2015-07-19 06:53:55

by Mike Krinkin

Subject: [PATCH] null_blk: fix use-after-free problem

end_cmd finishes request associated with nullb_cmd struct, so we
should save pointer to request_queue in a local variable before
calling end_cmd.

The problem was causes general protection fault with slab poisoning
enabled.

Fixes: 8b70f45e2eb2 ("null_blk: restart request processing on
completion handler")
Tested-by: Akinobu Mita <[email protected]>

Signed-off-by: Mike Krinkin <[email protected]>
---
drivers/block/null_blk.c | 18 +++++++++---------
1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c
index 69de41a..3177b24 100644
--- a/drivers/block/null_blk.c
+++ b/drivers/block/null_blk.c
@@ -240,19 +240,19 @@ static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer)
while ((entry = llist_del_all(&cq->list)) != NULL) {
entry = llist_reverse_order(entry);
do {
+ struct request_queue *q = NULL;
+
cmd = container_of(entry, struct nullb_cmd, ll_list);
entry = entry->next;
+ if (cmd->rq)
+ q = cmd->rq->q;
end_cmd(cmd);

- if (cmd->rq) {
- struct request_queue *q = cmd->rq->q;
-
- if (!q->mq_ops && blk_queue_stopped(q)) {
- spin_lock(q->queue_lock);
- if (blk_queue_stopped(q))
- blk_start_queue(q);
- spin_unlock(q->queue_lock);
- }
+ if (q && !q->mq_ops && blk_queue_stopped(q)) {
+ spin_lock(q->queue_lock);
+ if (blk_queue_stopped(q))
+ blk_start_queue(q);
+ spin_unlock(q->queue_lock);
}
} while (entry);
}
--
1.9.1

2015-07-22 19:31:33

by Jens Axboe

Subject: Re: [PATCH] null_blk: fix use-after-free problem

On 07/19/2015 12:53 AM, Mike Krinkin wrote:
> end_cmd finishes request associated with nullb_cmd struct, so we
> should save pointer to request_queue in a local variable before
> calling end_cmd.
>
> The problem was causes general protection fault with slab poisoning
> enabled.

Thanks, applied.

--
Jens Axboe