Merge branches 'hfi1' and 'mlx' into k.o/for-4.9-rc

drivers/infiniband/core/addr.c

@@ -699,13 +699,16 @@ EXPORT_SYMBOL(rdma_addr_cancel);
 struct resolve_cb_context {
 	struct rdma_dev_addr *addr;
 	struct completion comp;
+	int status;
 };
 
 static void resolve_cb(int status, struct sockaddr *src_addr,
 	     struct rdma_dev_addr *addr, void *context)
 {
-	memcpy(((struct resolve_cb_context *)context)->addr, addr, sizeof(struct
-				rdma_dev_addr));
+	if (!status)
+		memcpy(((struct resolve_cb_context *)context)->addr,
+		       addr, sizeof(struct rdma_dev_addr));
+	((struct resolve_cb_context *)context)->status = status;
 	complete(&((struct resolve_cb_context *)context)->comp);
 }
 
@@ -743,6 +746,10 @@ int rdma_addr_find_l2_eth_by_grh(const union ib_gid *sgid,
 
 	wait_for_completion(&ctx.comp);
 
+	ret = ctx.status;
+	if (ret)
+		return ret;
+
 	memcpy(dmac, dev_addr.dst_dev_addr, ETH_ALEN);
 	dev = dev_get_by_index(&init_net, dev_addr.bound_dev_if);
 	if (!dev)

drivers/infiniband/core/cm.c

@@ -80,6 +80,8 @@ static struct ib_cm {
 	__be32 random_id_operand;
 	struct list_head timewait_list;
 	struct workqueue_struct *wq;
+	/* Sync on cm change port state */
+	spinlock_t state_lock;
 } cm;
 
 /* Counter indexes ordered by attribute ID */
@@ -161,6 +163,8 @@ struct cm_port {
 	struct ib_mad_agent *mad_agent;
 	struct kobject port_obj;
 	u8 port_num;
+	struct list_head cm_priv_prim_list;
+	struct list_head cm_priv_altr_list;
 	struct cm_counter_group counter_group[CM_COUNTER_GROUPS];
 };
 
@@ -241,6 +245,12 @@ struct cm_id_private {
 	u8 service_timeout;
 	u8 target_ack_delay;
 
+	struct list_head prim_list;
+	struct list_head altr_list;
+	/* Indicates that the send port mad is registered and av is set */
+	int prim_send_port_not_ready;
+	int altr_send_port_not_ready;
+
 	struct list_head work_list;
 	atomic_t work_count;
 };
@@ -259,20 +269,47 @@ static int cm_alloc_msg(struct cm_id_private *cm_id_priv,
 	struct ib_mad_agent *mad_agent;
 	struct ib_mad_send_buf *m;
 	struct ib_ah *ah;
+	struct cm_av *av;
+	unsigned long flags, flags2;
+	int ret = 0;
 
+	/* don't let the port to be released till the agent is down */
+	spin_lock_irqsave(&cm.state_lock, flags2);
+	spin_lock_irqsave(&cm.lock, flags);
+	if (!cm_id_priv->prim_send_port_not_ready)
+		av = &cm_id_priv->av;
+	else if (!cm_id_priv->altr_send_port_not_ready &&
+		 (cm_id_priv->alt_av.port))
+		av = &cm_id_priv->alt_av;
+	else {
+		pr_info("%s: not valid CM id\n", __func__);
+		ret = -ENODEV;
+		spin_unlock_irqrestore(&cm.lock, flags);
+		goto out;
+	}
+	spin_unlock_irqrestore(&cm.lock, flags);
+	/* Make sure the port haven't released the mad yet */
 	mad_agent = cm_id_priv->av.port->mad_agent;
-	ah = ib_create_ah(mad_agent->qp->pd, &cm_id_priv->av.ah_attr);
-	if (IS_ERR(ah))
-		return PTR_ERR(ah);
+	if (!mad_agent) {
+		pr_info("%s: not a valid MAD agent\n", __func__);
+		ret = -ENODEV;
+		goto out;
+	}
+	ah = ib_create_ah(mad_agent->qp->pd, &av->ah_attr);
+	if (IS_ERR(ah)) {
+		ret = PTR_ERR(ah);
+		goto out;
+	}
 
 	m = ib_create_send_mad(mad_agent, cm_id_priv->id.remote_cm_qpn,
-			       cm_id_priv->av.pkey_index,
+			       av->pkey_index,
 			       0, IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA,
 			       GFP_ATOMIC,
 			       IB_MGMT_BASE_VERSION);
 	if (IS_ERR(m)) {
 		ib_destroy_ah(ah);
-		return PTR_ERR(m);
+		ret = PTR_ERR(m);
+		goto out;
 	}
 
