Re: [PATCH] ASoC: mediatek: mt8195: Re-add codec entry for ETDM1_OUT_BE dai link

[Chen-Yu Tsai <wenst@xxxxxxxxxxxx>]

On Thu, Jun 20, 2024 at 6:27 PM AngeloGioacchino Del Regno
<angelogioacchino.delregno@xxxxxxxxxxxxx> wrote:
>
> Il 20/06/24 11:25, Chen-Yu Tsai ha scritto:
> > This partially reverts commit e70b8dd26711704b1ff1f1b4eb3d048ba69e29da.
> >
> > Said commit removes the codec entry for the ETDM1_OUT_BE dai link for
> > some reason. This does not have the intended effect, as the remaining
> > DAILINK_COMP_ARRAY(COMP_EMPTY()) platform entry becomes the codec
> > entry, and the platform entry is completely gone.
> >
> > This causes in a KASAN out-of-bounds warning in mtk_soundcard_common_probe()
> > in sound/soc/mediatek/common/mtk-soundcard-driver.c:
> >
> >       for_each_card_prelinks(card, i, dai_link) {
> >               if (adsp_node && !strncmp(dai_link->name, "AFE_SOF", strlen("AFE_SOF")))
> >                       dai_link->platforms->of_node = adsp_node;
> >               else if (!dai_link->platforms->name && !dai_link->platforms->of_node)
> >                       dai_link->platforms->of_node = platform_node;
> >       }
> >
> > where the code expects the platforms array to have space for at least one entry.
> >
> > Re-add the entry so that dai_link->platforms has space.
> >
>
> Ok, but wait a minute... the commit that you're pointing at in the Fixes tag is
> a commit that fixes a problem identified in commit
>
> 13f58267cda3 ("ASoC: soc.h: don't create dummy Component via COMP_DUMMY()")
>
> to keep it short, after that one, without removing the COMP_DUMMY(), the audio
> was broken in .. some way, I don't currently remember specifically what was
> happening, but I had no sound at all.

Right. But this adds back "COMP_EMPTY()", not COMP_DUMMY().

I guess I shouldn't say this is a partial revert.

After painstakingly messing with the mixer controls, I tested all three
audio paths (speaker, headphone, DP) and they all work correctly.

I'll send a v2 updating the commit message.

ChenYu

> If the problem is not showing up anymore, backporting this commit to the kernels
> kernels affected by the issue that I solved... will break sound!
>
> So... well.. that's the "some reason".... :-)
>
> Cheers,
> Angelo
>
> > Fixes: e70b8dd26711 ("ASoC: mediatek: mt8195: Remove afe-dai component and rework codec link")
> > Signed-off-by: Chen-Yu Tsai <wenst@xxxxxxxxxxxx>
> > ---
> >   sound/soc/mediatek/mt8195/mt8195-mt6359.c | 1 +
> >   1 file changed, 1 insertion(+)
> >
> > diff --git a/sound/soc/mediatek/mt8195/mt8195-mt6359.c b/sound/soc/mediatek/mt8195/mt8195-mt6359.c
> > index ca8751190520..2832ef78eaed 100644
> > --- a/sound/soc/mediatek/mt8195/mt8195-mt6359.c
> > +++ b/sound/soc/mediatek/mt8195/mt8195-mt6359.c
> > @@ -827,6 +827,7 @@ SND_SOC_DAILINK_DEFS(ETDM2_IN_BE,
> >
> >   SND_SOC_DAILINK_DEFS(ETDM1_OUT_BE,
> >                    DAILINK_COMP_ARRAY(COMP_CPU("ETDM1_OUT")),
> > +                  DAILINK_COMP_ARRAY(COMP_EMPTY()),
> >                    DAILINK_COMP_ARRAY(COMP_EMPTY()));
> >
> >   SND_SOC_DAILINK_DEFS(ETDM2_OUT_BE,
>