diff options
Diffstat (limited to 'drivers/infiniband')
25 files changed, 374 insertions, 224 deletions
diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c index a3a2e9c1639b..df0c4f605a21 100644 --- a/drivers/infiniband/core/umem.c +++ b/drivers/infiniband/core/umem.c @@ -105,6 +105,7 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr, umem->length = size; umem->offset = addr & ~PAGE_MASK; umem->page_size = PAGE_SIZE; + umem->pid = get_task_pid(current, PIDTYPE_PID); /* * We ask for writable memory if any access flags other than * "remote read" are set. "Local write" and "remote write" @@ -198,6 +199,7 @@ out: if (ret < 0) { if (need_release) __ib_umem_release(context->device, umem, 0); + put_pid(umem->pid); kfree(umem); } else current->mm->pinned_vm = locked; @@ -230,15 +232,19 @@ void ib_umem_release(struct ib_umem *umem) { struct ib_ucontext *context = umem->context; struct mm_struct *mm; + struct task_struct *task; unsigned long diff; __ib_umem_release(umem->context->device, umem, 1); - mm = get_task_mm(current); - if (!mm) { - kfree(umem); - return; - } + task = get_pid_task(umem->pid, PIDTYPE_PID); + put_pid(umem->pid); + if (!task) + goto out; + mm = get_task_mm(task); + put_task_struct(task); + if (!mm) + goto out; diff = PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT; @@ -262,9 +268,10 @@ void ib_umem_release(struct ib_umem *umem) } else down_write(&mm->mmap_sem); - current->mm->pinned_vm -= diff; + mm->pinned_vm -= diff; up_write(&mm->mmap_sem); mmput(mm); +out: kfree(umem); } EXPORT_SYMBOL(ib_umem_release); diff --git a/drivers/infiniband/core/uverbs_marshall.c b/drivers/infiniband/core/uverbs_marshall.c index e7bee46868d1..abd97247443e 100644 --- a/drivers/infiniband/core/uverbs_marshall.c +++ b/drivers/infiniband/core/uverbs_marshall.c @@ -140,5 +140,9 @@ void ib_copy_path_rec_from_user(struct ib_sa_path_rec *dst, dst->packet_life_time = src->packet_life_time; dst->preference = src->preference; dst->packet_life_time_selector = src->packet_life_time_selector; + + memset(dst->smac, 0, sizeof(dst->smac)); + memset(dst->dmac, 0, sizeof(dst->dmac)); + dst->vlan_id = 0xffff; } EXPORT_SYMBOL(ib_copy_path_rec_from_user); diff --git a/drivers/infiniband/hw/ipath/ipath_user_pages.c b/drivers/infiniband/hw/ipath/ipath_user_pages.c index dc66c4506916..1da1252dcdb3 100644 --- a/drivers/infiniband/hw/ipath/ipath_user_pages.c +++ b/drivers/infiniband/hw/ipath/ipath_user_pages.c @@ -54,7 +54,7 @@ static void __ipath_release_user_pages(struct page **p, size_t num_pages, /* call with current->mm->mmap_sem held */ static int __ipath_get_user_pages(unsigned long start_page, size_t num_pages, - struct page **p, struct vm_area_struct **vma) + struct page **p) { unsigned long lock_limit; size_t got; @@ -74,7 +74,7 @@ static int __ipath_get_user_pages(unsigned long start_page, size_t num_pages, ret = get_user_pages(current, current->mm, start_page + got * PAGE_SIZE, num_pages - got, 1, 1, - p + got, vma); + p + got, NULL); if (ret < 0) goto bail_release; } @@ -165,7 +165,7 @@ int ipath_get_user_pages(unsigned long start_page, size_t num_pages, down_write(¤t->mm->mmap_sem); - ret = __ipath_get_user_pages(start_page, num_pages, p, NULL); + ret = __ipath_get_user_pages(start_page, num_pages, p); up_write(¤t->mm->mmap_sem); diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index af8256353c7d..bda5994ceb68 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -59,6 +59,7 @@ #define MLX4_IB_FLOW_MAX_PRIO 0xFFF #define MLX4_IB_FLOW_QPN_MASK 0xFFFFFF +#define MLX4_IB_CARD_REV_A0 0xA0 MODULE_AUTHOR("Roland Dreier"); MODULE_DESCRIPTION("Mellanox ConnectX HCA InfiniBand driver"); @@ -119,6 +120,17 @@ static int check_flow_steering_support(struct mlx4_dev *dev) return dmfs; } +static int num_ib_ports(struct mlx4_dev *dev) +{ + int ib_ports = 0; + int i; + + mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_IB) + ib_ports++; + + return ib_ports; +} + static int mlx4_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props) { @@ -126,6 +138,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; int err = -ENOMEM; + int have_ib_ports; in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL); out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL); @@ -142,6 +155,8 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, memset(props, 0, sizeof *props); + have_ib_ports = num_ib_ports(dev->dev); + props->fw_ver = dev->dev->caps.fw_ver; props->device_cap_flags = IB_DEVICE_CHANGE_PHY_PORT | IB_DEVICE_PORT_ACTIVE_EVENT | @@ -152,13 +167,15 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR; if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_QKEY_CNTR) props->device_cap_flags |= IB_DEVICE_BAD_QKEY_CNTR; - if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_APM) + if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_APM && have_ib_ports) props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG; if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_UD_AV_PORT) props->device_cap_flags |= IB_DEVICE_UD_AV_PORT_ENFORCE; if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_IPOIB_CSUM) props->device_cap_flags |= IB_DEVICE_UD_IP_CSUM; - if (dev->dev->caps.max_gso_sz && dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BLH) + if (dev->dev->caps.max_gso_sz && + (dev->dev->rev_id != MLX4_IB_CARD_REV_A0) && + (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BLH)) props->device_cap_flags |= IB_DEVICE_UD_TSO; if (dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_RESERVED_LKEY) props->device_cap_flags |= IB_DEVICE_LOCAL_DMA_LKEY; @@ -357,7 +374,7 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port, props->state = IB_PORT_DOWN; props->phys_state = state_to_phys_state(props->state); props->active_mtu = IB_MTU_256; - spin_lock(&iboe->lock); + spin_lock_bh(&iboe->lock); ndev = iboe->netdevs[port - 1]; if (!ndev) goto out_unlock; @@ -369,7 +386,7 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port, IB_PORT_ACTIVE : IB_PORT_DOWN; props->phys_state = state_to_phys_state(props->state); out_unlock: - spin_unlock(&iboe->lock); + spin_unlock_bh(&iboe->lock); out: mlx4_free_cmd_mailbox(mdev->dev, mailbox); return err; @@ -811,11 +828,11 @@ int mlx4_ib_add_mc(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp, if (!mqp->port) return 0; - spin_lock(&mdev->iboe.lock); + spin_lock_bh(&mdev->iboe.lock); ndev = mdev->iboe.netdevs[mqp->port - 1]; if (ndev) dev_hold(ndev); - spin_unlock(&mdev->iboe.lock); + spin_unlock_bh(&mdev->iboe.lock); if (ndev) { ret = 1; @@ -1292,11 +1309,11 @@ static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid) mutex_lock(&mqp->mutex); ge = find_gid_entry(mqp, gid->raw); if (ge) { - spin_lock(&mdev->iboe.lock); + spin_lock_bh(&mdev->iboe.lock); ndev = ge->added ? mdev->iboe.netdevs[ge->port - 1] : NULL; if (ndev) dev_hold(ndev); - spin_unlock(&mdev->iboe.lock); + spin_unlock_bh(&mdev->iboe.lock); if (ndev) dev_put(ndev); list_del(&ge->list); @@ -1417,6 +1434,9 @@ static void update_gids_task(struct work_struct *work) int err; struct mlx4_dev *dev = gw->dev->dev; + if (!gw->dev->ib_active) + return; + mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) { pr_warn("update gid table failed %ld\n", PTR_ERR(mailbox)); @@ -1447,6 +1467,9 @@ static void