Re: [PATCH v1] ASoC: support machine driver with TAS2781

[AngeloGioacchino Del Regno <angelogioacchino.delregno@xxxxxxxxxxxxx>]

Il 11/06/24 15:19, Mark Brown ha scritto:
> On Tue, Jun 11, 2024 at 02:38:09PM +0200, AngeloGioacchino Del Regno wrote:
> > Il 11/06/24 12:52, Mark Brown ha scritto:
> > > On Tue, Jun 11, 2024 at 10:22:27AM +0800, Rui Zhou wrote:
> > > > Support amp TAS2781 for mt8188 platform
>
> > > > +++ b/sound/soc/mediatek/Kconfig
> > > > @@ -233,6 +233,7 @@ config SND_SOC_MT8188_MT6359
> > > >    	select SND_SOC_HDMI_CODEC
> > > >    	select SND_SOC_DMIC
> > > >    	select SND_SOC_MAX98390
> > > > +	select SND_SOC_TAS2781_I2C
> > > >    	select SND_SOC_NAU8315
> > > >    	select SND_SOC_NAU8825
>
> > > Ideally this would be sorted to avoid spurious merge conficts.  Also do
> > > we really have dependencies here or should these just be being enabled
> > > by users - these drivers are all pretty generic now?
>
> > The only dependency is on the MT6359 PMIC's audio related IPs and nothing else.
>
> > I'd even remove the other select statements for the other codecs, honestly,
> > as there's no reason why MT8188 wouldn't support a codec or the other, and
> > there is no reason why that SoC needs exactly those codecs, apart from some
> > board (Chromebooks) having one, or the other.
>
> Right, that's what I meant - the drivers are now so generic that they
> probably shouldn't be selecting everything they could possibly work
> with.
>
> > Instead of adding a select statement on MT8188_MT6359, adding the TAS2781_I2C
> > driver as a module in defconfig is a better idea..... at least, IMO.
>
> Yes.
>
> > >   Not that this
> > > change in itself makes a huge difference to the existing state.
>
> > I agree it doesn't make any huge difference, but I also don't see why users
> > can't simply enable (y/m) that in the kernel configuration :-)
>
> I think this is a result of the drivers originally being very specific
> to a particular setup

Yeah, I agree.

>
> and gradually getting made more generic, if the
> driver can only be used on a specific machine then the selects make
> sense but that's no longer the case for these drivers I think.

No this is not going to be used only for a specific machine... so yeah your
thoughts are totally right.

Just as an example, MT8188 (Kompanio) and MT8390 (Genio) are the same SoC again,
like MT8195 (Kompanio) and MT8395 (Genio) - so this driver is also going to be
used for the Genio 700 EVK, other than some SBC, etc, which do anyway expose I2S
pins on some kind of header, so... :-)

Cheers!