diff options
Diffstat (limited to 'drivers/infiniband/core')
| -rw-r--r-- | drivers/infiniband/core/Makefile | 4 | ||||
| -rw-r--r-- | drivers/infiniband/core/addr.c | 22 | ||||
| -rw-r--r-- | drivers/infiniband/core/cache.c | 5 | ||||
| -rw-r--r-- | drivers/infiniband/core/cm.c | 66 | ||||
| -rw-r--r-- | drivers/infiniband/core/cma.c | 403 | ||||
| -rw-r--r-- | drivers/infiniband/core/device.c | 6 | ||||
| -rw-r--r-- | drivers/infiniband/core/iwcm.c | 1019 | ||||
| -rw-r--r-- | drivers/infiniband/core/iwcm.h | 62 | ||||
| -rw-r--r-- | drivers/infiniband/core/mad.c | 19 | ||||
| -rw-r--r-- | drivers/infiniband/core/mad_priv.h | 1 | ||||
| -rw-r--r-- | drivers/infiniband/core/mad_rmpp.c | 94 | ||||
| -rw-r--r-- | drivers/infiniband/core/sa_query.c | 67 | ||||
| -rw-r--r-- | drivers/infiniband/core/smi.c | 16 | ||||
| -rw-r--r-- | drivers/infiniband/core/sysfs.c | 13 | ||||
| -rw-r--r-- | drivers/infiniband/core/ucm.c | 9 | ||||
| -rw-r--r-- | drivers/infiniband/core/user_mad.c | 7 | ||||
| -rw-r--r-- | drivers/infiniband/core/uverbs_cmd.c | 64 | ||||
| -rw-r--r-- | drivers/infiniband/core/verbs.c | 21 |
18 files changed, 1731 insertions, 167 deletions
diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile index 68e73ec2d1f..163d991eb8c 100644 --- a/drivers/infiniband/core/Makefile +++ b/drivers/infiniband/core/Makefile @@ -1,7 +1,7 @@ infiniband-$(CONFIG_INFINIBAND_ADDR_TRANS) := ib_addr.o rdma_cm.o obj-$(CONFIG_INFINIBAND) += ib_core.o ib_mad.o ib_sa.o \ - ib_cm.o $(infiniband-y) + ib_cm.o iw_cm.o $(infiniband-y) obj-$(CONFIG_INFINIBAND_USER_MAD) += ib_umad.o obj-$(CONFIG_INFINIBAND_USER_ACCESS) += ib_uverbs.o ib_ucm.o @@ -14,6 +14,8 @@ ib_sa-y := sa_query.o ib_cm-y := cm.o +iw_cm-y := iwcm.o + rdma_cm-y := cma.o ib_addr-y := addr.o diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index 1205e802782..9cbf09e2052 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -61,12 +61,15 @@ static LIST_HEAD(req_list); static DECLARE_WORK(work, process_req, NULL); static struct workqueue_struct *addr_wq; -static int copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev, - unsigned char *dst_dev_addr) +int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev, + const unsigned char *dst_dev_addr) { switch (dev->type) { case ARPHRD_INFINIBAND: - dev_addr->dev_type = IB_NODE_CA; + dev_addr->dev_type = RDMA_NODE_IB_CA; + break; + case ARPHRD_ETHER: + dev_addr->dev_type = RDMA_NODE_RNIC; break; default: return -EADDRNOTAVAIL; @@ -78,6 +81,7 @@ static int copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev, memcpy(dev_addr->dst_dev_addr, dst_dev_addr, MAX_ADDR_LEN); return 0; } +EXPORT_SYMBOL(rdma_copy_addr); int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr) { @@ -89,7 +93,7 @@ int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr) if (!dev) return -EADDRNOTAVAIL; - ret = copy_addr(dev_addr, dev, NULL); + ret = rdma_copy_addr(dev_addr, dev, NULL); dev_put(dev); return ret; } @@ -161,7 +165,7 @@ static int addr_resolve_remote(struct sockaddr_in *src_in, /* If the device does ARP internally, return 'done' */ if (rt->idev->dev->flags & IFF_NOARP) { - copy_addr(addr, rt->idev->dev, NULL); + rdma_copy_addr(addr, rt->idev->dev, NULL); goto put; } @@ -181,7 +185,7 @@ static int addr_resolve_remote(struct sockaddr_in *src_in, src_in->sin_addr.s_addr = rt->rt_src; } - ret = copy_addr(addr, neigh->dev, neigh->ha); + ret = rdma_copy_addr(addr, neigh->dev, neigh->ha); release: neigh_release(neigh); put: @@ -245,7 +249,7 @@ static int addr_resolve_local(struct