Merge tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband

Pull infiniband updates from Roland Dreier:
 "First batch of InfiniBand/RDMA changes for the 3.7 merge window:
   - mlx4 IB support for SR-IOV
   - A couple of SRP initiator fixes
   - Batch of nes hardware driver fixes
   - Fix for long-standing use-after-free crash in IPoIB
   - Other miscellaneous fixes"

This merge also removes a new use of __cancel_delayed_work(), and
replaces it with the regular cancel_delayed_work() that is now irq-safe
thanks to the workqueue updates.

That said, I suspect the sequence in question should probably use
"mod_delayed_work()".  I just did the minimal "don't use deprecated
functions" fixup, though.

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (45 commits)
  IB/qib: Fix local access validation for user MRs
  mlx4_core: Disable SENSE_PORT for multifunction devices
  mlx4_core: Clean up enabling of SENSE_PORT for older (ConnectX-1/-2) HCAs
  mlx4_core: Stash PCI ID driver_data in mlx4_priv structure
  IB/srp: Avoid having aborted requests hang
  IB/srp: Fix use-after-free in srp_reset_req()
  IB/qib: Add a qib driver version
  RDMA/nes: Fix compilation error when nes_debug is enabled
  RDMA/nes: Print hardware resource type
  RDMA/nes: Fix for crash when TX checksum offload is off
  RDMA/nes: Cosmetic changes
  RDMA/nes: Fix for incorrect MSS when TSO is on
  RDMA/nes: Fix incorrect resolving of the loopback MAC address
  mlx4_core: Fix crash on uninitialized priv->cmd.slave_sem
  mlx4_core: Trivial cleanups to driver log messages
  mlx4_core: Trivial readability fix: "0X30" -> "0x30"
  IB/mlx4: Create paravirt contexts for VFs when master IB driver initializes
  mlx4: Modify proxy/tunnel QP mechanism so that guests do no calculations
  mlx4: Paravirtualize Node Guids for slaves
  mlx4: Activate SR-IOV mode for IB
  ...

drivers/infiniband/core/cache.c

@@ -167,6 +167,7 @@ int ib_find_cached_pkey(struct ib_device *device,
 	unsigned long flags;
 	int i;
 	int ret = -ENOENT;
+	int partial_ix = -1;
 
 	if (port_num < start_port(device) || port_num > end_port(device))
 		return -EINVAL;
@@ -179,6 +180,46 @@ int ib_find_cached_pkey(struct ib_device *device,
 
 	for (i = 0; i < cache->table_len; ++i)
 		if ((cache->table[i] & 0x7fff) == (pkey & 0x7fff)) {
+			if (cache->table[i] & 0x8000) {
+				*index = i;
+				ret = 0;
+				break;
+			} else
+				partial_ix = i;
+		}
+
+	if (ret && partial_ix >= 0) {
+		*index = partial_ix;
+		ret = 0;
+	}
+
+	read_unlock_irqrestore(&device->cache.lock, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_find_cached_pkey);
+
+int ib_find_exact_cached_pkey(struct ib_device *device,
+			      u8                port_num,
+			      u16               pkey,
+			      u16              *index)
+{
+	struct ib_pkey_cache *cache;
+	unsigned long flags;
+	int i;
+	int ret = -ENOENT;
+
+	if (port_num < start_port(device) || port_num > end_port(device))
+		return -EINVAL;
+
+	read_lock_irqsave(&device->cache.lock, flags);
+
+	cache = device->cache.pkey_cache[port_num - start_port(device)];
+
+	*index = -1;
+
+	for (i = 0; i < cache->table_len; ++i)
+		if (cache->table[i] == pkey) {
 			*index = i;
 			ret = 0;
 			break;
@@ -188,7 +229,7 @@ int ib_find_cached_pkey(struct ib_device *device,
 
 	return ret;
 }
-EXPORT_SYMBOL(ib_find_cached_pkey);
+EXPORT_SYMBOL(ib_find_exact_cached_pkey);
 
 int ib_get_cached_lmc(struct ib_device *device,
 		      u8                port_num,

drivers/infiniband/core/cma.c

@@ -3058,7 +3058,10 @@ static int cma_join_ib_multicast(struct rdma_id_private *id_priv,
 
 	if (id_priv->id.ps == RDMA_PS_IPOIB)
 		comp_mask |= IB_SA_MCMEMBER_REC_RATE |
-			     IB_SA_MCMEMBER_REC_RATE_SELECTOR;
+			     IB_SA_MCMEMBER_REC_RATE_SELECTOR |
+			     IB_SA_MCMEMBER_REC_MTU_SELECTOR |
+			     IB_SA_MCMEMBER_REC_MTU |
+			     IB_SA_MCMEMBER_REC_HOP_LIMIT;
 
 	mc->multicast.ib = ib_sa_join_multicast(&sa_client, id_priv->id.device,
 						id_priv->id.port_num, &rec,

drivers/infiniband/core/device.c

@@ -707,18 +707,28 @@ int ib_find_pkey(struct ib_device *device,
 {
 	int ret, i;
 	u16 tmp_pkey;
+	int partial_ix = -1;
 
 	for (i = 0; i < device->pkey_tbl_len[port_num - start_port(device)]; ++i) {
 		ret = ib_query_pkey(device, port_num, i, &tmp_pkey);
 		if (ret)
 			return ret;
-
 		if ((pkey & 0x7fff) == (tmp_pkey & 0x7fff)) {
-			*index = i;
-			return 0;
+			/* if there is full-member pkey take it.*/
+			if (tmp_pkey & 0x8000) {
+				*index = i;
+				return 0;
+			}
+			if (partial_ix < 0)
+				partial_ix = i;
 		}
 	}
 
+	/*no full-member, if exists take the limited*/
+	if (partial_ix >= 0) {
+		*index = partial_ix;
+		return 0;
+	}
 	return -ENOENT;
 }
 EXPORT_SYMBOL(ib_find_pkey);

drivers/infiniband/core/ucm.c

@@ -397,7 +397,6 @@ static ssize_t ib_ucm_event(struct ib_ucm_file *file,
 	struct ib_ucm_event_get cmd;
 	struct ib_ucm_event *uevent;
 	int result = 0;
-	DEFINE_WAIT(wait);
 
 	if (out_len < sizeof(struct ib_ucm_event_resp))
 		return -ENOSPC;

drivers/infiniband/core/ucma.c

@@ -310,7 +310,6 @@ static ssize_t ucma_get_event(struct ucma_file *file, const char __user *inbuf,
 	struct rdma_ucm_get_event cmd;
 	struct ucma_event *uevent;
 	int ret = 0;
-	DEFINE_WAIT(wait);
 
 	if (out_len < sizeof uevent->resp)
 		return -ENOSPC;

drivers/infiniband/hw/cxgb4/qp.c

@@ -137,19 +137,25 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 		return -ENOMEM;
 
 	wq->rq.qid = c4iw_get_qpid(rdev, uctx);
-	if (!wq->rq.qid)
-		goto err1;
+	if (!wq->rq.qid) {
+		ret = -ENOMEM;
+		goto free_sq_qid;
+	}
 
 	if (!user) {
 		wq->sq.sw_sq = kzalloc(wq->sq.size * sizeof *wq->sq.sw_sq,
 				 GFP_KERNEL);
-		if (!wq->sq.sw_sq)
-			goto err2;
+		if (!wq->sq.sw_sq) {
+			ret = -ENOMEM;
+			goto free_rq_qid;
+		}
 
 		wq->rq.sw_rq = kzalloc(wq->rq.size * sizeof *wq->rq.sw_rq,
 				 GFP_KERNEL);
-		if (!wq->rq.sw_rq)
-			goto err3;
+		if (!wq->rq.sw_rq) {
+			ret = -ENOMEM;
+			goto free_sw_sq;
+		}
 	}
 
