Re: [PATCH v6 12/16] ASoC: codecs: add MT6357 support

[Alexandre Mergnat <amergnat@baylibre.com>]

On 21/06/2024 14:44, Mark Brown wrote:
> On Wed, Jun 19, 2024 at 04:46:48PM +0200, amergnat@baylibre.com wrote:
> 
>> +	/* gain default values*/
>> +	regmap_update_bits(priv->regmap, MT6357_AUDENC_ANA_CON0, MT6357_AUDPREAMPLGAIN_MASK,
>> +			   UL_GAIN_0DB << MT6357_AUDPREAMPLGAIN_SFT);
>> +	regmap_update_bits(priv->regmap, MT6357_AUDENC_ANA_CON1, MT6357_AUDPREAMPRGAIN_MASK,
>> +			   UL_GAIN_0DB << MT6357_AUDPREAMPRGAIN_SFT);
>> +
>> +	regmap_update_bits(priv->regmap, MT6357_ZCD_CON1,
>> +			   MT6357_AUD_LOL_GAIN_MASK |
>> +			   MT6357_AUD_LOR_GAIN_MASK,
>> +			   DL_GAIN_0DB << MT6357_AUD_LOL_GAIN_SFT |
>> +			   DL_GAIN_0DB << MT6357_AUD_LOR_GAIN_SFT);
>> +
>> +	regmap_update_bits(priv->regmap, MT6357_ZCD_CON2,
>> +			   MT6357_AUD_HPL_GAIN_MASK |
>> +			   MT6357_AUD_HPR_GAIN_MASK,
>> +			   DL_GAIN_0DB << MT6357_AUD_HPL_GAIN_SFT |
>> +			   DL_GAIN_0DB << MT6357_AUD_HPR_GAIN_SFT);
>> +
>> +	regmap_update_bits(priv->regmap, MT6357_ZCD_CON3,
>> +			   MT6357_AUD_HS_GAIN_MASK, DL_GAIN_0DB);
> 
> We generally leave everything at chip defaults, why is this different?

Ok, after test, it isn't a problem to leave at chip defaults.

> 
>> +static int mt6357_codec_probe(struct snd_soc_component *codec)
>> +{
>> +	struct mt6357_priv *priv = snd_soc_component_get_drvdata(codec);
>> +
>> +	mt6357_codec_init(priv);
>> +	return 0;
>> +}
> 
> Why not just inline mt6357_codec_init() into the one user?

You're right.

> 
>> +static unsigned int mt6357_read(struct snd_soc_component *codec, unsigned int reg)
>> +{
>> +	struct mt6357_priv *priv = snd_soc_component_get_drvdata(codec);
>> +	unsigned int val;
>> +
>> +	regmap_read(priv->regmap, reg, &val);
>> +	return val;
>> +}
>> +
>> +static int mt6357_write(struct snd_soc_component *codec, unsigned int reg, unsigned int value)
>> +{
>> +	struct mt6357_priv *priv = snd_soc_component_get_drvdata(codec);
>> +
>> +	return regmap_update_bits(priv->regmap, reg, 0xffff, value);
>> +}
> 
> Why open code these, the core has standard adaptors for regmap?

You're right. snd_soc_component_write/read should be used instead

> 
>> +static const u32 micbias_values[MT6357_MICBIAS_ARRAY_SIZE] = {
>> +	1700000, 1800000, 1900000, 2000000,
>> +	2100000, 2500000, 2600000, 2700000
>> +};
> 
> Just use ARRAY_SIZE() for the size then the number can't be out of sync.
>
You're right.

-- 
Regards,
Alexandre