(struct isert_conn *isert_conn, { int ret; - isert_conn->login_req_buf = kzalloc(sizeof(*isert_conn->login_req_buf), + isert_conn->login_desc = kzalloc(sizeof(*isert_conn->login_desc), GFP_KERNEL); - if (!isert_conn->login_req_buf) + if (!isert_conn->login_desc) return -ENOMEM; - isert_conn->login_req_dma = ib_dma_map_single(ib_dev, - isert_conn->login_req_buf, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); - ret = ib_dma_mapping_error(ib_dev, isert_conn->login_req_dma); + isert_conn->login_desc->dma_addr = ib_dma_map_single(ib_dev, + isert_conn->login_desc->buf, + ISER_RX_SIZE, DMA_FROM_DEVICE); + ret = ib_dma_mapping_error(ib_dev, isert_conn->login_desc->dma_addr); if (ret) { - isert_err("login_req_dma mapping error: %d\n", ret); - isert_conn->login_req_dma = 0; - goto out_free_login_req_buf; + isert_err("login_desc dma mapping error: %d\n", ret); + isert_conn->login_desc->dma_addr = 0; + goto out_free_login_desc; } isert_conn->login_rsp_buf = kzalloc(ISER_RX_PAYLOAD_SIZE, GFP_KERNEL); if (!isert_conn->login_rsp_buf) { ret = -ENOMEM; - goto out_unmap_login_req_buf; + goto out_unmap_login_desc; } isert_conn->login_rsp_dma = ib_dma_map_single(ib_dev, @@ -455,11 +345,11 @@ isert_alloc_login_buf(struct isert_conn *isert_conn, out_free_login_rsp_buf: kfree(isert_conn->login_rsp_buf); -out_unmap_login_req_buf: - ib_dma_unmap_single(ib_dev, isert_conn->login_req_dma, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); -out_free_login_req_buf: - kfree(isert_conn->login_req_buf); +out_unmap_login_desc: + ib_dma_unmap_single(ib_dev, isert_conn->login_desc->dma_addr, + ISER_RX_SIZE, DMA_FROM_DEVICE); +out_free_login_desc: + kfree(isert_conn->login_desc); return ret; } @@ -489,6 +379,13 @@ isert_set_nego_params(struct isert_conn *isert_conn, } } +static void +isert_destroy_qp(struct isert_conn *isert_conn) +{ + ib_destroy_qp(isert_conn->qp); + ib_cq_pool_put(isert_conn->cq, isert_conn->cq_size); +} + static int isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) { @@ -502,7 +399,7 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) if (!np->enabled) { spin_unlock_bh(&np->np_thread_lock); isert_dbg("iscsi_np is not enabled, reject connect request\n"); - return rdma_reject(cma_id, NULL, 0); + return rdma_reject(cma_id, NULL, 0, IB_CM_REJ_CONSUMER_DEFINED); } spin_unlock_bh(&np->np_thread_lock); @@ -529,17 +426,19 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) isert_set_nego_params(isert_conn, &event->param.conn); - ret = isert_conn_setup_qp(isert_conn, cma_id); - if (ret) + isert_conn->qp = isert_create_qp(isert_conn, cma_id); + if (IS_ERR(isert_conn->qp)) { + ret = PTR_ERR(isert_conn->qp); goto out_conn_dev; + } ret = isert_login_post_recv(isert_conn); if (ret) - goto out_conn_dev; + goto out_destroy_qp; ret = isert_rdma_accept(isert_conn); if (ret) - goto out_conn_dev; + goto out_destroy_qp; mutex_lock(&isert_np->mutex); list_add_tail(&isert_conn->node, &isert_np->accepted); @@ -547,13 +446,15 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) return 0; +out_destroy_qp: + isert_destroy_qp(isert_conn); out_conn_dev: isert_device_put(device); out_rsp_dma_map: isert_free_login_buf(isert_conn); out: kfree(isert_conn); - rdma_reject(cma_id, NULL, 0); + rdma_reject(cma_id, NULL, 0, IB_CM_REJ_CONSUMER_DEFINED); return ret; } @@ -571,14 +472,10 @@ isert_connect_release(struct isert_conn *isert_conn) !isert_conn->dev_removed) rdma_destroy_id(isert_conn->cm_id); - if (isert_conn->qp) { - struct isert_comp *comp = isert_conn->qp->recv_cq->cq_context; - - isert_comp_put(comp); - ib_destroy_qp(isert_conn->qp); - } + if (isert_conn->qp) + isert_destroy_qp(isert_conn); - if (isert_conn->login_req_buf) + if (isert_conn->login_desc) isert_free_login_buf(isert_conn); isert_device_put(device); @@ -766,8 +663,8 @@ isert_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) case RDMA_CM_EVENT_ESTABLISHED: isert_connected_handler(cma_id); break; - case RDMA_CM_EVENT_ADDR_CHANGE: /* FALLTHRU */ - case RDMA_CM_EVENT_DISCONNECTED: /* FALLTHRU */ + case RDMA_CM_EVENT_ADDR_CHANGE: + case RDMA_CM_EVENT_DISCONNECTED: case RDMA_CM_EVENT_TIMEWAIT_EXIT: /* FALLTHRU */ ret = isert_disconnected_handler(cma_id, event->event); break; @@ -786,7 +683,7 @@ isert_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) case RDMA_CM_EVENT_REJECTED: isert_info("Connection rejected: %s\n", rdma_reject_msg(cma_id, event->status)); - /* fall through */ + fallthrough; case RDMA_CM_EVENT_UNREACHABLE: case RDMA_CM_EVENT_CONNECT_ERROR: ret = isert_connect_error(cma_id); @@ -964,17 +861,18 @@ isert_login_post_recv(struct isert_conn *isert_conn) int ret; memset(&sge, 0, sizeof(struct ib_sge)); - sge.addr = isert_conn->login_req_dma; + sge.addr = isert_conn->login_desc->dma_addr + + isert_get_hdr_offset(isert_conn->login_desc); sge.length = ISER_RX_PAYLOAD_SIZE; sge.lkey = isert_conn->device->pd->local_dma_lkey; isert_dbg("Setup sge: addr: %llx length: %d 0x%08x\n", sge.addr, sge.length, sge.lkey); - isert_conn->login_req_buf->rx_cqe.done = isert_login_recv_done; + isert_conn->login_desc->rx_cqe.done = isert_login_recv_done; memset(&rx_wr, 0, sizeof(struct ib_recv_wr)); - rx_wr.wr_cqe = &isert_conn->login_req_buf->rx_cqe; + rx_wr.wr_cqe = &isert_conn->login_desc->rx_cqe; rx_wr.sg_list = &sge; rx_wr.num_sge = 1; @@ -1051,7 +949,7 @@ post_send: static void isert_rx_login_req(struct isert_conn *isert_conn) { - struct iser_rx_desc *rx_desc = isert_conn->login_req_buf; + struct iser_rx_desc *rx_desc = isert_conn->login_desc; int rx_buflen = isert_conn->login_req_len; struct iscsi_conn *conn = isert_conn->conn; struct iscsi_login *login = conn->conn_login; @@ -1063,7 +961,7 @@ isert_rx_login_req(struct isert_conn *isert_conn) if (login->first_request) { struct iscsi_login_req *login_req = - (struct iscsi_login_req *)&rx_desc->iscsi_header; + (struct iscsi_login_req *)isert_get_iscsi_hdr(rx_desc); /* * Setup the initial iscsi_login values from the leading * login request PDU. @@ -1082,13 +980,13 @@ isert_rx_login_req(struct isert_conn *isert_conn) login->tsih = be16_to_cpu(login_req->tsih); } - memcpy(&login->req[0], (void *)&rx_desc->iscsi_header, ISCSI_HDR_LEN); + memcpy(&login->req[0], isert_get_iscsi_hdr(rx_desc), ISCSI_HDR_LEN); size = min(rx_buflen, MAX_KEY_VALUE_PAIRS); isert_dbg("Using login payload size: %d, rx_buflen: %d " "MAX_KEY_VALUE_PAIRS: %d\n", size, rx_buflen, MAX_KEY_VALUE_PAIRS); - memcpy(login->req_buf, &rx_desc->data[0], size); + memcpy(login->req_buf, isert_get_data(rx_desc), size); if (login->first_request) { complete(&isert_conn->login_comp); @@ -1153,14 +1051,15 @@ isert_handle_scsi_cmd(struct isert_conn *isert_conn, if (imm_data_len != data_len) { sg_nents = max(1UL, DIV_ROUND_UP(imm_data_len, PAGE_SIZE)); sg_copy_from_buffer(cmd->se_cmd.t_data_sg, sg_nents, - &rx_desc->data[0], imm_data_len); + isert_get_data(rx_desc), imm_data_len); isert_dbg("Copy Immediate sg_nents: %u imm_data_len: %d\n", sg_nents, imm_data_len); } else { sg_init_table(&isert_cmd->sg, 1); cmd->se_cmd.t_data_sg = &isert_cmd->sg; cmd->se_cmd.t_data_nents = 1; - sg_set_buf(&isert_cmd->sg, &rx_desc->data[0], imm_data_len); + sg_set_buf(&isert_cmd->sg, isert_get_data(rx_desc), + imm_data_len); isert_dbg("Transfer Immediate imm_data_len: %d\n", imm_data_len); } @@ -1229,9 +1128,9 @@ isert_handle_iscsi_dataout(struct isert_conn *isert_conn, } isert_dbg("Copying DataOut: sg_start: %p, sg_off: %u " "sg_nents: %u from %p %u\n", sg_start, sg_off, - sg_nents, &rx_desc->data[0], unsol_data_len); + sg_nents, isert_get_data(rx_desc), unsol_data_len); - sg_copy_from_buffer(sg_start, sg_nents, &rx_desc->data[0], + sg_copy_from_buffer(sg_start, sg_nents, isert_get_data(rx_desc), unsol_data_len); rc = iscsit_check_dataout_payload(cmd, hdr, false); @@ -1290,7 +1189,7 @@ isert_handle_text_cmd(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd } cmd->text_in_ptr = text_in; - memcpy(cmd->text_in_ptr, &rx_desc->data[0], payload_length); + memcpy(cmd->text_in_ptr, isert_get_data(rx_desc), payload_length); return iscsit_process_text_cmd(conn, cmd, hdr); } @@ -1300,7 +1199,7 @@ isert_rx_opcode(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc, uint32_t read_stag, uint64_t read_va, uint32_t write_stag, uint64_t write_va) { - struct iscsi_hdr *hdr = &rx_desc->iscsi_header; + struct iscsi_hdr *hdr = isert_get_iscsi_hdr(rx_desc); struct iscsi_conn *conn = isert_conn->conn; struct iscsi_cmd *cmd; struct isert_cmd *isert_cmd; @@ -1398,8 +1297,8 @@ isert_recv_done(struct ib_cq *cq, struct ib_wc *wc) struct isert_conn *isert_conn = wc->qp->qp_context; struct ib_device *ib_dev = isert_conn->cm_id->device; struct iser_rx_desc *rx_desc = cqe_to_rx_desc(wc->wr_cqe); - struct iscsi_hdr *hdr = &rx_desc->iscsi_header; - struct iser_ctrl *iser_ctrl = &rx_desc->iser_header; + struct iscsi_hdr *hdr = isert_get_iscsi_hdr(rx_desc); + struct iser_ctrl *iser_ctrl = isert_get_iser_hdr(rx_desc); uint64_t read_va = 0, write_va = 0; uint32_t read_stag = 0, write_stag = 0; @@ -1413,7 +1312,7 @@ isert_recv_done(struct ib_cq *cq, struct ib_wc *wc) rx_desc->in_use = true; ib_dma_sync_single_for_cpu(ib_dev, rx_desc->dma_addr, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + ISER_RX_SIZE, DMA_FROM_DEVICE); isert_dbg("DMA: 0x%llx, iSCSI opcode: 0x%02x, ITT: 0x%08x, flags: 0x%02x dlen: %d\n", rx_desc->dma_addr, hdr->opcode, hdr->itt, hdr->flags, @@ -1448,7 +1347,7 @@ isert_recv_done(struct ib_cq *cq, struct ib_wc *wc) read_stag, read_va, write_stag, write_va); ib_dma_sync_single_for_device(ib_dev, rx_desc->dma_addr, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + ISER_RX_SIZE, DMA_FROM_DEVICE); } static void @@ -1462,8 +1361,8 @@ isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc) return; } - ib_dma_sync_single_for_cpu(ib_dev, isert_conn->login_req_dma, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + ib_dma_sync_single_for_cpu(ib_dev, isert_conn->login_desc->dma_addr, + ISER_RX_SIZE, DMA_FROM_DEVICE); isert_conn->login_req_len = wc->byte_len - ISER_HEADERS_LEN; @@ -1478,8 +1377,8 @@ isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc) complete(&isert_conn->login_req_comp); mutex_unlock(&isert_conn->mutex); - ib_dma_sync_single_for_device(ib_dev, isert_conn->login_req_dma, - ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE); + ib_dma_sync_single_for_device(ib_dev, isert_conn->login_desc->dma_addr, + ISER_RX_SIZE, DMA_FROM_DEVICE); } static void @@ -1572,7 +1471,7 @@ isert_put_cmd(struct isert_cmd *isert_cmd, bool comp_err) transport_generic_free_cmd(&cmd->se_cmd, 0); break; } - /* fall through */ + fallthrough; default: iscsit_release_cmd(cmd); break; @@ -1750,7 +1649,7 @@ isert_do_control_comp(struct work_struct *work) switch (cmd->i_state) { case ISTATE_SEND_TASKMGTRSP: iscsit_tmr_post_handler(cmd, cmd->conn); - /* fall through */ + fallthrough; case ISTATE_SEND_REJECT: case ISTATE_SEND_TEXTRSP: cmd->i_state = ISTATE_SENT_STATUS; diff --git a/drivers/infiniband/ulp/isert/ib_isert.h b/drivers/infiniband/ulp/isert/ib_isert.h index 3b296bac4f60..7fee4a65e181 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.h +++ b/drivers/infiniband/ulp/isert/ib_isert.h @@ -59,11 +59,14 @@ ISERT_MAX_TX_MISC_PDUS + \ ISERT_MAX_RX_MISC_PDUS) -#define ISER_RX_PAD_SIZE (ISCSI_DEF_MAX_RECV_SEG_LEN + 4096 - \ - (ISER_RX_PAYLOAD_SIZE + sizeof(u64) + sizeof(struct ib_sge) + \ - sizeof(struct ib_cqe) + sizeof(bool))) +/* + * RX size is default of 8k plus headers, but data needs to align to + * 512 boundary, so use 1024 to have the extra space for alignment. + */ +#define ISER_RX_SIZE (ISCSI_DEF_MAX_RECV_SEG_LEN + 1024) -#define ISCSI_ISER_SG_TABLESIZE 256 +/* Maximum support is 16MB I/O size */ +#define ISCSI_ISER_MAX_SG_TABLESIZE 4096 enum isert_desc_type { ISCSI_TX_CONTROL, @@ -80,21 +83,41 @@ enum iser_conn_state { }; struct iser_rx_desc { - struct iser_ctrl iser_header; - struct iscsi_hdr iscsi_header; - char data[ISCSI_DEF_MAX_RECV_SEG_LEN]; + char buf[ISER_RX_SIZE]; u64 dma_addr; struct ib_sge rx_sg; struct ib_cqe rx_cqe; bool in_use; - char pad[ISER_RX_PAD_SIZE]; -} __packed; +}; static inline struct iser_rx_desc *cqe_to_rx_desc(struct ib_cqe *cqe) { return container_of(cqe, struct iser_rx_desc, rx_cqe); } +static void *isert_get_iser_hdr(struct iser_rx_desc *desc) +{ + return PTR_ALIGN(desc->buf + ISER_HEADERS_LEN, 512) - ISER_HEADERS_LEN; +} + +static size_t isert_get_hdr_offset(struct iser_rx_desc *desc) +{ + return isert_get_iser_hdr(desc) - (void *)desc->buf; +} + +static void *isert_get_iscsi_hdr(struct iser_rx_desc *desc) +{ + return isert_get_iser_hdr(desc) + sizeof(struct iser_ctrl); +} + +static void *isert_get_data(struct iser_rx_desc *desc) +{ + void *data = isert_get_iser_hdr(desc) + ISER_HEADERS_LEN; + + WARN_ON((uintptr_t)data & 511); + return data; +} + struct iser_tx_desc { struct iser_ctrl iser_header; struct iscsi_hdr iscsi_header; @@ -141,9 +164,8 @@ struct isert_conn { u32 responder_resources; u32 initiator_depth; bool pi_support; - struct iser_rx_desc *login_req_buf; + struct iser_rx_desc *login_desc; char *login_rsp_buf; - u64 login_req_dma; int login_req_len; u64 login_rsp_dma; struct iser_rx_desc *rx_descs; @@ -155,6 +177,8 @@ struct isert_conn { struct iser_tx_desc login_tx_desc; struct rdma_cm_id *cm_id; struct ib_qp *qp; + struct ib_cq *cq; + u32 cq_size; struct isert_device *device; struct mutex mutex; struct kref kref; @@ -165,22 +189,6 @@ struct isert_conn { bool dev_removed; }; -#define ISERT_MAX_CQ 64 - -/** - * struct isert_comp - iSER completion context - * - * @device: pointer to device handle - * @cq: completion queue - * @active_qps: Number of active QPs attached - * to completion context - */ -struct isert_comp { - struct isert_device *device; - struct ib_cq *cq; - int active_qps; -}; - struct isert_device { bool pi_capable; int refcount; diff --git a/drivers/infiniband/ulp/opa_vnic/Kconfig b/drivers/infiniband/ulp/opa_vnic/Kconfig index a1f266b9c0b2..e84248587187 100644 --- a/drivers/infiniband/ulp/opa_vnic/Kconfig +++ b/drivers/infiniband/ulp/opa_vnic/Kconfig @@ -2,7 +2,7 @@ config INFINIBAND_OPA_VNIC tristate "Intel OPA VNIC support" depends on X86_64 && INFINIBAND - ---help--- + help This is Omni-Path (OPA) Virtual Network Interface Controller (VNIC) driver for Ethernet over Omni-Path feature. It implements the HW independent VNIC functionality. It interfaces with Linux stack for diff --git a/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h b/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h index d324312a373c..f64519872297 100644 --- a/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h +++ b/drivers/infiniband/ulp/opa_vnic/opa_vnic_encap.h @@ -118,12 +118,17 @@ * struct opa_vesw_info - OPA vnic switch information * @fabric_id: 10-bit fabric id * @vesw_id: 12-bit virtual ethernet switch id + * @rsvd0: reserved bytes * @def_port_mask: bitmask of default ports + * @rsvd1: reserved bytes * @pkey: partition key + * @rsvd2: reserved bytes * @u_mcast_dlid: unknown multicast dlid * @u_ucast_dlid: array of unknown unicast dlids + * @rsvd3: reserved bytes * @rc: routing control * @eth_mtu: Ethernet MTU + * @rsvd4: reserved bytes */ struct opa_vesw_info { __be16 fabric_id; @@ -150,12 +155,14 @@ struct opa_vesw_info { * struct opa_per_veswport_info - OPA vnic per port information * @port_num: port number * @eth_link_status: current ethernet link state + * @rsvd0: reserved bytes * @base_mac_addr: base mac address * @config_state: configured port state * @oper_state: operational port state * @max_mac_tbl_ent: max number of mac table entries * @max_smac_ent: max smac entries in mac table * @mac_tbl_digest: mac table digest + * @rsvd1: reserved bytes * @encap_slid: base slid for the port * @pcp_to_sc_uc: sc by pcp index for unicast ethernet packets * @pcp_to_vl_uc: vl by pcp index for unicast ethernet packets @@ -165,8 +172,10 @@ struct opa_vesw_info { * @non_vlan_vl_uc: vl for non-vlan unicast ethernet packets * @non_vlan_sc_mc: sc for non-vlan multicast ethernet packets * @non_vlan_vl_mc: vl for non-vlan multicast ethernet packets + * @rsvd2: reserved bytes * @uc_macs_gen_count: generation count for unicast macs list * @mc_macs_gen_count: generation count for multicast macs list + * @rsvd3: reserved bytes */ struct opa_per_veswport_info { __be32 port_num; @@ -294,6 +303,7 @@ struct opa_veswport_mactable { * @rx_512_1023: received packet length is >=512 and < 1023 bytes * @rx_1024_1518: received packet length is >=1024 and < 1518 bytes * @rx_1519_max: received packet length >= 1519 bytes + * @reserved: reserved bytes * * All the above are counters of corresponding conditions. */ @@ -347,16 +357,26 @@ struct opa_veswport_summary_counters { * @veswport_num: virtual ethernet switch port number * @tx_errors: transmit errors * @rx_errors: receive errors + * @rsvd0: reserved bytes * @tx_smac_filt: smac filter errors + * @rsvd1: reserved bytes + * @rsvd2: reserved bytes + * @rsvd3: reserved bytes * @tx_dlid_zero: transmit packets with invalid dlid + * @rsvd4: reserved bytes * @tx_logic: other transmit errors + * @rsvd5: reserved bytes * @tx_drop_state: packet tansmission in non-forward port state * @rx_bad_veswid: received packet with invalid vesw id + * @rsvd6: reserved bytes * @rx_runt: received ethernet packet with length < 64 bytes * @rx_oversize: received ethernet packet with length > MTU size + * @rsvd7: reserved bytes * @rx_eth_down: received packets when interface is down * @rx_drop_state: received packets in non-forwarding port state * @rx_logic: other receive errors + * @rsvd8: reserved bytes + * @rsvd9: reserved bytes * * All the above are counters of corresponding error conditions. */ @@ -447,6 +467,7 @@ struct opa_veswport_iface_macs { * struct opa_vnic_vema_mad - Generic VEMA MAD * @mad_hdr: Generic MAD header * @rmpp_hdr: RMPP header for vendor specific MADs + * @reserved: reserved bytes * @oui: Unique org identifier * @data: MAD data */ @@ -467,6 +488,7 @@ struct opa_vnic_vema_mad { * @trap_num: Trap number * @toggle_count: Notice toggle bit and count value * @issuer_lid: Trap issuer's lid + * @reserved: reserved bytes * @issuer_gid: Issuer GID (only if Report method) * @raw_data: Trap message body */ @@ -487,6 +509,7 @@ struct opa_vnic_notice_attr { * struct opa_vnic_vema_mad_trap - Generic VEMA MAD Trap * @mad_hdr: Generic MAD header * @rmpp_hdr: RMPP header for vendor specific MADs + * @reserved: reserved bytes * @oui: Unique org identifier * @notice: Notice structure */ diff --git a/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c b/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c index 6e8d650c17c7..4933085a864a 100644 --- a/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c +++ b/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c @@ -113,7 +113,7 @@ struct opa_vnic_vema_port { struct mutex lock; }; -static void opa_vnic_vema_add_one(struct ib_device *device); +static int opa_vnic_vema_add_one(struct ib_device *device); static void opa_vnic_vema_rem_one(struct ib_device *device, void *client_data); @@ -547,7 +547,6 @@ static void vema_get(struct opa_vnic_vema_port *port, vema_get_mac_entries(port, recvd_mad, rsp_mad); break; case OPA_EM_ATTR_IFACE_UCAST_MACS: - /* fall through */ case OPA_EM_ATTR_IFACE_MCAST_MACS: vema_get_mac_list(port, recvd_mad, rsp_mad, attr_id); break; @@ -989,18 +988,18 @@ static void opa_vnic_ctrl_config_dev(struct opa_vnic_ctrl_port *cport, bool en) * * Allocate the vnic control port and initialize it. */ -static void opa_vnic_vema_add_one(struct ib_device *device) +static int opa_vnic_vema_add_one(struct ib_device *device) { struct opa_vnic_ctrl_port *cport; int rc, size = sizeof(*cport); if (!rdma_cap_opa_vnic(device)) - return; + return -EOPNOTSUPP; size += device->phys_port_cnt * sizeof(struct opa_vnic_vema_port); cport = kzalloc(size, GFP_KERNEL); if (!cport) - return; + return -ENOMEM; cport->num_ports = device->phys_port_cnt; cport->ibdev = device; @@ -1012,6 +1011,7 @@ static void opa_vnic_vema_add_one(struct ib_device *device) ib_set_client_data(device, &opa_vnic_client, cport); opa_vnic_ctrl_config_dev(cport, true); + return 0; } /** @@ -1026,9 +1026,6 @@ static void opa_vnic_vema_rem_one(struct ib_device *device, { struct opa_vnic_ctrl_port *cport = client_data; - if (!cport) - return; - c_info("removing VNIC client\n"); opa_vnic_ctrl_config_dev(cport, false); vema_unregister(cport); diff --git a/drivers/infiniband/ulp/rtrs/Kconfig b/drivers/infiniband/ulp/rtrs/Kconfig new file mode 100644 index 000000000000..9092b62e6dc8 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/Kconfig @@ -0,0 +1,27 @@ +# SPDX-License-Identifier: GPL-2.0-or-later + +config INFINIBAND_RTRS + tristate + depends on INFINIBAND_ADDR_TRANS + +config INFINIBAND_RTRS_CLIENT + tristate "RTRS client module" + depends on INFINIBAND_ADDR_TRANS + select INFINIBAND_RTRS + help + RDMA transport client module. + + RDMA Transport (RTRS) client implements a reliable transport layer + and also multipathing functionality and that it is intended to be + the base layer for a block storage initiator over RDMA. + +config INFINIBAND_RTRS_SERVER + tristate "RTRS server module" + depends on INFINIBAND_ADDR_TRANS + select INFINIBAND_RTRS + help + RDMA transport server module. + + RDMA Transport (RTRS) server module processing connection and IO + requests received from the RTRS client module, it will pass the + IO requests to its user eg. RNBD_server. diff --git a/drivers/infiniband/ulp/rtrs/Makefile b/drivers/infiniband/ulp/rtrs/Makefile new file mode 100644 index 000000000000..3898509be270 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/Makefile @@ -0,0 +1,15 @@ +# SPDX-License-Identifier: GPL-2.0-or-later + +rtrs-client-y := rtrs-clt.o \ + rtrs-clt-stats.o \ + rtrs-clt-sysfs.o + +rtrs-server-y := rtrs-srv.o \ + rtrs-srv-stats.o \ + rtrs-srv-sysfs.o + +rtrs-core-y := rtrs.o + +obj-$(CONFIG_INFINIBAND_RTRS) += rtrs-core.o +obj-$(CONFIG_INFINIBAND_RTRS_CLIENT) += rtrs-client.o +obj-$(CONFIG_INFINIBAND_RTRS_SERVER) += rtrs-server.o diff --git a/drivers/infiniband/ulp/rtrs/README b/drivers/infiniband/ulp/rtrs/README new file mode 100644 index 000000000000..5d9ea142e5dd --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/README @@ -0,0 +1,213 @@ +**************************** +RDMA Transport (RTRS) +**************************** + +RTRS (RDMA Transport) is a reliable high speed transport library +which provides support to establish optimal number of connections +between client and server machines using RDMA (InfiniBand, RoCE, iWarp) +transport. It is optimized to transfer (read/write) IO blocks. + +In its core interface it follows the BIO semantics of providing the +possibility to either write data from an sg list to the remote side +or to request ("read") data transfer from the remote side into a given +sg list. + +RTRS provides I/O fail-over and load-balancing capabilities by using +multipath I/O (see "add_path" and "mp_policy" configuration entries in +Documentation/ABI/testing/sysfs-class-rtrs-client). + +RTRS is used by the RNBD (RDMA Network Block Device) modules. + +================== +Transport protocol +================== + +Overview +-------- +An established connection between a client and a server is called rtrs +session. A session is associated with a set of memory chunks reserved on the +server side for a given client for rdma transfer. A session +consists of multiple paths, each representing a separate physical link +between client and server. Those are used for load balancing and failover. +Each path consists of as many connections (QPs) as there are cpus on +the client. + +When processing an incoming write or read request, rtrs client uses memory +chunks reserved for him on the server side. Their number, size and addresses +need to be exchanged between client and server during the connection +establishment phase. Apart from the memory related information client needs to +inform the server about the session name and identify each path and connection +individually. + +On an established session client sends to server write or read messages. +Server uses immediate field to tell the client which request is being +acknowledged and for errno. Client uses immediate field to tell the server +which of the memory chunks has been accessed and at which offset the message +can be found. + +Module parameter always_invalidate is introduced for the security problem +discussed in LPC RDMA MC 2019. When always_invalidate=Y, on the server side we +invalidate each rdma buffer before we hand it over to RNBD server and +then pass it to the block layer. A new rkey is generated and registered for the +buffer after it returns back from the block layer and RNBD server. +The new rkey is sent back to the client along with the IO result. +The procedure is the default behaviour of the driver. This invalidation and +registration on each IO causes performance drop of up to 20%. A user of the +driver may choose to load the modules with this mechanism switched off +(always_invalidate=N), if he understands and can take the risk of a malicious +client being able to corrupt memory of a server it is connected to. This might +be a reasonable option in a scenario where all the clients and all the servers +are located within a secure datacenter. + + +Connection establishment +------------------------ + +1. Client starts establishing connections belonging to a path of a session one +by one via attaching RTRS_MSG_CON_REQ messages to the rdma_connect requests. +Those include uuid of the session and uuid of the path to be +established. They are used by the server to find a persisting session/path or +to create a new one when necessary. The message also contains the protocol +version and magic for compatibility, total number of connections per session +(as many as cpus on the client), the id of the current connection and +the reconnect counter, which is used to resolve the situations where +client is trying to reconnect a path, while server is still destroying the old +one. + +2. Server accepts the connection requests one by one and attaches +RTRS_MSG_CONN_RSP messages to the rdma_accept. Apart from magic and +protocol version, the messages include error code, queue depth supported by +the server (number of memory chunks which are going to be allocated for that +session) and the maximum size of one io, RTRS_MSG_NEW_RKEY_F flags is set +when always_invalidate=Y. + +3. After all connections of a path are established client sends to server the +RTRS_MSG_INFO_REQ message, containing the name of the session. This message +requests the address information from the server. + +4. Server replies to the session info request message with RTRS_MSG_INFO_RSP, +which contains the addresses and keys of the RDMA buffers allocated for that +session. + +5. Session becomes connected after all paths to be established are connected +(i.e. steps 1-4 finished for all paths requested for a session) + +6. Server and client exchange periodically heartbeat messages (empty rdma +messages with an immediate field) which are used to detect a crash on remote +side or network outage in an absence of IO. + +7. On any RDMA related error or in the case of a heartbeat timeout, the +corresponding path is disconnected, all the inflight IO are failed over to a +healthy path, if any, and the reconnect mechanism is triggered. + +CLT SRV +*for each connection belonging to a path and for each path: +RTRS_MSG_CON_REQ -------------------> + <------------------- RTRS_MSG_CON_RSP +... +*after all connections are established: +RTRS_MSG_INFO_REQ -------------------> + <------------------- RTRS_MSG_INFO_RSP +*heartbeat is started from both sides: + -------------------> [RTRS_HB_MSG_IMM] +[RTRS_HB_MSG_ACK] <------------------- +[RTRS_HB_MSG_IMM] <------------------- + -------------------> [RTRS_HB_MSG_ACK] + +IO path +------- + +* Write (always_invalidate=N) * + +1. When processing a write request client selects one of the memory chunks +on the server side and rdma writes there the user data, user header and the +RTRS_MSG_RDMA_WRITE message. Apart from the type (write), the message only +contains size of the user header. The client tells the server which chunk has +been accessed and at what offset the RTRS_MSG_RDMA_WRITE can be found by +using the IMM field. + +2. When confirming a write request server sends an "empty" rdma message with +an immediate field. The 32 bit field is used to specify the outstanding +inflight IO and for the error code. + +CLT SRV +usr_data + usr_hdr + rtrs_msg_rdma_write -----------------> [RTRS_IO_REQ_IMM] +[RTRS_IO_RSP_IMM] <----------------- (id + errno) + +* Write (always_invalidate=Y) * + +1. When processing a write request client selects one of the memory chunks +on the server side and rdma writes there the user data, user header and the +RTRS_MSG_RDMA_WRITE message. Apart from the type (write), the message only +contains size of the user header. The client tells the server which chunk has +been accessed and at what offset the RTRS_MSG_RDMA_WRITE can be found by +using the IMM field, Server invalidate rkey associated to the memory chunks +first, when it finishes, pass the IO to RNBD server module. + +2. When confirming a write request server sends an "empty" rdma message with +an immediate field. The 32 bit field is used to specify the outstanding +inflight IO and for the error code. The new rkey is sent back using +SEND_WITH_IMM WR, client When it recived new rkey message, it validates +the message and finished IO after update rkey for the rbuffer, then post +back the recv buffer for later use. + +CLT SRV +usr_data + usr_hdr + rtrs_msg_rdma_write -----------------> [RTRS_IO_REQ_IMM] +[RTRS_MSG_RKEY_RSP] <----------------- (RTRS_MSG_RKEY_RSP) +[RTRS_IO_RSP_IMM] <----------------- (id + errno) + + +* Read (always_invalidate=N)* + +1. When processing a read request client selects one of the memory chunks +on the server side and rdma writes there the user header and the +RTRS_MSG_RDMA_READ message. This message contains the type (read), size of +the user header, flags (specifying if memory invalidation is necessary) and the +list of addresses along with keys for the data to be read into. + +2. When confirming a read request server transfers the requested data first, +attaches an invalidation message if requested and finally an "empty" rdma +message with an immediate field. The 32 bit field is used to specify the +outstanding inflight IO and the error code. + +CLT SRV +usr_hdr + rtrs_msg_rdma_read --------------> [RTRS_IO_REQ_IMM] +[RTRS_IO_RSP_IMM] <-------------- usr_data + (id + errno) +or in case client requested invalidation: +[RTRS_IO_RSP_IMM_W_INV] <-------------- usr_data + (INV) + (id + errno) + +* Read (always_invalidate=Y)* + +1. When processing a read request client selects one of the memory chunks +on the server side and rdma writes there the user header and the +RTRS_MSG_RDMA_READ message. This message contains the type (read), size of +the user header, flags (specifying if memory invalidation is necessary) and the +list of addresses along with keys for the data to be read into. +Server invalidate rkey associated to the memory chunks first, when it finishes, +passes the IO to RNBD server module. + +2. When confirming a read request server transfers the requested data first, +attaches an invalidation message if requested and finally an "empty" rdma +message with an immediate field. The 32 bit field is used to specify the +outstanding inflight IO and the error code. The new rkey is sent back using +SEND_WITH_IMM WR, client When it recived new rkey message, it validates +the message and finished IO after update rkey for the rbuffer, then post +back the recv buffer for later use. + +CLT SRV +usr_hdr + rtrs_msg_rdma_read --------------> [RTRS_IO_REQ_IMM] +[RTRS_IO_RSP_IMM] <-------------- usr_data + (id + errno) +[RTRS_MSG_RKEY_RSP] <----------------- (RTRS_MSG_RKEY_RSP) +or in case client requested invalidation: +[RTRS_IO_RSP_IMM_W_INV] <-------------- usr_data + (INV) + (id + errno) +========================================= +Contributors List(in alphabetical order) +========================================= +Danil Kipnis <danil.kipnis@profitbricks.com> +Fabian Holler <mail@fholler.de> +Guoqing Jiang <guoqing.jiang@cloud.ionos.com> +Jack Wang <jinpu.wang@profitbricks.com> +Kleber Souza <kleber.souza@profitbricks.com> +Lutz Pogrell <lutz.pogrell@cloud.ionos.com> +Milind Dumbare <Milind.dumbare@gmail.com> +Roman Penyaev <roman.penyaev@profitbricks.com> diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c b/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c new file mode 100644 index 000000000000..26bbe5d6dff5 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-stats.c @@ -0,0 +1,200 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include "rtrs-clt.h" + +void rtrs_clt_update_wc_stats(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_stats *stats = sess->stats; + struct rtrs_clt_stats_pcpu *s; + int cpu; + + cpu = raw_smp_processor_id(); + s = this_cpu_ptr(stats->pcpu_stats); + if (unlikely(con->cpu != cpu)) { + s->cpu_migr.to++; + + /* Careful here, override s pointer */ + s = per_cpu_ptr(stats->pcpu_stats, con->cpu); + atomic_inc(&s->cpu_migr.from); + } +} + +void rtrs_clt_inc_failover_cnt(struct rtrs_clt_stats *stats) +{ + struct rtrs_clt_stats_pcpu *s; + + s = this_cpu_ptr(stats->pcpu_stats); + s->rdma.failover_cnt++; +} + +int rtrs_clt_stats_migration_cnt_to_str(struct rtrs_clt_stats *stats, + char *buf, size_t len) +{ + struct rtrs_clt_stats_pcpu *s; + + size_t used; + int cpu; + + used = scnprintf(buf, len, " "); + for_each_possible_cpu(cpu) + used += scnprintf(buf + used, len - used, " CPU%u", cpu); + + used += scnprintf(buf + used, len - used, "\nfrom:"); + for_each_possible_cpu(cpu) { + s = per_cpu_ptr(stats->pcpu_stats, cpu); + used += scnprintf(buf + used, len - used, " %d", + atomic_read(&s->cpu_migr.from)); + } + + used += scnprintf(buf + used, len - used, "\nto :"); + for_each_possible_cpu(cpu) { + s = per_cpu_ptr(stats->pcpu_stats, cpu); + used += scnprintf(buf + used, len - used, " %d", + s->cpu_migr.to); + } + used += scnprintf(buf + used, len - used, "\n"); + + return used; +} + +int rtrs_clt_stats_reconnects_to_str(struct rtrs_clt_stats *stats, char *buf, + size_t len) +{ + return scnprintf(buf, len, "%d %d\n", + stats->reconnects.successful_cnt, + stats->reconnects.fail_cnt); +} + +ssize_t rtrs_clt_stats_rdma_to_str(struct rtrs_clt_stats *stats, + char *page, size_t len) +{ + struct rtrs_clt_stats_rdma sum; + struct rtrs_clt_stats_rdma *r; + int cpu; + + memset(&sum, 0, sizeof(sum)); + + for_each_possible_cpu(cpu) { + r = &per_cpu_ptr(stats->pcpu_stats, cpu)->rdma; + + sum.dir[READ].cnt += r->dir[READ].cnt; + sum.dir[READ].size_total += r->dir[READ].size_total; + sum.dir[WRITE].cnt += r->dir[WRITE].cnt; + sum.dir[WRITE].size_total += r->dir[WRITE].size_total; + sum.failover_cnt += r->failover_cnt; + } + + return scnprintf(page, len, "%llu %llu %llu %llu %u %llu\n", + sum.dir[READ].cnt, sum.dir[READ].size_total, + sum.dir[WRITE].cnt, sum.dir[WRITE].size_total, + atomic_read(&stats->inflight), sum.failover_cnt); +} + +ssize_t rtrs_clt_reset_all_help(struct rtrs_clt_stats *s, + char *page, size_t len) +{ + return scnprintf(page, len, "echo 1 to reset all statistics\n"); +} + +int rtrs_clt_reset_rdma_stats(struct rtrs_clt_stats *stats, bool enable) +{ + struct rtrs_clt_stats_pcpu *s; + int cpu; + + if (!enable) + return -EINVAL; + + for_each_possible_cpu(cpu) { + s = per_cpu_ptr(stats->pcpu_stats, cpu); + memset(&s->rdma, 0, sizeof(s->rdma)); + } + + return 0; +} + +int rtrs_clt_reset_cpu_migr_stats(struct rtrs_clt_stats *stats, bool enable) +{ + struct rtrs_clt_stats_pcpu *s; + int cpu; + + if (!enable) + return -EINVAL; + + for_each_possible_cpu(cpu) { + s = per_cpu_ptr(stats->pcpu_stats, cpu); + memset(&s->cpu_migr, 0, sizeof(s->cpu_migr)); + } + + return 0; +} + +int rtrs_clt_reset_reconnects_stat(struct rtrs_clt_stats *stats, bool enable) +{ + if (!enable) + return -EINVAL; + + memset(&stats->reconnects, 0, sizeof(stats->reconnects)); + + return 0; +} + +int rtrs_clt_reset_all_stats(struct rtrs_clt_stats *s, bool enable) +{ + if (enable) { + rtrs_clt_reset_rdma_stats(s, enable); + rtrs_clt_reset_cpu_migr_stats(s, enable); + rtrs_clt_reset_reconnects_stat(s, enable); + atomic_set(&s->inflight, 0); + return 0; + } + + return -EINVAL; +} + +static inline void rtrs_clt_update_rdma_stats(struct rtrs_clt_stats *stats, + size_t size, int d) +{ + struct rtrs_clt_stats_pcpu *s; + + s = this_cpu_ptr(stats->pcpu_stats); + s->rdma.dir[d].cnt++; + s->rdma.dir[d].size_total += size; +} + +void rtrs_clt_update_all_stats(struct rtrs_clt_io_req *req, int dir) +{ + struct rtrs_clt_con *con = req->con; + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt_stats *stats = sess->stats; + unsigned int len; + + len = req->usr_len + req->data_len; + rtrs_clt_update_rdma_stats(stats, len, dir); + if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT) + atomic_inc(&stats->inflight); +} + +int rtrs_clt_init_stats(struct rtrs_clt_stats *stats) +{ + stats->pcpu_stats = alloc_percpu(typeof(*stats->pcpu_stats)); + if (!stats->pcpu_stats) + return -ENOMEM; + + /* + * successful_cnt will be set to 0 after session + * is established for the first time + */ + stats->reconnects.successful_cnt = -1; + + return 0; +} diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c b/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c new file mode 100644 index 000000000000..298b747d0330 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c @@ -0,0 +1,483 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include "rtrs-pri.h" +#include "rtrs-clt.h" +#include "rtrs-log.h" + +#define MIN_MAX_RECONN_ATT -1 +#define MAX_MAX_RECONN_ATT 9999 + +static void rtrs_clt_sess_release(struct kobject *kobj) +{ + struct rtrs_clt_sess *sess; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + + free_sess(sess); +} + +static struct kobj_type ktype_sess = { + .sysfs_ops = &kobj_sysfs_ops, + .release = rtrs_clt_sess_release +}; + +static void rtrs_clt_sess_stats_release(struct kobject *kobj) +{ + struct rtrs_clt_stats *stats; + + stats = container_of(kobj, struct rtrs_clt_stats, kobj_stats); + + free_percpu(stats->pcpu_stats); + + kfree(stats); +} + +static struct kobj_type ktype_stats = { + .sysfs_ops = &kobj_sysfs_ops, + .release = rtrs_clt_sess_stats_release, +}; + +static ssize_t max_reconnect_attempts_show(struct device *dev, + struct device_attribute *attr, + char *page) +{ + struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev); + + return sprintf(page, "%d\n", rtrs_clt_get_max_reconnect_attempts(clt)); +} + +static ssize_t max_reconnect_attempts_store(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count) +{ + int value; + int ret; + struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev); + + ret = kstrtoint(buf, 10, &value); + if (ret) { + rtrs_err(clt, "%s: failed to convert string '%s' to int\n", + attr->attr.name, buf); + return ret; + } + if (value > MAX_MAX_RECONN_ATT || + value < MIN_MAX_RECONN_ATT) { + rtrs_err(clt, + "%s: invalid range (provided: '%s', accepted: min: %d, max: %d)\n", + attr->attr.name, buf, MIN_MAX_RECONN_ATT, + MAX_MAX_RECONN_ATT); + return -EINVAL; + } + rtrs_clt_set_max_reconnect_attempts(clt, value); + + return count; +} + +static DEVICE_ATTR_RW(max_reconnect_attempts); + +static ssize_t mpath_policy_show(struct device *dev, + struct device_attribute *attr, + char *page) +{ + struct rtrs_clt *clt; + + clt = container_of(dev, struct rtrs_clt, dev); + + switch (clt->mp_policy) { + case MP_POLICY_RR: + return sprintf(page, "round-robin (RR: %d)\n", clt->mp_policy); + case MP_POLICY_MIN_INFLIGHT: + return sprintf(page, "min-inflight (MI: %d)\n", clt->mp_policy); + default: + return sprintf(page, "Unknown (%d)\n", clt->mp_policy); + } +} + +static ssize_t mpath_policy_store(struct device *dev, + struct device_attribute *attr, + const char *buf, + size_t count) +{ + struct rtrs_clt *clt; + int value; + int ret; + + clt = container_of(dev, struct rtrs_clt, dev); + + ret = kstrtoint(buf, 10, &value); + if (!ret && (value == MP_POLICY_RR || + value == MP_POLICY_MIN_INFLIGHT)) { + clt->mp_policy = value; + return count; + } + + if (!strncasecmp(buf, "round-robin", 11) || + !strncasecmp(buf, "rr", 2)) + clt->mp_policy = MP_POLICY_RR; + else if (!strncasecmp(buf, "min-inflight", 12) || + !strncasecmp(buf, "mi", 2)) + clt->mp_policy = MP_POLICY_MIN_INFLIGHT; + else + return -EINVAL; + + return count; +} + +static DEVICE_ATTR_RW(mpath_policy); + +static ssize_t add_path_show(struct device *dev, + struct device_attribute *attr, char *page) +{ + return scnprintf(page, PAGE_SIZE, + "Usage: echo [<source addr>@]<destination addr> > %s\n\n*addr ::= [ ip:<ipv4|ipv6> | gid:<gid> ]\n", + attr->attr.name); +} + +static ssize_t add_path_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct sockaddr_storage srcaddr, dstaddr; + struct rtrs_addr addr = { + .src = &srcaddr, + .dst = &dstaddr + }; + struct rtrs_clt *clt; + const char *nl; + size_t len; + int err; + + clt = container_of(dev, struct rtrs_clt, dev); + + nl = strchr(buf, '\n'); + if (nl) + len = nl - buf; + else + len = count; + err = rtrs_addr_to_sockaddr(buf, len, clt->port, &addr); + if (err) + return -EINVAL; + + err = rtrs_clt_create_path_from_sysfs(clt, &addr); + if (err) + return err; + + return count; +} + +static DEVICE_ATTR_RW(add_path); + +static ssize_t rtrs_clt_state_show(struct kobject *kobj, + struct kobj_attribute *attr, char *page) +{ + struct rtrs_clt_sess *sess; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + if (sess->state == RTRS_CLT_CONNECTED) + return sprintf(page, "connected\n"); + + return sprintf(page, "disconnected\n"); +} + +static struct kobj_attribute rtrs_clt_state_attr = + __ATTR(state, 0444, rtrs_clt_state_show, NULL); + +static ssize_t rtrs_clt_reconnect_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + return scnprintf(page, PAGE_SIZE, "Usage: echo 1 > %s\n", + attr->attr.name); +} + +static ssize_t rtrs_clt_reconnect_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t count) +{ + struct rtrs_clt_sess *sess; + int ret; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + if (!sysfs_streq(buf, "1")) { + rtrs_err(sess->clt, "%s: unknown value: '%s'\n", + attr->attr.name, buf); + return -EINVAL; + } + ret = rtrs_clt_reconnect_from_sysfs(sess); + if (ret) + return ret; + + return count; +} + +static struct kobj_attribute rtrs_clt_reconnect_attr = + __ATTR(reconnect, 0644, rtrs_clt_reconnect_show, + rtrs_clt_reconnect_store); + +static ssize_t rtrs_clt_disconnect_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + return scnprintf(page, PAGE_SIZE, "Usage: echo 1 > %s\n", + attr->attr.name); +} + +static ssize_t rtrs_clt_disconnect_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t count) +{ + struct rtrs_clt_sess *sess; + int ret; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + if (!sysfs_streq(buf, "1")) { + rtrs_err(sess->clt, "%s: unknown value: '%s'\n", + attr->attr.name, buf); + return -EINVAL; + } + ret = rtrs_clt_disconnect_from_sysfs(sess); + if (ret) + return ret; + + return count; +} + +static struct kobj_attribute rtrs_clt_disconnect_attr = + __ATTR(disconnect, 0644, rtrs_clt_disconnect_show, + rtrs_clt_disconnect_store); + +static ssize_t rtrs_clt_remove_path_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + return scnprintf(page, PAGE_SIZE, "Usage: echo 1 > %s\n", + attr->attr.name); +} + +static ssize_t rtrs_clt_remove_path_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t count) +{ + struct rtrs_clt_sess *sess; + int ret; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + if (!sysfs_streq(buf, "1")) { + rtrs_err(sess->clt, "%s: unknown value: '%s'\n", + attr->attr.name, buf); + return -EINVAL; + } + ret = rtrs_clt_remove_path_from_sysfs(sess, &attr->attr); + if (ret) + return ret; + + return count; +} + +static struct kobj_attribute rtrs_clt_remove_path_attr = + __ATTR(remove_path, 0644, rtrs_clt_remove_path_show, + rtrs_clt_remove_path_store); + +STAT_ATTR(struct rtrs_clt_stats, cpu_migration, + rtrs_clt_stats_migration_cnt_to_str, + rtrs_clt_reset_cpu_migr_stats); + +STAT_ATTR(struct rtrs_clt_stats, reconnects, + rtrs_clt_stats_reconnects_to_str, + rtrs_clt_reset_reconnects_stat); + +STAT_ATTR(struct rtrs_clt_stats, rdma, + rtrs_clt_stats_rdma_to_str, + rtrs_clt_reset_rdma_stats); + +STAT_ATTR(struct rtrs_clt_stats, reset_all, + rtrs_clt_reset_all_help, + rtrs_clt_reset_all_stats); + +static struct attribute *rtrs_clt_stats_attrs[] = { + &cpu_migration_attr.attr, + &reconnects_attr.attr, + &rdma_attr.attr, + &reset_all_attr.attr, + NULL +}; + +static struct attribute_group rtrs_clt_stats_attr_group = { + .attrs = rtrs_clt_stats_attrs, +}; + +static ssize_t rtrs_clt_hca_port_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_clt_sess *sess; + + sess = container_of(kobj, typeof(*sess), kobj); + + return scnprintf(page, PAGE_SIZE, "%u\n", sess->hca_port); +} + +static struct kobj_attribute rtrs_clt_hca_port_attr = + __ATTR(hca_port, 0444, rtrs_clt_hca_port_show, NULL); + +static ssize_t rtrs_clt_hca_name_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_clt_sess *sess; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + + return scnprintf(page, PAGE_SIZE, "%s\n", sess->hca_name); +} + +static struct kobj_attribute rtrs_clt_hca_name_attr = + __ATTR(hca_name, 0444, rtrs_clt_hca_name_show, NULL); + +static ssize_t rtrs_clt_src_addr_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_clt_sess *sess; + int cnt; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + cnt = sockaddr_to_str((struct sockaddr *)&sess->s.src_addr, + page, PAGE_SIZE); + return cnt + scnprintf(page + cnt, PAGE_SIZE - cnt, "\n"); +} + +static struct kobj_attribute rtrs_clt_src_addr_attr = + __ATTR(src_addr, 0444, rtrs_clt_src_addr_show, NULL); + +static ssize_t rtrs_clt_dst_addr_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_clt_sess *sess; + int cnt; + + sess = container_of(kobj, struct rtrs_clt_sess, kobj); + cnt = sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, + page, PAGE_SIZE); + return cnt + scnprintf(page + cnt, PAGE_SIZE - cnt, "\n"); +} + +static struct kobj_attribute rtrs_clt_dst_addr_attr = + __ATTR(dst_addr, 0444, rtrs_clt_dst_addr_show, NULL); + +static struct attribute *rtrs_clt_sess_attrs[] = { + &rtrs_clt_hca_name_attr.attr, + &rtrs_clt_hca_port_attr.attr, + &rtrs_clt_src_addr_attr.attr, + &rtrs_clt_dst_addr_attr.attr, + &rtrs_clt_state_attr.attr, + &rtrs_clt_reconnect_attr.attr, + &rtrs_clt_disconnect_attr.attr, + &rtrs_clt_remove_path_attr.attr, + NULL, +}; + +static struct attribute_group rtrs_clt_sess_attr_group = { + .attrs = rtrs_clt_sess_attrs, +}; + +int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt *clt = sess->clt; + char str[NAME_MAX]; + int err, cnt; + + cnt = sockaddr_to_str((struct sockaddr *)&sess->s.src_addr, + str, sizeof(str)); + cnt += scnprintf(str + cnt, sizeof(str) - cnt, "@"); + sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, + str + cnt, sizeof(str) - cnt); + + err = kobject_init_and_add(&sess->kobj, &ktype_sess, clt->kobj_paths, + "%s", str); + if (err) { + pr_err("kobject_init_and_add: %d\n", err); + return err; + } + err = sysfs_create_group(&sess->kobj, &rtrs_clt_sess_attr_group); + if (err) { + pr_err("sysfs_create_group(): %d\n", err); + goto put_kobj; + } + err = kobject_init_and_add(&sess->stats->kobj_stats, &ktype_stats, + &sess->kobj, "stats"); + if (err) { + pr_err("kobject_init_and_add: %d\n", err); + goto remove_group; + } + + err = sysfs_create_group(&sess->stats->kobj_stats, + &rtrs_clt_stats_attr_group); + if (err) { + pr_err("failed to create stats sysfs group, err: %d\n", err); + goto put_kobj_stats; + } + + return 0; + +put_kobj_stats: + kobject_del(&sess->stats->kobj_stats); + kobject_put(&sess->stats->kobj_stats); +remove_group: + sysfs_remove_group(&sess->kobj, &rtrs_clt_sess_attr_group); +put_kobj: + kobject_del(&sess->kobj); + kobject_put(&sess->kobj); + + return err; +} + +void rtrs_clt_destroy_sess_files(struct rtrs_clt_sess *sess, + const struct attribute *sysfs_self) +{ + kobject_del(&sess->stats->kobj_stats); + kobject_put(&sess->stats->kobj_stats); + if (sysfs_self) + sysfs_remove_file_self(&sess->kobj, sysfs_self); + kobject_del(&sess->kobj); +} + +static struct attribute *rtrs_clt_attrs[] = { + &dev_attr_max_reconnect_attempts.attr, + &dev_attr_mpath_policy.attr, + &dev_attr_add_path.attr, + NULL, +}; + +static struct attribute_group rtrs_clt_attr_group = { + .attrs = rtrs_clt_attrs, +}; + +int rtrs_clt_create_sysfs_root_files(struct rtrs_clt *clt) +{ + return sysfs_create_group(&clt->dev.kobj, &rtrs_clt_attr_group); +} + +void rtrs_clt_destroy_sysfs_root_folders(struct rtrs_clt *clt) +{ + if (clt->kobj_paths) { + kobject_del(clt->kobj_paths); + kobject_put(clt->kobj_paths); + } +} + +void rtrs_clt_destroy_sysfs_root_files(struct rtrs_clt *clt) +{ + sysfs_remove_group(&clt->dev.kobj, &rtrs_clt_attr_group); +} diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.c b/drivers/infiniband/ulp/rtrs/rtrs-clt.c new file mode 100644 index 000000000000..776e89231c52 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.c @@ -0,0 +1,3002 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ + +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include <linux/module.h> +#include <linux/rculist.h> +#include <linux/random.h> + +#include "rtrs-clt.h" +#include "rtrs-log.h" + +#define RTRS_CONNECT_TIMEOUT_MS 30000 +/* + * Wait a bit before trying to reconnect after a failure + * in order to give server time to finish clean up which + * leads to "false positives" failed reconnect attempts + */ +#define RTRS_RECONNECT_BACKOFF 1000 +/* + * Wait for additional random time between 0 and 8 seconds + * before starting to reconnect to avoid clients reconnecting + * all at once in case of a major network outage + */ +#define RTRS_RECONNECT_SEED 8 + +MODULE_DESCRIPTION("RDMA Transport Client"); +MODULE_LICENSE("GPL"); + +static const struct rtrs_rdma_dev_pd_ops dev_pd_ops; +static struct rtrs_rdma_dev_pd dev_pd = { + .ops = &dev_pd_ops +}; + +static struct workqueue_struct *rtrs_wq; +static struct class *rtrs_clt_dev_class; + +static inline bool rtrs_clt_is_connected(const struct rtrs_clt *clt) +{ + struct rtrs_clt_sess *sess; + bool connected = false; + + rcu_read_lock(); + list_for_each_entry_rcu(sess, &clt->paths_list, s.entry) + connected |= READ_ONCE(sess->state) == RTRS_CLT_CONNECTED; + rcu_read_unlock(); + + return connected; +} + +static struct rtrs_permit * +__rtrs_get_permit(struct rtrs_clt *clt, enum rtrs_clt_con_type con_type) +{ + size_t max_depth = clt->queue_depth; + struct rtrs_permit *permit; + int bit; + + /* + * Adapted from null_blk get_tag(). Callers from different cpus may + * grab the same bit, since find_first_zero_bit is not atomic. + * But then the test_and_set_bit_lock will fail for all the + * callers but one, so that they will loop again. + * This way an explicit spinlock is not required. + */ + do { + bit = find_first_zero_bit(clt->permits_map, max_depth); + if (unlikely(bit >= max_depth)) + return NULL; + } while (unlikely(test_and_set_bit_lock(bit, clt->permits_map))); + + permit = get_permit(clt, bit); + WARN_ON(permit->mem_id != bit); + permit->cpu_id = raw_smp_processor_id(); + permit->con_type = con_type; + + return permit; +} + +static inline void __rtrs_put_permit(struct rtrs_clt *clt, + struct rtrs_permit *permit) +{ + clear_bit_unlock(permit->mem_id, clt->permits_map); +} + +/** + * rtrs_clt_get_permit() - allocates permit for future RDMA operation + * @clt: Current session + * @con_type: Type of connection to use with the permit + * @can_wait: Wait type + * + * Description: + * Allocates permit for the following RDMA operation. Permit is used + * to preallocate all resources and to propagate memory pressure + * up earlier. + * + * Context: + * Can sleep if @wait == RTRS_TAG_WAIT + */ +struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt *clt, + enum rtrs_clt_con_type con_type, + int can_wait) +{ + struct rtrs_permit *permit; + DEFINE_WAIT(wait); + + permit = __rtrs_get_permit(clt, con_type); + if (likely(permit) || !can_wait) + return permit; + + do { + prepare_to_wait(&clt->permits_wait, &wait, + TASK_UNINTERRUPTIBLE); + permit = __rtrs_get_permit(clt, con_type); + if (likely(permit)) + break; + + io_schedule(); + } while (1); + + finish_wait(&clt->permits_wait, &wait); + + return permit; +} +EXPORT_SYMBOL(rtrs_clt_get_permit); + +/** + * rtrs_clt_put_permit() - puts allocated permit + * @clt: Current session + * @permit: Permit to be freed + * + * Context: + * Does not matter + */ +void rtrs_clt_put_permit(struct rtrs_clt *clt, struct rtrs_permit *permit) +{ + if (WARN_ON(!test_bit(permit->mem_id, clt->permits_map))) + return; + + __rtrs_put_permit(clt, permit); + + /* + * rtrs_clt_get_permit() adds itself to the &clt->permits_wait list + * before calling schedule(). So if rtrs_clt_get_permit() is sleeping + * it must have added itself to &clt->permits_wait before + * __rtrs_put_permit() finished. + * Hence it is safe to guard wake_up() with a waitqueue_active() test. + */ + if (waitqueue_active(&clt->permits_wait)) + wake_up(&clt->permits_wait); +} +EXPORT_SYMBOL(rtrs_clt_put_permit); + +void *rtrs_permit_to_pdu(struct rtrs_permit *permit) +{ + return permit + 1; +} +EXPORT_SYMBOL(rtrs_permit_to_pdu); + +/** + * rtrs_permit_to_clt_con() - returns RDMA connection pointer by the permit + * @sess: client session pointer + * @permit: permit for the allocation of the RDMA buffer + * Note: + * IO connection starts from 1. + * 0 connection is for user messages. + */ +static +struct rtrs_clt_con *rtrs_permit_to_clt_con(struct rtrs_clt_sess *sess, + struct rtrs_permit *permit) +{ + int id = 0; + + if (likely(permit->con_type == RTRS_IO_CON)) + id = (permit->cpu_id % (sess->s.con_num - 1)) + 1; + + return to_clt_con(sess->s.con[id]); +} + +/** + * __rtrs_clt_change_state() - change the session state through session state + * machine. + * + * @sess: client session to change the state of. + * @new_state: state to change to. + * + * returns true if successful, false if the requested state can not be set. + * + * Locks: + * state_wq lock must be hold. + */ +static bool __rtrs_clt_change_state(struct rtrs_clt_sess *sess, + enum rtrs_clt_state new_state) +{ + enum rtrs_clt_state old_state; + bool changed = false; + + lockdep_assert_held(&sess->state_wq.lock); + + old_state = sess->state; + switch (new_state) { + case RTRS_CLT_CONNECTING: + switch (old_state) { + case RTRS_CLT_RECONNECTING: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_CLT_RECONNECTING: + switch (old_state) { + case RTRS_CLT_CONNECTED: + case RTRS_CLT_CONNECTING_ERR: + case RTRS_CLT_CLOSED: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_CLT_CONNECTED: + switch (old_state) { + case RTRS_CLT_CONNECTING: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_CLT_CONNECTING_ERR: + switch (old_state) { + case RTRS_CLT_CONNECTING: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_CLT_CLOSING: + switch (old_state) { + case RTRS_CLT_CONNECTING: + case RTRS_CLT_CONNECTING_ERR: + case RTRS_CLT_RECONNECTING: + case RTRS_CLT_CONNECTED: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_CLT_CLOSED: + switch (old_state) { + case RTRS_CLT_CLOSING: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_CLT_DEAD: + switch (old_state) { + case RTRS_CLT_CLOSED: + changed = true; + fallthrough; + default: + break; + } + break; + default: + break; + } + if (changed) { + sess->state = new_state; + wake_up_locked(&sess->state_wq); + } + + return changed; +} + +static bool rtrs_clt_change_state_from_to(struct rtrs_clt_sess *sess, + enum rtrs_clt_state old_state, + enum rtrs_clt_state new_state) +{ + bool changed = false; + + spin_lock_irq(&sess->state_wq.lock); + if (sess->state == old_state) + changed = __rtrs_clt_change_state(sess, new_state); + spin_unlock_irq(&sess->state_wq.lock); + + return changed; +} + +static void rtrs_rdma_error_recovery(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + + if (rtrs_clt_change_state_from_to(sess, + RTRS_CLT_CONNECTED, + RTRS_CLT_RECONNECTING)) { + struct rtrs_clt *clt = sess->clt; + unsigned int delay_ms; + + /* + * Normal scenario, reconnect if we were successfully connected + */ + delay_ms = clt->reconnect_delay_sec * 1000; + queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, + msecs_to_jiffies(delay_ms + + prandom_u32() % RTRS_RECONNECT_SEED)); + } else { + /* + * Error can happen just on establishing new connection, + * so notify waiter with error state, waiter is responsible + * for cleaning the rest and reconnect if needed. + */ + rtrs_clt_change_state_from_to(sess, + RTRS_CLT_CONNECTING, + RTRS_CLT_CONNECTING_ERR); + } +} + +static void rtrs_clt_fast_reg_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_clt_con *con = cq->cq_context; + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(con->c.sess, "Failed IB_WR_REG_MR: %s\n", + ib_wc_status_msg(wc->status)); + rtrs_rdma_error_recovery(con); + } +} + +static struct ib_cqe fast_reg_cqe = { + .done = rtrs_clt_fast_reg_done +}; + +static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno, + bool notify, bool can_wait); + +static void rtrs_clt_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_clt_io_req *req = + container_of(wc->wr_cqe, typeof(*req), inv_cqe); + struct rtrs_clt_con *con = cq->cq_context; + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(con->c.sess, "Failed IB_WR_LOCAL_INV: %s\n", + ib_wc_status_msg(wc->status)); + rtrs_rdma_error_recovery(con); + } + req->need_inv = false; + if (likely(req->need_inv_comp)) + complete(&req->inv_comp); + else + /* Complete request from INV callback */ + complete_rdma_req(req, req->inv_errno, true, false); +} + +static int rtrs_inv_rkey(struct rtrs_clt_io_req *req) +{ + struct rtrs_clt_con *con = req->con; + struct ib_send_wr wr = { + .opcode = IB_WR_LOCAL_INV, + .wr_cqe = &req->inv_cqe, + .send_flags = IB_SEND_SIGNALED, + .ex.invalidate_rkey = req->mr->rkey, + }; + req->inv_cqe.done = rtrs_clt_inv_rkey_done; + + return ib_post_send(con->c.qp, &wr, NULL); +} + +static void complete_rdma_req(struct rtrs_clt_io_req *req, int errno, + bool notify, bool can_wait) +{ + struct rtrs_clt_con *con = req->con; + struct rtrs_clt_sess *sess; + int err; + + if (WARN_ON(!req->in_use)) + return; + if (WARN_ON(!req->con)) + return; + sess = to_clt_sess(con->c.sess); + + if (req->sg_cnt) { + if (unlikely(req->dir == DMA_FROM_DEVICE && req->need_inv)) { + /* + * We are here to invalidate read requests + * ourselves. In normal scenario server should + * send INV for all read requests, but + * we are here, thus two things could happen: + * + * 1. this is failover, when errno != 0 + * and can_wait == 1, + * + * 2. something totally bad happened and + * server forgot to send INV, so we + * should do that ourselves. + */ + + if (likely(can_wait)) { + req->need_inv_comp = true; + } else { + /* This should be IO path, so always notify */ + WARN_ON(!notify); + /* Save errno for INV callback */ + req->inv_errno = errno; + } + + err = rtrs_inv_rkey(req); + if (unlikely(err)) { + rtrs_err(con->c.sess, "Send INV WR key=%#x: %d\n", + req->mr->rkey, err); + } else if (likely(can_wait)) { + wait_for_completion(&req->inv_comp); + } else { + /* + * Something went wrong, so request will be + * completed from INV callback. + */ + WARN_ON_ONCE(1); + + return; + } + } + ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist, + req->sg_cnt, req->dir); + } + if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT) + atomic_dec(&sess->stats->inflight); + + req->in_use = false; + req->con = NULL; + + if (notify) + req->conf(req->priv, errno); +} + +static int rtrs_post_send_rdma(struct rtrs_clt_con *con, + struct rtrs_clt_io_req *req, + struct rtrs_rbuf *rbuf, u32 off, + u32 imm, struct ib_send_wr *wr) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + enum ib_send_flags flags; + struct ib_sge sge; + + if (unlikely(!req->sg_size)) { + rtrs_wrn(con->c.sess, + "Doing RDMA Write failed, no data supplied\n"); + return -EINVAL; + } + + /* user data and user message in the first list element */ + sge.addr = req->iu->dma_addr; + sge.length = req->sg_size; + sge.lkey = sess->s.dev->ib_pd->local_dma_lkey; + + /* + * From time to time we have to post signalled sends, + * or send queue will fill up and only QP reset can help. + */ + flags = atomic_inc_return(&con->io_cnt) % sess->queue_depth ? + 0 : IB_SEND_SIGNALED; + + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, req->iu->dma_addr, + req->sg_size, DMA_TO_DEVICE); + + return rtrs_iu_post_rdma_write_imm(&con->c, req->iu, &sge, 1, + rbuf->rkey, rbuf->addr + off, + imm, flags, wr); +} + +static void process_io_rsp(struct rtrs_clt_sess *sess, u32 msg_id, + s16 errno, bool w_inval) +{ + struct rtrs_clt_io_req *req; + + if (WARN_ON(msg_id >= sess->queue_depth)) + return; + + req = &sess->reqs[msg_id]; + /* Drop need_inv if server responded with send with invalidation */ + req->need_inv &= !w_inval; + complete_rdma_req(req, errno, true, false); +} + +static void rtrs_clt_recv_done(struct rtrs_clt_con *con, struct ib_wc *wc) +{ + struct rtrs_iu *iu; + int err; + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + + WARN_ON(sess->flags != RTRS_MSG_NEW_RKEY_F); + iu = container_of(wc->wr_cqe, struct rtrs_iu, + cqe); + err = rtrs_iu_post_recv(&con->c, iu); + if (unlikely(err)) { + rtrs_err(con->c.sess, "post iu failed %d\n", err); + rtrs_rdma_error_recovery(con); + } +} + +static void rtrs_clt_rkey_rsp_done(struct rtrs_clt_con *con, struct ib_wc *wc) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_msg_rkey_rsp *msg; + u32 imm_type, imm_payload; + bool w_inval = false; + struct rtrs_iu *iu; + u32 buf_id; + int err; + + WARN_ON(sess->flags != RTRS_MSG_NEW_RKEY_F); + + iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); + + if (unlikely(wc->byte_len < sizeof(*msg))) { + rtrs_err(con->c.sess, "rkey response is malformed: size %d\n", + wc->byte_len); + goto out; + } + ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr, + iu->size, DMA_FROM_DEVICE); + msg = iu->buf; + if (unlikely(le16_to_cpu(msg->type) != RTRS_MSG_RKEY_RSP)) { + rtrs_err(sess->clt, "rkey response is malformed: type %d\n", + le16_to_cpu(msg->type)); + goto out; + } + buf_id = le16_to_cpu(msg->buf_id); + if (WARN_ON(buf_id >= sess->queue_depth)) + goto out; + + rtrs_from_imm(be32_to_cpu(wc->ex.imm_data), &imm_type, &imm_payload); + if (likely(imm_type == RTRS_IO_RSP_IMM || + imm_type == RTRS_IO_RSP_W_INV_IMM)) { + u32 msg_id; + + w_inval = (imm_type == RTRS_IO_RSP_W_INV_IMM); + rtrs_from_io_rsp_imm(imm_payload, &msg_id, &err); + + if (WARN_ON(buf_id != msg_id)) + goto out; + sess->rbufs[buf_id].rkey = le32_to_cpu(msg->rkey); + process_io_rsp(sess, msg_id, err, w_inval); + } + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, iu->dma_addr, + iu->size, DMA_FROM_DEVICE); + return rtrs_clt_recv_done(con, wc); +out: + rtrs_rdma_error_recovery(con); +} + +static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc); + +static struct ib_cqe io_comp_cqe = { + .done = rtrs_clt_rdma_done +}; + +/* + * Post x2 empty WRs: first is for this RDMA with IMM, + * second is for RECV with INV, which happened earlier. + */ +static int rtrs_post_recv_empty_x2(struct rtrs_con *con, struct ib_cqe *cqe) +{ + struct ib_recv_wr wr_arr[2], *wr; + int i; + + memset(wr_arr, 0, sizeof(wr_arr)); + for (i = 0; i < ARRAY_SIZE(wr_arr); i++) { + wr = &wr_arr[i]; + wr->wr_cqe = cqe; + if (i) + /* Chain backwards */ + wr->next = &wr_arr[i - 1]; + } + + return ib_post_recv(con->qp, wr, NULL); +} + +static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_clt_con *con = cq->cq_context; + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + u32 imm_type, imm_payload; + bool w_inval = false; + int err; + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + if (wc->status != IB_WC_WR_FLUSH_ERR) { + rtrs_err(sess->clt, "RDMA failed: %s\n", + ib_wc_status_msg(wc->status)); + rtrs_rdma_error_recovery(con); + } + return; + } + rtrs_clt_update_wc_stats(con); + + switch (wc->opcode) { + case IB_WC_RECV_RDMA_WITH_IMM: + /* + * post_recv() RDMA write completions of IO reqs (read/write) + * and hb + */ + if (WARN_ON(wc->wr_cqe->done != rtrs_clt_rdma_done)) + return; + rtrs_from_imm(be32_to_cpu(wc->ex.imm_data), + &imm_type, &imm_payload); + if (likely(imm_type == RTRS_IO_RSP_IMM || + imm_type == RTRS_IO_RSP_W_INV_IMM)) { + u32 msg_id; + + w_inval = (imm_type == RTRS_IO_RSP_W_INV_IMM); + rtrs_from_io_rsp_imm(imm_payload, &msg_id, &err); + + process_io_rsp(sess, msg_id, err, w_inval); + } else if (imm_type == RTRS_HB_MSG_IMM) { + WARN_ON(con->c.cid); + rtrs_send_hb_ack(&sess->s); + if (sess->flags == RTRS_MSG_NEW_RKEY_F) + return rtrs_clt_recv_done(con, wc); + } else if (imm_type == RTRS_HB_ACK_IMM) { + WARN_ON(con->c.cid); + sess->s.hb_missed_cnt = 0; + if (sess->flags == RTRS_MSG_NEW_RKEY_F) + return rtrs_clt_recv_done(con, wc); + } else { + rtrs_wrn(con->c.sess, "Unknown IMM type %u\n", + imm_type); + } + if (w_inval) + /* + * Post x2 empty WRs: first is for this RDMA with IMM, + * second is for RECV with INV, which happened earlier. + */ + err = rtrs_post_recv_empty_x2(&con->c, &io_comp_cqe); + else + err = rtrs_post_recv_empty(&con->c, &io_comp_cqe); + if (unlikely(err)) { + rtrs_err(con->c.sess, "rtrs_post_recv_empty(): %d\n", + err); + rtrs_rdma_error_recovery(con); + break; + } + break; + case IB_WC_RECV: + /* + * Key invalidations from server side + */ + WARN_ON(!