Pulseaudio dropping GPU audio

Pulseaudio dropping GPU audio

[Liam Hupfer]

Hello,

I’ve set up a Linux VM where I’m passing through an RX 580. I’m using the GPU
audio and my monitor’s audio jack. For some reason the Linux VM keeps losing
track of the audio output. It always shows up under `lspci' under ID `06:00.0'
using the `snd_hda_intel' driver, but `pulseaudio' (I also tried `pipewire',
same issue) keep losing it. Most times I can restart `pulseaudio' and it will
find the device again; occasionally, it takes multiple restarts.

I checked the Arch wiki and I found something[¹] on audio over HDMI (I’m using
DisplayPort, if that matters). I checked the device and it did report `Enable+'.
Even so, I added the kernel parameter `snd_hda_intel.enable_msi=1', but nothing
changed.

Here are some `journalctl --user -u pulseaudio' logs:
┌────
│ 20:37:31 host pulseaudio[2373]: ALSA woke us up to write new data to the device, but there was actually nothing to write.
│ 20:37:31 host pulseaudio[2373]: Most likely this is a bug in the ALSA driver 'snd_hda_intel'. Please report this issue to the ALSA developers.
│ 20:37:31 host pulseaudio[2373]: We were woken up with POLLOUT set -- however a subsequent snd_pcm_avail() returned 0 or another value < min_avail.
└────

Anyone have any suggestions for troubleshooting? `pulseaudio'’s logs suggested
it might have to do with ALSA, so I asked for help here.

Thank you!

—Liam


[¹] <https://wiki.archlinux.org/title/PCI_passthrough_via_OVMF#Slowed_down_audio_pumped_through_HDMI_on_the_video_card>

Re: Pulseaudio dropping GPU audio

[Takashi Iwai]

On Sat, 24 Jul 2021 23:50:09 +0200,
Liam Hupfer wrote:
> 
> Hello,
> 
> I’ve set up a Linux VM where I’m passing through an RX 580. I’m using the GPU
> audio and my monitor’s audio jack. For some reason the Linux VM keeps losing
> track of the audio output. It always shows up under `lspci' under ID `06:00.0'
> using the `snd_hda_intel' driver, but `pulseaudio' (I also tried `pipewire',
> same issue) keep losing it. Most times I can restart `pulseaudio' and it will
> find the device again; occasionally, it takes multiple restarts.
> 
> I checked the Arch wiki and I found something[¹] on audio over HDMI (I’m using
> DisplayPort, if that matters). I checked the device and it did report `Enable+'.
> Even so, I added the kernel parameter `snd_hda_intel.enable_msi=1', but nothing
> changed.
> 
> Here are some `journalctl --user -u pulseaudio' logs:
> ┌────
> │ 20:37:31 host pulseaudio[2373]: ALSA woke us up to write new data to the device, but there was actually nothing to write.
> │ 20:37:31 host pulseaudio[2373]: Most likely this is a bug in the ALSA driver 'snd_hda_intel'. Please report this issue to the ALSA developers.
> │ 20:37:31 host pulseaudio[2373]: We were woken up with POLLOUT set -- however a subsequent snd_pcm_avail() returned 0 or another value < min_avail.
> └────

The message is likely a red herring and irrelevant from your problem.

You can try at first the direct playback with aplay (specify the
proper option to access without pulse plugin), and confirm that the
device is set up.  If aplay works (at least detects), then check more
verbose log of pulseaudio.


Takashi

> Anyone have any suggestions for troubleshooting? `pulseaudio'’s logs suggested
> it might have to do with ALSA, so I asked for help here.
> 
> Thank you!
> 
> —Liam
> 
> 
> [¹] <https://wiki.archlinux.org/title/PCI_passthrough_via_OVMF#Slowed_down_audio_pumped_through_HDMI_on_the_video_card>
>