 	/*
@@ -157,15 +163,23 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 	 */
 	wq->rq.rqt_size = roundup_pow_of_two(wq->rq.size);
 	wq->rq.rqt_hwaddr = c4iw_rqtpool_alloc(rdev, wq->rq.rqt_size);
-	if (!wq->rq.rqt_hwaddr)
-		goto err4;
+	if (!wq->rq.rqt_hwaddr) {
+		ret = -ENOMEM;
+		goto free_sw_rq;
+	}
 
 	if (user) {
-		if (alloc_oc_sq(rdev, &wq->sq) && alloc_host_sq(rdev, &wq->sq))
-			goto err5;
+		ret = alloc_oc_sq(rdev, &wq->sq);
+		if (ret)
+			goto free_hwaddr;
+
+		ret = alloc_host_sq(rdev, &wq->sq);
+		if (ret)
+			goto free_sq;
 	} else
-		if (alloc_host_sq(rdev, &wq->sq))
-			goto err5;
+		ret = alloc_host_sq(rdev, &wq->sq);
+		if (ret)
+			goto free_hwaddr;
 	memset(wq->sq.queue, 0, wq->sq.memsize);
 	dma_unmap_addr_set(&wq->sq, mapping, wq->sq.dma_addr);
 
@@ -173,7 +187,7 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 					  wq->rq.memsize, &(wq->rq.dma_addr),
 					  GFP_KERNEL);
 	if (!wq->rq.queue)
-		goto err6;
+		goto free_sq;
 	PDBG("%s sq base va 0x%p pa 0x%llx rq base va 0x%p pa 0x%llx\n",
 		__func__, wq->sq.queue,
 		(unsigned long long)virt_to_phys(wq->sq.queue),
@@ -201,7 +215,7 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 	skb = alloc_skb(wr_len, GFP_KERNEL);
 	if (!skb) {
 		ret = -ENOMEM;
-		goto err7;
+		goto free_dma;
 	}
 	set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0);
 
@@ -266,33 +280,33 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 
 	ret = c4iw_ofld_send(rdev, skb);
 	if (ret)
-		goto err7;
+		goto free_dma;
 	ret = c4iw_wait_for_reply(rdev, &wr_wait, 0, wq->sq.qid, __func__);
 	if (ret)
-		goto err7;
+		goto free_dma;
 
 	PDBG("%s sqid 0x%x rqid 0x%x kdb 0x%p squdb 0x%llx rqudb 0x%llx\n",
 	     __func__, wq->sq.qid, wq->rq.qid, wq->db,
 	     (unsigned long long)wq->sq.udb, (unsigned long long)wq->rq.udb);
 
 	return 0;
-err7:
+free_dma:
 	dma_free_coherent(&(rdev->lldi.pdev->dev),
 			  wq->rq.memsize, wq->rq.queue,
 			  dma_unmap_addr(&wq->rq, mapping));
-err6:
+free_sq:
 	dealloc_sq(rdev, &wq->sq);
-err5:
+free_hwaddr:
 	c4iw_rqtpool_free(rdev, wq->rq.rqt_hwaddr, wq->rq.rqt_size);
-err4:
+free_sw_rq:
 	kfree(wq->rq.sw_rq);
-err3:
+free_sw_sq:
 	kfree(wq->sq.sw_sq);
-err2:
+free_rq_qid:
 	c4iw_put_qpid(rdev, wq->rq.qid, uctx);
-err1:
+free_sq_qid:
 	c4iw_put_qpid(rdev, wq->sq.qid, uctx);
-	return -ENOMEM;
+	return ret;
 }
 
 static int build_immd(struct t4_sq *sq, struct fw_ri_immd *immdp,

drivers/infiniband/hw/mlx4/Makefile

 obj-$(CONFIG_MLX4_INFINIBAND)	+= mlx4_ib.o
 
-mlx4_ib-y :=	ah.o cq.o doorbell.o mad.o main.o mr.o qp.o srq.o
+mlx4_ib-y :=	ah.o cq.o doorbell.o mad.o main.o mr.o qp.o srq.o mcg.o cm.o alias_GUID.o sysfs.o

