[RFC PATCH 1/1] ASoC: soc-core: symmetry checking for each DAIs separately

[Dong Aisheng-B29396]

> -----Original Message-----
> From: Lars-Peter Clausen [mailto:lars at metafoo.de]
> Sent: Friday, August 26, 2011 7:24 PM
> To: Dong Aisheng-B29396
> Cc: alsa-devel at alsa-project.org; linux-arm-kernel at lists.infradead.org;
> broonie at opensource.wolfsonmicro.com; lrg at ti.com; s.hauer at pengutronix.de;
> w.sang at pengutronix.de
> Subject: Re: [RFC PATCH 1/1] ASoC: soc-core: symmetry checking for each
> DAIs separately
> 
> On 08/26/2011 11:35 AM, Dong Aisheng wrote:
> > [...]
> >  	/* runtime devices */
> > diff --git a/sound/soc/soc-pcm.c b/sound/soc/soc-pcm.c index
> > 1aee9fc..3f7ded7 100644
> > --- a/sound/soc/soc-pcm.c
> > +++ b/sound/soc/soc-pcm.c
> > @@ -32,33 +32,54 @@ static int soc_pcm_apply_symmetry(struct
> snd_pcm_substream *substream)
> >  	struct snd_soc_pcm_runtime *rtd = substream->private_data;
> >  	struct snd_soc_dai *cpu_dai = rtd->cpu_dai;
> >  	struct snd_soc_dai *codec_dai = rtd->codec_dai;
> > +	unsigned int race;
> > +	unsigned int force_rate;
> >  	int ret;
> >
> > +	race = 0;
> > +	force_rate = 0;
> > +
> >  	if (!codec_dai->driver->symmetric_rates &&
> >  	    !cpu_dai->driver->symmetric_rates &&
> >  	    !rtd->dai_link->symmetric_rates)
> >  		return 0;
> >
> > +	if (codec_dai->active && codec_dai->driver->symmetric_rates ||
> > +	     codec_dai->active && rtd->dai_link->symmetric_rates) {
> 
> parenthesis, please, when mixing && and || in the same expression. Makes
> it easier to comprehend and protects against accidental mistakes.
Thanks for reminder, I will take it.

> > +		if (codec_dai->rate != 0)
> > +			force_rate = codec_dai->rate;
> > +		else
> > +			race = 1;
> > +	}
> > +
> > +	if (cpu_dai->active && cpu_dai->driver->symmetric_rates ||
> > +	    codec_dai->active && rtd->dai_link->symmetric_rates) {
> > +		if (cpu_dai->rate != 0)
> > +			force_rate = cpu_dai->rate;
> > +		else
> > +			race = 1;
> > +	}
> > +
> 
> If both dais are active and require symmetry we should call
> snd_pcm_hw_constraint_minmax for both rates. This will ensure that if
> both are already active and are running at different rates that there
> will be no valid rate for the new pcm stream. Maybe extend this function
> to take the dai as an parameter and call it twice, once for the codec_dai
> and once for the cpu_dai.
> This would allow to keep the current structure of the function.
I was doing like the way as you said before, however, I found the question
is that do we have to call snd_pcm_hw_constraint_minmax for the same substream
two times?

I just thought they should be running at the same rate if both are active.
Can you help point out in which case they may be different?
 
> > +	if (force_rate) {
> > +		dev_dbg(&rtd->dev, "Symmetry forces %dHz rate\n", rtd->rate);
> > +
> > +		ret = snd_pcm_hw_constraint_minmax(substream->runtime,
> > +				SNDRV_PCM_HW_PARAM_RATE,
> > +				force_rate, force_rate);
> > +		if (ret < 0) {
> > +			dev_err(&rtd->dev,
> > +				"Unable to apply rate symmetry constraint: %d\n",
> ret);
> > +			return ret;
> > +		}
> > +	}
> > +
> >  	/* This can happen if multiple streams are starting simultaneously
> -
> >  	 * the second can need to get its constraints before the first has
> >  	 * picked a rate.  Complain and allow the application to carry on.
> >  	 */
> > -	if (!rtd->rate) {
> > +	if (race)
> >  		dev_warn(&rtd->dev,
> > -			 "Not enforcing symmetric_rates due to race\n");
> > -		return 0;
> > -	}
> > -
> > -	dev_dbg(&rtd->dev, "Symmetry forces %dHz rate\n", rtd->rate);
> > -
> > -	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
> > -					   SNDRV_PCM_HW_PARAM_RATE,
> > -					   rtd->rate, rtd->rate);
> > -	if (ret < 0) {
> > -		dev_err(&rtd->dev,
> > -			"Unable to apply rate symmetry constraint: %d\n", ret);
> > -		return ret;
> > -	}
> > +			"Not enforcing symmetric_rates due to race\n");
> >
> >  	return 0;
> >  }
> > @@ -287,9 +308,14 @@ static int soc_pcm_close(struct snd_pcm_substream
> *substream)
> >  	cpu_dai->active--;
> >  	codec_dai->active--;
> >  	codec->active--;
> > +	rtd->active--;
> 
> I guess the line above is a leftover from the previous patch.
Yes, sorry for the mistake.

> > +
> > +	/* clear the corresponding DAIs rate when inactive */
> > +	if (!cpu_dai->active)
> > +		cpu_dai->rate = 0;
> >
> > -	if (!cpu_dai->active && !codec_dai->active)
> > -		rtd->rate = 0;
> > +	if (!codec_dai->active)
> > +		codec_dai->rate = 0;
> >
> >  	/* Muting the DAC suppresses artifacts caused during digital
> >  	 * shutdown, for example from stopping clocks.
> > @@ -447,7 +473,9 @@ static int soc_pcm_hw_params(struct
> snd_pcm_substream *substream,
> >  		}
> >  	}
> >
> > -	rtd->rate = params_rate(params);
> > +	/* store the rate for each DAIs */
> > +	cpu_dai->rate = params_rate(params);
> > +	codec_dai->rate = params_rate(params);
> >
> >  out:
> >  	mutex_unlock(&rtd->pcm_mutex);
>