[PATCH for-3.17 0/7] RoCE IP addressing fixes

[PATCH for-3.17 0/7] RoCE IP addressing fixes

[Or Gerlitz]

Hi Roland,

This series contains RoCE related fixes to the mlx4 IB driver 
(and one uverbs patch), which deal with the IP addressing logic.

Once applied into 3.17, we will further push them to -stable.

Or.

Jack Morgenstein (1):
  IB/mlx4: Fix lockdep splat for the iboe lock

Matan Barak (1):
  IB/core: When marsheling uverbs path, clear unused fields

Moni Shoua (5):
  IB/mlx4: Avoid null pointer dereference in mlx4_ib_scan_netdevs()
  IB/mlx4: Don't duplicate the default RoCE GID
  IB/mlx4: Reorder steps in RoCE GID table initialization
  IB/mlx4: Get upper dev addresses as RoCE GIDs when port comes up
  IB/mlx4: Avoid executing gid task when device is being removed

 drivers/infiniband/core/uverbs_marshall.c |    4 +
 drivers/infiniband/hw/mlx4/main.c         |  124 ++++++++++++++++++-----------
 2 files changed, 81 insertions(+), 47 deletions(-)

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 1/7] IB/mlx4: Avoid null pointer dereference in mlx4_ib_scan_netdevs()

[Or Gerlitz]

From: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

When Ethernet netdev is not present for a port (e.g. when the link
layer type of the port is InfiniBand) it's possible to dereference
a null pointer when we do netdevice scanning.

To fix that, we move a section of code that needs to run only when
netdev is present to a proper if () statement.

Fixes: ad4885d2 ('IB/mlx4: Build the port IBoE GID table properly under bonding')
Reported-by: Dan Carpenter <dan.carpenter-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c |   49 +++++++++++++++++++-----------------
 1 files changed, 26 insertions(+), 23 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index e1e558a..a6a9e57 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1786,31 +1786,34 @@ 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 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 && (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);
 		}
 	}
 
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 2/7] IB/mlx4: Don't duplicate the default RoCE GID

[Or Gerlitz]

From: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

When reading the IPv6 addresses from the net-device, make sure to avoid
adding duplicate entry to the GID table as of equality between the default
GID we generate to the default IPv6 link-local address of the device.

Fixes: acc4fcc ('IB/mlx4: Make sure GID index 0 is always occupied')
Signed-off-by: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index a6a9e57..61910f4 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1676,6 +1676,7 @@ 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;
 
@@ -1696,12 +1697,15 @@ 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);
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 3/7] IB/mlx4: Reorder steps in RoCE GID table initialization

[Or Gerlitz]

From: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

There's no need to reset the gid table twice and we need to do it only
for Ethernet ports. Also, no need to actively scan ndetdevs since it's being
done immediatly after we register netdev notifiers.

Signed-off-by: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c |   29 +++++++++++++++++------------
 1 files changed, 17 insertions(+), 12 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 61910f4..d95b5fe 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1727,24 +1727,33 @@ 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);
 
 	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);
 	read_unlock(&dev_base_lock);
-
-	return 0;
+out:
+	return err;
 }
 
 static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev,
@@ -2199,12 +2208,8 @@ 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) {
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 4/7] IB/mlx4: Get upper dev addresses as RoCE GIDs when port comes up

[Or Gerlitz]

From: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

When RoCE port becomes active and netdev of the port has upper
device (e.g bond/team), gids derived from the upper dev should
appear in the port's RoCE gid table.

Signed-off-by: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c |   25 +++++++++++++++++--------
 1 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index d95b5fe..dc0dac0 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1799,14 +1799,23 @@ 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);
-			/* 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 (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
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 5/7] IB/mlx4: Fix lockdep splat for the iboe lock

[Or Gerlitz]

From: Jack Morgenstein <jackm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>

Chuck Lever reported the following stack trace:

 =================================
 [ INFO: inconsistent lock state ]
 3.16.0-rc2-00024-g2e78883 #17 Tainted: G            E
 ---------------------------------
 inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
 swapper/0/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
 (&(&iboe->lock)->rlock){+.?...}, at: [<ffffffffa065f68b>] mlx4_ib_addr_event+0xdb/0x1a0 [mlx4_ib]
 {SOFTIRQ-ON-W} state was registered at:
  [<ffffffff810b3110>] mark_irqflags+0x110/0x170
  [<ffffffff810b4806>] __lock_acquire+0x2c6/0x5b0
  [<ffffffff810b4bd9>] lock_acquire+0xe9/0x120
  [<ffffffff815f7f6e>] _raw_spin_lock+0x3e/0x80
  [<ffffffffa0661084>] mlx4_ib_scan_netdevs+0x34/0x260 [mlx4_ib]
  [<ffffffffa06612db>] mlx4_ib_netdev_event+0x2b/0x40 [mlx4_ib]
  [<ffffffff81522219>] register_netdevice_notifier+0x99/0x1e0
  [<ffffffffa06626e3>] mlx4_ib_add+0x743/0xbc0 [mlx4_ib]
  [<ffffffffa05ec168>] mlx4_add_device+0x48/0xa0 [mlx4_core]
  [<ffffffffa05ec2c3>] mlx4_register_interface+0x73/0xb0 [mlx4_core]
  [<ffffffffa05c505e>] cm_req_handler+0x13e/0x460 [ib_cm]
  [<ffffffff810002e2>] do_one_initcall+0x112/0x1c0
  [<ffffffff810e8264>] do_init_module+0x34/0x190
  [<ffffffff810ea62f>] load_module+0x5cf/0x740
  [<ffffffff810ea939>] SyS_init_module+0x99/0xd0
  [<ffffffff815f8fd2>] system_call_fastpath+0x16/0x1b
 irq event stamp: 336142
 hardirqs last  enabled at (336142): [<ffffffff810612f5>] __local_bh_enable_ip+0xb5/0xc0
 hardirqs last disabled at (336141): [<ffffffff81061296>] __local_bh_enable_ip+0x56/0xc0
 softirqs last  enabled at (336004): [<ffffffff8106123a>] _local_bh_enable+0x4a/0x50
 softirqs last disabled at (336005): [<ffffffff810617a4>] irq_exit+0x44/0xd0