 	/* Timeout set by caller if response is expected. */
@@ -282,7 +319,10 @@ static int cm_alloc_msg(struct cm_id_private *cm_id_priv,
 	atomic_inc(&cm_id_priv->refcount);
 	m->context[0] = cm_id_priv;
 	*msg = m;
-	return 0;
+
+out:
+	spin_unlock_irqrestore(&cm.state_lock, flags2);
+	return ret;
 }
 
 static int cm_alloc_response_msg(struct cm_port *port,
@@ -352,7 +392,8 @@ static void cm_init_av_for_response(struct cm_port *port, struct ib_wc *wc,
 			   grh, &av->ah_attr);
 }
 
-static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av)
+static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av,
+			      struct cm_id_private *cm_id_priv)
 {
 	struct cm_device *cm_dev;
 	struct cm_port *port = NULL;
@@ -387,7 +428,17 @@ static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av)
 			     &av->ah_attr);
 	av->timeout = path->packet_life_time + 1;
 
-	return 0;
+	spin_lock_irqsave(&cm.lock, flags);
+	if (&cm_id_priv->av == av)
+		list_add_tail(&cm_id_priv->prim_list, &port->cm_priv_prim_list);
+	else if (&cm_id_priv->alt_av == av)
+		list_add_tail(&cm_id_priv->altr_list, &port->cm_priv_altr_list);
+	else
+		ret = -EINVAL;
+
+	spin_unlock_irqrestore(&cm.lock, flags);
+
+	return ret;
 }
 
 static int cm_alloc_id(struct cm_id_private *cm_id_priv)
@@ -677,6 +728,8 @@ struct ib_cm_id *ib_create_cm_id(struct ib_device *device,
 	spin_lock_init(&cm_id_priv->lock);
 	init_completion(&cm_id_priv->comp);
 	INIT_LIST_HEAD(&cm_id_priv->work_list);
+	INIT_LIST_HEAD(&cm_id_priv->prim_list);
+	INIT_LIST_HEAD(&cm_id_priv->altr_list);
 	atomic_set(&cm_id_priv->work_count, -1);
 	atomic_set(&cm_id_priv->refcount, 1);
 	return &cm_id_priv->id;
@@ -892,6 +945,15 @@ static void cm_destroy_id(struct ib_cm_id *cm_id, int err)
 		break;
 	}
 
+	spin_lock_irq(&cm.lock);
+	if (!list_empty(&cm_id_priv->altr_list) &&
+	    (!cm_id_priv->altr_send_port_not_ready))
+		list_del(&cm_id_priv->altr_list);
+	if (!list_empty(&cm_id_priv->prim_list) &&
+	    (!cm_id_priv->prim_send_port_not_ready))
+		list_del(&cm_id_priv->prim_list);
+	spin_unlock_irq(&cm.lock);
+
 	cm_free_id(cm_id->local_id);
 	cm_deref_id(cm_id_priv);
 	wait_for_completion(&cm_id_priv->comp);
@@ -1192,12 +1254,13 @@ int ib_send_cm_req(struct ib_cm_id *cm_id,
 		goto out;
 	}
 