reset_gids_task(struct work_struct *work) int err; struct mlx4_dev *dev = gw->dev->dev; + if (!gw->dev->ib_active) + return; + mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) { pr_warn("reset gid table failed\n"); @@ -1581,7 +1604,7 @@ static int mlx4_ib_addr_event(int event, struct net_device *event_netdev, return 0; iboe = &ibdev->iboe; - spin_lock(&iboe->lock); + spin_lock_bh(&iboe->lock); for (port = 1; port <= ibdev->dev->caps.num_ports; ++port) if ((netif_is_bond_master(real_dev) && @@ -1591,7 +1614,7 @@ static int mlx4_ib_addr_event(int event, struct net_device *event_netdev, update_gid_table(ibdev, port, gid, event == NETDEV_DOWN, 0); - spin_unlock(&iboe->lock); + spin_unlock_bh(&iboe->lock); return 0; } @@ -1664,13 +1687,21 @@ static void mlx4_ib_update_qps(struct mlx4_ib_dev *ibdev, new_smac = mlx4_mac_to_u64(dev->dev_addr); read_unlock(&dev_base_lock); + atomic64_set(&ibdev->iboe.mac[port - 1], new_smac); + + /* no need for update QP1 and mac registration in non-SRIOV */ + if (!mlx4_is_mfunc(ibdev->dev)) + return; + mutex_lock(&ibdev->qp1_proxy_lock[port - 1]); qp = ibdev->qp1_proxy[port - 1]; if (qp) { int new_smac_index; - u64 old_smac = qp->pri.smac; + u64 old_smac; struct mlx4_update_qp_params update_params; + mutex_lock(&qp->mutex); + old_smac = qp->pri.smac; if (new_smac == old_smac) goto unlock; @@ -1680,22 +1711,25 @@ static void mlx4_ib_update_qps(struct mlx4_ib_dev *ibdev, goto unlock; update_params.smac_index = new_smac_index; - if (mlx4_update_qp(ibdev->dev, &qp->mqp, MLX4_UPDATE_QP_SMAC, + if (mlx4_update_qp(ibdev->dev, qp->mqp.qpn, MLX4_UPDATE_QP_SMAC, &update_params)) { release_mac = new_smac; goto unlock; } - + /* if old port was zero, no mac was yet registered for this QP */ + if (qp->pri.smac_port) + release_mac = old_smac; qp->pri.smac = new_smac; + qp->pri.smac_port = port; qp->pri.smac_index = new_smac_index; - - release_mac = old_smac; } unlock: - mutex_unlock(&ibdev->qp1_proxy_lock[port - 1]); if (release_mac != MLX4_IB_INVALID_MAC) mlx4_unregister_mac(ibdev->dev, port, release_mac); + if (qp) + mutex_unlock(&qp->mutex); + mutex_unlock(&ibdev->qp1_proxy_lock[port - 1]); } static void mlx4_ib_get_dev_addr(struct net_device *dev, @@ -1706,6 +1740,7 @@ static void mlx4_ib_get_dev_addr(struct net_device *dev, struct inet6_dev *in6_dev; union ib_gid *pgid; struct inet6_ifaddr *ifp; + union ib_gid default_gid; #endif union ib_gid gid; @@ -1726,12 +1761,15 @@ static void mlx4_ib_get_dev_addr(struct net_device *dev, in_dev_put(in_dev); } #if IS_ENABLED(CONFIG_IPV6) + mlx4_make_default_gid(dev, &default_gid); /* IPv6 gids */ in6_dev = in6_dev_get(dev); if (in6_dev) { read_lock_bh(&in6_dev->lock); list_for_each_entry(ifp, &in6_dev->addr_list, if_list) { pgid = (union ib_gid *)&ifp->addr; + if (!memcmp(pgid, &default_gid, sizeof(*pgid))) + continue; update_gid_table(ibdev, port, pgid, 0, 0); } read_unlock_bh(&in6_dev->lock); @@ -1753,24 +1791,33 @@ static int mlx4_ib_init_gid_table(struct mlx4_ib_dev *ibdev) struct net_device *dev; struct mlx4_ib_iboe *iboe = &ibdev->iboe; int i; + int err = 0; - for (i = 1; i <= ibdev->num_ports; ++i) - if (reset_gid_table(ibdev, i)) - return -1; + for (i = 1; i <= ibdev->num_ports; ++i) { + if (rdma_port_get_link_layer(&ibdev->ib_dev, i) == + IB_LINK_LAYER_ETHERNET) { + err = reset_gid_table(ibdev, i); + if (err) + goto out; + } + } read_lock(&dev_base_lock); - spin_lock(&iboe->lock); + spin_lock_bh(&iboe->lock); for_each_netdev(&init_net, dev) { u8 port = mlx4_ib_get_dev_port(dev, ibdev); - if (port) + /* port will be non-zero only for ETH ports */ + if (port) { + mlx4_ib_set_default_gid(ibdev, dev, port); mlx4_ib_get_dev_addr(dev, ibdev, port); + } } - spin_unlock(&iboe->lock); + spin_unlock_bh(&iboe->lock); read_unlock(&dev_base_lock); - - return 0; +out: + return err; } static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev, @@ -1784,7 +1831,7 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev, iboe = &ibdev->iboe; - spin_lock(&iboe->lock); + spin_lock_bh(&iboe->lock); mlx4_foreach_ib_transport_port(port, ibdev->dev) { enum ib_port_state port_state = IB_PORT_NOP; struct net_device *old_master = iboe->masters[port - 1]; @@ -1816,35 +1863,47 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev, port_state = (netif_running(curr_netdev) && netif_carrier_ok(curr_netdev)) ? IB_PORT_ACTIVE : IB_PORT_DOWN; mlx4_ib_set_default_gid(ibdev, curr_netdev, port); - } else { - reset_gid_table(ibdev, port); - } - /* if using bonding/team and a slave port is down, we don't the bond IP - * based gids in the table since flows that select port by gid may get - * the down port. - */ - if (curr_master && (port_state == IB_PORT_DOWN)) { - reset_gid_table(ibdev, port); - mlx4_ib_set_default_gid(ibdev, curr_netdev, port); - } - /* if bonding is used it is possible that we add it to masters - * only after IP address is assigned to the net bonding - * interface. - */ - if (curr_master && (old_master != curr_master)) { - reset_gid_table(ibdev, port); - mlx4_ib_set_default_gid(ibdev, curr_netdev, port); - mlx4_ib_get_dev_addr(curr_master, ibdev, port); - } + if (curr_master) { + /* if using bonding/team and a slave port is down, we + * don't want the bond IP based gids in the table since + * flows that select port by gid may get the down port. + */ + if (port_state == IB_PORT_DOWN) { + reset_gid_table(ibdev, port); + mlx4_ib_set_default_gid(ibdev, + curr_netdev, + port); + } else { + /* gids from the upper dev (bond/team) + * should appear in port's gid table + */ + mlx4_ib_get_dev_addr(curr_master, + ibdev, port); + } + } + /* if bonding is used it is possible that we add it to + * masters only after IP address is assigned to the + * net bonding interface. + */ + if (curr_master && (old_master != curr_master)) { + reset_gid_table(ibdev, port); + mlx4_ib_set_default_gid(ibdev, + curr_netdev, port); + mlx4_ib_get_dev_addr(curr_master, ibdev, port); + } - if (!curr_master && (old_master != curr_master)) { + if (!curr_master && (old_master != curr_master)) { + reset_gid_table(ibdev, port); + mlx4_ib_set_default_gid(ibdev, + curr_netdev, port); + mlx4_ib_get_dev_addr(curr_netdev, ibdev, port); + } + } else { reset_gid_table(ibdev, port); - mlx4_ib_set_default_gid(ibdev, curr_netdev, port); - mlx4_ib_get_dev_addr(curr_netdev, ibdev, port); } } - spin_unlock(&iboe->lock); + spin_unlock_bh(&iboe->lock); if (update_qps_port > 0) mlx4_ib_update_qps(ibdev, dev, update_qps_port); @@ -2186,6 +2245,9 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) goto err_steer_free_bitmap; } + for (j = 1; j <= ibdev->dev->caps.num_ports; j++) + atomic64_set(&iboe->mac[j - 1], ibdev->dev->caps.def_mac[j]); + if (ib_register_device(&ibdev->ib_dev, NULL)) goto err_steer_free_bitmap; @@ -2222,12 +2284,8 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) } } #endif - for (i = 1 ; i <= ibdev->num_ports ; ++i) - reset_gid_table(ibdev, i); - rtnl_lock(); - mlx4_ib_scan_netdevs(ibdev, NULL, 0); - rtnl_unlock(); - mlx4_ib_init_gid_table(ibdev); + if (mlx4_ib_init_gid_table(ibdev)) + goto err_notif; } for (j = 0; j < ARRAY_SIZE(mlx4_class_attributes); ++j) { @@ -2375,6 +2433,9 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr) struct mlx4_ib_dev *ibdev = ibdev_ptr; int p; + ibdev->ib_active = false; + flush_workqueue(wq); + mlx4_ib_close_sriov(ibdev); mlx4_ib_mad_cleanup(ibdev); ib_unregister_device(&ibdev->ib_dev); diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index e8cad3926bfc..6eb743f65f6f 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -451,6 +451,7 @@ struct mlx4_ib_iboe { spinlock_t lock; struct net_device *netdevs[MLX4_MAX_PORTS]; struct net_device *masters[MLX4_MAX_PORTS]; + atomic64_t mac[MLX4_MAX_PORTS]; struct notifier_block nb; struct notifier_block nb_inet; struct notifier_block nb_inet6; diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c index 9b0e80e59b08..8f9325cfc85d 100644 --- a/drivers/infiniband/hw/mlx4/mr.c +++ b/drivers/infiniband/hw/mlx4/mr.c @@ -234,14 +234,13 @@ int mlx4_ib_rereg_user_mr(struct ib_mr *mr, int flags, 0); if (IS_ERR(mmr->umem)) { err = PTR_ERR(mmr->umem); + /* Prevent mlx4_ib_dereg_mr from free'ing invalid pointer */ mmr->umem = NULL; goto release_mpt_entry; } n = ib_umem_page_count(mmr->umem); shift = ilog2(mmr->umem->page_size); - mmr->mmr.iova = virt_addr; - mmr->mmr.size = length; err = mlx4_mr_rereg_mem_write(dev->dev, &mmr->mmr, virt_addr, length, n, shift, *pmpt_entry); @@ -249,6 +248,8 @@ int mlx4_ib_rereg_user_mr(struct ib_mr *mr, int flags, ib_umem_release(mmr->umem); goto release_mpt_entry; } + mmr->mmr.iova = virt_addr; + mmr->mmr.size = length; err = mlx4_ib_umem_write_mtt(dev, &mmr->mmr.mtt, mmr->umem); if (err) { @@ -262,6 +263,8 @@ int mlx4_ib_rereg_user_mr(struct ib_mr *mr, int flags, * return a failure. But dereg_mr will free the resources. */ err = mlx4_mr_hw_write_mpt(dev->dev, &mmr->mmr, pmpt_entry); + if (!err && flags & IB_MR_REREG_ACCESS) + mmr->mmr.access = mr_access_flags; release_mpt_entry: mlx4_mr_hw_put_mpt(dev->dev, pmpt_entry); diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index efb9eff8906c..9c5150c3cb31 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -964,9 +964,10 @@ static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, MLX4_QP_STATE_RST, NULL, 0, 0, &qp->mqp)) pr_warn("modify QP %06x to RESET failed.\n", qp->mqp.qpn); - if (qp->pri.smac) { + if (qp->pri.smac || (!qp->pri.smac && qp->pri.smac_port)) { mlx4_unregister_mac(dev->dev, qp->pri.smac_port, qp->pri.smac); qp->pri.smac = 0; + qp->pri.smac_port = 0; } if (qp->alt.smac) { mlx4_unregister_mac(dev->dev, qp->alt.smac_port, qp->alt.smac); @@ -1325,7 +1326,8 @@ static int _mlx4_set_path(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah, * If one was already assigned, but the new mac differs, * unregister the old one and register the new one. */ - if (!smac_info->smac || smac_info->smac != smac) { + if ((!smac_info->smac && !smac_info->smac_port) || + smac_info->smac != smac) { /* register candidate now, unreg if needed, after success */ smac_index = mlx4_register_mac(dev->dev, port, smac); if (smac_index >= 0) { @@ -1390,21 +1392,13 @@ static void update_mcg_macs(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp) static int handle_eth_ud_smac_index(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, u8 *smac, struct mlx4_qp_context *context) { - struct net_device *ndev; u64 u64_mac; int smac_index; - - ndev = dev->iboe.netdevs[qp->port - 1]; - if (ndev) { - smac = ndev->dev_addr; - u64_mac = mlx4_mac_to_u64(smac); - } else { - u64_mac = dev->dev->caps.def_mac[qp->port]; - } + u64_mac = atomic64_read(&dev->iboe.mac[qp->port - 1]); context->pri_path.sched_queue = MLX4_IB_DEFAULT_SCHED_QUEUE | ((qp->port - 1) << 6); - if (!qp->pri.smac) { + if (!qp->pri.smac && !qp->pri.smac_port) { smac_index = mlx4_register_mac(dev->dev, qp->port, u64_mac); if (smac_index >= 0) { qp->pri.candidate_smac_index = smac_index; @@ -1432,6 +1426,12 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, int steer_qp = 0; int err = -EINVAL; + /* APM is not supported under RoCE */ + if (attr_mask & IB_QP_ALT_PATH && + rdma_port_get_link_layer(&dev->ib_dev, qp->port) == + IB_LINK_LAYER_ETHERNET) + return -ENOTSUPP; + context = kzalloc(sizeof *context, GFP_KERNEL); if (!context) return -ENOMEM; @@ -1682,7 +1682,7 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, MLX4_IB_LINK_TYPE_ETH; if (dev->dev->caps.tunnel_offload_mode == MLX4_TUNNEL_OFFLOAD_MODE_VXLAN) { /* set QP to receive both tunneled & non-tunneled packets */ - if (!(context->flags & (1 << MLX4_RSS_QPC_FLAG_OFFSET))) + if (!(context->flags & cpu_to_be32(1 << MLX4_RSS_QPC_FLAG_OFFSET))) context->srqn = cpu_to_be32(7 << 28); } } @@ -1786,9 +1786,10 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp, if (qp->flags & MLX4_IB_QP_NETIF) mlx4_ib_steer_qp_reg(dev, qp, 0); } - if (qp->pri.smac) { + if (qp->pri.smac || (!qp->pri.smac && qp->pri.smac_port)) { mlx4_unregister_mac(dev->dev, qp->pri.smac_port, qp->pri.smac); qp->pri.smac = 0; + qp->pri.smac_port = 0; } if (qp->alt.smac) { mlx4_unregister_mac(dev->dev, qp->alt.smac_port, qp->alt.smac); @@ -1812,11 +1813,12 @@ out: if (err && steer_qp) mlx4_ib_steer_qp_reg(dev, qp, 0); kfree(context); - if (qp->pri.candidate_smac) { + if (qp->pri.candidate_smac || + (!qp->pri.candidate_smac && qp->pri.candidate_smac_port)) { if (err) { mlx4_unregister_mac(dev->dev, qp->pri.candidate_smac_port, qp->pri.candidate_smac); } else { - if (qp->pri.smac) + if (qp->pri.smac || (!qp->pri.smac && qp->pri.smac_port)) mlx4_unregister_mac(dev->dev, qp->pri.smac_port, qp->pri.smac); qp->pri.smac = qp->pri.candidate_smac; qp->pri.smac_index = qp->pri.candidate_smac_index; @@ -2089,6 +2091,16 @@ static int build_sriov_qp0_header(struct mlx4_ib_sqp *sqp, return 0; } +static void mlx4_u64_to_smac(u8 *dst_mac, u64 src_mac) +{ + int i; + + for (i = ETH_ALEN; i; i--) { + dst_mac[i - 1] = src_mac & 0xff; + src_mac >>= 8; + } +} + static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr, void *wqe, unsigned *mlx_seg_len) { @@ -2203,7 +2215,6 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr, } if (is_eth) { - u8 *smac; struct in6_addr in6; u16 pcp = (be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) >> 29) << 13; @@ -2216,12 +2227,17 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr, memcpy(&ctrl->imm, ah->av.eth.mac + 2, 4); memcpy(&in6, sgid.raw, sizeof(in6)); - if (!mlx4_is_mfunc(to_mdev(ib_dev)->dev)) - smac = to_mdev(sqp->qp.ibqp.device)-> - iboe.netdevs[sqp->qp.port - 1]->dev_addr; - else /* use the src mac of the tunnel */ - smac = ah->av.eth.s_mac; - memcpy(sqp->ud_header.eth.smac_h, smac, 6); + if (!mlx4_is_mfunc(to_mdev(ib_dev)->dev)) { + u64 mac = atomic64_read(&to_mdev(ib_dev)->iboe.mac[sqp->qp.port - 1]); + u8 smac[ETH_ALEN]; + + mlx4_u64_to_smac(smac, mac); + memcpy(sqp->ud_header.eth.smac_h, smac, ETH_ALEN); + } else { + /* use the src mac of the tunnel */ + memcpy(sqp->ud_header.eth.smac_h, ah->av.eth.s_mac, ETH_ALEN); + } + if (!memcmp(sqp->ud_header.eth.smac_h, sqp->ud_header.eth.dmac_h, 6)) mlx->flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK); if (!is_vlan) { diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c index e4056279166d..10cfce5119a9 100644 --- a/drivers/infiniband/hw/mlx5/cq.c +++ b/drivers/infiniband/hw/mlx5/cq.c @@ -752,7 +752,7 @@ struct ib_cq *mlx5_ib_create_cq(struct ib_device *ibdev, int entries, return ERR_PTR(-EINVAL); entries = roundup_pow_of_two(entries + 1); - if (entries > dev->mdev->caps.max_cqes) + if (entries > dev->mdev->caps.gen.max_cqes) return ERR_PTR(-EINVAL); cq = kzalloc(sizeof(*cq), GFP_KERNEL); @@ -919,7 +919,7 @@ int mlx5_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period) int err; u32 fsel; - if (!