diff options
Diffstat (limited to 'drivers/infiniband')
130 files changed, 4368 insertions, 2512 deletions
diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index f2a84c6..e9cf51b 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -37,6 +37,7 @@ #include <linux/inetdevice.h> #include <linux/slab.h> #include <linux/workqueue.h> +#include <linux/module.h> #include <net/arp.h> #include <net/neighbour.h> #include <net/route.h> diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index f9ba7d7..9353992 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -302,7 +302,8 @@ static void ib_cache_event(struct ib_event_handler *handler, event->event == IB_EVENT_LID_CHANGE || event->event == IB_EVENT_PKEY_CHANGE || event->event == IB_EVENT_SM_CHANGE || - event->event == IB_EVENT_CLIENT_REREGISTER) { + event->event == IB_EVENT_CLIENT_REREGISTER || + event->event == IB_EVENT_GID_CHANGE) { work = kmalloc(sizeof *work, GFP_ATOMIC); if (work) { INIT_WORK(&work->work, ib_cache_task); diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index fc0f2bd..8b72f39 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -36,6 +36,7 @@ #include <linux/completion.h> #include <linux/dma-mapping.h> #include <linux/device.h> +#include <linux/module.h> #include <linux/err.h> #include <linux/idr.h> #include <linux/interrupt.h> @@ -889,6 +890,8 @@ retest: break; case IB_CM_ESTABLISHED: spin_unlock_irq(&cm_id_priv->lock); + if (cm_id_priv->qp_type == IB_QPT_XRC_TGT) + break; ib_send_cm_dreq(cm_id, NULL, 0); goto retest; case IB_CM_DREQ_SENT: @@ -1008,7 +1011,6 @@ static void cm_format_req(struct cm_req_msg *req_msg, req_msg->service_id = param->service_id; req_msg->local_ca_guid = cm_id_priv->id.device->node_guid; cm_req_set_local_qpn(req_msg, cpu_to_be32(param->qp_num)); - cm_req_set_resp_res(req_msg, param->responder_resources); cm_req_set_init_depth(req_msg, param->initiator_depth); cm_req_set_remote_resp_timeout(req_msg, param->remote_cm_response_timeout); @@ -1017,12 +1019,16 @@ static void cm_format_req(struct cm_req_msg *req_msg, cm_req_set_starting_psn(req_msg, cpu_to_be32(param->starting_psn)); cm_req_set_local_resp_timeout(req_msg, param->local_cm_response_timeout); - cm_req_set_retry_count(req_msg, param->retry_count); req_msg->pkey = param->primary_path->pkey; cm_req_set_path_mtu(req_msg, param->primary_path->mtu); - cm_req_set_rnr_retry_count(req_msg, param->rnr_retry_count); cm_req_set_max_cm_retries(req_msg, param->max_cm_retries); - cm_req_set_srq(req_msg, param->srq); + + if (param->qp_type != IB_QPT_XRC_INI) { + cm_req_set_resp_res(req_msg, param->responder_resources); + cm_req_set_retry_count(req_msg, param->retry_count); + cm_req_set_rnr_retry_count(req_msg, param->rnr_retry_count); + cm_req_set_srq(req_msg, param->srq); + } if (pri_path->hop_limit <= 1) { req_msg->primary_local_lid = pri_path->slid; @@ -1080,7 +1086,8 @@ static int cm_validate_req_param(struct ib_cm_req_param *param) if (!param->primary_path) return -EINVAL; - if (param->qp_type != IB_QPT_RC && param->qp_type != IB_QPT_UC) + if (param->qp_type != IB_QPT_RC && param->qp_type != IB_QPT_UC && + param->qp_type != IB_QPT_XRC_INI) return -EINVAL; if (param->private_data && @@ -1601,18 +1608,24 @@ static void cm_format_rep(struct cm_rep_msg *rep_msg, cm_format_mad_hdr(&rep_msg->hdr, CM_REP_ATTR_ID, cm_id_priv->tid); rep_msg->local_comm_id = cm_id_priv->id.local_id; rep_msg->remote_comm_id = cm_id_priv->id.remote_id; - cm_rep_set_local_qpn(rep_msg, cpu_to_be32(param->qp_num)); cm_rep_set_starting_psn(rep_msg, cpu_to_be32(param->starting_psn)); rep_msg->resp_resources = param->responder_resources; - rep_msg->initiator_depth = param->initiator_depth; cm_rep_set_target_ack_delay(rep_msg, cm_id_priv->av.port->cm_dev->ack_delay); cm_rep_set_failover(rep_msg, param->failover_accepted); - cm_rep_set_flow_ctrl(rep_msg, param->flow_control); cm_rep_set_rnr_retry_count(rep_msg, param->rnr_retry_count); - cm_rep_set_srq(rep_msg, param->srq); rep_msg->local_ca_guid = cm_id_priv->id.device->node_guid; + if (cm_id_priv->qp_type != IB_QPT_XRC_TGT) { + rep_msg->initiator_depth = param->initiator_depth; + cm_rep_set_flow_ctrl(rep_msg, param->flow_control); + cm_rep_set_srq(rep_msg, param->srq); + cm_rep_set_local_qpn(rep_msg, cpu_to_be32(param->qp_num)); + } else { + cm_rep_set_srq(rep_msg, 1); + cm_rep_set_local_eecn(rep_msg, cpu_to_be32(param->qp_num)); + } + if (param->private_data && param->private_data_len) memcpy(rep_msg->private_data, param->private_data, param->private_data_len); @@ -1660,7 +1673,7 @@ int ib_send_cm_rep(struct ib_cm_id *cm_id, cm_id_priv->initiator_depth = param->initiator_depth; cm_id_priv->responder_resources = param->responder_resources; cm_id_priv->rq_psn = cm_rep_get_starting_psn(rep_msg); - cm_id_priv->local_qpn = cm_rep_get_local_qpn(rep_msg); + cm_id_priv->local_qpn = cpu_to_be32(param->qp_num & 0xFFFFFF); out: spin_unlock_irqrestore(&cm_id_priv->lock, flags); return ret; @@ -1731,7 +1744,7 @@ error: spin_unlock_irqrestore(&cm_id_priv->lock, flags); } EXPORT_SYMBOL(ib_send_cm_rtu); -static void cm_format_rep_event(struct cm_work *work) +static void cm_format_rep_event(struct cm_work *work, enum ib_qp_type qp_type) { struct cm_rep_msg *rep_msg; struct ib_cm_rep_event_param *param; @@ -1740,7 +1753,7 @@ static void cm_format_rep_event(struct cm_work *work) param = &work->cm_event.param.rep_rcvd; param->remote_ca_guid = rep_msg->local_ca_guid; param->remote_qkey = be32_to_cpu(rep_msg->local_qkey); - param->remote_qpn = be32_to_cpu(cm_rep_get_local_qpn(rep_msg)); + param->remote_qpn = be32_to_cpu(cm_rep_get_qpn(rep_msg, qp_type)); param->starting_psn = be32_to_cpu(cm_rep_get_starting_psn(rep_msg)); param->responder_resources = rep_msg->initiator_depth; param->initiator_depth = rep_msg->resp_resources; @@ -1808,7 +1821,7 @@ static int cm_rep_handler(struct cm_work *work) return -EINVAL; } - cm_format_rep_event(work); + cm_format_rep_event(work, cm_id_priv->qp_type); spin_lock_irq(&cm_id_priv->lock); switch (cm_id_priv->id.state) { @@ -1823,7 +1836,7 @@ static int cm_rep_handler(struct cm_work *work) cm_id_priv->timewait_info->work.remote_id = rep_msg->local_comm_id; cm_id_priv->timewait_info->remote_ca_guid = rep_msg->local_ca_guid; - cm_id_priv->timewait_info->remote_qpn = cm_rep_get_local_qpn(rep_msg); + cm_id_priv->timewait_info->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type); spin_lock(&cm.lock); /* Check for duplicate REP. */ @@ -1850,7 +1863,7 @@ static int cm_rep_handler(struct cm_work *work) cm_id_priv->id.state = IB_CM_REP_RCVD; cm_id_priv->id.remote_id = rep_msg->local_comm_id; - cm_id_priv->remote_qpn = cm_rep_get_local_qpn(rep_msg); + cm_id_priv->remote_qpn = cm_rep_get_qpn(rep_msg, cm_id_priv->qp_type); cm_id_priv->initiator_depth = rep_msg->resp_resources; cm_id_priv->responder_resources = rep_msg->initiator_depth; cm_id_priv->sq_psn = cm_rep_get_starting_psn(rep_msg); @@ -3492,7 +3505,8 @@ static int cm_init_qp_rtr_attr(struct cm_id_private *cm_id_priv, qp_attr->path_mtu = cm_id_priv->path_mtu; qp_attr->dest_qp_num = be32_to_cpu(cm_id_priv->remote_qpn); qp_attr->rq_psn = be32_to_cpu(cm_id_priv->rq_psn); - if (cm_id_priv->qp_type == IB_QPT_RC) { + if (cm_id_priv->qp_type == IB_QPT_RC || + cm_id_priv->qp_type == IB_QPT_XRC_TGT) { *qp_attr_mask |= IB_QP_MAX_DEST_RD_ATOMIC | IB_QP_MIN_RNR_TIMER; qp_attr->max_dest_rd_atomic = @@ -3537,15 +3551,21 @@ static int cm_init_qp_rts_attr(struct cm_id_private *cm_id_priv, if (cm_id_priv->id.lap_state == IB_CM_LAP_UNINIT) { *qp_attr_mask = IB_QP_STATE | IB_QP_SQ_PSN; qp_attr->sq_psn = be32_to_cpu(cm_id_priv->sq_psn); - if (cm_id_priv->qp_type == IB_QPT_RC) { - *qp_attr_mask |= IB_QP_TIMEOUT | IB_QP_RETRY_CNT | - IB_QP_RNR_RETRY | + switch (cm_id_priv->qp_type) { + case IB_QPT_RC: + case IB_QPT_XRC_INI: + *qp_attr_mask |= IB_QP_RETRY_CNT | IB_QP_RNR_RETRY | IB_QP_MAX_QP_RD_ATOMIC; - qp_attr->timeout = cm_id_priv->av.timeout; 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; + qp_attr->max_rd_atomic = cm_id_priv->initiator_depth; + /* fall through */ + case IB_QPT_XRC_TGT: + *qp_attr_mask |= IB_QP_TIMEOUT; + qp_attr->timeout = cm_id_priv->av.timeout; + break; + default: + break; } if (cm_id_priv->alt_av.ah_attr.dlid) { *qp_attr_mask |= IB_QP_PATH_MIG_STATE; diff --git a/drivers/infiniband/core/cm_msgs.h b/drivers/infiniband/core/cm_msgs.h index 7e63c08..505db2a 100644 --- a/drivers/infiniband/core/cm_msgs.h +++ b/drivers/infiniband/core/cm_msgs.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2004 Intel Corporation. All rights reserved. + * Copyright (c) 2004, 2011 Intel Corporation. All rights reserved. * Copyright (c) 2004 Topspin Corporation. All rights reserved. * Copyright (c) 2004 Voltaire Corporation. All rights reserved. * @@ -86,7 +86,7 @@ struct cm_req_msg { __be16 pkey; /* path MTU:4, RDC exists:1, RNR retry count:3. */ u8 offset50; - /* max CM Retries:4, SRQ:1, rsvd:3 */ + /* max CM Retries:4, SRQ:1, extended transport type:3 */ u8 offset51; __be16 primary_local_lid; @@ -175,6 +175,11 @@ static inline enum ib_qp_type cm_req_get_qp_type(struct cm_req_msg *req_msg) switch(transport_type) { case 0: return IB_QPT_RC; case 1: return IB_QPT_UC; + case 3: + switch (req_msg->offset51 & 0x7) { + case 1: return IB_QPT_XRC_TGT; + default: return 0; + } default: return 0; } } @@ -188,6 +193,12 @@ static inline void cm_req_set_qp_type(struct cm_req_msg *req_msg, req_msg->offset40) & 0xFFFFFFF9) | 0x2); break; + case IB_QPT_XRC_INI: + req_msg->offset40 = cpu_to_be32((be32_to_cpu( + req_msg->offset40) & + 0xFFFFFFF9) | 0x6); + req_msg->offset51 = (req_msg->offset51 & 0xF8) | 1; + break; default: req_msg->offset40 = cpu_to_be32(be32_to_cpu( req_msg->offset40) & @@ -527,6 +538,23 @@ static inline void cm_rep_set_local_qpn(struct cm_rep_msg *rep_msg, __be32 qpn) (be32_to_cpu(rep_msg->offset12) & 0x000000FF)); } +static inline __be32 cm_rep_get_local_eecn(struct cm_rep_msg *rep_msg) +{ + return cpu_to_be32(be32_to_cpu(rep_msg->offset16) >> 8); +} + +static inline void cm_rep_set_local_eecn(struct cm_rep_msg *rep_msg, __be32 eecn) +{ + rep_msg->offset16 = cpu_to_be32((be32_to_cpu(eecn) << 8) | + (be32_to_cpu(rep_msg->offset16) & 0x000000FF)); +} + +static inline __be32 cm_rep_get_qpn(struct cm_rep_msg *rep_msg, enum ib_qp_type qp_type) +{ + return (qp_type == IB_QPT_XRC_INI) ? + cm_rep_get_local_eecn(rep_msg) : cm_rep_get_local_qpn(rep_msg); +} + static inline __be32 cm_rep_get_starting_psn(struct cm_rep_msg *rep_msg) { return cpu_to_be32(be32_to_cpu(rep_msg->offset20) >> 8); diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index b6a33b3..d0d4aa9 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -41,6 +41,7 @@ #include <linux/idr.h> #include <linux/inetdevice.h> #include <linux/slab.h> +#include <linux/module.h> #include <net/tcp.h> #include <net/ipv6.h> @@ -81,6 +82,7 @@ static DEFINE_IDR(sdp_ps); static DEFINE_IDR(tcp_ps); static DEFINE_IDR(udp_ps); static DEFINE_IDR(ipoib_ps); +static DEFINE_IDR(ib_ps); struct cma_device { struct list_head list; @@ -359,6 +361,10 @@ static int cma_acquire_dev(struct rdma_id_private *id_priv) enum rdma_link_layer dev_ll = dev_addr->dev_type == ARPHRD_INFINIBAND ? IB_LINK_LAYER_INFINIBAND : IB_LINK_LAYER_ETHERNET; + if (dev_ll != IB_LINK_LAYER_INFINIBAND && + id_priv->id.ps == RDMA_PS_IPOIB) + return -EINVAL; + mutex_lock(&lock); iboe_addr_get_sgid(dev_addr, &iboe_gid); memcpy(&gid, dev_addr->src_dev_addr + @@ -406,11 +412,6 @@ static int cma_disable_callback(struct rdma_id_private *id_priv, return 0; } -static int cma_has_cm_dev(struct rdma_id_private *id_priv) -{ - return (id_priv->id.device && id_priv->cm_id.ib); -} - struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler, void *context, enum rdma_port_space ps, enum ib_qp_type qp_type) @@ -920,11 +921,11 @@ void rdma_destroy_id(struct rdma_cm_id *id) if (id_priv->cma_dev) { 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)) + if (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)) + if (id_priv->cm_id.iw) iw_destroy_cm_id(id_priv->cm_id.iw); break; default: @@ -1085,12 +1086,12 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id, if (cma_get_net_info(ib_event->private_data, listen_id->ps, &ip_ver, &port, &src, &dst)) - goto err; + return NULL; id = rdma_create_id(listen_id->event_handler, listen_id->context, listen_id->ps, ib_event->param.req_rcvd.qp_type); if (IS_ERR(id)) - goto err; + return NULL; cma_save_net_info(&id->route.addr, &listen_id->route.addr, ip_ver, port, src, dst); @@ -1100,7 +1101,7 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id, rt->path_rec = kmalloc(sizeof *rt->path_rec * rt->num_paths, GFP_KERNEL); if (!rt->path_rec) - goto destroy_id; + goto err; rt->path_rec[0] = *ib_event->param.req_rcvd.primary_path; if (rt->num_paths == 2) @@ -1114,7 +1115,7 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id, ret = rdma_translate_ip((struct sockaddr *) &rt->addr.src_addr, &rt->addr.dev_addr); if (ret) - goto destroy_id; + goto err; } rdma_addr_set_dgid(&rt->addr.dev_addr, &rt->path_rec[0].dgid); @@ -1122,9 +1123,8 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id, id_priv->state = RDMA_CM_CONNECT; return id_priv; -destroy_id: - rdma_destroy_id(id); err: + rdma_destroy_id(id); return NULL; } @@ -1181,6 +1181,15 @@ static void cma_set_req_event_data(struct rdma_cm_event *event, event->param.conn.qp_num = req_data->remote_qpn; } +static int cma_check_req_qp_type(struct rdma_cm_id *id, struct ib_cm_event *ib_event) +{ + return (((ib_event->event == IB_CM_REQ_RECEIVED) || + (ib_event->param.req_rcvd.qp_type == id->qp_type)) || + ((ib_event->event == IB_CM_SIDR_REQ_RECEIVED) && + (id->qp_type == IB_QPT_UD)) || + (!id->qp_type)); +} + static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event) { struct rdma_id_private *listen_id, *conn_id; @@ -1188,13 +1197,16 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event) int offset, ret; listen_id = cm_id->context; + if (!cma_check_req_qp_type(&listen_id->id, ib_event)) + return -EINVAL; + if (cma_disable_callback(listen_id, RDMA_CM_LISTEN)) return -ECONNABORTED; memset(&event, 0, sizeof event); offset = cma_user_data_offset(listen_id->id.ps); event.event = RDMA_CM_EVENT_CONNECT_REQUEST; - if (listen_id->id.qp_type == IB_QPT_UD) { + if (ib_event->event == IB_CM_SIDR_REQ_RECEIVED) { conn_id = cma_new_udp_id(&listen_id->id, ib_event); event.param.ud.private_data = ib_event->private_data + offset; event.param.ud.private_data_len = @@ -1330,6 +1342,8 @@ static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event) switch (iw_event->status) { case 0: event.event = RDMA_CM_EVENT_ESTABLISHED; + event.param.conn.initiator_depth = iw_event->ird; + event.param.conn.responder_resources = iw_event->ord; break; case -ECONNRESET: case -ECONNREFUSED: @@ -1345,6 +1359,8 @@ static int cma_iw_handler(struct iw_cm_id *iw_id, struct iw_cm_event *iw_event) break; case IW_CM_EVENT_ESTABLISHED: event.event = RDMA_CM_EVENT_ESTABLISHED; + event.param.conn.initiator_depth = iw_event->ird; + event.param.conn.responder_resources = iw_event->ord; break; default: BUG_ON(1); @@ -1435,8 +1451,8 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id, event.event = RDMA_CM_EVENT_CONNECT_REQUEST; event.param.conn.private_data = iw_event->private_data; event.param.conn.private_data_len = iw_event->private_data_len; - event.param.conn.initiator_depth = attr.max_qp_init_rd_atom; - event.param.conn.responder_resources = attr.max_qp_rd_atom; + event.param.conn.initiator_depth = iw_event->ird; + event.param.conn.responder_resources = iw_event->ord; /* * Protect against the user destroying conn_id from another thread @@ -1468,13 +1484,15 @@ static int cma_ib_listen(struct rdma_id_private *id_priv) { struct ib_cm_compare_data compare_data; struct sockaddr *addr; + struct ib_cm_id *id; __be64 svc_id; int ret; - id_priv->cm_id.ib = ib_create_cm_id(id_priv->id.device, cma_req_handler, - id_priv); - if (IS_ERR(id_priv->cm_id.ib)) - return PTR_ERR(id_priv->cm_id.ib); + id = ib_create_cm_id(id_priv->id.device, cma_req_handler, id_priv); + if (IS_ERR(id)) + return PTR_ERR(id); + + id_priv->cm_id.ib = id; addr = (struct sockaddr *) &id_priv->id.route.addr.src_addr; svc_id = cma_get_service_id(id_priv->id.ps, addr); @@ -1497,12 +1515,15 @@ static int cma_iw_listen(struct rdma_id_private *id_priv, int backlog) { int ret; struct sockaddr_in *sin; + struct iw_cm_id *id; + + id = iw_create_cm_id(id_priv->id.device, + iw_conn_req_handler, + id_priv); + if (IS_ERR(id)) + return PTR_ERR(id); - 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); + id_priv->cm_id.iw = id; sin = (struct sockaddr_in *) &id_priv->id.route.addr.src_addr; id_priv->cm_id.iw->local_addr = *sin; @@ -2231,6 +2252,9 @@ static int cma_get_port(struct rdma_id_private *id_priv) case RDMA_PS_IPOIB: ps = &ipoib_ps; break; + case RDMA_PS_IB: + ps = &ib_ps; + break; default: return -EPROTONOSUPPORT; } @@ -2484,10 +2508,14 @@ static int cma_resolve_ib_udp(struct rdma_id_private *id_priv, { struct ib_cm_sidr_req_param req; struct rdma_route *route; + struct ib_cm_id *id; int ret; req.private_data_len = sizeof(struct cma_hdr) + conn_param->private_data_len; + if (req.private_data_len < conn_param->private_data_len) + return -EINVAL; + req.private_data = kzalloc(req.private_data_len, GFP_ATOMIC); if (!req.private_data) return -ENOMEM; @@ -2501,12 +2529,13 @@ static int cma_resolve_ib_udp(struct rdma_id_private *id_priv, if (ret) goto out; - id_priv->cm_id.ib = ib_create_cm_id(id_priv->id.device, - cma_sidr_rep_handler, id_priv); - if (IS_ERR(id_priv->cm_id.ib)) { - ret = PTR_ERR(id_priv->cm_id.ib); + id = ib_create_cm_id(id_priv->id.device, cma_sidr_rep_handler, + id_priv); + if (IS_ERR(id)) { + ret = PTR_ERR(id); goto out; } + id_priv->cm_id.ib = id; req.path = route->path_rec; req.service_id = cma_get_service_id(id_priv->id.ps, @@ -2530,11 +2559,15 @@ static int cma_connect_ib(struct rdma_id_private *id_priv, struct ib_cm_req_param req; struct rdma_route *route; void *private_data; + struct ib_cm_id *id; int offset, ret; memset(&req, 0, sizeof req); offset = cma_user_data_offset(id_priv->id.ps); req.private_data_len = offset + conn_param->private_data_len; + if (req.private_data_len < conn_param->private_data_len) + return -EINVAL; + private_data = kzalloc(req.private_data_len, GFP_ATOMIC); if (!private_data) return -ENOMEM; @@ -2543,12 +2576,12 @@ static int cma_connect_ib(struct rdma_id_private *id_priv, memcpy(private_data + offset, conn_param->private_data, conn_param->private_data_len); - id_priv->cm_id.ib = ib_create_cm_id(id_priv->id.device, cma_ib_handler, - id_priv); - if (IS_ERR(id_priv->cm_id.ib)) { - ret = PTR_ERR(id_priv->cm_id.ib); + id = ib_create_cm_id(id_priv->id.device, cma_ib_handler, id_priv); + if (IS_ERR(id)) { + ret = PTR_ERR(id); goto out; } + id_priv->cm_id.ib = id; route = &id_priv->id.route; ret = cma_format_hdr(private_data, id_priv->id.ps, route); @@ -2563,7 +2596,7 @@ static int cma_connect_ib(struct rdma_id_private *id_priv, req.service_id = cma_get_service_id(id_priv->id.ps, (struct sockaddr *) &route->addr.dst_addr); req.qp_num = id_priv->qp_num; - req.qp_type = IB_QPT_RC; + req.qp_type = id_priv->id.qp_type; req.starting_psn = id_priv->seq_num; req.responder_resources = conn_param->responder_resources; req.initiator_depth = conn_param->initiator_depth; @@ -2577,8 +2610,8 @@ static int cma_connect_ib(struct rdma_id_private *id_priv, ret = ib_send_cm_req(id_priv->cm_id.ib, &req); out: - if (ret && !IS_ERR(id_priv->cm_id.ib)) { - ib_destroy_cm_id(id_priv->cm_id.ib); + if (ret && !IS_ERR(id)) { + ib_destroy_cm_id(id); id_priv->cm_id.ib = NULL; } @@ -2595,10 +2628,8 @@ static int cma_connect_iw(struct rdma_id_private *id_priv, 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; - } + if (IS_ERR(cm_id)) + return PTR_ERR(cm_id); id_priv->cm_id.iw = cm_id; @@ -2612,17 +2643,19 @@ static int cma_connect_iw(struct rdma_id_private *id_priv, if (ret) goto out; - 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) + if (conn_param) { + 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; + iw_param.qpn = id_priv->id.qp ? id_priv->qp_num : conn_param->qp_num; + } else { + memset(&iw_param, 0, sizeof iw_param); iw_param.qpn = id_priv->qp_num; - else - iw_param.qpn = conn_param->qp_num; + } ret = iw_cm_connect(cm_id, &iw_param); out: - if (ret && !IS_ERR(cm_id)) { + if (ret) { iw_destroy_cm_id(cm_id); id_priv->cm_id.iw = NULL; } @@ -2761,14 +2794,20 @@ int rdma_accept(struct rdma_cm_id *id, struct rdma_conn_param *conn_param) switch (rdma_node_get_transport(id->device->node_type)) { case RDMA_TRANSPORT_IB: - if (id->qp_type == IB_QPT_UD) - ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS, - conn_param->private_data, - conn_param->private_data_len); - else if (conn_param) - ret = cma_accept_ib(id_priv, conn_param); - else - ret = cma_rep_recv(id_priv); + if (id->qp_type == IB_QPT_UD) { + if (conn_param) + ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS, + conn_param->private_data, + conn_param->private_data_len); + else + ret = cma_send_sidr_rep(id_priv, IB_SIDR_SUCCESS, + NULL, 0); + } else { + 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); @@ -2795,7 +2834,7 @@ int rdma_notify(struct rdma_cm_id *id, enum ib_event_type event) int ret; id_priv = container_of(id, struct rdma_id_private, id); - if (!cma_has_cm_dev(id_priv)) + if (!id_priv->cm_id.ib) return -EINVAL; switch (id->device->node_type) { @@ -2817,7 +2856,7 @@ int rdma_reject(struct rdma_cm_id *id, const void *private_data, int ret; id_priv = container_of(id, struct rdma_id_private, id); - if (!cma_has_cm_dev(id_priv)) + if (!id_priv->cm_id.ib) return -EINVAL; switch (rdma_node_get_transport(id->device->node_type)) { @@ -2848,7 +2887,7 @@ int rdma_disconnect(struct rdma_cm_id *id) int ret; id_priv = container_of(id, struct rdma_id_private, id); - if (!cma_has_cm_dev(id_priv)) + if (!id_priv->cm_id.ib) return -EINVAL; switch (rdma_node_get_transport(id->device->node_type)) { @@ -3456,6 +3495,7 @@ static void __exit cma_cleanup(void) idr_destroy(&tcp_ps); idr_destroy(&udp_ps); idr_destroy(&ipoib_ps); + idr_destroy(&ib_ps); } module_init(cma_init); diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c index 4007f72..e711de4 100644 --- a/drivers/infiniband/core/device.c +++ b/drivers/infiniband/core/device.c @@ -627,6 +627,9 @@ int ib_modify_device(struct ib_device *device, int device_modify_mask, struct ib_device_modify *device_modify) { + if (!device->modify_device) + return -ENOSYS; + return device->modify_device(device, device_modify_mask, device_modify); } @@ -647,6 +650,9 @@ int ib_modify_port(struct ib_device *device, u8 port_num, int port_modify_mask, struct ib_port_modify *port_modify) { + if (!device->modify_port) + return -ENOSYS; + if (port_num < start_port(device) || port_num > end_port(device)) return -EINVAL; diff --git a/drivers/infiniband/core/fmr_pool.c b/drivers/infiniband/core/fmr_pool.c index 4507043..176c8f9 100644 --- a/drivers/infiniband/core/fmr_pool.c +++ b/drivers/infiniband/core/fmr_pool.c @@ -33,6 +33,7 @@ #include <linux/errno.h> #include <linux/spinlock.h> +#include <linux/export.h> #include <linux/slab.h> #include <linux/jhash.h> #include <linux/kthread.h> diff --git a/drivers/infiniband/core/iwcm.c b/drivers/infiniband/core/iwcm.c index a9c0423..2882556 100644 --- a/drivers/infiniband/core/iwcm.c +++ b/drivers/infiniband/core/iwcm.c @@ -45,6 +45,8 @@ #include <linux/workqueue.h> #include <linux/completion.h> #include <linux/slab.h> +#include <linux/module.h> +#include <linux/sysctl.h> #include <rdma/iw_cm.h> #include <rdma/ib_addr.h> @@ -64,6 +66,20 @@ struct iwcm_work { struct list_head free_list; }; +static unsigned int default_backlog = 256; + +static struct ctl_table_header *iwcm_ctl_table_hdr; +static struct ctl_table iwcm_ctl_table[] = { + { + .procname = "default_backlog", + .data = &default_backlog, + .maxlen = sizeof(default_backlog), + .mode = 0644, + .proc_handler = proc_dointvec, + }, + { } +}; + /* * The following services provide a mechanism for pre-allocating iwcm_work * elements. The design pre-allocates them based on the cm_id type: @@ -418,6 +434,9 @@ int iw_cm_listen(struct iw_cm_id *cm_id, int backlog) cm_id_priv = container_of(cm_id, struct iwcm_id_private, id); + if (!backlog) + backlog = default_backlog; + ret = alloc_work_entries(cm_id_priv, backlog); if (ret) return ret; @@ -1013,17 +1032,33 @@ int iw_cm_init_qp_attr(struct iw_cm_id *cm_id, } EXPORT_SYMBOL(iw_cm_init_qp_attr); +static struct ctl_path iwcm_ctl_path[] = { + { .procname = "net" }, + { .procname = "iw_cm" }, + { } +}; + static int __init iw_cm_init(void) { iwcm_wq = create_singlethread_workqueue("iw_cm_wq"); if (!iwcm_wq) return -ENOMEM; + iwcm_ctl_table_hdr = register_net_sysctl_table(&init_net, + iwcm_ctl_path, + iwcm_ctl_table); + if (!iwcm_ctl_table_hdr) { + pr_err("iw_cm: couldn't register sysctl paths\n"); + destroy_workqueue(iwcm_wq); + return -ENOMEM; + } + return 0; } static void __exit iw_cm_cleanup(void) { + unregister_net_sysctl_table(iwcm_ctl_table_hdr); destroy_workqueue(iwcm_wq); } diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c index b4d8672..2fe428b 100644 --- a/drivers/infiniband/core/mad.c +++ b/drivers/infiniband/core/mad.c @@ -35,6 +35,7 @@ */ #include <linux/dma-mapping.h> #include <linux/slab.h> +#include <linux/module.h> #include <rdma/ib_cache.h> #include "mad_priv.h" @@ -1596,6 +1597,9 @@ find_mad_agent(struct ib_mad_port_private *port_priv, mad->mad_hdr.class_version].class; if (!class) goto out; + if (convert_mgmt_class(mad->mad_hdr.mgmt_class) >= + IB_MGMT_MAX_METHODS) + goto out; method = class->method_table[convert_mgmt_class( mad->mad_hdr.mgmt_class)]; if (method) diff --git a/drivers/infiniband/core/multicast.c b/drivers/infiniband/core/multicast.c index 68b4162..d2360a8 100644 --- a/drivers/infiniband/core/multicast.c +++ b/drivers/infiniband/core/multicast.c @@ -34,6 +34,7 @@ #include <linux/dma-mapping.h> #include <linux/err.h> #include <linux/interrupt.h> +#include <linux/export.h> #include <linux/slab.h> #include <linux/bitops.h> #include <linux/random.h> diff --git a/drivers/infiniband/core/netlink.c b/drivers/infiniband/core/netlink.c index 9227f4a..d1c8196 100644 --- a/drivers/infiniband/core/netlink.c +++ b/drivers/infiniband/core/netlink.c @@ -32,6 +32,7 @@ #define pr_fmt(fmt) "%s:%s: " fmt, KBUILD_MODNAME, __func__ +#include <linux/export.h> #include <net/netlink.h> #include <net/net_namespace.h> #include <net/sock.h> diff --git a/drivers/infiniband/core/packer.c b/drivers/infiniband/core/packer.c index 019bd4b..1b65986 100644 --- a/drivers/infiniband/core/packer.c +++ b/drivers/infiniband/core/packer.c @@ -31,6 +31,7 @@ * SOFTWARE. */ +#include <linux/export.h> #include <linux/string.h> #include <rdma/ib_pack.h> diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c index 9ab5df7..c61bca3 100644 --- a/drivers/infiniband/core/sysfs.c +++ b/drivers/infiniband/core/sysfs.c @@ -35,6 +35,7 @@ #include "core_priv.h" #include <linux/slab.h> +#include <linux/stat.h> #include <linux/string.h> #include <rdma/ib_mad.h> @@ -185,17 +186,35 @@ static ssize_t rate_show(struct ib_port *p, struct port_attribute *unused, if (ret) return ret; + rate = (25 * attr.active_speed) / 10; + switch (attr.active_speed) { - case 2: speed = " DDR"; break; - case 4: speed = " QDR"; break; + case 2: + speed = " DDR"; + break; + case 4: + speed = " QDR"; + break; + case 8: + speed = " FDR10"; + rate = 10; + break; + case 16: + speed = " FDR"; + rate = 14; + break; + case 32: + speed = " EDR"; + rate = 25; + break; } - rate = 25 * ib_width_enum_to_int(attr.active_width) * attr.active_speed; + rate *= ib_width_enum_to_int(attr.active_width); if (rate < 0) return -EINVAL; return sprintf(buf, "%d%s Gb/sec (%dX%s)\n", - rate / 10, rate % 10 ? ".5" : "", + rate, (attr.active_speed == 1) ? ".5" : "", ib_width_enum_to_int(attr.active_width), speed); } diff --git a/drivers/infiniband/core/ucm.c b/drivers/infiniband/core/ucm.c index 08f948d..b8a0b4a 100644 --- a/drivers/infiniband/core/ucm.c +++ b/drivers/infiniband/core/ucm.c @@ -1122,7 +1122,7 @@ static ssize_t ib_ucm_write(struct file *filp, const char __user *buf, if (copy_from_user(&hdr, buf, sizeof(hdr))) return -EFAULT; - if (hdr.cmd < 0 || hdr.cmd >= ARRAY_SIZE(ucm_cmd_table)) + if (hdr.cmd >= ARRAY_SIZE(ucm_cmd_table)) return -EINVAL; if (hdr.in + sizeof(hdr) > len) diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 71be5ee..b37b0c0 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -41,6 +41,7 @@ #include <linux/miscdevice.h> #include <linux/slab.h> #include <linux/sysctl.h> +#include <linux/module.h> #include <rdma/rdma_user_cm.h> #include <rdma/ib_marshall.h> @@ -276,7 +277,7 @@ static int ucma_event_handler(struct rdma_cm_id *cm_id, ucma_set_event_context(ctx, event, uevent); uevent->resp.event = event->event; uevent->resp.status = event->status; - if (cm_id->ps == RDMA_PS_UDP || cm_id->ps == RDMA_PS_IPOIB) + if (cm_id->qp_type == IB_QPT_UD) ucma_copy_ud_event(&uevent->resp.param.ud, &event->param.ud); else ucma_copy_conn_event(&uevent->resp.param.conn, @@ -377,6 +378,9 @@ static int ucma_get_qp_type(struct rdma_ucm_create_id *cmd, enum ib_qp_type *qp_ case RDMA_PS_IPOIB: *qp_type = IB_QPT_UD; return 0; + case RDMA_PS_IB: + *qp_type = cmd->qp_type; + return 0; default: return -EINVAL; } @@ -1270,7 +1274,7 @@ static ssize_t ucma_write(struct file *filp, const char __user *buf, if (copy_from_user(&hdr, buf, sizeof(hdr))) return -EFAULT; - if (hdr.cmd < 0 || hdr.cmd >= ARRAY_SIZE(ucma_cmd_table)) + if (hdr.cmd >= ARRAY_SIZE(ucma_cmd_table)) return -EINVAL; if (hdr.in + sizeof(hdr) > len) diff --git a/drivers/infiniband/core/ud_header.c b/drivers/infiniband/core/ud_header.c index 9b737ff..72feee6 100644 --- a/drivers/infiniband/core/ud_header.c +++ b/drivers/infiniband/core/ud_header.c @@ -33,6 +33,7 @@ #include <linux/errno.h> #include <linux/string.h> +#include <linux/export.h> #include <linux/if_ether.h> #include <rdma/ib_pack.h> diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c index b645e55..c1fef27 100644 --- a/drivers/infiniband/core/umem.c +++ b/drivers/infiniband/core/umem.c @@ -35,6 +35,7 @@ #include <linux/mm.h> #include <linux/dma-mapping.h> #include <linux/sched.h> +#include <linux/export.h> #include <linux/hugetlb.h> #include <linux/dma-attrs.h> #include <linux/slab.h> @@ -93,6 +94,17 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr, if (dmasync) dma_set_attr(DMA_ATTR_WRITE_BARRIER, &attrs); + if (!size) + return ERR_PTR(-EINVAL); + + /* + * If the combination of the addr and size requested for this memory + * region causes an integer overflow, return error. + */ + if (((addr + size) < addr) || + PAGE_ALIGN(addr + size) < (addr + size)) + return ERR_PTR(-EINVAL); + if (!can_do_mlock()) return ERR_PTR(-EPERM); @@ -136,7 +148,7 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr, down_write(¤t->mm->mmap_sem); - locked = npages + current->mm->locked_vm; + locked = npages + current->mm->pinned_vm; lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT; if ((locked > lock_limit) && !capable(CAP_IPC_LOCK)) { @@ -206,7 +218,7 @@ out: __ib_umem_release(context->device, umem, 0); kfree(umem); } else - current->mm->locked_vm = locked; + current->mm->pinned_vm = locked; up_write(¤t->mm->mmap_sem); if (vma_list) @@ -222,7 +234,7 @@ static void ib_umem_account(struct work_struct *work) struct ib_umem *umem = container_of(work, struct ib_umem, work); down_write(&umem->mm->mmap_sem); - umem->mm->locked_vm -= umem->diff; + umem->mm->pinned_vm -= umem->diff; up_write(&umem->mm->mmap_sem); mmput(umem->mm); kfree(umem); @@ -268,7 +280,7 @@ void ib_umem_release(struct ib_umem *umem) } else down_write(&mm->mmap_sem); - current->mm->locked_vm -= diff; + current->mm->pinned_vm -= diff; up_write(&mm->mmap_sem); mmput(mm); kfree(umem); diff --git a/drivers/infiniband/core/user_mad.c b/drivers/infiniband/core/user_mad.c index 8d261b6..c685881 100644 --- a/drivers/infiniband/core/user_mad.c +++ b/drivers/infiniband/core/user_mad.c @@ -458,8 +458,7 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf, goto err; } - if (packet->mad.hdr.id < 0 || - packet->mad.hdr.id >= IB_UMAD_MAX_AGENTS) { + if (packet->mad.hdr.id >= IB_UMAD_MAX_AGENTS) { ret = -EINVAL; goto err; } @@ -703,7 +702,7 @@ static int ib_umad_unreg_agent(struct ib_umad_file *file, u32 __user *arg) mutex_lock(&file->port->file_mutex); mutex_lock(&file->mutex); - if (id < 0 || id >= IB_UMAD_MAX_AGENTS || !__get_agent(file, id)) { + if (id >= IB_UMAD_MAX_AGENTS || !__get_agent(file, id)) { ret = -EINVAL; goto out; } @@ -781,27 +780,19 @@ static int ib_umad_open(struct inode *inode, struct file *filp) { struct ib_umad_port *port; struct ib_umad_file *file; - int ret; + int ret = -ENXIO; port = container_of(inode->i_cdev, struct ib_umad_port, cdev); - if (port) - kref_get(&port->umad_dev->ref); - else - return -ENXIO; mutex_lock(&port->file_mutex); - if (!port->ib_dev) { - ret = -ENXIO; + if (!port->ib_dev) goto out; - } + ret = -ENOMEM; file = kzalloc(sizeof *file, GFP_KERNEL); - if (!file) { - kref_put(&port->umad_dev->ref, ib_umad_release_dev); - ret = -ENOMEM; + if (!file) goto out; - } mutex_init(&file->mutex); spin_lock_init(&file->send_lock); @@ -815,6 +806,13 @@ static int ib_umad_open(struct inode *inode, struct file *filp) list_add_tail(&file->port_list, &port->file_list); ret = nonseekable_open(inode, filp); + if (ret) { + list_del(&file->port_list); + kfree(file); + goto out; + } + + kref_get(&port->umad_dev->ref); out: mutex_unlock(&port->file_mutex); @@ -881,10 +879,6 @@ static int ib_umad_sm_open(struct inode *inode, struct file *filp) int ret; port = container_of(inode->i_cdev, struct ib_umad_port, sm_cdev); - if (port) - kref_get(&port->umad_dev->ref); - else - return -ENXIO; if (filp->f_flags & O_NONBLOCK) { if (down_trylock(&port->sm_sem)) { @@ -899,17 +893,27 @@ static int ib_umad_sm_open(struct inode *inode, struct file *filp) } ret = ib_modify_port(port->ib_dev, port->port_num, 0, &props); - if (ret) { - up(&port->sm_sem); - goto fail; - } + if (ret) + goto err_up_sem; filp->private_data = port; - return nonseekable_open(inode, filp); + ret = nonseekable_open(inode, filp); + if (ret) + goto err_clr_sm_cap; + + kref_get(&port->umad_dev->ref); + + return 0; + +err_clr_sm_cap: + swap(props.set_port_cap_mask, props.clr_port_cap_mask); + ib_modify_port(port->ib_dev, port->port_num, 0, &props); + +err_up_sem: + up(&port->sm_sem); fail: - kref_put(&port->umad_dev->ref, ib_umad_release_dev); return ret; } diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h index a078e56..228af18 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -69,13 +69,16 @@ */ struct ib_uverbs_device { - struct kref ref; + atomic_t refcount; int num_comp_vectors; struct completion comp; struct device *dev; struct ib_device *ib_dev; int devnum; struct cdev cdev; + struct rb_root xrcd_tree; + struct mutex xrcd_tree_mutex; + struct kobject kobj; }; struct ib_uverbs_event_file { @@ -120,6 +123,16 @@ struct ib_uevent_object { u32 events_reported; }; +struct ib_uxrcd_object { + struct ib_uobject uobject; + atomic_t refcnt; +}; + +struct ib_usrq_object { + struct ib_uevent_object uevent; + struct ib_uxrcd_object *uxrcd; +}; + struct ib_uqp_object { struct ib_uevent_object uevent; struct list_head mcast_list; @@ -142,6 +155,7 @@ extern struct idr ib_uverbs_ah_idr; extern struct idr ib_uverbs_cq_idr; extern struct idr ib_uverbs_qp_idr; extern struct idr ib_uverbs_srq_idr; +extern struct idr ib_uverbs_xrcd_idr; void idr_remove_uobj(struct idr *idp, struct ib_uobject *uobj); @@ -161,6 +175,7 @@ void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr); void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr); void ib_uverbs_event_handler(struct ib_event_handler *handler, struct ib_event *event); +void ib_uverbs_dealloc_xrcd(struct ib_uverbs_device *dev, struct ib_xrcd *xrcd); #define IB_UVERBS_DECLARE_CMD(name) \ ssize_t ib_uverbs_##name(struct ib_uverbs_file *file, \ @@ -181,6 +196,7 @@ IB_UVERBS_DECLARE_CMD(poll_cq); IB_UVERBS_DECLARE_CMD(req_notify_cq); IB_UVERBS_DECLARE_CMD(destroy_cq); IB_UVERBS_DECLARE_CMD(create_qp); +IB_UVERBS_DECLARE_CMD(open_qp); IB_UVERBS_DECLARE_CMD(query_qp); IB_UVERBS_DECLARE_CMD(modify_qp); IB_UVERBS_DECLARE_CMD(destroy_qp); @@ -195,5 +211,8 @@ IB_UVERBS_DECLARE_CMD(create_srq); IB_UVERBS_DECLARE_CMD(modify_srq); IB_UVERBS_DECLARE_CMD(query_srq); IB_UVERBS_DECLARE_CMD(destroy_srq); +IB_UVERBS_DECLARE_CMD(create_xsrq); +IB_UVERBS_DECLARE_CMD(open_xrcd); +IB_UVERBS_DECLARE_CMD(close_xrcd); #endif /* UVERBS_H */ diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index c426992..3be21aa 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -47,6 +47,7 @@ static struct lock_class_key cq_lock_key; static struct lock_class_key qp_lock_key; static struct lock_class_key ah_lock_key; static struct lock_class_key srq_lock_key; +static struct lock_class_key xrcd_lock_key; #define INIT_UDATA(udata, ibuf, obuf, ilen, olen) \ do { \ @@ -240,11 +241,24 @@ static struct ib_qp *idr_read_qp(int qp_handle, struct ib_ucontext *context) return idr_read_obj(&ib_uverbs_qp_idr, qp_handle, context, 0); } +static struct ib_qp *idr_write_qp(int qp_handle, struct ib_ucontext *context) +{ + struct ib_uobject *uobj; + + uobj = idr_write_uobj(&ib_uverbs_qp_idr, qp_handle, context); + return uobj ? uobj->object : NULL; +} + static void put_qp_read(struct ib_qp *qp) { put_uobj_read(qp->uobject); } +static void put_qp_write(struct ib_qp *qp) +{ + put_uobj_write(qp->uobject); +} + static struct ib_srq *idr_read_srq(int srq_handle, struct ib_ucontext *context) { return idr_read_obj(&ib_uverbs_srq_idr, srq_handle, context, 0); @@ -255,6 +269,18 @@ static void put_srq_read(struct ib_srq *srq) put_uobj_read(srq->uobject); } +static struct ib_xrcd *idr_read_xrcd(int xrcd_handle, struct ib_ucontext *context, + struct ib_uobject **uobj) +{ + *uobj = idr_read_uobj(&ib_uverbs_xrcd_idr, xrcd_handle, context, 0); + return *uobj ? (*uobj)->object : NULL; +} + +static void put_xrcd_read(struct ib_uobject *uobj) +{ + put_uobj_read(uobj); +} + ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file, const char __user *buf, int in_len, int out_len) @@ -298,6 +324,7 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file, INIT_LIST_HEAD(&ucontext->qp_list); INIT_LIST_HEAD(&ucontext->srq_list); INIT_LIST_HEAD(&ucontext->ah_list); + INIT_LIST_HEAD(&ucontext->xrcd_list); ucontext->closing = 0; resp.num_comp_vectors = file->device->num_comp_vectors; @@ -579,6 +606,310 @@ ssize_t ib_uverbs_dealloc_pd(struct ib_uverbs_file *file, return in_len; } +struct xrcd_table_entry { + struct rb_node node; + struct ib_xrcd *xrcd; + struct inode *inode; +}; + +static int xrcd_table_insert(struct ib_uverbs_device *dev, + struct inode *inode, + struct ib_xrcd *xrcd) +{ + struct xrcd_table_entry *entry, *scan; + struct rb_node **p = &dev->xrcd_tree.rb_node; + struct rb_node *parent = NULL; + + entry = kmalloc(sizeof *entry, GFP_KERNEL); + if (!entry) + return -ENOMEM; + + entry->xrcd = xrcd; + entry->inode = inode; + + while (*p) { + parent = *p; + scan = rb_entry(parent, struct xrcd_table_entry, node); + + if (inode < scan->inode) { + p = &(*p)->rb_left; + } else if (inode > scan->inode) { + p = &(*p)->rb_right; + } else { + kfree(entry); + return -EEXIST; + } + } + + rb_link_node(&entry->node, parent, p); + rb_insert_color(&entry->node, &dev->xrcd_tree); + igrab(inode); + return 0; +} + +static struct xrcd_table_entry *xrcd_table_search(struct ib_uverbs_device *dev, + struct inode *inode) +{ + struct xrcd_table_entry *entry; + struct rb_node *p = dev->xrcd_tree.rb_node; + + while (p) { + entry = rb_entry(p, struct xrcd_table_entry, node); + + if (inode < entry->inode) + p = p->rb_left; + else if (inode > entry->inode) + p = p->rb_right; + else + return entry; + } + + return NULL; +} + +static struct ib_xrcd *find_xrcd(struct ib_uverbs_device *dev, struct inode *inode) +{ + struct xrcd_table_entry *entry; + + entry = xrcd_table_search(dev, inode); + if (!entry) + return NULL; + + return entry->xrcd; +} + +static void xrcd_table_delete(struct ib_uverbs_device *dev, + struct inode *inode) +{ + struct xrcd_table_entry *entry; + + entry = xrcd_table_search(dev, inode); + if (entry) { + iput(inode); + rb_erase(&entry->node, &dev->xrcd_tree); + kfree(entry); + } +} + +ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, + const char __user *buf, int in_len, + int out_len) +{ + struct ib_uverbs_open_xrcd cmd; + struct ib_uverbs_open_xrcd_resp resp; + struct ib_udata udata; + struct ib_uxrcd_object *obj; + struct ib_xrcd *xrcd = NULL; + struct file *f = NULL; + struct inode *inode = NULL; + int ret = 0; + int new_xrcd = 0; + + if (out_len < sizeof resp) + return -ENOSPC; + + if (copy_from_user(&cmd, buf, sizeof cmd)) + return -EFAULT; + + INIT_UDATA(&udata, buf + sizeof cmd, + (unsigned long) cmd.response + sizeof resp, + in_len - sizeof cmd, out_len - sizeof resp); + + mutex_lock(&file->device->xrcd_tree_mutex); + + if (cmd.fd != -1) { + /* search for file descriptor */ + f = fget(cmd.fd); + if (!f) { + ret = -EBADF; + goto err_tree_mutex_unlock; + } + + inode = f->f_dentry->d_inode; + if (!inode) { + ret = -EBADF; + goto err_tree_mutex_unlock; + } + + xrcd = find_xrcd(file->device, inode); + if (!xrcd && !(cmd.oflags & O_CREAT)) { + /* no file descriptor. Need CREATE flag */ + ret = -EAGAIN; + goto err_tree_mutex_unlock; + } + + if (xrcd && cmd.oflags & O_EXCL) { + ret = -EINVAL; + goto err_tree_mutex_unlock; + } + } + + obj = kmalloc(sizeof *obj, GFP_KERNEL); + if (!obj) { + ret = -ENOMEM; + goto err_tree_mutex_unlock; + } + + init_uobj(&obj->uobject, 0, file->ucontext, &xrcd_lock_key); + + down_write(&obj->uobject.mutex); + + if (!xrcd) { + xrcd = file->device->ib_dev->alloc_xrcd(file->device->ib_dev, + file->ucontext, &udata); + if (IS_ERR(xrcd)) { + ret = PTR_ERR(xrcd); + goto err; + } + + xrcd->inode = inode; + xrcd->device = file->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; + ret = idr_add_uobj(&ib_uverbs_xrcd_idr, &obj->uobject); + if (ret) + goto err_idr; + + memset(&resp, 0, sizeof resp); + resp.xrcd_handle = obj->uobject.id; + + if (inode) { + if (new_xrcd) { + /* create new inode/xrcd table entry */ + ret = xrcd_table_insert(file->device, inode, xrcd); + if (ret) + goto err_insert_xrcd; + } + atomic_inc(&xrcd->usecnt); + } + + if (copy_to_user((void __user *) (unsigned long) cmd.response, + &resp, sizeof resp)) { + ret = -EFAULT; + goto err_copy; + } + + if (f) + fput(f); + + mutex_lock(&file->mutex); + list_add_tail(&obj->uobject.list, &file->ucontext->xrcd_list); + mutex_unlock(&file->mutex); + + obj->uobject.live = 1; + up_write(&obj->uobject.mutex); + + mutex_unlock(&file->device->xrcd_tree_mutex); + return in_len; + +err_copy: + if (inode) { + if (new_xrcd) + xrcd_table_delete(file->device, inode); + atomic_dec(&xrcd->usecnt); + } + +err_insert_xrcd: + idr_remove_uobj(&ib_uverbs_xrcd_idr, &obj->uobject); + +err_idr: + ib_dealloc_xrcd(xrcd); + +err: + put_uobj_write(&obj->uobject); + +err_tree_mutex_unlock: + if (f) + fput(f); + + mutex_unlock(&file->device->xrcd_tree_mutex); + + return ret; +} + +ssize_t ib_uverbs_close_xrcd(struct ib_uverbs_file *file, + const char __user *buf, int in_len, + int out_len) +{ + struct ib_uverbs_close_xrcd cmd; + struct ib_uobject *uobj; + struct ib_xrcd *xrcd = NULL; + struct inode *inode = NULL; + struct ib_uxrcd_object *obj; + int live; + int ret = 0; + + if (copy_from_user(&cmd, buf, sizeof cmd)) + return -EFAULT; + + mutex_lock(&file->device->xrcd_tree_mutex); + uobj = idr_write_uobj(&ib_uverbs_xrcd_idr, cmd.xrcd_handle, file->ucontext); + if (!uobj) { + ret = -EINVAL; + goto out; + } + + xrcd = uobj->object; + inode = xrcd->inode; + obj = container_of(uobj, struct ib_uxrcd_object, uobject); + if (atomic_read(&obj->refcnt)) { + put_uobj_write(uobj); + ret = -EBUSY; + goto out; + } + + if (!inode || atomic_dec_and_test(&xrcd->usecnt)) { + ret = ib_dealloc_xrcd(uobj->object); + if (!ret) + uobj->live = 0; + } + + live = uobj->live; + if (inode && ret) + atomic_inc(&xrcd->usecnt); + + put_uobj_write(uobj); + + if (ret) + goto out; + + if (inode && !live) + xrcd_table_delete(file->device, inode); + + idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj); + mutex_lock(&file->mutex); + list_del(&uobj->list); + mutex_unlock(&file->mutex); + + put_uobj(uobj); + ret = in_len; + +out: + mutex_unlock(&file->device->xrcd_tree_mutex); + return ret; +} + +void ib_uverbs_dealloc_xrcd(struct ib_uverbs_device *dev, + struct ib_xrcd *xrcd) +{ + struct inode *inode; + + inode = xrcd->inode; + if (inode && !atomic_dec_and_test(&xrcd->usecnt)) + return; + + ib_dealloc_xrcd(xrcd); + + if (inode) + xrcd_table_delete(dev, inode); +} + ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file, const char __user *buf, int in_len, int out_len) @@ -1052,9 +1383,12 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, struct ib_uverbs_create_qp_resp resp; struct ib_udata udata; struct ib_uqp_object *obj; - struct ib_pd *pd; - struct ib_cq *scq, *rcq; - struct ib_srq *srq; + struct ib_device *device; + struct ib_pd *pd = NULL; + struct ib_xrcd *xrcd = NULL; + struct ib_uobject *uninitialized_var(xrcd_uobj); + struct ib_cq *scq = NULL, *rcq = NULL; + struct ib_srq *srq = NULL; struct ib_qp *qp; struct ib_qp_init_attr attr; int ret; @@ -1076,15 +1410,39 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, init_uobj(&obj->uevent.uobject, cmd.user_handle, file->ucontext, &qp_lock_key); down_write(&obj->uevent.uobject.mutex); - srq = cmd.is_srq ? idr_read_srq(cmd.srq_handle, file->ucontext) : NULL; - pd = idr_read_pd(cmd.pd_handle, file->ucontext); - scq = idr_read_cq(cmd.send_cq_handle, file->ucontext, 0); - rcq = cmd.recv_cq_handle == cmd.send_cq_handle ? - scq : idr_read_cq(cmd.recv_cq_handle, file->ucontext, 1); + if (cmd.qp_type == IB_QPT_XRC_TGT) { + xrcd = idr_read_xrcd(cmd.pd_handle, file->ucontext, &xrcd_uobj); + if (!xrcd) { + ret = -EINVAL; + goto err_put; + } + device = xrcd->device; + } else { + pd = idr_read_pd(cmd.pd_handle, file->ucontext); + scq = idr_read_cq(cmd.send_cq_handle, file->ucontext, 0); + if (!pd || !scq) { + ret = -EINVAL; + goto err_put; + } - if (!pd || !scq || !rcq || (cmd.is_srq && !srq)) { - ret = -EINVAL; - goto err_put; + if (cmd.qp_type == IB_QPT_XRC_INI) { + cmd.max_recv_wr = cmd.max_recv_sge = 0; + } else { + if (cmd.is_srq) { + srq = idr_read_srq(cmd.srq_handle, file->ucontext); + if (!srq || srq->srq_type != IB_SRQT_BASIC) { + ret = -EINVAL; + goto err_put; + } + } + rcq = (cmd.recv_cq_handle == cmd.send_cq_handle) ? + scq : idr_read_cq(cmd.recv_cq_handle, file->ucontext, 1); + if (!rcq) { + ret = -EINVAL; + goto err_put; + } + } + device = pd->device; } attr.event_handler = ib_uverbs_qp_event_handler; @@ -1092,6 +1450,7 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, attr.send_cq = scq; attr.recv_cq = rcq; attr.srq = srq; + attr.xrcd = xrcd; attr.sq_sig_type = cmd.sq_sig_all ? IB_SIGNAL_ALL_WR : IB_SIGNAL_REQ_WR; attr.qp_type = cmd.qp_type; attr.create_flags = 0; @@ -1106,26 +1465,35 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, INIT_LIST_HEAD(&obj->uevent.event_list); INIT_LIST_HEAD(&obj->mcast_list); - qp = pd->device->create_qp(pd, &attr, &udata); + if (cmd.qp_type == IB_QPT_XRC_TGT) + qp = ib_create_qp(pd, &attr); + else + qp = device->create_qp(pd, &attr, &udata); + if (IS_ERR(qp)) { ret = PTR_ERR(qp); goto err_put; } - qp->device = pd->device; - qp->pd = pd; - qp->send_cq = attr.send_cq; - qp->recv_cq = attr.recv_cq; - qp->srq = attr.srq; - qp->uobject = &obj->uevent.uobject; - qp->event_handler = attr.event_handler; - qp->qp_context = attr.qp_context; - qp->qp_type = attr.qp_type; - atomic_inc(&pd->usecnt); - atomic_inc(&attr.send_cq->usecnt); - atomic_inc(&attr.recv_cq->usecnt); - if (attr.srq) - atomic_inc(&attr.srq->usecnt); + if (cmd.qp_type != IB_QPT_XRC_TGT) { + qp->real_qp = qp; + qp->device = device; + qp->pd = pd; + qp->send_cq = attr.send_cq; + qp->recv_cq = attr.recv_cq; + qp->srq = attr.srq; + qp->event_handler = attr.event_handler; + qp->qp_context = attr.qp_context; + qp->qp_type = attr.qp_type; + atomic_set(&qp->usecnt, 0); + atomic_inc(&pd->usecnt); + atomic_inc(&attr.send_cq->usecnt); + if (attr.recv_cq) + atomic_inc(&attr.recv_cq->usecnt); + if (attr.srq) + atomic_inc(&attr.srq->usecnt); + } + qp->uobject = &obj->uevent.uobject; obj->uevent.uobject.object = qp; ret = idr_add_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject); @@ -1147,9 +1515,13 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, goto err_copy; } - put_pd_read(pd); - put_cq_read(scq); - if (rcq != scq) + if (xrcd) + put_xrcd_read(xrcd_uobj); + if (pd) + put_pd_read(pd); + if (scq) + put_cq_read(scq); + if (rcq && rcq != scq) put_cq_read(rcq); if (srq) put_srq_read(srq); @@ -1171,6 +1543,8 @@ err_destroy: ib_destroy_qp(qp); err_put: + if (xrcd) + put_xrcd_read(xrcd_uobj); if (pd) put_pd_read(pd); if (scq) @@ -1184,6 +1558,98 @@ err_put: return ret; } +ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file, + const char __user *buf, int in_len, int out_len) +{ + struct ib_uverbs_open_qp cmd; + struct ib_uverbs_create_qp_resp resp; + struct ib_udata udata; + 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; + + if (out_len < sizeof resp) + return -ENOSPC; + + if (copy_from_user(&cmd, buf, sizeof cmd)) + return -EFAULT; + + INIT_UDATA(&udata, buf + sizeof cmd, + (unsigned long) cmd.response + sizeof resp, + in_len - sizeof cmd, out_len - sizeof resp); + + obj = kmalloc(sizeof *obj, GFP_KERNEL); + if (!obj) + return -ENOMEM; + + init_uobj(&obj->uevent.uobject, cmd.user_handle, file->ucontext, &qp_lock_key); + down_write(&obj->uevent.uobject.mutex); + + xrcd = idr_read_xrcd(cmd.pd_handle, file->ucontext, &xrcd_uobj); + if (!xrcd) { + ret = -EINVAL; + goto err_put; + } + + attr.event_handler = ib_uverbs_qp_event_handler; + attr.qp_context = file; + attr.qp_num = cmd.qpn; + attr.qp_type = cmd.qp_type; + + obj->uevent.events_reported = 0; + INIT_LIST_HEAD(&obj->uevent.event_list); + INIT_LIST_HEAD(&obj->mcast_list); + + qp = ib_open_qp(xrcd, &attr); + if (IS_ERR(qp)) { + ret = PTR_ERR(qp); + goto err_put; + } + + qp->uobject = &obj->uevent.uobject; + + obj->uevent.uobject.object = qp; + ret = idr_add_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject); + if (ret) + goto err_destroy; + + memset(&resp, 0, sizeof resp); + resp.qpn = qp->qp_num; + resp.qp_handle = obj->uevent.uobject.id; + + if (copy_to_user((void __user *) (unsigned long) cmd.response, + &resp, sizeof resp)) { + ret = -EFAULT; + goto err_remove; + } + + put_xrcd_read(xrcd_uobj); + + mutex_lock(&file->mutex); + list_add_tail(&obj->uevent.uobject.list, &file->ucontext->qp_list); + mutex_unlock(&file->mutex); + + obj->uevent.uobject.live = 1; + + up_write(&obj->uevent.uobject.mutex); + + return in_len; + +err_remove: + idr_remove_uobj(&ib_uverbs_qp_idr, &obj->uevent.uobject); + +err_destroy: + ib_destroy_qp(qp); + +err_put: + put_xrcd_read(xrcd_uobj); + put_uobj_write(&obj->uevent.uobject); + return ret; +} + ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file, const char __user *buf, int in_len, int out_len) @@ -1284,6 +1750,20 @@ out: return ret ? ret : in_len; } +/* Remove ignored fields set in the attribute mask */ +static int modify_qp_mask(enum ib_qp_type qp_type, int mask) +{ + switch (qp_type) { + case IB_QPT_XRC_INI: + return mask & ~(IB_QP_MAX_DEST_RD_ATOMIC | IB_QP_MIN_RNR_TIMER); + case IB_QPT_XRC_TGT: + return mask & ~(IB_QP_MAX_QP_RD_ATOMIC | IB_QP_RETRY_CNT | + IB_QP_RNR_RETRY); + default: + return mask; + } +} + ssize_t ib_uverbs_modify_qp(struct ib_uverbs_file *file, const char __user *buf, int in_len, int out_len) @@ -1356,7 +1836,12 @@ ssize_t ib_uverbs_modify_qp(struct ib_uverbs_file *file, attr->alt_ah_attr.ah_flags = cmd.alt_dest.is_global ? IB_AH_GRH : 0; attr->alt_ah_attr.port_num = cmd.alt_dest.port_num; - ret = qp->device->modify_qp(qp, attr, cmd.attr_mask, &udata); + if (qp->real_qp == qp) { + ret = qp->device->modify_qp(qp, attr, + modify_qp_mask(qp->qp_type, cmd.attr_mask), &udata); + } else { + ret = ib_modify_qp(qp, attr, modify_qp_mask(qp->qp_type, cmd.attr_mask)); + } put_qp_read(qp); @@ -1494,6 +1979,12 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, next->send_flags = user_wr->send_flags; if (is_ud) { + if (next->opcode != IB_WR_SEND && + next->opcode != IB_WR_SEND_WITH_IMM) { + ret = -EINVAL; + goto out_put; + } + next->wr.ud.ah = idr_read_ah(user_wr->wr.ud.ah, file->ucontext); if (!next->wr.ud.ah) { @@ -1530,9 +2021,11 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, user_wr->wr.atomic.compare_add; next->wr.atomic.swap = user_wr->wr.atomic.swap; next->wr.atomic.rkey = user_wr->wr.atomic.rkey; + case IB_WR_SEND: break; default: - break; + ret = -EINVAL; + goto out_put; } } @@ -1553,7 +2046,7 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, } resp.bad_wr = 0; - ret = qp->device->post_send(qp, wr, &bad_wr); + ret = qp->device->post_send(qp->real_qp, wr, &bad_wr); if (ret) for (next = wr; next; next = next->next) { ++resp.bad_wr; @@ -1691,7 +2184,7 @@ ssize_t ib_uverbs_post_recv(struct ib_uverbs_file *file, goto out; resp.bad_wr = 0; - ret = qp->device->post_recv(qp, wr, &bad_wr); + ret = qp->device->post_recv(qp->real_qp, wr, &bad_wr); put_qp_read(qp); @@ -1904,7 +2397,7 @@ ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file, if (copy_from_user(&cmd, buf, sizeof cmd)) return -EFAULT; - qp = idr_read_qp(cmd.qp_handle, file->ucontext); + qp = idr_write_qp(cmd.qp_handle, file->ucontext); if (!qp) return -EINVAL; @@ -1933,7 +2426,7 @@ ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file, kfree(mcast); out_put: - put_qp_read(qp); + put_qp_write(qp); return ret ? ret : in_len; } @@ -1951,7 +2444,7 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file, if (copy_from_user(&cmd, buf, sizeof cmd)) return -EFAULT; - qp = idr_read_qp(cmd.qp_handle, file->ucontext); + qp = idr_write_qp(cmd.qp_handle, file->ucontext); if (!qp) return -EINVAL; @@ -1970,112 +2463,204 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file, } out_put: - put_qp_read(qp); + put_qp_write(qp); return ret ? ret : in_len; } -ssize_t ib_uverbs_create_srq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +int __uverbs_create_xsrq(struct ib_uverbs_file *file, + struct ib_uverbs_create_xsrq *cmd, + struct ib_udata *udata) { - struct ib_uverbs_create_srq cmd; struct ib_uverbs_create_srq_resp resp; - struct ib_udata udata; - struct ib_uevent_object *obj; + 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; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - INIT_UDATA(&udata, buf + sizeof cmd, - (unsigned long) cmd.response + sizeof resp, - in_len - sizeof cmd, out_len - sizeof resp); - obj = kmalloc(sizeof *obj, GFP_KERNEL); if (!obj) return -ENOMEM; - init_uobj(&obj->uobject, cmd.user_handle, file->ucontext, &srq_lock_key); - down_write(&obj->uobject.mutex); + init_uobj(&obj->uevent.uobject, cmd->user_handle, file->ucontext, &srq_lock_key); + down_write(&obj->uevent.uobject.mutex); - pd = idr_read_pd(cmd.pd_handle, file->ucontext); + pd = idr_read_pd(cmd->pd_handle, file->ucontext); if (!pd) { ret = -EINVAL; goto err; } + if (cmd->srq_type == IB_SRQT_XRC) { + attr.ext.xrc.cq = idr_read_cq(cmd->cq_handle, file->ucontext, 0); + if (!attr.ext.xrc.cq) { + ret = -EINVAL; + goto err_put_pd; + } + + attr.ext.xrc.xrcd = idr_read_xrcd(cmd->xrcd_handle, file->ucontext, &xrcd_uobj); + if (!attr.ext.xrc.xrcd) { + ret = -EINVAL; + goto err_put_cq; + } + + obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, uobject); + atomic_inc(&obj->uxrcd->refcnt); + } + attr.event_handler = ib_uverbs_srq_event_handler; attr.srq_context = file; - attr.attr.max_wr = cmd.max_wr; - attr.attr.max_sge = cmd.max_sge; - attr.attr.srq_limit = cmd.srq_limit; + 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; - obj->events_reported = 0; - INIT_LIST_HEAD(&obj->event_list); + obj->uevent.events_reported = 0; + INIT_LIST_HEAD(&obj->uevent.event_list); - srq = pd->device->create_srq(pd, &attr, &udata); + srq = pd->device->create_srq(pd, &attr, udata); if (IS_ERR(srq)) { ret = PTR_ERR(srq); goto err_put; } - srq->device = pd->device; - srq->pd = pd; - srq->uobject = &obj->uobject; + srq->device = pd->device; + srq->pd = pd; + srq->srq_type = cmd->srq_type; + srq->uobject = &obj->uevent.uobject; srq->event_handler = attr.event_handler; srq->srq_context = attr.srq_context; + + if (cmd->srq_type == IB_SRQT_XRC) { + srq->ext.xrc.cq = attr.ext.xrc.cq; + srq->ext.xrc.xrcd = attr.ext.xrc.xrcd; + atomic_inc(&attr.ext.xrc.cq->usecnt); + atomic_inc(&attr.ext.xrc.xrcd->usecnt); + } + atomic_inc(&pd->usecnt); atomic_set(&srq->usecnt, 0); - obj->uobject.object = srq; - ret = idr_add_uobj(&ib_uverbs_srq_idr, &obj->uobject); + obj->uevent.uobject.object = srq; + ret = idr_add_uobj(&ib_uverbs_srq_idr, &obj->uevent.uobject); if (ret) goto err_destroy; memset(&resp, 0, sizeof resp); - resp.srq_handle = obj->uobject.id; + 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; - if (copy_to_user((void __user *) (unsigned long) cmd.response, + if (copy_to_user((void __user *) (unsigned long) cmd->response, &resp, sizeof resp)) { ret = -EFAULT; goto err_copy; } + if (cmd->srq_type == IB_SRQT_XRC) { + put_uobj_read(xrcd_uobj); + put_cq_read(attr.ext.xrc.cq); + } put_pd_read(pd); mutex_lock(&file->mutex); - list_add_tail(&obj->uobject.list, &file->ucontext->srq_list); + list_add_tail(&obj->uevent.uobject.list, &file->ucontext->srq_list); mutex_unlock(&file->mutex); - obj->uobject.live = 1; + obj->uevent.uobject.live = 1; - up_write(&obj->uobject.mutex); + up_write(&obj->uevent.uobject.mutex); - return in_len; + return 0; err_copy: - idr_remove_uobj(&ib_uverbs_srq_idr, &obj->uobject); + idr_remove_uobj(&ib_uverbs_srq_idr, &obj->uevent.uobject); err_destroy: ib_destroy_srq(srq); err_put: + if (cmd->srq_type == IB_SRQT_XRC) { + atomic_dec(&obj->uxrcd->refcnt); + put_uobj_read(xrcd_uobj); + } + +err_put_cq: + if (cmd->srq_type == IB_SRQT_XRC) + put_cq_read(attr.ext.xrc.cq); + +err_put_pd: put_pd_read(pd); err: - put_uobj_write(&obj->uobject); + put_uobj_write(&obj->uevent.uobject); return ret; } +ssize_t ib_uverbs_create_srq(struct ib_uverbs_file *file, + const char __user *buf, int in_len, + int out_len) +{ + struct ib_uverbs_create_srq cmd; + struct ib_uverbs_create_xsrq xcmd; + struct ib_uverbs_create_srq_resp resp; + struct ib_udata udata; + int ret; + + if (out_len < sizeof resp) + return -ENOSPC; + + if (copy_from_user(&cmd, buf, sizeof cmd)) + return -EFAULT; + + xcmd.response = cmd.response; + xcmd.user_handle = cmd.user_handle; + xcmd.srq_type = IB_SRQT_BASIC; + xcmd.pd_handle = cmd.pd_handle; + xcmd.max_wr = cmd.max_wr; + xcmd.max_sge = cmd.max_sge; + xcmd.srq_limit = cmd.srq_limit; + + INIT_UDATA(&udata, buf + sizeof cmd, + (unsigned long) cmd.response + sizeof resp, + in_len - sizeof cmd, out_len - sizeof resp); + + ret = __uverbs_create_xsrq(file, &xcmd, &udata); + if (ret) + return ret; + + return in_len; +} + +ssize_t ib_uverbs_create_xsrq(struct ib_uverbs_file *file, + const char __user *buf, int in_len, int out_len) +{ + struct ib_uverbs_create_xsrq cmd; + struct ib_uverbs_create_srq_resp resp; + struct ib_udata udata; + int ret; + + if (out_len < sizeof resp) + return -ENOSPC; + + if (copy_from_user(&cmd, buf, sizeof cmd)) + return -EFAULT; + + INIT_UDATA(&udata, buf + sizeof cmd, + (unsigned long) cmd.response + sizeof resp, + in_len - sizeof cmd, out_len - sizeof resp); + + ret = __uverbs_create_xsrq(file, &cmd, &udata); + if (ret) + return ret; + + return in_len; +} + ssize_t ib_uverbs_modify_srq(struct ib_uverbs_file *file, const char __user *buf, int in_len, int out_len) diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 56898b6..f07c6e3 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -72,6 +72,7 @@ DEFINE_IDR(ib_uverbs_ah_idr); DEFINE_IDR(ib_uverbs_cq_idr); DEFINE_IDR(ib_uverbs_qp_idr); DEFINE_IDR(ib_uverbs_srq_idr); +DEFINE_IDR(ib_uverbs_xrcd_idr); static DEFINE_SPINLOCK(map_lock); static DECLARE_BITMAP(dev_map, IB_UVERBS_MAX_DEVICES); @@ -107,19 +108,27 @@ static ssize_t (*uverbs_cmd_table[])(struct ib_uverbs_file *file, [IB_USER_VERBS_CMD_MODIFY_SRQ] = ib_uverbs_modify_srq, [IB_USER_VERBS_CMD_QUERY_SRQ] = ib_uverbs_query_srq, [IB_USER_VERBS_CMD_DESTROY_SRQ] = ib_uverbs_destroy_srq, + [IB_USER_VERBS_CMD_OPEN_XRCD] = ib_uverbs_open_xrcd, + [IB_USER_VERBS_CMD_CLOSE_XRCD] = ib_uverbs_close_xrcd, + [IB_USER_VERBS_CMD_CREATE_XSRQ] = ib_uverbs_create_xsrq, + [IB_USER_VERBS_CMD_OPEN_QP] = ib_uverbs_open_qp }; static void ib_uverbs_add_one(struct ib_device *device); static void ib_uverbs_remove_one(struct ib_device *device); -static void ib_uverbs_release_dev(struct kref *ref) +static void ib_uverbs_release_dev(struct kobject *kobj) { struct ib_uverbs_device *dev = - container_of(ref, struct ib_uverbs_device, ref); + container_of(kobj, struct ib_uverbs_device, kobj); - complete(&dev->comp); + kfree(dev); } +static struct kobj_type ib_uverbs_dev_ktype = { + .release = ib_uverbs_release_dev, +}; + static void ib_uverbs_release_event_file(struct kref *ref) { struct ib_uverbs_event_file *file = @@ -202,8 +211,12 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file, container_of(uobj, struct ib_uqp_object, uevent.uobject); idr_remove_uobj(&ib_uverbs_qp_idr, uobj); - ib_uverbs_detach_umcast(qp, uqp); - ib_destroy_qp(qp); + if (qp != qp->real_qp) { + ib_close_qp(qp); + } else { + ib_uverbs_detach_umcast(qp, uqp); + ib_destroy_qp(qp); + } ib_uverbs_release_uevent(file, &uqp->uevent); kfree(uqp); } @@ -241,6 +254,18 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file, kfree(uobj); } + mutex_lock(&file->device->xrcd_tree_mutex); + list_for_each_entry_safe(uobj, tmp, &context->xrcd_list, list) { + struct ib_xrcd *xrcd = uobj->object; + struct ib_uxrcd_object *uxrcd = + container_of(uobj, struct ib_uxrcd_object, uobject); + + idr_remove_uobj(&ib_uverbs_xrcd_idr, uobj); + ib_uverbs_dealloc_xrcd(file->device, xrcd); + kfree(uxrcd); + } + mutex_unlock(&file->device->xrcd_tree_mutex); + list_for_each_entry_safe(uobj, tmp, &context->pd_list, list) { struct ib_pd *pd = uobj->object; @@ -252,13 +277,19 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file, return context->device->dealloc_ucontext(context); } +static void ib_uverbs_comp_dev(struct ib_uverbs_device *dev) +{ + complete(&dev->comp); +} + static void ib_uverbs_release_file(struct kref *ref) { struct ib_uverbs_file *file = container_of(ref, struct ib_uverbs_file, ref); module_put(file->device->ib_dev->owner); - kref_put(&file->device->ref, ib_uverbs_release_dev); + if (atomic_dec_and_test(&file->device->refcount)) + ib_uverbs_comp_dev(file->device); kfree(file); } @@ -430,6 +461,7 @@ static void ib_uverbs_async_handler(struct ib_uverbs_file *file, entry->desc.async.element = element; entry->desc.async.event_type = event; + entry->desc.async.reserved = 0; entry->counter = counter; list_add_tail(&entry->list, &file->async_file->event_list); @@ -557,8 +589,7 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, if (hdr.in_words * 4 != count) return -EINVAL; - if (hdr.command < 0 || - hdr.command >= ARRAY_SIZE(uverbs_cmd_table) || + if (hdr.command >= ARRAY_SIZE(uverbs_cmd_table) || !uverbs_cmd_table[hdr.command]) return -EINVAL; @@ -600,9 +631,7 @@ static int ib_uverbs_open(struct inode *inode, struct file *filp) int ret; dev = container_of(inode->i_cdev, struct ib_uverbs_device, cdev); - if (dev) - kref_get(&dev->ref); - else + if (!atomic_inc_not_zero(&dev->refcount)) return -ENXIO; if (!try_module_get(dev->ib_dev->owner)) { @@ -623,6 +652,7 @@ static int ib_uverbs_open(struct inode *inode, struct file *filp) mutex_init(&file->mutex); filp->private_data = file; + kobject_get(&dev->kobj); return nonseekable_open(inode, filp); @@ -630,13 +660,16 @@ err_module: module_put(dev->ib_dev->owner); err: - kref_put(&dev->ref, ib_uverbs_release_dev); + if (atomic_dec_and_test(&dev->refcount)) + ib_uverbs_comp_dev(dev); + return ret; } 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; ib_uverbs_cleanup_ucontext(file, file->ucontext); @@ -644,6 +677,7 @@ static int ib_uverbs_close(struct inode *inode, struct file *filp) kref_put(&file->async_file->ref, ib_uverbs_release_event_file); kref_put(&file->ref, ib_uverbs_release_file); + kobject_put(&dev->kobj); return 0; } @@ -739,8 +773,11 @@ static void ib_uverbs_add_one(struct ib_device *device) if (!uverbs_dev) return; - kref_init(&uverbs_dev->ref); + atomic_set(&uverbs_dev->refcount, 1); init_completion(&uverbs_dev->comp); + uverbs_dev->xrcd_tree = RB_ROOT; + mutex_init(&uverbs_dev->xrcd_tree_mutex); + kobject_init(&uverbs_dev->kobj, &ib_uverbs_dev_ktype); spin_lock(&map_lock); devnum = find_first_zero_bit(dev_map, IB_UVERBS_MAX_DEVICES); @@ -767,6 +804,7 @@ static void ib_uverbs_add_one(struct ib_device *device) cdev_init(&uverbs_dev->cdev, NULL); uverbs_dev->cdev.owner = THIS_MODULE; uverbs_dev->cdev.ops = device->mmap ? &uverbs_mmap_fops : &uverbs_fops; + uverbs_dev->cdev.kobj.parent = &uverbs_dev->kobj; kobject_set_name(&uverbs_dev->cdev.kobj, "uverbs%d", uverbs_dev->devnum); if (cdev_add(&uverbs_dev->cdev, base, 1)) goto err_cdev; @@ -797,9 +835,10 @@ err_cdev: clear_bit(devnum, overflow_map); err: - kref_put(&uverbs_dev->ref, ib_uverbs_release_dev); + if (atomic_dec_and_test(&uverbs_dev->refcount)) + ib_uverbs_comp_dev(uverbs_dev); wait_for_completion(&uverbs_dev->comp); - kfree(uverbs_dev); + kobject_put(&uverbs_dev->kobj); return; } @@ -819,9 +858,10 @@ static void ib_uverbs_remove_one(struct ib_device *device) else clear_bit(uverbs_dev->devnum - IB_UVERBS_MAX_DEVICES, overflow_map); - kref_put(&uverbs_dev->ref, ib_uverbs_release_dev); + if (atomic_dec_and_test(&uverbs_dev->refcount)) + ib_uverbs_comp_dev(uverbs_dev); wait_for_completion(&uverbs_dev->comp); - kfree(uverbs_dev); + kobject_put(&uverbs_dev->kobj); } static char *uverbs_devnode(struct device *dev, mode_t *mode) diff --git a/drivers/infiniband/core/uverbs_marshall.c b/drivers/infiniband/core/uverbs_marshall.c index 1b1146f..e7bee46 100644 --- a/drivers/infiniband/core/uverbs_marshall.c +++ b/drivers/infiniband/core/uverbs_marshall.c @@ -30,6 +30,7 @@ * SOFTWARE. */ +#include <linux/export.h> #include <rdma/ib_marshall.h> void ib_copy_ah_attr_to_user(struct ib_uverbs_ah_attr *dst, diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index af7a8b0..575b780 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -38,7 +38,9 @@ #include <linux/errno.h> #include <linux/err.h> +#include <linux/export.h> #include <linux/string.h> +#include <linux/slab.h> #include <rdma/ib_verbs.h> #include <rdma/ib_cache.h> @@ -77,6 +79,31 @@ enum ib_rate mult_to_ib_rate(int mult) } EXPORT_SYMBOL(mult_to_ib_rate); +int ib_rate_to_mbps(enum ib_rate rate) +{ + switch (rate) { + case IB_RATE_2_5_GBPS: return 2500; + case IB_RATE_5_GBPS: return 5000; + case IB_RATE_10_GBPS: return 10000; + case IB_RATE_20_GBPS: return 20000; + case IB_RATE_30_GBPS: return 30000; + case IB_RATE_40_GBPS: return 40000; + case IB_RATE_60_GBPS: return 60000; + case IB_RATE_80_GBPS: return 80000; + case IB_RATE_120_GBPS: return 120000; + case IB_RATE_14_GBPS: return 14062; + case IB_RATE_56_GBPS: return 56250; + case IB_RATE_112_GBPS: return 112500; + case IB_RATE_168_GBPS: return 168750; + case IB_RATE_25_GBPS: return 25781; + case IB_RATE_100_GBPS: return 103125; + case IB_RATE_200_GBPS: return 206250; + case IB_RATE_300_GBPS: return 309375; + default: return -1; + } +} +EXPORT_SYMBOL(ib_rate_to_mbps); + enum rdma_transport_type rdma_node_get_transport(enum rdma_node_type node_type) { @@ -250,6 +277,13 @@ struct ib_srq *ib_create_srq(struct ib_pd *pd, srq->uobject = NULL; srq->event_handler = srq_init_attr->event_handler; srq->srq_context = srq_init_attr->srq_context; + srq->srq_type = srq_init_attr->srq_type; + if (srq->srq_type == IB_SRQT_XRC) { + srq->ext.xrc.xrcd = srq_init_attr->ext.xrc.xrcd; + srq->ext.xrc.cq = srq_init_attr->ext.xrc.cq; + atomic_inc(&srq->ext.xrc.xrcd->usecnt); + atomic_inc(&srq->ext.xrc.cq->usecnt); + } atomic_inc(&pd->usecnt); atomic_set(&srq->usecnt, 0); } @@ -279,16 +313,29 @@ EXPORT_SYMBOL(ib_query_srq); int ib_destroy_srq(struct ib_srq *srq) { struct ib_pd *pd; + enum ib_srq_type srq_type; + struct ib_xrcd *uninitialized_var(xrcd); + struct ib_cq *uninitialized_var(cq); int ret; if (atomic_read(&srq->usecnt)) return -EBUSY; pd = srq->pd; + srq_type = srq->srq_type; + if (srq_type == IB_SRQT_XRC) { + xrcd = srq->ext.xrc.xrcd; + cq = srq->ext.xrc.cq; + } ret = srq->device->destroy_srq(srq); - if (!ret) + if (!ret) { atomic_dec(&pd->usecnt); + if (srq_type == IB_SRQT_XRC) { + atomic_dec(&xrcd->usecnt); + atomic_dec(&cq->usecnt); + } + } return ret; } @@ -296,28 +343,123 @@ EXPORT_SYMBOL(ib_destroy_srq); /* Queue pairs */ +static void __ib_shared_qp_event_handler(struct ib_event *event, void *context) +{ + struct ib_qp *qp = context; + + list_for_each_entry(event->element.qp, &qp->open_list, open_list) + event->element.qp->event_handler(event, event->element.qp->qp_context); +} + +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) +{ + struct ib_qp *qp; + unsigned long flags; + + qp = kzalloc(sizeof *qp, GFP_KERNEL); + if (!qp) + return ERR_PTR(-ENOMEM); + + qp->real_qp = real_qp; + atomic_inc(&real_qp->usecnt); + qp->device = real_qp->device; + qp->event_handler = event_handler; + qp->qp_context = qp_context; + qp->qp_num = real_qp->qp_num; + qp->qp_type = real_qp->qp_type; + + spin_lock_irqsave(&real_qp->device->event_handler_lock, flags); + list_add(&qp->open_list, &real_qp->open_list); + spin_unlock_irqrestore(&real_qp->device->event_handler_lock, flags); + + return qp; +} + +struct ib_qp *ib_open_qp(struct ib_xrcd *xrcd, + struct ib_qp_open_attr *qp_open_attr) +{ + struct ib_qp *qp, *real_qp; + + 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; + } + } + mutex_unlock(&xrcd->tgt_qp_mutex); + return qp; +} +EXPORT_SYMBOL(ib_open_qp); + struct ib_qp *ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *qp_init_attr) { - struct ib_qp *qp; + struct ib_qp *qp, *real_qp; + struct ib_device *device; - qp = pd->device->create_qp(pd, qp_init_attr, NULL); + device = pd ? pd->device : qp_init_attr->xrcd->device; + qp = device->create_qp(pd, qp_init_attr, NULL); if (!IS_ERR(qp)) { - qp->device = pd->device; - qp->pd = pd; - qp->send_cq = qp_init_attr->send_cq; - qp->recv_cq = qp_init_attr->recv_cq; - qp->srq = qp_init_attr->srq; - qp->uobject = NULL; - qp->event_handler = qp_init_attr->event_handler; - qp->qp_context = qp_init_attr->qp_context; - qp->qp_type = qp_init_attr->qp_type; - atomic_inc(&pd->usecnt); - atomic_inc(&qp_init_attr->send_cq->usecnt); - atomic_inc(&qp_init_attr->recv_cq->usecnt); - if (qp_init_attr->srq) - atomic_inc(&qp_init_attr->srq->usecnt); + qp->device = device; + qp->real_qp = qp; + qp->uobject = NULL; + qp->qp_type = qp_init_attr->qp_type; + + atomic_set(&qp->usecnt, 0); + if (qp_init_attr->qp_type == IB_QPT_XRC_TGT) { + qp->event_handler = __ib_shared_qp_event_handler; + qp->qp_context = qp; + qp->pd = NULL; + qp->send_cq = qp->recv_cq = NULL; + qp->srq = NULL; + qp->xrcd = qp_init_attr->xrcd; + atomic_inc(&qp_init_attr->xrcd->usecnt); + INIT_LIST_HEAD(&qp->open_list); + + real_qp = qp; + qp = __ib_open_qp(real_qp, qp_init_attr->event_handler, + qp_init_attr->qp_context); + if (!IS_ERR(qp)) + __ib_insert_xrcd_qp(qp_init_attr->xrcd, real_qp); + else + real_qp->device->destroy_qp(real_qp); + } else { + qp->event_handler = qp_init_attr->event_handler; + qp->qp_context = qp_init_attr->qp_context; + if (qp_init_attr->qp_type == IB_QPT_XRC_INI) { + qp->recv_cq = NULL; + qp->srq = NULL; + } else { + qp->recv_cq = qp_init_attr->recv_cq; + atomic_inc(&qp_init_attr->recv_cq->usecnt); + qp->srq = qp_init_attr->srq; + if (qp->srq) + atomic_inc(&qp_init_attr->srq->usecnt); + } + + qp->pd = pd; + qp->send_cq = qp_init_attr->send_cq; + qp->xrcd = NULL; + + atomic_inc(&pd->usecnt); + atomic_inc(&qp_init_attr->send_cq->usecnt); + } } return qp; @@ -326,8 +468,8 @@ EXPORT_SYMBOL(ib_create_qp); static const struct { int valid; - enum ib_qp_attr_mask req_param[IB_QPT_RAW_ETHERTYPE + 1]; - enum ib_qp_attr_mask opt_param[IB_QPT_RAW_ETHERTYPE + 1]; + enum ib_qp_attr_mask req_param[IB_QPT_MAX]; + enum ib_qp_attr_mask opt_param[IB_QPT_MAX]; } qp_state_table[IB_QPS_ERR + 1][IB_QPS_ERR + 1] = { [IB_QPS_RESET] = { [IB_QPS_RESET] = { .valid = 1 }, @@ -343,6 +485,12 @@ static const struct { [IB_QPT_RC] = (IB_QP_PKEY_INDEX | IB_QP_PORT | IB_QP_ACCESS_FLAGS), + [IB_QPT_XRC_INI] = (IB_QP_PKEY_INDEX | + IB_QP_PORT | + IB_QP_ACCESS_FLAGS), + [IB_QPT_XRC_TGT] = (IB_QP_PKEY_INDEX | + IB_QP_PORT | + IB_QP_ACCESS_FLAGS), [IB_QPT_SMI] = (IB_QP_PKEY_INDEX | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_PKEY_INDEX | @@ -365,6 +513,12 @@ static const struct { [IB_QPT_RC] = (IB_QP_PKEY_INDEX | IB_QP_PORT | IB_QP_ACCESS_FLAGS), + [IB_QPT_XRC_INI] = (IB_QP_PKEY_INDEX | + IB_QP_PORT | + IB_QP_ACCESS_FLAGS), + [IB_QPT_XRC_TGT] = (IB_QP_PKEY_INDEX | + IB_QP_PORT | + IB_QP_ACCESS_FLAGS), [IB_QPT_SMI] = (IB_QP_PKEY_INDEX | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_PKEY_INDEX | @@ -384,6 +538,16 @@ static const struct { IB_QP_RQ_PSN | IB_QP_MAX_DEST_RD_ATOMIC | IB_QP_MIN_RNR_TIMER), + [IB_QPT_XRC_INI] = (IB_QP_AV | + IB_QP_PATH_MTU | + IB_QP_DEST_QPN | + IB_QP_RQ_PSN), + [IB_QPT_XRC_TGT] = (IB_QP_AV | + IB_QP_PATH_MTU | + IB_QP_DEST_QPN | + IB_QP_RQ_PSN | + IB_QP_MAX_DEST_RD_ATOMIC | + IB_QP_MIN_RNR_TIMER), }, .opt_param = { [IB_QPT_UD] = (IB_QP_PKEY_INDEX | @@ -394,6 +558,12 @@ static const struct { [IB_QPT_RC] = (IB_QP_ALT_PATH | IB_QP_ACCESS_FLAGS | IB_QP_PKEY_INDEX), + [IB_QPT_XRC_INI] = (IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_PKEY_INDEX), + [IB_QPT_XRC_TGT] = (IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_PKEY_INDEX), [IB_QPT_SMI] = (IB_QP_PKEY_INDEX | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_PKEY_INDEX | @@ -414,6 +584,13 @@ static const struct { IB_QP_RNR_RETRY | IB_QP_SQ_PSN | IB_QP_MAX_QP_RD_ATOMIC), + [IB_QPT_XRC_INI] = (IB_QP_TIMEOUT | + IB_QP_RETRY_CNT | + IB_QP_RNR_RETRY | + IB_QP_SQ_PSN | + IB_QP_MAX_QP_RD_ATOMIC), + [IB_QPT_XRC_TGT] = (IB_QP_TIMEOUT | + IB_QP_SQ_PSN), [IB_QPT_SMI] = IB_QP_SQ_PSN, [IB_QPT_GSI] = IB_QP_SQ_PSN, }, @@ -429,6 +606,15 @@ static const struct { IB_QP_ACCESS_FLAGS | IB_QP_MIN_RNR_TIMER | IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_INI] = (IB_QP_CUR_STATE | + IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_TGT] = (IB_QP_CUR_STATE | + IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_MIN_RNR_TIMER | + IB_QP_PATH_MIG_STATE), [IB_QPT_SMI] = (IB_QP_CUR_STATE | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_CUR_STATE | @@ -453,6 +639,15 @@ static const struct { IB_QP_ALT_PATH | IB_QP_PATH_MIG_STATE | IB_QP_MIN_RNR_TIMER), + [IB_QPT_XRC_INI] = (IB_QP_CUR_STATE | + IB_QP_ACCESS_FLAGS | + IB_QP_ALT_PATH | + IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_TGT] = (IB_QP_CUR_STATE | + IB_QP_ACCESS_FLAGS | + IB_QP_ALT_PATH | + IB_QP_PATH_MIG_STATE | + IB_QP_MIN_RNR_TIMER), [IB_QPT_SMI] = (IB_QP_CUR_STATE | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_CUR_STATE | @@ -465,6 +660,8 @@ static const struct { [IB_QPT_UD] = IB_QP_EN_SQD_ASYNC_NOTIFY, [IB_QPT_UC] = IB_QP_EN_SQD_ASYNC_NOTIFY, [IB_QPT_RC] = IB_QP_EN_SQD_ASYNC_NOTIFY, + [IB_QPT_XRC_INI] = IB_QP_EN_SQD_ASYNC_NOTIFY, + [IB_QPT_XRC_TGT] = IB_QP_EN_SQD_ASYNC_NOTIFY, /* ??? */ [IB_QPT_SMI] = IB_QP_EN_SQD_ASYNC_NOTIFY, [IB_QPT_GSI] = IB_QP_EN_SQD_ASYNC_NOTIFY } @@ -487,6 +684,15 @@ static const struct { IB_QP_ACCESS_FLAGS | IB_QP_MIN_RNR_TIMER | IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_INI] = (IB_QP_CUR_STATE | + IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_TGT] = (IB_QP_CUR_STATE | + IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_MIN_RNR_TIMER | + IB_QP_PATH_MIG_STATE), [IB_QPT_SMI] = (IB_QP_CUR_STATE | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_CUR_STATE | @@ -515,6 +721,25 @@ static const struct { IB_QP_PKEY_INDEX | IB_QP_MIN_RNR_TIMER | IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_INI] = (IB_QP_PORT | + IB_QP_AV | + IB_QP_TIMEOUT | + IB_QP_RETRY_CNT | + IB_QP_RNR_RETRY | + IB_QP_MAX_QP_RD_ATOMIC | + IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_PKEY_INDEX | + IB_QP_PATH_MIG_STATE), + [IB_QPT_XRC_TGT] = (IB_QP_PORT | + IB_QP_AV | + IB_QP_TIMEOUT | + IB_QP_MAX_DEST_RD_ATOMIC | + IB_QP_ALT_PATH | + IB_QP_ACCESS_FLAGS | + IB_QP_PKEY_INDEX | + IB_QP_MIN_RNR_TIMER | + IB_QP_PATH_MIG_STATE), [IB_QPT_SMI] = (IB_QP_PKEY_INDEX | IB_QP_QKEY), [IB_QPT_GSI] = (IB_QP_PKEY_INDEX | @@ -579,7 +804,7 @@ int ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *qp_attr, int qp_attr_mask) { - return qp->device->modify_qp(qp, qp_attr, qp_attr_mask, NULL); + return qp->device->modify_qp(qp->real_qp, qp_attr, qp_attr_mask, NULL); } EXPORT_SYMBOL(ib_modify_qp); @@ -589,11 +814,59 @@ int ib_query_qp(struct ib_qp *qp, struct ib_qp_init_attr *qp_init_attr) { return qp->device->query_qp ? - qp->device->query_qp(qp, qp_attr, qp_attr_mask, qp_init_attr) : + qp->device->query_qp(qp->real_qp, qp_attr, qp_attr_mask, qp_init_attr) : -ENOSYS; } EXPORT_SYMBOL(ib_query_qp); +int ib_close_qp(struct ib_qp *qp) +{ + struct ib_qp *real_qp; + unsigned long flags; + + real_qp = qp->real_qp; + if (real_qp == qp) + return -EINVAL; + + spin_lock_irqsave(&real_qp->device->event_handler_lock, flags); + list_del(&qp->open_list); + spin_unlock_irqrestore(&real_qp->device->event_handler_lock, flags); + + atomic_dec(&real_qp->usecnt); + kfree(qp); + + return 0; +} +EXPORT_SYMBOL(ib_close_qp); + +static int __ib_destroy_shared_qp(struct ib_qp *qp) +{ + struct ib_xrcd *xrcd; + struct ib_qp *real_qp; + int ret; + + real_qp = qp->real_qp; + xrcd = real_qp->xrcd; + + mutex_lock(&xrcd->tgt_qp_mutex); + ib_close_qp(qp); + if (atomic_read(&real_qp->usecnt) == 0) + list_del(&real_qp->xrcd_list); + else + real_qp = NULL; + mutex_unlock(&xrcd->tgt_qp_mutex); + + 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; +} + int ib_destroy_qp(struct ib_qp *qp) { struct ib_pd *pd; @@ -601,16 +874,25 @@ int ib_destroy_qp(struct ib_qp *qp) struct ib_srq *srq; int ret; - pd = qp->pd; - scq = qp->send_cq; - rcq = qp->recv_cq; - srq = qp->srq; + if (atomic_read(&qp->usecnt)) + return -EBUSY; + + if (qp->real_qp != qp) + return __ib_destroy_shared_qp(qp); + + pd = qp->pd; + scq = qp->send_cq; + rcq = qp->recv_cq; + srq = qp->srq; ret = qp->device->destroy_qp(qp); if (!ret) { - atomic_dec(&pd->usecnt); - atomic_dec(&scq->usecnt); - atomic_dec(&rcq->usecnt); + if (pd) + atomic_dec(&pd->usecnt); + if (scq) + atomic_dec(&scq->usecnt); + if (rcq) + atomic_dec(&rcq->usecnt); if (srq) atomic_dec(&srq->usecnt); } @@ -920,3 +1202,42 @@ int ib_detach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid) return qp->device->detach_mcast(qp, gid, lid); } EXPORT_SYMBOL(ib_detach_mcast); + +struct ib_xrcd *ib_alloc_xrcd(struct ib_device *device) +{ + struct ib_xrcd *xrcd; + + if (!device->alloc_xrcd) + return ERR_PTR(-ENOSYS); + + xrcd = device->alloc_xrcd(device, NULL, 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); + } + + return xrcd; +} +EXPORT_SYMBOL(ib_alloc_xrcd); + +int ib_dealloc_xrcd(struct ib_xrcd *xrcd) +{ + 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; + } + + return xrcd->device->dealloc_xrcd(xrcd); +} +EXPORT_SYMBOL(ib_dealloc_xrcd); diff --git a/drivers/infiniband/hw/amso1100/c2.c b/drivers/infiniband/hw/amso1100/c2.c index 0cfc455..5ce7b9e 100644 --- a/drivers/infiniband/hw/amso1100/c2.c +++ b/drivers/infiniband/hw/amso1100/c2.c @@ -36,6 +36,7 @@ #include <linux/netdevice.h> #include <linux/etherdevice.h> #include <linux/inetdevice.h> +#include <linux/interrupt.h> #include <linux/delay.h> #include <linux/ethtool.h> #include <linux/mii.h> @@ -799,13 +800,10 @@ static int c2_xmit_frame(struct sk_buff *skb, struct net_device *netdev) /* Loop thru additional data fragments and queue them */ if (skb_shinfo(skb)->nr_frags) { for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) { - skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; - maplen = frag->size; - mapaddr = - pci_map_page(c2dev->pcidev, frag->page, - frag->page_offset, maplen, - PCI_DMA_TODEVICE); - + const skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; + maplen = skb_frag_size(frag); + mapaddr = skb_frag_dma_map(&c2dev->pcidev->dev, frag, + 0, maplen, DMA_TO_DEVICE); elem = elem->next; elem->skb = NULL; elem->mapaddr = mapaddr; diff --git a/drivers/infiniband/hw/amso1100/c2_ae.c b/drivers/infiniband/hw/amso1100/c2_ae.c index 24f9e3a..32d34e8 100644 --- a/drivers/infiniband/hw/amso1100/c2_ae.c +++ b/drivers/infiniband/hw/amso1100/c2_ae.c @@ -288,6 +288,11 @@ void c2_ae_event(struct c2_dev *c2dev, u32 mq_index) cm_event.private_data_len = be32_to_cpu(req->private_data_length); cm_event.private_data = req->private_data; + /* + * Until ird/ord negotiation via MPAv2 support is added, send + * max supported values + */ + cm_event.ird = cm_event.ord = 128; if (cm_id->event_handler) cm_id->event_handler(cm_id, &cm_event); diff --git a/drivers/infiniband/hw/amso1100/c2_intr.c b/drivers/infiniband/hw/amso1100/c2_intr.c index 0ebe4e8..8951db4 100644 --- a/drivers/infiniband/hw/amso1100/c2_intr.c +++ b/drivers/infiniband/hw/amso1100/c2_intr.c @@ -183,6 +183,11 @@ static void handle_vq(struct c2_dev *c2dev, u32 mq_index) case IW_CM_EVENT_ESTABLISHED: c2_set_qp_state(req->qp, C2_QP_STATE_RTS); + /* + * Until ird/ord negotiation via MPAv2 support is added, send + * max supported values + */ + cm_event.ird = cm_event.ord = 128; case IW_CM_EVENT_CLOSE: /* diff --git a/drivers/infiniband/hw/amso1100/c2_provider.c b/drivers/infiniband/hw/amso1100/c2_provider.c index aeebc4d..12f923d 100644 --- a/drivers/infiniband/hw/amso1100/c2_provider.c +++ b/drivers/infiniband/hw/amso1100/c2_provider.c @@ -99,14 +99,6 @@ static int c2_query_port(struct ib_device *ibdev, return 0; } -static int c2_modify_port(struct ib_device *ibdev, - u8 port, int port_modify_mask, - struct ib_port_modify *props) -{ - pr_debug("%s:%u\n", __func__, __LINE__); - return 0; -} - static int c2_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 * pkey) { @@ -761,10 +753,7 @@ static struct net_device *c2_pseudo_netdev_init(struct c2_dev *c2dev) memcpy_fromio(netdev->dev_addr, c2dev->kva + C2_REGS_RDMA_ENADDR, 6); /* Print out the MAC address */ - pr_debug("%s: MAC %02X:%02X:%02X:%02X:%02X:%02X\n", - netdev->name, - netdev->dev_addr[0], netdev->dev_addr[1], netdev->dev_addr[2], - netdev->dev_addr[3], netdev->dev_addr[4], netdev->dev_addr[5]); + pr_debug("%s: MAC %pM\n", netdev->name, netdev->dev_addr); #if 0 /* Disable network packets */ @@ -817,7 +806,6 @@ int c2_register_device(struct c2_dev *dev) dev->ibdev.dma_device = &dev->pcidev->dev; dev->ibdev.query_device = c2_query_device; dev->ibdev.query_port = c2_query_port; - dev->ibdev.modify_port = c2_modify_port; dev->ibdev.query_pkey = c2_query_pkey; dev->ibdev.query_gid = c2_query_gid; dev->ibdev.alloc_ucontext = c2_alloc_ucontext; diff --git a/drivers/infiniband/hw/cxgb3/Makefile b/drivers/infiniband/hw/cxgb3/Makefile index 621619c..2761364 100644 --- a/drivers/infiniband/hw/cxgb3/Makefile +++ b/drivers/infiniband/hw/cxgb3/Makefile @@ -1,4 +1,4 @@ -ccflags-y := -Idrivers/net/cxgb3 +ccflags-y := -Idrivers/net/ethernet/chelsio/cxgb3 obj-$(CONFIG_INFINIBAND_CXGB3) += iw_cxgb3.o diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c index e55ce7a..c88b12b 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_cm.c +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c @@ -753,6 +753,11 @@ static void connect_request_upcall(struct iwch_ep *ep) event.private_data_len = ep->plen; event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); event.provider_data = ep; + /* + * Until ird/ord negotiation via MPAv2 support is added, send max + * supported values + */ + event.ird = event.ord = 8; if (state_read(&ep->parent_ep->com) != DEAD) { get_ep(&ep->com); ep->parent_ep->com.cm_id->event_handler( @@ -770,6 +775,11 @@ static void established_upcall(struct iwch_ep *ep) PDBG("%s ep %p\n", __func__, ep); memset(&event, 0, sizeof(event)); event.event = IW_CM_EVENT_ESTABLISHED; + /* + * Until ird/ord negotiation via MPAv2 support is added, send max + * supported values + */ + event.ird = event.ord = 8; if (ep->com.cm_id) { PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid); ep->com.cm_id->event_handler(ep->com.cm_id, &event); diff --git a/drivers/infiniband/hw/cxgb3/iwch_ev.c b/drivers/infiniband/hw/cxgb3/iwch_ev.c index 71e0d84..abcc9e7 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_ev.c +++ b/drivers/infiniband/hw/cxgb3/iwch_ev.c @@ -46,6 +46,7 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp, struct ib_event event; struct iwch_qp_attributes attrs; struct iwch_qp *qhp; + unsigned long flag; spin_lock(&rnicp->lock); qhp = get_qhp(rnicp, CQE_QPID(rsp_msg->cqe)); @@ -94,7 +95,9 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp, if (qhp->ibqp.event_handler) (*qhp->ibqp.event_handler)(&event, qhp->ibqp.qp_context); + spin_lock_irqsave(&chp->comp_handler_lock, flag); (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); + spin_unlock_irqrestore(&chp->comp_handler_lock, flag); if (atomic_dec_and_test(&qhp->refcnt)) wake_up(&qhp->wait); @@ -107,6 +110,7 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb) struct iwch_cq *chp; struct iwch_qp *qhp; u32 cqid = RSPQ_CQID(rsp_msg); + unsigned long flag; rnicp = (struct iwch_dev *) rdev_p->ulp; spin_lock(&rnicp->lock); @@ -170,7 +174,9 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb) */ if (qhp->ep && SQ_TYPE(rsp_msg->cqe)) dst_confirm(qhp->ep->dst); + spin_lock_irqsave(&chp->comp_handler_lock, flag); (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); + spin_unlock_irqrestore(&chp->comp_handler_lock, flag); break; case TPT_ERR_STAG: diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index 2e27413..37c224f 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -61,13 +61,6 @@ #include "iwch_user.h" #include "common.h" -static int iwch_modify_port(struct ib_device *ibdev, - u8 port, int port_modify_mask, - struct ib_port_modify *props) -{ - return -ENOSYS; -} - static struct ib_ah *iwch_ah_create(struct ib_pd *pd, struct ib_ah_attr *ah_attr) { @@ -197,6 +190,7 @@ static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, int ve chp->rhp = rhp; chp->ibcq.cqe = 1 << chp->cq.size_log2; spin_lock_init(&chp->lock); + spin_lock_init(&chp->comp_handler_lock); atomic_set(&chp->refcnt, 1); init_waitqueue_head(&chp->wait); if (insert_handle(rhp, &rhp->cqidr, chp, chp->cq.cqid)) { @@ -1392,7 +1386,6 @@ int iwch_register_device(struct iwch_dev *dev) dev->ibdev.dma_device = &(dev->rdev.rnic_info.pdev->dev); dev->ibdev.query_device = iwch_query_device; dev->ibdev.query_port = iwch_query_port; - dev->ibdev.modify_port = iwch_modify_port; dev->ibdev.query_pkey = iwch_query_pkey; dev->ibdev.query_gid = iwch_query_gid; dev->ibdev.alloc_ucontext = iwch_alloc_ucontext; diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.h b/drivers/infiniband/hw/cxgb3/iwch_provider.h index 9a342c9..87c14b0 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.h +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.h @@ -103,6 +103,7 @@ struct iwch_cq { struct iwch_dev *rhp; struct t3_cq cq; spinlock_t lock; + spinlock_t comp_handler_lock; atomic_t refcnt; wait_queue_head_t wait; u32 __user *user_rptr_addr; diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c index ecd313f..bea5839 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_qp.c +++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c @@ -822,8 +822,11 @@ static void __flush_qp(struct iwch_qp *qhp, struct iwch_cq *rchp, flushed = cxio_flush_rq(&qhp->wq, &rchp->cq, count); spin_unlock(&qhp->lock); spin_unlock_irqrestore(&rchp->lock, *flag); - if (flushed) + if (flushed) { + spin_lock_irqsave(&rchp->comp_handler_lock, *flag); (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, *flag); + } /* locking hierarchy: cq lock first, then qp lock. */ spin_lock_irqsave(&schp->lock, *flag); @@ -833,8 +836,11 @@ static void __flush_qp(struct iwch_qp *qhp, struct iwch_cq *rchp, flushed = cxio_flush_sq(&qhp->wq, &schp->cq, count); spin_unlock(&qhp->lock); spin_unlock_irqrestore(&schp->lock, *flag); - if (flushed) + if (flushed) { + spin_lock_irqsave(&schp->comp_handler_lock, *flag); (*schp->ibcq.comp_handler)(&schp->ibcq, schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, *flag); + } /* deref */ if (atomic_dec_and_test(&qhp->refcnt)) @@ -853,11 +859,15 @@ static void flush_qp(struct iwch_qp *qhp, unsigned long *flag) if (qhp->ibqp.uobject) { cxio_set_wq_in_error(&qhp->wq); cxio_set_cq_in_error(&rchp->cq); + spin_lock_irqsave(&rchp->comp_handler_lock, *flag); (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, *flag); if (schp != rchp) { cxio_set_cq_in_error(&schp->cq); + spin_lock_irqsave(&schp->comp_handler_lock, *flag); (*schp->ibcq.comp_handler)(&schp->ibcq, schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, *flag); } return; } diff --git a/drivers/infiniband/hw/cxgb4/Makefile b/drivers/infiniband/hw/cxgb4/Makefile index cd20b13..46b878c 100644 --- a/drivers/infiniband/hw/cxgb4/Makefile +++ b/drivers/infiniband/hw/cxgb4/Makefile @@ -1,4 +1,4 @@ -ccflags-y := -Idrivers/net/cxgb4 +ccflags-y := -Idrivers/net/ethernet/chelsio/cxgb4 obj-$(CONFIG_INFINIBAND_CXGB4) += iw_cxgb4.o diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c index 267005d..2e6a538 100644 --- a/drivers/infiniband/hw/cxgb4/cm.c +++ b/drivers/infiniband/hw/cxgb4/cm.c @@ -103,7 +103,8 @@ MODULE_PARM_DESC(ep_timeout_secs, "CM Endpoint operation timeout " static int mpa_rev = 1; module_param(mpa_rev, int, 0644); MODULE_PARM_DESC(mpa_rev, "MPA Revision, 0 supports amso1100, " - "1 is spec compliant. (default=1)"); + "1 is RFC0544 spec compliant, 2 is IETF MPA Peer Connect Draft" + " compliant (default=1)"); static int markers_enabled; module_param(markers_enabled, int, 0644); @@ -497,17 +498,21 @@ static int send_connect(struct c4iw_ep *ep) return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t); } -static void send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb) +static void send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb, + u8 mpa_rev_to_use) { int mpalen, wrlen; struct fw_ofld_tx_data_wr *req; struct mpa_message *mpa; + struct mpa_v2_conn_params mpa_v2_params; PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen); BUG_ON(skb_cloned(skb)); mpalen = sizeof(*mpa) + ep->plen; + if (mpa_rev_to_use == 2) + mpalen += sizeof(struct mpa_v2_conn_params); wrlen = roundup(mpalen + sizeof *req, 16); skb = get_skb(skb, wrlen, GFP_KERNEL); if (!skb) { @@ -533,12 +538,41 @@ static void send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb) mpa = (struct mpa_message *)(req + 1); memcpy(mpa->key, MPA_KEY_REQ, sizeof(mpa->key)); mpa->flags = (crc_enabled ? MPA_CRC : 0) | - (markers_enabled ? MPA_MARKERS : 0); + (markers_enabled ? MPA_MARKERS : 0) | + (mpa_rev_to_use == 2 ? MPA_ENHANCED_RDMA_CONN : 0); mpa->private_data_size = htons(ep->plen); - mpa->revision = mpa_rev; + mpa->revision = mpa_rev_to_use; + if (mpa_rev_to_use == 1) { + ep->tried_with_mpa_v1 = 1; + ep->retry_with_mpa_v1 = 0; + } + + if (mpa_rev_to_use == 2) { + mpa->private_data_size += + htons(sizeof(struct mpa_v2_conn_params)); + mpa_v2_params.ird = htons((u16)ep->ird); + mpa_v2_params.ord = htons((u16)ep->ord); + + if (peer2peer) { + mpa_v2_params.ird |= htons(MPA_V2_PEER2PEER_MODEL); + if (p2p_type == FW_RI_INIT_P2PTYPE_RDMA_WRITE) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_WRITE_RTR); + else if (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_READ_RTR); + } + memcpy(mpa->private_data, &mpa_v2_params, + sizeof(struct mpa_v2_conn_params)); - if (ep->plen) - memcpy(mpa->private_data, ep->mpa_pkt + sizeof(*mpa), ep->plen); + if (ep->plen) + memcpy(mpa->private_data + + sizeof(struct mpa_v2_conn_params), + ep->mpa_pkt + sizeof(*mpa), ep->plen); + } else + if (ep->plen) + memcpy(mpa->private_data, + ep->mpa_pkt + sizeof(*mpa), ep->plen); /* * Reference the mpa skb. This ensures the data area @@ -562,10 +596,13 @@ static int send_mpa_reject(struct c4iw_ep *ep, const void *pdata, u8 plen) struct fw_ofld_tx_data_wr *req; struct mpa_message *mpa; struct sk_buff *skb; + struct mpa_v2_conn_params mpa_v2_params; PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen); mpalen = sizeof(*mpa) + plen; + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) + mpalen += sizeof(struct mpa_v2_conn_params); wrlen = roundup(mpalen + sizeof *req, 16); skb = get_skb(NULL, wrlen, GFP_KERNEL); @@ -595,8 +632,29 @@ static int send_mpa_reject(struct c4iw_ep *ep, const void *pdata, u8 plen) mpa->flags = MPA_REJECT; mpa->revision = mpa_rev; mpa->private_data_size = htons(plen); - if (plen) - memcpy(mpa->private_data, pdata, plen); + + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) { + mpa->flags |= MPA_ENHANCED_RDMA_CONN; + mpa->private_data_size += + htons(sizeof(struct mpa_v2_conn_params)); + mpa_v2_params.ird = htons(((u16)ep->ird) | + (peer2peer ? MPA_V2_PEER2PEER_MODEL : + 0)); + mpa_v2_params.ord = htons(((u16)ep->ord) | (peer2peer ? + (p2p_type == + FW_RI_INIT_P2PTYPE_RDMA_WRITE ? + MPA_V2_RDMA_WRITE_RTR : p2p_type == + FW_RI_INIT_P2PTYPE_READ_REQ ? + MPA_V2_RDMA_READ_RTR : 0) : 0)); + memcpy(mpa->private_data, &mpa_v2_params, + sizeof(struct mpa_v2_conn_params)); + + if (ep->plen) + memcpy(mpa->private_data + + sizeof(struct mpa_v2_conn_params), pdata, plen); + } else + if (plen) + memcpy(mpa->private_data, pdata, plen); /* * Reference the mpa skb again. This ensures the data area @@ -617,10 +675,13 @@ static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen) struct fw_ofld_tx_data_wr *req; struct mpa_message *mpa; struct sk_buff *skb; + struct mpa_v2_conn_params mpa_v2_params; PDBG("%s ep %p tid %u pd_len %d\n", __func__, ep, ep->hwtid, ep->plen); mpalen = sizeof(*mpa) + plen; + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) + mpalen += sizeof(struct mpa_v2_conn_params); wrlen = roundup(mpalen + sizeof *req, 16); skb = get_skb(NULL, wrlen, GFP_KERNEL); @@ -649,10 +710,36 @@ static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen) memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key)); mpa->flags = (ep->mpa_attr.crc_enabled ? MPA_CRC : 0) | (markers_enabled ? MPA_MARKERS : 0); - mpa->revision = mpa_rev; + mpa->revision = ep->mpa_attr.version; mpa->private_data_size = htons(plen); - if (plen) - memcpy(mpa->private_data, pdata, plen); + + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) { + mpa->flags |= MPA_ENHANCED_RDMA_CONN; + mpa->private_data_size += + htons(sizeof(struct mpa_v2_conn_params)); + mpa_v2_params.ird = htons((u16)ep->ird); + mpa_v2_params.ord = htons((u16)ep->ord); + if (peer2peer && (ep->mpa_attr.p2p_type != + FW_RI_INIT_P2PTYPE_DISABLED)) { + mpa_v2_params.ird |= htons(MPA_V2_PEER2PEER_MODEL); + + if (p2p_type == FW_RI_INIT_P2PTYPE_RDMA_WRITE) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_WRITE_RTR); + else if (p2p_type == FW_RI_INIT_P2PTYPE_READ_REQ) + mpa_v2_params.ord |= + htons(MPA_V2_RDMA_READ_RTR); + } + + memcpy(mpa->private_data, &mpa_v2_params, + sizeof(struct mpa_v2_conn_params)); + + if (ep->plen) + memcpy(mpa->private_data + + sizeof(struct mpa_v2_conn_params), pdata, plen); + } else + if (plen) + memcpy(mpa->private_data, pdata, plen); /* * Reference the mpa skb. This ensures the data area @@ -695,7 +782,10 @@ static int act_establish(struct c4iw_dev *dev, struct sk_buff *skb) /* start MPA negotiation */ send_flowc(ep, NULL); - send_mpa_req(ep, skb); + if (ep->retry_with_mpa_v1) + send_mpa_req(ep, skb, 1); + else + send_mpa_req(ep, skb, mpa_rev); return 0; } @@ -769,8 +859,19 @@ static void connect_reply_upcall(struct c4iw_ep *ep, int status) event.remote_addr = ep->com.remote_addr; if ((status == 0) || (status == -ECONNREFUSED)) { - event.private_data_len = ep->plen; - event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); + if (!ep->tried_with_mpa_v1) { + /* this means MPA_v2 is used */ + event.private_data_len = ep->plen - + sizeof(struct mpa_v2_conn_params); + event.private_data = ep->mpa_pkt + + sizeof(struct mpa_message) + + sizeof(struct mpa_v2_conn_params); + } else { + /* this means MPA_v1 is used */ + event.private_data_len = ep->plen; + event.private_data = ep->mpa_pkt + + sizeof(struct mpa_message); + } } PDBG("%s ep %p tid %u status %d\n", __func__, ep, @@ -793,9 +894,22 @@ static void connect_request_upcall(struct c4iw_ep *ep) event.event = IW_CM_EVENT_CONNECT_REQUEST; event.local_addr = ep->com.local_addr; event.remote_addr = ep->com.remote_addr; - event.private_data_len = ep->plen; - event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); event.provider_data = ep; + if (!ep->tried_with_mpa_v1) { + /* this means MPA_v2 is used */ + event.ord = ep->ord; + event.ird = ep->ird; + event.private_data_len = ep->plen - + sizeof(struct mpa_v2_conn_params); + event.private_data = ep->mpa_pkt + sizeof(struct mpa_message) + + sizeof(struct mpa_v2_conn_params); + } else { + /* this means MPA_v1 is used. Send max supported */ + event.ord = c4iw_max_read_depth; + event.ird = c4iw_max_read_depth; + event.private_data_len = ep->plen; + event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); + } if (state_read(&ep->parent_ep->com) != DEAD) { c4iw_get_ep(&ep->com); ep->parent_ep->com.cm_id->event_handler( @@ -813,6 +927,8 @@ static void established_upcall(struct c4iw_ep *ep) PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); memset(&event, 0, sizeof(event)); event.event = IW_CM_EVENT_ESTABLISHED; + event.ird = ep->ird; + event.ord = ep->ord; if (ep->com.cm_id) { PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); ep->com.cm_id->event_handler(ep->com.cm_id, &event); @@ -848,7 +964,10 @@ static int update_rx_credits(struct c4iw_ep *ep, u32 credits) static void process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb) { struct mpa_message *mpa; + struct mpa_v2_conn_params *mpa_v2_params; u16 plen; + u16 resp_ird, resp_ord; + u8 rtr_mismatch = 0, insuff_ird = 0; struct c4iw_qp_attributes attrs; enum c4iw_qp_attr_mask mask; int err; @@ -888,7 +1007,9 @@ static void process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb) mpa = (struct mpa_message *) ep->mpa_pkt; /* Validate MPA header. */ - if (mpa->revision != mpa_rev) { + if (mpa->revision > mpa_rev) { + printk(KERN_ERR MOD "%s MPA version mismatch. Local = %d," + " Received = %d\n", __func__, mpa_rev, mpa->revision); err = -EPROTO; goto err; } @@ -938,13 +1059,66 @@ static void process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb) ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0; ep->mpa_attr.recv_marker_enabled = markers_enabled; ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0; - ep->mpa_attr.version = mpa_rev; - ep->mpa_attr.p2p_type = peer2peer ? p2p_type : - FW_RI_INIT_P2PTYPE_DISABLED; + ep->mpa_attr.version = mpa->revision; + ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED; + + if (mpa->revision == 2) { + ep->mpa_attr.enhanced_rdma_conn = + mpa->flags & MPA_ENHANCED_RDMA_CONN ? 1 : 0; + if (ep->mpa_attr.enhanced_rdma_conn) { + mpa_v2_params = (struct mpa_v2_conn_params *) + (ep->mpa_pkt + sizeof(*mpa)); + resp_ird = ntohs(mpa_v2_params->ird) & + MPA_V2_IRD_ORD_MASK; + resp_ord = ntohs(mpa_v2_params->ord) & + MPA_V2_IRD_ORD_MASK; + + /* + * This is a double-check. Ideally, below checks are + * not required since ird/ord stuff has been taken + * care of in c4iw_accept_cr + */ + if ((ep->ird < resp_ord) || (ep->ord > resp_ird)) { + err = -ENOMEM; + ep->ird = resp_ord; + ep->ord = resp_ird; + insuff_ird = 1; + } + + if (ntohs(mpa_v2_params->ird) & + MPA_V2_PEER2PEER_MODEL) { + if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_WRITE_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_RDMA_WRITE; + else if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_READ_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_READ_REQ; + } + } + } else if (mpa->revision == 1) + if (peer2peer) + ep->mpa_attr.p2p_type = p2p_type; + PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, " - "xmit_marker_enabled=%d, version=%d\n", __func__, - ep->mpa_attr.crc_enabled, ep->mpa_attr.recv_marker_enabled, - ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version); + "xmit_marker_enabled=%d, version=%d p2p_type=%d local-p2p_type = " + "%d\n", __func__, ep->mpa_attr.crc_enabled, + ep->mpa_attr.recv_marker_enabled, + ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version, + ep->mpa_attr.p2p_type, p2p_type); + + /* + * If responder's RTR does not match with that of initiator, assign + * FW_RI_INIT_P2PTYPE_DISABLED in mpa attributes so that RTR is not + * generated when moving QP to RTS state. + * A TERM message will be sent after QP has moved to RTS state + */ + if ((ep->mpa_attr.version == 2) && + (ep->mpa_attr.p2p_type != p2p_type)) { + ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED; + rtr_mismatch = 1; + } attrs.mpa_attr = ep->mpa_attr; attrs.max_ird = ep->ird; @@ -961,6 +1135,39 @@ static void process_mpa_reply(struct c4iw_ep *ep, struct sk_buff *skb) ep->com.qp, mask, &attrs, 1); if (err) goto err; + + /* + * If responder's RTR requirement did not match with what initiator + * supports, generate TERM message + */ + if (rtr_mismatch) { + printk(KERN_ERR "%s: RTR mismatch, sending TERM\n", __func__); + attrs.layer_etype = LAYER_MPA | DDP_LLP; + attrs.ecode = MPA_NOMATCH_RTR; + attrs.next_state = C4IW_QP_STATE_TERMINATE; + err = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 0); + err = -ENOMEM; + goto out; + } + + /* + * Generate TERM if initiator IRD is not sufficient for responder + * provided ORD. Currently, we do the same behaviour even when + * responder provided IRD is also not sufficient as regards to + * initiator ORD. + */ + if (insuff_ird) { + printk(KERN_ERR "%s: Insufficient IRD, sending TERM\n", + __func__); + attrs.layer_etype = LAYER_MPA | DDP_LLP; + attrs.ecode = MPA_INSUFF_IRD; + attrs.next_state = C4IW_QP_STATE_TERMINATE; + err = c4iw_modify_qp(ep->com.qp->rhp, ep->com.qp, + C4IW_QP_ATTR_NEXT_STATE, &attrs, 0); + err = -ENOMEM; + goto out; + } goto out; err: state_set(&ep->com, ABORTING); @@ -973,6 +1180,7 @@ out: static void process_mpa_request(struct c4iw_ep *ep, struct sk_buff *skb) { struct mpa_message *mpa; + struct mpa_v2_conn_params *mpa_v2_params; u16 plen; PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid); @@ -1013,7 +1221,9 @@ static void process_mpa_request(struct c4iw_ep *ep, struct sk_buff *skb) /* * Validate MPA Header. */ - if (mpa->revision != mpa_rev) { + if (mpa->revision > mpa_rev) { + printk(KERN_ERR MOD "%s MPA version mismatch. Local = %d," + " Received = %d\n", __func__, mpa_rev, mpa->revision); abort_connection(ep, skb, GFP_KERNEL); return; } @@ -1056,9 +1266,37 @@ static void process_mpa_request(struct c4iw_ep *ep, struct sk_buff *skb) ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0; ep->mpa_attr.recv_marker_enabled = markers_enabled; ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0; - ep->mpa_attr.version = mpa_rev; - ep->mpa_attr.p2p_type = peer2peer ? p2p_type : - FW_RI_INIT_P2PTYPE_DISABLED; + ep->mpa_attr.version = mpa->revision; + if (mpa->revision == 1) + ep->tried_with_mpa_v1 = 1; + ep->mpa_attr.p2p_type = FW_RI_INIT_P2PTYPE_DISABLED; + + if (mpa->revision == 2) { + ep->mpa_attr.enhanced_rdma_conn = + mpa->flags & MPA_ENHANCED_RDMA_CONN ? 1 : 0; + if (ep->mpa_attr.enhanced_rdma_conn) { + mpa_v2_params = (struct mpa_v2_conn_params *) + (ep->mpa_pkt + sizeof(*mpa)); + ep->ird = ntohs(mpa_v2_params->ird) & + MPA_V2_IRD_ORD_MASK; + ep->ord = ntohs(mpa_v2_params->ord) & + MPA_V2_IRD_ORD_MASK; + if (ntohs(mpa_v2_params->ird) & MPA_V2_PEER2PEER_MODEL) + if (peer2peer) { + if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_WRITE_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_RDMA_WRITE; + else if (ntohs(mpa_v2_params->ord) & + MPA_V2_RDMA_READ_RTR) + ep->mpa_attr.p2p_type = + FW_RI_INIT_P2PTYPE_READ_REQ; + } + } + } else if (mpa->revision == 1) + if (peer2peer) + ep->mpa_attr.p2p_type = p2p_type; + PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, " "xmit_marker_enabled=%d, version=%d p2p_type=%d\n", __func__, ep->mpa_attr.crc_enabled, ep->mpa_attr.recv_marker_enabled, @@ -1552,6 +1790,114 @@ static int is_neg_adv_abort(unsigned int status) status == CPL_ERR_PERSIST_NEG_ADVICE; } +static int c4iw_reconnect(struct c4iw_ep *ep) +{ + int err = 0; + struct rtable *rt; + struct net_device *pdev; + struct neighbour *neigh; + int step; + + PDBG("%s qp %p cm_id %p\n", __func__, ep->com.qp, ep->com.cm_id); + init_timer(&ep->timer); + + /* + * Allocate an active TID to initiate a TCP connection. + */ + ep->atid = cxgb4_alloc_atid(ep->com.dev->rdev.lldi.tids, ep); + if (ep->atid == -1) { + printk(KERN_ERR MOD "%s - cannot alloc atid.\n", __func__); + err = -ENOMEM; + goto fail2; + } + + /* find a route */ + rt = find_route(ep->com.dev, + ep->com.cm_id->local_addr.sin_addr.s_addr, + ep->com.cm_id->remote_addr.sin_addr.s_addr, + ep->com.cm_id->local_addr.sin_port, + ep->com.cm_id->remote_addr.sin_port, 0); + if (!rt) { + printk(KERN_ERR MOD "%s - cannot find route.\n", __func__); + err = -EHOSTUNREACH; + goto fail3; + } + ep->dst = &rt->dst; + + rcu_read_lock(); + neigh = dst_get_neighbour(ep->dst); + + /* get a l2t entry */ + if (neigh->dev->flags & IFF_LOOPBACK) { + PDBG("%s LOOPBACK\n", __func__); + pdev = ip_dev_find(&init_net, + ep->com.cm_id->remote_addr.sin_addr.s_addr); + ep->l2t = cxgb4_l2t_get(ep->com.dev->rdev.lldi.l2t, + neigh, pdev, 0); + ep->mtu = pdev->mtu; + ep->tx_chan = cxgb4_port_chan(pdev); + ep->smac_idx = (cxgb4_port_viid(pdev) & 0x7F) << 1; + step = ep->com.dev->rdev.lldi.ntxq / + ep->com.dev->rdev.lldi.nchan; + ep->txq_idx = cxgb4_port_idx(pdev) * step; + step = ep->com.dev->rdev.lldi.nrxq / + ep->com.dev->rdev.lldi.nchan; + ep->ctrlq_idx = cxgb4_port_idx(pdev); + ep->rss_qid = ep->com.dev->rdev.lldi.rxq_ids[ + cxgb4_port_idx(pdev) * step]; + dev_put(pdev); + } else { + ep->l2t = cxgb4_l2t_get(ep->com.dev->rdev.lldi.l2t, + neigh, neigh->dev, 0); + ep->mtu = dst_mtu(ep->dst); + ep->tx_chan = cxgb4_port_chan(neigh->dev); + ep->smac_idx = (cxgb4_port_viid(neigh->dev) & 0x7F) << 1; + step = ep->com.dev->rdev.lldi.ntxq / + ep->com.dev->rdev.lldi.nchan; + ep->txq_idx = cxgb4_port_idx(neigh->dev) * step; + ep->ctrlq_idx = cxgb4_port_idx(neigh->dev); + step = ep->com.dev->rdev.lldi.nrxq / + ep->com.dev->rdev.lldi.nchan; + ep->rss_qid = ep->com.dev->rdev.lldi.rxq_ids[ + cxgb4_port_idx(neigh->dev) * step]; + } + rcu_read_unlock(); + if (!ep->l2t) { + printk(KERN_ERR MOD "%s - cannot alloc l2e.\n", __func__); + err = -ENOMEM; + goto fail4; + } + + PDBG("%s txq_idx %u tx_chan %u smac_idx %u rss_qid %u l2t_idx %u\n", + __func__, ep->txq_idx, ep->tx_chan, ep->smac_idx, ep->rss_qid, + ep->l2t->idx); + + state_set(&ep->com, CONNECTING); + ep->tos = 0; + + /* send connect request to rnic */ + err = send_connect(ep); + if (!err) + goto out; + + cxgb4_l2t_release(ep->l2t); +fail4: + dst_release(ep->dst); +fail3: + cxgb4_free_atid(ep->com.dev->rdev.lldi.tids, ep->atid); +fail2: + /* + * remember to send notification to upper layer. + * We are in here so the upper layer is not aware that this is + * re-connect attempt and so, upper layer is still waiting for + * response of 1st connect request. + */ + connect_reply_upcall(ep, -ECONNRESET); + c4iw_put_ep(&ep->com); +out: + return err; +} + static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) { struct cpl_abort_req_rss *req = cplhdr(skb); @@ -1575,8 +1921,11 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) /* * Wake up any threads in rdma_init() or rdma_fini(). + * However, this is not needed if com state is just + * MPA_REQ_SENT */ - c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); + if (ep->com.state != MPA_REQ_SENT) + c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET); mutex_lock(&ep->com.mutex); switch (ep->com.state) { @@ -1587,7 +1936,21 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) break; case MPA_REQ_SENT: stop_ep_timer(ep); - connect_reply_upcall(ep, -ECONNRESET); + if (mpa_rev == 2 && ep->tried_with_mpa_v1) + connect_reply_upcall(ep, -ECONNRESET); + else { + /* + * we just don't send notification upwards because we + * want to retry with mpa_v1 without upper layers even + * knowing it. + * + * do some housekeeping so as to re-initiate the + * connection + */ + PDBG("%s: mpa_rev=%d. Retrying with mpav1\n", __func__, + mpa_rev); + ep->retry_with_mpa_v1 = 1; + } break; case MPA_REP_SENT: break; @@ -1623,7 +1986,9 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) dst_confirm(ep->dst); if (ep->com.state != ABORTING) { __state_set(&ep->com, DEAD); - release = 1; + /* we don't release if we want to retry with mpa_v1 */ + if (!ep->retry_with_mpa_v1) + release = 1; } mutex_unlock(&ep->com.mutex); @@ -1643,6 +2008,15 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) out: if (release) release_ep_resources(ep); + + /* retry with mpa-v1 */ + if (ep && ep->retry_with_mpa_v1) { + cxgb4_remove_tid(ep->com.dev->rdev.lldi.tids, 0, ep->hwtid); + dst_release(ep->dst); + cxgb4_l2t_release(ep->l2t); + c4iw_reconnect(ep); + } + return 0; } @@ -1794,18 +2168,40 @@ int c4iw_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) goto err; } - cm_id->add_ref(cm_id); - ep->com.cm_id = cm_id; - ep->com.qp = qp; + if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) { + if (conn_param->ord > ep->ird) { + ep->ird = conn_param->ird; + ep->ord = conn_param->ord; + send_mpa_reject(ep, conn_param->private_data, + conn_param->private_data_len); + abort_connection(ep, NULL, GFP_KERNEL); + err = -ENOMEM; + goto err; + } + if (conn_param->ird > ep->ord) { + if (!ep->ord) + conn_param->ird = 1; + else { + abort_connection(ep, NULL, GFP_KERNEL); + err = -ENOMEM; + goto err; + } + } + } ep->ird = conn_param->ird; ep->ord = conn_param->ord; - if (peer2peer && ep->ird == 0) - ep->ird = 1; + if (ep->mpa_attr.version != 2) + if (peer2peer && ep->ird == 0) + ep->ird = 1; PDBG("%s %d ird %d ord %d\n", __func__, __LINE__, ep->ird, ep->ord); + cm_id->add_ref(cm_id); + ep->com.cm_id = cm_id; + ep->com.qp = qp; + /* bind QP to EP and move to RTS */ attrs.mpa_attr = ep->mpa_attr; attrs.max_ird = ep->ird; @@ -1947,6 +2343,8 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) ep->com.dev->rdev.lldi.nchan; ep->rss_qid = ep->com.dev->rdev.lldi.rxq_ids[ cxgb4_port_idx(neigh->dev) * step]; + ep->retry_with_mpa_v1 = 0; + ep->tried_with_mpa_v1 = 0; } rcu_read_unlock(); if (!ep->l2t) { diff --git a/drivers/infiniband/hw/cxgb4/cq.c b/drivers/infiniband/hw/cxgb4/cq.c index 1720dc7..3bd6b69 100644 --- a/drivers/infiniband/hw/cxgb4/cq.c +++ b/drivers/infiniband/hw/cxgb4/cq.c @@ -185,7 +185,7 @@ static void insert_recv_cqe(struct t4_wq *wq, struct t4_cq *cq) V_CQE_OPCODE(FW_RI_SEND) | V_CQE_TYPE(0) | V_CQE_SWCQE(1) | - V_CQE_QPID(wq->rq.qid)); + V_CQE_QPID(wq->sq.qid)); cqe.bits_type_ts = cpu_to_be64(V_CQE_GENBIT((u64)cq->gen)); cq->sw_queue[cq->sw_pidx] = cqe; t4_swcq_produce(cq); @@ -311,7 +311,7 @@ void c4iw_count_rcqes(struct t4_cq *cq, struct t4_wq *wq, int *count) while (ptr != cq->sw_pidx) { cqe = &cq->sw_queue[ptr]; if (RQ_TYPE(cqe) && (CQE_OPCODE(cqe) != FW_RI_READ_RESP) && - (CQE_QPID(cqe) == wq->rq.qid) && cqe_completes_wr(cqe, wq)) + (CQE_QPID(cqe) == wq->sq.qid) && cqe_completes_wr(cqe, wq)) (*count)++; if (++ptr == cq->size) ptr = 0; @@ -818,6 +818,7 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, int entries, chp->cq.size--; /* status page */ chp->ibcq.cqe = entries - 2; spin_lock_init(&chp->lock); + spin_lock_init(&chp->comp_handler_lock); atomic_set(&chp->refcnt, 1); init_waitqueue_head(&chp->wait); ret = insert_handle(rhp, &rhp->cqidr, chp, chp->cq.cqid); @@ -842,7 +843,8 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev, int entries, uresp.gts_key = ucontext->key; ucontext->key += PAGE_SIZE; spin_unlock(&ucontext->mmap_lock); - ret = ib_copy_to_udata(udata, &uresp, sizeof uresp); + ret = ib_copy_to_udata(udata, &uresp, + sizeof(uresp) - sizeof(uresp.reserved)); if (ret) goto err5; diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c index 40a13cc..6d0df6e 100644 --- a/drivers/infiniband/hw/cxgb4/device.c +++ b/drivers/infiniband/hw/cxgb4/device.c @@ -376,10 +376,8 @@ struct uld_ctx { struct c4iw_dev *dev; }; -static void c4iw_remove(struct uld_ctx *ctx) +static void c4iw_dealloc(struct uld_ctx *ctx) { - PDBG("%s c4iw_dev %p\n", __func__, ctx->dev); - c4iw_unregister_device(ctx->dev); c4iw_rdev_close(&ctx->dev->rdev); idr_destroy(&ctx->dev->cqidr); idr_destroy(&ctx->dev->qpidr); @@ -389,11 +387,30 @@ static void c4iw_remove(struct uld_ctx *ctx) ctx->dev = NULL; } +static void c4iw_remove(struct uld_ctx *ctx) +{ + PDBG("%s c4iw_dev %p\n", __func__, ctx->dev); + c4iw_unregister_device(ctx->dev); + c4iw_dealloc(ctx); +} + +static int rdma_supported(const struct cxgb4_lld_info *infop) +{ + return infop->vr->stag.size > 0 && infop->vr->pbl.size > 0 && + infop->vr->rq.size > 0 && infop->vr->qp.size > 0 && + infop->vr->cq.size > 0 && infop->vr->ocq.size > 0; +} + static struct c4iw_dev *c4iw_alloc(const struct cxgb4_lld_info *infop) { struct c4iw_dev *devp; int ret; + if (!rdma_supported(infop)) { + printk(KERN_INFO MOD "%s: RDMA not supported on this device.\n", + pci_name(infop->pdev)); + return ERR_PTR(-ENOSYS); + } devp = (struct c4iw_dev *)ib_alloc_device(sizeof(*devp)); if (!devp) { printk(KERN_ERR MOD "Cannot allocate ib device\n"); @@ -414,7 +431,6 @@ static struct c4iw_dev *c4iw_alloc(const struct cxgb4_lld_info *infop) ret = c4iw_rdev_open(&devp->rdev); if (ret) { - mutex_unlock(&dev_mutex); printk(KERN_ERR MOD "Unable to open CXIO rdev err %d\n", ret); ib_dealloc_device(&devp->ibdev); return ERR_PTR(ret); @@ -519,15 +535,24 @@ static int c4iw_uld_state_change(void *handle, enum cxgb4_state new_state) case CXGB4_STATE_UP: printk(KERN_INFO MOD "%s: Up\n", pci_name(ctx->lldi.pdev)); if (!ctx->dev) { - int ret = 0; + int ret; ctx->dev = c4iw_alloc(&ctx->lldi); - if (!IS_ERR(ctx->dev)) - ret = c4iw_register_device(ctx->dev); - if (IS_ERR(ctx->dev) || ret) + if (IS_ERR(ctx->dev)) { + printk(KERN_ERR MOD + "%s: initialization failed: %ld\n", + pci_name(ctx->lldi.pdev), + PTR_ERR(ctx->dev)); + ctx->dev = NULL; + break; + } + ret = c4iw_register_device(ctx->dev); + if (ret) { printk(KERN_ERR MOD "%s: RDMA registration failed: %d\n", pci_name(ctx->lldi.pdev), ret); + c4iw_dealloc(ctx); + } } break; case CXGB4_STATE_DOWN: diff --git a/drivers/infiniband/hw/cxgb4/ev.c b/drivers/infiniband/hw/cxgb4/ev.c index c13041a..397cb36 100644 --- a/drivers/infiniband/hw/cxgb4/ev.c +++ b/drivers/infiniband/hw/cxgb4/ev.c @@ -42,6 +42,7 @@ static void post_qp_event(struct c4iw_dev *dev, struct c4iw_cq *chp, { struct ib_event event; struct c4iw_qp_attributes attrs; + unsigned long flag; if ((qhp->attr.state == C4IW_QP_STATE_ERROR) || (qhp->attr.state == C4IW_QP_STATE_TERMINATE)) { @@ -72,7 +73,9 @@ static void post_qp_event(struct c4iw_dev *dev, struct c4iw_cq *chp, if (qhp->ibqp.event_handler) (*qhp->ibqp.event_handler)(&event, qhp->ibqp.qp_context); + spin_lock_irqsave(&chp->comp_handler_lock, flag); (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); + spin_unlock_irqrestore(&chp->comp_handler_lock, flag); } void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe) @@ -183,11 +186,14 @@ out: int c4iw_ev_handler(struct c4iw_dev *dev, u32 qid) { struct c4iw_cq *chp; + unsigned long flag; chp = get_chp(dev, qid); - if (chp) + if (chp) { + spin_lock_irqsave(&chp->comp_handler_lock, flag); (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); - else + spin_unlock_irqrestore(&chp->comp_handler_lock, flag); + } else PDBG("%s unknown cqid 0x%x\n", __func__, qid); return 0; } diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h index 4f04537..1357c5b 100644 --- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h +++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h @@ -309,6 +309,7 @@ struct c4iw_cq { struct c4iw_dev *rhp; struct t4_cq cq; spinlock_t lock; + spinlock_t comp_handler_lock; atomic_t refcnt; wait_queue_head_t wait; }; @@ -323,6 +324,7 @@ struct c4iw_mpa_attributes { u8 recv_marker_enabled; u8 xmit_marker_enabled; u8 crc_enabled; + u8 enhanced_rdma_conn; u8 version; u8 p2p_type; }; @@ -349,6 +351,8 @@ struct c4iw_qp_attributes { u8 is_terminate_local; struct c4iw_mpa_attributes mpa_attr; struct c4iw_ep *llp_stream_handle; + u8 layer_etype; + u8 ecode; }; struct c4iw_qp { @@ -501,11 +505,18 @@ enum c4iw_mmid_state { #define MPA_KEY_REP "MPA ID Rep Frame" #define MPA_MAX_PRIVATE_DATA 256 +#define MPA_ENHANCED_RDMA_CONN 0x10 #define MPA_REJECT 0x20 #define MPA_CRC 0x40 #define MPA_MARKERS 0x80 #define MPA_FLAGS_MASK 0xE0 +#define MPA_V2_PEER2PEER_MODEL 0x8000 +#define MPA_V2_ZERO_LEN_FPDU_RTR 0x4000 +#define MPA_V2_RDMA_WRITE_RTR 0x8000 +#define MPA_V2_RDMA_READ_RTR 0x4000 +#define MPA_V2_IRD_ORD_MASK 0x3FFF + #define c4iw_put_ep(ep) { \ PDBG("put_ep (via %s:%u) ep %p refcnt %d\n", __func__, __LINE__, \ ep, atomic_read(&((ep)->kref.refcount))); \ @@ -528,6 +539,11 @@ struct mpa_message { u8 private_data[0]; }; +struct mpa_v2_conn_params { + __be16 ird; + __be16 ord; +}; + struct terminate_message { u8 layer_etype; u8 ecode; @@ -580,7 +596,10 @@ enum c4iw_ddp_ecodes { enum c4iw_mpa_ecodes { MPA_CRC_ERR = 0x02, - MPA_MARKER_ERR = 0x03 + MPA_MARKER_ERR = 0x03, + MPA_LOCAL_CATA = 0x05, + MPA_INSUFF_IRD = 0x06, + MPA_NOMATCH_RTR = 0x07, }; enum c4iw_ep_state { @@ -651,6 +670,8 @@ struct c4iw_ep { u16 txq_idx; u16 ctrlq_idx; u8 tos; + u8 retry_with_mpa_v1; + u8 tried_with_mpa_v1; }; static inline struct c4iw_ep *to_ep(struct iw_cm_id *cm_id) diff --git a/drivers/infiniband/hw/cxgb4/mem.c b/drivers/infiniband/hw/cxgb4/mem.c index 0347eed..40c8353 100644 --- a/drivers/infiniband/hw/cxgb4/mem.c +++ b/drivers/infiniband/hw/cxgb4/mem.c @@ -31,7 +31,7 @@ */ #include <rdma/ib_umem.