drivers/infiniband/hw/mlx4/cm.c

+/*
+ * Copyright (c) 2012 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <rdma/ib_mad.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/rbtree.h>
+#include <linux/idr.h>
+#include <rdma/ib_cm.h>
+
+#include "mlx4_ib.h"
+
+#define CM_CLEANUP_CACHE_TIMEOUT  (5 * HZ)
+
+struct id_map_entry {
+	struct rb_node node;
+
+	u32 sl_cm_id;
+	u32 pv_cm_id;
+	int slave_id;
+	int scheduled_delete;
+	struct mlx4_ib_dev *dev;
+
+	struct list_head list;
+	struct delayed_work timeout;
+};
+
+struct cm_generic_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+};
+
+struct cm_req_msg {
+	unsigned char unused[0x60];
+	union ib_gid primary_path_sgid;
+};
+
+
+static void set_local_comm_id(struct ib_mad *mad, u32 cm_id)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+	msg->local_comm_id = cpu_to_be32(cm_id);
+}
+
+static u32 get_local_comm_id(struct ib_mad *mad)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+
+	return be32_to_cpu(msg->local_comm_id);
+}
+
+static void set_remote_comm_id(struct ib_mad *mad, u32 cm_id)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+	msg->remote_comm_id = cpu_to_be32(cm_id);
+}
+
+static u32 get_remote_comm_id(struct ib_mad *mad)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+
+	return be32_to_cpu(msg->remote_comm_id);
+}
+
+static union ib_gid gid_from_req_msg(struct ib_device *ibdev, struct ib_mad *mad)
+{
+	struct cm_req_msg *msg = (struct cm_req_msg *)mad;
+
+	return msg->primary_path_sgid;
+}
+
+/* Lock should be taken before called */
+static struct id_map_entry *
+id_map_find_by_sl_id(struct ib_device *ibdev, u32 slave_id, u32 sl_cm_id)
+{
+	struct rb_root *sl_id_map = &to_mdev(ibdev)->sriov.sl_id_map;
+	struct rb_node *node = sl_id_map->rb_node;
+
+	while (node) {
+		struct id_map_entry *id_map_entry =
+			rb_entry(node, struct id_map_entry, node);
+
+		if (id_map_entry->sl_cm_id > sl_cm_id)
+			node = node->rb_left;
+		else if (id_map_entry->sl_cm_id < sl_cm_id)
+			node = node->rb_right;
+		else if (id_map_entry->slave_id > slave_id)
+			node = node->rb_left;
+		else if (id_map_entry->slave_id < slave_id)
+			node = node->rb_right;
+		else
+			return id_map_entry;
+	}
+	return NULL;
+}
+
+static void id_map_ent_timeout(struct work_struct *work)
+{
+	struct delayed_work *delay = to_delayed_work(work);
+	struct id_map_entry *ent = container_of(delay, struct id_map_entry, timeout);
+	struct id_map_entry *db_ent, *found_ent;
+	struct mlx4_ib_dev *dev = ent->dev;
+	struct mlx4_ib_sriov *sriov = &dev->sriov;
+	struct rb_root *sl_id_map = &sriov->sl_id_map;
+	int pv_id = (int) ent->pv_cm_id;
+
+	spin_lock(&sriov->id_map_lock);
+	db_ent = (struct id_map_entry *)idr_find(&sriov->pv_id_table, pv_id);
+	if (!db_ent)
+		goto out;
+	found_ent = id_map_find_by_sl_id(&dev->ib_dev, ent->slave_id, ent->sl_cm_id);
+	if (found_ent && found_ent == ent)
+		rb_erase(&found_ent->node, sl_id_map);
+	idr_remove(&sriov->pv_id_table, pv_id);
+
+out:
+	list_del(&ent->list);
+	spin_unlock(&sriov->id_map_lock);
+	kfree(ent);
+}
+
+static void id_map_find_del(struct ib_device *ibdev, int pv_cm_id)
+{
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+	struct rb_root *sl_id_map = &sriov->sl_id_map;
+	struct id_map_entry *ent, *found_ent;
+
+	spin_lock(&sriov->id_map_lock);
+	ent = (struct id_map_entry *)idr_find(&sriov->pv_id_table, pv_cm_id);
+	if (!ent)
+		goto out;
+	found_ent = id_map_find_by_sl_id(ibdev, ent->slave_id, ent->sl_cm_id);
+	if (found_ent && found_ent == ent)
+		rb_erase(&found_ent->node, sl_id_map);
+	idr_remove(&sriov->pv_id_table, pv_cm_id);
+out:
+	spin_unlock(&sriov->id_map_lock);
+}
+
+static void sl_id_map_add(struct ib_device *ibdev, struct id_map_entry *new)
+{
+	struct rb_root *sl_id_map = &to_mdev(ibdev)->sriov.sl_id_map;
+	struct rb_node **link = &sl_id_map->rb_node, *parent = NULL;
+	struct id_map_entry *ent;
+	int slave_id = new->slave_id;
+	int sl_cm_id = new->sl_cm_id;
+
+	ent = id_map_find_by_sl_id(ibdev, slave_id, sl_cm_id);
+	if (ent) {
+		pr_debug("overriding existing sl_id_map entry (cm_id = %x)\n",
+			 sl_cm_id);
+
+		rb_replace_node(&ent->node, &new->node, sl_id_map);
+		return;
+	}
+
+	/* Go to the bottom of the tree */
+	while (*link) {
+		parent = *link;
+		ent = rb_entry(parent, struct id_map_entry, node);
+
+		if (ent->sl_cm_id > sl_cm_id || (ent->sl_cm_id == sl_cm_id && ent->slave_id > slave_id))
+			link = &(*link)->rb_left;
+		else
+			link = &(*link)->rb_right;
+	}
+
+	rb_link_node(&new->node, parent, link);
+	rb_insert_color(&new->node, sl_id_map);
+}
+
+static struct id_map_entry *
+id_map_alloc(struct ib_device *ibdev, int slave_id, u32 sl_cm_id)
+{
+	int ret, id;
+	static int next_id;
+	struct id_map_entry *ent;
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+
+	ent = kmalloc(sizeof (struct id_map_entry), GFP_KERNEL);
+	if (!ent) {
+		mlx4_ib_warn(ibdev, "Couldn't allocate id cache entry - out of memory\n");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	ent->sl_cm_id = sl_cm_id;
+	ent->slave_id = slave_id;
+	ent->scheduled_delete = 0;
+	ent->dev = to_mdev(ibdev);
+	INIT_DELAYED_WORK(&ent->timeout, id_map_ent_timeout);
+
+	do {
+		spin_lock(&to_mdev(ibdev)->sriov.id_map_lock);
+		ret = idr_get_new_above(&sriov->pv_id_table, ent,
+					next_id, &id);
+		if (!ret) {
+			next_id = ((unsigned) id + 1) & MAX_ID_MASK;
+			ent->pv_cm_id = (u32)id;
+			sl_id_map_add(ibdev, ent);
+		}
+
+		spin_unlock(&sriov->id_map_lock);
+	} while (ret == -EAGAIN && idr_pre_get(&sriov->pv_id_table, GFP_KERNEL));
+	/*the function idr_get_new_above can return -ENOSPC, so don't insert in that case.*/
+	if (!ret) {
+		spin_lock(&sriov->id_map_lock);
+		list_add_tail(&ent->list, &sriov->cm_list);
+		spin_unlock(&sriov->id_map_lock);
+		return ent;
+	}
+	/*error flow*/
+	kfree(ent);
+	mlx4_ib_warn(ibdev, "No more space in the idr (err:0x%x)\n", ret);
+	return ERR_PTR(-ENOMEM);
+}
+
+static struct id_map_entry *
+id_map_get(struct ib_device *ibdev, int *pv_cm_id, int sl_cm_id, int slave_id)
+{
+	struct id_map_entry *ent;
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+
+	spin_lock(&sriov->id_map_lock);
+	if (*pv_cm_id == -1) {
+		ent = id_map_find_by_sl_id(ibdev, sl_cm_id, slave_id);
+		if (ent)
+			*pv_cm_id = (int) ent->pv_cm_id;
+	} else
+		ent = (struct id_map_entry *)idr_find(&sriov->pv_id_table, *pv_cm_id);
+	spin_unlock(&sriov->id_map_lock);
+
+	return ent;
+}
+
+static void schedule_delayed(struct ib_device *ibdev, struct id_map_entry *id)
+{
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+	unsigned long flags;
+
+	spin_lock_irqsave(&sriov->going_down_lock, flags);
+	spin_lock(&sriov->id_map_lock);
+	/*make sure that there is no schedule inside the scheduled work.*/
+	if (!sriov->is_going_down) {
+		id->scheduled_delete = 1;
+		schedule_delayed_work(&id->timeout, CM_CLEANUP_CACHE_TIMEOUT);
+	}
+	spin_unlock(&sriov->id_map_lock);
+	spin_unlock_irqrestore(&sriov->going_down_lock, flags);
+}
+
+int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id,
+		struct ib_mad *mad)
+{
+	struct id_map_entry *id;
+	u32 sl_cm_id;
+	int pv_cm_id = -1;
+
+	sl_cm_id = get_local_comm_id(mad);
+
+	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID ||
+			mad->mad_hdr.attr_id == CM_REP_ATTR_ID) {
+		id = id_map_alloc(ibdev, slave_id, sl_cm_id);
+		if (IS_ERR(id)) {
+			mlx4_ib_warn(ibdev, "%s: id{slave: %d, sl_cm_id: 0x%x} Failed to id_map_alloc\n",
+				__func__, slave_id, sl_cm_id);
+			return PTR_ERR(id);
+		}
+	} else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID) {
+		return 0;
+	} else {
+		id = id_map_get(ibdev, &pv_cm_id, slave_id, sl_cm_id);
+	}
+
+	if (!id) {
+		pr_debug("id{slave: %d, sl_cm_id: 0x%x} is NULL!\n",
+			 slave_id, sl_cm_id);
+		return -EINVAL;
+	}
+
+	set_local_comm_id(mad, id->pv_cm_id);
+
+	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
+		schedule_delayed(ibdev, id);
+	else if (mad->mad_hdr.attr_id == CM_DREP_ATTR_ID)
+		id_map_find_del(ibdev, pv_cm_id);
+
+	return 0;
+}
+
+int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
+							     struct ib_mad *mad)
+{
+	u32 pv_cm_id;
+	struct id_map_entry *id;
+
+	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID) {
+		union ib_gid gid;
+
+		gid = gid_from_req_msg(ibdev, mad);
+		*slave = mlx4_ib_find_real_gid(ibdev, port, gid.global.interface_id);
+		if (*slave < 0) {
+			mlx4_ib_warn(ibdev, "failed matching slave_id by gid (0x%llx)\n",
+					gid.global.interface_id);
+			return -ENOENT;
+		}
+		return 0;
+	}
+
+	pv_cm_id = get_remote_comm_id(mad);
+	id = id_map_get(ibdev, (int *)&pv_cm_id, -1, -1);
+
+	if (!id) {
+		pr_debug("Couldn't find an entry for pv_cm_id 0x%x\n", pv_cm_id);
+		return -ENOENT;
+	}
+
+	*slave = id->slave_id;
+	set_remote_comm_id(mad, id->sl_cm_id);
+
+	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
+		schedule_delayed(ibdev, id);
+	else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID ||
+			mad->mad_hdr.attr_id == CM_DREP_ATTR_ID) {
+		id_map_find_del(ibdev, (int) pv_cm_id);
+	}
+
+	return 0;
+}
+
+void mlx4_ib_cm_paravirt_init(struct mlx4_ib_dev *dev)
+{
+	spin_lock_init(&dev->sriov.id_map_lock);
+	INIT_LIST_HEAD(&dev->sriov.cm_list);
+	dev->sriov.sl_id_map = RB_ROOT;
+	idr_init(&dev->sriov.pv_id_table);
+	idr_pre_get(&dev->sriov.pv_id_table, GFP_KERNEL);
+}
+
+/* slave = -1 ==> all slaves */
+/* TBD -- call paravirt clean for single slave.  Need for slave RESET event */
+void mlx4_ib_cm_paravirt_clean(struct mlx4_ib_dev *dev, int slave)
+{
+	struct mlx4_ib_sriov *sriov = &dev->sriov;
+	struct rb_root *sl_id_map = &sriov->sl_id_map;
+	struct list_head lh;
+	struct rb_node *nd;
+	int need_flush = 1;
+	struct id_map_entry *map, *tmp_map;
+	/* cancel all delayed work queue entries */
+	INIT_LIST_HEAD(&lh);
+	spin_lock(&sriov->id_map_lock);
+	list_for_each_entry_safe(map, tmp_map, &dev->sriov.cm_list, list) {
+		if (slave < 0 || slave == map->slave_id) {
+			if (map->scheduled_delete)
+				need_flush &= !!cancel_delayed_work(&map->timeout);
+		}
+	}
+
+	spin_unlock(&sriov->id_map_lock);
+
+	if (!need_flush)
+		flush_scheduled_work(); /* make sure all timers were flushed */
+
+	/* now, remove all leftover entries from databases*/
+	spin_lock(&sriov->id_map_lock);
+	if (slave < 0) {
+		while (rb_first(sl_id_map)) {
+			struct id_map_entry *ent =
+				rb_entry(rb_first(sl_id_map),
+					 struct id_map_entry, node);
+
+			rb_erase(&ent->node, sl_id_map);
+			idr_remove(&sriov->pv_id_table, (int) ent->pv_cm_id);
+		}
+		list_splice_init(&dev->sriov.cm_list, &lh);
+	} else {
+		/* first, move nodes belonging to slave to db remove list */
+		nd = rb_first(sl_id_map);
+		while (nd) {
+			struct id_map_entry *ent =
+				rb_entry(nd, struct id_map_entry, node);
+			nd = rb_next(nd);
+			if (ent->slave_id == slave)
+				list_move_tail(&ent->list, &lh);
+		}
+		/* remove those nodes from databases */
+		list_for_each_entry_safe(map, tmp_map, &lh, list) {
+			rb_erase(&map->node, sl_id_map);
+			idr_remove(&sriov->pv_id_table, (int) map->pv_cm_id);
+		}
+
+		/* add remaining nodes from cm_list */
+		list_for_each_entry_safe(map, tmp_map, &dev->sriov.cm_list, list) {
+			if (slave == map->slave_id)
+				list_move_tail(&map->list, &lh);
+		}
+	}
+
+	spin_unlock(&sriov->id_map_lock);
+
+	/* free any map entries left behind due to cancel_delayed_work above */
+	list_for_each_entry_safe(map, tmp_map, &lh, list) {
+		list_del(&map->list);
+		kfree(map);
+	}
+}