(dev->mdev->caps.flags & MLX5_DEV_CAP_FLAG_CQ_MODER)) + if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_CQ_MODER)) return -ENOSYS; in = kzalloc(sizeof(*in), GFP_KERNEL); @@ -1074,7 +1074,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) int uninitialized_var(cqe_size); unsigned long flags; - if (!(dev->mdev->caps.flags & MLX5_DEV_CAP_FLAG_RESIZE_CQ)) { + if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_RESIZE_CQ)) { pr_info("Firmware does not support resize CQ\n"); return -ENOSYS; } @@ -1083,7 +1083,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata) return -EINVAL; entries = roundup_pow_of_two(entries + 1); - if (entries > dev->mdev->caps.max_cqes + 1) + if (entries > dev->mdev->caps.gen.max_cqes + 1) return -EINVAL; if (entries == ibcq->cqe + 1) diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c index b514bbb5610f..657af9a1167c 100644 --- a/drivers/infiniband/hw/mlx5/mad.c +++ b/drivers/infiniband/hw/mlx5/mad.c @@ -129,7 +129,7 @@ int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port) packet_error = be16_to_cpu(out_mad->status); - dev->mdev->caps.ext_port_cap[port - 1] = (!err && !packet_error) ? + dev->mdev->caps.gen.ext_port_cap[port - 1] = (!err && !packet_error) ? MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO : 0; out: diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index d8907b20522a..f3114d1132fb 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -157,11 +157,13 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, struct mlx5_ib_dev *dev = to_mdev(ibdev); struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; + struct mlx5_general_caps *gen; int err = -ENOMEM; int max_rq_sg; int max_sq_sg; u64 flags; + gen = &dev->mdev->caps.gen; in_mad = kzalloc(sizeof(*in_mad), GFP_KERNEL); out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL); if (!in_mad || !out_mad) @@ -183,7 +185,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, IB_DEVICE_PORT_ACTIVE_EVENT | IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN; - flags = dev->mdev->caps.flags; + flags = gen->flags; if (flags & MLX5_DEV_CAP_FLAG_BAD_PKEY_CNTR) props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR; if (flags & MLX5_DEV_CAP_FLAG_BAD_QKEY_CNTR) @@ -213,30 +215,31 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, memcpy(&props->sys_image_guid, out_mad->data + 4, 8); props->max_mr_size = ~0ull; - props->page_size_cap = dev->mdev->caps.min_page_sz; - props->max_qp = 1 << dev->mdev->caps.log_max_qp; - props->max_qp_wr = dev->mdev->caps.max_wqes; - max_rq_sg = dev->mdev->caps.max_rq_desc_sz / sizeof(struct mlx5_wqe_data_seg); - max_sq_sg = (dev->mdev->caps.max_sq_desc_sz - sizeof(struct mlx5_wqe_ctrl_seg)) / + props->page_size_cap = gen->min_page_sz; + props->max_qp = 1 << gen->log_max_qp; + props->max_qp_wr = gen->max_wqes; + max_rq_sg = gen->max_rq_desc_sz / sizeof(struct mlx5_wqe_data_seg); + max_sq_sg = (gen->max_sq_desc_sz - sizeof(struct mlx5_wqe_ctrl_seg)) / sizeof(struct mlx5_wqe_data_seg); props->max_sge = min(max_rq_sg, max_sq_sg); - props->max_cq = 1 << dev->mdev->caps.log_max_cq; - props->max_cqe = dev->mdev->caps.max_cqes - 1; - props->max_mr = 1 << dev->mdev->caps.log_max_mkey; - props->max_pd = 1 << dev->mdev->caps.log_max_pd; - props->max_qp_rd_atom = dev->mdev->caps.max_ra_req_qp; - props->max_qp_init_rd_atom = dev->mdev->caps.max_ra_res_qp; + props->max_cq = 1 << gen->log_max_cq; + props->max_cqe = gen->max_cqes - 1; + props->max_mr = 1 << gen->log_max_mkey; + props->max_pd = 1 << gen->log_max_pd; + props->max_qp_rd_atom = 1 << gen->log_max_ra_req_qp; + props->max_qp_init_rd_atom = 1 << gen->log_max_ra_res_qp; + props->max_srq = 1 << gen->log_max_srq; + props->max_srq_wr = gen->max_srq_wqes - 1; + props->local_ca_ack_delay = gen->local_ca_ack_delay; props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp; - props->max_srq = 1 << dev->mdev->caps.log_max_srq; - props->max_srq_wr = dev->mdev->caps.max_srq_wqes - 1; props->max_srq_sge = max_rq_sg - 1; props->max_fast_reg_page_list_len = (unsigned int)-1; - props->local_ca_ack_delay = dev->mdev->caps.local_ca_ack_delay; + props->local_ca_ack_delay = gen->local_ca_ack_delay; props->atomic_cap = IB_ATOMIC_NONE; props->masked_atomic_cap = IB_ATOMIC_NONE; props->max_pkeys = be16_to_cpup((__be16 *)(out_mad->data + 28)); - props->max_mcast_grp = 1 << dev->mdev->caps.log_max_mcg; - props->max_mcast_qp_attach = dev->mdev->caps.max_qp_mcg; + props->max_mcast_grp = 1 << gen->log_max_mcg; + props->max_mcast_qp_attach = gen->max_qp_mcg; props->max_total_mcast_qp_attach = props->max_mcast_qp_attach * props->max_mcast_grp; props->max_map_per_fmr = INT_MAX; /* no limit in ConnectIB */ @@ -254,10 +257,12 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, struct mlx5_ib_dev *dev = to_mdev(ibdev); struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; + struct mlx5_general_caps *gen; int ext_active_speed; int err = -ENOMEM; - if (port < 1 || port > dev->mdev->caps.num_ports) { + gen = &dev->mdev->caps.gen; + if (port < 1 || port > gen->num_ports) { mlx5_ib_warn(dev, "invalid port number %d\n", port); return -EINVAL; } @@ -288,8 +293,8 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, props->phys_state = out_mad->data[33] >> 4; props->port_cap_flags = be32_to_cpup((__be32 *)(out_mad->data + 20)); props->gid_tbl_len = out_mad->data[50]; - props->max_msg_sz = 1 << to_mdev(ibdev)->mdev->caps.log_max_msg; - props->pkey_tbl_len = to_mdev(ibdev)->mdev->caps.port[port - 1].pkey_table_len; + props->max_msg_sz = 1 << gen->log_max_msg; + props->pkey_tbl_len = gen->port[port - 1].pkey_table_len; props->bad_pkey_cntr = be16_to_cpup((__be16 *)(out_mad->data + 46)); props->qkey_viol_cntr = be16_to_cpup((__be16 *)(out_mad->data + 48)); props->active_width = out_mad->data[31] & 0xf; @@ -316,7 +321,7 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, /* If reported active speed is QDR, check if is FDR-10 */ if (props->active_speed == 4) { - if (dev->mdev->caps.ext_port_cap[port - 1] & + if (gen->ext_port_cap[port - 1] & MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO) { init_query_mad(in_mad); in_mad->attr_id = MLX5_ATTR_EXTENDED_PORT_INFO; @@ -470,6 +475,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, struct mlx5_ib_alloc_ucontext_req_v2 req; struct mlx5_ib_alloc_ucontext_resp resp; struct mlx5_ib_ucontext *context; + struct mlx5_general_caps *gen; struct mlx5_uuar_info *uuari; struct mlx5_uar *uars; int gross_uuars; @@ -480,6 +486,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, int i; size_t reqlen; + gen = &dev->mdev->caps.gen; if (!dev->ib_active) return ERR_PTR(-EAGAIN); @@ -512,14 +519,14 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, num_uars = req.total_num_uuars / MLX5_NON_FP_BF_REGS_PER_PAGE; gross_uuars = num_uars * MLX5_BF_REGS_PER_PAGE; - resp.qp_tab_size = 1 << dev->mdev->caps.log_max_qp; - resp.bf_reg_size = dev->mdev->caps.bf_reg_size; + resp.qp_tab_size = 1 << gen->log_max_qp; + resp.bf_reg_size = gen->bf_reg_size; resp.cache_line_size = L1_CACHE_BYTES; - resp.max_sq_desc_sz = dev->mdev->caps.max_sq_desc_sz; - resp.max_rq_desc_sz = dev->mdev->caps.max_rq_desc_sz; - resp.max_send_wqebb = dev->mdev->caps.max_wqes; - resp.max_recv_wr = dev->mdev->caps.max_wqes; - resp.max_srq_recv_wr = dev->mdev->caps.max_srq_wqes; + resp.max_sq_desc_sz = gen->max_sq_desc_sz; + resp.max_rq_desc_sz = gen->max_rq_desc_sz; + resp.max_send_wqebb = gen->max_wqes; + resp.max_recv_wr = gen->max_wqes; + resp.max_srq_recv_wr = gen->max_srq_wqes; context = kzalloc(sizeof(*context), GFP_KERNEL); if (!context) @@ -565,7 +572,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, mutex_init(&context->db_page_mutex); resp.tot_uuars = req.total_num_uuars; - resp.num_ports = dev->mdev->caps.num_ports; + resp.num_ports = gen->num_ports; err = ib_copy_to_udata(udata, &resp, sizeof(resp) - sizeof(resp.reserved)); if (err) @@ -967,9 +974,11 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context, static void get_ext_port_caps(struct mlx5_ib_dev *dev) { + struct mlx5_general_caps *gen; int port; - for (port = 1; port <= dev->mdev->caps.num_ports; port++) + gen = &dev->mdev->caps.gen; + for (port = 1; port <= gen->num_ports; port++) mlx5_query_ext_port_caps(dev, port); } @@ -977,9 +986,11 @@ static int get_port_caps(struct mlx5_ib_dev *dev) { struct ib_device_attr *dprops = NULL; struct ib_port_attr *pprops = NULL; + struct mlx5_general_caps *gen; int err = 0; int port; + gen = &dev->mdev->caps.gen; pprops = kmalloc(sizeof(*pprops), GFP_KERNEL); if (!pprops) goto out; @@ -994,14 +1005,14 @@ static int get_port_caps(struct mlx5_ib_dev *dev) goto out; } - for (port = 1; port <= dev->mdev->caps.num_ports; port++) { + for (port = 1; port <= gen->num_ports; port++) { err = mlx5_ib_query_port(&dev->ib_dev, port, pprops); if (err) { mlx5_ib_warn(dev, "query_port %d failed %d\n", port, err); break; } - dev->mdev->caps.port[port - 1].pkey_table_len = dprops->max_pkeys; - dev->mdev->caps.port[port - 1].gid_table_len = pprops->gid_tbl_len; + gen->port[port - 1].pkey_table_len = dprops->max_pkeys; + gen->port[port - 1].gid_table_len = pprops->gid_tbl_len; mlx5_ib_dbg(dev, "pkey_table_len %d, gid_table_len %d\n", dprops->max_pkeys, pprops->gid_tbl_len); } @@ -1279,8 +1290,8 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) strlcpy(dev->ib_dev.name, "mlx5_%d", IB_DEVICE_NAME_MAX); dev->ib_dev.owner = THIS_MODULE; dev->ib_dev.node_type = RDMA_NODE_IB_CA; - dev->ib_dev.local_dma_lkey = mdev->caps.reserved_lkey; - dev->num_ports = mdev->caps.num_ports; + dev->ib_dev.local_dma_lkey = mdev->caps.gen.reserved_lkey; + dev->num_ports = mdev->caps.gen.num_ports; dev->ib_dev.phys_port_cnt = dev->num_ports; dev->ib_dev.num_comp_vectors = dev->num_comp_vectors; dev->ib_dev.dma_device = &mdev->pdev->dev; @@ -1355,7 +1366,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) dev->ib_dev.free_fast_reg_page_list = mlx5_ib_free_fast_reg_page_list; dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status; - if (mdev->caps.flags & MLX5_DEV_CAP_FLAG_XRC) { + if (mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_XRC) { dev->ib_dev.alloc_xrcd = mlx5_ib_alloc_xrcd; dev->ib_dev.dealloc_xrcd = mlx5_ib_dealloc_xrcd; dev->ib_dev.uverbs_cmd_mask |= diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 8c574b63d77b..dbfe498870c1 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -158,11 +158,13 @@ static void mlx5_ib_qp_event(struct mlx5_core_qp *qp, int type) static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap, int has_rq, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd) { + struct mlx5_general_caps *gen; int wqe_size; int wq_size; + gen = &dev->mdev->caps.gen; /* Sanity check RQ size before proceeding */ - if (cap->max_recv_wr > dev->mdev->caps.max_wqes) + if (cap->max_recv_wr > gen->max_wqes) return -EINVAL; if (!has_rq) { @@ -182,10 +184,10 @@ static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap, wq_size = roundup_pow_of_two(cap->max_recv_wr) * wqe_size; wq_size = max_t(int, wq_size, MLX5_SEND_WQE_BB); qp->rq.wqe_cnt = wq_size / wqe_size; - if (wqe_size > dev->mdev->caps.max_rq_desc_sz) { + if (wqe_size > gen->max_rq_desc_sz) { mlx5_ib_dbg(dev, "wqe_size %d, max %d\n", wqe_size, - dev->mdev->caps.max_rq_desc_sz); + gen->max_rq_desc_sz); return -EINVAL; } qp->rq.wqe_shift = ilog2(wqe_size); @@ -266,9 +268,11 @@ static int calc_send_wqe(struct ib_qp_init_attr *attr) static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, struct mlx5_ib_qp *qp) { + struct mlx5_general_caps *gen; int wqe_size; int wq_size; + gen = &dev->mdev->caps.gen; if (!attr->cap.max_send_wr) return 0; @@ -277,9 +281,9 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, if (wqe_size < 0) return wqe_size; - if (wqe_size > dev->mdev->caps.max_sq_desc_sz) { + if (wqe_size > gen->max_sq_desc_sz) { mlx5_ib_dbg(dev, "wqe_size(%d) > max_sq_desc_sz(%d)\n", - wqe_size, dev->mdev->caps.max_sq_desc_sz); + wqe_size, gen->max_sq_desc_sz); return -EINVAL; } @@ -292,9 +296,9 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, wq_size = roundup_pow_of_two(attr->cap.max_send_wr * wqe_size); qp->sq.wqe_cnt = wq_size / MLX5_SEND_WQE_BB; - if (qp->sq.wqe_cnt > dev->mdev->caps.max_wqes) { + if (qp->sq.wqe_cnt > gen->max_wqes) { mlx5_ib_dbg(dev, "wqe count(%d) exceeds limits(%d)\n", - qp->sq.wqe_cnt, dev->mdev->caps.max_wqes); + qp->sq.wqe_cnt, gen->max_wqes); return -ENOMEM; } qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB); @@ -309,11 +313,13 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd) { + struct mlx5_general_caps *gen; int desc_sz = 1 << qp->sq.wqe_shift; - if (desc_sz > dev->mdev->caps.max_sq_desc_sz) { + gen = &dev->mdev->caps.gen; + if (desc_sz > gen->max_sq_desc_sz) { mlx5_ib_warn(dev, "desc_sz %d, max_sq_desc_sz %d\n", - desc_sz, dev->mdev->caps.max_sq_desc_sz); + desc_sz, gen->max_sq_desc_sz); return -EINVAL; } @@ -325,9 +331,9 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev, qp->sq.wqe_cnt = ucmd->sq_wqe_count; - if (qp->sq.wqe_cnt > dev->mdev->caps.max_wqes) { + if (qp->sq.wqe_cnt > gen->max_wqes) { mlx5_ib_warn(dev, "wqe_cnt %d, max_wqes %d\n", - qp->sq.wqe_cnt, dev->mdev->caps.max_wqes); + qp->sq.wqe_cnt, gen->max_wqes); return -EINVAL; } @@ -803,16 +809,18 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_resources *devr = &dev->devr; struct mlx5_ib_create_qp_resp resp; struct mlx5_create_qp_mbox_in *in; + struct mlx5_general_caps *gen; struct mlx5_ib_create_qp ucmd; int inlen = sizeof(*in); int err; + gen = &dev->mdev->caps.gen; mutex_init(&qp->mutex); spin_lock_init(&qp->sq.lock); spin_lock_init(&qp->rq.lock); if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) { - if (!(dev->mdev->caps.flags & MLX5_DEV_CAP_FLAG_BLOCK_MCAST)) { + if (!(gen->flags & MLX5_DEV_CAP_FLAG_BLOCK_MCAST)) { mlx5_ib_dbg(dev, "block multicast loopback isn't supported\n"); return -EINVAL; } else { @@ -851,9 +859,9 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, mlx5_ib_dbg(dev, "invalid rq params\n"); return -EINVAL; } - if (ucmd.sq_wqe_count > dev->mdev->caps.max_wqes) { + if (ucmd.sq_wqe_count > gen->max_wqes) { mlx5_ib_dbg(dev, "requested sq_wqe_count (%d) > max allowed (%d)\n", - ucmd.sq_wqe_count, dev->mdev->caps.max_wqes); + ucmd.sq_wqe_count, gen->max_wqes); return -EINVAL; } err = create_user_qp(dev, pd, qp, udata, &in, &resp, &inlen); @@ -1144,6 +1152,7 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata) { + struct mlx5_general_caps *gen; struct mlx5_ib_dev *dev; struct mlx5_ib_qp *qp; u16 xrcdn = 0; @@ -1161,11 +1170,12 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd, } dev = to_mdev(to_mxrcd(init_attr->xrcd)->ibxrcd.device); } + gen = &dev->mdev->caps.gen; switch (init_attr->qp_type) { case IB_QPT_XRC_TGT: case IB_QPT_XRC_INI: - if (!