diff options
Diffstat (limited to 'drivers/infiniband/hw')
44 files changed, 791 insertions, 458 deletions
diff --git a/drivers/infiniband/hw/amso1100/c2_provider.c b/drivers/infiniband/hw/amso1100/c2_provider.c index 69580e282af0..5119d6508181 100644 --- a/drivers/infiniband/hw/amso1100/c2_provider.c +++ b/drivers/infiniband/hw/amso1100/c2_provider.c @@ -653,7 +653,7 @@ static int c2_service_destroy(struct iw_cm_id *cm_id) static int c2_pseudo_up(struct net_device *netdev) { struct in_device *ind; - struct c2_dev *c2dev = netdev->priv; + struct c2_dev *c2dev = netdev->ml_priv; ind = in_dev_get(netdev); if (!ind) @@ -678,7 +678,7 @@ static int c2_pseudo_up(struct net_device *netdev) static int c2_pseudo_down(struct net_device *netdev) { struct in_device *ind; - struct c2_dev *c2dev = netdev->priv; + struct c2_dev *c2dev = netdev->ml_priv; ind = in_dev_get(netdev); if (!ind) @@ -746,14 +746,14 @@ static struct net_device *c2_pseudo_netdev_init(struct c2_dev *c2dev) /* change ethxxx to iwxxx */ strcpy(name, "iw"); strcat(name, &c2dev->netdev->name[3]); - netdev = alloc_netdev(sizeof(*netdev), name, setup); + netdev = alloc_netdev(0, name, setup); if (!netdev) { printk(KERN_ERR PFX "%s - etherdev alloc failed", __func__); return NULL; } - netdev->priv = c2dev; + netdev->ml_priv = c2dev; SET_NETDEV_DEV(netdev, &c2dev->pcidev->dev); diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c index c325c44807e8..44e936e48a31 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_cm.c +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c @@ -1942,6 +1942,7 @@ fail4: fail3: cxgb3_free_atid(ep->com.tdev, ep->atid); fail2: + cm_id->rem_ref(cm_id); put_ep(&ep->com); out: return err; diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index ecff98043589..160ef482712d 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -1102,9 +1102,7 @@ static u64 fw_vers_string_to_u64(struct iwch_dev *iwch_dev) char *cp, *next; unsigned fw_maj, fw_min, fw_mic; - rtnl_lock(); lldev->ethtool_ops->get_drvinfo(lldev, &info); - rtnl_unlock(); next = info.fw_version + 1; cp = strsep(&next, "."); @@ -1192,9 +1190,7 @@ static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, ch struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev; PDBG("%s dev 0x%p\n", __func__, dev); - rtnl_lock(); lldev->ethtool_ops->get_drvinfo(lldev, &info); - rtnl_unlock(); return sprintf(buf, "%s\n", info.fw_version); } @@ -1207,9 +1203,7 @@ static ssize_t show_hca(struct device *dev, struct device_attribute *attr, struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev; PDBG("%s dev 0x%p\n", __func__, dev); - rtnl_lock(); lldev->ethtool_ops->get_drvinfo(lldev, &info); - rtnl_unlock(); return sprintf(buf, "%s\n", info.driver); } diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c index 3e4585c2318a..19661b2f0406 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_qp.c +++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c @@ -745,7 +745,6 @@ int iwch_post_zb_read(struct iwch_qp *qhp) wqe->read.rdmaop = T3_READ_REQ; wqe->read.reserved[0] = 0; wqe->read.reserved[1] = 0; - wqe->read.reserved[2] = 0; wqe->read.rem_stag = cpu_to_be32(1); wqe->read.rem_to = cpu_to_be64(1); wqe->read.local_stag = cpu_to_be32(1); diff --git a/drivers/infiniband/hw/ehca/ehca_classes.h b/drivers/infiniband/hw/ehca/ehca_classes.h index 5d7b7855afb9..c825142a2fb7 100644 --- a/drivers/infiniband/hw/ehca/ehca_classes.h +++ b/drivers/infiniband/hw/ehca/ehca_classes.h @@ -128,6 +128,8 @@ struct ehca_shca { /* MR pgsize: bit 0-3 means 4K, 64K, 1M, 16M respectively */ u32 hca_cap_mr_pgsize; int max_mtu; + int max_num_qps; + int max_num_cqs; atomic_t num_cqs; atomic_t num_qps; }; @@ -161,7 +163,8 @@ struct ehca_mod_qp_parm { /* struct for tracking if cqes have been reported to the application */ struct ehca_qmap_entry { u16 app_wr_id; - u16 reported; + u8 reported; + u8 cqe_req; }; struct ehca_queue_map { @@ -169,8 +172,16 @@ struct ehca_queue_map { unsigned int entries; unsigned int tail; unsigned int left_to_poll; + unsigned int next_wqe_idx; /* Idx to first wqe to be flushed */ }; +/* function to calculate the next index for the qmap */ +static inline unsigned int next_index(unsigned int cur_index, unsigned int limit) +{ + unsigned int temp = cur_index + 1; + return (temp == limit) ? 0 : temp; +} + struct ehca_qp { union { struct ib_qp ib_qp; diff --git a/drivers/infiniband/hw/ehca/ehca_cq.c b/drivers/infiniband/hw/ehca/ehca_cq.c index 33647a95eb9a..2f4c28a30271 100644 --- a/drivers/infiniband/hw/ehca/ehca_cq.c +++ b/drivers/infiniband/hw/ehca/ehca_cq.c @@ -132,9 +132,9 @@ struct ib_cq *ehca_create_cq(struct ib_device *device, int cqe, int comp_vector, if (cqe >= 0xFFFFFFFF - 64 - additional_cqe) return ERR_PTR(-EINVAL); - if (!atomic_add_unless(&shca->num_cqs, 1, ehca_max_cq)) { + if (!atomic_add_unless(&shca->num_cqs, 1, shca->max_num_cqs)) { ehca_err(device, "Unable to create CQ, max number of %i " - "CQs reached.", ehca_max_cq); + "CQs reached.", shca->max_num_cqs); ehca_err(device, "To increase the maximum number of CQs " "use the number_of_cqs module parameter.\n"); return ERR_PTR(-ENOSPC); diff --git a/drivers/infiniband/hw/ehca/ehca_eq.c b/drivers/infiniband/hw/ehca/ehca_eq.c index 49660dfa1867..523e733c630e 100644 --- a/drivers/infiniband/hw/ehca/ehca_eq.c +++ b/drivers/infiniband/hw/ehca/ehca_eq.c @@ -113,7 +113,7 @@ int ehca_create_eq(struct ehca_shca *shca, if (h_ret != H_SUCCESS || vpage) goto create_eq_exit2; } else { - if (h_ret != H_PAGE_REGISTERED || !vpage) + if (h_ret != H_PAGE_REGISTERED) goto create_eq_exit2; } } diff --git a/drivers/infiniband/hw/ehca/ehca_irq.c b/drivers/infiniband/hw/ehca/ehca_irq.c index cb55be04442c..757035ea246f 100644 --- a/drivers/infiniband/hw/ehca/ehca_irq.c +++ b/drivers/infiniband/hw/ehca/ehca_irq.c @@ -359,36 +359,48 @@ static void notify_port_conf_change(struct ehca_shca *shca, int port_num) *old_attr = new_attr; } +/* replay modify_qp for sqps -- return 0 if all is well, 1 if AQP1 destroyed */ +static int replay_modify_qp(struct ehca_sport *sport) +{ + int aqp1_destroyed; + unsigned long flags; + + spin_lock_irqsave(&sport->mod_sqp_lock, flags); + + aqp1_destroyed = !sport->ibqp_sqp[IB_QPT_GSI]; + + if (sport->ibqp_sqp[IB_QPT_SMI]) + ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_SMI]); + if (!aqp1_destroyed) + ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_GSI]); + + spin_unlock_irqrestore(&sport->mod_sqp_lock, flags); + + return aqp1_destroyed; +} + static void parse_ec(struct ehca_shca *shca, u64 eqe) { u8 ec = EHCA_BMASK_GET(NEQE_EVENT_CODE, eqe); u8 port = EHCA_BMASK_GET(NEQE_PORT_NUMBER, eqe); u8 spec_event; struct ehca_sport *sport = &shca->sport[port - 1]; - unsigned long flags; switch (ec) { case 0x30: /* port availability change */ if (EHCA_BMASK_GET(NEQE_PORT_AVAILABILITY, eqe)) { - int suppress_event; - /* replay modify_qp for sqps */ - spin_lock_irqsave(&sport->mod_sqp_lock, flags); - suppress_event = !sport->ibqp_sqp[IB_QPT_GSI]; - if (sport->ibqp_sqp[IB_QPT_SMI]) - ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_SMI]); - if (!suppress_event) - ehca_recover_sqp(sport->ibqp_sqp[IB_QPT_GSI]); - spin_unlock_irqrestore(&sport->mod_sqp_lock, flags); - - /* AQP1 was destroyed, ignore this event */ - if (suppress_event) - break; + /* only replay modify_qp calls in autodetect mode; + * if AQP1 was destroyed, the port is already down + * again and we can drop the event. + */ + if (ehca_nr_ports < 0) + if (replay_modify_qp(sport)) + break; sport->port_state = IB_PORT_ACTIVE; dispatch_port_event(shca, port, IB_EVENT_PORT_ACTIVE, "is active"); - ehca_query_sma_attr(shca, port, - &sport->saved_attr); + ehca_query_sma_attr(shca, port, &sport->saved_attr); } else { sport->port_state = IB_PORT_DOWN; dispatch_port_event(shca, port, IB_EVENT_PORT_ERR, diff --git a/drivers/infiniband/hw/ehca/ehca_main.c b/drivers/infiniband/hw/ehca/ehca_main.c index 598844d2edc9..3b77b674cbf6 100644 --- a/drivers/infiniband/hw/ehca/ehca_main.c +++ b/drivers/infiniband/hw/ehca/ehca_main.c @@ -44,6 +44,8 @@ #include <linux/slab.h> #endif +#include <linux/notifier.h> +#include <linux/memory.h> #include "ehca_classes.h" #include "ehca_iverbs.h" #include "ehca_mrmw.h" @@ -366,22 +368,23 @@ static int ehca_sense_attributes(struct ehca_shca *shca) shca->hca_cap_mr_pgsize |= pgsize_map[i + 1]; /* Set maximum number of CQs and QPs to calculate EQ size */ - if (ehca_max_qp == -1) - ehca_max_qp = min_t(int, rblock->max_qp, EHCA_MAX_NUM_QUEUES); - else if (ehca_max_qp < 1 || ehca_max_qp > rblock->max_qp) { - ehca_gen_err("Requested number of QPs is out of range (1 - %i) " - "specified by HW", rblock->max_qp); - ret = -EINVAL; - goto sense_attributes1; + if (shca->max_num_qps == -1) + shca->max_num_qps = min_t(int, rblock->max_qp, + EHCA_MAX_NUM_QUEUES); + else if (shca->max_num_qps < 1 || shca->max_num_qps > rblock->max_qp) { + ehca_gen_warn("The requested number of QPs is out of range " + "(1 - %i) specified by HW. Value is set to %i", + rblock->max_qp, rblock->max_qp); + shca->max_num_qps = rblock->max_qp; } - if (ehca_max_cq == -1) - ehca_max_cq = min_t(int, rblock->max_cq, EHCA_MAX_NUM_QUEUES); - else if (ehca_max_cq < 1 || ehca_max_cq > rblock->max_cq) { - ehca_gen_err("Requested number of CQs is out of range (1 - %i) " - "specified by HW", rblock->max_cq); - ret = -EINVAL; - goto sense_attributes1; + if (shca->max_num_cqs == -1) + shca->max_num_cqs = min_t(int, rblock->max_cq, + EHCA_MAX_NUM_QUEUES); + else if (shca->max_num_cqs < 1 || shca->max_num_cqs > rblock->max_cq) { + ehca_gen_warn("The requested number of CQs is out of range " + "(1 - %i) specified by HW. Value is set to %i", + rblock->max_cq, rblock->max_cq); } /* query max MTU from first port -- it's the same for all ports */ @@ -714,6 +717,7 @@ static int __devinit ehca_probe(struct of_device *dev, const u64 *handle; struct ib_pd *ibpd; int ret, i, eq_size; + unsigned long flags; handle = of_get_property(dev->node, "ibm,hca-handle", NULL); if (!handle) { @@ -733,9 +737,13 @@ static int __devinit ehca_probe(struct of_device *dev, ehca_gen_err("Cannot allocate shca memory."); return -ENOMEM; } + mutex_init(&shca->modify_mutex); atomic_set(&shca->num_cqs, 0); atomic_set(&shca->num_qps, 0); + shca->max_num_qps = ehca_max_qp; + shca->max_num_cqs = ehca_max_cq; + for (i = 0; i < ARRAY_SIZE(shca->sport); i++) spin_lock_init(&shca->sport[i].mod_sqp_lock); @@ -755,7 +763,7 @@ static int __devinit ehca_probe(struct of_device *dev, goto probe1; } - eq_size = 2 * ehca_max_cq + 4 * ehca_max_qp; + eq_size = 2 * shca->max_num_cqs + 4 * shca->max_num_qps; /* create event queues */ ret = ehca_create_eq(shca, &shca->eq, EHCA_EQ, eq_size); if (ret) { @@ -823,9 +831,9 @@ static int __devinit ehca_probe(struct of_device *dev, ehca_err(&shca->ib_device, "Cannot create device attributes ret=%d", ret); - spin_lock(&shca_list_lock); + spin_lock_irqsave(&shca_list_lock, flags); list_add(&shca->shca_list, &shca_list); - spin_unlock(&shca_list_lock); + spin_unlock_irqrestore(&shca_list_lock, flags); return 0; @@ -871,6 +879,7 @@ probe1: static int __devexit ehca_remove(struct of_device *dev) { struct ehca_shca *shca = dev->dev.driver_data; + unsigned long flags; int ret; sysfs_remove_group(&dev->dev.kobj, &ehca_dev_attr_grp); @@ -908,9 +917,9 @@ static int __devexit ehca_remove(struct of_device *dev) ib_dealloc_device(&shca->ib_device); - spin_lock(&shca_list_lock); + spin_lock_irqsave(&shca_list_lock, flags); list_del(&shca->shca_list); - spin_unlock(&shca_list_lock); + spin_unlock_irqrestore(&shca_list_lock, flags); return ret; } @@ -964,6 +973,41 @@ void ehca_poll_eqs(unsigned long data) spin_unlock(&shca_list_lock); } +static int ehca_mem_notifier(struct notifier_block *nb, + unsigned long action, void *data) +{ + static unsigned long ehca_dmem_warn_time; + unsigned long flags; + + switch (action) { + case MEM_CANCEL_OFFLINE: + case MEM_CANCEL_ONLINE: + case MEM_ONLINE: + case MEM_OFFLINE: + return NOTIFY_OK; + case MEM_GOING_ONLINE: + case MEM_GOING_OFFLINE: + /* only ok if no hca is attached to the lpar */ + spin_lock_irqsave(&shca_list_lock, flags); + if (list_empty(&shca_list)) { + spin_unlock_irqrestore(&shca_list_lock, flags); + return