2021-12-21 08:11:28

by Hao Peng

Subject: [PATCH] block/elevator: handle possible null pointer

There is a check for q->tag_set in the front of elevator_get_default,
and there should be a check here too.

Signed-off-by: Peng Hao <[email protected]>
---
block/elevator.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/block/elevator.c b/block/elevator.c
index 1f39f6e8ebb9..b7d0bead680c 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -636,7 +636,7 @@ static struct elevator_type *elevator_get_default(struct request_queue *q)
if (q->tag_set && q->tag_set->flags & BLK_MQ_F_NO_SCHED_BY_DEFAULT)
return NULL;

- if (q->nr_hw_queues != 1 &&
+ if (q->nr_hw_queues != 1 && q->tag_set &&
!blk_mq_is_shared_tags(q->tag_set->flags))
return NULL;

--
2.27.0

2021-12-21 15:26:30

by Jens Axboe

Subject: Re: [PATCH] block/elevator: handle possible null pointer

On 12/21/21 1:10 AM, Peng Hao wrote:
> There is a check for q->tag_set in the front of elevator_get_default,
> and there should be a check here too.

I always get suspicious when I see patches like that. Is the other check
valid? Why does it need to get checked? There's really no meat on the
bone in this commit message.

--
Jens Axboe