drivers/infiniband/hw/mlx4/cq.c

@@ -547,6 +547,26 @@ static int mlx4_ib_ipoib_csum_ok(__be16 status, __be16 checksum)
 		checksum == cpu_to_be16(0xffff);
 }
 
+static int use_tunnel_data(struct mlx4_ib_qp *qp, struct mlx4_ib_cq *cq, struct ib_wc *wc,
+			   unsigned tail, struct mlx4_cqe *cqe)
+{
+	struct mlx4_ib_proxy_sqp_hdr *hdr;
+
+	ib_dma_sync_single_for_cpu(qp->ibqp.device,
+				   qp->sqp_proxy_rcv[tail].map,
+				   sizeof (struct mlx4_ib_proxy_sqp_hdr),
+				   DMA_FROM_DEVICE);
+	hdr = (struct mlx4_ib_proxy_sqp_hdr *) (qp->sqp_proxy_rcv[tail].addr);
+	wc->pkey_index	= be16_to_cpu(hdr->tun.pkey_index);
+	wc->slid	= be16_to_cpu(hdr->tun.slid_mac_47_32);
+	wc->sl		= (u8) (be16_to_cpu(hdr->tun.sl_vid) >> 12);
+	wc->src_qp	= be32_to_cpu(hdr->tun.flags_src_qp) & 0xFFFFFF;
+	wc->wc_flags   |= (hdr->tun.g_ml_path & 0x80) ? (IB_WC_GRH) : 0;
+	wc->dlid_path_bits = 0;
+
+	return 0;
+}
+
 static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 			    struct mlx4_ib_qp **cur_qp,
 			    struct ib_wc *wc)
@@ -559,6 +579,7 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 	int is_error;
 	u32 g_mlpath_rqpn;
 	u16 wqe_ctr;
+	unsigned tail = 0;
 
 repoll:
 	cqe = next_cqe_sw(cq);
@@ -634,7 +655,8 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 		mlx4_ib_free_srq_wqe(srq, wqe_ctr);
 	} else {
 		wq	  = &(*cur_qp)->rq;
-		wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
+		tail	  = wq->tail & (wq->wqe_cnt - 1);
+		wc->wr_id = wq->wrid[tail];
 		++wq->tail;
 	}
 
@@ -717,6 +739,13 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 			break;
 		}
 