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include "iw_cxgb4.h" diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c index 5b9e422..247fe70 100644 --- a/drivers/infiniband/hw/cxgb4/provider.c +++ b/drivers/infiniband/hw/cxgb4/provider.c @@ -58,13 +58,6 @@ static int fastreg_support = 1; module_param(fastreg_support, int, 0644); MODULE_PARM_DESC(fastreg_support, "Advertise fastreg support (default=1)"); -static int c4iw_modify_port(struct ib_device *ibdev, - u8 port, int port_modify_mask, - struct ib_port_modify *props) -{ - return -ENOSYS; -} - static struct ib_ah *c4iw_ah_create(struct ib_pd *pd, struct ib_ah_attr *ah_attr) { @@ -456,7 +449,6 @@ int c4iw_register_device(struct c4iw_dev *dev) dev->ibdev.dma_device = &(dev->rdev.lldi.pdev->dev); dev->ibdev.query_device = c4iw_query_device; dev->ibdev.query_port = c4iw_query_port; - dev->ibdev.modify_port = c4iw_modify_port; dev->ibdev.query_pkey = c4iw_query_pkey; dev->ibdev.query_gid = c4iw_query_gid; dev->ibdev.alloc_ucontext = c4iw_alloc_ucontext; diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c index a41578e..5f940ae 100644 --- a/drivers/infiniband/hw/cxgb4/qp.c +++ b/drivers/infiniband/hw/cxgb4/qp.c @@ -29,6 +29,9 @@ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ + +#include <linux/module.h> + #include "iw_cxgb4.h" static int ocqp_support = 1; @@ -917,7 +920,11 @@ static void post_terminate(struct c4iw_qp *qhp, struct t4_cqe *err_cqe, wqe->u.terminate.type = FW_RI_TYPE_TERMINATE; wqe->u.terminate.immdlen = cpu_to_be32(sizeof *term); term = (struct terminate_message *)wqe->u.terminate.termmsg; - build_term_codes(err_cqe, &term->layer_etype, &term->ecode); + if (qhp->attr.layer_etype == (LAYER_MPA|DDP_LLP)) { + term->layer_etype = qhp->attr.layer_etype; + term->ecode = qhp->attr.ecode; + } else + build_term_codes(err_cqe, &term->layer_etype, &term->ecode); c4iw_ofld_send(&qhp->rhp->rdev, skb); } @@ -941,8 +948,11 @@ static void __flush_qp(struct c4iw_qp *qhp, struct c4iw_cq *rchp, flushed = c4iw_flush_rq(&qhp->wq, &rchp->cq, count); spin_unlock(&qhp->lock); spin_unlock_irqrestore(&rchp->lock, flag); - if (flushed) + if (flushed) { + spin_lock_irqsave(&rchp->comp_handler_lock, flag); (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + } /* locking hierarchy: cq lock first, then qp lock. */ spin_lock_irqsave(&schp->lock, flag); @@ -952,13 +962,17 @@ static void __flush_qp(struct c4iw_qp *qhp, struct c4iw_cq *rchp, flushed = c4iw_flush_sq(&qhp->wq, &schp->cq, count); spin_unlock(&qhp->lock); spin_unlock_irqrestore(&schp->lock, flag); - if (flushed) + if (flushed) { + spin_lock_irqsave(&schp->comp_handler_lock, flag); (*schp->ibcq.comp_handler)(&schp->ibcq, schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, flag); + } } static void flush_qp(struct c4iw_qp *qhp) { struct c4iw_cq *rchp, *schp; + unsigned long flag; rchp = get_chp(qhp->rhp, qhp->attr.rcq); schp = get_chp(qhp->rhp, qhp->attr.scq); @@ -966,8 +980,16 @@ static void flush_qp(struct c4iw_qp *qhp) if (qhp->ibqp.uobject) { t4_set_wq_in_error(&qhp->wq); t4_set_cq_in_error(&rchp->cq); - if (schp != rchp) + spin_lock_irqsave(&rchp->comp_handler_lock, flag); + (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + if (schp != rchp) { t4_set_cq_in_error(&schp->cq); + spin_lock_irqsave(&schp->comp_handler_lock, flag); + (*schp->ibcq.comp_handler)(&schp->ibcq, + schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, flag); + } return; } __flush_qp(qhp, rchp, schp); @@ -1012,6 +1034,7 @@ out: static void build_rtr_msg(u8 p2p_type, struct fw_ri_init *init) { + PDBG("%s p2p_type = %d\n", __func__, p2p_type); memset(&init->u, 0, sizeof init->u); switch (p2p_type) { case FW_RI_INIT_P2PTYPE_RDMA_WRITE: @@ -1206,12 +1229,16 @@ int c4iw_modify_qp(struct c4iw_dev *rhp, struct c4iw_qp *qhp, disconnect = 1; c4iw_get_ep(&qhp->ep->com); } + if (qhp->ibqp.uobject) + t4_set_wq_in_error(&qhp->wq); ret = rdma_fini(rhp, qhp, ep); if (ret) goto err; break; case C4IW_QP_STATE_TERMINATE: set_state(qhp, C4IW_QP_STATE_TERMINATE); + qhp->attr.layer_etype = attrs->layer_etype; + qhp->attr.ecode = attrs->ecode; if (qhp->ibqp.uobject) t4_set_wq_in_error(&qhp->wq); ep = qhp->ep; @@ -1222,6 +1249,8 @@ int c4iw_modify_qp(struct c4iw_dev *rhp, struct c4iw_qp *qhp, break; case C4IW_QP_STATE_ERROR: set_state(qhp, C4IW_QP_STATE_ERROR); + if (qhp->ibqp.uobject) + t4_set_wq_in_error(&qhp->wq); if (!internal) { abort = 1; disconnect = 1; @@ -1334,7 +1363,10 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp) rhp = qhp->rhp; attrs.next_state = C4IW_QP_STATE_ERROR; - c4iw_modify_qp(rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 0); + if (qhp->attr.state == C4IW_QP_STATE_TERMINATE) + c4iw_modify_qp(rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 1); + else + c4iw_modify_qp(rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 0); wait_event(qhp->wait, !qhp->ep); remove_handle(rhp, &rhp->qpidr, qhp->wq.sq.qid); diff --git a/drivers/infiniband/hw/cxgb4/resource.c b/drivers/infiniband/hw/cxgb4/resource.c index 4fb50d5..407ff39 100644 --- a/drivers/infiniband/hw/cxgb4/resource.c +++ b/drivers/infiniband/hw/cxgb4/resource.c @@ -37,6 +37,7 @@ #include <linux/spinlock.h> #include <linux/errno.h> #include <linux/genalloc.h> +#include <linux/ratelimit.h> #include "iw_cxgb4.h" #define RANDOM_SIZE 16 @@ -311,8 +312,8 @@ u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size) { unsigned long addr = gen_pool_alloc(rdev->pbl_pool, size); PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size); - if (!addr && printk_ratelimit()) - printk(KERN_WARNING MOD "%s: Out of PBL memory\n", + if (!addr) + printk_ratelimited(KERN_WARNING MOD "%s: Out of PBL memory\n", pci_name(rdev->lldi.pdev)); return (u32)addr; } @@ -373,8 +374,8 @@ u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size) { unsigned long addr = gen_pool_alloc(rdev->rqt_pool, size << 6); PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size << 6); - if (!addr && printk_ratelimit()) - printk(KERN_WARNING MOD "%s: Out of RQT memory\n", + if (!addr) + printk_ratelimited(KERN_WARNING MOD "%s: Out of RQT memory\n", pci_name(rdev->lldi.pdev)); return (u32)addr; } diff --git a/drivers/infiniband/hw/cxgb4/user.h b/drivers/infiniband/hw/cxgb4/user.h index e6669d5..06e8938 100644 --- a/drivers/infiniband/hw/cxgb4/user.h +++ b/drivers/infiniband/hw/cxgb4/user.h @@ -48,6 +48,7 @@ struct c4iw_create_cq_resp { __u32 cqid; __u32 size; __u32 qid_mask; + __u32 reserved; /* explicit padding (optional for i386) */ }; diff --git a/drivers/infiniband/hw/ehca/ehca_cq.c b/drivers/infiniband/hw/ehca/ehca_cq.c index d9b0ebc..6eeb84d 100644 --- a/drivers/infiniband/hw/ehca/ehca_cq.c +++ b/drivers/infiniband/hw/ehca/ehca_cq.c @@ -296,6 +296,7 @@ struct ib_cq *ehca_create_cq(struct ib_device *device, int cqe, int comp_vector, (my_cq->galpas.user.fw_handle & (PAGE_SIZE - 1)); if (ib_copy_to_udata(udata, &resp, sizeof(resp))) { ehca_err(device, "Copy to udata failed."); + cq = ERR_PTR(-EFAULT); goto create_cq_exit4; } } diff --git a/drivers/infiniband/hw/ehca/ehca_eq.c b/drivers/infiniband/hw/ehca/ehca_eq.c index d9b1bb4..818d721 100644 --- a/drivers/infiniband/hw/ehca/ehca_eq.c +++ b/drivers/infiniband/hw/ehca/ehca_eq.c @@ -125,7 +125,7 @@ int ehca_create_eq(struct ehca_shca *shca, tasklet_init(&eq->interrupt_task, ehca_tasklet_eq, (long)shca); ret = ibmebus_request_irq(eq->ist, ehca_interrupt_eq, - IRQF_DISABLED, "ehca_eq", + 0, "ehca_eq", (void *)shca); if (ret < 0) ehca_err(ib_dev, "Can't map interrupt handler."); @@ -133,7 +133,7 @@ int ehca_create_eq(struct ehca_shca *shca, tasklet_init(&eq->interrupt_task, ehca_tasklet_neq, (long)shca); ret = ibmebus_request_irq(eq->ist, ehca_interrupt_neq, - IRQF_DISABLED, "ehca_neq", + 0, "ehca_neq", (void *)shca); if (ret < 0) ehca_err(ib_dev, "Can't map interrupt handler."); diff --git a/drivers/infiniband/hw/ehca/ehca_qp.c b/drivers/infiniband/hw/ehca/ehca_qp.c index 32fb342..964f855 100644 --- a/drivers/infiniband/hw/ehca/ehca_qp.c +++ b/drivers/infiniband/hw/ehca/ehca_qp.c @@ -977,6 +977,9 @@ struct ib_srq *ehca_create_srq(struct ib_pd *pd, struct hcp_modify_qp_control_block *mqpcb; u64 hret, update_mask; + if (srq_init_attr->srq_type != IB_SRQT_BASIC) + return ERR_PTR(-ENOSYS); + /* For common attributes, internal_create_qp() takes its info * out of qp_init_attr, so copy all common attrs there. */ diff --git a/drivers/infiniband/hw/ehca/ehca_tools.h b/drivers/infiniband/hw/ehca/ehca_tools.h index f09914c..54c0d23 100644 --- a/drivers/infiniband/hw/ehca/ehca_tools.h +++ b/drivers/infiniband/hw/ehca/ehca_tools.h @@ -58,7 +58,7 @@ #include <linux/cpu.h> #include <linux/device.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include <asm/abs_addr.h> #include <asm/ibmebus.h> #include <asm/io.h> diff --git a/drivers/infiniband/hw/ipath/ipath_diag.c b/drivers/infiniband/hw/ipath/ipath_diag.c index daef61d..45802e9 100644 --- a/drivers/infiniband/hw/ipath/ipath_diag.c +++ b/drivers/infiniband/hw/ipath/ipath_diag.c @@ -45,6 +45,7 @@ #include <linux/pci.h> #include <linux/vmalloc.h> #include <linux/fs.h> +#include <linux/export.h> #include <asm/uaccess.h> #include "ipath_kernel.h" @@ -325,7 +326,7 @@ static ssize_t ipath_diagpkt_write(struct file *fp, size_t count, loff_t *off) { u32 __iomem *piobuf; - u32 plen, clen, pbufn; + u32 plen, pbufn, maxlen_reserve; struct ipath_diag_pkt odp; struct ipath_diag_xpkt dp; u32 *tmpbuf = NULL; @@ -334,42 +335,24 @@ static ssize_t ipath_diagpkt_write(struct file *fp, u64 val; u32 l_state, lt_state; /* LinkState, LinkTrainingState */ - if (count < sizeof(odp)) { - ret = -EINVAL; - goto bail; - } if (count == sizeof(dp)) { if (copy_from_user(&dp, data, sizeof(dp))) { ret = -EFAULT; goto bail; } - } else if (copy_from_user(&odp, data, sizeof(odp))) { - ret = -EFAULT; - goto bail; - } - - /* - * Due to padding/alignment issues (lessened with new struct) - * the old and new structs are the same length. We need to - * disambiguate them, which we can do because odp.len has never - * been less than the total of LRH+BTH+DETH so far, while - * dp.unit (same offset) unit is unlikely to get that high. - * Similarly, dp.data, the pointer to user at the same offset - * as odp.unit, is almost certainly at least one (512byte)page - * "above" NULL. The if-block below can be omitted if compatibility - * between a new driver and older diagnostic code is unimportant. - * compatibility the other direction (new diags, old driver) is - * handled in the diagnostic code, with a warning. - */ - if (dp.unit >= 20 && dp.data < 512) { - /* very probable version mismatch. Fix it up */ - memcpy(&odp, &dp, sizeof(odp)); - /* We got a legacy dp, copy elements to dp */ + } else if (count == sizeof(odp)) { + if (copy_from_user(&odp, data, sizeof(odp))) { + ret = -EFAULT; + goto bail; + } + dp.len = odp.len; dp.unit = odp.unit; dp.data = odp.data; - dp.len = odp.len; - dp.pbc_wd = 0; /* Indicate we need to compute PBC wd */ + dp.pbc_wd = 0; + } else { + ret = -EINVAL; + goto bail; } /* send count must be an exact number of dwords */ @@ -378,7 +361,7 @@ static ssize_t ipath_diagpkt_write(struct file *fp, goto bail; } - clen = dp.len >> 2; + plen = dp.len >> 2; dd = ipath_lookup(dp.unit); if (!dd || !(dd->ipath_flags & IPATH_PRESENT) || @@ -421,16 +404,22 @@ static ssize_t ipath_diagpkt_write(struct file *fp, goto bail; } - /* need total length before first word written */ - /* +1 word is for the qword padding */ - plen = sizeof(u32) + dp.len; - - if ((plen + 4) > dd->ipath_ibmaxlen) { + /* + * need total length before first word written, plus 2 Dwords. One Dword + * is for padding so we get the full user data when not aligned on + * a word boundary. The other Dword is to make sure we have room for the + * ICRC which gets tacked on later. + */ + maxlen_reserve = 2 * sizeof(u32); + if (dp.len > dd->ipath_ibmaxlen - maxlen_reserve) { ipath_dbg("Pkt len 0x%x > ibmaxlen %x\n", - plen - 4, dd->ipath_ibmaxlen); + dp.len, dd->ipath_ibmaxlen); ret = -EINVAL; - goto bail; /* before writing pbc */ + goto bail; } + + plen = sizeof(u32) + dp.len; + tmpbuf = vmalloc(plen); if (!tmpbuf) { dev_info(&dd->pcidev->dev, "Unable to allocate tmp buffer, " @@ -472,11 +461,11 @@ static ssize_t ipath_diagpkt_write(struct file *fp, */ if (dd->ipath_flags & IPATH_PIO_FLUSH_WC) { ipath_flush_wc(); - __iowrite32_copy(piobuf + 2, tmpbuf, clen - 1); + __iowrite32_copy(piobuf + 2, tmpbuf, plen - 1); ipath_flush_wc(); - __raw_writel(tmpbuf[clen - 1], piobuf + clen + 1); + __raw_writel(tmpbuf[plen - 1], piobuf + plen + 1); } else - __iowrite32_copy(piobuf + 2, tmpbuf, clen); + __iowrite32_copy(piobuf + 2, tmpbuf, plen); ipath_flush_wc(); diff --git a/drivers/infiniband/hw/ipath/ipath_driver.c b/drivers/infiniband/hw/ipath/ipath_driver.c index be24ac7..bfca37b 100644 --- a/drivers/infiniband/hw/ipath/ipath_driver.c +++ b/drivers/infiniband/hw/ipath/ipath_driver.c @@ -41,6 +41,7 @@ #include <linux/vmalloc.h> #include <linux/bitmap.h> #include <linux/slab.h> +#include <linux/module.h> #include "ipath_kernel.h" #include "ipath_verbs.h" diff --git a/drivers/infiniband/hw/ipath/ipath_file_ops.c b/drivers/infiniband/hw/ipath/ipath_file_ops.c index ee79a2d..736d9ed 100644 --- a/drivers/infiniband/hw/ipath/ipath_file_ops.c +++ b/drivers/infiniband/hw/ipath/ipath_file_ops.c @@ -35,11 +35,13 @@ #include <linux/poll.h> #include <linux/cdev.h> #include <linux/swap.h> +#include <linux/export.h> #include <linux/vmalloc.h> #include <linux/slab.h> #include <linux/highmem.h> #include <linux/io.h> #include <linux/jiffies.h> +#include <linux/cpu.h> #include <asm/pgtable.h> #include "ipath_kernel.h" @@ -1684,17 +1686,19 @@ static int find_best_unit(struct file *fp, * information. There may be some issues with dual core numbering * as well. This needs more work prior to release. */ - if (!cpumask_empty(¤t->cpus_allowed) && - !cpumask_full(¤t->cpus_allowed)) { + if (!cpumask_empty(tsk_cpus_allowed(current)) && + !cpumask_full(tsk_cpus_allowed(current))) { int ncpus = num_online_cpus(), curcpu = -1, nset = 0; - for (i = 0; i < ncpus; i++) - if (cpumask_test_cpu(i, ¤t->cpus_allowed)) { + get_online_cpus(); + for_each_online_cpu(i) + if (cpumask_test_cpu(i, tsk_cpus_allowed(current))) { ipath_cdbg(PROC, "%s[%u] affinity set for " "cpu %d/%d\n", current->comm, current->pid, i, ncpus); curcpu = i; nset++; } + put_online_cpus(); if (curcpu != -1 && nset != ncpus) { if (npresent) { prefunit = curcpu / (ncpus / npresent); diff --git a/drivers/infiniband/hw/ipath/ipath_init_chip.c b/drivers/infiniband/hw/ipath/ipath_init_chip.c index 7c1eebe..49b09c6 100644 --- a/drivers/infiniband/hw/ipath/ipath_init_chip.c +++ b/drivers/infiniband/hw/ipath/ipath_init_chip.c @@ -33,7 +33,9 @@ #include <linux/pci.h> #include <linux/netdevice.h> +#include <linux/moduleparam.h> #include <linux/slab.h> +#include <linux/stat.h> #include <linux/vmalloc.h> #include "ipath_kernel.h" diff --git a/drivers/infiniband/hw/ipath/ipath_mad.c b/drivers/infiniband/hw/ipath/ipath_mad.c index ceb98ee..43f2d04 100644 --- a/drivers/infiniband/hw/ipath/ipath_mad.c +++ b/drivers/infiniband/hw/ipath/ipath_mad.c @@ -32,6 +32,7 @@ */ #include <rdma/ib_smi.h> +#include <rdma/ib_pma.h> #include "ipath_kernel.h" #include "ipath_verbs.h" @@ -789,151 +790,18 @@ static int recv_subn_set_pkeytable(struct ib_smp *smp, return recv_subn_get_pkeytable(smp, ibdev); } -#define IB_PMA_CLASS_PORT_INFO cpu_to_be16(0x0001) -#define IB_PMA_PORT_SAMPLES_CONTROL cpu_to_be16(0x0010) -#define IB_PMA_PORT_SAMPLES_RESULT cpu_to_be16(0x0011) -#define IB_PMA_PORT_COUNTERS cpu_to_be16(0x0012) -#define IB_PMA_PORT_COUNTERS_EXT cpu_to_be16(0x001D) -#define IB_PMA_PORT_SAMPLES_RESULT_EXT cpu_to_be16(0x001E) - -struct ib_perf { - u8 base_version; - u8 mgmt_class; - u8 class_version; - u8 method; - __be16 status; - __be16 unused; - __be64 tid; - __be16 attr_id; - __be16 resv; - __be32 attr_mod; - u8 reserved[40]; - u8 data[192]; -} __attribute__ ((packed)); - -struct ib_pma_classportinfo { - u8 base_version; - u8 class_version; - __be16 cap_mask; - u8 reserved[3]; - u8 resp_time_value; /* only lower 5 bits */ - union ib_gid redirect_gid; - __be32 redirect_tc_sl_fl; /* 8, 4, 20 bits respectively */ - __be16 redirect_lid; - __be16 redirect_pkey; - __be32 redirect_qp; /* only lower 24 bits */ - __be32 redirect_qkey; - union ib_gid trap_gid; - __be32 trap_tc_sl_fl; /* 8, 4, 20 bits respectively */ - __be16 trap_lid; - __be16 trap_pkey; - __be32 trap_hl_qp; /* 8, 24 bits respectively */ - __be32 trap_qkey; -} __attribute__ ((packed)); - -struct ib_pma_portsamplescontrol { - u8 opcode; - u8 port_select; - u8 tick; - u8 counter_width; /* only lower 3 bits */ - __be32 counter_mask0_9; /* 2, 10 * 3, bits */ - __be16 counter_mask10_14; /* 1, 5 * 3, bits */ - u8 sample_mechanisms; - u8 sample_status; /* only lower 2 bits */ - __be64 option_mask; - __be64 vendor_mask; - __be32 sample_start; - __be32 sample_interval; - __be16 tag; - __be16 counter_select[15]; -} __attribute__ ((packed)); - -struct ib_pma_portsamplesresult { - __be16 tag; - __be16 sample_status; /* only lower 2 bits */ - __be32 counter[15]; -} __attribute__ ((packed)); - -struct ib_pma_portsamplesresult_ext { - __be16 tag; - __be16 sample_status; /* only lower 2 bits */ - __be32 extended_width; /* only upper 2 bits */ - __be64 counter[15]; -} __attribute__ ((packed)); - -struct ib_pma_portcounters { - u8 reserved; - u8 port_select; - __be16 counter_select; - __be16 symbol_error_counter; - u8 link_error_recovery_counter; - u8 link_downed_counter; - __be16 port_rcv_errors; - __be16 port_rcv_remphys_errors; - __be16 port_rcv_switch_relay_errors; - __be16 port_xmit_discards; - u8 port_xmit_constraint_errors; - u8 port_rcv_constraint_errors; - u8 reserved1; - u8 lli_ebor_errors; /* 4, 4, bits */ - __be16 reserved2; - __be16 vl15_dropped; - __be32 port_xmit_data; - __be32 port_rcv_data; - __be32 port_xmit_packets; - __be32 port_rcv_packets; -} __attribute__ ((packed)); - -#define IB_PMA_SEL_SYMBOL_ERROR cpu_to_be16(0x0001) -#define IB_PMA_SEL_LINK_ERROR_RECOVERY cpu_to_be16(0x0002) -#define IB_PMA_SEL_LINK_DOWNED cpu_to_be16(0x0004) -#define IB_PMA_SEL_PORT_RCV_ERRORS cpu_to_be16(0x0008) -#define IB_PMA_SEL_PORT_RCV_REMPHYS_ERRORS cpu_to_be16(0x0010) -#define IB_PMA_SEL_PORT_XMIT_DISCARDS cpu_to_be16(0x0040) -#define IB_PMA_SEL_LOCAL_LINK_INTEGRITY_ERRORS cpu_to_be16(0x0200) -#define IB_PMA_SEL_EXCESSIVE_BUFFER_OVERRUNS cpu_to_be16(0x0400) -#define IB_PMA_SEL_PORT_VL15_DROPPED cpu_to_be16(0x0800) -#define IB_PMA_SEL_PORT_XMIT_DATA cpu_to_be16(0x1000) -#define IB_PMA_SEL_PORT_RCV_DATA cpu_to_be16(0x2000) -#define IB_PMA_SEL_PORT_XMIT_PACKETS cpu_to_be16(0x4000) -#define IB_PMA_SEL_PORT_RCV_PACKETS cpu_to_be16(0x8000) - -struct ib_pma_portcounters_ext { - u8 reserved; - u8 port_select; - __be16 counter_select; - __be32 reserved1; - __be64 port_xmit_data; - __be64 port_rcv_data; - __be64 port_xmit_packets; - __be64 port_rcv_packets; - __be64 port_unicast_xmit_packets; - __be64 port_unicast_rcv_packets; - __be64 port_multicast_xmit_packets; - __be64 port_multicast_rcv_packets; -} __attribute__ ((packed)); - -#define IB_PMA_SELX_PORT_XMIT_DATA cpu_to_be16(0x0001) -#define IB_PMA_SELX_PORT_RCV_DATA cpu_to_be16(0x0002) -#define IB_PMA_SELX_PORT_XMIT_PACKETS cpu_to_be16(0x0004) -#define IB_PMA_SELX_PORT_RCV_PACKETS cpu_to_be16(0x0008) -#define IB_PMA_SELX_PORT_UNI_XMIT_PACKETS cpu_to_be16(0x0010) -#define IB_PMA_SELX_PORT_UNI_RCV_PACKETS cpu_to_be16(0x0020) -#define IB_PMA_SELX_PORT_MULTI_XMIT_PACKETS cpu_to_be16(0x0040) -#define IB_PMA_SELX_PORT_MULTI_RCV_PACKETS cpu_to_be16(0x0080) - -static int recv_pma_get_classportinfo(struct ib_perf *pmp) +static int recv_pma_get_classportinfo(struct ib_pma_mad *pmp) { - struct ib_pma_classportinfo *p = - (struct ib_pma_classportinfo *)pmp->data; + struct ib_class_port_info *p = + (struct ib_class_port_info *)pmp->data; memset(pmp->data, 0, sizeof(pmp->data)); - if (pmp->attr_mod != 0) - pmp->status |= IB_SMP_INVALID_FIELD; + if (pmp->mad_hdr.attr_mod != 0) + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; /* Indicate AllPortSelect is valid (only one port anyway) */ - p->cap_mask = cpu_to_be16(1 << 8); + p->capability_mask = cpu_to_be16(1 << 8); p->base_version = 1; p->class_version = 1; /* @@ -957,7 +825,7 @@ static int recv_pma_get_classportinfo(struct ib_perf *pmp) COUNTER_MASK(1, 3) | \ COUNTER_MASK(1, 4)) -static int recv_pma_get_portsamplescontrol(struct ib_perf *pmp, +static int recv_pma_get_portsamplescontrol(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portsamplescontrol *p = @@ -970,9 +838,9 @@ static int recv_pma_get_portsamplescontrol(struct ib_perf *pmp, memset(pmp->data, 0, sizeof(pmp->data)); p->port_select = port_select; - if (pmp->attr_mod != 0 || + if (pmp->mad_hdr.attr_mod != 0 || (port_select != port && port_select != 0xFF)) - pmp->status |= IB_SMP_INVALID_FIELD; + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; /* * Ticks are 10x the link transfer period which for 2.5Gbs is 4 * nsec. 0 == 4 nsec., 1 == 8 nsec., ..., 255 == 1020 nsec. Sample @@ -1006,7 +874,7 @@ static int recv_pma_get_portsamplescontrol(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int recv_pma_set_portsamplescontrol(struct ib_perf *pmp, +static int recv_pma_set_portsamplescontrol(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portsamplescontrol *p = @@ -1017,9 +885,9 @@ static int recv_pma_set_portsamplescontrol(struct ib_perf *pmp, u8 status; int ret; - if (pmp->attr_mod != 0 || + if (pmp->mad_hdr.attr_mod != 0 || (p->port_select != port && p->port_select != 0xFF)) { - pmp->status |= IB_SMP_INVALID_FIELD; + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; ret = reply((struct ib_smp *) pmp); goto bail; } @@ -1093,7 +961,7 @@ static u64 get_counter(struct ipath_ibdev *dev, return ret; } -static int recv_pma_get_portsamplesresult(struct ib_perf *pmp, +static int recv_pma_get_portsamplesresult(struct ib_pma_mad *pmp, struct ib_device *ibdev) { struct ib_pma_portsamplesresult *p = @@ -1118,7 +986,7 @@ static int recv_pma_get_portsamplesresult(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int recv_pma_get_portsamplesresult_ext(struct ib_perf *pmp, +static int recv_pma_get_portsamplesresult_ext(struct ib_pma_mad *pmp, struct ib_device *ibdev) { struct ib_pma_portsamplesresult_ext *p = @@ -1145,7 +1013,7 @@ static int recv_pma_get_portsamplesresult_ext(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int recv_pma_get_portcounters(struct ib_perf *pmp, +static int recv_pma_get_portcounters(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters *p = (struct ib_pma_portcounters *) @@ -1179,9 +1047,9 @@ static int recv_pma_get_portcounters(struct ib_perf *pmp, memset(pmp->data, 0, sizeof(pmp->data)); p->port_select = port_select; - if (pmp->attr_mod != 0 || + if (pmp->mad_hdr.attr_mod != 0 || (port_select != port && port_select != 0xFF)) - pmp->status |= IB_SMP_INVALID_FIELD; + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; if (cntrs.symbol_error_counter > 0xFFFFUL) p->symbol_error_counter = cpu_to_be16(0xFFFF); @@ -1216,7 +1084,7 @@ static int recv_pma_get_portcounters(struct ib_perf *pmp, cntrs.local_link_integrity_errors = 0xFUL; if (cntrs.excessive_buffer_overrun_errors > 0xFUL) cntrs.excessive_buffer_overrun_errors = 0xFUL; - p->lli_ebor_errors = (cntrs.local_link_integrity_errors << 4) | + p->link_overrun_errors = (cntrs.local_link_integrity_errors << 4) | cntrs.excessive_buffer_overrun_errors; if (cntrs.vl15_dropped > 0xFFFFUL) p->vl15_dropped = cpu_to_be16(0xFFFF); @@ -1244,7 +1112,7 @@ static int recv_pma_get_portcounters(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int recv_pma_get_portcounters_ext(struct ib_perf *pmp, +static int recv_pma_get_portcounters_ext(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters_ext *p = @@ -1265,9 +1133,9 @@ static int recv_pma_get_portcounters_ext(struct ib_perf *pmp, memset(pmp->data, 0, sizeof(pmp->data)); p->port_select = port_select; - if (pmp->attr_mod != 0 || + if (pmp->mad_hdr.attr_mod != 0 || (port_select != port && port_select != 0xFF)) - pmp->status |= IB_SMP_INVALID_FIELD; + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; p->port_xmit_data = cpu_to_be64(swords); p->port_rcv_data = cpu_to_be64(rwords); @@ -1281,7 +1149,7 @@ static int recv_pma_get_portcounters_ext(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int recv_pma_set_portcounters(struct ib_perf *pmp, +static int recv_pma_set_portcounters(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters *p = (struct ib_pma_portcounters *) @@ -1344,7 +1212,7 @@ static int recv_pma_set_portcounters(struct ib_perf *pmp, return recv_pma_get_portcounters(pmp, ibdev, port); } -static int recv_pma_set_portcounters_ext(struct ib_perf *pmp, +static int recv_pma_set_portcounters_ext(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters *p = (struct ib_pma_portcounters *) @@ -1518,19 +1386,19 @@ static int process_perf(struct ib_device *ibdev, u8 port_num, struct ib_mad *in_mad, struct ib_mad *out_mad) { - struct ib_perf *pmp = (struct ib_perf *)out_mad; + struct ib_pma_mad *pmp = (struct ib_pma_mad *)out_mad; int ret; *out_mad = *in_mad; - if (pmp->class_version != 1) { - pmp->status |= IB_SMP_UNSUP_VERSION; + if (pmp->mad_hdr.class_version != 1) { + pmp->mad_hdr.status |= IB_SMP_UNSUP_VERSION; ret = reply((struct ib_smp *) pmp); goto bail; } - switch (pmp->method) { + switch (pmp->mad_hdr.method) { case IB_MGMT_METHOD_GET: - switch (pmp->attr_id) { + switch (pmp->mad_hdr.attr_id) { case IB_PMA_CLASS_PORT_INFO: ret = recv_pma_get_classportinfo(pmp); goto bail; @@ -1554,13 +1422,13 @@ static int process_perf(struct ib_device *ibdev, u8 port_num, port_num); goto bail; default: - pmp->status |= IB_SMP_UNSUP_METH_ATTR; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_smp *) pmp); goto bail; } case IB_MGMT_METHOD_SET: - switch (pmp->attr_id) { + switch (pmp->mad_hdr.attr_id) { case IB_PMA_PORT_SAMPLES_CONTROL: ret = recv_pma_set_portsamplescontrol(pmp, ibdev, port_num); @@ -1574,7 +1442,7 @@ static int process_perf(struct ib_device *ibdev, u8 port_num, port_num); goto bail; default: - pmp->status |= IB_SMP_UNSUP_METH_ATTR; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_smp *) pmp); goto bail; } @@ -1588,7 +1456,7 @@ static int process_perf(struct ib_device *ibdev, u8 port_num, ret = IB_MAD_RESULT_SUCCESS; goto bail; default: - pmp->status |= IB_SMP_UNSUP_METHOD; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METHOD; ret = reply((struct ib_smp *) pmp); } diff --git a/drivers/infiniband/hw/ipath/ipath_srq.c b/drivers/infiniband/hw/ipath/ipath_srq.c index 386e2c7..2627198 100644 --- a/drivers/infiniband/hw/ipath/ipath_srq.c +++ b/drivers/infiniband/hw/ipath/ipath_srq.c @@ -107,6 +107,11 @@ struct ib_srq *ipath_create_srq(struct ib_pd *ibpd, u32 sz; struct ib_srq *ret; + if (srq_init_attr->srq_type != IB_SRQT_BASIC) { + ret = ERR_PTR(-ENOSYS); + goto done; + } + if (srq_init_attr->attr.max_wr == 0) { ret = ERR_PTR(-EINVAL); goto done; diff --git a/drivers/infiniband/hw/ipath/ipath_sysfs.c b/drivers/infiniband/hw/ipath/ipath_sysfs.c index 8991677..75558f3 100644 --- a/drivers/infiniband/hw/ipath/ipath_sysfs.c +++ b/drivers/infiniband/hw/ipath/ipath_sysfs.c @@ -32,6 +32,7 @@ */ #include <linux/ctype.h> +#include <linux/stat.h> #include "ipath_kernel.h" #include "ipath_verbs.h" diff --git a/drivers/infiniband/hw/ipath/ipath_user_pages.c b/drivers/infiniband/hw/ipath/ipath_user_pages.c index cfed539..dc66c45 100644 --- a/drivers/infiniband/hw/ipath/ipath_user_pages.c +++ b/drivers/infiniband/hw/ipath/ipath_user_pages.c @@ -79,7 +79,7 @@ static int __ipath_get_user_pages(unsigned long start_page, size_t num_pages, goto bail_release; } - current->mm->locked_vm += num_pages; + current->mm->pinned_vm += num_pages; ret = 0; goto bail; @@ -178,7 +178,7 @@ void ipath_release_user_pages(struct page **p, size_t num_pages) __ipath_release_user_pages(p, num_pages, 1); - current->mm->locked_vm -= num_pages; + current->mm->pinned_vm -= num_pages; up_write(¤t->mm->mmap_sem); } @@ -195,7 +195,7 @@ static void user_pages_account(struct work_struct *_work) container_of(_work, struct ipath_user_pages_work, work); down_write(&work->mm->mmap_sem); - work->mm->locked_vm -= work->num_pages; + work->mm->pinned_vm -= work->num_pages; up_write(&work->mm->mmap_sem); mmput(work->mm); kfree(work); diff --git a/drivers/infiniband/hw/ipath/ipath_user_sdma.c b/drivers/infiniband/hw/ipath/ipath_user_sdma.c index f5cb13b..cc04b7b 100644 --- a/drivers/infiniband/hw/ipath/ipath_user_sdma.c +++ b/drivers/infiniband/hw/ipath/ipath_user_sdma.c @@ -280,9 +280,7 @@ static int ipath_user_sdma_pin_pages(const struct ipath_devdata *dd, int j; int ret; - ret = get_user_pages(current, current->mm, addr, - npages, 0, 1, pages, NULL); - + ret = get_user_pages_fast(addr, npages, 0, pages); if (ret != npages) { int i; @@ -811,10 +809,7 @@ int ipath_user_sdma_writev(struct ipath_devdata *dd, while (dim) { const int mxp = 8; - down_write(¤t->mm->mmap_sem); ret = ipath_user_sdma_queue_pkts(dd, pq, &list, iov, dim, mxp); - up_write(¤t->mm->mmap_sem); - if (ret <= 0) goto done_unlock; else { diff --git a/drivers/infiniband/hw/ipath/ipath_verbs.c b/drivers/infiniband/hw/ipath/ipath_verbs.c index dd7f26d..439c35d 100644 --- a/drivers/infiniband/hw/ipath/ipath_verbs.c +++ b/drivers/infiniband/hw/ipath/ipath_verbs.c @@ -35,6 +35,7 @@ #include <rdma/ib_user_verbs.h> #include <linux/io.h> #include <linux/slab.h> +#include <linux/module.h> #include <linux/utsname.h> #include <linux/rculist.h> diff --git a/drivers/infiniband/hw/mlx4/Kconfig b/drivers/infiniband/hw/mlx4/Kconfig index bd995b2..24ab11a 100644 --- a/drivers/infiniband/hw/mlx4/Kconfig +++ b/drivers/infiniband/hw/mlx4/Kconfig @@ -1,6 +1,7 @@ config MLX4_INFINIBAND tristate "Mellanox ConnectX HCA support" - depends on NETDEVICES && NETDEV_10000 && PCI + depends on NETDEVICES && ETHERNET && PCI + select NET_VENDOR_MELLANOX select MLX4_CORE ---help--- This driver provides low-level InfiniBand support for diff --git a/drivers/infiniband/hw/mlx4/ah.c b/drivers/infiniband/hw/mlx4/ah.c index 4b8f9c4..5426ccf 100644 --- a/drivers/infiniband/hw/mlx4/ah.c +++ b/drivers/infiniband/hw/mlx4/ah.c @@ -169,9 +169,13 @@ int mlx4_ib_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr) enum rdma_link_layer ll; memset(ah_attr, 0, sizeof *ah_attr); - ah_attr->sl = be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) >> 28; ah_attr->port_num = be32_to_cpu(ah->av.ib.port_pd) >> 24; ll = rdma_port_get_link_layer(ibah->device, ah_attr->port_num); + if (ll == IB_LINK_LAYER_ETHERNET) + ah_attr->sl = be32_to_cpu(ah->av.eth.sl_tclass_flowlabel) >> 29; + else + ah_attr->sl = be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) >> 28; + ah_attr->dlid = ll == IB_LINK_LAYER_INFINIBAND ? be16_to_cpu(ah->av.ib.dlid) : 0; if (ah->av.ib.stat_rate) ah_attr->static_rate = ah->av.ib.stat_rate - MLX4_STAT_RATE_OFFSET; diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index 44fc310..46b291e 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -35,6 +35,7 @@ #include <linux/mlx4/cmd.h> #include <linux/gfp.h> +#include <rdma/ib_pma.h> #include "mlx4_ib.h" @@ -43,6 +44,14 @@ enum { MLX4_IB_VENDOR_CLASS2 = 0xa }; +/* Counters should be saturate once they reach their maximum value */ +#define ASSIGN_32BIT_COUNTER(counter, value) do {\ + if ((value) > (u32)~0U) \ + counter = cpu_to_be32((u32)~0U); \ + else \ + counter = cpu_to_be32(value); \ +} while (0) + int mlx4_MAD_IFC(struct mlx4_ib_dev *dev, int ignore_mkey, int ignore_bkey, int port, struct ib_wc *in_wc, struct ib_grh *in_grh, void *in_mad, void *response_mad) @@ -232,7 +241,7 @@ static void forward_trap(struct mlx4_ib_dev *dev, u8 port_num, struct ib_mad *ma } } -int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, +static int ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, struct ib_wc *in_wc, struct ib_grh *in_grh, struct ib_mad *in_mad, struct ib_mad *out_mad) { @@ -299,6 +308,75 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; } +static void edit_counter(struct mlx4_counter *cnt, + struct ib_pma_portcounters *pma_cnt) +{ + ASSIGN_32BIT_COUNTER(pma_cnt->port_xmit_data, + (be64_to_cpu(cnt->tx_bytes) >> 2)); + ASSIGN_32BIT_COUNTER(pma_cnt->port_rcv_data, + (be64_to_cpu(cnt->rx_bytes) >> 2)); + ASSIGN_32BIT_COUNTER(pma_cnt->port_xmit_packets, + be64_to_cpu(cnt->tx_frames)); + ASSIGN_32BIT_COUNTER(pma_cnt->port_rcv_packets, + be64_to_cpu(cnt->rx_frames)); +} + +static int iboe_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, + struct ib_wc *in_wc, struct ib_grh *in_grh, + struct ib_mad *in_mad, struct ib_mad *out_mad) +{ + struct mlx4_cmd_mailbox *mailbox; + struct mlx4_ib_dev *dev = to_mdev(ibdev); + int err; + u32 inmod = dev->counters[port_num - 1] & 0xffff; + u8 mode; + + if (in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_PERF_MGMT) + return -EINVAL; + + mailbox = mlx4_alloc_cmd_mailbox(dev->dev); + if (IS_ERR(mailbox)) + return IB_MAD_RESULT_FAILURE; + + err = mlx4_cmd_box(dev->dev, 0, mailbox->dma, inmod, 0, + MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C); + if (err) + err = IB_MAD_RESULT_FAILURE; + else { + memset(out_mad->data, 0, sizeof out_mad->data); + mode = ((struct mlx4_counter *)mailbox->buf)->counter_mode; + switch (mode & 0xf) { + case 0: + edit_counter(mailbox->buf, + (void *)(out_mad->data + 40)); + err = IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; + break; + default: + err = IB_MAD_RESULT_FAILURE; + } + } + + mlx4_free_cmd_mailbox(dev->dev, mailbox); + + return err; +} + +int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, + struct ib_wc *in_wc, struct ib_grh *in_grh, + struct ib_mad *in_mad, struct ib_mad *out_mad) +{ + switch (rdma_port_get_link_layer(ibdev, port_num)) { + case IB_LINK_LAYER_INFINIBAND: + return ib_process_mad(ibdev, mad_flags, port_num, in_wc, + in_grh, in_mad, out_mad); + case IB_LINK_LAYER_ETHERNET: + return iboe_process_mad(ibdev, mad_flags, port_num, in_wc, + in_grh, in_mad, out_mad); + default: + return -EINVAL; + } +} + static void send_handler(struct ib_mad_agent *agent, struct ib_mad_send_wc *mad_send_wc) { diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index fbe1973..18836cd 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -128,6 +128,8 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, (dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_REMOTE_INV) && (dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_FAST_REG_WR)) props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS; + if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) + props->device_cap_flags |= IB_DEVICE_XRC; props->vendor_id = be32_to_cpup((__be32 *) (out_mad->data + 36)) & 0xffffff; @@ -181,8 +183,12 @@ mlx4_ib_port_link_layer(struct ib_device *device, u8 port_num) static int ib_link_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props, + struct ib_smp *in_mad, struct ib_smp *out_mad) { + int ext_active_speed; + int err; + props->lid = be16_to_cpup((__be16 *) (out_mad->data + 16)); props->lmc = out_mad->data[34] & 0x7; props->sm_lid = be16_to_cpup((__be16 *) (out_mad->data + 18)); @@ -203,6 +209,39 @@ static int ib_link_query_port(struct ib_device *ibdev, u8 port, props->max_vl_num = out_mad->data[37] >> 4; props->init_type_reply = out_mad->data[41] >> 4; + /* Check if extended speeds (EDR/FDR/...) are supported */ + if (props->port_cap_flags & IB_PORT_EXTENDED_SPEEDS_SUP) { + ext_active_speed = out_mad->data[62] >> 4; + + switch (ext_active_speed) { + case 1: + props->active_speed = 16; /* FDR */ + break; + case 2: + props->active_speed = 32; /* EDR */ + break; + } + } + + /* If reported active speed is QDR, check if is FDR-10 */ + if (props->active_speed == 4) { + if (to_mdev(ibdev)->dev->caps.ext_port_cap[port] & + MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO) { + init_query_mad(in_mad); + in_mad->attr_id = MLX4_ATTR_EXTENDED_PORT_INFO; + in_mad->attr_mod = cpu_to_be32(port); + + err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, + NULL, NULL, in_mad, out_mad); + if (err) + return err; + + /* Checking LinkSpeedActive for FDR-10 */ + if (out_mad->data[15] & 0x1) + props->active_speed = 8; + } + } + return 0; } @@ -227,7 +266,7 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port, props->pkey_tbl_len = 1; props->bad_pkey_cntr = be16_to_cpup((__be16 *) (out_mad->data + 46)); props->qkey_viol_cntr = be16_to_cpup((__be16 *) (out_mad->data + 48)); - props->max_mtu = IB_MTU_2048; + props->max_mtu = IB_MTU_4096; props->subnet_timeout = 0; props->max_vl_num = out_mad->data[37] >> 4; props->init_type_reply = 0; @@ -274,7 +313,7 @@ static int mlx4_ib_query_port(struct ib_device *ibdev, u8 port, goto out; err = mlx4_ib_port_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND ? - ib_link_query_port(ibdev, port, props, out_mad) : + ib_link_query_port(ibdev, port, props, in_mad, out_mad) : eth_link_query_port(ibdev, port, props, out_mad); out: @@ -566,6 +605,57 @@ static int mlx4_ib_dealloc_pd(struct ib_pd *pd) return 0; } +static struct ib_xrcd *mlx4_ib_alloc_xrcd(struct ib_device *ibdev, + struct ib_ucontext *context, + struct ib_udata *udata) +{ + struct mlx4_ib_xrcd *xrcd; + 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); + + err = mlx4_xrcd_alloc(to_mdev(ibdev)->dev, &xrcd->xrcdn); + if (err) + goto err1; + + xrcd->pd = ib_alloc_pd(ibdev); + if (IS_ERR(xrcd->pd)) { + err = PTR_ERR(xrcd->pd); + goto err2; + } + + xrcd->cq = ib_create_cq(ibdev, NULL, NULL, xrcd, 1, 0); + if (IS_ERR(xrcd->cq)) { + err = PTR_ERR(xrcd->cq); + goto err3; + } + + return &xrcd->ibxrcd; + +err3: + ib_dealloc_pd(xrcd->pd); +err2: + mlx4_xrcd_free(to_mdev(ibdev)->dev, xrcd->xrcdn); +err1: + kfree(xrcd); + return ERR_PTR(err); +} + +static int mlx4_ib_dealloc_xrcd(struct ib_xrcd *xrcd) +{ + 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) { struct mlx4_ib_qp *mqp = to_mqp(ibqp); @@ -816,7 +906,7 @@ static void update_gids_task(struct work_struct *work) memcpy(gw->dev->iboe.gid_table[gw->port - 1], gw->gids, sizeof gw->gids); event.device = &gw->dev->ib_dev; event.element.port_num = gw->port; - event.event = IB_EVENT_LID_CHANGE; + event.event = IB_EVENT_GID_CHANGE; ib_dispatch_event(&event); } @@ -1044,7 +1134,9 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) (1ull << IB_USER_VERBS_CMD_CREATE_SRQ) | (1ull << IB_USER_VERBS_CMD_MODIFY_SRQ) | (1ull << IB_USER_VERBS_CMD_QUERY_SRQ) | - (1ull << IB_USER_VERBS_CMD_DESTROY_SRQ); + (1ull << IB_USER_VERBS_CMD_DESTROY_SRQ) | + (1ull << IB_USER_VERBS_CMD_CREATE_XSRQ) | + (1ull << IB_USER_VERBS_CMD_OPEN_QP); ibdev->ib_dev.query_device = mlx4_ib_query_device; ibdev->ib_dev.query_port = mlx4_ib_query_port; @@ -1093,16 +1185,34 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) ibdev->ib_dev.unmap_fmr = mlx4_ib_unmap_fmr; ibdev->ib_dev.dealloc_fmr = mlx4_ib_fmr_dealloc; + if (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) { + ibdev->ib_dev.alloc_xrcd = mlx4_ib_alloc_xrcd; + ibdev->ib_dev.dealloc_xrcd = mlx4_ib_dealloc_xrcd; + ibdev->ib_dev.uverbs_cmd_mask |= + (1ull << IB_USER_VERBS_CMD_OPEN_XRCD) | + (1ull << IB_USER_VERBS_CMD_CLOSE_XRCD); + } + spin_lock_init(&iboe->lock); if (init_node_data(ibdev)) goto err_map; + for (i = 0; i < ibdev->num_ports; ++i) { + if (mlx4_ib_port_link_layer(&ibdev->ib_dev, i + 1) == + IB_LINK_LAYER_ETHERNET) { + err = mlx4_counter_alloc(ibdev->dev, &ibdev->counters[i]); + if (err) + ibdev->counters[i] = -1; + } else + ibdev->counters[i] = -1; + } + spin_lock_init(&ibdev->sm_lock); mutex_init(&ibdev->cap_mask_mutex); if (ib_register_device(&ibdev->ib_dev, NULL)) - goto err_map; + goto err_counter; if (mlx4_ib_mad_init(ibdev)) goto err_reg; @@ -1132,6 +1242,11 @@ err_notif: err_reg: ib_unregister_device(&ibdev->ib_dev); +err_counter: + for (; i; --i) + if (ibdev->counters[i - 1] != -1) + mlx4_counter_free(ibdev->dev, ibdev->counters[i - 1]); + err_map: iounmap(ibdev->uar_map); @@ -1160,7 +1275,9 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr) ibdev->iboe.nb.notifier_call = NULL; } iounmap(ibdev->uar_map); - + for (p = 0; p < ibdev->num_ports; ++p) + if (ibdev->counters[p] != -1) + mlx4_counter_free(ibdev->dev, ibdev->counters[p]); mlx4_foreach_port(p, dev, MLX4_PORT_TYPE_IB) mlx4_CLOSE_PORT(dev, p); diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 2a322f2..ed80345 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -56,6 +56,13 @@ struct mlx4_ib_pd { u32 pdn; }; +struct mlx4_ib_xrcd { + struct ib_xrcd ibxrcd; + u32 xrcdn; + struct ib_pd *pd; + struct ib_cq *cq; +}; + struct mlx4_ib_cq_buf { struct mlx4_buf buf; struct mlx4_mtt mtt; @@ -138,6 +145,7 @@ struct mlx4_ib_qp { struct mlx4_mtt mtt; int buf_size; struct mutex mutex; + u16 xrcdn; u32 flags; u8 port; u8 alt_port; @@ -193,6 +201,7 @@ struct mlx4_ib_dev { struct mutex cap_mask_mutex; bool ib_active; struct mlx4_ib_iboe iboe; + int counters[MLX4_MAX_PORTS]; }; static inline struct mlx4_ib_dev *to_mdev(struct ib_device *ibdev) @@ -210,6 +219,11 @@ static inline struct mlx4_ib_pd *to_mpd(struct ib_pd *ibpd) return container_of(ibpd, struct mlx4_ib_pd, ibpd); } +static inline struct mlx4_ib_xrcd *to_mxrcd(struct ib_xrcd *ibxrcd) +{ + return container_of(ibxrcd, struct mlx4_ib_xrcd, ibxrcd); +} + static inline struct mlx4_ib_cq *to_mcq(struct ib_cq *ibcq) { return container_of(ibcq, struct mlx4_ib_cq, ibcq); diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index 23c04ff..2ed14a7 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -302,15 +302,14 @@ static int send_wqe_overhead(enum ib_qp_type type, u32 flags) } static int set_rq_size(struct mlx4_ib_dev *dev, struct ib_qp_cap *cap, - int is_user, int has_srq, struct mlx4_ib_qp *qp) + int is_user, int has_rq, struct mlx4_ib_qp *qp) { /* Sanity check RQ size before proceeding */ if (cap->max_recv_wr > dev->dev->caps.max_wqes || cap->max_recv_sge > dev->dev->caps.max_rq_sg) return -EINVAL; - if (has_srq) { - /* QPs attached to an SRQ should have no RQ */ + if (!has_rq) { if (cap->max_recv_wr) return -EINVAL; @@ -463,6 +462,14 @@ static int set_user_sq_size(struct mlx4_ib_dev *dev, return 0; } +static int qp_has_rq(struct ib_qp_init_attr *attr) +{ + if (attr->qp_type == IB_QPT_XRC_INI || attr->qp_type == IB_QPT_XRC_TGT) + return 0; + + return !attr->srq; +} + static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata, int sqpn, struct mlx4_ib_qp *qp) @@ -479,7 +486,7 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) qp->sq_signal_bits = cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE); - err = set_rq_size(dev, &init_attr->cap, !!pd->uobject, !!init_attr->srq, qp); + err = set_rq_size(dev, &init_attr->cap, !!pd->uobject, qp_has_rq(init_attr), qp); if (err) goto err; @@ -513,7 +520,7 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, if (err) goto err_mtt; - if (!init_attr->srq) { + if (qp_has_rq(init_attr)) { err = mlx4_ib_db_map_user(to_mucontext(pd->uobject->context), ucmd.db_addr, &qp->db); if (err) @@ -532,7 +539,7 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, if (err) goto err; - if (!init_attr->srq) { + if (qp_has_rq(init_attr)) { err = mlx4_db_alloc(dev->dev, &qp->db, 0); if (err) goto err; @@ -575,6 +582,9 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, if (err) goto err_qpn; + if (init_attr->qp_type == IB_QPT_XRC_TGT) + qp->mqp.qpn |= (1 << 23); + /* * Hardware wants QPN written in big-endian order (after * shifting) for send doorbell. Precompute this value to save @@ -592,9 +602,8 @@ err_qpn: err_wrid: if (pd->uobject) { - if (!init_attr->srq) - mlx4_ib_db_unmap_user(to_mucontext(pd->uobject->context), - &qp->db); + if (qp_has_rq(init_attr)) + mlx4_ib_db_unmap_user(to_mucontext(pd->uobject->context), &qp->db); } else { kfree(qp->sq.wrid); kfree(qp->rq.wrid); @@ -610,7 +619,7 @@ err_buf: mlx4_buf_free(dev->dev, qp->buf_size, &qp->buf); err_db: - if (!pd->uobject && !init_attr->srq) + if (!pd->uobject && qp_has_rq(init_attr)) mlx4_db_free(dev->dev, &qp->db); err: @@ -671,6 +680,33 @@ static void del_gid_entries(struct mlx4_ib_qp *qp) } } +static struct mlx4_ib_pd *get_pd(struct mlx4_ib_qp *qp) +{ + if (qp->ibqp.qp_type == IB_QPT_XRC_TGT) + return to_mpd(to_mxrcd(qp->ibqp.xrcd)->pd); + else + return to_mpd(qp->ibqp.pd); +} + +static void get_cqs(struct mlx4_ib_qp *qp, + struct mlx4_ib_cq **send_cq, struct mlx4_ib_cq **recv_cq) +{ + switch (qp->ibqp.qp_type) { + case IB_QPT_XRC_TGT: + *send_cq = to_mcq(to_mxrcd(qp->ibqp.xrcd)->cq); + *recv_cq = *send_cq; + break; + case IB_QPT_XRC_INI: + *send_cq = to_mcq(qp->ibqp.send_cq); + *recv_cq = *send_cq; + break; + default: + *send_cq = to_mcq(qp->ibqp.send_cq); + *recv_cq = to_mcq(qp->ibqp.recv_cq); + break; + } +} + static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, int is_user) { @@ -682,8 +718,7 @@ static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, printk(KERN_WARNING "mlx4_ib: modify QP %06x to RESET failed.\n", qp->mqp.qpn); - send_cq = to_mcq(qp->ibqp.send_cq); - recv_cq = to_mcq(qp->ibqp.recv_cq); + get_cqs(qp, &send_cq, &recv_cq); mlx4_ib_lock_cqs(send_cq, recv_cq); @@ -706,7 +741,7 @@ static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, mlx4_mtt_cleanup(dev->dev, &qp->mtt); if (is_user) { - if (!qp->ibqp.srq) + if (qp->rq.wqe_cnt) mlx4_ib_db_unmap_user(to_mucontext(qp->ibqp.uobject->context), &qp->db); ib_umem_release(qp->umem); @@ -714,7 +749,7 @@ static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, kfree(qp->sq.wrid); kfree(qp->rq.wrid); mlx4_buf_free(dev->dev, qp->buf_size, &qp->buf); - if (!qp->ibqp.srq) + if (qp->rq.wqe_cnt) mlx4_db_free(dev->dev, &qp->db); } @@ -725,10 +760,10 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata) { - struct mlx4_ib_dev *dev = to_mdev(pd->device); struct mlx4_ib_sqp *sqp; struct mlx4_ib_qp *qp; int err; + u16 xrcdn = 0; /* * We only support LSO and multicast loopback blocking, and @@ -739,10 +774,20 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd, return ERR_PTR(-EINVAL); if (init_attr->create_flags && - (pd->uobject || init_attr->qp_type != IB_QPT_UD)) + (udata || init_attr->qp_type != IB_QPT_UD)) return ERR_PTR(-EINVAL); switch (init_attr->qp_type) { + case IB_QPT_XRC_TGT: + 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 */ + 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 */ case IB_QPT_RC: case IB_QPT_UC: case IB_QPT_UD: @@ -751,13 +796,14 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd, if (!qp) return ERR_PTR(-ENOMEM); - err = create_qp_common(dev, pd, init_attr, udata, 0, qp); + err = create_qp_common(to_mdev(pd->device), pd, init_attr, udata, 0, qp); if (err) { kfree(qp); return ERR_PTR(err); } qp->ibqp.qp_num = qp->mqp.qpn; + qp->xrcdn = xrcdn; break; } @@ -765,7 +811,7 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd, case IB_QPT_GSI: { /* Userspace is not allowed to create special QPs: */ - if (pd->uobject) + if (udata) return ERR_PTR(-EINVAL); sqp = kzalloc(sizeof *sqp, GFP_KERNEL); @@ -774,8 +820,8 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd, qp = &sqp->qp; - err = create_qp_common(dev, pd, init_attr, udata, - dev->dev->caps.sqp_start + + err = create_qp_common(to_mdev(pd->device), pd, init_attr, udata, + to_mdev(pd->device)->dev->caps.sqp_start + (init_attr->qp_type == IB_QPT_SMI ? 0 : 2) + init_attr->port_num - 1, qp); @@ -801,11 +847,13 @@ int mlx4_ib_destroy_qp(struct ib_qp *qp) { struct mlx4_ib_dev *dev = to_mdev(qp->device); struct mlx4_ib_qp *mqp = to_mqp(qp); + struct mlx4_ib_pd *pd; if (is_qp0(dev, mqp)) mlx4_CLOSE_PORT(dev->dev, mqp->port); - destroy_qp_common(dev, mqp, !!qp->pd->uobject); + pd = get_pd(mqp); + destroy_qp_common(dev, mqp, !!pd->ibpd.uobject); if (is_sqp(dev, mqp)) kfree(to_msqp(mqp)); @@ -821,6 +869,8 @@ static int to_mlx4_st(enum ib_qp_type type) case IB_QPT_RC: return MLX4_QP_ST_RC; case IB_QPT_UC: return MLX4_QP_ST_UC; case IB_QPT_UD: return MLX4_QP_ST_UD; + case IB_QPT_XRC_INI: + case IB_QPT_XRC_TGT: return MLX4_QP_ST_XRC; case IB_QPT_SMI: case IB_QPT_GSI: return MLX4_QP_ST_MLX; default: return -1; @@ -893,7 +943,6 @@ static int mlx4_set_path(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah, --path->static_rate; } else path->static_rate = 0; - path->counter_index = 0xff; if (ah->ah_flags & IB_AH_GRH) { if (ah->grh.sgid_index >= dev->dev->caps.gid_table_len[port]) { @@ -960,6 +1009,8 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, { struct mlx4_ib_dev *dev = to_mdev(ibqp->device); struct mlx4_ib_qp *qp = to_mqp(ibqp); + struct mlx4_ib_pd *pd; + struct mlx4_ib_cq *send_cq, *recv_cq; struct mlx4_qp_context *context; enum mlx4_qp_optpar optpar = 0; int sqd_event; @@ -1015,8 +1066,10 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, context->sq_size_stride = ilog2(qp->sq.wqe_cnt) << 3; context->sq_size_stride |= qp->sq.wqe_shift - 4; - if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) + if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { context->sq_size_stride |= !!qp->sq_no_prefetch << 7; + context->xrcd = cpu_to_be32((u32) qp->xrcdn); + } if (qp->ibqp.uobject) context->usr_page = cpu_to_be32(to_mucontext(ibqp->uobject->context)->uar.index); @@ -1034,6 +1087,15 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, } } + if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { + if (dev->counters[qp->port - 1] != -1) { + context->pri_path.counter_index = + dev->counters[qp->port - 1]; + optpar |= MLX4_QP_OPTPAR_COUNTER_INDEX; + } else + context->pri_path.counter_index = 0xff; + } + if (attr_mask & IB_QP_PKEY_INDEX) { context->pri_path.pkey_index = attr->pkey_index; optpar |= MLX4_QP_OPTPAR_PKEY_INDEX; @@ -1071,8 +1133,12 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, optpar |= MLX4_QP_OPTPAR_ALT_ADDR_PATH; } - context->pd = cpu_to_be32(to_mpd(ibqp->pd)->pdn); - context->params1 = cpu_to_be32(MLX4_IB_ACK_REQ_FREQ << 28); + pd = get_pd(qp); + get_cqs(qp, &send_cq, &recv_cq); + context->pd = cpu_to_be32(pd->pdn); + context->cqn_send = cpu_to_be32(send_cq->mcq.cqn); + context->cqn_recv = cpu_to_be32(recv_cq->mcq.cqn); + context->params1 = cpu_to_be32(MLX4_IB_ACK_REQ_FREQ << 28); /* Set "fast registration enabled" for all kernel QPs */ if (!qp->ibqp.uobject) @@ -1098,8 +1164,6 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, if (attr_mask & IB_QP_SQ_PSN) context->next_send_psn = cpu_to_be32(attr->sq_psn); - context->cqn_send = cpu_to_be32(to_mcq(ibqp->send_cq)->mcq.cqn); - if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) { if (attr->max_dest_rd_atomic) context->params2 |= @@ -1122,8 +1186,6 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, if (attr_mask & IB_QP_RQ_PSN) context->rnr_nextrecvpsn |= cpu_to_be32(attr->rq_psn); - context->cqn_recv = cpu_to_be32(to_mcq(ibqp->recv_cq)->mcq.cqn); - if (attr_mask & IB_QP_QKEY) { context->qkey = cpu_to_be32(attr->qkey); optpar |= MLX4_QP_OPTPAR_Q_KEY; @@ -1132,7 +1194,7 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, if (ibqp->srq) context->srqn = cpu_to_be32(1 << 24 | to_msrq(ibqp->srq)->msrq.srqn); - if (!ibqp->srq && cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) + 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); if (cur_state == IB_QPS_INIT && @@ -1217,17 +1279,17 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, * entries and reinitialize the QP. */ if (new_state == IB_QPS_RESET && !ibqp->uobject) { - mlx4_ib_cq_clean(to_mcq(ibqp->recv_cq), qp->mqp.qpn, + mlx4_ib_cq_clean(recv_cq, qp->mqp.qpn, ibqp->srq ? to_msrq(ibqp->srq): NULL); - if (ibqp->send_cq != ibqp->recv_cq) - mlx4_ib_cq_clean(to_mcq(ibqp->send_cq), qp->mqp.qpn, NULL); + if (send_cq != recv_cq) + mlx4_ib_cq_clean(send_cq, qp->mqp.qpn, NULL); qp->rq.head = 0; qp->rq.tail = 0; qp->sq.head = 0; qp->sq.tail = 0; qp->sq_next_wqe = 0; - if (!ibqp->srq) + if (qp->rq.wqe_cnt) *qp->db.db = 0; } @@ -1301,7 +1363,7 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr, int is_eth; int is_vlan = 0; int is_grh; - u16 vlan = 0; + u16 vlan; send_size = 0; for (i = 0; i < wr->num_sge; ++i) @@ -1539,14 +1601,13 @@ static void set_masked_atomic_seg(struct mlx4_wqe_masked_atomic_seg *aseg, } static void set_datagram_seg(struct mlx4_wqe_datagram_seg *dseg, - struct ib_send_wr *wr, __be16 *vlan) + struct ib_send_wr *wr) { memcpy(dseg->av, &to_mah(wr->wr.ud.ah)->av, sizeof (struct mlx4_av)); dseg->dqpn = cpu_to_be32(wr->wr.ud.remote_qpn); dseg->qkey = cpu_to_be32(wr->wr.ud.remote_qkey); dseg->vlan = to_mah(wr->wr.ud.ah)->av.eth.vlan; memcpy(dseg->mac, to_mah(wr->wr.ud.ah)->av.eth.mac, 6); - *vlan = dseg->vlan; } static void set_mlx_icrc_seg(void *dseg) @@ -1609,8 +1670,7 @@ static int build_lso_seg(struct mlx4_wqe_lso_seg *wqe, struct ib_send_wr *wr, memcpy(wqe->header, wr->wr.ud.header, wr->wr.ud.hlen); - *lso_hdr_sz = cpu_to_be32((wr->wr.ud.mss - wr->wr.ud.hlen) << 16 | - wr->wr.ud.hlen); + *lso_hdr_sz = cpu_to_be32(wr->wr.ud.mss << 16 | wr->wr.ud.hlen); *lso_seg_len = halign; return 0; } @@ -1649,7 +1709,6 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, __be32 uninitialized_var(lso_hdr_sz); __be32 blh; int i; - __be16 vlan = cpu_to_be16(0xffff); spin_lock_irqsave(&qp->sq.lock, flags); @@ -1753,7 +1812,7 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, break; case IB_QPT_UD: - set_datagram_seg(wqe, wr, &vlan); + set_datagram_seg(wqe, wr); wqe += sizeof (struct mlx4_wqe_datagram_seg); size += sizeof (struct mlx4_wqe_datagram_seg) / 16; @@ -1816,11 +1875,6 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, ctrl->fence_size = (wr->send_flags & IB_SEND_FENCE ? MLX4_WQE_CTRL_FENCE : 0) | size; - if (be16_to_cpu(vlan) < 0x1000) { - ctrl->ins_vlan = 1 << 6; - ctrl->vlan_tag = vlan; - } - /* * Make sure descriptor is fully written before * setting ownership bit (because HW can start diff --git a/drivers/infiniband/hw/mlx4/srq.c b/drivers/infiniband/hw/mlx4/srq.c index 818b7ec..39542f3 100644 --- a/drivers/infiniband/hw/mlx4/srq.c +++ b/drivers/infiniband/hw/mlx4/srq.c @@ -76,6 +76,8 @@ struct ib_srq *mlx4_ib_create_srq(struct ib_pd *pd, struct mlx4_ib_srq *srq; struct mlx4_wqe_srq_next_seg *next; struct mlx4_wqe_data_seg *scatter; + u32 cqn; + u16 xrcdn; int desc_size; int buf_size; int err; @@ -174,12 +176,18 @@ struct ib_srq *mlx4_ib_create_srq(struct ib_pd *pd, } } - err = mlx4_srq_alloc(dev->dev, to_mpd(pd)->pdn, &srq->mtt, + cqn = (init_attr->srq_type == IB_SRQT_XRC) ? + to_mcq(init_attr->ext.xrc.cq)->mcq.cqn : 0; + xrcdn = (init_attr->srq_type == IB_SRQT_XRC) ? + to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn : + (u16) dev->dev->caps.reserved_xrcds; + err = mlx4_srq_alloc(dev->dev, to_mpd(pd)->pdn, cqn, xrcdn, &srq->mtt, srq->db.dma, &srq->msrq); if (err) goto err_wrid; srq->msrq.event = mlx4_ib_srq_event; + srq->ibsrq.ext.xrc.srq_num = srq->msrq.srqn; if (pd->uobject) if (ib_copy_to_udata(udata, &srq->msrq.srqn, sizeof (__u32))) { diff --git a/drivers/infiniband/hw/mthca/mthca_catas.c b/drivers/infiniband/hw/mthca/mthca_catas.c index e4a08c2..712d2a3 100644 --- a/drivers/infiniband/hw/mthca/mthca_catas.c +++ b/drivers/infiniband/hw/mthca/mthca_catas.c @@ -31,6 +31,7 @@ */ #include <linux/jiffies.h> +#include <linux/module.h> #include <linux/timer.h> #include <linux/workqueue.h> diff --git a/drivers/infiniband/hw/mthca/mthca_cmd.c b/drivers/infiniband/hw/mthca/mthca_cmd.c index 7bfa2a1..9d3e5c1 100644 --- a/drivers/infiniband/hw/mthca/mthca_cmd.c +++ b/drivers/infiniband/hw/mthca/mthca_cmd.c @@ -36,6 +36,7 @@ #include <linux/pci.h> #include <linux/errno.h> #include <linux/sched.h> +#include <linux/module.h> #include <linux/slab.h> #include <asm/io.h> #include <rdma/ib_mad.h> @@ -301,6 +302,38 @@ static int mthca_cmd_post(struct mthca_dev *dev, return err; } + +static int mthca_status_to_errno(u8 status) +{ + static const int trans_table[] = { + [MTHCA_CMD_STAT_INTERNAL_ERR] = -EIO, + [MTHCA_CMD_STAT_BAD_OP] = -EPERM, + [MTHCA_CMD_STAT_BAD_PARAM] = -EINVAL, + [MTHCA_CMD_STAT_BAD_SYS_STATE] = -ENXIO, + [MTHCA_CMD_STAT_BAD_RESOURCE] = -EBADF, + [MTHCA_CMD_STAT_RESOURCE_BUSY] = -EBUSY, + [MTHCA_CMD_STAT_DDR_MEM_ERR] = -ENOMEM, + [MTHCA_CMD_STAT_EXCEED_LIM] = -ENOMEM, + [MTHCA_CMD_STAT_BAD_RES_STATE] = -EBADF, + [MTHCA_CMD_STAT_BAD_INDEX] = -EBADF, + [MTHCA_CMD_STAT_BAD_NVMEM] = -EFAULT, + [MTHCA_CMD_STAT_BAD_QPEE_STATE] = -EINVAL, + [MTHCA_CMD_STAT_BAD_SEG_PARAM] = -EFAULT, + [MTHCA_CMD_STAT_REG_BOUND] = -EBUSY, + [MTHCA_CMD_STAT_LAM_NOT_PRE] = -EAGAIN, + [MTHCA_CMD_STAT_BAD_PKT] = -EBADMSG, + [MTHCA_CMD_STAT_BAD_SIZE] = -ENOMEM, + }; + + if (status >= ARRAY_SIZE(trans_table) || + (status != MTHCA_CMD_STAT_OK + && trans_table[status] == 0)) + return -EINVAL; + + return trans_table[status]; +} + + static int mthca_cmd_poll(struct mthca_dev *dev, u64 in_param, u64 *out_param, @@ -308,11 +341,11 @@ static int mthca_cmd_poll(struct mthca_dev *dev, u32 in_modifier, u8 op_modifier, u16 op, - unsigned long timeout, - u8 *status) + unsigned