NOTIFY_OK; + } else { + spin_unlock_irqrestore(&shca_list_lock, flags); + if (printk_timed_ratelimit(&ehca_dmem_warn_time, + 30 * 1000)) + ehca_gen_err("DMEM operations are not allowed" + "in conjunction with eHCA"); + return NOTIFY_BAD; + } + } + return NOTIFY_OK; +} + +static struct notifier_block ehca_mem_nb = { + .notifier_call = ehca_mem_notifier, +}; + static int __init ehca_module_init(void) { int ret; @@ -991,6 +1035,12 @@ static int __init ehca_module_init(void) goto module_init2; } + ret = register_memory_notifier(&ehca_mem_nb); + if (ret) { + ehca_gen_err("Failed registering memory add/remove notifier"); + goto module_init3; + } + if (ehca_poll_all_eqs != 1) { ehca_gen_err("WARNING!!!"); ehca_gen_err("It is possible to lose interrupts."); @@ -1003,6 +1053,9 @@ static int __init ehca_module_init(void) return 0; +module_init3: + ibmebus_unregister_driver(&ehca_driver); + module_init2: ehca_destroy_slab_caches(); @@ -1018,6 +1071,8 @@ static void __exit ehca_module_exit(void) ibmebus_unregister_driver(&ehca_driver); + unregister_memory_notifier(&ehca_mem_nb); + ehca_destroy_slab_caches(); ehca_destroy_comp_pool(); diff --git a/drivers/infiniband/hw/ehca/ehca_qp.c b/drivers/infiniband/hw/ehca/ehca_qp.c index 4dbe2870e014..f161cf173dbe 100644 --- a/drivers/infiniband/hw/ehca/ehca_qp.c +++ b/drivers/infiniband/hw/ehca/ehca_qp.c @@ -435,9 +435,13 @@ static void reset_queue_map(struct ehca_queue_map *qmap) { int i; - qmap->tail = 0; - for (i = 0; i < qmap->entries; i++) + qmap->tail = qmap->entries - 1; + qmap->left_to_poll = 0; + qmap->next_wqe_idx = 0; + for (i = 0; i < qmap->entries; i++) { qmap->map[i].reported = 1; + qmap->map[i].cqe_req = 0; + } } /* @@ -465,9 +469,9 @@ static struct ehca_qp *internal_create_qp( u32 swqe_size = 0, rwqe_size = 0, ib_qp_num; unsigned long flags; - if (!atomic_add_unless(&shca->num_qps, 1, ehca_max_qp)) { + if (!atomic_add_unless(&shca->num_qps, 1, shca->max_num_qps)) { ehca_err(pd->device, "Unable to create QP, max number of %i " - "QPs reached.", ehca_max_qp); + "QPs reached.", shca->max_num_qps); ehca_err(pd->device, "To increase the maximum number of QPs " "use the number_of_qps module parameter.\n"); return ERR_PTR(-ENOSPC); @@ -502,6 +506,12 @@ static struct ehca_qp *internal_create_qp( if (init_attr->srq) { my_srq = container_of(init_attr->srq, struct ehca_qp, ib_srq); + if (qp_type == IB_QPT_UC) { + ehca_err(pd->device, "UC with SRQ not supported"); + atomic_dec(&shca->num_qps); + return ERR_PTR(-EINVAL); + } + has_srq = 1; parms.ext_type = EQPT_SRQBASE; parms.srq_qpn = my_srq->real_qp_num; @@ -854,6 +864,11 @@ static struct ehca_qp *internal_create_qp( if (qp_type == IB_QPT_GSI) { h_ret = ehca_define_sqp(shca, my_qp, init_attr); if (h_ret != H_SUCCESS) { + kfree(my_qp->mod_qp_parm); + my_qp->mod_qp_parm = NULL; + /* the QP pointer is no longer valid */ + shca->sport[init_attr->port_num - 1].ibqp_sqp[qp_type] = + NULL; ret = ehca2ib_return_code(h_ret); goto create_qp_exit6; } @@ -1110,6 +1125,7 @@ static int calc_left_cqes(u64 wqe_p, struct ipz_queue *ipz_queue, void *wqe_v; u64 q_ofs; u32 wqe_idx; + unsigned int tail_idx; /* convert real to abs address */ wqe_p = wqe_p & (~(1UL << 63)); @@ -1122,12 +1138,17 @@ static int calc_left_cqes(u64 wqe_p, struct ipz_queue *ipz_queue, return -EFAULT; } + tail_idx = next_index(qmap->tail, qmap->entries); wqe_idx = q_ofs / ipz_queue->qe_size; - if (wqe_idx < qmap->tail) - qmap->left_to_poll = (qmap->entries - qmap->tail) + wqe_idx; - else - qmap->left_to_poll = wqe_idx - qmap->tail; + /* check all processed wqes, whether a cqe is requested or not */ + while (tail_idx != wqe_idx) { + if (qmap->map[tail_idx].cqe_req) + qmap->left_to_poll++; + tail_idx = next_index(tail_idx, qmap->entries); + } + /* save index in queue, where we have to start flushing */ + qmap->next_wqe_idx = wqe_idx; return 0; } @@ -1174,10 +1195,14 @@ static int check_for_left_cqes(struct ehca_qp *my_qp, struct ehca_shca *shca) } else { spin_lock_irqsave(&my_qp->send_cq->spinlock, flags); my_qp->sq_map.left_to_poll = 0; + my_qp->sq_map.next_wqe_idx = next_index(my_qp->sq_map.tail, + my_qp->sq_map.entries); spin_unlock_irqrestore(&my_qp->send_cq->spinlock, flags); spin_lock_irqsave(&my_qp->recv_cq->spinlock, flags); my_qp->rq_map.left_to_poll = 0; + my_qp->rq_map.next_wqe_idx = next_index(my_qp->rq_map.tail, + my_qp->rq_map.entries); spin_unlock_irqrestore(&my_qp->recv_cq->spinlock, flags); } diff --git a/drivers/infiniband/hw/ehca/ehca_reqs.c b/drivers/infiniband/hw/ehca/ehca_reqs.c index 64928079eafa..c7112686782f 100644 --- a/drivers/infiniband/hw/ehca/ehca_reqs.c +++ b/drivers/infiniband/hw/ehca/ehca_reqs.c @@ -179,6 +179,7 @@ static inline int ehca_write_swqe(struct ehca_qp *qp, qmap_entry->app_wr_id = get_app_wr_id(send_wr->wr_id); qmap_entry->reported = 0; + qmap_entry->cqe_req = 0; switch (send_wr->opcode) { case IB_WR_SEND: @@ -203,8 +204,10 @@ static inline int ehca_write_swqe(struct ehca_qp *qp, if ((send_wr->send_flags & IB_SEND_SIGNALED || qp->init_attr.sq_sig_type == IB_SIGNAL_ALL_WR) - && !hidden) + && !hidden) { wqe_p->wr_flag |= WQE_WRFLAG_REQ_SIGNAL_COM; + qmap_entry->cqe_req = 1; + } if (send_wr->opcode == IB_WR_SEND_WITH_IMM || send_wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM) { @@ -569,6 +572,7 @@ static int internal_post_recv(struct ehca_qp *my_qp, qmap_entry = &my_qp->rq_map.map[rq_map_idx]; qmap_entry->app_wr_id = get_app_wr_id(cur_recv_wr->wr_id); qmap_entry->reported = 0; + qmap_entry->cqe_req = 1; wqe_cnt++; } /* eof for cur_recv_wr */ @@ -706,27 +710,34 @@ repoll: goto repoll; wc->qp = &my_qp->ib_qp; + qmap_tail_idx = get_app_wr_id(cqe->work_request_id); + if (!(cqe->w_completion_flags & WC_SEND_RECEIVE_BIT)) + /* We got a send completion. */ + qmap = &my_qp->sq_map; + else + /* We got a receive completion. */ + qmap = &my_qp->rq_map; + + /* advance the tail pointer */ + qmap->tail = qmap_tail_idx; + if (is_error) { /* * set left_to_poll to 0 because in error state, we will not * get any additional CQEs */ - ehca_add_to_err_list(my_qp, 1); + my_qp->sq_map.next_wqe_idx = next_index(my_qp->sq_map.tail, + my_qp->sq_map.entries); my_qp->sq_map.left_to_poll = 0; + ehca_add_to_err_list(my_qp, 1); + my_qp->rq_map.next_wqe_idx = next_index(my_qp->rq_map.tail, + my_qp->rq_map.entries); + my_qp->rq_map.left_to_poll = 0; if (HAS_RQ(my_qp)) ehca_add_to_err_list(my_qp, 0); - my_qp->rq_map.left_to_poll = 0; } - qmap_tail_idx = get_app_wr_id(cqe->work_request_id); - if (!(cqe->w_completion_flags & WC_SEND_RECEIVE_BIT)) - /* We got a send completion. */ - qmap = &my_qp->sq_map; - else - /* We got a receive completion. */ - qmap = &my_qp->rq_map; - qmap_entry = &qmap->map[qmap_tail_idx]; if (qmap_entry->reported) { ehca_warn(cq->device, "Double cqe on qp_num=%#x", @@ -738,10 +749,6 @@ repoll: wc->wr_id = replace_wr_id(cqe->work_request_id, qmap_entry->app_wr_id); qmap_entry->reported = 1; - /* this is a proper completion, we need to advance the tail pointer */ - if (++qmap->tail == qmap->entries) - qmap->tail = 0; - /* if left_to_poll is decremented to 0, add the QP to the error list */ if (qmap->left_to_poll > 0) { qmap->left_to_poll--; @@ -805,13 +812,14 @@ static int generate_flush_cqes(struct ehca_qp *my_qp, struct ib_cq *cq, else qmap = &my_qp->rq_map; - qmap_entry = &qmap->map[qmap->tail]; + qmap_entry = &qmap->map[qmap->next_wqe_idx]; while ((nr < num_entries) && (qmap_entry->reported == 0)) { /* generate flush CQE */ + memset(wc, 0, sizeof(*wc)); - offset = qmap->tail * ipz_queue->qe_size; + offset = qmap->next_wqe_idx * ipz_queue->qe_size; wqe = (struct ehca_wqe *)ipz_qeit_calc(ipz_queue, offset); if (!wqe) { ehca_err(cq->device, "Invalid wqe offset=%#lx on " @@ -850,11 +858,11 @@ static int generate_flush_cqes(struct ehca_qp *my_qp, struct ib_cq *cq, wc->qp = &my_qp->ib_qp; - /* mark as reported and advance tail pointer */ + /* mark as reported and advance next_wqe pointer */ qmap_entry->reported = 1; - if (++qmap->tail == qmap->entries) - qmap->tail = 0; - qmap_entry = &qmap->map[qmap->tail]; + qmap->next_wqe_idx = next_index(qmap->next_wqe_idx, + qmap->entries); + qmap_entry = &qmap->map[qmap->next_wqe_idx]; wc++; nr++; } diff --git a/drivers/infiniband/hw/ipath/ipath_driver.c b/drivers/infiniband/hw/ipath/ipath_driver.c index ad0aab60b051..69c0ce321b4e 100644 --- a/drivers/infiniband/hw/ipath/ipath_driver.c +++ b/drivers/infiniband/hw/ipath/ipath_driver.c @@ -661,6 +661,8 @@ bail: static void __devexit cleanup_device(struct ipath_devdata *dd) { int port; + struct ipath_portdata **tmp; + unsigned long flags; if (*dd->ipath_statusp & IPATH_STATUS_CHIP_PRESENT) { /* can't do anything more with chip; needs re-init */ @@ -742,20 +744,21 @@ static void __devexit cleanup_device(struct ipath_devdata *dd) /* * free any resources still in use (usually just kernel ports) - * at unload; we do for portcnt, not cfgports, because cfgports - * could have changed while we were loaded. + * at unload; we do for portcnt, because that's what we allocate. + * We acquire lock to be really paranoid that ipath_pd isn't being + * accessed from some interrupt-related code (that should not happen, + * but best to be sure). */ + spin_lock_irqsave(&dd->ipath_uctxt_lock, flags); + tmp = dd->ipath_pd; + dd->ipath_pd = NULL; + spin_unlock_irqrestore(&dd->ipath_uctxt_lock, flags); for (port = 0; port < dd->ipath_portcnt; port++) { - struct ipath_portdata *pd = dd->ipath_pd[port]; - dd->ipath_pd[port] = NULL; + struct ipath_portdata *pd = tmp[port]; + tmp[port] = NULL; /* debugging paranoia */ ipath_free_pddata(dd, pd); } - kfree(dd->ipath_pd); - /* - * debuggability, in case some cleanup path tries to use it - * after this - */ - dd->ipath_pd = NULL; + kfree(tmp); } static void __devexit ipath_remove_one(struct pci_dev *pdev) @@ -2586,6 +2589,7 @@ int ipath_reset_device(int unit) { int ret, i; struct ipath_devdata *dd = ipath_lookup(unit); + unsigned long flags; if (!dd) { ret = -ENODEV; @@ -2611,18 +2615,21 @@ int ipath_reset_device(int unit) goto bail; } + spin_lock_irqsave(&dd->ipath_uctxt_lock, flags); if (dd->ipath_pd) for (i = 1; i < dd->ipath_cfgports; i++) { - if (dd->ipath_pd[i] && dd->ipath_pd[i]->port_cnt) { - ipath_dbg("unit %u port %d is in use " - "(PID %u cmd %s), can't reset\n", - unit, i, - pid_nr(dd->ipath_pd[i]->port_pid), - dd->ipath_pd[i]->port_comm); - ret = -EBUSY; - goto bail; - } + if (!dd->ipath_pd[i] || !dd->ipath_pd[i]->port_cnt) + continue; + spin_unlock_irqrestore(&dd->ipath_uctxt_lock, flags); + ipath_dbg("unit %u port %d is in use " + "(PID %u cmd %s), can't reset\n", + unit, i, + pid_nr(dd->ipath_pd[i]->port_pid), + dd->ipath_pd[i]->port_comm); + ret = -EBUSY; + goto bail; } + spin_unlock_irqrestore(&dd->ipath_uctxt_lock, flags); if (dd->ipath_flags & IPATH_HAS_SEND_DMA) teardown_sdma(dd); @@ -2656,9 +2663,12 @@ static int ipath_signal_procs(struct ipath_devdata *dd, int sig) { int i, sub, any = 0; struct pid *pid; + unsigned long flags; if (!dd->ipath_pd) return 0; + + spin_lock_irqsave(&dd->ipath_uctxt_lock, flags); for (i = 1; i < dd->ipath_cfgports; i++) { if (!dd->ipath_pd[i] || !dd->ipath_pd[i]->port_cnt) continue; @@ -2682,6 +2692,7 @@ static int ipath_signal_procs(struct ipath_devdata *dd, int sig) any++; } } + spin_unlock_irqrestore(&dd->ipath_uctxt_lock, flags); return any; } diff --git a/drivers/infiniband/hw/ipath/ipath_file_ops.c b/drivers/infiniband/hw/ipath/ipath_file_ops.c index 56c0eda3c077..239d4e8068ac 100644 --- a/drivers/infiniband/hw/ipath/ipath_file_ops.c +++ b/drivers/infiniband/hw/ipath/ipath_file_ops.c @@ -223,8 +223,13 @@ static int ipath_get_base_info(struct file *fp, (unsigned long long) kinfo->spi_subport_rcvhdr_base); } - kinfo->spi_pioindex = (kinfo->spi_piobufbase - dd->ipath_piobufbase) / - dd->ipath_palign; + /* + * All user buffers are 2KB buffers. If we ever support + * giving 4KB buffers to user processes, this will need some + * work. + */ + kinfo->spi_pioindex = (kinfo->spi_piobufbase - + (dd->ipath_piobufbase & 0xffffffff)) / dd->ipath_palign; kinfo->spi_pioalign = dd->ipath_palign; kinfo->spi_qpair = IPATH_KD_QP; @@ -2041,7 +2046,9 @@ static int ipath_close(struct inode *in, struct file *fp) struct ipath_filedata *fd; struct ipath_portdata *pd; struct ipath_devdata *dd; + unsigned long flags; unsigned port; + struct pid *pid; ipath_cdbg(VERBOSE, "close on dev %lx, private data %p\n", (long)in->i_rdev, fp->private_data); @@ -2074,14 +2081,13 @@ static int ipath_close(struct inode *in, struct file *fp) mutex_unlock(&ipath_mutex); goto bail; } + /* early; no interrupt users after this */ + spin_lock_irqsave(&dd->ipath_uctxt_lock, flags); port = pd->port_port; - - if (pd->port_hdrqfull) { - ipath_cdbg(PROC, "%s[%u] had %u rcvhdrqfull errors " - "during run\n", pd->port_comm, pid_nr(pd->port_pid), - pd->port_hdrqfull); - pd->port_hdrqfull = 0; - } + dd->ipath_pd[port] = NULL; + pid = pd->port_pid; + pd->port_pid = NULL; + spin_unlock_irqrestore(&dd->ipath_uctxt_lock, flags); if (pd->port_rcvwait_to || pd->port_piowait_to || pd->port_rcvnowait || pd->port_pionowait) { @@ -2138,13 +2144,11 @@ static int ipath_close(struct inode *in, struct file *fp) unlock_expected_tids(pd); ipath_stats.sps_ports--; ipath_cdbg(PROC, "%s[%u] closed port %u:%u\n", - pd->port_comm, pid_nr(pd->port_pid), + pd->port_comm, pid_nr(pid), dd->ipath_unit, port); } - put_pid(pd->port_pid); - pd->port_pid = NULL; - dd->ipath_pd[pd->port_port] = NULL; /* before releasing mutex */ + put_pid(pid); mutex_unlock(&ipath_mutex); ipath_free_pddata(dd, pd); /* after releasing the mutex */ @@ -2455,7 +2459,7 @@ static int init_cdev(int minor, char *name, const struct file_operations *fops, goto err_cdev; } - device = device_create_drvdata(ipath_class, NULL, dev, NULL, name); + device = device_create(ipath_class, NULL, dev, NULL, name); if (IS_ERR(device)) { ret = PTR_ERR(device); diff --git a/drivers/infiniband/hw/ipath/ipath_fs.c b/drivers/infiniband/hw/ipath/ipath_fs.c index 8bb5170b4e41..53912c327bfe 100644 --- a/drivers/infiniband/hw/ipath/ipath_fs.c +++ b/drivers/infiniband/hw/ipath/ipath_fs.c @@ -86,7 +86,7 @@ static int create_file(const char *name, mode_t mode, *dentry = NULL; mutex_lock(&parent->d_inode->i_mutex); *dentry = lookup_one_len(name, parent, strlen(name)); - if (!IS_ERR(dentry)) + if (!IS_ERR(*dentry)) error = ipathfs_mknod(parent->d_inode, *dentry, mode, fops, data); else diff --git a/drivers/infiniband/hw/ipath/ipath_iba6120.c b/drivers/infiniband/hw/ipath/ipath_iba6120.c index 421cc2af891f..fbf8c5379ea8 100644 --- a/drivers/infiniband/hw/ipath/ipath_iba6120.c +++ b/drivers/infiniband/hw/ipath/ipath_iba6120.c @@ -721,6 +721,12 @@ static int ipath_pe_bringup_serdes(struct ipath_devdata *dd) INFINIPATH_HWE_SERDESPLLFAILED); } + dd->ibdeltainprog = 1; + dd->ibsymsnap = + ipath_read_creg32(dd, dd->ipath_cregs->cr_ibsymbolerrcnt); + dd->iblnkerrsnap = + ipath_read_creg32(dd, dd->ipath_cregs->cr_iblinkerrrecovcnt); + val = ipath_read_kreg64(dd, dd->ipath_kregs->kr_serdesconfig0); config1 = ipath_read_kreg64(dd, dd->ipath_kregs->kr_serdesconfig1); @@ -810,6 +816,36 @@ static void ipath_pe_quiet_serdes(struct ipath_devdata *dd) { u64 val = ipath_read_kreg64(dd, dd->ipath_kregs->kr_serdesconfig0); + if (dd->ibsymdelta || dd->iblnkerrdelta || + dd->ibdeltainprog) { + u64 diagc; + /* enable counter writes */ + diagc = ipath_read_kreg64(dd, dd->ipath_kregs->kr_hwdiagctrl); + ipath_write_kreg(dd, dd->ipath_kregs->kr_hwdiagctrl, + diagc | INFINIPATH_DC_COUNTERWREN); + + if (dd->ibsymdelta || dd->ibdeltainprog) { + val = ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt); + if (dd->ibdeltainprog) + val -= val - dd->ibsymsnap; + val -= dd->ibsymdelta; + ipath_write_creg(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt, val); + } + if (dd->iblnkerrdelta || dd->ibdeltainprog) { + val = ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt); + if (dd->ibdeltainprog) + val -= val - dd->iblnkerrsnap; + val -= dd->iblnkerrdelta; + ipath_write_creg(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt, val); + } + + /* and disable counter writes */ + ipath_write_kreg(dd, dd->ipath_kregs->kr_hwdiagctrl, diagc); + } val |= INFINIPATH_SERDC0_TXIDLE; ipath_dbg("Setting TxIdleEn on serdes (config0 = %llx)\n", (unsigned long long) val); @@ -1749,6 +1785,31 @@ static void ipath_pe_config_jint(struct ipath_devdata *dd, u16 a, u16 b) static int ipath_pe_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) { + if (ibup) { + if (dd->ibdeltainprog) { + dd->ibdeltainprog = 0; + dd->ibsymdelta += + ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt) - + dd->ibsymsnap; + dd->iblnkerrdelta += + ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt) - + dd->iblnkerrsnap; + } + } else { + dd->ipath_lli_counter = 0; + if (!dd->ibdeltainprog) { + dd->ibdeltainprog = 1; + dd->ibsymsnap = + ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt); + dd->iblnkerrsnap = + ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt); + } + } + ipath_setup_pe_setextled(dd, ipath_ib_linkstate(dd, ibcs), ipath_ib_linktrstate(dd, ibcs)); return 0; diff --git a/drivers/infiniband/hw/ipath/ipath_iba7220.c b/drivers/infiniband/hw/ipath/ipath_iba7220.c index 9839e20119bc..b2a9d4c155d1 100644 --- a/drivers/infiniband/hw/ipath/ipath_iba7220.c +++ b/drivers/infiniband/hw/ipath/ipath_iba7220.c @@ -951,6 +951,12 @@ static int ipath_7220_bringup_serdes(struct ipath_devdata *dd) INFINIPATH_HWE_SERDESPLLFAILED); } + dd->ibdeltainprog = 1; + dd->ibsymsnap = + ipath_read_creg32(dd, dd->ipath_cregs->cr_ibsymbolerrcnt); + dd->iblnkerrsnap = + ipath_read_creg32(dd, dd->ipath_cregs->cr_iblinkerrrecovcnt); + if (!dd->ipath_ibcddrctrl) { /* not on re-init after reset */ dd->ipath_ibcddrctrl = @@ -1084,6 +1090,37 @@ static void ipath_7220_config_jint(struct ipath_devdata *dd, static void ipath_7220_quiet_serdes(struct ipath_devdata *dd) { u64 val; + if (dd->ibsymdelta || dd->iblnkerrdelta || + dd->ibdeltainprog) { + u64 diagc; + /* enable counter writes */ + diagc = ipath_read_kreg64(dd, dd->ipath_kregs->kr_hwdiagctrl); + ipath_write_kreg(dd, dd->ipath_kregs->kr_hwdiagctrl, + diagc | INFINIPATH_DC_COUNTERWREN); + + if (dd->ibsymdelta || dd->ibdeltainprog) { + val = ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt); + if (dd->ibdeltainprog) + val -= val - dd->ibsymsnap; + val -= dd->ibsymdelta; + ipath_write_creg(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt, val); + } + if (dd->iblnkerrdelta || dd->ibdeltainprog) { + val = ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt); + if (dd->ibdeltainprog) + val -= val - dd->iblnkerrsnap; + val -= dd->iblnkerrdelta; + ipath_write_creg(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt, val); + } + + /* and disable counter writes */ + ipath_write_kreg(dd, dd->ipath_kregs->kr_hwdiagctrl, diagc); + } + dd->ipath_flags &= ~IPATH_IB_AUTONEG_INPROG; wake_up(&dd->ipath_autoneg_wait); cancel_delayed_work(&dd->ipath_autoneg_work); @@ -2325,7 +2362,7 @@ static void try_auto_neg(struct ipath_devdata *dd) static int ipath_7220_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) { - int ret = 0; + int ret = 0, symadj = 0; u32 ltstate = ipath_ib_linkstate(dd, ibcs); dd->ipath_link_width_active = @@ -2368,6 +2405,13 @@ static int ipath_7220_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) ipath_dbg("DDR negotiation try, %u/%u\n", dd->ipath_autoneg_tries, IPATH_AUTONEG_TRIES); + if (!dd->ibdeltainprog) { + dd->ibdeltainprog = 1; + dd->ibsymsnap = ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt); + dd->iblnkerrsnap = ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt); + } try_auto_neg(dd); ret = 1; /* no other IB status change processing */ } else if ((dd->ipath_flags & IPATH_IB_AUTONEG_INPROG) @@ -2388,6 +2432,7 @@ static int ipath_7220_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) set_speed_fast(dd, dd->ipath_link_speed_enabled); wake_up(&dd->ipath_autoneg_wait); + symadj = 1; } else if (dd->ipath_flags & IPATH_IB_AUTONEG_FAILED) { /* * clear autoneg failure flag, and do setup @@ -2403,22 +2448,28 @@ static int ipath_7220_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) IBA7220_IBC_IBTA_1_2_MASK; ipath_write_kreg(dd, IPATH_KREG_OFFSET(IBNCModeCtrl), 0); + symadj = 1; } } /* - * if we are in 1X, and are in autoneg width, it - * could be due to an xgxs problem, so if we haven't + * if we are in 1X on rev1 only, and are in autoneg width, + * it could be due to an xgxs problem, so if we haven't * already tried, try twice to get to 4X; if we * tried, and couldn't, report it, since it will * probably not be what is desired. */ - if ((dd->ipath_link_width_enabled & (IB_WIDTH_1X | + if (dd->ipath_minrev == 1 && + (dd->ipath_link_width_enabled & (IB_WIDTH_1X | IB_WIDTH_4X)) == (IB_WIDTH_1X | IB_WIDTH_4X) && dd->ipath_link_width_active == IB_WIDTH_1X && dd->ipath_x1_fix_tries < 3) { - if (++dd->ipath_x1_fix_tries == 3) + if (++dd->ipath_x1_fix_tries == 3) { dev_info(&dd->pcidev->dev, "IB link is in 1X mode\n"); + if (!(dd->ipath_flags & + IPATH_IB_AUTONEG_INPROG)) + symadj = 1; + } else { ipath_cdbg(VERBOSE, "IB 1X in " "auto-width, try %u to be " @@ -2429,7 +2480,8 @@ static int ipath_7220_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) dd->ipath_f_xgxs_reset(dd); ret = 1; /* skip other processing */ } - } + } else if (!(dd->ipath_flags & IPATH_IB_AUTONEG_INPROG)) + symadj = 1; if (!ret) { dd->delay_mult = rate_to_delay @@ -2440,6 +2492,25 @@ static int ipath_7220_ib_updown(struct ipath_devdata *dd, int ibup, u64 ibcs) } } + if (symadj) { + if (dd->ibdeltainprog) { + dd->ibdeltainprog = 0; + dd->ibsymdelta += ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt) - + dd->ibsymsnap; + dd->iblnkerrdelta += ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt) - + dd->iblnkerrsnap; + } + } else if (!ibup && !dd->ibdeltainprog + && !(dd->ipath_flags & IPATH_IB_AUTONEG_INPROG)) { + dd->ibdeltainprog = 1; + dd->ibsymsnap = ipath_read_creg32(dd, + dd->ipath_cregs->cr_ibsymbolerrcnt); + dd->iblnkerrsnap = ipath_read_creg32(dd, + dd->ipath_cregs->cr_iblinkerrrecovcnt); + } + if (!ret) ipath_setup_7220_setextled(dd, ipath_ib_linkstate(dd, ibcs), ltstate); diff --git a/drivers/infiniband/hw/ipath/ipath_init_chip.c b/drivers/infiniband/hw/ipath/ipath_init_chip.c index 3e5baa43fc82..64aeefbd2a5d 100644 --- a/drivers/infiniband/hw/ipath/ipath_init_chip.c +++ b/drivers/infiniband/hw/ipath/ipath_init_chip.c @@ -229,6 +229,7 @@ static int init_chip_first(struct ipath_devdata *dd) spin_lock_init(&dd->ipath_kernel_tid_lock); spin_lock_init(&dd->ipath_user_tid_lock); spin_lock_init(&dd->ipath_sendctrl_lock); + spin_lock_init(&dd->ipath_uctxt_lock); spin_lock_init(&dd->ipath_sdma_lock); spin_lock_init(&dd->ipath_gpio_lock); spin_lock_init(&dd->ipath_eep_st_lock); diff --git a/drivers/infiniband/hw/ipath/ipath_kernel.h b/drivers/infiniband/hw/ipath/ipath_kernel.h index 0bd8bcb184a1..6ba4861dd6ac 100644 --- a/drivers/infiniband/hw/ipath/ipath_kernel.h +++ b/drivers/infiniband/hw/ipath/ipath_kernel.h @@ -355,6 +355,19 @@ struct ipath_devdata { /* errors masked because they occur too fast */ ipath_err_t ipath_maskederrs; u64 ipath_lastlinkrecov; /* link recoveries at last ACTIVE */ + /* these 5 fields are used to establish deltas for IB Symbol + * errors and linkrecovery errors. They can be reported on + * some chips during link negotiation prior to INIT, and with + * DDR when faking DDR negotiations with non-IBTA switches. + * The chip counters are adjusted at driver unload if there is + * a non-zero delta. + */ + u64 ibdeltainprog; + u64 ibsymdelta; + u64 ibsymsnap; + u64 iblnkerrdelta; + u64 iblnkerrsnap; + /* time in jiffies at which to re-enable maskederrs */ unsigned long ipath_unmasktime; /* count of egrfull errors, combined for all ports */ @@ -464,6 +477,8 @@ struct ipath_devdata { spinlock_t ipath_kernel_tid_lock; spinlock_t ipath_user_tid_lock; spinlock_t ipath_sendctrl_lock; + /* around ipath_pd and (user ports) port_cnt use (intr vs free) */ + spinlock_t ipath_uctxt_lock; /* * IPATH_STATUS_*, diff --git a/drivers/infiniband/hw/ipath/ipath_keys.c b/drivers/infiniband/hw/ipath/ipath_keys.c index 8f32b17a5eed..c0e933fec218 100644 --- a/drivers/infiniband/hw/ipath/ipath_keys.c +++ b/drivers/infiniband/hw/ipath/ipath_keys.c @@ -132,6 +132,7 @@ int ipath_lkey_ok(struct ipath_qp *qp, struct ipath_sge *isge, * (see ipath_get_dma_mr and ipath_dma.c). */ if (sge->lkey == 0) { + /* always a kernel port, no locking