On Fri, Dec 10, 2021 at 09:53:39AM -0600, Pierre-Louis Bossart wrote:
> On 12/10/21 9:10 AM, Shumin Chen wrote:
> > +static int es8156_resume(struct snd_soc_component *codec)
> > +{
> > + return 0;
> es8156_set_bias_level(codec, SND_SOC_BIAS_ON);
> for symmetry with suspend?
Shouldn't be _ON since we're resuming so there should be no running
audio - _STANDBY would be more normal.
> > +static int es8156_probe(struct snd_soc_component *codec)
> > +{
> > + struct es8156_priv *es8156 = snd_soc_component_get_drvdata(codec);
> > + int ret = 0;
> > +
> > + es8156_codec = codec;
> > +
> > +#if MCLK
> > + es8156->mclk = devm_clk_get(codec->dev, "mclk");
> > + if (PTR_ERR(es8156->mclk) == -EPROBE_DEFER)
> > + return -EPROBE_DEFER;
> > + ret = clk_prepare_enable(es8156->mclk);
> > +#endif
> Unclear how MCLK will be enabled in a build, did you mean
> es8156->mclk = devm_clk_get_optional(codec->dev, "mclk");
Please also move all resource acquisition to the I2C level probe, that
means we don't get the card trying to instantiate repeatedly while the
CODEC isn't ready.
Attachment:
signature.asc
Description: PGP signature
