Re: [PATCH] ASoC: core: Move all users to deferrable card binding

[Alexander Stein <alexander.stein@ew.tq-group.com>]

Hi Cezary,

Am Donnerstag, 21. Mai 2026, 16:41:04 CEST schrieb Cezary Rojewski:
> On 5/21/2026 12:11 PM, Alexander Stein wrote:
> > Am Donnerstag, 21. Mai 2026, 10:13:45 CEST schrieb Cezary Rojewski:
> 
> ...
> 
> >>>> Now, listing the clocks won't help here unfortunately. The patch works
> >>>> on the component-level, clock details do not interest it. Also, could
> >>>> you specify the scenario? Is just a simple platform boot causing the
> >>>> problem -or- do you perform some specific test steps to reproduce?
> >>>
> >>> There is no specific scenario, just booting the board causes this error
> >>> during probe.
> >> Thank you for the confirmation.
> >>
> >> I've forgotten to mention - please also attach the equivalnet but from
> >> the system running the previous kernel (one that causes no trouble).
> >> Now, would it be possible to receive the excerpts as attachments? The
> >> timestamps are also welcome - in the above snapshop, all of them are cut
> >> off.
> > 
> > I stripped the timestamps deliberately, people told me it's just noise.
> > But here we go
> 
> ...
> > [   15.360319] tlv320aic32x4 1-0018: ASoC error (-22): at snd_soc_dai_set_sysclk() on tlv320aic32x4-hifi
> > [   15.360342] fsl-asoc-card sound: failed to set sysclk in fsl_asoc_card_late_probe
> > [   15.360360] fsl-asoc-card sound: ASoC error (-22): at snd_soc_card_late_probe() on tqm-tlv320aic32
> > [   15.558546] snd_soc_core:snd_soc_register_dai: hdmi-audio-codec hdmi-audio-codec.1: ASoC: Registered DAI 'i2s-hifi'
> The framework' behavior looks just fine. I'll need some additional help 
> from your side here.
> 
> What does -22 from snd_soc_dai_set_sysclk() actually mean here? The 
> codec's ->set_sysclk looks as such:
> 
> static int aic32x4_set_dai_sysclk(struct snd_soc_dai *codec_dai,
>                                    int clk_id, unsigned int freq, int dir)
> {
>          struct snd_soc_component *component = codec_dai->component;
>          struct clk *mclk;
>          struct clk *pll;
> 
>          pll = devm_clk_get(component->dev, "pll");
>          if (IS_ERR(pll))
>                  return PTR_ERR(pll);
> 
>          mclk = clk_get_parent(pll);
> 
>          return clk_set_rate(mclk, freq);
> }
> 
> Do we even get here? If so, does -22 is caused by devm_clk_get(), 
> clk_get_parent() or clk_set_rate() ?

AFAICS the error comes from clk_set_rate(). For some reason sai3_mclk1_cg is
to be set to 24 MHz which is not possible, because the parent has 12288000 Hz.

> I'm also up for a call (e.g.: Teams call) to debug the issue. Typically 
> such problems arise due to dependency problems between the components 
> (e.g.: platform into codec works, the other way around does not) but 
> I've only spent an hour or two analyzing fsl <> tlv320aic32x4 stack and 
> analyzing dependencies usually takes more than that : )

I'm open for that as well if you see this is a better way to approach this.

Thanks and best regards
Alexander
-- 
TQ-Systems GmbH | Mühlstraße 2, Gut Delling | 82229 Seefeld, Germany
Amtsgericht München, HRB 105018
Geschäftsführer: Detlef Schneider, Rüdiger Stahl, Stefan Schneider
http://www.tq-group.com/