No sound and alsactl init errors after updating to 1.2.13

[wastlnd]

I think I have kind of a regression after updating to the latest version. Running alsactl restore as a runit service, I have this error at startup:

Found hardware: "acp" "" "" "" ""
Hardware is initialized using a generic method

And I have no sound at all. I have tried to apply the latest commit fixing the udev rule, but it didn't work. Downgrading to 1.2.12 fixes the issue.
I am running an Arch based distro (kernel 6.6) on AMD Ryzen 7 5825U CPU with Radeon graphics. My soundcard is Realtek and I have an AMD microphone array. I don't run Pulseaudio or Pipewire, just plain ALSA.

[perexg]

Could you attach 90-alsa-restore.rules file from 1.2.12 and 1.2.13 (with updates applied - f90124c and 6f7ce73) from your distro to check ?

EDIT: Also, please test, if you replace the mentioned file with 1.2.12 version in your distro for 1.2.13 package, if things are working correctly.

[wastlnd]

Thanks, these is the 90-alsa-restore.rules from 1.2.12:

ACTION=="add", SUBSYSTEM=="sound", KERNEL=="controlC*", KERNELS!="card*", TEST=="/usr/bin", TEST=="/usr/share/alsa", GOTO="alsa_restore_go"
GOTO="alsa_restore_end"

LABEL="alsa_restore_go"
TEST!="/etc/alsa/state-daemon.conf", RUN+="/usr/bin/alsactl restore $devnode"
TEST=="/etc/alsa/state-daemon.conf", RUN+="/usr/bin/alsactl nrestore $devnode"

LABEL="alsa_restore_end"

And this is the same file from 1.2.13:

ACTION=="add", SUBSYSTEM=="sound", KERNEL=="controlC*", KERNELS!="card*", TEST=="/usr/bin", TEST=="/usr/share/alsa", GOTO="alsa_restore_go"
GOTO="alsa_restore_end"

ENV{ALSA_CARD_NUMBER}="$attr{device/number}"

# mark HDA analog card; HDMI/DP card does not have capture devices
DRIVERS=="snd_hda_intel", TEST=="device/pcmC$env{ALSA_CARD_NUMBER}D0p", RUN+="/bin/sh -c 'echo ALSA_CARD_HDA_ANALOG=$env{ALSA_CARD_NUMBER} >> /run/udev/alsa-hda-analog-card'"

# check for ACP hardware
TEST=="device/device/acp3x-dmic-capture", GOTO="alsa_hda_analog"
TEST=="device/device/acp6x-dmic-capture", GOTO="alsa_hda_analog"
TEST=="device/device/acp63-dmic-capture", GOTO="alsa_hda_analog"
TEST=="device/device/acp-pdm-dmic", GOTO="alsa_hda_analog"
GOTO="alsa_restore_std"

LABEL="alsa_hda_analog"
# restore configuration for profile with combined cards (HDA + digital mic)
TEST!="/run/udev/alsa-hda-analog-card", GOTO="alsa_restore_std"
IMPORT{program}="/usr/bin/cat /run/udev/alsa-hda-analog-card"
ENV{ALSA_CARD_HDA_ANALOG}!="", ENV{ALSA_CARD_NUMBER}="$env{ALSA_CARD_HDA_ANALOG}"

LABEL="alsa_restore_go"
TEST!="/etc/alsa/state-daemon.conf", RUN+="/usr/bin/alsactl restore $env{ALSA_CARD_NUMBER}"
TEST=="/etc/alsa/state-daemon.conf", RUN+="/usr/bin/alsactl nrestore $env{ALSA_CARD_NUMBER}"

LABEL="alsa_restore_end"

Arch Linux has not a git version, indeed it looks like f90124c and 6f7ce73 are not there yet. I just tried to manually edit the .rules file according to the second commit, it didn't work but FWIW I missed the first commit. I will try asap and let you know.

[wastlnd]

Ok, done and reporting.

1. applying the two commits to the 90-alsa-restore-rules file in1.2.13 has no effect, the error persists;
2. copying the 90-alsa-restore-rules file from 1.2.12 to 1.2.13 ends up in a weird behavior: the same error shows up but I have sound, I just can't change the volume level, if I try they immediately bounce back to the default level.

[perexg]

It's really confusing. You may check which other applications are using sound device like fuser /dev/snd/control* and try to stop them.

Also, if the mentioned information appears, UCM files are not installed or UCM code is not enabled in alsa-lib. It's not error just information.

[wastlnd]

You may check which other applications are using sound device like fuser /dev/snd/control* and try to stop them

The only process using the device is the taskbar, killing it didn't make the difference. I also tried to run alsactl restore with the --no-ucm option, no change. FWIW, the system gets messed up only if the command is run automatically at boot, running it manually on termianl makes the audio work again.

[perexg]

I think that it's a dup of alsa-project/alsa-ucm-conf#472 . UCM files must be patched, too.

[wastlnd]

Ha ok, I saw that that issues has been fixed but just for Intel devices, right? I have an AMD chipset with Realtek ACL236 card.

[perexg]

HDA config files are shared.

[perexg]

My mistake, it's for Intel's SOF driver, so it won't help.

[wastlnd]

It's mostly about the alsa service at boot, it looks like it fails and is stuck on an endless loop. Yet downgroading alsa-utils to 1.2.12 and leaving the 1.2.13 versions of alsa-lib and alsa-ucm-conf fixes the issue. Also, I wonder whether there's another way to run alsactl restore at boot.

[wastlnd]

It looks like I am at least narrowing down the issue: the problematic module is snd_apc3x_rn, it is detected but for some reason it makes the service crash. As a temporary workaround I blacklisted it and the service is working again. Weird, I think that the linux 6.6 kernel should have this supported by default...

[perexg]

You refer probably udev rule, so could you move this file to another location to inactivate it and run 'alsactl restore 0` for all cards in the system? Zero in the command is card number. Hopefully, the endless loop can be reproduced with it and if you add '-d' argument, it may help to identify what is going on.

[wastlnd]

Actually I just created a blacklist.conf file in /etc/modprobe.d with blacklist snd_apc3x_rn. How could I achieve this result in the udev rule? The related card number seems to be "2", and card 1 is the working default one.