-	ret = cm_init_av_by_path(param->primary_path, &cm_id_priv->av);
+	ret = cm_init_av_by_path(param->primary_path, &cm_id_priv->av,
+				 cm_id_priv);
 	if (ret)
 		goto error1;
 	if (param->alternate_path) {
 		ret = cm_init_av_by_path(param->alternate_path,
-					 &cm_id_priv->alt_av);
+					 &cm_id_priv->alt_av, cm_id_priv);
 		if (ret)
 			goto error1;
 	}
@@ -1653,7 +1716,8 @@ static int cm_req_handler(struct cm_work *work)
 			dev_put(gid_attr.ndev);
 		}
 		work->path[0].gid_type = gid_attr.gid_type;
-		ret = cm_init_av_by_path(&work->path[0], &cm_id_priv->av);
+		ret = cm_init_av_by_path(&work->path[0], &cm_id_priv->av,
+					 cm_id_priv);
 	}
 	if (ret) {
 		int err = ib_get_cached_gid(work->port->cm_dev->ib_device,
@@ -1672,7 +1736,8 @@ static int cm_req_handler(struct cm_work *work)
 		goto rejected;
 	}
 	if (req_msg->alt_local_lid) {
-		ret = cm_init_av_by_path(&work->path[1], &cm_id_priv->alt_av);
+		ret = cm_init_av_by_path(&work->path[1], &cm_id_priv->alt_av,
+					 cm_id_priv);
 		if (ret) {
 			ib_send_cm_rej(cm_id, IB_CM_REJ_INVALID_ALT_GID,
 				       &work->path[0].sgid,
@@ -2727,7 +2792,8 @@ int ib_send_cm_lap(struct ib_cm_id *cm_id,
 		goto out;
 	}
 
-	ret = cm_init_av_by_path(alternate_path, &cm_id_priv->alt_av);
+	ret = cm_init_av_by_path(alternate_path, &cm_id_priv->alt_av,
+				 cm_id_priv);
 	if (ret)
 		goto out;
 	cm_id_priv->alt_av.timeout =
@@ -2839,7 +2905,8 @@ static int cm_lap_handler(struct cm_work *work)
 	cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
 				work->mad_recv_wc->recv_buf.grh,
 				&cm_id_priv->av);
-	cm_init_av_by_path(param->alternate_path, &cm_id_priv->alt_av);
+	cm_init_av_by_path(param->alternate_path, &cm_id_priv->alt_av,
+			   cm_id_priv);
 	ret = atomic_inc_and_test(&cm_id_priv->work_count);
 	if (!ret)
 		list_add_tail(&work->list, &cm_id_priv->work_list);
@@ -3031,7 +3098,7 @@ int ib_send_cm_sidr_req(struct ib_cm_id *cm_id,
 		return -EINVAL;
 
 	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
-	ret = cm_init_av_by_path(param->path, &cm_id_priv->av);
+	ret = cm_init_av_by_path(param->path, &cm_id_priv->av, cm_id_priv);
 	if (ret)
 		goto out;
 
@@ -3468,7 +3535,9 @@ static int cm_establish(struct ib_cm_id *cm_id)
 static int cm_migrate(struct ib_cm_id *cm_id)
 {
 	struct cm_id_private *cm_id_priv;
+	struct cm_av tmp_av;
 	unsigned long flags;
+	int tmp_send_port_not_ready;
 	int ret = 0;
 
 	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
@@ -3477,7 +3546,14 @@ static int cm_migrate(struct ib_cm_id *cm_id)
 	    (cm_id->lap_state == IB_CM_LAP_UNINIT ||
 	     cm_id->lap_state == IB_CM_LAP_IDLE)) {
 		cm_id->lap_state = IB_CM_LAP_IDLE;
+		/* Swap address vector */
+		tmp_av = cm_id_priv->av;
 		cm_id_priv->av = cm_id_priv->alt_av;
+		cm_id_priv->alt_av = tmp_av;
+		/* Swap port send ready state */
+		tmp_send_port_not_ready = cm_id_priv->prim_send_port_not_ready;
+		cm_id_priv->prim_send_port_not_ready = cm_id_priv->altr_send_port_not_ready;
+		cm_id_priv->altr_send_port_not_ready = tmp_send_port_not_ready;
 	} else
 		ret = -EINVAL;
 	spin_unlock_irqrestore(&cm_id_priv->lock, flags);
@@ -3888,6 +3964,9 @@ static void cm_add_one(struct ib_device *ib_device)
 		port->cm_dev = cm_dev;
 		port->port_num = i;
 