long timeout) { int err = 0; unsigned long end; + u8 status; down(&dev->cmd.poll_sem); @@ -341,7 +374,12 @@ static int mthca_cmd_poll(struct mthca_dev *dev, (u64) be32_to_cpu((__force __be32) __raw_readl(dev->hcr + HCR_OUT_PARAM_OFFSET + 4)); - *status = be32_to_cpu((__force __be32) __raw_readl(dev->hcr + HCR_STATUS_OFFSET)) >> 24; + status = be32_to_cpu((__force __be32) __raw_readl(dev->hcr + HCR_STATUS_OFFSET)) >> 24; + if (status) { + mthca_dbg(dev, "Command %02x completed with status %02x\n", + op, status); + err = mthca_status_to_errno(status); + } out: up(&dev->cmd.poll_sem); @@ -374,8 +412,7 @@ static int mthca_cmd_wait(struct mthca_dev *dev, u32 in_modifier, u8 op_modifier, u16 op, - unsigned long timeout, - u8 *status) + unsigned long timeout) { int err = 0; struct mthca_cmd_context *context; @@ -407,10 +444,11 @@ static int mthca_cmd_wait(struct mthca_dev *dev, if (err) goto out; - *status = context->status; - if (*status) + if (context->status) { mthca_dbg(dev, "Command %02x completed with status %02x\n", - op, *status); + op, context->status); + err = mthca_status_to_errno(context->status); + } if (out_is_imm) *out_param = context->out_param; @@ -432,17 +470,16 @@ static int mthca_cmd_box(struct mthca_dev *dev, u32 in_modifier, u8 op_modifier, u16 op, - unsigned long timeout, - u8 *status) + unsigned long timeout) { if (dev->cmd.flags & MTHCA_CMD_USE_EVENTS) return mthca_cmd_wait(dev, in_param, &out_param, 0, in_modifier, op_modifier, op, - timeout, status); + timeout); else return mthca_cmd_poll(dev, in_param, &out_param, 0, in_modifier, op_modifier, op, - timeout, status); + timeout); } /* Invoke a command with no output parameter */ @@ -451,11 +488,10 @@ static int mthca_cmd(struct mthca_dev *dev, u32 in_modifier, u8 op_modifier, u16 op, - unsigned long timeout, - u8 *status) + unsigned long timeout) { return mthca_cmd_box(dev, in_param, 0, in_modifier, - op_modifier, op, timeout, status); + op_modifier, op, timeout); } /* @@ -469,17 +505,16 @@ static int mthca_cmd_imm(struct mthca_dev *dev, u32 in_modifier, u8 op_modifier, u16 op, - unsigned long timeout, - u8 *status) + unsigned long timeout) { if (dev->cmd.flags & MTHCA_CMD_USE_EVENTS) return mthca_cmd_wait(dev, in_param, out_param, 1, in_modifier, op_modifier, op, - timeout, status); + timeout); else return mthca_cmd_poll(dev, in_param, out_param, 1, in_modifier, op_modifier, op, - timeout, status); + timeout); } int mthca_cmd_init(struct mthca_dev *dev) @@ -596,14 +631,14 @@ void mthca_free_mailbox(struct mthca_dev *dev, struct mthca_mailbox *mailbox) kfree(mailbox); } -int mthca_SYS_EN(struct mthca_dev *dev, u8 *status) +int mthca_SYS_EN(struct mthca_dev *dev) { u64 out; int ret; - ret = mthca_cmd_imm(dev, 0, &out, 0, 0, CMD_SYS_EN, CMD_TIME_CLASS_D, status); + ret = mthca_cmd_imm(dev, 0, &out, 0, 0, CMD_SYS_EN, CMD_TIME_CLASS_D); - if (*status == MTHCA_CMD_STAT_DDR_MEM_ERR) + if (ret == -ENOMEM) mthca_warn(dev, "SYS_EN DDR error: syn=%x, sock=%d, " "sladdr=%d, SPD source=%s\n", (int) (out >> 6) & 0xf, (int) (out >> 4) & 3, @@ -612,13 +647,13 @@ int mthca_SYS_EN(struct mthca_dev *dev, u8 *status) return ret; } -int mthca_SYS_DIS(struct mthca_dev *dev, u8 *status) +int mthca_SYS_DIS(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0, 0, CMD_SYS_DIS, CMD_TIME_CLASS_C, status); + return mthca_cmd(dev, 0, 0, 0, CMD_SYS_DIS, CMD_TIME_CLASS_C); } static int mthca_map_cmd(struct mthca_dev *dev, u16 op, struct mthca_icm *icm, - u64 virt, u8 *status) + u64 virt) { struct mthca_mailbox *mailbox; struct mthca_icm_iter iter; @@ -666,8 +701,8 @@ static int mthca_map_cmd(struct mthca_dev *dev, u16 op, struct mthca_icm *icm, if (++nent == MTHCA_MAILBOX_SIZE / 16) { err = mthca_cmd(dev, mailbox->dma, nent, 0, op, - CMD_TIME_CLASS_B, status); - if (err || *status) + CMD_TIME_CLASS_B); + if (err) goto out; nent = 0; } @@ -676,7 +711,7 @@ static int mthca_map_cmd(struct mthca_dev *dev, u16 op, struct mthca_icm *icm, if (nent) err = mthca_cmd(dev, mailbox->dma, nent, 0, op, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); switch (op) { case CMD_MAP_FA: @@ -696,19 +731,19 @@ out: return err; } -int mthca_MAP_FA(struct mthca_dev *dev, struct mthca_icm *icm, u8 *status) +int mthca_MAP_FA(struct mthca_dev *dev, struct mthca_icm *icm) { - return mthca_map_cmd(dev, CMD_MAP_FA, icm, -1, status); + return mthca_map_cmd(dev, CMD_MAP_FA, icm, -1); } -int mthca_UNMAP_FA(struct mthca_dev *dev, u8 *status) +int mthca_UNMAP_FA(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0, 0, CMD_UNMAP_FA, CMD_TIME_CLASS_B, status); + return mthca_cmd(dev, 0, 0, 0, CMD_UNMAP_FA, CMD_TIME_CLASS_B); } -int mthca_RUN_FW(struct mthca_dev *dev, u8 *status) +int mthca_RUN_FW(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0, 0, CMD_RUN_FW, CMD_TIME_CLASS_A, status); + return mthca_cmd(dev, 0, 0, 0, CMD_RUN_FW, CMD_TIME_CLASS_A); } static void mthca_setup_cmd_doorbells(struct mthca_dev *dev, u64 base) @@ -737,7 +772,7 @@ static void mthca_setup_cmd_doorbells(struct mthca_dev *dev, u64 base) mthca_dbg(dev, "Mapped doorbell page for posting FW commands\n"); } -int mthca_QUERY_FW(struct mthca_dev *dev, u8 *status) +int mthca_QUERY_FW(struct mthca_dev *dev) { struct mthca_mailbox *mailbox; u32 *outbox; @@ -771,7 +806,7 @@ int mthca_QUERY_FW(struct mthca_dev *dev, u8 *status) outbox = mailbox->buf; err = mthca_cmd_box(dev, 0, mailbox->dma, 0, 0, CMD_QUERY_FW, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); if (err) goto out; @@ -843,7 +878,7 @@ out: return err; } -int mthca_ENABLE_LAM(struct mthca_dev *dev, u8 *status) +int mthca_ENABLE_LAM(struct mthca_dev *dev) { struct mthca_mailbox *mailbox; u8 info; @@ -864,14 +899,11 @@ int mthca_ENABLE_LAM(struct mthca_dev *dev, u8 *status) outbox = mailbox->buf; err = mthca_cmd_box(dev, 0, mailbox->dma, 0, 0, CMD_ENABLE_LAM, - CMD_TIME_CLASS_C, status); + CMD_TIME_CLASS_C); if (err) goto out; - if (*status == MTHCA_CMD_STAT_LAM_NOT_PRE) - goto out; - MTHCA_GET(dev->ddr_start, outbox, ENABLE_LAM_START_OFFSET); MTHCA_GET(dev->ddr_end, outbox, ENABLE_LAM_END_OFFSET); MTHCA_GET(info, outbox, ENABLE_LAM_INFO_OFFSET); @@ -896,12 +928,12 @@ out: return err; } -int mthca_DISABLE_LAM(struct mthca_dev *dev, u8 *status) +int mthca_DISABLE_LAM(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0, 0, CMD_SYS_DIS, CMD_TIME_CLASS_C, status); + return mthca_cmd(dev, 0, 0, 0, CMD_SYS_DIS, CMD_TIME_CLASS_C); } -int mthca_QUERY_DDR(struct mthca_dev *dev, u8 *status) +int mthca_QUERY_DDR(struct mthca_dev *dev) { struct mthca_mailbox *mailbox; u8 info; @@ -922,7 +954,7 @@ int mthca_QUERY_DDR(struct mthca_dev *dev, u8 *status) outbox = mailbox->buf; err = mthca_cmd_box(dev, 0, mailbox->dma, 0, 0, CMD_QUERY_DDR, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); if (err) goto out; @@ -952,7 +984,7 @@ out: } int mthca_QUERY_DEV_LIM(struct mthca_dev *dev, - struct mthca_dev_lim *dev_lim, u8 *status) + struct mthca_dev_lim *dev_lim) { struct mthca_mailbox *mailbox; u32 *outbox; @@ -1028,7 +1060,7 @@ int mthca_QUERY_DEV_LIM(struct mthca_dev *dev, outbox = mailbox->buf; err = mthca_cmd_box(dev, 0, mailbox->dma, 0, 0, CMD_QUERY_DEV_LIM, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); if (err) goto out; @@ -1232,7 +1264,7 @@ static void get_board_id(void *vsd, char *board_id) } int mthca_QUERY_ADAPTER(struct mthca_dev *dev, - struct mthca_adapter *adapter, u8 *status) + struct mthca_adapter *adapter) { struct mthca_mailbox *mailbox; u32 *outbox; @@ -1251,7 +1283,7 @@ int mthca_QUERY_ADAPTER(struct mthca_dev *dev, outbox = mailbox->buf; err = mthca_cmd_box(dev, 0, mailbox->dma, 0, 0, CMD_QUERY_ADAPTER, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); if (err) goto out; @@ -1275,8 +1307,7 @@ out: } int mthca_INIT_HCA(struct mthca_dev *dev, - struct mthca_init_hca_param *param, - u8 *status) + struct mthca_init_hca_param *param) { struct mthca_mailbox *mailbox; __be32 *inbox; @@ -1393,7 +1424,8 @@ int mthca_INIT_HCA(struct mthca_dev *dev, MTHCA_PUT(inbox, param->uarc_base, INIT_HCA_UAR_CTX_BASE_OFFSET); } - err = mthca_cmd(dev, mailbox->dma, 0, 0, CMD_INIT_HCA, CMD_TIME_CLASS_D, status); + err = mthca_cmd(dev, mailbox->dma, 0, 0, + CMD_INIT_HCA, CMD_TIME_CLASS_D); mthca_free_mailbox(dev, mailbox); return err; @@ -1401,7 +1433,7 @@ int mthca_INIT_HCA(struct mthca_dev *dev, int mthca_INIT_IB(struct mthca_dev *dev, struct mthca_init_ib_param *param, - int port, u8 *status) + int port) { struct mthca_mailbox *mailbox; u32 *inbox; @@ -1445,24 +1477,24 @@ int mthca_INIT_IB(struct mthca_dev *dev, MTHCA_PUT(inbox, param->si_guid, INIT_IB_SI_GUID_OFFSET); err = mthca_cmd(dev, mailbox->dma, port, 0, CMD_INIT_IB, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); mthca_free_mailbox(dev, mailbox); return err; } -int mthca_CLOSE_IB(struct mthca_dev *dev, int port, u8 *status) +int mthca_CLOSE_IB(struct mthca_dev *dev, int port) { - return mthca_cmd(dev, 0, port, 0, CMD_CLOSE_IB, CMD_TIME_CLASS_A, status); + return mthca_cmd(dev, 0, port, 0, CMD_CLOSE_IB, CMD_TIME_CLASS_A); } -int mthca_CLOSE_HCA(struct mthca_dev *dev, int panic, u8 *status) +int mthca_CLOSE_HCA(struct mthca_dev *dev, int panic) { - return mthca_cmd(dev, 0, 0, panic, CMD_CLOSE_HCA, CMD_TIME_CLASS_C, status); + return mthca_cmd(dev, 0, 0, panic, CMD_CLOSE_HCA, CMD_TIME_CLASS_C); } int mthca_SET_IB(struct mthca_dev *dev, struct mthca_set_ib_param *param, - int port, u8 *status) + int port) { struct mthca_mailbox *mailbox; u32 *inbox; @@ -1491,18 +1523,18 @@ int mthca_SET_IB(struct mthca_dev *dev, struct mthca_set_ib_param *param, MTHCA_PUT(inbox, param->si_guid, SET_IB_SI_GUID_OFFSET); err = mthca_cmd(dev, mailbox->dma, port, 0, CMD_SET_IB, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); mthca_free_mailbox(dev, mailbox); return err; } -int mthca_MAP_ICM(struct mthca_dev *dev, struct mthca_icm *icm, u64 virt, u8 *status) +int mthca_MAP_ICM(struct mthca_dev *dev, struct mthca_icm *icm, u64 virt) { - return mthca_map_cmd(dev, CMD_MAP_ICM, icm, virt, status); + return mthca_map_cmd(dev, CMD_MAP_ICM, icm, virt); } -int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt, u8 *status) +int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt) { struct mthca_mailbox *mailbox; __be64 *inbox; @@ -1517,7 +1549,7 @@ int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt, u8 *status inbox[1] = cpu_to_be64(dma_addr); err = mthca_cmd(dev, mailbox->dma, 1, 0, CMD_MAP_ICM, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); mthca_free_mailbox(dev, mailbox); @@ -1528,31 +1560,31 @@ int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt, u8 *status return err; } -int mthca_UNMAP_ICM(struct mthca_dev *dev, u64 virt, u32 page_count, u8 *status) +int mthca_UNMAP_ICM(struct mthca_dev *dev, u64 virt, u32 page_count) { mthca_dbg(dev, "Unmapping %d pages at %llx from ICM.\n", page_count, (unsigned long long) virt); - return mthca_cmd(dev, virt, page_count, 0, CMD_UNMAP_ICM, CMD_TIME_CLASS_B, status); + return mthca_cmd(dev, virt, page_count, 0, + CMD_UNMAP_ICM, CMD_TIME_CLASS_B); } -int mthca_MAP_ICM_AUX(struct mthca_dev *dev, struct mthca_icm *icm, u8 *status) +int mthca_MAP_ICM_AUX(struct mthca_dev *dev, struct mthca_icm *icm) { - return mthca_map_cmd(dev, CMD_MAP_ICM_AUX, icm, -1, status); + return mthca_map_cmd(dev, CMD_MAP_ICM_AUX, icm, -1); } -int mthca_UNMAP_ICM_AUX(struct mthca_dev *dev, u8 *status) +int mthca_UNMAP_ICM_AUX(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0, 0, CMD_UNMAP_ICM_AUX, CMD_TIME_CLASS_B, status); + return mthca_cmd(dev, 0, 0, 0, CMD_UNMAP_ICM_AUX, CMD_TIME_CLASS_B); } -int mthca_SET_ICM_SIZE(struct mthca_dev *dev, u64 icm_size, u64 *aux_pages, - u8 *status) +int mthca_SET_ICM_SIZE(struct mthca_dev *dev, u64 icm_size, u64 *aux_pages) { - int ret = mthca_cmd_imm(dev, icm_size, aux_pages, 0, 0, CMD_SET_ICM_SIZE, - CMD_TIME_CLASS_A, status); + int ret = mthca_cmd_imm(dev, icm_size, aux_pages, 0, + 0, CMD_SET_ICM_SIZE, CMD_TIME_CLASS_A); - if (ret || status) + if (ret) return ret; /* @@ -1566,74 +1598,73 @@ int mthca_SET_ICM_SIZE(struct mthca_dev *dev, u64 icm_size, u64 *aux_pages, } int mthca_SW2HW_MPT(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int mpt_index, u8 *status) + int mpt_index) { return mthca_cmd(dev, mailbox->dma, mpt_index, 0, CMD_SW2HW_MPT, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); } int mthca_HW2SW_MPT(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int mpt_index, u8 *status) + int mpt_index) { return mthca_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, mpt_index, !mailbox, CMD_HW2SW_MPT, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); } int mthca_WRITE_MTT(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int num_mtt, u8 *status) + int num_mtt) { return mthca_cmd(dev, mailbox->dma, num_mtt, 0, CMD_WRITE_MTT, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); } -int mthca_SYNC_TPT(struct mthca_dev *dev, u8 *status) +int mthca_SYNC_TPT(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0, 0, CMD_SYNC_TPT, CMD_TIME_CLASS_B, status); + return mthca_cmd(dev, 0, 0, 0, CMD_SYNC_TPT, CMD_TIME_CLASS_B); } int mthca_MAP_EQ(struct mthca_dev *dev, u64 event_mask, int unmap, - int eq_num, u8 *status) + int eq_num) { mthca_dbg(dev, "%s mask %016llx for eqn %d\n", unmap ? "Clearing" : "Setting", (unsigned long long) event_mask, eq_num); return mthca_cmd(dev, event_mask, (unmap << 31) | eq_num, - 0, CMD_MAP_EQ, CMD_TIME_CLASS_B, status); + 0, CMD_MAP_EQ, CMD_TIME_CLASS_B); } int mthca_SW2HW_EQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int eq_num, u8 *status) + int eq_num) { return mthca_cmd(dev, mailbox->dma, eq_num, 0, CMD_SW2HW_EQ, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } int mthca_HW2SW_EQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int eq_num, u8 *status) + int eq_num) { return mthca_cmd_box(dev, 0, mailbox->dma, eq_num, 0, CMD_HW2SW_EQ, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } int mthca_SW2HW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int cq_num, u8 *status) + int cq_num) { return mthca_cmd(dev, mailbox->dma, cq_num, 0, CMD_SW2HW_CQ, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } int mthca_HW2SW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int cq_num, u8 *status) + int cq_num) { return mthca_cmd_box(dev, 0, mailbox->dma, cq_num, 0, CMD_HW2SW_CQ, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } -int mthca_RESIZE_CQ(struct mthca_dev *dev, int cq_num, u32 lkey, u8 log_size, - u8 *status) +int mthca_RESIZE_CQ(struct mthca_dev *dev, int cq_num, u32 lkey, u8 log_size) { struct mthca_mailbox *mailbox; __be32 *inbox; @@ -1657,44 +1688,43 @@ int mthca_RESIZE_CQ(struct mthca_dev *dev, int cq_num, u32 lkey, u8 log_size, MTHCA_PUT(inbox, lkey, RESIZE_CQ_LKEY_OFFSET); err = mthca_cmd(dev, mailbox->dma, cq_num, 1, CMD_RESIZE_CQ, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); mthca_free_mailbox(dev, mailbox); return err; } int mthca_SW2HW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int srq_num, u8 *status) + int srq_num) { return mthca_cmd(dev, mailbox->dma, srq_num, 0, CMD_SW2HW_SRQ, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } int mthca_HW2SW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int srq_num, u8 *status) + int srq_num) { return mthca_cmd_box(dev, 0, mailbox->dma, srq_num, 0, CMD_HW2SW_SRQ, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } int mthca_QUERY_SRQ(struct mthca_dev *dev, u32 num, - struct mthca_mailbox *mailbox, u8 *status) + struct mthca_mailbox *mailbox) { return mthca_cmd_box(dev, 0, mailbox->dma, num, 0, - CMD_QUERY_SRQ, CMD_TIME_CLASS_A, status); + CMD_QUERY_SRQ, CMD_TIME_CLASS_A); } -int mthca_ARM_SRQ(struct mthca_dev *dev, int srq_num, int limit, u8 *status) +int mthca_ARM_SRQ(struct mthca_dev *dev, int srq_num, int limit) { return mthca_cmd(dev, limit, srq_num, 0, CMD_ARM_SRQ, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); } int mthca_MODIFY_QP(struct mthca_dev *dev, enum ib_qp_state cur, enum ib_qp_state next, u32 num, int is_ee, - struct mthca_mailbox *mailbox, u32 optmask, - u8 *status) + struct mthca_mailbox *mailbox, u32 optmask) { static const u16 op[IB_QPS_ERR + 1][IB_QPS_ERR + 1] = { [IB_QPS_RESET] = { @@ -1755,7 +1785,7 @@ int mthca_MODIFY_QP(struct mthca_dev *dev, enum ib_qp_state cur, err = mthca_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, (!!is_ee << 24) | num, op_mod, - op[cur][next], CMD_TIME_CLASS_C, status); + op[cur][next], CMD_TIME_CLASS_C); if (0 && mailbox) { int i; @@ -1789,21 +1819,20 @@ int mthca_MODIFY_QP(struct mthca_dev *dev, enum ib_qp_state cur, } err = mthca_cmd(dev, mailbox->dma, optmask | (!!is_ee << 24) | num, - op_mod, op[cur][next], CMD_TIME_CLASS_C, status); + op_mod, op[cur][next], CMD_TIME_CLASS_C); } return err; } int mthca_QUERY_QP(struct mthca_dev *dev, u32 num, int is_ee, - struct mthca_mailbox *mailbox, u8 *status) + struct mthca_mailbox *mailbox) { return mthca_cmd_box(dev, 0, mailbox->dma, (!!is_ee << 24) | num, 0, - CMD_QUERY_QPEE, CMD_TIME_CLASS_A, status); + CMD_QUERY_QPEE, CMD_TIME_CLASS_A); } -int mthca_CONF_SPECIAL_QP(struct mthca_dev *dev, int type, u32 qpn, - u8 *status) +int mthca_CONF_SPECIAL_QP(struct mthca_dev *dev, int type, u32 qpn) { u8 op_mod; @@ -1825,12 +1854,12 @@ int mthca_CONF_SPECIAL_QP(struct mthca_dev *dev, int type, u32 qpn, } return mthca_cmd(dev, 0, qpn, op_mod, CMD_CONF_SPECIAL_QP, - CMD_TIME_CLASS_B, status); + CMD_TIME_CLASS_B); } int mthca_MAD_IFC(struct mthca_dev *dev, int ignore_mkey, int ignore_bkey, int port, struct ib_wc *in_wc, struct ib_grh *in_grh, - void *in_mad, void *response_mad, u8 *status) + void *in_mad, void *response_mad) { struct mthca_mailbox *inmailbox, *outmailbox; void *inbox; @@ -1897,9 +1926,9 @@ int mthca_MAD_IFC(struct mthca_dev *dev, int ignore_mkey, int ignore_bkey, err = mthca_cmd_box(dev, inmailbox->dma, outmailbox->dma, in_modifier, op_modifier, - CMD_MAD_IFC, CMD_TIME_CLASS_C, status); + CMD_MAD_IFC, CMD_TIME_CLASS_C); - if (!err && !*status) + if (!err) memcpy(response_mad, outmailbox->buf, 256); mthca_free_mailbox(dev, inmailbox); @@ -1908,33 +1937,33 @@ int mthca_MAD_IFC(struct mthca_dev *dev, int ignore_mkey, int ignore_bkey, } int mthca_READ_MGM(struct mthca_dev *dev, int index, - struct mthca_mailbox *mailbox, u8 *status) + struct mthca_mailbox *mailbox) { return mthca_cmd_box(dev, 0, mailbox->dma, index, 0, - CMD_READ_MGM, CMD_TIME_CLASS_A, status); + CMD_READ_MGM, CMD_TIME_CLASS_A); } int mthca_WRITE_MGM(struct mthca_dev *dev, int index, - struct mthca_mailbox *mailbox, u8 *status) + struct mthca_mailbox *mailbox) { return mthca_cmd(dev, mailbox->dma, index, 0, CMD_WRITE_MGM, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); } int mthca_MGID_HASH(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - u16 *hash, u8 *status) + u16 *hash) { u64 imm; int err; err = mthca_cmd_imm(dev, mailbox->dma, &imm, 0, 0, CMD_MGID_HASH, - CMD_TIME_CLASS_A, status); + CMD_TIME_CLASS_A); *hash = imm; return err; } -int mthca_NOP(struct mthca_dev *dev, u8 *status) +int mthca_NOP(struct mthca_dev *dev) { - return mthca_cmd(dev, 0, 0x1f, 0, CMD_NOP, msecs_to_jiffies(100), status); + return mthca_cmd(dev, 0, 0x1f, 0, CMD_NOP, msecs_to_jiffies(100)); } diff --git a/drivers/infiniband/hw/mthca/mthca_cmd.h b/drivers/infiniband/hw/mthca/mthca_cmd.h index 6efd326..f952244 100644 --- a/drivers/infiniband/hw/mthca/mthca_cmd.h +++ b/drivers/infiniband/hw/mthca/mthca_cmd.h @@ -252,79 +252,74 @@ struct mthca_mailbox *mthca_alloc_mailbox(struct mthca_dev *dev, gfp_t gfp_mask); void mthca_free_mailbox(struct mthca_dev *dev, struct mthca_mailbox *mailbox); -int mthca_SYS_EN(struct mthca_dev *dev, u8 *status); -int mthca_SYS_DIS(struct mthca_dev *dev, u8 *status); -int mthca_MAP_FA(struct mthca_dev *dev, struct mthca_icm *icm, u8 *status); -int mthca_UNMAP_FA(struct mthca_dev *dev, u8 *status); -int mthca_RUN_FW(struct mthca_dev *dev, u8 *status); -int mthca_QUERY_FW(struct mthca_dev *dev, u8 *status); -int mthca_ENABLE_LAM(struct mthca_dev *dev, u8 *status); -int mthca_DISABLE_LAM(struct mthca_dev *dev, u8 *status); -int mthca_QUERY_DDR(struct mthca_dev *dev, u8 *status); +int mthca_SYS_EN(struct mthca_dev *dev); +int mthca_SYS_DIS(struct mthca_dev *dev); +int mthca_MAP_FA(struct mthca_dev *dev, struct mthca_icm *icm); +int mthca_UNMAP_FA(struct mthca_dev *dev); +int mthca_RUN_FW(struct mthca_dev *dev); +int mthca_QUERY_FW(struct mthca_dev *dev); +int mthca_ENABLE_LAM(struct mthca_dev *dev); +int mthca_DISABLE_LAM(struct mthca_dev *dev); +int mthca_QUERY_DDR(struct mthca_dev *dev); int mthca_QUERY_DEV_LIM(struct mthca_dev *dev, - struct mthca_dev_lim *dev_lim, u8 *status); + struct mthca_dev_lim *dev_lim); int mthca_QUERY_ADAPTER(struct mthca_dev *dev, - struct mthca_adapter *adapter, u8 *status); + struct mthca_adapter *adapter); int mthca_INIT_HCA(struct mthca_dev *dev, - struct mthca_init_hca_param *param, - u8 *status); + struct mthca_init_hca_param *param); int mthca_INIT_IB(struct mthca_dev *dev, struct mthca_init_ib_param *param, - int port, u8 *status); -int mthca_CLOSE_IB(struct mthca_dev *dev, int port, u8 *status); -int mthca_CLOSE_HCA(struct mthca_dev *dev, int panic, u8 *status); + int port); +int mthca_CLOSE_IB(struct mthca_dev *dev, int port); +int mthca_CLOSE_HCA(struct mthca_dev *dev, int panic); int mthca_SET_IB(struct mthca_dev *dev, struct mthca_set_ib_param *param, - int port, u8 *status); -int mthca_MAP_ICM(struct mthca_dev *dev, struct mthca_icm *icm, u64 virt, u8 *status); -int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt, u8 *status); -int mthca_UNMAP_ICM(struct mthca_dev *dev, u64 virt, u32 page_count, u8 *status); -int mthca_MAP_ICM_AUX(struct mthca_dev *dev, struct mthca_icm *icm, u8 *status); -int mthca_UNMAP_ICM_AUX(struct mthca_dev *dev, u8 *status); -int mthca_SET_ICM_SIZE(struct mthca_dev *dev, u64 icm_size, u64 *aux_pages, - u8 *status); + int port); +int mthca_MAP_ICM(struct mthca_dev *dev, struct mthca_icm *icm, u64 virt); +int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt); +int mthca_UNMAP_ICM(struct mthca_dev *dev, u64 virt, u32 page_count); +int mthca_MAP_ICM_AUX(struct mthca_dev *dev, struct mthca_icm *icm); +int mthca_UNMAP_ICM_AUX(struct mthca_dev *dev); +int mthca_SET_ICM_SIZE(struct mthca_dev *dev, u64 icm_size, u64 *aux_pages); int mthca_SW2HW_MPT(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int mpt_index, u8 *status); + int mpt_index); int mthca_HW2SW_MPT(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int mpt_index, u8 *status); + int mpt_index); int mthca_WRITE_MTT(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int num_mtt, u8 *status); -int mthca_SYNC_TPT(struct mthca_dev *dev, u8 *status); + int num_mtt); +int mthca_SYNC_TPT(struct mthca_dev *dev); int mthca_MAP_EQ(struct mthca_dev *dev, u64 event_mask, int unmap, - int eq_num, u8 *status); + int eq_num); int mthca_SW2HW_EQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int eq_num, u8 *status); + int eq_num); int mthca_HW2SW_EQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int eq_num, u8 *status); + int eq_num); int mthca_SW2HW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int cq_num, u8 *status); + int cq_num); int mthca_HW2SW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int cq_num, u8 *status); -int mthca_RESIZE_CQ(struct mthca_dev *dev, int cq_num, u32 lkey, u8 log_size, - u8 *status); + int cq_num); +int mthca_RESIZE_CQ(struct mthca_dev *dev, int cq_num, u32 lkey, u8 log_size); int mthca_SW2HW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int srq_num, u8 *status); + int srq_num); int mthca_HW2SW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - int srq_num, u8 *status); + int srq_num); int mthca_QUERY_SRQ(struct mthca_dev *dev, u32 num, - struct mthca_mailbox *mailbox, u8 *status); -int mthca_ARM_SRQ(struct mthca_dev *dev, int srq_num, int limit, u8 *status); + struct mthca_mailbox *mailbox); +int mthca_ARM_SRQ(struct mthca_dev *dev, int srq_num, int limit); int mthca_MODIFY_QP(struct mthca_dev *dev, enum ib_qp_state cur, enum ib_qp_state next, u32 num, int is_ee, - struct mthca_mailbox *mailbox, u32 optmask, - u8 *status); + struct mthca_mailbox *mailbox, u32 optmask); int mthca_QUERY_QP(struct mthca_dev *dev, u32 num, int is_ee, - struct mthca_mailbox *mailbox, u8 *status); -int mthca_CONF_SPECIAL_QP(struct mthca_dev *dev, int type, u32 qpn, - u8 *status); + struct mthca_mailbox *mailbox); +int mthca_CONF_SPECIAL_QP(struct mthca_dev *dev, int type, u32 qpn); int mthca_MAD_IFC(struct mthca_dev *dev, int ignore_mkey, int ignore_bkey, int port, struct ib_wc *in_wc, struct ib_grh *in_grh, - void *in_mad, void *response_mad, u8 *status); + void *in_mad, void *response_mad); int mthca_READ_MGM(struct mthca_dev *dev, int index, - struct mthca_mailbox *mailbox, u8 *status); + struct mthca_mailbox *mailbox); int mthca_WRITE_MGM(struct mthca_dev *dev, int index, - struct mthca_mailbox *mailbox, u8 *status); + struct mthca_mailbox *mailbox); int mthca_MGID_HASH(struct mthca_dev *dev, struct mthca_mailbox *mailbox, - u16 *hash, u8 *status); -int mthca_NOP(struct mthca_dev *dev, u8 *status); + u16 *hash); +int mthca_NOP(struct mthca_dev *dev); #endif /* MTHCA_CMD_H */ diff --git a/drivers/infiniband/hw/mthca/mthca_cq.c b/drivers/infiniband/hw/mthca/mthca_cq.c index 18ee3fa..53157b8 100644 --- a/drivers/infiniband/hw/mthca/mthca_cq.c +++ b/drivers/infiniband/hw/mthca/mthca_cq.c @@ -779,7 +779,6 @@ int mthca_init_cq(struct mthca_dev *dev, int nent, struct mthca_mailbox *mailbox; struct mthca_cq_context *cq_context; int err = -ENOMEM; - u8 status; cq->ibcq.cqe = nent - 1; cq->is_kernel = !ctx; @@ -847,19 +846,12 @@ int mthca_init_cq(struct mthca_dev *dev, int nent, cq_context->state_db = cpu_to_be32(cq->arm_db_index); } - err = mthca_SW2HW_CQ(dev, mailbox, cq->cqn, &status); + err = mthca_SW2HW_CQ(dev, mailbox, cq->cqn); if (err) { mthca_warn(dev, "SW2HW_CQ failed (%d)\n", err); goto err_out_free_mr; } - if (status) { - mthca_warn(dev, "SW2HW_CQ returned status 0x%02x\n", - status); - err = -EINVAL; - goto err_out_free_mr; - } - spin_lock_irq(&dev->cq_table.lock); if (mthca_array_set(&dev->cq_table.cq, cq->cqn & (dev->limits.num_cqs - 1), @@ -915,7 +907,6 @@ void mthca_free_cq(struct mthca_dev *dev, { struct mthca_mailbox *mailbox; int err; - u8 status; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); if (IS_ERR(mailbox)) { @@ -923,11 +914,9 @@ void mthca_free_cq(struct mthca_dev *dev, return; } - err = mthca_HW2SW_CQ(dev, mailbox, cq->cqn, &status); + err = mthca_HW2SW_CQ(dev, mailbox, cq->cqn); if (err) mthca_warn(dev, "HW2SW_CQ failed (%d)\n", err); - else if (status) - mthca_warn(dev, "HW2SW_CQ returned status 0x%02x\n", status); if (0) { __be32 *ctx = mailbox->buf; diff --git a/drivers/infiniband/hw/mthca/mthca_eq.c b/drivers/infiniband/hw/mthca/mthca_eq.c index 76785c6..7c9d35f 100644 --- a/drivers/infiniband/hw/mthca/mthca_eq.c +++ b/drivers/infiniband/hw/mthca/mthca_eq.c @@ -474,7 +474,6 @@ static int mthca_create_eq(struct mthca_dev *dev, struct mthca_eq_context *eq_context; int err = -ENOMEM; int i; - u8 status; eq->dev = dev; eq->nent = roundup_pow_of_two(max(nent, 2)); @@ -543,15 +542,9 @@ static int mthca_create_eq(struct mthca_dev *dev, eq_context->intr = intr; eq_context->lkey = cpu_to_be32(eq->mr.ibmr.lkey); - err = mthca_SW2HW_EQ(dev, mailbox, eq->eqn, &status); + err = mthca_SW2HW_EQ(dev, mailbox, eq->eqn); if (err) { - mthca_warn(dev, "SW2HW_EQ failed (%d)\n", err); - goto err_out_free_mr; - } - if (status) { - mthca_warn(dev, "SW2HW_EQ returned status 0x%02x\n", - status); - err = -EINVAL; + mthca_warn(dev, "SW2HW_EQ returned %d\n", err); goto err_out_free_mr; } @@ -597,7 +590,6 @@ static void mthca_free_eq(struct mthca_dev *dev, { struct mthca_mailbox *mailbox; int err; - u8 status; int npages = (eq->nent * MTHCA_EQ_ENTRY_SIZE + PAGE_SIZE - 1) / PAGE_SIZE; int i; @@ -606,11 +598,9 @@ static void mthca_free_eq(struct mthca_dev *dev, if (IS_ERR(mailbox)) return; - err = mthca_HW2SW_EQ(dev, mailbox, eq->eqn, &status); + err = mthca_HW2SW_EQ(dev, mailbox, eq->eqn); if (err) - mthca_warn(dev, "HW2SW_EQ failed (%d)\n", err); - if (status) - mthca_warn(dev, "HW2SW_EQ returned status 0x%02x\n", status); + mthca_warn(dev, "HW2SW_EQ returned %d\n", err); dev->eq_table.arm_mask &= ~eq->eqn_mask; @@ -738,7 +728,6 @@ static void mthca_unmap_eq_regs(struct mthca_dev *dev) int mthca_map_eq_icm(struct mthca_dev *dev, u64 icm_virt) { int ret; - u8 status; /* * We assume that mapping one page is enough for the whole EQ @@ -757,9 +746,7 @@ int mthca_map_eq_icm(struct mthca_dev *dev, u64 icm_virt) return -ENOMEM; } - ret = mthca_MAP_ICM_page(dev, dev->eq_table.icm_dma, icm_virt, &status); - if (!ret && status) - ret = -EINVAL; + ret = mthca_MAP_ICM_page(dev, dev->eq_table.icm_dma, icm_virt); if (ret) { pci_unmap_page(dev->pdev, dev->eq_table.icm_dma, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); @@ -771,9 +758,7 @@ int mthca_map_eq_icm(struct mthca_dev *dev, u64 icm_virt) void mthca_unmap_eq_icm(struct mthca_dev *dev) { - u8 status; - - mthca_UNMAP_ICM(dev, dev->eq_table.icm_virt, 1, &status); + mthca_UNMAP_ICM(dev, dev->eq_table.icm_virt, 1); pci_unmap_page(dev->pdev, dev->eq_table.icm_dma, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); __free_page(dev->eq_table.icm_page); @@ -782,7 +767,6 @@ void mthca_unmap_eq_icm(struct mthca_dev *dev) int mthca_init_eq_table(struct mthca_dev *dev) { int err; - u8 status; u8 intr; int i; @@ -864,22 +848,16 @@ int mthca_init_eq_table(struct mthca_dev *dev) } err = mthca_MAP_EQ(dev, async_mask(dev), - 0, dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, &status); + 0, dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn); if (err) mthca_warn(dev, "MAP_EQ for async EQ %d failed (%d)\n", dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, err); - if (status) - mthca_warn(dev, "MAP_EQ for async EQ %d returned status 0x%02x\n", - dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, status); err = mthca_MAP_EQ(dev, MTHCA_CMD_EVENT_MASK, - 0, dev->eq_table.eq[MTHCA_EQ_CMD].eqn, &status); + 0, dev->eq_table.eq[MTHCA_EQ_CMD].eqn); if (err) mthca_warn(dev, "MAP_EQ for cmd EQ %d failed (%d)\n", dev->eq_table.eq[MTHCA_EQ_CMD].eqn, err); - if (status) - mthca_warn(dev, "MAP_EQ for cmd EQ %d returned status 0x%02x\n", - dev->eq_table.eq[MTHCA_EQ_CMD].eqn, status); for (i = 0; i < MTHCA_NUM_EQ; ++i) if (mthca_is_memfree(dev)) @@ -909,15 +887,14 @@ err_out_free: void mthca_cleanup_eq_table(struct mthca_dev *dev) { - u8 status; int i; mthca_free_irqs(dev); mthca_MAP_EQ(dev, async_mask(dev), - 1, dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn, &status); + 1, dev->eq_table.eq[MTHCA_EQ_ASYNC].eqn); mthca_MAP_EQ(dev, MTHCA_CMD_EVENT_MASK, - 1, dev->eq_table.eq[MTHCA_EQ_CMD].eqn, &status); + 1, dev->eq_table.eq[MTHCA_EQ_CMD].eqn); for (i = 0; i < MTHCA_NUM_EQ; ++i) mthca_free_eq(dev, &dev->eq_table.eq[i]); diff --git a/drivers/infiniband/hw/mthca/mthca_mad.c b/drivers/infiniband/hw/mthca/mthca_mad.c index 03a59534..b6f7f45 100644 --- a/drivers/infiniband/hw/mthca/mthca_mad.c +++ b/drivers/infiniband/hw/mthca/mthca_mad.c @@ -201,7 +201,6 @@ int mthca_process_mad(struct ib_device *ibdev, struct ib_mad *out_mad) { int err; - u8 status; u16 slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE); u16 prev_lid = 0; struct ib_port_attr pattr; @@ -252,17 +251,11 @@ int mthca_process_mad(struct ib_device *ibdev, err = mthca_MAD_IFC(to_mdev(ibdev), mad_flags & IB_MAD_IGNORE_MKEY, mad_flags & IB_MAD_IGNORE_BKEY, - port_num, in_wc, in_grh, in_mad, out_mad, - &status); - if (err) { - mthca_err(to_mdev(ibdev), "MAD_IFC failed\n"); - return IB_MAD_RESULT_FAILURE; - } - if (status == MTHCA_CMD_STAT_BAD_PKT) + port_num, in_wc, in_grh, in_mad, out_mad); + if (err == -EBADMSG) return IB_MAD_RESULT_SUCCESS; - if (status) { - mthca_err(to_mdev(ibdev), "MAD_IFC returned status %02x\n", - status); + else if (err) { + mthca_err(to_mdev(ibdev), "MAD_IFC returned %d\n", err); return IB_MAD_RESULT_FAILURE; } diff --git a/drivers/infiniband/hw/mthca/mthca_main.c b/drivers/infiniband/hw/mthca/mthca_main.c index f24b79b..aa12a53 100644 --- a/drivers/infiniband/hw/mthca/mthca_main.c +++ b/drivers/infiniband/hw/mthca/mthca_main.c @@ -149,7 +149,7 @@ static int mthca_tune_pci(struct mthca_dev *mdev) } else if (!(mdev->mthca_flags & MTHCA_FLAG_PCIE)) mthca_info(mdev, "No PCI-X capability, not setting RBC.\n"); - if (pci_find_capability(mdev->pdev, PCI_CAP_ID_EXP)) { + if (pci_is_pcie(mdev->pdev)) { if (pcie_set_readrq(mdev->pdev, 4096)) { mthca_err(mdev, "Couldn't write PCI Express read request, " "aborting.\n"); @@ -165,19 +165,14 @@ static int mthca_tune_pci(struct mthca_dev *mdev) static int mthca_dev_lim(struct mthca_dev *mdev, struct mthca_dev_lim *dev_lim) { int err; - u8 status; mdev->limits.mtt_seg_size = (1 << log_mtts_per_seg) * 8; - err = mthca_QUERY_DEV_LIM(mdev, dev_lim, &status); + err = mthca_QUERY_DEV_LIM(mdev, dev_lim); if (err) { - mthca_err(mdev, "QUERY_DEV_LIM command failed, aborting.\n"); + mthca_err(mdev, "QUERY_DEV_LIM command returned %d" + ", aborting.\n", err); return err; } - if (status) { - mthca_err(mdev, "QUERY_DEV_LIM returned status 0x%02x, " - "aborting.\n", status); - return -EINVAL; - } if (dev_lim->min_page_sz > PAGE_SIZE) { mthca_err(mdev, "HCA minimum page size of %d bigger than " "kernel PAGE_SIZE of %ld, aborting.\n", @@ -293,49 +288,32 @@ static int mthca_dev_lim(struct mthca_dev *mdev, struct mthca_dev_lim *dev_lim) static int mthca_init_tavor(struct mthca_dev *mdev) { s64 size; - u8 status; int err; struct mthca_dev_lim dev_lim; struct mthca_profile profile; struct mthca_init_hca_param init_hca; - err = mthca_SYS_EN(mdev, &status); + err = mthca_SYS_EN(mdev); if (err) { - mthca_err(mdev, "SYS_EN command failed, aborting.\n"); + mthca_err(mdev, "SYS_EN command returned %d, aborting.\n", err); return err; } - if (status) { - mthca_err(mdev, "SYS_EN returned status 0x%02x, " - "aborting.\n", status); - return -EINVAL; - } - err = mthca_QUERY_FW(mdev, &status); + err = mthca_QUERY_FW(mdev); if (err) { - mthca_err(mdev, "QUERY_FW command failed, aborting.\n"); - goto err_disable; - } - if (status) { - mthca_err(mdev, "QUERY_FW returned status 0x%02x, " - "aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "QUERY_FW command returned %d," + " aborting.\n", err); goto err_disable; } - err = mthca_QUERY_DDR(mdev, &status); + err = mthca_QUERY_DDR(mdev); if (err) { - mthca_err(mdev, "QUERY_DDR command failed, aborting.\n"); - goto err_disable; - } - if (status) { - mthca_err(mdev, "QUERY_DDR returned status 0x%02x, " - "aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "QUERY_DDR command returned %d, aborting.\n", err); goto err_disable; } err = mthca_dev_lim(mdev, &dev_lim); if (err) { - mthca_err(mdev, "QUERY_DEV_LIM command failed, aborting.\n"); + mthca_err(mdev, "QUERY_DEV_LIM command returned %d, aborting.\n", err); goto err_disable; } @@ -351,29 +329,22 @@ static int mthca_init_tavor(struct mthca_dev *mdev) goto err_disable; } - err = mthca_INIT_HCA(mdev, &init_hca, &status); + err = mthca_INIT_HCA(mdev, &init_hca); if (err) { - mthca_err(mdev, "INIT_HCA command failed, aborting.\n"); - goto err_disable; - } - if (status) { - mthca_err(mdev, "INIT_HCA returned status 0x%02x, " - "aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "INIT_HCA command returned %d, aborting.\n", err); goto err_disable; } return 0; err_disable: - mthca_SYS_DIS(mdev, &status); + mthca_SYS_DIS(mdev); return err; } static int mthca_load_fw(struct mthca_dev *mdev) { - u8 status; int err; /* FIXME: use HCA-attached memory for FW if present */ @@ -386,31 +357,21 @@ static int mthca_load_fw(struct mthca_dev *mdev) return -ENOMEM; } - err = mthca_MAP_FA(mdev, mdev->fw.arbel.fw_icm, &status); + err = mthca_MAP_FA(mdev, mdev->fw.arbel.fw_icm); if (err) { - mthca_err(mdev, "MAP_FA command failed, aborting.\n"); - goto err_free; - } - if (status) { - mthca_err(mdev, "MAP_FA returned status 0x%02x, aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "MAP_FA command returned %d, aborting.\n", err); goto err_free; } - err = mthca_RUN_FW(mdev, &status); + err = mthca_RUN_FW(mdev); if (err) { - mthca_err(mdev, "RUN_FW command failed, aborting.\n"); - goto err_unmap_fa; - } - if (status) { - mthca_err(mdev, "RUN_FW returned status 0x%02x, aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "RUN_FW command returned %d, aborting.\n", err); goto err_unmap_fa; } return 0; err_unmap_fa: - mthca_UNMAP_FA(mdev, &status); + mthca_UNMAP_FA(mdev); err_free: mthca_free_icm(mdev, mdev->fw.arbel.fw_icm, 0); @@ -423,19 +384,13 @@ static int mthca_init_icm(struct mthca_dev *mdev, u64 icm_size) { u64 aux_pages; - u8 status; int err; - err = mthca_SET_ICM_SIZE(mdev, icm_size, &aux_pages, &status); + err = mthca_SET_ICM_SIZE(mdev, icm_size, &aux_pages); if (err) { - mthca_err(mdev, "SET_ICM_SIZE command failed, aborting.\n"); + mthca_err(mdev, "SET_ICM_SIZE command returned %d, aborting.\n", err); return err; } - if (status) { - mthca_err(mdev, "SET_ICM_SIZE returned status 0x%02x, " - "aborting.\n", status); - return -EINVAL; - } mthca_dbg(mdev, "%lld KB of HCA context requires %lld KB aux memory.\n", (unsigned long long) icm_size >> 10, @@ -448,14 +403,9 @@ static int mthca_init_icm(struct mthca_dev *mdev, return -ENOMEM; } - err = mthca_MAP_ICM_AUX(mdev, mdev->fw.arbel.aux_icm, &status); + err = mthca_MAP_ICM_AUX(mdev, mdev->fw.arbel.aux_icm); if (err) { - mthca_err(mdev, "MAP_ICM_AUX command failed, aborting.\n"); - goto err_free_aux; - } - if (status) { - mthca_err(mdev, "MAP_ICM_AUX returned status 0x%02x, aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "MAP_ICM_AUX returned %d, aborting.\n", err); goto err_free_aux; } @@ -596,7 +546,7 @@ err_unmap_eq: mthca_unmap_eq_icm(mdev); err_unmap_aux: - mthca_UNMAP_ICM_AUX(mdev, &status); + mthca_UNMAP_ICM_AUX(mdev); err_free_aux: mthca_free_icm(mdev, mdev->fw.arbel.aux_icm, 0); @@ -606,7 +556,6 @@ err_free_aux: static void mthca_free_icms(struct mthca_dev *mdev) { - u8 status; mthca_free_icm_table(mdev, mdev->mcg_table.table); if (mdev->mthca_flags & MTHCA_FLAG_SRQ) @@ -619,7 +568,7 @@ static void mthca_free_icms(struct mthca_dev *mdev) mthca_free_icm_table(mdev, mdev->mr_table.mtt_table); mthca_unmap_eq_icm(mdev); - mthca_UNMAP_ICM_AUX(mdev, &status); + mthca_UNMAP_ICM_AUX(mdev); mthca_free_icm(mdev, mdev->fw.arbel.aux_icm, 0); } @@ -629,43 +578,32 @@ static int mthca_init_arbel(struct mthca_dev *mdev) struct mthca_profile profile; struct mthca_init_hca_param init_hca; s64 icm_size; - u8 status; int err; - err = mthca_QUERY_FW(mdev, &status); + err = mthca_QUERY_FW(mdev); if (err) { - mthca_err(mdev, "QUERY_FW command failed, aborting.\n"); + mthca_err(mdev, "QUERY_FW command failed %d, aborting.\n", err); return err; } - if (status) { - mthca_err(mdev, "QUERY_FW returned status 0x%02x, " - "aborting.\n", status); - return -EINVAL; - } - err = mthca_ENABLE_LAM(mdev, &status); - if (err) { - mthca_err(mdev, "ENABLE_LAM command failed, aborting.\n"); - return err; - } - if (status == MTHCA_CMD_STAT_LAM_NOT_PRE) { + err = mthca_ENABLE_LAM(mdev); + if (err == -EAGAIN) { mthca_dbg(mdev, "No HCA-attached memory (running in MemFree mode)\n"); mdev->mthca_flags |= MTHCA_FLAG_NO_LAM; - } else if (status) { - mthca_err(mdev, "ENABLE_LAM returned status 0x%02x, " - "aborting.\n", status); - return -EINVAL; + } else if (err) { + mthca_err(mdev, "ENABLE_LAM returned %d, aborting.\n", err); + return err; } err = mthca_load_fw(mdev); if (err) { - mthca_err(mdev, "Failed to start FW, aborting.\n"); + mthca_err(mdev, "Loading FW returned %d, aborting.\n", err); goto err_disable; } err = mthca_dev_lim(mdev, &dev_lim); if (err) { - mthca_err(mdev, "QUERY_DEV_LIM command failed, aborting.\n"); + mthca_err(mdev, "QUERY_DEV_LIM returned %d, aborting.\n", err); goto err_stop_fw; } @@ -685,15 +623,9 @@ static int mthca_init_arbel(struct mthca_dev *mdev) if (err) goto err_stop_fw; - err = mthca_INIT_HCA(mdev, &init_hca, &status); + err = mthca_INIT_HCA(mdev, &init_hca); if (err) { - mthca_err(mdev, "INIT_HCA command failed, aborting.\n"); - goto err_free_icm; - } - if (status) { - mthca_err(mdev, "INIT_HCA returned status 0x%02x, " - "aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "INIT_HCA command returned %d, aborting.\n", err); goto err_free_icm; } @@ -703,37 +635,34 @@ err_free_icm: mthca_free_icms(mdev); err_stop_fw: - mthca_UNMAP_FA(mdev, &status); + mthca_UNMAP_FA(mdev); mthca_free_icm(mdev, mdev->fw.arbel.fw_icm, 0); err_disable: if (!(mdev->mthca_flags & MTHCA_FLAG_NO_LAM)) - mthca_DISABLE_LAM(mdev, &status); + mthca_DISABLE_LAM(mdev); return err; } static void mthca_close_hca(struct mthca_dev *mdev) { - u8 status; - - mthca_CLOSE_HCA(mdev, 0, &status); + mthca_CLOSE_HCA(mdev, 0); if (mthca_is_memfree(mdev)) { mthca_free_icms(mdev); - mthca_UNMAP_FA(mdev, &status); + mthca_UNMAP_FA(mdev); mthca_free_icm(mdev, mdev->fw.arbel.fw_icm, 0); if (!(mdev->mthca_flags & MTHCA_FLAG_NO_LAM)) - mthca_DISABLE_LAM(mdev, &status); + mthca_DISABLE_LAM(mdev); } else - mthca_SYS_DIS(mdev, &status); + mthca_SYS_DIS(mdev); } static int mthca_init_hca(struct mthca_dev *mdev) { - u8 status; int err; struct mthca_adapter adapter; @@ -745,15 +674,9 @@ static int mthca_init_hca(struct mthca_dev *mdev) if (err) return err; - err = mthca_QUERY_ADAPTER(mdev, &adapter, &status); + err = mthca_QUERY_ADAPTER(mdev, &adapter); if (err) { - mthca_err(mdev, "QUERY_ADAPTER command failed, aborting.\n"); - goto err_close; - } - if (status) { - mthca_err(mdev, "QUERY_ADAPTER returned status 0x%02x, " - "aborting.\n", status); - err = -EINVAL; + mthca_err(mdev, "QUERY_ADAPTER command returned %d, aborting.\n", err); goto err_close; } @@ -772,7 +695,6 @@ err_close: static int mthca_setup_hca(struct mthca_dev *dev) { int err; - u8 status; MTHCA_INIT_DOORBELL_LOCK(&dev->doorbell_lock); @@ -833,8 +755,8 @@ static int mthca_setup_hca(struct mthca_dev *dev) goto err_eq_table_free; } - err = mthca_NOP(dev, &status); - if (err || status) { + err = mthca_NOP(dev); + if (err) { if (dev->mthca_flags & MTHCA_FLAG_MSI_X) { mthca_warn(dev, "NOP command failed to generate interrupt " "(IRQ %d).\n", @@ -1166,7 +1088,6 @@ err_disable_pdev: static void __mthca_remove_one(struct pci_dev *pdev) { struct mthca_dev *mdev = pci_get_drvdata(pdev); - u8 status; int p; if (mdev) { @@ -1174,7 +1095,7 @@ static void __mthca_remove_one(struct pci_dev *pdev) mthca_unregister_device(mdev); for (p = 1; p <= mdev->limits.num_ports; ++p) - mthca_CLOSE_IB(mdev, p, &status); + mthca_CLOSE_IB(mdev, p); mthca_cleanup_mcg_table(mdev); mthca_cleanup_av_table(mdev); diff --git a/drivers/infiniband/hw/mthca/mthca_mcg.c b/drivers/infiniband/hw/mthca/mthca_mcg.c index 515790a..6304ae8 100644 --- a/drivers/infiniband/hw/mthca/mthca_mcg.c +++ b/drivers/infiniband/hw/mthca/mthca_mcg.c @@ -68,7 +68,6 @@ static int find_mgm(struct mthca_dev *dev, struct mthca_mgm *mgm = mgm_mailbox->buf; u8 *mgid; int err; - u8 status; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); if (IS_ERR(mailbox)) @@ -77,12 +76,9 @@ static int find_mgm(struct mthca_dev *dev, memcpy(mgid, gid, 16); - err = mthca_MGID_HASH(dev, mailbox, hash, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "MGID_HASH returned status %02x\n", status); - err = -EINVAL; + err = mthca_MGID_HASH(dev, mailbox, hash); + if (err) { + mthca_err(dev, "MGID_HASH failed (%d)\n", err); goto out; } @@ -93,12 +89,9 @@ static int find_mgm(struct mthca_dev *dev, *prev = -1; do { - err = mthca_READ_MGM(dev, *index, mgm_mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "READ_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_READ_MGM(dev, *index, mgm_mailbox); + if (err) { + mthca_err(dev, "READ_MGM failed (%d)\n", err); goto out; } @@ -134,7 +127,6 @@ int mthca_multicast_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) int link = 0; int i; int err; - u8 status; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); if (IS_ERR(mailbox)) @@ -160,12 +152,9 @@ int mthca_multicast_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) goto out; } - err = mthca_READ_MGM(dev, index, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "READ_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_READ_MGM(dev, index, mailbox); + if (err) { + mthca_err(dev, "READ_MGM failed (%d)\n", err); goto out; } memset(mgm, 0, sizeof *mgm); @@ -189,11 +178,9 @@ int mthca_multicast_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) goto out; } - err = mthca_WRITE_MGM(dev, index, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "WRITE_MGM returned status %02x\n", status); + err = mthca_WRITE_MGM(dev, index, mailbox); + if (err) { + mthca_err(dev, "WRITE_MGM failed %d\n", err); err = -EINVAL; goto out; } @@ -201,24 +188,17 @@ int mthca_multicast_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) if (!link) goto out; - err = mthca_READ_MGM(dev, prev, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "READ_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_READ_MGM(dev, prev, mailbox); + if (err) { + mthca_err(dev, "READ_MGM failed %d\n", err); goto out; } mgm->next_gid_index = cpu_to_be32(index << 6); - err = mthca_WRITE_MGM(dev, prev, mailbox, &status); + err = mthca_WRITE_MGM(dev, prev, mailbox); if (err) - goto out; - if (status) { - mthca_err(dev, "WRITE_MGM returned status %02x\n", status); - err = -EINVAL; - } + mthca_err(dev, "WRITE_MGM returned %d\n", err); out: if (err && link && index != -1) { @@ -240,7 +220,6 @@ int mthca_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) int prev, index; int i, loc; int err; - u8 status; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); if (IS_ERR(mailbox)) @@ -275,12 +254,9 @@ int mthca_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) mgm->qp[loc] = mgm->qp[i - 1]; mgm->qp[i - 1] = 0; - err = mthca_WRITE_MGM(dev, index, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "WRITE_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_WRITE_MGM(dev, index, mailbox); + if (err) { + mthca_err(dev, "WRITE_MGM returned %d\n", err); goto out; } @@ -292,24 +268,17 @@ int mthca_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) int amgm_index_to_free = be32_to_cpu(mgm->next_gid_index) >> 6; if (amgm_index_to_free) { err = mthca_READ_MGM(dev, amgm_index_to_free, - mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "READ_MGM returned status %02x\n", - status); - err = -EINVAL; + mailbox); + if (err) { + mthca_err(dev, "READ_MGM returned %d\n", err); goto out; } } else memset(mgm->gid, 0, 16); - err = mthca_WRITE_MGM(dev, index, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "WRITE_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_WRITE_MGM(dev, index, mailbox); + if (err) { + mthca_err(dev, "WRITE_MGM returned %d\n", err); goto out; } if (amgm_index_to_free) { @@ -319,23 +288,17 @@ int mthca_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) } else { /* Remove entry from AMGM */ int curr_next_index = be32_to_cpu(mgm->next_gid_index) >> 6; - err = mthca_READ_MGM(dev, prev, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "READ_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_READ_MGM(dev, prev, mailbox); + if (err) { + mthca_err(dev, "READ_MGM returned %d\n", err); goto out; } mgm->next_gid_index = cpu_to_be32(curr_next_index << 6); - err = mthca_WRITE_MGM(dev, prev, mailbox, &status); - if (err) - goto out; - if (status) { - mthca_err(dev, "WRITE_MGM returned status %02x\n", status); - err = -EINVAL; + err = mthca_WRITE_MGM(dev, prev, mailbox); + if (err) { + mthca_err(dev, "WRITE_MGM returned %d\n", err); goto out; } BUG_ON(index < dev->limits.num_mgms); diff --git a/drivers/infiniband/hw/mthca/mthca_memfree.c b/drivers/infiniband/hw/mthca/mthca_memfree.c index 8c2a837..7d2e42d 100644 --- a/drivers/infiniband/hw/mthca/mthca_memfree.c +++ b/drivers/infiniband/hw/mthca/mthca_memfree.c @@ -223,7 +223,6 @@ int mthca_table_get(struct mthca_dev *dev, struct mthca_icm_table *table, int ob { int i = (obj & (table->num_obj - 1)) * table->obj_size / MTHCA_TABLE_CHUNK_SIZE; int ret = 0; - u8 status; mutex_lock(&table->mutex); @@ -240,8 +239,8 @@ int mthca_table_get(struct mthca_dev *dev, struct mthca_icm_table *table, int ob goto out; } - if (mthca_MAP_ICM(dev, table->icm[i], table->virt + i * MTHCA_TABLE_CHUNK_SIZE, - &status) || status) { + if (mthca_MAP_ICM(dev, table->icm[i], + table->virt + i * MTHCA_TABLE_CHUNK_SIZE)) { mthca_free_icm(dev, table->icm[i], table->coherent); table->icm[i] = NULL; ret = -ENOMEM; @@ -258,7 +257,6 @@ out: void mthca_table_put(struct mthca_dev *dev, struct mthca_icm_table *table, int obj) { int i; - u8 status; if (!mthca_is_memfree(dev)) return; @@ -269,8 +267,7 @@ void mthca_table_put(struct mthca_dev *dev, struct mthca_icm_table *table, int o if (--table->icm[i]->refcount == 0) { mthca_UNMAP_ICM(dev, table->virt + i * MTHCA_TABLE_CHUNK_SIZE, - MTHCA_TABLE_CHUNK_SIZE / MTHCA_ICM_PAGE_SIZE, - &status); + MTHCA_TABLE_CHUNK_SIZE / MTHCA_ICM_PAGE_SIZE); mthca_free_icm(dev, table->icm[i], table->coherent); table->icm[i] = NULL; } @@ -366,7 +363,6 @@ struct mthca_icm_table *mthca_alloc_icm_table(struct mthca_dev *dev, int num_icm; unsigned chunk_size; int i; - u8 status; obj_per_chunk = MTHCA_TABLE_CHUNK_SIZE / obj_size; num_icm = DIV_ROUND_UP(nobj, obj_per_chunk); @@ -396,8 +392,8 @@ struct mthca_icm_table *mthca_alloc_icm_table(struct mthca_dev *dev, __GFP_NOWARN, use_coherent); if (!table->icm[i]) goto err; - if (mthca_MAP_ICM(dev, table->icm[i], virt + i * MTHCA_TABLE_CHUNK_SIZE, - &status) || status) { + if (mthca_MAP_ICM(dev, table->icm[i], + virt + i * MTHCA_TABLE_CHUNK_SIZE)) { mthca_free_icm(dev, table->icm[i], table->coherent); table->icm[i] = NULL; goto err; @@ -416,8 +412,7 @@ err: for (i = 0; i < num_icm; ++i) if (table->icm[i]) { mthca_UNMAP_ICM(dev, virt + i * MTHCA_TABLE_CHUNK_SIZE, - MTHCA_TABLE_CHUNK_SIZE / MTHCA_ICM_PAGE_SIZE, - &status); + MTHCA_TABLE_CHUNK_SIZE / MTHCA_ICM_PAGE_SIZE); mthca_free_icm(dev, table->icm[i], table->coherent); } @@ -429,13 +424,12 @@ err: void mthca_free_icm_table(struct mthca_dev *dev, struct mthca_icm_table *table) { int i; - u8 status; for (i = 0; i < table->num_icm; ++i) if (table->icm[i]) { - mthca_UNMAP_ICM(dev, table->virt + i * MTHCA_TABLE_CHUNK_SIZE, - MTHCA_TABLE_CHUNK_SIZE / MTHCA_ICM_PAGE_SIZE, - &status); + mthca_UNMAP_ICM(dev, + table->virt + i * MTHCA_TABLE_CHUNK_SIZE, + MTHCA_TABLE_CHUNK_SIZE / MTHCA_ICM_PAGE_SIZE); mthca_free_icm(dev, table->icm[i], table->coherent); } @@ -454,7 +448,6 @@ int mthca_map_user_db(struct mthca_dev *dev, struct mthca_uar *uar, { struct page *pages[1]; int ret = 0; - u8 status; int i; if (!mthca_is_memfree(dev)) @@ -494,9 +487,7 @@ int mthca_map_user_db(struct mthca_dev *dev, struct mthca_uar *uar, } ret = mthca_MAP_ICM_page(dev, sg_dma_address(&db_tab->page[i].mem), - mthca_uarc_virt(dev, uar, i), &status); - if (!ret && status) - ret = -EINVAL; + mthca_uarc_virt(dev, uar, i)); if (ret) { pci_unmap_sg(dev->pdev, &db_tab->page[i].mem, 1, PCI_DMA_TODEVICE); put_page(sg_page(&db_tab->page[i].mem)); @@ -557,14 +548,13 @@ void mthca_cleanup_user_db_tab(struct mthca_dev *dev, struct mthca_uar *uar, struct mthca_user_db_table *db_tab) { int i; - u8 status; if (!mthca_is_memfree(dev)) return; for (i = 0; i < dev->uar_table.uarc_size / MTHCA_ICM_PAGE_SIZE; ++i) { if (db_tab->page[i].uvirt) { - mthca_UNMAP_ICM(dev, mthca_uarc_virt(dev, uar, i), 1, &status); + mthca_UNMAP_ICM(dev, mthca_uarc_virt(dev, uar, i), 1); pci_unmap_sg(dev->pdev, &db_tab->page[i].mem, 1, PCI_DMA_TODEVICE); put_page(sg_page(&db_tab->page[i].mem)); } @@ -581,7 +571,6 @@ int mthca_alloc_db(struct mthca_dev *dev, enum mthca_db_type type, int i, j; struct mthca_db_page *page; int ret = 0; - u8 status; mutex_lock(&dev->db_tab->mutex); @@ -644,9 +633,7 @@ alloc: memset(page->db_rec, 0, MTHCA_ICM_PAGE_SIZE); ret = mthca_MAP_ICM_page(dev, page->mapping, - mthca_uarc_virt(dev, &dev->driver_uar, i), &status); - if (!ret && status) - ret = -EINVAL; + mthca_uarc_virt(dev, &dev->driver_uar, i)); if (ret) { dma_free_coherent(&dev->pdev->dev, MTHCA_ICM_PAGE_SIZE, page->db_rec, page->mapping); @@ -678,7 +665,6 @@ void mthca_free_db(struct mthca_dev *dev, int type, int db_index) { int i, j; struct mthca_db_page *page; - u8 status; i = db_index / MTHCA_DB_REC_PER_PAGE; j = db_index % MTHCA_DB_REC_PER_PAGE; @@ -694,7 +680,7 @@ void mthca_free_db(struct mthca_dev *dev, int type, int db_index) if (bitmap_empty(page->used, MTHCA_DB_REC_PER_PAGE) && i >= dev->db_tab->max_group1 - 1) { - mthca_UNMAP_ICM(dev, mthca_uarc_virt(dev, &dev->driver_uar, i), 1, &status); + mthca_UNMAP_ICM(dev, mthca_uarc_virt(dev, &dev->driver_uar, i), 1); dma_free_coherent(&dev->pdev->dev, MTHCA_ICM_PAGE_SIZE, page->db_rec, page->mapping); @@ -745,7 +731,6 @@ int mthca_init_db_tab(struct mthca_dev *dev) void mthca_cleanup_db_tab(struct mthca_dev *dev) { int i; - u8 status; if (!mthca_is_memfree(dev)) return; @@ -763,7 +748,7 @@ void mthca_cleanup_db_tab(struct mthca_dev *dev) if (!bitmap_empty(dev->db_tab->page[i].used, MTHCA_DB_REC_PER_PAGE)) mthca_warn(dev, "Kernel UARC page %d not empty\n", i); - mthca_UNMAP_ICM(dev, mthca_uarc_virt(dev, &dev->driver_uar, i), 1, &status); + mthca_UNMAP_ICM(dev, mthca_uarc_virt(dev, &dev->driver_uar, i), 1); dma_free_coherent(&dev->pdev->dev, MTHCA_ICM_PAGE_SIZE, dev->db_tab->page[i].db_rec, diff --git a/drivers/infiniband/hw/mthca/mthca_mr.c b/drivers/infiniband/hw/mthca/mthca_mr.c index 44045c8..ed9a989 100644 --- a/drivers/infiniband/hw/mthca/mthca_mr.c +++ b/drivers/infiniband/hw/mthca/mthca_mr.c @@ -146,7 +146,7 @@ static int mthca_buddy_init(struct mthca_buddy *buddy, int max_order) buddy->bits = kzalloc((buddy->max_order + 1) * sizeof (long *), GFP_KERNEL); - buddy->num_free = kzalloc((buddy->max_order + 1) * sizeof (int *), + buddy->num_free = kcalloc((buddy->max_order + 1), sizeof *buddy->num_free, GFP_KERNEL); if (!buddy->bits || !buddy->num_free) goto err_out; @@ -257,7 +257,6 @@ static int __mthca_write_mtt(struct mthca_dev *dev, struct mthca_mtt *mtt, struct mthca_mailbox *mailbox; __be64 *mtt_entry; int err = 0; - u8 status; int i; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); @@ -281,17 +280,11 @@ static int __mthca_write_mtt(struct mthca_dev *dev, struct mthca_mtt *mtt, if (i & 1) mtt_entry[i + 2] = 0; - err = mthca_WRITE_MTT(dev, mailbox, (i + 1) & ~1, &status); + err = mthca_WRITE_MTT(dev, mailbox, (i + 1) & ~1); if (err) { mthca_warn(dev, "WRITE_MTT failed (%d)\n", err); goto out; } - if (status) { - mthca_warn(dev, "WRITE_MTT returned status 0x%02x\n", - status); - err = -EINVAL; - goto out; - } list_len -= i; start_index += i; @@ -441,7 +434,6 @@ int mthca_mr_alloc(struct mthca_dev *dev, u32 pd, int buffer_size_shift, u32 key; int i; int err; - u8 status; WARN_ON(buffer_size_shift >= 32); @@ -497,16 +489,10 @@ int mthca_mr_alloc(struct mthca_dev *dev, u32 pd, int buffer_size_shift, } err = mthca_SW2HW_MPT(dev, mailbox, - key & (dev->limits.num_mpts - 1), - &status); + key & (dev->limits.num_mpts - 1)); if (err) { mthca_warn(dev, "SW2HW_MPT failed (%d)\n", err); goto err_out_mailbox; - } else if (status) { - mthca_warn(dev, "SW2HW_MPT returned status 0x%02x\n", - status); - err = -EINVAL; - goto err_out_mailbox; } mthca_free_mailbox(dev, mailbox); @@ -567,17 +553,12 @@ static void mthca_free_region(struct mthca_dev *dev, u32 lkey) void mthca_free_mr(struct mthca_dev *dev, struct mthca_mr *mr) { int err; - u8 status; err = mthca_HW2SW_MPT(dev, NULL, key_to_hw_index(dev, mr->ibmr.lkey) & - (dev->limits.num_mpts - 1), - &status); + (dev->limits.num_mpts - 1)); if (err) mthca_warn(dev, "HW2SW_MPT failed (%d)\n", err); - else if (status) - mthca_warn(dev, "HW2SW_MPT returned status 0x%02x\n", - status); mthca_free_region(dev, mr->ibmr.lkey); mthca_free_mtt(dev, mr->mtt); @@ -590,7 +571,6 @@ int mthca_fmr_alloc(struct mthca_dev *dev, u32 pd, struct mthca_mailbox *mailbox; u64 mtt_seg; u32 key, idx; - u8 status; int list_len = mr->attr.max_pages; int err = -ENOMEM; int i; @@ -672,18 +652,11 @@ int mthca_fmr_alloc(struct mthca_dev *dev, u32 pd, } err = mthca_SW2HW_MPT(dev, mailbox, - key & (dev->limits.num_mpts - 1), - &status); + key & (dev->limits.num_mpts - 1)); if (err) { mthca_warn(dev, "SW2HW_MPT failed (%d)\n", err); goto err_out_mailbox_free; } - if (status) { - mthca_warn(dev, "SW2HW_MPT returned status 0x%02x\n", - status); - err = -EINVAL; - goto err_out_mailbox_free; - } mthca_free_mailbox(dev, mailbox); return 0; diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c index 1e0b4b6..42dde06 100644 --- a/drivers/infiniband/hw/mthca/mthca_provider.c +++ b/drivers/infiniband/hw/mthca/mthca_provider.c @@ -40,7 +40,9 @@ #include <linux/sched.h> #include <linux/slab.h> +#include <linux/stat.h> #include <linux/mm.h> +#include <linux/export.h> #include "mthca_dev.h" #include "mthca_cmd.h" @@ -63,8 +65,6 @@ static int mthca_query_device(struct ib_device *ibdev, int err = -ENOMEM; struct mthca_dev *mdev = to_mdev(ibdev); - u8 status; - in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL); out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL); if (!in_mad || !out_mad) @@ -78,14 +78,9 @@ static int mthca_query_device(struct ib_device *ibdev, in_mad->attr_id = IB_SMP_ATTR_NODE_INFO; err = mthca_MAD_IFC(mdev, 1, 1, - 1, NULL, NULL, in_mad, out_mad, - &status); + 1, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } props->device_cap_flags = mdev->device_cap_flags; props->vendor_id = be32_to_cpup((__be32 *) (out_mad->data + 36)) & @@ -141,7 +136,6 @@ static int mthca_query_port(struct ib_device *ibdev, struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; - u8 status; in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL); out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL); @@ -155,14 +149,9 @@ static int mthca_query_port(struct ib_device *ibdev, in_mad->attr_mod = cpu_to_be32(port); err = mthca_MAD_IFC(to_mdev(ibdev), 1, 1, - port, NULL, NULL, in_mad, out_mad, - &status); + port, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } props->lid = be16_to_cpup((__be16 *) (out_mad->data + 16)); props->lmc = out_mad->data[34] & 0x7; @@ -214,7 +203,6 @@ static int mthca_modify_port(struct ib_device *ibdev, struct mthca_set_ib_param set_ib; struct ib_port_attr attr; int err; - u8 status; if (mutex_lock_interruptible(&to_mdev(ibdev)->cap_mask_mutex)) return -ERESTARTSYS; @@ -229,14 +217,9 @@ static int mthca_modify_port(struct ib_device *ibdev, set_ib.cap_mask = (attr.port_cap_flags | props->set_port_cap_mask) & ~props->clr_port_cap_mask; - err = mthca_SET_IB(to_mdev(ibdev), &set_ib, port, &status); + err = mthca_SET_IB(to_mdev(ibdev), &set_ib, port); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } - out: mutex_unlock(&to_mdev(ibdev)->cap_mask_mutex); return err; @@ -248,7 +231,6 @@ static int mthca_query_pkey(struct ib_device *ibdev, struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; - u8 status; in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL); out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL); @@ -260,14 +242,9 @@ static int mthca_query_pkey(struct ib_device *ibdev, in_mad->attr_mod = cpu_to_be32(index / 32); err = mthca_MAD_IFC(to_mdev(ibdev), 1, 1, - port, NULL, NULL, in_mad, out_mad, - &status); + port, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } *pkey = be16_to_cpu(((__be16 *) out_mad->data)[index % 32]); @@ -283,7 +260,6 @@ static int mthca_query_gid(struct ib_device *ibdev, u8 port, struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; - u8 status; in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL); out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL); @@ -295,14 +271,9 @@ static int mthca_query_gid(struct ib_device *ibdev, u8 port, in_mad->attr_mod = cpu_to_be32(port); err = mthca_MAD_IFC(to_mdev(ibdev), 1, 1, - port, NULL, NULL, in_mad, out_mad, - &status); + port, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } memcpy(gid->raw, out_mad->data + 8, 8); @@ -311,14 +282,9 @@ static int mthca_query_gid(struct ib_device *ibdev, u8 port, in_mad->attr_mod = cpu_to_be32(index / 8); err = mthca_MAD_IFC(to_mdev(ibdev), 1, 1, - port, NULL, NULL, in_mad, out_mad, - &status); + port, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8); @@ -474,6 +440,9 @@ static struct ib_srq *mthca_create_srq(struct ib_pd *pd, struct mthca_srq *srq; int err; + if (init_attr->srq_type != IB_SRQT_BASIC) + return ERR_PTR(-ENOSYS); + srq = kmalloc(sizeof *srq, GFP_KERNEL); if (!srq) return ERR_PTR(-ENOMEM); @@ -726,6 +695,7 @@ static struct ib_cq *mthca_create_cq(struct ib_device *ibdev, int entries, if (context && ib_copy_to_udata(udata, &cq->cqn, sizeof (__u32))) { mthca_free_cq(to_mdev(ibdev), cq); + err = -EFAULT; goto err_free; } @@ -800,7 +770,6 @@ static int mthca_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *uda struct mthca_cq *cq = to_mcq(ibcq); struct mthca_resize_cq ucmd; u32 lkey; - u8 status; int ret; if (entries < 1 || entries > dev->limits.max_cqes) @@ -827,9 +796,7 @@ static int mthca_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *uda lkey = ucmd.lkey; } - ret = mthca_RESIZE_CQ(dev, cq->cqn, lkey, ilog2(entries), &status); - if (status) - ret = -EINVAL; + ret = mthca_RESIZE_CQ(dev, cq->cqn, lkey, ilog2(entries)); if (ret) { if (cq->resize_buf) { @@ -1161,7 +1128,6 @@ static int mthca_unmap_fmr(struct list_head *fmr_list) { struct ib_fmr *fmr; int err; - u8 status; struct mthca_dev *mdev = NULL; list_for_each_entry(fmr, fmr_list, list) { @@ -1182,12 +1148,8 @@ static int mthca_unmap_fmr(struct list_head *fmr_list) list_for_each_entry(fmr, fmr_list, list) mthca_tavor_fmr_unmap(mdev, to_mfmr(fmr)); - err = mthca_SYNC_TPT(mdev, &status); - if (err) - return err; - if (status) - return -EINVAL; - return 0; + err = mthca_SYNC_TPT(mdev); + return err; } static ssize_t show_rev(struct device *device, struct device_attribute *attr, @@ -1253,7 +1215,6 @@ static int mthca_init_node_data(struct mthca_dev *dev) struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; - u8 status; in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL); out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL); @@ -1264,28 +1225,18 @@ static int mthca_init_node_data(struct mthca_dev *dev) in_mad->attr_id = IB_SMP_ATTR_NODE_DESC; err = mthca_MAD_IFC(dev, 1, 1, - 1, NULL, NULL, in_mad, out_mad, - &status); + 1, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } memcpy(dev->ib_dev.node_desc, out_mad->data, 64); in_mad->attr_id = IB_SMP_ATTR_NODE_INFO; err = mthca_MAD_IFC(dev, 1, 1, - 1, NULL, NULL, in_mad, out_mad, - &status); + 1, NULL, NULL, in_mad, out_mad); if (err) goto out; - if (status) { - err = -EINVAL; - goto out; - } if (mthca_is_memfree(dev)) dev->rev_id = be32_to_cpup((__be32 *) (out_mad->data + 32)); diff --git a/drivers/infiniband/hw/mthca/mthca_qp.c b/drivers/infiniband/hw/mthca/mthca_qp.c index a34c9d3..9601049 100644 --- a/drivers/infiniband/hw/mthca/mthca_qp.c +++ b/drivers/infiniband/hw/mthca/mthca_qp.c @@ -308,7 +308,6 @@ static void store_attrs(struct mthca_sqp *sqp, const struct ib_qp_attr *attr, static void init_port(struct mthca_dev *dev, int port) { int err; - u8 status; struct mthca_init_ib_param param; memset(¶m, 0, sizeof param); @@ -319,11 +318,9 @@ static void init_port(struct mthca_dev *dev, int port) param.gid_cap = dev->limits.gid_table_len; param.pkey_cap = dev->limits.pkey_table_len; - err = mthca_INIT_IB(dev, ¶m, port, &status); + err = mthca_INIT_IB(dev, ¶m, port); if (err) mthca_warn(dev, "INIT_IB failed, return code %d.