summaryrefslogtreecommitdiff
path: root/sound/usb
diff options
context:
space:
mode:
authorTakashi Iwai <tiwai@suse.de>2023-06-12 15:28:18 +0200
committerTakashi Iwai <tiwai@suse.de>2023-06-16 09:28:08 +0200
commit8ba61c9f6c9bdfbf9d197b0282641d24ae909778 (patch)
tree9a72e60dd5757c0b8929a64c6884017073f665a2 /sound/usb
parent297224fc0922e7385573a30c29ffdabb67f27b7d (diff)
downloadlwn-8ba61c9f6c9bdfbf9d197b0282641d24ae909778.tar.gz
lwn-8ba61c9f6c9bdfbf9d197b0282641d24ae909778.zip
ALSA: usb-audio: Fix broken resume due to UAC3 power state
As reported in the bugzilla below, the PM resume of a UAC3 device may fail due to the incomplete power state change, stuck at D1. The reason is that the driver expects the full D0 power state change only at hw_params, while the normal PCM resume procedure doesn't call hw_params. For fixing the bug, we add the same power state update to D0 at the prepare callback, which is certainly called by the resume procedure. Note that, with this change, the power state change in the hw_params becomes almost redundant, since snd_usb_hw_params() doesn't touch the parameters (at least it tires so). But dropping it is still a bit risky (e.g. we have the media-driver binding), so I leave the D0 power state change in snd_usb_hw_params() as is for now. Fixes: a0a4959eb4e9 ("ALSA: usb-audio: Operate UAC3 Power Domains in PCM callbacks") Cc: <stable@vger.kernel.org> Link: https://bugzilla.kernel.org/show_bug.cgi?id=217539 Link: https://lore.kernel.org/r/20230612132818.29486-1-tiwai@suse.de Signed-off-by: Takashi Iwai <tiwai@suse.de>
Diffstat (limited to 'sound/usb')
-rw-r--r--sound/usb/pcm.c4
1 files changed, 4 insertions, 0 deletions
diff --git a/sound/usb/pcm.c b/sound/usb/pcm.c
index eec5232f9fb2..08bf535ed163 100644
--- a/sound/usb/pcm.c
+++ b/sound/usb/pcm.c
@@ -650,6 +650,10 @@ static int snd_usb_pcm_prepare(struct snd_pcm_substream *substream)
goto unlock;
}
+ ret = snd_usb_pcm_change_state(subs, UAC3_PD_STATE_D0);
+ if (ret < 0)
+ goto unlock;
+
again:
if (subs->sync_endpoint) {
ret = snd_usb_endpoint_prepare(chip, subs->sync_endpoint);