+		INIT_LIST_HEAD(&port->cm_priv_prim_list);
+		INIT_LIST_HEAD(&port->cm_priv_altr_list);
+
 		ret = cm_create_port_fs(port);
 		if (ret)
 			goto error1;
@@ -3945,6 +4024,8 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data)
 {
 	struct cm_device *cm_dev = client_data;
 	struct cm_port *port;
+	struct cm_id_private *cm_id_priv;
+	struct ib_mad_agent *cur_mad_agent;
 	struct ib_port_modify port_modify = {
 		.clr_port_cap_mask = IB_PORT_CM_SUP
 	};
@@ -3968,15 +4049,27 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data)
 
 		port = cm_dev->port[i-1];
 		ib_modify_port(ib_device, port->port_num, 0, &port_modify);
+		/* Mark all the cm_id's as not valid */
+		spin_lock_irq(&cm.lock);
+		list_for_each_entry(cm_id_priv, &port->cm_priv_altr_list, altr_list)
+			cm_id_priv->altr_send_port_not_ready = 1;
+		list_for_each_entry(cm_id_priv, &port->cm_priv_prim_list, prim_list)
+			cm_id_priv->prim_send_port_not_ready = 1;
+		spin_unlock_irq(&cm.lock);
 		/*
 		 * We flush the queue here after the going_down set, this
 		 * verify that no new works will be queued in the recv handler,
 		 * after that we can call the unregister_mad_agent
 		 */
 		flush_workqueue(cm.wq);
-		ib_unregister_mad_agent(port->mad_agent);
+		spin_lock_irq(&cm.state_lock);
+		cur_mad_agent = port->mad_agent;
+		port->mad_agent = NULL;
+		spin_unlock_irq(&cm.state_lock);
+		ib_unregister_mad_agent(cur_mad_agent);
 		cm_remove_port_fs(port);
 	}
+
 	device_unregister(cm_dev->device);
 	kfree(cm_dev);
 }
@@ -3989,6 +4082,7 @@ static int __init ib_cm_init(void)
 	INIT_LIST_HEAD(&cm.device_list);
 	rwlock_init(&cm.device_lock);
 	spin_lock_init(&cm.lock);
+	spin_lock_init(&cm.state_lock);
 	cm.listen_service_table = RB_ROOT;
 	cm.listen_service_id = be64_to_cpu(IB_CM_ASSIGN_SERVICE_ID);
 	cm.remote_id_table = RB_ROOT;

drivers/infiniband/core/cma.c

@@ -2436,6 +2436,18 @@ static int iboe_tos_to_sl(struct net_device *ndev, int tos)
 	return 0;
 }
 
+static enum ib_gid_type cma_route_gid_type(enum rdma_network_type network_type,
+					   unsigned long supported_gids,
+					   enum ib_gid_type default_gid)
+{
+	if ((network_type == RDMA_NETWORK_IPV4 ||
+	     network_type == RDMA_NETWORK_IPV6) &&
+	    test_bit(IB_GID_TYPE_ROCE_UDP_ENCAP, &supported_gids))
+		return IB_GID_TYPE_ROCE_UDP_ENCAP;
+
+	return default_gid;
+}
+
 static int cma_resolve_iboe_route(struct rdma_id_private *id_priv)
 {
 	struct rdma_route *route = &id_priv->id.route;
@@ -2461,6 +2473,8 @@ static int cma_resolve_iboe_route(struct rdma_id_private *id_priv)
 	route->num_paths = 1;
 
 	if (addr->dev_addr.bound_dev_if) {
+		unsigned long supported_gids;
+
 		ndev = dev_get_by_index(&init_net, addr->dev_addr.bound_dev_if);
 		if (!ndev) {
 			ret = -ENODEV;
@@ -2484,7 +2498,12 @@ static int cma_resolve_iboe_route(struct rdma_id_private *id_priv)
 
 		route->path_rec->net = &init_net;
 		route->path_rec->ifindex = ndev->ifindex;
-		route->path_rec->gid_type = id_priv->gid_type;
+		supported_gids = roce_gid_type_mask_support(id_priv->id.device,
+							    id_priv->id.port_num);
+		route->path_rec->gid_type =
+			cma_route_gid_type(addr->dev_addr.network,
+					   supported_gids,
+					   id_priv->gid_type);
 	}
 	if (!ndev) {
 		ret = -ENODEV;

drivers/infiniband/core/umem.c

@@ -175,7 +175,7 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
 
 	cur_base = addr & PAGE_MASK;
 
