Re: [PATCH 2/2] ASoC: tlv320adcx140: Add the tlv320adcx140 codec driver family

[Dan Murphy]

Mark

Thanks for the review

On 2/10/20 7:52 AM, Mark Brown wrote:
> On Fri, Feb 07, 2020 at 01:45:33PM -0600, Dan Murphy wrote:
>
> > +	/* interface format */
> > +	switch (fmt & SND_SOC_DAIFMT_FORMAT_MASK) {
> > +	case SND_SOC_DAIFMT_I2S:
> > +		iface_reg1 |= ADCX140_I2S_MODE_BIT;
> > +		break;
> > +	case SND_SOC_DAIFMT_LEFT_J:
> > +		iface_reg1 |= ADCX140_LEFT_JUST_BIT;
> > +		break;
> > +	case SND_SOC_DAIFMT_DSP_A:
> > +	case SND_SOC_DAIFMT_DSP_B:
> > +		break;
> _DSP_A and _DSP_B are two different format so I'd expect the device to
> be configured differently for them, or for only one to be supported.
Ack.  I will need to adjust the TX offset for these.  I will do add a 
prepare call back and set the offset there.  These can fall through 
since TDM is default.
> > +static int adcx140_mute(struct snd_soc_dai *codec_dai, int mute)
> > +{
> > +	struct snd_soc_component *component = codec_dai->component;
> > +	int config_reg;
> > +	int mic_enable;
> > +	int i;
> > +
> > +	/* There is not a single register to mute.  Each enabled path has to be
> > +	 * muted individually.  Read which path is enabled and mute it.
> > +	 */
> > +	snd_soc_component_read(component, ADCX140_IN_CH_EN, &mic_enable);
> > +	if (!mic_enable)
> > +		return 0;
> You could also just offer this control to userspace, it's not
> *essential* to have this operation though it can help with glitching
> during stream startup.
>
> > +
> > +	for (i = 0; i < ADCX140_MAX_CHANNELS; i++) {
> > +		config_reg = ADCX140_CH8_CFG2 - (5 * i);
> > +		if (!(mic_enable & BIT(i)))
> > +			continue;
> > +
> > +		if (mute)
> > +			snd_soc_component_write(component, config_reg, 0);
> > +	}
> How does the unmute work?
This is unmuted through volume control.  I can remove this as the device 
is muted when the volume register is set to 0.  That will force the user 
to have to set a volume.
> > +	internal_reg = device_property_present(adcx140->dev,
> > +					       "ti,use-internal-areg");
> > +
> > +	if (internal_reg)
> > +		sleep_cfg_val |= ADCX140_AREG_INTERNAL;
> Does this actually need a specific property or could you support the
> regulator API and then use regulator_get_optional() to figure out if an
> external AVDD is attached?

Yes we could set internal AREG bit if no supply is given.

> > +static int adcx140_codec_probe(struct snd_soc_component *component)
> > +{
> > +	struct adcx140_priv *adcx140 = snd_soc_component_get_drvdata(component);
> > +
> > +	return adc5410_init(adcx140);
> > +}
> Does the separate init function buy us anything?

No that is an artifact of my development I can move everything to 
codec_probe

Dan