\n", err); - if (status) - mthca_warn(dev, "INIT_IB returned status %02x.\n", status); } static __be32 get_hw_access_flags(struct mthca_qp *qp, const struct ib_qp_attr *attr, @@ -433,7 +430,6 @@ int mthca_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr_m struct mthca_qp_param *qp_param; struct mthca_qp_context *context; int mthca_state; - u8 status; mutex_lock(&qp->mutex); @@ -448,12 +444,9 @@ int mthca_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr_m goto out; } - err = mthca_QUERY_QP(dev, qp->qpn, 0, mailbox, &status); - if (err) - goto out_mailbox; - if (status) { - mthca_warn(dev, "QUERY_QP returned status %02x\n", status); - err = -EINVAL; + err = mthca_QUERY_QP(dev, qp->qpn, 0, mailbox); + if (err) { + mthca_warn(dev, "QUERY_QP failed (%d)\n", err); goto out_mailbox; } @@ -555,7 +548,6 @@ static int __mthca_modify_qp(struct ib_qp *ibqp, struct mthca_qp_param *qp_param; struct mthca_qp_context *qp_context; u32 sqd_event = 0; - u8 status; int err = -EINVAL; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); @@ -781,13 +773,10 @@ static int __mthca_modify_qp(struct ib_qp *ibqp, sqd_event = 1 << 31; err = mthca_MODIFY_QP(dev, cur_state, new_state, qp->qpn, 0, - mailbox, sqd_event, &status); - if (err) - goto out_mailbox; - if (status) { - mthca_warn(dev, "modify QP %d->%d returned status %02x.\n", - cur_state, new_state, status); - err = -EINVAL; + mailbox, sqd_event); + if (err) { + mthca_warn(dev, "modify QP %d->%d returned %d.\n", + cur_state, new_state, err); goto out_mailbox; } @@ -817,7 +806,7 @@ static int __mthca_modify_qp(struct ib_qp *ibqp, cur_state != IB_QPS_ERR && (new_state == IB_QPS_RESET || new_state == IB_QPS_ERR)) - mthca_CLOSE_IB(dev, qp->port, &status); + mthca_CLOSE_IB(dev, qp->port); } /* @@ -1429,7 +1418,6 @@ static inline int get_qp_refcount(struct mthca_dev *dev, struct mthca_qp *qp) void mthca_free_qp(struct mthca_dev *dev, struct mthca_qp *qp) { - u8 status; struct mthca_cq *send_cq; struct mthca_cq *recv_cq; @@ -1454,7 +1442,7 @@ void mthca_free_qp(struct mthca_dev *dev, if (qp->state != IB_QPS_RESET) mthca_MODIFY_QP(dev, qp->state, IB_QPS_RESET, qp->qpn, 0, - NULL, 0, &status); + NULL, 0); /* * If this is a userspace QP, the buffers, MR, CQs and so on @@ -2263,7 +2251,6 @@ void mthca_free_err_wqe(struct mthca_dev *dev, struct mthca_qp *qp, int is_send, int mthca_init_qp_table(struct mthca_dev *dev) { int err; - u8 status; int i; spin_lock_init(&dev->qp_table.lock); @@ -2290,15 +2277,10 @@ int mthca_init_qp_table(struct mthca_dev *dev) for (i = 0; i < 2; ++i) { err = mthca_CONF_SPECIAL_QP(dev, i ? IB_QPT_GSI : IB_QPT_SMI, - dev->qp_table.sqp_start + i * 2, - &status); - if (err) - goto err_out; - if (status) { + dev->qp_table.sqp_start + i * 2); + if (err) { mthca_warn(dev, "CONF_SPECIAL_QP returned " - "status %02x, aborting.\n", - status); - err = -EINVAL; + "%d, aborting.\n", err); goto err_out; } } @@ -2306,7 +2288,7 @@ int mthca_init_qp_table(struct mthca_dev *dev) err_out: for (i = 0; i < 2; ++i) - mthca_CONF_SPECIAL_QP(dev, i, 0, &status); + mthca_CONF_SPECIAL_QP(dev, i, 0); mthca_array_cleanup(&dev->qp_table.qp, dev->limits.num_qps); mthca_alloc_cleanup(&dev->qp_table.alloc); @@ -2317,10 +2299,9 @@ int mthca_init_qp_table(struct mthca_dev *dev) void mthca_cleanup_qp_table(struct mthca_dev *dev) { int i; - u8 status; for (i = 0; i < 2; ++i) - mthca_CONF_SPECIAL_QP(dev, i, 0, &status); + mthca_CONF_SPECIAL_QP(dev, i, 0); mthca_array_cleanup(&dev->qp_table.qp, dev->limits.num_qps); mthca_alloc_cleanup(&dev->qp_table.alloc); diff --git a/drivers/infiniband/hw/mthca/mthca_reset.c b/drivers/infiniband/hw/mthca/mthca_reset.c index 2a13a16..4fa3534 100644 --- a/drivers/infiniband/hw/mthca/mthca_reset.c +++ b/drivers/infiniband/hw/mthca/mthca_reset.c @@ -113,7 +113,7 @@ int mthca_reset(struct mthca_dev *mdev) } hca_pcix_cap = pci_find_capability(mdev->pdev, PCI_CAP_ID_PCIX); - hca_pcie_cap = pci_find_capability(mdev->pdev, PCI_CAP_ID_EXP); + hca_pcie_cap = pci_pcie_cap(mdev->pdev); if (bridge) { bridge_header = kmalloc(256, GFP_KERNEL); diff --git a/drivers/infiniband/hw/mthca/mthca_srq.c b/drivers/infiniband/hw/mthca/mthca_srq.c index 4fabe62..d22f970 100644 --- a/drivers/infiniband/hw/mthca/mthca_srq.c +++ b/drivers/infiniband/hw/mthca/mthca_srq.c @@ -200,7 +200,6 @@ int mthca_alloc_srq(struct mthca_dev *dev, struct mthca_pd *pd, struct ib_srq_attr *attr, struct mthca_srq *srq) { struct mthca_mailbox *mailbox; - u8 status; int ds; int err; @@ -266,18 +265,12 @@ int mthca_alloc_srq(struct mthca_dev *dev, struct mthca_pd *pd, else mthca_tavor_init_srq_context(dev, pd, srq, mailbox->buf); - err = mthca_SW2HW_SRQ(dev, mailbox, srq->srqn, &status); + err = mthca_SW2HW_SRQ(dev, mailbox, srq->srqn); if (err) { mthca_warn(dev, "SW2HW_SRQ failed (%d)\n", err); goto err_out_free_buf; } - if (status) { - mthca_warn(dev, "SW2HW_SRQ returned status 0x%02x\n", - status); - err = -EINVAL; - goto err_out_free_buf; - } spin_lock_irq(&dev->srq_table.lock); if (mthca_array_set(&dev->srq_table.srq, @@ -299,11 +292,9 @@ int mthca_alloc_srq(struct mthca_dev *dev, struct mthca_pd *pd, return 0; err_out_free_srq: - err = mthca_HW2SW_SRQ(dev, mailbox, srq->srqn, &status); + err = mthca_HW2SW_SRQ(dev, mailbox, srq->srqn); if (err) mthca_warn(dev, "HW2SW_SRQ failed (%d)\n", err); - else if (status) - mthca_warn(dev, "HW2SW_SRQ returned status 0x%02x\n", status); err_out_free_buf: if (!pd->ibpd.uobject) @@ -340,7 +331,6 @@ void mthca_free_srq(struct mthca_dev *dev, struct mthca_srq *srq) { struct mthca_mailbox *mailbox; int err; - u8 status; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); if (IS_ERR(mailbox)) { @@ -348,11 +338,9 @@ void mthca_free_srq(struct mthca_dev *dev, struct mthca_srq *srq) return; } - err = mthca_HW2SW_SRQ(dev, mailbox, srq->srqn, &status); + err = mthca_HW2SW_SRQ(dev, mailbox, srq->srqn); if (err) mthca_warn(dev, "HW2SW_SRQ failed (%d)\n", err); - else if (status) - mthca_warn(dev, "HW2SW_SRQ returned status 0x%02x\n", status); spin_lock_irq(&dev->srq_table.lock); mthca_array_clear(&dev->srq_table.srq, @@ -378,8 +366,7 @@ int mthca_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr, { struct mthca_dev *dev = to_mdev(ibsrq->device); struct mthca_srq *srq = to_msrq(ibsrq); - int ret; - u8 status; + int ret = 0; /* We don't support resizing SRQs (yet?) */ if (attr_mask & IB_SRQ_MAX_WR) @@ -391,16 +378,11 @@ int mthca_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr, return -EINVAL; mutex_lock(&srq->mutex); - ret = mthca_ARM_SRQ(dev, srq->srqn, attr->srq_limit, &status); + ret = mthca_ARM_SRQ(dev, srq->srqn, attr->srq_limit); mutex_unlock(&srq->mutex); - - if (ret) - return ret; - if (status) - return -EINVAL; } - return 0; + return ret; } int mthca_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr) @@ -410,14 +392,13 @@ int mthca_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr) struct mthca_mailbox *mailbox; struct mthca_arbel_srq_context *arbel_ctx; struct mthca_tavor_srq_context *tavor_ctx; - u8 status; int err; mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL); if (IS_ERR(mailbox)) return PTR_ERR(mailbox); - err = mthca_QUERY_SRQ(dev, srq->srqn, mailbox, &status); + err = mthca_QUERY_SRQ(dev, srq->srqn, mailbox); if (err) goto out; diff --git a/drivers/infiniband/hw/nes/Makefile b/drivers/infiniband/hw/nes/Makefile index 3514851..97820c2 100644 --- a/drivers/infiniband/hw/nes/Makefile +++ b/drivers/infiniband/hw/nes/Makefile @@ -1,3 +1,3 @@ obj-$(CONFIG_INFINIBAND_NES) += iw_nes.o -iw_nes-objs := nes.o nes_hw.o nes_nic.o nes_utils.o nes_verbs.o nes_cm.o +iw_nes-objs := nes.o nes_hw.o nes_nic.o nes_utils.o nes_verbs.o nes_cm.o nes_mgt.o diff --git a/drivers/infiniband/hw/nes/nes.c b/drivers/infiniband/hw/nes/nes.c index 2d668c6..5965b3d 100644 --- a/drivers/infiniband/hw/nes/nes.c +++ b/drivers/infiniband/hw/nes/nes.c @@ -84,7 +84,7 @@ module_param(send_first, int, 0644); MODULE_PARM_DESC(send_first, "Send RDMA Message First on Active Connection"); -unsigned int nes_drv_opt = 0; +unsigned int nes_drv_opt = NES_DRV_OPT_DISABLE_INT_MOD | NES_DRV_OPT_ENABLE_PAU; module_param(nes_drv_opt, int, 0644); MODULE_PARM_DESC(nes_drv_opt, "Driver option parameters"); @@ -130,9 +130,6 @@ static struct notifier_block nes_net_notifier = { .notifier_call = nes_net_event }; - - - /** * nes_inetaddr_event */ @@ -321,6 +318,9 @@ void nes_rem_ref(struct ib_qp *ibqp) } if (atomic_dec_and_test(&nesqp->refcount)) { + if (nesqp->pau_mode) + nes_destroy_pau_qp(nesdev, nesqp); + /* Destroy the QP */ cqp_request = nes_get_cqp_request(nesdev); if (cqp_request == NULL) { diff --git a/drivers/infiniband/hw/nes/nes.h b/drivers/infiniband/hw/nes/nes.h index 6e30273..3ade373 100644 --- a/drivers/infiniband/hw/nes/nes.h +++ b/drivers/infiniband/hw/nes/nes.h @@ -102,6 +102,7 @@ #define NES_DRV_OPT_NO_INLINE_DATA 0x00000080 #define NES_DRV_OPT_DISABLE_INT_MOD 0x00000100 #define NES_DRV_OPT_DISABLE_VIRT_WQ 0x00000200 +#define NES_DRV_OPT_ENABLE_PAU 0x00000400 #define NES_AEQ_EVENT_TIMEOUT 2500 #define NES_DISCONNECT_EVENT_TIMEOUT 2000 @@ -128,6 +129,7 @@ #define NES_DBG_IW_RX 0x00020000 #define NES_DBG_IW_TX 0x00040000 #define NES_DBG_SHUTDOWN 0x00080000 +#define NES_DBG_PAU 0x00100000 #define NES_DBG_RSVD1 0x10000000 #define NES_DBG_RSVD2 0x20000000 #define NES_DBG_RSVD3 0x40000000 @@ -162,6 +164,7 @@ do { \ #include "nes_context.h" #include "nes_user.h" #include "nes_cm.h" +#include "nes_mgt.h" extern int max_mtu; #define max_frame_len (max_mtu+ETH_HLEN) @@ -202,6 +205,8 @@ extern atomic_t cm_nodes_created; extern atomic_t cm_nodes_destroyed; extern atomic_t cm_accel_dropped_pkts; extern atomic_t cm_resets_recvd; +extern atomic_t pau_qps_created; +extern atomic_t pau_qps_destroyed; extern u32 int_mod_timer_init; extern u32 int_mod_cq_depth_256; @@ -273,6 +278,14 @@ struct nes_device { u8 link_recheck; }; +/* Receive skb private area - must fit in skb->cb area */ +struct nes_rskb_cb { + u64 busaddr; + u32 maplen; + u32 seqnum; + u8 *data_start; + struct nes_qp *nesqp; +}; static inline __le32 get_crc_value(struct nes_v4_quad *nes_quad) { @@ -305,8 +318,8 @@ set_wqe_32bit_value(__le32 *wqe_words, u32 index, u32 value) static inline void nes_fill_init_cqp_wqe(struct nes_hw_cqp_wqe *cqp_wqe, struct nes_device *nesdev) { - set_wqe_64bit_value(cqp_wqe->wqe_words, NES_CQP_WQE_COMP_CTX_LOW_IDX, - (u64)((unsigned long) &nesdev->cqp)); + cqp_wqe->wqe_words[NES_CQP_WQE_COMP_CTX_LOW_IDX] = 0; + cqp_wqe->wqe_words[NES_CQP_WQE_COMP_CTX_HIGH_IDX] = 0; cqp_wqe->wqe_words[NES_CQP_WQE_COMP_SCRATCH_LOW_IDX] = 0; cqp_wqe->wqe_words[NES_CQP_WQE_COMP_SCRATCH_HIGH_IDX] = 0; cqp_wqe->wqe_words[NES_CQP_STAG_WQE_PBL_BLK_COUNT_IDX] = 0; diff --git a/drivers/infiniband/hw/nes/nes_cm.c b/drivers/infiniband/hw/nes/nes_cm.c index a1f74f6..0a52d72 100644 --- a/drivers/infiniband/hw/nes/nes_cm.c +++ b/drivers/infiniband/hw/nes/nes_cm.c @@ -34,7 +34,7 @@ #define TCPOPT_TIMESTAMP 8 -#include <asm/atomic.h> +#include <linux/atomic.h> #include <linux/skbuff.h> #include <linux/ip.h> #include <linux/tcp.h> @@ -77,26 +77,19 @@ atomic_t cm_nodes_destroyed; atomic_t cm_accel_dropped_pkts; atomic_t cm_resets_recvd; -static inline int mini_cm_accelerated(struct nes_cm_core *, - struct nes_cm_node *); -static struct nes_cm_listener *mini_cm_listen(struct nes_cm_core *, - struct nes_vnic *, struct nes_cm_info *); +static inline int mini_cm_accelerated(struct nes_cm_core *, struct nes_cm_node *); +static struct nes_cm_listener *mini_cm_listen(struct nes_cm_core *, struct nes_vnic *, struct nes_cm_info *); static int mini_cm_del_listen(struct nes_cm_core *, struct nes_cm_listener *); -static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *, - struct nes_vnic *, u16, void *, struct nes_cm_info *); +static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *, struct nes_vnic *, u16, void *, struct nes_cm_info *); static int mini_cm_close(struct nes_cm_core *, struct nes_cm_node *); -static int mini_cm_accept(struct nes_cm_core *, struct ietf_mpa_frame *, - struct nes_cm_node *); -static int mini_cm_reject(struct nes_cm_core *, struct ietf_mpa_frame *, - struct nes_cm_node *); -static int mini_cm_recv_pkt(struct nes_cm_core *, struct nes_vnic *, - struct sk_buff *); +static int mini_cm_accept(struct nes_cm_core *, struct nes_cm_node *); +static int mini_cm_reject(struct nes_cm_core *, struct nes_cm_node *); +static int mini_cm_recv_pkt(struct nes_cm_core *, struct nes_vnic *, struct sk_buff *); static int mini_cm_dealloc_core(struct nes_cm_core *); static int mini_cm_get(struct nes_cm_core *); static int mini_cm_set(struct nes_cm_core *, u32, u32); -static void form_cm_frame(struct sk_buff *, struct nes_cm_node *, - void *, u32, void *, u32, u8); +static void form_cm_frame(struct sk_buff *, struct nes_cm_node *, void *, u32, void *, u32, u8); static int add_ref_cm_node(struct nes_cm_node *); static int rem_ref_cm_node(struct nes_cm_core *, struct nes_cm_node *); @@ -111,16 +104,14 @@ static int send_syn(struct nes_cm_node *, u32, struct sk_buff *); static int send_reset(struct nes_cm_node *, struct sk_buff *); static int send_ack(struct nes_cm_node *cm_node, struct sk_buff *skb); static int send_fin(struct nes_cm_node *cm_node, struct sk_buff *skb); -static void process_packet(struct nes_cm_node *, struct sk_buff *, - struct nes_cm_core *); +static void process_packet(struct nes_cm_node *, struct sk_buff *, struct nes_cm_core *); static void active_open_err(struct nes_cm_node *, struct sk_buff *, int); static void passive_open_err(struct nes_cm_node *, struct sk_buff *, int); static void cleanup_retrans_entry(struct nes_cm_node *); static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *); static void free_retrans_entry(struct nes_cm_node *cm_node); -static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, - struct sk_buff *skb, int optionsize, int passive); +static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, struct sk_buff *skb, int optionsize, int passive); /* CM event handler functions */ static void cm_event_connected(struct nes_cm_event *); @@ -130,6 +121,12 @@ static void cm_event_mpa_req(struct nes_cm_event *); static void cm_event_mpa_reject(struct nes_cm_event *); static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node); +/* MPA build functions */ +static int cm_build_mpa_frame(struct nes_cm_node *, u8 **, u16 *, u8 *, u8); +static void build_mpa_v2(struct nes_cm_node *, void *, u8); +static void build_mpa_v1(struct nes_cm_node *, void *, u8); +static void build_rdma0_msg(struct nes_cm_node *, struct nes_qp **); + static void print_core(struct nes_cm_core *core); /* External CM API Interface */ @@ -159,12 +156,21 @@ atomic_t cm_connecteds; atomic_t cm_connect_reqs; atomic_t cm_rejects; +int nes_add_ref_cm_node(struct nes_cm_node *cm_node) +{ + return add_ref_cm_node(cm_node); +} + +int nes_rem_ref_cm_node(struct nes_cm_node *cm_node) +{ + return rem_ref_cm_node(cm_node->cm_core, cm_node); +} /** * create_event */ -static struct nes_cm_event *create_event(struct nes_cm_node *cm_node, - enum nes_cm_event_type type) +static struct nes_cm_event *create_event(struct nes_cm_node * cm_node, + enum nes_cm_event_type type) { struct nes_cm_event *event; @@ -186,10 +192,10 @@ static struct nes_cm_event *create_event(struct nes_cm_node *cm_node, event->cm_info.cm_id = cm_node->cm_id; nes_debug(NES_DBG_CM, "cm_node=%p Created event=%p, type=%u, " - "dst_addr=%08x[%x], src_addr=%08x[%x]\n", - cm_node, event, type, event->cm_info.loc_addr, - event->cm_info.loc_port, event->cm_info.rem_addr, - event->cm_info.rem_port); + "dst_addr=%08x[%x], src_addr=%08x[%x]\n", + cm_node, event, type, event->cm_info.loc_addr, + event->cm_info.loc_port, event->cm_info.rem_addr, + event->cm_info.rem_port); nes_cm_post_event(event); return event; @@ -201,14 +207,19 @@ static struct nes_cm_event *create_event(struct nes_cm_node *cm_node, */ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb) { + u8 start_addr = 0; + u8 *start_ptr = &start_addr; + u8 **start_buff = &start_ptr; + u16 buff_len = 0; + if (!skb) { nes_debug(NES_DBG_CM, "skb set to NULL\n"); return -1; } /* send an MPA Request frame */ - form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame, - cm_node->mpa_frame_size, SET_ACK); + cm_build_mpa_frame(cm_node, start_buff, &buff_len, NULL, MPA_KEY_REQUEST); + form_cm_frame(skb, cm_node, NULL, 0, *start_buff, buff_len, SET_ACK); return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0); } @@ -217,7 +228,11 @@ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb) static int send_mpa_reject(struct nes_cm_node *cm_node) { - struct sk_buff *skb = NULL; + struct sk_buff *skb = NULL; + u8 start_addr = 0; + u8 *start_ptr = &start_addr; + u8 **start_buff = &start_ptr; + u16 buff_len = 0; skb = dev_alloc_skb(MAX_CM_BUFFER); if (!skb) { @@ -226,8 +241,8 @@ static int send_mpa_reject(struct nes_cm_node *cm_node) } /* send an MPA reject frame */ - form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame, - cm_node->mpa_frame_size, SET_ACK | SET_FIN); + cm_build_mpa_frame(cm_node, start_buff, &buff_len, NULL, MPA_KEY_REPLY); + form_cm_frame(skb, cm_node, NULL, 0, *start_buff, buff_len, SET_ACK | SET_FIN); cm_node->state = NES_CM_STATE_FIN_WAIT1; return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0); @@ -239,24 +254,31 @@ static int send_mpa_reject(struct nes_cm_node *cm_node) * IETF MPA frame */ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type, - u32 len) + u32 len) { - struct ietf_mpa_frame *mpa_frame; + struct ietf_mpa_v1 *mpa_frame; + struct ietf_mpa_v2 *mpa_v2_frame; + struct ietf_rtr_msg *rtr_msg; + int mpa_hdr_len; + int priv_data_len; *type = NES_MPA_REQUEST_ACCEPT; /* assume req frame is in tcp data payload */ - if (len < sizeof(struct ietf_mpa_frame)) { + if (len < sizeof(struct ietf_mpa_v1)) { nes_debug(NES_DBG_CM, "The received ietf buffer was too small (%x)\n", len); return -EINVAL; } - mpa_frame = (struct ietf_mpa_frame *)buffer; - cm_node->mpa_frame_size = ntohs(mpa_frame->priv_data_len); + /* points to the beginning of the frame, which could be MPA V1 or V2 */ + mpa_frame = (struct ietf_mpa_v1 *)buffer; + mpa_hdr_len = sizeof(struct ietf_mpa_v1); + priv_data_len = ntohs(mpa_frame->priv_data_len); + /* make sure mpa private data len is less than 512 bytes */ - if (cm_node->mpa_frame_size > IETF_MAX_PRIV_DATA_LEN) { + if (priv_data_len > IETF_MAX_PRIV_DATA_LEN) { nes_debug(NES_DBG_CM, "The received Length of Private" - " Data field exceeds 512 octets\n"); + " Data field exceeds 512 octets\n"); return -EINVAL; } /* @@ -264,11 +286,22 @@ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type, * received MPA version and MPA key information * */ - if (mpa_frame->rev != mpa_version) { + if (mpa_frame->rev != IETF_MPA_V1 && mpa_frame->rev != IETF_MPA_V2) { + nes_debug(NES_DBG_CM, "The received mpa version" + " is not supported\n"); + return -EINVAL; + } + /* + * backwards compatibility only + */ + if (mpa_frame->rev > cm_node->mpa_frame_rev) { nes_debug(NES_DBG_CM, "The received mpa version" - " can not be interoperated\n"); + " can not be interoperated\n"); return -EINVAL; + } else { + cm_node->mpa_frame_rev = mpa_frame->rev; } + if (cm_node->state != NES_CM_STATE_MPAREQ_SENT) { if (memcmp(mpa_frame->key, IEFT_MPA_KEY_REQ, IETF_MPA_KEY_SIZE)) { nes_debug(NES_DBG_CM, "Unexpected MPA Key received \n"); @@ -281,25 +314,75 @@ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type, } } - if (cm_node->mpa_frame_size + sizeof(struct ietf_mpa_frame) != len) { + + if (priv_data_len + mpa_hdr_len != len) { nes_debug(NES_DBG_CM, "The received ietf buffer was not right" - " complete (%x + %x != %x)\n", - cm_node->mpa_frame_size, - (u32)sizeof(struct ietf_mpa_frame), len); + " complete (%x + %x != %x)\n", + priv_data_len, mpa_hdr_len, len); return -EINVAL; } /* make sure it does not exceed the max size */ if (len > MAX_CM_BUFFER) { nes_debug(NES_DBG_CM, "The received ietf buffer was too large" - " (%x + %x != %x)\n", - cm_node->mpa_frame_size, - (u32)sizeof(struct ietf_mpa_frame), len); + " (%x + %x != %x)\n", + priv_data_len, mpa_hdr_len, len); return -EINVAL; } + cm_node->mpa_frame_size = priv_data_len; + + switch (mpa_frame->rev) { + case IETF_MPA_V2: { + u16 ird_size; + u16 ord_size; + mpa_v2_frame = (struct ietf_mpa_v2 *)buffer; + mpa_hdr_len += IETF_RTR_MSG_SIZE; + cm_node->mpa_frame_size -= IETF_RTR_MSG_SIZE; + rtr_msg = &mpa_v2_frame->rtr_msg; + + /* parse rtr message */ + rtr_msg->ctrl_ird = ntohs(rtr_msg->ctrl_ird); + rtr_msg->ctrl_ord = ntohs(rtr_msg->ctrl_ord); + ird_size = rtr_msg->ctrl_ird & IETF_NO_IRD_ORD; + ord_size = rtr_msg->ctrl_ord & IETF_NO_IRD_ORD; + + if (!(rtr_msg->ctrl_ird & IETF_PEER_TO_PEER)) { + /* send reset */ + return -EINVAL; + } + + if (cm_node->state != NES_CM_STATE_MPAREQ_SENT) { + /* responder */ + if (cm_node->ord_size > ird_size) + cm_node->ord_size = ird_size; + } else { + /* initiator */ + if (cm_node->ord_size > ird_size) + cm_node->ord_size = ird_size; + + if (cm_node->ird_size < ord_size) { + /* no resources available */ + /* send terminate message */ + return -EINVAL; + } + } + + if (rtr_msg->ctrl_ord & IETF_RDMA0_READ) { + cm_node->send_rdma0_op = SEND_RDMA_READ_ZERO; + } else if (rtr_msg->ctrl_ord & IETF_RDMA0_WRITE) { + cm_node->send_rdma0_op = SEND_RDMA_WRITE_ZERO; + } else { /* Not supported RDMA0 operation */ + return -EINVAL; + } + break; + } + case IETF_MPA_V1: + default: + break; + } + /* copy entire MPA frame to our cm_node's frame */ - memcpy(cm_node->mpa_frame_buf, buffer + sizeof(struct ietf_mpa_frame), - cm_node->mpa_frame_size); + memcpy(cm_node->mpa_frame_buf, buffer + mpa_hdr_len, cm_node->mpa_frame_size); if (mpa_frame->flags & IETF_MPA_FLAGS_REJECT) *type = NES_MPA_REQUEST_REJECT; @@ -312,8 +395,8 @@ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type, * node info to build. */ static void form_cm_frame(struct sk_buff *skb, - struct nes_cm_node *cm_node, void *options, u32 optionsize, - void *data, u32 datasize, u8 flags) + struct nes_cm_node *cm_node, void *options, u32 optionsize, + void *data, u32 datasize, u8 flags) { struct tcphdr *tcph; struct iphdr *iph; @@ -322,14 +405,14 @@ static void form_cm_frame(struct sk_buff *skb, u16 packetsize = sizeof(*iph); packetsize += sizeof(*tcph); - packetsize += optionsize + datasize; + packetsize += optionsize + datasize; + skb_trim(skb, 0); memset(skb->data, 0x00, ETH_HLEN + sizeof(*iph) + sizeof(*tcph)); - skb->len = 0; buf = skb_put(skb, packetsize + ETH_HLEN); - ethh = (struct ethhdr *) buf; + ethh = (struct ethhdr *)buf; buf += ETH_HLEN; iph = (struct iphdr *)buf; @@ -337,7 +420,7 @@ static void form_cm_frame(struct sk_buff *skb, tcph = (struct tcphdr *)buf; skb_reset_mac_header(skb); skb_set_network_header(skb, ETH_HLEN); - skb_set_transport_header(skb, ETH_HLEN+sizeof(*iph)); + skb_set_transport_header(skb, ETH_HLEN + sizeof(*iph)); buf += sizeof(*tcph); skb->ip_summed = CHECKSUM_PARTIAL; @@ -350,14 +433,14 @@ static void form_cm_frame(struct sk_buff *skb, ethh->h_proto = htons(0x0800); iph->version = IPVERSION; - iph->ihl = 5; /* 5 * 4Byte words, IP headr len */ + iph->ihl = 5; /* 5 * 4Byte words, IP headr len */ iph->tos = 0; iph->tot_len = htons(packetsize); iph->id = htons(++cm_node->tcp_cntxt.loc_id); iph->frag_off = htons(0x4000); iph->ttl = 0x40; - iph->protocol = 0x06; /* IPPROTO_TCP */ + iph->protocol = 0x06; /* IPPROTO_TCP */ iph->saddr = htonl(cm_node->loc_addr); iph->daddr = htonl(cm_node->rem_addr); @@ -370,14 +453,16 @@ static void form_cm_frame(struct sk_buff *skb, cm_node->tcp_cntxt.loc_ack_num = cm_node->tcp_cntxt.rcv_nxt; tcph->ack_seq = htonl(cm_node->tcp_cntxt.loc_ack_num); tcph->ack = 1; - } else + } else { tcph->ack_seq = 0; + } if (flags & SET_SYN) { cm_node->tcp_cntxt.loc_seq_num++; tcph->syn = 1; - } else + } else { cm_node->tcp_cntxt.loc_seq_num += datasize; + } if (flags & SET_FIN) { cm_node->tcp_cntxt.loc_seq_num++; @@ -398,10 +483,8 @@ static void form_cm_frame(struct sk_buff *skb, skb_shinfo(skb)->nr_frags = 0; cm_packets_created++; - } - /** * print_core - dump a cm core */ @@ -413,7 +496,7 @@ static void print_core(struct nes_cm_core *core) return; nes_debug(NES_DBG_CM, "---------------------------------------------\n"); - nes_debug(NES_DBG_CM, "State : %u \n", core->state); + nes_debug(NES_DBG_CM, "State : %u \n", core->state); nes_debug(NES_DBG_CM, "Listen Nodes : %u \n", atomic_read(&core->listen_node_cnt)); nes_debug(NES_DBG_CM, "Active Nodes : %u \n", atomic_read(&core->node_cnt)); @@ -423,6 +506,147 @@ static void print_core(struct nes_cm_core *core) nes_debug(NES_DBG_CM, "-------------- end core ---------------\n"); } +/** + * cm_build_mpa_frame - build a MPA V1 frame or MPA V2 frame + */ +static int cm_build_mpa_frame(struct nes_cm_node *cm_node, u8 **start_buff, + u16 *buff_len, u8 *pci_mem, u8 mpa_key) +{ + int ret = 0; + + *start_buff = (pci_mem) ? pci_mem : &cm_node->mpa_frame_buf[0]; + + switch (cm_node->mpa_frame_rev) { + case IETF_MPA_V1: + *start_buff = (u8 *)*start_buff + sizeof(struct ietf_rtr_msg); + *buff_len = sizeof(struct ietf_mpa_v1) + cm_node->mpa_frame_size; + build_mpa_v1(cm_node, *start_buff, mpa_key); + break; + case IETF_MPA_V2: + *buff_len = sizeof(struct ietf_mpa_v2) + cm_node->mpa_frame_size; + build_mpa_v2(cm_node, *start_buff, mpa_key); + break; + default: + ret = -EINVAL; + } + return ret; +} + +/** + * build_mpa_v2 - build a MPA V2 frame + */ +static void build_mpa_v2(struct nes_cm_node *cm_node, + void *start_addr, u8 mpa_key) +{ + struct ietf_mpa_v2 *mpa_frame = (struct ietf_mpa_v2 *)start_addr; + struct ietf_rtr_msg *rtr_msg = &mpa_frame->rtr_msg; + + /* initialize the upper 5 bytes of the frame */ + build_mpa_v1(cm_node, start_addr, mpa_key); + mpa_frame->flags |= IETF_MPA_V2_FLAG; /* set a bit to indicate MPA V2 */ + mpa_frame->priv_data_len += htons(IETF_RTR_MSG_SIZE); + + /* initialize RTR msg */ + rtr_msg->ctrl_ird = (cm_node->ird_size > IETF_NO_IRD_ORD) ? + IETF_NO_IRD_ORD : cm_node->ird_size; + rtr_msg->ctrl_ord = (cm_node->ord_size > IETF_NO_IRD_ORD) ? + IETF_NO_IRD_ORD : cm_node->ord_size; + + rtr_msg->ctrl_ird |= IETF_PEER_TO_PEER; + rtr_msg->ctrl_ird |= IETF_FLPDU_ZERO_LEN; + + switch (mpa_key) { + case MPA_KEY_REQUEST: + rtr_msg->ctrl_ord |= IETF_RDMA0_WRITE; + rtr_msg->ctrl_ord |= IETF_RDMA0_READ; + break; + case MPA_KEY_REPLY: + switch (cm_node->send_rdma0_op) { + case SEND_RDMA_WRITE_ZERO: + rtr_msg->ctrl_ord |= IETF_RDMA0_WRITE; + break; + case SEND_RDMA_READ_ZERO: + rtr_msg->ctrl_ord |= IETF_RDMA0_READ; + break; + } + } + rtr_msg->ctrl_ird = htons(rtr_msg->ctrl_ird); + rtr_msg->ctrl_ord = htons(rtr_msg->ctrl_ord); +} + +/** + * build_mpa_v1 - build a MPA V1 frame + */ +static void build_mpa_v1(struct nes_cm_node *cm_node, void *start_addr, u8 mpa_key) +{ + struct ietf_mpa_v1 *mpa_frame = (struct ietf_mpa_v1 *)start_addr; + + switch (mpa_key) { + case MPA_KEY_REQUEST: + memcpy(mpa_frame->key, IEFT_MPA_KEY_REQ, IETF_MPA_KEY_SIZE); + break; + case MPA_KEY_REPLY: + memcpy(mpa_frame->key, IEFT_MPA_KEY_REP, IETF_MPA_KEY_SIZE); + break; + } + mpa_frame->flags = IETF_MPA_FLAGS_CRC; + mpa_frame->rev = cm_node->mpa_frame_rev; + mpa_frame->priv_data_len = htons(cm_node->mpa_frame_size); +} + +static void build_rdma0_msg(struct nes_cm_node *cm_node, struct nes_qp **nesqp_addr) +{ + u64 u64temp; + struct nes_qp *nesqp = *nesqp_addr; + struct nes_hw_qp_wqe *wqe = &nesqp->hwqp.sq_vbase[0]; + + u64temp = (unsigned long)nesqp; + u64temp |= NES_SW_CONTEXT_ALIGN >> 1; + set_wqe_64bit_value(wqe->wqe_words, NES_IWARP_SQ_WQE_COMP_CTX_LOW_IDX, u64temp); + + wqe->wqe_words[NES_IWARP_SQ_WQE_FRAG0_LOW_IDX] = 0; + wqe->wqe_words[NES_IWARP_SQ_WQE_FRAG0_HIGH_IDX] = 0; + + switch (cm_node->send_rdma0_op) { + case SEND_RDMA_WRITE_ZERO: + nes_debug(NES_DBG_CM, "Sending first write.\n"); + wqe->wqe_words[NES_IWARP_SQ_WQE_MISC_IDX] = + cpu_to_le32(NES_IWARP_SQ_OP_RDMAW); + wqe->wqe_words[NES_IWARP_SQ_WQE_TOTAL_PAYLOAD_IDX] = 0; + wqe->wqe_words[NES_IWARP_SQ_WQE_LENGTH0_IDX] = 0; + wqe->wqe_words[NES_IWARP_SQ_WQE_STAG0_IDX] = 0; + break; + + case SEND_RDMA_READ_ZERO: + default: + if (cm_node->send_rdma0_op != SEND_RDMA_READ_ZERO) { + printk(KERN_ERR "%s[%u]: Unsupported RDMA0 len operation=%u\n", + __func__, __LINE__, cm_node->send_rdma0_op); + WARN_ON(1); + } + nes_debug(NES_DBG_CM, "Sending first rdma operation.\n"); + wqe->wqe_words[NES_IWARP_SQ_WQE_MISC_IDX] = + cpu_to_le32(NES_IWARP_SQ_OP_RDMAR); + wqe->wqe_words[NES_IWARP_SQ_WQE_RDMA_TO_LOW_IDX] = 1; + wqe->wqe_words[NES_IWARP_SQ_WQE_RDMA_TO_HIGH_IDX] = 0; + wqe->wqe_words[NES_IWARP_SQ_WQE_RDMA_LENGTH_IDX] = 0; + wqe->wqe_words[NES_IWARP_SQ_WQE_RDMA_STAG_IDX] = 1; + wqe->wqe_words[NES_IWARP_SQ_WQE_STAG0_IDX] = 1; + break; + } + + if (nesqp->sq_kmapped) { + nesqp->sq_kmapped = 0; + kunmap(nesqp->page); + } + + /*use the reserved spot on the WQ for the extra first WQE*/ + nesqp->nesqp_context->ird_ord_sizes &= cpu_to_le32(~(NES_QPCONTEXT_ORDIRD_LSMM_PRESENT | + NES_QPCONTEXT_ORDIRD_WRPDU | + NES_QPCONTEXT_ORDIRD_ALSMM)); + nesqp->skip_lsmm = 1; + nesqp->hwqp.sq_tail = 0; +} /** * schedule_nes_timer @@ -430,10 +654,10 @@ static void print_core(struct nes_cm_core *core) * rem_ref_cm_node(cm_core, cm_node);add_ref_cm_node(cm_node); */ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, - enum nes_timer_type type, int send_retrans, - int close_when_complete) + enum nes_timer_type type, int send_retrans, + int close_when_complete) { - unsigned long flags; + unsigned long flags; struct nes_cm_core *cm_core = cm_node->cm_core; struct nes_timer_entry *new_send; int ret = 0; @@ -454,7 +678,7 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, new_send->close_when_complete = close_when_complete; if (type == NES_TIMER_TYPE_CLOSE) { - new_send->timetosend += (HZ/10); + new_send->timetosend += (HZ / 10); if (cm_node->recv_entry) { kfree(new_send); WARN_ON(1); @@ -475,7 +699,7 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, ret = nes_nic_cm_xmit(new_send->skb, cm_node->netdev); if (ret != NETDEV_TX_OK) { nes_debug(NES_DBG_CM, "Error sending packet %p " - "(jiffies = %lu)\n", new_send, jiffies); + "(jiffies = %lu)\n", new_send, jiffies); new_send->timetosend = jiffies; ret = NETDEV_TX_OK; } else { @@ -504,6 +728,7 @@ static void nes_retrans_expired(struct nes_cm_node *cm_node) struct iw_cm_id *cm_id = cm_node->cm_id; enum nes_cm_node_state state = cm_node->state; cm_node->state = NES_CM_STATE_CLOSED; + switch (state) { case NES_CM_STATE_SYN_RCVD: case NES_CM_STATE_CLOSING: @@ -536,10 +761,10 @@ static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node) spin_lock_irqsave(&nesqp->lock, qplockflags); if (nesqp->cm_id) { nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, " - "refcount = %d: HIT A " - "NES_TIMER_TYPE_CLOSE with something " - "to do!!!\n", nesqp->hwqp.qp_id, cm_id, - atomic_read(&nesqp->refcount)); + "refcount = %d: HIT A " + "NES_TIMER_TYPE_CLOSE with something " + "to do!!!\n", nesqp->hwqp.qp_id, cm_id, + atomic_read(&nesqp->refcount)); nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED; nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT; nesqp->ibqp_state = IB_QPS_ERR; @@ -548,10 +773,10 @@ static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node) } else { spin_unlock_irqrestore(&nesqp->lock, qplockflags); nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, " - "refcount = %d: HIT A " - "NES_TIMER_TYPE_CLOSE with nothing " - "to do!!!\n", nesqp->hwqp.qp_id, cm_id, - atomic_read(&nesqp->refcount)); + "refcount = %d: HIT A " + "NES_TIMER_TYPE_CLOSE with nothing " + "to do!!!\n", nesqp->hwqp.qp_id, cm_id, + atomic_read(&nesqp->refcount)); } } else if (rem_node) { /* TIME_WAIT state */ @@ -580,11 +805,12 @@ static void nes_cm_timer_tick(unsigned long pass) int ret = NETDEV_TX_OK; struct list_head timer_list; + INIT_LIST_HEAD(&timer_list); spin_lock_irqsave(&cm_core->ht_lock, flags); list_for_each_safe(list_node, list_core_temp, - &cm_core->connected_nodes) { + &cm_core->connected_nodes) { cm_node = container_of(list_node, struct nes_cm_node, list); if ((cm_node->recv_entry) || (cm_node->send_entry)) { add_ref_cm_node(cm_node); @@ -595,18 +821,19 @@ static void nes_cm_timer_tick(unsigned long pass) list_for_each_safe(list_node, list_core_temp, &timer_list) { cm_node = container_of(list_node, struct nes_cm_node, - timer_entry); + timer_entry); recv_entry = cm_node->recv_entry; if (recv_entry) { if (time_after(recv_entry->timetosend, jiffies)) { if (nexttimeout > recv_entry->timetosend || - !settimer) { + !settimer) { nexttimeout = recv_entry->timetosend; settimer = 1; } - } else + } else { handle_recv_entry(cm_node, 1); + } } spin_lock_irqsave(&cm_node->retrans_list_lock, flags); @@ -617,8 +844,8 @@ static void nes_cm_timer_tick(unsigned long pass) if (time_after(send_entry->timetosend, jiffies)) { if (cm_node->state != NES_CM_STATE_TSA) { if ((nexttimeout > - send_entry->timetosend) || - !settimer) { + send_entry->timetosend) || + !settimer) { nexttimeout = send_entry->timetosend; settimer = 1; @@ -630,13 +857,13 @@ static void nes_cm_timer_tick(unsigned long pass) } if ((cm_node->state == NES_CM_STATE_TSA) || - (cm_node->state == NES_CM_STATE_CLOSED)) { + (cm_node->state == NES_CM_STATE_CLOSED)) { free_retrans_entry(cm_node); break; } if (!send_entry->retranscount || - !send_entry->retrycount) { + !send_entry->retrycount) { cm_packets_dropped++; free_retrans_entry(cm_node); @@ -645,28 +872,28 @@ static void nes_cm_timer_tick(unsigned long pass) nes_retrans_expired(cm_node); cm_node->state = NES_CM_STATE_CLOSED; spin_lock_irqsave(&cm_node->retrans_list_lock, - flags); + flags); break; } atomic_inc(&send_entry->skb->users); cm_packets_retrans++; nes_debug(NES_DBG_CM, "Retransmitting send_entry %p " - "for node %p, jiffies = %lu, time to send = " - "%lu, retranscount = %u, send_entry->seq_num = " - "0x%08X, cm_node->tcp_cntxt.rem_ack_num = " - "0x%08X\n", send_entry, cm_node, jiffies, - send_entry->timetosend, - send_entry->retranscount, - send_entry->seq_num, - cm_node->tcp_cntxt.rem_ack_num); + "for node %p, jiffies = %lu, time to send = " + "%lu, retranscount = %u, send_entry->seq_num = " + "0x%08X, cm_node->tcp_cntxt.rem_ack_num = " + "0x%08X\n", send_entry, cm_node, jiffies, + send_entry->timetosend, + send_entry->retranscount, + send_entry->seq_num, + cm_node->tcp_cntxt.rem_ack_num); spin_unlock_irqrestore(&cm_node->retrans_list_lock, - flags); + flags); ret = nes_nic_cm_xmit(send_entry->skb, cm_node->netdev); spin_lock_irqsave(&cm_node->retrans_list_lock, flags); if (ret != NETDEV_TX_OK) { nes_debug(NES_DBG_CM, "rexmit failed for " - "node=%p\n", cm_node); + "node=%p\n", cm_node); cm_packets_bounced++; send_entry->retrycount--; nexttimeout = jiffies + NES_SHORT_TIME; @@ -676,18 +903,18 @@ static void nes_cm_timer_tick(unsigned long pass) cm_packets_sent++; } nes_debug(NES_DBG_CM, "Packet Sent: retrans count = " - "%u, retry count = %u.\n", - send_entry->retranscount, - send_entry->retrycount); + "%u, retry count = %u.\n", + send_entry->retranscount, + send_entry->retrycount); if (send_entry->send_retrans) { send_entry->retranscount--; timetosend = (NES_RETRY_TIMEOUT << - (NES_DEFAULT_RETRANS - send_entry->retranscount)); + (NES_DEFAULT_RETRANS - send_entry->retranscount)); send_entry->timetosend = jiffies + - min(timetosend, NES_MAX_TIMEOUT); + min(timetosend, NES_MAX_TIMEOUT); if (nexttimeout > send_entry->timetosend || - !settimer) { + !settimer) { nexttimeout = send_entry->timetosend; settimer = 1; } @@ -696,11 +923,11 @@ static void nes_cm_timer_tick(unsigned long pass) close_when_complete = send_entry->close_when_complete; nes_debug(NES_DBG_CM, "cm_node=%p state=%d\n", - cm_node, cm_node->state); + cm_node, cm_node->state); free_retrans_entry(cm_node); if (close_when_complete) rem_ref_cm_node(cm_node->cm_core, - cm_node); + cm_node); } } while (0); @@ -710,7 +937,7 @@ static void nes_cm_timer_tick(unsigned long pass) if (settimer) { if (!timer_pending(&cm_core->tcp_timer)) { - cm_core->tcp_timer.expires = nexttimeout; + cm_core->tcp_timer.expires = nexttimeout; add_timer(&cm_core->tcp_timer); } } @@ -721,13 +948,13 @@ static void nes_cm_timer_tick(unsigned long pass) * send_syn */ static int send_syn(struct nes_cm_node *cm_node, u32 sendack, - struct sk_buff *skb) + struct sk_buff *skb) { int ret; int flags = SET_SYN; char optionsbuffer[sizeof(struct option_mss) + - sizeof(struct option_windowscale) + sizeof(struct option_base) + - TCP_OPTIONS_PADDING]; + sizeof(struct option_windowscale) + sizeof(struct option_base) + + TCP_OPTIONS_PADDING]; int optionssize = 0; /* Sending MSS option */ @@ -854,7 +1081,7 @@ static int send_fin(struct nes_cm_node *cm_node, struct sk_buff *skb) * find_node - find a cm node that matches the reference cm node */ static struct nes_cm_node *find_node(struct nes_cm_core *cm_core, - u16 rem_port, nes_addr_t rem_addr, u16 loc_port, nes_addr_t loc_addr) + u16 rem_port, nes_addr_t rem_addr, u16 loc_port, nes_addr_t loc_addr) { unsigned long flags; struct list_head *hte; @@ -868,12 +1095,12 @@ static struct nes_cm_node *find_node(struct nes_cm_core *cm_core, list_for_each_entry(cm_node, hte, list) { /* compare quad, return node handle if a match */ nes_debug(NES_DBG_CM, "finding node %x:%x =? %x:%x ^ %x:%x =? %x:%x\n", - cm_node->loc_addr, cm_node->loc_port, - loc_addr, loc_port, - cm_node->rem_addr, cm_node->rem_port, - rem_addr, rem_port); + cm_node->loc_addr, cm_node->loc_port, + loc_addr, loc_port, + cm_node->rem_addr, cm_node->rem_port, + rem_addr, rem_port); if ((cm_node->loc_addr == loc_addr) && (cm_node->loc_port == loc_port) && - (cm_node->rem_addr == rem_addr) && (cm_node->rem_port == rem_port)) { + (cm_node->rem_addr == rem_addr) && (cm_node->rem_port == rem_port)) { add_ref_cm_node(cm_node); spin_unlock_irqrestore(&cm_core->ht_lock, flags); return cm_node; @@ -890,7 +1117,7 @@ static struct nes_cm_node *find_node(struct nes_cm_core *cm_core, * find_listener - find a cm node listening on this addr-port pair */ static struct nes_cm_listener *find_listener(struct nes_cm_core *cm_core, - nes_addr_t dst_addr, u16 dst_port, enum nes_cm_listener_state listener_state) + nes_addr_t dst_addr, u16 dst_port, enum nes_cm_listener_state listener_state) { unsigned long flags; struct nes_cm_listener *listen_node; @@ -900,9 +1127,9 @@ static struct nes_cm_listener *find_listener(struct nes_cm_core *cm_core, list_for_each_entry(listen_node, &cm_core->listen_list.list, list) { /* compare node pair, return node handle if a match */ if (((listen_node->loc_addr == dst_addr) || - listen_node->loc_addr == 0x00000000) && - (listen_node->loc_port == dst_port) && - (listener_state & listen_node->listener_state)) { + listen_node->loc_addr == 0x00000000) && + (listen_node->loc_port == dst_port) && + (listener_state & listen_node->listener_state)) { atomic_inc(&listen_node->ref_count); spin_unlock_irqrestore(&cm_core->listen_list_lock, flags); return listen_node; @@ -927,7 +1154,7 @@ static int add_hte_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node return -EINVAL; nes_debug(NES_DBG_CM, "Adding Node %p to Active Connection HT\n", - cm_node); + cm_node); spin_lock_irqsave(&cm_core->ht_lock, flags); @@ -946,7 +1173,7 @@ static int add_hte_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node * mini_cm_dec_refcnt_listen */ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, - struct nes_cm_listener *listener, int free_hanging_nodes) + struct nes_cm_listener *listener, int free_hanging_nodes) { int ret = -EINVAL; int err = 0; @@ -957,8 +1184,8 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, struct list_head reset_list; nes_debug(NES_DBG_CM, "attempting listener= %p free_nodes= %d, " - "refcnt=%d\n", listener, free_hanging_nodes, - atomic_read(&listener->ref_count)); + "refcnt=%d\n", listener, free_hanging_nodes, + atomic_read(&listener->ref_count)); /* free non-accelerated child nodes for this listener */ INIT_LIST_HEAD(&reset_list); if (free_hanging_nodes) { @@ -966,7 +1193,7 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, list_for_each_safe(list_pos, list_temp, &g_cm_core->connected_nodes) { cm_node = container_of(list_pos, struct nes_cm_node, - list); + list); if ((cm_node->listener == listener) && (!cm_node->accelerated)) { add_ref_cm_node(cm_node); @@ -978,7 +1205,7 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, list_for_each_safe(list_pos, list_temp, &reset_list) { cm_node = container_of(list_pos, struct nes_cm_node, - reset_entry); + reset_entry); { struct nes_cm_node *loopback = cm_node->loopbackpartner; enum nes_cm_node_state old_state; @@ -990,7 +1217,7 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, err = send_reset(cm_node, NULL); if (err) { cm_node->state = - NES_CM_STATE_CLOSED; + NES_CM_STATE_CLOSED; WARN_ON(1); } else { old_state = cm_node->state; @@ -1035,10 +1262,9 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, spin_unlock_irqrestore(&cm_core->listen_list_lock, flags); - if (listener->nesvnic) { + if (listener->nesvnic) nes_manage_apbvt(listener->nesvnic, listener->loc_port, - PCI_FUNC(listener->nesvnic->nesdev->pcidev->devfn), NES_MANAGE_APBVT_DEL); - } + PCI_FUNC(listener->nesvnic->nesdev->pcidev->devfn), NES_MANAGE_APBVT_DEL); nes_debug(NES_DBG_CM, "destroying listener (%p)\n", listener); @@ -1052,8 +1278,8 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, if (listener) { if (atomic_read(&listener->pend_accepts_cnt) > 0) nes_debug(NES_DBG_CM, "destroying listener (%p)" - " with non-zero pending accepts=%u\n", - listener, atomic_read(&listener->pend_accepts_cnt)); + " with non-zero pending accepts=%u\n", + listener, atomic_read(&listener->pend_accepts_cnt)); } return ret; @@ -1064,7 +1290,7 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, * mini_cm_del_listen */ static int mini_cm_del_listen(struct nes_cm_core *cm_core, - struct nes_cm_listener *listener) + struct nes_cm_listener *listener) { listener->listener_state = NES_CM_LISTENER_PASSIVE_STATE; listener->cm_id = NULL; /* going to be destroyed pretty soon */ @@ -1076,9 +1302,10 @@ static int mini_cm_del_listen(struct nes_cm_core *cm_core, * mini_cm_accelerated */ static inline int mini_cm_accelerated(struct nes_cm_core *cm_core, - struct nes_cm_node *cm_node) + struct nes_cm_node *cm_node) { u32 was_timer_set; + cm_node->accelerated = 1; if (cm_node->accept_pend) { @@ -1112,7 +1339,7 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip, int arpi rt = ip_route_output(&init_net, htonl(dst_ip), 0, 0, 0); if (IS_ERR(rt)) { printk(KERN_ERR "%s: ip_route_output_key failed for 0x%08X\n", - __func__, dst_ip); + __func__, dst_ip); return rc; } @@ -1130,7 +1357,7 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip, int arpi if (arpindex >= 0) { if (!memcmp(nesadapter->arp_table[arpindex].mac_addr, - neigh->ha, ETH_ALEN)){ + neigh->ha, ETH_ALEN)) { /* Mac address same as in nes_arp_table */ neigh_release(neigh); ip_rt_put(rt); @@ -1138,8 +1365,8 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip, int arpi } nes_manage_arp_cache(nesvnic->netdev, - nesadapter->arp_table[arpindex].mac_addr, - dst_ip, NES_ARP_DELETE); + nesadapter->arp_table[arpindex].mac_addr, + dst_ip, NES_ARP_DELETE); } nes_manage_arp_cache(nesvnic->netdev, neigh->ha, @@ -1163,8 +1390,8 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip, int arpi * make_cm_node - create a new instance of a cm node */ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, - struct nes_vnic *nesvnic, struct nes_cm_info *cm_info, - struct nes_cm_listener *listener) + struct nes_vnic *nesvnic, struct nes_cm_info *cm_info, + struct nes_cm_listener *listener) { struct nes_cm_node *cm_node; struct timespec ts; @@ -1183,7 +1410,12 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, cm_node->rem_addr = cm_info->rem_addr; cm_node->loc_port = cm_info->loc_port; cm_node->rem_port = cm_info->rem_port; - cm_node->send_write0 = send_first; + + cm_node->mpa_frame_rev = mpa_version; + cm_node->send_rdma0_op = SEND_RDMA_READ_ZERO; + cm_node->ird_size = IETF_NO_IRD_ORD; + cm_node->ord_size = IETF_NO_IRD_ORD; + nes_debug(NES_DBG_CM, "Make node addresses : loc = %pI4:%x, rem = %pI4:%x\n", &cm_node->loc_addr, cm_node->loc_port, &cm_node->rem_addr, cm_node->rem_port); @@ -1193,7 +1425,7 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, memcpy(cm_node->loc_mac, nesvnic->netdev->dev_addr, ETH_ALEN); nes_debug(NES_DBG_CM, "listener=%p, cm_id=%p\n", cm_node->listener, - cm_node->cm_id); + cm_node->cm_id); spin_lock_init(&cm_node->retrans_list_lock); @@ -1204,11 +1436,11 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, cm_node->tcp_cntxt.loc_id = NES_CM_DEF_LOCAL_ID; cm_node->tcp_cntxt.rcv_wscale = NES_CM_DEFAULT_RCV_WND_SCALE; cm_node->tcp_cntxt.rcv_wnd = NES_CM_DEFAULT_RCV_WND_SCALED >> - NES_CM_DEFAULT_RCV_WND_SCALE; + NES_CM_DEFAULT_RCV_WND_SCALE; ts = current_kernel_time(); cm_node->tcp_cntxt.loc_seq_num = htonl(ts.tv_nsec); cm_node->tcp_cntxt.mss = nesvnic->max_frame_size - sizeof(struct iphdr) - - sizeof(struct tcphdr) - ETH_HLEN - VLAN_HLEN; + sizeof(struct tcphdr) - ETH_HLEN - VLAN_HLEN; cm_node->tcp_cntxt.rcv_nxt = 0; /* get a unique session ID , add thread_id to an upcounter to handle race */ atomic_inc(&cm_core->node_cnt); @@ -1224,12 +1456,11 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, cm_node->loopbackpartner = NULL; /* get the mac addr for the remote node */ - if (ipv4_is_loopback(htonl(cm_node->rem_addr))) + if (ipv4_is_loopback(htonl(cm_node->rem_addr))) { arpindex = nes_arp_table(nesdev, ntohl(nesvnic->local_ipaddr), NULL, NES_ARP_RESOLVE); - else { + } else { oldarpindex = nes_arp_table(nesdev, cm_node->rem_addr, NULL, NES_ARP_RESOLVE); arpindex = nes_addr_resolve_neigh(nesvnic, cm_info->rem_addr, oldarpindex); - } if (arpindex < 0) { kfree(cm_node); @@ -1262,7 +1493,7 @@ static int add_ref_cm_node(struct nes_cm_node *cm_node) * rem_ref_cm_node - destroy an instance of a cm node */ static int rem_ref_cm_node(struct nes_cm_core *cm_core, - struct nes_cm_node *cm_node) + struct nes_cm_node *cm_node) { unsigned long flags; struct nes_qp *nesqp; @@ -1293,9 +1524,9 @@ static int rem_ref_cm_node(struct nes_cm_core *cm_core, } else { if (cm_node->apbvt_set && cm_node->nesvnic) { nes_manage_apbvt(cm_node->nesvnic, cm_node->loc_port, - PCI_FUNC( - cm_node->nesvnic->nesdev->pcidev->devfn), - NES_MANAGE_APBVT_DEL); + PCI_FUNC( + cm_node->nesvnic->nesdev->pcidev->devfn), + NES_MANAGE_APBVT_DEL); } } @@ -1316,7 +1547,7 @@ static int rem_ref_cm_node(struct nes_cm_core *cm_core, * process_options */ static int process_options(struct nes_cm_node *cm_node, u8 *optionsloc, - u32 optionsize, u32 syn_packet) + u32 optionsize, u32 syn_packet) { u32 tmp; u32 offset = 0; @@ -1334,15 +1565,15 @@ static int process_options(struct nes_cm_node *cm_node, u8 *optionsloc, continue; case OPTION_NUMBER_MSS: nes_debug(NES_DBG_CM, "%s: MSS Length: %d Offset: %d " - "Size: %d\n", __func__, - all_options->as_mss.length, offset, optionsize); + "Size: %d\n", __func__, + all_options->as_mss.length, offset, optionsize); got_mss_option = 1; if (all_options->as_mss.length != 4) { return 1; } else { tmp = ntohs(all_options->as_mss.mss); if (tmp > 0 && tmp < - cm_node->tcp_cntxt.mss) + cm_node->tcp_cntxt.mss) cm_node->tcp_cntxt.mss = tmp; } break; @@ -1350,12 +1581,9 @@ static int process_options(struct nes_cm_node *cm_node, u8 *optionsloc, cm_node->tcp_cntxt.snd_wscale = all_options->as_windowscale.shiftcount; break; - case OPTION_NUMBER_WRITE0: - cm_node->send_write0 = 1; - break; default: nes_debug(NES_DBG_CM, "TCP Option not understood: %x\n", - all_options->as_base.optionnum); + all_options->as_base.optionnum); break; } offset += all_options->as_base.length; @@ -1374,8 +1602,8 @@ static void drop_packet(struct sk_buff *skb) static void handle_fin_pkt(struct nes_cm_node *cm_node) { nes_debug(NES_DBG_CM, "Received FIN, cm_node = %p, state = %u. " - "refcnt=%d\n", cm_node, cm_node->state, - atomic_read(&cm_node->ref_count)); + "refcnt=%d\n", cm_node, cm_node->state, + atomic_read(&cm_node->ref_count)); switch (cm_node->state) { case NES_CM_STATE_SYN_RCVD: case NES_CM_STATE_SYN_SENT: @@ -1441,7 +1669,20 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, nes_debug(NES_DBG_CM, "%s[%u] create abort for cm_node=%p " "listener=%p state=%d\n", __func__, __LINE__, cm_node, cm_node->listener, cm_node->state); - active_open_err(cm_node, skb, reset); + switch (cm_node->mpa_frame_rev) { + case IETF_MPA_V2: + cm_node->mpa_frame_rev = IETF_MPA_V1; + /* send a syn and goto syn sent state */ + cm_node->state = NES_CM_STATE_SYN_SENT; + if (send_syn(cm_node, 0, NULL)) { + active_open_err(cm_node, skb, reset); + } + break; + case IETF_MPA_V1: + default: + active_open_err(cm_node, skb, reset); + break; + } break; case NES_CM_STATE_MPAREQ_RCVD: atomic_inc(&cm_node->passive_state); @@ -1477,21 +1718,21 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb) { - - int ret = 0; + int ret = 0; int datasize = skb->len; u8 *dataloc = skb->data; enum nes_cm_event_type type = NES_CM_EVENT_UNKNOWN; - u32 res_type; + u32 res_type; + ret = parse_mpa(cm_node, dataloc, &res_type, datasize); if (ret) { nes_debug(NES_DBG_CM, "didn't like MPA Request\n"); if (cm_node->state == NES_CM_STATE_MPAREQ_SENT) { nes_debug(NES_DBG_CM, "%s[%u] create abort for " - "cm_node=%p listener=%p state=%d\n", __func__, - __LINE__, cm_node, cm_node->listener, - cm_node->state); + "cm_node=%p listener=%p state=%d\n", __func__, + __LINE__, cm_node, cm_node->listener, + cm_node->state); active_open_err(cm_node, skb, 1); } else { passive_open_err(cm_node, skb, 1); @@ -1501,16 +1742,15 @@ static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb) switch (cm_node->state) { case NES_CM_STATE_ESTABLISHED: - if (res_type == NES_MPA_REQUEST_REJECT) { + if (res_type == NES_MPA_REQUEST_REJECT) /*BIG problem as we are receiving the MPA.. So should - * not be REJECT.. This is Passive Open.. We can - * only receive it Reject for Active Open...*/ + * not be REJECT.. This is Passive Open.. We can + * only receive it Reject for Active Open...*/ WARN_ON(1); - } cm_node->state = NES_CM_STATE_MPAREQ_RCVD; type = NES_CM_EVENT_MPA_REQ; atomic_set(&cm_node->passive_state, - NES_PASSIVE_STATE_INDICATED); + NES_PASSIVE_STATE_INDICATED); break; case NES_CM_STATE_MPAREQ_SENT: cleanup_retrans_entry(cm_node); @@ -1537,8 +1777,8 @@ static void indicate_pkt_err(struct nes_cm_node *cm_node, struct sk_buff *skb) case NES_CM_STATE_SYN_SENT: case NES_CM_STATE_MPAREQ_SENT: nes_debug(NES_DBG_CM, "%s[%u] create abort for cm_node=%p " - "listener=%p state=%d\n", __func__, __LINE__, cm_node, - cm_node->listener, cm_node->state); + "listener=%p state=%d\n", __func__, __LINE__, cm_node, + cm_node->listener, cm_node->state); active_open_err(cm_node, skb, 1); break; case NES_CM_STATE_ESTABLISHED: @@ -1552,11 +1792,11 @@ static void indicate_pkt_err(struct nes_cm_node *cm_node, struct sk_buff *skb) } static int check_syn(struct nes_cm_node *cm_node, struct tcphdr *tcph, - struct sk_buff *skb) + struct sk_buff *skb) { int err; - err = ((ntohl(tcph->ack_seq) == cm_node->tcp_cntxt.loc_seq_num))? 0 : 1; + err = ((ntohl(tcph->ack_seq) == cm_node->tcp_cntxt.loc_seq_num)) ? 