+		if (mlx4_is_mfunc(to_mdev(cq->ibcq.device)->dev)) {
+			if ((*cur_qp)->mlx4_ib_qp_type &
+			    (MLX4_IB_QPT_PROXY_SMI_OWNER |
+			     MLX4_IB_QPT_PROXY_SMI | MLX4_IB_QPT_PROXY_GSI))
+				return use_tunnel_data(*cur_qp, cq, wc, tail, cqe);
+		}
+
 		wc->slid	   = be16_to_cpu(cqe->rlid);
 		g_mlpath_rqpn	   = be32_to_cpu(cqe->g_mlpath_rqpn);
 		wc->src_qp	   = g_mlpath_rqpn & 0xffffff;

drivers/infiniband/hw/mlx4/main.c

@@ -59,6 +59,10 @@ MODULE_DESCRIPTION("Mellanox ConnectX HCA InfiniBand driver");
 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_VERSION(DRV_VERSION);
 
+int mlx4_ib_sm_guid_assign = 1;
+module_param_named(sm_guid_assign, mlx4_ib_sm_guid_assign, int, 0444);
+MODULE_PARM_DESC(sm_guid_assign, "Enable SM alias_GUID assignment if sm_guid_assign > 0 (Default: 1)");
+
 static const char mlx4_ib_version[] =
 	DRV_NAME ": Mellanox ConnectX InfiniBand driver v"
 	DRV_VERSION " (" DRV_RELDATE ")\n";
@@ -70,6 +74,8 @@ struct update_gid_work {
 	int			port;
 };
 
+static void do_slave_init(struct mlx4_ib_dev *ibdev, int slave, int do_init);
+
 static struct workqueue_struct *wq;
 
 static void init_query_mad(struct ib_smp *mad)
@@ -98,7 +104,8 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
 	init_query_mad(in_mad);
 	in_mad->attr_id = IB_SMP_ATTR_NODE_INFO;
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, 1, NULL, NULL, in_mad, out_mad);
+	err = mlx4_MAD_IFC(to_mdev(ibdev), MLX4_MAD_IFC_IGNORE_KEYS,
+			   1, NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
 
@@ -133,7 +140,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
 
 	props->vendor_id	   = be32_to_cpup((__be32 *) (out_mad->data + 36)) &
 		0xffffff;
-	props->vendor_part_id	   = be16_to_cpup((__be16 *) (out_mad->data + 30));
+	props->vendor_part_id	   = dev->dev->pdev->device;
 	props->hw_ver		   = be32_to_cpup((__be32 *) (out_mad->data + 32));
 	memcpy(&props->sys_image_guid, out_mad->data +	4, 8);
 
@@ -182,11 +189,12 @@ mlx4_ib_port_link_layer(struct ib_device *device, u8 port_num)
 }
 
 static int ib_link_query_port(struct ib_device *ibdev, u8 port,
-			      struct ib_port_attr *props)
+			      struct ib_port_attr *props, int netw_view)
 {
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
 	int ext_active_speed;
+	int mad_ifc_flags = MLX4_MAD_IFC_IGNORE_KEYS;
 	int err = -ENOMEM;
 
 	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
@@ -198,7 +206,10 @@ static int ib_link_query_port(struct ib_device *ibdev, u8 port,
 	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
 	in_mad->attr_mod = cpu_to_be32(port);
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL,
+	if (mlx4_is_mfunc(to_mdev(ibdev)->dev) && netw_view)
+		mad_ifc_flags |= MLX4_MAD_IFC_NET_VIEW;
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), mad_ifc_flags, port, NULL, NULL,
 				in_mad, out_mad);
 	if (err)
 		goto out;
@@ -211,7 +222,10 @@ static int ib_link_query_port(struct ib_device *ibdev, u8 port,
 	props->state		= out_mad->data[32] & 0xf;
 	props->phys_state	= out_mad->data[33] >> 4;
 	props->port_cap_flags	= be32_to_cpup((__be32 *) (out_mad->data + 20));
-	props->gid_tbl_len	= to_mdev(ibdev)->dev->caps.gid_table_len[port];
+	if (netw_view)
+		props->gid_tbl_len = out_mad->data[50];
+	else
+		props->gid_tbl_len = to_mdev(ibdev)->dev->caps.gid_table_len[port];
 	props->max_msg_sz	= to_mdev(ibdev)->dev->caps.max_msg_sz;
 	props->pkey_tbl_len	= to_mdev(ibdev)->dev->caps.pkey_table_len[port];
 	props->bad_pkey_cntr	= be16_to_cpup((__be16 *) (out_mad->data + 46));
@@ -244,7 +258,7 @@ static int ib_link_query_port(struct ib_device *ibdev, u8 port,
 		in_mad->attr_id = MLX4_ATTR_EXTENDED_PORT_INFO;
 		in_mad->attr_mod = cpu_to_be32(port);
 
-		err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port,
+		err = mlx4_MAD_IFC(to_mdev(ibdev), mad_ifc_flags, port,
 				   NULL, NULL, in_mad, out_mad);
 		if (err)
 			goto out;
@@ -270,7 +284,7 @@ static u8 state_to_phys_state(enum ib_port_state state)
 }
 
 static int eth_link_query_port(struct ib_device *ibdev, u8 port,
-			       struct ib_port_attr *props)
+			       struct ib_port_attr *props, int netw_view)
 {
 
 	struct mlx4_ib_dev *mdev = to_mdev(ibdev);
@@ -320,26 +334,36 @@ static int eth_link_query_port(struct ib_device *ibdev, u8 port,
 	return err;
 }
 
-static int mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
-			      struct ib_port_attr *props)
+int __mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
+			 struct ib_port_attr *props, int netw_view)
 {
 	int err;
 
 	memset(props, 0, sizeof *props);
 
 	err = mlx4_ib_port_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND ?
-		ib_link_query_port(ibdev, port, props) :
-				eth_link_query_port(ibdev, port, props);
+		ib_link_query_port(ibdev, port, props, netw_view) :
+				eth_link_query_port(ibdev, port, props, netw_view);
 
 	return err;
 }
 
-static int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
-			       union ib_gid *gid)
+static int mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
+			      struct ib_port_attr *props)
+{
+	/* returns host view */
+	return __mlx4_ib_query_port(ibdev, port, props, 0);
+}
+
+int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
+			union ib_gid *gid, int netw_view)
 {
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
 	int err = -ENOMEM;
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	int clear = 0;
+	int mad_ifc_flags = MLX4_MAD_IFC_IGNORE_KEYS;
 
 	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
 	out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
@@ -350,23 +374,38 @@ static int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 	in_mad->attr_id  = IB_SMP_ATTR_PORT_INFO;
 	in_mad->attr_mod = cpu_to_be32(port);
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	if (mlx4_is_mfunc(dev->dev) && netw_view)
+		mad_ifc_flags |= MLX4_MAD_IFC_NET_VIEW;
+
+	err = mlx4_MAD_IFC(dev, mad_ifc_flags, port, NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
 
 	memcpy(gid->raw, out_mad->data + 8, 8);
 
