Lockdep splat in sound core

The following lockdep splat occurs when the EMU10K1 driver is loaded. It actually did that for a long time already, it's about time I finally report it:

=============================================
[ INFO: possible recursive locking detected ]
3.1.0-rc2-00171-gbed8cad #88
---------------------------------------------
pulseaudio/3074 is trying to acquire lock:
(&(&substream->self_group.lock)->rlock/1){......}, at: [<ffffffffa012b0c9>] snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]

but task is already holding lock:
(&(&substream->self_group.lock)->rlock/1){......}, at: [<ffffffffa012b0c9>] snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0
       ----

lock(&(&substream->self_group.lock)->rlock);
lock(&(&substream->self_group.lock)->rlock);

*** DEADLOCK ***

May be due to missing lock nesting notation

4 locks held by pulseaudio/3074:

 #0:  (snd_pcm_link_rwlock){.-.-..}, at: [<ffffffffa012a0b1>] snd_pcm_stream_lock_irq+0x15/0x25 [snd_pcm]
 #1:  (&(&substream->self_group.lock)->rlock){-.-...}, at: [<ffffffffa012a0bd>] snd_pcm_stream_lock_irq+0x21/0x25 [snd_pcm]
 #2:  (&(&substream->group->lock)->rlock){......}, at: [<ffffffffa012ca6d>] snd_pcm_action+0x2e/0x8b [snd_pcm]
 #3:  (&(&substream->self_group.lock)->rlock/1){......}, at: [<ffffffffa012b0c9>] snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]

stack backtrace:
Pid: 3074, comm: pulseaudio Not tainted 3.1.0-rc2-00171-gbed8cad #88
Call Trace:
 [<ffffffff8106b728>] print_deadlock_bug+0xca/0xd5
 [<ffffffff8106c567>] validate_chain.isra.34+0x31c/0x39d
 [<ffffffff8106cc7c>] __lock_acquire+0x36f/0x3e8
 [<ffffffffa012b0c9>] ? snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]
 [<ffffffff8106d1a9>] lock_acquire+0xc4/0x112
 [<ffffffffa012b0c9>] ? snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]
 [<ffffffff813d1494>] _raw_spin_lock_nested+0x3a/0x49
 [<ffffffffa012b0c9>] ? snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]
 [<ffffffffa012b0c9>] snd_pcm_action_group+0x4f/0x1d6 [snd_pcm]
 [<ffffffffa012caa3>] snd_pcm_action+0x64/0x8b [snd_pcm]
 [<ffffffffa012a0bd>] ? snd_pcm_stream_lock_irq+0x21/0x25 [snd_pcm]
 [<ffffffffa012cadf>] snd_pcm_stop+0x15/0x17 [snd_pcm]
 [<ffffffffa012cdce>] snd_pcm_drop+0x5e/0x6d [snd_pcm]
 [<ffffffffa012dc6e>] snd_pcm_common_ioctl1+0x53e/0x580 [snd_pcm]
 [<ffffffff8106d652>] ? trace_hardirqs_on_caller.part.36+0xbd/0xf4
 [<ffffffffa012e12d>] snd_pcm_playback_ioctl1+0x21d/0x232 [snd_pcm]
 [<ffffffff810f307d>] ? slab_free_hook.isra.64+0x50/0x59
 [<ffffffffa012e16c>] snd_pcm_playback_ioctl+0x2a/0x2e [snd_pcm]
 [<ffffffff8110bb1a>] do_vfs_ioctl+0x267/0x281
 [<ffffffff810febc5>] ? fget_light+0x3b/0x9a
 [<ffffffff8110bb85>] sys_ioctl+0x51/0x74
 [<ffffffff813d2912>] system_call_fastpath+0x16/0x1b

Andreas Bombe

[...]

I created a Bugzilla entry at
https://bugzilla.kernel.org/show_bug.cgi?id=42072
for your bug report, please add your address to the CC list in there, thanks!

Maciej Rutecki
http://www.maciek.unixy.pl