On Mon, 07 Mar 2022 09:05:20 +0100, Hillf Danton wrote:
Walk around the deadlock by trying to lock tasklist_lock for write on timer irq and scheduling workqueue work if any lock owner detected.
Oh no, that's toooo ugly.
And the problem isn't only here; take a look at commits f671a691e299 and 2f488f698fda. There are other users of kill_fasync() with the hard-IRQ disabled, too.
So, IMO, the handling of tasklist_lock around kill_fasync() looks broken and the fix should be needed there (or other core part), instead of messing round each caller's code.
thanks,
Takashi
Only for thoughts now.
Hillf
#syz test: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/ 38f80f42147f
--- x/sound/core/timer.c +++ y/sound/core/timer.c @@ -916,7 +916,14 @@ void snd_timer_interrupt(struct snd_time }
/* now process all fast callbacks */
- snd_timer_process_callbacks(timer, &timer->ack_list_head);
if (write_trylock(&tasklist_lock)) {
write_unlock(&tasklist_lock);snd_timer_process_callbacks(timer, &timer->ack_list_head);} else {
/* go the slow path to avoid deadlock by calling kill_fasync() */list_splice_init(&timer->ack_list_head,&timer->sack_list_head);}
/* do we have any slow callbacks? */ use_work = !list_empty(&timer->sack_list_head);
--