On 2023-01-30 12:27 PM, Linux kernel regression tracking (Thorsten
Leemhuis) wrote:
On 30.01.23 07:33, Jason Montleon wrote:
I ran a bisect back further while patching in
f2bd1c5ae2cb0cf9525c9bffc0038c12dd7e1338.
Cezary Rojewski, you authored this change which appears to cause a
regression. Do you maybe have an idea what's wrong here?
Hello,
Sitting with Lukasz and debugging this.
Please note that messages seen in the logs:
[ 8.538645] snd_soc_skl 0000:00:1f.3: ASoC: no sink widget found for
AEC Capture
[ 8.538654] snd_soc_skl 0000:00:1f.3: ASoC: Failed to add route
echo_ref_out cpr 7 -> direct -> AEC Capture
and:
[ 8.614993] rt5663 i2c-10EC5663:00: sysclk < 384 x fs, disable i2s asrc
[ 8.617460] HDMI HDA Codec ehdaudio0D2: hdac_hdmi_present_sense:
disconnect for pin:port 5:0
[ 8.617581] HDMI HDA Codec ehdaudio0D2: hdac_hdmi_present_sense:
disconnect for pin:port 6:0
[ 8.617712] HDMI HDA Codec ehdaudio0D2: hdac_hdmi_present_sense:
disconnect for pin:port 7:0
seem to be just noise, these are present even in working configuration.
Looks like topology and AudioDSP firmware loads fine. However, one or
more of sound cards component is deferred when probing what results in
no error in the logs but still lack of audio.
We managed to reproduce the problem on Lukasz' Eve machine. If hardware
cooperates, we should be able to narrow down the problem quickly from
here. Our suspect - something (a patch or two) is missing in sound/soc
in Fedora tree. Our CI which also covers upstream skylake-driver with
HDAudio (e.g.: rt274/rt286/rt298) and I2S (same examples) configurations
show no signs of regression. Thus, this is probably Chromebook-specific.
My opinion stands at - none of the patches mentioned during the
discussion is the real problem. All of these were in fact addressing
real problems with audio stack stability/reloading.
Will reply once more is known.
Kind regards,
Czarek