 other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(&(&iboe->lock)->rlock);
  <Interrupt>
    lock(&(&iboe->lock)->rlock);

 *** DEADLOCK ***

The above problem was caused by the spin lock being taken both in the process
context and in a soft-irq context (in a netdev notifier handler).

The required fix is to use spin_lock/unlock_bh() instead of spin_lock/unlock
on the iboe lock.

Reported-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Jack Morgenstein <jackm-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c |   24 ++++++++++++------------
 1 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index dc0dac0..5179504 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -357,7 +357,7 @@ 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 +369,7 @@ 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 +811,11 @@ 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;
@@ -1262,11 +1262,11 @@ 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);
@@ -1551,7 +1551,7 @@ 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) &&
@@ -1561,7 +1561,7 @@ 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;
 
 }
@@ -1739,7 +1739,7 @@ static int mlx4_ib_init_gid_table(struct mlx4_ib_dev *ibdev)
 	}
 
 	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);
@@ -1750,7 +1750,7 @@ static int mlx4_ib_init_gid_table(struct mlx4_ib_dev *ibdev)
 		}
 	}
 
-	spin_unlock(&iboe->lock);
+	spin_unlock_bh(&iboe->lock);
 	read_unlock(&dev_base_lock);
 out:
 	return err;
@@ -1767,7 +1767,7 @@ 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];
@@ -1839,7 +1839,7 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev,
 		}
 	}
 
-	spin_unlock(&iboe->lock);
+	spin_unlock_bh(&iboe->lock);
 
 	if (update_qps_port > 0)
 		mlx4_ib_update_qps(ibdev, dev, update_qps_port);
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 6/7] IB/mlx4: Avoid executing gid task when device is being removed

[Or Gerlitz]

From: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

When device is being removed (e.g during VPI port link type change from ETH
to IB), tasks for gid table changes should not be executed.

Flush the current queue of tasks and block further tasks from entering the queue.

Signed-off-by: Moni Shoua <monis-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 5179504..ea30f9e 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1387,6 +1387,9 @@ 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));
@@ -1417,6 +1420,9 @@ 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");
@@ -2366,6 +2372,9 @@ 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);
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH for-3.17 7/7] IB/core: When marsheling uverbs path, clear unused fields

[Or Gerlitz]

From: Matan Barak <matanb-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

When marsheling a user path to the kernel struct ib_sa_path, need
to zero smac, dmac and set the vlan id to the "no vlan" value.

Fixes: dd5f03b ('IB/core: Ethernet L2 attributes in verbs/cm structures')
Reported-by: Aleksey Senin <alekseys-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Matan Barak <matanb-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Or Gerlitz <ogerlitz-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 drivers/infiniband/core/uverbs_marshall.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/drivers/infiniband/core/uverbs_marshall.c b/drivers/infiniband/core/uverbs_marshall.c
index e7bee46..5c4a040 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 = be16_to_cpu(0xffff);
 }
 EXPORT_SYMBOL(ib_copy_path_rec_from_user);
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH for-3.17 0/7] RoCE IP addressing fixes

[Or Gerlitz]

On 8/21/2014 2:28 PM, Or Gerlitz wrote:
> Hi Roland,
>
> This series contains RoCE related fixes to the mlx4 IB driver
> (and one uverbs patch), which deal with the IP addressing logic.
>
> Once applied into 3.17, we will further push them to -stable.

Roland,

This series addresses real bugs whose fixes should go upstream and from 
there into inbox drivers, can you please act to send post-rc1 update 
which contains these patches and whatever else is in the queue for 3.17-rc?

Note that for patch 7/7 there's V1, I sent it to you today.

Or.

>
> Or.
>
> Jack Morgenstein (1):
>    IB/mlx4: Fix lockdep splat for the iboe lock
>
> Matan Barak (1):
>    IB/core: When marsheling uverbs path, clear unused fields
>
> Moni Shoua (5):
>    IB/mlx4: Avoid null pointer dereference in mlx4_ib_scan_netdevs()
>    IB/mlx4: Don't duplicate the default RoCE GID
>    IB/mlx4: Reorder steps in RoCE GID table initialization
>    IB/mlx4: Get upper dev addresses as RoCE GIDs when port comes up
>    IB/mlx4: Avoid executing gid task when device is being removed
>
>   drivers/infiniband/core/uverbs_marshall.c |    4 +
>   drivers/infiniband/hw/mlx4/main.c         |  124 ++++++++++++++++++-----------
>   2 files changed, 81 insertions(+), 47 deletions(-)
>

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html