Endless spam in dmesg: ASoC error (-19): at snd_soc_dai_prepare() on i2s-hifi

[Diederik de Haas]

On Wed Oct 15, 2025 at 5:30 PM CEST, Diederik de Haas wrote:
> On several of my Rockchip based devices I get ASoC errors, where the one
> from the subject occurs the most, but there are others as well.
> While it should ofc be reported when an error occurs, the volume was
> bad, then really bad and now it's just absurd:
>
> root at rock5b:~# dmesg | grep -c "ASoC error (-19): at snd_soc_dai_prepare"
> 983
>
> If 'dmesg' buffer can hold 1000 lines, then I have 17 of these as well:
> hdmi-audio-codec hdmi-audio-codec.8.auto: HDMI: Unknown ELD version 0
>
> This means that after just a few minutes, 'dmesg' has now become utterly
> useless as it ONLY contains these errors/lines.
>
> Booting into older kernels didn't seem to have an effect. If it had, I
> could start to bisect it. I have no idea where it comes from.
> So now I have removed pipewire ... and everything that depends on it ...
> of which 'mpv' is the most important to me.
> Now those errors are actually gone. And dmesg actually useful again.

I installed pipewire 1.4.2 from Debian Stable (previously 1.4.9 from
Sid/Unstable) and now the count is steady at 120.

> So I guess it's triggered by (a change in) pipewire, but still the
> kernel should not be generating endless copies of the same error. Once
> or twice should be enough.

120 is still too much, but now dmesg isn't entirely useless anymore.

> Can something be done about this?

Cheers,
  Diederik