block: init flush rq ref count to 1
PATCH 1/1 block: init flush rq ref count to 1
We discovered a problem in newer kernels where a disconnect of a NBD device while the flush request was pending would result in a hang. This is because the blk mq timeout handler does
if (!refcount_inc_not_zero(&rq->ref))
return true;to determine if it's ok to run the timeout handler for the request. Flush_rq's don't have a ref count set, so we'd skip running the timeout handler for this request and it would just sit there in limbo forever.
Fix this by always setting the refcount of any request going through blk_init_rq() to 1. I tested this with a nbd-server that dropped flush requests to verify that it hung, and then tested with this patch to verify I got the timeout as expected and the error handling kicked in. Thanks,
--- block/blk-core.c | 1 + 1 file changed, 1 insertion(+)
block/blk-core.c +1 -0
@@ -116,6 +116,7 @@ void blk_rq_init(struct request_queue *q, struct request *rq) rq->internal_tag =-1 ; rq->start_time_ns =ktime_get_ns (); rq->part =NULL ; +refcount_set (&rq->ref,1 ); }EXPORT_SYMBOL (blk_rq_init);
Signature · 1 line
2.14.3
Jens Axboe
maintainerQuoted 11 lines · Josef Bacik
We discovered a problem in newer kernels where a disconnect of a NBD device while the flush request was pending would result in a hang. This is because the blk mq timeout handler does
if (!refcount_inc_not_zero(&rq->ref)) return true;to determine if it's ok to run the timeout handler for the request. Flush_rq's don't have a ref count set, so we'd skip running the timeout handler for this request and it would just sit there in limbo forever.
Quoted 5 lines
Fix this by always setting the refcount of any request going through
blk_init_rq()to 1. I tested this with a nbd-server that dropped flush requests to verify that it hung, and then tested with this patch to verify I got the timeout as expected and the error handling kicked in. Thanks,
Looks good to me, thanks Josef.
Signature · 1 line
Jens Axboe