(wc->wc_flags & IB_WC_WITH_INVALIDATE || + wc->wc_flags & IB_WC_WITH_IMM)); + WARN_ON(wc->wr_cqe->done != rtrs_clt_rdma_done); + if (sess->flags == RTRS_MSG_NEW_RKEY_F) { + if (wc->wc_flags & IB_WC_WITH_INVALIDATE) + return rtrs_clt_recv_done(con, wc); + + return rtrs_clt_rkey_rsp_done(con, wc); + } + break; + case IB_WC_RDMA_WRITE: + /* + * post_send() RDMA write completions of IO reqs (read/write) + * and hb + */ + break; + + default: + rtrs_wrn(sess->clt, "Unexpected WC type: %d\n", wc->opcode); + return; + } +} + +static int post_recv_io(struct rtrs_clt_con *con, size_t q_size) +{ + int err, i; + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + + for (i = 0; i < q_size; i++) { + if (sess->flags == RTRS_MSG_NEW_RKEY_F) { + struct rtrs_iu *iu = &con->rsp_ius[i]; + + err = rtrs_iu_post_recv(&con->c, iu); + } else { + err = rtrs_post_recv_empty(&con->c, &io_comp_cqe); + } + if (unlikely(err)) + return err; + } + + return 0; +} + +static int post_recv_sess(struct rtrs_clt_sess *sess) +{ + size_t q_size = 0; + int err, cid; + + for (cid = 0; cid < sess->s.con_num; cid++) { + if (cid == 0) + q_size = SERVICE_CON_QUEUE_DEPTH; + else + q_size = sess->queue_depth; + + /* + * x2 for RDMA read responses + FR key invalidations, + * RDMA writes do not require any FR registrations. + */ + q_size *= 2; + + err = post_recv_io(to_clt_con(sess->s.con[cid]), q_size); + if (unlikely(err)) { + rtrs_err(sess->clt, "post_recv_io(), err: %d\n", err); + return err; + } + } + + return 0; +} + +struct path_it { + int i; + struct list_head skip_list; + struct rtrs_clt *clt; + struct rtrs_clt_sess *(*next_path)(struct path_it *it); +}; + +/** + * list_next_or_null_rr_rcu - get next list element in round-robin fashion. + * @head: the head for the list. + * @ptr: the list head to take the next element from. + * @type: the type of the struct this is embedded in. + * @memb: the name of the list_head within the struct. + * + * Next element returned in round-robin fashion, i.e. head will be skipped, + * but if list is observed as empty, NULL will be returned. + * + * This primitive may safely run concurrently with the _rcu list-mutation + * primitives such as list_add_rcu() as long as it's guarded by rcu_read_lock(). + */ +#define list_next_or_null_rr_rcu(head, ptr, type, memb) \ +({ \ + list_next_or_null_rcu(head, ptr, type, memb) ?: \ + list_next_or_null_rcu(head, READ_ONCE((ptr)->next), \ + type, memb); \ +}) + +/** + * get_next_path_rr() - Returns path in round-robin fashion. + * @it: the path pointer + * + * Related to @MP_POLICY_RR + * + * Locks: + * rcu_read_lock() must be hold. + */ +static struct rtrs_clt_sess *get_next_path_rr(struct path_it *it) +{ + struct rtrs_clt_sess __rcu **ppcpu_path; + struct rtrs_clt_sess *path; + struct rtrs_clt *clt; + + clt = it->clt; + + /* + * Here we use two RCU objects: @paths_list and @pcpu_path + * pointer. See rtrs_clt_remove_path_from_arr() for details + * how that is handled. + */ + + ppcpu_path = this_cpu_ptr(clt->pcpu_path); + path = rcu_dereference(*ppcpu_path); + if (unlikely(!path)) + path = list_first_or_null_rcu(&clt->paths_list, + typeof(*path), s.entry); + else + path = list_next_or_null_rr_rcu(&clt->paths_list, + &path->s.entry, + typeof(*path), + s.entry); + rcu_assign_pointer(*ppcpu_path, path); + + return path; +} + +/** + * get_next_path_min_inflight() - Returns path with minimal inflight count. + * @it: the path pointer + * + * Related to @MP_POLICY_MIN_INFLIGHT + * + * Locks: + * rcu_read_lock() must be hold. + */ +static struct rtrs_clt_sess *get_next_path_min_inflight(struct path_it *it) +{ + struct rtrs_clt_sess *min_path = NULL; + struct rtrs_clt *clt = it->clt; + struct rtrs_clt_sess *sess; + int min_inflight = INT_MAX; + int inflight; + + list_for_each_entry_rcu(sess, &clt->paths_list, s.entry) { + if (unlikely(!list_empty(raw_cpu_ptr(sess->mp_skip_entry)))) + continue; + + inflight = atomic_read(&sess->stats->inflight); + + if (inflight < min_inflight) { + min_inflight = inflight; + min_path = sess; + } + } + + /* + * add the path to the skip list, so that next time we can get + * a different one + */ + if (min_path) + list_add(raw_cpu_ptr(min_path->mp_skip_entry), &it->skip_list); + + return min_path; +} + +static inline void path_it_init(struct path_it *it, struct rtrs_clt *clt) +{ + INIT_LIST_HEAD(&it->skip_list); + it->clt = clt; + it->i = 0; + + if (clt->mp_policy == MP_POLICY_RR) + it->next_path = get_next_path_rr; + else + it->next_path = get_next_path_min_inflight; +} + +static inline void path_it_deinit(struct path_it *it) +{ + struct list_head *skip, *tmp; + /* + * The skip_list is used only for the MIN_INFLIGHT policy. + * We need to remove paths from it, so that next IO can insert + * paths (->mp_skip_entry) into a skip_list again. + */ + list_for_each_safe(skip, tmp, &it->skip_list) + list_del_init(skip); +} + +/** + * rtrs_clt_init_req() Initialize an rtrs_clt_io_req holding information + * about an inflight IO. + * The user buffer holding user control message (not data) is copied into + * the corresponding buffer of rtrs_iu (req->iu->buf), which later on will + * also hold the control message of rtrs. + * @req: an io request holding information about IO. + * @sess: client session + * @conf: conformation callback function to notify upper layer. + * @permit: permit for allocation of RDMA remote buffer + * @priv: private pointer + * @vec: kernel vector containing control message + * @usr_len: length of the user message + * @sg: scater list for IO data + * @sg_cnt: number of scater list entries + * @data_len: length of the IO data + * @dir: direction of the IO. + */ +static void rtrs_clt_init_req(struct rtrs_clt_io_req *req, + struct rtrs_clt_sess *sess, + void (*conf)(void *priv, int errno), + struct rtrs_permit *permit, void *priv, + const struct kvec *vec, size_t usr_len, + struct scatterlist *sg, size_t sg_cnt, + size_t data_len, int dir) +{ + struct iov_iter iter; + size_t len; + + req->permit = permit; + req->in_use = true; + req->usr_len = usr_len; + req->data_len = data_len; + req->sglist = sg; + req->sg_cnt = sg_cnt; + req->priv = priv; + req->dir = dir; + req->con = rtrs_permit_to_clt_con(sess, permit); + req->conf = conf; + req->need_inv = false; + req->need_inv_comp = false; + req->inv_errno = 0; + + iov_iter_kvec(&iter, READ, vec, 1, usr_len); + len = _copy_from_iter(req->iu->buf, usr_len, &iter); + WARN_ON(len != usr_len); + + reinit_completion(&req->inv_comp); +} + +static struct rtrs_clt_io_req * +rtrs_clt_get_req(struct rtrs_clt_sess *sess, + void (*conf)(void *priv, int errno), + struct rtrs_permit *permit, void *priv, + const struct kvec *vec, size_t usr_len, + struct scatterlist *sg, size_t sg_cnt, + size_t data_len, int dir) +{ + struct rtrs_clt_io_req *req; + + req = &sess->reqs[permit->mem_id]; + rtrs_clt_init_req(req, sess, conf, permit, priv, vec, usr_len, + sg, sg_cnt, data_len, dir); + return req; +} + +static struct rtrs_clt_io_req * +rtrs_clt_get_copy_req(struct rtrs_clt_sess *alive_sess, + struct rtrs_clt_io_req *fail_req) +{ + struct rtrs_clt_io_req *req; + struct kvec vec = { + .iov_base = fail_req->iu->buf, + .iov_len = fail_req->usr_len + }; + + req = &alive_sess->reqs[fail_req->permit->mem_id]; + rtrs_clt_init_req(req, alive_sess, fail_req->conf, fail_req->permit, + fail_req->priv, &vec, fail_req->usr_len, + fail_req->sglist, fail_req->sg_cnt, + fail_req->data_len, fail_req->dir); + return req; +} + +static int rtrs_post_rdma_write_sg(struct rtrs_clt_con *con, + struct rtrs_clt_io_req *req, + struct rtrs_rbuf *rbuf, + u32 size, u32 imm) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct ib_sge *sge = req->sge; + enum ib_send_flags flags; + struct scatterlist *sg; + size_t num_sge; + int i; + + for_each_sg(req->sglist, sg, req->sg_cnt, i) { + sge[i].addr = sg_dma_address(sg); + sge[i].length = sg_dma_len(sg); + sge[i].lkey = sess->s.dev->ib_pd->local_dma_lkey; + } + sge[i].addr = req->iu->dma_addr; + sge[i].length = size; + sge[i].lkey = sess->s.dev->ib_pd->local_dma_lkey; + + num_sge = 1 + req->sg_cnt; + + /* + * From time to time we have to post signalled sends, + * or send queue will fill up and only QP reset can help. + */ + flags = atomic_inc_return(&con->io_cnt) % sess->queue_depth ? + 0 : IB_SEND_SIGNALED; + + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, req->iu->dma_addr, + size, DMA_TO_DEVICE); + + return rtrs_iu_post_rdma_write_imm(&con->c, req->iu, sge, num_sge, + rbuf->rkey, rbuf->addr, imm, + flags, NULL); +} + +static int rtrs_clt_write_req(struct rtrs_clt_io_req *req) +{ + struct rtrs_clt_con *con = req->con; + struct rtrs_sess *s = con->c.sess; + struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rtrs_msg_rdma_write *msg; + + struct rtrs_rbuf *rbuf; + int ret, count = 0; + u32 imm, buf_id; + + const size_t tsize = sizeof(*msg) + req->data_len + req->usr_len; + + if (unlikely(tsize > sess->chunk_size)) { + rtrs_wrn(s, "Write request failed, size too big %zu > %d\n", + tsize, sess->chunk_size); + return -EMSGSIZE; + } + if (req->sg_cnt) { + count = ib_dma_map_sg(sess->s.dev->ib_dev, req->sglist, + req->sg_cnt, req->dir); + if (unlikely(!count)) { + rtrs_wrn(s, "Write request failed, map failed\n"); + return -EINVAL; + } + } + /* put rtrs msg after sg and user message */ + msg = req->iu->buf + req->usr_len; + msg->type = cpu_to_le16(RTRS_MSG_WRITE); + msg->usr_len = cpu_to_le16(req->usr_len); + + /* rtrs message on server side will be after user data and message */ + imm = req->permit->mem_off + req->data_len + req->usr_len; + imm = rtrs_to_io_req_imm(imm); + buf_id = req->permit->mem_id; + req->sg_size = tsize; + rbuf = &sess->rbufs[buf_id]; + + /* + * Update stats now, after request is successfully sent it is not + * safe anymore to touch it. + */ + rtrs_clt_update_all_stats(req, WRITE); + + ret = rtrs_post_rdma_write_sg(req->con, req, rbuf, + req->usr_len + sizeof(*msg), + imm); + if (unlikely(ret)) { + rtrs_err(s, "Write request failed: %d\n", ret); + if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT) + atomic_dec(&sess->stats->inflight); + if (req->sg_cnt) + ib_dma_unmap_sg(sess->s.dev->ib_dev, req->sglist, + req->sg_cnt, req->dir); + } + + return ret; +} + +static int rtrs_map_sg_fr(struct rtrs_clt_io_req *req, size_t count) +{ + int nr; + + /* Align the MR to a 4K page size to match the block virt boundary */ + nr = ib_map_mr_sg(req->mr, req->sglist, count, NULL, SZ_4K); + if (nr < 0) + return nr; + if (unlikely(nr < req->sg_cnt)) + return -EINVAL; + ib_update_fast_reg_key(req->mr, ib_inc_rkey(req->mr->rkey)); + + return nr; +} + +static int rtrs_clt_read_req(struct rtrs_clt_io_req *req) +{ + struct rtrs_clt_con *con = req->con; + struct rtrs_sess *s = con->c.sess; + struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rtrs_msg_rdma_read *msg; + struct rtrs_ib_dev *dev; + + struct ib_reg_wr rwr; + struct ib_send_wr *wr = NULL; + + int ret, count = 0; + u32 imm, buf_id; + + const size_t tsize = sizeof(*msg) + req->data_len + req->usr_len; + + s = &sess->s; + dev = sess->s.dev; + + if (unlikely(tsize > sess->chunk_size)) { + rtrs_wrn(s, + "Read request failed, message size is %zu, bigger than CHUNK_SIZE %d\n", + tsize, sess->chunk_size); + return -EMSGSIZE; + } + + if (req->sg_cnt) { + count = ib_dma_map_sg(dev->ib_dev, req->sglist, req->sg_cnt, + req->dir); + if (unlikely(!count)) { + rtrs_wrn(s, + "Read request failed, dma map failed\n"); + return -EINVAL; + } + } + /* put our message into req->buf after user message*/ + msg = req->iu->buf + req->usr_len; + msg->type = cpu_to_le16(RTRS_MSG_READ); + msg->usr_len = cpu_to_le16(req->usr_len); + + if (count) { + ret = rtrs_map_sg_fr(req, count); + if (ret < 0) { + rtrs_err_rl(s, + "Read request failed, failed to map fast reg. data, err: %d\n", + ret); + ib_dma_unmap_sg(dev->ib_dev, req->sglist, req->sg_cnt, + req->dir); + return ret; + } + rwr = (struct ib_reg_wr) { + .wr.opcode = IB_WR_REG_MR, + .wr.wr_cqe = &fast_reg_cqe, + .mr = req->mr, + .key = req->mr->rkey, + .access = (IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE), + }; + wr = &rwr.wr; + + msg->sg_cnt = cpu_to_le16(1); + msg->flags = cpu_to_le16(RTRS_MSG_NEED_INVAL_F); + + msg->desc[0].addr = cpu_to_le64(req->mr->iova); + msg->desc[0].key = cpu_to_le32(req->mr->rkey); + msg->desc[0].len = cpu_to_le32(req->mr->length); + + /* Further invalidation is required */ + req->need_inv = !!RTRS_MSG_NEED_INVAL_F; + + } else { + msg->sg_cnt = 0; + msg->flags = 0; + } + /* + * rtrs message will be after the space reserved for disk data and + * user message + */ + imm = req->permit->mem_off + req->data_len + req->usr_len; + imm = rtrs_to_io_req_imm(imm); + buf_id = req->permit->mem_id; + + req->sg_size = sizeof(*msg); + req->sg_size += le16_to_cpu(msg->sg_cnt) * sizeof(struct rtrs_sg_desc); + req->sg_size += req->usr_len; + + /* + * Update stats now, after request is successfully sent it is not + * safe anymore to touch it. + */ + rtrs_clt_update_all_stats(req, READ); + + ret = rtrs_post_send_rdma(req->con, req, &sess->rbufs[buf_id], + req->data_len, imm, wr); + if (unlikely(ret)) { + rtrs_err(s, "Read request failed: %d\n", ret); + if (sess->clt->mp_policy == MP_POLICY_MIN_INFLIGHT) + atomic_dec(&sess->stats->inflight); + req->need_inv = false; + if (req->sg_cnt) + ib_dma_unmap_sg(dev->ib_dev, req->sglist, + req->sg_cnt, req->dir); + } + + return ret; +} + +/** + * rtrs_clt_failover_req() Try to find an active path for a failed request + * @clt: clt context + * @fail_req: a failed io request. + */ +static int rtrs_clt_failover_req(struct rtrs_clt *clt, + struct rtrs_clt_io_req *fail_req) +{ + struct rtrs_clt_sess *alive_sess; + struct rtrs_clt_io_req *req; + int err = -ECONNABORTED; + struct path_it it; + + rcu_read_lock(); + for (path_it_init(&it, clt); + (alive_sess = it.next_path(&it)) && it.i < it.clt->paths_num; + it.i++) { + if (unlikely(READ_ONCE(alive_sess->state) != + RTRS_CLT_CONNECTED)) + continue; + req = rtrs_clt_get_copy_req(alive_sess, fail_req); + if (req->dir == DMA_TO_DEVICE) + err = rtrs_clt_write_req(req); + else + err = rtrs_clt_read_req(req); + if (unlikely(err)) { + req->in_use = false; + continue; + } + /* Success path */ + rtrs_clt_inc_failover_cnt(alive_sess->stats); + break; + } + path_it_deinit(&it); + rcu_read_unlock(); + + return err; +} + +static void fail_all_outstanding_reqs(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_io_req *req; + int i, err; + + if (!sess->reqs) + return; + for (i = 0; i < sess->queue_depth; ++i) { + req = &sess->reqs[i]; + if (!req->in_use) + continue; + + /* + * Safely (without notification) complete failed request. + * After completion this request is still useble and can + * be failovered to another path. + */ + complete_rdma_req(req, -ECONNABORTED, false, true); + + err = rtrs_clt_failover_req(clt, req); + if (unlikely(err)) + /* Failover failed, notify anyway */ + req->conf(req->priv, err); + } +} + +static void free_sess_reqs(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt_io_req *req; + int i; + + if (!sess->reqs) + return; + for (i = 0; i < sess->queue_depth; ++i) { + req = &sess->reqs[i]; + if (req->mr) + ib_dereg_mr(req->mr); + kfree(req->sge); + rtrs_iu_free(req->iu, DMA_TO_DEVICE, + sess->s.dev->ib_dev, 1); + } + kfree(sess->reqs); + sess->reqs = NULL; +} + +static int alloc_sess_reqs(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt_io_req *req; + struct rtrs_clt *clt = sess->clt; + int i, err = -ENOMEM; + + sess->reqs = kcalloc(sess->queue_depth, sizeof(*sess->reqs), + GFP_KERNEL); + if (!sess->reqs) + return -ENOMEM; + + for (i = 0; i < sess->queue_depth; ++i) { + req = &sess->reqs[i]; + req->iu = rtrs_iu_alloc(1, sess->max_hdr_size, GFP_KERNEL, + sess->s.dev->ib_dev, + DMA_TO_DEVICE, + rtrs_clt_rdma_done); + if (!req->iu) + goto out; + + req->sge = kmalloc_array(clt->max_segments + 1, + sizeof(*req->sge), GFP_KERNEL); + if (!req->sge) + goto out; + + req->mr = ib_alloc_mr(sess->s.dev->ib_pd, IB_MR_TYPE_MEM_REG, + sess->max_pages_per_mr); + if (IS_ERR(req->mr)) { + err = PTR_ERR(req->mr); + req->mr = NULL; + pr_err("Failed to alloc sess->max_pages_per_mr %d\n", + sess->max_pages_per_mr); + goto out; + } + + init_completion(&req->inv_comp); + } + + return 0; + +out: + free_sess_reqs(sess); + + return err; +} + +static int alloc_permits(struct rtrs_clt *clt) +{ + unsigned int chunk_bits; + int err, i; + + clt->permits_map = kcalloc(BITS_TO_LONGS(clt->queue_depth), + sizeof(long), GFP_KERNEL); + if (!clt->permits_map) { + err = -ENOMEM; + goto out_err; + } + clt->permits = kcalloc(clt->queue_depth, permit_size(clt), GFP_KERNEL); + if (!clt->permits) { + err = -ENOMEM; + goto err_map; + } + chunk_bits = ilog2(clt->queue_depth - 1) + 1; + for (i = 0; i < clt->queue_depth; i++) { + struct rtrs_permit *permit; + + permit = get_permit(clt, i); + permit->mem_id = i; + permit->mem_off = i << (MAX_IMM_PAYL_BITS - chunk_bits); + } + + return 0; + +err_map: + kfree(clt->permits_map); + clt->permits_map = NULL; +out_err: + return err; +} + +static void free_permits(struct rtrs_clt *clt) +{ + kfree(clt->permits_map); + clt->permits_map = NULL; + kfree(clt->permits); + clt->permits = NULL; +} + +static void query_fast_reg_mode(struct rtrs_clt_sess *sess) +{ + struct ib_device *ib_dev; + u64 max_pages_per_mr; + int mr_page_shift; + + ib_dev = sess->s.dev->ib_dev; + + /* + * Use the smallest page size supported by the HCA, down to a + * minimum of 4096 bytes. We're unlikely to build large sglists + * out of smaller entries. + */ + mr_page_shift = max(12, ffs(ib_dev->attrs.page_size_cap) - 1); + max_pages_per_mr = ib_dev->attrs.max_mr_size; + do_div(max_pages_per_mr, (1ull << mr_page_shift)); + sess->max_pages_per_mr = + min3(sess->max_pages_per_mr, (u32)max_pages_per_mr, + ib_dev->attrs.max_fast_reg_page_list_len); + sess->max_send_sge = ib_dev->attrs.max_send_sge; +} + +static bool rtrs_clt_change_state_get_old(struct rtrs_clt_sess *sess, + enum rtrs_clt_state new_state, + enum rtrs_clt_state *old_state) +{ + bool changed; + + spin_lock_irq(&sess->state_wq.lock); + *old_state = sess->state; + changed = __rtrs_clt_change_state(sess, new_state); + spin_unlock_irq(&sess->state_wq.lock); + + return changed; +} + +static bool rtrs_clt_change_state(struct rtrs_clt_sess *sess, + enum rtrs_clt_state new_state) +{ + enum rtrs_clt_state old_state; + + return rtrs_clt_change_state_get_old(sess, new_state, &old_state); +} + +static void rtrs_clt_hb_err_handler(struct rtrs_con *c) +{ + struct rtrs_clt_con *con = container_of(c, typeof(*con), c); + + rtrs_rdma_error_recovery(con); +} + +static void rtrs_clt_init_hb(struct rtrs_clt_sess *sess) +{ + rtrs_init_hb(&sess->s, &io_comp_cqe, + RTRS_HB_INTERVAL_MS, + RTRS_HB_MISSED_MAX, + rtrs_clt_hb_err_handler, + rtrs_wq); +} + +static void rtrs_clt_start_hb(struct rtrs_clt_sess *sess) +{ + rtrs_start_hb(&sess->s); +} + +static void rtrs_clt_stop_hb(struct rtrs_clt_sess *sess) +{ + rtrs_stop_hb(&sess->s); +} + +static void rtrs_clt_reconnect_work(struct work_struct *work); +static void rtrs_clt_close_work(struct work_struct *work); + +static struct rtrs_clt_sess *alloc_sess(struct rtrs_clt *clt, + const struct rtrs_addr *path, + size_t con_num, u16 max_segments, + size_t max_segment_size) +{ + struct rtrs_clt_sess *sess; + int err = -ENOMEM; + int cpu; + + sess = kzalloc(sizeof(*sess), GFP_KERNEL); + if (!sess) + goto err; + + /* Extra connection for user messages */ + con_num += 1; + + sess->s.con = kcalloc(con_num, sizeof(*sess->s.con), GFP_KERNEL); + if (!sess->s.con) + goto err_free_sess; + + sess->stats = kzalloc(sizeof(*sess->stats), GFP_KERNEL); + if (!sess->stats) + goto err_free_con; + + mutex_init(&sess->init_mutex); + uuid_gen(&sess->s.uuid); + memcpy(&sess->s.dst_addr, path->dst, + rdma_addr_size((struct sockaddr *)path->dst)); + + /* + * rdma_resolve_addr() passes src_addr to cma_bind_addr, which + * checks the sa_family to be non-zero. If user passed src_addr=NULL + * the sess->src_addr will contain only zeros, which is then fine. + */ + if (path->src) + memcpy(&sess->s.src_addr, path->src, + rdma_addr_size((struct sockaddr *)path->src)); + strlcpy(sess->s.sessname, clt->sessname, sizeof(sess->s.sessname)); + sess->s.con_num = con_num; + sess->clt = clt; + sess->max_pages_per_mr = max_segments * max_segment_size >> 12; + init_waitqueue_head(&sess->state_wq); + sess->state = RTRS_CLT_CONNECTING; + atomic_set(&sess->connected_cnt, 0); + INIT_WORK(&sess->close_work, rtrs_clt_close_work); + INIT_DELAYED_WORK(&sess->reconnect_dwork, rtrs_clt_reconnect_work); + rtrs_clt_init_hb(sess); + + sess->mp_skip_entry = alloc_percpu(typeof(*sess->mp_skip_entry)); + if (!sess->mp_skip_entry) + goto err_free_stats; + + for_each_possible_cpu(cpu) + INIT_LIST_HEAD(per_cpu_ptr(sess->mp_skip_entry, cpu)); + + err = rtrs_clt_init_stats(sess->stats); + if (err) + goto err_free_percpu; + + return sess; + +err_free_percpu: + free_percpu(sess->mp_skip_entry); +err_free_stats: + kfree(sess->stats); +err_free_con: + kfree(sess->s.con); +err_free_sess: + kfree(sess); +err: + return ERR_PTR(err); +} + +void free_sess(struct rtrs_clt_sess *sess) +{ + free_percpu(sess->mp_skip_entry); + mutex_destroy(&sess->init_mutex); + kfree(sess->s.con); + kfree(sess->rbufs); + kfree(sess); +} + +static int create_con(struct rtrs_clt_sess *sess, unsigned int cid) +{ + struct rtrs_clt_con *con; + + con = kzalloc(sizeof(*con), GFP_KERNEL); + if (!con) + return -ENOMEM; + + /* Map first two connections to the first CPU */ + con->cpu = (cid ? cid - 1 : 0) % nr_cpu_ids; + con->c.cid = cid; + con->c.sess = &sess->s; + atomic_set(&con->io_cnt, 0); + + sess->s.con[cid] = &con->c; + + return 0; +} + +static void destroy_con(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + + sess->s.con[con->c.cid] = NULL; + kfree(con); +} + +static int create_con_cq_qp(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + u16 wr_queue_size; + int err, cq_vector; + struct rtrs_msg_rkey_rsp *rsp; + + /* + * This function can fail, but still destroy_con_cq_qp() should + * be called, this is because create_con_cq_qp() is called on cm + * event path, thus caller/waiter never knows: have we failed before + * create_con_cq_qp() or after. To solve this dilemma without + * creating any additional flags just allow destroy_con_cq_qp() be + * called many times. + */ + + if (con->c.cid == 0) { + /* + * One completion for each receive and two for each send + * (send request + registration) + * + 2 for drain and heartbeat + * in case qp gets into error state + */ + wr_queue_size = SERVICE_CON_QUEUE_DEPTH * 3 + 2; + /* We must be the first here */ + if (WARN_ON(sess->s.dev)) + return -EINVAL; + + /* + * The whole session uses device from user connection. + * Be careful not to close user connection before ib dev + * is gracefully put. + */ + sess->s.dev = rtrs_ib_dev_find_or_add(con->c.cm_id->device, + &dev_pd); + if (!sess->s.dev) { + rtrs_wrn(sess->clt, + "rtrs_ib_dev_find_get_or_add(): no memory\n"); + return -ENOMEM; + } + sess->s.dev_ref = 1; + query_fast_reg_mode(sess); + } else { + /* + * Here we assume that session members are correctly set. + * This is always true if user connection (cid == 0) is + * established first. + */ + if (WARN_ON(!sess->s.dev)) + return -EINVAL; + if (WARN_ON(!sess->queue_depth)) + return -EINVAL; + + /* Shared between connections */ + sess->s.dev_ref++; + wr_queue_size = + min_t(int, sess->s.dev->ib_dev->attrs.max_qp_wr, + /* QD * (REQ + RSP + FR REGS or INVS) + drain */ + sess->queue_depth * 3 + 1); + } + /* alloc iu to recv new rkey reply when server reports flags set */ + if (sess->flags == RTRS_MSG_NEW_RKEY_F || con->c.cid == 0) { + con->rsp_ius = rtrs_iu_alloc(wr_queue_size, sizeof(*rsp), + GFP_KERNEL, sess->s.dev->ib_dev, + DMA_FROM_DEVICE, + rtrs_clt_rdma_done); + if (!con->rsp_ius) + return -ENOMEM; + con->queue_size = wr_queue_size; + } + cq_vector = con->cpu % sess->s.dev->ib_dev->num_comp_vectors; + err = rtrs_cq_qp_create(&sess->s, &con->c, sess->max_send_sge, + cq_vector, wr_queue_size, wr_queue_size, + IB_POLL_SOFTIRQ); + /* + * In case of error we do not bother to clean previous allocations, + * since destroy_con_cq_qp() must be called. + */ + return err; +} + +static void destroy_con_cq_qp(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + + /* + * Be careful here: destroy_con_cq_qp() can be called even + * create_con_cq_qp() failed, see comments there. + */ + + rtrs_cq_qp_destroy(&con->c); + if (con->rsp_ius) { + rtrs_iu_free(con->rsp_ius, DMA_FROM_DEVICE, + sess->s.dev->ib_dev, con->queue_size); + con->rsp_ius = NULL; + con->queue_size = 0; + } + if (sess->s.dev_ref && !--sess->s.dev_ref) { + rtrs_ib_dev_put(sess->s.dev); + sess->s.dev = NULL; + } +} + +static void stop_cm(struct rtrs_clt_con *con) +{ + rdma_disconnect(con->c.cm_id); + if (con->c.qp) + ib_drain_qp(con->c.qp); +} + +static void destroy_cm(struct rtrs_clt_con *con) +{ + rdma_destroy_id(con->c.cm_id); + con->c.cm_id = NULL; +} + +static int rtrs_rdma_addr_resolved(struct rtrs_clt_con *con) +{ + struct rtrs_sess *s = con->c.sess; + int err; + + err = create_con_cq_qp(con); + if (err) { + rtrs_err(s, "create_con_cq_qp(), err: %d\n", err); + return err; + } + err = rdma_resolve_route(con->c.cm_id, RTRS_CONNECT_TIMEOUT_MS); + if (err) { + rtrs_err(s, "Resolving route failed, err: %d\n", err); + destroy_con_cq_qp(con); + } + + return err; +} + +static int rtrs_rdma_route_resolved(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt *clt = sess->clt; + struct rtrs_msg_conn_req msg; + struct rdma_conn_param param; + + int err; + + param = (struct rdma_conn_param) { + .retry_count = 7, + .rnr_retry_count = 7, + .private_data = &msg, + .private_data_len = sizeof(msg), + }; + + msg = (struct rtrs_msg_conn_req) { + .magic = cpu_to_le16(RTRS_MAGIC), + .version = cpu_to_le16(RTRS_PROTO_VER), + .cid = cpu_to_le16(con->c.cid), + .cid_num = cpu_to_le16(sess->s.con_num), + .recon_cnt = cpu_to_le16(sess->s.recon_cnt), + }; + uuid_copy(&msg.sess_uuid, &sess->s.uuid); + uuid_copy(&msg.paths_uuid, &clt->paths_uuid); + + err = rdma_connect(con->c.cm_id, ¶m); + if (err) + rtrs_err(clt, "rdma_connect(): %d\n", err); + + return err; +} + +static int rtrs_rdma_conn_established(struct rtrs_clt_con *con, + struct rdma_cm_event *ev) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_clt *clt = sess->clt; + const struct rtrs_msg_conn_rsp *msg; + u16 version, queue_depth; + int errno; + u8 len; + + msg = ev->param.conn.private_data; + len = ev->param.conn.private_data_len; + if (len < sizeof(*msg)) { + rtrs_err(clt, "Invalid RTRS connection response\n"); + return -ECONNRESET; + } + if (le16_to_cpu(msg->magic) != RTRS_MAGIC) { + rtrs_err(clt, "Invalid RTRS magic\n"); + return -ECONNRESET; + } + version = le16_to_cpu(msg->version); + if (version >> 8 != RTRS_PROTO_VER_MAJOR) { + rtrs_err(clt, "Unsupported major RTRS version: %d, expected %d\n", + version >> 8, RTRS_PROTO_VER_MAJOR); + return -ECONNRESET; + } + errno = le16_to_cpu(msg->errno); + if (errno) { + rtrs_err(clt, "Invalid RTRS message: errno %d\n", + errno); + return -ECONNRESET; + } + if (con->c.cid == 0) { + queue_depth = le16_to_cpu(msg->queue_depth); + + if (queue_depth > MAX_SESS_QUEUE_DEPTH) { + rtrs_err(clt, "Invalid RTRS message: queue=%d\n", + queue_depth); + return -ECONNRESET; + } + if (!sess->rbufs || sess->queue_depth < queue_depth) { + kfree(sess->rbufs); + sess->rbufs = kcalloc(queue_depth, sizeof(*sess->rbufs), + GFP_KERNEL); + if (!sess->rbufs) + return -ENOMEM; + } + sess->queue_depth = queue_depth; + sess->max_hdr_size = le32_to_cpu(msg->max_hdr_size); + sess->max_io_size = le32_to_cpu(msg->max_io_size); + sess->flags = le32_to_cpu(msg->flags); + sess->chunk_size = sess->max_io_size + sess->max_hdr_size; + + /* + * Global queue depth and IO size is always a minimum. + * If while a reconnection server sends us a value a bit + * higher - client does not care and uses cached minimum. + * + * Since we can have several sessions (paths) restablishing + * connections in parallel, use lock. + */ + mutex_lock(&clt->paths_mutex); + clt->queue_depth = min_not_zero(sess->queue_depth, + clt->queue_depth); + clt->max_io_size = min_not_zero(sess->max_io_size, + clt->max_io_size); + mutex_unlock(&clt->paths_mutex); + + /* + * Cache the hca_port and hca_name for sysfs + */ + sess->hca_port = con->c.cm_id->port_num; + scnprintf(sess->hca_name, sizeof(sess->hca_name), + sess->s.dev->ib_dev->name); + sess->s.src_addr = con->c.cm_id->route.addr.src_addr; + } + + return 0; +} + +static inline void flag_success_on_conn(struct rtrs_clt_con *con) +{ + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + + atomic_inc(&sess->connected_cnt); + con->cm_err = 1; +} + +static int rtrs_rdma_conn_rejected(struct rtrs_clt_con *con, + struct rdma_cm_event *ev) +{ + struct rtrs_sess *s = con->c.sess; + const struct rtrs_msg_conn_rsp *msg; + const char *rej_msg; + int status, errno; + u8 data_len; + + status = ev->status; + rej_msg = rdma_reject_msg(con->c.cm_id, status); + msg = rdma_consumer_reject_data(con->c.cm_id, ev, &data_len); + + if (msg && data_len >= sizeof(*msg)) { + errno = (int16_t)le16_to_cpu(msg->errno); + if (errno == -EBUSY) + rtrs_err(s, + "Previous session is still exists on the server, please reconnect later\n"); + else + rtrs_err(s, + "Connect rejected: status %d (%s), rtrs errno %d\n", + status, rej_msg, errno); + } else { + rtrs_err(s, + "Connect rejected but with malformed message: status %d (%s)\n", + status, rej_msg); + } + + return -ECONNRESET; +} + +static void rtrs_clt_close_conns(struct rtrs_clt_sess *sess, bool wait) +{ + if (rtrs_clt_change_state(sess, RTRS_CLT_CLOSING)) + queue_work(rtrs_wq, &sess->close_work); + if (wait) + flush_work(&sess->close_work); +} + +static inline void flag_error_on_conn(struct rtrs_clt_con *con, int cm_err) +{ + if (con->cm_err == 1) { + struct rtrs_clt_sess *sess; + + sess = to_clt_sess(con->c.sess); + if (atomic_dec_and_test(&sess->connected_cnt)) + + wake_up(&sess->state_wq); + } + con->cm_err = cm_err; +} + +static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id, + struct rdma_cm_event *ev) +{ + struct rtrs_clt_con *con = cm_id->context; + struct rtrs_sess *s = con->c.sess; + struct rtrs_clt_sess *sess = to_clt_sess(s); + int cm_err = 0; + + switch (ev->event) { + case RDMA_CM_EVENT_ADDR_RESOLVED: + cm_err = rtrs_rdma_addr_resolved(con); + break; + case RDMA_CM_EVENT_ROUTE_RESOLVED: + cm_err = rtrs_rdma_route_resolved(con); + break; + case RDMA_CM_EVENT_ESTABLISHED: + con->cm_err = rtrs_rdma_conn_established(con, ev); + if (likely(!con->cm_err)) { + /* + * Report success and wake up. Here we abuse state_wq, + * i.e. wake up without state change, but we set cm_err. + */ + flag_success_on_conn(con); + wake_up(&sess->state_wq); + return 0; + } + break; + case RDMA_CM_EVENT_REJECTED: + cm_err = rtrs_rdma_conn_rejected(con, ev); + break; + case RDMA_CM_EVENT_CONNECT_ERROR: + case RDMA_CM_EVENT_UNREACHABLE: + rtrs_wrn(s, "CM error event %d\n", ev->event); + cm_err = -ECONNRESET; + break; + case RDMA_CM_EVENT_ADDR_ERROR: + case RDMA_CM_EVENT_ROUTE_ERROR: + cm_err = -EHOSTUNREACH; + break; + case RDMA_CM_EVENT_DISCONNECTED: + case RDMA_CM_EVENT_ADDR_CHANGE: + case RDMA_CM_EVENT_TIMEWAIT_EXIT: + cm_err = -ECONNRESET; + break; + case RDMA_CM_EVENT_DEVICE_REMOVAL: + /* + * Device removal is a special case. Queue close and return 0. + */ + rtrs_clt_close_conns(sess, false); + return 0; + default: + rtrs_err(s, "Unexpected RDMA CM event (%d)\n", ev->event); + cm_err = -ECONNRESET; + break; + } + + if (cm_err) { + /* + * cm error makes sense only on connection establishing, + * in other cases we rely on normal procedure of reconnecting. + */ + flag_error_on_conn(con, cm_err); + rtrs_rdma_error_recovery(con); + } + + return 0; +} + +static int create_cm(struct rtrs_clt_con *con) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_clt_sess *sess = to_clt_sess(s); + struct rdma_cm_id *cm_id; + int err; + + cm_id = rdma_create_id(&init_net, rtrs_clt_rdma_cm_handler, con, + sess->s.dst_addr.ss_family == AF_IB ? + RDMA_PS_IB : RDMA_PS_TCP, IB_QPT_RC); + if (IS_ERR(cm_id)) { + err = PTR_ERR(cm_id); + rtrs_err(s, "Failed to create CM ID, err: %d\n", err); + + return err; + } + con->c.cm_id = cm_id; + con->cm_err = 0; + /* allow the port to be reused */ + err = rdma_set_reuseaddr(cm_id, 1); + if (err != 0) { + rtrs_err(s, "Set address reuse failed, err: %d\n", err); + goto destroy_cm; + } + err = rdma_resolve_addr(cm_id, (struct sockaddr *)&sess->s.src_addr, + (struct sockaddr *)&sess->s.dst_addr, + RTRS_CONNECT_TIMEOUT_MS); + if (err) { + rtrs_err(s, "Failed to resolve address, err: %d\n", err); + goto destroy_cm; + } + /* + * Combine connection status and session events. This is needed + * for waiting two possible cases: cm_err has something meaningful + * or session state was really changed to error by device removal. + */ + err = wait_event_interruptible_timeout( + sess->state_wq, + con->cm_err || sess->state != RTRS_CLT_CONNECTING, + msecs_to_jiffies(RTRS_CONNECT_TIMEOUT_MS)); + if (err == 0 || err == -ERESTARTSYS) { + if (err == 0) + err = -ETIMEDOUT; + /* Timedout or interrupted */ + goto errr; + } + if (con->cm_err < 0) { + err = con->cm_err; + goto errr; + } + if (READ_ONCE(sess->state) != RTRS_CLT_CONNECTING) { + /* Device removal */ + err = -ECONNABORTED; + goto errr; + } + + return 0; + +errr: + stop_cm(con); + /* Is safe to call destroy if cq_qp is not inited */ + destroy_con_cq_qp(con); +destroy_cm: + destroy_cm(con); + + return err; +} + +static void rtrs_clt_sess_up(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt *clt = sess->clt; + int up; + + /* + * We can fire RECONNECTED event only when all paths were + * connected on rtrs_clt_open(), then each was disconnected + * and the first one connected again. That's why this nasty + * game with counter value. + */ + + mutex_lock(&clt->paths_ev_mutex); + up = ++clt->paths_up; + /* + * Here it is safe to access paths num directly since up counter + * is greater than MAX_PATHS_NUM only while rtrs_clt_open() is + * in progress, thus paths removals are impossible. + */ + if (up > MAX_PATHS_NUM && up == MAX_PATHS_NUM + clt->paths_num) + clt->paths_up = clt->paths_num; + else if (up == 1) + clt->link_ev(clt->priv, RTRS_CLT_LINK_EV_RECONNECTED); + mutex_unlock(&clt->paths_ev_mutex); + + /* Mark session as established */ + sess->established = true; + sess->reconnect_attempts = 0; + sess->stats->reconnects.successful_cnt++; +} + +static void rtrs_clt_sess_down(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt *clt = sess->clt; + + if (!sess->established) + return; + + sess->established = false; + mutex_lock(&clt->paths_ev_mutex); + WARN_ON(!clt->paths_up); + if (--clt->paths_up == 0) + clt->link_ev(clt->priv, RTRS_CLT_LINK_EV_DISCONNECTED); + mutex_unlock(&clt->paths_ev_mutex); +} + +static void rtrs_clt_stop_and_destroy_conns(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt_con *con; + unsigned int cid; + + WARN_ON(READ_ONCE(sess->state) == RTRS_CLT_CONNECTED); + + /* + * Possible race with rtrs_clt_open(), when DEVICE_REMOVAL comes + * exactly in between. Start destroying after it finishes. + */ + mutex_lock(&sess->init_mutex); + mutex_unlock(&sess->init_mutex); + + /* + * All IO paths must observe !CONNECTED state before we + * free everything. + */ + synchronize_rcu(); + + rtrs_clt_stop_hb(sess); + + /* + * The order it utterly crucial: firstly disconnect and complete all + * rdma requests with error (thus set in_use=false for requests), + * then fail outstanding requests checking in_use for each, and + * eventually notify upper layer about session disconnection. + */ + + for (cid = 0; cid < sess->s.con_num; cid++) { + if (!sess->s.con[cid]) + break; + con = to_clt_con(sess->s.con[cid]); + stop_cm(con); + } + fail_all_outstanding_reqs(sess); + free_sess_reqs(sess); + rtrs_clt_sess_down(sess); + + /* + * Wait for graceful shutdown, namely when peer side invokes + * rdma_disconnect(). 