ALSA: hda: Fix hang at HD-audio codec unbinding due to refcount saturation
authorTakashi Iwai <tiwai@suse.de>
Sat, 10 Sep 2022 14:25:50 +0000 (16:25 +0200)
committerTakashi Iwai <tiwai@suse.de>
Sat, 10 Sep 2022 14:26:19 +0000 (16:26 +0200)
commitead3d3c5b54f76da79c079e61bacb4279ec56965
treee92e5050c3bd274c2a837e3aebd3a8a83b4b2d8a
parent09e3e3159cd4d3c9f3a1f025cb8e635d93c67c9a
ALSA: hda: Fix hang at HD-audio codec unbinding due to refcount saturation

We fixed the potential deadlock at dynamic unbinding the HD-audio
codec at the commit 7206998f578d ("ALSA: hda: Fix potential deadlock
at codec unbinding"), but ironically, this caused another potential
deadlock.  The current code uses refcount_dec() and waits for the
pending task with wait_event for dropping the refcount to 0.  This
works fine when PCMs are assigned and actually waiting for the
refcount drop.

Meanwhile, when there was no PCM assigned, the refcount_dec() call
itself was supposed to drop to zero -- alas, it doesn't in reality;
refcount_dec() complains, spews kernel warning and it saturates
instead of dropping to 0, due to the nature of refcount_dec()
implementation.  This eventually blocks the wait_event() wakeup and
the code get stuck there.

For avoiding the problem, we call refcount_dec_and_test() and skips
the sync-wait if it already reaches to zero.

The patch does a slight code reshuffling to make sure to invoke other
disconnect calls before the sync-wait, too.

Fixes: 7206998f578d ("ALSA: hda: Fix potential deadlock at codec unbinding")
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/YxtflWQnslMHVlU7@intel.com
Link: https://lore.kernel.org/r/20220910142550.28494-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
sound/pci/hda/hda_bind.c