Radxa Zero 3W - sound devices mixed up after 9.20

[radio-miskovice]

After the upgrade and fixing WiFi I was surprised that Direwolf stopped working. (Direwolf is a sound modem used for packet radio applications). The reason why it was failing was that it did not find the configured sound card (which worked before upgrade, of course).

After some investigation I found this:

Before upgrade the on-board HDMI was hw:0, while the USB adapter connected to USB port was plughw:1 (Digirig Mobile, it contains USB hub chip, CM108 and CP2102).

After the upgrade it is swapped. This happened even though during my testing to revive the WiFi module (BTW, thanks MichaIng) the USB adapter was not connected...

Two questions:

1. Is there a way to configure the system that it puts on-board devices first, thus avoiding ambiguity in card and device indexes?
2. I use a special ALSA config to map the USB device to a synthetic device pasym0 which should enable, in the future, if required, that multiple programs share the same sound card. Is there a way to use e.g. USB PID & VID to identify plughw devices, which would also made the configuration "stick" to a particular USB device and avoid the described situation in the future?

Now as I read what I wrote it looks like a question to StackOverflow rather than a bug report but I will keep it here just for reference if anybody had a similar problem.

I solved the problem by modifying /etc/asound.conf. Now it looks like this:

pcm.digirig {
  type hw
  card 0
}

pcm.digirig-dmix {
  type dmix
  ipc_key 2511291
  slave {
    pcm "digirig"
  }
}

pcm.digirig-dsnoop {
  type dsnoop
  ipc_key 2511292
  slave {
    pcm "digirig"
  }
}

pcm.asym0 {
  type asym
  playback.pcm "digirig-dmix"
  capture.pcm "digirig-dsnoop"
}

pcm.pasym0 {
  type plug
  slave.pcm "asym0"
}

ctl.pasym0 {
  type hw
  card 0
}

Before the upgrade it worked with "card 1" instead of "card 0"

[MichaIng]

Thanks for your report.

I remember I saw a changelog somewhere, either in ALSA or related to Debian Trixie, maybe even Bookworm already, that USB audio devices may appear with indexes below other/onboard devices. But I cannot find it anymore.

But since you upgraded the kernel only, while this has probably become possible since a longer time, mainline Linux seems to iterate them in different order. Sadly I have no real solution to fix the card numbers in the first place. You could write an overlay to disable the HDMI audio device if not needed: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/arch/arm64/boot/dts/rockchip/rk3566-radxa-zero-3.dtsi#n203

If no audio-capable monitor is attached, the HDMI audio device usually is not present either, so either way round depending on attached hardware, those card numbers change. Theoretically udev rules could set them, but it is probably easier to use other identifiers (which you would use in the udev rules) in the ALSA config instead.

It is possible o use card names instead of numbers, isn't it? Yes, just tested. See cat /proc/asound/card0/id and use that value instead of 0 in your asound.conf to be independent of the number. Just tested with card Intel on a VM where card 0 has the ID "Intel".