'connected_cnt' is decremented only on + * CM events, thus if other side had crashed and hb has detected + * something is wrong, here we will stuck for exactly timeout ms, + * since CM does not fire anything. That is fine, we are not in + * hurry. + */ + wait_event_timeout(sess->state_wq, !atomic_read(&sess->connected_cnt), + msecs_to_jiffies(RTRS_CONNECT_TIMEOUT_MS)); + + for (cid = 0; cid < sess->s.con_num; cid++) { + if (!sess->s.con[cid]) + break; + con = to_clt_con(sess->s.con[cid]); + destroy_con_cq_qp(con); + destroy_cm(con); + destroy_con(con); + } +} + +static inline bool xchg_sessions(struct rtrs_clt_sess __rcu **rcu_ppcpu_path, + struct rtrs_clt_sess *sess, + struct rtrs_clt_sess *next) +{ + struct rtrs_clt_sess **ppcpu_path; + + /* Call cmpxchg() without sparse warnings */ + ppcpu_path = (typeof(ppcpu_path))rcu_ppcpu_path; + return sess == cmpxchg(ppcpu_path, sess, next); +} + +static void rtrs_clt_remove_path_from_arr(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt *clt = sess->clt; + struct rtrs_clt_sess *next; + bool wait_for_grace = false; + int cpu; + + mutex_lock(&clt->paths_mutex); + list_del_rcu(&sess->s.entry); + + /* Make sure everybody observes path removal. */ + synchronize_rcu(); + + /* + * At this point nobody sees @sess in the list, but still we have + * dangling pointer @pcpu_path which _can_ point to @sess. Since + * nobody can observe @sess in the list, we guarantee that IO path + * will not assign @sess to @pcpu_path, i.e. @pcpu_path can be equal + * to @sess, but can never again become @sess. + */ + + /* + * Decrement paths number only after grace period, because + * caller of do_each_path() must firstly observe list without + * path and only then decremented paths number. + * + * Otherwise there can be the following situation: + * o Two paths exist and IO is coming. + * o One path is removed: + * CPU#0 CPU#1 + * do_each_path(): rtrs_clt_remove_path_from_arr(): + * path = get_next_path() + * ^^^ list_del_rcu(path) + * [!CONNECTED path] clt->paths_num-- + * ^^^^^^^^^ + * load clt->paths_num from 2 to 1 + * ^^^^^^^^^ + * sees 1 + * + * path is observed as !CONNECTED, but do_each_path() loop + * ends, because expression i < clt->paths_num is false. + */ + clt->paths_num--; + + /* + * Get @next connection from current @sess which is going to be + * removed. If @sess is the last element, then @next is NULL. + */ + rcu_read_lock(); + next = list_next_or_null_rr_rcu(&clt->paths_list, &sess->s.entry, + typeof(*next), s.entry); + rcu_read_unlock(); + + /* + * @pcpu paths can still point to the path which is going to be + * removed, so change the pointer manually. + */ + for_each_possible_cpu(cpu) { + struct rtrs_clt_sess __rcu **ppcpu_path; + + ppcpu_path = per_cpu_ptr(clt->pcpu_path, cpu); + if (rcu_dereference_protected(*ppcpu_path, + lockdep_is_held(&clt->paths_mutex)) != sess) + /* + * synchronize_rcu() was called just after deleting + * entry from the list, thus IO code path cannot + * change pointer back to the pointer which is going + * to be removed, we are safe here. + */ + continue; + + /* + * We race with IO code path, which also changes pointer, + * thus we have to be careful not to overwrite it. + */ + if (xchg_sessions(ppcpu_path, sess, next)) + /* + * @ppcpu_path was successfully replaced with @next, + * that means that someone could also pick up the + * @sess and dereferencing it right now, so wait for + * a grace period is required. + */ + wait_for_grace = true; + } + if (wait_for_grace) + synchronize_rcu(); + + mutex_unlock(&clt->paths_mutex); +} + +static void rtrs_clt_add_path_to_arr(struct rtrs_clt_sess *sess, + struct rtrs_addr *addr) +{ + struct rtrs_clt *clt = sess->clt; + + mutex_lock(&clt->paths_mutex); + clt->paths_num++; + + list_add_tail_rcu(&sess->s.entry, &clt->paths_list); + mutex_unlock(&clt->paths_mutex); +} + +static void rtrs_clt_close_work(struct work_struct *work) +{ + struct rtrs_clt_sess *sess; + + sess = container_of(work, struct rtrs_clt_sess, close_work); + + cancel_delayed_work_sync(&sess->reconnect_dwork); + rtrs_clt_stop_and_destroy_conns(sess); + rtrs_clt_change_state(sess, RTRS_CLT_CLOSED); +} + +static int init_conns(struct rtrs_clt_sess *sess) +{ + unsigned int cid; + int err; + + /* + * On every new session connections increase reconnect counter + * to avoid clashes with previous sessions not yet closed + * sessions on a server side. + */ + sess->s.recon_cnt++; + + /* Establish all RDMA connections */ + for (cid = 0; cid < sess->s.con_num; cid++) { + err = create_con(sess, cid); + if (err) + goto destroy; + + err = create_cm(to_clt_con(sess->s.con[cid])); + if (err) { + destroy_con(to_clt_con(sess->s.con[cid])); + goto destroy; + } + } + err = alloc_sess_reqs(sess); + if (err) + goto destroy; + + rtrs_clt_start_hb(sess); + + return 0; + +destroy: + while (cid--) { + struct rtrs_clt_con *con = to_clt_con(sess->s.con[cid]); + + stop_cm(con); + destroy_con_cq_qp(con); + destroy_cm(con); + destroy_con(con); + } + /* + * If we've never taken async path and got an error, say, + * doing rdma_resolve_addr(), switch to CONNECTION_ERR state + * manually to keep reconnecting. + */ + rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR); + + return err; +} + +static void rtrs_clt_info_req_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_clt_con *con = cq->cq_context; + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_iu *iu; + + iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); + rtrs_iu_free(iu, DMA_TO_DEVICE, sess->s.dev->ib_dev, 1); + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(sess->clt, "Sess info request send failed: %s\n", + ib_wc_status_msg(wc->status)); + rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR); + return; + } + + rtrs_clt_update_wc_stats(con); +} + +static int process_info_rsp(struct rtrs_clt_sess *sess, + const struct rtrs_msg_info_rsp *msg) +{ + unsigned int sg_cnt, total_len; + int i, sgi; + + sg_cnt = le16_to_cpu(msg->sg_cnt); + if (unlikely(!sg_cnt)) + return -EINVAL; + /* + * Check if IB immediate data size is enough to hold the mem_id and + * the offset inside the memory chunk. + */ + if (unlikely((ilog2(sg_cnt - 1) + 1) + + (ilog2(sess->chunk_size - 1) + 1) > + MAX_IMM_PAYL_BITS)) { + rtrs_err(sess->clt, + "RDMA immediate size (%db) not enough to encode %d buffers of size %dB\n", + MAX_IMM_PAYL_BITS, sg_cnt, sess->chunk_size); + return -EINVAL; + } + if (unlikely(!sg_cnt || (sess->queue_depth % sg_cnt))) { + rtrs_err(sess->clt, "Incorrect sg_cnt %d, is not multiple\n", + sg_cnt); + return -EINVAL; + } + total_len = 0; + for (sgi = 0, i = 0; sgi < sg_cnt && i < sess->queue_depth; sgi++) { + const struct rtrs_sg_desc *desc = &msg->desc[sgi]; + u32 len, rkey; + u64 addr; + + addr = le64_to_cpu(desc->addr); + rkey = le32_to_cpu(desc->key); + len = le32_to_cpu(desc->len); + + total_len += len; + + if (unlikely(!len || (len % sess->chunk_size))) { + rtrs_err(sess->clt, "Incorrect [%d].len %d\n", sgi, + len); + return -EINVAL; + } + for ( ; len && i < sess->queue_depth; i++) { + sess->rbufs[i].addr = addr; + sess->rbufs[i].rkey = rkey; + + len -= sess->chunk_size; + addr += sess->chunk_size; + } + } + /* Sanity check */ + if (unlikely(sgi != sg_cnt || i != sess->queue_depth)) { + rtrs_err(sess->clt, "Incorrect sg vector, not fully mapped\n"); + return -EINVAL; + } + if (unlikely(total_len != sess->chunk_size * sess->queue_depth)) { + rtrs_err(sess->clt, "Incorrect total_len %d\n", total_len); + return -EINVAL; + } + + return 0; +} + +static void rtrs_clt_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_clt_con *con = cq->cq_context; + struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); + struct rtrs_msg_info_rsp *msg; + enum rtrs_clt_state state; + struct rtrs_iu *iu; + size_t rx_sz; + int err; + + state = RTRS_CLT_CONNECTING_ERR; + + WARN_ON(con->c.cid); + iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(sess->clt, "Sess info response recv failed: %s\n", + ib_wc_status_msg(wc->status)); + goto out; + } + WARN_ON(wc->opcode != IB_WC_RECV); + + if (unlikely(wc->byte_len < sizeof(*msg))) { + rtrs_err(sess->clt, "Sess info response is malformed: size %d\n", + wc->byte_len); + goto out; + } + ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr, + iu->size, DMA_FROM_DEVICE); + msg = iu->buf; + if (unlikely(le16_to_cpu(msg->type) != RTRS_MSG_INFO_RSP)) { + rtrs_err(sess->clt, "Sess info response is malformed: type %d\n", + le16_to_cpu(msg->type)); + goto out; + } + rx_sz = sizeof(*msg); + rx_sz += sizeof(msg->desc[0]) * le16_to_cpu(msg->sg_cnt); + if (unlikely(wc->byte_len < rx_sz)) { + rtrs_err(sess->clt, "Sess info response is malformed: size %d\n", + wc->byte_len); + goto out; + } + err = process_info_rsp(sess, msg); + if (unlikely(err)) + goto out; + + err = post_recv_sess(sess); + if (unlikely(err)) + goto out; + + state = RTRS_CLT_CONNECTED; + +out: + rtrs_clt_update_wc_stats(con); + rtrs_iu_free(iu, DMA_FROM_DEVICE, sess->s.dev->ib_dev, 1); + rtrs_clt_change_state(sess, state); +} + +static int rtrs_send_sess_info(struct rtrs_clt_sess *sess) +{ + struct rtrs_clt_con *usr_con = to_clt_con(sess->s.con[0]); + struct rtrs_msg_info_req *msg; + struct rtrs_iu *tx_iu, *rx_iu; + size_t rx_sz; + int err; + + rx_sz = sizeof(struct rtrs_msg_info_rsp); + rx_sz += sizeof(u64) * MAX_SESS_QUEUE_DEPTH; + + tx_iu = rtrs_iu_alloc(1, sizeof(struct rtrs_msg_info_req), GFP_KERNEL, + sess->s.dev->ib_dev, DMA_TO_DEVICE, + rtrs_clt_info_req_done); + rx_iu = rtrs_iu_alloc(1, rx_sz, GFP_KERNEL, sess->s.dev->ib_dev, + DMA_FROM_DEVICE, rtrs_clt_info_rsp_done); + if (unlikely(!tx_iu || !rx_iu)) { + err = -ENOMEM; + goto out; + } + /* Prepare for getting info response */ + err = rtrs_iu_post_recv(&usr_con->c, rx_iu); + if (unlikely(err)) { + rtrs_err(sess->clt, "rtrs_iu_post_recv(), err: %d\n", err); + goto out; + } + rx_iu = NULL; + + msg = tx_iu->buf; + msg->type = cpu_to_le16(RTRS_MSG_INFO_REQ); + memcpy(msg->sessname, sess->s.sessname, sizeof(msg->sessname)); + + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, tx_iu->dma_addr, + tx_iu->size, DMA_TO_DEVICE); + + /* Send info request */ + err = rtrs_iu_post_send(&usr_con->c, tx_iu, sizeof(*msg), NULL); + if (unlikely(err)) { + rtrs_err(sess->clt, "rtrs_iu_post_send(), err: %d\n", err); + goto out; + } + tx_iu = NULL; + + /* Wait for state change */ + wait_event_interruptible_timeout(sess->state_wq, + sess->state != RTRS_CLT_CONNECTING, + msecs_to_jiffies( + RTRS_CONNECT_TIMEOUT_MS)); + if (unlikely(READ_ONCE(sess->state) != RTRS_CLT_CONNECTED)) { + if (READ_ONCE(sess->state) == RTRS_CLT_CONNECTING_ERR) + err = -ECONNRESET; + else + err = -ETIMEDOUT; + goto out; + } + +out: + if (tx_iu) + rtrs_iu_free(tx_iu, DMA_TO_DEVICE, sess->s.dev->ib_dev, 1); + if (rx_iu) + rtrs_iu_free(rx_iu, DMA_FROM_DEVICE, sess->s.dev->ib_dev, 1); + if (unlikely(err)) + /* If we've never taken async path because of malloc problems */ + rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR); + + return err; +} + +/** + * init_sess() - establishes all session connections and does handshake + * @sess: client session. + * In case of error full close or reconnect procedure should be taken, + * because reconnect or close async works can be started. + */ +static int init_sess(struct rtrs_clt_sess *sess) +{ + int err; + + mutex_lock(&sess->init_mutex); + err = init_conns(sess); + if (err) { + rtrs_err(sess->clt, "init_conns(), err: %d\n", err); + goto out; + } + err = rtrs_send_sess_info(sess); + if (err) { + rtrs_err(sess->clt, "rtrs_send_sess_info(), err: %d\n", err); + goto out; + } + rtrs_clt_sess_up(sess); +out: + mutex_unlock(&sess->init_mutex); + + return err; +} + +static void rtrs_clt_reconnect_work(struct work_struct *work) +{ + struct rtrs_clt_sess *sess; + struct rtrs_clt *clt; + unsigned int delay_ms; + int err; + + sess = container_of(to_delayed_work(work), struct rtrs_clt_sess, + reconnect_dwork); + clt = sess->clt; + + if (READ_ONCE(sess->state) != RTRS_CLT_RECONNECTING) + return; + + if (sess->reconnect_attempts >= clt->max_reconnect_attempts) { + /* Close a session completely if max attempts is reached */ + rtrs_clt_close_conns(sess, false); + return; + } + sess->reconnect_attempts++; + + /* Stop everything */ + rtrs_clt_stop_and_destroy_conns(sess); + msleep(RTRS_RECONNECT_BACKOFF); + if (rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING)) { + err = init_sess(sess); + if (err) + goto reconnect_again; + } + + return; + +reconnect_again: + if (rtrs_clt_change_state(sess, RTRS_CLT_RECONNECTING)) { + sess->stats->reconnects.fail_cnt++; + delay_ms = clt->reconnect_delay_sec * 1000; + queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, + msecs_to_jiffies(delay_ms + + prandom_u32() % + RTRS_RECONNECT_SEED)); + } +} + +static void rtrs_clt_dev_release(struct device *dev) +{ + struct rtrs_clt *clt = container_of(dev, struct rtrs_clt, dev); + + kfree(clt); +} + +static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num, + u16 port, size_t pdu_sz, void *priv, + void (*link_ev)(void *priv, + enum rtrs_clt_link_ev ev), + unsigned int max_segments, + size_t max_segment_size, + unsigned int reconnect_delay_sec, + unsigned int max_reconnect_attempts) +{ + struct rtrs_clt *clt; + int err; + + if (!paths_num || paths_num > MAX_PATHS_NUM) + return ERR_PTR(-EINVAL); + + if (strlen(sessname) >= sizeof(clt->sessname)) + return ERR_PTR(-EINVAL); + + clt = kzalloc(sizeof(*clt), GFP_KERNEL); + if (!clt) + return ERR_PTR(-ENOMEM); + + clt->pcpu_path = alloc_percpu(typeof(*clt->pcpu_path)); + if (!clt->pcpu_path) { + kfree(clt); + return ERR_PTR(-ENOMEM); + } + + uuid_gen(&clt->paths_uuid); + INIT_LIST_HEAD_RCU(&clt->paths_list); + clt->paths_num = paths_num; + clt->paths_up = MAX_PATHS_NUM; + clt->port = port; + clt->pdu_sz = pdu_sz; + clt->max_segments = max_segments; + clt->max_segment_size = max_segment_size; + clt->reconnect_delay_sec = reconnect_delay_sec; + clt->max_reconnect_attempts = max_reconnect_attempts; + clt->priv = priv; + clt->link_ev = link_ev; + clt->mp_policy = MP_POLICY_MIN_INFLIGHT; + strlcpy(clt->sessname, sessname, sizeof(clt->sessname)); + init_waitqueue_head(&clt->permits_wait); + mutex_init(&clt->paths_ev_mutex); + mutex_init(&clt->paths_mutex); + + clt->dev.class = rtrs_clt_dev_class; + clt->dev.release = rtrs_clt_dev_release; + err = dev_set_name(&clt->dev, "%s", sessname); + if (err) { + free_percpu(clt->pcpu_path); + kfree(clt); + return ERR_PTR(err); + } + /* + * Suppress user space notification until + * sysfs files are created + */ + dev_set_uevent_suppress(&clt->dev, true); + err = device_register(&clt->dev); + if (err) { + free_percpu(clt->pcpu_path); + put_device(&clt->dev); + return ERR_PTR(err); + } + + clt->kobj_paths = kobject_create_and_add("paths", &clt->dev.kobj); + if (!clt->kobj_paths) { + free_percpu(clt->pcpu_path); + device_unregister(&clt->dev); + return NULL; + } + err = rtrs_clt_create_sysfs_root_files(clt); + if (err) { + free_percpu(clt->pcpu_path); + kobject_del(clt->kobj_paths); + kobject_put(clt->kobj_paths); + device_unregister(&clt->dev); + return ERR_PTR(err); + } + dev_set_uevent_suppress(&clt->dev, false); + kobject_uevent(&clt->dev.kobj, KOBJ_ADD); + + return clt; +} + +static void wait_for_inflight_permits(struct rtrs_clt *clt) +{ + if (clt->permits_map) { + size_t sz = clt->queue_depth; + + wait_event(clt->permits_wait, + find_first_bit(clt->permits_map, sz) >= sz); + } +} + +static void free_clt(struct rtrs_clt *clt) +{ + wait_for_inflight_permits(clt); + free_permits(clt); + free_percpu(clt->pcpu_path); + mutex_destroy(&clt->paths_ev_mutex); + mutex_destroy(&clt->paths_mutex); + /* release callback will free clt in last put */ + device_unregister(&clt->dev); +} + +/** + * rtrs_clt_open() - Open a session to an RTRS server + * @ops: holds the link event callback and the private pointer. + * @sessname: name of the session + * @paths: Paths to be established defined by their src and dst addresses + * @paths_num: Number of elements in the @paths array + * @port: port to be used by the RTRS session + * @pdu_sz: Size of extra payload which can be accessed after permit allocation. + * @reconnect_delay_sec: time between reconnect tries + * @max_segments: Max. number of segments per IO request + * @max_segment_size: Max. size of one segment + * @max_reconnect_attempts: Number of times to reconnect on error before giving + * up, 0 for * disabled, -1 for forever + * + * Starts session establishment with the rtrs_server. The function can block + * up to ~2000ms before it returns. + * + * Return a valid pointer on success otherwise PTR_ERR. + */ +struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops, + const char *sessname, + const struct rtrs_addr *paths, + size_t paths_num, u16 port, + size_t pdu_sz, u8 reconnect_delay_sec, + u16 max_segments, + size_t max_segment_size, + s16 max_reconnect_attempts) +{ + struct rtrs_clt_sess *sess, *tmp; + struct rtrs_clt *clt; + int err, i; + + clt = alloc_clt(sessname, paths_num, port, pdu_sz, ops->priv, + ops->link_ev, + max_segments, max_segment_size, reconnect_delay_sec, + max_reconnect_attempts); + if (IS_ERR(clt)) { + err = PTR_ERR(clt); + goto out; + } + for (i = 0; i < paths_num; i++) { + struct rtrs_clt_sess *sess; + + sess = alloc_sess(clt, &paths[i], nr_cpu_ids, + max_segments, max_segment_size); + if (IS_ERR(sess)) { + err = PTR_ERR(sess); + goto close_all_sess; + } + list_add_tail_rcu(&sess->s.entry, &clt->paths_list); + + err = init_sess(sess); + if (err) { + list_del_rcu(&sess->s.entry); + rtrs_clt_close_conns(sess, true); + free_sess(sess); + goto close_all_sess; + } + + err = rtrs_clt_create_sess_files(sess); + if (err) { + list_del_rcu(&sess->s.entry); + rtrs_clt_close_conns(sess, true); + free_sess(sess); + goto close_all_sess; + } + } + err = alloc_permits(clt); + if (err) + goto close_all_sess; + + return clt; + +close_all_sess: + list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) { + rtrs_clt_destroy_sess_files(sess, NULL); + rtrs_clt_close_conns(sess, true); + kobject_put(&sess->kobj); + } + rtrs_clt_destroy_sysfs_root_files(clt); + rtrs_clt_destroy_sysfs_root_folders(clt); + free_clt(clt); + +out: + return ERR_PTR(err); +} +EXPORT_SYMBOL(rtrs_clt_open); + +/** + * rtrs_clt_close() - Close a session + * @clt: Session handle. Session is freed upon return. + */ +void rtrs_clt_close(struct rtrs_clt *clt) +{ + struct rtrs_clt_sess *sess, *tmp; + + /* Firstly forbid sysfs access */ + rtrs_clt_destroy_sysfs_root_files(clt); + rtrs_clt_destroy_sysfs_root_folders(clt); + + /* Now it is safe to iterate over all paths without locks */ + list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) { + rtrs_clt_destroy_sess_files(sess, NULL); + rtrs_clt_close_conns(sess, true); + kobject_put(&sess->kobj); + } + free_clt(clt); +} +EXPORT_SYMBOL(rtrs_clt_close); + +int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_sess *sess) +{ + enum rtrs_clt_state old_state; + int err = -EBUSY; + bool changed; + + changed = rtrs_clt_change_state_get_old(sess, RTRS_CLT_RECONNECTING, + &old_state); + if (changed) { + sess->reconnect_attempts = 0; + queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, 0); + } + if (changed || old_state == RTRS_CLT_RECONNECTING) { + /* + * flush_delayed_work() queues pending work for immediate + * execution, so do the flush if we have queued something + * right now or work is pending. + */ + flush_delayed_work(&sess->reconnect_dwork); + err = (READ_ONCE(sess->state) == + RTRS_CLT_CONNECTED ? 0 : -ENOTCONN); + } + + return err; +} + +int rtrs_clt_disconnect_from_sysfs(struct rtrs_clt_sess *sess) +{ + rtrs_clt_close_conns(sess, true); + + return 0; +} + +int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_sess *sess, + const struct attribute *sysfs_self) +{ + enum rtrs_clt_state old_state; + bool changed; + + /* + * Continue stopping path till state was changed to DEAD or + * state was observed as DEAD: + * 1. State was changed to DEAD - we were fast and nobody + * invoked rtrs_clt_reconnect(), which can again start + * reconnecting. + * 2. State was observed as DEAD - we have someone in parallel + * removing the path. + */ + do { + rtrs_clt_close_conns(sess, true); + changed = rtrs_clt_change_state_get_old(sess, + RTRS_CLT_DEAD, + &old_state); + } while (!changed && old_state != RTRS_CLT_DEAD); + + if (likely(changed)) { + rtrs_clt_destroy_sess_files(sess, sysfs_self); + rtrs_clt_remove_path_from_arr(sess); + kobject_put(&sess->kobj); + } + + return 0; +} + +void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt *clt, int value) +{ + clt->max_reconnect_attempts = (unsigned int)value; +} + +int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt *clt) +{ + return (int)clt->max_reconnect_attempts; +} + +/** + * rtrs_clt_request() - Request data transfer to/from server via RDMA. + * + * @dir: READ/WRITE + * @ops: callback function to be called as confirmation, and the pointer. + * @clt: Session + * @permit: Preallocated permit + * @vec: Message that is sent to server together with the request. + * Sum of len of all @vec elements limited to <= IO_MSG_SIZE. + * Since the msg is copied internally it can be allocated on stack. + * @nr: Number of elements in @vec. + * @data_len: length of data sent to/from server + * @sg: Pages to be sent/received to/from server. + * @sg_cnt: Number of elements in the @sg + * + * Return: + * 0: Success + * <0: Error + * + * On dir=READ rtrs client will request a data transfer from Server to client. + * The data that the server will respond with will be stored in @sg when + * the user receives an %RTRS_CLT_RDMA_EV_RDMA_REQUEST_WRITE_COMPL event. + * On dir=WRITE rtrs client will rdma write data in sg to server side. + */ +int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops, + struct rtrs_clt *clt, struct rtrs_permit *permit, + const struct kvec *vec, size_t nr, size_t data_len, + struct scatterlist *sg, unsigned int sg_cnt) +{ + struct rtrs_clt_io_req *req; + struct rtrs_clt_sess *sess; + + enum dma_data_direction dma_dir; + int err = -ECONNABORTED, i; + size_t usr_len, hdr_len; + struct path_it it; + + /* Get kvec length */ + for (i = 0, usr_len = 0; i < nr; i++) + usr_len += vec[i].iov_len; + + if (dir == READ) { + hdr_len = sizeof(struct rtrs_msg_rdma_read) + + sg_cnt * sizeof(struct rtrs_sg_desc); + dma_dir = DMA_FROM_DEVICE; + } else { + hdr_len = sizeof(struct rtrs_msg_rdma_write); + dma_dir = DMA_TO_DEVICE; + } + + rcu_read_lock(); + for (path_it_init(&it, clt); + (sess = it.next_path(&it)) && it.i < it.clt->paths_num; it.i++) { + if (unlikely(READ_ONCE(sess->state) != RTRS_CLT_CONNECTED)) + continue; + + if (unlikely(usr_len + hdr_len > sess->max_hdr_size)) { + rtrs_wrn_rl(sess->clt, + "%s request failed, user message size is %zu and header length %zu, but max size is %u\n", + dir == READ ? "Read" : "Write", + usr_len, hdr_len, sess->max_hdr_size); + err = -EMSGSIZE; + break; + } + req = rtrs_clt_get_req(sess, ops->conf_fn, permit, ops->priv, + vec, usr_len, sg, sg_cnt, data_len, + dma_dir); + if (dir == READ) + err = rtrs_clt_read_req(req); + else + err = rtrs_clt_write_req(req); + if (unlikely(err)) { + req->in_use = false; + continue; + } + /* Success path */ + break; + } + path_it_deinit(&it); + rcu_read_unlock(); + + return err; +} +EXPORT_SYMBOL(rtrs_clt_request); + +/** + * rtrs_clt_query() - queries RTRS session attributes + *@clt: session pointer + *@attr: query results for session attributes. + * Returns: + * 0 on success + * -ECOMM no connection to the server + */ +int rtrs_clt_query(struct rtrs_clt *clt, struct rtrs_attrs *attr) +{ + if (!rtrs_clt_is_connected(clt)) + return -ECOMM; + + attr->queue_depth = clt->queue_depth; + attr->max_io_size = clt->max_io_size; + attr->sess_kobj = &clt->dev.kobj; + strlcpy(attr->sessname, clt->sessname, sizeof(attr->sessname)); + + return 0; +} +EXPORT_SYMBOL(rtrs_clt_query); + +int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt, + struct rtrs_addr *addr) +{ + struct rtrs_clt_sess *sess; + int err; + + sess = alloc_sess(clt, addr, nr_cpu_ids, clt->max_segments, + clt->max_segment_size); + if (IS_ERR(sess)) + return PTR_ERR(sess); + + /* + * It is totally safe to add path in CONNECTING state: coming + * IO will never grab it. Also it is very important to add + * path before init, since init fires LINK_CONNECTED event. + */ + rtrs_clt_add_path_to_arr(sess, addr); + + err = init_sess(sess); + if (err) + goto close_sess; + + err = rtrs_clt_create_sess_files(sess); + if (err) + goto close_sess; + + return 0; + +close_sess: + rtrs_clt_remove_path_from_arr(sess); + rtrs_clt_close_conns(sess, true); + free_sess(sess); + + return err; +} + +static int rtrs_clt_ib_dev_init(struct rtrs_ib_dev *dev) +{ + if (!(dev->ib_dev->attrs.device_cap_flags & + IB_DEVICE_MEM_MGT_EXTENSIONS)) { + pr_err("Memory registrations not supported.