Re: [syzbot] [sound?] possible deadlock in snd_pcm_period_elapsed (4)

[Takashi Iwai <tiwai@suse.de>]

On Fri, 15 Mar 2024 11:00:31 +0100,
syzbot wrote:
(snip)
> the shortest dependencies between 2nd lock and 1st lock:
>  -> (&timer->lock){+.+.}-{2:2} {
>     HARDIRQ-ON-W at:
>                       lock_acquire+0x1e4/0x530 kernel/locking/lockdep.c:5754
>                       __raw_spin_lock include/linux/spinlock_api_smp.h:133 [inline]
>                       _raw_spin_lock+0x2e/0x40 kernel/locking/spinlock.c:154
>                       spin_lock include/linux/spinlock.h:351 [inline]
>                       class_spinlock_constructor include/linux/spinlock.h:561 [inline]
>                       snd_timer_close_locked+0x53/0x8d0 sound/core/timer.c:412

Ouch, I incorrectly converted from spin_lock_irq() to
guard(spinlock).  It should have been guard(spinlock_irq), of course.

Will submit the fix patch.


thanks,

Takashi