+	if (mlx4_is_mfunc(dev->dev) && !netw_view) {
+		if (index) {
+			/* For any index > 0, return the null guid */
+			err = 0;
+			clear = 1;
+			goto out;
+		}
+	}
+
 	init_query_mad(in_mad);
 	in_mad->attr_id  = IB_SMP_ATTR_GUID_INFO;
 	in_mad->attr_mod = cpu_to_be32(index / 8);
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	err = mlx4_MAD_IFC(dev, mad_ifc_flags, port,
+			   NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
 
 	memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8);
 
 out:
+	if (clear)
+		memset(gid->raw + 8, 0, 8);
 	kfree(in_mad);
 	kfree(out_mad);
 	return err;
@@ -386,16 +425,17 @@ static int mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
 			     union ib_gid *gid)
 {
 	if (rdma_port_get_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND)
-		return __mlx4_ib_query_gid(ibdev, port, index, gid);
+		return __mlx4_ib_query_gid(ibdev, port, index, gid, 0);
 	else
 		return iboe_query_gid(ibdev, port, index, gid);
 }
 
-static int mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
-			      u16 *pkey)
+int __mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
+			 u16 *pkey, int netw_view)
 {
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
+	int mad_ifc_flags = MLX4_MAD_IFC_IGNORE_KEYS;
 	int err = -ENOMEM;
 
 	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
@@ -407,7 +447,11 @@ static int mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
 	in_mad->attr_id  = IB_SMP_ATTR_PKEY_TABLE;
 	in_mad->attr_mod = cpu_to_be32(index / 32);
 
-	err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
+	if (mlx4_is_mfunc(to_mdev(ibdev)->dev) && netw_view)
+		mad_ifc_flags |= MLX4_MAD_IFC_NET_VIEW;
+
+	err = mlx4_MAD_IFC(to_mdev(ibdev), mad_ifc_flags, port, NULL, NULL,
+			   in_mad, out_mad);
 	if (err)
 		goto out;
 
@@ -419,6 +463,11 @@ static int mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
 	return err;
 }
 
+static int mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey)
+{
+	return __mlx4_ib_query_pkey(ibdev, port, index, pkey, 0);
+}
+
 static int mlx4_ib_modify_device(struct ib_device *ibdev, int mask,
 				 struct ib_device_modify *props)
 {
@@ -431,6 +480,9 @@ static int mlx4_ib_modify_device(struct ib_device *ibdev, int mask,
 	if (!(mask & IB_DEVICE_MODIFY_NODE_DESC))
 		return 0;
 
+	if (mlx4_is_slave(to_mdev(ibdev)->dev))
+		return -EOPNOTSUPP;
+
 	spin_lock_irqsave(&to_mdev(ibdev)->sm_lock, flags);
 	memcpy(ibdev->node_desc, props->node_desc, 64);
 	spin_unlock_irqrestore(&to_mdev(ibdev)->sm_lock, flags);
@@ -446,7 +498,7 @@ static int mlx4_ib_modify_device(struct ib_device *ibdev, int mask,
 	memset(mailbox->buf, 0, 256);
 	memcpy(mailbox->buf, props->node_desc, 64);
 	mlx4_cmd(to_mdev(ibdev)->dev, mailbox->dma, 1, 0,
-		 MLX4_CMD_SET_NODE, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+		 MLX4_CMD_SET_NODE, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
 
 	mlx4_free_cmd_mailbox(to_mdev(ibdev)->dev, mailbox);
 
@@ -849,6 +901,7 @@ static int init_node_data(struct mlx4_ib_dev *dev)
 {
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
+	int mad_ifc_flags = MLX4_MAD_IFC_IGNORE_KEYS;
 	int err = -ENOMEM;
 
 	in_mad  = kzalloc(sizeof *in_mad, GFP_KERNEL);
@@ -858,8 +911,10 @@ static int init_node_data(struct mlx4_ib_dev *dev)
 
 	init_query_mad(in_mad);
 	in_mad->attr_id = IB_SMP_ATTR_NODE_DESC;
+	if (mlx4_is_master(dev->dev))
+		mad_ifc_flags |= MLX4_MAD_IFC_NET_VIEW;
 
-	err = mlx4_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad);
+	err = mlx4_MAD_IFC(dev, mad_ifc_flags, 1, NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
 
@@ -867,10 +922,11 @@ static int init_node_data(struct mlx4_ib_dev *dev)
 
 	in_mad->attr_id = IB_SMP_ATTR_NODE_INFO;
 
-	err = mlx4_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad);
+	err = mlx4_MAD_IFC(dev, mad_ifc_flags, 1, NULL, NULL, in_mad, out_mad);
 	if (err)
 		goto out;
 
+	dev->dev->rev_id = be32_to_cpup((__be32 *) (out_mad->data + 32));
 	memcpy(&dev->ib_dev.node_guid, out_mad->data + 12, 8);
 
 out:
@@ -959,7 +1015,7 @@ static void update_gids_task(struct work_struct *work)
 
 	err = mlx4_cmd(dev, mailbox->dma, MLX4_SET_PORT_GID_TABLE << 8 | gw->port,
 		       1, MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
-		       MLX4_CMD_NATIVE);
+		       MLX4_CMD_WRAPPED);
 	if (err)
 		pr_warn("set port command failed\n");
 	else {
@@ -1121,6 +1177,38 @@ static int mlx4_ib_netdev_event(struct notifier_block *this, unsigned long event
 	return NOTIFY_DONE;
 }
 
+static void init_pkeys(struct mlx4_ib_dev *ibdev)
+{
+	int port;
+	int slave;
+	int i;
+
+	if (mlx4_is_master(ibdev->dev)) {
+		for (slave = 0; slave <= ibdev->dev->num_vfs; ++slave) {
+			for (port = 1; port <= ibdev->dev->caps.num_ports; ++port) {
+				for (i = 0;
+				     i < ibdev->dev->phys_caps.pkey_phys_table_len[port];
+				     ++i) {
+					ibdev->pkeys.virt2phys_pkey[slave][port - 1][i] =
+					/* master has the identity virt2phys pkey mapping */
+						(slave == mlx4_master_func_num(ibdev->dev) || !i) ? i :
+							ibdev->dev->phys_caps.pkey_phys_table_len[port] - 1;
+					mlx4_sync_pkey_table(ibdev->dev, slave, port, i,
+							     ibdev->pkeys.virt2phys_pkey[slave][port - 1][i]);
+				}
+			}
+		}
+		/* initialize pkey cache */
+		for (port = 1; port <= ibdev->dev->caps.num_ports; ++port) {
+			for (i = 0;
+			     i < ibdev->dev->phys_caps.pkey_phys_table_len[port];
+			     ++i)
+				ibdev->pkeys.phys_pkey_cache[port-1][i] =
+					(i) ? 0 : 0xFFFF;
+		}
+	}
+}
+
 static void mlx4_ib_alloc_eqs(struct mlx4_dev *dev, struct mlx4_ib_dev *ibdev)
 {
 	char name[32];
@@ -1207,11 +1295,15 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 
 	pr_info_once("%s", mlx4_ib_version);
 
-	if (mlx4_is_mfunc(dev)) {
-		pr_warn("IB not yet supported in SRIOV\n");
+	mlx4_foreach_non_ib_transport_port(i, dev)
+		num_ports++;
+
+	if (mlx4_is_mfunc(dev) && num_ports) {
+		dev_err(&dev->pdev->dev, "RoCE is not supported over SRIOV as yet\n");
 		return NULL;
 	}
 
+	num_ports = 0;
 	mlx4_foreach_ib_transport_port(i, dev)
 		num_ports++;
 