0 : 1; if (err) active_open_err(cm_node, skb, 1); @@ -1564,7 +1804,7 @@ static int check_syn(struct nes_cm_node *cm_node, struct tcphdr *tcph, } static int check_seq(struct nes_cm_node *cm_node, struct tcphdr *tcph, - struct sk_buff *skb) + struct sk_buff *skb) { int err = 0; u32 seq; @@ -1572,21 +1812,22 @@ static int check_seq(struct nes_cm_node *cm_node, struct tcphdr *tcph, u32 loc_seq_num = cm_node->tcp_cntxt.loc_seq_num; u32 rcv_nxt = cm_node->tcp_cntxt.rcv_nxt; u32 rcv_wnd; + seq = ntohl(tcph->seq); ack_seq = ntohl(tcph->ack_seq); rcv_wnd = cm_node->tcp_cntxt.rcv_wnd; if (ack_seq != loc_seq_num) err = 1; - else if (!between(seq, rcv_nxt, (rcv_nxt+rcv_wnd))) + else if (!between(seq, rcv_nxt, (rcv_nxt + rcv_wnd))) err = 1; if (err) { nes_debug(NES_DBG_CM, "%s[%u] create abort for cm_node=%p " - "listener=%p state=%d\n", __func__, __LINE__, cm_node, - cm_node->listener, cm_node->state); + "listener=%p state=%d\n", __func__, __LINE__, cm_node, + cm_node->listener, cm_node->state); indicate_pkt_err(cm_node, skb); nes_debug(NES_DBG_CM, "seq ERROR cm_node =%p seq=0x%08X " - "rcv_nxt=0x%08X rcv_wnd=0x%x\n", cm_node, seq, rcv_nxt, - rcv_wnd); + "rcv_nxt=0x%08X rcv_wnd=0x%x\n", cm_node, seq, rcv_nxt, + rcv_wnd); } return err; } @@ -1596,9 +1837,8 @@ static int check_seq(struct nes_cm_node *cm_node, struct tcphdr *tcph, * is created with a listener or it may comein as rexmitted packet which in * that case will be just dropped. */ - static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, - struct tcphdr *tcph) + struct tcphdr *tcph) { int ret; u32 inc_sequence; @@ -1617,15 +1857,15 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, case NES_CM_STATE_LISTENING: /* Passive OPEN */ if (atomic_read(&cm_node->listener->pend_accepts_cnt) > - cm_node->listener->backlog) { + cm_node->listener->backlog) { nes_debug(NES_DBG_CM, "drop syn due to backlog " - "pressure \n"); + "pressure \n"); cm_backlog_drops++; passive_open_err(cm_node, skb, 0); break; } ret = handle_tcp_options(cm_node, tcph, skb, optionsize, - 1); + 1); if (ret) { passive_open_err(cm_node, skb, 0); /* drop pkt */ @@ -1659,9 +1899,8 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, } static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, - struct tcphdr *tcph) + struct tcphdr *tcph) { - int ret; u32 inc_sequence; int optionsize; @@ -1680,7 +1919,7 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, ret = handle_tcp_options(cm_node, tcph, skb, optionsize, 0); if (ret) { nes_debug(NES_DBG_CM, "cm_node=%p tcp_options failed\n", - cm_node); + cm_node); break; } cleanup_retrans_entry(cm_node); @@ -1719,12 +1958,13 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, } static int handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, - struct tcphdr *tcph) + struct tcphdr *tcph) { int datasize = 0; u32 inc_sequence; int ret = 0; int optionsize; + optionsize = (tcph->doff << 2) - sizeof(struct tcphdr); if (check_seq(cm_node, tcph, skb)) @@ -1745,8 +1985,9 @@ static int handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; handle_rcv_mpa(cm_node, skb); - } else /* rcvd ACK only */ + } else { /* rcvd ACK only */ dev_kfree_skb_any(skb); + } break; case NES_CM_STATE_ESTABLISHED: /* Passive OPEN */ @@ -1754,16 +1995,18 @@ static int handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; handle_rcv_mpa(cm_node, skb); - } else + } else { drop_packet(skb); + } break; case NES_CM_STATE_MPAREQ_SENT: cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq); if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; handle_rcv_mpa(cm_node, skb); - } else /* Could be just an ack pkt.. */ + } else { /* Could be just an ack pkt.. */ dev_kfree_skb_any(skb); + } break; case NES_CM_STATE_LISTENING: cleanup_retrans_entry(cm_node); @@ -1804,14 +2047,15 @@ static int handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, - struct sk_buff *skb, int optionsize, int passive) + struct sk_buff *skb, int optionsize, int passive) { u8 *optionsloc = (u8 *)&tcph[1]; + if (optionsize) { if (process_options(cm_node, optionsloc, optionsize, - (u32)tcph->syn)) { + (u32)tcph->syn)) { nes_debug(NES_DBG_CM, "%s: Node %p, Sending RESET\n", - __func__, cm_node); + __func__, cm_node); if (passive) passive_open_err(cm_node, skb, 1); else @@ -1821,7 +2065,7 @@ static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, } cm_node->tcp_cntxt.snd_wnd = ntohs(tcph->window) << - cm_node->tcp_cntxt.snd_wscale; + cm_node->tcp_cntxt.snd_wscale; if (cm_node->tcp_cntxt.snd_wnd > cm_node->tcp_cntxt.max_snd_wnd) cm_node->tcp_cntxt.max_snd_wnd = cm_node->tcp_cntxt.snd_wnd; @@ -1832,18 +2076,18 @@ static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, * active_open_err() will send reset() if flag set.. * It will also send ABORT event. */ - static void active_open_err(struct nes_cm_node *cm_node, struct sk_buff *skb, - int reset) + int reset) { cleanup_retrans_entry(cm_node); if (reset) { nes_debug(NES_DBG_CM, "ERROR active err called for cm_node=%p, " - "state=%d\n", cm_node, cm_node->state); + "state=%d\n", cm_node, cm_node->state); add_ref_cm_node(cm_node); send_reset(cm_node, skb); - } else + } else { dev_kfree_skb_any(skb); + } cm_node->state = NES_CM_STATE_CLOSED; create_event(cm_node, NES_CM_EVENT_ABORTED); @@ -1853,15 +2097,14 @@ static void active_open_err(struct nes_cm_node *cm_node, struct sk_buff *skb, * passive_open_err() will either do a reset() or will free up the skb and * remove the cm_node. */ - static void passive_open_err(struct nes_cm_node *cm_node, struct sk_buff *skb, - int reset) + int reset) { cleanup_retrans_entry(cm_node); cm_node->state = NES_CM_STATE_CLOSED; if (reset) { nes_debug(NES_DBG_CM, "passive_open_err sending RST for " - "cm_node=%p state =%d\n", cm_node, cm_node->state); + "cm_node=%p state =%d\n", cm_node, cm_node->state); send_reset(cm_node, skb); } else { dev_kfree_skb_any(skb); @@ -1876,6 +2119,7 @@ static void passive_open_err(struct nes_cm_node *cm_node, struct sk_buff *skb, static void free_retrans_entry(struct nes_cm_node *cm_node) { struct nes_timer_entry *send_entry; + send_entry = cm_node->send_entry; if (send_entry) { cm_node->send_entry = NULL; @@ -1899,26 +2143,28 @@ static void cleanup_retrans_entry(struct nes_cm_node *cm_node) * Returns skb if to be freed, else it will return NULL if already used.. */ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb, - struct nes_cm_core *cm_core) + struct nes_cm_core *cm_core) { - enum nes_tcpip_pkt_type pkt_type = NES_PKT_TYPE_UNKNOWN; + enum nes_tcpip_pkt_type pkt_type = NES_PKT_TYPE_UNKNOWN; struct tcphdr *tcph = tcp_hdr(skb); - u32 fin_set = 0; + u32 fin_set = 0; int ret = 0; + skb_pull(skb, ip_hdr(skb)->ihl << 2); nes_debug(NES_DBG_CM, "process_packet: cm_node=%p state =%d syn=%d " - "ack=%d rst=%d fin=%d\n", cm_node, cm_node->state, tcph->syn, - tcph->ack, tcph->rst, tcph->fin); + "ack=%d rst=%d fin=%d\n", cm_node, cm_node->state, tcph->syn, + tcph->ack, tcph->rst, tcph->fin); - if (tcph->rst) + if (tcph->rst) { pkt_type = NES_PKT_TYPE_RST; - else if (tcph->syn) { + } else if (tcph->syn) { pkt_type = NES_PKT_TYPE_SYN; if (tcph->ack) pkt_type = NES_PKT_TYPE_SYNACK; - } else if (tcph->ack) + } else if (tcph->ack) { pkt_type = NES_PKT_TYPE_ACK; + } if (tcph->fin) fin_set = 1; @@ -1949,17 +2195,17 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb, * mini_cm_listen - create a listen node with params */ static struct nes_cm_listener *mini_cm_listen(struct nes_cm_core *cm_core, - struct nes_vnic *nesvnic, struct nes_cm_info *cm_info) + struct nes_vnic *nesvnic, struct nes_cm_info *cm_info) { struct nes_cm_listener *listener; unsigned long flags; nes_debug(NES_DBG_CM, "Search for 0x%08x : 0x%04x\n", - cm_info->loc_addr, cm_info->loc_port); + cm_info->loc_addr, cm_info->loc_port); /* cannot have multiple matching listeners */ listener = find_listener(cm_core, htonl(cm_info->loc_addr), - htons(cm_info->loc_port), NES_CM_LISTENER_EITHER_STATE); + htons(cm_info->loc_port), NES_CM_LISTENER_EITHER_STATE); if (listener && listener->listener_state == NES_CM_LISTENER_ACTIVE_STATE) { /* find automatically incs ref count ??? */ atomic_dec(&listener->ref_count); @@ -2005,9 +2251,9 @@ static struct nes_cm_listener *mini_cm_listen(struct nes_cm_core *cm_core, } nes_debug(NES_DBG_CM, "Api - listen(): addr=0x%08X, port=0x%04x," - " listener = %p, backlog = %d, cm_id = %p.\n", - cm_info->loc_addr, cm_info->loc_port, - listener, listener->backlog, listener->cm_id); + " listener = %p, backlog = %d, cm_id = %p.\n", + cm_info->loc_addr, cm_info->loc_port, + listener, listener->backlog, listener->cm_id); return listener; } @@ -2017,26 +2263,20 @@ static struct nes_cm_listener *mini_cm_listen(struct nes_cm_core *cm_core, * mini_cm_connect - make a connection node with params */ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, - struct nes_vnic *nesvnic, u16 private_data_len, - void *private_data, struct nes_cm_info *cm_info) + struct nes_vnic *nesvnic, u16 private_data_len, + void *private_data, struct nes_cm_info *cm_info) { int ret = 0; struct nes_cm_node *cm_node; struct nes_cm_listener *loopbackremotelistener; struct nes_cm_node *loopbackremotenode; struct nes_cm_info loopback_cm_info; - u16 mpa_frame_size = sizeof(struct ietf_mpa_frame) + private_data_len; - struct ietf_mpa_frame *mpa_frame = NULL; + u8 *start_buff; /* create a CM connection node */ cm_node = make_cm_node(cm_core, nesvnic, cm_info, NULL); if (!cm_node) return NULL; - mpa_frame = &cm_node->mpa_frame; - memcpy(mpa_frame->key, IEFT_MPA_KEY_REQ, IETF_MPA_KEY_SIZE); - mpa_frame->flags = IETF_MPA_FLAGS_CRC; - mpa_frame->rev = IETF_MPA_VERSION; - mpa_frame->priv_data_len = htons(private_data_len); /* set our node side to client (active) side */ cm_node->tcp_cntxt.client = 1; @@ -2044,8 +2284,8 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, if (cm_info->loc_addr == cm_info->rem_addr) { loopbackremotelistener = find_listener(cm_core, - ntohl(nesvnic->local_ipaddr), cm_node->rem_port, - NES_CM_LISTENER_ACTIVE_STATE); + ntohl(nesvnic->local_ipaddr), cm_node->rem_port, + NES_CM_LISTENER_ACTIVE_STATE); if (loopbackremotelistener == NULL) { create_event(cm_node, NES_CM_EVENT_ABORTED); } else { @@ -2054,7 +2294,7 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, loopback_cm_info.rem_port = cm_info->loc_port; loopback_cm_info.cm_id = loopbackremotelistener->cm_id; loopbackremotenode = make_cm_node(cm_core, nesvnic, - &loopback_cm_info, loopbackremotelistener); + &loopback_cm_info, loopbackremotelistener); if (!loopbackremotenode) { rem_ref_cm_node(cm_node->cm_core, cm_node); return NULL; @@ -2065,7 +2305,7 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, NES_CM_DEFAULT_RCV_WND_SCALE; cm_node->loopbackpartner = loopbackremotenode; memcpy(loopbackremotenode->mpa_frame_buf, private_data, - private_data_len); + private_data_len); loopbackremotenode->mpa_frame_size = private_data_len; /* we are done handling this state. */ @@ -2093,12 +2333,10 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, return cm_node; } - /* set our node side to client (active) side */ - cm_node->tcp_cntxt.client = 1; - /* init our MPA frame ptr */ - memcpy(mpa_frame->priv_data, private_data, private_data_len); + start_buff = &cm_node->mpa_frame_buf[0] + sizeof(struct ietf_mpa_v2); + cm_node->mpa_frame_size = private_data_len; - cm_node->mpa_frame_size = mpa_frame_size; + memcpy(start_buff, private_data, private_data_len); /* send a syn and goto syn sent state */ cm_node->state = NES_CM_STATE_SYN_SENT; @@ -2107,18 +2345,19 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, if (ret) { /* error in sending the syn free up the cm_node struct */ nes_debug(NES_DBG_CM, "Api - connect() FAILED: dest " - "addr=0x%08X, port=0x%04x, cm_node=%p, cm_id = %p.\n", - cm_node->rem_addr, cm_node->rem_port, cm_node, - cm_node->cm_id); + "addr=0x%08X, port=0x%04x, cm_node=%p, cm_id = %p.\n", + cm_node->rem_addr, cm_node->rem_port, cm_node, + cm_node->cm_id); rem_ref_cm_node(cm_node->cm_core, cm_node); cm_node = NULL; } - if (cm_node) + if (cm_node) { nes_debug(NES_DBG_CM, "Api - connect(): dest addr=0x%08X," - "port=0x%04x, cm_node=%p, cm_id = %p.\n", - cm_node->rem_addr, cm_node->rem_port, cm_node, - cm_node->cm_id); + "port=0x%04x, cm_node=%p, cm_id = %p.\n", + cm_node->rem_addr, cm_node->rem_port, cm_node, + cm_node->cm_id); + } return cm_node; } @@ -2128,8 +2367,7 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, * mini_cm_accept - accept a connection * This function is never called */ -static int mini_cm_accept(struct nes_cm_core *cm_core, - struct ietf_mpa_frame *mpa_frame, struct nes_cm_node *cm_node) +static int mini_cm_accept(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node) { return 0; } @@ -2138,8 +2376,7 @@ static int mini_cm_accept(struct nes_cm_core *cm_core, /** * mini_cm_reject - reject and teardown a connection */ -static int mini_cm_reject(struct nes_cm_core *cm_core, - struct ietf_mpa_frame *mpa_frame, struct nes_cm_node *cm_node) +static int mini_cm_reject(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node) { int ret = 0; int err = 0; @@ -2149,7 +2386,7 @@ static int mini_cm_reject(struct nes_cm_core *cm_core, struct nes_cm_node *loopback = cm_node->loopbackpartner; nes_debug(NES_DBG_CM, "%s cm_node=%p type=%d state=%d\n", - __func__, cm_node, cm_node->tcp_cntxt.client, cm_node->state); + __func__, cm_node, cm_node->tcp_cntxt.client, cm_node->state); if (cm_node->tcp_cntxt.client) return ret; @@ -2170,8 +2407,9 @@ static int mini_cm_reject(struct nes_cm_core *cm_core, err = send_reset(cm_node, NULL); if (err) WARN_ON(1); - } else + } else { cm_id->add_ref(cm_id); + } } } } else { @@ -2246,7 +2484,7 @@ static int mini_cm_close(struct nes_cm_core *cm_core, struct nes_cm_node *cm_nod case NES_CM_STATE_TSA: if (cm_node->send_entry) printk(KERN_ERR "ERROR Close got called from STATE_TSA " - "send_entry=%p\n", cm_node->send_entry); + "send_entry=%p\n", cm_node->send_entry); ret = rem_ref_cm_node(cm_core, cm_node); break; } @@ -2259,7 +2497,7 @@ static int mini_cm_close(struct nes_cm_core *cm_core, struct nes_cm_node *cm_nod * node state machine */ static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, - struct nes_vnic *nesvnic, struct sk_buff *skb) + struct nes_vnic *nesvnic, struct sk_buff *skb) { struct nes_cm_node *cm_node = NULL; struct nes_cm_listener *listener = NULL; @@ -2271,9 +2509,8 @@ static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, if (!skb) return 0; - if (skb->len < sizeof(struct iphdr) + sizeof(struct tcphdr)) { + if (skb->len < sizeof(struct iphdr) + sizeof(struct tcphdr)) return 0; - } iph = (struct iphdr *)skb->data; tcph = (struct tcphdr *)(skb->data + sizeof(struct iphdr)); @@ -2291,8 +2528,8 @@ static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, do { cm_node = find_node(cm_core, - nfo.rem_port, nfo.rem_addr, - nfo.loc_port, nfo.loc_addr); + nfo.rem_port, nfo.rem_addr, + nfo.loc_port, nfo.loc_addr); if (!cm_node) { /* Only type of packet accepted are for */ @@ -2302,8 +2539,8 @@ static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, break; } listener = find_listener(cm_core, nfo.loc_addr, - nfo.loc_port, - NES_CM_LISTENER_ACTIVE_STATE); + nfo.loc_port, + NES_CM_LISTENER_ACTIVE_STATE); if (!listener) { nfo.cm_id = NULL; nfo.conn_type = 0; @@ -2314,10 +2551,10 @@ static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, nfo.cm_id = listener->cm_id; nfo.conn_type = listener->conn_type; cm_node = make_cm_node(cm_core, nesvnic, &nfo, - listener); + listener); if (!cm_node) { nes_debug(NES_DBG_CM, "Unable to allocate " - "node\n"); + "node\n"); cm_packets_dropped++; atomic_dec(&listener->ref_count); dev_kfree_skb_any(skb); @@ -2333,9 +2570,13 @@ static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, } add_ref_cm_node(cm_node); } else if (cm_node->state == NES_CM_STATE_TSA) { - rem_ref_cm_node(cm_core, cm_node); - atomic_inc(&cm_accel_dropped_pkts); - dev_kfree_skb_any(skb); + if (cm_node->nesqp->pau_mode) + nes_queue_mgt_skbs(skb, nesvnic, cm_node->nesqp); + else { + rem_ref_cm_node(cm_core, cm_node); + atomic_inc(&cm_accel_dropped_pkts); + dev_kfree_skb_any(skb); + } break; } skb_reset_network_header(skb); @@ -2365,7 +2606,7 @@ static struct nes_cm_core *nes_cm_alloc_core(void) init_timer(&cm_core->tcp_timer); cm_core->tcp_timer.function = nes_cm_timer_tick; - cm_core->mtu = NES_CM_DEFAULT_MTU; + cm_core->mtu = NES_CM_DEFAULT_MTU; cm_core->state = NES_CM_STATE_INITED; cm_core->free_tx_pkt_max = NES_CM_DEFAULT_FREE_PKTS; @@ -2403,9 +2644,8 @@ static int mini_cm_dealloc_core(struct nes_cm_core *cm_core) barrier(); - if (timer_pending(&cm_core->tcp_timer)) { + if (timer_pending(&cm_core->tcp_timer)) del_timer(&cm_core->tcp_timer); - } destroy_workqueue(cm_core->event_wq); destroy_workqueue(cm_core->disconn_wq); @@ -2460,8 +2700,8 @@ static int nes_cm_init_tsa_conn(struct nes_qp *nesqp, struct nes_cm_node *cm_nod return -EINVAL; nesqp->nesqp_context->misc |= cpu_to_le32(NES_QPCONTEXT_MISC_IPV4 | - NES_QPCONTEXT_MISC_NO_NAGLE | NES_QPCONTEXT_MISC_DO_NOT_FRAG | - NES_QPCONTEXT_MISC_DROS); + NES_QPCONTEXT_MISC_NO_NAGLE | NES_QPCONTEXT_MISC_DO_NOT_FRAG | + NES_QPCONTEXT_MISC_DROS); if (cm_node->tcp_cntxt.snd_wscale || cm_node->tcp_cntxt.rcv_wscale) nesqp->nesqp_context->misc |= cpu_to_le32(NES_QPCONTEXT_MISC_WSCALE); @@ -2471,15 +2711,15 @@ static int nes_cm_init_tsa_conn(struct nes_qp *nesqp, struct nes_cm_node *cm_nod nesqp->nesqp_context->mss |= cpu_to_le32(((u32)cm_node->tcp_cntxt.mss) << 16); nesqp->nesqp_context->tcp_state_flow_label |= cpu_to_le32( - (u32)NES_QPCONTEXT_TCPSTATE_EST << NES_QPCONTEXT_TCPFLOW_TCP_STATE_SHIFT); + (u32)NES_QPCONTEXT_TCPSTATE_EST << NES_QPCONTEXT_TCPFLOW_TCP_STATE_SHIFT); nesqp->nesqp_context->pd_index_wscale |= cpu_to_le32( - (cm_node->tcp_cntxt.snd_wscale << NES_QPCONTEXT_PDWSCALE_SND_WSCALE_SHIFT) & - NES_QPCONTEXT_PDWSCALE_SND_WSCALE_MASK); + (cm_node->tcp_cntxt.snd_wscale << NES_QPCONTEXT_PDWSCALE_SND_WSCALE_SHIFT) & + NES_QPCONTEXT_PDWSCALE_SND_WSCALE_MASK); nesqp->nesqp_context->pd_index_wscale |= cpu_to_le32( - (cm_node->tcp_cntxt.rcv_wscale << NES_QPCONTEXT_PDWSCALE_RCV_WSCALE_SHIFT) & - NES_QPCONTEXT_PDWSCALE_RCV_WSCALE_MASK); + (cm_node->tcp_cntxt.rcv_wscale << NES_QPCONTEXT_PDWSCALE_RCV_WSCALE_SHIFT) & + NES_QPCONTEXT_PDWSCALE_RCV_WSCALE_MASK); nesqp->nesqp_context->keepalive = cpu_to_le32(0x80); nesqp->nesqp_context->ts_recent = 0; @@ -2488,24 +2728,24 @@ static int nes_cm_init_tsa_conn(struct nes_qp *nesqp, struct nes_cm_node *cm_nod nesqp->nesqp_context->snd_wnd = cpu_to_le32(cm_node->tcp_cntxt.snd_wnd); nesqp->nesqp_context->rcv_nxt = cpu_to_le32(cm_node->tcp_cntxt.rcv_nxt); nesqp->nesqp_context->rcv_wnd = cpu_to_le32(cm_node->tcp_cntxt.rcv_wnd << - cm_node->tcp_cntxt.rcv_wscale); + cm_node->tcp_cntxt.rcv_wscale); nesqp->nesqp_context->snd_max = cpu_to_le32(cm_node->tcp_cntxt.loc_seq_num); nesqp->nesqp_context->snd_una = cpu_to_le32(cm_node->tcp_cntxt.loc_seq_num); nesqp->nesqp_context->srtt = 0; nesqp->nesqp_context->rttvar = cpu_to_le32(0x6); nesqp->nesqp_context->ssthresh = cpu_to_le32(0x3FFFC000); - nesqp->nesqp_context->cwnd = cpu_to_le32(2*cm_node->tcp_cntxt.mss); + nesqp->nesqp_context->cwnd = cpu_to_le32(2 * cm_node->tcp_cntxt.mss); nesqp->nesqp_context->snd_wl1 = cpu_to_le32(cm_node->tcp_cntxt.rcv_nxt); nesqp->nesqp_context->snd_wl2 = cpu_to_le32(cm_node->tcp_cntxt.loc_seq_num); nesqp->nesqp_context->max_snd_wnd = cpu_to_le32(cm_node->tcp_cntxt.max_snd_wnd); nes_debug(NES_DBG_CM, "QP%u: rcv_nxt = 0x%08X, snd_nxt = 0x%08X," - " Setting MSS to %u, PDWscale = 0x%08X, rcv_wnd = %u, context misc = 0x%08X.\n", - nesqp->hwqp.qp_id, le32_to_cpu(nesqp->nesqp_context->rcv_nxt), - le32_to_cpu(nesqp->nesqp_context->snd_nxt), - cm_node->tcp_cntxt.mss, le32_to_cpu(nesqp->nesqp_context->pd_index_wscale), - le32_to_cpu(nesqp->nesqp_context->rcv_wnd), - le32_to_cpu(nesqp->nesqp_context->misc)); + " Setting MSS to %u, PDWscale = 0x%08X, rcv_wnd = %u, context misc = 0x%08X.\n", + nesqp->hwqp.qp_id, le32_to_cpu(nesqp->nesqp_context->rcv_nxt), + le32_to_cpu(nesqp->nesqp_context->snd_nxt), + cm_node->tcp_cntxt.mss, le32_to_cpu(nesqp->nesqp_context->pd_index_wscale), + le32_to_cpu(nesqp->nesqp_context->rcv_wnd), + le32_to_cpu(nesqp->nesqp_context->misc)); nes_debug(NES_DBG_CM, " snd_wnd = 0x%08X.\n", le32_to_cpu(nesqp->nesqp_context->snd_wnd)); nes_debug(NES_DBG_CM, " snd_cwnd = 0x%08X.\n", le32_to_cpu(nesqp->nesqp_context->cwnd)); nes_debug(NES_DBG_CM, " max_swnd = 0x%08X.\n", le32_to_cpu(nesqp->nesqp_context->max_snd_wnd)); @@ -2526,7 +2766,7 @@ int nes_cm_disconn(struct nes_qp *nesqp) work = kzalloc(sizeof *work, GFP_ATOMIC); if (!work) - return -ENOMEM; /* Timer will clean up */ + return -ENOMEM; /* Timer will clean up */ nes_add_ref(&nesqp->ibqp); work->nesqp = nesqp; @@ -2546,7 +2786,7 @@ static void nes_disconnect_worker(struct work_struct *work) kfree(dwork); nes_debug(NES_DBG_CM, "processing AEQE id 0x%04X for QP%u.\n", - nesqp->last_aeq, nesqp->hwqp.qp_id); + nesqp->last_aeq, nesqp->hwqp.qp_id); nes_cm_disconn_true(nesqp); nes_rem_ref(&nesqp->ibqp); } @@ -2582,7 +2822,7 @@ static int nes_cm_disconn_true(struct nes_qp *nesqp) /* make sure we havent already closed this connection */ if (!cm_id) { nes_debug(NES_DBG_CM, "QP%u disconnect_worker cmid is NULL\n", - nesqp->hwqp.qp_id); + nesqp->hwqp.qp_id); spin_unlock_irqrestore(&nesqp->lock, flags); return -1; } @@ -2591,7 +2831,7 @@ static int nes_cm_disconn_true(struct nes_qp *nesqp) nes_debug(NES_DBG_CM, "Disconnecting QP%u\n", nesqp->hwqp.qp_id); original_hw_tcp_state = nesqp->hw_tcp_state; - original_ibqp_state = nesqp->ibqp_state; + original_ibqp_state = nesqp->ibqp_state; last_ae = nesqp->last_aeq; if (nesqp->term_flags) { @@ -2649,16 +2889,16 @@ static int nes_cm_disconn_true(struct nes_qp *nesqp) cm_event.private_data_len = 0; nes_debug(NES_DBG_CM, "Generating a CM Disconnect Event" - " for QP%u, SQ Head = %u, SQ Tail = %u. " - "cm_id = %p, refcount = %u.\n", - nesqp->hwqp.qp_id, nesqp->hwqp.sq_head, - nesqp->hwqp.sq_tail, cm_id, - atomic_read(&nesqp->refcount)); + " for QP%u, SQ Head = %u, SQ Tail = %u. " + "cm_id = %p, refcount = %u.\n", + nesqp->hwqp.qp_id, nesqp->hwqp.sq_head, + nesqp->hwqp.sq_tail, cm_id, + atomic_read(&nesqp->refcount)); ret = cm_id->event_handler(cm_id, &cm_event); if (ret) nes_debug(NES_DBG_CM, "OFA CM event_handler " - "returned, ret=%d\n", ret); + "returned, ret=%d\n", ret); } if (issue_close) { @@ -2676,9 +2916,8 @@ static int nes_cm_disconn_true(struct nes_qp *nesqp) cm_event.private_data_len = 0; ret = cm_id->event_handler(cm_id, &cm_event); - if (ret) { + if (ret) nes_debug(NES_DBG_CM, "OFA CM event_handler returned, ret=%d\n", ret); - } cm_id->rem_ref(cm_id); } @@ -2718,8 +2957,8 @@ static int nes_disconnect(struct nes_qp *nesqp, int abrupt) if (nesqp->lsmm_mr) nesibdev->ibdev.dereg_mr(nesqp->lsmm_mr); pci_free_consistent(nesdev->pcidev, - nesqp->private_data_len+sizeof(struct ietf_mpa_frame), - nesqp->ietf_frame, nesqp->ietf_frame_pbase); + nesqp->private_data_len + nesqp->ietf_frame_size, + nesqp->ietf_frame, nesqp->ietf_frame_pbase); } } @@ -2758,6 +2997,12 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) struct ib_phys_buf ibphysbuf; struct nes_pd *nespd; u64 tagged_offset; + u8 mpa_frame_offset = 0; + struct ietf_mpa_v2 *mpa_v2_frame; + u8 start_addr = 0; + u8 *start_ptr = &start_addr; + u8 **start_buff = &start_ptr; + u16 buff_len = 0; ibqp = nes_get_qp(cm_id->device, conn_param->qpn); if (!ibqp) @@ -2798,53 +3043,49 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) nes_debug(NES_DBG_CM, "netdev refcnt = %u.\n", netdev_refcnt_read(nesvnic->netdev)); + nesqp->ietf_frame_size = sizeof(struct ietf_mpa_v2); /* allocate the ietf frame and space for private data */ nesqp->ietf_frame = pci_alloc_consistent(nesdev->pcidev, - sizeof(struct ietf_mpa_frame) + conn_param->private_data_len, - &nesqp->ietf_frame_pbase); + nesqp->ietf_frame_size + conn_param->private_data_len, + &nesqp->ietf_frame_pbase); if (!nesqp->ietf_frame) { - nes_debug(NES_DBG_CM, "Unable to allocate memory for private " - "data\n"); + nes_debug(NES_DBG_CM, "Unable to allocate memory for private data\n"); return -ENOMEM; } + mpa_v2_frame = (struct ietf_mpa_v2 *)nesqp->ietf_frame; + if (cm_node->mpa_frame_rev == IETF_MPA_V1) + mpa_frame_offset = 4; - /* setup the MPA frame */ - nesqp->private_data_len = conn_param->private_data_len; - memcpy(nesqp->ietf_frame->key, IEFT_MPA_KEY_REP, IETF_MPA_KEY_SIZE); - - memcpy(nesqp->ietf_frame->priv_data, conn_param->private_data, - conn_param->private_data_len); + memcpy(mpa_v2_frame->priv_data, conn_param->private_data, + conn_param->private_data_len); - nesqp->ietf_frame->priv_data_len = - cpu_to_be16(conn_param->private_data_len); - nesqp->ietf_frame->rev = mpa_version; - nesqp->ietf_frame->flags = IETF_MPA_FLAGS_CRC; + cm_build_mpa_frame(cm_node, start_buff, &buff_len, nesqp->ietf_frame, MPA_KEY_REPLY); + nesqp->private_data_len = conn_param->private_data_len; /* setup our first outgoing iWarp send WQE (the IETF frame response) */ wqe = &nesqp->hwqp.sq_vbase[0]; if (cm_id->remote_addr.sin_addr.s_addr != - cm_id->local_addr.sin_addr.s_addr) { + cm_id->local_addr.sin_addr.s_addr) { u64temp = (unsigned long)nesqp; nesibdev = nesvnic->nesibdev; nespd = nesqp->nespd; - ibphysbuf.addr = nesqp->ietf_frame_pbase; - ibphysbuf.size = conn_param->private_data_len + - sizeof(struct ietf_mpa_frame); - tagged_offset = (u64)(unsigned long)nesqp->ietf_frame; + ibphysbuf.addr = nesqp->ietf_frame_pbase + mpa_frame_offset; + ibphysbuf.size = buff_len; + tagged_offset = (u64)(unsigned long)*start_buff; ibmr = nesibdev->ibdev.reg_phys_mr((struct ib_pd *)nespd, - &ibphysbuf, 1, - IB_ACCESS_LOCAL_WRITE, - &tagged_offset); + &ibphysbuf, 1, + IB_ACCESS_LOCAL_WRITE, + &tagged_offset); if (!ibmr) { nes_debug(NES_DBG_CM, "Unable to register memory region" - "for lSMM for cm_node = %p \n", - cm_node); + "for lSMM for cm_node = %p \n", + cm_node); pci_free_consistent(nesdev->pcidev, - nesqp->private_data_len+sizeof(struct ietf_mpa_frame), - nesqp->ietf_frame, nesqp->ietf_frame_pbase); + nesqp->private_data_len + nesqp->ietf_frame_size, + nesqp->ietf_frame, nesqp->ietf_frame_pbase); return -ENOMEM; } @@ -2852,22 +3093,20 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) ibmr->device = nespd->ibpd.device; nesqp->lsmm_mr = ibmr; - u64temp |= NES_SW_CONTEXT_ALIGN>>1; + u64temp |= NES_SW_CONTEXT_ALIGN >> 1; set_wqe_64bit_value(wqe->wqe_words, - NES_IWARP_SQ_WQE_COMP_CTX_LOW_IDX, - u64temp); + NES_IWARP_SQ_WQE_COMP_CTX_LOW_IDX, + u64temp); wqe->wqe_words[NES_IWARP_SQ_WQE_MISC_IDX] = cpu_to_le32(NES_IWARP_SQ_WQE_STREAMING | - NES_IWARP_SQ_WQE_WRPDU); + NES_IWARP_SQ_WQE_WRPDU); wqe->wqe_words[NES_IWARP_SQ_WQE_TOTAL_PAYLOAD_IDX] = - cpu_to_le32(conn_param->private_data_len + - sizeof(struct ietf_mpa_frame)); + cpu_to_le32(buff_len); set_wqe_64bit_value(wqe->wqe_words, - NES_IWARP_SQ_WQE_FRAG0_LOW_IDX, - (u64)(unsigned long)nesqp->ietf_frame); + NES_IWARP_SQ_WQE_FRAG0_LOW_IDX, + (u64)(unsigned long)(*start_buff)); wqe->wqe_words[NES_IWARP_SQ_WQE_LENGTH0_IDX] = - cpu_to_le32(conn_param->private_data_len + - sizeof(struct ietf_mpa_frame)); + cpu_to_le32(buff_len); wqe->wqe_words[NES_IWARP_SQ_WQE_STAG0_IDX] = ibmr->lkey; if (nesqp->sq_kmapped) { nesqp->sq_kmapped = 0; @@ -2876,7 +3115,7 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) nesqp->nesqp_context->ird_ord_sizes |= cpu_to_le32(NES_QPCONTEXT_ORDIRD_LSMM_PRESENT | - NES_QPCONTEXT_ORDIRD_WRPDU); + NES_QPCONTEXT_ORDIRD_WRPDU); } else { nesqp->nesqp_context->ird_ord_sizes |= cpu_to_le32(NES_QPCONTEXT_ORDIRD_WRPDU); @@ -2890,11 +3129,11 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) /* nesqp->cm_node = (void *)cm_id->provider_data; */ cm_id->provider_data = nesqp; - nesqp->active_conn = 0; + nesqp->active_conn = 0; if (cm_node->state == NES_CM_STATE_TSA) nes_debug(NES_DBG_CM, "Already state = TSA for cm_node=%p\n", - cm_node); + cm_node); nes_cm_init_tsa_conn(nesqp, cm_node); @@ -2911,13 +3150,13 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) cpu_to_le32(ntohl(cm_id->remote_addr.sin_addr.s_addr)); nesqp->nesqp_context->misc2 |= cpu_to_le32( - (u32)PCI_FUNC(nesdev->pcidev->devfn) << - NES_QPCONTEXT_MISC2_SRC_IP_SHIFT); + (u32)PCI_FUNC(nesdev->pcidev->devfn) << + NES_QPCONTEXT_MISC2_SRC_IP_SHIFT); nesqp->nesqp_context->arp_index_vlan |= cpu_to_le32(nes_arp_table(nesdev, - le32_to_cpu(nesqp->nesqp_context->ip0), NULL, - NES_ARP_RESOLVE) << 16); + le32_to_cpu(nesqp->nesqp_context->ip0), NULL, + NES_ARP_RESOLVE) << 16); nesqp->nesqp_context->ts_val_delta = cpu_to_le32( jiffies - nes_read_indexed(nesdev, NES_IDX_TCP_NOW)); @@ -2943,7 +3182,7 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) crc_value = get_crc_value(&nes_quad); nesqp->hte_index = cpu_to_be32(crc_value ^ 0xffffffff); nes_debug(NES_DBG_CM, "HTE Index = 0x%08X, CRC = 0x%08X\n", - nesqp->hte_index, nesqp->hte_index & adapter->hte_index_mask); + nesqp->hte_index, nesqp->hte_index & adapter->hte_index_mask); nesqp->hte_index &= adapter->hte_index_mask; nesqp->nesqp_context->hte_index = cpu_to_le32(nesqp->hte_index); @@ -2951,17 +3190,15 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) cm_node->cm_core->api->accelerated(cm_node->cm_core, cm_node); nes_debug(NES_DBG_CM, "QP%u, Destination IP = 0x%08X:0x%04X, local = " - "0x%08X:0x%04X, rcv_nxt=0x%08X, snd_nxt=0x%08X, mpa + " - "private data length=%zu.\n", nesqp->hwqp.qp_id, - ntohl(cm_id->remote_addr.sin_addr.s_addr), - ntohs(cm_id->remote_addr.sin_port), - ntohl(cm_id->local_addr.sin_addr.s_addr), - ntohs(cm_id->local_addr.sin_port), - le32_to_cpu(nesqp->nesqp_context->rcv_nxt), - le32_to_cpu(nesqp->nesqp_context->snd_nxt), - conn_param->private_data_len + - sizeof(struct ietf_mpa_frame)); - + "0x%08X:0x%04X, rcv_nxt=0x%08X, snd_nxt=0x%08X, mpa + " + "private data length=%u.\n", nesqp->hwqp.qp_id, + ntohl(cm_id->remote_addr.sin_addr.s_addr), + ntohs(cm_id->remote_addr.sin_port), + ntohl(cm_id->local_addr.sin_addr.s_addr), + ntohs(cm_id->local_addr.sin_port), + le32_to_cpu(nesqp->nesqp_context->rcv_nxt), + le32_to_cpu(nesqp->nesqp_context->snd_nxt), + buff_len); /* notify OF layer that accept event was successful */ cm_id->add_ref(cm_id); @@ -2982,12 +3219,12 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) nesqp->private_data_len; /* copy entire MPA frame to our cm_node's frame */ memcpy(cm_node->loopbackpartner->mpa_frame_buf, - nesqp->ietf_frame->priv_data, nesqp->private_data_len); + conn_param->private_data, conn_param->private_data_len); create_event(cm_node->loopbackpartner, NES_CM_EVENT_CONNECTED); } if (ret) printk(KERN_ERR "%s[%u] OFA CM event_handler returned, " - "ret=%d\n", __func__, __LINE__, ret); + "ret=%d\n", __func__, __LINE__, ret); return 0; } @@ -3000,34 +3237,28 @@ int nes_reject(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len) { struct nes_cm_node *cm_node; struct nes_cm_node *loopback; - struct nes_cm_core *cm_core; + u8 *start_buff; atomic_inc(&cm_rejects); - cm_node = (struct nes_cm_node *) cm_id->provider_data; + cm_node = (struct nes_cm_node *)cm_id->provider_data; loopback = cm_node->loopbackpartner; cm_core = cm_node->cm_core; cm_node->cm_id = cm_id; - cm_node->mpa_frame_size = sizeof(struct ietf_mpa_frame) + pdata_len; - if (cm_node->mpa_frame_size > MAX_CM_BUFFER) + if (pdata_len + sizeof(struct ietf_mpa_v2) > MAX_CM_BUFFER) return -EINVAL; - memcpy(&cm_node->mpa_frame.key[0], IEFT_MPA_KEY_REP, IETF_MPA_KEY_SIZE); if (loopback) { memcpy(&loopback->mpa_frame.priv_data, pdata, pdata_len); loopback->mpa_frame.priv_data_len = pdata_len; - loopback->mpa_frame_size = sizeof(struct ietf_mpa_frame) + - pdata_len; + loopback->mpa_frame_size = pdata_len; } else { - memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len); - cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len); + start_buff = &cm_node->mpa_frame_buf[0] + sizeof(struct ietf_mpa_v2); + cm_node->mpa_frame_size = pdata_len; + memcpy(start_buff, pdata, pdata_len); } - - cm_node->mpa_frame.rev = mpa_version; - cm_node->mpa_frame.flags = IETF_MPA_FLAGS_CRC | IETF_MPA_FLAGS_REJECT; - - return cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node); + return cm_core->api->reject(cm_core, cm_node); } @@ -3054,7 +3285,7 @@ int nes_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) nesvnic = to_nesvnic(nesqp->ibqp.device); if (!nesvnic) return -EINVAL; - nesdev = nesvnic->nesdev; + nesdev = nesvnic->nesdev; if (!nesdev) return -EINVAL; @@ -3062,12 +3293,12 @@ int nes_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) return -EINVAL; nes_debug(NES_DBG_CM, "QP%u, current IP = 0x%08X, Destination IP = " - "0x%08X:0x%04X, local = 0x%08X:0x%04X.\n", nesqp->hwqp.qp_id, - ntohl(nesvnic->local_ipaddr), - ntohl(cm_id->remote_addr.sin_addr.s_addr), - ntohs(cm_id->remote_addr.sin_port), - ntohl(cm_id->local_addr.sin_addr.s_addr), - ntohs(cm_id->local_addr.sin_port)); + "0x%08X:0x%04X, local = 0x%08X:0x%04X.\n", nesqp->hwqp.qp_id, + ntohl(nesvnic->local_ipaddr), + ntohl(cm_id->remote_addr.sin_addr.s_addr), + ntohs(cm_id->remote_addr.sin_port), + ntohl(cm_id->local_addr.sin_addr.s_addr), + ntohs(cm_id->local_addr.sin_port)); atomic_inc(&cm_connects); nesqp->active_conn = 1; @@ -3081,12 +3312,12 @@ int nes_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) nesqp->nesqp_context->ird_ord_sizes |= cpu_to_le32((u32)conn_param->ord); nes_debug(NES_DBG_CM, "requested ord = 0x%08X.\n", (u32)conn_param->ord); nes_debug(NES_DBG_CM, "mpa private data len =%u\n", - conn_param->private_data_len); + conn_param->private_data_len); if (cm_id->local_addr.sin_addr.s_addr != - cm_id->remote_addr.sin_addr.s_addr) { + cm_id->remote_addr.sin_addr.s_addr) { nes_manage_apbvt(nesvnic, ntohs(cm_id->local_addr.sin_port), - PCI_FUNC(nesdev->pcidev->devfn), NES_MANAGE_APBVT_ADD); + PCI_FUNC(nesdev->pcidev->devfn), NES_MANAGE_APBVT_ADD); apbvt_set = 1; } @@ -3102,13 +3333,13 @@ int nes_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) /* create a connect CM node connection */ cm_node = g_cm_core->api->connect(g_cm_core, nesvnic, - conn_param->private_data_len, (void *)conn_param->private_data, - &cm_info); + conn_param->private_data_len, (void *)conn_param->private_data, + &cm_info); if (!cm_node) { if (apbvt_set) nes_manage_apbvt(nesvnic, ntohs(cm_id->local_addr.sin_port), - PCI_FUNC(nesdev->pcidev->devfn), - NES_MANAGE_APBVT_DEL); + PCI_FUNC(nesdev->pcidev->devfn), + NES_MANAGE_APBVT_DEL); cm_id->rem_ref(cm_id); return -ENOMEM; @@ -3158,7 +3389,7 @@ int nes_create_listen(struct iw_cm_id *cm_id, int backlog) cm_node = g_cm_core->api->listen(g_cm_core, nesvnic, &cm_info); if (!cm_node) { printk(KERN_ERR "%s[%u] Error returned from listen API call\n", - __func__, __LINE__); + __func__, __LINE__); return -ENOMEM; } @@ -3166,12 +3397,12 @@ int nes_create_listen(struct iw_cm_id *cm_id, int backlog) if (!cm_node->reused_node) { err = nes_manage_apbvt(nesvnic, - ntohs(cm_id->local_addr.sin_port), - PCI_FUNC(nesvnic->nesdev->pcidev->devfn), - NES_MANAGE_APBVT_ADD); + ntohs(cm_id->local_addr.sin_port), + PCI_FUNC(nesvnic->nesdev->pcidev->devfn), + NES_MANAGE_APBVT_ADD); if (err) { printk(KERN_ERR "nes_manage_apbvt call returned %d.\n", - err); + err); g_cm_core->api->stop_listener(g_cm_core, (void *)cm_node); return err; } @@ -3208,13 +3439,13 @@ int nes_destroy_listen(struct iw_cm_id *cm_id) int nes_cm_recv(struct sk_buff *skb, struct net_device *netdevice) { int rc = 0; + cm_packets_received++; - if ((g_cm_core) && (g_cm_core->api)) { + if ((g_cm_core) && (g_cm_core->api)) rc = g_cm_core->api->recv_pkt(g_cm_core, netdev_priv(netdevice), skb); - } else { + else nes_debug(NES_DBG_CM, "Unable to process packet for CM," - " cm is not setup properly.\n"); - } + " cm is not setup properly.\n"); return rc; } @@ -3229,11 +3460,10 @@ int nes_cm_start(void) nes_debug(NES_DBG_CM, "\n"); /* create the primary CM core, pass this handle to subsequent core inits */ g_cm_core = nes_cm_alloc_core(); - if (g_cm_core) { + if (g_cm_core) return 0; - } else { + else return -ENOMEM; - } } @@ -3254,7 +3484,6 @@ int nes_cm_stop(void) */ static void cm_event_connected(struct nes_cm_event *event) { - u64 u64temp; struct nes_qp *nesqp; struct nes_vnic *nesvnic; struct nes_device *nesdev; @@ -3263,7 +3492,6 @@ static void cm_event_connected(struct nes_cm_event *event) struct ib_qp_attr attr; struct iw_cm_id *cm_id; struct iw_cm_event cm_event; - struct nes_hw_qp_wqe *wqe; struct nes_v4_quad nes_quad; u32 crc_value; int ret; @@ -3277,17 +3505,16 @@ static void cm_event_connected(struct nes_cm_event *event) nesdev = nesvnic->nesdev; nesadapter = nesdev->nesadapter; - if (nesqp->destroyed) { + if (nesqp->destroyed) return; - } atomic_inc(&cm_connecteds); nes_debug(NES_DBG_CM, "QP%u attempting to connect to 0x%08X:0x%04X on" - " local port 0x%04X. jiffies = %lu.\n", - nesqp->hwqp.qp_id, - ntohl(cm_id->remote_addr.sin_addr.s_addr), - ntohs(cm_id->remote_addr.sin_port), - ntohs(cm_id->local_addr.sin_port), - jiffies); + " local port 0x%04X. jiffies = %lu.\n", + nesqp->hwqp.qp_id, + ntohl(cm_id->remote_addr.sin_addr.s_addr), + ntohs(cm_id->remote_addr.sin_port), + ntohs(cm_id->local_addr.sin_port), + jiffies); nes_cm_init_tsa_conn(nesqp, cm_node); @@ -3318,40 +3545,12 @@ static void cm_event_connected(struct nes_cm_event *event) NES_QPCONTEXT_ORDIRD_IWARP_MODE_SHIFT); /* Adjust tail for not having a LSMM */ - nesqp->hwqp.sq_tail = 1; + /*nesqp->hwqp.sq_tail = 1;*/ -#if defined(NES_SEND_FIRST_WRITE) - if (cm_node->send_write0) { - nes_debug(NES_DBG_CM, "Sending first write.\n"); - wqe = &nesqp->hwqp.sq_vbase[0]; - u64temp = (unsigned long)nesqp; - u64temp |= NES_SW_CONTEXT_ALIGN>>1; - set_wqe_64bit_value(wqe->wqe_words, - NES_IWARP_SQ_WQE_COMP_CTX_LOW_IDX, u64temp); - wqe->wqe_words[NES_IWARP_SQ_WQE_MISC_IDX] = - cpu_to_le32(NES_IWARP_SQ_OP_RDMAW); - wqe->wqe_words[NES_IWARP_SQ_WQE_TOTAL_PAYLOAD_IDX] = 0; - wqe->wqe_words[NES_IWARP_SQ_WQE_FRAG0_LOW_IDX] = 0; - wqe->wqe_words[NES_IWARP_SQ_WQE_FRAG0_HIGH_IDX] = 0; - wqe->wqe_words[NES_IWARP_SQ_WQE_LENGTH0_IDX] = 0; - wqe->wqe_words[NES_IWARP_SQ_WQE_STAG0_IDX] = 0; + build_rdma0_msg(cm_node, &nesqp); - if (nesqp->sq_kmapped) { - nesqp->sq_kmapped = 0; - kunmap(nesqp->page); - } - - /* use the reserved spot on the WQ for the extra first WQE */ - nesqp->nesqp_context->ird_ord_sizes &= - cpu_to_le32(~(NES_QPCONTEXT_ORDIRD_LSMM_PRESENT | - NES_QPCONTEXT_ORDIRD_WRPDU | - NES_QPCONTEXT_ORDIRD_ALSMM)); - nesqp->skip_lsmm = 1; - nesqp->hwqp.sq_tail = 0; - nes_write32(nesdev->regs + NES_WQE_ALLOC, - (1 << 24) | 0x00800000 | nesqp->hwqp.qp_id); - } -#endif + nes_write32(nesdev->regs + NES_WQE_ALLOC, + (1 << 24) | 0x00800000 | nesqp->hwqp.qp_id); memset(&nes_quad, 0, sizeof(nes_quad)); @@ -3368,13 +3567,13 @@ static void cm_event_connected(struct nes_cm_event *event) crc_value = get_crc_value(&nes_quad); nesqp->hte_index = cpu_to_be32(crc_value ^ 0xffffffff); nes_debug(NES_DBG_CM, "HTE Index = 0x%08X, After CRC = 0x%08X\n", - nesqp->hte_index, nesqp->hte_index & nesadapter->hte_index_mask); + nesqp->hte_index, nesqp->hte_index & nesadapter->hte_index_mask); nesqp->hte_index &= nesadapter->hte_index_mask; nesqp->nesqp_context->hte_index = cpu_to_le32(nesqp->hte_index); nesqp->ietf_frame = &cm_node->mpa_frame; - nesqp->private_data_len = (u8) cm_node->mpa_frame_size; + nesqp->private_data_len = (u8)cm_node->mpa_frame_size; cm_node->cm_core->api->accelerated(cm_node->cm_core, cm_node); /* notify OF layer we successfully created the requested connection */ @@ -3386,7 +3585,9 @@ static void cm_event_connected(struct nes_cm_event *event) cm_event.remote_addr = cm_id->remote_addr; cm_event.private_data = (void *)event->cm_node->mpa_frame_buf; - cm_event.private_data_len = (u8) event->cm_node->mpa_frame_size; + cm_event.private_data_len = (u8)event->cm_node->mpa_frame_size; + cm_event.ird = cm_node->ird_size; + cm_event.ord = cm_node->ord_size; cm_event.local_addr.sin_addr.s_addr = event->cm_info.rem_addr; ret = cm_id->event_handler(cm_id, &cm_event); @@ -3394,12 +3595,12 @@ static void cm_event_connected(struct nes_cm_event *event) if (ret) printk(KERN_ERR "%s[%u] OFA CM event_handler returned, " - "ret=%d\n", __func__, __LINE__, ret); + "ret=%d\n", __func__, __LINE__, ret); attr.qp_state = IB_QPS_RTS; nes_modify_qp(&nesqp->ibqp, &attr, IB_QP_STATE, NULL); nes_debug(NES_DBG_CM, "Exiting connect thread for QP%u. jiffies = " - "%lu\n", nesqp->hwqp.qp_id, jiffies); + "%lu\n", nesqp->hwqp.qp_id, jiffies); return; } @@ -3420,16 +3621,14 @@ static void cm_event_connect_error(struct nes_cm_event *event) return; cm_id = event->cm_node->cm_id; - if (!cm_id) { + if (!cm_id) return; - } nes_debug(NES_DBG_CM, "cm_node=%p, cm_id=%p\n", event->cm_node, cm_id); nesqp = cm_id->provider_data; - if (!nesqp) { + if (!nesqp) return; - } /* notify OF layer about this connection error event */ /* cm_id->rem_ref(cm_id); */ @@ -3444,14 +3643,14 @@ static void cm_event_connect_error(struct nes_cm_event *event) cm_event.private_data_len = 0; nes_debug(NES_DBG_CM, "call CM_EVENT REJECTED, local_addr=%08x, " - "remove_addr=%08x\n", cm_event.local_addr.sin_addr.s_addr, - cm_event.remote_addr.sin_addr.s_addr); + "remove_addr=%08x\n", cm_event.local_addr.sin_addr.s_addr, + cm_event.remote_addr.sin_addr.s_addr); ret = cm_id->event_handler(cm_id, &cm_event); nes_debug(NES_DBG_CM, "OFA CM event_handler returned, ret=%d\n", ret); if (ret) printk(KERN_ERR "%s[%u] OFA CM event_handler returned, " - "ret=%d\n", __func__, __LINE__, ret); + "ret=%d\n", __func__, __LINE__, ret); cm_id->rem_ref(cm_id); rem_ref_cm_node(event->cm_node->cm_core, event->cm_node); @@ -3521,7 +3720,7 @@ static void cm_event_reset(struct nes_cm_event *event) */ static void cm_event_mpa_req(struct nes_cm_event *event) { - struct iw_cm_id *cm_id; + struct iw_cm_id *cm_id; struct iw_cm_event cm_event; int ret; struct nes_cm_node *cm_node; @@ -3533,7 +3732,7 @@ static void cm_event_mpa_req(struct nes_cm_event *event) atomic_inc(&cm_connect_reqs); nes_debug(NES_DBG_CM, "cm_node = %p - cm_id = %p, jiffies = %lu\n", - cm_node, cm_id, jiffies); + cm_node, cm_id, jiffies); cm_event.event = IW_CM_EVENT_CONNECT_REQUEST; cm_event.status = 0; @@ -3547,19 +3746,21 @@ static void cm_event_mpa_req(struct nes_cm_event *event) cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port); cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr); cm_event.private_data = cm_node->mpa_frame_buf; - cm_event.private_data_len = (u8) cm_node->mpa_frame_size; + cm_event.private_data_len = (u8)cm_node->mpa_frame_size; + cm_event.ird = cm_node->ird_size; + cm_event.ord = cm_node->ord_size; ret = cm_id->event_handler(cm_id, &cm_event); if (ret) printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n", - __func__, __LINE__, ret); + __func__, __LINE__, ret); return; } static void cm_event_mpa_reject(struct nes_cm_event *event) { - struct iw_cm_id *cm_id; + struct iw_cm_id *cm_id; struct iw_cm_event cm_event; struct nes_cm_node *cm_node; int ret; @@ -3571,7 +3772,7 @@ static void cm_event_mpa_reject(struct nes_cm_event *event) atomic_inc(&cm_connect_reqs); nes_debug(NES_DBG_CM, "cm_node = %p - cm_id = %p, jiffies = %lu\n", - cm_node, cm_id, jiffies); + cm_node, cm_id, jiffies); cm_event.event = IW_CM_EVENT_CONNECT_REPLY; cm_event.status = -ECONNREFUSED; @@ -3586,17 +3787,17 @@ static void cm_event_mpa_reject(struct nes_cm_event *event) cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr); cm_event.private_data = cm_node->mpa_frame_buf; - cm_event.private_data_len = (u8) cm_node->mpa_frame_size; + cm_event.private_data_len = (u8)cm_node->mpa_frame_size; nes_debug(NES_DBG_CM, "call CM_EVENT_MPA_REJECTED, local_addr=%08x, " - "remove_addr=%08x\n", - cm_event.local_addr.sin_addr.s_addr, - cm_event.remote_addr.sin_addr.s_addr); + "remove_addr=%08x\n", + cm_event.local_addr.sin_addr.s_addr, + cm_event.remote_addr.sin_addr.s_addr); ret = cm_id->event_handler(cm_id, &cm_event); if (ret) printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n", - __func__, __LINE__, ret); + __func__, __LINE__, ret); return; } @@ -3615,7 +3816,7 @@ static int nes_cm_post_event(struct nes_cm_event *event) event->cm_info.cm_id->add_ref(event->cm_info.cm_id); INIT_WORK(&event->event_work, nes_cm_event_handler); nes_debug(NES_DBG_CM, "cm_node=%p queue_work, event=%p\n", - event->cm_node, event); + event->cm_node, event); queue_work(event->cm_node->cm_core->event_wq, &event->event_work); @@ -3632,7 +3833,7 @@ static int nes_cm_post_event(struct nes_cm_event *event) static void nes_cm_event_handler(struct work_struct *work) { struct nes_cm_event *event = container_of(work, struct nes_cm_event, - event_work); + event_work); struct nes_cm_core *cm_core; if ((!event) || (!event->cm_node) || (!event->cm_node->cm_core)) @@ -3640,29 +3841,29 @@ static void nes_cm_event_handler(struct work_struct *work) cm_core = event->cm_node->cm_core; nes_debug(NES_DBG_CM, "event=%p, event->type=%u, events posted=%u\n", - event, event->type, atomic_read(&cm_core->events_posted)); + event, event->type, atomic_read(&cm_core->events_posted)); switch (event->type) { case NES_CM_EVENT_MPA_REQ: cm_event_mpa_req(event); nes_debug(NES_DBG_CM, "cm_node=%p CM Event: MPA REQUEST\n", - event->cm_node); + event->cm_node); break; case NES_CM_EVENT_RESET: nes_debug(NES_DBG_CM, "cm_node = %p CM Event: RESET\n", - event->cm_node); + event->cm_node); cm_event_reset(event); break; case NES_CM_EVENT_CONNECTED: if ((!event->cm_node->cm_id) || - (event->cm_node->state != NES_CM_STATE_TSA)) + (event->cm_node->state != NES_CM_STATE_TSA)) break; cm_event_connected(event); nes_debug(NES_DBG_CM, "CM Event: CONNECTED\n"); break; case NES_CM_EVENT_MPA_REJECT: if ((!event->cm_node->cm_id) || - (event->cm_node->state == NES_CM_STATE_TSA)) + (event->cm_node->state == NES_CM_STATE_TSA)) break; cm_event_mpa_reject(event); nes_debug(NES_DBG_CM, "CM Event: REJECT\n"); @@ -3670,7 +3871,7 @@ static void nes_cm_event_handler(struct work_struct *work) case NES_CM_EVENT_ABORTED: if ((!event->cm_node->cm_id) || - (event->cm_node->state == NES_CM_STATE_TSA)) + (event->cm_node->state == NES_CM_STATE_TSA)) break; cm_event_connect_error(event); nes_debug(NES_DBG_CM, "CM Event: ABORTED\n"); diff --git a/drivers/infiniband/hw/nes/nes_cm.h b/drivers/infiniband/hw/nes/nes_cm.h index d9825fd..bdfa1fb 100644 --- a/drivers/infiniband/hw/nes/nes_cm.h +++ b/drivers/infiniband/hw/nes/nes_cm.h @@ -48,7 +48,16 @@ #define IETF_MPA_KEY_SIZE 16 #define IETF_MPA_VERSION 1 #define IETF_MAX_PRIV_DATA_LEN 512 -#define IETF_MPA_FRAME_SIZE 20 +#define IETF_MPA_FRAME_SIZE 20 +#define IETF_RTR_MSG_SIZE 4 +#define IETF_MPA_V2_FLAG 0x10 + +/* IETF RTR MSG Fields */ +#define IETF_PEER_TO_PEER 0x8000 +#define IETF_FLPDU_ZERO_LEN 0x4000 +#define IETF_RDMA0_WRITE 0x8000 +#define IETF_RDMA0_READ 0x4000 +#define IETF_NO_IRD_ORD 0x3FFF enum ietf_mpa_flags { IETF_MPA_FLAGS_MARKERS = 0x80, /* receive Markers */ @@ -56,7 +65,7 @@ enum ietf_mpa_flags { IETF_MPA_FLAGS_REJECT = 0x20, /* Reject */ }; -struct ietf_mpa_frame { +struct ietf_mpa_v1 { u8 key[IETF_MPA_KEY_SIZE]; u8 flags; u8 rev; @@ -66,6 +75,20 @@ struct ietf_mpa_frame { #define ietf_mpa_req_resp_frame ietf_mpa_frame +struct ietf_rtr_msg { + __be16 ctrl_ird; + __be16 ctrl_ord; +}; + +struct ietf_mpa_v2 { + u8 key[IETF_MPA_KEY_SIZE]; + u8 flags; + u8 rev; + __be16 priv_data_len; + struct ietf_rtr_msg rtr_msg; + u8 priv_data[0]; +}; + struct nes_v4_quad { u32 rsvd0; __le32 DstIpAdrIndex; /* Only most significant 5 bits are valid */ @@ -171,8 +194,7 @@ struct nes_timer_entry { #define NES_CM_DEF_SEQ2 0x18ed5740 #define NES_CM_DEF_LOCAL_ID2 0xb807 -#define MAX_CM_BUFFER (IETF_MPA_FRAME_SIZE + IETF_MAX_PRIV_DATA_LEN) - +#define MAX_CM_BUFFER (IETF_MPA_FRAME_SIZE + IETF_RTR_MSG_SIZE + IETF_MAX_PRIV_DATA_LEN) typedef u32 nes_addr_t; @@ -204,6 +226,21 @@ enum nes_cm_node_state { NES_CM_STATE_CLOSED }; +enum mpa_frame_version { + IETF_MPA_V1 = 1, + IETF_MPA_V2 = 2 +}; + +enum mpa_frame_key { + MPA_KEY_REQUEST, + MPA_KEY_REPLY +}; + +enum send_rdma0 { + SEND_RDMA_READ_ZERO = 1, + SEND_RDMA_WRITE_ZERO = 2 +}; + enum nes_tcpip_pkt_type { NES_PKT_TYPE_UNKNOWN, NES_PKT_TYPE_SYN, @@ -245,9 +282,9 @@ struct nes_cm_tcp_context { enum nes_cm_listener_state { - NES_CM_LISTENER_PASSIVE_STATE=1, - NES_CM_LISTENER_ACTIVE_STATE=2, - NES_CM_LISTENER_EITHER_STATE=3 + NES_CM_LISTENER_PASSIVE_STATE = 1, + NES_CM_LISTENER_ACTIVE_STATE = 2, + NES_CM_LISTENER_EITHER_STATE = 3 }; struct nes_cm_listener { @@ -283,16 +320,20 @@ struct nes_cm_node { struct nes_cm_node *loopbackpartner; - struct nes_timer_entry *send_entry; - + struct nes_timer_entry *send_entry; + struct nes_timer_entry *recv_entry; spinlock_t retrans_list_lock; - struct nes_timer_entry *recv_entry; + enum send_rdma0 send_rdma0_op; - int send_write0; union { - struct ietf_mpa_frame mpa_frame; - u8 mpa_frame_buf[MAX_CM_BUFFER]; + struct ietf_mpa_v1 mpa_frame; + struct ietf_mpa_v2 mpa_v2_frame; + u8 mpa_frame_buf[MAX_CM_BUFFER]; }; + enum mpa_frame_version mpa_frame_rev; + u16 ird_size; + u16 ord_size; + u16 mpa_frame_size; struct iw_cm_id *cm_id; struct list_head list; @@ -399,10 +440,8 @@ struct nes_cm_ops { struct nes_vnic *, u16, void *, struct nes_cm_info *); int (*close)(struct nes_cm_core *, struct nes_cm_node *); - int (*accept)(struct nes_cm_core *, struct ietf_mpa_frame *, - struct nes_cm_node *); - int (*reject)(struct nes_cm_core *, struct ietf_mpa_frame *, - struct nes_cm_node *); + int (*accept)(struct nes_cm_core *, struct nes_cm_node *); + int (*reject)(struct nes_cm_core *, struct nes_cm_node *); int (*recv_pkt)(struct nes_cm_core *, struct nes_vnic *, struct sk_buff *); int (*destroy_cm_core)(struct nes_cm_core *); @@ -422,5 +461,7 @@ int nes_destroy_listen(struct iw_cm_id *); int nes_cm_recv(struct sk_buff *, struct net_device *); int nes_cm_start(void); int nes_cm_stop(void); +int nes_add_ref_cm_node(struct nes_cm_node *cm_node); +int nes_rem_ref_cm_node(struct nes_cm_node *cm_node); #endif /* NES_CM_H */ diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c index ba4814a..4cd1bf7 100644 --- a/drivers/infiniband/hw/nes/nes_hw.c +++ b/drivers/infiniband/hw/nes/nes_hw.c @@ -109,6 +109,14 @@ static unsigned char *nes_tcp_state_str[] = { }; #endif +static inline void print_ip(struct nes_cm_node *cm_node) +{ + unsigned char *rem_addr; + if (cm_node) { + rem_addr = (unsigned char *)&cm_node->rem_addr; + printk(KERN_ERR PFX "Remote IP addr: %pI4\n", rem_addr); + } +} /** * nes_nic_init_timer_defaults @@ -1554,6 +1562,7 @@ static void nes_replenish_nic_rq(struct nes_vnic *nesvnic) struct nes_hw_nic_rq_wqe *nic_rqe; struct nes_hw_nic *nesnic; struct nes_device *nesdev; + struct nes_rskb_cb *cb; u32 rx_wqes_posted = 0; nesnic = &nesvnic->nic; @@ -1579,6 +1588,9 @@ static void nes_replenish_nic_rq(struct nes_vnic *nesvnic) bus_address = pci_map_single(nesdev->pcidev, skb->data, nesvnic->max_frame_size, PCI_DMA_FROMDEVICE); + cb = (struct nes_rskb_cb *)&skb->cb[0]; + cb->busaddr = bus_address; + cb->maplen = nesvnic->max_frame_size; nic_rqe = &nesnic->rq_vbase[nesvnic->nic.rq_head]; nic_rqe->wqe_words[NES_NIC_RQ_WQE_LENGTH_1_0_IDX] = @@ -1668,6 +1680,7 @@ int nes_init_nic_qp(struct nes_device *nesdev, struct net_device *netdev) u32 cqp_head; u32 counter; u32 wqe_count; + struct nes_rskb_cb *cb; u8 jumbomode=0; /* Allocate fragment, SQ, RQ, and CQ; Reuse CEQ based on the PCI function */ @@ -1844,6 +1857,9 @@ int nes_init_nic_qp(struct nes_device *nesdev, struct net_device *netdev) pmem = pci_map_single(nesdev->pcidev, skb->data, nesvnic->max_frame_size, PCI_DMA_FROMDEVICE); + cb = (struct nes_rskb_cb *)&skb->cb[0]; + cb->busaddr = pmem; + cb->maplen = nesvnic->max_frame_size; nic_rqe = &nesvnic->nic.rq_vbase[counter]; nic_rqe->wqe_words[NES_NIC_RQ_WQE_LENGTH_1_0_IDX] = cpu_to_le32(nesvnic->max_frame_size); @@ -1872,6 +1888,13 @@ int nes_init_nic_qp(struct nes_device *nesdev, struct net_device *netdev) jumbomode = 1; nes_nic_init_timer_defaults(nesdev, jumbomode); } + if ((nesdev->nesadapter->allow_unaligned_fpdus) && + (nes_init_mgt_qp(nesdev, netdev, nesvnic))) { + nes_debug(NES_DBG_INIT, "%s: Out of memory for pau nic\n", netdev->name); + nes_destroy_nic_qp(nesvnic); + return -ENOMEM; + } + nesvnic->lro_mgr.max_aggr = nes_lro_max_aggr; nesvnic->lro_mgr.max_desc = NES_MAX_LRO_DESCRIPTORS; nesvnic->lro_mgr.lro_arr = nesvnic->lro_desc; @@ -1894,28 +1917,29 @@ void nes_destroy_nic_qp(struct nes_vnic *nesvnic) struct nes_device *nesdev = nesvnic->nesdev; struct nes_hw_cqp_wqe *cqp_wqe; struct nes_hw_nic_sq_wqe *nic_sqe; - struct nes_hw_nic_rq_wqe *nic_rqe; __le16 *wqe_fragment_length; u16 wqe_fragment_index; - u64 wqe_frag; u32 cqp_head; u32 wqm_cfg0; unsigned long flags; + struct sk_buff *rx_skb; + struct nes_rskb_cb *cb; int ret; + if (nesdev->nesadapter->allow_unaligned_fpdus) + nes_destroy_mgt(nesvnic); + /* clear wqe stall before destroying NIC QP */ wqm_cfg0 = nes_read_indexed(nesdev, NES_IDX_WQM_CONFIG0); nes_write_indexed(nesdev, NES_IDX_WQM_CONFIG0, wqm_cfg0 & 0xFFFF7FFF); /* Free remaining NIC receive buffers */ while (nesvnic->nic.rq_head != nesvnic->nic.rq_tail) { - nic_rqe = &nesvnic->nic.rq_vbase[nesvnic->nic.rq_tail]; - wqe_frag = (u64)le32_to_cpu( - nic_rqe->wqe_words[NES_NIC_RQ_WQE_FRAG0_LOW_IDX]); - wqe_frag |= ((u64)le32_to_cpu( - nic_rqe->wqe_words[NES_NIC_RQ_WQE_FRAG0_HIGH_IDX]))<<32; - pci_unmap_single(nesdev->pcidev, (dma_addr_t)wqe_frag, - nesvnic->max_frame_size, PCI_DMA_FROMDEVICE); + rx_skb = nesvnic->nic.rx_skb[nesvnic->nic.rq_tail]; + cb = (struct nes_rskb_cb *)&rx_skb->cb[0]; + pci_unmap_single(nesdev->pcidev, cb->busaddr, cb->maplen, + PCI_DMA_FROMDEVICE); + dev_kfree_skb(nesvnic->nic.rx_skb[nesvnic->nic.rq_tail++]); nesvnic->nic.rq_tail &= (nesvnic->nic.rq_size - 1); } @@ -2774,6 +2798,7 @@ void nes_nic_ce_handler(struct nes_device *nesdev, struct nes_hw_nic_cq *cq) struct nes_hw_nic_sq_wqe *nic_sqe; struct sk_buff *skb; struct sk_buff *rx_skb; + struct nes_rskb_cb *cb; __le16 *wqe_fragment_length; u32 head; u32 cq_size; @@ -2858,6 +2883,8 @@ void nes_nic_ce_handler(struct nes_device *nesdev, struct nes_hw_nic_cq *cq) bus_address += ((u64)le32_to_cpu(nic_rqe->wqe_words[NES_NIC_RQ_WQE_FRAG0_HIGH_IDX])) << 32; pci_unmap_single(nesdev->pcidev, bus_address, nesvnic->max_frame_size, PCI_DMA_FROMDEVICE); + cb = (struct nes_rskb_cb *)&rx_skb->cb[0]; + cb->busaddr = 0; /* rx_skb->tail = rx_skb->data + rx_pkt_size; */ /* rx_skb->len = rx_pkt_size; */ rx_skb->len = 0; /* TODO: see if this is necessary */ @@ -2916,24 +2943,19 @@ void nes_nic_ce_handler(struct nes_device *nesdev, struct nes_hw_nic_cq *cq) goto skip_rx_indicate0; - if ((cqe_misc & NES_NIC_CQE_TAG_VALID) && - (nesvnic->vlan_grp != NULL)) { + if (cqe_misc & NES_NIC_CQE_TAG_VALID) { vlan_tag = (u16)(le32_to_cpu( cq->cq_vbase[head].cqe_words[NES_NIC_CQE_TAG_PKT_TYPE_IDX]) >> 16); nes_debug(NES_DBG_CQ, "%s: Reporting stripped VLAN packet. Tag = 0x%04X\n", nesvnic->netdev->name, vlan_tag); - if (nes_use_lro) - lro_vlan_hwaccel_receive_skb(&nesvnic->lro_mgr, rx_skb, - nesvnic->vlan_grp, vlan_tag, NULL); - else - nes_vlan_rx(rx_skb, nesvnic->vlan_grp, vlan_tag); - } else { - if (nes_use_lro) - lro_receive_skb(&nesvnic->lro_mgr, rx_skb, NULL); - else - nes_netif_rx(rx_skb); + + __vlan_hwaccel_put_tag(rx_skb, vlan_tag); } + if (nes_use_lro) + lro_receive_skb(&nesvnic->lro_mgr, rx_skb, NULL); + else + netif_receive_skb(rx_skb); skip_rx_indicate0: ; @@ -2987,6 +3009,7 @@ skip_rx_indicate0: } + /** * nes_cqp_ce_handler */ @@ -3001,6 +3024,8 @@ static void nes_cqp_ce_handler(struct nes_device *nesdev, struct nes_hw_cq *cq) u32 cq_size; u32 cqe_count=0; u32 error_code; + u32 opcode; + u32 ctx_index; /* u32 counter; */ head = cq->cq_head; @@ -3011,12 +3036,9 @@ static void nes_cqp_ce_handler(struct nes_device *nesdev, struct nes_hw_cq *cq) /* nes_debug(NES_DBG_CQP, "head=%u cqe_words=%08X\n", head, le32_to_cpu(cq->cq_vbase[head].cqe_words[NES_CQE_OPCODE_IDX])); */ - if (le32_to_cpu(cq->cq_vbase[head].cqe_words[NES_CQE_OPCODE_IDX]) & NES_CQE_VALID) { - u64temp = (((u64)(le32_to_cpu(cq->cq_vbase[head]. - cqe_words[NES_CQE_COMP_COMP_CTX_HIGH_IDX]))) << 32) | - ((u64)(le32_to_cpu(cq->cq_vbase[head]. - cqe_words[NES_CQE_COMP_COMP_CTX_LOW_IDX]))); - cqp = *((struct nes_hw_cqp **)&u64temp); + opcode = le32_to_cpu(cq->cq_vbase[head].cqe_words[NES_CQE_OPCODE_IDX]); + if (opcode & NES_CQE_VALID) { + cqp = &nesdev->cqp; error_code = le32_to_cpu(cq->cq_vbase[head].cqe_words[NES_CQE_ERROR_CODE_IDX]); if (error_code) { @@ -3025,15 +3047,14 @@ static void nes_cqp_ce_handler(struct nes_device *nesdev, struct nes_hw_cq *cq) le32_to_cpu(cq->cq_vbase[head].cqe_words[NES_CQE_OPCODE_IDX])&0x3f, (u16)(error_code >> 16), (u16)error_code); - nes_debug(NES_DBG_CQP, "cqp: qp_id=%u, sq_head=%u, sq_tail=%u\n", - cqp->qp_id, cqp->sq_head, cqp->sq_tail); } - u64temp = (((u64)(le32_to_cpu(nesdev->cqp.sq_vbase[cqp->sq_tail]. - wqe_words[NES_CQP_WQE_COMP_SCRATCH_HIGH_IDX]))) << 32) | - ((u64)(le32_to_cpu(nesdev->cqp.sq_vbase[cqp->sq_tail]. - wqe_words[NES_CQP_WQE_COMP_SCRATCH_LOW_IDX]))); - cqp_request = *((struct nes_cqp_request **)&u64temp); + u64temp = (((u64)(le32_to_cpu(cq->cq_vbase[head]. + cqe_words[NES_CQE_COMP_COMP_CTX_HIGH_IDX]))) << 32) | + ((u64)(le32_to_cpu(cq->cq_vbase[head]. + cqe_words[NES_CQE_COMP_COMP_CTX_LOW_IDX]))); + + cqp_request = (struct nes_cqp_request *)(unsigned long)u64temp; if (cqp_request) { if (cqp_request->waiting) { /* nes_debug(NES_DBG_CQP, "%s: Waking up requestor\n"); */ @@ -3079,9 +3100,15 @@ static void nes_cqp_ce_handler(struct nes_device *nesdev, struct nes_hw_cq *cq) cqp_wqe = &nesdev->cqp.sq_vbase[head]; memcpy(cqp_wqe, &cqp_request->cqp_wqe, sizeof(*cqp_wqe)); barrier(); - cqp_wqe->wqe_words[NES_CQP_WQE_COMP_SCRATCH_LOW_IDX] = + + opcode = cqp_wqe->wqe_words[NES_CQP_WQE_OPCODE_IDX]; + if ((opcode & NES_CQP_OPCODE_MASK) == NES_CQP_DOWNLOAD_SEGMENT) + ctx_index = NES_CQP_WQE_DL_COMP_CTX_LOW_IDX; + else + ctx_index = NES_CQP_WQE_COMP_CTX_LOW_IDX; + cqp_wqe->wqe_words[ctx_index] = cpu_to_le32((u32)((unsigned long)cqp_request)); - cqp_wqe->wqe_words[NES_CQP_WQE_COMP_SCRATCH_HIGH_IDX] = + cqp_wqe->wqe_words[ctx_index + 1] = cpu_to_le32((u32)(upper_32_bits((unsigned long)cqp_request))); nes_debug(NES_DBG_CQP, "CQP request %p (opcode 0x%02X) put on CQPs SQ wqe%u.