needed */ struct ipath_pd *pd = to_ipd(qp->ibqp.pd); if (pd->user) { @@ -211,6 +212,7 @@ int ipath_rkey_ok(struct ipath_qp *qp, struct ipath_sge_state *ss, * (see ipath_get_dma_mr and ipath_dma.c). */ if (rkey == 0) { + /* always a kernel port, no locking needed */ struct ipath_pd *pd = to_ipd(qp->ibqp.pd); if (pd->user) { diff --git a/drivers/infiniband/hw/ipath/ipath_mad.c b/drivers/infiniband/hw/ipath/ipath_mad.c index be4fc9ada8e7..17a123197477 100644 --- a/drivers/infiniband/hw/ipath/ipath_mad.c +++ b/drivers/infiniband/hw/ipath/ipath_mad.c @@ -348,6 +348,7 @@ bail: */ static int get_pkeys(struct ipath_devdata *dd, u16 * pkeys) { + /* always a kernel port, no locking needed */ struct ipath_portdata *pd = dd->ipath_pd[0]; memcpy(pkeys, pd->port_pkeys, sizeof(pd->port_pkeys)); @@ -730,6 +731,7 @@ static int set_pkeys(struct ipath_devdata *dd, u16 *pkeys) int i; int changed = 0; + /* always a kernel port, no locking needed */ pd = dd->ipath_pd[0]; for (i = 0; i < ARRAY_SIZE(pd->port_pkeys); i++) { diff --git a/drivers/infiniband/hw/ipath/ipath_qp.c b/drivers/infiniband/hw/ipath/ipath_qp.c index 4715911101e4..3a5a89b609c4 100644 --- a/drivers/infiniband/hw/ipath/ipath_qp.c +++ b/drivers/infiniband/hw/ipath/ipath_qp.c @@ -745,6 +745,7 @@ struct ib_qp *ipath_create_qp(struct ib_pd *ibpd, struct ipath_swqe *swq = NULL; struct ipath_ibdev *dev; size_t sz; + size_t sg_list_sz; struct ib_qp *ret; if (init_attr->create_flags) { @@ -789,19 +790,31 @@ struct ib_qp *ipath_create_qp(struct ib_pd *ibpd, goto bail; } sz = sizeof(*qp); + sg_list_sz = 0; if (init_attr->srq) { struct ipath_srq *srq = to_isrq(init_attr->srq); - sz += sizeof(*qp->r_sg_list) * - srq->rq.max_sge; - } else - sz += sizeof(*qp->r_sg_list) * - init_attr->cap.max_recv_sge; - qp = kmalloc(sz, GFP_KERNEL); + if (srq->rq.max_sge > 1) + sg_list_sz = sizeof(*qp->r_sg_list) * + (srq->rq.max_sge - 1); + } else if (init_attr->cap.max_recv_sge > 1) + sg_list_sz = sizeof(*qp->r_sg_list) * + (init_attr->cap.max_recv_sge - 1); + qp = kmalloc(sz + sg_list_sz, GFP_KERNEL); if (!qp) { ret = ERR_PTR(-ENOMEM); goto bail_swq; } + if (sg_list_sz && (init_attr->qp_type == IB_QPT_UD || + init_attr->qp_type == IB_QPT_SMI || + init_attr->qp_type == IB_QPT_GSI)) { + qp->r_ud_sg_list = kmalloc(sg_list_sz, GFP_KERNEL); + if (!qp->r_ud_sg_list) { + ret = ERR_PTR(-ENOMEM); + goto bail_qp; + } + } else + qp->r_ud_sg_list = NULL; if (init_attr->srq) { sz = 0; qp->r_rq.size = 0; @@ -818,7 +831,7 @@ struct ib_qp *ipath_create_qp(struct ib_pd *ibpd, qp->r_rq.size * sz); if (!qp->r_rq.wq) { ret = ERR_PTR(-ENOMEM); - goto bail_qp; + goto bail_sg_list; } } @@ -848,7 +861,7 @@ struct ib_qp *ipath_create_qp(struct ib_pd *ibpd, if (err) { ret = ERR_PTR(err); vfree(qp->r_rq.wq); - goto bail_qp; + goto bail_sg_list; } qp->ip = NULL; qp->s_tx = NULL; @@ -925,6 +938,8 @@ bail_ip: vfree(qp->r_rq.wq); ipath_free_qp(&dev->qp_table, qp); free_qpn(&dev->qp_table, qp->ibqp.qp_num); +bail_sg_list: + kfree(qp->r_ud_sg_list); bail_qp: kfree(qp); bail_swq: @@ -989,6 +1004,7 @@ int ipath_destroy_qp(struct ib_qp *ibqp) kref_put(&qp->ip->ref, ipath_release_mmap_info); else vfree(qp->r_rq.wq); + kfree(qp->r_ud_sg_list); vfree(qp->s_wq); kfree(qp); return 0; diff --git a/drivers/infiniband/hw/ipath/ipath_rc.c b/drivers/infiniband/hw/ipath/ipath_rc.c index 7b93cda1a4bd..9170710b950d 100644 --- a/drivers/infiniband/hw/ipath/ipath_rc.c +++ b/drivers/infiniband/hw/ipath/ipath_rc.c @@ -573,9 +573,8 @@ int ipath_make_rc_req(struct ipath_qp *qp) ohdr->u.rc.reth.length = cpu_to_be32(qp->s_len); qp->s_state = OP(RDMA_READ_REQUEST); hwords += sizeof(ohdr->u.rc.reth) / sizeof(u32); - bth2 = qp->s_psn++ & IPATH_PSN_MASK; - if (ipath_cmp24(qp->s_psn, qp->s_next_psn) > 0) - qp->s_next_psn = qp->s_psn; + bth2 = qp->s_psn & IPATH_PSN_MASK; + qp->s_psn = wqe->lpsn + 1; ss = NULL; len = 0; qp->s_cur++; diff --git a/drivers/infiniband/hw/ipath/ipath_ruc.c b/drivers/infiniband/hw/ipath/ipath_ruc.c index fc0f6d9e6030..2296832f94da 100644 --- a/drivers/infiniband/hw/ipath/ipath_ruc.c +++ b/drivers/infiniband/hw/ipath/ipath_ruc.c @@ -156,7 +156,7 @@ bail: /** * ipath_get_rwqe - copy the next RWQE into the QP's RWQE * @qp: the QP - * @wr_id_only: update wr_id only, not SGEs + * @wr_id_only: update qp->r_wr_id only, not qp->r_sge * * Return 0 if no RWQE is available, otherwise return 1. * @@ -173,8 +173,6 @@ int ipath_get_rwqe(struct ipath_qp *qp, int wr_id_only) u32 tail; int ret; - qp->r_sge.sg_list = qp->r_sg_list; - if (qp->ibqp.srq) { srq = to_isrq(qp->ibqp.srq); handler = srq->ibsrq.event_handler; @@ -206,8 +204,10 @@ int ipath_get_rwqe(struct ipath_qp *qp, int wr_id_only) wqe = get_rwqe_ptr(rq, tail); if (++tail >= rq->size) tail = 0; - } while (!wr_id_only && !ipath_init_sge(qp, wqe, &qp->r_len, - &qp->r_sge)); + if (wr_id_only) + break; + qp->r_sge.sg_list = qp->r_sg_list; + } while (!ipath_init_sge(qp, wqe, &qp->r_len, &qp->r_sge)); qp->r_wr_id = wqe->wr_id; wq->tail = tail; diff --git a/drivers/infiniband/hw/ipath/ipath_sdma.c b/drivers/infiniband/hw/ipath/ipath_sdma.c index 284c9bca517e..8e255adf5d9b 100644 --- a/drivers/infiniband/hw/ipath/ipath_sdma.c +++ b/drivers/infiniband/hw/ipath/ipath_sdma.c @@ -698,10 +698,8 @@ retry: addr = dma_map_single(&dd->pcidev->dev, tx->txreq.map_addr, tx->map_len, DMA_TO_DEVICE); - if (dma_mapping_error(&dd->pcidev->dev, addr)) { - ret = -EIO; - goto unlock; - } + if (dma_mapping_error(&dd->pcidev->dev, addr)) + goto ioerr; dwoffset = tx->map_len >> 2; make_sdma_desc(dd, sdmadesc, (u64) addr, dwoffset, 0); @@ -741,6 +739,8 @@ retry: dw = (len + 3) >> 2; addr = dma_map_single(&dd->pcidev->dev, sge->vaddr, dw << 2, DMA_TO_DEVICE); + if (dma_mapping_error(&dd->pcidev->dev, addr)) + goto unmap; make_sdma_desc(dd, sdmadesc, (u64) addr, dw, dwoffset); /* SDmaUseLargeBuf has to be set in every descriptor */ if (tx->txreq.flags & IPATH_SDMA_TXREQ_F_USELARGEBUF) @@ -798,7 +798,18 @@ retry: list_add_tail(&tx->txreq.list, &dd->ipath_sdma_activelist); if (tx->txreq.flags & IPATH_SDMA_TXREQ_F_VL15) vl15_watchdog_enq(dd); - + goto unlock; + +unmap: + while (tail != dd->ipath_sdma_descq_tail) { + if (!tail) + tail = dd->ipath_sdma_descq_cnt - 1; + else + tail--; + unmap_desc(dd, tail); + } +ioerr: + ret = -EIO; unlock: spin_unlock_irqrestore(&dd->ipath_sdma_lock, flags); fail: diff --git a/drivers/infiniband/hw/ipath/ipath_stats.c b/drivers/infiniband/hw/ipath/ipath_stats.c index c8e3d65f0de8..f63e143e3292 100644 --- a/drivers/infiniband/hw/ipath/ipath_stats.c +++ b/drivers/infiniband/hw/ipath/ipath_stats.c @@ -112,6 +112,14 @@ u64 ipath_snap_cntr(struct ipath_devdata *dd, ipath_creg creg) dd->ipath_lastrpkts = val; } val64 = dd->ipath_rpkts; + } else if (creg == dd->ipath_cregs->cr_ibsymbolerrcnt) { + if (dd->ibdeltainprog) + val64 -= val64 - dd->ibsymsnap; + val64 -= dd->ibsymdelta; + } else if (creg == dd->ipath_cregs->cr_iblinkerrrecovcnt) { + if (dd->ibdeltainprog) + val64 -= val64 - dd->iblnkerrsnap; + val64 -= dd->iblnkerrdelta; } else val64 = (u64) val; diff --git a/drivers/infiniband/hw/ipath/ipath_ud.c b/drivers/infiniband/hw/ipath/ipath_ud.c index 729446f56aab..91c74cc797ae 100644 --- a/drivers/infiniband/hw/ipath/ipath_ud.c +++ b/drivers/infiniband/hw/ipath/ipath_ud.c @@ -70,8 +70,6 @@ static void ipath_ud_loopback(struct ipath_qp *sqp, struct ipath_swqe *swqe) goto done; } - rsge.sg_list = NULL; - /* * Check that the qkey matches (except for QP0, see 9.6.1.4.1). * Qkeys with the high order bit set mean use the @@ -115,21 +113,6 @@ static void ipath_ud_loopback(struct ipath_qp *sqp, struct ipath_swqe *swqe) rq = &qp->r_rq; } - if (rq->max_sge > 1) { - /* - * XXX We could use GFP_KERNEL if ipath_do_send() - * was always called from the tasklet instead of - * from ipath_post_send(). - */ - rsge.sg_list = kmalloc((rq->max_sge - 1) * - sizeof(struct ipath_sge), - GFP_ATOMIC); - if (!rsge.sg_list) { - dev->n_pkt_drops++; - goto drop; - } - } - /* * Get the next work request entry to find where to put the data. * Note that it is safe to drop the lock after changing rq->tail @@ -147,6 +130,7 @@ static void ipath_ud_loopback(struct ipath_qp *sqp, struct ipath_swqe *swqe) goto drop; } wqe = get_rwqe_ptr(rq, tail); + rsge.sg_list = qp->r_ud_sg_list; if (!ipath_init_sge(qp, wqe, &rlen, &rsge)) { spin_unlock_irqrestore(&rq->lock, flags); dev->n_pkt_drops++; @@ -242,7 +226,6 @@ static void ipath_ud_loopback(struct ipath_qp *sqp, struct ipath_swqe *swqe) ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc, swqe->wr.send_flags & IB_SEND_SOLICITED); drop: - kfree(rsge.sg_list); if (atomic_dec_and_test(&qp->refcount)) wake_up(&qp->wait); done:; diff --git a/drivers/infiniband/hw/ipath/ipath_verbs.c b/drivers/infiniband/hw/ipath/ipath_verbs.c index eabc4247860b..cdf0e6abd34d 100644 --- a/drivers/infiniband/hw/ipath/ipath_verbs.c +++ b/drivers/infiniband/hw/ipath/ipath_verbs.c @@ -1852,7 +1852,7 @@ unsigned ipath_get_npkeys(struct ipath_devdata *dd) } /** - * ipath_get_pkey - return the indexed PKEY from the port 0 PKEY table + * ipath_get_pkey - return the indexed PKEY from the port PKEY table * @dd: the infinipath device * @index: the PKEY index */ @@ -1860,6 +1860,7 @@ unsigned ipath_get_pkey(struct ipath_devdata *dd, unsigned index) { unsigned ret; + /* always a kernel port, no locking needed */ if (index >= ARRAY_SIZE(dd->ipath_pd[0]->port_pkeys)) ret = 0; else diff --git a/drivers/infiniband/hw/ipath/ipath_verbs.h b/drivers/infiniband/hw/ipath/ipath_verbs.h index 9d12ae8a778e..11e3f613df93 100644 --- a/drivers/infiniband/hw/ipath/ipath_verbs.h +++ b/drivers/infiniband/hw/ipath/ipath_verbs.h @@ -431,6 +431,7 @@ struct ipath_qp { u32 s_lsn; /* limit sequence number (credit) */ struct ipath_swqe *s_wq; /* send work queue */ struct ipath_swqe *s_wqe; + struct ipath_sge *r_ud_sg_list; struct ipath_rq r_rq; /* receive work queue */ struct ipath_sge r_sg_list[0]; /* verified SGEs */ }; diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c index d0866a3636e2..a3c5af1d7ec0 100644 --- a/drivers/infiniband/hw/mlx4/cq.c +++ b/drivers/infiniband/hw/mlx4/cq.c @@ -222,7 +222,7 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector } err = mlx4_cq_alloc(dev->dev, entries, &cq->buf.mtt, uar, - cq->db.dma, &cq->mcq, 0); + cq->db.dma, &cq->mcq, vector, 0); if (err) goto err_dbmap; @@ -325,15 +325,17 @@ static int mlx4_ib_get_outstanding_cqes(struct mlx4_ib_cq *cq) static void mlx4_ib_cq_resize_copy_cqes(struct mlx4_ib_cq *cq) { - struct mlx4_cqe *cqe; + struct mlx4_cqe *cqe, *new_cqe; int i; i = cq->mcq.cons_index; cqe = get_cqe(cq, i & cq->ibcq.cqe); while ((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) != MLX4_CQE_OPCODE_RESIZE) { - memcpy(get_cqe_from_buf(&cq->resize_buf->buf, - (i + 1) & cq->resize_buf->cqe), - get_cqe(cq, i & cq->ibcq.cqe), sizeof(struct mlx4_cqe)); + new_cqe = get_cqe_from_buf(&cq->resize_buf->buf, + (i + 1) & cq->resize_buf->cqe); + memcpy(new_cqe, get_cqe(cq, i & cq->ibcq.cqe), sizeof(struct mlx4_cqe)); + new_cqe->owner_sr_opcode = (cqe->owner_sr_opcode & ~MLX4_CQE_OWNER_MASK) | + (((i + 1) & (cq->resize_buf->cqe + 1)) ? MLX4_CQE_OWNER_MASK : 0); cqe = get_cqe(cq, ++i & cq->ibcq.cqe); } ++cq->mcq.cons_index; @@ -343,6 +345,7 @@ int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) { struct mlx4_ib_dev *dev = to_mdev(ibcq->device); struct mlx4_ib_cq *cq = to_mcq(ibcq); + struct mlx4_mtt mtt; int outst_cqe; int err; @@ -376,10 +379,13 @@ int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) goto out; } + mtt = cq->buf.mtt; + err = mlx4_cq_resize(dev->dev, &cq->mcq, entries, &cq->resize_buf->buf.mtt); if (err) goto err_buf; + mlx4_mtt_cleanup(dev->dev, &mtt); if (ibcq->uobject) { cq->buf = cq->resize_buf->buf; cq->ibcq.cqe = cq->resize_buf->cqe; @@ -406,6 +412,7 @@ int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) goto out; err_buf: + mlx4_mtt_cleanup(dev->dev, &cq->resize_buf->buf.mtt); if (!ibcq->uobject) mlx4_ib_free_cq_buf(dev, &cq->resize_buf->buf, cq->resize_buf->cqe); @@ -692,7 +699,7 @@ repoll: } wc->slid = be16_to_cpu(cqe->rlid); - wc->sl = be16_to_cpu(cqe->sl_vid >> 12); + wc->sl = be16_to_cpu(cqe->sl_vid) >> 12; g_mlpath_rqpn = be32_to_cpu(cqe->g_mlpath_rqpn); wc->src_qp = g_mlpath_rqpn & 