Re: [PATCH 25/53] ASoC: switch over to use snd_soc_register_component() on omap mcbsp

[Peter Ujfalusi <peter.ujfalusi@ti.com>]

On 03/18/2013 07:25 AM, Kuninori Morimoto wrote:
> Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> ---
>  sound/soc/omap/omap-mcbsp.c |    9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/sound/soc/omap/omap-mcbsp.c b/sound/soc/omap/omap-mcbsp.c
> index 8d2defd..f51685d 100644
> --- a/sound/soc/omap/omap-mcbsp.c
> +++ b/sound/soc/omap/omap-mcbsp.c
> @@ -586,6 +586,10 @@ static struct snd_soc_dai_driver omap_mcbsp_dai = {
>  	.ops = &mcbsp_dai_ops,
>  };
>  
> +static const struct snd_soc_component_driver omap_mcbsp_component = {
> +	.name		= "omap-mcbsp",
> +};
> +
>  static int omap_mcbsp_st_info_volsw(struct snd_kcontrol *kcontrol,
>  			struct snd_ctl_elem_info *uinfo)
>  {
> @@ -793,7 +797,8 @@ static int asoc_mcbsp_probe(struct platform_device *pdev)
>  
>  	ret = omap_mcbsp_init(pdev);
>  	if (!ret)
> -		return snd_soc_register_dai(&pdev->dev, &omap_mcbsp_dai);
> +		return snd_soc_register_component(&pdev->dev, &omap_mcbsp_component,
> +						  &omap_mcbsp_dai, 1);

This certainly breaks OMAP audio without the 'ASoC:
snd_soc_register_component() uses properly snd_soc_register_dai[s]()' patch
sent within the thread. But the comment from Lars stands for the users who
registers 1 dai with snd_soc_register_dais() call.

On the other hand I'm not sure what is the long time aim with the
snd_soc_register_component().
In essence it is just a wrapper for snd_soc_register_dai/s() without any
advantage, except that we register the DAIs with a new api which does not give
clue what it is doing.

Or is the aim is to eventually merge the snd_soc_codec_driver struct into
snd_soc_component_driver struct and use the same structure for the CPU/DAI and
CODEC/DAI registration and share the same code to do it?

I'm sure this had been discussed at some conference but unfortunately I could
not attend to them lately.

All in all this is not going to break OMAP, but as of now I don't know the
reasoning why this is a good thing...

>  	return ret;
>  }
> @@ -802,7 +807,7 @@ static int asoc_mcbsp_remove(struct platform_device *pdev)
>  {
>  	struct omap_mcbsp *mcbsp = platform_get_drvdata(pdev);
>  
> -	snd_soc_unregister_dai(&pdev->dev);
> +	snd_soc_unregister_component(&pdev->dev);
>  
>  	if (mcbsp->pdata->ops && mcbsp->pdata->ops->free)
>  		mcbsp->pdata->ops->free(mcbsp->id);
> 


-- 
Péter