(dev->mdev->caps.flags & MLX5_DEV_CAP_FLAG_XRC)) { + if (!(gen->flags & MLX5_DEV_CAP_FLAG_XRC)) { mlx5_ib_dbg(dev, "XRC not supported\n"); return ERR_PTR(-ENOSYS); } @@ -1272,6 +1282,9 @@ enum { static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate) { + struct mlx5_general_caps *gen; + + gen = &dev->mdev->caps.gen; if (rate == IB_RATE_PORT_CURRENT) { return 0; } else if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_300_GBPS) { @@ -1279,7 +1292,7 @@ static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate) } else { while (rate != IB_RATE_2_5_GBPS && !(1 << (rate + MLX5_STAT_RATE_OFFSET) & - dev->mdev->caps.stat_rate_support)) + gen->stat_rate_support)) --rate; } @@ -1290,8 +1303,10 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, const struct ib_ah_attr *ah, struct mlx5_qp_path *path, u8 port, int attr_mask, u32 path_flags, const struct ib_qp_attr *attr) { + struct mlx5_general_caps *gen; int err; + gen = &dev->mdev->caps.gen; path->fl = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0; path->free_ar = (path_flags & MLX5_PATH_FLAG_FREE_AR) ? 0x80 : 0; @@ -1318,9 +1333,9 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, const struct ib_ah_attr *ah, path->port = port; if (ah->ah_flags & IB_AH_GRH) { - if (ah->grh.sgid_index >= dev->mdev->caps.port[port - 1].gid_table_len) { + if (ah->grh.sgid_index >= gen->port[port - 1].gid_table_len) { pr_err(KERN_ERR "sgid_index (%u) too large. max is %d\n", - ah->grh.sgid_index, dev->mdev->caps.port[port - 1].gid_table_len); + ah->grh.sgid_index, gen->port[port - 1].gid_table_len); return -EINVAL; } @@ -1492,6 +1507,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, struct mlx5_ib_qp *qp = to_mqp(ibqp); struct mlx5_ib_cq *send_cq, *recv_cq; struct mlx5_qp_context *context; + struct mlx5_general_caps *gen; struct mlx5_modify_qp_mbox_in *in; struct mlx5_ib_pd *pd; enum mlx5_qp_state mlx5_cur, mlx5_new; @@ -1500,6 +1516,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, int mlx5_st; int err; + gen = &dev->mdev->caps.gen; in = kzalloc(sizeof(*in), GFP_KERNEL); if (!in) return -ENOMEM; @@ -1539,7 +1556,7 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, err = -EINVAL; goto out; } - context->mtu_msgmax = (attr->path_mtu << 5) | dev->mdev->caps.log_max_msg; + context->mtu_msgmax = (attr->path_mtu << 5) | gen->log_max_msg; } if (attr_mask & IB_QP_DEST_QPN) @@ -1685,9 +1702,11 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, struct mlx5_ib_dev *dev = to_mdev(ibqp->device); struct mlx5_ib_qp *qp = to_mqp(ibqp); enum ib_qp_state cur_state, new_state; + struct mlx5_general_caps *gen; int err = -EINVAL; int port; + gen = &dev->mdev->caps.gen; mutex_lock(&qp->mutex); cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state; @@ -1699,21 +1718,21 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, goto out; if ((attr_mask & IB_QP_PORT) && - (attr->port_num == 0 || attr->port_num > dev->mdev->caps.num_ports)) + (attr->port_num == 0 || attr->port_num > gen->num_ports)) goto out; if (attr_mask & IB_QP_PKEY_INDEX) { port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port; - if (attr->pkey_index >= dev->mdev->caps.port[port - 1].pkey_table_len) + if (attr->pkey_index >= gen->port[port - 1].pkey_table_len) goto out; } if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && - attr->max_rd_atomic > dev->mdev->caps.max_ra_res_qp) + attr->max_rd_atomic > (1 << gen->log_max_ra_res_qp)) goto out; if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && - attr->max_dest_rd_atomic > dev->mdev->caps.max_ra_req_qp) + attr->max_dest_rd_atomic > (1 << gen->log_max_ra_req_qp)) goto out; if (cur_state == new_state && cur_state == IB_QPS_RESET) { @@ -2893,7 +2912,8 @@ static void to_ib_ah_attr(struct mlx5_ib_dev *ibdev, struct ib_ah_attr *ib_ah_at memset(ib_ah_attr, 0, sizeof(*ib_ah_attr)); ib_ah_attr->port_num = path->port; - if (ib_ah_attr->port_num == 0 || ib_ah_attr->port_num > dev->caps.num_ports) + if (ib_ah_attr->port_num == 0 || + ib_ah_attr->port_num > dev->caps.gen.num_ports) return; ib_ah_attr->sl = path->sl & 0xf; @@ -3011,10 +3031,12 @@ struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(ibdev); + struct mlx5_general_caps *gen; struct mlx5_ib_xrcd *xrcd; int err; - if (!(dev->mdev->caps.flags & MLX5_DEV_CAP_FLAG_XRC)) + gen = &dev->mdev->caps.gen; + if (!(gen->flags & MLX5_DEV_CAP_FLAG_XRC)) return ERR_PTR(-ENOSYS); xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL); diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c index 70bd131ba646..97cc1baaa8e3 100644 --- a/drivers/infiniband/hw/mlx5/srq.c +++ b/drivers/infiniband/hw/mlx5/srq.c @@ -238,6 +238,7 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, struct ib_udata *udata) { struct mlx5_ib_dev *dev = to_mdev(pd->device); + struct mlx5_general_caps *gen; struct mlx5_ib_srq *srq; int desc_size; int buf_size; @@ -247,11 +248,12 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, int is_xrc; u32 flgs, xrcdn; + gen = &dev->mdev->caps.gen; /* Sanity check SRQ size before proceeding */ - if (init_attr->attr.max_wr >= dev->mdev->caps.max_srq_wqes) { + if (init_attr->attr.max_wr >= gen->max_srq_wqes) { mlx5_ib_dbg(dev, "max_wr %d, cap %d\n", init_attr->attr.max_wr, - dev->mdev->caps.max_srq_wqes); + gen->max_srq_wqes); return ERR_PTR(-EINVAL); } diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c index 40f8536c10b0..ac02ce4e8040 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c @@ -38,7 +38,7 @@ #define OCRDMA_VID_PCP_SHIFT 0xD static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah, - struct ib_ah_attr *attr, int pdid) + struct ib_ah_attr *attr, union ib_gid *sgid, int pdid) { int status = 0; u16 vlan_tag; bool vlan_enabled = false; @@ -49,8 +49,7 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah, memset(ð, 0, sizeof(eth)); memset(&grh, 0, sizeof(grh)); - ah->sgid_index = attr->grh.sgid_index; - + /* VLAN */ vlan_tag = attr->vlan_id; if (!vlan_tag || (vlan_tag > 0xFFF)) vlan_tag = dev->pvid; @@ -65,15 +64,14 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah, eth.eth_type = cpu_to_be16(OCRDMA_ROCE_ETH_TYPE); eth_sz = sizeof(struct ocrdma_eth_basic); } + /* MAC */ memcpy(ð.smac[0], &dev->nic_info.mac_addr[0], ETH_ALEN); - memcpy(ð.dmac[0], attr->dmac, ETH_ALEN); status = ocrdma_resolve_dmac(dev, attr, ð.dmac[0]); if (status) return status; - status = ocrdma_query_gid(&dev->ibdev, 1, attr->grh.sgid_index, - (union ib_gid *)&grh.sgid[0]); - if (status) - return status; + ah->sgid_index = attr->grh.sgid_index; + memcpy(&grh.sgid[0], sgid->raw, sizeof(union ib_gid)); + memcpy(&grh.dgid[0], attr->grh.dgid.raw, sizeof(attr->grh.dgid.raw)); grh.tclass_flow = cpu_to_be32((6 << 28) | (attr->grh.traffic_class << 24) | @@ -81,8 +79,7 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah, /* 0x1b is next header value in GRH */ grh.pdid_hoplimit = cpu_to_be32((pdid << 16) | (0x1b << 8) | attr->grh.hop_limit); - - memcpy(&grh.dgid[0], attr->grh.dgid.raw, sizeof(attr->grh.dgid.raw)); + /* Eth HDR */ memcpy(&ah->av->eth_hdr, ð, eth_sz); memcpy((u8 *)ah->av + eth_sz, &grh, sizeof(struct ocrdma_grh)); if (vlan_enabled) @@ -98,6 +95,8 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd, struct ib_ah_attr *attr) struct ocrdma_ah *ah; struct ocrdma_pd *pd = get_ocrdma_pd(ibpd); struct ocrdma_dev *dev = get_ocrdma_dev(ibpd->device); + union ib_gid sgid; + u8 zmac[ETH_ALEN]; if (!