diff options
59 files changed, 13520 insertions, 74 deletions
diff --git a/MAINTAINERS b/MAINTAINERS index e1b090f86e0d..4daa6712eac1 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -7444,6 +7444,15 @@ W: http://www.mellanox.com Q: http://patchwork.ozlabs.org/project/netdev/list/ F: drivers/net/ethernet/mellanox/mlxsw/ +SOFT-ROCE DRIVER (rxe) +M: Moni Shoua <monis@mellanox.com> +L: linux-rdma@vger.kernel.org +S: Supported +W: https://github.com/SoftRoCE/rxe-dev/wiki/rxe-dev:-Home +Q: http://patchwork.kernel.org/project/linux-rdma/list/ +F: drivers/infiniband/hw/rxe/ +F: include/uapi/rdma/rdma_user_rxe.h + MEMBARRIER SUPPORT M: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> M: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> diff --git a/drivers/infiniband/Kconfig b/drivers/infiniband/Kconfig index 2137adfbd8c3..e9b7dc037ff8 100644 --- a/drivers/infiniband/Kconfig +++ b/drivers/infiniband/Kconfig @@ -84,6 +84,7 @@ source "drivers/infiniband/ulp/iser/Kconfig" source "drivers/infiniband/ulp/isert/Kconfig" source "drivers/infiniband/sw/rdmavt/Kconfig" +source "drivers/infiniband/sw/rxe/Kconfig" source "drivers/infiniband/hw/hfi1/Kconfig" diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index f0c91ba3178a..0451307bea18 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -68,6 +68,7 @@ MODULE_DESCRIPTION("Generic RDMA CM Agent"); MODULE_LICENSE("Dual BSD/GPL"); #define CMA_CM_RESPONSE_TIMEOUT 20 +#define CMA_QUERY_CLASSPORT_INFO_TIMEOUT 3000 #define CMA_MAX_CM_RETRIES 15 #define CMA_CM_MRA_SETTING (IB_CM_MRA_FLAG_DELAY | 24) #define CMA_IBOE_PACKET_LIFETIME 18 @@ -162,6 +163,14 @@ struct rdma_bind_list { unsigned short port; }; +struct class_port_info_context { + struct ib_class_port_info *class_port_info; + struct ib_device *device; + struct completion done; + struct ib_sa_query *sa_query; + u8 port_num; +}; + static int cma_ps_alloc(struct net *net, enum rdma_port_space ps, struct rdma_bind_list *bind_list, int snum) { @@ -306,6 +315,7 @@ struct cma_multicast { struct sockaddr_storage addr; struct kref mcref; bool igmp_joined; + u8 join_state; }; struct cma_work { @@ -3754,10 +3764,63 @@ static void cma_set_mgid(struct rdma_id_private *id_priv, } } +static void cma_query_sa_classport_info_cb(int status, + struct ib_class_port_info *rec, + void *context) +{ + struct class_port_info_context *cb_ctx = context; + + WARN_ON(!context); + + if (status || !rec) { + pr_debug("RDMA CM: %s port %u failed query ClassPortInfo status: %d\n", + cb_ctx->device->name, cb_ctx->port_num, status); + goto out; + } + + memcpy(cb_ctx->class_port_info, rec, sizeof(struct ib_class_port_info)); + +out: + complete(&cb_ctx->done); +} + +static int cma_query_sa_classport_info(struct ib_device *device, u8 port_num, + struct ib_class_port_info *class_port_info) +{ + struct class_port_info_context *cb_ctx; + int ret; + + cb_ctx = kmalloc(sizeof(*cb_ctx), GFP_KERNEL); + if (!cb_ctx) + return -ENOMEM; + + cb_ctx->device = device; + cb_ctx->class_port_info = class_port_info; + cb_ctx->port_num = port_num; + init_completion(&cb_ctx->done); + + ret = ib_sa_classport_info_rec_query(&sa_client, device, port_num, + CMA_QUERY_CLASSPORT_INFO_TIMEOUT, + GFP_KERNEL, cma_query_sa_classport_info_cb, + cb_ctx, &cb_ctx->sa_query); + if (ret < 0) { + pr_err("RDMA CM: %s port %u failed to send ClassPortInfo query, ret: %d\n", + device->name, port_num, ret); + goto out; + } + + wait_for_completion(&cb_ctx->done); + +out: + kfree(cb_ctx); + return ret; +} + static int cma_join_ib_multicast(struct rdma_id_private *id_priv, struct cma_multicast *mc) { struct ib_sa_mcmember_rec rec; + struct ib_class_port_info class_port_info; struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr; ib_sa_comp_mask comp_mask; int ret; @@ -3776,7 +3839,24 @@ static int cma_join_ib_multicast(struct rdma_id_private *id_priv, rec.qkey = cpu_to_be32(id_priv->qkey); rdma_addr_get_sgid(dev_addr, &rec.port_gid); rec.pkey = cpu_to_be16(ib_addr_get_pkey(dev_addr)); - rec.join_state = 1; + rec.join_state = mc->join_state; + + if (rec.join_state == BIT(SENDONLY_FULLMEMBER_JOIN)) { + ret = cma_query_sa_classport_info(id_priv->id.device, + id_priv->id.port_num, + &class_port_info); + + if (ret) + return ret; + + if (!(ib_get_cpi_capmask2(&class_port_info) & + IB_SA_CAP_MASK2_SENDONLY_FULL_MEM_SUPPORT)) { + pr_warn("RDMA CM: %s port %u Unable to multicast join\n" + "RDMA CM: SM doesn't support Send Only Full Member option\n", + id_priv->id.device->name, id_priv->id.port_num); + return -EOPNOTSUPP; + } + } comp_mask = IB_SA_MCMEMBER_REC_MGID | IB_SA_MCMEMBER_REC_PORT_GID | IB_SA_MCMEMBER_REC_PKEY | IB_SA_MCMEMBER_REC_JOIN_STATE | @@ -3845,6 +3925,9 @@ static int cma_iboe_join_multicast(struct rdma_id_private *id_priv, struct sockaddr *addr = (struct sockaddr *)&mc->addr; struct net_device *ndev = NULL; enum ib_gid_type gid_type; + bool send_only; + + send_only = mc->join_state == BIT(SENDONLY_FULLMEMBER_JOIN); if (cma_zero_addr((struct sockaddr *)&mc->addr)) return -EINVAL; @@ -3878,12 +3961,14 @@ static int cma_iboe_join_multicast(struct rdma_id_private *id_priv, gid_type = id_priv->cma_dev->default_gid_type[id_priv->id.port_num - rdma_start_port(id_priv->cma_dev->device)]; if (addr->sa_family == AF_INET) { - if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) - err = cma_igmp_send(ndev, &mc->multicast.ib->rec.mgid, - true); - if (!err) { - mc->igmp_joined = true; + if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) { mc->multicast.ib->rec.hop_limit = IPV6_DEFAULT_HOPLIMIT; + if (!send_only) { + err = cma_igmp_send(ndev, &mc->multicast.ib->rec.mgid, + true); + if (!err) + mc->igmp_joined = true; + } } } else { if (gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) @@ -3913,7 +3998,7 @@ out1: } int rdma_join_multicast(struct rdma_cm_id *id, struct sockaddr *addr, - void *context) + u8 join_state, void *context) { struct rdma_id_private *id_priv; struct cma_multicast *mc; @@ -3932,6 +4017,7 @@ int rdma_join_multicast(struct rdma_cm_id *id, struct sockaddr *addr, mc->context = context; mc->id_priv = id_priv; mc->igmp_joined = false; + mc->join_state = join_state; spin_lock(&id_priv->lock); list_add(&mc->list, &id_priv->mc_list); spin_unlock(&id_priv->lock); diff --git a/drivers/infiniband/core/iwpm_util.c b/drivers/infiniband/core/iwpm_util.c index b65e06c560d7..ade71e7f0131 100644 --- a/drivers/infiniband/core/iwpm_util.c +++ b/drivers/infiniband/core/iwpm_util.c @@ -37,6 +37,7 @@ #define IWPM_MAPINFO_HASH_MASK (IWPM_MAPINFO_HASH_SIZE - 1) #define IWPM_REMINFO_HASH_SIZE 64 #define IWPM_REMINFO_HASH_MASK (IWPM_REMINFO_HASH_SIZE - 1) +#define IWPM_MSG_SIZE 512 static LIST_HEAD(iwpm_nlmsg_req_list); static DEFINE_SPINLOCK(iwpm_nlmsg_req_lock); @@ -452,7 +453,7 @@ struct sk_buff *iwpm_create_nlmsg(u32 nl_op, struct nlmsghdr **nlh, { struct sk_buff *skb = NULL; - skb = dev_alloc_skb(NLMSG_GOODSIZE); + skb = dev_alloc_skb(IWPM_MSG_SIZE); if (!skb) { pr_err("%s Unable to allocate skb\n", __func__); goto create_nlmsg_exit; diff --git a/drivers/infiniband/core/multicast.c b/drivers/infiniband/core/multicast.c index a83ec28a147b..3a3c5d73bbfc 100644 --- a/drivers/infiniband/core/multicast.c +++ b/drivers/infiniband/core/multicast.c @@ -93,18 +93,6 @@ enum { struct mcast_member; -/* -* There are 4 types of join states: -* FullMember, NonMember, SendOnlyNonMember, SendOnlyFullMember. -*/ -enum { - FULLMEMBER_JOIN, - NONMEMBER_JOIN, - SENDONLY_NONMEBER_JOIN, - SENDONLY_FULLMEMBER_JOIN, - NUM_JOIN_MEMBERSHIP_TYPES, -}; - struct mcast_group { struct ib_sa_mcmember_rec rec; struct rb_node node; diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index e95538650dc6..b9bf7aa055e7 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -65,10 +65,17 @@ struct ib_sa_sm_ah { u8 src_path_mask; }; +struct ib_sa_classport_cache { + bool valid; + struct ib_class_port_info data; +}; + struct ib_sa_port { struct ib_mad_agent *agent; struct ib_sa_sm_ah *sm_ah; struct work_struct update_task; + struct ib_sa_classport_cache classport_info; + spinlock_t classport_lock; /* protects class port info set */ spinlock_t ah_lock; u8 port_num; }; @@ -998,6 +1005,13 @@ static void ib_sa_event(struct ib_event_handler *handler, struct ib_event *event port->sm_ah = NULL; spin_unlock_irqrestore(&port->ah_lock, flags); + if (event->event == IB_EVENT_SM_CHANGE || + event->event == IB_EVENT_CLIENT_REREGISTER || + event->event == IB_EVENT_LID_CHANGE) { + spin_lock_irqsave(&port->classport_lock, flags); + port->classport_info.valid = false; + spin_unlock_irqrestore(&port->classport_lock, flags); + } queue_work(ib_wq, &sa_dev->port[event->element.port_num - sa_dev->start_port].update_task); } @@ -1719,6 +1733,7 @@ static void ib_sa_classport_info_rec_callback(struct ib_sa_query *sa_query, int status, struct ib_sa_mad *mad) { + unsigned long flags; struct ib_sa_classport_info_query *query = container_of(sa_query, struct ib_sa_classport_info_query, sa_query); @@ -1728,6 +1743,16 @@ static void ib_sa_classport_info_rec_callback(struct ib_sa_query *sa_query, ib_unpack(classport_info_rec_table, ARRAY_SIZE(classport_info_rec_table), mad->data, &rec); + + spin_lock_irqsave(&sa_query->port->classport_lock, flags); + if (!status && !sa_query->port->classport_info.valid) { + memcpy(&sa_query->port->classport_info.data, &rec, + sizeof(sa_query->port->classport_info.data)); + + sa_query->port->classport_info.valid = true; + } + spin_unlock_irqrestore(&sa_query->port->classport_lock, flags); + query->callback(status, &rec, query->context); } else { query->callback(status, NULL, query->context); @@ -1754,7 +1779,9 @@ int ib_sa_classport_info_rec_query(struct ib_sa_client *client, struct ib_sa_port *port; struct ib_mad_agent *agent; struct ib_sa_mad *mad; + struct ib_class_port_info cached_class_port_info; int ret; + unsigned long flags; if (!sa_dev) return -ENODEV; @@ -1762,6 +1789,17 @@ int ib_sa_classport_info_rec_query(struct ib_sa_client *client, port = &sa_dev->port[port_num - sa_dev->start_port]; agent = port->agent; + /* Use cached ClassPortInfo attribute if valid instead of sending mad */ + spin_lock_irqsave(&port->classport_lock, flags); + if (port->classport_info.valid && callback) { + memcpy(&cached_class_port_info, &port->classport_info.data, + sizeof(cached_class_port_info)); + spin_unlock_irqrestore(&port->classport_lock, flags); + callback(0, &cached_class_port_info, context); + return 0; + } + spin_unlock_irqrestore(&port->classport_lock, flags); + query = kzalloc(sizeof(*query), gfp_mask); if (!query) return -ENOMEM; @@ -1885,6 +1923,9 @@ static void ib_sa_add_one(struct ib_device *device) sa_dev->port[i].sm_ah = NULL; sa_dev->port[i].port_num = i + s; + spin_lock_init(&sa_dev->port[i].classport_lock); + sa_dev->port[i].classport_info.valid = false; + sa_dev->port[i].agent = ib_register_mad_agent(device, i + s, IB_QPT_GSI, NULL, 0, send_handler, diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index c0f3826abb30..2825ece91d3c 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -106,6 +106,7 @@ struct ucma_multicast { int events_reported; u64 uid; + u8 join_state; struct list_head list; struct sockaddr_storage addr; }; @@ -1317,12 +1318,20 @@ static ssize_t ucma_process_join(struct ucma_file *file, struct ucma_multicast *mc; struct sockaddr *addr; int ret; + u8 join_state; if (out_len < sizeof(resp)) return -ENOSPC; addr = (struct sockaddr *) &cmd->addr; - if (cmd->reserved || !cmd->addr_size || (cmd->addr_size != rdma_addr_size(addr))) + if (!cmd->addr_size || (cmd->addr_size != rdma_addr_size(addr))) + return -EINVAL; + + if (cmd->join_flags == RDMA_MC_JOIN_FLAG_FULLMEMBER) + join_state = BIT(FULLMEMBER_JOIN); + else if (cmd->join_flags == RDMA_MC_JOIN_FLAG_SENDONLY_FULLMEMBER) + join_state = BIT(SENDONLY_FULLMEMBER_JOIN); + else return -EINVAL; ctx = ucma_get_ctx(file, cmd->id); @@ -1335,10 +1344,11 @@ static ssize_t ucma_process_join(struct ucma_file *file, ret = -ENOMEM; goto err1; } - + mc->join_state = join_state; mc->uid = cmd->uid; memcpy(&mc->addr, addr, cmd->addr_size); - ret = rdma_join_multicast(ctx->cm_id, (struct sockaddr *) &mc->addr, mc); + ret = rdma_join_multicast(ctx->cm_id, (struct sockaddr *)&mc->addr, + join_state, mc); if (ret) goto err2; @@ -1382,7 +1392,7 @@ static ssize_t ucma_join_ip_multicast(struct ucma_file *file, join_cmd.uid = cmd.uid; join_cmd.id = cmd.id; join_cmd.addr_size = rdma_addr_size((struct sockaddr *) &cmd.addr); - join_cmd.reserved = 0; + join_cmd.join_flags = RDMA_MC_JOIN_FLAG_FULLMEMBER; memcpy(&join_cmd.addr, &cmd.addr, join_cmd.addr_size); return ucma_process_join(file, &join_cmd, out_len); diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h index b7f3b8d359b9..df26a741cda6 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -116,6 +116,7 @@ struct ib_uverbs_event_file { struct ib_uverbs_file { struct kref ref; struct mutex mutex; + struct mutex cleanup_mutex; /* protect cleanup */ struct ib_uverbs_device *device; struct ib_ucontext *ucontext; struct ib_event_handler event_handler; diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 426e0ac203fc..0012fa58c105 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -969,6 +969,7 @@ static int ib_uverbs_open(struct inode *inode, struct file *filp) file->async_file = NULL; kref_init(&file->ref); mutex_init(&file->mutex); + mutex_init(&file->cleanup_mutex); filp->private_data = file; kobject_get(&dev->kobj); @@ -994,18 +995,20 @@ static int ib_uverbs_close(struct inode *inode, struct file *filp) { struct ib_uverbs_file *file = filp->private_data; struct ib_uverbs_device *dev = file->device; - struct ib_ucontext *ucontext = NULL; + + mutex_lock(&file->cleanup_mutex); + if (file->ucontext) { + ib_uverbs_cleanup_ucontext(file, file->ucontext); + file->ucontext = NULL; + } + mutex_unlock(&file->cleanup_mutex); mutex_lock(&file->device->lists_mutex); - ucontext = file->ucontext; - file->ucontext = NULL; if (!file->is_closed) { list_del(&file->list); file->is_closed = 1; } mutex_unlock(&file->device->lists_mutex); - if (ucontext) - ib_uverbs_cleanup_ucontext(file, ucontext); if (file->async_file) kref_put(&file->async_file->ref, ib_uverbs_release_event_file); @@ -1219,22 +1222,30 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev, mutex_lock(&uverbs_dev->lists_mutex); while (!list_empty(&uverbs_dev->uverbs_file_list)) { struct ib_ucontext *ucontext; - file = list_first_entry(&uverbs_dev->uverbs_file_list, struct ib_uverbs_file, list); file->is_closed = 1; - ucontext = file->ucontext; list_del(&file->list); - file->ucontext = NULL; kref_get(&file->ref); mutex_unlock(&uverbs_dev->lists_mutex); - /* We must release the mutex before going ahead and calling - * disassociate_ucontext. disassociate_ucontext might end up - * indirectly calling uverbs_close, for example due to freeing - * the resources (e.g mmput). - */ + ib_uverbs_event_handler(&file->event_handler, &event); + + mutex_lock(&file->cleanup_mutex); + ucontext = file->ucontext; + file->ucontext = NULL; + mutex_unlock(&file->cleanup_mutex); + + /* At this point ib_uverbs_close cannot be running + * ib_uverbs_cleanup_ucontext + */ if (ucontext) { + /* We must release the mutex before going ahead and + * calling disassociate_ucontext. disassociate_ucontext + * might end up indirectly calling uverbs_close, + * for example due to freeing the resources + * (e.g mmput). + */ ib_dev->disassociate_ucontext(ucontext); ib_uverbs_cleanup_ucontext(file, ucontext); } diff --git a/drivers/infiniband/hw/hfi1/Kconfig b/drivers/infiniband/hw/hfi1/Kconfig index a925fb0db706..f846fd51b85b 100644 --- a/drivers/infiniband/hw/hfi1/Kconfig +++ b/drivers/infiniband/hw/hfi1/Kconfig @@ -3,7 +3,6 @@ config INFINIBAND_HFI1 depends on X86_64 && INFINIBAND_RDMAVT select MMU_NOTIFIER select CRC32 - default m ---help--- This is a low-level driver for Intel OPA Gen1 adapter. config HFI1_DEBUG_SDMA_ORDER diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c index 7a5b0e676cc7..3b792c5023d0 100644 --- a/drivers/infiniband/hw/hfi1/file_ops.c +++ b/drivers/infiniband/hw/hfi1/file_ops.c @@ -225,7 +225,7 @@ static long hfi1_file_ioctl(struct file *fp, unsigned int cmd, sizeof(struct hfi1_base_info)); break; case HFI1_IOCTL_CREDIT_UPD: - if (uctxt && uctxt->sc) + if (uctxt) sc_return_credits(uctxt->sc); break; diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c index 9f8b516eb2b0..d6fc8a6e8c33 100644 --- a/drivers/infiniband/hw/mlx4/cq.c +++ b/drivers/infiniband/hw/mlx4/cq.c @@ -288,7 +288,7 @@ static int mlx4_alloc_resize_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq, if (cq->resize_buf) return -EBUSY; - cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC); + cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_KERNEL); if (!cq->resize_buf) return -ENOMEM; @@ -316,7 +316,7 @@ static int mlx4_alloc_resize_umem(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) return -EFAULT; - cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC); + cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_KERNEL); if (!cq->resize_buf) return -ENOMEM; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 1cbd075ce7e4..ee3669486996 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -2049,6 +2049,195 @@ static struct device_attribute *mlx4_class_attributes[] = { &dev_attr_board_id }; +struct diag_counter { + const char *name; + u32 offset; +}; + +#define DIAG_COUNTER(_name, _offset) \ + { .name = #_name, .offset = _offset } + +static const struct diag_counter diag_basic[] = { + DIAG_COUNTER(rq_num_lle, 0x00), + DIAG_COUNTER(sq_num_lle, 0x04), + DIAG_COUNTER(rq_num_lqpoe, 0x08), + DIAG_COUNTER(sq_num_lqpoe, 0x0C), + DIAG_COUNTER(rq_num_lpe, 0x18), + DIAG_COUNTER(sq_num_lpe, 0x1C), + DIAG_COUNTER(rq_num_wrfe, 0x20), + DIAG_COUNTER(sq_num_wrfe, 0x24), + DIAG_COUNTER(sq_num_mwbe, 0x2C), + DIAG_COUNTER(sq_num_bre, 0x34), + DIAG_COUNTER(sq_num_rire, 0x44), + DIAG_COUNTER(rq_num_rire, 0x48), + DIAG_COUNTER(sq_num_rae, 0x4C), + DIAG_COUNT |