[alsa-devel] [PATCH] ASoC: fsl-asoc-card: add wm8960 support

[Nicolin Chen]

On Wed, Aug 12, 2015 at 12:45:10PM +0800, Zidan Wang wrote:

> > > +		/*
> > > +		 * for imx6ul we should set SAI2_MCLK_DIR to enable
> > > +		 * codec MCLK
> > > +		 */
> > > +		if (of_device_is_compatible(np, "fsl,imx6ul-evk-wm8960"))
> > > +			regmap_update_bits(priv->gpr, 4, 1 << 20, 1 << 20);

> > What's this SAI2_MCLK_DIR? Can you provide the description of it?
> SAI2_MCLK_DIR: SD1_CLK data direction control when sai2.MCLK is selected (ALT2)
> 0 SD1_CLK output driver is disabled when configured for ALT2
> 1 SD1_CLK output driver is enabled when configured for ALT2

> We are using SAI2 now, so we should enable SAI2_MCLK_DIR. If we want to
> use SAI1 and SAI3, we should also enable SAI1_MCLK_DIR and SAI3_MCLK_DIR.
> I don't consider SAI1 and SAI3 case, should i add it like below:

Yes. Just think about a general interface/configuration that can
satisfy all the MCLK cases you can imagine.

Since it's a clock dealing with SAI, I think it would be better to
put the configuration inside:

494         } else if (strstr(cpu_np->name, "sai")) {
495                 priv->cpu_priv.sysclk_id[1] = FSL_SAI_CLK_MAST1;
496                 priv->cpu_priv.sysclk_id[0] = FSL_SAI_CLK_MAST1;
497         }

And I still don't understand the meaning of this MCLK at all.
Does it indicate SAI's MCLK or an MCLK for external CODEC chips?
Will it be possible that a user needs this MCLK bit to be unset?

Can you share the document to me privately so I can give you a
suggestion in detail?

> > > -	priv->dai_link[0].codec_dai_name = codec_dev->name;
> > > +	if (strstr(codec_dev->name, "wm8960"))
> > > +		priv->dai_link[0].codec_dai_name = "wm8960-hifi";
> > > +	else
> > > +		priv->dai_link[0].codec_dai_name = codec_dev->name;

> > What happens to the codec_dev->name over here so that you need to
> > specify the name?
> In wm8960 codec driver, codec dai name is "wm8960-hifi".
> Using codec_dev->name will cauce register card failed.

When we meet a problem, the first response should be to find out
the route cause instead of fixing it with a work around directly.
If the codec_dev->name caused a card registration failure, you
need to find out why. There must be something wrong with its codec
driver or this sound card driver.

Nicolin