(attr->ah_flags & IB_AH_GRH)) return ERR_PTR(-EINVAL); @@ -111,7 +110,27 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd, struct ib_ah_attr *attr) status = ocrdma_alloc_av(dev, ah); if (status) goto av_err; - status = set_av_attr(dev, ah, attr, pd->id); + + status = ocrdma_query_gid(&dev->ibdev, 1, attr->grh.sgid_index, &sgid); + if (status) { + pr_err("%s(): Failed to query sgid, status = %d\n", + __func__, status); + goto av_conf_err; + } + + memset(&zmac, 0, ETH_ALEN); + if (pd->uctx && + memcmp(attr->dmac, &zmac, ETH_ALEN)) { + status = rdma_addr_find_dmac_by_grh(&sgid, &attr->grh.dgid, + attr->dmac, &attr->vlan_id); + if (status) { + pr_err("%s(): Failed to resolve dmac from gid." + "status = %d\n", __func__, status); + goto av_conf_err; + } + } + + status = set_av_attr(dev, ah, attr, &sgid, pd->id); if (status) goto av_conf_err; @@ -145,7 +164,7 @@ int ocrdma_query_ah(struct ib_ah *ibah, struct ib_ah_attr *attr) struct ocrdma_av *av = ah->av; struct ocrdma_grh *grh; attr->ah_flags |= IB_AH_GRH; - if (ah->av->valid & Bit(1)) { + if (ah->av->valid & OCRDMA_AV_VALID) { grh = (struct ocrdma_grh *)((u8 *)ah->av + sizeof(struct ocrdma_eth_vlan)); attr->sl = be16_to_cpu(av->eth_hdr.vlan_tag) >> 13; diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c index acb434d16903..e8b8569788c0 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c @@ -101,7 +101,7 @@ int ocrdma_query_device(struct ib_device *ibdev, struct ib_device_attr *attr) attr->max_srq_sge = dev->attr.max_srq_sge; attr->max_srq_wr = dev->attr.max_rqe; attr->local_ca_ack_delay = dev->attr.local_ca_ack_delay; - attr->max_fast_reg_page_list_len = 0; + attr->max_fast_reg_page_list_len = dev->attr.max_pages_per_frmr; attr->max_pkeys = 1; return 0; } @@ -1870,7 +1870,7 @@ static int ocrdma_build_inline_sges(struct ocrdma_qp *qp, hdr->total_len = ocrdma_sglist_len(wr->sg_list, wr->num_sge); if (unlikely(hdr->total_len > qp->max_inline_data)) { pr_err("%s() supported_len=0x%x,\n" - " unspported len req=0x%x\n", __func__, + " unsupported len req=0x%x\n", __func__, qp->max_inline_data, hdr->total_len); return -EINVAL; } @@ -2846,11 +2846,9 @@ int ocrdma_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags cq_flags) if (cq->first_arm) { ocrdma_ring_cq_db(dev, cq_id, arm_needed, sol_needed, 0); cq->first_arm = false; - goto skip_defer; } - cq->deferred_arm = true; -skip_defer: + cq->deferred_arm = true; cq->deferred_sol = sol_needed; spin_unlock_irqrestore(&cq->cq_lock, flags); diff --git a/drivers/infiniband/hw/qib/qib_debugfs.c b/drivers/infiniband/hw/qib/qib_debugfs.c index 799a0c3bffc4..6abd3ed3cd51 100644 --- a/drivers/infiniband/hw/qib/qib_debugfs.c +++ b/drivers/infiniband/hw/qib/qib_debugfs.c @@ -193,6 +193,7 @@ static void *_qp_stats_seq_start(struct seq_file *s, loff_t *pos) struct qib_qp_iter *iter; loff_t n = *pos; + rcu_read_lock(); iter = qib_qp_iter_init(s->private); if (!iter) return NULL; @@ -224,7 +225,7 @@ static void *_qp_stats_seq_next(struct seq_file *s, void *iter_ptr, static void _qp_stats_seq_stop(struct seq_file *s, void *iter_ptr) { - /* nothing for now */ + rcu_read_unlock(); } static int _qp_stats_seq_show(struct seq_file *s, void *iter_ptr) diff --git a/drivers/infiniband/hw/qib/qib_qp.c b/drivers/infiniband/hw/qib/qib_qp.c index 7fcc150d603c..6ddc0264aad2 100644 --- a/drivers/infiniband/hw/qib/qib_qp.c +++ b/drivers/infiniband/hw/qib/qib_qp.c @@ -1325,7 +1325,6 @@ int qib_qp_iter_next(struct qib_qp_iter *iter) struct qib_qp *pqp = iter->qp; struct qib_qp *qp; - rcu_read_lock(); for (; n < dev->qp_table_size; n++) { if (pqp) qp = rcu_dereference(pqp->next); @@ -1333,18 +1332,11 @@ int qib_qp_iter_next(struct qib_qp_iter *iter) qp = rcu_dereference(dev->qp_table[n]); pqp = qp; if (qp) { - if (iter->qp) - atomic_dec(&iter->qp->refcount); - atomic_inc(&qp->refcount); - rcu_read_unlock(); iter->qp = qp; iter->n = n; return 0; } } - rcu_read_unlock(); - if (iter->qp) - atomic_dec(&iter->qp->refcount); return ret; } diff --git a/drivers/infiniband/hw/qib/qib_user_pages.c b/drivers/infiniband/hw/qib/qib_user_pages.c index 2bc1d2b96298..74f90b2619f6 100644 --- a/drivers/infiniband/hw/qib/qib_user_pages.c +++ b/drivers/infiniband/hw/qib/qib_user_pages.c @@ -52,7 +52,7 @@ static void __qib_release_user_pages(struct page **p, size_t num_pages, * Call with current->mm->mmap_sem held. */ static int __qib_get_user_pages(unsigned long start_page, size_t num_pages, - struct page **p, struct vm_area_struct **vma) + struct page **p) { unsigned long lock_limit; size_t got; @@ -69,7 +69,7 @@ static int __qib_get_user_pages(unsigned long start_page, size_t num_pages, ret = get_user_pages(current, current->mm, start_page + got * PAGE_SIZE, num_pages - got, 1, 1, - p + got, vma); + p + got, NULL); if (ret < 0) goto bail_release; } @@ -136,7 +136,7 @@ int qib_get_user_pages(unsigned long start_page, size_t num_pages, down_write(¤t->mm->mmap_sem); - ret = __qib_get_user_pages(start_page, num_pages, p, NULL); + ret = __qib_get_user_pages(start_page, num_pages, p); up_write(¤t->mm->mmap_sem); diff --git a/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c b/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c index f8dfd76be89f..db3588df3546 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c @@ -511,7 +511,7 @@ int usnic_ib_qp_grp_modify(struct usnic_ib_qp_grp *qp_grp, usnic_ib_qp_grp_state_to_string(old_state), usnic_ib_qp_grp_state_to_string(new_state)); } else { - usnic_err("Failed to transistion %u from %s to %s", + usnic_err("Failed to transition %u from %s to %s", qp_grp->grp_id, usnic_ib_qp_grp_state_to_string(old_state), usnic_ib_qp_grp_state_to_string(new_state)); diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index 3edce617c31b..d7562beb5423 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -131,6 +131,12 @@ struct ipoib_cb { u8 hwaddr[INFINIBAND_ALEN]; }; +static inline struct ipoib_cb *ipoib_skb_cb(const struct sk_buff *skb) +{ + BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct ipoib_cb)); + return (struct ipoib_cb *)skb->cb; +} + /* Used for all multicast joins (broadcast, IPv4 mcast and IPv6 mcast) */ struct ipoib_mcast { struct ib_sa_mcmember_rec mcmember; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c index 1310acf6bf92..58b5aa3b6f2d 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c @@ -716,7 +716,7 @@ static int ipoib_start_xmit(struct sk_buff *skb, struct net_device *dev) { struct ipoib_dev_priv *priv = netdev_priv(dev); struct ipoib_neigh *neigh; - struct ipoib_cb *cb = (struct ipoib_cb *) skb->cb; + struct ipoib_cb *cb = ipoib_skb_cb(skb); struct ipoib_header *header; unsigned