diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2012-07-24 13:56:26 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2012-07-24 13:56:26 -0700 |
commit | 5dedb9f3bd5bcb186313ea0c0cff8f2c525d4122 (patch) | |
tree | 88514547a6e95176e7a9dc2fbdc7fa7c1bec3107 /drivers/infiniband/core | |
parent | ddb03448274b95bff6df2a2f1a74d7eb4be529d3 (diff) | |
parent | 089117e1ad265625b523a4168f77f2521b18fd32 (diff) | |
download | op-kernel-dev-5dedb9f3bd5bcb186313ea0c0cff8f2c525d4122.zip op-kernel-dev-5dedb9f3bd5bcb186313ea0c0cff8f2c525d4122.tar.gz |
Merge tag 'rdma-for-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband
Pull InfiniBand/RDMA changes from Roland Dreier:
- Updates to the qib low-level driver
- First chunk of changes for SR-IOV support for mlx4 IB
- RDMA CM support for IPv6-only binding
- Other misc cleanups and fixes
Fix up some add-add conflicts in include/linux/mlx4/device.h and
drivers/net/ethernet/mellanox/mlx4/main.c
* tag 'rdma-for-3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (30 commits)
IB/qib: checkpatch fixes
IB/qib: Add congestion control agent implementation
IB/qib: Reduce sdma_lock contention
IB/qib: Fix an incorrect log message
IB/qib: Fix QP RCU sparse warnings
mlx4: Put physical GID and P_Key table sizes in mlx4_phys_caps struct and paravirtualize them
mlx4_core: Allow guests to have IB ports
mlx4_core: Implement mechanism for reserved Q_Keys
net/mlx4_core: Free ICM table in case of error
IB/cm: Destroy idr as part of the module init error flow
mlx4_core: Remove double function declarations
IB/mlx4: Fill the masked_atomic_cap attribute in query device
IB/mthca: Fill in sq_sig_type in query QP
IB/mthca: Warning about event for non-existent QPs should show event type
IB/qib: Fix sparse RCU warnings in qib_keys.c
net/mlx4_core: Initialize IB port capabilities for all slaves
mlx4: Use port management change event instead of smp_snoop
IB/qib: RCU locking for MR validation
IB/qib: Avoid returning EBUSY from MR deregister
IB/qib: Fix UC MR refs for immediate operations
...
Diffstat (limited to 'drivers/infiniband/core')
-rw-r--r-- | drivers/infiniband/core/addr.c | 4 | ||||
-rw-r--r-- | drivers/infiniband/core/cm.c | 16 | ||||
-rw-r--r-- | drivers/infiniband/core/cm_msgs.h | 12 | ||||
-rw-r--r-- | drivers/infiniband/core/cma.c | 77 | ||||
-rw-r--r-- | drivers/infiniband/core/sa_query.c | 133 | ||||
-rw-r--r-- | drivers/infiniband/core/ucma.c | 7 |
6 files changed, 211 insertions, 38 deletions
diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index 6ef660c..28058ae 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -129,7 +129,7 @@ int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr) dev_put(dev); break; -#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) +#if IS_ENABLED(CONFIG_IPV6) case AF_INET6: rcu_read_lock(); for_each_netdev_rcu(&init_net, dev) { @@ -243,7 +243,7 @@ out: return ret; } -#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) +#if IS_ENABLED(CONFIG_IPV6) static int addr6_resolve(struct sockaddr_in6 *src_in, struct sockaddr_in6 *dst_in, struct rdma_dev_addr *addr) diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index c889aae..d67999f 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -3848,24 +3848,28 @@ static int __init ib_cm_init(void) INIT_LIST_HEAD(&cm.timewait_list); ret = class_register(&cm_class); - if (ret) - return -ENOMEM; + if (ret) { + ret = -ENOMEM; + goto error1; + } cm.wq = create_workqueue("ib_cm"); if (!cm.wq) { ret = -ENOMEM; - goto error1; + goto error2; } ret = ib_register_client(&cm_client); if (ret) - goto error2; + goto error3; return 0; -error2: +error3: destroy_workqueue(cm.wq); -error1: +error2: class_unregister(&cm_class); +error1: + idr_destroy(&cm.local_id_table); return ret; } diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h index 7da9b21..be068f4 100644 --- a/drivers/infiniband/core/cm_msgs.h +++ b/drivers/infiniband/core/cm_msgs.h @@ -44,18 +44,6 @@ #define IB_CM_CLASS_VERSION 2 /* IB specification 1.2 */ -#define CM_REQ_ATTR_ID cpu_to_be16(0x0010) -#define CM_MRA_ATTR_ID cpu_to_be16(0x0011) -#define CM_REJ_ATTR_ID cpu_to_be16(0x0012) -#define CM_REP_ATTR_ID cpu_to_be16(0x0013) -#define CM_RTU_ATTR_ID cpu_to_be16(0x0014) -#define CM_DREQ_ATTR_ID cpu_to_be16(0x0015) -#define CM_DREP_ATTR_ID cpu_to_be16(0x0016) -#define CM_SIDR_REQ_ATTR_ID cpu_to_be16(0x0017) -#define CM_SIDR_REP_ATTR_ID cpu_to_be16(0x0018) -#define CM_LAP_ATTR_ID cpu_to_be16(0x0019) -#define CM_APR_ATTR_ID cpu_to_be16(0x001A) - enum cm_msg_sequence { CM_MSG_SEQUENCE_REQ, CM_MSG_SEQUENCE_LAP, diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 2e826f9..5a335b5 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -99,6 +99,10 @@ struct rdma_bind_list { unsigned short port; }; +enum { + CMA_OPTION_AFONLY, +}; + /* * Device removal can occur at anytime, so we need extra handling to * serialize notifying the user of device removal with other callbacks. @@ -137,9 +141,11 @@ struct rdma_id_private { u32 qkey; u32 qp_num; pid_t owner; + u32 options; u8 srq; u8 tos; u8 reuseaddr; + u8 afonly; }; struct cma_multicast { @@ -1297,8 +1303,10 @@ static void cma_set_compare_data(enum rdma_port_space ps, struct sockaddr *addr, } else { cma_set_ip_ver(cma_data, 4); cma_set_ip_ver(cma_mask, 0xF); - cma_data->dst_addr.ip4.addr = ip4_addr; - cma_mask->dst_addr.ip4.addr = htonl(~0); + if (!cma_any_addr(addr)) { + cma_data->dst_addr.ip4.addr = ip4_addr; + cma_mask->dst_addr.ip4.addr = htonl(~0); + } } break; case AF_INET6: @@ -1312,9 +1320,11 @@ static void cma_set_compare_data(enum rdma_port_space ps, struct sockaddr *addr, } else { cma_set_ip_ver(cma_data, 6); cma_set_ip_ver(cma_mask, 0xF); - cma_data->dst_addr.ip6 = ip6_addr; - memset(&cma_mask->dst_addr.ip6, 0xFF, - sizeof cma_mask->dst_addr.ip6); + if (!cma_any_addr(addr)) { + cma_data->dst_addr.ip6 = ip6_addr; + memset(&cma_mask->dst_addr.ip6, 0xFF, + sizeof cma_mask->dst_addr.ip6); + } } break; default: @@ -1499,7 +1509,7 @@ static int cma_ib_listen(struct rdma_id_private *id_priv) addr = (struct sockaddr *) &id_priv->id.route.addr.src_addr; svc_id = cma_get_service_id(id_priv->id.ps, addr); - if (cma_any_addr(addr)) + if (cma_any_addr(addr) && !id_priv->afonly) ret = ib_cm_listen(id_priv->cm_id.ib, svc_id, 0, NULL); else { cma_set_compare_data(id_priv->id.ps, addr, &compare_data); @@ -1573,6 +1583,7 @@ static void cma_listen_on_dev(struct rdma_id_private *id_priv, list_add_tail(&dev_id_priv->listen_list, &id_priv->listen_list); atomic_inc(&id_priv->refcount); dev_id_priv->internal_id = 1; + dev_id_priv->afonly = id_priv->afonly; ret = rdma_listen(id, id_priv->backlog); if (ret) @@ -2098,6 +2109,26 @@ int rdma_set_reuseaddr(struct rdma_cm_id *id, int reuse) } EXPORT_SYMBOL(rdma_set_reuseaddr); +int rdma_set_afonly(struct