Received-SPF: pass (google.com: domain of linux-kernel+bounces-109589-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Message-ID: <tencent_AC7946DBB367A364897F43B2045A37625705@qq.com>
From: Edward Adam Davis <eadavis@qq.com>
To: syzbot+18840ef96e57b83b7fea@syzkaller.appspotmail.com
Cc: linux-kernel@vger.kernel.org,
	linux-sound@vger.kernel.org,
	perex@perex.cz,
	syzkaller-bugs@googlegroups.com,
	tiwai@suse.com
Subject: [PATCH] ALSA: timer: fix deadlock in _snd_pcm_stream_lock_irqsave
Date: Thu, 21 Mar 2024 10:22:24 +0800
In-Reply-To: <00000000000096e5980613f4cb96@google.com>
References: <00000000000096e5980613f4cb96@google.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

[Syzbot reported]
swapper/2/0 just changed the state of lock:
ffff88802a304110 (&group->lock#2){..-.}-{2:2}, at: _snd_pcm_stream_lock_irqsave+0xa0/0xd0 sound/core/pcm_native.c:170
but this lock took another, SOFTIRQ-unsafe lock in the past:
 (&timer->lock){+.+.}-{2:2}


and interrupts could create inverse lock ordering between them.


other info that might help us debug this:
 Possible interrupt unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(&timer->lock);
                               local_irq_disable();
                               lock(&group->lock#2);
                               lock(&timer->lock);
  <Interrupt>
    lock(&group->lock#2);

 *** DEADLOCK ***
[Fix]
Ensure that the context interrupt state is the same before and after using the
timer->lock.

Fixes: beb45974dd49 ("ALSA: timer: Use guard() for locking")
Reported-and-tested-by: syzbot+18840ef96e57b83b7fea@syzkaller.appspotmail.com
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
---
 sound/core/timer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sound/core/timer.c b/sound/core/timer.c
index 15b07d09c4b7..c501faa30040 100644
--- a/sound/core/timer.c
+++ b/sound/core/timer.c
@@ -409,7 +409,7 @@ static void snd_timer_close_locked(struct snd_timer_instance *timeri,
 	struct snd_timer *timer = timeri->timer;
 
 	if (timer) {
-		guard(spinlock)(&timer->lock);
+		guard(spinlock_irqsave)(&timer->lock);
 		timeri->flags |= SNDRV_TIMER_IFLG_DEAD;
 	}
 
-- 
2.43.0