From 65d56083b4617a4cade0cff68cbbaf68114169d6 Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Wed, 19 Dec 2012 13:51:44 -0800
Subject: [PATCH] Fix zpool on zvol lock inversion deadlock

In all but one case the spa_namespace_lock is taken before the
bdev->bd_mutex lock.  But Linux __blkdev_get() function calls
fops->open() with the bdev->bd_mutex lock held and we must
somehow still safely acquire the spa_namespace_lock.

To avoid a potential lock inversion deadlock we preemptively
try to take the spa_namespace_lock().  Normally it will not
be contended and this is safe because spa_open_common() handles
the case where the caller already holds the spa_namespace_lock.

When it is contended we risk a lock inversion if we were to
block waiting for the lock.  Luckily, the __blkdev_get()
function allows us to return -ERESTARTSYS which will result in
bdev->bd_mutex being dropped, reacquired, and fops->open() being
called again.  This process can be repeated safely until both
locks are acquired.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Jorgen Lundman <lundman@lundman.net>
Closes #612
---
 module/zfs/zvol.c | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/module/zfs/zvol.c b/module/zfs/zvol.c
index 7a448f194..d4d533f02 100644
--- a/module/zfs/zvol.c
+++ b/module/zfs/zvol.c
@@ -891,11 +891,39 @@ zvol_first_open(zvol_state_t *zv)
 {
 	objset_t *os;
 	uint64_t volsize;
+	int locked = 0;
 	int error;
 	uint64_t ro;
 
+	/*
+	 * In all other cases the spa_namespace_lock is taken before the
+	 * bdev->bd_mutex lock.  But in this case the Linux __blkdev_get()
+	 * function calls fops->open() with the bdev->bd_mutex lock held.
+	 *
+	 * To avoid a potential lock inversion deadlock we preemptively
+	 * try to take the spa_namespace_lock().  Normally it will not
+	 * be contended and this is safe because spa_open_common() handles
+	 * the case where the caller already holds the spa_namespace_lock.
+	 *
+	 * When it is contended we risk a lock inversion if we were to
+	 * block waiting for the lock.  Luckily, the __blkdev_get()
+	 * function allows us to return -ERESTARTSYS which will result in
+	 * bdev->bd_mutex being dropped, reacquired, and fops->open() being
+	 * called again.  This process can be repeated safely until both
+	 * locks are acquired.
+	 */
+	if (!mutex_owned(&spa_namespace_lock)) {
+		locked = mutex_tryenter(&spa_namespace_lock);
+		if (!locked)
+			return (-ERESTARTSYS);
+	}
+
 	/* lie and say we're read-only */
 	error = dmu_objset_own(zv->zv_name, DMU_OST_ZVOL, 1, zvol_tag, &os);
+
+	if (locked)
+		mutex_exit(&spa_namespace_lock);
+
 	if (error)
 		return (-error);