dm thin: irqsave must always be used with the pool->lock spinlock

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>

1 files changed, 3 insertions, 2 deletions

diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 53728be..ae5fd0b 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -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