sockaddr_in *src_in, if (ZERONET(src_ip)) { src_in->sin_family = dst_in->sin_family; src_in->sin_addr.s_addr = dst_ip; - ret = copy_addr(addr, dev, dev->dev_addr); + ret = rdma_copy_addr(addr, dev, dev->dev_addr); } else if (LOOPBACK(src_ip)) { ret = rdma_translate_ip((struct sockaddr *)dst_in, addr); if (!ret) @@ -327,10 +331,10 @@ void rdma_addr_cancel(struct rdma_dev_addr *addr) } EXPORT_SYMBOL(rdma_addr_cancel); -static int netevent_callback(struct notifier_block *self, unsigned long event, +static int netevent_callback(struct notifier_block *self, unsigned long event, void *ctx) { - if (event == NETEVENT_NEIGH_UPDATE) { + if (event == NETEVENT_NEIGH_UPDATE) { struct neighbour *neigh = ctx; if (neigh->dev->type == ARPHRD_INFINIBAND && diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index 75313ade2e0..20e9f64e67a 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -62,12 +62,13 @@ struct ib_update_work { static inline int start_port(struct ib_device *device) { - return device->node_type == IB_NODE_SWITCH ? 0 : 1; + return (device->node_type == RDMA_NODE_IB_SWITCH) ? 0 : 1; } static inline int end_port(struct ib_device *device) { - return device->node_type == IB_NODE_SWITCH ? 0 : device->phys_port_cnt; + return (device->node_type == RDMA_NODE_IB_SWITCH) ? + 0 : device->phys_port_cnt; } int ib_get_cached_gid(struct ib_device *device, diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index 0de335b7bfc..f35fcc4c063 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2004, 2005 Intel Corporation. All rights reserved. + * Copyright (c) 2004-2006 Intel Corporation. All rights reserved. * Copyright (c) 2004 Topspin Corporation. All rights reserved. * Copyright (c) 2004, 2005 Voltaire Corporation. All rights reserved. * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved. @@ -41,6 +41,7 @@ #include <linux/idr.h> #include <linux/interrupt.h> #include <linux/pci.h> +#include <linux/random.h> #include <linux/rbtree.h> #include <linux/spinlock.h> #include <linux/workqueue.h> @@ -73,6 +74,7 @@ static struct ib_cm { struct rb_root remote_id_table; struct rb_root remote_sidr_table; struct idr local_id_table; + __be32 random_id_operand; struct workqueue_struct *wq; } cm; @@ -177,7 +179,7 @@ static int cm_alloc_msg(struct cm_id_private *cm_id_priv, if (IS_ERR(ah)) return PTR_ERR(ah); - m = ib_create_send_mad(mad_agent, cm_id_priv->id.remote_cm_qpn, + m = ib_create_send_mad(mad_agent, cm_id_priv->id.remote_cm_qpn, cm_id_priv->av.pkey_index, 0, IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA, GFP_ATOMIC); @@ -299,15 +301,17 @@ static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av) static int cm_alloc_id(struct cm_id_private *cm_id_priv) { unsigned long flags; - int ret; + int ret, id; static int next_id; do { spin_lock_irqsave(&cm.lock, flags); - ret = idr_get_new_above(&cm.local_id_table, cm_id_priv, next_id++, - (__force int *) &cm_id_priv->id.local_id); + ret = idr_get_new_above(&cm.local_id_table, cm_id_priv, + next_id++, &id); spin_unlock_irqrestore(&cm.lock, flags); } while( (ret == -EAGAIN) && idr_pre_get(&cm.local_id_table, GFP_KERNEL) ); + + cm_id_priv->id.local_id = (__force __be32) (id ^ cm.random_id_operand); return ret; } @@ -316,7 +320,8 @@ static void cm_free_id(__be32 local_id) unsigned long flags; spin_lock_irqsave(&cm.lock, flags); - idr_remove(&cm.local_id_table, (__force int) local_id); + idr_remove(&cm.local_id_table, + (__force int) (local_id ^ cm.random_id_operand)); spin_unlock_irqrestore(&cm.lock, flags); } @@ -324,7 +329,8 @@ static struct cm_id_private * cm_get_id(__be32 local_id, __be32 remote_id) { struct cm_id_private *cm_id_priv; - cm_id_priv = idr_find(&cm.local_id_table, (__force int) local_id); + cm_id_priv = idr_find(&cm.local_id_table, + (__force int) (local_id ^ cm.random_id_operand)); if (cm_id_priv) { if (cm_id_priv->id.remote_id == remote_id) atomic_inc(&cm_id_priv->refcount); @@ -679,6 +685,8 @@ static void cm_enter_timewait(struct cm_id_private *cm_id_priv) { int wait_time; + cm_cleanup_timewait(cm_id_priv->timewait_info); + /* * The cm_id could be destroyed by the user before we exit timewait. * To protect against this, we search for the cm_id after exiting @@ -1354,7 +1362,7 @@ static int cm_req_handler(struct cm_work *work) id.local_id); if (IS_ERR(cm_id_priv->timewait_info)) { ret = PTR_ERR(cm_id_priv->timewait_info); - goto error1; + goto destroy; } cm_id_priv->timewait_info->work.remote_id = req_msg->local_comm_id; cm_id_priv->timewait_info->remote_ca_guid = req_msg->local_ca_guid; @@ -1363,7 +1371,8 @@ static int cm_req_handler(struct cm_work *work) listen_cm_id_priv = cm_match_req(work, cm_id_priv); if (!listen_cm_id_priv) { ret = -EINVAL; - goto error2; + kfree(cm_id_priv->timewait_info); + goto destroy; } cm_id_priv->id.cm_handler = listen_cm_id_priv->id.cm_handler; @@ -1373,12 +1382,22 @@ static int cm_req_handler(struct cm_work *work) cm_format_paths_from_req(req_msg, &work->path[0], &work->path[1]); ret = cm_init_av_by_path(&work->path[0], &cm_id_priv->av); - if (ret) - goto error3; + if (ret) { + ib_get_cached_gid(work->port->cm_dev->device, + work->port->port_num, 0, &work->path[0].sgid); + ib_send_cm_rej(cm_id, IB_CM_REJ_INVALID_GID, + &work->path[0].sgid, sizeof work->path[0].sgid, + NULL, 0); + goto rejected; + } if (req_msg->alt_local_lid) { ret = cm_init_av_by_path(&work->path[1], &cm_id_priv->alt_av); - if (ret) - goto error3; + if (ret) { + ib_send_cm_rej(cm_id, IB_CM_REJ_INVALID_ALT_GID, + &work->path[0].sgid, + sizeof work->path[0].sgid, NULL, 0); + goto rejected; + } } cm_id_priv->tid = req_msg->hdr.tid; cm_id_priv->timeout_ms = cm_convert_to_ms( @@ -1400,12 +1419,11 @@ static int cm_req_handler(struct cm_work *work) cm_deref_id(listen_cm_id_priv); return 0; -error3: atomic_dec(&cm_id_priv->refcount); +rejected: + atomic_dec(&cm_id_priv->refcount); cm_deref_id(listen_cm_id_priv); - cm_cleanup_timewait(cm_id_priv->timewait_info); -error2: kfree(cm_id_priv->timewait_info); - cm_id_priv->timewait_info = NULL; -error1: ib_destroy_cm_id(&cm_id_priv->id); +destroy: + ib_destroy_cm_id(cm_id); return ret; } @@ -2072,8 +2090,9 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg) spin_unlock_irqrestore(&cm.lock, flags); return NULL; } - cm_id_priv = idr_find(&cm.local_id_table, - (__force int) timewait_info->work.local_id); + cm_id_priv = idr_find(&cm.local_id_table, (__force int) + (timewait_info->work.local_id ^ + cm.random_id_operand)); if (cm_id_priv) { if (cm_id_priv->id.remote_id == remote_id) atomic_inc(&cm_id_priv->refcount); @@ -3125,7 +3144,8 @@ static int cm_init_qp_init_attr(struct cm_id_private *cm_id_priv, qp_attr->qp_access_flags = IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE; if (cm_id_priv->responder_resources) - qp_attr->qp_access_flags |= IB_ACCESS_REMOTE_READ; + qp_attr->qp_access_flags |= IB_ACCESS_REMOTE_READ | + IB_ACCESS_REMOTE_ATOMIC; qp_attr->pkey_index = cm_id_priv->av.pkey_index; qp_attr->port_num = cm_id_priv->av.port->port_num; ret = 0; @@ -3262,6 +3282,9 @@ static void cm_add_one(struct ib_device *device) int ret; u8 i; + if (rdma_node_get_transport(device->node_type) != RDMA_TRANSPORT_IB) + return; + cm_dev = kmalloc(sizeof(*cm_dev) + sizeof(*port) * device->phys_port_cnt, GFP_KERNEL); if (!cm_dev) @@ -3349,6 +3372,7 @@ static int __init ib_cm_init(void) cm.remote_qp_table = RB_ROOT; cm.remote_sidr_table = RB_ROOT; idr_init(&cm.local_id_table); + get_random_bytes(&cm.random_id_operand, sizeof cm.random_id_operand); idr_pre_get(&cm.local_id_table, GFP_KERNEL); cm.wq = create_workqueue("ib_cm"); diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 5d625a81193..1178bd434d1 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -35,6 +35,7 @@ #include <linux/mutex.h> #include <linux/random.h> #include <linux/idr.h> +#include <linux/inetdevice.h> #include <net/tcp.h> @@ -43,6 +44,7 @@ #include <rdma/ib_cache.h> #include <rdma/ib_cm.h> #include <rdma/ib_sa.h> +#include <rdma/iw_cm.h> MODULE_AUTHOR("Sean Hefty"); MODULE_DESCRIPTION("Generic RDMA CM Agent"); @@ -60,6 +62,7 @@ static struct ib_client cma_client = { .remove = cma_remove_one }; +static struct ib_sa_client sa_client; static LIST_HEAD(dev_list); static LIST_HEAD(listen_any_list); static DEFINE_MUTEX(lock); @@ -124,6 +127,7 @@ struct rdma_id_private { int query_id; union { struct ib_cm_id *ib; + struct iw_cm_id *iw; } cm_id; u32 seq_num; @@ -259,15 +263,24 @@ static void cma_detach_from_dev(struct rdma_id_private *id_priv) id_priv->cma_dev = NULL; } -static int cma_acquire_ib_dev(struct rdma_id_private *id_priv) +static int cma_acquire_dev(struct rdma_id_private *id_priv) { + enum rdma_node_type dev_type = id_priv->id.route.addr.dev_addr.dev_type; struct cma_device *cma_dev; union ib_gid gid; int ret = -ENODEV; - ib_addr_get_sgid(&id_priv->id.route.addr.dev_addr, &gid), + switch (rdma_node_get_transport(dev_type)) { + case RDMA_TRANSPORT_IB: + ib_addr_get_sgid(&id_priv->id.route.addr.dev_addr, &gid); + break; + case RDMA_TRANSPORT_IWARP: + iw_addr_get_sgid(&id_priv->id.route.addr.dev_addr, &gid); + break; + default: + return -ENODEV; + } - mutex_lock(&lock); list_for_each_entry(cma_dev, &dev_list, list) { ret = ib_find_cached_gid(cma_dev->device, &gid, &id_priv->id.port_num, NULL); @@ -276,20 +289,9 @@ static int cma_acquire_ib_dev(struct rdma_id_private *id_priv) break; } } - mutex_unlock(&lock); return ret; } -static int cma_acquire_dev(struct rdma_id_private *id_priv) -{ - switch (id_priv->id.route.addr.dev_addr.dev_type) { - case IB_NODE_CA: - return cma_acquire_ib_dev(id_priv); - default: - return -ENODEV; - } -} - static void cma_deref_id(struct rdma_id_private *id_priv) { if (atomic_dec_and_test(&id_priv->refcount)) @@ -347,6 +349,16 @@ static int cma_init_ib_qp(struct rdma_id_private *id_priv, struct ib_qp *qp) IB_QP_PKEY_INDEX | IB_QP_PORT); } +static int cma_init_iw_qp(struct rdma_id_private *id_priv, struct ib_qp *qp) +{ + struct ib_qp_attr qp_attr; + + qp_attr.qp_state = IB_QPS_INIT; + qp_attr.qp_access_flags = IB_ACCESS_LOCAL_WRITE; + + return ib_modify_qp(qp, &qp_attr, IB_QP_STATE | IB_QP_ACCESS_FLAGS); +} + int rdma_create_qp(struct rdma_cm_id *id, struct ib_pd *pd, struct ib_qp_init_attr *qp_init_attr) { @@ -362,10 +374,13 @@ int rdma_create_qp(struct rdma_cm_id *id, struct ib_pd *pd, if (IS_ERR(qp)) return PTR_ERR(qp); - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: ret = cma_init_ib_qp(id_priv, qp); break; + case RDMA_TRANSPORT_IWARP: + ret = cma_init_iw_qp(id_priv, qp); + break; default: ret = -ENOSYS; break; @@ -451,13 +466,17 @@ int rdma_init_qp_attr(struct rdma_cm_id *id, struct ib_qp_attr *qp_attr, int ret; id_priv = container_of(id, struct rdma_id_private, id); - switch (id_priv->id.device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id_priv->id.device->node_type)) { + case RDMA_TRANSPORT_IB: ret = ib_cm_init_qp_attr(id_priv->cm_id.ib, qp_attr, qp_attr_mask); if (qp_attr->qp_state == IB_QPS_RTR) qp_attr->rq_psn = id_priv->seq_num; break; + case RDMA_TRANSPORT_IWARP: + ret = iw_cm_init_qp_attr(id_priv->cm_id.iw, qp_attr, + qp_attr_mask); + break; default: ret = -ENOSYS; break; @@ -590,8 +609,8 @@ static int cma_notify_user(struct rdma_id_private *id_priv, static void cma_cancel_route(struct rdma_id_private *id_priv) { - switch (id_priv->id.device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id_priv->id.device->node_type)) { + case RDMA_TRANSPORT_IB: if (id_priv->query) ib_sa_cancel_query(id_priv->query_id, id_priv->query); break; @@ -611,11 +630,15 @@ static void cma_destroy_listen(struct rdma_id_private *id_priv) cma_exch(id_priv, CMA_DESTROYING); if (id_priv->cma_dev) { - switch (id_priv->id.device->node_type) { - case IB_NODE_CA: - if (id_priv->cm_id.ib && !IS_ERR(id_priv->cm_id.ib)) + switch (rdma_node_get_transport(id_priv->id.device->node_type)) { + case RDMA_TRANSPORT_IB: + if (id_priv->cm_id.ib && !IS_ERR(id_priv->cm_id.ib)) ib_destroy_cm_id(id_priv->cm_id.ib); break; + case RDMA_TRANSPORT_IWARP: + if (id_priv->cm_id.iw && !IS_ERR(id_priv->cm_id.iw)) + iw_destroy_cm_id(id_priv->cm_id.iw); + break; default: break; } @@ -689,19 +712,25 @@ void rdma_destroy_id(struct rdma_cm_id *id) state = cma_exch(id_priv, CMA_DESTROYING); cma_cancel_operation(id_priv, state); + mutex_lock(&lock); if (id_priv->cma_dev) { - switch (id->device->node_type) { - case IB_NODE_CA: - if (id_priv->cm_id.ib && !IS_ERR(id_priv->cm_id.ib)) + mutex_unlock(&lock); + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: + if (id_priv->cm_id.ib && !IS_ERR(id_priv->cm_id.ib)) ib_destroy_cm_id(id_priv->cm_id.ib); break; + case RDMA_TRANSPORT_IWARP: + if (id_priv->cm_id.iw && !IS_ERR(id_priv->cm_id.iw)) + iw_destroy_cm_id(id_priv->cm_id.iw); + break; default: break; } - mutex_lock(&lock); + mutex_lock(&lock); cma_detach_from_dev(id_priv); - mutex_unlock(&lock); } + mutex_unlock(&lock); cma_release_port(id_priv); cma_deref_id(id_priv); @@ -869,7 +898,7 @@ static struct rdma_id_private *cma_new_id(struct rdma_cm_id *listen_id, ib_addr_set_sgid(&rt->addr.dev_addr, &rt->path_rec[0].sgid); ib_addr_set_dgid(&rt->addr.dev_addr, &rt->path_rec[0].dgid); ib_addr_set_pkey(&rt->addr.dev_addr, be16_to_cpu(rt->path_rec[0].pkey)); - rt->addr.dev_addr.dev_type = IB_NODE_CA; + rt->addr.dev_addr.dev_type = RDMA_NODE_IB_CA; id_priv = container_of(id, struct rdma_id_private, id); id_priv->state = CMA_CONNECT; @@ -898,7 +927,9 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event) } atomic_inc(&conn_id->dev_remove); - ret = cma_acquire_ib_dev(conn_id); + mutex_lock(&lock); + ret = cma_acquire_dev(conn_id); + mutex_unlock(&lock); if (ret) { ret = -ENODEV; cma_release_remove(conn_id); @@ -982,6 +1013,130 @@ static void cma_set_compare_data(enum rdma_port_space ps, struct sockaddr *addr, } } +static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event) +{ + struct rdma_id_private *id_priv = iw_id->context; + enum rdma_cm_event_type event = 0; + struct sockaddr_in *sin; + int ret = 0; + + atomic_inc(&id_priv->dev_remove); + + switch (iw_event->event) { + case IW_CM_EVENT_CLOSE: + event = RDMA_CM_EVENT_DISCONNECTED; + break; + case IW_CM_EVENT_CONNECT_REPLY: + sin = (struct sockaddr_in *) &id_priv->id.route.addr.src_addr; + *sin = iw_event->local_addr; + sin = (struct sockaddr_in *) &id_priv->id.route.addr.dst_addr; + *sin = iw_event->remote_addr; + if (iw_event->status) + event = RDMA_CM_EVENT_REJECTED; + else + event = RDMA_CM_EVENT_ESTABLISHED; + break; + case IW_CM_EVENT_ESTABLISHED: + event = RDMA_CM_EVENT_ESTABLISHED; + break; + default: + BUG_ON(1); + } + + ret = cma_notify_user(id_priv, event, iw_event->status, + iw_event->private_data, + iw_event->private_data_len); + if (ret) { + /* Destroy the CM ID by returning a non-zero value. */ + id_priv->cm_id.iw = NULL; + cma_exch(id_priv, CMA_DESTROYING); + cma_release_remove(id_priv); + rdma_destroy_id(&id_priv->id); + return ret; + } + + cma_release_remove(id_priv); + return ret; +} + +static int iw_conn_req_handler(struct iw_cm_id *cm_id, + struct iw_cm_event *iw_event) +{ + struct rdma_cm_id *new_cm_id; + struct rdma_id_private *listen_id, *conn_id; + struct sockaddr_in *sin; + struct net_device *dev = NULL; + int ret; + + listen_id = cm_id->context; + atomic_inc(&listen_id->dev_remove); + if (!cma_comp(listen_id, CMA_LISTEN)) { + ret = -ECONNABORTED; + goto out; + } + + /* Create a new RDMA id for the new IW CM ID */ + new_cm_id = rdma_create_id(listen_id->id.event_handler, + listen_id->id.context, + RDMA_PS_TCP); + if (!new_cm_id) { + ret = -ENOMEM; + goto out; + } + conn_id = container_of(new_cm_id, struct rdma_id_private, id); + atomic_inc(&conn_id->dev_remove); + conn_id->state = CMA_CONNECT; + + dev = ip_dev_find(iw_event->local_addr.sin_addr.s_addr); + if (!dev) { + ret = -EADDRNOTAVAIL; + cma_release_remove(conn_id); + rdma_destroy_id(new_cm_id); + goto out; + } + ret = rdma_copy_addr(&conn_id->id.route.addr.dev_addr, dev, NULL); + if (ret) { + cma_release_remove(conn_id); + rdma_destroy_id(new_cm_id); + goto out; + } + + mutex_lock(&lock); + ret = cma_acquire_dev(conn_id); + mutex_unlock(&lock); + if (ret) { + cma_release_remove(conn_id); + rdma_destroy_id(new_cm_id); + goto out; + } + + conn_id->cm_id.iw = cm_id; + cm_id->context = conn_id; + cm_id->cm_handler = cma_iw_handler; + + sin = (struct sockaddr_in *) &new_cm_id->route.addr.src_addr; + *sin = iw_event->local_addr; + sin = (struct sockaddr_in *) &new_cm_id->route.addr.dst_addr; + *sin = iw_event->remote_addr; + + ret = cma_notify_user(conn_id, RDMA_CM_EVENT_CONNECT_REQUEST, 0, + iw_event->private_data, + iw_event->private_data_len); + if (ret) { + /* User wants to destroy the CM ID */ + conn_id->cm_id.iw = NULL; + cma_exch(conn_id, CMA_DESTROYING); + cma_release_remove(conn_id); + rdma_destroy_id(&conn_id->id); + } + +out: + if (dev) + dev_put(dev); + cma_release_remove(listen_id); + return ret; +} + static int cma_ib_listen(struct rdma_id_private *id_priv) { struct ib_cm_compare_data compare_data; @@ -1011,6 +1166,30 @@ static int cma_ib_listen(struct rdma_id_private *id_priv) return ret; } +static int cma_iw_listen(struct rdma_id_private *id_priv, int backlog) +{ + int ret; + struct sockaddr_in *sin; + + id_priv->cm_id.iw = iw_create_cm_id(id_priv->id.device, + iw_conn_req_handler, + id_priv); + if (IS_ERR(id_priv->cm_id.iw)) + return PTR_ERR(id_priv->cm_id.iw); + + sin = (struct sockaddr_in *) &id_priv->id.route.addr.src_addr; + id_priv->cm_id.iw->local_addr = *sin; + + ret = iw_cm_listen(id_priv->cm_id.iw, backlog); + + if (ret) { + iw_destroy_cm_id(id_priv->cm_id.iw); + id_priv->cm_id.iw = NULL; + } + + return ret; +} + static int cma_listen_handler(struct rdma_cm_id *id, struct