\n"); + return -ENOTSUPP; + } + + return 0; +} + +static const struct rtrs_rdma_dev_pd_ops dev_pd_ops = { + .init = rtrs_clt_ib_dev_init +}; + +static int __init rtrs_client_init(void) +{ + rtrs_rdma_dev_pd_init(0, &dev_pd); + + rtrs_clt_dev_class = class_create(THIS_MODULE, "rtrs-client"); + if (IS_ERR(rtrs_clt_dev_class)) { + pr_err("Failed to create rtrs-client dev class\n"); + return PTR_ERR(rtrs_clt_dev_class); + } + rtrs_wq = alloc_workqueue("rtrs_client_wq", 0, 0); + if (!rtrs_wq) { + class_destroy(rtrs_clt_dev_class); + return -ENOMEM; + } + + return 0; +} + +static void __exit rtrs_client_exit(void) +{ + destroy_workqueue(rtrs_wq); + class_destroy(rtrs_clt_dev_class); + rtrs_rdma_dev_pd_deinit(&dev_pd); +} + +module_init(rtrs_client_init); +module_exit(rtrs_client_exit); diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.h b/drivers/infiniband/ulp/rtrs/rtrs-clt.h new file mode 100644 index 000000000000..167acd3c90fc --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.h @@ -0,0 +1,252 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ + +#ifndef RTRS_CLT_H +#define RTRS_CLT_H + +#include <linux/device.h> +#include "rtrs-pri.h" + +/** + * enum rtrs_clt_state - Client states. + */ +enum rtrs_clt_state { + RTRS_CLT_CONNECTING, + RTRS_CLT_CONNECTING_ERR, + RTRS_CLT_RECONNECTING, + RTRS_CLT_CONNECTED, + RTRS_CLT_CLOSING, + RTRS_CLT_CLOSED, + RTRS_CLT_DEAD, +}; + +enum rtrs_mp_policy { + MP_POLICY_RR, + MP_POLICY_MIN_INFLIGHT, +}; + +/* see Documentation/ABI/testing/sysfs-class-rtrs-client for details */ +struct rtrs_clt_stats_reconnects { + int successful_cnt; + int fail_cnt; +}; + +/* see Documentation/ABI/testing/sysfs-class-rtrs-client for details */ +struct rtrs_clt_stats_cpu_migr { + atomic_t from; + int to; +}; + +/* stats for Read and write operation. + * see Documentation/ABI/testing/sysfs-class-rtrs-client for details + */ +struct rtrs_clt_stats_rdma { + struct { + u64 cnt; + u64 size_total; + } dir[2]; + + u64 failover_cnt; +}; + +struct rtrs_clt_stats_pcpu { + struct rtrs_clt_stats_cpu_migr cpu_migr; + struct rtrs_clt_stats_rdma rdma; +}; + +struct rtrs_clt_stats { + struct kobject kobj_stats; + struct rtrs_clt_stats_pcpu __percpu *pcpu_stats; + struct rtrs_clt_stats_reconnects reconnects; + atomic_t inflight; +}; + +struct rtrs_clt_con { + struct rtrs_con c; + struct rtrs_iu *rsp_ius; + u32 queue_size; + unsigned int cpu; + atomic_t io_cnt; + int cm_err; +}; + +/** + * rtrs_permit - permits the memory allocation for future RDMA operation. + * Combine with irq pinning to keep IO on same CPU. + */ +struct rtrs_permit { + enum rtrs_clt_con_type con_type; + unsigned int cpu_id; + unsigned int mem_id; + unsigned int mem_off; +}; + +/** + * rtrs_clt_io_req - describes one inflight IO request + */ +struct rtrs_clt_io_req { + struct list_head list; + struct rtrs_iu *iu; + struct scatterlist *sglist; /* list holding user data */ + unsigned int sg_cnt; + unsigned int sg_size; + unsigned int data_len; + unsigned int usr_len; + void *priv; + bool in_use; + struct rtrs_clt_con *con; + struct rtrs_sg_desc *desc; + struct ib_sge *sge; + struct rtrs_permit *permit; + enum dma_data_direction dir; + void (*conf)(void *priv, int errno); + unsigned long start_jiffies; + + struct ib_mr *mr; + struct ib_cqe inv_cqe; + struct completion inv_comp; + int inv_errno; + bool need_inv_comp; + bool need_inv; +}; + +struct rtrs_rbuf { + u64 addr; + u32 rkey; +}; + +struct rtrs_clt_sess { + struct rtrs_sess s; + struct rtrs_clt *clt; + wait_queue_head_t state_wq; + enum rtrs_clt_state state; + atomic_t connected_cnt; + struct mutex init_mutex; + struct rtrs_clt_io_req *reqs; + struct delayed_work reconnect_dwork; + struct work_struct close_work; + unsigned int reconnect_attempts; + bool established; + struct rtrs_rbuf *rbufs; + size_t max_io_size; + u32 max_hdr_size; + u32 chunk_size; + size_t queue_depth; + u32 max_pages_per_mr; + int max_send_sge; + u32 flags; + struct kobject kobj; + struct rtrs_clt_stats *stats; + /* cache hca_port and hca_name to display in sysfs */ + u8 hca_port; + char hca_name[IB_DEVICE_NAME_MAX]; + struct list_head __percpu + *mp_skip_entry; +}; + +struct rtrs_clt { + struct list_head paths_list; /* rcu protected list */ + size_t paths_num; + struct rtrs_clt_sess + __rcu * __percpu *pcpu_path; + uuid_t paths_uuid; + int paths_up; + struct mutex paths_mutex; + struct mutex paths_ev_mutex; + char sessname[NAME_MAX]; + u16 port; + unsigned int max_reconnect_attempts; + unsigned int reconnect_delay_sec; + unsigned int max_segments; + size_t max_segment_size; + void *permits; + unsigned long *permits_map; + size_t queue_depth; + size_t max_io_size; + wait_queue_head_t permits_wait; + size_t pdu_sz; + void *priv; + void (*link_ev)(void *priv, + enum rtrs_clt_link_ev ev); + struct device dev; + struct kobject *kobj_paths; + enum rtrs_mp_policy mp_policy; +}; + +static inline struct rtrs_clt_con *to_clt_con(struct rtrs_con *c) +{ + return container_of(c, struct rtrs_clt_con, c); +} + +static inline struct rtrs_clt_sess *to_clt_sess(struct rtrs_sess *s) +{ + return container_of(s, struct rtrs_clt_sess, s); +} + +static inline int permit_size(struct rtrs_clt *clt) +{ + return sizeof(struct rtrs_permit) + clt->pdu_sz; +} + +static inline struct rtrs_permit *get_permit(struct rtrs_clt *clt, int idx) +{ + return (struct rtrs_permit *)(clt->permits + permit_size(clt) * idx); +} + +int rtrs_clt_reconnect_from_sysfs(struct rtrs_clt_sess *sess); +int rtrs_clt_disconnect_from_sysfs(struct rtrs_clt_sess *sess); +int rtrs_clt_create_path_from_sysfs(struct rtrs_clt *clt, + struct rtrs_addr *addr); +int rtrs_clt_remove_path_from_sysfs(struct rtrs_clt_sess *sess, + const struct attribute *sysfs_self); + +void rtrs_clt_set_max_reconnect_attempts(struct rtrs_clt *clt, int value); +int rtrs_clt_get_max_reconnect_attempts(const struct rtrs_clt *clt); +void free_sess(struct rtrs_clt_sess *sess); + +/* rtrs-clt-stats.c */ + +int rtrs_clt_init_stats(struct rtrs_clt_stats *stats); + +void rtrs_clt_inc_failover_cnt(struct rtrs_clt_stats *s); + +void rtrs_clt_update_wc_stats(struct rtrs_clt_con *con); +void rtrs_clt_update_all_stats(struct rtrs_clt_io_req *req, int dir); + +int rtrs_clt_reset_rdma_lat_distr_stats(struct rtrs_clt_stats *stats, + bool enable); +ssize_t rtrs_clt_stats_rdma_lat_distr_to_str(struct rtrs_clt_stats *stats, + char *page, size_t len); +int rtrs_clt_reset_cpu_migr_stats(struct rtrs_clt_stats *stats, bool enable); +int rtrs_clt_stats_migration_cnt_to_str(struct rtrs_clt_stats *stats, char *buf, + size_t len); +int rtrs_clt_reset_reconnects_stat(struct rtrs_clt_stats *stats, bool enable); +int rtrs_clt_stats_reconnects_to_str(struct rtrs_clt_stats *stats, char *buf, + size_t len); +int rtrs_clt_reset_wc_comp_stats(struct rtrs_clt_stats *stats, bool enable); +int rtrs_clt_stats_wc_completion_to_str(struct rtrs_clt_stats *stats, char *buf, + size_t len); +int rtrs_clt_reset_rdma_stats(struct rtrs_clt_stats *stats, bool enable); +ssize_t rtrs_clt_stats_rdma_to_str(struct rtrs_clt_stats *stats, + char *page, size_t len); +int rtrs_clt_reset_all_stats(struct rtrs_clt_stats *stats, bool enable); +ssize_t rtrs_clt_reset_all_help(struct rtrs_clt_stats *stats, + char *page, size_t len); + +/* rtrs-clt-sysfs.c */ + +int rtrs_clt_create_sysfs_root_files(struct rtrs_clt *clt); +void rtrs_clt_destroy_sysfs_root_folders(struct rtrs_clt *clt); +void rtrs_clt_destroy_sysfs_root_files(struct rtrs_clt *clt); + +int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess); +void rtrs_clt_destroy_sess_files(struct rtrs_clt_sess *sess, + const struct attribute *sysfs_self); + +#endif /* RTRS_CLT_H */ diff --git a/drivers/infiniband/ulp/rtrs/rtrs-log.h b/drivers/infiniband/ulp/rtrs/rtrs-log.h new file mode 100644 index 000000000000..53c785b992f2 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-log.h @@ -0,0 +1,28 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#ifndef RTRS_LOG_H +#define RTRS_LOG_H + +#define rtrs_log(fn, obj, fmt, ...) \ + fn("<%s>: " fmt, obj->sessname, ##__VA_ARGS__) + +#define rtrs_err(obj, fmt, ...) \ + rtrs_log(pr_err, obj, fmt, ##__VA_ARGS__) +#define rtrs_err_rl(obj, fmt, ...) \ + rtrs_log(pr_err_ratelimited, obj, fmt, ##__VA_ARGS__) +#define rtrs_wrn(obj, fmt, ...) \ + rtrs_log(pr_warn, obj, fmt, ##__VA_ARGS__) +#define rtrs_wrn_rl(obj, fmt, ...) \ + rtrs_log(pr_warn_ratelimited, obj, fmt, ##__VA_ARGS__) +#define rtrs_info(obj, fmt, ...) \ + rtrs_log(pr_info, obj, fmt, ##__VA_ARGS__) +#define rtrs_info_rl(obj, fmt, ...) \ + rtrs_log(pr_info_ratelimited, obj, fmt, ##__VA_ARGS__) + +#endif /* RTRS_LOG_H */ diff --git a/drivers/infiniband/ulp/rtrs/rtrs-pri.h b/drivers/infiniband/ulp/rtrs/rtrs-pri.h new file mode 100644 index 000000000000..0a93c87ef92b --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-pri.h @@ -0,0 +1,399 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ + +#ifndef RTRS_PRI_H +#define RTRS_PRI_H + +#include <linux/uuid.h> +#include <rdma/rdma_cm.h> +#include <rdma/ib_verbs.h> +#include <rdma/ib.h> + +#include "rtrs.h" + +#define RTRS_PROTO_VER_MAJOR 2 +#define RTRS_PROTO_VER_MINOR 0 + +#define RTRS_PROTO_VER_STRING __stringify(RTRS_PROTO_VER_MAJOR) "." \ + __stringify(RTRS_PROTO_VER_MINOR) + +enum rtrs_imm_const { + MAX_IMM_TYPE_BITS = 4, + MAX_IMM_TYPE_MASK = ((1 << MAX_IMM_TYPE_BITS) - 1), + MAX_IMM_PAYL_BITS = 28, + MAX_IMM_PAYL_MASK = ((1 << MAX_IMM_PAYL_BITS) - 1), +}; + +enum rtrs_imm_type { + RTRS_IO_REQ_IMM = 0, /* client to server */ + RTRS_IO_RSP_IMM = 1, /* server to client */ + RTRS_IO_RSP_W_INV_IMM = 2, /* server to client */ + + RTRS_HB_MSG_IMM = 8, /* HB: HeartBeat */ + RTRS_HB_ACK_IMM = 9, + + RTRS_LAST_IMM, +}; + +enum { + SERVICE_CON_QUEUE_DEPTH = 512, + + MAX_PATHS_NUM = 128, + + /* + * With the size of struct rtrs_permit allocated on the client, 4K + * is the maximum number of rtrs_permits we can allocate. This number is + * also used on the client to allocate the IU for the user connection + * to receive the RDMA addresses from the server. + */ + MAX_SESS_QUEUE_DEPTH = 4096, + + RTRS_HB_INTERVAL_MS = 5000, + RTRS_HB_MISSED_MAX = 5, + + RTRS_MAGIC = 0x1BBD, + RTRS_PROTO_VER = (RTRS_PROTO_VER_MAJOR << 8) | RTRS_PROTO_VER_MINOR, +}; + +struct rtrs_ib_dev; + +struct rtrs_rdma_dev_pd_ops { + struct rtrs_ib_dev *(*alloc)(void); + void (*free)(struct rtrs_ib_dev *dev); + int (*init)(struct rtrs_ib_dev *dev); + void (*deinit)(struct rtrs_ib_dev *dev); +}; + +struct rtrs_rdma_dev_pd { + struct mutex mutex; + struct list_head list; + enum ib_pd_flags pd_flags; + const struct rtrs_rdma_dev_pd_ops *ops; +}; + +struct rtrs_ib_dev { + struct ib_device *ib_dev; + struct ib_pd *ib_pd; + struct kref ref; + struct list_head entry; + struct rtrs_rdma_dev_pd *pool; +}; + +struct rtrs_con { + struct rtrs_sess *sess; + struct ib_qp *qp; + struct ib_cq *cq; + struct rdma_cm_id *cm_id; + unsigned int cid; +}; + +struct rtrs_sess { + struct list_head entry; + struct sockaddr_storage dst_addr; + struct sockaddr_storage src_addr; + char sessname[NAME_MAX]; + uuid_t uuid; + struct rtrs_con **con; + unsigned int con_num; + unsigned int recon_cnt; + struct rtrs_ib_dev *dev; + int dev_ref; + struct ib_cqe *hb_cqe; + void (*hb_err_handler)(struct rtrs_con *con); + struct workqueue_struct *hb_wq; + struct delayed_work hb_dwork; + unsigned int hb_interval_ms; + unsigned int hb_missed_cnt; + unsigned int hb_missed_max; +}; + +/* rtrs information unit */ +struct rtrs_iu { + struct list_head list; + struct ib_cqe cqe; + dma_addr_t dma_addr; + void *buf; + size_t size; + enum dma_data_direction direction; +}; + +/** + * enum rtrs_msg_types - RTRS message types, see also rtrs/README + * @RTRS_MSG_INFO_REQ: Client additional info request to the server + * @RTRS_MSG_INFO_RSP: Server additional info response to the client + * @RTRS_MSG_WRITE: Client writes data per RDMA to server + * @RTRS_MSG_READ: Client requests data transfer from server + * @RTRS_MSG_RKEY_RSP: Server refreshed rkey for rbuf + */ +enum rtrs_msg_types { + RTRS_MSG_INFO_REQ, + RTRS_MSG_INFO_RSP, + RTRS_MSG_WRITE, + RTRS_MSG_READ, + RTRS_MSG_RKEY_RSP, +}; + +/** + * enum rtrs_msg_flags - RTRS message flags. + * @RTRS_NEED_INVAL: Send invalidation in response. + * @RTRS_MSG_NEW_RKEY_F: Send refreshed rkey in response. + */ +enum rtrs_msg_flags { + RTRS_MSG_NEED_INVAL_F = 1 << 0, + RTRS_MSG_NEW_RKEY_F = 1 << 1, +}; + +/** + * struct rtrs_sg_desc - RDMA-Buffer entry description + * @addr: Address of RDMA destination buffer + * @key: Authorization rkey to write to the buffer + * @len: Size of the buffer + */ +struct rtrs_sg_desc { + __le64 addr; + __le32 key; + __le32 len; +}; + +/** + * struct rtrs_msg_conn_req - Client connection request to the server + * @magic: RTRS magic + * @version: RTRS protocol version + * @cid: Current connection id + * @cid_num: Number of connections per session + * @recon_cnt: Reconnections counter + * @sess_uuid: UUID of a session (path) + * @paths_uuid: UUID of a group of sessions (paths) + * + * NOTE: max size 56 bytes, see man rdma_connect(). + */ +struct rtrs_msg_conn_req { + /* Is set to 0 by cma.c in case of AF_IB, do not touch that. + * see https://www.spinics.net/lists/linux-rdma/msg22397.html + */ + u8 __cma_version; + /* On sender side that should be set to 0, or cma_save_ip_info() + * extract garbage and will fail. + */ + u8 __ip_version; + __le16 magic; + __le16 version; + __le16 cid; + __le16 cid_num; + __le16 recon_cnt; + uuid_t sess_uuid; + uuid_t paths_uuid; + u8 reserved[12]; +}; + +/** + * struct rtrs_msg_conn_rsp - Server connection response to the client + * @magic: RTRS magic + * @version: RTRS protocol version + * @errno: If rdma_accept() then 0, if rdma_reject() indicates error + * @queue_depth: max inflight messages (queue-depth) in this session + * @max_io_size: max io size server supports + * @max_hdr_size: max msg header size server supports + * + * NOTE: size is 56 bytes, max possible is 136 bytes, see man rdma_accept(). + */ +struct rtrs_msg_conn_rsp { + __le16 magic; + __le16 version; + __le16 errno; + __le16 queue_depth; + __le32 max_io_size; + __le32 max_hdr_size; + __le32 flags; + u8 reserved[36]; +}; + +/** + * struct rtrs_msg_info_req + * @type: @RTRS_MSG_INFO_REQ + * @sessname: Session name chosen by client + */ +struct rtrs_msg_info_req { + __le16 type; + u8 sessname[NAME_MAX]; + u8 reserved[15]; +}; + +/** + * struct rtrs_msg_info_rsp + * @type: @RTRS_MSG_INFO_RSP + * @sg_cnt: Number of @desc entries + * @desc: RDMA buffers where the client can write to server + */ +struct rtrs_msg_info_rsp { + __le16 type; + __le16 sg_cnt; + u8 reserved[4]; + struct rtrs_sg_desc desc[]; +}; + +/** + * struct rtrs_msg_rkey_rsp + * @type: @RTRS_MSG_RKEY_RSP + * @buf_id: RDMA buf_id of the new rkey + * @rkey: new remote key for RDMA buffers id from server + */ +struct rtrs_msg_rkey_rsp { + __le16 type; + __le16 buf_id; + __le32 rkey; +}; + +/** + * struct rtrs_msg_rdma_read - RDMA data transfer request from client + * @type: always @RTRS_MSG_READ + * @usr_len: length of user payload + * @sg_cnt: number of @desc entries + * @desc: RDMA buffers where the server can write the result to + */ +struct rtrs_msg_rdma_read { + __le16 type; + __le16 usr_len; + __le16 flags; + __le16 sg_cnt; + struct rtrs_sg_desc desc[]; +}; + +/** + * struct_msg_rdma_write - Message transferred to server with RDMA-Write + * @type: always @RTRS_MSG_WRITE + * @usr_len: length of user payload + */ +struct rtrs_msg_rdma_write { + __le16 type; + __le16 usr_len; +}; + +/** + * struct_msg_rdma_hdr - header for read or write request + * @type: @RTRS_MSG_WRITE | @RTRS_MSG_READ + */ +struct rtrs_msg_rdma_hdr { + __le16 type; +}; + +/* rtrs.c */ + +struct rtrs_iu *rtrs_iu_alloc(u32 queue_size, size_t size, gfp_t t, + struct ib_device *dev, enum dma_data_direction, + void (*done)(struct ib_cq *cq, struct ib_wc *wc)); +void rtrs_iu_free(struct rtrs_iu *iu, enum dma_data_direction dir, + struct ib_device *dev, u32 queue_size); +int rtrs_iu_post_recv(struct rtrs_con *con, struct rtrs_iu *iu); +int rtrs_iu_post_send(struct rtrs_con *con, struct rtrs_iu *iu, size_t size, + struct ib_send_wr *head); +int rtrs_iu_post_rdma_write_imm(struct rtrs_con *con, struct rtrs_iu *iu, + struct ib_sge *sge, unsigned int num_sge, + u32 rkey, u64 rdma_addr, u32 imm_data, + enum ib_send_flags flags, + struct ib_send_wr *head); + +int rtrs_post_recv_empty(struct rtrs_con *con, struct ib_cqe *cqe); +int rtrs_post_rdma_write_imm_empty(struct rtrs_con *con, struct ib_cqe *cqe, + u32 imm_data, enum ib_send_flags flags, + struct ib_send_wr *head); + +int rtrs_cq_qp_create(struct rtrs_sess *rtrs_sess, struct rtrs_con *con, + u32 max_send_sge, int cq_vector, u16 cq_size, + u16 wr_queue_size, enum ib_poll_context poll_ctx); +void rtrs_cq_qp_destroy(struct rtrs_con *con); + +void rtrs_init_hb(struct rtrs_sess *sess, struct ib_cqe *cqe, + unsigned int interval_ms, unsigned int missed_max, + void (*err_handler)(struct rtrs_con *con), + struct workqueue_struct *wq); +void rtrs_start_hb(struct rtrs_sess *sess); +void rtrs_stop_hb(struct rtrs_sess *sess); +void rtrs_send_hb_ack(struct rtrs_sess *sess); + +void rtrs_rdma_dev_pd_init(enum ib_pd_flags pd_flags, + struct rtrs_rdma_dev_pd *pool); +void rtrs_rdma_dev_pd_deinit(struct rtrs_rdma_dev_pd *pool); + +struct rtrs_ib_dev *rtrs_ib_dev_find_or_add(struct ib_device *ib_dev, + struct rtrs_rdma_dev_pd *pool); +int rtrs_ib_dev_put(struct rtrs_ib_dev *dev); + +static inline u32 rtrs_to_imm(u32 type, u32 payload) +{ + BUILD_BUG_ON(MAX_IMM_PAYL_BITS + MAX_IMM_TYPE_BITS != 32); + BUILD_BUG_ON(RTRS_LAST_IMM > (1<<MAX_IMM_TYPE_BITS)); + return ((type & MAX_IMM_TYPE_MASK) << MAX_IMM_PAYL_BITS) | + (payload & MAX_IMM_PAYL_MASK); +} + +static inline void rtrs_from_imm(u32 imm, u32 *type, u32 *payload) +{ + *payload = imm & MAX_IMM_PAYL_MASK; + *type = imm >> MAX_IMM_PAYL_BITS; +} + +static inline u32 rtrs_to_io_req_imm(u32 addr) +{ + return rtrs_to_imm(RTRS_IO_REQ_IMM, addr); +} + +static inline u32 rtrs_to_io_rsp_imm(u32 msg_id, int errno, bool w_inval) +{ + enum rtrs_imm_type type; + u32 payload; + + /* 9 bits for errno, 19 bits for msg_id */ + payload = (abs(errno) & 0x1ff) << 19 | (msg_id & 0x7ffff); + type = w_inval ? RTRS_IO_RSP_W_INV_IMM : RTRS_IO_RSP_IMM; + + return rtrs_to_imm(type, payload); +} + +static inline void rtrs_from_io_rsp_imm(u32 payload, u32 *msg_id, int *errno) +{ + /* 9 bits for errno, 19 bits for msg_id */ + *msg_id = payload & 0x7ffff; + *errno = -(int)((payload >> 19) & 0x1ff); +} + +#define STAT_STORE_FUNC(type, set_value, reset) \ +static ssize_t set_value##_store(struct kobject *kobj, \ + struct kobj_attribute *attr, \ + const char *buf, size_t count) \ +{ \ + int ret = -EINVAL; \ + type *stats = container_of(kobj, type, kobj_stats); \ + \ + if (sysfs_streq(buf, "1")) \ + ret = reset(stats, true); \ + else if (sysfs_streq(buf, "0")) \ + ret = reset(stats, false); \ + if (ret) \ + return ret; \ + \ + return count; \ +} + +#define STAT_SHOW_FUNC(type, get_value, print) \ +static ssize_t get_value##_show(struct kobject *kobj, \ + struct kobj_attribute *attr, \ + char *page) \ +{ \ + type *stats = container_of(kobj, type, kobj_stats); \ + \ + return print(stats, page, PAGE_SIZE); \ +} + +#define STAT_ATTR(type, stat, print, reset) \ +STAT_STORE_FUNC(type, stat, reset) \ +STAT_SHOW_FUNC(type, stat, print) \ +static struct kobj_attribute stat##_attr = __ATTR_RW(stat) + +#endif /* RTRS_PRI_H */ diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c b/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c new file mode 100644 index 000000000000..e102b1368d0c --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-stats.c @@ -0,0 +1,38 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include "rtrs-srv.h" + +int rtrs_srv_reset_rdma_stats(struct rtrs_srv_stats *stats, bool enable) +{ + if (enable) { + struct rtrs_srv_stats_rdma_stats *r = &stats->rdma_stats; + + memset(r, 0, sizeof(*r)); + return 0; + } + + return -EINVAL; +} + +ssize_t rtrs_srv_stats_rdma_to_str(struct rtrs_srv_stats *stats, + char *page, size_t len) +{ + struct rtrs_srv_stats_rdma_stats *r = &stats->rdma_stats; + struct rtrs_srv_sess *sess = stats->sess; + + return scnprintf(page, len, "%lld %lld %lld %lld %u\n", + (s64)atomic64_read(&r->dir[READ].cnt), + (s64)atomic64_read(&r->dir[READ].size_total), + (s64)atomic64_read(&r->dir[WRITE].cnt), + (s64)atomic64_read(&r->dir[WRITE].size_total), + atomic_read(&sess->ids_inflight)); +} diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c b/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c new file mode 100644 index 000000000000..cf6a2be61695 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c @@ -0,0 +1,313 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include "rtrs-pri.h" +#include "rtrs-srv.h" +#include "rtrs-log.h" + +static void rtrs_srv_release(struct kobject *kobj) +{ + struct rtrs_srv_sess *sess; + + sess = container_of(kobj, struct rtrs_srv_sess, kobj); + kfree(sess); +} + +static struct kobj_type ktype = { + .sysfs_ops = &kobj_sysfs_ops, + .release = rtrs_srv_release, +}; + +static ssize_t rtrs_srv_disconnect_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + return scnprintf(page, PAGE_SIZE, "Usage: echo 1 > %s\n", + attr->attr.name); +} + +static ssize_t rtrs_srv_disconnect_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t count) +{ + struct rtrs_srv_sess *sess; + struct rtrs_sess *s; + char str[MAXHOSTNAMELEN]; + + sess = container_of(kobj, struct rtrs_srv_sess, kobj); + s = &sess->s; + if (!sysfs_streq(buf, "1")) { + rtrs_err(s, "%s: invalid value: '%s'\n", + attr->attr.name, buf); + return -EINVAL; + } + + sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, str, sizeof(str)); + + rtrs_info(s, "disconnect for path %s requested\n", str); + close_sess(sess); + + return count; +} + +static struct kobj_attribute rtrs_srv_disconnect_attr = + __ATTR(disconnect, 0644, + rtrs_srv_disconnect_show, rtrs_srv_disconnect_store); + +static ssize_t rtrs_srv_hca_port_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_srv_sess *sess; + struct rtrs_con *usr_con; + + sess = container_of(kobj, typeof(*sess), kobj); + usr_con = sess->s.con[0]; + + return scnprintf(page, PAGE_SIZE, "%u\n", + usr_con->cm_id->port_num); +} + +static struct kobj_attribute rtrs_srv_hca_port_attr = + __ATTR(hca_port, 0444, rtrs_srv_hca_port_show, NULL); + +static ssize_t rtrs_srv_hca_name_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_srv_sess *sess; + + sess = container_of(kobj, struct rtrs_srv_sess, kobj); + + return scnprintf(page, PAGE_SIZE, "%s\n", + sess->s.dev->ib_dev->name); +} + +static struct kobj_attribute rtrs_srv_hca_name_attr = + __ATTR(hca_name, 0444, rtrs_srv_hca_name_show, NULL); + +static ssize_t rtrs_srv_src_addr_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_srv_sess *sess; + int cnt; + + sess = container_of(kobj, struct rtrs_srv_sess, kobj); + cnt = sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, + page, PAGE_SIZE); + return cnt + scnprintf(page + cnt, PAGE_SIZE - cnt, "\n"); +} + +static struct kobj_attribute rtrs_srv_src_addr_attr = + __ATTR(src_addr, 0444, rtrs_srv_src_addr_show, NULL); + +static ssize_t rtrs_srv_dst_addr_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *page) +{ + struct rtrs_srv_sess *sess; + int cnt; + + sess = container_of(kobj, struct rtrs_srv_sess, kobj); + cnt = sockaddr_to_str((struct sockaddr *)&sess->s.src_addr, + page, PAGE_SIZE); + return cnt + scnprintf(page + cnt, PAGE_SIZE - cnt, "\n"); +} + +static struct kobj_attribute rtrs_srv_dst_addr_attr = + __ATTR(dst_addr, 0444, rtrs_srv_dst_addr_show, NULL); + +static struct attribute *rtrs_srv_sess_attrs[] = { + &rtrs_srv_hca_name_attr.attr, + &rtrs_srv_hca_port_attr.attr, + &rtrs_srv_src_addr_attr.attr, + &rtrs_srv_dst_addr_attr.attr, + &rtrs_srv_disconnect_attr.attr, + NULL, +}; + +static struct attribute_group rtrs_srv_sess_attr_group = { + .attrs = rtrs_srv_sess_attrs, +}; + +STAT_ATTR(struct rtrs_srv_stats, rdma, + rtrs_srv_stats_rdma_to_str, + rtrs_srv_reset_rdma_stats); + +static struct attribute *rtrs_srv_stats_attrs[] = { + &rdma_attr.attr, + NULL, +}; + +static struct attribute_group rtrs_srv_stats_attr_group = { + .attrs = rtrs_srv_stats_attrs, +}; + +static int rtrs_srv_create_once_sysfs_root_folders(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + int err = 0; + + mutex_lock(&srv->paths_mutex); + if (srv->dev_ref++) { + /* + * Device needs to be registered only on the first session + */ + goto unlock; + } + srv->dev.class = rtrs_dev_class; + err = dev_set_name(&srv->dev, "%s", sess->s.sessname); + if (err) + goto unlock; + + /* + * Suppress user space notification until + * sysfs files are created + */ + dev_set_uevent_suppress(&srv->dev, true); + err = device_add(&srv->dev); + if (err) { + pr_err("device_add(): %d\n", err); + goto put; + } + srv->kobj_paths = kobject_create_and_add("paths", &srv->dev.kobj); + if (!srv->kobj_paths) { + err = -ENOMEM; + pr_err("kobject_create_and_add(): %d\n", err); + device_del(&srv->dev); + goto unlock; + } + dev_set_uevent_suppress(&srv->dev, false); + kobject_uevent(&srv->dev.kobj, KOBJ_ADD); + goto unlock; + +put: + put_device(&srv->dev); +unlock: + mutex_unlock(&srv->paths_mutex); + + return err; +} + +static void +rtrs_srv_destroy_once_sysfs_root_folders(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + + mutex_lock(&srv->paths_mutex); + if (!--srv->dev_ref) { + kobject_del(srv->kobj_paths); + kobject_put(srv->kobj_paths); + mutex_unlock(&srv->paths_mutex); + device_del(&srv->dev); + } else { + mutex_unlock(&srv->paths_mutex); + } +} + +static void rtrs_srv_sess_stats_release(struct kobject *kobj) +{ + struct rtrs_srv_stats *stats; + + stats = container_of(kobj, struct rtrs_srv_stats, kobj_stats); + + kfree(stats); +} + +static struct kobj_type ktype_stats = { + .sysfs_ops = &kobj_sysfs_ops, + .release = rtrs_srv_sess_stats_release, +}; + +static int rtrs_srv_create_stats_files(struct rtrs_srv_sess *sess) +{ + int err; + struct rtrs_sess *s = &sess->s; + + err = kobject_init_and_add(&sess->stats->kobj_stats, &ktype_stats, + &sess->kobj, "stats"); + if (err) { + rtrs_err(s, "kobject_init_and_add(): %d\n", err); + return err; + } + err = sysfs_create_group(&sess->stats->kobj_stats, + &rtrs_srv_stats_attr_group); + if (err) { + rtrs_err(s, "sysfs_create_group(): %d\n", err); + goto err; + } + + return 0; + +err: + kobject_del(&sess->stats->kobj_stats); + kobject_put(&sess->stats->kobj_stats); + + return err; +} + +int rtrs_srv_create_sess_files(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_sess *s = &sess->s; + char str[NAME_MAX]; + int err, cnt; + + cnt = sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, + str, sizeof(str)); + cnt += scnprintf(str + cnt, sizeof(str) - cnt, "@"); + sockaddr_to_str((struct sockaddr *)&sess->s.src_addr, + str + cnt, sizeof(str) - cnt); + + err = rtrs_srv_create_once_sysfs_root_folders(sess); + if (err) + return err; + + err = kobject_init_and_add(&sess->kobj, &ktype, srv->kobj_paths, + "%s", str); + if (err) { + rtrs_err(s, "kobject_init_and_add(): %d\n", err); + goto destroy_root; + } + err = sysfs_create_group(&sess->kobj, &rtrs_srv_sess_attr_group); + if (err) { + rtrs_err(s, "sysfs_create_group(): %d\n", err); + goto put_kobj; + } + err = rtrs_srv_create_stats_files(sess); + if (err) + goto remove_group; + + return 0; + +remove_group: + sysfs_remove_group(&sess->kobj, &rtrs_srv_sess_attr_group); +put_kobj: + kobject_del(&sess->kobj); + kobject_put(&sess->kobj); +destroy_root: + rtrs_srv_destroy_once_sysfs_root_folders(sess); + + return err; +} + +void rtrs_srv_destroy_sess_files(struct rtrs_srv_sess *sess) +{ + if (sess->kobj.state_in_sysfs) { + kobject_del(&sess->stats->kobj_stats); + kobject_put(&sess->stats->kobj_stats); + kobject_del(&sess->kobj); + kobject_put(&sess->kobj); + + rtrs_srv_destroy_once_sysfs_root_folders(sess); + } +} diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv.c b/drivers/infiniband/ulp/rtrs/rtrs-srv.c new file mode 100644 index 000000000000..28f6414dfa3d --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv.c @@ -0,0 +1,2187 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ + +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include <linux/module.h> +#include <linux/mempool.h> + +#include "rtrs-srv.h" +#include "rtrs-log.h" +#include <rdma/ib_cm.h> + +MODULE_DESCRIPTION("RDMA Transport Server"); +MODULE_LICENSE("GPL"); + +/* Must be power of 2, see mask from mr->page_size in ib_sg_to_pages() */ +#define DEFAULT_MAX_CHUNK_SIZE (128 << 10) +#define DEFAULT_SESS_QUEUE_DEPTH 512 +#define MAX_HDR_SIZE PAGE_SIZE + +/* We guarantee to serve 10 paths at least */ +#define CHUNK_POOL_SZ 10 + +static struct rtrs_rdma_dev_pd dev_pd; +static mempool_t *chunk_pool; +struct class *rtrs_dev_class; + +static int __read_mostly max_chunk_size = DEFAULT_MAX_CHUNK_SIZE; +static int __read_mostly sess_queue_depth = DEFAULT_SESS_QUEUE_DEPTH; + +static bool always_invalidate = true; +module_param(always_invalidate, bool, 0444); +MODULE_PARM_DESC(always_invalidate, + "Invalidate memory registration for contiguous memory regions before accessing."); + +module_param_named(max_chunk_size, max_chunk_size, int, 0444); +MODULE_PARM_DESC(max_chunk_size, + "Max size for each IO request, when change the unit is in byte (default: " + __stringify(DEFAULT_MAX_CHUNK_SIZE) "KB)"); + +module_param_named(sess_queue_depth, sess_queue_depth, int, 0444); +MODULE_PARM_DESC(sess_queue_depth, + "Number of buffers for pending I/O requests to allocate per session. Maximum: " + __stringify(MAX_SESS_QUEUE_DEPTH) " (default: " + __stringify(DEFAULT_SESS_QUEUE_DEPTH) ")"); + +static cpumask_t cq_affinity_mask = { CPU_BITS_ALL }; + +static struct workqueue_struct *rtrs_wq; + +static inline struct rtrs_srv_con *to_srv_con(struct rtrs_con *c) +{ + return container_of(c, struct rtrs_srv_con, c); +} + +static inline struct rtrs_srv_sess *to_srv_sess(struct rtrs_sess *s) +{ + return container_of(s, struct rtrs_srv_sess, s); +} + +static bool __rtrs_srv_change_state(struct rtrs_srv_sess *sess, + enum rtrs_srv_state new_state) +{ + enum rtrs_srv_state old_state; + bool changed = false; + + lockdep_assert_held(&sess->state_lock); + old_state = sess->state; + switch (new_state) { + case RTRS_SRV_CONNECTED: + switch (old_state) { + case RTRS_SRV_CONNECTING: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_SRV_CLOSING: + switch (old_state) { + case RTRS_SRV_CONNECTING: + case RTRS_SRV_CONNECTED: + changed = true; + fallthrough; + default: + break; + } + break; + case RTRS_SRV_CLOSED: + switch (old_state) { + case RTRS_SRV_CLOSING: + changed = true; + fallthrough; + default: + break; + } + break; + default: + break; + } + if (changed) + sess->state = new_state; + + return changed; +} + +static bool rtrs_srv_change_state_get_old(struct rtrs_srv_sess *sess, + enum rtrs_srv_state new_state, + enum rtrs_srv_state *old_state) +{ + bool changed; + + spin_lock_irq(&sess->state_lock); + *old_state = sess->state; + changed = __rtrs_srv_change_state(sess, new_state); + spin_unlock_irq(&sess->state_lock); + + return changed; +} + +static bool rtrs_srv_change_state(struct rtrs_srv_sess *sess, + enum rtrs_srv_state new_state) +{ + enum rtrs_srv_state old_state; + + return rtrs_srv_change_state_get_old(sess, new_state, &old_state); +} + +static void free_id(struct rtrs_srv_op *id) +{ + if (!id) + return; + kfree(id); +} + +static void rtrs_srv_free_ops_ids(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + int i; + + WARN_ON(atomic_read(&sess->ids_inflight)); + if (sess->ops_ids) { + for (i = 0; i < srv->queue_depth; i++) + free_id(sess->ops_ids[i]); + kfree(sess->ops_ids); + sess->ops_ids = NULL; + } +} + +static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc); + +static struct ib_cqe io_comp_cqe = { + .done = rtrs_srv_rdma_done +}; + +static int rtrs_srv_alloc_ops_ids(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_op *id; + int i; + + sess->ops_ids = kcalloc(srv->queue_depth, sizeof(*sess->ops_ids), + GFP_KERNEL); + if (!sess->ops_ids) + goto err; + + for (i = 0; i < srv->queue_depth; ++i) { + id = kzalloc(sizeof(*id), GFP_KERNEL); + if (!id) + goto err; + + sess->ops_ids[i] = id; + } + init_waitqueue_head(&sess->ids_waitq); + atomic_set(&sess->ids_inflight, 0); + + return 0; + +err: + rtrs_srv_free_ops_ids(sess); + return -ENOMEM; +} + +static inline void rtrs_srv_get_ops_ids(struct rtrs_srv_sess *sess) +{ + atomic_inc(&sess->ids_inflight); +} + +static inline void rtrs_srv_put_ops_ids(struct rtrs_srv_sess *sess) +{ + if (atomic_dec_and_test(&sess->ids_inflight)) + wake_up(&sess->ids_waitq); +} + +static void rtrs_srv_wait_ops_ids(struct rtrs_srv_sess *sess) +{ + wait_event(sess->ids_waitq, !atomic_read(&sess->ids_inflight)); +} + + +static void rtrs_srv_reg_mr_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_srv_con *con = cq->cq_context; + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(s, "REG MR failed: %s\n", + ib_wc_status_msg(wc->status)); + close_sess(sess); + return; + } +} + +static struct ib_cqe local_reg_cqe = { + .done = rtrs_srv_reg_mr_done +}; + +static int rdma_write_sg(struct rtrs_srv_op *id) +{ + struct rtrs_sess *s = id->con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + dma_addr_t dma_addr = sess->dma_addr[id->msg_id]; + struct rtrs_srv_mr *srv_mr; + struct rtrs_srv *srv = sess->srv; + struct ib_send_wr inv_wr, imm_wr; + struct ib_rdma_wr *wr = NULL; + enum ib_send_flags flags; + size_t sg_cnt; + int err, offset; + bool need_inval; + u32 rkey = 0; + struct ib_reg_wr rwr; + struct ib_sge *plist; + struct ib_sge list; + + sg_cnt = le16_to_cpu(id->rd_msg->sg_cnt); + need_inval = le16_to_cpu(id->rd_msg->flags) & RTRS_MSG_NEED_INVAL_F; + if (unlikely(sg_cnt != 1)) + return -EINVAL; + + offset = 0; + + wr = &id->tx_wr; + plist = &id->tx_sg; + plist->addr = dma_addr + offset; + plist->length = le32_to_cpu(id->rd_msg->desc[0].len); + + /* WR will fail with length error + * if this is 0 + */ + if (unlikely(plist->length == 0)) { + rtrs_err(s, "Invalid RDMA-Write sg list length 0\n"); + return -EINVAL; + } + + plist->lkey = sess->s.dev->ib_pd->local_dma_lkey; + offset += plist->length; + + wr->wr.sg_list = plist; + wr->wr.num_sge = 1; + wr->remote_addr = le64_to_cpu(id->rd_msg->desc[0].addr); + wr->rkey = le32_to_cpu(id->rd_msg->desc[0].key); + if (rkey == 0) + rkey = wr->rkey; + else + /* Only one key is actually used */ + WARN_ON_ONCE(rkey != wr->rkey); + + wr->wr.opcode = IB_WR_RDMA_WRITE; + wr->wr.ex.imm_data = 0; + wr->wr.send_flags = 0; + + if (need_inval && always_invalidate) { + wr->wr.next = &rwr.wr; + rwr.wr.next = &inv_wr; + inv_wr.next = &imm_wr; + } else if (always_invalidate) { + wr->wr.next = &rwr.wr; + rwr.wr.next = &imm_wr; + } else if (need_inval) { + wr->wr.next = &inv_wr; + inv_wr.next = &imm_wr; + } else { + wr->wr.next = &imm_wr; + } + /* + * From time to time we have to post signaled sends, + * or send queue will fill up and only QP reset can help. + */ + flags = (atomic_inc_return(&id->con->wr_cnt) % srv->queue_depth) ? + 0 : IB_SEND_SIGNALED; + + if (need_inval) { + inv_wr.sg_list = NULL; + inv_wr.num_sge = 0; + inv_wr.opcode = IB_WR_SEND_WITH_INV; + inv_wr.send_flags = 0; + inv_wr.ex.invalidate_rkey = rkey; + } + + imm_wr.next = NULL; + if (always_invalidate) { + struct rtrs_msg_rkey_rsp *msg; + + srv_mr = &sess->mrs[id->msg_id]; + rwr.wr.opcode = IB_WR_REG_MR; + rwr.wr.num_sge = 0; + rwr.mr = srv_mr->mr; + rwr.wr.send_flags = 0; + rwr.key = srv_mr->mr->rkey; + rwr.access = (IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE); + msg = srv_mr->iu->buf; + msg->buf_id = cpu_to_le16(id->msg_id); + msg->type = cpu_to_le16(RTRS_MSG_RKEY_RSP); + msg->rkey = cpu_to_le32(srv_mr->mr->rkey); + + list.addr = srv_mr->iu->dma_addr; + list.length = sizeof(*msg); + list.lkey = sess->s.dev->ib_pd->local_dma_lkey; + imm_wr.sg_list = &list; + imm_wr.num_sge = 1; + imm_wr.opcode = IB_WR_SEND_WITH_IMM; + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, + srv_mr->iu->dma_addr, + srv_mr->iu->size, DMA_TO_DEVICE); + } else { + imm_wr.sg_list = NULL; + imm_wr.num_sge = 0; + imm_wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM; + } + imm_wr.send_flags = flags; + imm_wr.ex.imm_data = cpu_to_be32(rtrs_to_io_rsp_imm(id->msg_id, + 0, need_inval)); + + imm_wr.wr_cqe = &io_comp_cqe; + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, dma_addr, + offset, DMA_BIDIRECTIONAL); + + err = ib_post_send(id->con->c.qp, &id->tx_wr.wr, NULL); + if (unlikely(err)) + rtrs_err(s, + "Posting RDMA-Write-Request to QP failed, err: %d\n", + err); + + return err; +} + +/** + * send_io_resp_imm() - respond to client with empty IMM on failed READ/WRITE + * requests or on successful WRITE request. + * @con: the connection to send back result + * @id: the id associated with the IO + * @errno: the error number of the IO. + * + * Return 0 on success, errno otherwise. + */ +static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, + int errno) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct ib_send_wr inv_wr, imm_wr, *wr = NULL; + struct ib_reg_wr rwr; + struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_mr *srv_mr; + bool need_inval = false; + enum ib_send_flags flags; + u32 imm; + int err; + + if (id->dir == READ) { + struct rtrs_msg_rdma_read *rd_msg = id->rd_msg; + size_t sg_cnt; + + need_inval = le16_to_cpu(rd_msg->flags) & + RTRS_MSG_NEED_INVAL_F; + sg_cnt = le16_to_cpu(rd_msg->sg_cnt); + + if (need_inval) { + if (likely(sg_cnt)) { + inv_wr.sg_list = NULL; + inv_wr.num_sge = 0; + inv_wr.opcode = IB_WR_SEND_WITH_INV; + inv_wr.send_flags = 0; + /* Only one key is actually used */ + inv_wr.ex.invalidate_rkey = + le32_to_cpu(rd_msg->desc[0].key); + } else { + WARN_ON_ONCE(1); + need_inval = false; + } + } + } + + if (need_inval && always_invalidate) { + wr = &inv_wr; + inv_wr.next = &rwr.wr; + rwr.wr.next = &imm_wr; + } else if (always_invalidate) { + wr = &rwr.wr; + rwr.wr.next = &imm_wr; + } else if (need_inval) { + wr = &inv_wr; + inv_wr.next = &imm_wr; + } else { + wr = &imm_wr; + } + /* + * From time to time we have to post signalled sends, + * or send queue will fill up and only QP reset can help. + */ + flags = (atomic_inc_return(&con->wr_cnt) % srv->queue_depth) ? + 0 : IB_SEND_SIGNALED; + imm = rtrs_to_io_rsp_imm(id->msg_id, errno, need_inval); + imm_wr.next = NULL; + if (always_invalidate) { + struct ib_sge list; + struct rtrs_msg_rkey_rsp *msg; + + srv_mr = &sess->mrs[id->msg_id]; + rwr.wr.next = &imm_wr; + rwr.wr.opcode = IB_WR_REG_MR; + rwr.wr.num_sge = 0; + rwr.wr.send_flags = 0; + rwr.mr = srv_mr->mr; + rwr.key = srv_mr->mr->rkey; + rwr.access = (IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE); + msg = srv_mr->iu->buf; + msg->buf_id = cpu_to_le16(id->msg_id); + msg->type = cpu_to_le16(RTRS_MSG_RKEY_RSP); + msg->rkey = cpu_to_le32(srv_mr->mr->rkey); + + list.addr = srv_mr->iu->dma_addr; + list.length = sizeof(*msg); + list.lkey = sess->s.dev->ib_pd->local_dma_lkey; + imm_wr.sg_list = &list; + imm_wr.num_sge = 1; + imm_wr.opcode = IB_WR_SEND_WITH_IMM; + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, + srv_mr->iu->dma_addr, + srv_mr->iu->size, DMA_TO_DEVICE); + } else { + imm_wr.sg_list = NULL; + imm_wr.num_sge = 0; + imm_wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM; + } + imm_wr.send_flags = flags; + imm_wr.wr_cqe = &io_comp_cqe; + + imm_wr.ex.imm_data = cpu_to_be32(imm); + + err = ib_post_send(id->con->c.qp, wr, NULL); + if (unlikely(err)) + rtrs_err_rl(s, "Posting RDMA-Reply to QP failed, err: %d\n", + err); + + return err; +} + +void close_sess(struct rtrs_srv_sess *sess) +{ + enum rtrs_srv_state old_state; + + if (rtrs_srv_change_state_get_old(sess, RTRS_SRV_CLOSING, + &old_state)) + queue_work(rtrs_wq, &sess->close_work); + WARN_ON(sess->state != RTRS_SRV_CLOSING); +} + +static inline const char *rtrs_srv_state_str(enum rtrs_srv_state state) +{ + switch (state) { + case RTRS_SRV_CONNECTING: + return "RTRS_SRV_CONNECTING"; + case RTRS_SRV_CONNECTED: + return "RTRS_SRV_CONNECTED"; + case RTRS_SRV_CLOSING: + return "RTRS_SRV_CLOSING"; + case RTRS_SRV_CLOSED: + return "RTRS_SRV_CLOSED"; + default: + return "UNKNOWN"; + } +} + +/** + * rtrs_srv_resp_rdma() - Finish an RDMA request + * + * @id: Internal RTRS operation identifier + * @status: Response Code sent to the other side for this operation. + * 0 = success, <=0 error + * Context: any + * + * Finish a RDMA operation. A message is sent to the client and the + * corresponding memory areas will be released. + */ +bool rtrs_srv_resp_rdma(struct rtrs_srv_op *id, int status) +{ + struct rtrs_srv_sess *sess; + struct rtrs_srv_con *con; + struct rtrs_sess *s; + int err; + + if (WARN_ON(!id)) + return true; + + con = id->con; + s = con->c.sess; + sess = to_srv_sess(s); + + id->status = status; + + if (unlikely(sess->state != RTRS_SRV_CONNECTED)) { + rtrs_err_rl(s, + "Sending I/O response failed, session is disconnected, sess state %s\n", + rtrs_srv_state_str(sess->state)); + goto out; + } + if (always_invalidate) { + struct rtrs_srv_mr *mr = &sess->mrs[id->msg_id]; + + ib_update_fast_reg_key(mr->mr, ib_inc_rkey(mr->mr->rkey)); + } + if (unlikely(atomic_sub_return(1, + &con->sq_wr_avail) < 0)) { + pr_err("IB send queue full\n"); + atomic_add(1, &con->sq_wr_avail); + spin_lock(&con->rsp_wr_wait_lock); + list_add_tail(&id->wait_list, &con->rsp_wr_wait_list); + spin_unlock(&con->rsp_wr_wait_lock); + return false; + } + + if (status || id->dir == WRITE || !id->rd_msg->sg_cnt) + err = send_io_resp_imm(con, id, status); + else + err = rdma_write_sg(id); + + if (unlikely(err)) { + rtrs_err_rl(s, "IO response failed: %d\n", err); + close_sess(sess); + } +out: + rtrs_srv_put_ops_ids(sess); + return true; +} +EXPORT_SYMBOL(rtrs_srv_resp_rdma); + +/** + * rtrs_srv_set_sess_priv() - Set private pointer in rtrs_srv. + * @srv: Session pointer + * @priv: The private pointer that is associated with the session. + */ +void rtrs_srv_set_sess_priv(struct rtrs_srv *srv, void *priv) +{ + srv->priv = priv; +} +EXPORT_SYMBOL(rtrs_srv_set_sess_priv); + +static void unmap_cont_bufs(struct rtrs_srv_sess *sess) +{ + int i; + + for (i = 0; i < sess->mrs_num; i++) { + struct rtrs_srv_mr *srv_mr; + + srv_mr = &sess->mrs[i]; + rtrs_iu_free(srv_mr->iu, DMA_TO_DEVICE, + sess->s.dev->ib_dev, 1); + ib_dereg_mr(srv_mr->mr); + ib_dma_unmap_sg(sess->s.dev->ib_dev, srv_mr->sgt.sgl, + srv_mr->sgt.nents, DMA_BIDIRECTIONAL); + sg_free_table(&srv_mr->sgt); + } + kfree(sess->mrs); +} + +static int map_cont_bufs(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_sess *ss = &sess->s; + int i, mri, err, mrs_num; + unsigned int chunk_bits; + int chunks_per_mr = 1; + + /* + * Here we map queue_depth chunks to MR. Firstly we have to + * figure out how many chunks can we map per MR. + */ + if (always_invalidate) { + /* + * in order to do invalidate for each chunks of memory, we needs + * more memory regions. + */ + mrs_num = srv->queue_depth; + } else { + chunks_per_mr = + sess->s.dev->ib_dev->attrs.max_fast_reg_page_list_len; + mrs_num = DIV_ROUND_UP(srv->queue_depth, chunks_per_mr); + chunks_per_mr = DIV_ROUND_UP(srv->queue_depth, mrs_num); + } + + sess->mrs = kcalloc(mrs_num, sizeof(*sess->mrs), GFP_KERNEL); + if (!sess->mrs) + return -ENOMEM; + + sess->mrs_num = mrs_num; + + for (mri = 0; mri < mrs_num; mri++) { + struct rtrs_srv_mr *srv_mr = &sess->mrs[mri]; + struct sg_table *sgt = &srv_mr->sgt; + struct scatterlist *s; + struct ib_mr *mr; + int nr, chunks; + + chunks = chunks_per_mr * mri; + if (!always_invalidate) + chunks_per_mr = min_t(int, chunks_per_mr, + srv->queue_depth - chunks); + + err = sg_alloc_table(sgt, chunks_per_mr, GFP_KERNEL); + if (err) + goto err; + + for_each_sg(sgt->sgl, s, chunks_per_mr, i) + sg_set_page(s, srv->chunks[chunks + i], + max_chunk_size, 0); + + nr = ib_dma_map_sg(sess->s.dev->ib_dev, sgt->sgl, + sgt->nents, DMA_BIDIRECTIONAL); + if (nr < sgt->nents) { + err = nr < 0 ? nr : -EINVAL; + goto free_sg; + } + mr = ib_alloc_mr(sess->s.dev->ib_pd, IB_MR_TYPE_MEM_REG, + sgt->nents); + if (IS_ERR(mr)) { + err = PTR_ERR(mr); + goto unmap_sg; + } + nr = ib_map_mr_sg(mr, sgt->sgl, sgt->nents, + NULL, max_chunk_size); + if (nr < 0 || nr < sgt->nents) { + err = nr < 0 ? nr : -EINVAL; + goto dereg_mr; + } + + if (always_invalidate) { + srv_mr->iu = rtrs_iu_alloc(1, + sizeof(struct rtrs_msg_rkey_rsp), + GFP_KERNEL, sess->s.dev->ib_dev, + DMA_TO_DEVICE, rtrs_srv_rdma_done); + if (!srv_mr->iu) { + err = -ENOMEM; + rtrs_err(ss, "rtrs_iu_alloc(), err: %d\n", err); + goto free_iu; + } + } + /* Eventually dma addr for each chunk can be cached */ + for_each_sg(sgt->sgl, s, sgt->orig_nents, i) + sess->dma_addr[chunks + i] = sg_dma_address(s); + + ib_update_fast_reg_key(mr, ib_inc_rkey(mr->rkey)); + srv_mr->mr = mr; + + continue; +err: + while (mri--) { + srv_mr = &sess->mrs[mri]; + sgt = &srv_mr->sgt; + mr = srv_mr->mr; +free_iu: + rtrs_iu_free(srv_mr->iu, DMA_TO_DEVICE, + sess->s.dev->ib_dev, 1); +dereg_mr: + ib_dereg_mr(mr); +unmap_sg: + ib_dma_unmap_sg(sess->s.dev->ib_dev, sgt->sgl, + sgt->nents, DMA_BIDIRECTIONAL); +free_sg: + sg_free_table(sgt); + } + kfree(sess->mrs); + + return err; + } + + chunk_bits = ilog2(srv->queue_depth - 1) + 1; + sess->mem_bits = (MAX_IMM_PAYL_BITS - chunk_bits); + + return 0; +} + +static void rtrs_srv_hb_err_handler(struct rtrs_con *c) +{ + close_sess(to_srv_sess(c->sess)); +} + +static void rtrs_srv_init_hb(struct rtrs_srv_sess *sess) +{ + rtrs_init_hb(&sess->s, &io_comp_cqe, + RTRS_HB_INTERVAL_MS, + RTRS_HB_MISSED_MAX, + rtrs_srv_hb_err_handler, + rtrs_wq); +} + +static void rtrs_srv_start_hb(struct rtrs_srv_sess *sess) +{ + rtrs_start_hb(&sess->s); +} + +static void rtrs_srv_stop_hb(struct rtrs_srv_sess *sess) +{ + rtrs_stop_hb(&sess->s); +} + +static void rtrs_srv_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_srv_con *con = cq->cq_context; + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_iu *iu; + + iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); + rtrs_iu_free(iu, DMA_TO_DEVICE, sess->s.dev->ib_dev, 1); + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(s, "Sess info response send failed: %s\n", + ib_wc_status_msg(wc->status)); + close_sess(sess); + return; + } + WARN_ON(wc->opcode != IB_WC_SEND); +} + +static void rtrs_srv_sess_up(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_ctx *ctx = srv->ctx; + int up; + + mutex_lock(&srv->paths_ev_mutex); + up = ++srv->paths_up; + if (up == 1) + ctx->ops.link_ev(srv, RTRS_SRV_LINK_EV_CONNECTED, NULL); + mutex_unlock(&srv->paths_ev_mutex); + + /* Mark session as established */ + sess->established = true; +} + +static void rtrs_srv_sess_down(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_ctx *ctx = srv->ctx; + + if (!sess->established) + return; + + sess->established = false; + mutex_lock(&srv->paths_ev_mutex); + WARN_ON(!srv->paths_up); + if (--srv->paths_up == 0) + ctx->ops.link_ev(srv, RTRS_SRV_LINK_EV_DISCONNECTED, srv->priv); + mutex_unlock(&srv->paths_ev_mutex); +} + +static int post_recv_sess(struct rtrs_srv_sess *sess); + +static int process_info_req(struct rtrs_srv_con *con, + struct rtrs_msg_info_req *msg) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct ib_send_wr *reg_wr = NULL; + struct rtrs_msg_info_rsp *rsp; + struct rtrs_iu *tx_iu; + struct ib_reg_wr *rwr; + int mri, err; + size_t tx_sz; + + err = post_recv_sess(sess); + if (unlikely(err)) { + rtrs_err(s, "post_recv_sess(), err: %d\n", err); + return err; + } + rwr = kcalloc(sess->mrs_num, sizeof(*rwr), GFP_KERNEL); + if (unlikely(!rwr)) + return -ENOMEM; + strlcpy(sess->s.sessname, msg->sessname, sizeof(sess->s.sessname)); + + tx_sz = sizeof(*rsp); + tx_sz += sizeof(rsp->desc[0]) * sess->mrs_num; + tx_iu = rtrs_iu_alloc(1, tx_sz, GFP_KERNEL, sess->s.dev->ib_dev, + DMA_TO_DEVICE, rtrs_srv_info_rsp_done); + if (unlikely(!tx_iu)) { + err = -ENOMEM; + goto rwr_free; + } + + rsp = tx_iu->buf; + rsp->type = cpu_to_le16(RTRS_MSG_INFO_RSP); + rsp->sg_cnt = cpu_to_le16(sess->mrs_num); + + for (mri = 0; mri < sess->mrs_num; mri++) { + struct ib_mr *mr = sess->mrs[mri].mr; + + rsp->desc[mri].addr = cpu_to_le64(mr->iova); + rsp->desc[mri].key = cpu_to_le32(mr->rkey); + rsp->desc[mri].len = cpu_to_le32(mr->length); + + /* + * Fill in reg MR request and chain them *backwards* + */ + rwr[mri].wr.next = mri ? &rwr[mri - 1].wr : NULL; + rwr[mri].wr.opcode = IB_WR_REG_MR; + rwr[mri].wr.wr_cqe = &local_reg_cqe; + rwr[mri].wr.num_sge = 0; + rwr[mri].wr.send_flags = mri ? 0 : IB_SEND_SIGNALED; + rwr[mri].mr = mr; + rwr[mri].key = mr->rkey; + rwr[mri].access = (IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_WRITE); + reg_wr = &rwr[mri].wr; + } + + err = rtrs_srv_create_sess_files(sess); + if (unlikely(err)) + goto iu_free; + kobject_get(&sess->kobj); + get_device(&sess->srv->dev); + rtrs_srv_change_state(sess, RTRS_SRV_CONNECTED); + rtrs_srv_start_hb(sess); + + /* + * We do not account number of established connections at the current + * moment, we rely on the client, which should send info request when + * all connections are successfully established. Thus, simply notify + * listener with a proper event if we are the first path. + */ + rtrs_srv_sess_up(sess); + + ib_dma_sync_single_for_device(sess->s.dev->ib_dev, tx_iu->dma_addr, + tx_iu->size, DMA_TO_DEVICE); + + /* Send info response */ + err = rtrs_iu_post_send(&con->c, tx_iu, tx_sz, reg_wr); + if (unlikely(err)) { + rtrs_err(s, "rtrs_iu_post_send(), err: %d\n", err); +iu_free: + rtrs_iu_free(tx_iu, DMA_TO_DEVICE, sess->s.dev->ib_dev, 1); + } +rwr_free: + kfree(rwr); + + return err; +} + +static void rtrs_srv_info_req_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_srv_con *con = cq->cq_context; + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_msg_info_req *msg; + struct rtrs_iu *iu; + int err; + + WARN_ON(con->c.cid); + + iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(s, "Sess info request receive failed: %s\n", + ib_wc_status_msg(wc->status)); + goto close; + } + WARN_ON(wc->opcode != IB_WC_RECV); + + if (unlikely(wc->byte_len < sizeof(*msg))) { + rtrs_err(s, "Sess info request is malformed: size %d\n", + wc->byte_len); + goto close; + } + ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, iu->dma_addr, + iu->size, DMA_FROM_DEVICE); + msg = iu->buf; + if (unlikely(le16_to_cpu(msg->type) != RTRS_MSG_INFO_REQ)) { + rtrs_err(s, "Sess info request is malformed: type %d\n", + le16_to_cpu(msg->type)); + goto close; + } + err = process_info_req(con, msg); + if (unlikely(err)) + goto close; + +out: + rtrs_iu_free(iu, DMA_FROM_DEVICE, sess->s.dev->ib_dev, 1); + return; +close: + close_sess(sess); + goto out; +} + +static int post_recv_info_req(struct rtrs_srv_con *con) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_iu *rx_iu; + int err; + + rx_iu = rtrs_iu_alloc(1, sizeof(struct rtrs_msg_info_req), + GFP_KERNEL, sess->s.dev->ib_dev, + DMA_FROM_DEVICE, rtrs_srv_info_req_done); + if (unlikely(!rx_iu)) + return -ENOMEM; + /* Prepare for getting info response */ + err = rtrs_iu_post_recv(&con->c, rx_iu); + if (unlikely(err)) { + rtrs_err(s, "rtrs_iu_post_recv(), err: %d\n", err); + rtrs_iu_free(rx_iu, DMA_FROM_DEVICE, sess->s.dev->ib_dev, 1); + return err; + } + + return 0; +} + +static int post_recv_io(struct rtrs_srv_con *con, size_t q_size) +{ + int i, err; + + for (i = 0; i < q_size; i++) { + err = rtrs_post_recv_empty(&con->c, &io_comp_cqe); + if (unlikely(err)) + return err; + } + + return 0; +} + +static int post_recv_sess(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_sess *s = &sess->s; + size_t q_size; + int err, cid; + + for (cid = 0; cid < sess->s.con_num; cid++) { + if (cid == 0) + q_size = SERVICE_CON_QUEUE_DEPTH; + else + q_size = srv->queue_depth; + + err = post_recv_io(to_srv_con(sess->s.con[cid]), q_size); + if (unlikely(err)) { + rtrs_err(s, "post_recv_io(), err: %d\n", err); + return err; + } + } + + return 0; +} + +static void process_read(struct rtrs_srv_con *con, + struct rtrs_msg_rdma_read *msg, + u32 buf_id, u32 off) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_ctx *ctx = srv->ctx; + struct rtrs_srv_op *id; + + size_t usr_len, data_len; + void *data; + int ret; + + if (unlikely(sess->state != RTRS_SRV_CONNECTED)) { + rtrs_err_rl(s, + "Processing read request failed, session is disconnected, sess state %s\n", + rtrs_srv_state_str(sess->state)); + return; + } + if (unlikely(msg->sg_cnt != 1 && msg->sg_cnt != 0)) { + rtrs_err_rl(s, + "Processing read request failed, invalid message\n"); + return; + } + rtrs_srv_get_ops_ids(sess); + rtrs_srv_update_rdma_stats(sess->stats, off, READ); + id = sess->ops_ids[buf_id]; + id->con = con; + id->dir = READ; + id->msg_id = buf_id; + id->rd_msg = msg; + usr_len = le16_to_cpu(msg->usr_len); + data_len = off - usr_len; + data = page_address(srv->chunks[buf_id]); + ret = ctx->ops.rdma_ev(srv, srv->priv, id, READ, data, data_len, + data + data_len, usr_len); + + if (unlikely(ret)) { + rtrs_err_rl(s, + "Processing read request failed, user module cb reported for msg_id %d, err: %d\n", + buf_id, ret); + goto send_err_msg; + } + + return; + +send_err_msg: + ret = send_io_resp_imm(con, id, ret); + if (ret < 0) { + rtrs_err_rl(s, + "Sending err msg for failed RDMA-Write-Req failed, msg_id %d, err: %d\n", + buf_id, ret); + close_sess(sess); + } + rtrs_srv_put_ops_ids(sess); +} + +static void process_write(struct rtrs_srv_con *con, + struct rtrs_msg_rdma_write *req, + u32 buf_id, u32 off) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_srv *srv = sess->srv; + struct rtrs_srv_ctx *ctx = srv->ctx; + struct rtrs_srv_op *id; + + size_t data_len, usr_len; + void *data; + int ret; + + if (unlikely(sess->state != RTRS_SRV_CONNECTED)) { + rtrs_err_rl(s, + "Processing write request failed, session is disconnected, sess state %s\n", + rtrs_srv_state_str(sess->state)); + return; + } + rtrs_srv_get_ops_ids(sess); + rtrs_srv_update_rdma_stats(sess->stats, off, WRITE); + id = sess->ops_ids[buf_id]; + id->con = con; + id->dir = WRITE; + id->msg_id = buf_id; + + usr_len = le16_to_cpu(req->usr_len); + data_len = off - usr_len; + data = page_address(srv->chunks[buf_id]); + ret = ctx->ops.rdma_ev(srv, srv->priv, id, WRITE, data, data_len, + data + data_len, usr_len); + if (unlikely(ret)) { + rtrs_err_rl(s, + "Processing write request failed, user module callback reports err: %d\n", + ret); + goto send_err_msg; + } + + return; + +send_err_msg: + ret = send_io_resp_imm(con, id, ret); + if (ret < 0) { + rtrs_err_rl(s, + "Processing write request failed, sending I/O response failed, msg_id %d, err: %d\n", + buf_id, ret); + close_sess(sess); + } + rtrs_srv_put_ops_ids(sess); +} + +static void process_io_req(struct rtrs_srv_con *con, void *msg, + u32 id, u32 off) +{ + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_msg_rdma_hdr *hdr; + unsigned int type; + + ib_dma_sync_single_for_cpu(sess->s.dev->ib_dev, sess->dma_addr[id], + max_chunk_size, DMA_BIDIRECTIONAL); + hdr = msg; + type = le16_to_cpu(hdr->type); + + switch (type) { + case RTRS_MSG_WRITE: + process_write(con, msg, id, off); + break; + case RTRS_MSG_READ: + process_read(con, msg, id, off); + break; + default: + rtrs_err(s, + "Processing I/O request failed, unknown message type received: 0x%02x\n", + type); + goto err; + } + + return; + +err: + close_sess(sess); +} + +static void rtrs_srv_inv_rkey_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_srv_mr *mr = + container_of(wc->wr_cqe, typeof(*mr), inv_cqe); + struct rtrs_srv_con *con = cq->cq_context; + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_srv *srv = sess->srv; + u32 msg_id, off; + void *data; + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + rtrs_err(s, "Failed IB_WR_LOCAL_INV: %s\n", + ib_wc_status_msg(wc->status)); + close_sess(sess); + } + msg_id = mr->msg_id; + off = mr->msg_off; + data = page_address(srv->chunks[msg_id]) + off; + process_io_req(con, data, msg_id, off); +} + +static int rtrs_srv_inv_rkey(struct rtrs_srv_con *con, + struct rtrs_srv_mr *mr) +{ + struct ib_send_wr wr = { + .opcode = IB_WR_LOCAL_INV, + .wr_cqe = &mr->inv_cqe, + .send_flags = IB_SEND_SIGNALED, + .ex.invalidate_rkey = mr->mr->rkey, + }; + mr->inv_cqe.done = rtrs_srv_inv_rkey_done; + + return ib_post_send(con->c.qp, &wr, NULL); +} + +static void rtrs_rdma_process_wr_wait_list(struct rtrs_srv_con *con) +{ + spin_lock(&con->rsp_wr_wait_lock); + while (!list_empty(&con->rsp_wr_wait_list)) { + struct rtrs_srv_op *id; + int ret; + + id = list_entry(con->rsp_wr_wait_list.next, + struct rtrs_srv_op, wait_list); + list_del(&id->wait_list); + + spin_unlock(&con->rsp_wr_wait_lock); + ret = rtrs_srv_resp_rdma(id, id->status); + spin_lock(&con->rsp_wr_wait_lock); + + if (!ret) { + list_add(&id->wait_list, &con->rsp_wr_wait_list); + break; + } + } + spin_unlock(&con->rsp_wr_wait_lock); +} + +static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rtrs_srv_con *con = cq->cq_context; + struct rtrs_sess *s = con->c.sess; + struct rtrs_srv_sess *sess = to_srv_sess(s); + struct rtrs_srv *srv = sess->srv; + u32 imm_type, imm_payload; + int err; + + if (unlikely(wc->status != IB_WC_SUCCESS)) { + if (wc->status != IB_WC_WR_FLUSH_ERR) { + rtrs_err(s, + "%s (wr_cqe: %p, type: %d, vendor_err: 0x%x, len: %u)\n", + ib_wc_status_msg(wc->status), wc->wr_cqe, + wc->opcode, wc->vendor_err, wc->byte_len); + close_sess(sess); + } + return; + } + + switch (wc->opcode) { + case IB_WC_RECV_RDMA_WITH_IMM: + /* + * post_recv() RDMA write completions of IO reqs (read/write) + * and hb + */ + if (WARN_ON(wc->wr_cqe != &io_comp_cqe)) + return; + err = rtrs_post_recv_empty(&con->c, &io_comp_cqe); + if (unlikely(err)) { + rtrs_err(s, "rtrs_post_recv(), err: %d\n", err); + close_sess(sess); + break; + } + rtrs_from_imm(be32_to_cpu(wc->ex.imm_data), + &imm_type, &imm_payload); + if (likely(imm_type == RTRS_IO_REQ_IMM)) { + u32 msg_id, off; + void *data; + + msg_id = imm_payload >> sess->mem_bits; + off = imm_payload & ((1 << sess->mem_bits) - 1); + if (unlikely(msg_id >= srv->queue_depth || + off >= max_chunk_size)) { + rtrs_err(s, "Wrong msg_id %u, off %u\n", + msg_id, off); + close_sess(sess); + return; + } + if (always_invalidate) { + struct rtrs_srv_mr *mr = &sess->mrs[msg_id]; + + mr->msg_off = off; + mr->msg_id = msg_id; + err = rtrs_srv_inv_rkey(con, mr); + if (unlikely(err)) { + rtrs_err(s, "rtrs_post_recv(), err: %d\n", + err); + close_sess(sess); + break; + } + } else { + data = page_address(srv->chunks[msg_id]) + off; + process_io_req(con, data, msg_id, off); + } + } else if (imm_type == RTRS_HB_MSG_IMM) { + WARN_ON(con->c.cid); + rtrs_send_hb_ack(&sess->s); + } else if (imm_type == RTRS_HB_ACK_IMM) { + WARN_ON(con->c.cid); + sess->s.hb_missed_cnt = 0; + } else { + rtrs_wrn(s, "Unknown IMM type %u\n", imm_type); + } + break; + case IB_WC_RDMA_WRITE: + case IB_WC_SEND: + /* + * post_send() RDMA write completions of IO reqs (read/write) + * and hb + */ + atomic_add(srv->queue_depth, &con->sq_wr_avail); + + if (unlikely(!list_empty_careful(&con->rsp_wr_wait_list))) + rtrs_rdma_process_wr_wait_list(con); + + break; + default: + rtrs_wrn(s, "Unexpected WC type: %d\n", wc->opcode); + return; + } +} + +/** + * rtrs_srv_get_sess_name() - Get rtrs_srv peer hostname. + * @srv: Session + * @sessname: Sessname buffer + * @len: Length of sessname buffer + */ +int rtrs_srv_get_sess_name(struct rtrs_srv *srv, char *sessname, size_t len) +{ + struct rtrs_srv_sess *sess; + int err = -ENOTCONN; + + mutex_lock(&srv->paths_mutex); + list_for_each_entry(sess, &srv->paths_list, s.entry) { + if (sess->state != RTRS_SRV_CONNECTED) + continue; + strlcpy(sessname, sess->s.sessname, + min_t(size_t, sizeof(sess->s.sessname), len)); + err = 0; + break; + } + mutex_unlock(&srv->paths_mutex); + + return err; +} +EXPORT_SYMBOL(rtrs_srv_get_sess_name); + +/** + * rtrs_srv_get_sess_qdepth() - Get rtrs_srv qdepth. + * @srv: Session + */ +int rtrs_srv_get_queue_depth(struct rtrs_srv *srv) +{ + return srv->queue_depth; +} +EXPORT_SYMBOL(rtrs_srv_get_queue_depth); + +static int find_next_bit_ring(struct rtrs_srv_sess *sess) +{ + struct ib_device *ib_dev = sess->s.dev->ib_dev; + int v; + + v = cpumask_next(sess->cur_cq_vector, &cq_affinity_mask); + if (v >= nr_cpu_ids || v >= ib_dev->num_comp_vectors) + v = cpumask_first(&cq_affinity_mask); + return v; +} + +static int rtrs_srv_get_next_cq_vector(struct rtrs_srv_sess *sess) +{ + sess->cur_cq_vector = find_next_bit_ring(sess); + + return sess->cur_cq_vector; +} + +static void rtrs_srv_dev_release(struct device *dev) +{ + struct rtrs_srv *srv = container_of(dev, struct rtrs_srv, dev); + + kfree(srv); +} + +static struct rtrs_srv *__alloc_srv(struct rtrs_srv_ctx *ctx, + const uuid_t *paths_uuid) +{ + struct rtrs_srv *srv; + int i; + + srv = kzalloc(sizeof(*srv), GFP_KERNEL); + if (!srv) + return NULL; + + refcount_set(&srv->refcount, 1); + INIT_LIST_HEAD(&srv->paths_list); + mutex_init(&srv->paths_mutex); + mutex_init(&srv->paths_ev_mutex); + uuid_copy(&srv->paths_uuid, paths_uuid); + srv->queue_depth = sess_queue_depth; + srv->ctx = ctx; + device_initialize(&srv->dev); + srv->dev.release = rtrs_srv_dev_release; + + srv->chunks = kcalloc(srv->queue_depth, sizeof(*srv->chunks), + GFP_KERNEL); + if (!srv->chunks) + goto err_free_srv; + + for (i = 0; i < srv->queue_depth; i++) { + srv->chunks[i] = mempool_alloc(chunk_pool, GFP_KERNEL); + if (!srv->chunks[i]) + goto err_free_chunks; + } + list_add(&srv->ctx_list, &ctx->srv_list); + + return srv; + +err_free_chunks: + while (i--) + mempool_free(srv->chunks[i], chunk_pool); + kfree(srv->chunks); + +err_free_srv: + kfree(srv); + + return NULL; +} + +static void free_srv(struct rtrs_srv *srv) +{ + int i; + + WARN_ON(refcount_read(&srv->refcount)); + for (i = 0; i < srv->queue_depth; i++) + mempool_free(srv->chunks[i], chunk_pool); + kfree(srv->chunks); + mutex_destroy(&srv->paths_mutex); + mutex_destroy(&srv->paths_ev_mutex); + /* last put to release the srv structure */ + put_device(&srv->dev); +} + +static inline struct rtrs_srv *__find_srv_and_get(struct rtrs_srv_ctx *ctx, + const uuid_t *paths_uuid) +{ + struct rtrs_srv *srv; + + list_for_each_entry(srv, &ctx->srv_list, ctx_list) { + if (uuid_equal(&srv->paths_uuid, paths_uuid) && + refcount_inc_not_zero(&srv->refcount)) + return srv; + } + + return NULL; +} + +static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, + const uuid_t *paths_uuid) +{ + struct rtrs_srv *srv; + + mutex_lock(&ctx->srv_mutex); + srv = __find_srv_and_get(ctx, paths_uuid); + if (!srv) + srv = __alloc_srv(ctx, paths_uuid); + mutex_unlock(&ctx->srv_mutex); + + return srv; +} + +static void put_srv(struct rtrs_srv *srv) +{ + if (refcount_dec_and_test(&srv->refcount)) { + struct rtrs_srv_ctx *ctx = srv->ctx; + + WARN_ON(srv->dev.kobj.state_in_sysfs); + + mutex_lock(&ctx->srv_mutex); + list_del(&srv->ctx_list); + mutex_unlock(&ctx->srv_mutex); + free_srv(srv); + } +} + +static void __add_path_to_srv(struct rtrs_srv *srv, + struct rtrs_srv_sess *sess) +{ + list_add_tail(&sess->s.entry, &srv->paths_list); + srv->paths_num++; + WARN_ON(srv->paths_num >= MAX_PATHS_NUM); +} + +static void del_path_from_srv(struct rtrs_srv_sess *sess) +{ + struct rtrs_srv *srv = sess->srv; + + if (WARN_ON(!srv)) + return; + + mutex_lock(&srv->paths_mutex); + list_del(&sess->s.entry); + WARN_ON(!srv->paths_num); + srv->paths_num--; + mutex_unlock(&srv->paths_mutex); +} + +/* return true if addresses are the same, error other wise */ +static int sockaddr_cmp(const struct sockaddr *a, const struct sockaddr *b) +{ + switch (a->sa_family) { + case AF_IB: + return memcmp(&((struct sockaddr_ib *)a)->sib_addr, + &((struct sockaddr_ib *)b)->sib_addr, + sizeof(struct ib_addr)) && + (b->sa_family == AF_IB); + case AF_INET: + return memcmp(&((struct sockaddr_in *)a)->sin_addr, + &((struct sockaddr_in *)b)->sin_addr, + sizeof(struct in_addr)) && + (b->sa_family == AF_INET); + case AF_INET6: + return memcmp(&((struct sockaddr_in6 *)a)->sin6_addr, + &((struct sockaddr_in6 *)b)->sin6_addr, + sizeof(struct in6_addr)) && + (b->sa_family == AF_INET6); + default: + return -ENOENT; + } +} + +static bool __is_path_w_addr_exists(struct rtrs_srv *srv, + struct rdma_addr *addr) +{ + struct rtrs_srv_sess *sess; + + list_for_each_entry(sess, &srv->paths_list, s.entry) + if (!sockaddr_cmp((struct sockaddr *)&sess->s.dst_addr, + (struct sockaddr *)&addr->dst_addr) && + !sockaddr_cmp((struct sockaddr *)&sess->s.src_addr, + (struct sockaddr *)&addr->src_addr)) + return true; + + return false; +} + +static void free_sess(struct rtrs_srv_sess *sess) +{ + if (sess->kobj.state_in_sysfs) + kobject_put(&sess->kobj); + else + kfree(sess); +} + +static void rtrs_srv_close_work(struct work_struct *work) +{ + struct rtrs_srv_sess *sess; + struct rtrs_srv_con *con; + int i; + + sess = container_of(work, typeof(*sess), close_work); + + rtrs_srv_destroy_sess_files(sess); + rtrs_srv_stop_hb(sess); + + for (i = 0; i < sess->s.con_num; i++) { + if (!sess->s.con[i]) + continue; + con = to_srv_con(sess->s.con[i]); + rdma_disconnect(con->c.cm_id); + ib_drain_qp(con->c.qp); + } + /* Wait for all inflights */ + rtrs_srv_wait_ops_ids(sess); + + /* Notify upper layer if we are the last path */ + rtrs_srv_sess_down(sess); + + unmap_cont_bufs(sess); + rtrs_srv_free_ops_ids(sess); + + for (i = 0; i < sess->s.con_num; i++) { + if (!sess->s.con[i]) + continue; + con = to_srv_con(sess->s.con[i]); + rtrs_cq_qp_destroy(&con->c); + rdma_destroy_id(con->c.cm_id); + kfree(con); + } + rtrs_ib_dev_put(sess->s.dev); + + del_path_from_srv(sess); + put_srv(sess->srv); + sess->srv = NULL; + rtrs_srv_change_state(sess, RTRS_SRV_CLOSED); + + kfree(sess->dma_addr); + kfree(sess->s.con); + free_sess(sess); +} + +static int rtrs_rdma_do_accept(struct rtrs_srv_sess *sess, + struct rdma_cm_id *cm_id) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_msg_conn_rsp msg; + struct rdma_conn_param param; + int err; + + param = (struct rdma_conn_param) { + .rnr_retry_count = 7, + .private_data = &msg, + .private_data_len = sizeof(msg), + }; + + msg = (struct rtrs_msg_conn_rsp) { + .magic = cpu_to_le16(RTRS_MAGIC), + .version = cpu_to_le16(RTRS_PROTO_VER), + .queue_depth = cpu_to_le16(srv->queue_depth), + .max_io_size = cpu_to_le32(max_chunk_size - MAX_HDR_SIZE), + .max_hdr_size = cpu_to_le32(MAX_HDR_SIZE), + }; + + if (always_invalidate) + msg.flags = cpu_to_le32(RTRS_MSG_NEW_RKEY_F); + + err = rdma_accept(cm_id, ¶m); + if (err) + pr_err("rdma_accept(), err: %d\n", err); + + return err; +} + +static int rtrs_rdma_do_reject(struct rdma_cm_id *cm_id, int errno) +{ + struct rtrs_msg_conn_rsp msg; + int err; + + msg = (struct rtrs_msg_conn_rsp) { + .magic = cpu_to_le16(RTRS_MAGIC), + .version = cpu_to_le16(RTRS_PROTO_VER), + .errno = cpu_to_le16(errno), + }; + + err = rdma_reject(cm_id, &msg, sizeof(msg), IB_CM_REJ_CONSUMER_DEFINED); + if (err) + pr_err("rdma_reject(), err: %d\n", err); + + /* Bounce errno back */ + return errno; +} + +static struct rtrs_srv_sess * +__find_sess(struct rtrs_srv *srv, const uuid_t *sess_uuid) +{ + struct rtrs_srv_sess *sess; + + list_for_each_entry(sess, &srv->paths_list, s.entry) { + if (uuid_equal(&sess->s.uuid, sess_uuid)) + return sess; + } + + return NULL; +} + +static int create_con(struct rtrs_srv_sess *sess, + struct rdma_cm_id *cm_id, + unsigned int cid) +{ + struct rtrs_srv *srv = sess->srv; + struct rtrs_sess *s = &sess->s; + struct rtrs_srv_con *con; + + u16 cq_size, wr_queue_size; + int err, cq_vector; + + con = kzalloc(sizeof(*con), GFP_KERNEL); + if (!con) { + err = -ENOMEM; + goto err; + } + + spin_lock_init(&con->rsp_wr_wait_lock); + INIT_LIST_HEAD(&con->rsp_wr_wait_list); + con->c.cm_id = cm_id; + con->c.sess = &sess->s; + con->c.cid = cid; + atomic_set(&con->wr_cnt, 0); + + if (con->c.cid == 0) { + /* + * All receive and all send (each requiring invalidate) + * + 2 for drain and heartbeat + */ + wr_queue_size = SERVICE_CON_QUEUE_DEPTH * 3 + 2; + cq_size = wr_queue_size; + } else { + /* + * If we have all receive requests posted and + * all write requests posted and each read request + * requires an invalidate request + drain + * and qp gets into error state. + */ + cq_size = srv->queue_depth * 3 + 1; + /* + * In theory we might have queue_depth * 32 + * outstanding requests if