search

heikomat / linux

Linux kernel source tree
Other
43 stars 8 forks source link

Sound stops after ~3 seconds when playing AoE II HD #1

Open heikomat opened 6 years ago

heikomat commented 6 years ago

When playing AoE II HD edition via playonlinux, the sound initially works, but after about 3 seconds in the main menu, it stops until the game is restarted.

Tested on Debian testing

heikomat commented 6 years ago

Disabling suspend-on-idle as described here fixes the problem :)

Will add this to the e200ha-sound-manual and probably the automated setup script

heikomat commented 6 years ago

This fix seems to cause pulseaudio to utilize a whole core all of the time. needs further investigation

ThisIsSet-L commented 6 years ago

Check the Pulse Audio log to make sure it isn't hammering the HDMI audio outputs. Mine went nuts when I applied the pulse audio config files, and I had to unload the HDMI sound card module.

Edit:I'll try to be more specific in a bit, I just woke up.

heikomat commented 6 years ago

@ThisIsSet-L thanks for the hint! it seems to indeed be the HDMI audio output. When enabling debug logs, my syslog is spammed with these lines over and over:

Jul 14 12:46:55 heiko-e200ha pulseaudio[1772]: I: [alsa-sink-HdmiLpeAudio] alsa-sink.c: Starting playback.
Jul 14 12:46:55 heiko-e200ha pulseaudio[1772]: D: [alsa-sink-HdmiLpeAudio] alsa-util.c: Got POLLERR from ALSA
Jul 14 12:46:55 heiko-e200ha pulseaudio[1772]: D: [alsa-sink-HdmiLpeAudio] alsa-util.c: Got POLLOUT from ALSA
Jul 14 12:46:55 heiko-e200ha pulseaudio[1772]: D: [alsa-sink-HdmiLpeAudio] alsa-util.c: PCM state is XRUN
heikomat commented 6 years ago

listing the sinkns with pactl list sinks and unloading the owner-module of the HDMI sink as described here did stop the syslog spamming and fixed the cpu usage, but i'm sure that now my hdmi audio is no longer working.

It's a step forward, but maybe we can find a way to make HDMI sound work without the insane cpu usage, and without the builtin audio being wrongfully suspended :)

ThisIsSet-L commented 6 years ago

This issue was seen on the patched 4.17 kernel I was using under Arch. And yes, I did the same thing to stop the same problem. It's a very narrow issue that I haven't been able to find anywhere else. Glad to know it wasn't entirely my fault.