@@ -1318,10 +1410,12 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	ibdev->ib_dev.detach_mcast	= mlx4_ib_mcg_detach;
 	ibdev->ib_dev.process_mad	= mlx4_ib_process_mad;
 
-	ibdev->ib_dev.alloc_fmr		= mlx4_ib_fmr_alloc;
-	ibdev->ib_dev.map_phys_fmr	= mlx4_ib_map_phys_fmr;
-	ibdev->ib_dev.unmap_fmr		= mlx4_ib_unmap_fmr;
-	ibdev->ib_dev.dealloc_fmr	= mlx4_ib_fmr_dealloc;
+	if (!mlx4_is_slave(ibdev->dev)) {
+		ibdev->ib_dev.alloc_fmr		= mlx4_ib_fmr_alloc;
+		ibdev->ib_dev.map_phys_fmr	= mlx4_ib_map_phys_fmr;
+		ibdev->ib_dev.unmap_fmr		= mlx4_ib_unmap_fmr;
+		ibdev->ib_dev.dealloc_fmr	= mlx4_ib_fmr_dealloc;
+	}
 
 	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) {
 		ibdev->ib_dev.alloc_xrcd = mlx4_ib_alloc_xrcd;
@@ -1357,11 +1451,14 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	if (mlx4_ib_mad_init(ibdev))
 		goto err_reg;
 
+	if (mlx4_ib_init_sriov(ibdev))
+		goto err_mad;
+
 	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_IBOE && !iboe->nb.notifier_call) {
 		iboe->nb.notifier_call = mlx4_ib_netdev_event;
 		err = register_netdevice_notifier(&iboe->nb);
 		if (err)
-			goto err_reg;
+			goto err_sriov;
 	}
 
 	for (j = 0; j < ARRAY_SIZE(mlx4_class_attributes); ++j) {
@@ -1372,6 +1469,18 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 
 	ibdev->ib_active = true;
 
+	if (mlx4_is_mfunc(ibdev->dev))
+		init_pkeys(ibdev);
+
+	/* create paravirt contexts for any VFs which are active */
+	if (mlx4_is_master(ibdev->dev)) {
+		for (j = 0; j < MLX4_MFUNC_MAX; j++) {
+			if (j == mlx4_master_func_num(ibdev->dev))
+				continue;
+			if (mlx4_is_slave_active(ibdev->dev, j))
+				do_slave_init(ibdev, j, 1);
+		}
+	}
 	return ibdev;
 
 err_notif:
@@ -1379,6 +1488,12 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 		pr_warn("failure unregistering notifier\n");
 	flush_workqueue(wq);
 
+err_sriov:
+	mlx4_ib_close_sriov(ibdev);
+
+err_mad:
+	mlx4_ib_mad_cleanup(ibdev);
+
 err_reg:
 	ib_unregister_device(&ibdev->ib_dev);
 
@@ -1407,6 +1522,7 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
 	struct mlx4_ib_dev *ibdev = ibdev_ptr;
 	int p;
 
+	mlx4_ib_close_sriov(ibdev);
 	mlx4_ib_mad_cleanup(ibdev);
 	ib_unregister_device(&ibdev->ib_dev);
 	if (ibdev->iboe.nb.notifier_call) {
@@ -1428,6 +1544,51 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
 	ib_dealloc_device(&ibdev->ib_dev);
 }
 
+static void do_slave_init(struct mlx4_ib_dev *ibdev, int slave, int do_init)
+{
+	struct mlx4_ib_demux_work **dm = NULL;
+	struct mlx4_dev *dev = ibdev->dev;
+	int i;
+	unsigned long flags;
+
+	if (!mlx4_is_master(dev))
+		return;
+
+	dm = kcalloc(dev->caps.num_ports, sizeof *dm, GFP_ATOMIC);
+	if (!dm) {
+		pr_err("failed to allocate memory for tunneling qp update\n");
+		goto out;
+	}
+
+	for (i = 0; i < dev->caps.num_ports; i++) {
+		dm[i] = kmalloc(sizeof (struct mlx4_ib_demux_work), GFP_ATOMIC);
+		if (!dm[i]) {
+			pr_err("failed to allocate memory for tunneling qp update work struct\n");
+			for (i = 0; i < dev->caps.num_ports; i++) {
+				if (dm[i])
+					kfree(dm[i]);
+			}
+			goto out;
+		}
+	}
+	/* initialize or tear down tunnel QPs for the slave */
+	for (i = 0; i < dev->caps.num_ports; i++) {
+		INIT_WORK(&dm[i]->work, mlx4_ib_tunnels_update_work);
+		dm[i]->port = i + 1;
+		dm[i]->slave = slave;
+		dm[i]->do_init = do_init;
+		dm[i]->dev = ibdev;
+		spin_lock_irqsave(&ibdev->sriov.going_down_lock, flags);
+		if (!ibdev->sriov.is_going_down)
+			queue_work(ibdev->sriov.demux[i].ud_wq, &dm[i]->work);
+		spin_unlock_irqrestore(&ibdev->sriov.going_down_lock, flags);
+	}
+out:
+	if (dm)
+		kfree(dm);
+	return;
+}
+
 static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
 			  enum mlx4_dev_event event, unsigned long param)
 {
@@ -1435,22 +1596,28 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
 	struct mlx4_ib_dev *ibdev = to_mdev((struct ib_device *) ibdev_ptr);
 	struct mlx4_eqe *eqe = NULL;
 	struct ib_event_work *ew;
-	int port = 0;
+	int p = 0;
 
 	if (event == MLX4_DEV_EVENT_PORT_MGMT_CHANGE)
 		eqe = (struct mlx4_eqe *)param;
 	else
-		port = (u8)param;
-
-	if (port > ibdev->num_ports)
-		return;
+		p = (int) param;
 
 	switch (event) {
 	case MLX4_DEV_EVENT_PORT_UP:
+		if (p > ibdev->num_ports)
+			return;
+		if (mlx4_is_master(dev) &&
+		    rdma_port_get_link_layer(&ibdev->ib_dev, p) ==
+			IB_LINK_LAYER_INFINIBAND) {
+			mlx4_ib_invalidate_all_guid_record(ibdev, p);
+		}
 		ibev.event = IB_EVENT_PORT_ACTIVE;
 		break;
 
 	case MLX4_DEV_EVENT_PORT_DOWN:
+		if (p > ibdev->num_ports)
+			return;
 		ibev.event = IB_EVENT_PORT_ERR;
 		break;
 
@@ -1469,7 +1636,21 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
 		INIT_WORK(&ew->work, handle_port_mgmt_change_event);
 		memcpy(&ew->ib_eqe, eqe, sizeof *eqe);
 		ew->ib_dev = ibdev;
-		handle_port_mgmt_change_event(&ew->work);
+		/* need to queue only for port owner, which uses GEN_EQE */
+		if (mlx4_is_master(dev))
+			queue_work(wq, &ew->work);
+		else
+			handle_port_mgmt_change_event(&ew->work);
+		return;
+
+	case MLX4_DEV_EVENT_SLAVE_INIT:
+		/* here, p is the slave id */
+		do_slave_init(ibdev, p, 1);
+		return;
+
+	case MLX4_DEV_EVENT_SLAVE_SHUTDOWN:
+		/* here, p is the slave id */
+		do_slave_init(ibdev, p, 0);
 		return;
 
 	default:
@@ -1477,7 +1658,7 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
 	}
 
 	ibev.device	      = ibdev_ptr;
-	ibev.element.port_num = port;
+	ibev.element.port_num = (u8) p;
 
 	ib_dispatch_event(&ibev);
 }
@@ -1497,18 +1678,28 @@ static int __init mlx4_ib_init(void)
 	if (!wq)
 		return -ENOMEM;
 