an unsafe global key is used + * and we have queue_depth read requests each consisting + * of 32 different addresses. div 3 for mlx5. + */ + wr_queue_size = sess->s.dev->ib_dev->attrs.max_qp_wr / 3; + } + atomic_set(&con->sq_wr_avail, wr_queue_size); + cq_vector = rtrs_srv_get_next_cq_vector(sess); + + /* TODO: SOFTIRQ can be faster, but be careful with softirq context */ + err = rtrs_cq_qp_create(&sess->s, &con->c, 1, cq_vector, cq_size, + wr_queue_size, IB_POLL_WORKQUEUE); + if (err) { + rtrs_err(s, "rtrs_cq_qp_create(), err: %d\n", err); + goto free_con; + } + if (con->c.cid == 0) { + err = post_recv_info_req(con); + if (err) + goto free_cqqp; + } + WARN_ON(sess->s.con[cid]); + sess->s.con[cid] = &con->c; + + /* + * Change context from server to current connection. The other + * way is to use cm_id->qp->qp_context, which does not work on OFED. + */ + cm_id->context = &con->c; + + return 0; + +free_cqqp: + rtrs_cq_qp_destroy(&con->c); +free_con: + kfree(con); + +err: + return err; +} + +static struct rtrs_srv_sess *__alloc_sess(struct rtrs_srv *srv, + struct rdma_cm_id *cm_id, + unsigned int con_num, + unsigned int recon_cnt, + const uuid_t *uuid) +{ + struct rtrs_srv_sess *sess; + int err = -ENOMEM; + + if (srv->paths_num >= MAX_PATHS_NUM) { + err = -ECONNRESET; + goto err; + } + if (__is_path_w_addr_exists(srv, &cm_id->route.addr)) { + err = -EEXIST; + pr_err("Path with same addr exists\n"); + goto err; + } + sess = kzalloc(sizeof(*sess), GFP_KERNEL); + if (!sess) + goto err; + + sess->stats = kzalloc(sizeof(*sess->stats), GFP_KERNEL); + if (!sess->stats) + goto err_free_sess; + + sess->stats->sess = sess; + + sess->dma_addr = kcalloc(srv->queue_depth, sizeof(*sess->dma_addr), + GFP_KERNEL); + if (!sess->dma_addr) + goto err_free_stats; + + sess->s.con = kcalloc(con_num, sizeof(*sess->s.con), GFP_KERNEL); + if (!sess->s.con) + goto err_free_dma_addr; + + sess->state = RTRS_SRV_CONNECTING; + sess->srv = srv; + sess->cur_cq_vector = -1; + sess->s.dst_addr = cm_id->route.addr.dst_addr; + sess->s.src_addr = cm_id->route.addr.src_addr; + sess->s.con_num = con_num; + sess->s.recon_cnt = recon_cnt; + uuid_copy(&sess->s.uuid, uuid); + spin_lock_init(&sess->state_lock); + INIT_WORK(&sess->close_work, rtrs_srv_close_work); + rtrs_srv_init_hb(sess); + + sess->s.dev = rtrs_ib_dev_find_or_add(cm_id->device, &dev_pd); + if (!sess->s.dev) { + err = -ENOMEM; + goto err_free_con; + } + err = map_cont_bufs(sess); + if (err) + goto err_put_dev; + + err = rtrs_srv_alloc_ops_ids(sess); + if (err) + goto err_unmap_bufs; + + __add_path_to_srv(srv, sess); + + return sess; + +err_unmap_bufs: + unmap_cont_bufs(sess); +err_put_dev: + rtrs_ib_dev_put(sess->s.dev); +err_free_con: + kfree(sess->s.con); +err_free_dma_addr: + kfree(sess->dma_addr); +err_free_stats: + kfree(sess->stats); +err_free_sess: + kfree(sess); +err: + return ERR_PTR(err); +} + +static int rtrs_rdma_connect(struct rdma_cm_id *cm_id, + const struct rtrs_msg_conn_req *msg, + size_t len) +{ + struct rtrs_srv_ctx *ctx = cm_id->context; + struct rtrs_srv_sess *sess; + struct rtrs_srv *srv; + + u16 version, con_num, cid; + u16 recon_cnt; + int err; + + if (len < sizeof(*msg)) { + pr_err("Invalid RTRS connection request\n"); + goto reject_w_econnreset; + } + if (le16_to_cpu(msg->magic) != RTRS_MAGIC) { + pr_err("Invalid RTRS magic\n"); + goto reject_w_econnreset; + } + version = le16_to_cpu(msg->version); + if (version >> 8 != RTRS_PROTO_VER_MAJOR) { + pr_err("Unsupported major RTRS version: %d, expected %d\n", + version >> 8, RTRS_PROTO_VER_MAJOR); + goto reject_w_econnreset; + } + con_num = le16_to_cpu(msg->cid_num); + if (con_num > 4096) { + /* Sanity check */ + pr_err("Too many connections requested: %d\n", con_num); + goto reject_w_econnreset; + } + cid = le16_to_cpu(msg->cid); + if (cid >= con_num) { + /* Sanity check */ + pr_err("Incorrect cid: %d >= %d\n", cid, con_num); + goto reject_w_econnreset; + } + recon_cnt = le16_to_cpu(msg->recon_cnt); + srv = get_or_create_srv(ctx, &msg->paths_uuid); + if (!srv) { + err = -ENOMEM; + goto reject_w_err; + } + mutex_lock(&srv->paths_mutex); + sess = __find_sess(srv, &msg->sess_uuid); + if (sess) { + struct rtrs_sess *s = &sess->s; + + /* Session already holds a reference */ + put_srv(srv); + + if (sess->state != RTRS_SRV_CONNECTING) { + rtrs_err(s, "Session in wrong state: %s\n", + rtrs_srv_state_str(sess->state)); + mutex_unlock(&srv->paths_mutex); + goto reject_w_econnreset; + } + /* + * Sanity checks + */ + if (con_num != s->con_num || cid >= s->con_num) { + rtrs_err(s, "Incorrect request: %d, %d\n", + cid, con_num); + mutex_unlock(&srv->paths_mutex); + goto reject_w_econnreset; + } + if (s->con[cid]) { + rtrs_err(s, "Connection already exists: %d\n", + cid); + mutex_unlock(&srv->paths_mutex); + goto reject_w_econnreset; + } + } else { + sess = __alloc_sess(srv, cm_id, con_num, recon_cnt, + &msg->sess_uuid); + if (IS_ERR(sess)) { + mutex_unlock(&srv->paths_mutex); + put_srv(srv); + err = PTR_ERR(sess); + goto reject_w_err; + } + } + err = create_con(sess, cm_id, cid); + if (err) { + (void)rtrs_rdma_do_reject(cm_id, err); + /* + * Since session has other connections we follow normal way + * through workqueue, but still return an error to tell cma.c + * to call rdma_destroy_id() for current connection. + */ + goto close_and_return_err; + } + err = rtrs_rdma_do_accept(sess, cm_id); + if (err) { + (void)rtrs_rdma_do_reject(cm_id, err); + /* + * Since current connection was successfully added to the + * session we follow normal way through workqueue to close the + * session, thus return 0 to tell cma.c we call + * rdma_destroy_id() ourselves. + */ + err = 0; + goto close_and_return_err; + } + mutex_unlock(&srv->paths_mutex); + + return 0; + +reject_w_err: + return rtrs_rdma_do_reject(cm_id, err); + +reject_w_econnreset: + return rtrs_rdma_do_reject(cm_id, -ECONNRESET); + +close_and_return_err: + close_sess(sess); + mutex_unlock(&srv->paths_mutex); + + return err; +} + +static int rtrs_srv_rdma_cm_handler(struct rdma_cm_id *cm_id, + struct rdma_cm_event *ev) +{ + struct rtrs_srv_sess *sess = NULL; + struct rtrs_sess *s = NULL; + + if (ev->event != RDMA_CM_EVENT_CONNECT_REQUEST) { + struct rtrs_con *c = cm_id->context; + + s = c->sess; + sess = to_srv_sess(s); + } + + switch (ev->event) { + case RDMA_CM_EVENT_CONNECT_REQUEST: + /* + * In case of error cma.c will destroy cm_id, + * see cma_process_remove() + */ + return rtrs_rdma_connect(cm_id, ev->param.conn.private_data, + ev->param.conn.private_data_len); + case RDMA_CM_EVENT_ESTABLISHED: + /* Nothing here */ + break; + case RDMA_CM_EVENT_REJECTED: + case RDMA_CM_EVENT_CONNECT_ERROR: + case RDMA_CM_EVENT_UNREACHABLE: + rtrs_err(s, "CM error (CM event: %s, err: %d)\n", + rdma_event_msg(ev->event), ev->status); + close_sess(sess); + break; + case RDMA_CM_EVENT_DISCONNECTED: + case RDMA_CM_EVENT_ADDR_CHANGE: + case RDMA_CM_EVENT_TIMEWAIT_EXIT: + close_sess(sess); + break; + case RDMA_CM_EVENT_DEVICE_REMOVAL: + close_sess(sess); + break; + default: + pr_err("Ignoring unexpected CM event %s, err %d\n", + rdma_event_msg(ev->event), ev->status); + break; + } + + return 0; +} + +static struct rdma_cm_id *rtrs_srv_cm_init(struct rtrs_srv_ctx *ctx, + struct sockaddr *addr, + enum rdma_ucm_port_space ps) +{ + struct rdma_cm_id *cm_id; + int ret; + + cm_id = rdma_create_id(&init_net, rtrs_srv_rdma_cm_handler, + ctx, ps, IB_QPT_RC); + if (IS_ERR(cm_id)) { + ret = PTR_ERR(cm_id); + pr_err("Creating id for RDMA connection failed, err: %d\n", + ret); + goto err_out; + } + ret = rdma_bind_addr(cm_id, addr); + if (ret) { + pr_err("Binding RDMA address failed, err: %d\n", ret); + goto err_cm; + } + ret = rdma_listen(cm_id, 64); + if (ret) { + pr_err("Listening on RDMA connection failed, err: %d\n", + ret); + goto err_cm; + } + + return cm_id; + +err_cm: + rdma_destroy_id(cm_id); +err_out: + + return ERR_PTR(ret); +} + +static int rtrs_srv_rdma_init(struct rtrs_srv_ctx *ctx, u16 port) +{ + struct sockaddr_in6 sin = { + .sin6_family = AF_INET6, + .sin6_addr = IN6ADDR_ANY_INIT, + .sin6_port = htons(port), + }; + struct sockaddr_ib sib = { + .sib_family = AF_IB, + .sib_sid = cpu_to_be64(RDMA_IB_IP_PS_IB | port), + .sib_sid_mask = cpu_to_be64(0xffffffffffffffffULL), + .sib_pkey = cpu_to_be16(0xffff), + }; + struct rdma_cm_id *cm_ip, *cm_ib; + int ret; + + /* + * We accept both IPoIB and IB connections, so we need to keep + * two cm id's, one for each socket type and port space. + * If the cm initialization of one of the id's fails, we abort + * everything. + */ + cm_ip = rtrs_srv_cm_init(ctx, (struct sockaddr *)&sin, RDMA_PS_TCP); + if (IS_ERR(cm_ip)) + return PTR_ERR(cm_ip); + + cm_ib = rtrs_srv_cm_init(ctx, (struct sockaddr *)&sib, RDMA_PS_IB); + if (IS_ERR(cm_ib)) { + ret = PTR_ERR(cm_ib); + goto free_cm_ip; + } + + ctx->cm_id_ip = cm_ip; + ctx->cm_id_ib = cm_ib; + + return 0; + +free_cm_ip: + rdma_destroy_id(cm_ip); + + return ret; +} + +static struct rtrs_srv_ctx *alloc_srv_ctx(struct rtrs_srv_ops *ops) +{ + struct rtrs_srv_ctx *ctx; + + ctx = kzalloc(sizeof(*ctx), GFP_KERNEL); + if (!ctx) + return NULL; + + ctx->ops = *ops; + mutex_init(&ctx->srv_mutex); + INIT_LIST_HEAD(&ctx->srv_list); + + return ctx; +} + +static void free_srv_ctx(struct rtrs_srv_ctx *ctx) +{ + WARN_ON(!list_empty(&ctx->srv_list)); + mutex_destroy(&ctx->srv_mutex); + kfree(ctx); +} + +/** + * rtrs_srv_open() - open RTRS server context + * @ops: callback functions + * @port: port to listen on + * + * Creates server context with specified callbacks. + * + * Return a valid pointer on success otherwise PTR_ERR. + */ +struct rtrs_srv_ctx *rtrs_srv_open(struct rtrs_srv_ops *ops, u16 port) +{ + struct rtrs_srv_ctx *ctx; + int err; + + ctx = alloc_srv_ctx(ops); + if (!ctx) + return ERR_PTR(-ENOMEM); + + err = rtrs_srv_rdma_init(ctx, port); + if (err) { + free_srv_ctx(ctx); + return ERR_PTR(err); + } + + return ctx; +} +EXPORT_SYMBOL(rtrs_srv_open); + +static void close_sessions(struct rtrs_srv *srv) +{ + struct rtrs_srv_sess *sess; + + mutex_lock(&srv->paths_mutex); + list_for_each_entry(sess, &srv->paths_list, s.entry) + close_sess(sess); + mutex_unlock(&srv->paths_mutex); +} + +static void close_ctx(struct rtrs_srv_ctx *ctx) +{ + struct rtrs_srv *srv; + + mutex_lock(&ctx->srv_mutex); + list_for_each_entry(srv, &ctx->srv_list, ctx_list) + close_sessions(srv); + mutex_unlock(&ctx->srv_mutex); + flush_workqueue(rtrs_wq); +} + +/** + * rtrs_srv_close() - close RTRS server context + * @ctx: pointer to server context + * + * Closes RTRS server context with all client sessions. + */ +void rtrs_srv_close(struct rtrs_srv_ctx *ctx) +{ + rdma_destroy_id(ctx->cm_id_ip); + rdma_destroy_id(ctx->cm_id_ib); + close_ctx(ctx); + free_srv_ctx(ctx); +} +EXPORT_SYMBOL(rtrs_srv_close); + +static int check_module_params(void) +{ + if (sess_queue_depth < 1 || sess_queue_depth > MAX_SESS_QUEUE_DEPTH) { + pr_err("Invalid sess_queue_depth value %d, has to be >= %d, <= %d.\n", + sess_queue_depth, 1, MAX_SESS_QUEUE_DEPTH); + return -EINVAL; + } + if (max_chunk_size < 4096 || !is_power_of_2(max_chunk_size)) { + pr_err("Invalid max_chunk_size value %d, has to be >= %d and should be power of two.\n", + max_chunk_size, 4096); + return -EINVAL; + } + + /* + * Check if IB immediate data size is enough to hold the mem_id and the + * offset inside the memory chunk + */ + if ((ilog2(sess_queue_depth - 1) + 1) + + (ilog2(max_chunk_size - 1) + 1) > MAX_IMM_PAYL_BITS) { + pr_err("RDMA immediate size (%db) not enough to encode %d buffers of size %dB. Reduce 'sess_queue_depth' or 'max_chunk_size' parameters.\n", + MAX_IMM_PAYL_BITS, sess_queue_depth, max_chunk_size); + return -EINVAL; + } + + return 0; +} + +static int __init rtrs_server_init(void) +{ + int err; + + pr_info("Loading module %s, proto %s: (max_chunk_size: %d (pure IO %ld, headers %ld) , sess_queue_depth: %d, always_invalidate: %d)\n", + KBUILD_MODNAME, RTRS_PROTO_VER_STRING, + max_chunk_size, max_chunk_size - MAX_HDR_SIZE, MAX_HDR_SIZE, + sess_queue_depth, always_invalidate); + + rtrs_rdma_dev_pd_init(0, &dev_pd); + + err = check_module_params(); + if (err) { + pr_err("Failed to load module, invalid module parameters, err: %d\n", + err); + return err; + } + chunk_pool = mempool_create_page_pool(sess_queue_depth * CHUNK_POOL_SZ, + get_order(max_chunk_size)); + if (!chunk_pool) + return -ENOMEM; + rtrs_dev_class = class_create(THIS_MODULE, "rtrs-server"); + if (IS_ERR(rtrs_dev_class)) { + err = PTR_ERR(rtrs_dev_class); + goto out_chunk_pool; + } + rtrs_wq = alloc_workqueue("rtrs_server_wq", 0, 0); + if (!rtrs_wq) { + err = -ENOMEM; + goto out_dev_class; + } + + return 0; + +out_dev_class: + class_destroy(rtrs_dev_class); +out_chunk_pool: + mempool_destroy(chunk_pool); + + return err; +} + +static void __exit rtrs_server_exit(void) +{ + destroy_workqueue(rtrs_wq); + class_destroy(rtrs_dev_class); + mempool_destroy(chunk_pool); + rtrs_rdma_dev_pd_deinit(&dev_pd); +} + +module_init(rtrs_server_init); +module_exit(rtrs_server_exit); diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv.h b/drivers/infiniband/ulp/rtrs/rtrs-srv.h new file mode 100644 index 000000000000..dc95b0932f0d --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv.h @@ -0,0 +1,148 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ + +#ifndef RTRS_SRV_H +#define RTRS_SRV_H + +#include <linux/device.h> +#include <linux/refcount.h> +#include "rtrs-pri.h" + +/* + * enum rtrs_srv_state - Server states. + */ +enum rtrs_srv_state { + RTRS_SRV_CONNECTING, + RTRS_SRV_CONNECTED, + RTRS_SRV_CLOSING, + RTRS_SRV_CLOSED, +}; + +/* stats for Read and write operation. + * see Documentation/ABI/testing/sysfs-class-rtrs-server for details + */ +struct rtrs_srv_stats_rdma_stats { + struct { + atomic64_t cnt; + atomic64_t size_total; + } dir[2]; +}; + +struct rtrs_srv_stats { + struct kobject kobj_stats; + struct rtrs_srv_stats_rdma_stats rdma_stats; + struct rtrs_srv_sess *sess; +}; + +struct rtrs_srv_con { + struct rtrs_con c; + atomic_t wr_cnt; + atomic_t sq_wr_avail; + struct list_head rsp_wr_wait_list; + spinlock_t rsp_wr_wait_lock; +}; + +/* IO context in rtrs_srv, each io has one */ +struct rtrs_srv_op { + struct rtrs_srv_con *con; + u32 msg_id; + u8 dir; + struct rtrs_msg_rdma_read *rd_msg; + struct ib_rdma_wr tx_wr; + struct ib_sge tx_sg; + struct list_head wait_list; + int status; +}; + +/* + * server side memory region context, when always_invalidate=Y, we need + * queue_depth of memory regrion to invalidate each memory region. + */ +struct rtrs_srv_mr { + struct ib_mr *mr; + struct sg_table sgt; + struct ib_cqe inv_cqe; /* only for always_invalidate=true */ + u32 msg_id; /* only for always_invalidate=true */ + u32 msg_off; /* only for always_invalidate=true */ + struct rtrs_iu *iu; /* send buffer for new rkey msg */ +}; + +struct rtrs_srv_sess { + struct rtrs_sess s; + struct rtrs_srv *srv; + struct work_struct close_work; + enum rtrs_srv_state state; + spinlock_t state_lock; + int cur_cq_vector; + struct rtrs_srv_op **ops_ids; + atomic_t ids_inflight; + wait_queue_head_t ids_waitq; + struct rtrs_srv_mr *mrs; + unsigned int mrs_num; + dma_addr_t *dma_addr; + bool established; + unsigned int mem_bits; + struct kobject kobj; + struct rtrs_srv_stats *stats; +}; + +struct rtrs_srv { + struct list_head paths_list; + int paths_up; + struct mutex paths_ev_mutex; + size_t paths_num; + struct mutex paths_mutex; + uuid_t paths_uuid; + refcount_t refcount; + struct rtrs_srv_ctx *ctx; + struct list_head ctx_list; + void *priv; + size_t queue_depth; + struct page **chunks; + struct device dev; + unsigned int dev_ref; + struct kobject *kobj_paths; +}; + +struct rtrs_srv_ctx { + struct rtrs_srv_ops ops; + struct rdma_cm_id *cm_id_ip; + struct rdma_cm_id *cm_id_ib; + struct mutex srv_mutex; + struct list_head srv_list; +}; + +extern struct class *rtrs_dev_class; + +void close_sess(struct rtrs_srv_sess *sess); + +static inline void rtrs_srv_update_rdma_stats(struct rtrs_srv_stats *s, + size_t size, int d) +{ + atomic64_inc(&s->rdma_stats.dir[d].cnt); + atomic64_add(size, &s->rdma_stats.dir[d].size_total); +} + +/* functions which are implemented in rtrs-srv-stats.c */ +int rtrs_srv_reset_rdma_stats(struct rtrs_srv_stats *stats, bool enable); +ssize_t rtrs_srv_stats_rdma_to_str(struct rtrs_srv_stats *stats, + char *page, size_t len); +int rtrs_srv_reset_wc_completion_stats(struct rtrs_srv_stats *stats, + bool enable); +int rtrs_srv_stats_wc_completion_to_str(struct rtrs_srv_stats *stats, char *buf, + size_t len); +int rtrs_srv_reset_all_stats(struct rtrs_srv_stats *stats, bool enable); +ssize_t rtrs_srv_reset_all_help(struct rtrs_srv_stats *stats, + char *page, size_t len); + +/* functions which are implemented in rtrs-srv-sysfs.c */ +int rtrs_srv_create_sess_files(struct rtrs_srv_sess *sess); +void rtrs_srv_destroy_sess_files(struct rtrs_srv_sess *sess); + +#endif /* RTRS_SRV_H */ diff --git a/drivers/infiniband/ulp/rtrs/rtrs.c b/drivers/infiniband/ulp/rtrs/rtrs.c new file mode 100644 index 000000000000..ff1093d6e4bc --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs.c @@ -0,0 +1,612 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#undef pr_fmt +#define pr_fmt(fmt) KBUILD_MODNAME " L" __stringify(__LINE__) ": " fmt + +#include <linux/module.h> +#include <linux/inet.h> + +#include "rtrs-pri.h" +#include "rtrs-log.h" + +MODULE_DESCRIPTION("RDMA Transport Core"); +MODULE_LICENSE("GPL"); + +struct rtrs_iu *rtrs_iu_alloc(u32 queue_size, size_t size, gfp_t gfp_mask, + struct ib_device *dma_dev, + enum dma_data_direction dir, + void (*done)(struct ib_cq *cq, struct ib_wc *wc)) +{ + struct rtrs_iu *ius, *iu; + int i; + + ius = kcalloc(queue_size, sizeof(*ius), gfp_mask); + if (!ius) + return NULL; + for (i = 0; i < queue_size; i++) { + iu = &ius[i]; + iu->buf = kzalloc(size, gfp_mask); + if (!iu->buf) + goto err; + + iu->dma_addr = ib_dma_map_single(dma_dev, iu->buf, size, dir); + if (ib_dma_mapping_error(dma_dev, iu->dma_addr)) + goto err; + + iu->cqe.done = done; + iu->size = size; + iu->direction = dir; + } + return ius; +err: + rtrs_iu_free(ius, dir, dma_dev, i); + return NULL; +} +EXPORT_SYMBOL_GPL(rtrs_iu_alloc); + +void rtrs_iu_free(struct rtrs_iu *ius, enum dma_data_direction dir, + struct ib_device *ibdev, u32 queue_size) +{ + struct rtrs_iu *iu; + int i; + + if (!ius) + return; + + for (i = 0; i < queue_size; i++) { + iu = &ius[i]; + ib_dma_unmap_single(ibdev, iu->dma_addr, iu->size, dir); + kfree(iu->buf); + } + kfree(ius); +} +EXPORT_SYMBOL_GPL(rtrs_iu_free); + +int rtrs_iu_post_recv(struct rtrs_con *con, struct rtrs_iu *iu) +{ + struct rtrs_sess *sess = con->sess; + struct ib_recv_wr wr; + struct ib_sge list; + + list.addr = iu->dma_addr; + list.length = iu->size; + list.lkey = sess->dev->ib_pd->local_dma_lkey; + + if (list.length == 0) { + rtrs_wrn(con->sess, + "Posting receive work request failed, sg list is empty\n"); + return -EINVAL; + } + wr = (struct ib_recv_wr) { + .wr_cqe = &iu->cqe, + .sg_list = &list, + .num_sge = 1, + }; + + return ib_post_recv(con->qp, &wr, NULL); +} +EXPORT_SYMBOL_GPL(rtrs_iu_post_recv); + +int rtrs_post_recv_empty(struct rtrs_con *con, struct ib_cqe *cqe) +{ + struct ib_recv_wr wr; + + wr = (struct ib_recv_wr) { + .wr_cqe = cqe, + }; + + return ib_post_recv(con->qp, &wr, NULL); +} +EXPORT_SYMBOL_GPL(rtrs_post_recv_empty); + +int rtrs_iu_post_send(struct rtrs_con *con, struct rtrs_iu *iu, size_t size, + struct ib_send_wr *head) +{ + struct rtrs_sess *sess = con->sess; + struct ib_send_wr wr; + struct ib_sge list; + + if (WARN_ON(size == 0)) + return -EINVAL; + + list.addr = iu->dma_addr; + list.length = size; + list.lkey = sess->dev->ib_pd->local_dma_lkey; + + wr = (struct ib_send_wr) { + .wr_cqe = &iu->cqe, + .sg_list = &list, + .num_sge = 1, + .opcode = IB_WR_SEND, + .send_flags = IB_SEND_SIGNALED, + }; + + if (head) { + struct ib_send_wr *tail = head; + + while (tail->next) + tail = tail->next; + tail->next = ≀ + } else { + head = ≀ + } + + return ib_post_send(con->qp, head, NULL); +} +EXPORT_SYMBOL_GPL(rtrs_iu_post_send); + +int rtrs_iu_post_rdma_write_imm(struct rtrs_con *con, struct rtrs_iu *iu, + struct ib_sge *sge, unsigned int num_sge, + u32 rkey, u64 rdma_addr, u32 imm_data, + enum ib_send_flags flags, + struct ib_send_wr *head) +{ + struct ib_rdma_wr wr; + int i; + + wr = (struct ib_rdma_wr) { + .wr.wr_cqe = &iu->cqe, + .wr.sg_list = sge, + .wr.num_sge = num_sge, + .rkey = rkey, + .remote_addr = rdma_addr, + .wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM, + .wr.ex.imm_data = cpu_to_be32(imm_data), + .wr.send_flags = flags, + }; + + /* + * If one of the sges has 0 size, the operation will fail with a + * length error + */ + for (i = 0; i < num_sge; i++) + if (WARN_ON(sge[i].length == 0)) + return -EINVAL; + + if (head) { + struct ib_send_wr *tail = head; + + while (tail->next) + tail = tail->next; + tail->next = &wr.wr; + } else { + head = &wr.wr; + } + + return ib_post_send(con->qp, head, NULL); +} +EXPORT_SYMBOL_GPL(rtrs_iu_post_rdma_write_imm); + +int rtrs_post_rdma_write_imm_empty(struct rtrs_con *con, struct ib_cqe *cqe, + u32 imm_data, enum ib_send_flags flags, + struct ib_send_wr *head) +{ + struct ib_send_wr wr; + + wr = (struct ib_send_wr) { + .wr_cqe = cqe, + .send_flags = flags, + .opcode = IB_WR_RDMA_WRITE_WITH_IMM, + .ex.imm_data = cpu_to_be32(imm_data), + }; + + if (head) { + struct ib_send_wr *tail = head; + + while (tail->next) + tail = tail->next; + tail->next = ≀ + } else { + head = ≀ + } + + return ib_post_send(con->qp, head, NULL); +} +EXPORT_SYMBOL_GPL(rtrs_post_rdma_write_imm_empty); + +static void qp_event_handler(struct ib_event *ev, void *ctx) +{ + struct rtrs_con *con = ctx; + + switch (ev->event) { + case IB_EVENT_COMM_EST: + rtrs_info(con->sess, "QP event %s (%d) received\n", + ib_event_msg(ev->event), ev->event); + rdma_notify(con->cm_id, IB_EVENT_COMM_EST); + break; + default: + rtrs_info(con->sess, "Unhandled QP event %s (%d) received\n", + ib_event_msg(ev->event), ev->event); + break; + } +} + +static int create_cq(struct rtrs_con *con, int cq_vector, u16 cq_size, + enum ib_poll_context poll_ctx) +{ + struct rdma_cm_id *cm_id = con->cm_id; + struct ib_cq *cq; + + cq = ib_alloc_cq(cm_id->device, con, cq_size, + cq_vector, poll_ctx); + if (IS_ERR(cq)) { + rtrs_err(con->sess, "Creating completion queue failed, errno: %ld\n", + PTR_ERR(cq)); + return PTR_ERR(cq); + } + con->cq = cq; + + return 0; +} + +static int create_qp(struct rtrs_con *con, struct ib_pd *pd, + u16 wr_queue_size, u32 max_sge) +{ + struct ib_qp_init_attr init_attr = {NULL}; + struct rdma_cm_id *cm_id = con->cm_id; + int ret; + + init_attr.cap.max_send_wr = wr_queue_size; + init_attr.cap.max_recv_wr = wr_queue_size; + init_attr.cap.max_recv_sge = 1; + init_attr.event_handler = qp_event_handler; + init_attr.qp_context = con; + init_attr.cap.max_send_sge = max_sge; + + init_attr.qp_type = IB_QPT_RC; + init_attr.send_cq = con->cq; + init_attr.recv_cq = con->cq; + init_attr.sq_sig_type = IB_SIGNAL_REQ_WR; + + ret = rdma_create_qp(cm_id, pd, &init_attr); + if (ret) { + rtrs_err(con->sess, "Creating QP failed, err: %d\n", ret); + return ret; + } + con->qp = cm_id->qp; + + return ret; +} + +int rtrs_cq_qp_create(struct rtrs_sess *sess, struct rtrs_con *con, + u32 max_send_sge, int cq_vector, u16 cq_size, + u16 wr_queue_size, enum ib_poll_context poll_ctx) +{ + int err; + + err = create_cq(con, cq_vector, cq_size, poll_ctx); + if (err) + return err; + + err = create_qp(con, sess->dev->ib_pd, wr_queue_size, max_send_sge); + if (err) { + ib_free_cq(con->cq); + con->cq = NULL; + return err; + } + con->sess = sess; + + return 0; +} +EXPORT_SYMBOL_GPL(rtrs_cq_qp_create); + +void rtrs_cq_qp_destroy(struct rtrs_con *con) +{ + if (con->qp) { + rdma_destroy_qp(con->cm_id); + con->qp = NULL; + } + if (con->cq) { + ib_free_cq(con->cq); + con->cq = NULL; + } +} +EXPORT_SYMBOL_GPL(rtrs_cq_qp_destroy); + +static void schedule_hb(struct rtrs_sess *sess) +{ + queue_delayed_work(sess->hb_wq, &sess->hb_dwork, + msecs_to_jiffies(sess->hb_interval_ms)); +} + +void rtrs_send_hb_ack(struct rtrs_sess *sess) +{ + struct rtrs_con *usr_con = sess->con[0]; + u32 imm; + int err; + + imm = rtrs_to_imm(RTRS_HB_ACK_IMM, 0); + err = rtrs_post_rdma_write_imm_empty(usr_con, sess->hb_cqe, imm, + IB_SEND_SIGNALED, NULL); + if (err) { + sess->hb_err_handler(usr_con); + return; + } +} +EXPORT_SYMBOL_GPL(rtrs_send_hb_ack); + +static void hb_work(struct work_struct *work) +{ + struct rtrs_con *usr_con; + struct rtrs_sess *sess; + u32 imm; + int err; + + sess = container_of(to_delayed_work(work), typeof(*sess), hb_dwork); + usr_con = sess->con[0]; + + if (sess->hb_missed_cnt > sess->hb_missed_max) { + sess->hb_err_handler(usr_con); + return; + } + if (sess->hb_missed_cnt++) { + /* Reschedule work without sending hb */ + schedule_hb(sess); + return; + } + imm = rtrs_to_imm(RTRS_HB_MSG_IMM, 0); + err = rtrs_post_rdma_write_imm_empty(usr_con, sess->hb_cqe, imm, + IB_SEND_SIGNALED, NULL); + if (err) { + sess->hb_err_handler(usr_con); + return; + } + + schedule_hb(sess); +} + +void rtrs_init_hb(struct rtrs_sess *sess, struct ib_cqe *cqe, + unsigned int interval_ms, unsigned int missed_max, + void (*err_handler)(struct rtrs_con *con), + struct workqueue_struct *wq) +{ + sess->hb_cqe = cqe; + sess->hb_interval_ms = interval_ms; + sess->hb_err_handler = err_handler; + sess->hb_wq = wq; + sess->hb_missed_max = missed_max; + sess->hb_missed_cnt = 0; + INIT_DELAYED_WORK(&sess->hb_dwork, hb_work); +} +EXPORT_SYMBOL_GPL(rtrs_init_hb); + +void rtrs_start_hb(struct rtrs_sess *sess) +{ + schedule_hb(sess); +} +EXPORT_SYMBOL_GPL(rtrs_start_hb); + +void rtrs_stop_hb(struct rtrs_sess *sess) +{ + cancel_delayed_work_sync(&sess->hb_dwork); + sess->hb_missed_cnt = 0; + sess->hb_missed_max = 0; +} +EXPORT_SYMBOL_GPL(rtrs_stop_hb); + +static int rtrs_str_gid_to_sockaddr(const char *addr, size_t len, + short port, struct sockaddr_storage *dst) +{ + struct sockaddr_ib *dst_ib = (struct sockaddr_ib *)dst; + int ret; + + /* + * We can use some of the IPv6 functions since GID is a valid + * IPv6 address format + */ + ret = in6_pton(addr, len, dst_ib->sib_addr.sib_raw, '\0', NULL); + if (ret == 0) + return -EINVAL; + + dst_ib->sib_family = AF_IB; + /* + * Use the same TCP server port number as the IB service ID + * on the IB port space range + */ + dst_ib->sib_sid = cpu_to_be64(RDMA_IB_IP_PS_IB | port); + dst_ib->sib_sid_mask = cpu_to_be64(0xffffffffffffffffULL); + dst_ib->sib_pkey = cpu_to_be16(0xffff); + + return 0; +} + +/** + * rtrs_str_to_sockaddr() - Convert rtrs address string to sockaddr + * @addr: String representation of an addr (IPv4, IPv6 or IB GID): + * - "ip:192.168.1.1" + * - "ip:fe80::200:5aee:feaa:20a2" + * - "gid:fe80::200:5aee:feaa:20a2" + * @len: String address length + * @port: Destination port + * @dst: Destination sockaddr structure + * + * Returns 0 if conversion successful. Non-zero on error. + */ +static int rtrs_str_to_sockaddr(const char *addr, size_t len, + u16 port, struct sockaddr_storage *dst) +{ + if (strncmp(addr, "gid:", 4) == 0) { + return rtrs_str_gid_to_sockaddr(addr + 4, len - 4, port, dst); + } else if (strncmp(addr, "ip:", 3) == 0) { + char port_str[8]; + char *cpy; + int err; + + snprintf(port_str, sizeof(port_str), "%u", port); + cpy = kstrndup(addr + 3, len - 3, GFP_KERNEL); + err = cpy ? inet_pton_with_scope(&init_net, AF_UNSPEC, + cpy, port_str, dst) : -ENOMEM; + kfree(cpy); + + return err; + } + return -EPROTONOSUPPORT; +} + +/** + * sockaddr_to_str() - convert sockaddr to a string. + * @addr: the sockadddr structure to be converted. + * @buf: string containing socket addr. + * @len: string length. + * + * The return value is the number of characters written into buf not + * including the trailing '\0'. If len is == 0 the function returns 0.. + */ +int sockaddr_to_str(const struct sockaddr *addr, char *buf, size_t len) +{ + + switch (addr->sa_family) { + case AF_IB: + return scnprintf(buf, len, "gid:%pI6", + &((struct sockaddr_ib *)addr)->sib_addr.sib_raw); + case AF_INET: + return scnprintf(buf, len, "ip:%pI4", + &((struct sockaddr_in *)addr)->sin_addr); + case AF_INET6: + return scnprintf(buf, len, "ip:%pI6c", + &((struct sockaddr_in6 *)addr)->sin6_addr); + } + return scnprintf(buf, len, "<invalid address family>"); +} +EXPORT_SYMBOL(sockaddr_to_str); + +/** + * rtrs_addr_to_sockaddr() - convert path string "src,dst" or "src@dst" + * to sockaddreses + * @str: string containing source and destination addr of a path + * separated by ',' or '@' I.e. "ip:1.1.1.1,ip:1.1.1.2" or + * "ip:1.1.1.1@ip:1.1.1.2". If str contains only one address it's + * considered to be destination. + * @len: string length + * @port: Destination port number. + * @addr: will be set to the source/destination address or to NULL + * if str doesn't contain any source address. + * + * Returns zero if conversion successful. Non-zero otherwise. + */ +int rtrs_addr_to_sockaddr(const char *str, size_t len, u16 port, + struct rtrs_addr *addr) +{ + const char *d; + + d = strchr(str, ','); + if (!d) + d = strchr(str, '@'); + if (d) { + if (rtrs_str_to_sockaddr(str, d - str, 0, addr->src)) + return -EINVAL; + d += 1; + len -= d - str; + str = d; + + } else { + addr->src = NULL; + } + return rtrs_str_to_sockaddr(str, len, port, addr->dst); +} +EXPORT_SYMBOL(rtrs_addr_to_sockaddr); + +void rtrs_rdma_dev_pd_init(enum ib_pd_flags pd_flags, + struct rtrs_rdma_dev_pd *pool) +{ + WARN_ON(pool->ops && (!pool->ops->alloc ^ !pool->ops->free)); + INIT_LIST_HEAD(&pool->list); + mutex_init(&pool->mutex); + pool->pd_flags = pd_flags; +} +EXPORT_SYMBOL(rtrs_rdma_dev_pd_init); + +void rtrs_rdma_dev_pd_deinit(struct rtrs_rdma_dev_pd *pool) +{ + mutex_destroy(&pool->mutex); + WARN_ON(!list_empty(&pool->list)); +} +EXPORT_SYMBOL(rtrs_rdma_dev_pd_deinit); + +static void dev_free(struct kref *ref) +{ + struct rtrs_rdma_dev_pd *pool; + struct rtrs_ib_dev *dev; + + dev = container_of(ref, typeof(*dev), ref); + pool = dev->pool; + + mutex_lock(&pool->mutex); + list_del(&dev->entry); + mutex_unlock(&pool->mutex); + + if (pool->ops && pool->ops->deinit) + pool->ops->deinit(dev); + + ib_dealloc_pd(dev->ib_pd); + + if (pool->ops && pool->ops->free) + pool->ops->free(dev); + else + kfree(dev); +} + +int rtrs_ib_dev_put(struct rtrs_ib_dev *dev) +{ + return kref_put(&dev->ref, dev_free); +} +EXPORT_SYMBOL(rtrs_ib_dev_put); + +static int rtrs_ib_dev_get(struct rtrs_ib_dev *dev) +{ + return kref_get_unless_zero(&dev->ref); +} + +struct rtrs_ib_dev * +rtrs_ib_dev_find_or_add(struct ib_device *ib_dev, + struct rtrs_rdma_dev_pd *pool) +{ + struct rtrs_ib_dev *dev; + + mutex_lock(&pool->mutex); + list_for_each_entry(dev, &pool->list, entry) { + if (dev->ib_dev->node_guid == ib_dev->node_guid && + rtrs_ib_dev_get(dev)) + goto out_unlock; + } + mutex_unlock(&pool->mutex); + if (pool->ops && pool->ops->alloc) + dev = pool->ops->alloc(); + else + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (IS_ERR_OR_NULL(dev)) + goto out_err; + + kref_init(&dev->ref); + dev->pool = pool; + dev->ib_dev = ib_dev; + dev->ib_pd = ib_alloc_pd(ib_dev, pool->pd_flags); + if (IS_ERR(dev->ib_pd)) + goto out_free_dev; + + if (pool->ops && pool->ops->init && pool->ops->init(dev)) + goto out_free_pd; + + mutex_lock(&pool->mutex); + list_add(&dev->entry, &pool->list); +out_unlock: + mutex_unlock(&pool->mutex); + return dev; + +out_free_pd: + ib_dealloc_pd(dev->ib_pd); +out_free_dev: + if (pool->ops && pool->ops->free) + pool->ops->free(dev); + else + kfree(dev); +out_err: + return NULL; +} +EXPORT_SYMBOL(rtrs_ib_dev_find_or_add); diff --git a/drivers/infiniband/ulp/rtrs/rtrs.h b/drivers/infiniband/ulp/rtrs/rtrs.h new file mode 100644 index 000000000000..9af750f4d783 --- /dev/null +++ b/drivers/infiniband/ulp/rtrs/rtrs.h @@ -0,0 +1,196 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * RDMA Transport Layer + * + * Copyright (c) 2014 - 2018 ProfitBricks GmbH. All rights reserved. + * Copyright (c) 2018 - 2019 1&1 IONOS Cloud GmbH. All rights reserved. + * Copyright (c) 2019 - 2020 1&1 IONOS SE. All rights reserved. + */ +#ifndef RTRS_H +#define RTRS_H + +#include <linux/socket.h> +#include <linux/scatterlist.h> + +struct rtrs_permit; +struct rtrs_clt; +struct rtrs_srv_ctx; +struct rtrs_srv; +struct rtrs_srv_op; + +/* + * RDMA transport (RTRS) client API + */ + +/** + * enum rtrs_clt_link_ev - Events about connectivity state of a client + * @RTRS_CLT_LINK_EV_RECONNECTED Client was