Re: bio linked list corruption.

[Jens Axboe]

On 10/26/2016 04:58 PM, Linus Torvalds wrote:
> On Wed, Oct 26, 2016 at 3:51 PM, Linus Torvalds
> <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> > Dave: it might be a good idea to split that "WARN_ON_ONCE()" in
> > blk_mq_merge_queue_io() into two
>
> I did that myself too, since Dave sees this during boot.
>
> But I'm not getting the warning ;(
>
> Dave gets it with ext4, and thats' what I have too, so I'm not sure
> what the required trigger would be.

Actually, I think I see what might trigger it. You are on nvme, iirc,
and that has a deep queue. Dave, are you testing on a sata drive or
something similar with a shallower queue depth? If we end up sleeping
for a request, I think we could trigger data->ctx being different.

Dave, can you hit the warnings with this? Totally untested...


diff --git a/block/blk-mq.c b/block/blk-mq.c
index ddc2eed64771..80a9c45a9235 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1217,9 +1217,7 @@ static struct request *blk_mq_map_request(struct 
request_queue *q,
 	blk_mq_set_alloc_data(&alloc_data, q, 0, ctx, hctx);
 	rq = __blk_mq_alloc_request(&alloc_data, op, op_flags);

-	hctx->queued++;
-	data->hctx = hctx;
-	data->ctx = ctx;
+	data->hctx->queued++;
 	return rq;
 }


--
Jens Axboe