After upgrade kernel from 6.0.2 to 6.0.3 on Lenovo T14 Gen2i, sound stopped working. dmesg: paź 21 21:11:45 kernel: snd_hda_codec_hdmi ehdaudio0D2: failed to create hda codec -12 paź 21 21:11:45 kernel: snd_hda_codec_hdmi ehdaudio0D2: ASoC: error at snd_soc_component_probe on ehdaudio0D2: -12 paź 21 21:11:45 kernel: skl_hda_dsp_generic skl_hda_dsp_generic: ASoC: failed to instantiate card -12
Created attachment 303070 [details] trace
Could you try to revert the commit 7494e2e6c55ed192f2b91c821fd6832744ba8741?
Revert commit: 7494e2e6c55ed192f2b91c821fd6832744ba874 fixes the problem, the sound works
Reverting the mentioned commit also works in my case, I was having the same problem
related: https://github.com/thesofproject/linux/issues/3764
Adding Cezary to Cc. I don't remember exactly which more commits to be needed. Meanwhile, can people do a quick check whether the issue is present on 6.1-rc1 kernel, too?
yes, on 6.1-rc1 (9abf2313adc1ca1b6180c508c25f22f9395cc780) the sound works fine
Thanks, that's a good news. So we just miss some commits.
Hello, Related patchset on the alsa-devel: https://lore.kernel.org/alsa-devel/20220816111727.3218543-1-cezary.rojewski@intel.com/. All of the patches in the series are a dependency for the last one. Is "ALSA: hda: Fix page fault in snd_hda_codec_shutdown()" in kernel tree on its own?
Hi! (In reply to Cezary Rojewski from comment #9) > Related patchset on the alsa-devel: > https://lore.kernel.org/alsa-devel/20220816111727.3218543-1-cezary. > rojewski@intel.com/. > > All of the patches in the series are a dependency for the last one. Is > "ALSA: hda: Fix page fault in snd_hda_codec_shutdown()" in kernel tree on > its own? It seems only this patch was backported to 6.0.3: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.0.3&id=7494e2e6c55ed192f2b91c821fd6832744ba8741
IMO, that's too many changes as a stable material, so maybe it's better to drop the patch from 6.0.y...
Both solutions are fine by me. The page fault that appears during snd_hda_codec_shutdown() and is being fixed here, requires user to perform module reload scenario under certain conditions.
(finishing the above one..) So, it's not a common case by any means.
OK, I'm going to submit the revert patch for 6.0.y.
The revert is queued for the next 6.0.y stable.