On Tue, Aug 23, 2016 at 01:33:43PM +0800, Oder Chiou wrote:
+- realtek,use-ldo2
This would more normally be done by making LDO2 a regulator via the normal regulator framework - look at how the arizona drivers handle DCVDD for an example, it provides a default supply mapping which makes things really easy for users. Right now it's not clear to me looking at the code that the driver works for systems that don't use an internal LDO as there's nothing that enables MICVDD.
+static int rt5660_probe(struct snd_soc_codec *codec) +{
- struct rt5660_priv *rt5660 = snd_soc_codec_get_drvdata(codec);
- /* Check if MCLK provided */
- rt5660->mclk = devm_clk_get(codec->dev, "mclk");
- if (PTR_ERR(rt5660->mclk) == -EPROBE_DEFER)
return -EPROBE_DEFER;
This should be in the I2C level probe so we don't have the device showing as probing there without the resources it needs being present as that makes things confusing for people.
- if (rt5660->pdata.dmic1_data_pin) {
regmap_update_bits(rt5660->regmap, RT5660_GPIO_CTRL1,RT5660_GP1_PIN_MASK, RT5660_GP1_PIN_DMIC1_SCL);if (rt5660->pdata.dmic1_data_pin == RT5660_DMIC1_DATA_GPIO2)regmap_update_bits(rt5660->regmap, RT5660_DMIC_CTRL1,RT5660_SEL_DMIC_DATA_MASK,RT5660_SEL_DMIC_DATA_GPIO2);else if (rt5660->pdata.dmic1_data_pin == RT5660_DMIC1_DATA_IN1P)regmap_update_bits(rt5660->regmap, RT5660_DMIC_CTRL1,RT5660_SEL_DMIC_DATA_MASK,RT5660_SEL_DMIC_DATA_IN1P);- }
We do a reset when we unregister at the CODEC level - that'll overwrite these settings won't it?
Otherwise this looks good.