+	err = mlx4_ib_mcg_init();
+	if (err)
+		goto clean_wq;
+
 	err = mlx4_register_interface(&mlx4_ib_interface);
-	if (err) {
-		destroy_workqueue(wq);
-		return err;
-	}
+	if (err)
+		goto clean_mcg;
 
 	return 0;
+
+clean_mcg:
+	mlx4_ib_mcg_destroy();
+
+clean_wq:
+	destroy_workqueue(wq);
+	return err;
 }
 
 static void __exit mlx4_ib_cleanup(void)
 {
 	mlx4_unregister_interface(&mlx4_ib_interface);
+	mlx4_ib_mcg_destroy();
 	destroy_workqueue(wq);
 }
 

drivers/infiniband/hw/nes/nes.h

@@ -399,11 +399,20 @@ static inline void nes_write8(void __iomem *addr, u8 val)
 	writeb(val, addr);
 }
 
-
+enum nes_resource {
+	NES_RESOURCE_MW = 1,
+	NES_RESOURCE_FAST_MR,
+	NES_RESOURCE_PHYS_MR,
+	NES_RESOURCE_USER_MR,
+	NES_RESOURCE_PD,
+	NES_RESOURCE_QP,
+	NES_RESOURCE_CQ,
+	NES_RESOURCE_ARP
+};
 
 static inline int nes_alloc_resource(struct nes_adapter *nesadapter,
 		unsigned long *resource_array, u32 max_resources,
-		u32 *req_resource_num, u32 *next)
+		u32 *req_resource_num, u32 *next, enum nes_resource resource_type)
 {
 	unsigned long flags;
 	u32 resource_num;
@@ -414,7 +423,7 @@ static inline int nes_alloc_resource(struct nes_adapter *nesadapter,
 	if (resource_num >= max_resources) {
 		resource_num = find_first_zero_bit(resource_array, max_resources);
 		if (resource_num >= max_resources) {
-			printk(KERN_ERR PFX "%s: No available resourcess.\n", __func__);
+			printk(KERN_ERR PFX "%s: No available resources [type=%u].\n", __func__, resource_type);
 			spin_unlock_irqrestore(&nesadapter->resource_lock, flags);
 			return -EMFILE;
 		}

drivers/infiniband/hw/nes/nes_hw.c

@@ -3575,10 +3575,10 @@ static void nes_process_iwarp_aeqe(struct nes_device *nesdev,
 	tcp_state = (aeq_info & NES_AEQE_TCP_STATE_MASK) >> NES_AEQE_TCP_STATE_SHIFT;
 	iwarp_state = (aeq_info & NES_AEQE_IWARP_STATE_MASK) >> NES_AEQE_IWARP_STATE_SHIFT;
 	nes_debug(NES_DBG_AEQ, "aeid = 0x%04X, qp-cq id = %d, aeqe = %p,"
-			" Tcp state = %s, iWARP state = %s\n",
+			" Tcp state = %d, iWARP state = %d\n",
 			async_event_id,
 			le32_to_cpu(aeqe->aeqe_words[NES_AEQE_COMP_QP_CQ_ID_IDX]), aeqe,
-			nes_tcp_state_str[tcp_state], nes_iwarp_state_str[iwarp_state]);
+			tcp_state, iwarp_state);
 
 	aeqe_cq_id = le32_to_cpu(aeqe->aeqe_words[NES_AEQE_COMP_QP_CQ_ID_IDX]);
 	if (aeq_info & NES_AEQE_QP) {

drivers/infiniband/hw/nes/nes_utils.c

@@ -699,7 +699,7 @@ int nes_arp_table(struct nes_device *nesdev, u32 ip_addr, u8 *mac_addr, u32 acti
 
 		arp_index = 0;
 		err = nes_alloc_resource(nesadapter, nesadapter->allocated_arps,
-				nesadapter->arp_table_size, (u32 *)&arp_index, &nesadapter->next_arp_index);
+				nesadapter->arp_table_size, (u32 *)&arp_index, &nesadapter->next_arp_index, NES_RESOURCE_ARP);
 		if (err) {
 			nes_debug(NES_DBG_NETDEV, "nes_alloc_resource returned error = %u\n", err);
 			return err;

drivers/infiniband/hw/qib/qib_common.h

@@ -285,7 +285,6 @@ struct qib_base_info {
 
 #ifndef QIB_KERN_TYPE
 #define QIB_KERN_TYPE 0
-#define QIB_IDSTR "QLogic kernel.org driver"
 #endif
 
 /*
@@ -301,6 +300,19 @@ struct qib_base_info {
 */
 #define QIB_KERN_SWVERSION ((QIB_KERN_TYPE << 31) | QIB_USER_SWVERSION)
 
+/*
+ * Define the driver version number.  This is something that refers only
+ * to the driver itself, not the software interfaces it supports.
+ */
+#define QIB_DRIVER_VERSION_BASE "1.11"
+
+/* create the final driver version string */
+#ifdef QIB_IDSTR
+#define QIB_DRIVER_VERSION QIB_DRIVER_VERSION_BASE " " QIB_IDSTR
+#else
+#define QIB_DRIVER_VERSION QIB_DRIVER_VERSION_BASE
+#endif
+
 /*
  * If the unit is specified via open, HCA choice is fixed.  If port is
  * specified, it's also fixed.  Otherwise we try to spread contexts

drivers/infiniband/hw/qib/qib_driver.c

@@ -46,7 +46,7 @@
  * The size has to be longer than this string, so we can append
  * board/chip information to it in the init code.
  */
-const char ib_qib_version[] = QIB_IDSTR "\n";
+const char ib_qib_version[] = QIB_DRIVER_VERSION "\n";
 
 DEFINE_SPINLOCK(qib_devs_lock);
 LIST_HEAD(qib_dev_list);
@@ -65,6 +65,7 @@ MODULE_PARM_DESC(compat_ddr_negotiate,
 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_AUTHOR("QLogic <support@qlogic.com>");
 MODULE_DESCRIPTION("QLogic IB driver");
+MODULE_VERSION(QIB_DRIVER_VERSION);
 
 /*
  * QIB_PIO_MAXIBHDR is the max IB header size allowed for in our

drivers/infiniband/hw/qib/qib_keys.c

@@ -186,8 +186,9 @@ int qib_lkey_ok(struct qib_lkey_table *rkt, struct qib_pd *pd,
 		goto bail;
 
 	off = sge->addr - mr->user_base;
-	if (unlikely(sge->addr < mr->iova || off + sge->length > mr->length ||
-		     (mr->access_flags & acc) == 0))
+	if (unlikely(sge->addr < mr->user_base ||
+		     off + sge->length > mr->length ||
+		     (mr->access_flags & acc) != acc))
 		goto bail;
 	if (unlikely(!atomic_inc_not_zero(&mr->refcount)))
 		goto bail;