\n", cqp_request, le32_to_cpu(cqp_wqe->wqe_words[NES_CQP_WQE_OPCODE_IDX])&0x3f, head); @@ -3097,7 +3124,6 @@ static void nes_cqp_ce_handler(struct nes_device *nesdev, struct nes_hw_cq *cq) nes_read32(nesdev->regs+NES_CQE_ALLOC); } - static u8 *locate_mpa(u8 *pkt, u32 aeq_info) { if (aeq_info & NES_AEQE_Q2_DATA_ETHERNET) { @@ -3553,9 +3579,9 @@ static void nes_process_iwarp_aeqe(struct nes_device *nesdev, aeqe_cq_id = le32_to_cpu(aeqe->aeqe_words[NES_AEQE_COMP_QP_CQ_ID_IDX]); if (aeq_info & NES_AEQE_QP) { - if ((!nes_is_resource_allocated(nesadapter, nesadapter->allocated_qps, - aeqe_cq_id)) || - (atomic_read(&nesqp->close_timer_started))) + if (!nes_is_resource_allocated(nesadapter, + nesadapter->allocated_qps, + aeqe_cq_id)) return; } @@ -3566,8 +3592,7 @@ static void nes_process_iwarp_aeqe(struct nes_device *nesdev, if (atomic_inc_return(&nesqp->close_timer_started) == 1) { if ((tcp_state == NES_AEQE_TCP_STATE_CLOSE_WAIT) && - (nesqp->ibqp_state == IB_QPS_RTS) && - ((nesadapter->eeprom_version >> 16) != NES_A0)) { + (nesqp->ibqp_state == IB_QPS_RTS)) { spin_lock_irqsave(&nesqp->lock, flags); nesqp->hw_iwarp_state = iwarp_state; nesqp->hw_tcp_state = tcp_state; @@ -3594,9 +3619,10 @@ static void nes_process_iwarp_aeqe(struct nes_device *nesdev, return; } spin_lock_irqsave(&nesqp->lock, flags); - nesqp->hw_iwarp_state = NES_AEQE_IWARP_STATE_CLOSING; + nesqp->hw_iwarp_state = iwarp_state; + nesqp->hw_tcp_state = tcp_state; + nesqp->last_aeq = async_event_id; spin_unlock_irqrestore(&nesqp->lock, flags); - nes_hw_modify_qp(nesdev, nesqp, NES_CQP_QP_IWARP_STATE_CLOSING, 0, 0); nes_cm_disconn(nesqp); break; @@ -3694,7 +3720,9 @@ static void nes_process_iwarp_aeqe(struct nes_device *nesdev, case NES_AEQE_AEID_ROE_INVALID_RDMA_WRITE_OR_READ_RESP: printk(KERN_ERR PFX "QP[%u] async_event_id=0x%04X IB_EVENT_QP_FATAL\n", nesqp->hwqp.qp_id, async_event_id); - nes_terminate_connection(nesdev, nesqp, aeqe, IB_EVENT_QP_FATAL); + print_ip(nesqp->cm_node); + if (!atomic_read(&nesqp->close_timer_started)) + nes_terminate_connection(nesdev, nesqp, aeqe, IB_EVENT_QP_FATAL); break; case NES_AEQE_AEID_CQ_OPERATION_ERROR: diff --git a/drivers/infiniband/hw/nes/nes_hw.h b/drivers/infiniband/hw/nes/nes_hw.h index 9159411..0b590e1 100644 --- a/drivers/infiniband/hw/nes/nes_hw.h +++ b/drivers/infiniband/hw/nes/nes_hw.h @@ -47,6 +47,11 @@ #define NES_MULTICAST_PF_MAX 8 #define NES_A0 3 +#define NES_ENABLE_PAU 0x07000001 +#define NES_DISABLE_PAU 0x07000000 +#define NES_PAU_COUNTER 10 +#define NES_CQP_OPCODE_MASK 0x3f + enum pci_regs { NES_INT_STAT = 0x0000, NES_INT_MASK = 0x0004, @@ -73,8 +78,10 @@ enum indexed_regs { NES_IDX_QP_CONTROL = 0x0040, NES_IDX_FLM_CONTROL = 0x0080, NES_IDX_INT_CPU_STATUS = 0x00a0, + NES_IDX_GPR_TRIGGER = 0x00bc, NES_IDX_GPIO_CONTROL = 0x00f0, NES_IDX_GPIO_DATA = 0x00f4, + NES_IDX_GPR2 = 0x010c, NES_IDX_TCP_CONFIG0 = 0x01e4, NES_IDX_TCP_TIMER_CONFIG = 0x01ec, NES_IDX_TCP_NOW = 0x01f0, @@ -202,6 +209,7 @@ enum nes_cqp_opcodes { NES_CQP_REGISTER_SHARED_STAG = 0x0c, NES_CQP_DEALLOCATE_STAG = 0x0d, NES_CQP_MANAGE_ARP_CACHE = 0x0f, + NES_CQP_DOWNLOAD_SEGMENT = 0x10, NES_CQP_SUSPEND_QPS = 0x11, NES_CQP_UPLOAD_CONTEXT = 0x13, NES_CQP_CREATE_CEQ = 0x16, @@ -210,7 +218,8 @@ enum nes_cqp_opcodes { NES_CQP_DESTROY_AEQ = 0x1b, NES_CQP_LMI_ACCESS = 0x20, NES_CQP_FLUSH_WQES = 0x22, - NES_CQP_MANAGE_APBVT = 0x23 + NES_CQP_MANAGE_APBVT = 0x23, + NES_CQP_MANAGE_QUAD_HASH = 0x25 }; enum nes_cqp_wqe_word_idx { @@ -222,6 +231,14 @@ enum nes_cqp_wqe_word_idx { NES_CQP_WQE_COMP_SCRATCH_HIGH_IDX = 5, }; +enum nes_cqp_wqe_word_download_idx { /* format differs from other cqp ops */ + NES_CQP_WQE_DL_OPCODE_IDX = 0, + NES_CQP_WQE_DL_COMP_CTX_LOW_IDX = 1, + NES_CQP_WQE_DL_COMP_CTX_HIGH_IDX = 2, + NES_CQP_WQE_DL_LENGTH_0_TOTAL_IDX = 3 + /* For index values 4-15 use NES_NIC_SQ_WQE_ values */ +}; + enum nes_cqp_cq_wqeword_idx { NES_CQP_CQ_WQE_PBL_LOW_IDX = 6, NES_CQP_CQ_WQE_PBL_HIGH_IDX = 7, @@ -242,6 +259,7 @@ enum nes_cqp_stag_wqeword_idx { NES_CQP_STAG_WQE_PBL_LEN_IDX = 14 }; +#define NES_CQP_OP_LOGICAL_PORT_SHIFT 26 #define NES_CQP_OP_IWARP_STATE_SHIFT 28 #define NES_CQP_OP_TERMLEN_SHIFT 28 @@ -599,6 +617,7 @@ enum nes_nic_sq_wqe_bits { enum nes_nic_cqe_word_idx { NES_NIC_CQE_ACCQP_ID_IDX = 0, + NES_NIC_CQE_HASH_RCVNXT = 1, NES_NIC_CQE_TAG_PKT_TYPE_IDX = 2, NES_NIC_CQE_MISC_IDX = 3, }; @@ -1005,6 +1024,11 @@ struct nes_arp_entry { #define NES_NIC_CQ_DOWNWARD_TREND 16 #define NES_PFT_SIZE 48 +#define NES_MGT_WQ_COUNT 32 +#define NES_MGT_CTX_SIZE ((NES_NIC_CTX_RQ_SIZE_32) | (NES_NIC_CTX_SQ_SIZE_32)) +#define NES_MGT_QP_OFFSET 36 +#define NES_MGT_QP_COUNT 4 + struct nes_hw_tune_timer { /* u16 cq_count; */ u16 threshold_low; @@ -1118,6 +1142,7 @@ struct nes_adapter { u32 et_rate_sample_interval; u32 timer_int_limit; u32 wqm_quanta; + u8 allow_unaligned_fpdus; /* Adapter base MAC address */ u32 mac_addr_low; @@ -1211,7 +1236,6 @@ struct nes_vnic { /* void *mem; */ struct nes_device *nesdev; struct net_device *netdev; - struct vlan_group *vlan_grp; atomic_t rx_skbs_needed; atomic_t rx_skb_timer_running; int budget; @@ -1252,6 +1276,14 @@ struct nes_vnic { enum ib_event_type delayed_event; enum ib_event_type last_dispatched_event; spinlock_t port_ibevent_lock; + u32 mgt_mem_size; + void *mgt_vbase; + dma_addr_t mgt_pbase; + struct nes_vnic_mgt *mgtvnic[NES_MGT_QP_COUNT]; + struct task_struct *mgt_thread; + wait_queue_head_t mgt_wait_queue; + struct sk_buff_head mgt_skb_list; + }; struct nes_ib_device { @@ -1357,7 +1389,4 @@ struct nes_terminate_hdr { #define NES_LINK_RECHECK_DELAY msecs_to_jiffies(50) #define NES_LINK_RECHECK_MAX 60 -#define nes_vlan_rx vlan_hwaccel_receive_skb -#define nes_netif_rx netif_receive_skb - #endif /* __NES_HW_H */ diff --git a/drivers/infiniband/hw/nes/nes_nic.c b/drivers/infiniband/hw/nes/nes_nic.c index d3a1c41..c00d2f3 100644 --- a/drivers/infiniband/hw/nes/nes_nic.c +++ b/drivers/infiniband/hw/nes/nes_nic.c @@ -441,13 +441,13 @@ static int nes_nic_send(struct sk_buff *skb, struct net_device *netdev) nesnic->tx_skb[nesnic->sq_head] = skb; for (skb_fragment_index = 0; skb_fragment_index < skb_shinfo(skb)->nr_frags; skb_fragment_index++) { - bus_address = pci_map_page( nesdev->pcidev, - skb_shinfo(skb)->frags[skb_fragment_index].page, - skb_shinfo(skb)->frags[skb_fragment_index].page_offset, - skb_shinfo(skb)->frags[skb_fragment_index].size, - PCI_DMA_TODEVICE); + skb_frag_t *frag = + &skb_shinfo(skb)->frags[skb_fragment_index]; + bus_address = skb_frag_dma_map(&nesdev->pcidev->dev, + frag, 0, skb_frag_size(frag), + DMA_TO_DEVICE); wqe_fragment_length[wqe_fragment_index] = - cpu_to_le16(skb_shinfo(skb)->frags[skb_fragment_index].size); + cpu_to_le16(skb_frag_size(&skb_shinfo(skb)->frags[skb_fragment_index])); set_wqe_64bit_value(nic_sqe->wqe_words, NES_NIC_SQ_WQE_FRAG0_LOW_IDX+(2*wqe_fragment_index), bus_address); wqe_fragment_index++; @@ -561,11 +561,12 @@ tso_sq_no_longer_full: /* Map all the buffers */ for (tso_frag_count=0; tso_frag_count < skb_shinfo(skb)->nr_frags; tso_frag_count++) { - tso_bus_address[tso_frag_count] = pci_map_page( nesdev->pcidev, - skb_shinfo(skb)->frags[tso_frag_count].page, - skb_shinfo(skb)->frags[tso_frag_count].page_offset, - skb_shinfo(skb)->frags[tso_frag_count].size, - PCI_DMA_TODEVICE); + skb_frag_t *frag = + &skb_shinfo(skb)->frags[tso_frag_count]; + tso_bus_address[tso_frag_count] = + skb_frag_dma_map(&nesdev->pcidev->dev, + frag, 0, skb_frag_size(frag), + DMA_TO_DEVICE); } tso_frag_index = 0; @@ -636,11 +637,11 @@ tso_sq_no_longer_full: } while (wqe_fragment_index < 5) { wqe_fragment_length[wqe_fragment_index] = - cpu_to_le16(skb_shinfo(skb)->frags[tso_frag_index].size); + cpu_to_le16(skb_frag_size(&skb_shinfo(skb)->frags[tso_frag_index])); set_wqe_64bit_value(nic_sqe->wqe_words, NES_NIC_SQ_WQE_FRAG0_LOW_IDX+(2*wqe_fragment_index), (u64)tso_bus_address[tso_frag_index]); wqe_fragment_index++; - tso_wqe_length += skb_shinfo(skb)->frags[tso_frag_index++].size; + tso_wqe_length += skb_frag_size(&skb_shinfo(skb)->frags[tso_frag_index++]); if (wqe_fragment_index < 5) wqe_fragment_length[wqe_fragment_index] = 0; if (tso_frag_index == tso_frag_count) @@ -1090,6 +1091,8 @@ static const char nes_ethtool_stringset[][ETH_GSTRING_LEN] = { "LRO aggregated", "LRO flushed", "LRO no_desc", + "PAU CreateQPs", + "PAU DestroyQPs", }; #define NES_ETHTOOL_STAT_COUNT ARRAY_SIZE(nes_ethtool_stringset) @@ -1305,6 +1308,8 @@ static void nes_netdev_get_ethtool_stats(struct net_device *netdev, target_stat_values[++index] = nesvnic->lro_mgr.stats.aggregated; target_stat_values[++index] = nesvnic->lro_mgr.stats.flushed; target_stat_values[++index] = nesvnic->lro_mgr.stats.no_desc; + target_stat_values[++index] = atomic_read(&pau_qps_created); + target_stat_values[++index] = atomic_read(&pau_qps_destroyed); } /** @@ -1584,23 +1589,19 @@ static const struct ethtool_ops nes_ethtool_ops = { .set_pauseparam = nes_netdev_set_pauseparam, }; - -static void nes_netdev_vlan_rx_register(struct net_device *netdev, struct vlan_group *grp) +static void nes_vlan_mode(struct net_device *netdev, struct nes_device *nesdev, u32 features) { - struct nes_vnic *nesvnic = netdev_priv(netdev); - struct nes_device *nesdev = nesvnic->nesdev; struct nes_adapter *nesadapter = nesdev->nesadapter; u32 u32temp; unsigned long flags; spin_lock_irqsave(&nesadapter->phy_lock, flags); - nesvnic->vlan_grp = grp; nes_debug(NES_DBG_NETDEV, "%s: %s\n", __func__, netdev->name); /* Enable/Disable VLAN Stripping */ u32temp = nes_read_indexed(nesdev, NES_IDX_PCIX_DIAG); - if (grp) + if (features & NETIF_F_HW_VLAN_RX) u32temp &= 0xfdffffff; else u32temp |= 0x02000000; @@ -1609,17 +1610,44 @@ static void nes_netdev_vlan_rx_register(struct net_device *netdev, struct vlan_g spin_unlock_irqrestore(&nesadapter->phy_lock, flags); } +static u32 nes_fix_features(struct net_device *netdev, u32 features) +{ + /* + * Since there is no support for separate rx/tx vlan accel + * enable/disable make sure tx flag is always in same state as rx. + */ + if (features & NETIF_F_HW_VLAN_RX) + features |= NETIF_F_HW_VLAN_TX; + else + features &= ~NETIF_F_HW_VLAN_TX; + + return features; +} + +static int nes_set_features(struct net_device *netdev, u32 features) +{ + struct nes_vnic *nesvnic = netdev_priv(netdev); + struct nes_device *nesdev = nesvnic->nesdev; + u32 changed = netdev->features ^ features; + + if (changed & NETIF_F_HW_VLAN_RX) + nes_vlan_mode(netdev, nesdev, features); + + return 0; +} + static const struct net_device_ops nes_netdev_ops = { - .ndo_open = nes_netdev_open, + .ndo_open = nes_netdev_open, .ndo_stop = nes_netdev_stop, - .ndo_start_xmit = nes_netdev_start_xmit, + .ndo_start_xmit = nes_netdev_start_xmit, .ndo_get_stats = nes_netdev_get_stats, - .ndo_tx_timeout = nes_netdev_tx_timeout, + .ndo_tx_timeout = nes_netdev_tx_timeout, .ndo_set_mac_address = nes_netdev_set_mac_address, - .ndo_set_multicast_list = nes_netdev_set_multicast_list, + .ndo_set_rx_mode = nes_netdev_set_multicast_list, .ndo_change_mtu = nes_netdev_change_mtu, .ndo_validate_addr = eth_validate_addr, - .ndo_vlan_rx_register = nes_netdev_vlan_rx_register, + .ndo_fix_features = nes_fix_features, + .ndo_set_features = nes_set_features, }; /** @@ -1656,7 +1684,7 @@ struct net_device *nes_netdev_init(struct nes_device *nesdev, netdev->ethtool_ops = &nes_ethtool_ops; netif_napi_add(netdev, &nesvnic->napi, nes_netdev_poll, 128); nes_debug(NES_DBG_INIT, "Enabling VLAN Insert/Delete.\n"); - netdev->features |= NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX; + netdev->features |= NETIF_F_HW_VLAN_TX; /* Fill in the port structure */ nesvnic->netdev = netdev; @@ -1683,7 +1711,8 @@ struct net_device *nes_netdev_init(struct nes_device *nesdev, netdev->dev_addr[5] = (u8)u64temp; memcpy(netdev->perm_addr, netdev->dev_addr, 6); - netdev->hw_features = NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_IP_CSUM; + netdev->hw_features = NETIF_F_RXCSUM | NETIF_F_SG | NETIF_F_IP_CSUM | + NETIF_F_HW_VLAN_RX; if ((nesvnic->logical_port < 2) || (nesdev->nesadapter->hw_rev != NE020_REV)) netdev->hw_features |= NETIF_F_TSO; netdev->features |= netdev->hw_features; @@ -1815,6 +1844,8 @@ struct net_device *nes_netdev_init(struct nes_device *nesdev, nes_init_phy(nesdev); } + nes_vlan_mode(netdev, nesdev, netdev->features); + return netdev; } diff --git a/drivers/infiniband/hw/nes/nes_utils.c b/drivers/infiniband/hw/nes/nes_utils.c index f9c417c..cd10968 100644 --- a/drivers/infiniband/hw/nes/nes_utils.c +++ b/drivers/infiniband/hw/nes/nes_utils.c @@ -51,13 +51,34 @@ #include "nes.h" - - static u16 nes_read16_eeprom(void __iomem *addr, u16 offset); u32 mh_detected; u32 mh_pauses_sent; +u32 nes_set_pau(struct nes_device *nesdev) +{ + u32 ret = 0; + u32 counter; + + nes_write_indexed(nesdev, NES_IDX_GPR2, NES_ENABLE_PAU); + nes_write_indexed(nesdev, NES_IDX_GPR_TRIGGER, 1); + + for (counter = 0; counter < NES_PAU_COUNTER; counter++) { + udelay(30); + if (!nes_read_indexed(nesdev, NES_IDX_GPR2)) { + printk(KERN_INFO PFX "PAU is supported.\n"); + break; + } + nes_write_indexed(nesdev, NES_IDX_GPR_TRIGGER, 1); + } + if (counter == NES_PAU_COUNTER) { + printk(KERN_INFO PFX "PAU is not supported.\n"); + return -EPERM; + } + return ret; +} + /** * nes_read_eeprom_values - */ @@ -187,6 +208,11 @@ int nes_read_eeprom_values(struct nes_device *nesdev, struct nes_adapter *nesada if (((major_ver == 3) && (minor_ver >= 16)) || (major_ver > 3)) nesadapter->send_term_ok = 1; + if (nes_drv_opt & NES_DRV_OPT_ENABLE_PAU) { + if (!nes_set_pau(nesdev)) + nesadapter->allow_unaligned_fpdus = 1; + } + nesadapter->firmware_version = (((u32)(u8)(eeprom_data>>8)) << 16) + (u32)((u8)eeprom_data); @@ -594,6 +620,7 @@ void nes_put_cqp_request(struct nes_device *nesdev, nes_free_cqp_request(nesdev, cqp_request); } + /** * nes_post_cqp_request */ @@ -604,6 +631,8 @@ void nes_post_cqp_request(struct nes_device *nesdev, unsigned long flags; u32 cqp_head; u64 u64temp; + u32 opcode; + int ctx_index = NES_CQP_WQE_COMP_CTX_LOW_IDX; spin_lock_irqsave(&nesdev->cqp.lock, flags); @@ -614,17 +643,20 @@ void nes_post_cqp_request(struct nes_device *nesdev, nesdev->cqp.sq_head &= nesdev->cqp.sq_size-1; cqp_wqe = &nesdev->cqp.sq_vbase[cqp_head]; memcpy(cqp_wqe, &cqp_request->cqp_wqe, sizeof(*cqp_wqe)); + opcode = le32_to_cpu(cqp_wqe->wqe_words[NES_CQP_WQE_OPCODE_IDX]); + if ((opcode & NES_CQP_OPCODE_MASK) == NES_CQP_DOWNLOAD_SEGMENT) + ctx_index = NES_CQP_WQE_DL_COMP_CTX_LOW_IDX; barrier(); u64temp = (unsigned long)cqp_request; - set_wqe_64bit_value(cqp_wqe->wqe_words, NES_CQP_WQE_COMP_SCRATCH_LOW_IDX, - u64temp); + set_wqe_64bit_value(cqp_wqe->wqe_words, ctx_index, u64temp); nes_debug(NES_DBG_CQP, "CQP request (opcode 0x%02X), line 1 = 0x%08X put on CQPs SQ," - " request = %p, cqp_head = %u, cqp_tail = %u, cqp_size = %u," - " waiting = %d, refcount = %d.\n", - le32_to_cpu(cqp_wqe->wqe_words[NES_CQP_WQE_OPCODE_IDX])&0x3f, - le32_to_cpu(cqp_wqe->wqe_words[NES_CQP_WQE_ID_IDX]), cqp_request, - nesdev->cqp.sq_head, nesdev->cqp.sq_tail, nesdev->cqp.sq_size, - cqp_request->waiting, atomic_read(&cqp_request->refcount)); + " request = %p, cqp_head = %u, cqp_tail = %u, cqp_size = %u," + " waiting = %d, refcount = %d.\n", + opcode & NES_CQP_OPCODE_MASK, + le32_to_cpu(cqp_wqe->wqe_words[NES_CQP_WQE_ID_IDX]), cqp_request, + nesdev->cqp.sq_head, nesdev->cqp.sq_tail, nesdev->cqp.sq_size, + cqp_request->waiting, atomic_read(&cqp_request->refcount)); + barrier(); /* Ring doorbell (1 WQEs) */ @@ -645,7 +677,6 @@ void nes_post_cqp_request(struct nes_device *nesdev, return; } - /** * nes_arp_table */ diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c index 59db49f..330eb6e 100644 --- a/drivers/infiniband/hw/nes/nes_verbs.c +++ b/drivers/infiniband/hw/nes/nes_verbs.c @@ -605,16 +605,6 @@ static int nes_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr /** - * nes_modify_port - */ -static int nes_modify_port(struct ib_device *ibdev, u8 port, - int port_modify_mask, struct ib_port_modify *props) -{ - return 0; -} - - -/** * nes_query_pkey */ static int nes_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey) @@ -1193,7 +1183,7 @@ static struct ib_qp *nes_create_qp(struct ib_pd *ibpd, nes_free_resource(nesadapter, nesadapter->allocated_qps, qp_num); kfree(nesqp->allocated_buffer); nes_debug(NES_DBG_QP, "ib_copy_from_udata() Failed \n"); - return NULL; + return ERR_PTR(-EFAULT); } if (req.user_wqe_buffers) { virt_wqs = 1; @@ -1470,7 +1460,7 @@ static int nes_destroy_qp(struct ib_qp *ibqp) struct ib_qp_attr attr; struct iw_cm_id *cm_id; struct iw_cm_event cm_event; - int ret; + int ret = 0; atomic_inc(&sw_qps_destroyed); nesqp->destroyed = 1; @@ -1523,7 +1513,6 @@ static int nes_destroy_qp(struct ib_qp *ibqp) if ((nesqp->nesrcq) && (nesqp->nesrcq != nesqp->nesscq)) nes_clean_cq(nesqp, nesqp->nesrcq); } - nes_rem_ref(&nesqp->ibqp); return 0; } @@ -2350,8 +2339,10 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, skip_pages = ((u32)region->offset) >> 12; - if (ib_copy_from_udata(&req, udata, sizeof(req))) + if (ib_copy_from_udata(&req, udata, sizeof(req))) { + ib_umem_release(region); return ERR_PTR(-EFAULT); + } nes_debug(NES_DBG_MR, "Memory Registration type = %08X.\n", req.reg_type); switch (req.reg_type) { @@ -2648,6 +2639,7 @@ static struct ib_mr *nes_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, return &nesmr->ibmr; } + ib_umem_release(region); return ERR_PTR(-ENOSYS); } @@ -3889,7 +3881,6 @@ struct nes_ib_device *nes_init_ofa_device(struct net_device *netdev) nesibdev->ibdev.dev.parent = &nesdev->pcidev->dev; nesibdev->ibdev.query_device = nes_query_device; nesibdev->ibdev.query_port = nes_query_port; - nesibdev->ibdev.modify_port = nes_modify_port; nesibdev->ibdev.query_pkey = nes_query_pkey; nesibdev->ibdev.query_gid = nes_query_gid; nesibdev->ibdev.alloc_ucontext = nes_alloc_ucontext; diff --git a/drivers/infiniband/hw/nes/nes_verbs.h b/drivers/infiniband/hw/nes/nes_verbs.h index 2df9993e..fe6b6e9 100644 --- a/drivers/infiniband/hw/nes/nes_verbs.h +++ b/drivers/infiniband/hw/nes/nes_verbs.h @@ -139,7 +139,8 @@ struct nes_qp { struct nes_cq *nesrcq; struct nes_pd *nespd; void *cm_node; /* handle of the node this QP is associated with */ - struct ietf_mpa_frame *ietf_frame; + void *ietf_frame; + u8 ietf_frame_size; dma_addr_t ietf_frame_pbase; struct ib_mr *lsmm_mr; struct nes_hw_qp hwqp; @@ -154,6 +155,7 @@ struct nes_qp { u32 mmap_sq_db_index; u32 mmap_rq_db_index; spinlock_t lock; + spinlock_t pau_lock; struct nes_qp_context *nesqp_context; dma_addr_t nesqp_context_pbase; void *pbl_vbase; @@ -161,6 +163,8 @@ struct nes_qp { struct page *page; struct timer_list terminate_timer; enum ib_event_type terminate_eventtype; + struct sk_buff_head pau_list; + u32 pau_rcv_nxt; u16 active_conn:1; u16 skip_lsmm:1; u16 user_mode:1; @@ -168,7 +172,8 @@ struct nes_qp { u16 flush_issued:1; u16 destroyed:1; u16 sig_all:1; - u16 rsvd:9; + u16 pau_mode:1; + u16 rsvd:8; u16 private_data_len; u16 term_sq_flush_code; u16 term_rq_flush_code; @@ -176,5 +181,8 @@ struct nes_qp { u8 hw_tcp_state; u8 term_flags; u8 sq_kmapped; + u8 pau_busy; + u8 pau_pending; + u8 pau_state; }; #endif /* NES_VERBS_H */ diff --git a/drivers/infiniband/hw/qib/qib.h b/drivers/infiniband/hw/qib/qib.h index 769a1d9..c0b72a6 100644 --- a/drivers/infiniband/hw/qib/qib.h +++ b/drivers/infiniband/hw/qib/qib.h @@ -171,7 +171,9 @@ struct qib_ctxtdata { /* how many alloc_pages() chunks in rcvegrbuf_pages */ u32 rcvegrbuf_chunks; /* how many egrbufs per chunk */ - u32 rcvegrbufs_perchunk; + u16 rcvegrbufs_perchunk; + /* ilog2 of above */ + u16 rcvegrbufs_perchunk_shift; /* order for rcvegrbuf_pages */ size_t rcvegrbuf_size; /* rcvhdrq size (for freeing) */ @@ -221,6 +223,9 @@ struct qib_ctxtdata { /* ctxt rcvhdrq head offset */ u32 head; u32 pkt_count; + /* lookaside fields */ + struct qib_qp *lookaside_qp; + u32 lookaside_qpn; /* QPs waiting for context processing */ struct list_head qp_wait_list; }; @@ -807,6 +812,10 @@ struct qib_devdata { * supports, less gives more pio bufs/ctxt, etc. */ u32 cfgctxts; + /* + * number of ctxts available for PSM open + */ + u32 freectxts; /* * hint that we should update pioavailshadow before @@ -936,7 +945,9 @@ struct qib_devdata { /* chip address space used by 4k pio buffers */ u32 align4k; /* size of each rcvegrbuffer */ - u32 rcvegrbufsize; + u16 rcvegrbufsize; + /* log2 of above */ + u16 rcvegrbufsize_shift; /* localbus width (1, 2,4,8,16,32) from config space */ u32 lbus_width; /* localbus speed in MHz */ @@ -994,12 +1005,6 @@ struct qib_devdata { /* control high-level access to EEPROM */ struct mutex eep_lock; uint64_t traffic_wds; - /* active time is kept in seconds, but logged in hours */ - atomic_t active_time; - /* Below are nominal shadow of EEPROM, new since last EEPROM update */ - uint8_t eep_st_errs[QIB_EEP_LOG_CNT]; - uint8_t eep_st_new_errs[QIB_EEP_LOG_CNT]; - uint16_t eep_hrs; /* * masks for which bits of errs, hwerrs that cause * each of the counters to increment. @@ -1012,6 +1017,8 @@ struct qib_devdata { u8 psxmitwait_supported; /* cycle length of PS* counters in HW (in picoseconds) */ u16 psxmitwait_check_rate; + /* high volume overflow errors defered to tasklet */ + struct tasklet_struct error_tasklet; }; /* hol_state values */ @@ -1214,8 +1221,7 @@ int qib_twsi_blk_rd(struct qib_devdata *dd, int dev, int addr, void *buffer, int qib_twsi_blk_wr(struct qib_devdata *dd, int dev, int addr, const void *buffer, int len); void qib_get_eeprom_info(struct qib_devdata *); -int qib_update_eeprom_log(struct qib_devdata *dd); -void qib_inc_eeprom_err(struct qib_devdata *dd, u32 eidx, u32 incr); +#define qib_inc_eeprom_err(dd, eidx, incr) void qib_dump_lookup_output_queue(struct qib_devdata *); void qib_force_pio_avail_update(struct qib_devdata *); void qib_clear_symerror_on_linkup(unsigned long opaque); @@ -1415,6 +1421,10 @@ extern struct mutex qib_mutex; qib_get_unit_name((dd)->unit), ##__VA_ARGS__); \ } while (0) +#define qib_dev_warn(dd, fmt, ...) \ + dev_warn(&(dd)->pcidev->dev, "%s: " fmt, \ + qib_get_unit_name((dd)->unit), ##__VA_ARGS__) + #define qib_dev_porterr(dd, port, fmt, ...) \ do { \ dev_err(&(dd)->pcidev->dev, "%s: IB%u:%u " fmt, \ @@ -1433,6 +1443,7 @@ extern struct mutex qib_mutex; struct qib_hwerror_msgs { u64 mask; const char *msg; + size_t sz; }; #define QLOGIC_IB_HWE_MSG(a, b) { .mask = a, .msg = b } diff --git a/drivers/infiniband/hw/qib/qib_diag.c b/drivers/infiniband/hw/qib/qib_diag.c index 204c4dd..9892456 100644 --- a/drivers/infiniband/hw/qib/qib_diag.c +++ b/drivers/infiniband/hw/qib/qib_diag.c @@ -46,6 +46,7 @@ #include <linux/pci.h> #include <linux/poll.h> #include <linux/vmalloc.h> +#include <linux/export.h> #include <linux/fs.h> #include <linux/uaccess.h> diff --git a/drivers/infiniband/hw/qib/qib_driver.c b/drivers/infiniband/hw/qib/qib_driver.c index 23e584f..c90a55f 100644 --- a/drivers/infiniband/hw/qib/qib_driver.c +++ b/drivers/infiniband/hw/qib/qib_driver.c @@ -37,6 +37,7 @@ #include <linux/delay.h> #include <linux/netdevice.h> #include <linux/vmalloc.h> +#include <linux/module.h> #include "qib.h" @@ -279,10 +280,10 @@ bail: */ static inline void *qib_get_egrbuf(const struct qib_ctxtdata *rcd, u32 etail) { - const u32 chunk = etail / rcd->rcvegrbufs_perchunk; - const u32 idx = etail % rcd->rcvegrbufs_perchunk; + const u32 chunk = etail >> rcd->rcvegrbufs_perchunk_shift; + const u32 idx = etail & ((u32)rcd->rcvegrbufs_perchunk - 1); - return rcd->rcvegrbuf[chunk] + idx * rcd->dd->rcvegrbufsize; + return rcd->rcvegrbuf[chunk] + (idx << rcd->dd->rcvegrbufsize_shift); } /* @@ -310,7 +311,6 @@ static u32 qib_rcv_hdrerr(struct qib_ctxtdata *rcd, struct qib_pportdata *ppd, u32 opcode; u32 psn; int diff; - unsigned long flags; /* Sanity check packet */ if (tlen < 24) @@ -365,7 +365,6 @@ static u32 qib_rcv_hdrerr(struct qib_ctxtdata *rcd, struct qib_pportdata *ppd, switch (qp->ibqp.qp_type) { case IB_QPT_RC: - spin_lock_irqsave(&qp->s_lock, flags); ruc_res = qib_ruc_check_hdr( ibp, hdr, @@ -373,11 +372,8 @@ static u32 qib_rcv_hdrerr(struct qib_ctxtdata *rcd, struct qib_pportdata *ppd, qp, be32_to_cpu(ohdr->bth[0])); if (ruc_res) { - spin_unlock_irqrestore(&qp->s_lock, - flags); goto unlock; } - spin_unlock_irqrestore(&qp->s_lock, flags); /* Only deal with RDMA Writes for now */ if (opcode < @@ -547,6 +543,15 @@ move_along: updegr = 0; } } + /* + * Notify qib_destroy_qp() if it is waiting + * for lookaside_qp to finish. + */ + if (rcd->lookaside_qp) { + if (atomic_dec_and_test(&rcd->lookaside_qp->refcount)) + wake_up(&rcd->lookaside_qp->wait); + rcd->lookaside_qp = NULL; + } rcd->head = l; rcd->pkt_count += i; diff --git a/drivers/infiniband/hw/qib/qib_eeprom.c b/drivers/infiniband/hw/qib/qib_eeprom.c index 92d9cfe..6359c2f 100644 --- a/drivers/infiniband/hw/qib/qib_eeprom.c +++ b/drivers/infiniband/hw/qib/qib_eeprom.c @@ -263,189 +263,9 @@ void qib_get_eeprom_info(struct qib_devdata *dd) qib_dev_err(dd, "Board SN %s did not pass functional " "test: %s\n", dd->serial, ifp->if_comment); - memcpy(&dd->eep_st_errs, &ifp->if_errcntp, QIB_EEP_LOG_CNT); - /* - * Power-on (actually "active") hours are kept as little-endian value - * in EEPROM, but as seconds in a (possibly as small as 24-bit) - * atomic_t while running. - */ - atomic_set(&dd->active_time, 0); - dd->eep_hrs = ifp->if_powerhour[0] | (ifp->if_powerhour[1] << 8); - done: vfree(buf); bail:; } -/** - * qib_update_eeprom_log - copy active-time and error counters to eeprom - * @dd: the qlogic_ib device - * - * Although the time is kept as seconds in the qib_devdata struct, it is - * rounded to hours for re-write, as we have only 16 bits in EEPROM. - * First-cut code reads whole (expected) struct qib_flash, modifies, - * re-writes. Future direction: read/write only what we need, assuming - * that the EEPROM had to have been "good enough" for driver init, and - * if not, we aren't making it worse. - * - */ -int qib_update_eeprom_log(struct qib_devdata *dd) -{ - void *buf; - struct qib_flash *ifp; - int len, hi_water; - uint32_t new_time, new_hrs; - u8 csum; - int ret, idx; - unsigned long flags; - - /* first, check if we actually need to do anything. */ - ret = 0; - for (idx = 0; idx < QIB_EEP_LOG_CNT; ++idx) { - if (dd->eep_st_new_errs[idx]) { - ret = 1; - break; - } - } - new_time = atomic_read(&dd->active_time); - - if (ret == 0 && new_time < 3600) - goto bail; - - /* - * The quick-check above determined that there is something worthy - * of logging, so get current contents and do a more detailed idea. - * read full flash, not just currently used part, since it may have - * been written with a newer definition - */ - len = sizeof(struct qib_flash); - buf = vmalloc(len); - ret = 1; - if (!buf) { - qib_dev_err(dd, "Couldn't allocate memory to read %u " - "bytes from eeprom for logging\n", len); - goto bail; - } - - /* Grab semaphore and read current EEPROM. If we get an - * error, let go, but if not, keep it until we finish write. - */ - ret = mutex_lock_interruptible(&dd->eep_lock); - if (ret) { - qib_dev_err(dd, "Unable to acquire EEPROM for logging\n"); - goto free_bail; - } - ret = qib_twsi_blk_rd(dd, dd->twsi_eeprom_dev, 0, buf, len); - if (ret) { - mutex_unlock(&dd->eep_lock); - qib_dev_err(dd, "Unable read EEPROM for logging\n"); - goto free_bail; - } - ifp = (struct qib_flash *)buf; - - csum = flash_csum(ifp, 0); - if (csum != ifp->if_csum) { - mutex_unlock(&dd->eep_lock); - qib_dev_err(dd, "EEPROM cks err (0x%02X, S/B 0x%02X)\n", - csum, ifp->if_csum); - ret = 1; - goto free_bail; - } - hi_water = 0; - spin_lock_irqsave(&dd->eep_st_lock, flags); - for (idx = 0; idx < QIB_EEP_LOG_CNT; ++idx) { - int new_val = dd->eep_st_new_errs[idx]; - if (new_val) { - /* - * If we have seen any errors, add to EEPROM values - * We need to saturate at 0xFF (255) and we also - * would need to adjust the checksum if we were - * trying to minimize EEPROM traffic - * Note that we add to actual current count in EEPROM, - * in case it was altered while we were running. - */ - new_val += ifp->if_errcntp[idx]; - if (new_val > 0xFF) - new_val = 0xFF; - if (ifp->if_errcntp[idx] != new_val) { - ifp->if_errcntp[idx] = new_val; - hi_water = offsetof(struct qib_flash, - if_errcntp) + idx; - } - /* - * update our shadow (used to minimize EEPROM - * traffic), to match what we are about to write. - */ - dd->eep_st_errs[idx] = new_val; - dd->eep_st_new_errs[idx] = 0; - } - } - /* - * Now update active-time. We would like to round to the nearest hour - * but unless atomic_t are sure to be proper signed ints we cannot, - * because we need to account for what we "transfer" to EEPROM and - * if we log an hour at 31 minutes, then we would need to set - * active_time to -29 to accurately count the _next_ hour. - */ - if (new_time >= 3600) { - new_hrs = new_time / 3600; - atomic_sub((new_hrs * 3600), &dd->active_time); - new_hrs += dd->eep_hrs; - if (new_hrs > 0xFFFF) - new_hrs = 0xFFFF; - dd->eep_hrs = new_hrs; - if ((new_hrs & 0xFF) != ifp->if_powerhour[0]) { - ifp->if_powerhour[0] = new_hrs & 0xFF; - hi_water = offsetof(struct qib_flash, if_powerhour); - } - if ((new_hrs >> 8) != ifp->if_powerhour[1]) { - ifp->if_powerhour[1] = new_hrs >> 8; - hi_water = offsetof(struct qib_flash, if_powerhour) + 1; - } - } - /* - * There is a tiny possibility that we could somehow fail to write - * the EEPROM after updating our shadows, but problems from holding - * the spinlock too long are a much bigger issue. - */ - spin_unlock_irqrestore(&dd->eep_st_lock, flags); - if (hi_water) { - /* we made some change to the data, uopdate cksum and write */ - csum = flash_csum(ifp, 1); - ret = eeprom_write_with_enable(dd, 0, buf, hi_water + 1); - } - mutex_unlock(&dd->eep_lock); - if (ret) - qib_dev_err(dd, "Failed updating EEPROM\n"); - -free_bail: - vfree(buf); -bail: - return ret; -} - -/** - * qib_inc_eeprom_err - increment one of the four error counters - * that are logged to EEPROM. - * @dd: the qlogic_ib device - * @eidx: 0..3, the counter to increment - * @incr: how much to add - * - * Each counter is 8-bits, and saturates at 255 (0xFF). They - * are copied to the EEPROM (aka flash) whenever qib_update_eeprom_log() - * is called, but it can only be called in a context that allows sleep. - * This function can be called even at interrupt level. - */ -void qib_inc_eeprom_err(struct qib_devdata *dd, u32 eidx, u32 incr) -{ - uint new_val; - unsigned long flags; - - spin_lock_irqsave(&dd->eep_st_lock, flags); - new_val = dd->eep_st_new_errs[eidx] + incr; - if (new_val > 255) - new_val = 255; - dd->eep_st_new_errs[eidx] = new_val; - spin_unlock_irqrestore(&dd->eep_st_lock, flags); -} diff --git a/drivers/infiniband/hw/qib/qib_file_ops.c b/drivers/infiniband/hw/qib/qib_file_ops.c index 406fca5..a740324 100644 --- a/drivers/infiniband/hw/qib/qib_file_ops.c +++ b/drivers/infiniband/hw/qib/qib_file_ops.c @@ -43,6 +43,7 @@ #include <linux/jiffies.h> #include <asm/pgtable.h> #include <linux/delay.h> +#include <linux/export.h> #include "qib.h" #include "qib_common.h" @@ -1284,6 +1285,7 @@ static int setup_ctxt(struct qib_pportdata *ppd, int ctxt, strlcpy(rcd->comm, current->comm, sizeof(rcd->comm)); ctxt_fp(fp) = rcd; qib_stats.sps_ctxts++; + dd->freectxts--; ret = 0; goto bail; @@ -1527,6 +1529,7 @@ done_chk_sdma: struct qib_filedata *fd = fp->private_data; const struct qib_ctxtdata *rcd = fd->rcd; const struct qib_devdata *dd = rcd->dd; + unsigned int weight; if (dd->flags & QIB_HAS_SEND_DMA) { fd->pq = qib_user_sdma_queue_create(&dd->pcidev->dev, @@ -1545,8 +1548,8 @@ done_chk_sdma: * it just means that sooner or later we don't recommend * a cpu, and let the scheduler do it's best. */ - if (!ret && cpus_weight(current->cpus_allowed) >= - qib_cpulist_count) { + weight = cpumask_weight(tsk_cpus_allowed(current)); + if (!ret && weight >= qib_cpulist_count) { int cpu; cpu = find_first_zero_bit(qib_cpulist, qib_cpulist_count); @@ -1554,13 +1557,13 @@ done_chk_sdma: __set_bit(cpu, qib_cpulist); fd->rec_cpu_num = cpu; } - } else if (cpus_weight(current->cpus_allowed) == 1 && - test_bit(first_cpu(current->cpus_allowed), + } else if (weight == 1 && + test_bit(cpumask_first(tsk_cpus_allowed(current)), qib_cpulist)) qib_devinfo(dd->pcidev, "%s PID %u affinity " "set to cpu %d; already allocated\n", current->comm, current->pid, - first_cpu(current->cpus_allowed)); + cpumask_first(tsk_cpus_allowed(current))); } mutex_unlock(&qib_mutex); @@ -1791,6 +1794,7 @@ static int qib_close(struct inode *in, struct file *fp) if (dd->pageshadow) unlock_expected_tids(rcd); qib_stats.sps_ctxts--; + dd->freectxts++; } mutex_unlock(&qib_mutex); @@ -1904,8 +1908,9 @@ int qib_set_uevent_bits(struct qib_pportdata *ppd, const int evtbit) struct qib_ctxtdata *rcd; unsigned ctxt; int ret = 0; + unsigned long flags; - spin_lock(&ppd->dd->uctxt_lock); + spin_lock_irqsave(&ppd->dd->uctxt_lock, flags); for (ctxt = ppd->dd->first_user_ctxt; ctxt < ppd->dd->cfgctxts; ctxt++) { rcd = ppd->dd->rcd[ctxt]; @@ -1924,7 +1929,7 @@ int qib_set_uevent_bits(struct qib_pportdata *ppd, const int evtbit) ret = 1; break; } - spin_unlock(&ppd->dd->uctxt_lock); + spin_unlock_irqrestore(&ppd->dd->uctxt_lock, flags); return ret; } diff --git a/drivers/infiniband/hw/qib/qib_iba6120.c b/drivers/infiniband/hw/qib/qib_iba6120.c index 65df26c..2a0f7de 100644 --- a/drivers/infiniband/hw/qib/qib_iba6120.c +++ b/drivers/infiniband/hw/qib/qib_iba6120.c @@ -2674,8 +2674,6 @@ static void qib_get_6120_faststats(unsigned long opaque) spin_lock_irqsave(&dd->eep_st_lock, flags); traffic_wds -= dd->traffic_wds; dd->traffic_wds += traffic_wds; - if (traffic_wds >= QIB_TRAFFIC_ACTIVE_THRESHOLD) - atomic_add(5, &dd->active_time); /* S/B #define */ spin_unlock_irqrestore(&dd->eep_st_lock, flags); qib_chk_6120_errormask(dd); @@ -3275,6 +3273,8 @@ static int init_6120_variables(struct qib_devdata *dd) /* we always allocate at least 2048 bytes for eager buffers */ ret = ib_mtu_enum_to_int(qib_ibmtu); dd->rcvegrbufsize = ret != -1 ? max(ret, 2048) : QIB_DEFAULT_MTU; + BUG_ON(!is_power_of_2(dd->rcvegrbufsize)); + dd->rcvegrbufsize_shift = ilog2(dd->rcvegrbufsize); qib_6120_tidtemplate(dd); diff --git a/drivers/infiniband/hw/qib/qib_iba7220.c b/drivers/infiniband/hw/qib/qib_iba7220.c index 759bb63..4dc04e3 100644 --- a/drivers/infiniband/hw/qib/qib_iba7220.c +++ b/drivers/infiniband/hw/qib/qib_iba7220.c @@ -39,6 +39,7 @@ #include <linux/interrupt.h> #include <linux/pci.h> #include <linux/delay.h> +#include <linux/module.h> #include <linux/io.h> #include <rdma/ib_verbs.h> @@ -2434,6 +2435,7 @@ static int qib_7220_set_ib_cfg(struct qib_pportdata *ppd, int which, u32 val) int lsb, ret = 0, setforce = 0; u16 lcmd, licmd; unsigned long flags; + u32 tmp = 0; switch (which) { case QIB_IB_CFG_LIDLMC: @@ -2467,9 +2469,6 @@ static int qib_7220_set_ib_cfg(struct qib_pportdata *ppd, int which, u32 val) maskr = IBA7220_IBC_WIDTH_MASK; lsb = IBA7220_IBC_WIDTH_SHIFT; setforce = 1; - spin_lock_irqsave(&ppd->lflags_lock, flags); - ppd->lflags |= QIBL_IB_FORCE_NOTIFY; - spin_unlock_irqrestore(&ppd->lflags_lock, flags); break; case QIB_IB_CFG_SPD_ENB: /* set allowed Link speeds */ @@ -2643,6 +2642,28 @@ static int qib_7220_set_ib_cfg(struct qib_pportdata *ppd, int which, u32 val) goto bail; } qib_set_ib_7220_lstate(ppd, lcmd, licmd); + + maskr = IBA7220_IBC_WIDTH_MASK; + lsb = IBA7220_IBC_WIDTH_SHIFT; + tmp = (ppd->cpspec->ibcddrctrl >> lsb) & maskr; + /* If the width active on the chip does not match the + * width in the shadow register, write the new active + * width to the chip. + * We don't have to worry about speed as the speed is taken + * care of by set_7220_ibspeed_fast called by ib_updown. + */ + if (ppd->link_width_enabled-1 != tmp) { + ppd->cpspec->ibcddrctrl &= ~(maskr << lsb); + ppd->cpspec->ibcddrctrl |= + (((u64)(ppd->link_width_enabled-1) & maskr) << + lsb); + qib_write_kreg(dd, kr_ibcddrctrl, + ppd->cpspec->ibcddrctrl); + qib_write_kreg(dd, kr_scratch, 0); + spin_lock_irqsave(&ppd->lflags_lock, flags); + ppd->lflags |= QIBL_IB_FORCE_NOTIFY; + spin_unlock_irqrestore(&ppd->lflags_lock, flags); + } goto bail; case QIB_IB_CFG_HRTBT: /* set Heartbeat off/enable/auto */ @@ -3271,8 +3292,6 @@ static void qib_get_7220_faststats(unsigned long opaque) spin_lock_irqsave(&dd->eep_st_lock, flags); traffic_wds -= dd->traffic_wds; dd->traffic_wds += traffic_wds; - if (traffic_wds >= QIB_TRAFFIC_ACTIVE_THRESHOLD) - atomic_add(5, &dd->active_time); /* S/B #define */ spin_unlock_irqrestore(&dd->eep_st_lock, flags); done: mod_timer(&dd->stats_timer, jiffies + HZ * ACTIVITY_TIMER); @@ -4067,6 +4086,8 @@ static int qib_init_7220_variables(struct qib_devdata *dd) /* we always allocate at least 2048 bytes for eager buffers */ ret = ib_mtu_enum_to_int(qib_ibmtu); dd->rcvegrbufsize = ret != -1 ? max(ret, 2048) : QIB_DEFAULT_MTU; + BUG_ON(!is_power_of_2(dd->rcvegrbufsize)); + dd->rcvegrbufsize_shift = ilog2(dd->rcvegrbufsize); qib_7220_tidtemplate(dd); diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c index 49e4a58..44180c6 100644 --- a/drivers/infiniband/hw/qib/qib_iba7322.c +++ b/drivers/infiniband/hw/qib/qib_iba7322.c @@ -40,6 +40,7 @@ #include <linux/delay.h> #include <linux/io.h> #include <linux/jiffies.h> +#include <linux/module.h> #include <rdma/ib_verbs.h> #include <rdma/ib_smi.h> @@ -114,6 +115,10 @@ static ushort qib_singleport; module_param_named(singleport, qib_singleport, ushort, S_IRUGO); MODULE_PARM_DESC(singleport, "Use only IB port 1; more per-port buffer space"); +static ushort qib_krcvq01_no_msi; +module_param_named(krcvq01_no_msi, qib_krcvq01_no_msi, ushort, S_IRUGO); +MODULE_PARM_DESC(krcvq01_no_msi, "No MSI for kctx < 2"); + /* * Receive header queue sizes */ @@ -397,7 +402,6 @@ MODULE_PARM_DESC(txselect, \ #define crp_txdroppedpkt CREG_IDX(TxDroppedPktCnt) #define crp_txhdrerr CREG_IDX(TxHeadersErrCnt) #define crp_txlenerr CREG_IDX(TxLenErrCnt) -#define crp_txlenerr CREG_IDX(TxLenErrCnt) #define crp_txminmaxlenerr CREG_IDX(TxMaxMinLenErrCnt) #define crp_txsdmadesc CREG_IDX(TxSDmaDescCnt) #define crp_txunderrun CREG_IDX(TxUnderrunCnt) @@ -1107,9 +1111,9 @@ static inline u32 read_7322_creg32_port(const struct qib_pportdata *ppd, #define AUTONEG_TRIES 3 /* sequential retries to negotiate DDR */ #define HWE_AUTO(fldname) { .mask = SYM_MASK(HwErrMask, fldname##Mask), \ - .msg = #fldname } + .msg = #fldname , .sz = sizeof(#fldname) } #define HWE_AUTO_P(fldname, port) { .mask = SYM_MASK(HwErrMask, \ - fldname##Mask##_##port), .msg = #fldname } + fldname##Mask##_##port), .msg = #fldname , .sz = sizeof(#fldname) } static const struct qib_hwerror_msgs qib_7322_hwerror_msgs[] = { HWE_AUTO_P(IBSerdesPClkNotDetect, 1), HWE_AUTO_P(IBSerdesPClkNotDetect, 0), @@ -1127,14 +1131,16 @@ static const struct qib_hwerror_msgs qib_7322_hwerror_msgs[] = { HWE_AUTO_P(IBCBusFromSPCParityErr, 0), HWE_AUTO(statusValidNoEop), HWE_AUTO(LATriggered), - { .mask = 0 } + { .mask = 0, .sz = 0 } }; #define E_AUTO(fldname) { .mask = SYM_MASK(ErrMask, fldname##Mask), \ - .msg = #fldname } + .msg = #fldname, .sz = sizeof(#fldname) } #define E_P_AUTO(fldname) { .mask = SYM_MASK(ErrMask_0, fldname##Mask), \ - .msg = #fldname } + .msg = #fldname, .sz = sizeof(#fldname) } static const struct qib_hwerror_msgs qib_7322error_msgs[] = { + E_AUTO(RcvEgrFullErr), + E_AUTO(RcvHdrFullErr), E_AUTO(ResetNegated), E_AUTO(HardwareErr), E_AUTO(InvalidAddrErr), @@ -1147,9 +1153,7 @@ static const struct qib_hwerror_msgs qib_7322error_msgs[] = { E_AUTO(SendSpecialTriggerErr), E_AUTO(SDmaWrongPortErr), E_AUTO(SDmaBufMaskDuplicateErr), - E_AUTO(RcvHdrFullErr), - E_AUTO(RcvEgrFullErr), - { .mask = 0 } + { .mask = 0, .sz = 0 } }; static const struct qib_hwerror_msgs qib_7322p_error_msgs[] = { @@ -1159,7 +1163,8 @@ static const struct qib_hwerror_msgs qib_7322p_error_msgs[] = { /* * SDmaHaltErr is not really an error, make it clearer; */ - {.mask = SYM_MASK(ErrMask_0, SDmaHaltErrMask), .msg = "SDmaHalted"}, + {.mask = SYM_MASK(ErrMask_0, SDmaHaltErrMask), .msg = "SDmaHalted", + .sz = 11}, E_P_AUTO(SDmaDescAddrMisalignErr), E_P_AUTO(SDmaUnexpDataErr), E_P_AUTO(SDmaMissingDwErr), @@ -1195,7 +1200,7 @@ static const struct qib_hwerror_msgs qib_7322p_error_msgs[] = { E_P_AUTO(RcvICRCErr), E_P_AUTO(RcvVCRCErr), E_P_AUTO(RcvFormatErr), - { .mask = 0 } + { .mask = 0, .sz = 0 } }; /* @@ -1203,17 +1208,17 @@ static const struct qib_hwerror_msgs qib_7322p_error_msgs[] = { * context */ #define INTR_AUTO(fldname) { .mask = SYM_MASK(IntMask, fldname##Mask), \ - .msg = #fldname } + .msg = #fldname, .sz = sizeof(#fldname) } /* Below generates "auto-message" for interrupts specific to a port */ #define INTR_AUTO_P(fldname) { .mask = MASK_ACROSS(\ SYM_LSB(IntMask, fldname##Mask##_0), \ SYM_LSB(IntMask, fldname##Mask##_1)), \ - .msg = #fldname "_P" } + .msg = #fldname "_P", .sz = sizeof(#fldname "_P") } /* For some reason, the SerDesTrimDone bits are reversed */ #define INTR_AUTO_PI(fldname) { .mask = MASK_ACROSS(\ SYM_LSB(IntMask, fldname##Mask##_1), \ SYM_LSB(IntMask, fldname##Mask##_0)), \ - .msg = #fldname "_P" } + .msg = #fldname "_P", .sz = sizeof(#fldname "_P") } /* * Below generates "auto-message" for interrupts specific to a context, * with ctxt-number appended @@ -1221,7 +1226,7 @@ static const struct qib_hwerror_msgs qib_7322p_error_msgs[] = { #define INTR_AUTO_C(fldname) { .mask = MASK_ACROSS(\ SYM_LSB(IntMask, fldname##0IntMask), \ SYM_LSB(IntMask, fldname##17IntMask)), \ - .msg = #fldname "_C"} + .msg = #fldname "_C", .sz = sizeof(#fldname "_C") } static const struct qib_hwerror_msgs qib_7322_intr_msgs[] = { INTR_AUTO_P(SDmaInt), @@ -1235,11 +1240,12 @@ static const struct qib_hwerror_msgs qib_7322_intr_msgs[] = { INTR_AUTO_P(SendDoneInt), INTR_AUTO(SendBufAvailInt), INTR_AUTO_C(RcvAvail), - { .mask = 0 } + { .mask = 0, .sz = 0 } }; #define TXSYMPTOM_AUTO_P(fldname) \ - { .mask = SYM_MASK(SendHdrErrSymptom_0, fldname), .msg = #fldname } + { .mask = SYM_MASK(SendHdrErrSymptom_0, fldname), \ + .msg = #fldname, .sz = sizeof(#fldname) } static const struct qib_hwerror_msgs hdrchk_msgs[] = { TXSYMPTOM_AUTO_P(NonKeyPacket), TXSYMPTOM_AUTO_P(GRHFail), @@ -1248,7 +1254,7 @@ static const struct qib_hwerror_msgs hdrchk_msgs[] = { TXSYMPTOM_AUTO_P(SLIDFail), TXSYMPTOM_AUTO_P(RawIPV6), TXSYMPTOM_AUTO_P(PacketTooSmall), - { .mask = 0 } + { .mask = 0, .sz = 0 } }; #define IBA7322_HDRHEAD_PKTINT_SHIFT 32 /* interrupt cnt in upper 32 bits */ @@ -1293,7 +1299,7 @@ static void err_decode(char *msg, size_t len, u64 errs, u64 these, lmask; int took, multi, n = 0; - while (msp && msp->mask) { + while (errs && msp && msp->mask) { multi = (msp->mask & (msp->mask - 1)); while (errs & msp->mask) { these = (errs & msp->mask); @@ -1304,9 +1310,14 @@ static void err_decode(char *msg, size_t len, u64 errs, *msg++ = ','; len--; } - took = scnprintf(msg, len, "%s", msp->msg); + BUG_ON(!msp->sz); + /* msp->sz counts the nul */ + took = min_t(size_t, msp->sz - (size_t)1, len); + memcpy(msg, msp->msg, took); len -= took; msg += took; + if (len) + *msg = '\0'; } errs &= ~lmask; if (len && multi) { @@ -1644,6 +1655,14 @@ done: return; } +static void qib_error_tasklet(unsigned long data) +{ + struct qib_devdata *dd = (struct qib_devdata *)data; + + handle_7322_errors(dd); + qib_write_kreg(dd, kr_errmask, dd->cspec->errormask); +} + static void reenable_chase(unsigned long opaque) { struct qib_pportdata *ppd = (struct qib_pportdata *)opaque; @@ -2260,6 +2279,11 @@ static int qib_7322_bringup_serdes(struct qib_pportdata *ppd) qib_write_kreg_port(ppd, krp_ibcctrl_a, ppd->cpspec->ibcctrl_a); qib_write_kreg(dd, kr_scratch, 0ULL); + /* ensure previous Tx parameters are not still forced */ + qib_write_kreg_port(ppd, krp_tx_deemph_override, + SYM_MASK(IBSD_TX_DEEMPHASIS_OVERRIDE_0, + reset_tx_deemphasis_override)); + if (qib_compat_ddr_negotiate) { ppd->cpspec->ibdeltainprog = 1; ppd->cpspec->ibsymsnap = read_7322_creg32_port(ppd, @@ -2288,16 +2312,11 @@ static int qib_7322_bringup_serdes(struct qib_pportdata *ppd) SYM_LSB(IBCCtrlA_0, MaxPktLen); ppd->cpspec->ibcctrl_a = ibc; /* without linkcmd or linkinitcmd! */ - /* initially come up waiting for TS1, without sending anything. */ - val = ppd->cpspec->ibcctrl_a | (QLOGIC_IB_IBCC_LINKINITCMD_DISABLE << - QLOGIC_IB_IBCC_LINKINITCMD_SHIFT); - /* * Reset the PCS interface to the serdes (and also ibc, which is still * in reset from above). Writes new value of ibcctrl_a as last step. */ qib_7322_mini_pcs_reset(ppd); - qib_write_kreg(dd, kr_scratch, 0ULL); if (!ppd->cpspec->ibcctrl_b) { unsigned lse = ppd->link_speed_enabled; @@ -2363,17 +2382,20 @@ static int qib_7322_bringup_serdes(struct qib_pportdata *ppd) ppd->cpspec->ibcctrl_a |= SYM_MASK(IBCCtrlA_0, IBLinkEn); set_vls(ppd); + /* initially come up DISABLED, without sending anything. */ + val = ppd->cpspec->ibcctrl_a | (QLOGIC_IB_IBCC_LINKINITCMD_DISABLE << + QLOGIC_IB_IBCC_LINKINITCMD_SHIFT); + qib_write_kreg_port(ppd, krp_ibcctrl_a, val); + qib_write_kreg(dd, kr_scratch, 0ULL); + /* clear the linkinit cmds */ + ppd->cpspec->ibcctrl_a = val & ~SYM_MASK(IBCCtrlA_0, LinkInitCmd); + /* be paranoid against later code motion, etc. */ spin_lock_irqsave(&dd->cspec->rcvmod_lock, flags); ppd->p_rcvctrl |= SYM_MASK(RcvCtrl_0, RcvIBPortEnable); qib_write_kreg_port(ppd, krp_rcvctrl, ppd->p_rcvctrl); spin_unlock_irqrestore(&dd->cspec->rcvmod_lock, flags); - /* Hold the link state machine for mezz boards */ - if (IS_QMH(dd) || IS_QME(dd)) - qib_set_ib_7322_lstate(ppd, 0, - QLOGIC_IB_IBCC_LINKINITCMD_DISABLE); - /* Also enable IBSTATUSCHG interrupt. */ val = qib_read_kreg_port(ppd, krp_errmask); qib_write_kreg_port(ppd, krp_errmask, @@ -2725,8 +2747,10 @@ static noinline void unlikely_7322_intr(struct qib_devdata *dd, u64 istat) unknown_7322_ibits(dd, istat); if (istat & QIB_I_GPIO) unknown_7322_gpio_intr(dd); - if (istat & QIB_I_C_ERROR) - handle_7322_errors(dd); + if (istat & QIB_I_C_ERROR) { + qib_write_kreg(dd, kr_errmask, 0ULL); + tasklet_schedule(&dd->error_tasklet); + } if (istat & INT_MASK_P(Err, 0) && dd->rcd[0]) handle_7322_p_errors(dd->rcd[0]->ppd); if (istat & INT_MASK_P(Err, 1) && dd->rcd[1]) @@ -2833,9 +2857,8 @@ static irqreturn_t qib_7322intr(int irq, void *data) for (i = 0; i < dd->first_user_ctxt; i++) { if (ctxtrbits & rmask) { ctxtrbits &= ~rmask; - if (dd->rcd[i]) { + if (dd->rcd[i]) qib_kreceive(dd->rcd[i], NULL, &npkts); - } } rmask <<= 1; } @@ -3125,6 +3148,8 @@ try_intx: arg = dd->rcd[ctxt]; if (!arg) continue; + if (qib_krcvq01_no_msi && ctxt < 2) + continue; lsb = QIB_I_RCVAVAIL_LSB + ctxt; handler = qib_7322pintr; name = QIB_DRV_NAME " (kctx)"; @@ -3159,6 +3184,8 @@ try_intx: for (i = 0; i < ARRAY_SIZE(redirect); i++) qib_write_kreg(dd, kr_intredirect + i, redirect[i]); dd->cspec->main_int_mask = mask; + tasklet_init(&dd->error_tasklet, qib_error_tasklet, + (unsigned long)dd); bail:; } @@ -4766,8 +4793,6 @@ static void qib_get_7322_faststats(unsigned long opaque) spin_lock_irqsave(&ppd->dd->eep_st_lock, flags); traffic_wds -= ppd->dd->traffic_wds; ppd->dd->traffic_wds += traffic_wds; - if (traffic_wds >= QIB_TRAFFIC_ACTIVE_THRESHOLD) - atomic_add(ACTIVITY_TIMER, &ppd->dd->active_time); spin_unlock_irqrestore(&ppd->dd->eep_st_lock, flags); if (ppd->cpspec->qdr_dfe_on && (ppd->link_speed_active & QIB_IB_QDR) && @@ -5208,6 +5233,8 @@ static int qib_7322_ib_updown(struct qib_pportdata *ppd, int ibup, u64 ibcs) QIBL_IB_AUTONEG_INPROG))) set_7322_ibspeed_fast(ppd, ppd->link_speed_enabled); if (!(ppd->lflags & QIBL_IB_AUTONEG_INPROG)) { + struct qib_qsfp_data *qd = + &ppd->cpspec->qsfp_data; /* unlock the Tx settings, speed may change */ qib_write_kreg_port(ppd, krp_tx_deemph_override, SYM_MASK(IBSD_TX_DEEMPHASIS_OVERRIDE_0, @@ -5215,6 +5242,12 @@ static int qib_7322_ib_updown(struct qib_pportdata *ppd, int ibup, u64 ibcs) qib_cancel_sends(ppd); /* on link down, ensure sane pcs state */ qib_7322_mini_pcs_reset(ppd); + /* schedule the qsfp refresh which should turn the link + off */ + if (ppd->dd->flags & QIB_HAS_QSFP) { + qd->t_insert = get_jiffies_64(); + queue_work(ib_wq, &qd->work); + } spin_lock_irqsave(&ppd->sdma_lock, flags); if (__qib_sdma_running(ppd)) __qib_sdma_process_event(ppd, @@ -5565,43 +5598,79 @@ static void qsfp_7322_event(struct work_struct *work) struct qib_qsfp_data *qd; struct qib_pportdata *ppd; u64 pwrup; + unsigned long flags; int ret; u32 le2; qd = container_of(work, struct qib_qsfp_data, work); ppd = qd->ppd; - pwrup = qd->t_insert + msecs_to_jiffies(QSFP_PWR_LAG_MSEC); + pwrup = qd->t_insert + + msecs_to_jiffies(QSFP_PWR_LAG_MSEC - QSFP_MODPRS_LAG_MSEC); - /* - * Some QSFP's not only do not respond until the full power-up - * time, but may behave badly if we try. So hold off responding - * to insertion. - */ - while (1) { - u64 now = get_jiffies_64(); - if (time_after64(now, pwrup)) - break; - msleep(20); - } - ret = qib_refresh_qsfp_cache(ppd, &qd->cache); - /* - * Need to change LE2 back to defaults if we couldn't - * read the cable type (to handle cable swaps), so do this - * even on failure to read cable information. We don't - * get here for QME, so IS_QME check not needed here. - */ - if (!ret && !ppd->dd->cspec->r1) { - if (QSFP_IS_ACTIVE_FAR(qd->cache.tech)) - le2 = LE2_QME; - else if (qd->cache.atten[1] >= qib_long_atten && - QSFP_IS_CU(qd->cache.tech)) - le2 = LE2_5m; - else + /* Delay for 20 msecs to allow ModPrs resistor to setup */ + mdelay(QSFP_MODPRS_LAG_MSEC); + + if (!qib_qsfp_mod_present(ppd)) { + ppd->cpspec->qsfp_data.modpresent = 0; + /* Set the physical link to disabled */ + qib_set_ib_7322_lstate(ppd, 0, + QLOGIC_IB_IBCC_LINKINITCMD_DISABLE); + spin_lock_irqsave(&ppd->lflags_lock, flags); + ppd->lflags &= ~QIBL_LINKV; + spin_unlock_irqrestore(&ppd->lflags_lock, flags); + } else { + /* + * Some QSFP's not only do not respond until the full power-up + * time, but may behave badly if we try. So hold off responding + * to insertion. + */ + while (1) { + u64 now = get_jiffies_64(); + if (time_after64(now, pwrup)) + break; + msleep(20); + } + + ret = qib_refresh_qsfp_cache(ppd, &qd->cache); + + /* + * Need to change LE2 back to defaults if we couldn't + * read the cable type (to handle cable swaps), so do this + * even on failure to read cable information. We don't + * get here for QME, so IS_QME check not needed here. + */ + if (!ret && !ppd->dd->cspec->r1) { + if (QSFP_IS_ACTIVE_FAR(qd->cache.tech)) + le2 = LE2_QME; + else if (qd->cache.atten[1] >= qib_long_atten && + QSFP_IS_CU(qd->cache.tech)) + le2 = LE2_5m; + else + le2 = LE2_DEFAULT; + } else le2 = LE2_DEFAULT; - } else - le2 = LE2_DEFAULT; - ibsd_wr_allchans(ppd, 13, (le2 << 7), BMASK(9, 7)); - init_txdds_table(ppd, 0); + ibsd_wr_allchans(ppd, 13, (le2 << 7), BMASK(9, 7)); + /* + * We always change parameteters, since we can choose + * values for cables without eeproms, and the cable may have + * changed from a cable with full or partial eeprom content + * to one with partial or no content. + */ + init_txdds_table(ppd, 0); + /* The physical link is being re-enabled only when the + * previous state was DISABLED and the VALID bit is not + * set. This should only happen when the cable has been + * physically pulled. */ + if (!ppd->cpspec->qsfp_data.modpresent && + (ppd->lflags & (QIBL_LINKV | QIBL_IB_LINK_DISABLED))) { + ppd->cpspec->qsfp_data.modpresent = 1; + qib_set_ib_7322_lstate(ppd, 0, + QLOGIC_IB_IBCC_LINKINITCMD_SLEEP); + spin_lock_irqsave(&ppd->lflags_lock, flags); + ppd->lflags |= QIBL_LINKV; + spin_unlock_irqrestore(&ppd->lflags_lock, flags); + } + } } /* @@ -5705,7 +5774,8 @@ static void set_no_qsfp_atten(struct qib_devdata *dd, int change) /* now change the IBC and serdes, overriding generic */ init_txdds_table(ppd, 1); /* Re-enable the physical state machine on mezz boards - * now that the correct settings have been set. */ + * now that the correct settings have been set. + * QSFP boards are handles by the QSFP event handler */ if (IS_QMH(dd) || IS_QME(dd)) qib_set_ib_7322_lstate(ppd, 0, QLOGIC_IB_IBCC_LINKINITCMD_SLEEP); @@ -6183,6 +6253,8 @@ static int qib_init_7322_variables(struct qib_devdata *dd) /* we always allocate at least 2048 bytes for eager buffers */ dd->rcvegrbufsize = max(mtu, 2048); + BUG_ON(!is_power_of_2(dd->rcvegrbufsize)); + dd->rcvegrbufsize_shift = ilog2(dd->rcvegrbufsize); qib_7322_tidtemplate(dd); @@ -6790,6 +6862,10 @@ struct qib_devdata *qib_init_iba7322_funcs(struct pci_dev *pdev, (i >= ARRAY_SIZE(irq_table) && dd->rcd[i - ARRAY_SIZE(irq_table)])) actual_cnt++; + /* reduce by ctxt's < 2 */ + if (qib_krcvq01_no_msi) + actual_cnt -= dd->num_pports; + tabsize = actual_cnt; dd->cspec->msix_entries = kmalloc(tabsize * sizeof(struct msix_entry), GFP_KERNEL); @@ -7121,7 +7197,8 @@ static void find_best_ent(struct qib_pportdata *ppd, } } - /* Lookup serdes setting by cable type and attenuation */ + /* Active cables don't have attenuation so we only set SERDES + * settings to account for the attenuation of the board traces. */ if (!override && QSFP_IS_ACTIVE(qd->tech)) { *sdr_dds = txdds_sdr + ppd->dd->board_atten; *ddr_dds = txdds_ddr + ppd->dd->board_atten; @@ -7438,12 +7515,6 @@ static int serdes_7322_init_new(struct qib_pportdata *ppd) u32 le_val, rxcaldone; int chan, chan_done = (1 << SERDES_CHANS) - 1; - /* - * Initialize the Tx DDS tables. Also done every QSFP event, - * for adapters with QSFP - */ - init_txdds_table(ppd, 0); - /* Clear cmode-override, may be set from older driver */ ahb_mod(ppd->dd, IBSD(ppd->hw_pidx), 5, 10, 0 << 14, 1 << 14); @@ -7629,6 +7700,12 @@ static int serdes_7322_init_new(struct qib_pportdata *ppd) /* VGA output common mode */ ibsd_wr_allchans(ppd, 12, (3 << 2), BMASK(3, 2)); + /* + * Initialize the Tx DDS tables. Also done every QSFP event, + * for adapters with QSFP + */ + init_txdds_table(ppd, 0); + return 0; } diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c index a01f3fc..9966ec2 100644 --- a/drivers/infiniband/hw/qib/qib_init.c +++ b/drivers/infiniband/hw/qib/qib_init.c @@ -37,6 +37,7 @@ #include <linux/vmalloc.h> #include <linux/delay.h> #include <linux/idr.h> +#include <linux/module.h> #include "qib.h" #include "qib_common.h" @@ -183,6 +184,9 @@ struct qib_ctxtdata *qib_create_ctxtdata(struct qib_pportdata *ppd, u32 ctxt) rcd->rcvegrbuf_chunks = (rcd->rcvegrcnt + rcd->rcvegrbufs_perchunk - 1) / rcd->rcvegrbufs_perchunk; + BUG_ON(!is_power_of_2(rcd->rcvegrbufs_perchunk)); + rcd->rcvegrbufs_perchunk_shift = + ilog2(rcd->rcvegrbufs_perchunk); } return rcd; } @@ -398,6 +402,7 @@ static void enable_chip(struct qib_devdata *dd) if (rcd) dd->f_rcvctrl(rcd->ppd, rcvmask, i); } + dd->freectxts = dd->cfgctxts - dd->first_user_ctxt; } static void verify_interrupt(unsigned long opaque) @@ -581,10 +586,6 @@ int qib_init(struct qib_devdata *dd, int reinit) continue; } - /* let link come up, and enable IBC */ - spin_lock_irqsave(&ppd->lflags_lock, flags); - ppd->lflags &= ~QIBL_IB_LINK_DISABLED; - spin_unlock_irqrestore(&ppd->lflags_lock, flags); portok++; } @@ -764,7 +765,6 @@ static void qib_shutdown_device(struct qib_devdata *dd) dd->f_quiet_serdes(ppd); } - qib_update_eeprom_log(dd); } /** diff --git a/drivers/infiniband/hw/qib/qib_keys.c b/drivers/infiniband/hw/qib/qib_keys.c index 8fd19a4..ca6e6cf 100644 --- a/drivers/infiniband/hw/qib/qib_keys.c +++ b/drivers/infiniband/hw/qib/qib_keys.c @@ -69,6 +69,10 @@ int qib_alloc_lkey(struct qib_lkey_table *rkt, struct qib_mregion *mr) * unrestricted LKEY. */ rkt->gen++; + /* + * bits are capped in qib_verbs.c to insure enough bits + * for generation number + */ mr->lkey = (r << (32 - ib_qib_lkey_table_size)) | ((((1 << (24 - ib_qib_lkey_table_size)) - 1) & rkt->gen) << 8); diff --git a/drivers/infiniband/hw/qib/qib_mad.c b/drivers/infiniband/hw/qib/qib_mad.c index 8fd3df5..dd2ad6d3 100644 --- a/drivers/infiniband/hw/qib/qib_mad.c +++ b/drivers/infiniband/hw/qib/qib_mad.c @@ -1007,7 +1007,7 @@ static int set_pkeys(struct qib_devdata *dd, u8 port, u16 *pkeys) event.event = IB_EVENT_PKEY_CHANGE; event.device = &dd->verbs_dev.ibdev; - event.element.port_num = 1; + event.element.port_num = port; ib_dispatch_event(&event); } return 0; @@ -1125,22 +1125,22 @@ static int subn_trap_repress(struct ib_smp *smp, struct ib_device *ibdev, return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED; } -static int pma_get_classportinfo(struct ib_perf *pmp, +static int pma_get_classportinfo(struct ib_pma_mad *pmp, struct ib_device *ibdev) { - struct ib_pma_classportinfo *p = - (struct ib_pma_classportinfo *)pmp->data; + struct ib_class_port_info *p = + (struct ib_class_port_info *)pmp->data; struct qib_devdata *dd = dd_from_ibdev(ibdev); memset(pmp->data, 0, sizeof(pmp->data)); - if (pmp->attr_mod != 0) - pmp->status |= IB_SMP_INVALID_FIELD; + if (pmp->mad_hdr.attr_mod != 0) + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; /* Note that AllPortSelect is not valid */ p->base_version = 1; p->class_version = 1; - p->cap_mask = IB_PMA_CLASS_CAP_EXT_WIDTH; + p->capability_mask = IB_PMA_CLASS_CAP_EXT_WIDTH; /* * Set the most significant bit of CM2 to indicate support for * congestion statistics @@ -1154,7 +1154,7 @@ static int pma_get_classportinfo(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int pma_get_portsamplescontrol(struct ib_perf *pmp, +static int pma_get_portsamplescontrol(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portsamplescontrol *p = @@ -1169,8 +1169,8 @@ static int pma_get_portsamplescontrol(struct ib_perf *pmp, memset(pmp->data, 0, sizeof(pmp->data)); p->port_select = port_select; - if (pmp->attr_mod != 0 || port_select != port) { - pmp->status |= IB_SMP_INVALID_FIELD; + if (pmp->mad_hdr.attr_mod != 0 || port_select != port) { + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; goto bail; } spin_lock_irqsave(&ibp->lock, flags); @@ -1192,7 +1192,7 @@ bail: return reply((struct ib_smp *) pmp); } -static int pma_set_portsamplescontrol(struct ib_perf *pmp, +static int pma_set_portsamplescontrol(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portsamplescontrol *p = @@ -1205,8 +1205,8 @@ static int pma_set_portsamplescontrol(struct ib_perf *pmp, u8 status, xmit_flags; int ret; - if (pmp->attr_mod != 0 || p->port_select != port) { - pmp->status |= IB_SMP_INVALID_FIELD; + if (pmp->mad_hdr.attr_mod != 0 || p->port_select != port) { + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; ret = reply((struct ib_smp *) pmp); goto bail; } @@ -1321,7 +1321,7 @@ static u64 get_cache_hw_sample_counters(struct qib_pportdata *ppd, return ret; } -static int pma_get_portsamplesresult(struct ib_perf *pmp, +static int pma_get_portsamplesresult(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portsamplesresult *p = @@ -1360,7 +1360,7 @@ static int pma_get_portsamplesresult(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int pma_get_portsamplesresult_ext(struct ib_perf *pmp, +static int pma_get_portsamplesresult_ext(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portsamplesresult_ext *p = @@ -1402,7 +1402,7 @@ static int pma_get_portsamplesresult_ext(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int pma_get_portcounters(struct ib_perf *pmp, +static int pma_get_portcounters(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters *p = (struct ib_pma_portcounters *) @@ -1436,8 +1436,8 @@ static int pma_get_portcounters(struct ib_perf *pmp, memset(pmp->data, 0, sizeof(pmp->data)); p->port_select = port_select; - if (pmp->attr_mod != 0 || port_select != port) - pmp->status |= IB_SMP_INVALID_FIELD; + if (pmp->mad_hdr.attr_mod != 0 || port_select != port) + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; if (cntrs.symbol_error_counter > 0xFFFFUL) p->symbol_error_counter = cpu_to_be16(0xFFFF); @@ -1472,7 +1472,7 @@ static int pma_get_portcounters(struct ib_perf *pmp, cntrs.local_link_integrity_errors = 0xFUL; if (cntrs.excessive_buffer_overrun_errors > 0xFUL) cntrs.excessive_buffer_overrun_errors = 0xFUL; - p->lli_ebor_errors = (cntrs.local_link_integrity_errors << 4) | + p->link_overrun_errors = (cntrs.local_link_integrity_errors << 4) | cntrs.excessive_buffer_overrun_errors; if (cntrs.vl15_dropped > 0xFFFFUL) p->vl15_dropped = cpu_to_be16(0xFFFF); @@ -1500,7 +1500,7 @@ static int pma_get_portcounters(struct ib_perf *pmp, return reply((struct ib_smp *) pmp); } -static int pma_get_portcounters_cong(struct ib_perf *pmp, +static int