rdma_cm_id *id, int afonly) +{ + struct rdma_id_private *id_priv; + unsigned long flags; + int ret; + + id_priv = container_of(id, struct rdma_id_private, id); + spin_lock_irqsave(&id_priv->lock, flags); + if (id_priv->state == RDMA_CM_IDLE || id_priv->state == RDMA_CM_ADDR_BOUND) { + id_priv->options |= (1 << CMA_OPTION_AFONLY); + id_priv->afonly = afonly; + ret = 0; + } else { + ret = -EINVAL; + } + spin_unlock_irqrestore(&id_priv->lock, flags); + return ret; +} +EXPORT_SYMBOL(rdma_set_afonly); + static void cma_bind_port(struct rdma_bind_list *bind_list, struct rdma_id_private *id_priv) { @@ -2187,22 +2218,24 @@ static int cma_check_port(struct rdma_bind_list *bind_list, struct hlist_node *node; addr = (struct sockaddr *) &id_priv->id.route.addr.src_addr; - if (cma_any_addr(addr) && !reuseaddr) - return -EADDRNOTAVAIL; - hlist_for_each_entry(cur_id, node, &bind_list->owners, node) { if (id_priv == cur_id) continue; - if ((cur_id->state == RDMA_CM_LISTEN) || - !reuseaddr || !cur_id->reuseaddr) { - cur_addr = (struct sockaddr *) &cur_id->id.route.addr.src_addr; - if (cma_any_addr(cur_addr)) - return -EADDRNOTAVAIL; + if ((cur_id->state != RDMA_CM_LISTEN) && reuseaddr && + cur_id->reuseaddr) + continue; - if (!cma_addr_cmp(addr, cur_addr)) - return -EADDRINUSE; - } + cur_addr = (struct sockaddr *) &cur_id->id.route.addr.src_addr; + if (id_priv->afonly && cur_id->afonly && + (addr->sa_family != cur_addr->sa_family)) + continue; + + if (cma_any_addr(addr) || cma_any_addr(cur_addr)) + return -EADDRNOTAVAIL; + + if (!cma_addr_cmp(addr, cur_addr)) + return -EADDRINUSE; } return 0; } @@ -2278,7 +2311,7 @@ static int cma_get_port(struct rdma_id_private *id_priv) static int cma_check_linklocal(struct rdma_dev_addr *dev_addr, struct sockaddr *addr) { -#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) +#if IS_ENABLED(CONFIG_IPV6) struct sockaddr_in6 *sin6; if (addr->sa_family != AF_INET6) @@ -2371,6 +2404,14 @@ int rdma_bind_addr(struct rdma_cm_id *id, struct sockaddr *addr) } memcpy(&id->route.addr.src_addr, addr, ip_addr_size(addr)); + if (!(id_priv->options & (1 << CMA_OPTION_AFONLY))) { + if (addr->sa_family == AF_INET) + id_priv->afonly = 1; +#if IS_ENABLED(CONFIG_IPV6) + else if (addr->sa_family == AF_INET6) + id_priv->afonly = init_net.ipv6.sysctl.bindv6only; +#endif + } ret = cma_get_port(id_priv); if (ret) goto err2; diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index fbbfa24..a8905ab 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -94,6 +94,12 @@ struct ib_sa_path_query { struct ib_sa_query sa_query; }; +struct ib_sa_guidinfo_query { + void (*callback)(int, struct ib_sa_guidinfo_rec *, void *); + void *context; + struct ib_sa_query sa_query; +}; + struct ib_sa_mcmember_query { void (*callback)(int, struct ib_sa_mcmember_rec *, void *); void *context; @@ -347,6 +353,34 @@ static const struct ib_field service_rec_table[] = { .size_bits = 2*64 }, }; +#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, \ + .field_name = "sa_guidinfo_rec:" #field + +static const struct ib_field guidinfo_rec_table[] = { + { GUIDINFO_REC_FIELD(lid), + .offset_words = 0, + .offset_bits = 0, + .size_bits = 16 }, + { GUIDINFO_REC_FIELD(block_num), + .offset_words = 0, + .offset_bits = 16, + .size_bits = 8 }, + { GUIDINFO_REC_FIELD(res1), + .offset_words = 0, + .offset_bits = 24, + .size_bits = 8 }, + { GUIDINFO_REC_FIELD(res2), + .offset_words = 1, + .offset_bits = 0, + .size_bits = 32 }, + { GUIDINFO_REC_FIELD(guid_info_list), + .offset_words = 2, + .offset_bits = 0, + .size_bits = 512 }, +}; + static void free_sm_ah(struct kref *kref) { struct ib_sa_sm_ah *sm_ah = container_of(kref, struct ib_sa_sm_ah, ref); @@ -945,6 +979,105 @@ err1: return ret; } +/* Support GuidInfoRecord */ +static void ib_sa_guidinfo_rec_callback(struct ib_sa_query *sa_query, + int status, + struct ib_sa_mad *mad) +{ + struct ib_sa_guidinfo_query *query = + container_of(sa_query, struct ib_sa_guidinfo_query, sa_query); + + if (mad) { + struct ib_sa_guidinfo_rec rec; + + ib_unpack(guidinfo_rec_table, ARRAY_SIZE(guidinfo_rec_table), + mad->data, &rec); + query->callback(status, &rec, query->context); + } else + query->callback(status, NULL, query->context); +} + +static void ib_sa_guidinfo_rec_release(struct ib_sa_query *sa_query) +{ + kfree(container_of(sa_query, struct ib_sa_guidinfo_query, sa_query)); +} + +int ib_sa_guid_info_rec_query(struct ib_sa_client *client, + struct ib_device *device, u8 port_num, + struct ib_sa_guidinfo_rec *rec, + ib_sa_comp_mask comp_mask, u8 method, + int timeout_ms, gfp_t gfp_mask, + void (*callback)(int status, + struct ib_sa_guidinfo_rec *resp, + void *context), + void *context, + struct ib_sa_query **sa_query) +{ + struct ib_sa_guidinfo_query *query; + struct ib_sa_device *sa_dev = ib_get_client_data(device, &sa_client); + struct ib_sa_port *port; + struct ib_mad_agent *agent; + struct ib_sa_mad *mad; + int ret; + + if (!sa_dev) + return -ENODEV; + + if (method != IB_MGMT_METHOD_GET && + method != IB_MGMT_METHOD_SET && + method != IB_SA_METHOD_DELETE) { + return -EINVAL; + } + + port = &sa_dev->port[port_num - sa_dev->start_port]; + agent = port->agent; + + query = kmalloc(sizeof *query, gfp_mask); + if (!query) + return -ENOMEM; + + query->sa_query.port = port; + ret = alloc_mad(&query->sa_query, gfp_mask); + if (ret) + goto err1; + + ib_sa_client_get(client); + query->sa_query.client = client; + query->callback = callback; + query->context = context; + + mad = query->sa_query.mad_buf->mad; + init_mad(mad, agent); + + query->sa_query.callback = callback ? ib_sa_guidinfo_rec_callback : NULL; + query->sa_query.release = ib_sa_guidinfo_rec_release; + + mad->mad_hdr.method = method; + mad->mad_hdr.attr_id = cpu_to_be16(IB_SA_ATTR_GUID_INFO_REC); + mad->sa_hdr.comp_mask = comp_mask; + + ib_pack(guidinfo_rec_table, ARRAY_SIZE(guidinfo_rec_table), rec, + mad->data); + + *sa_query = &query->sa_query; + + ret = send_mad(&query->sa_query, timeout_ms, gfp_mask); + if (ret < 0) + goto err2; + + return ret; + +err2: + *sa_query = NULL; + ib_sa_client_put(query->sa_query.client); + free_mad(&query->sa_query); + +err1: + kfree(query); + return ret; +} +EXPORT_SYMBOL(ib_sa_guid_info_rec_query); + static void send_handler(struct ib_mad_agent *agent, struct ib_mad_send_wc *mad_send_wc) { diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 8002ae6..893cb87 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -909,6 +909,13 @@ static int ucma_set_option_id(struct ucma_context *ctx, int optname, } ret = rdma_set_reuseaddr(ctx->cm_id, *((int *) optval) ? 1 : 0); break; + case RDMA_OPTION_ID_AFONLY: + if (optlen != sizeof(int)) { + ret = -EINVAL; + break; + } + ret = rdma_set_afonly(ctx->cm_id, *((int *) optval) ? 1 : 0); + break; default: ret = -ENOSYS; } |