ALSA: hda: update the power_state during the direct-complete
authorHui Wang <hui.wang@canonical.com>
Wed, 2 Jun 2021 14:54:24 +0000 (22:54 +0800)
committerTakashi Iwai <tiwai@suse.de>
Thu, 3 Jun 2021 07:40:28 +0000 (09:40 +0200)
commitb8b90c17602689eeaa5b219d104bbc215d1225cc
treee17a2ae3a2a3de126e3e4012d3ace8049787a04e
parent9c1fe96bded935369f8340c2ac2e9e189f697d5d
ALSA: hda: update the power_state during the direct-complete

The patch_realtek.c needs to check if the power_state.event equals
PM_EVENT_SUSPEND, after using the direct-complete, the suspend() and
resume() will be skipped if the codec is already rt_suspended, in this
case, the patch_realtek.c will always get PM_EVENT_ON even the system
is really resumed from S3.

We could set power_state to PMSG_SUSPEND in the prepare(), if other
PM functions are called before complete(), those functions will
override power_state; if no other PM functions are called before
complete(), we could know the suspend() and resume() are skipped since
only S3 pm functions could be skipped by direct-complete, in this case
set power_state to PMSG_RESUME in the complete(). This could guarantee
the first time of calling hda_codec_runtime_resume() after complete()
has the correct power_state.

Fixes: 215a22ed31a1 ("ALSA: hda: Refactor codec PM to use direct-complete optimization")
Cc: <stable@vger.kernel.org>
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Link: https://lore.kernel.org/r/20210602145424.3132-1-hui.wang@canonical.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
sound/pci/hda/hda_codec.c