I can confirm that this bug still affects my Debian 10 systems that have kernel
from Debian 12 (currently 6.1.0-20-amd64) and use Haswell CPU (e.g. Intel Core
i74800MQ, i74810MQ, etc.). I have "VT-d" enabled in the BIOS of these laptops
and I tested that this bug occurs when I connect external monitor via:
- HDMI port on the laptop to HDMI port on the monitor
- DP or DVI port on the laptop docking station to DP or HDMI port on the monitor
and then I ensure PULSEAUDIO outputs sound to the monitor DP or HDMI port. I
also tested that passing "intel_iommu=on,igfx_off" to the kernel command line
(in GRUB config or command line) still works as a workaround.
Also on my systems I noticed that if I set PULSEAUDIO to output sound to the
monitor DP or HDMI and then start a YOUTUBE playback the audio/video starts for
a brief moment normally and then it starts to stutter and then it quickly fails
completely. Basically it looks like the playback speed starts as normal, but
then quickly increases until it is too fast for the system to handle. To test
this I switched PULSEAUDIO (using PAVUCONTROL) to "Analog Stereo Duplex" and
then back to the monitor DP or HDMI ("Digital Stereo (HDMI*)") after each
YOUTUBE playback attempt failed.
After reading this bug and the related kernel bug
(https://bugzilla.kernel.org/show_bug.cgi?id=60769) I thing that Bryan Cebuliak
and me experienced that kernel IOMMU bug that affects systems with Haswell
CPUs. However the initial reporter Daniel Pocock and Mike Fuller could had a
slightly different issue with HDMI audio. But it is hard to tell as they did
not describe the symptoms in much detail. So we may never know unless they
report back whether they still have problems with HDMI audio and whether the
"intel_iommu=on,igfx_off" kernel parameter works as a workaround for them too.
Anyway, I would like to thank Bryan Cebuliak as the information he provided
helped me find the workaround faster.
Regards,
Jmkr