0xffffff; wc->dlid_path_bits = (g_mlpath_rqpn >> 24) & 0x7f; diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index cdca3a511e1c..606f1e2ef284 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -298,7 +298,7 @@ int mlx4_ib_mad_init(struct mlx4_ib_dev *dev) int p, q; int ret; - for (p = 0; p < dev->dev->caps.num_ports; ++p) + for (p = 0; p < dev->num_ports; ++p) for (q = 0; q <= 1; ++q) { agent = ib_register_mad_agent(&dev->ib_dev, p + 1, q ? IB_QPT_GSI : IB_QPT_SMI, @@ -314,7 +314,7 @@ int mlx4_ib_mad_init(struct mlx4_ib_dev *dev) return 0; err: - for (p = 0; p < dev->dev->caps.num_ports; ++p) + for (p = 0; p < dev->num_ports; ++p) for (q = 0; q <= 1; ++q) if (dev->send_agent[p][q]) ib_unregister_mad_agent(dev->send_agent[p][q]); @@ -327,7 +327,7 @@ void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev) struct ib_mad_agent *agent; int p, q; - for (p = 0; p < dev->dev->caps.num_ports; ++p) { + for (p = 0; p < dev->num_ports; ++p) { for (q = 0; q <= 1; ++q) { agent = dev->send_agent[p][q]; dev->send_agent[p][q] = NULL; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index a3c2851c0545..dcefe1fceb5c 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -574,8 +574,11 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) ibdev->ib_dev.owner = THIS_MODULE; ibdev->ib_dev.node_type = RDMA_NODE_IB_CA; ibdev->ib_dev.local_dma_lkey = dev->caps.reserved_lkey; - ibdev->ib_dev.phys_port_cnt = dev->caps.num_ports; - ibdev->ib_dev.num_comp_vectors = 1; + ibdev->num_ports = 0; + mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB) + ibdev->num_ports++; + ibdev->ib_dev.phys_port_cnt = ibdev->num_ports; + ibdev->ib_dev.num_comp_vectors = dev->caps.num_comp_vectors; ibdev->ib_dev.dma_device = &dev->pdev->dev; ibdev->ib_dev.uverbs_abi_ver = MLX4_IB_UVERBS_ABI_VERSION; @@ -691,7 +694,7 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr) struct mlx4_ib_dev *ibdev = ibdev_ptr; int p; - for (p = 1; p <= dev->caps.num_ports; ++p) + for (p = 1; p <= ibdev->num_ports; ++p) mlx4_CLOSE_PORT(dev, p); mlx4_ib_mad_cleanup(ibdev); @@ -706,6 +709,10 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr, enum mlx4_dev_event event, int port) { struct ib_event ibev; + struct mlx4_ib_dev *ibdev = to_mdev((struct ib_device *) ibdev_ptr); + + if (port > ibdev->num_ports) + return; switch (event) { case MLX4_DEV_EVENT_PORT_UP: diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 6e2b0dc21b61..9974e886b8de 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -162,6 +162,7 @@ struct mlx4_ib_ah { struct mlx4_ib_dev { struct ib_device ib_dev; struct mlx4_dev *dev; + int num_ports; void __iomem *uar_map; struct mlx4_uar priv_uar; diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c index 87f5c5a87b98..8e4d26d56a95 100644 --- a/drivers/infiniband/hw/mlx4/mr.c +++ b/drivers/infiniband/hw/mlx4/mr.c @@ -205,6 +205,7 @@ struct ib_mr *mlx4_ib_alloc_fast_reg_mr(struct ib_pd *pd, goto err_mr; mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key; + mr->umem = NULL; return &mr->ibmr; diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index baa01deb2436..39167a797f99 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -451,6 +451,7 @@ 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) { + int qpn; int err; mutex_init(&qp->mutex); @@ -545,9 +546,17 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, } } - err = mlx4_qp_alloc(dev->dev, sqpn, &qp->mqp); + if (sqpn) { + qpn = sqpn; + } else { + err = mlx4_qp_reserve_range(dev->dev, 1, 1, &qpn); + if (err) + goto err_wrid; + } + + err = mlx4_qp_alloc(dev->dev, qpn, &qp->mqp); if (err) - goto err_wrid; + goto err_qpn; /* * Hardware wants QPN written in big-endian order (after @@ -560,6 +569,10 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, return 0; +err_qpn: + if (!sqpn) + mlx4_qp_release_range(dev->dev, qpn, 1); + err_wrid: if (pd->uobject) { if (!init_attr->srq) @@ -655,6 +668,10 @@ static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, mlx4_ib_unlock_cqs(send_cq, recv_cq); mlx4_qp_free(dev->dev, &qp->mqp); + + if (!is_sqp(dev, qp)) + mlx4_qp_release_range(dev->dev, qp->mqp.qpn, 1); + mlx4_mtt_cleanup(dev->dev, &qp->mtt); if (is_user) { diff --git a/drivers/infiniband/hw/mthca/mthca_mcg.c b/drivers/infiniband/hw/mthca/mthca_mcg.c index 3f5f94879208..d4c81053e439 100644 --- a/drivers/infiniband/hw/mthca/mthca_mcg.c +++ b/drivers/infiniband/hw/mthca/mthca_mcg.c @@ -87,17 +87,7 @@ static int find_mgm(struct mthca_dev *dev, } if (0) - mthca_dbg(dev, "Hash for %04x:%04x:%04x:%04x:" - "%04x:%04x:%04x:%04x is %04x\n", - be16_to_cpu(((__be16 *) gid)[0]), - be16_to_cpu(((__be16 *) gid)[1]), - be16_to_cpu(((__be16 *) gid)[2]), - be16_to_cpu(((__be16 *) gid)[3]), - be16_to_cpu(((__be16 *) gid)[4]), - be16_to_cpu(((__be16 *) gid)[5]), - be16_to_cpu(((__be16 *) gid)[6]), - be16_to_cpu(((__be16 *) gid)[7]), - *hash); + mthca_dbg(dev, "Hash for %pI6 is %04x\n", gid, *hash); *index = *hash; *prev = -1; @@ -264,16 +254,7 @@ int mthca_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) goto out; if (index == -1) { - mthca_err(dev, "MGID %04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x " - "not found\n", - be16_to_cpu(((__be16 *) gid->raw)[0]), - be16_to_cpu(((__be16 *) gid->raw)[1]), - be16_to_cpu(((__be16 *) gid->raw)[2]), - be16_to_cpu(((__be16 *) gid->raw)[3]), - be16_to_cpu(((__be16 *) gid->raw)[4]), - be16_to_cpu(((__be16 *) gid->raw)[5]), - be16_to_cpu(((__be16 *) gid->raw)[6]), - be16_to_cpu(((__be16 *) gid->raw)[7])); + mthca_err(dev, "MGID %pI6 not found\n", gid->raw); err = -EINVAL; goto out; } diff --git a/drivers/infiniband/hw/nes/nes.c b/drivers/infiniband/hw/nes/nes.c index a2b04d62b1a4..b9611ade9eab 100644 --- a/drivers/infiniband/hw/nes/nes.c +++ b/drivers/infiniband/hw/nes/nes.c @@ -95,6 +95,10 @@ unsigned int wqm_quanta = 0x10000; module_param(wqm_quanta, int, 0644); MODULE_PARM_DESC(wqm_quanta, "WQM quanta"); +static unsigned int limit_maxrdreqsz; +module_param(limit_maxrdreqsz, bool, 0644); +MODULE_PARM_DESC(limit_maxrdreqsz, "Limit max read request size to 256 Bytes"); + LIST_HEAD(nes_adapter_list); static LIST_HEAD(nes_dev_list); @@ -138,14 +142,9 @@ static int nes_inetaddr_event(struct notifier_block *notifier, struct nes_device *nesdev; struct net_device *netdev; struct nes_vnic *nesvnic; - unsigned int addr; - unsigned int mask; - - addr = ntohl(ifa->ifa_address); - mask = ntohl(ifa->ifa_mask); - nes_debug(NES_DBG_NETDEV, "nes_inetaddr_event: ip address " NIPQUAD_FMT - ", netmask " NIPQUAD_FMT ".\n", - HIPQUAD(addr), HIPQUAD(mask)); + + nes_debug(NES_DBG_NETDEV, "nes_inetaddr_event: ip address %pI4, netmask %pI4.\n", + &ifa->ifa_address, &ifa->ifa_mask); list_for_each_entry(nesdev, &nes_dev_list, list) { nes_debug(NES_DBG_NETDEV, "Nesdev list entry = 0x%p. (%s)\n", nesdev, nesdev->netdev[0]->name); @@ -356,10 +355,8 @@ struct ib_qp *nes_get_qp(struct ib_device *device, int qpn) */ static void nes_print_macaddr(struct net_device *netdev) { - DECLARE_MAC_BUF(mac); - - nes_debug(NES_DBG_INIT, "%s: %s, IRQ %u\n", - netdev->name, print_mac(mac, netdev->dev_addr), netdev->irq); + nes_debug(NES_DBG_INIT, "%s: %pM, IRQ %u\n", + netdev->name, netdev->dev_addr, netdev->irq); } /** @@ -588,6 +585,18 @@ static int __devinit nes_probe(struct pci_dev *pcidev, const struct pci_device_i nesdev->nesadapter->port_count; } + if ((limit_maxrdreqsz || + ((nesdev->nesadapter->phy_type[0] == NES_PHY_TYPE_GLADIUS) && + (hw_rev == NE020_REV1))) && + (pcie_get_readrq(pcidev) > 256)) { + if (pcie_set_readrq(pcidev, 256)) + printk(KERN_ERR PFX "Unable to set max read request" + " to 256 bytes\n"); + else + nes_debug(NES_DBG_INIT, "Max read request size set" + " to 256 bytes\n"); + } + tasklet_init(&nesdev->dpc_tasklet, nes_dpc, (unsigned long)nesdev); /* bring up the Control QP */ diff --git a/drivers/infiniband/hw/nes/nes.h b/drivers/infiniband/hw/nes/nes.h index 1595dc7bba9d..13a5bb1a7bcf 100644 --- a/drivers/infiniband/hw/nes/nes.h +++ b/drivers/infiniband/hw/nes/nes.h @@ -137,14 +137,18 @@ #ifdef CONFIG_INFINIBAND_NES_DEBUG #define nes_debug(level, fmt, args...) \ +do { \ if (level & nes_debug_level) \ - printk(KERN_ERR PFX "%s[%u]: " fmt, __func__, __LINE__, ##args) - -#define assert(expr) \ -if (!(expr)) { \ - printk(KERN_ERR PFX "Assertion failed! %s, %s, %s, line %d\n", \ - #expr, __FILE__, __func__, __LINE__); \ -} + printk(KERN_ERR PFX "%s[%u]: " fmt, __func__, __LINE__, ##args); \ +} while (0) + +#define assert(expr) \ +do { \ + if (!(expr)) { \ + printk(KERN_ERR PFX "Assertion failed! %s, %s, %s, line %d\n", \ + #expr, __FILE__, __func__, __LINE__); \ + } \ +} while (0) #define NES_EVENT_TIMEOUT 1200000 #else diff --git a/drivers/infiniband/hw/nes/nes_cm.c b/drivers/infiniband/hw/nes/nes_cm.c index 2caf9da81ad5..a812db243477 100644 --- a/drivers/infiniband/hw/nes/nes_cm.c +++ b/drivers/infiniband/hw/nes/nes_cm.c @@ -86,15 +86,14 @@ 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 void mini_cm_recv_pkt(struct nes_cm_core *, struct nes_vnic *, +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 struct sk_buff *form_cm_frame(struct sk_buff *, struct nes_cm_node *, +static void form_cm_frame(struct sk_buff *, struct nes_cm_node *, void *, u32, void *, u32, u8); -static struct sk_buff *get_free_pkt(struct nes_cm_node *cm_node); static int add_ref_cm_node(struct nes_cm_node *); static int rem_ref_cm_node(struct nes_cm_core *, struct nes_cm_node *); @@ -251,7 +250,7 @@ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 len) * form_cm_frame - get a free packet and build empty frame Use * node info to build. */ -static struct sk_buff *form_cm_frame(struct sk_buff *skb, +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) { @@ -339,7 +338,6 @@ static struct sk_buff *form_cm_frame(struct sk_buff *skb, skb_shinfo(skb)->nr_frags = 0; cm_packets_created++; - return skb; } @@ -356,7 +354,6 @@ static void print_core(struct nes_cm_core *core) nes_debug(NES_DBG_CM, "State : %u \n", core->state); - nes_debug(NES_DBG_CM, "Tx Free cnt : %u \n", skb_queue_len(&core->tx_free_list)); 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)); @@ -381,8 +378,6 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, int ret = 0; u32 was_timer_set; - if (!cm_node) - return -EINVAL; new_send = kzalloc(sizeof(*new_send), GFP_ATOMIC); if (!new_send) return -1; @@ -459,13 +454,23 @@ static void nes_cm_timer_tick(unsigned long pass) int ret = NETDEV_TX_OK; enum nes_cm_node_state last_state; + 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); - add_ref_cm_node(cm_node); - spin_unlock_irqrestore(&cm_core->ht_lock, flags); + if (!list_empty(&cm_node->recv_list) || (cm_node->send_entry)) { + add_ref_cm_node(cm_node); + list_add(&cm_node->timer_entry, &timer_list); + } + } + spin_unlock_irqrestore(&cm_core->ht_lock, flags); + + list_for_each_safe(list_node, list_core_temp, &timer_list) { + cm_node = container_of(list_node, struct nes_cm_node, + timer_entry); spin_lock_irqsave(&cm_node->recv_list_lock, flags); list_for_each_safe(list_core, list_node_temp, &cm_node->recv_list) { @@ -519,7 +524,7 @@ static void nes_cm_timer_tick(unsigned long pass) do { send_entry = cm_node->send_entry; if (!send_entry) - continue; + break; if (time_after(send_entry->timetosend, jiffies)) { if (cm_node->state != NES_CM_STATE_TSA) { if ((nexttimeout > @@ -528,18 +533,18 @@ static void nes_cm_timer_tick(unsigned long pass) nexttimeout = send_entry->timetosend; settimer = 1; - continue; + break; } } else { free_retrans_entry(cm_node); - continue; + break; } } if ((cm_node->state == NES_CM_STATE_TSA) || (cm_node->state == NES_CM_STATE_CLOSED)) { free_retrans_entry(cm_node); - continue; + break; } if (!send_entry->retranscount || @@ -557,7 +562,7 @@ static void nes_cm_timer_tick(unsigned long pass) NES_CM_EVENT_ABORTED); spin_lock_irqsave(&cm_node->retrans_list_lock, flags); - continue; + break; } atomic_inc(&send_entry->skb->users); cm_packets_retrans++; @@ -583,7 +588,7 @@ static void nes_cm_timer_tick(unsigned long pass) send_entry->retrycount--; nexttimeout = jiffies + NES_SHORT_TIME; settimer = 1; - continue; + break; } else { cm_packets_sent++; } @@ -615,14 +620,12 @@ static void nes_cm_timer_tick(unsigned long