Sound randomly stops working until vm restarted (Fedora vm on Debian host running VirtualBox 5.1.18)

[Philip Seeger]

It happens very often that the sound stops working after a vm has been running for more than a day.

To clarify: "Sound stops working" means just that. For example, an audio file would be played in VLC and no sound would be produced (volume settings have not been changed).

There's a hint that it must have something to do with VirtualBox: After restarting the vm, it works again. It's not even necessary to restart the guest system. The vm can be suspended while VLC is running, it's then immediately resumed and as soon as it's running again, the sound is working again, the music played by VLC will come out.

This is an annoying bug because the vm has to be suspended/resumed all the time and sometimes that's not possible when certain connections should no be disrupted.

Nothing is logged in VBox.log. Host audio driver is PulseAudio, audio controller is ICH AC97 (default settings). Guest OS is Fedora 25. Host OS is Debian 8. VirtualBox 5.1.18.

[JacekP]

I am experiencing it too.

A big problem for me as I am using VirtualBox for Skype and WebEx on Linux.

Issue has started around 2018-01-01 after series of Windows 7 and Linux/VirtualBox updates.

[William Ruppel]

I am having a similar issue.

I'm running an Oracle Linux 7.6 guest on a Windows 10 1809 host using the Intel HD Audio controller.

Since I've upgraded to 6.0.4 (was using 5.2.12), all multimedia sounds stop working after some random amount of time (hours, usually). If I restart pulseaudio, sometimes (but rarely) the sound will begin working again (although some apps require a restart). Restarting the VM resolves the issue. Hence, this appears to be a regression in VirtualBox.

I've attached a recent /var/log/messages, and there are some interesting things to point out:

1) See the errors between lines 954-963 inclusive. 2) See the error at line 1885

In an attempt to workaround the following errors:

snd_hda_intel 0000:00:05.0: Invalid position buffer, using LPIB read method instead. snd_hda_intel 0000:00:05.0: IRQ timing workaround is activated for card #0. Suggest a bigger bdl_pos_adj.

I am setting the following options on the snd_hda_intel module via /etc/modprobe.d/sound.conf:

options snd_hda_intel bdl_pos_adj=32,32 position_fix=1,1

However, this has not helped. Seems that when the sound stops, it coincides with the following message:

snd_hda_intel 0000:00:05.0: IRQ timing workaround is activated for card #0. Suggest a bigger bdl_pos_adj.

Thanks.

[William Ruppel]

Kernel messages

[Socratis]

Hey William, you found a matching ticket! :)

Maybe you should reference ticket 29324472 from the Oracle BugDB that you mentioned you filed...