Re: reply: [syzbot] [block?] BUG: corrupted list in loop_process_work

[Ming Lei <ming.lei@redhat.com>]

Hello Zhaoyang,

On Tue, Feb 18, 2025 at 10:49:04AM +0800, Zhaoyang Huang wrote:
> > Hello,
> >
> > syzbot found the following issue on:
> >
> > HEAD commit:    c674aa7c289e Add linux-next specific files for 20250212
> > git tree:       linux-next
> > console+strace: https://syzkaller.appspot.com/x/log.txt?x=125063f8580000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=a0fd539126ae5541
> > dashboard link: https://syzkaller.appspot.com/bug?extid=c104904eeb2c0edbdb06
> > compiler:       Debian clang version 15.0.6, GNU ld (GNU Binutils for Debian) 2.40
> > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=158a3bdf980000
> > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=17e18aa4580000
>  #syz test
> 
> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> index 68c943a77e41..354d77f9228b 100644
> --- a/drivers/block/loop.c
> +++ b/drivers/block/loop.c
> @@ -1972,7 +1972,8 @@ static void loop_process_work(struct loop_worker *worker,
>          * *and* the worker will not run again which ensures that it
>          * is safe to free any worker on the idle list
>          */
> -       if (worker && !work_pending(&worker->work)) {
> +       if (worker && !work_pending(&worker->work)
> +               && list_empty(&worker->idle_list)) {
>                 worker->last_ran_at = jiffies;
>                 list_add_tail(&worker->idle_list, &lo->idle_worker_list);
>                 loop_set_timer(lo);

The `work` to be queued may originate from RB tree or lo->rootcg_work, so it may
be freed during queuing without the lock.

I think you may need to revert the patch.


Thanks,
Ming