long flags; @@ -813,7 +813,7 @@ static int ipoib_hard_header(struct sk_buff *skb, const void *daddr, const void *saddr, unsigned len) { struct ipoib_header *header; - struct ipoib_cb *cb = (struct ipoib_cb *) skb->cb; + struct ipoib_cb *cb = ipoib_skb_cb(skb); header = (struct ipoib_header *) skb_push(skb, sizeof *header); @@ -1364,7 +1364,7 @@ void ipoib_setup(struct net_device *dev) dev->tx_queue_len = ipoib_sendq_size * 2; dev->features = (NETIF_F_VLAN_CHALLENGED | NETIF_F_HIGHDMA); - dev->priv_flags &= ~IFF_XMIT_DST_RELEASE; + netif_keep_dst(dev); memcpy(dev->broadcast, ipv4_bcast_addr, INFINIBAND_ALEN); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index d4e005720d01..ffb83b5f7e80 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -529,21 +529,13 @@ void ipoib_mcast_join_task(struct work_struct *work) port_attr.state); return; } + priv->local_lid = port_attr.lid; if (ib_query_gid(priv->ca, priv->port, 0, &priv->local_gid)) ipoib_warn(priv, "ib_query_gid() failed\n"); else memcpy(priv->dev->dev_addr + 4, priv->local_gid.raw, sizeof (union ib_gid)); - { - struct ib_port_attr attr; - - if (!ib_query_port(priv->ca, priv->port, &attr)) - priv->local_lid = attr.lid; - else - ipoib_warn(priv, "ib_query_port failed\n"); - } - if (!priv->broadcast) { struct ipoib_mcast *broadcast; diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c index 61ee91d88380..93ce62fe1594 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.c +++ b/drivers/infiniband/ulp/iser/iscsi_iser.c @@ -344,7 +344,6 @@ iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session, int is_leading) { struct iscsi_conn *conn = cls_conn->dd_data; - struct iscsi_session *session; struct iser_conn *ib_conn; struct iscsi_endpoint *ep; int error; @@ -363,9 +362,17 @@ iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session, } ib_conn = ep->dd_data; - session = conn->session; - if (iser_alloc_rx_descriptors(ib_conn, session)) - return -ENOMEM; + mutex_lock(&ib_conn->state_mutex); + if (ib_conn->state != ISER_CONN_UP) { + error = -EINVAL; + iser_err("iser_conn %p state is %d, teardown started\n", + ib_conn, ib_conn->state); + goto out; + } + + error = iser_alloc_rx_descriptors(ib_conn, conn->session); + if (error) + goto out; /* binds the iSER connection retrieved from the previously * connected ep_handle to the iSCSI layer connection. exchanges @@ -375,7 +382,9 @@ iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session, conn->dd_data = ib_conn; ib_conn->iscsi_conn = conn; - return 0; +out: + mutex_unlock(&ib_conn->state_mutex); + return error; } static int diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h index c877dad381cb..9f0e0e34d6ca 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.h +++ b/drivers/infiniband/ulp/iser/iscsi_iser.h @@ -69,7 +69,7 @@ #define DRV_NAME "iser" #define PFX DRV_NAME ": " -#define DRV_VER "1.4" +#define DRV_VER "1.4.1" #define iser_dbg(fmt, arg...) \ do { \ diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index 3ef167f97d6f..3bfec4bbda52 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -73,7 +73,7 @@ static int iser_create_device_ib_res(struct iser_device *device) { struct iser_cq_desc *cq_desc; struct ib_device_attr *dev_attr = &device->dev_attr; - int ret, i, j; + int ret, i; ret = ib_query_device(device->ib_device, dev_attr); if (ret) { @@ -125,16 +125,20 @@ static int iser_create_device_ib_res(struct iser_device *device) iser_cq_event_callback, (void *)&cq_desc[i], ISER_MAX_RX_CQ_LEN, i); - if (IS_ERR(device->rx_cq[i])) + if (IS_ERR(device->rx_cq[i])) { + device->rx_cq[i] = NULL; goto cq_err; + } device->tx_cq[i] = ib_create_cq(device->ib_device, NULL, iser_cq_event_callback, (void *)&cq_desc[i], ISER_MAX_TX_CQ_LEN, i); - if (IS_ERR(device->tx_cq[i])) + if (IS_ERR(device->tx_cq[i])) { + device->tx_cq[i] = NULL; goto cq_err; + } if (ib_req_notify_cq(device->rx_cq[i], IB_CQ_NEXT_COMP)) goto cq_err; @@ -160,14 +164,14 @@ static int iser_create_device_ib_res(struct iser_device *device) handler_err: ib_dereg_mr(device->mr); dma_mr_err: - for (j = 0; j < device->cqs_used; j++) - tasklet_kill(&device->cq_tasklet[j]); + for (i = 0; i < device->cqs_used; i++) + tasklet_kill(&device->cq_tasklet[i]); cq_err: - for (j = 0; j < i; j++) { - if (device->tx_cq[j]) - ib_destroy_cq(device->tx_cq[j]); - if (device->rx_cq[j]) - ib_destroy_cq(device->rx_cq[j]); + for (i = 0; i < device->cqs_used; i++) { + if (device->tx_cq[i]) + ib_destroy_cq(device->tx_cq[i]); + if (device->rx_cq[i]) + ib_destroy_cq(device->rx_cq[i]); } ib_dealloc_pd(device->pd); pd_err: diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index d4c7928a0f36..da8ff124762a 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -586,17 +586,12 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) init_completion(&isert_conn->conn_wait); init_completion(&isert_conn->conn_wait_comp_err); kref_init(&isert_conn->conn_kref); - kref_get(&isert_conn->conn_kref); mutex_init(&isert_conn->conn_mutex); spin_lock_init(&isert_conn->conn_lock); INIT_LIST_HEAD(&isert_conn->conn_fr_pool); cma_id->context = isert_conn; isert_conn->conn_cm_id = cma_id; - isert_conn->responder_resources = event->param.conn.responder_resources; - isert_conn->initiator_depth = event->param.conn.initiator_depth; - pr_debug("Using responder_resources: %u initiator_depth: %u\n", - isert_conn->responder_resources, isert_conn->initiator_depth); isert_conn->login_buf = kzalloc(ISCSI_DEF_MAX_RECV_SEG_LEN + ISER_RX_LOGIN_SIZE, GFP_KERNEL); @@ -643,6 +638,12 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) goto out_rsp_dma_map; } + /* Set max inflight RDMA READ requests */ + isert_conn->initiator_depth = min_t(u8, + event->param.conn.initiator_depth, + device->dev_attr.max_qp_init_rd_atom); + pr_debug("Using initiator_depth: %u\n", isert_conn->initiator_depth); + isert_conn->conn_device = device; isert_conn->conn_pd = ib_alloc_pd(isert_conn->conn_device->ib_device); if (IS_ERR(isert_conn->conn_pd)) { @@ -746,7 +747,9 @@ isert_connect_release(struct isert_conn *isert_conn) static void isert_connected_handler(struct rdma_cm_id *cma_id) { - return; + struct isert_conn *isert_conn = cma_id->context; + + kref_get(&isert_conn->conn_kref); } static void @@ -798,7 +801,6 @@ isert_disconnect_work(struct work_struct *work) wake_up: complete(&isert_conn->conn_wait); - isert_put_conn(isert_conn); } static void @@ -3067,7 +3069,6 @@ isert_rdma_accept(struct isert_conn *isert_conn) int ret; memset(&cp, 0, sizeof(struct rdma_conn_param)); - cp.responder_resources = isert_conn->responder_resources; cp.initiator_depth = isert_conn->initiator_depth; cp.retry_count = 7; cp.rnr_retry_count = 7; @@ -3215,7 +3216,7 @@ static void isert_wait_conn(struct iscsi_conn *conn) pr_debug("isert_wait_conn: Starting \n"); mutex_lock(&isert_conn->conn_mutex); - if (isert_conn->conn_cm_id) { + if (isert_conn->conn_cm_id && !isert_conn->disconnect) { pr_debug("Calling rdma_disconnect from isert_wait_conn\n"); rdma_disconnect(isert_conn->conn_cm_id); } @@ -3234,6 +3235,7 @@ static void isert_wait_conn(struct iscsi_conn *conn) wait_for_completion(&isert_conn->conn_wait_comp_err); wait_for_completion(&isert_conn->conn_wait); + isert_put_conn(isert_conn); } static void isert_free_conn(struct iscsi_conn *conn) |