pass) spin_unlock_irqrestore(&cm_node->retrans_list_lock, flags); rem_ref_cm_node(cm_node->cm_core, cm_node); - spin_lock_irqsave(&cm_core->ht_lock, flags); if (ret != NETDEV_TX_OK) { nes_debug(NES_DBG_CM, "rexmit failed for cm_node=%p\n", cm_node); break; } } - spin_unlock_irqrestore(&cm_core->ht_lock, flags); if (settimer) { if (!timer_pending(&cm_core->tcp_timer)) { @@ -683,7 +686,7 @@ static int send_syn(struct nes_cm_node *cm_node, u32 sendack, optionssize += 1; if (!skb) - skb = get_free_pkt(cm_node); + skb = dev_alloc_skb(MAX_CM_BUFFER); if (!skb) { nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n"); return -1; @@ -708,7 +711,7 @@ static int send_reset(struct nes_cm_node *cm_node, struct sk_buff *skb) int flags = SET_RST | SET_ACK; if (!skb) - skb = get_free_pkt(cm_node); + skb = dev_alloc_skb(MAX_CM_BUFFER); if (!skb) { nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n"); return -1; @@ -729,7 +732,7 @@ static int send_ack(struct nes_cm_node *cm_node, struct sk_buff *skb) int ret; if (!skb) - skb = get_free_pkt(cm_node); + skb = dev_alloc_skb(MAX_CM_BUFFER); if (!skb) { nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n"); @@ -752,7 +755,7 @@ static int send_fin(struct nes_cm_node *cm_node, struct sk_buff *skb) /* if we didn't get a frame get one */ if (!skb) - skb = get_free_pkt(cm_node); + skb = dev_alloc_skb(MAX_CM_BUFFER); if (!skb) { nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n"); @@ -767,64 +770,20 @@ static int send_fin(struct nes_cm_node *cm_node, struct sk_buff *skb) /** - * get_free_pkt - */ -static struct sk_buff *get_free_pkt(struct nes_cm_node *cm_node) -{ - struct sk_buff *skb, *new_skb; - - /* check to see if we need to repopulate the free tx pkt queue */ - if (skb_queue_len(&cm_node->cm_core->tx_free_list) < NES_CM_FREE_PKT_LO_WATERMARK) { - while (skb_queue_len(&cm_node->cm_core->tx_free_list) < - cm_node->cm_core->free_tx_pkt_max) { - /* replace the frame we took, we won't get it back */ - new_skb = dev_alloc_skb(cm_node->cm_core->mtu); - BUG_ON(!new_skb); - /* add a replacement frame to the free tx list head */ - skb_queue_head(&cm_node->cm_core->tx_free_list, new_skb); - } - } - - skb = skb_dequeue(&cm_node->cm_core->tx_free_list); - - return skb; -} - - -/** - * make_hashkey - generate hash key from node tuple - */ -static inline int make_hashkey(u16 loc_port, nes_addr_t loc_addr, u16 rem_port, - nes_addr_t rem_addr) -{ - u32 hashkey = 0; - - hashkey = loc_addr + rem_addr + loc_port + rem_port; - hashkey = (hashkey % NES_CM_HASHTABLE_SIZE); - - return hashkey; -} - - -/** * 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) { unsigned long flags; - u32 hashkey; struct list_head *hte; struct nes_cm_node *cm_node; - /* make a hash index key for this packet */ - hashkey = make_hashkey(loc_port, loc_addr, rem_port, rem_addr); - /* get a handle on the hte */ hte = &cm_core->connected_nodes; - nes_debug(NES_DBG_CM, "Searching for an owner node: " NIPQUAD_FMT ":%x from core %p->%p\n", - HIPQUAD(loc_addr), loc_port, cm_core, hte); + nes_debug(NES_DBG_CM, "Searching for an owner node: %pI4:%x from core %p->%p\n", + &loc_addr, loc_port, cm_core, hte); /* walk list and find cm_node associated with this session ID */ spin_lock_irqsave(&cm_core->ht_lock, flags); @@ -873,8 +832,8 @@ static struct nes_cm_listener *find_listener(struct nes_cm_core *cm_core, } spin_unlock_irqrestore(&cm_core->listen_list_lock, flags); - nes_debug(NES_DBG_CM, "Unable to find listener for " NIPQUAD_FMT ":%x\n", - HIPQUAD(dst_addr), dst_port); + nes_debug(NES_DBG_CM, "Unable to find listener for %pI4:%x\n", + &dst_addr, dst_port); /* no listener */ return NULL; @@ -887,7 +846,6 @@ static struct nes_cm_listener *find_listener(struct nes_cm_core *cm_core, static int add_hte_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node) { unsigned long flags; - u32 hashkey; struct list_head *hte; if (!cm_node || !cm_core) @@ -896,11 +854,6 @@ static int add_hte_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node nes_debug(NES_DBG_CM, "Adding Node %p to Active Connection HT\n", cm_node); - /* first, make an index into our hash table */ - hashkey = make_hashkey(cm_node->loc_port, cm_node->loc_addr, - cm_node->rem_port, cm_node->rem_addr); - cm_node->hashkey = hashkey; - spin_lock_irqsave(&cm_core->ht_lock, flags); /* get a handle on the hash table element (list head for this slot) */ @@ -925,28 +878,36 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, struct list_head *list_pos = NULL; struct list_head *list_temp = NULL; struct nes_cm_node *cm_node = NULL; + 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)); /* free non-accelerated child nodes for this listener */ + INIT_LIST_HEAD(&reset_list); if (free_hanging_nodes) { spin_lock_irqsave(&cm_core->ht_lock, flags); list_for_each_safe(list_pos, list_temp, - &g_cm_core->connected_nodes) { + &g_cm_core->connected_nodes) { cm_node = container_of(list_pos, struct nes_cm_node, list); if ((cm_node->listener == listener) && - (!cm_node->accelerated)) { - cleanup_retrans_entry(cm_node); - spin_unlock_irqrestore(&cm_core->ht_lock, - flags); - send_reset(cm_node, NULL); - spin_lock_irqsave(&cm_core->ht_lock, flags); + (!cm_node->accelerated)) { + add_ref_cm_node(cm_node); + list_add(&cm_node->reset_entry, &reset_list); } } spin_unlock_irqrestore(&cm_core->ht_lock, flags); } + + list_for_each_safe(list_pos, list_temp, &reset_list) { + cm_node = container_of(list_pos, struct nes_cm_node, + reset_entry); + cleanup_retrans_entry(cm_node); + send_reset(cm_node, NULL); + rem_ref_cm_node(cm_node->cm_core, cm_node); + } + spin_lock_irqsave(&cm_core->listen_list_lock, flags); if (!atomic_dec_return(&listener->ref_count)) { list_del(&listener->list); @@ -1027,7 +988,6 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip) struct flowi fl; struct neighbour *neigh; int rc = -1; - DECLARE_MAC_BUF(mac); memset(&fl, 0, sizeof fl); fl.nl_u.ip4_u.daddr = htonl(dst_ip); @@ -1041,8 +1001,8 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip) if (neigh) { if (neigh->nud_state & NUD_VALID) { nes_debug(NES_DBG_CM, "Neighbor MAC address for 0x%08X" - " is %s, Gateway is 0x%08X \n", dst_ip, - print_mac(mac, neigh->ha), ntohl(rt->rt_gateway)); + " is %pM, Gateway is 0x%08X \n", dst_ip, + neigh->ha, ntohl(rt->rt_gateway)); nes_manage_arp_cache(nesvnic->netdev, neigh->ha, dst_ip, NES_ARP_ADD); rc = nes_arp_table(nesvnic->nesdev, dst_ip, NULL, @@ -1071,7 +1031,6 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, int arpindex = 0; struct nes_device *nesdev; struct nes_adapter *nesadapter; - DECLARE_MAC_BUF(mac); /* create an hte and cm_node for this instance */ cm_node = kzalloc(sizeof(*cm_node), GFP_ATOMIC); @@ -1084,10 +1043,9 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, cm_node->loc_port = cm_info->loc_port; cm_node->rem_port = cm_info->rem_port; cm_node->send_write0 = send_first; - nes_debug(NES_DBG_CM, "Make node addresses : loc = " NIPQUAD_FMT - ":%x, rem = " NIPQUAD_FMT ":%x\n", - HIPQUAD(cm_node->loc_addr), cm_node->loc_port, - HIPQUAD(cm_node->rem_addr), cm_node->rem_port); + 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); cm_node->listener = listener; cm_node->netdev = nesvnic->netdev; cm_node->cm_id = cm_info->cm_id; @@ -1126,7 +1084,10 @@ 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 */ - arpindex = nes_arp_table(nesdev, cm_node->rem_addr, NULL, NES_ARP_RESOLVE); + if (ipv4_is_loopback(htonl(cm_node->rem_addr))) + arpindex = nes_arp_table(nesdev, ntohl(nesvnic->local_ipaddr), NULL, NES_ARP_RESOLVE); + else + arpindex = nes_arp_table(nesdev, cm_node->rem_addr, NULL, NES_ARP_RESOLVE); if (arpindex < 0) { arpindex = nes_addr_resolve_neigh(nesvnic, cm_info->rem_addr); if (arpindex < 0) { @@ -1137,8 +1098,8 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, /* copy the mac addr to node context */ memcpy(cm_node->rem_mac, nesadapter->arp_table[arpindex].mac_addr, ETH_ALEN); - nes_debug(NES_DBG_CM, "Remote mac addr from arp table: %s\n", - print_mac(mac, cm_node->rem_mac)); + nes_debug(NES_DBG_CM, "Remote mac addr from arp table: %pM\n", + cm_node->rem_mac); add_hte_node(cm_core, cm_node); atomic_inc(&cm_nodes_created); @@ -1306,7 +1267,6 @@ static void drop_packet(struct sk_buff *skb) static void handle_fin_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, struct tcphdr *tcph) { - atomic_inc(&cm_resets_recvd); 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)); @@ -1344,6 +1304,7 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, { int reset = 0; /* whether to send reset in case of err.. */ + int passive_state; atomic_inc(&cm_resets_recvd); nes_debug(NES_DBG_CM, "Received Reset, cm_node = %p, state = %u." " refcnt=%d\n", cm_node, cm_node->state, @@ -1357,7 +1318,14 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, cm_node->listener, cm_node->state); active_open_err(cm_node, skb, reset); break; - /* For PASSIVE open states, remove the cm_node event */ + case NES_CM_STATE_MPAREQ_RCVD: + passive_state = atomic_add_return(1, &cm_node->passive_state); + if (passive_state == NES_SEND_RESET_EVENT) + create_event(cm_node, NES_CM_EVENT_RESET); + cleanup_retrans_entry(cm_node); + cm_node->state = NES_CM_STATE_CLOSED; + dev_kfree_skb_any(skb); + break; case NES_CM_STATE_ESTABLISHED: case NES_CM_STATE_SYN_RCVD: case NES_CM_STATE_LISTENING: @@ -1365,7 +1333,14 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, passive_open_err(cm_node, skb, reset); break; case NES_CM_STATE_TSA: + active_open_err(cm_node, skb, reset); + break; + case NES_CM_STATE_CLOSED: + cleanup_retrans_entry(cm_node); + drop_packet(skb); + break; default: + drop_packet(skb); break; } } @@ -1394,6 +1369,9 @@ static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb, dev_kfree_skb_any(skb); if (type == NES_CM_EVENT_CONNECTED) cm_node->state = NES_CM_STATE_TSA; + else + atomic_set(&cm_node->passive_state, + NES_PASSIVE_STATE_INDICATED); create_event(cm_node, type); } @@ -1474,7 +1452,7 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, int optionsize; optionsize = (tcph->doff << 2) - sizeof(struct tcphdr); - skb_pull(skb, tcph->doff << 2); + skb_trim(skb, 0); inc_sequence = ntohl(tcph->seq); switch (cm_node->state) { @@ -1507,6 +1485,10 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, cm_node->state = NES_CM_STATE_SYN_RCVD; send_syn(cm_node, 1, skb); break; + case NES_CM_STATE_CLOSED: + cleanup_retrans_entry(cm_node); + send_reset(cm_node, skb); + break; case NES_CM_STATE_TSA: case NES_CM_STATE_ESTABLISHED: case NES_CM_STATE_FIN_WAIT1: @@ -1515,7 +1497,6 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, case NES_CM_STATE_LAST_ACK: case NES_CM_STATE_CLOSING: case NES_CM_STATE_UNKNOWN: - case NES_CM_STATE_CLOSED: default: drop_packet(skb); break; @@ -1531,7 +1512,7 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, int optionsize; optionsize = (tcph->doff << 2) - sizeof(struct tcphdr); - skb_pull(skb, tcph->doff << 2); + skb_trim(skb, 0); inc_sequence = ntohl(tcph->seq); switch (cm_node->state) { case NES_CM_STATE_SYN_SENT: @@ -1555,6 +1536,12 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, /* passive open, so should not be here */ passive_open_err(cm_node, skb, 1); break; + case NES_CM_STATE_LISTENING: + case NES_CM_STATE_CLOSED: + cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq); + cleanup_retrans_entry(cm_node); + send_reset(cm_node, skb); + break; case NES_CM_STATE_ESTABLISHED: case NES_CM_STATE_FIN_WAIT1: case NES_CM_STATE_FIN_WAIT2: @@ -1562,7 +1549,6 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, case NES_CM_STATE_TSA: case NES_CM_STATE_CLOSING: case NES_CM_STATE_UNKNOWN: - case NES_CM_STATE_CLOSED: case NES_CM_STATE_MPAREQ_SENT: default: drop_packet(skb); @@ -1577,6 +1563,13 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, u32 inc_sequence; u32 rem_seq_ack; u32 rem_seq; + int ret; + int optionsize; + u32 temp_seq = cm_node->tcp_cntxt.loc_seq_num; + + optionsize = (tcph->doff << 2) - sizeof(struct tcphdr); + cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq); + if (check_seq(cm_node, tcph, skb)) return; @@ -1589,7 +1582,18 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, switch (cm_node->state) { case NES_CM_STATE_SYN_RCVD: /* Passive OPEN */ + ret = handle_tcp_options(cm_node, tcph, skb, optionsize, 