diff mbox series

[v4,01/10] scsi: core: Fix a race between scsi_done() and scsi_timeout()

Message ID 20221018202958.1902564-2-bvanassche@acm.org
State New
Headers show
Series Fix a deadlock in the UFS driver | expand

Commit Message

Bart Van Assche Oct. 18, 2022, 8:29 p.m. UTC
If there is a race between scsi_done() and scsi_timeout() and if
scsi_timeout() loses the race, scsi_timeout() should not reset the
request timer. Hence change the return value for this case from
BLK_EH_RESET_TIMER into BLK_EH_DONE.

Although the block layer holds a reference on a request (req->ref) while
calling a timeout handler, restarting the timer (blk_add_timer()) while
a request is being completed is racy.

Reviewed-by: Mike Christie <michael.christie@oracle.com>
Cc: Keith Busch <kbusch@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: John Garry <john.garry@huawei.com>
Cc: Hannes Reinecke <hare@suse.de>
Reported-by: Adrian Hunter <adrian.hunter@intel.com>
Fixes: 15f73f5b3e59 ("blk-mq: move failure injection out of blk_mq_complete_request")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
---
 drivers/scsi/scsi_error.c | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)
diff mbox series

Patch

diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c
index 6995c8979230..02520f912306 100644
--- a/drivers/scsi/scsi_error.c
+++ b/drivers/scsi/scsi_error.c
@@ -343,19 +343,11 @@  enum blk_eh_timer_return scsi_timeout(struct request *req)
 
 	if (rtn == BLK_EH_DONE) {
 		/*
-		 * Set the command to complete first in order to prevent a real
-		 * completion from releasing the command while error handling
-		 * is using it. If the command was already completed, then the
-		 * lower level driver beat the timeout handler, and it is safe
-		 * to return without escalating error recovery.
-		 *
-		 * If timeout handling lost the race to a real completion, the
-		 * block layer may ignore that due to a fake timeout injection,
-		 * so return RESET_TIMER to allow error handling another shot
-		 * at this command.
+		 * If scsi_done() has already set SCMD_STATE_COMPLETE, do not
+		 * modify *scmd.
 		 */
 		if (test_and_set_bit(SCMD_STATE_COMPLETE, &scmd->state))
-			return BLK_EH_RESET_TIMER;
+			return BLK_EH_DONE;
 		if (scsi_abort_command(scmd) != SUCCESS) {
 			set_host_byte(scmd, DID_TIME_OUT);
 			scsi_eh_scmd_add(scmd);