reconnected. + * @RTRS_CLT_LINK_EV_DISCONNECTED Client was disconnected. + */ +enum rtrs_clt_link_ev { + RTRS_CLT_LINK_EV_RECONNECTED, + RTRS_CLT_LINK_EV_DISCONNECTED, +}; + +/** + * Source and destination address of a path to be established + */ +struct rtrs_addr { + struct sockaddr_storage *src; + struct sockaddr_storage *dst; +}; + +/** + * rtrs_clt_ops - it holds the link event callback and private pointer. + * @priv: User supplied private data. + * @link_ev: Event notification callback function for connection state changes + * @priv: User supplied data that was passed to rtrs_clt_open() + * @ev: Occurred event + */ +struct rtrs_clt_ops { + void *priv; + void (*link_ev)(void *priv, enum rtrs_clt_link_ev ev); +}; + +struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops, + const char *sessname, + const struct rtrs_addr *paths, + size_t path_cnt, u16 port, + size_t pdu_sz, u8 reconnect_delay_sec, + u16 max_segments, + size_t max_segment_size, + s16 max_reconnect_attempts); + +void rtrs_clt_close(struct rtrs_clt *sess); + +/** + * rtrs_permit_to_pdu() - converts rtrs_permit to opaque pdu pointer + * @permit: RTRS permit pointer, it associates the memory allocation for future + * RDMA operation. + */ +void *rtrs_permit_to_pdu(struct rtrs_permit *permit); + +enum { + RTRS_PERMIT_NOWAIT = 0, + RTRS_PERMIT_WAIT = 1, +}; + +/** + * enum rtrs_clt_con_type() type of ib connection to use with a given + * rtrs_permit + * @ADMIN_CON - use connection reserved for "service" messages + * @IO_CON - use a connection reserved for IO + */ +enum rtrs_clt_con_type { + RTRS_ADMIN_CON, + RTRS_IO_CON +}; + +struct rtrs_permit *rtrs_clt_get_permit(struct rtrs_clt *sess, + enum rtrs_clt_con_type con_type, + int wait); + +void rtrs_clt_put_permit(struct rtrs_clt *sess, struct rtrs_permit *permit); + +/** + * rtrs_clt_req_ops - it holds the request confirmation callback + * and a private pointer. + * @priv: User supplied private data. + * @conf_fn: callback function to be called as confirmation + * @priv: User provided data, passed back with corresponding + * @(conf) confirmation. + * @errno: error number. + */ +struct rtrs_clt_req_ops { + void *priv; + void (*conf_fn)(void *priv, int errno); +}; + +int rtrs_clt_request(int dir, struct rtrs_clt_req_ops *ops, + struct rtrs_clt *sess, struct rtrs_permit *permit, + const struct kvec *vec, size_t nr, size_t len, + struct scatterlist *sg, unsigned int sg_cnt); + +/** + * rtrs_attrs - RTRS session attributes + */ +struct rtrs_attrs { + u32 queue_depth; + u32 max_io_size; + u8 sessname[NAME_MAX]; + struct kobject *sess_kobj; +}; + +int rtrs_clt_query(struct rtrs_clt *sess, struct rtrs_attrs *attr); + +/* + * Here goes RTRS server API + */ + +/** + * enum rtrs_srv_link_ev - Server link events + * @RTRS_SRV_LINK_EV_CONNECTED: Connection from client established + * @RTRS_SRV_LINK_EV_DISCONNECTED: Connection was disconnected, all + * connection RTRS resources were freed. + */ +enum rtrs_srv_link_ev { + RTRS_SRV_LINK_EV_CONNECTED, + RTRS_SRV_LINK_EV_DISCONNECTED, +}; + +struct rtrs_srv_ops { + /** + * rdma_ev(): Event notification for RDMA operations + * If the callback returns a value != 0, an error + * message for the data transfer will be sent to + * the client. + + * @sess: Session + * @priv: Private data set by rtrs_srv_set_sess_priv() + * @id: internal RTRS operation id + * @dir: READ/WRITE + * @data: Pointer to (bidirectional) rdma memory area: + * - in case of %RTRS_SRV_RDMA_EV_RECV contains + * data sent by the client + * - in case of %RTRS_SRV_RDMA_EV_WRITE_REQ points + * to the memory area where the response is to be + * written to + * @datalen: Size of the memory area in @data + * @usr: The extra user message sent by the client (%vec) + * @usrlen: Size of the user message + */ + int (*rdma_ev)(struct rtrs_srv *sess, void *priv, + struct rtrs_srv_op *id, int dir, + void *data, size_t datalen, const void *usr, + size_t usrlen); + /** + * link_ev(): Events about connectivity state changes + * If the callback returns != 0 and the event + * %RTRS_SRV_LINK_EV_CONNECTED the corresponding + * session will be destroyed. + * @sess: Session + * @ev: event + * @priv: Private data from user if previously set with + * rtrs_srv_set_sess_priv() + */ + int (*link_ev)(struct rtrs_srv *sess, enum rtrs_srv_link_ev ev, + void *priv); +}; + +struct rtrs_srv_ctx *rtrs_srv_open(struct rtrs_srv_ops *ops, u16 port); + +void rtrs_srv_close(struct rtrs_srv_ctx *ctx); + +bool rtrs_srv_resp_rdma(struct rtrs_srv_op *id, int errno); + +void rtrs_srv_set_sess_priv(struct rtrs_srv *sess, void *priv); + +int rtrs_srv_get_sess_name(struct rtrs_srv *sess, char *sessname, size_t len); + +int rtrs_srv_get_queue_depth(struct rtrs_srv *sess); + +int rtrs_addr_to_sockaddr(const char *str, size_t len, u16 port, + struct rtrs_addr *addr); + +int sockaddr_to_str(const struct sockaddr *addr, char *buf, size_t len); +#endif diff --git a/drivers/infiniband/ulp/srp/Kconfig b/drivers/infiniband/ulp/srp/Kconfig index 6f5e7b3a3864..67cd63d1399c 100644 --- a/drivers/infiniband/ulp/srp/Kconfig +++ b/drivers/infiniband/ulp/srp/Kconfig @@ -3,7 +3,7 @@ config INFINIBAND_SRP tristate "InfiniBand SCSI RDMA Protocol" depends on SCSI && INFINIBAND_ADDR_TRANS select SCSI_SRP_ATTRS - ---help--- + help Support for the SCSI RDMA Protocol over InfiniBand. This allows you to access storage devices that speak SRP over InfiniBand. diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index cd1181c39ed2..d8fcd21ab472 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -71,7 +71,6 @@ static unsigned int srp_sg_tablesize; static unsigned int cmd_sg_entries; static unsigned int indirect_sg_entries; static bool allow_ext_sg; -static bool prefer_fr = true; static bool register_always = true; static bool never_register; static int topspin_workarounds = 1; @@ -95,10 +94,6 @@ module_param(topspin_workarounds, int, 0444); MODULE_PARM_DESC(topspin_workarounds, "Enable workarounds for Topspin/Cisco SRP target bugs if != 0"); -module_param(prefer_fr, bool, 0444); -MODULE_PARM_DESC(prefer_fr, -"Whether to use fast registration if both FMR and fast registration are supported"); - module_param(register_always, bool, 0444); MODULE_PARM_DESC(register_always, "Use memory registration even for contiguous memory regions"); @@ -146,7 +141,7 @@ module_param(ch_count, uint, 0444); MODULE_PARM_DESC(ch_count, "Number of RDMA channels to use for communication with an SRP target. Using more than one channel improves performance if the HCA supports multiple completion vectors. The default value is the minimum of four times the number of online CPU sockets and the number of completion vectors supported by the HCA."); -static void srp_add_one(struct ib_device *device); +static int srp_add_one(struct ib_device *device); static void srp_remove_one(struct ib_device *device, void *client_data); static void srp_rename_dev(struct ib_device *device, void *client_data); static void srp_recv_done(struct ib_cq *cq, struct ib_wc *wc); @@ -388,24 +383,6 @@ static int srp_new_cm_id(struct srp_rdma_ch *ch) srp_new_ib_cm_id(ch); } -static struct ib_fmr_pool *srp_alloc_fmr_pool(struct srp_target_port *target) -{ - struct srp_device *dev = target->srp_host->srp_dev; - struct ib_fmr_pool_param fmr_param; - - memset(&fmr_param, 0, sizeof(fmr_param)); - fmr_param.pool_size = target->mr_pool_size; - fmr_param.dirty_watermark = fmr_param.pool_size / 4; - fmr_param.cache = 1; - fmr_param.max_pages_per_fmr = dev->max_pages_per_mr; - fmr_param.page_shift = ilog2(dev->mr_page_size); - fmr_param.access = (IB_ACCESS_LOCAL_WRITE | - IB_ACCESS_REMOTE_WRITE | - IB_ACCESS_REMOTE_READ); - - return ib_create_fmr_pool(dev->pd, &fmr_param); -} - /** * srp_destroy_fr_pool() - free the resources owned by a pool * @pool: Fast registration pool to be destroyed. @@ -556,7 +533,6 @@ static int srp_create_ch_ib(struct srp_rdma_ch *ch) struct ib_qp_init_attr *init_attr; struct ib_cq *recv_cq, *send_cq; struct ib_qp *qp; - struct ib_fmr_pool *fmr_pool = NULL; struct srp_fr_pool *fr_pool = NULL; const int m = 1 + dev->use_fast_reg * target->mr_per_cmd * 2; int ret; @@ -619,14 +595,6 @@ static int srp_create_ch_ib(struct srp_rdma_ch *ch) "FR pool allocation failed (%d)\n", ret); goto err_qp; } - } else if (dev->use_fmr) { - fmr_pool = srp_alloc_fmr_pool(target); - if (IS_ERR(fmr_pool)) { - ret = PTR_ERR(fmr_pool); - shost_printk(KERN_WARNING, target->scsi_host, PFX - "FMR pool allocation failed (%d)\n", ret); - goto err_qp; - } } if (ch->qp) @@ -644,10 +612,6 @@ static int srp_create_ch_ib(struct srp_rdma_ch *ch) if (ch->fr_pool) srp_destroy_fr_pool(ch->fr_pool); ch->fr_pool = fr_pool; - } else if (dev->use_fmr) { - if (ch->fmr_pool) - ib_destroy_fmr_pool(ch->fmr_pool); - ch->fmr_pool = fmr_pool; } kfree(init_attr); @@ -702,9 +666,6 @@ static void srp_free_ch_ib(struct srp_target_port *target, if (dev->use_fast_reg) { if (ch->fr_pool) srp_destroy_fr_pool(ch->fr_pool); - } else if (dev->use_fmr) { - if (ch->fmr_pool) - ib_destroy_fmr_pool(ch->fmr_pool); } srp_destroy_qp(ch); @@ -1017,12 +978,8 @@ static void srp_free_req_data(struct srp_target_port *target, for (i = 0; i < target->req_ring_size; ++i) { req = &ch->req_ring[i]; - if (dev->use_fast_reg) { + if (dev->use_fast_reg) kfree(req->fr_list); - } else { - kfree(req->fmr_list); - kfree(req->map_page); - } if (req->indirect_dma_addr) { ib_dma_unmap_single(ibdev, req->indirect_dma_addr, target->indirect_size, @@ -1056,16 +1013,8 @@ static int srp_alloc_req_data(struct srp_rdma_ch *ch) GFP_KERNEL); if (!mr_list) goto out; - if (srp_dev->use_fast_reg) { + if (srp_dev->use_fast_reg) req->fr_list = mr_list; - } else { - req->fmr_list = mr_list; - req->map_page = kmalloc_array(srp_dev->max_pages_per_mr, - sizeof(void *), - GFP_KERNEL); - if (!req->map_page) - goto out; - } req->indirect_desc = kmalloc(target->indirect_size, GFP_KERNEL); if (!req->indirect_desc) goto out; @@ -1272,11 +1221,6 @@ static void srp_unmap_data(struct scsi_cmnd *scmnd, if (req->nmdesc) srp_fr_pool_put(ch->fr_pool, req->fr_list, req->nmdesc); - } else if (dev->use_fmr) { - struct ib_pool_fmr **pfmr; - - for (i = req->nmdesc, pfmr = req->fmr_list; i > 0; i--, pfmr++) - ib_fmr_pool_unmap(*pfmr); } ib_dma_unmap_sg(ibdev, scsi_sglist(scmnd), scsi_sg_count(scmnd), @@ -1472,50 +1416,6 @@ static void srp_map_desc(struct srp_map_state *state, dma_addr_t dma_addr, state->ndesc++; } -static int srp_map_finish_fmr(struct srp_map_state *state, - struct srp_rdma_ch *ch) -{ - struct srp_target_port *target = ch->target; - struct srp_device *dev = target->srp_host->srp_dev; - struct ib_pool_fmr *fmr; - u64 io_addr = 0; - - if (state->fmr.next >= state->fmr.end) { - shost_printk(KERN_ERR, ch->target->scsi_host, - PFX "Out of MRs (mr_per_cmd = %d)\n", - ch->target->mr_per_cmd); - return -ENOMEM; - } - - WARN_ON_ONCE(!dev->use_fmr); - - if (state->npages == 0) - return 0; - - if (state->npages == 1 && target->global_rkey) { - srp_map_desc(state, state->base_dma_addr, state->dma_len, - target->global_rkey); - goto reset_state; - } - - fmr = ib_fmr_pool_map_phys(ch->fmr_pool, state->pages, - state->npages, io_addr); - if (IS_ERR(fmr)) - return PTR_ERR(fmr); - - *state->fmr.next++ = fmr; - state->nmdesc++; - - srp_map_desc(state, state->base_dma_addr & ~dev->mr_page_mask, - state->dma_len, fmr->fmr->rkey); - -reset_state: - state->npages = 0; - state->dma_len = 0; - - return 0; -} - static void srp_reg_mr_err_done(struct ib_cq *cq, struct ib_wc *wc) { srp_handle_qp_err(cq, wc, "FAST REG"); @@ -1606,74 +1506,6 @@ static int srp_map_finish_fr(struct srp_map_state *state, return n; } -static int srp_map_sg_entry(struct srp_map_state *state, - struct srp_rdma_ch *ch, - struct scatterlist *sg) -{ - struct srp_target_port *target = ch->target; - struct srp_device *dev = target->srp_host->srp_dev; - dma_addr_t dma_addr = sg_dma_address(sg); - unsigned int dma_len = sg_dma_len(sg); - unsigned int len = 0; - int ret; - - WARN_ON_ONCE(!dma_len); - - while (dma_len) { - unsigned offset = dma_addr & ~dev->mr_page_mask; - - if (state->npages == dev->max_pages_per_mr || - (state->npages > 0 && offset != 0)) { - ret = srp_map_finish_fmr(state, ch); - if (ret) - return ret; - } - - len = min_t(unsigned int, dma_len, dev->mr_page_size - offset); - - if (!state->npages) - state->base_dma_addr = dma_addr; - state->pages[state->npages++] = dma_addr & dev->mr_page_mask; - state->dma_len += len; - dma_addr += len; - dma_len -= len; - } - - /* - * If the end of the MR is not on a page boundary then we need to - * close it out and start a new one -- we can only merge at page - * boundaries. - */ - ret = 0; - if ((dma_addr & ~dev->mr_page_mask) != 0) - ret = srp_map_finish_fmr(state, ch); - return ret; -} - -static int srp_map_sg_fmr(struct srp_map_state *state, struct srp_rdma_ch *ch, - struct srp_request *req, struct scatterlist *scat, - int count) -{ - struct scatterlist *sg; - int i, ret; - - state->pages = req->map_page; - state->fmr.next = req->fmr_list; - state->fmr.end = req->fmr_list + ch->target->mr_per_cmd; - - for_each_sg(scat, sg, count, i) { - ret = srp_map_sg_entry(state, ch, sg); - if (ret) - return ret; - } - - ret = srp_map_finish_fmr(state, ch); - if (ret) - return ret; - - return 0; -} - static int srp_map_sg_fr(struct srp_map_state *state, struct srp_rdma_ch *ch, struct srp_request *req, struct scatterlist *scat, int count) @@ -1733,7 +1565,6 @@ static int srp_map_idb(struct srp_rdma_ch *ch, struct srp_request *req, struct srp_device *dev = target->srp_host->srp_dev; struct srp_map_state state; struct srp_direct_buf idb_desc; - u64 idb_pages[1]; struct scatterlist idb_sg[1]; int ret; @@ -1756,14 +1587,6 @@ static int srp_map_idb(struct srp_rdma_ch *ch, struct srp_request *req, if (ret < 0) return ret; WARN_ON_ONCE(ret < 1); - } else if (dev->use_fmr) { - state.pages = idb_pages; - state.pages[0] = (req->indirect_dma_addr & - dev->mr_page_mask); - state.npages = 1; - ret = srp_map_finish_fmr(&state, ch); - if (ret < 0) - return ret; } else { return -EINVAL; } @@ -1787,9 +1610,6 @@ static void srp_check_mapping(struct srp_map_state *state, if (dev->use_fast_reg) for (i = 0, pfr = req->fr_list; i < state->nmdesc; i++, pfr++) mr_len += (*pfr)->mr->length; - else if (dev->use_fmr) - for (i = 0; i < state->nmdesc; i++) - mr_len += be32_to_cpu(req->indirect_desc[i].len); if (desc_len != scsi_bufflen(req->scmnd) || mr_len > scsi_bufflen(req->scmnd)) pr_err("Inconsistent: scsi len %d <> desc len %lld <> mr len %lld; ndesc %d; nmdesc = %d\n", @@ -1904,8 +1724,6 @@ static int srp_map_data(struct scsi_cmnd *scmnd, struct srp_rdma_ch *ch, state.desc = req->indirect_desc; if (dev->use_fast_reg) ret = srp_map_sg_fr(&state, ch, req, scat, count); - else if (dev->use_fmr) - ret = srp_map_sg_fmr(&state, ch, req, scat, count); else ret = srp_map_sg_dma(&state, ch, req, scat, count); req->nmdesc = state.nmdesc; @@ -3424,6 +3242,7 @@ enum { SRP_OPT_IP_DEST = 1 << 16, SRP_OPT_TARGET_CAN_QUEUE= 1 << 17, SRP_OPT_MAX_IT_IU_SIZE = 1 << 18, + SRP_OPT_CH_COUNT = 1 << 19, }; static unsigned int srp_opt_mandatory[] = { @@ -3457,6 +3276,7 @@ static const match_table_t srp_opt_tokens = { { SRP_OPT_IP_SRC, "src=%s" }, { SRP_OPT_IP_DEST, "dest=%s" }, { SRP_OPT_MAX_IT_IU_SIZE, "max_it_iu_size=%d" }, + { SRP_OPT_CH_COUNT, "ch_count=%u", }, { SRP_OPT_ERR, NULL } }; @@ -3758,6 +3578,14 @@ static int srp_parse_options(struct net *net, const char *buf, target->max_it_iu_size = token; break; + case SRP_OPT_CH_COUNT: + if (match_int(args, &token) || token < 1) { + pr_warn("bad channel count %s\n", p); + goto out; + } + target->ch_count = token; + break; + default: pr_warn("unknown parameter or missing value '%s' in target creation request\n", p); @@ -3864,13 +3692,13 @@ static ssize_t srp_create_target(struct device *dev, goto out; } - if (!srp_dev->has_fmr && !srp_dev->has_fr && !target->allow_ext_sg && + if (!srp_dev->has_fr && !target->allow_ext_sg && target->cmd_sg_cnt < target->sg_tablesize) { pr_warn("No MR pool and no external indirect descriptors, limiting sg_tablesize to cmd_sg_cnt\n"); target->sg_tablesize = target->cmd_sg_cnt; } - if (srp_dev->use_fast_reg || srp_dev->use_fmr) { + if (srp_dev->use_fast_reg) { bool gaps_reg = (ibdev->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG); @@ -3878,12 +3706,12 @@ static ssize_t srp_create_target(struct device *dev, (ilog2(srp_dev->mr_page_size) - 9); if (!gaps_reg) { /* - * FR and FMR can only map one HCA page per entry. If - * the start address is not aligned on a HCA page - * boundary two entries will be used for the head and - * the tail although these two entries combined - * contain at most one HCA page of data. Hence the "+ - * 1" in the calculation below. + * FR can only map one HCA page per entry. If the start + * address is not aligned on a HCA page boundary two + * entries will be used for the head and the tail + * although these two entries combined contain at most + * one HCA page of data. Hence the "+ 1" in the + * calculation below. * * The indirect data buffer descriptor is contiguous * so the memory for that buffer will only be @@ -3921,11 +3749,13 @@ static ssize_t srp_create_target(struct device *dev, goto out; ret = -ENOMEM; - target->ch_count = max_t(unsigned, num_online_nodes(), - min(ch_count ? : - min(4 * num_online_nodes(), - ibdev->num_comp_vectors), - num_online_cpus())); + if (target->ch_count == 0) + target->ch_count = + max_t(unsigned int, num_online_nodes(), + min(ch_count ?: + min(4 * num_online_nodes(), + ibdev->num_comp_vectors), + num_online_cpus())); target->ch = kcalloc(target->ch_count, sizeof(*target->ch), GFP_KERNEL); if (!target->ch) @@ -4132,7 +3962,7 @@ static void srp_rename_dev(struct ib_device *device, void *client_data) } } -static void srp_add_one(struct ib_device *device) +static int srp_add_one(struct ib_device *device) { struct srp_device *srp_dev; struct ib_device_attr *attr = &device->attrs; @@ -4144,7 +3974,7 @@ static void srp_add_one(struct ib_device *device) srp_dev = kzalloc(sizeof(*srp_dev), GFP_KERNEL); if (!srp_dev) - return; + return -ENOMEM; /* * Use the smallest page size supported by the HCA, down to a @@ -4162,23 +3992,15 @@ static void srp_add_one(struct ib_device *device) srp_dev->max_pages_per_mr = min_t(u64, SRP_MAX_PAGES_PER_MR, max_pages_per_mr); - srp_dev->has_fmr = (device->ops.alloc_fmr && - device->ops.dealloc_fmr && - device->ops.map_phys_fmr && - device->ops.unmap_fmr); srp_dev->has_fr = (attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS); - if (!never_register && !srp_dev->has_fmr && !srp_dev->has_fr) { - dev_warn(&device->dev, "neither FMR nor FR is supported\n"); - } else if (!never_register && - attr->max_mr_size >= 2 * srp_dev->mr_page_size) { - srp_dev->use_fast_reg = (srp_dev->has_fr && - (!srp_dev->has_fmr || prefer_fr)); - srp_dev->use_fmr = !srp_dev->use_fast_reg && srp_dev->has_fmr; - } + if (!never_register && !srp_dev->has_fr) + dev_warn(&device->dev, "FR is not supported\n"); + else if (!never_register && + attr->max_mr_size >= 2 * srp_dev->mr_page_size) + srp_dev->use_fast_reg = srp_dev->has_fr; - if (never_register || !register_always || - (!srp_dev->has_fmr && !srp_dev->has_fr)) + if (never_register || !register_always || !srp_dev->has_fr) flags |= IB_PD_UNSAFE_GLOBAL_RKEY; if (srp_dev->use_fast_reg) { @@ -4197,8 +4019,12 @@ static void srp_add_one(struct ib_device *device) srp_dev->dev = device; srp_dev->pd = ib_alloc_pd(device, flags); - if (IS_ERR(srp_dev->pd)) - goto free_dev; + if (IS_ERR(srp_dev->pd)) { + int ret = PTR_ERR(srp_dev->pd); + + kfree(srp_dev); + return ret; + } if (flags & IB_PD_UNSAFE_GLOBAL_RKEY) { srp_dev->global_rkey = srp_dev->pd->unsafe_global_rkey; @@ -4212,10 +4038,7 @@ static void srp_add_one(struct ib_device *device) } ib_set_client_data(device, &srp_client, srp_dev); - return; - -free_dev: - kfree(srp_dev); + return 0; } static void srp_remove_one(struct ib_device *device, void *client_data) @@ -4225,8 +4048,6 @@ static void srp_remove_one(struct ib_device *device, void *client_data) struct srp_target_port *target; srp_dev = client_data; - if (!srp_dev) - return; list_for_each_entry_safe(host, tmp_host, &srp_dev->dev_list, list) { device_unregister(&host->dev); diff --git a/drivers/infiniband/ulp/srp/ib_srp.h b/drivers/infiniband/ulp/srp/ib_srp.h index 6fabcc2faf1f..6818cac0a3b7 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.h +++ b/drivers/infiniband/ulp/srp/ib_srp.h @@ -44,7 +44,6 @@ #include <rdma/ib_verbs.h> #include <rdma/ib_sa.h> #include <rdma/ib_cm.h> -#include <rdma/ib_fmr_pool.h> #include <rdma/rdma_cm.h> enum { @@ -95,8 +94,7 @@ enum srp_iu_type { /* * @mr_page_mask: HCA memory registration page mask. * @mr_page_size: HCA memory registration page size. - * @mr_max_size: Maximum size in bytes of a single FMR / FR registration - * request. + * @mr_max_size: Maximum size in bytes of a single FR registration request. */ struct srp_device { struct list_head dev_list; @@ -107,9 +105,7 @@ struct srp_device { int mr_page_size; int mr_max_size; int max_pages_per_mr; - bool has_fmr; bool has_fr; - bool use_fmr; bool use_fast_reg; }; @@ -127,11 +123,7 @@ struct srp_host { struct srp_request { struct scsi_cmnd *scmnd; struct srp_iu *cmd; - union { - struct ib_pool_fmr **fmr_list; - struct srp_fr_desc **fr_list; - }; - u64 *map_page; + struct srp_fr_desc **fr_list; struct srp_direct_buf *indirect_desc; dma_addr_t indirect_dma_addr; short nmdesc; @@ -155,10 +147,7 @@ struct srp_rdma_ch { struct ib_cq *send_cq; struct ib_cq *recv_cq; struct ib_qp *qp; - union { - struct ib_fmr_pool *fmr_pool; - struct srp_fr_pool *fr_pool; - }; + struct srp_fr_pool *fr_pool; uint32_t max_it_iu_len; uint32_t max_ti_iu_len; u8 max_imm_sge; @@ -319,20 +308,16 @@ struct srp_fr_pool { * @pages: Array with DMA addresses of pages being considered for * memory registration. * @base_dma_addr: DMA address of the first page that has not yet been mapped. - * @dma_len: Number of bytes that will be registered with the next - * FMR or FR memory registration call. + * @dma_len: Number of bytes that will be registered with the next FR + * memory registration call. * @total_len: Total number of bytes in the sg-list being mapped. * @npages: Number of page addresses in the pages[] array. - * @nmdesc: Number of FMR or FR memory descriptors used for mapping. + * @nmdesc: Number of FR memory descriptors used for mapping. * @ndesc: Number of SRP buffer descriptors that have been filled in. */ struct srp_map_state { union { struct { - struct ib_pool_fmr **next; - struct ib_pool_fmr **end; - } fmr; - struct { struct srp_fr_desc **next; struct srp_fr_desc **end; } fr; diff --git a/drivers/infiniband/ulp/srpt/Kconfig b/drivers/infiniband/ulp/srpt/Kconfig index ce7567cea9f6..4b5d9b792cfa 100644 --- a/drivers/infiniband/ulp/srpt/Kconfig +++ b/drivers/infiniband/ulp/srpt/Kconfig @@ -2,7 +2,7 @@ config INFINIBAND_SRPT tristate "InfiniBand SCSI RDMA Protocol target support" depends on INFINIBAND && INFINIBAND_ADDR_TRANS && TARGET_CORE - ---help--- + help Support for the SCSI RDMA Protocol (SRP) Target driver. The SRP protocol is a protocol that allows an initiator to access diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index 98552749d71c..0065eb17ae36 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -81,7 +81,7 @@ MODULE_PARM_DESC(srpt_srq_size, static int srpt_get_u64_x(char *buffer, const struct kernel_param *kp) { - return sprintf(buffer, "0x%016llx", *(u64 *)kp->arg); + return sprintf(buffer, "0x%016llx\n", *(u64 *)kp->arg); } module_param_call(srpt_service_guid, NULL, srpt_get_u64_x, &srpt_service_guid, 0444); @@ -135,14 +135,11 @@ static bool srpt_set_ch_state(struct srpt_rdma_ch *ch, enum rdma_ch_state new) static void srpt_event_handler(struct ib_event_handler *handler, struct ib_event *event) { - struct srpt_device *sdev; + struct srpt_device *sdev = + container_of(handler, struct srpt_device, event_handler); struct srpt_port *sport; u8 port_num; - sdev = ib_get_client_data(event->device, &srpt_client); - if (!sdev || sdev->device != event->device) - return; - pr_debug("ASYNC event= %d on device= %s\n", event->event, dev_name(&sdev->device->dev)); @@ -217,8 +214,9 @@ static const char *get_ch_state_name(enum rdma_ch_state s) */ static void srpt_qp_event(struct ib_event *event, struct srpt_rdma_ch *ch) { - pr_debug("QP event %d on ch=%p sess_name=%s state=%d\n", - event->event, ch, ch->sess_name, ch->state); + pr_debug("QP event %d on ch=%p sess_name=%s-%d state=%s\n", + event->event, ch, ch->sess_name, ch->qp->qp_num, + get_ch_state_name(ch->state)); switch (event->event) { case IB_EVENT_COMM_EST: @@ -610,6 +608,11 @@ static int srpt_refresh_port(struct srpt_port *sport) dev_name(&sport->sdev->device->dev), sport->port, PTR_ERR(sport->mad_agent)); sport->mad_agent = NULL; + memset(&port_modify, 0, sizeof(port_modify)); + port_modify.clr_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP; + ib_modify_port(sport->sdev->device, sport->port, 0, + &port_modify); + } } @@ -633,9 +636,8 @@ static void srpt_unregister_mad_agent(struct srpt_device *sdev) for (i = 1; i <= sdev->device->phys_port_cnt; i++) { sport = &sdev->port[i - 1]; WARN_ON(sport->port != i); - if (ib_modify_port(sdev->device, i, 0, &port_modify) < 0) - pr_err("disabling MAD processing failed.\n"); if (sport->mad_agent) { + ib_modify_port(sdev->device, i, 0, &port_modify); ib_unregister_mad_agent(sport->mad_agent); sport->mad_agent = NULL; } @@ -867,7 +869,7 @@ static int srpt_zerolength_write(struct srpt_rdma_ch *ch) static void srpt_zerolength_write_done(struct ib_cq *cq, struct ib_wc *wc) { - struct srpt_rdma_ch *ch = cq->cq_context; + struct srpt_rdma_ch *ch = wc->qp->qp_context; pr_debug("%s-%d wc->status %d\n", ch->sess_name, ch->qp->qp_num, wc->status); @@ -1320,7 +1322,7 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx) */ static void srpt_rdma_read_done(struct ib_cq *cq, struct ib_wc *wc) { - struct srpt_rdma_ch *ch = cq->cq_context; + struct srpt_rdma_ch *ch = wc->qp->qp_context; struct srpt_send_ioctx *ioctx = container_of(wc->wr_cqe, struct srpt_send_ioctx, rdma_cqe); @@ -1681,7 +1683,7 @@ push: static void srpt_recv_done(struct ib_cq *cq, struct ib_wc *wc) { - struct srpt_rdma_ch *ch = cq->cq_context; + struct srpt_rdma_ch *ch = wc->qp->qp_context; struct srpt_recv_ioctx *ioctx = container_of(wc->wr_cqe, struct srpt_recv_ioctx, ioctx.cqe); @@ -1742,7 +1744,7 @@ static void srpt_process_wait_list(struct srpt_rdma_ch *ch) */ static void srpt_send_done(struct ib_cq *cq, struct ib_wc *wc) { - struct srpt_rdma_ch *ch = cq->cq_context; + struct srpt_rdma_ch *ch = wc->qp->qp_context; struct srpt_send_ioctx *ioctx = container_of(wc->wr_cqe, struct srpt_send_ioctx, ioctx.cqe); enum srpt_command_state state; @@ -1789,7 +1791,7 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch) goto out; retry: - ch->cq = ib_alloc_cq_any(sdev->device, ch, ch->rq_size + sq_size, + ch->cq = ib_cq_pool_get(sdev->device, ch->rq_size + sq_size, -1, IB_POLL_WORKQUEUE); if (IS_ERR(ch->cq)) { ret = PTR_ERR(ch->cq); @@ -1797,6 +1799,7 @@ retry: ch->rq_size + sq_size, ret); goto out; } + ch->cq_size = ch->rq_size + sq_size; qp_init->qp_context = (void *)ch; qp_init->event_handler @@ -1814,18 +1817,13 @@ retry: */ qp_init->cap.max_send_wr = min(sq_size / 2, attrs->max_qp_wr); qp_init->cap.max_rdma_ctxs = sq_size / 2; - qp_init->cap.max_send_sge = min(attrs->max_send_sge, - SRPT_MAX_SG_PER_WQE); - qp_init->cap.max_recv_sge = min(attrs->max_recv_sge, - SRPT_MAX_SG_PER_WQE); + qp_init->cap.max_send_sge = attrs->max_send_sge; + qp_init->cap.max_recv_sge = 1; qp_init->port_num = ch->sport->port; - if (sdev->use_srq) { + if (sdev->use_srq) qp_init->srq = sdev->srq; - } else { + else qp_init->cap.max_recv_wr = ch->rq_size; - qp_init->cap.max_recv_sge = min(attrs->max_recv_sge, - SRPT_MAX_SG_PER_WQE); - } if (ch->using_rdma_cm) { ret = rdma_create_qp(ch->rdma_cm.cm_id, sdev->pd, qp_init); @@ -1846,7 +1844,7 @@ retry: if (retry) { pr_debug("failed to create queue pair with sq_size = %d (%d) - retrying\n", sq_size, ret); - ib_free_cq(ch->cq); + ib_cq_pool_put(ch->cq, ch->cq_size); sq_size = max(sq_size / 2, MIN_SRPT_SQ_SIZE); goto retry; } else { @@ -1872,14 +1870,14 @@ out: err_destroy_cq: ch->qp = NULL; - ib_free_cq(ch->cq); + ib_cq_pool_put(ch->cq, ch->cq_size); goto out; } static void srpt_destroy_ch_ib(struct srpt_rdma_ch *ch) { ib_destroy_qp(ch->qp); - ib_free_cq(ch->cq); + ib_cq_pool_put(ch->cq, ch->cq_size); } /** @@ -1984,8 +1982,8 @@ static void __srpt_close_all_ch(struct srpt_port *sport) list_for_each_entry(nexus, &sport->nexus_list, entry) { list_for_each_entry(ch, &nexus->ch_list, list) { if (srpt_disconnect_ch(ch) >= 0) - pr_info("Closing channel %s because target %s_%d has been disabled\n", - ch->sess_name, + pr_info("Closing channel %s-%d because target %s_%d has been disabled\n", + ch->sess_name, ch->qp->qp_num, dev_name(&sport->sdev->device->dev), sport->port); srpt_close_ch(ch); @@ -2159,9 +2157,6 @@ static int srpt_cm_req_recv(struct srpt_device *const sdev, WARN_ON_ONCE(irqs_disabled()); - if (WARN_ON(!sdev || !req)) - return -EINVAL; - it_iu_len = be32_to_cpu(req->req_it_iu_len); pr_info("Received SRP_LOGIN_REQ with i_port_id %pI6, t_port_id %pI6 and it_iu_len %d on port %d (guid=%pI6); pkey %#04x\n", @@ -2496,7 +2491,8 @@ reject: SRP_BUF_FORMAT_INDIRECT); if (rdma_cm_id) - rdma_reject(rdma_cm_id, rej, sizeof(*rej)); + rdma_reject(rdma_cm_id, rej, sizeof(*rej), + IB_CM_REJ_CONSUMER_DEFINED); else ib_send_cm_rej(ib_cm_id, IB_CM_REJ_CONSUMER_DEFINED, NULL, 0, rej, sizeof(*rej)); @@ -3104,7 +3100,7 @@ static int srpt_use_srq(struct srpt_device *sdev, bool use_srq) * srpt_add_one - InfiniBand device addition callback function * @device: Describes a HCA. */ -static void srpt_add_one(struct ib_device *device) +static int srpt_add_one(struct ib_device *device) { struct srpt_device *sdev; struct srpt_port *sport; @@ -3115,14 +3111,16 @@ static void srpt_add_one(struct ib_device *device) sdev = kzalloc(struct_size(sdev, port, device->phys_port_cnt), GFP_KERNEL); if (!sdev) - goto err; + return -ENOMEM; sdev->device = device; mutex_init(&sdev->sdev_mutex); sdev->pd = ib_alloc_pd(device, 0); - if (IS_ERR(sdev->pd)) + if (IS_ERR(sdev->pd)) { + ret = PTR_ERR(sdev->pd); goto free_dev; + } sdev->lkey = sdev->pd->local_dma_lkey; @@ -3138,6 +3136,7 @@ static void srpt_add_one(struct ib_device *device) if (IS_ERR(sdev->cm_id)) { pr_info("ib_create_cm_id() failed: %ld\n", PTR_ERR(sdev->cm_id)); + ret = PTR_ERR(sdev->cm_id); sdev->cm_id = NULL; if (!rdma_cm_id) goto err_ring; @@ -3182,7 +3181,8 @@ static void srpt_add_one(struct ib_device *device) mutex_init(&sport->port_gid_id.mutex); INIT_LIST_HEAD(&sport->port_gid_id.tpg_list); - if (srpt_refresh_port(sport)) { + ret = srpt_refresh_port(sport); + if (ret) { pr_err("MAD registration failed for %s-%d.\n", dev_name(&sdev->device->dev), i); goto err_event; @@ -3193,10 +3193,9 @@ static void srpt_add_one(struct ib_device *device) list_add_tail(&sdev->list, &srpt_dev_list); spin_unlock(&srpt_dev_lock); -out: ib_set_client_data(device, &srpt_client, sdev); pr_debug("added %s.\n", dev_name(&device->dev)); - return; + return 0; err_event: ib_unregister_event_handler(&sdev->event_handler); @@ -3208,10 +3207,8 @@ err_ring: ib_dealloc_pd(sdev->pd); free_dev: kfree(sdev); -err: - sdev = NULL; pr_info("%s(%s) failed.\n", __func__, dev_name(&device->dev)); - goto out; + return ret; } /** @@ -3224,12 +3221,6 @@ static void srpt_remove_one(struct ib_device *device, void *client_data) struct srpt_device *sdev = client_data; int i; - if (!sdev) { - pr_info("%s(%s): nothing to do.\n", __func__, - dev_name(&device->dev)); - return; - } - srpt_unregister_mad_agent(sdev); ib_unregister_event_handler(&sdev->event_handler); diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.h b/drivers/infiniband/ulp/srpt/ib_srpt.h index 2e1a69840857..41435a699b53 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.h +++ b/drivers/infiniband/ulp/srpt/ib_srpt.h @@ -105,11 +105,6 @@ enum { SRP_CMD_ACA = 0x4, SRPT_DEF_SG_TABLESIZE = 128, - /* - * An experimentally determined value that avoids that QP creation - * fails due to "swiotlb buffer is full" on systems using the swiotlb. - */ - SRPT_MAX_SG_PER_WQE = 16, MIN_SRPT_SQ_SIZE = 16, DEF_SRPT_SQ_SIZE = 4096, @@ -305,6 +300,7 @@ struct srpt_rdma_ch { } rdma_cm; }; struct ib_cq *cq; + u32 cq_size; struct ib_cqe zw_cqe; struct rcu_head rcu; struct kref kref; |