dm thin: irqsave must always be used with the pool->lock spinlock
authorJoe Thornber <ejt@redhat.com>
Tue, 8 Apr 2014 10:08:41 +0000 (11:08 +0100)
committerMike Snitzer <snitzer@redhat.com>
Tue, 8 Apr 2014 14:10:51 +0000 (10:10 -0400)
Commit c140e1c4e23 ("dm thin: use per thin device deferred bio lists")
incorrectly stopped disabling irqs when taking the pool's spinlock.

Irqs must be disabled when taking the pool's spinlock otherwise a thread
could spin_lock(), then get interrupted to service thin_endio() in
interrupt context, which would then deadlock in spin_lock_irqsave().

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
drivers/md/dm-thin.c

index 53728be84dee35ac8dfabbf48087919841049f1a..ae5fd0b9c75ccf525efb8d36c81bacafb304d49c 100644 (file)
@@ -3101,6 +3101,7 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv)
        struct thin_c *tc;
        struct dm_dev *pool_dev, *origin_dev;
        struct mapped_device *pool_md;
+       unsigned long flags;
 
        mutex_lock(&dm_thin_pool_table.mutex);
 
@@ -3191,9 +3192,9 @@ static int thin_ctr(struct dm_target *ti, unsigned argc, char **argv)
 
        mutex_unlock(&dm_thin_pool_table.mutex);
 
-       spin_lock(&tc->pool->lock);
+       spin_lock_irqsave(&tc->pool->lock, flags);
        list_add_tail_rcu(&tc->list, &tc->pool->active_thins);
-       spin_unlock(&tc->pool->lock);
+       spin_unlock_irqrestore(&tc->pool->lock, flags);
        /*
         * This synchronize_rcu() call is needed here otherwise we risk a
         * wake_worker() call finding no bios to process (because the newly