rdma_cm_event *event) { @@ -1087,12 +1266,17 @@ int rdma_listen(struct rdma_cm_id *id, int backlog) id_priv->backlog = backlog; if (id->device) { - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: ret = cma_ib_listen(id_priv); if (ret) goto err; break; + case RDMA_TRANSPORT_IWARP: + ret = cma_iw_listen(id_priv, backlog); + if (ret) + goto err; + break; default: ret = -ENOSYS; goto err; @@ -1140,7 +1324,7 @@ static int cma_query_ib_route(struct rdma_id_private *id_priv, int timeout_ms, path_rec.pkey = cpu_to_be16(ib_addr_get_pkey(addr)); path_rec.numb_path = 1; - id_priv->query_id = ib_sa_path_rec_get(id_priv->id.device, + id_priv->query_id = ib_sa_path_rec_get(&sa_client, id_priv->id.device, id_priv->id.port_num, &path_rec, IB_SA_PATH_REC_DGID | IB_SA_PATH_REC_SGID | IB_SA_PATH_REC_PKEY | IB_SA_PATH_REC_NUMB_PATH, @@ -1231,6 +1415,23 @@ err: } EXPORT_SYMBOL(rdma_set_ib_paths); +static int cma_resolve_iw_route(struct rdma_id_private *id_priv, int timeout_ms) +{ + struct cma_work *work; + + work = kzalloc(sizeof *work, GFP_KERNEL); + if (!work) + return -ENOMEM; + + work->id = id_priv; + INIT_WORK(&work->work, cma_work_handler, work); + work->old_state = CMA_ROUTE_QUERY; + work->new_state = CMA_ROUTE_RESOLVED; + work->event.event = RDMA_CM_EVENT_ROUTE_RESOLVED; + queue_work(cma_wq, &work->work); + return 0; +} + int rdma_resolve_route(struct rdma_cm_id *id, int timeout_ms) { struct rdma_id_private *id_priv; @@ -1241,10 +1442,13 @@ int rdma_resolve_route(struct rdma_cm_id *id, int timeout_ms) return -EINVAL; atomic_inc(&id_priv->refcount); - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: ret = cma_resolve_ib_route(id_priv, timeout_ms); break; + case RDMA_TRANSPORT_IWARP: + ret = cma_resolve_iw_route(id_priv, timeout_ms); + break; default: ret = -ENOSYS; break; @@ -1309,16 +1513,26 @@ static void addr_handler(int status, struct sockaddr *src_addr, enum rdma_cm_event_type event; atomic_inc(&id_priv->dev_remove); - if (!id_priv->cma_dev && !status) + + /* + * Grab mutex to block rdma_destroy_id() from removing the device while + * we're trying to acquire it. + */ + mutex_lock(&lock); + if (!cma_comp_exch(id_priv, CMA_ADDR_QUERY, CMA_ADDR_RESOLVED)) { + mutex_unlock(&lock); + goto out; + } + + if (!status && !id_priv->cma_dev) status = cma_acquire_dev(id_priv); + mutex_unlock(&lock); if (status) { - if (!cma_comp_exch(id_priv, CMA_ADDR_QUERY, CMA_ADDR_BOUND)) + if (!cma_comp_exch(id_priv, CMA_ADDR_RESOLVED, CMA_ADDR_BOUND)) goto out; event = RDMA_CM_EVENT_ADDR_ERROR; } else { - if (!cma_comp_exch(id_priv, CMA_ADDR_QUERY, CMA_ADDR_RESOLVED)) - goto out; memcpy(&id_priv->id.route.addr.src_addr, src_addr, ip_addr_size(src_addr)); event = RDMA_CM_EVENT_ADDR_RESOLVED; @@ -1492,7 +1706,7 @@ static int cma_use_port(struct idr *ps, struct rdma_id_private *id_priv) hlist_for_each_entry(cur_id, node, &bind_list->owners, node) { if (cma_any_addr(&cur_id->id.route.addr.src_addr)) return -EADDRNOTAVAIL; - + cur_sin = (struct sockaddr_in *) &cur_id->id.route.addr.src_addr; if (sin->sin_addr.s_addr == cur_sin->sin_addr.s_addr) return -EADDRINUSE; @@ -1542,8 +1756,11 @@ int rdma_bind_addr(struct rdma_cm_id *id, struct sockaddr *addr) if (!cma_any_addr(addr)) { ret = rdma_translate_ip(addr, &id->route.addr.dev_addr); - if (!ret) + if (!ret) { + mutex_lock(&lock); ret = cma_acquire_dev(id_priv); + mutex_unlock(&lock); + } if (ret) goto err; } @@ -1649,6 +1866,47 @@ out: return ret; } +static int cma_connect_iw(struct rdma_id_private *id_priv, + struct rdma_conn_param *conn_param) +{ + struct iw_cm_id *cm_id; + struct sockaddr_in* sin; + int ret; + struct iw_cm_conn_param iw_param; + + cm_id = iw_create_cm_id(id_priv->id.device, cma_iw_handler, id_priv); + if (IS_ERR(cm_id)) { + ret = PTR_ERR(cm_id); + goto out; + } + + id_priv->cm_id.iw = cm_id; + + sin = (struct sockaddr_in*) &id_priv->id.route.addr.src_addr; + cm_id->local_addr = *sin; + + sin = (struct sockaddr_in*) &id_priv->id.route.addr.dst_addr; + cm_id->remote_addr = *sin; + + ret = cma_modify_qp_rtr(&id_priv->id); + if (ret) { + iw_destroy_cm_id(cm_id); + return ret; + } + + iw_param.ord = conn_param->initiator_depth; + iw_param.ird = conn_param->responder_resources; + iw_param.private_data = conn_param->private_data; + iw_param.private_data_len = conn_param->private_data_len; + if (id_priv->id.qp) + iw_param.qpn = id_priv->qp_num; + else + iw_param.qpn = conn_param->qp_num; + ret = iw_cm_connect(cm_id, &iw_param); +out: + return ret; +} + int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param) { struct rdma_id_private *id_priv; @@ -1664,10 +1922,13 @@ int rdma_connect(struct rdma_cm_id *id, struct rdma_conn_param *conn_param) id_priv->srq = conn_param->srq; } - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: ret = cma_connect_ib(id_priv, conn_param); break; + case RDMA_TRANSPORT_IWARP: + ret = cma_connect_iw(id_priv, conn_param); + break; default: ret = -ENOSYS; break; @@ -1708,6 +1969,28 @@ static int cma_accept_ib(struct rdma_id_private *id_priv, return ib_send_cm_rep(id_priv->cm_id.ib, &rep); } +static int cma_accept_iw(struct rdma_id_private *id_priv, + struct rdma_conn_param *conn_param) +{ + struct iw_cm_conn_param iw_param; + int ret; + + ret = cma_modify_qp_rtr(&id_priv->id); + if (ret) + return ret; + + iw_param.ord = conn_param->initiator_depth; + iw_param.ird = conn_param->responder_resources; + iw_param.private_data = conn_param->private_data; + iw_param.private_data_len = conn_param->private_data_len; + if (id_priv->id.qp) { + iw_param.qpn = id_priv->qp_num; + } else + iw_param.qpn = conn_param->qp_num; + + return iw_cm_accept(id_priv->cm_id.iw, &iw_param); +} + int rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param) { struct rdma_id_private *id_priv; @@ -1723,13 +2006,16 @@ int rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param) id_priv->srq = conn_param->srq; } - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: if (conn_param) ret = cma_accept_ib(id_priv, conn_param); else ret = cma_rep_recv(id_priv); break; + case RDMA_TRANSPORT_IWARP: + ret = cma_accept_iw(id_priv, conn_param); + break; default: ret = -ENOSYS; break; @@ -1756,12 +2042,16 @@ int rdma_reject(struct rdma_cm_id *id, const void *private_data, if (!cma_comp(id_priv, CMA_CONNECT)) return -EINVAL; - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: ret = ib_send_cm_rej(id_priv->cm_id.ib, IB_CM_REJ_CONSUMER_DEFINED, NULL, 0, private_data, private_data_len); break; + case RDMA_TRANSPORT_IWARP: + ret = iw_cm_reject(id_priv->cm_id.iw, + private_data, private_data_len); + break; default: ret = -ENOSYS; break; @@ -1780,17 +2070,20 @@ int rdma_disconnect(struct rdma_cm_id *id) !cma_comp(id_priv, CMA_DISCONNECT)) return -EINVAL; - ret = cma_modify_qp_err(id); - if (ret) - goto out; - - switch (id->device->node_type) { - case IB_NODE_CA: + switch (rdma_node_get_transport(id->device->node_type)) { + case RDMA_TRANSPORT_IB: + ret = cma_modify_qp_err(id); + if (ret) + goto out; /* Initiate or respond to a disconnect. */ if (ib_send_cm_dreq(id_priv->cm_id.ib, NULL |