[PATCH 3.16 035/212] IB/ipoib: Fix deadlock between rmmod and set_mode

From: Ben Hutchings
Date: Thu Jun 01 2017 - 12:29:01 EST


3.16.44-rc1 review patch. If anyone has any objections, please let me know.

------------------

From: Feras Daoud <ferasda@xxxxxxxxxxxx>

commit 0a0007f28304cb9fc87809c86abb80ec71317f20 upstream.

When calling set_mode from sys/fs, the call flow locks the sys/fs lock
first and then tries to lock rtnl_lock (when calling ipoib_set_mod).
On the other hand, the rmmod call flow takes the rtnl_lock first
(when calling unregister_netdev) and then tries to take the sys/fs
lock. Deadlock a->b, b->a.

The problem starts when ipoib_set_mod frees it's rtnl_lck and tries
to get it after that.

set_mod:
[<ffffffff8104f2bd>] ? check_preempt_curr+0x6d/0x90
[<ffffffff814fee8e>] __mutex_lock_slowpath+0x13e/0x180
[<ffffffff81448655>] ? __rtnl_unlock+0x15/0x20
[<ffffffff814fed2b>] mutex_lock+0x2b/0x50
[<ffffffff81448675>] rtnl_lock+0x15/0x20
[<ffffffffa02ad807>] ipoib_set_mode+0x97/0x160 [ib_ipoib]
[<ffffffffa02b5f5b>] set_mode+0x3b/0x80 [ib_ipoib]
[<ffffffff8134b840>] dev_attr_store+0x20/0x30
[<ffffffff811f0fe5>] sysfs_write_file+0xe5/0x170
[<ffffffff8117b068>] vfs_write+0xb8/0x1a0
[<ffffffff8117ba81>] sys_write+0x51/0x90
[<ffffffff8100b0f2>] system_call_fastpath+0x16/0x1b

rmmod:
[<ffffffff81279ffc>] ? put_dec+0x10c/0x110
[<ffffffff8127a2ee>] ? number+0x2ee/0x320
[<ffffffff814fe6a5>] schedule_timeout+0x215/0x2e0
[<ffffffff8127cc04>] ? vsnprintf+0x484/0x5f0
[<ffffffff8127b550>] ? string+0x40/0x100
[<ffffffff814fe323>] wait_for_common+0x123/0x180
[<ffffffff81060250>] ? default_wake_function+0x0/0x20
[<ffffffff8119661e>] ? ifind_fast+0x5e/0xb0
[<ffffffff814fe43d>] wait_for_completion+0x1d/0x20
[<ffffffff811f2e68>] sysfs_addrm_finish+0x228/0x270
[<ffffffff811f2fb3>] sysfs_remove_dir+0xa3/0xf0
[<ffffffff81273f66>] kobject_del+0x16/0x40
[<ffffffff8134cd14>] device_del+0x184/0x1e0
[<ffffffff8144e59b>] netdev_unregister_kobject+0xab/0xc0
[<ffffffff8143c05e>] rollback_registered+0xae/0x130
[<ffffffff8143c102>] unregister_netdevice+0x22/0x70
[<ffffffff8143c16e>] unregister_netdev+0x1e/0x30
[<ffffffffa02a91b0>] ipoib_remove_one+0xe0/0x120 [ib_ipoib]
[<ffffffffa01ed95f>] ib_unregister_device+0x4f/0x100 [ib_core]
[<ffffffffa021f5e1>] mlx4_ib_remove+0x41/0x180 [mlx4_ib]
[<ffffffffa01ab771>] mlx4_remove_device+0x71/0x90 [mlx4_core]

Fixes: 862096a8bbf8 ("IB/ipoib: Add more rtnl_link_ops callbacks")
Cc: Or Gerlitz <ogerlitz@xxxxxxxxxxxx>
Signed-off-by: Feras Daoud <ferasda@xxxxxxxxxxxx>
Signed-off-by: Erez Shitrit <erezsh@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leon@xxxxxxxxxx>
Signed-off-by: Doug Ledford <dledford@xxxxxxxxxx>
Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
---
drivers/infiniband/ulp/ipoib/ipoib_cm.c | 12 +++++++-----
drivers/infiniband/ulp/ipoib/ipoib_main.c | 6 ++----
2 files changed, 9 insertions(+), 9 deletions(-)

--- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
@@ -1496,12 +1496,14 @@ static ssize_t set_mode(struct device *d

ret = ipoib_set_mode(dev, buf);

- rtnl_unlock();
+ /* The assumption is that the function ipoib_set_mode returned
+ * with the rtnl held by it, if not the value -EBUSY returned,
+ * then no need to rtnl_unlock
+ */
+ if (ret != -EBUSY)
+ rtnl_unlock();

- if (!ret)
- return count;
-
- return ret;
+ return (!ret || ret == -EBUSY) ? count : ret;
}

static DEVICE_ATTR(mode, S_IWUSR | S_IRUGO, show_mode, set_mode);
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -243,8 +243,7 @@ int ipoib_set_mode(struct net_device *de
priv->tx_wr.send_flags &= ~IB_SEND_IP_CSUM;

ipoib_flush_paths(dev);
- rtnl_lock();
- return 0;
+ return (!rtnl_trylock()) ? -EBUSY : 0;
}

if (!strcmp(buf, "datagram\n")) {
@@ -253,8 +252,7 @@ int ipoib_set_mode(struct net_device *de
dev_set_mtu(dev, min(priv->mcast_mtu, dev->mtu));
rtnl_unlock();
ipoib_flush_paths(dev);
- rtnl_lock();
- return 0;
+ return (!rtnl_trylock()) ? -EBUSY : 0;
}

return -EINVAL;