Since blk_queue_enter() uses rcu_read_lock_sched() scsi_device_quiesce()
must use synchronize_sched().
Reported-by: Tejun Heo <tj@xxxxxxxxxx>
Fixes: 3a0a529971ec ("block, scsi: Make SCSI quiesce and resume work reliably")
Signed-off-by: Bart Van Assche <bart.vanassche@xxxxxxx>
Cc: Hannes Reinecke <hare@xxxxxxxx>
Cc: Ming Lei <ming.lei@xxxxxxxxxx>
Cc: Christoph Hellwig <hch@xxxxxx>
Cc: Johannes Thumshirn <jthumshirn@xxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: Oleksandr Natalenko <oleksandr@xxxxxxxxxxxxxx>
Cc: Martin Steigerwald <martin@xxxxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx # v4.15
---
drivers/scsi/scsi_lib.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index 1d83f29aee74..0b99ee2fbbb5 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -3014,7 +3014,7 @@ scsi_device_quiesce(struct scsi_device *sdev)
* unfreeze even if the queue was already frozen before this function
* was called. See also https://lwn.net/Articles/573497/.
*/
- synchronize_rcu();
+ synchronize_sched();
blk_mq_unfreeze_queue(q);
mutex_lock(&sdev->state_mutex);
--
2.16.2