pma_get_portcounters_cong(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { /* Congestion PMA packets start at offset 24 not 64 */ @@ -1510,7 +1510,7 @@ static int pma_get_portcounters_cong(struct ib_perf *pmp, struct qib_ibport *ibp = to_iport(ibdev, port); struct qib_pportdata *ppd = ppd_from_ibp(ibp); struct qib_devdata *dd = dd_from_ppd(ppd); - u32 port_select = be32_to_cpu(pmp->attr_mod) & 0xFF; + u32 port_select = be32_to_cpu(pmp->mad_hdr.attr_mod) & 0xFF; u64 xmit_wait_counter; unsigned long flags; @@ -1519,9 +1519,9 @@ static int pma_get_portcounters_cong(struct ib_perf *pmp, * SET method ends up calling this anyway. */ if (!dd->psxmitwait_supported) - pmp->status |= IB_SMP_UNSUP_METH_ATTR; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR; if (port_select != port) - pmp->status |= IB_SMP_INVALID_FIELD; + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; qib_get_counters(ppd, &cntrs); spin_lock_irqsave(&ppd->ibport_data.lock, flags); @@ -1603,7 +1603,7 @@ static int pma_get_portcounters_cong(struct ib_perf *pmp, cntrs.local_link_integrity_errors = 0xFUL; if (cntrs.excessive_buffer_overrun_errors > 0xFUL) cntrs.excessive_buffer_overrun_errors = 0xFUL; - p->lli_ebor_errors = (cntrs.local_link_integrity_errors << 4) | + p->link_overrun_errors = (cntrs.local_link_integrity_errors << 4) | cntrs.excessive_buffer_overrun_errors; if (cntrs.vl15_dropped > 0xFFFFUL) p->vl15_dropped = cpu_to_be16(0xFFFF); @@ -1613,7 +1613,7 @@ static int pma_get_portcounters_cong(struct ib_perf *pmp, return reply((struct ib_smp *)pmp); } -static int pma_get_portcounters_ext(struct ib_perf *pmp, +static int pma_get_portcounters_ext(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters_ext *p = @@ -1626,8 +1626,8 @@ static int pma_get_portcounters_ext(struct ib_perf *pmp, memset(pmp->data, 0, sizeof(pmp->data)); p->port_select = port_select; - if (pmp->attr_mod != 0 || port_select != port) { - pmp->status |= IB_SMP_INVALID_FIELD; + if (pmp->mad_hdr.attr_mod != 0 || port_select != port) { + pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD; goto bail; } @@ -1652,7 +1652,7 @@ bail: return reply((struct ib_smp *) pmp); } -static int pma_set_portcounters(struct ib_perf *pmp, +static int pma_set_portcounters(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters *p = (struct ib_pma_portcounters *) @@ -1715,14 +1715,14 @@ static int pma_set_portcounters(struct ib_perf *pmp, return pma_get_portcounters(pmp, ibdev, port); } -static int pma_set_portcounters_cong(struct ib_perf *pmp, +static int pma_set_portcounters_cong(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct qib_ibport *ibp = to_iport(ibdev, port); struct qib_pportdata *ppd = ppd_from_ibp(ibp); struct qib_devdata *dd = dd_from_ppd(ppd); struct qib_verbs_counters cntrs; - u32 counter_select = (be32_to_cpu(pmp->attr_mod) >> 24) & 0xFF; + u32 counter_select = (be32_to_cpu(pmp->mad_hdr.attr_mod) >> 24) & 0xFF; int ret = 0; unsigned long flags; @@ -1766,7 +1766,7 @@ static int pma_set_portcounters_cong(struct ib_perf *pmp, return ret; } -static int pma_set_portcounters_ext(struct ib_perf *pmp, +static int pma_set_portcounters_ext(struct ib_pma_mad *pmp, struct ib_device *ibdev, u8 port) { struct ib_pma_portcounters *p = (struct ib_pma_portcounters *) @@ -1959,19 +1959,19 @@ static int process_perf(struct ib_device *ibdev, u8 port, struct ib_mad *in_mad, struct ib_mad *out_mad) { - struct ib_perf *pmp = (struct ib_perf *)out_mad; + struct ib_pma_mad *pmp = (struct ib_pma_mad *)out_mad; int ret; *out_mad = *in_mad; - if (pmp->class_version != 1) { - pmp->status |= IB_SMP_UNSUP_VERSION; + if (pmp->mad_hdr.class_version != 1) { + pmp->mad_hdr.status |= IB_SMP_UNSUP_VERSION; ret = reply((struct ib_smp *) pmp); goto bail; } - switch (pmp->method) { + switch (pmp->mad_hdr.method) { case IB_MGMT_METHOD_GET: - switch (pmp->attr_id) { + switch (pmp->mad_hdr.attr_id) { case IB_PMA_CLASS_PORT_INFO: ret = pma_get_classportinfo(pmp, ibdev); goto bail; @@ -1994,13 +1994,13 @@ static int process_perf(struct ib_device *ibdev, u8 port, ret = pma_get_portcounters_cong(pmp, ibdev, port); goto bail; default: - pmp->status |= IB_SMP_UNSUP_METH_ATTR; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_smp *) pmp); goto bail; } case IB_MGMT_METHOD_SET: - switch (pmp->attr_id) { + switch (pmp->mad_hdr.attr_id) { case IB_PMA_PORT_SAMPLES_CONTROL: ret = pma_set_portsamplescontrol(pmp, ibdev, port); goto bail; @@ -2014,7 +2014,7 @@ static int process_perf(struct ib_device *ibdev, u8 port, ret = pma_set_portcounters_cong(pmp, ibdev, port); goto bail; default: - pmp->status |= IB_SMP_UNSUP_METH_ATTR; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR; ret = reply((struct ib_smp *) pmp); goto bail; } @@ -2030,7 +2030,7 @@ static int process_perf(struct ib_device *ibdev, u8 port, goto bail; default: - pmp->status |= IB_SMP_UNSUP_METHOD; + pmp->mad_hdr.status |= IB_SMP_UNSUP_METHOD; ret = reply((struct ib_smp *) pmp); } diff --git a/drivers/infiniband/hw/qib/qib_mad.h b/drivers/infiniband/hw/qib/qib_mad.h index 7840ab5..ecc416c 100644 --- a/drivers/infiniband/hw/qib/qib_mad.h +++ b/drivers/infiniband/hw/qib/qib_mad.h @@ -32,6 +32,8 @@ * SOFTWARE. */ +#include <rdma/ib_pma.h> + #define IB_SMP_UNSUP_VERSION cpu_to_be16(0x0004) #define IB_SMP_UNSUP_METHOD cpu_to_be16(0x0008) #define IB_SMP_UNSUP_METH_ATTR cpu_to_be16(0x000C) @@ -180,109 +182,8 @@ struct ib_vl_weight_elem { #define IB_VLARB_HIGHPRI_0_31 3 #define IB_VLARB_HIGHPRI_32_63 4 -/* - * PMA class portinfo capability mask bits - */ -#define IB_PMA_CLASS_CAP_ALLPORTSELECT cpu_to_be16(1 << 8) -#define IB_PMA_CLASS_CAP_EXT_WIDTH cpu_to_be16(1 << 9) -#define IB_PMA_CLASS_CAP_XMIT_WAIT cpu_to_be16(1 << 12) - -#define IB_PMA_CLASS_PORT_INFO cpu_to_be16(0x0001) -#define IB_PMA_PORT_SAMPLES_CONTROL cpu_to_be16(0x0010) -#define IB_PMA_PORT_SAMPLES_RESULT cpu_to_be16(0x0011) -#define IB_PMA_PORT_COUNTERS cpu_to_be16(0x0012) -#define IB_PMA_PORT_COUNTERS_EXT cpu_to_be16(0x001D) -#define IB_PMA_PORT_SAMPLES_RESULT_EXT cpu_to_be16(0x001E) #define IB_PMA_PORT_COUNTERS_CONG cpu_to_be16(0xFF00) -struct ib_perf { - u8 base_version; - u8 mgmt_class; - u8 class_version; - u8 method; - __be16 status; - __be16 unused; - __be64 tid; - __be16 attr_id; - __be16 resv; - __be32 attr_mod; - u8 reserved[40]; - u8 data[192]; -} __attribute__ ((packed)); - -struct ib_pma_classportinfo { - u8 base_version; - u8 class_version; - __be16 cap_mask; - u8 reserved[3]; - u8 resp_time_value; /* only lower 5 bits */ - union ib_gid redirect_gid; - __be32 redirect_tc_sl_fl; /* 8, 4, 20 bits respectively */ - __be16 redirect_lid; - __be16 redirect_pkey; - __be32 redirect_qp; /* only lower 24 bits */ - __be32 redirect_qkey; - union ib_gid trap_gid; - __be32 trap_tc_sl_fl; /* 8, 4, 20 bits respectively */ - __be16 trap_lid; - __be16 trap_pkey; - __be32 trap_hl_qp; /* 8, 24 bits respectively */ - __be32 trap_qkey; -} __attribute__ ((packed)); - -struct ib_pma_portsamplescontrol { - u8 opcode; - u8 port_select; - u8 tick; - u8 counter_width; /* only lower 3 bits */ - __be32 counter_mask0_9; /* 2, 10 * 3, bits */ - __be16 counter_mask10_14; /* 1, 5 * 3, bits */ - u8 sample_mechanisms; - u8 sample_status; /* only lower 2 bits */ - __be64 option_mask; - __be64 vendor_mask; - __be32 sample_start; - __be32 sample_interval; - __be16 tag; - __be16 counter_select[15]; -} __attribute__ ((packed)); - -struct ib_pma_portsamplesresult { - __be16 tag; - __be16 sample_status; /* only lower 2 bits */ - __be32 counter[15]; -} __attribute__ ((packed)); - -struct ib_pma_portsamplesresult_ext { - __be16 tag; - __be16 sample_status; /* only lower 2 bits */ - __be32 extended_width; /* only upper 2 bits */ - __be64 counter[15]; -} __attribute__ ((packed)); - -struct ib_pma_portcounters { - u8 reserved; - u8 port_select; - __be16 counter_select; - __be16 symbol_error_counter; - u8 link_error_recovery_counter; - u8 link_downed_counter; - __be16 port_rcv_errors; - __be16 port_rcv_remphys_errors; - __be16 port_rcv_switch_relay_errors; - __be16 port_xmit_discards; - u8 port_xmit_constraint_errors; - u8 port_rcv_constraint_errors; - u8 reserved1; - u8 lli_ebor_errors; /* 4, 4, bits */ - __be16 reserved2; - __be16 vl15_dropped; - __be32 port_xmit_data; - __be32 port_rcv_data; - __be32 port_xmit_packets; - __be32 port_rcv_packets; -} __attribute__ ((packed)); - struct ib_pma_portcounters_cong { u8 reserved; u8 reserved1; @@ -297,7 +198,7 @@ struct ib_pma_portcounters_cong { u8 port_xmit_constraint_errors; u8 port_rcv_constraint_errors; u8 reserved2; - u8 lli_ebor_errors; /* 4, 4, bits */ + u8 link_overrun_errors; /* LocalLink: 7:4, BufferOverrun: 3:0 */ __be16 reserved3; __be16 vl15_dropped; __be64 port_xmit_data; @@ -316,49 +217,11 @@ struct ib_pma_portcounters_cong { /* number of 4nsec cycles equaling 2secs */ #define QIB_CONG_TIMER_PSINTERVAL 0x1DCD64EC -#define IB_PMA_SEL_SYMBOL_ERROR cpu_to_be16(0x0001) -#define IB_PMA_SEL_LINK_ERROR_RECOVERY cpu_to_be16(0x0002) -#define IB_PMA_SEL_LINK_DOWNED cpu_to_be16(0x0004) -#define IB_PMA_SEL_PORT_RCV_ERRORS cpu_to_be16(0x0008) -#define IB_PMA_SEL_PORT_RCV_REMPHYS_ERRORS cpu_to_be16(0x0010) -#define IB_PMA_SEL_PORT_XMIT_DISCARDS cpu_to_be16(0x0040) -#define IB_PMA_SEL_LOCAL_LINK_INTEGRITY_ERRORS cpu_to_be16(0x0200) -#define IB_PMA_SEL_EXCESSIVE_BUFFER_OVERRUNS cpu_to_be16(0x0400) -#define IB_PMA_SEL_PORT_VL15_DROPPED cpu_to_be16(0x0800) -#define IB_PMA_SEL_PORT_XMIT_DATA cpu_to_be16(0x1000) -#define IB_PMA_SEL_PORT_RCV_DATA cpu_to_be16(0x2000) -#define IB_PMA_SEL_PORT_XMIT_PACKETS cpu_to_be16(0x4000) -#define IB_PMA_SEL_PORT_RCV_PACKETS cpu_to_be16(0x8000) - #define IB_PMA_SEL_CONG_ALL 0x01 #define IB_PMA_SEL_CONG_PORT_DATA 0x02 #define IB_PMA_SEL_CONG_XMIT 0x04 #define IB_PMA_SEL_CONG_ROUTING 0x08 -struct ib_pma_portcounters_ext { - u8 reserved; - u8 port_select; - __be16 counter_select; - __be32 reserved1; - __be64 port_xmit_data; - __be64 port_rcv_data; - __be64 port_xmit_packets; - __be64 port_rcv_packets; - __be64 port_unicast_xmit_packets; - __be64 port_unicast_rcv_packets; - __be64 port_multicast_xmit_packets; - __be64 port_multicast_rcv_packets; -} __attribute__ ((packed)); - -#define IB_PMA_SELX_PORT_XMIT_DATA cpu_to_be16(0x0001) -#define IB_PMA_SELX_PORT_RCV_DATA cpu_to_be16(0x0002) -#define IB_PMA_SELX_PORT_XMIT_PACKETS cpu_to_be16(0x0004) -#define IB_PMA_SELX_PORT_RCV_PACKETS cpu_to_be16(0x0008) -#define IB_PMA_SELX_PORT_UNI_XMIT_PACKETS cpu_to_be16(0x0010) -#define IB_PMA_SELX_PORT_UNI_RCV_PACKETS cpu_to_be16(0x0020) -#define IB_PMA_SELX_PORT_MULTI_XMIT_PACKETS cpu_to_be16(0x0040) -#define IB_PMA_SELX_PORT_MULTI_RCV_PACKETS cpu_to_be16(0x0080) - /* * The PortSamplesControl.CounterMasks field is an array of 3 bit fields * which specify the N'th counter's capabilities. See ch. 16.1.3.2. diff --git a/drivers/infiniband/hw/qib/qib_pcie.c b/drivers/infiniband/hw/qib/qib_pcie.c index 891cc2f..97a8bdf 100644 --- a/drivers/infiniband/hw/qib/qib_pcie.c +++ b/drivers/infiniband/hw/qib/qib_pcie.c @@ -35,6 +35,7 @@ #include <linux/delay.h> #include <linux/vmalloc.h> #include <linux/aer.h> +#include <linux/module.h> #include "qib.h" @@ -255,7 +256,7 @@ int qib_pcie_params(struct qib_devdata *dd, u32 minw, u32 *nent, u16 linkstat, speed; int pos = 0, pose, ret = 1; - pose = pci_find_capability(dd->pcidev, PCI_CAP_ID_EXP); + pose = pci_pcie_cap(dd->pcidev); if (!pose) { qib_dev_err(dd, "Can't find PCI Express capability!\n"); /* set up something... */ @@ -509,7 +510,7 @@ static int qib_tune_pcie_coalesce(struct qib_devdata *dd) qib_devinfo(dd->pcidev, "Parent not root\n"); return 1; } - ppos = pci_find_capability(parent, PCI_CAP_ID_EXP); + ppos = pci_pcie_cap(parent); if (!ppos) return 1; if (parent->vendor != 0x8086) @@ -578,14 +579,14 @@ static int qib_tune_pcie_caps(struct qib_devdata *dd) qib_devinfo(dd->pcidev, "Parent not root\n"); goto bail; } - ppos = pci_find_capability(parent, PCI_CAP_ID_EXP); + ppos = pci_pcie_cap(parent); if (ppos) { pci_read_config_word(parent, ppos + PCI_EXP_DEVCAP, &pcaps); pci_read_config_word(parent, ppos + PCI_EXP_DEVCTL, &pctl); } else goto bail; /* Find out supported and configured values for endpoint (us) */ - epos = pci_find_capability(dd->pcidev, PCI_CAP_ID_EXP); + epos = pci_pcie_cap(dd->pcidev); if (epos) { pci_read_config_word(dd->pcidev, epos + PCI_EXP_DEVCAP, &ecaps); pci_read_config_word(dd->pcidev, epos + PCI_EXP_DEVCTL, &ectl); diff --git a/drivers/infiniband/hw/qib/qib_qp.c b/drivers/infiniband/hw/qib/qib_qp.c index e16751f..7e7e16f 100644 --- a/drivers/infiniband/hw/qib/qib_qp.c +++ b/drivers/infiniband/hw/qib/qib_qp.c @@ -34,6 +34,7 @@ #include <linux/err.h> #include <linux/vmalloc.h> +#include <linux/jhash.h> #include "qib.h" @@ -204,6 +205,13 @@ static void free_qpn(struct qib_qpn_table *qpt, u32 qpn) clear_bit(qpn & BITS_PER_PAGE_MASK, map->page); } +static inline unsigned qpn_hash(struct qib_ibdev *dev, u32 qpn) +{ + return jhash_1word(qpn, dev->qp_rnd) & + (dev->qp_table_size - 1); +} + + /* * Put the QP into the hash table. * The hash table holds a reference to the QP. @@ -211,22 +219,23 @@ static void free_qpn(struct qib_qpn_table *qpt, u32 qpn) static void insert_qp(struct qib_ibdev *dev, struct qib_qp *qp) { struct qib_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num); - unsigned n = qp->ibqp.qp_num % dev->qp_table_size; unsigned long flags; + unsigned n = qpn_hash(dev, qp->ibqp.qp_num); spin_lock_irqsave(&dev->qpt_lock, flags); + atomic_inc(&qp->refcount); if (qp->ibqp.qp_num == 0) - ibp->qp0 = qp; + rcu_assign_pointer(ibp->qp0, qp); else if (qp->ibqp.qp_num == 1) - ibp->qp1 = qp; + rcu_assign_pointer(ibp->qp1, qp); else { qp->next = dev->qp_table[n]; - dev->qp_table[n] = qp; + rcu_assign_pointer(dev->qp_table[n], qp); } - atomic_inc(&qp->refcount); spin_unlock_irqrestore(&dev->qpt_lock, flags); + synchronize_rcu(); } /* @@ -236,29 +245,32 @@ static void insert_qp(struct qib_ibdev *dev, struct qib_qp *qp) static void remove_qp(struct qib_ibdev *dev, struct qib_qp *qp) { struct qib_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num); - struct qib_qp *q, **qpp; + unsigned n = qpn_hash(dev, qp->ibqp.qp_num); unsigned long flags; - qpp = &dev->qp_table[qp->ibqp.qp_num % dev->qp_table_size]; - spin_lock_irqsave(&dev->qpt_lock, flags); if (ibp->qp0 == qp) { - ibp->qp0 = NULL; atomic_dec(&qp->refcount); + rcu_assign_pointer(ibp->qp0, NULL); } else if (ibp->qp1 == qp) { - ibp->qp1 = NULL; atomic_dec(&qp->refcount); - } else + rcu_assign_pointer(ibp->qp1, NULL); + } else { + struct qib_qp *q, **qpp; + + qpp = &dev->qp_table[n]; for (; (q = *qpp) != NULL; qpp = &q->next) if (q == qp) { - *qpp = qp->next; - qp->next = NULL; atomic_dec(&qp->refcount); + rcu_assign_pointer(*qpp, qp->next); + qp->next = NULL; break; } + } spin_unlock_irqrestore(&dev->qpt_lock, flags); + synchronize_rcu(); } /** @@ -280,21 +292,24 @@ unsigned qib_free_all_qps(struct qib_devdata *dd) if (!qib_mcast_tree_empty(ibp)) qp_inuse++; - if (ibp->qp0) + rcu_read_lock(); + if (rcu_dereference(ibp->qp0)) qp_inuse++; - if (ibp->qp1) + if (rcu_dereference(ibp->qp1)) qp_inuse++; + rcu_read_unlock(); } spin_lock_irqsave(&dev->qpt_lock, flags); for (n = 0; n < dev->qp_table_size; n++) { qp = dev->qp_table[n]; - dev->qp_table[n] = NULL; + rcu_assign_pointer(dev->qp_table[n], NULL); for (; qp; qp = qp->next) qp_inuse++; } spin_unlock_irqrestore(&dev->qpt_lock, flags); + synchronize_rcu(); return qp_inuse; } @@ -309,25 +324,28 @@ unsigned qib_free_all_qps(struct qib_devdata *dd) */ struct qib_qp *qib_lookup_qpn(struct qib_ibport *ibp, u32 qpn) { - struct qib_ibdev *dev = &ppd_from_ibp(ibp)->dd->verbs_dev; - unsigned long flags; - struct qib_qp *qp; + struct qib_qp *qp = NULL; - spin_lock_irqsave(&dev->qpt_lock, flags); + if (unlikely(qpn <= 1)) { + rcu_read_lock(); + if (qpn == 0) + qp = rcu_dereference(ibp->qp0); + else + qp = rcu_dereference(ibp->qp1); + } else { + struct qib_ibdev *dev = &ppd_from_ibp(ibp)->dd->verbs_dev; + unsigned n = qpn_hash(dev, qpn); - if (qpn == 0) - qp = ibp->qp0; - else if (qpn == 1) - qp = ibp->qp1; - else - for (qp = dev->qp_table[qpn % dev->qp_table_size]; qp; - qp = qp->next) + rcu_read_lock(); + for (qp = dev->qp_table[n]; rcu_dereference(qp); qp = qp->next) if (qp->ibqp.qp_num == qpn) break; + } if (qp) - atomic_inc(&qp->refcount); + if (unlikely(!atomic_inc_not_zero(&qp->refcount))) + qp = NULL; - spin_unlock_irqrestore(&dev->qpt_lock, flags); + rcu_read_unlock(); return qp; } @@ -765,8 +783,10 @@ int qib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, } } - if (attr_mask & IB_QP_PATH_MTU) + if (attr_mask & IB_QP_PATH_MTU) { qp->path_mtu = pmtu; + qp->pmtu = ib_mtu_enum_to_int(pmtu); + } if (attr_mask & IB_QP_RETRY_CNT) { qp->s_retry_cnt = attr->retry_cnt; @@ -781,8 +801,12 @@ int qib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, if (attr_mask & IB_QP_MIN_RNR_TIMER) qp->r_min_rnr_timer = attr->min_rnr_timer; - if (attr_mask & IB_QP_TIMEOUT) + if (attr_mask & IB_QP_TIMEOUT) { qp->timeout = attr->timeout; + qp->timeout_jiffies = + usecs_to_jiffies((4096UL * (1UL << qp->timeout)) / + 1000UL); + } if (attr_mask & IB_QP_QKEY) qp->qkey = attr->qkey; @@ -1013,6 +1037,10 @@ struct ib_qp *qib_create_qp(struct ib_pd *ibpd, ret = ERR_PTR(-ENOMEM); goto bail_swq; } + RCU_INIT_POINTER(qp->next, NULL); + qp->timeout_jiffies = + usecs_to_jiffies((4096UL * (1UL << qp->timeout)) / + 1000UL); if (init_attr->srq) sz = 0; else { diff --git a/drivers/infiniband/hw/qib/qib_qsfp.c b/drivers/infiniband/hw/qib/qib_qsfp.c index 3374a52..fa71b1e 100644 --- a/drivers/infiniband/hw/qib/qib_qsfp.c +++ b/drivers/infiniband/hw/qib/qib_qsfp.c @@ -273,18 +273,12 @@ int qib_refresh_qsfp_cache(struct qib_pportdata *ppd, struct qib_qsfp_cache *cp) int ret; int idx; u16 cks; - u32 mask; u8 peek[4]; /* ensure sane contents on invalid reads, for cable swaps */ memset(cp, 0, sizeof(*cp)); - mask = QSFP_GPIO_MOD_PRS_N; - if (ppd->hw_pidx) - mask <<= QSFP_GPIO_PORT2_SHIFT; - - ret = ppd->dd->f_gpio_mod(ppd->dd, 0, 0, 0); - if (ret & mask) { + if (!qib_qsfp_mod_present(ppd)) { ret = -ENODEV; goto bail; } @@ -444,6 +438,19 @@ const char * const qib_qsfp_devtech[16] = { static const char *pwr_codes = "1.5W2.0W2.5W3.5W"; +int qib_qsfp_mod_present(struct qib_pportdata *ppd) +{ + u32 mask; + int ret; + + mask = QSFP_GPIO_MOD_PRS_N << + (ppd->hw_pidx * QSFP_GPIO_PORT2_SHIFT); + ret = ppd->dd->f_gpio_mod(ppd->dd, 0, 0, 0); + + return !((ret & mask) >> + ((ppd->hw_pidx * QSFP_GPIO_PORT2_SHIFT) + 3)); +} + /* * Initialize structures that control access to QSFP. Called once per port * on cards that support QSFP. @@ -452,7 +459,6 @@ void qib_qsfp_init(struct qib_qsfp_data *qd, void (*fevent)(struct work_struct *)) { u32 mask, highs; - int pins; struct qib_devdata *dd = qd->ppd->dd; @@ -474,19 +480,6 @@ void qib_qsfp_init(struct qib_qsfp_data *qd, udelay(20); /* Generous RST dwell */ dd->f_gpio_mod(dd, mask, mask, mask); - /* Spec says module can take up to two seconds! */ - mask = QSFP_GPIO_MOD_PRS_N; - if (qd->ppd->hw_pidx) - mask <<= QSFP_GPIO_PORT2_SHIFT; - - /* Do not try to wait here. Better to let event handle it */ - pins = dd->f_gpio_mod(dd, 0, 0, 0); - if (pins & mask) - goto bail; - /* We see a module, but it may be unwise to look yet. Just schedule */ - qd->t_insert = get_jiffies_64(); - queue_work(ib_wq, &qd->work); -bail: return; } diff --git a/drivers/infiniband/hw/qib/qib_qsfp.h b/drivers/infiniband/hw/qib/qib_qsfp.h index c109bbd..46002a9 100644 --- a/drivers/infiniband/hw/qib/qib_qsfp.h +++ b/drivers/infiniband/hw/qib/qib_qsfp.h @@ -34,6 +34,7 @@ #define QSFP_DEV 0xA0 #define QSFP_PWR_LAG_MSEC 2000 +#define QSFP_MODPRS_LAG_MSEC 20 /* * Below are masks for various QSFP signals, for Port 1. @@ -177,10 +178,12 @@ struct qib_qsfp_data { struct work_struct work; struct qib_qsfp_cache cache; u64 t_insert; + u8 modpresent; }; extern int qib_refresh_qsfp_cache(struct qib_pportdata *ppd, struct qib_qsfp_cache *cp); +extern int qib_qsfp_mod_present(struct qib_pportdata *ppd); extern void qib_qsfp_init(struct qib_qsfp_data *qd, void (*fevent)(struct work_struct *)); extern void qib_qsfp_deinit(struct qib_qsfp_data *qd); diff --git a/drivers/infiniband/hw/qib/qib_rc.c b/drivers/infiniband/hw/qib/qib_rc.c index eca0c41..894afac 100644 --- a/drivers/infiniband/hw/qib/qib_rc.c +++ b/drivers/infiniband/hw/qib/qib_rc.c @@ -59,8 +59,7 @@ static void start_timer(struct qib_qp *qp) qp->s_flags |= QIB_S_TIMER; qp->s_timer.function = rc_timeout; /* 4.096 usec. * (1 << qp->timeout) */ - qp->s_timer.expires = jiffies + - usecs_to_jiffies((4096UL * (1UL << qp->timeout)) / 1000UL); + qp->s_timer.expires = jiffies + qp->timeout_jiffies; add_timer(&qp->s_timer); } @@ -239,7 +238,7 @@ int qib_make_rc_req(struct qib_qp *qp) u32 len; u32 bth0; u32 bth2; - u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu); + u32 pmtu = qp->pmtu; char newreq; unsigned long flags; int ret = 0; @@ -272,13 +271,9 @@ int qib_make_rc_req(struct qib_qp *qp) goto bail; } wqe = get_swqe_ptr(qp, qp->s_last); - while (qp->s_last != qp->s_acked) { - qib_send_complete(qp, wqe, IB_WC_SUCCESS); - if (++qp->s_last >= qp->s_size) - qp->s_last = 0; - wqe = get_swqe_ptr(qp, qp->s_last); - } - qib_send_complete(qp, wqe, IB_WC_WR_FLUSH_ERR); + qib_send_complete(qp, wqe, qp->s_last != qp->s_acked ? + IB_WC_SUCCESS : IB_WC_WR_FLUSH_ERR); + /* will get called again */ goto done; } @@ -1519,9 +1514,7 @@ read_middle: * 4.096 usec. * (1 << qp->timeout) */ qp->s_flags |= QIB_S_TIMER; - mod_timer(&qp->s_timer, jiffies + - usecs_to_jiffies((4096UL * (1UL << qp->timeout)) / - 1000UL)); + mod_timer(&qp->s_timer, jiffies + qp->timeout_jiffies); if (qp->s_flags & QIB_S_WAIT_ACK) { qp->s_flags &= ~QIB_S_WAIT_ACK; qib_schedule_send(qp); @@ -1732,7 +1725,7 @@ static int qib_rc_rcv_error(struct qib_other_headers *ohdr, * same request. */ offset = ((psn - e->psn) & QIB_PSN_MASK) * - ib_mtu_enum_to_int(qp->path_mtu); + qp->pmtu; len = be32_to_cpu(reth->length); if (unlikely(offset + len != e->rdma_sge.sge_length)) goto unlock_done; @@ -1876,7 +1869,7 @@ void qib_rc_rcv(struct qib_ctxtdata *rcd, struct qib_ib_header *hdr, u32 psn; u32 pad; struct ib_wc wc; - u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu); + u32 pmtu = qp->pmtu; int diff; struct ib_reth *reth; unsigned long flags; @@ -1892,10 +1885,8 @@ void qib_rc_rcv(struct qib_ctxtdata *rcd, struct qib_ib_header *hdr, } opcode = be32_to_cpu(ohdr->bth[0]); - spin_lock_irqsave(&qp->s_lock, flags); if (qib_ruc_check_hdr(ibp, hdr, has_grh, qp, opcode)) - goto sunlock; - spin_unlock_irqrestore(&qp->s_lock, flags); + return; psn = be32_to_cpu(ohdr->bth[2]); opcode >>= 24; @@ -1955,8 +1946,6 @@ void qib_rc_rcv(struct qib_ctxtdata *rcd, struct qib_ib_header *hdr, break; } - memset(&wc, 0, sizeof wc); - if (qp->state == IB_QPS_RTR && !(qp->r_flags & QIB_R_COMM_EST)) { qp->r_flags |= QIB_R_COMM_EST; if (qp->ibqp.event_handler) { @@ -2009,16 +1998,19 @@ send_middle: goto rnr_nak; qp->r_rcv_len = 0; if (opcode == OP(SEND_ONLY)) - goto send_last; - /* FALLTHROUGH */ + goto no_immediate_data; + /* FALLTHROUGH for SEND_ONLY_WITH_IMMEDIATE */ case OP(SEND_LAST_WITH_IMMEDIATE): send_last_imm: wc.ex.imm_data = ohdr->u.imm_data; hdrsize += 4; wc.wc_flags = IB_WC_WITH_IMM; - /* FALLTHROUGH */ + goto send_last; case OP(SEND_LAST): case OP(RDMA_WRITE_LAST): +no_immediate_data: + wc.wc_flags = 0; + wc.ex.imm_data = 0; send_last: /* Get the number of bytes the message was padded by. */ pad = (be32_to_cpu(ohdr->bth[0]) >> 20) & 3; @@ -2051,6 +2043,12 @@ send_last: wc.src_qp = qp->remote_qpn; wc.slid = qp->remote_ah_attr.dlid; wc.sl = qp->remote_ah_attr.sl; + /* zero fields that are N/A */ + wc.vendor_err = 0; + wc.pkey_index = 0; + wc.dlid_path_bits = 0; + wc.port_num = 0; + wc.csum_ok = 0; /* Signal completion event if the solicited bit is set. */ qib_cq_enter(to_icq(qp->ibqp.recv_cq), &wc, (ohdr->bth[0] & @@ -2089,7 +2087,7 @@ send_last: if (opcode == OP(RDMA_WRITE_FIRST)) goto send_middle; else if (opcode == OP(RDMA_WRITE_ONLY)) - goto send_last; + goto no_immediate_data; ret = qib_get_rwqe(qp, 1); if (ret < 0) goto nack_op_err; diff --git a/drivers/infiniband/hw/qib/qib_ruc.c b/drivers/infiniband/hw/qib/qib_ruc.c index eb78d93..b4b37e4 100644 --- a/drivers/infiniband/hw/qib/qib_ruc.c +++ b/drivers/infiniband/hw/qib/qib_ruc.c @@ -260,12 +260,15 @@ static int gid_ok(union ib_gid *gid, __be64 gid_prefix, __be64 id) /* * - * This should be called with the QP s_lock held. + * This should be called with the QP r_lock held. + * + * The s_lock will be acquired around the qib_migrate_qp() call. */ int qib_ruc_check_hdr(struct qib_ibport *ibp, struct qib_ib_header *hdr, int has_grh, struct qib_qp *qp, u32 bth0) { __be64 guid; + unsigned long flags; if (qp->s_mig_state == IB_MIG_ARMED && (bth0 & IB_BTH_MIG_REQ)) { if (!has_grh) { @@ -295,7 +298,9 @@ int qib_ruc_check_hdr(struct qib_ibport *ibp, struct qib_ib_header *hdr, if (be16_to_cpu(hdr->lrh[3]) != qp->alt_ah_attr.dlid || ppd_from_ibp(ibp)->port != qp->alt_ah_attr.port_num) goto err; + spin_lock_irqsave(&qp->s_lock, flags); qib_migrate_qp(qp); + spin_unlock_irqrestore(&qp->s_lock, flags); } else { if (!has_grh) { if (qp->remote_ah_attr.ah_flags & IB_AH_GRH) diff --git a/drivers/infiniband/hw/qib/qib_sd7220.c b/drivers/infiniband/hw/qib/qib_sd7220.c index e9f9f8b..de1a4b2 100644 --- a/drivers/infiniband/hw/qib/qib_sd7220.c +++ b/drivers/infiniband/hw/qib/qib_sd7220.c @@ -38,6 +38,7 @@ #include <linux/pci.h> #include <linux/delay.h> +#include <linux/module.h> #include <linux/firmware.h> #include "qib.h" diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c index cad4449..12a9604 100644 --- a/drivers/infiniband/hw/qib/qib_sdma.c +++ b/drivers/infiniband/hw/qib/qib_sdma.c @@ -32,6 +32,7 @@ #include <linux/spinlock.h> #include <linux/netdevice.h> +#include <linux/moduleparam.h> #include "qib.h" #include "qib_common.h" diff --git a/drivers/infiniband/hw/qib/qib_srq.c b/drivers/infiniband/hw/qib/qib_srq.c index c3ec8ef..d623593 100644 --- a/drivers/infiniband/hw/qib/qib_srq.c +++ b/drivers/infiniband/hw/qib/qib_srq.c @@ -107,6 +107,11 @@ struct ib_srq *qib_create_srq(struct ib_pd *ibpd, u32 sz; struct ib_srq *ret; + if (srq_init_attr->srq_type != IB_SRQT_BASIC) { + ret = ERR_PTR(-ENOSYS); + goto done; + } + if (srq_init_attr->attr.max_sge == 0 || srq_init_attr->attr.max_sge > ib_qib_max_srq_sges || srq_init_attr->attr.max_wr == 0 || diff --git a/drivers/infiniband/hw/qib/qib_sysfs.c b/drivers/infiniband/hw/qib/qib_sysfs.c index d50a33f..c97224a 100644 --- a/drivers/infiniband/hw/qib/qib_sysfs.c +++ b/drivers/infiniband/hw/qib/qib_sysfs.c @@ -507,6 +507,17 @@ static ssize_t show_nctxts(struct device *device, dd->first_user_ctxt); } +static ssize_t show_nfreectxts(struct device *device, + struct device_attribute *attr, char *buf) +{ + struct qib_ibdev *dev = + container_of(device, struct qib_ibdev, ibdev.dev); + struct qib_devdata *dd = dd_from_dev(dev); + + /* Return the number of free user ports (contexts) available. */ + return scnprintf(buf, PAGE_SIZE, "%u\n", dd->freectxts); +} + static ssize_t show_serial(struct device *device, struct device_attribute *attr, char *buf) { @@ -539,28 +550,6 @@ bail: return ret < 0 ? ret : count; } -static ssize_t show_logged_errs(struct device *device, - struct device_attribute *attr, char *buf) -{ - struct qib_ibdev *dev = - container_of(device, struct qib_ibdev, ibdev.dev); - struct qib_devdata *dd = dd_from_dev(dev); - int idx, count; - - /* force consistency with actual EEPROM */ - if (qib_update_eeprom_log(dd) != 0) - return -ENXIO; - - count = 0; - for (idx = 0; idx < QIB_EEP_LOG_CNT; ++idx) { - count += scnprintf(buf + count, PAGE_SIZE - count, "%d%c", - dd->eep_st_errs[idx], - idx == (QIB_EEP_LOG_CNT - 1) ? '\n' : ' '); - } - - return count; -} - /* * Dump tempsense regs. in decimal, to ease shell-scripts. */ @@ -604,9 +593,9 @@ static DEVICE_ATTR(hca_type, S_IRUGO, show_hca, NULL); static DEVICE_ATTR(board_id, S_IRUGO, show_hca, NULL); static DEVICE_ATTR(version, S_IRUGO, show_version, NULL); static DEVICE_ATTR(nctxts, S_IRUGO, show_nctxts, NULL); +static DEVICE_ATTR(nfreectxts, S_IRUGO, show_nfreectxts, NULL); static DEVICE_ATTR(serial, S_IRUGO, show_serial, NULL); static DEVICE_ATTR(boardversion, S_IRUGO, show_boardversion, NULL); -static DEVICE_ATTR(logged_errors, S_IRUGO, show_logged_errs, NULL); static DEVICE_ATTR(tempsense, S_IRUGO, show_tempsense, NULL); static DEVICE_ATTR(localbus_info, S_IRUGO, show_localbus_info, NULL); static DEVICE_ATTR(chip_reset, S_IWUSR, NULL, store_chip_reset); @@ -617,9 +606,9 @@ static struct device_attribute *qib_attributes[] = { &dev_attr_board_id, &dev_attr_version, &dev_attr_nctxts, + &dev_attr_nfreectxts, &dev_attr_serial, &dev_attr_boardversion, - &dev_attr_logged_errors, &dev_attr_tempsense, &dev_attr_localbus_info, &dev_attr_chip_reset, diff --git a/drivers/infiniband/hw/qib/qib_tx.c b/drivers/infiniband/hw/qib/qib_tx.c index 7f36454..1bf626c 100644 --- a/drivers/infiniband/hw/qib/qib_tx.c +++ b/drivers/infiniband/hw/qib/qib_tx.c @@ -36,6 +36,7 @@ #include <linux/delay.h> #include <linux/netdevice.h> #include <linux/vmalloc.h> +#include <linux/moduleparam.h> #include "qib.h" diff --git a/drivers/infiniband/hw/qib/qib_uc.c b/drivers/infiniband/hw/qib/qib_uc.c index 32ccf3c..847e7af 100644 --- a/drivers/infiniband/hw/qib/qib_uc.c +++ b/drivers/infiniband/hw/qib/qib_uc.c @@ -51,7 +51,7 @@ int qib_make_uc_req(struct qib_qp *qp) u32 hwords; u32 bth0; u32 len; - u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu); + u32 pmtu = qp->pmtu; int ret = 0; spin_lock_irqsave(&qp->s_lock, flags); @@ -243,13 +243,12 @@ void qib_uc_rcv(struct qib_ibport *ibp, struct qib_ib_header *hdr, int has_grh, void *data, u32 tlen, struct qib_qp *qp) { struct qib_other_headers *ohdr; - unsigned long flags; u32 opcode; u32 hdrsize; u32 psn; u32 pad; struct ib_wc wc; - u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu); + u32 pmtu = qp->pmtu; struct ib_reth *reth; int ret; @@ -263,14 +262,11 @@ void qib_uc_rcv(struct qib_ibport *ibp, struct qib_ib_header *hdr, } opcode = be32_to_cpu(ohdr->bth[0]); - spin_lock_irqsave(&qp->s_lock, flags); if (qib_ruc_check_hdr(ibp, hdr, has_grh, qp, opcode)) - goto sunlock; - spin_unlock_irqrestore(&qp->s_lock, flags); + return; psn = be32_to_cpu(ohdr->bth[2]); opcode >>= 24; - memset(&wc, 0, sizeof wc); /* Compare the PSN verses the expected PSN. */ if (unlikely(qib_cmp24(psn, qp->r_psn) != 0)) { @@ -370,7 +366,7 @@ send_first: } qp->r_rcv_len = 0; if (opcode == OP(SEND_ONLY)) - goto send_last; + goto no_immediate_data; else if (opcode == OP(SEND_ONLY_WITH_IMMEDIATE)) goto send_last_imm; /* FALLTHROUGH */ @@ -389,8 +385,11 @@ send_last_imm: wc.ex.imm_data = ohdr->u.imm_data; hdrsize += 4; wc.wc_flags = IB_WC_WITH_IMM; - /* FALLTHROUGH */ + goto send_last; case OP(SEND_LAST): +no_immediate_data: + wc.ex.imm_data = 0; + wc.wc_flags = 0; send_last: /* Get the number of bytes the message was padded by. */ pad = (be32_to_cpu(ohdr->bth[0]) >> 20) & 3; @@ -418,6 +417,12 @@ last_imm: wc.src_qp = qp->remote_qpn; wc.slid = qp->remote_ah_attr.dlid; wc.sl = qp->remote_ah_attr.sl; + /* zero fields that are N/A */ + wc.vendor_err = 0; + wc.pkey_index = 0; + wc.dlid_path_bits = 0; + wc.port_num = 0; + wc.csum_ok = 0; /* Signal completion event if the solicited bit is set. */ qib_cq_enter(to_icq(qp->ibqp.recv_cq), &wc, (ohdr->bth[0] & @@ -546,6 +551,4 @@ op_err: qib_rc_error(qp, IB_WC_LOC_QP_OP_ERR); return; -sunlock: - spin_unlock_irqrestore(&qp->s_lock, flags); } diff --git a/drivers/infiniband/hw/qib/qib_ud.c b/drivers/infiniband/hw/qib/qib_ud.c index 828609f..c0d93d4 100644 --- a/drivers/infiniband/hw/qib/qib_ud.c +++ b/drivers/infiniband/hw/qib/qib_ud.c @@ -57,13 +57,20 @@ static void qib_ud_loopback(struct qib_qp *sqp, struct qib_swqe *swqe) struct qib_sge *sge; struct ib_wc wc; u32 length; + enum ib_qp_type sqptype, dqptype; qp = qib_lookup_qpn(ibp, swqe->wr.wr.ud.remote_qpn); if (!qp) { ibp->n_pkt_drops++; return; } - if (qp->ibqp.qp_type != sqp->ibqp.qp_type || + + sqptype = sqp->ibqp.qp_type == IB_QPT_GSI ? + IB_QPT_UD : sqp->ibqp.qp_type; + dqptype = qp->ibqp.qp_type == IB_QPT_GSI ? + IB_QPT_UD : qp->ibqp.qp_type; + + if (dqptype != sqptype || !(ib_qib_state_ops[qp->state] & QIB_PROCESS_RECV_OK)) { ibp->n_pkt_drops++; goto drop; diff --git a/drivers/infiniband/hw/qib/qib_user_pages.c b/drivers/infiniband/hw/qib/qib_user_pages.c index 7689e49..2bc1d2b 100644 --- a/drivers/infiniband/hw/qib/qib_user_pages.c +++ b/drivers/infiniband/hw/qib/qib_user_pages.c @@ -74,7 +74,7 @@ static int __qib_get_user_pages(unsigned long start_page, size_t num_pages, goto bail_release; } - current->mm->locked_vm += num_pages; + current->mm->pinned_vm += num_pages; ret = 0; goto bail; @@ -151,7 +151,7 @@ void qib_release_user_pages(struct page **p, size_t num_pages) __qib_release_user_pages(p, num_pages, 1); if (current->mm) { - current->mm->locked_vm -= num_pages; + current->mm->pinned_vm -= num_pages; up_write(¤t->mm->mmap_sem); } } diff --git a/drivers/infiniband/hw/qib/qib_user_sdma.c b/drivers/infiniband/hw/qib/qib_user_sdma.c index 8244208..573b460 100644 --- a/drivers/infiniband/hw/qib/qib_user_sdma.c +++ b/drivers/infiniband/hw/qib/qib_user_sdma.c @@ -284,8 +284,7 @@ static int qib_user_sdma_pin_pages(const struct qib_devdata *dd, int j; int ret; - ret = get_user_pages(current, current->mm, addr, - npages, 0, 1, pages, NULL); + ret = get_user_pages_fast(addr, npages, 0, pages); if (ret != npages) { int i; @@ -830,10 +829,7 @@ int qib_user_sdma_writev(struct qib_ctxtdata *rcd, while (dim) { const int mxp = 8; - down_write(¤t->mm->mmap_sem); ret = qib_user_sdma_queue_pkts(dd, pq, &list, iov, dim, mxp); - up_write(¤t->mm->mmap_sem); - if (ret <= 0) goto done_unlock; else { diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c index 9fab404..c51a6f9 100644 --- a/drivers/infiniband/hw/qib/qib_verbs.c +++ b/drivers/infiniband/hw/qib/qib_verbs.c @@ -35,14 +35,17 @@ #include <rdma/ib_mad.h> #include <rdma/ib_user_verbs.h> #include <linux/io.h> +#include <linux/module.h> #include <linux/utsname.h> #include <linux/rculist.h> #include <linux/mm.h> +#include <linux/random.h> +#include <linux/vmalloc.h> #include "qib.h" #include "qib_common.h" -static unsigned int ib_qib_qp_table_size = 251; +static unsigned int ib_qib_qp_table_size = 256; module_param_named(qp_table_size, ib_qib_qp_table_size, uint, S_IRUGO); MODULE_PARM_DESC(qp_table_size, "QP table size"); @@ -659,17 +662,25 @@ void qib_ib_rcv(struct qib_ctxtdata *rcd, void *rhdr, void *data, u32 tlen) if (atomic_dec_return(&mcast->refcount) <= 1) wake_up(&mcast->wait); } else { - qp = qib_lookup_qpn(ibp, qp_num); - if (!qp) - goto drop; + if (rcd->lookaside_qp) { + if (rcd->lookaside_qpn != qp_num) { + if (atomic_dec_and_test( + &rcd->lookaside_qp->refcount)) + wake_up( + &rcd->lookaside_qp->wait); + rcd->lookaside_qp = NULL; + } + } + if (!rcd->lookaside_qp) { + qp = qib_lookup_qpn(ibp, qp_num); + if (!qp) + goto drop; + rcd->lookaside_qp = qp; + rcd->lookaside_qpn = qp_num; + } else + qp = rcd->lookaside_qp; ibp->n_unicast_rcv++; qib_qp_rcv(rcd, hdr, lnh == QIB_LRH_GRH, data, tlen, qp); - /* - * Notify qib_destroy_qp() if it is waiting - * for us to finish. - */ - if (atomic_dec_and_test(&qp->refcount)) - wake_up(&qp->wait); } return; @@ -1974,6 +1985,8 @@ static void init_ibport(struct qib_pportdata *ppd) ibp->z_excessive_buffer_overrun_errors = cntrs.excessive_buffer_overrun_errors; ibp->z_vl15_dropped = cntrs.vl15_dropped; + RCU_INIT_POINTER(ibp->qp0, NULL); + RCU_INIT_POINTER(ibp->qp1, NULL); } /** @@ -1990,12 +2003,15 @@ int qib_register_ib_device(struct qib_devdata *dd) int ret; dev->qp_table_size = ib_qib_qp_table_size; - dev->qp_table = kzalloc(dev->qp_table_size * sizeof *dev->qp_table, + get_random_bytes(&dev->qp_rnd, sizeof(dev->qp_rnd)); + dev->qp_table = kmalloc(dev->qp_table_size * sizeof *dev->qp_table, GFP_KERNEL); if (!dev->qp_table) { ret = -ENOMEM; goto err_qpt; } + for (i = 0; i < dev->qp_table_size; i++) + RCU_INIT_POINTER(dev->qp_table[i], NULL); for (i = 0; i < dd->num_pports; i++) init_ibport(ppd + i); @@ -2020,10 +2036,16 @@ int qib_register_ib_device(struct qib_devdata *dd) * the LKEY). The remaining bits act as a generation number or tag. */ spin_lock_init(&dev->lk_table.lock); + /* insure generation is at least 4 bits see keys.c */ + if (ib_qib_lkey_table_size > MAX_LKEY_TABLE_BITS) { + qib_dev_warn(dd, "lkey bits %u too large, reduced to %u\n", + ib_qib_lkey_table_size, MAX_LKEY_TABLE_BITS); + ib_qib_lkey_table_size = MAX_LKEY_TABLE_BITS; + } dev->lk_table.max = 1 << ib_qib_lkey_table_size; lk_tab_size = dev->lk_table.max * sizeof(*dev->lk_table.table); dev->lk_table.table = (struct qib_mregion **) - __get_free_pages(GFP_KERNEL, get_order(lk_tab_size)); + vmalloc(lk_tab_size); if (dev->lk_table.table == NULL) { ret = -ENOMEM; goto err_lk; @@ -2193,7 +2215,7 @@ err_tx: sizeof(struct qib_pio_header), dev->pio_hdrs, dev->pio_hdrs_phys); err_hdrs: - free_pages((unsigned long) dev->lk_table.table, get_order(lk_tab_size)); + vfree(dev->lk_table.table); err_lk: kfree(dev->qp_table); err_qpt: @@ -2247,7 +2269,6 @@ void qib_unregister_ib_device(struct qib_devdata *dd) sizeof(struct qib_pio_header), dev->pio_hdrs, dev->pio_hdrs_phys); lk_tab_size = dev->lk_table.max * sizeof(*dev->lk_table.table); - free_pages((unsigned long) dev->lk_table.table, - get_order(lk_tab_size)); + vfree(dev->lk_table.table); kfree(dev->qp_table); } diff --git a/drivers/infiniband/hw/qib/qib_verbs.h b/drivers/infiniband/hw/qib/qib_verbs.h index 95e5b47..66f7f62 100644 --- a/drivers/infiniband/hw/qib/qib_verbs.h +++ b/drivers/infiniband/hw/qib/qib_verbs.h @@ -485,6 +485,7 @@ struct qib_qp { u8 alt_timeout; /* Alternate path timeout for this QP */ u8 port_num; enum ib_mtu path_mtu; + u32 pmtu; /* decoded from path_mtu */ u32 remote_qpn; u32 qkey; /* QKEY for this QP (for UD or RD) */ u32 s_size; /* send work queue size */ @@ -495,6 +496,7 @@ struct qib_qp { u32 s_last; /* last completed entry */ u32 s_ssn; /* SSN of tail entry */ u32 s_lsn; /* limit sequence number (credit) */ + unsigned long timeout_jiffies; /* computed from timeout */ struct qib_swqe *s_wq; /* send work queue */ struct qib_swqe *s_wqe; struct qib_rq r_rq; /* receive work queue */ @@ -620,6 +622,8 @@ struct qib_qpn_table { struct qpn_map map[QPNMAP_ENTRIES]; }; +#define MAX_LKEY_TABLE_BITS 23 + struct qib_lkey_table { spinlock_t lock; /* protect changes in this struct */ u32 next; /* next unused index (speeds search) */ @@ -723,7 +727,8 @@ struct qib_ibdev { dma_addr_t pio_hdrs_phys; /* list of QPs waiting for RNR timer */ spinlock_t pending_lock; /* protect wait lists, PMA counters, etc. */ - unsigned qp_table_size; /* size of the hash table */ + u32 qp_table_size; /* size of the hash table */ + u32 qp_rnd; /* random bytes for hash */ spinlock_t qpt_lock; u32 n_piowait; diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index 936804e..86df632 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -46,7 +46,7 @@ #include <net/neighbour.h> #include <net/sch_generic.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include <rdma/ib_verbs.h> #include <rdma/ib_pack.h> diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c index 073acdf..3767853 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c @@ -37,6 +37,7 @@ #include <linux/delay.h> #include <linux/slab.h> #include <linux/vmalloc.h> +#include <linux/moduleparam.h> #include "ipoib.h" @@ -84,7 +85,7 @@ static void ipoib_cm_dma_unmap_rx(struct ipoib_dev_priv *priv, int frags, ib_dma_unmap_single(priv->ca, mapping[0], IPOIB_CM_HEAD_SIZE, DMA_FROM_DEVICE); for (i = 0; i < frags; ++i) - ib_dma_unmap_single(priv->ca, mapping[i + 1], PAGE_SIZE, DMA_FROM_DEVICE); + ib_dma_unmap_page(priv->ca, mapping[i + 1], PAGE_SIZE, DMA_FROM_DEVICE); } static int ipoib_cm_post_receive_srq(struct net_device *dev, int id) @@ -169,7 +170,7 @@ static struct sk_buff *ipoib_cm_alloc_rx_skb(struct net_device *dev, goto partial_error; skb_fill_page_desc(skb, i, page, 0, PAGE_SIZE); - mapping[i + 1] = ib_dma_map_page(priv->ca, skb_shinfo(skb)->frags[i].page, + mapping[i + 1] = ib_dma_map_page(priv->ca, page, 0, PAGE_SIZE, DMA_FROM_DEVICE); if (unlikely(ib_dma_mapping_error(priv->ca, mapping[i + 1]))) goto partial_error; @@ -183,7 +184,7 @@ partial_error: ib_dma_unmap_single(priv->ca, mapping[0], IPOIB_CM_HEAD_SIZE, DMA_FROM_DEVICE); for (; i > 0; --i) - ib_dma_unmap_single(priv->ca, mapping[i], PAGE_SIZE, DMA_FROM_DEVICE); + ib_dma_unmap_page(priv->ca, mapping[i], PAGE_SIZE, DMA_FROM_DEVICE); dev_kfree_skb_any(skb); return NULL; @@ -537,12 +538,13 @@ static void skb_put_frags(struct sk_buff *skb, unsigned int hdr_space, if (length == 0) { /* don't need this page */ - skb_fill_page_desc(toskb, i, frag->page, 0, PAGE_SIZE); + skb_fill_page_desc(toskb, i, skb_frag_page(frag), + 0, PAGE_SIZE); --skb_shinfo(skb)->nr_frags; } else { size = min(length, (unsigned) PAGE_SIZE); - frag->size = size; + skb_frag_size_set(frag, size); skb->data_len += size; skb->truesize += size; skb->len += size; @@ -1500,6 +1502,7 @@ static void ipoib_cm_create_srq(struct net_device *dev, int max_sge) { struct ipoib_dev_priv *priv = netdev_priv(dev); struct ib_srq_init_attr srq_init_attr = { + .srq_type = IB_SRQT_BASIC, .attr = { .max_wr = ipoib_recvq_size, .max_sge = max_sge diff --git a/drivers/infiniband/ulp/ipoib/ipoib_fs.c b/drivers/infiniband/ulp/ipoib/ipoib_fs.c index 86eae22..5006185 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_fs.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_fs.c @@ -37,6 +37,7 @@ struct file_operations; #include <linux/debugfs.h> +#include <linux/export.h> #include "ipoib.h" @@ -212,16 +213,15 @@ static int ipoib_path_seq_show(struct seq_file *file, void *iter_ptr) gid_buf, path.pathrec.dlid ? "yes" : "no"); if (path.pathrec.dlid) { - rate = ib_rate_to_mult(path.pathrec.rate) * 25; + rate = ib_rate_to_mbps(path.pathrec.rate); seq_printf(file, " DLID: 0x%04x\n" " SL: %12d\n" - " rate: %*d%s Gb/sec\n", + " rate: %8d.%d Gb/sec\n", be16_to_cpu(path.pathrec.dlid), path.pathrec.sl, - 10 - ((rate % 10) ? 2 : 0), - rate / 10, rate % 10 ? ".5" : ""); + rate / 1000, rate % 1000); } seq_putc(file, '\n'); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c index 81ae61d..4115be5 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c @@ -34,6 +34,7 @@ */ #include <linux/delay.h> +#include <linux/moduleparam.h> #include <linux/dma-mapping.h> #include <linux/slab.h> @@ -56,21 +57,24 @@ struct ipoib_ah *ipoib_create_ah(struct net_device *dev, struct ib_pd *pd, struct ib_ah_attr *attr) { struct ipoib_ah *ah; + struct ib_ah *vah; ah = kmalloc(sizeof *ah, GFP_KERNEL); if (!ah) - return NULL; + return ERR_PTR(-ENOMEM); ah->dev = dev; ah->last_send = 0; kref_init(&ah->ref); - ah->ah = ib_create_ah(pd, attr); - if (IS_ERR(ah->ah)) { + vah = ib_create_ah(pd, attr); + if (IS_ERR(vah)) { kfree(ah); - ah = NULL; - } else + ah = (struct ipoib_ah *)vah; + } else { + ah->ah = vah; ipoib_dbg(netdev_priv(dev), "Created ah %p\n", ah->ah); + } return ah; } @@ -117,7 +121,7 @@ static void ipoib_ud_skb_put_frags(struct ipoib_dev_priv *priv, size = length - IPOIB_UD_HEAD_SIZE; - frag->size = size; + skb_frag_size_set(frag, size); skb->data_len += size; skb->truesize += size; } else @@ -182,7 +186,7 @@ static struct sk_buff *ipoib_alloc_rx_skb(struct net_device *dev, int id) goto partial_error; skb_fill_page_desc(skb, 0, page, 0, PAGE_SIZE); mapping[1] = - ib_dma_map_page(priv->ca, skb_shinfo(skb)->frags[0].page, + ib_dma_map_page(priv->ca, page, 0, PAGE_SIZE, DMA_FROM_DEVICE); if (unlikely(ib_dma_mapping_error(priv->ca, mapping[1]))) goto partial_error; @@ -322,9 +326,10 @@ static int ipoib_dma_map_tx(struct ib_device *ca, off = 0; for (i = 0; i < skb_shinfo(skb)->nr_frags; ++i) { - skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; - mapping[i + off] = ib_dma_map_page(ca, frag->page, - frag->page_offset, frag->size, + const skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; + mapping[i + off] = ib_dma_map_page(ca, + skb_frag_page(frag), + frag->page_offset, skb_frag_size(frag), DMA_TO_DEVICE); if (unlikely(ib_dma_mapping_error(ca, mapping[i + off]))) goto partial_error; @@ -333,8 +338,9 @@ static int ipoib_dma_map_tx(struct ib_device *ca, partial_error: for (; i > 0; --i) { - skb_frag_t *frag = &skb_shinfo(skb)->frags[i - 1]; - ib_dma_unmap_page(ca, mapping[i - !off], frag->size, DMA_TO_DEVICE); + const skb_frag_t *frag = &skb_shinfo(skb)->frags[i - 1]; + + ib_dma_unmap_page(ca, mapping[i - !off], skb_frag_size(frag), DMA_TO_DEVICE); } if (off) @@ -358,8 +364,9 @@ static void ipoib_dma_unmap_tx(struct ib_device *ca, off = 0; for (i = 0; i < skb_shinfo(skb)->nr_frags; ++i) { - skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; - ib_dma_unmap_page(ca, mapping[i + off], frag->size, + const skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; + + ib_dma_unmap_page(ca, mapping[i + off], skb_frag_size(frag), DMA_TO_DEVICE); } } @@ -509,7 +516,7 @@ static inline int post_send(struct ipoib_dev_priv *priv, for (i = 0; i < nr_frags; ++i) { priv->tx_sge[i + off].addr = mapping[i + off]; - priv->tx_sge[i + off].length = frags[i].size; + priv->tx_sge[i + off].length = skb_frag_size(&frags[i]); } priv->tx_wr.num_sge = nr_frags + off; priv->tx_wr.wr_id = wr_id; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c index 6ea9600..1740b82 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c @@ -432,7 +432,7 @@ static void path_rec_completion(int status, spin_lock_irqsave(&priv->lock, flags); - if (ah) { + if (!IS_ERR_OR_NULL(ah)) { path->pathrec = *pathrec; old_ah = path->ah; @@ -989,7 +989,7 @@ static const struct net_device_ops ipoib_netdev_ops = { .ndo_fix_features = ipoib_fix_features, .ndo_start_xmit = ipoib_start_xmit, .ndo_tx_timeout = ipoib_timeout, - .ndo_set_multicast_list = ipoib_set_mcast_list, + .ndo_set_rx_mode = ipoib_set_mcast_list, .ndo_neigh_setup = ipoib_neigh_setup_dev, }; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index fc04594..80799c0 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -34,6 +34,7 @@ #include <linux/skbuff.h> #include <linux/rtnetlink.h> +#include <linux/moduleparam.h> #include <linux/ip.h> #include <linux/in.h> #include <linux/igmp.h> @@ -248,8 +249,11 @@ static int ipoib_mcast_join_finish(struct ipoib_mcast *mcast, av.grh.dgid = mcast->mcmember.mgid; ah = ipoib_create_ah(dev, priv->pd, &av); - if (!ah) { - ipoib_warn(priv, "ib_address_create failed\n"); + if (IS_ERR(ah)) { + ipoib_warn(priv, "ib_address_create failed %ld\n", + -PTR_ERR(ah)); + /* use original error */ + return PTR_ERR(ah); } else { spin_lock_irq(&priv->lock); mcast->ah = ah; diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c index f8f5758..d5f3b69 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.c +++ b/drivers/infiniband/ulp/iser/iscsi_iser.c @@ -57,6 +57,7 @@ #include <linux/scatterlist.h> #include <linux/delay.h> #include <linux/slab.h> +#include <linux/module.h> #include <net/sock.h> @@ -101,13 +102,17 @@ iscsi_iser_recv(struct iscsi_conn *conn, /* verify PDU length */ datalen = ntoh24(hdr->dlength); - if (datalen != rx_data_len) { - printk(KERN_ERR "iscsi_iser: datalen %d (hdr) != %d (IB) \n", - datalen, rx_data_len); + if (datalen > rx_data_len || (datalen + 4) < rx_data_len) { + iser_err("wrong datalen %d (hdr), %d (IB)\n", + datalen, rx_data_len); rc = ISCSI_ERR_DATALEN; goto error; } + if (datalen != rx_data_len) + iser_dbg("aligned datalen (%d) hdr, %d (IB)\n", + datalen, rx_data_len); + /* read AHS */ ahslen = hdr->hlength * 4; @@ -147,7 +152,6 @@ int iser_initialize_task_headers(struct iscsi_task *task, tx_desc->tx_sg[0].length = ISER_HEADERS_LEN; tx_desc->tx_sg[0].lkey = device->mr->lkey; - iser_task->headers_initialized = 1; iser_task->iser_conn = iser_conn; return 0; } @@ -162,8 +166,7 @@ iscsi_iser_task_init(struct iscsi_task *task) { struct iscsi_iser_task *iser_task = task->dd_data; - if (!iser_task->headers_initialized) - if (iser_initialize_task_headers(task, &iser_task->desc)) + if (iser_initialize_task_headers(task, &iser_task->desc)) return -ENOMEM; /* mgmt task */ @@ -274,6 +277,13 @@ iscsi_iser_task_xmit(struct iscsi_task *task) static void iscsi_iser_cleanup_task(struct iscsi_task *task) { struct iscsi_iser_task *iser_task = task->dd_data; + struct iser_tx_desc *tx_desc = &iser_task->desc; + + struct iscsi_iser_conn *iser_conn = task->conn->dd_data; + struct iser_device *device = iser_conn->ib_conn->device; + + ib_dma_unmap_single(device->ib_device, + tx_desc->dma_addr, ISER_HEADERS_LEN, DMA_TO_DEVICE); /* mgmt tasks do not need special cleanup */ if (!task->sc) @@ -618,6 +628,59 @@ iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep) iser_conn_terminate(ib_conn); } +static mode_t iser_attr_is_visible(int param_type, int param) +{ + switch (param_type) { + case ISCSI_HOST_PARAM: + switch (param) { + case ISCSI_HOST_PARAM_NETDEV_NAME: + case ISCSI_HOST_PARAM_HWADDRESS: + case ISCSI_HOST_PARAM_INITIATOR_NAME: + return S_IRUGO; + default: + return 0; + } + case ISCSI_PARAM: + switch (param) { + case ISCSI_PARAM_MAX_RECV_DLENGTH: + case ISCSI_PARAM_MAX_XMIT_DLENGTH: + case ISCSI_PARAM_HDRDGST_EN: + case ISCSI_PARAM_DATADGST_EN: + case ISCSI_PARAM_CONN_ADDRESS: + case ISCSI_PARAM_CONN_PORT: + case ISCSI_PARAM_EXP_STATSN: + case ISCSI_PARAM_PERSISTENT_ADDRESS: + case ISCSI_PARAM_PERSISTENT_PORT: + case ISCSI_PARAM_PING_TMO: + case ISCSI_PARAM_RECV_TMO: + case ISCSI_PARAM_INITIAL_R2T_EN: + case ISCSI_PARAM_MAX_R2T: + case ISCSI_PARAM_IMM_DATA_EN: + case ISCSI_PARAM_FIRST_BURST: + case ISCSI_PARAM_MAX_BURST: + case ISCSI_PARAM_PDU_INORDER_EN: + case ISCSI_PARAM_DATASEQ_INORDER_EN: + case ISCSI_PARAM_TARGET_NAME: + case ISCSI_PARAM_TPGT: + case ISCSI_PARAM_USERNAME: + case ISCSI_PARAM_PASSWORD: + case ISCSI_PARAM_USERNAME_IN: + case ISCSI_PARAM_PASSWORD_IN: + case ISCSI_PARAM_FAST_ABORT: + case ISCSI_PARAM_ABORT_TMO: + case ISCSI_PARAM_LU_RESET_TMO: + case ISCSI_PARAM_TGT_RESET_TMO: + case ISCSI_PARAM_IFACE_NAME: + case ISCSI_PARAM_INITIATOR_NAME: + return S_IRUGO; + default: + return 0; + } + } + + return 0; +} + static struct scsi_host_template iscsi_iser_sht = { .module = THIS_MODULE, .name = "iSCSI Initiator over iSER, v." DRV_VER, @@ -639,32 +702,6 @@ static struct iscsi_transport iscsi_iser_transport = { .owner = THIS_MODULE, .name = "iser", .caps = CAP_RECOVERY_L0 | CAP_MULTI_R2T, - .param_mask = ISCSI_MAX_RECV_DLENGTH | - ISCSI_MAX_XMIT_DLENGTH | - ISCSI_HDRDGST_EN | - ISCSI_DATADGST_EN | - ISCSI_INITIAL_R2T_EN | - ISCSI_MAX_R2T | - ISCSI_IMM_DATA_EN | - ISCSI_FIRST_BURST | - ISCSI_MAX_BURST | - ISCSI_PDU_INORDER_EN | - ISCSI_DATASEQ_INORDER_EN | - ISCSI_CONN_PORT | - ISCSI_CONN_ADDRESS | - ISCSI_EXP_STATSN | - ISCSI_PERSISTENT_PORT | - ISCSI_PERSISTENT_ADDRESS | - ISCSI_TARGET_NAME | ISCSI_TPGT | - ISCSI_USERNAME | ISCSI_PASSWORD | - ISCSI_USERNAME_IN | ISCSI_PASSWORD_IN | - ISCSI_FAST_ABORT | ISCSI_ABORT_TMO | - ISCSI_LU_RESET_TMO | ISCSI_TGT_RESET_TMO | - ISCSI_PING_TMO | ISCSI_RECV_TMO | - ISCSI_IFACE_NAME | ISCSI_INITIATOR_NAME, - .host_param_mask = ISCSI_HOST_HWADDRESS | - ISCSI_HOST_NETDEV_NAME | - ISCSI_HOST_INITIATOR_NAME, /* session management */ .create_session = iscsi_iser_session_create, .destroy_session = iscsi_iser_session_destroy, @@ -672,6 +709,7 @@ static struct iscsi_transport iscsi_iser_transport = { .create_conn = iscsi_iser_conn_create, .bind_conn = iscsi_iser_conn_bind, .destroy_conn = iscsi_iser_conn_destroy, + .attr_is_visible = iser_attr_is_visible, .set_param = iscsi_iser_set_param, .get_conn_param = iscsi_conn_get_param, .get_ep_param = iscsi_iser_get_ep_param, diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h index 634aef0..296be43 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.h +++ b/drivers/infiniband/ulp/iser/iscsi_iser.h @@ -45,6 +45,7 @@ #include <scsi/libiscsi.h> #include <scsi/scsi_transport_iscsi.h> +#include <linux/interrupt.h> #include <linux/wait.h> #include <linux/sched.h> #include <linux/list.h> @@ -88,7 +89,7 @@ } while (0) #define SHIFT_4K 12 -#define SIZE_4K (1UL << SHIFT_4K) +#define SIZE_4K (1ULL << SHIFT_4K) #define MASK_4K (~(SIZE_4K-1)) /* support up to 512KB in one RDMA */ @@ -256,7 +257,8 @@ struct iser_conn { struct list_head conn_list; /* entry in ig conn list */ char *login_buf; - u64 login_dma; + char *login_req_buf, *login_resp_buf; + u64 login_req_dma, login_resp_dma; unsigned int rx_desc_head; struct iser_rx_desc *rx_descs; struct ib_recv_wr rx_wr[ISER_MIN_POSTED_RX]; @@ -276,7 +278,6 @@ struct iscsi_iser_task { struct iser_regd_buf rdma_regd[ISER_DIRS_NUM];/* regd rdma buf */ struct iser_data_buf data[ISER_DIRS_NUM]; /* orig. data des*/ struct iser_data_buf data_copy[ISER_DIRS_NUM];/* contig. copy */ - int headers_initialized; }; struct iser_page_vec { diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c index eb1ee6f..738a149 100644 --- a/drivers/infiniband/ulp/iser/iser_initiator.c +++ b/drivers/infiniband/ulp/iser/iser_initiator.c @@ -221,8 +221,14 @@ void iser_free_rx_descriptors(struct iser_conn *ib_conn) struct iser_device *device = ib_conn->device; if (ib_conn->login_buf) { - ib_dma_unmap_single(device->ib_device, ib_conn->login_dma, - ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE); + if (ib_conn->login_req_dma) + ib_dma_unmap_single(device->ib_device, + ib_conn->login_req_dma, + ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE); + if (ib_conn->login_resp_dma) + ib_dma_unmap_single(device->ib_device, + ib_conn->login_resp_dma, + ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE); kfree(ib_conn->login_buf); } @@ -272,7 +278,7 @@ int iser_send_command(struct iscsi_conn *conn, unsigned long edtl; int err; struct iser_data_buf *data_buf; - struct iscsi_cmd *hdr = (struct iscsi_cmd *)task->hdr; + struct iscsi_scsi_req *hdr = (struct iscsi_scsi_req *)task->hdr; struct scsi_cmnd *sc = task->sc; struct iser_tx_desc *tx_desc = &iser_task->desc; @@ -395,6 +401,7 @@ int iser_send_control(struct iscsi_conn *conn, unsigned long data_seg_len; int err = 0; struct iser_device *device; + struct iser_conn *ib_conn = iser_conn->ib_conn; /* build the tx desc regd header and add it to the tx desc dto */ mdesc->type = ISCSI_TX_CONTROL; @@ -410,10 +417,20 @@ int iser_send_control(struct iscsi_conn *conn, iser_err("data present on non login task!!!\n"); goto send_control_error; } - memcpy(iser_conn->ib_conn->login_buf, task->data, + + ib_dma_sync_single_for_cpu(device->ib_device, + ib_conn->login_req_dma, task->data_count, + DMA_TO_DEVICE); + + memcpy(iser_conn->ib_conn->login_req_buf, task->data, task->data_count); - tx_dsg->addr = iser_conn->ib_conn->login_dma; - tx_dsg->length = data_seg_len; + + ib_dma_sync_single_for_device(device->ib_device, + ib_conn->login_req_dma, task->data_count, + DMA_TO_DEVICE); + + tx_dsg->addr = iser_conn->ib_conn->login_req_dma; + tx_dsg->length = task->data_count; tx_dsg->lkey = device->mr->lkey; mdesc->num_sge = 2; } @@ -449,8 +466,8 @@ void iser_rcv_completion(struct iser_rx_desc *rx_desc, int rx_buflen, outstanding, count, err; /* differentiate between login to all other PDUs */ - if ((char *)rx_desc == ib_conn->login_buf) { - rx_dma = ib_conn->login_dma; + if ((char *)rx_desc == ib_conn->login_resp_buf) { + rx_dma = ib_conn->login_resp_dma; rx_buflen = ISER_RX_LOGIN_SIZE; } else { rx_dma = rx_desc->dma_addr; @@ -477,7 +494,7 @@ void iser_rcv_completion(struct iser_rx_desc *rx_desc, * for the posted rx bufs refcount to become zero handles everything */ conn->ib_conn->post_recv_buf_count--; - if (rx_dma == ib_conn->login_dma) + if (rx_dma == ib_conn->login_resp_dma) return; outstanding = ib_conn->post_recv_buf_count; diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index ede1475..e28877c 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -155,20 +155,39 @@ static int iser_create_ib_conn_res(struct iser_conn *ib_conn) { struct iser_device *device; struct ib_qp_init_attr init_attr; - int ret = -ENOMEM; + int req_err, resp_err, ret = -ENOMEM; struct ib_fmr_pool_param params; BUG_ON(ib_conn->device == NULL); device = ib_conn->device; - ib_conn->login_buf = kmalloc(ISER_RX_LOGIN_SIZE, GFP_KERNEL); + ib_conn->login_buf = kmalloc(ISCSI_DEF_MAX_RECV_SEG_LEN + + ISER_RX_LOGIN_SIZE, GFP_KERNEL); if (!ib_conn->login_buf) goto out_err; - ib_conn->login_dma = ib_dma_map_single(ib_conn->device->ib_device, - (void *)ib_conn->login_buf, ISER_RX_LOGIN_SIZE, - DMA_FROM_DEVICE); + ib_conn->login_req_buf = ib_conn->login_buf; + ib_conn->login_resp_buf = ib_conn->login_buf + ISCSI_DEF_MAX_RECV_SEG_LEN; + + ib_conn->login_req_dma = ib_dma_map_single(ib_conn->device->ib_device, + (void *)ib_conn->login_req_buf, + ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE); + + ib_conn->login_resp_dma = ib_dma_map_single(ib_conn->device->ib_device, + (void *)ib_conn->login_resp_buf, + ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE); + + req_err = ib_dma_mapping_error(device->ib_device, ib_conn->login_req_dma); + resp_err = ib_dma_mapping_error(device->ib_device, ib_conn->login_resp_dma); + + if (req_err || resp_err) { + if (req_err) + ib_conn->login_req_dma = 0; + if (resp_err) + ib_conn->login_resp_dma = 0; + goto out_err; + } ib_conn->page_vec = kmalloc(sizeof(struct iser_page_vec) + (sizeof(u64) * (ISCSI_ISER_SG_TABLESIZE +1)), @@ -658,11 +677,11 @@ int iser_post_recvl(struct iser_conn *ib_conn) struct ib_sge sge; int ib_ret; - sge.addr = ib_conn->login_dma; + sge.addr = ib_conn->login_resp_dma; sge.length = ISER_RX_LOGIN_SIZE; sge.lkey = ib_conn->device->mr->lkey; - rx_wr.wr_id = (unsigned long)ib_conn->login_buf; + rx_wr.wr_id = (unsigned long)ib_conn->login_resp_buf; rx_wr.sg_list = &sge; rx_wr.num_sge = 1; rx_wr.next = NULL; diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index aa5eafa..8770d44 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -39,7 +39,7 @@ #include <linux/random.h> #include <linux/jiffies.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include <scsi/scsi.h> #include <scsi/scsi_device.h> @@ -1353,6 +1353,12 @@ err_unmap: err_iu: srp_put_tx_iu(target, iu, SRP_IU_CMD); + /* + * Avoid that the loops that iterate over the request ring can + * encounter a dangling SCSI command pointer. + */ + req->scmnd = NULL; + spin_lock_irqsave(&target->lock, flags); list_add(&req->list, &target->free_reqs); |