null_blk: fix use-after-free error
authorMike Krinkin <krinkin.m.u@gmail.com>
Tue, 15 Dec 2015 09:56:40 +0000 (12:56 +0300)
committerJens Axboe <axboe@fb.com>
Tue, 22 Dec 2015 17:42:48 +0000 (10:42 -0700)
blk_end_request_all may free request, so we need to save
request_queue pointer before blk_end_request_all call.

The problem was introduced in commit cf8ecc5a8455266f8d51
("null_blk: guarantee device restart in all irq modes")
and causes general protection fault with slab poisoning
enabled.

Fixes: cf8ecc5a8455266f8d51 ("null_blk: guarantee device
       restart in all irq modes")

Signed-off-by: Mike Krinkin <krinkin.m.u@gmail.com>
Reviewed-by: Ming Lei <tom.leiming@gmail.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
drivers/block/null_blk.c

index 8162475d96b55471befacf889acf8dfa28e274ea..a428e4ef71fd74ce2d569b9cfd407a7f6b0e3eb0 100644 (file)
@@ -219,6 +219,9 @@ static void end_cmd(struct nullb_cmd *cmd)
 {
        struct request_queue *q = NULL;
 
+       if (cmd->rq)
+               q = cmd->rq->q;
+
        switch (queue_mode)  {
        case NULL_Q_MQ:
                blk_mq_end_request(cmd->rq, 0);
@@ -232,9 +235,6 @@ static void end_cmd(struct nullb_cmd *cmd)
                goto free_cmd;
        }
 
-       if (cmd->rq)
-               q = cmd->rq->q;
-
        /* Restart queue if needed, as we are freeing a tag */
        if (q && !q->mq_ops && blk_queue_stopped(q)) {
                unsigned long flags;