-	if (npages == 0) {
+	if (npages == 0 || npages > UINT_MAX) {
 		ret = -EINVAL;
 		goto out;
 	}

drivers/infiniband/core/uverbs_main.c

@@ -262,12 +262,9 @@ static int ib_uverbs_cleanup_ucontext(struct ib_uverbs_file *file,
 			container_of(uobj, struct ib_uqp_object, uevent.uobject);
 
 		idr_remove_uobj(&ib_uverbs_qp_idr, uobj);
-		if (qp != qp->real_qp) {
-			ib_close_qp(qp);
-		} else {
+		if (qp == qp->real_qp)
 			ib_uverbs_detach_umcast(qp, uqp);
-			ib_destroy_qp(qp);
-		}
+		ib_destroy_qp(qp);
 		ib_uverbs_release_uevent(file, &uqp->uevent);
 		kfree(uqp);
 	}

drivers/infiniband/hw/mlx4/ah.c

@@ -102,7 +102,10 @@ static struct ib_ah *create_iboe_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr
 	if (vlan_tag < 0x1000)
 		vlan_tag |= (ah_attr->sl & 7) << 13;
 	ah->av.eth.port_pd = cpu_to_be32(to_mpd(pd)->pdn | (ah_attr->port_num << 24));
-	ah->av.eth.gid_index = mlx4_ib_gid_index_to_real_index(ibdev, ah_attr->port_num, ah_attr->grh.sgid_index);
+	ret = mlx4_ib_gid_index_to_real_index(ibdev, ah_attr->port_num, ah_attr->grh.sgid_index);
+	if (ret < 0)
+		return ERR_PTR(ret);
+	ah->av.eth.gid_index = ret;
 	ah->av.eth.vlan = cpu_to_be16(vlan_tag);
 	ah->av.eth.hop_limit = ah_attr->grh.hop_limit;
 	if (ah_attr->static_rate) {

drivers/infiniband/hw/mlx4/cq.c

@@ -253,11 +253,14 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev,
 	if (context)
 		if (ib_copy_to_udata(udata, &cq->mcq.cqn, sizeof (__u32))) {
 			err = -EFAULT;
-			goto err_dbmap;
+			goto err_cq_free;
 		}
 
 	return &cq->ibcq;
 
+err_cq_free:
+	mlx4_cq_free(dev->dev, &cq->mcq);
+
 err_dbmap:
 	if (context)
 		mlx4_ib_db_unmap_user(to_mucontext(context), &cq->db);

drivers/infiniband/hw/mlx5/cq.c

@@ -932,8 +932,7 @@ struct ib_cq *mlx5_ib_create_cq(struct ib_device *ibdev,
 		if (err)
 			goto err_create;
 	} else {
-		/* for now choose 64 bytes till we have a proper interface */
-		cqe_size = 64;
+		cqe_size = cache_line_size() == 128 ? 128 : 64;
 		err = create_cq_kernel(dev, cq, entries, cqe_size, &cqb,
 				       &index, &inlen);
 		if (err)

drivers/infiniband/hw/mlx5/main.c

@@ -2311,14 +2311,14 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context,
 {
 	struct mlx5_ib_dev *ibdev = (struct mlx5_ib_dev *)context;
 	struct ib_event ibev;
-
+	bool fatal = false;
 	u8 port = 0;
 
 	switch (event) {
 	case MLX5_DEV_EVENT_SYS_ERROR:
-		ibdev->ib_active = false;
 		ibev.event = IB_EVENT_DEVICE_FATAL;
 		mlx5_ib_handle_internal_error(ibdev);
+		fatal = true;
 		break;
 
 	case MLX5_DEV_EVENT_PORT_UP:
@@ -2370,6 +2370,9 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context,
 