1); + if (ret) + break; cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq); + cm_node->tcp_cntxt.loc_seq_num = temp_seq; + if (cm_node->tcp_cntxt.rem_ack_num != + cm_node->tcp_cntxt.loc_seq_num) { + nes_debug(NES_DBG_CM, "rem_ack_num != loc_seq_num\n"); + cleanup_retrans_entry(cm_node); + send_reset(cm_node, skb); + return; + } cm_node->state = NES_CM_STATE_ESTABLISHED; if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; @@ -1621,11 +1625,15 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, dev_kfree_skb_any(skb); } break; + case NES_CM_STATE_LISTENING: + case NES_CM_STATE_CLOSED: + cleanup_retrans_entry(cm_node); + send_reset(cm_node, skb); + break; case NES_CM_STATE_FIN_WAIT1: case NES_CM_STATE_SYN_SENT: case NES_CM_STATE_FIN_WAIT2: case NES_CM_STATE_TSA: - case NES_CM_STATE_CLOSED: case NES_CM_STATE_MPAREQ_RCVD: case NES_CM_STATE_LAST_ACK: case NES_CM_STATE_CLOSING: @@ -1648,9 +1656,9 @@ static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, nes_debug(NES_DBG_CM, "%s: Node %p, Sending RESET\n", __func__, cm_node); if (passive) - passive_open_err(cm_node, skb, 0); + passive_open_err(cm_node, skb, 1); else - active_open_err(cm_node, skb, 0); + active_open_err(cm_node, skb, 1); return 1; } } @@ -1970,6 +1978,7 @@ static int mini_cm_reject(struct nes_cm_core *cm_core, struct ietf_mpa_frame *mpa_frame, struct nes_cm_node *cm_node) { int ret = 0; + int passive_state; 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); @@ -1977,9 +1986,13 @@ static int mini_cm_reject(struct nes_cm_core *cm_core, if (cm_node->tcp_cntxt.client) return ret; cleanup_retrans_entry(cm_node); - cm_node->state = NES_CM_STATE_CLOSED; - ret = send_reset(cm_node, NULL); + passive_state = atomic_add_return(1, &cm_node->passive_state); + cm_node->state = NES_CM_STATE_CLOSED; + if (passive_state == NES_SEND_RESET_EVENT) + rem_ref_cm_node(cm_core, cm_node); + else + ret = send_reset(cm_node, NULL); return ret; } @@ -2037,7 +2050,7 @@ static int mini_cm_close(struct nes_cm_core *cm_core, struct nes_cm_node *cm_nod * recv_pkt - recv an ETHERNET packet, and process it through CM * node state machine */ -static void mini_cm_recv_pkt(struct nes_cm_core *cm_core, +static int mini_cm_recv_pkt(struct nes_cm_core *cm_core, struct nes_vnic *nesvnic, struct sk_buff *skb) { struct nes_cm_node *cm_node = NULL; @@ -2045,33 +2058,24 @@ static void mini_cm_recv_pkt(struct nes_cm_core *cm_core, struct iphdr *iph; struct tcphdr *tcph; struct nes_cm_info nfo; + int skb_handled = 1; if (!skb) - return; + return 0; if (skb->len < sizeof(struct iphdr) + sizeof(struct tcphdr)) { - dev_kfree_skb_any(skb); - return; + return 0; } iph = (struct iphdr *)skb->data; tcph = (struct tcphdr *)(skb->data + sizeof(struct iphdr)); - skb_reset_network_header(skb); - skb_set_transport_header(skb, sizeof(*tcph)); - if (!tcph) { - dev_kfree_skb_any(skb); - return; - } - skb->len = ntohs(iph->tot_len); nfo.loc_addr = ntohl(iph->daddr); nfo.loc_port = ntohs(tcph->dest); nfo.rem_addr = ntohl(iph->saddr); nfo.rem_port = ntohs(tcph->source); - nes_debug(NES_DBG_CM, "Received packet: dest=" NIPQUAD_FMT - ":0x%04X src=" NIPQUAD_FMT ":0x%04X\n", - NIPQUAD(iph->daddr), tcph->dest, - NIPQUAD(iph->saddr), tcph->source); + nes_debug(NES_DBG_CM, "Received packet: dest=%pI4:0x%04X src=%pI4:0x%04X\n", + &iph->daddr, tcph->dest, &iph->saddr, tcph->source); do { cm_node = find_node(cm_core, @@ -2082,23 +2086,21 @@ static void mini_cm_recv_pkt(struct nes_cm_core *cm_core, /* Only type of packet accepted are for */ /* the PASSIVE open (syn only) */ if ((!tcph->syn) || (tcph->ack)) { - cm_packets_dropped++; + skb_handled = 0; break; } listener = find_listener(cm_core, nfo.loc_addr, nfo.loc_port, NES_CM_LISTENER_ACTIVE_STATE); - if (listener) { - nfo.cm_id = listener->cm_id; - nfo.conn_type = listener->conn_type; - } else { - nes_debug(NES_DBG_CM, "Unable to find listener " - "for the pkt\n"); - cm_packets_dropped++; - dev_kfree_skb_any(skb); + if (!listener) { + nfo.cm_id = NULL; + nfo.conn_type = 0; + nes_debug(NES_DBG_CM, "Unable to find listener for the pkt\n"); + skb_handled = 0; break; } - + nfo.cm_id = listener->cm_id; + nfo.conn_type = listener->conn_type; cm_node = make_cm_node(cm_core, nesvnic, &nfo, listener); if (!cm_node) { @@ -2124,9 +2126,13 @@ static void mini_cm_recv_pkt(struct nes_cm_core *cm_core, dev_kfree_skb_any(skb); break; } + skb_reset_network_header(skb); + skb_set_transport_header(skb, sizeof(*tcph)); + skb->len = ntohs(iph->tot_len); process_packet(cm_node, skb, cm_core); rem_ref_cm_node(cm_core, cm_node); } while (0); + return skb_handled; } @@ -2135,10 +2141,7 @@ static void mini_cm_recv_pkt(struct nes_cm_core *cm_core, */ static struct nes_cm_core *nes_cm_alloc_core(void) { - int i; - struct nes_cm_core *cm_core; - struct sk_buff *skb = NULL; /* setup the CM core */ /* alloc top level core control structure */ @@ -2156,19 +2159,6 @@ static struct nes_cm_core *nes_cm_alloc_core(void) atomic_set(&cm_core->events_posted, 0); - /* init the packet lists */ - skb_queue_head_init(&cm_core->tx_free_list); - - for (i = 0; i < NES_CM_DEFAULT_FRAME_CNT; i++) { - skb = dev_alloc_skb(cm_core->mtu); - if (!skb) { - kfree(cm_core); - return NULL; - } - /* add 'raw' skb to free frame list */ - skb_queue_head(&cm_core->tx_free_list, skb); - } - cm_core->api = &nes_cm_api; spin_lock_init(&cm_core->ht_lock); @@ -2397,7 +2387,6 @@ static int nes_cm_disconn_true(struct nes_qp *nesqp) atomic_inc(&cm_disconnects); cm_event.event = IW_CM_EVENT_DISCONNECT; if (last_ae == NES_AEQE_AEID_LLP_CONNECTION_RESET) { - issued_disconnect_reset = 1; cm_event.status = IW_CM_EVENT_STATUS_RESET; nes_debug(NES_DBG_CM, "Generating a CM " "Disconnect Event (status reset) for " @@ -2547,6 +2536,7 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) struct nes_v4_quad nes_quad; u32 crc_value; int ret; + int passive_state; ibqp = nes_get_qp(cm_id->device, conn_param->qpn); if (!ibqp) @@ -2714,8 +2704,6 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) conn_param->private_data_len + sizeof(struct ietf_mpa_frame)); - attr.qp_state = IB_QPS_RTS; - nes_modify_qp(&nesqp->ibqp, &attr, IB_QP_STATE, NULL); /* notify OF layer that accept event was successfull */ cm_id->add_ref(cm_id); @@ -2728,6 +2716,8 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) cm_event.private_data = NULL; cm_event.private_data_len = 0; ret = cm_id->event_handler(cm_id, &cm_event); + attr.qp_state = IB_QPS_RTS; + nes_modify_qp(&nesqp->ibqp, &attr, IB_QP_STATE, NULL); if (cm_node->loopbackpartner) { cm_node->loopbackpartner->mpa_frame_size = nesqp->private_data_len; @@ -2740,6 +2730,9 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) printk(KERN_ERR "%s[%u] OFA CM event_handler returned, " "ret=%d\n", __func__, __LINE__, ret); + passive_state = atomic_add_return(1, &cm_node->passive_state); + if (passive_state == NES_SEND_RESET_EVENT) + create_event(cm_node, NES_CM_EVENT_RESET); return 0; } @@ -2943,15 +2936,16 @@ 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)) { - g_cm_core->api->recv_pkt(g_cm_core, netdev_priv(netdevice), skb); + rc = g_cm_core->api->recv_pkt(g_cm_core, netdev_priv(netdevice), skb); } else { nes_debug(NES_DBG_CM, "Unable to process packet for CM," " cm is not setup properly.\n"); } - return 0; + return rc; } @@ -3222,6 +3216,18 @@ static void cm_event_reset(struct nes_cm_event *event) cm_event.private_data_len = 0; ret = cm_id->event_handler(cm_id, &cm_event); + cm_id->add_ref(cm_id); + atomic_inc(&cm_closes); + cm_event.event = IW_CM_EVENT_CLOSE; + cm_event.status = IW_CM_EVENT_STATUS_OK; + cm_event.provider_data = cm_id->provider_data; + cm_event.local_addr = cm_id->local_addr; + cm_event.remote_addr = cm_id->remote_addr; + cm_event.private_data = NULL; + cm_event.private_data_len = 0; + nes_debug(NES_DBG_CM, "NODE %p Generating CLOSE\n", event->cm_node); + ret = cm_id->event_handler(cm_id, &cm_event); + nes_debug(NES_DBG_CM, "OFA CM event_handler returned, ret=%d\n", ret); diff --git a/drivers/infiniband/hw/nes/nes_cm.h b/drivers/infiniband/hw/nes/nes_cm.h index 367b3d290140..fafa35042ebd 100644 --- a/drivers/infiniband/hw/nes/nes_cm.h +++ b/drivers/infiniband/hw/nes/nes_cm.h @@ -76,6 +76,10 @@ enum nes_timer_type { NES_TIMER_TYPE_CLOSE, }; +#define NES_PASSIVE_STATE_INDICATED 0 +#define NES_DO_NOT_SEND_RESET_EVENT 1 +#define NES_SEND_RESET_EVENT 2 + #define MAX_NES_IFS 4 #define SET_ACK 1 @@ -161,6 +165,8 @@ struct nes_timer_entry { #define NES_CM_DEF_SEQ2 0x18ed5740 #define NES_CM_DEF_LOCAL_ID2 0xb807 +#define MAX_CM_BUFFER 512 + typedef u32 nes_addr_t; @@ -254,8 +260,6 @@ struct nes_cm_listener { /* per connection node and node state information */ struct nes_cm_node { - u32 hashkey; - nes_addr_t loc_addr, rem_addr; u16 loc_port, rem_port; @@ -292,7 +296,10 @@ struct nes_cm_node { int apbvt_set; int accept_pend; int freed; + struct list_head timer_entry; + struct list_head reset_entry; struct nes_qp *nesqp; + atomic_t passive_state; }; /* structure for client or CM to fill when making CM api calls. */ @@ -350,7 +357,6 @@ struct nes_cm_core { u32 mtu; u32 free_tx_pkt_max; u32 rx_pkt_posted; - struct sk_buff_head tx_free_list; atomic_t ht_node_cnt; struct list_head connected_nodes; /* struct list_head hashtable[NES_CM_HASHTABLE_SIZE]; */ @@ -390,7 +396,7 @@ struct nes_cm_ops { struct nes_cm_node *); int (*reject)(struct nes_cm_core *, struct ietf_mpa_frame *, struct nes_cm_node *); - void (*recv_pkt)(struct nes_cm_core *, struct nes_vnic *, + int (*recv_pkt)(struct nes_cm_core *, struct nes_vnic *, struct sk_buff *); int (*destroy_cm_core)(struct nes_cm_core *); int (*get)(struct nes_cm_core *); diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c index 7c49cc882d75..5d139db1b771 100644 --- a/drivers/infiniband/hw/nes/nes_hw.c +++ b/drivers/infiniband/hw/nes/nes_hw.c @@ -2541,7 +2541,7 @@ static void nes_nic_napi_ce_handler(struct nes_device *nesdev, struct nes_hw_nic { struct nes_vnic *nesvnic = container_of(cq, struct nes_vnic, nic_cq); - netif_rx_schedule(nesdev->netdev[nesvnic->netdev_index], &nesvnic->napi); + netif_rx_schedule(&nesvnic->napi); } @@ -2700,27 +2700,33 @@ void nes_nic_ce_handler(struct nes_device *nesdev, struct nes_hw_nic_cq *cq) pkt_type, (pkt_type & NES_PKT_TYPE_APBVT_MASK)); */ if ((pkt_type & NES_PKT_TYPE_APBVT_MASK) == NES_PKT_TYPE_APBVT_BITS) { - nes_cm_recv(rx_skb, nesvnic->netdev); + if (nes_cm_recv(rx_skb, nesvnic->netdev)) + rx_skb = NULL; + } + if (rx_skb == NULL) + goto skip_rx_indicate0; + + + if ((cqe_misc & NES_NIC_CQE_TAG_VALID) && + (nesvnic->vlan_grp != NULL)) { + 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 ((cqe_misc & NES_NIC_CQE_TAG_VALID) && (nesvnic->vlan_grp != NULL)) { - 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); - } + if (nes_use_lro) + lro_receive_skb(&nesvnic->lro_mgr, rx_skb, NULL); + else + nes_netif_rx(rx_skb); } +skip_rx_indicate0: nesvnic->netdev->last_rx = jiffies; /* nesvnic->netstats.rx_packets++; */ /* nesvnic->netstats.rx_bytes += rx_pkt_size; */ diff --git a/drivers/infiniband/hw/nes/nes_hw.h b/drivers/infiniband/hw/nes/nes_hw.h index 610b9d859597..bc0b4de04450 100644 --- a/drivers/infiniband/hw/nes/nes_hw.h +++ b/drivers/infiniband/hw/nes/nes_hw.h @@ -40,6 +40,7 @@ #define NES_PHY_TYPE_ARGUS 4 #define NES_PHY_TYPE_PUMA_1G 5 #define NES_PHY_TYPE_PUMA_10G 6 +#define NES_PHY_TYPE_GLADIUS 7 #define NES_MULTICAST_PF_MAX 8 diff --git a/drivers/infiniband/hw/nes/nes_nic.c b/drivers/infiniband/hw/nes/nes_nic.c index 730358637bb6..57a47cf7e513 100644 --- a/drivers/infiniband/hw/nes/nes_nic.c +++ b/drivers/infiniband/hw/nes/nes_nic.c @@ -99,7 +99,6 @@ static int nics_per_function = 1; static int nes_netdev_poll(struct napi_struct *napi, int budget) { struct nes_vnic *nesvnic = container_of(napi, struct nes_vnic, napi); - struct net_device *netdev = nesvnic->netdev; struct nes_device *nesdev = nesvnic->nesdev; struct nes_hw_nic_cq *nescq = &nesvnic->nic_cq; @@ -112,7 +111,7 @@ static int nes_netdev_poll(struct napi_struct *napi, int budget) nes_nic_ce_handler(nesdev, nescq); if (nescq->cqes_pending == 0) { - netif_rx_complete(netdev, napi); + netif_rx_complete(napi); /* clear out completed cqes and arm */ nes_write32(nesdev->regs+NES_CQE_ALLOC, NES_CQE_ALLOC_NOTIFY_NEXT | nescq->cq_number | (nescq->cqe_allocs_pending << 16)); @@ -797,14 +796,13 @@ static int nes_netdev_set_mac_address(struct net_device *netdev, void *p) int i; u32 macaddr_low; u16 macaddr_high; - DECLARE_MAC_BUF(mac); if (!is_valid_ether_addr(mac_addr->sa_data)) return -EADDRNOTAVAIL; memcpy(netdev->dev_addr, mac_addr->sa_data, netdev->addr_len); - printk(PFX "%s: Address length = %d, Address = %s\n", - __func__, netdev->addr_len, print_mac(mac, mac_addr->sa_data)); + printk(PFX "%s: Address length = %d, Address = %pM\n", + __func__, netdev->addr_len, mac_addr->sa_data); macaddr_high = ((u16)netdev->dev_addr[0]) << 8; macaddr_high += (u16)netdev->dev_addr[1]; macaddr_low = ((u32)netdev->dev_addr[2]) << 24; @@ -909,9 +907,8 @@ static void nes_netdev_set_multicast_list(struct net_device *netdev) if (mc_index >= max_pft_entries_avaiable) break; if (multicast_addr) { - DECLARE_MAC_BUF(mac); - nes_debug(NES_DBG_NIC_RX, "Assigning MC Address %s to register 0x%04X nic_idx=%d\n", - print_mac(mac, multicast_addr->dmi_addr), + nes_debug(NES_DBG_NIC_RX, "Assigning MC Address %pM to register 0x%04X nic_idx=%d\n", + multicast_addr->dmi_addr, perfect_filter_register_address+(mc_index * 8), mc_nic_index); macaddr_high = ((u16)multicast_addr->dmi_addr[0]) << 8; diff --git a/drivers/infiniband/hw/nes/nes_utils.c b/drivers/infiniband/hw/nes/nes_utils.c index fb8cbd71a2ef..aa9b7348c728 100644 --- a/drivers/infiniband/hw/nes/nes_utils.c +++ b/drivers/infiniband/hw/nes/nes_utils.c @@ -540,11 +540,14 @@ struct nes_cqp_request *nes_get_cqp_request(struct nes_device *nesdev) if (!list_empty(&nesdev->cqp_avail_reqs)) { spin_lock_irqsave(&nesdev->cqp.lock, flags); - cqp_request = list_entry(nesdev->cqp_avail_reqs.next, + if (!list_empty(&nesdev->cqp_avail_reqs)) { + cqp_request = list_entry(nesdev->cqp_avail_reqs.next, struct nes_cqp_request, list); - list_del_init(&cqp_request->list); + list_del_init(&cqp_request->list); + } spin_unlock_irqrestore(&nesdev->cqp.lock, flags); - } else { + } + if (cqp_request == NULL) { cqp_request = kzalloc(sizeof(struct nes_cqp_request), GFP_KERNEL); if (cqp_request) { cqp_request->dynamic = 1; @@ -679,9 +682,8 @@ int nes_arp_table(struct nes_device *nesdev, u32 ip_addr, u8 *mac_addr, u32 acti /* DELETE or RESOLVE */ if (arp_index == nesadapter->arp_table_size) { - nes_debug(NES_DBG_NETDEV, "MAC for " NIPQUAD_FMT " not in ARP table - cannot %s\n", - HIPQUAD(ip_addr), - action == NES_ARP_RESOLVE ? "resolve" : "delete"); + nes_debug(NES_DBG_NETDEV, "MAC for %pI4 not in ARP table - cannot %s\n", + &ip_addr, action == NES_ARP_RESOLVE ? "resolve" : "delete"); return -1; } diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c index 932e56fcf774..4fdb72454f94 100644 --- a/drivers/infiniband/hw/nes/nes_verbs.c +++ b/drivers/infiniband/hw/nes/nes_verbs.c @@ -220,14 +220,14 @@ static int nes_bind_mw(struct ib_qp *ibqp, struct ib_mw *ibmw, if (nesqp->ibqp_state > IB_QPS_RTS) return -EINVAL; - spin_lock_irqsave(&nesqp->lock, flags); + spin_lock_irqsave(&nesqp->lock, flags); head = nesqp->hwqp.sq_head; qsize = nesqp->hwqp.sq_tail; /* Check for SQ overflow */ if (((head + (2 * qsize) - nesqp->hwqp.sq_tail) % qsize) == (qsize - 1)) { - spin_unlock_irqrestore(&nesqp->lock, flags); + spin_unlock_irqrestore(&nesqp->lock, flags); return -EINVAL; } @@ -269,7 +269,7 @@ static int nes_bind_mw(struct ib_qp *ibqp, struct ib_mw *ibmw, nes_write32(nesdev->regs+NES_WQE_ALLOC, (1 << 24) | 0x00800000 | nesqp->hwqp.qp_id); - spin_unlock_irqrestore(&nesqp->lock, flags); + spin_unlock_irqrestore(&nesqp->lock, flags); return 0; } @@ -349,7 +349,7 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd, if (nesfmr->nesmr.pbls_used > nesadapter->free_4kpbl) { spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); ret = -ENOMEM; - goto failed_vpbl_alloc; + goto failed_vpbl_avail; } else { nesadapter->free_4kpbl -= nesfmr->nesmr.pbls_used; } @@ -357,7 +357,7 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd, if (nesfmr->nesmr.pbls_used > nesadapter->free_256pbl) { spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); ret = -ENOMEM; - goto failed_vpbl_alloc; + goto failed_vpbl_avail; } else { nesadapter->free_256pbl -= nesfmr->nesmr.pbls_used; } @@ -391,14 +391,14 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd, goto failed_vpbl_alloc; } - nesfmr->root_vpbl.leaf_vpbl = kzalloc(sizeof(*nesfmr->root_vpbl.leaf_vpbl)*1024, GFP_KERNEL); + nesfmr->leaf_pbl_cnt = nesfmr->nesmr.pbls_used-1; + nesfmr->root_vpbl.leaf_vpbl = kzalloc(sizeof(*nesfmr->root_vpbl.leaf_vpbl)*1024, GFP_ATOMIC); if (!nesfmr->root_vpbl.leaf_vpbl) { spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); ret = -ENOMEM; goto failed_leaf_vpbl_alloc; } - nesfmr->leaf_pbl_cnt = nesfmr->nesmr.pbls_used-1; nes_debug(NES_DBG_MR, "two level pbl, root_vpbl.pbl_vbase=%p" " leaf_pbl_cnt=%d root_vpbl.leaf_vpbl=%p\n", nesfmr->root_vpbl.pbl_vbase, nesfmr->leaf_pbl_cnt, nesfmr->root_vpbl.leaf_vpbl); @@ -519,6 +519,16 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd, nesfmr->root_vpbl.pbl_pbase); failed_vpbl_alloc: + if (nesfmr->nesmr.pbls_used != 0) { + spin_lock_irqsave(&nesadapter->pbl_lock, flags); + if (nesfmr->nesmr.pbl_4k) + nesadapter->free_4kpbl += nesfmr->nesmr.pbls_used; + else + nesadapter->free_256pbl += nesfmr->nesmr.pbls_used; + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + } + +failed_vpbl_avail: kfree(nesfmr); failed_fmr_alloc: @@ -534,18 +544,14 @@ static struct ib_fmr *nes_alloc_fmr(struct ib_pd *ibpd, */ static int nes_dealloc_fmr(struct ib_fmr *ibfmr) { + unsigned long flags; struct nes_mr *nesmr = to_nesmr_from_ibfmr(ibfmr); struct nes_fmr *nesfmr = to_nesfmr(nesmr); struct nes_vnic *nesvnic = to_nesvnic(ibfmr->device); struct nes_device *nesdev = nesvnic->nesdev; - struct nes_mr temp_nesmr = *nesmr; + struct nes_adapter *nesadapter = nesdev->nesadapter; int i = 0; - temp_nesmr.ibmw.device = ibfmr->device; - temp_nesmr.ibmw.pd = ibfmr->pd; - temp_nesmr.ibmw.rkey = ibfmr->rkey; - temp_nesmr.ibmw.uobject = NULL; - /* free the resources */ if (nesfmr->leaf_pbl_cnt == 0) { /* single PBL case */ @@ -561,8 +567,24 @@ static int nes_dealloc_fmr(struct ib_fmr *ibfmr) pci_free_consistent(nesdev->pcidev, 8192, nesfmr->root_vpbl.pbl_vbase, nesfmr->root_vpbl.pbl_pbase); } + nesmr->ibmw.device = ibfmr->device; + nesmr->ibmw.pd = ibfmr->pd; + nesmr->ibmw.rkey = ibfmr->rkey; + nesmr->ibmw.uobject = NULL; + + if (nesfmr->nesmr.pbls_used != 0) { + spin_lock_irqsave(&nesadapter->pbl_lock, flags); + if (nesfmr->nesmr.pbl_4k) { + nesadapter->free_4kpbl += nesfmr->nesmr.pbls_used; + WARN_ON(nesadapter->free_4kpbl > nesadapter->max_4kpbl); + } else { + nesadapter->free_256pbl += nesfmr->nesmr.pbls_used; + WARN_ON(nesadapter->free_256pbl > nesadapter->max_256pbl); + } + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + } - return nes_dealloc_mw(&temp_nesmr.ibmw); + return nes_dealloc_mw(&nesmr->ibmw); } @@ -1595,7 +1617,7 @@ static struct ib_cq *nes_create_cq(struct ib_device *ibdev, int entries, nes_ucontext->mcrqf = req.mcrqf; if (nes_ucontext->mcrqf) { if (nes_ucontext->mcrqf & 0x80000000) - nescq->hw_cq.cq_number = nesvnic->nic.qp_id + 12 + (nes_ucontext->mcrqf & 0xf) - 1; + nescq->hw_cq.cq_number = nesvnic->nic.qp_id + 28 + 2 * ((nes_ucontext->mcrqf & 0xf) - 1); else if (nes_ucontext->mcrqf & 0x40000000) nescq->hw_cq.cq_number = nes_ucontext->mcrqf & 0xffff; else @@ -1673,13 +1695,8 @@ static struct ib_cq *nes_create_cq(struct ib_device *ibdev, int entries, /* use 4k pbl */ nes_debug(NES_DBG_CQ, "pbl_entries=%u, use a 4k PBL\n", pbl_entries); if (nesadapter->free_4kpbl == 0) { - if (cqp_request->dynamic) { - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - kfree(cqp_request); - } else { - list_add_tail(&cqp_request->list, &nesdev->cqp_avail_reqs); - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - } + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + nes_free_cqp_request(nesdev, cqp_request); if (!context) pci_free_consistent(nesdev->pcidev, nescq->cq_mem_size, mem, nescq->hw_cq.cq_pbase); @@ -1695,13 +1712,8 @@ static struct ib_cq *nes_create_cq(struct ib_device *ibdev, int entries, /* use 256 byte pbl */ nes_debug(NES_DBG_CQ, "pbl_entries=%u, use a 256 byte PBL\n", pbl_entries); if (nesadapter->free_256pbl == 0) { - if (cqp_request->dynamic) { - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - kfree(cqp_request); - } else { - list_add_tail(&cqp_request->list, &nesdev->cqp_avail_reqs); - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - } + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + nes_free_cqp_request(nesdev, cqp_request); if (!context) pci_free_consistent(nesdev->pcidev, nescq->cq_mem_size, mem, nescq->hw_cq.cq_pbase); @@ -1906,13 +1918,8 @@ static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd, /* Two level PBL */ if ((pbl_count+1) > nesadapter->free_4kpbl) { nes_debug(NES_DBG_MR, "Out of 4KB Pbls for two level request.\n"); - if (cqp_request->dynamic) { - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - kfree(cqp_request); - } else { - list_add_tail(&cqp_request->list, &nesdev->cqp_avail_reqs); - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - } + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + nes_free_cqp_request(nesdev, cqp_request); return -ENOMEM; } else { nesadapter->free_4kpbl -= pbl_count+1; @@ -1920,13 +1927,8 @@ static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd, } else if (residual_page_count > 32) { if (pbl_count > nesadapter->free_4kpbl) { nes_debug(NES_DBG_MR, "Out of 4KB Pbls.\n"); - if (cqp_request->dynamic) { - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - kfree(cqp_request); - } else { - list_add_tail(&cqp_request->list, &nesdev->cqp_avail_reqs); - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - } + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + nes_free_cqp_request(nesdev, cqp_request); return -ENOMEM; } else { nesadapter->free_4kpbl -= pbl_count; @@ -1934,13 +1936,8 @@ static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd, } else { if (pbl_count > nesadapter->free_256pbl) { nes_debug(NES_DBG_MR, "Out of 256B Pbls.\n"); - if (cqp_request->dynamic) { - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - kfree(cqp_request); - } else { - list_add_tail(&cqp_request->list, &nesdev->cqp_avail_reqs); - spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); - } + spin_unlock_irqrestore(&nesadapter->pbl_lock, flags); + nes_free_cqp_request(nesdev, cqp_request); return -ENOMEM; } else { nesadapter->free_256pbl -= pbl_count; @@ -3212,7 +3209,7 @@ static int nes_post_send(struct ib_qp *ibqp, struct ib_send_wr *ib_wr, if (nesqp->ibqp_state > IB_QPS_RTS) return -EINVAL; - spin_lock_irqsave(&nesqp->lock, flags); + spin_lock_irqsave(&nesqp->lock, flags); head = nesqp->hwqp.sq_head; @@ -3337,7 +3334,7 @@ static int nes_post_send(struct ib_qp *ibqp, struct ib_send_wr *ib_wr, (counter << 24) | 0x00800000 | nesqp->hwqp.qp_id); } - spin_unlock_irqrestore(&nesqp->lock, flags); + spin_unlock_irqrestore(&nesqp->lock, flags); if (err) *bad_wr = ib_wr; @@ -3368,7 +3365,7 @@ static int nes_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *ib_wr, if (nesqp->ibqp_state > IB_QPS_RTS) return -EINVAL; - spin_lock_irqsave(&nesqp->lock, flags); + spin_lock_irqsave(&nesqp->lock, flags); head = nesqp->hwqp.rq_head; @@ -3421,7 +3418,7 @@ static int nes_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *ib_wr, nes_write32(nesdev->regs+NES_WQE_ALLOC, (counter<<24) | nesqp->hwqp.qp_id); } - spin_unlock_irqrestore(&nesqp->lock, flags); + spin_unlock_irqrestore(&nesqp->lock, flags); if (err) *bad_wr = ib_wr; @@ -3453,7 +3450,7 @@ static int nes_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry) nes_debug(NES_DBG_CQ, "\n"); - spin_lock_irqsave(&nescq->lock, flags); + spin_lock_irqsave(&nescq->lock, flags); head = nescq->hw_cq.cq_head; cq_size = nescq->hw_cq.cq_size; @@ -3562,7 +3559,7 @@ static int nes_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry) nes_debug(NES_DBG_CQ, "Reporting %u completions for CQ%u.\n", cqe_count, nescq->hw_cq.cq_number); - spin_unlock_irqrestore(&nescq->lock, flags); + spin_unlock_irqrestore(&nescq->lock, flags); return cqe_count; } |