[alsa-devel] [PATCH 4/5] arch/powerpc/boot/dts pcm030 add mpc5200-soc-audio node

[Stephen Warren]

On 09/11/2012 08:14 PM, Eric Millbrandt wrote:
> Describe the audio codec on the pcm030 baseboard.

> +++ b/arch/powerpc/boot/dts/pcm030.dts

> +	sound {
> +		compatible = "fsl,mpc5200b-soc-audio","fsl,mpc5200-soc-audio";
> +		card-name = "pcm030";
> +		audio-platform = <&audio_platform>;
> +		number-of-dais = <2>;
> +
> +		analog at 0 {

Purely from a DT perspective (i.e. I didn't look at the code that parses
this), you don't need to include the "@0" and "@1" in the node names,
because the two node names "analog" and "digital" are already unique.

However, in general I can see that you might want multiple analog DAIs.
There are a couple choices for differentiating the node names in that case:

1) If you want to use the "@0" unit address syntax in the node names to
differentiate them, each child node needs a reg property containing the
same value, and the parent node needs properties #address-cells=<1>,
#size-cells=<0>;

2) Or, since this is within an individual device binding rather than
something within a standardized bus, you can simply choose to make the
node names unique in some other way, such as "analog0", "analog1", i.e.
without the "@"; I believe the "@" syntax would be explicitly reserved
for representing a unit address as in (1).

Of course, I could be wrong about this assertion that the "@n" is
reserved for the unit address even with the privacy of an individual
binding; it'd be best to validate it by posting to the
devicetree-discuss mailing list.

> +			stream-name = "AC97 Analog";
> +			codec-name = "wm9712-codec.0";
> +			codec-dai-name = "wm9712-hifi";
> +			cpu-dai-name = "mpc5200-psc-ac97.0";
> +		};
> +
> +		digital at 1 {
> +			stream-name = "AC97 IEC958";
> +			codec-name = "wm9712-codec.0";
> +			codec-dai-name = "wm9712-aux";
> +			cpu-dai-name = "mpc5200-psc-ac97.0";
> +		};
> +	};
>  };