On Sat, Mar 15, 2014 at 12:30:05PM +0100, Jean-Francois Moine wrote:
> There may be many couples of CPU/CODEC DAI links.
> The example 2 is extracted from the Cubox DT.

Oh, here's some documentation - please include the documentation
before the code, without the documentation the reader is going to have
no idea what the code is supposed to be implementing.

> +  There may be one or many couples (simple-audio-card,cpu, simple-audio-card,codec)
> +  (see example 2).

This doesn't mention how they're matched up.

> +sound {
> +	compatible = "simple-audio-card";
> +	simple-audio-card,name = "Cubox Audio";
> +
> +	simple-audio-card,cpu@0 {		/* I2S - HDMI */
> +		sound-dai = <&audio1 0>;
> +		format = "i2s";
> +	};
> +	simple-audio-card,codec@0 {
> +		sound-dai = <&tda998x 0>;
> +	};
> +
> +	simple-audio-card,cpu@1 {		/* S/PDIF - HDMI */
> +		sound-dai = <&audio1 1>;
> +	};
> +	simple-audio-card,codec@1 {
> +		sound-dai = <&tda998x 1>;
> +	};

So, this is not exactly pretty as a binding.  I would expect to see the
links explicitly represented in the DT so you see the two DAIs in each
link grouped into a container, the above isn't very easy to read and as
Jyri says this lack of clarity also causes practical problems in that
there's nowhere to place link specific parameters.

I think what I'd expect to see here is that the simple card can either
be a container with a link in it directly or a container of links.