 	if (ibdev->ib_active)
 		ib_dispatch_event(&ibev);
+
+	if (fatal)
+		ibdev->ib_active = false;
 }
 
 static void get_ext_port_caps(struct mlx5_ib_dev *dev)
@@ -3115,7 +3118,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	}
 	err = init_node_data(dev);
 	if (err)
-		goto err_dealloc;
+		goto err_free_port;
 
 	mutex_init(&dev->flow_db.lock);
 	mutex_init(&dev->cap_mask_mutex);
@@ -3125,7 +3128,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	if (ll == IB_LINK_LAYER_ETHERNET) {
 		err = mlx5_enable_roce(dev);
 		if (err)
-			goto err_dealloc;
+			goto err_free_port;
 	}
 
 	err = create_dev_resources(&dev->devr);

drivers/infiniband/hw/mlx5/mlx5_ib.h

@@ -626,6 +626,8 @@ struct mlx5_ib_dev {
 	struct mlx5_ib_resources	devr;
 	struct mlx5_mr_cache		cache;
 	struct timer_list		delay_timer;
+	/* Prevents soft lock on massive reg MRs */
+	struct mutex			slow_path_mutex;
 	int				fill_delay;
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	struct ib_odp_caps	odp_caps;

drivers/infiniband/hw/mlx5/mr.c

@@ -610,6 +610,7 @@ int mlx5_mr_cache_init(struct mlx5_ib_dev *dev)
 	int err;
 	int i;
 
+	mutex_init(&dev->slow_path_mutex);
 	cache->wq = alloc_ordered_workqueue("mkey_cache", WQ_MEM_RECLAIM);
 	if (!cache->wq) {
 		mlx5_ib_warn(dev, "failed to create work queue\n");
@@ -1182,9 +1183,12 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 		goto error;
 	}
 
-	if (!mr)
+	if (!mr) {
+		mutex_lock(&dev->slow_path_mutex);
 		mr = reg_create(NULL, pd, virt_addr, length, umem, ncont,
 				page_shift, access_flags);
+		mutex_unlock(&dev->slow_path_mutex);
+	}
 
 	if (IS_ERR(mr)) {
 		err = PTR_ERR(mr);

drivers/infiniband/hw/mlx5/qp.c

@@ -2052,8 +2052,8 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd,
 
 		mlx5_ib_dbg(dev, "ib qpnum 0x%x, mlx qpn 0x%x, rcqn 0x%x, scqn 0x%x\n",
 			    qp->ibqp.qp_num, qp->trans_qp.base.mqp.qpn,
-			    to_mcq(init_attr->recv_cq)->mcq.cqn,
-			    to_mcq(init_attr->send_cq)->mcq.cqn);
+			    init_attr->recv_cq ? to_mcq(init_attr->recv_cq)->mcq.cqn : -1,
+			    init_attr->send_cq ? to_mcq(init_attr->send_cq)->mcq.cqn : -1);
 
 		qp->trans_qp.xrcdn = xrcdn;
 
@@ -4815,6 +4815,14 @@ struct ib_rwq_ind_table *mlx5_ib_create_rwq_ind_table(struct ib_device *device,
 				 udata->inlen))
 		return ERR_PTR(-EOPNOTSUPP);
 
+	if (init_attr->log_ind_tbl_size >
+	    MLX5_CAP_GEN(dev->mdev, log_max_rqt_size)) {
+		mlx5_ib_dbg(dev, "log_ind_tbl_size = %d is bigger than supported = %d\n",
+			    init_attr->log_ind_tbl_size,
+			    MLX5_CAP_GEN(dev->mdev, log_max_rqt_size));
+		return ERR_PTR(-EINVAL);
+	}
+
 	min_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved);
 	if (udata->outlen && udata->outlen < min_resp_len)
 		return ERR_PTR(-EINVAL);

drivers/infiniband/sw/rxe/rxe_net.c

@@ -243,10 +243,8 @@ static struct socket *rxe_setup_udp_tunnel(struct net *net, __be16 port,
 {
 	int err;
 	struct socket *sock;
-	struct udp_port_cfg udp_cfg;
-	struct udp_tunnel_sock_cfg tnl_cfg;
-
-	memset(&udp_cfg, 0, sizeof(udp_cfg));
+	struct udp_port_cfg udp_cfg = {0};
+	struct udp_tunnel_sock_cfg tnl_cfg = {0};
 
 	if (ipv6) {
 		udp_cfg.family = AF_INET6;
@@ -264,10 +262,8 @@ static struct socket *rxe_setup_udp_tunnel(struct net *net, __be16 port,
 		return ERR_PTR(err);
 	}
 
-	tnl_cfg.sk_user_data = NULL;
 	tnl_cfg.encap_type = 1;
 	tnl_cfg.encap_rcv = rxe_udp_encap_recv;
-	tnl_cfg.encap_destroy = NULL;
 
 	/* Setup UDP tunnel */
 	setup_udp_tunnel_sock(net, sock, &tnl_cfg);

drivers/infiniband/sw/rxe/rxe_qp.c

@@ -522,6 +522,7 @@ static void rxe_qp_reset(struct rxe_qp *qp)
 	if (qp->sq.queue) {
 		__rxe_do_task(&qp->comp.task);
 		__rxe_do_task(&qp->req.task);
+		rxe_queue_reset(qp->sq.queue);
 	}
 
 	/* cleanup attributes */
@@ -573,6 +574,7 @@ void rxe_qp_error(struct rxe_qp *qp)
 {
 	qp->req.state = QP_STATE_ERROR;
 	qp->resp.state = QP_STATE_ERROR;
+	qp->attr.qp_state = IB_QPS_ERR;
 
 	/* drain work and packet queues */
 	rxe_run_task(&qp->resp.task, 1);

drivers/infiniband/sw/rxe/rxe_queue.c

@@ -84,6 +84,15 @@ int do_mmap_info(struct rxe_dev *rxe,
 	return -EINVAL;
 }
 
+inline void rxe_queue_reset(struct rxe_queue *q)
+{
+	/* queue is comprised from header and the memory
+	 * of the actual queue. See "struct rxe_queue_buf" in rxe_queue.h
+	 * reset only the queue itself and not the management header
+	 */
+	memset(q->buf->data, 0, q->buf_size - sizeof(struct rxe_queue_buf));
+}
+
 struct rxe_queue *rxe_queue_init(struct rxe_dev *rxe,
 				 int *num_elem,
 				 unsigned int elem_size)

drivers/infiniband/sw/rxe/rxe_queue.h

@@ -84,6 +84,8 @@ int do_mmap_info(struct rxe_dev *rxe,
 		 size_t buf_size,
 		 struct rxe_mmap_info **ip_p);
 
+void rxe_queue_reset(struct rxe_queue *q);
+
 struct rxe_queue *rxe_queue_init(struct rxe_dev *rxe,
 				 int *num_elem,
 				 unsigned int elem_size);

drivers/infiniband/sw/rxe/rxe_req.c

@@ -696,7 +696,8 @@ int rxe_requester(void *arg)
 						       qp->req.wqe_index);
 			wqe->state = wqe_state_done;
 			wqe->status = IB_WC_SUCCESS;
-			goto complete;
+			__rxe_do_task(&qp->comp.task);
+			return 0;
 		}
 		payload = mtu;
 	}
@@ -745,13 +746,17 @@ int rxe_requester(void *arg)
 	wqe->status = IB_WC_LOC_PROT_ERR;
 	wqe->state = wqe_state_error;
 
-complete:
-	if (qp_type(qp) != IB_QPT_RC) {
-		while (rxe_completer(qp) == 0)
-			;
-	}
-
-	return 0;
+	/*
+	 * IBA Spec. Section 10.7.3.1 SIGNALED COMPLETIONS
+	 * ---------8<---------8<-------------
+	 * ...Note that if a completion error occurs, a Work Completion
+	 * will always be generated, even if the signaling
+	 * indicator requests an Unsignaled Completion.
+	 * ---------8<---------8<-------------
+	 */
+	wqe->wr.send_flags |= IB_SEND_SIGNALED;
+	__rxe_do_task(&qp->comp.task);
+	return -EAGAIN;
 
 exit:
 	return -EAGAIN;