mbox series

[0/4] ASoC: codecs: add WSA883x support

Message ID 20220629090644.67982-1-srinivas.kandagatla@linaro.org
Headers show
Series ASoC: codecs: add WSA883x support | expand

Message

Srinivas Kandagatla June 29, 2022, 9:06 a.m. UTC
This patchset adds support for WSA883x smart speaker amplifier codec
connected via SoundWire. This codec also has a temperature sensor used
for speaker protection, support for this is not added yet.

Most of the code is derived from Qualcomm downstream msm-5.10 kernel.
Thanks to Patrick Lai's Team.

This codec is tested on SM8450 MTP.

Thanks,
Srini

Srinivas Kandagatla (4):
  ASoC: dt-bindings: Add WSA883x bindings
  ASoC: codecs: add wsa883x amplifier support
  ASoC: codecs: wsa883x: add control, dapm widgets and map
  MAINTAINERS: add ASoC Qualcomm codecs

 .../bindings/sound/qcom,wsa883x.yaml          |   74 +
 MAINTAINERS                                   |    4 +
 sound/soc/codecs/Kconfig                      |   10 +
 sound/soc/codecs/Makefile                     |    2 +
 sound/soc/codecs/wsa883x.c                    | 1498 +++++++++++++++++
 5 files changed, 1588 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/sound/qcom,wsa883x.yaml
 create mode 100644 sound/soc/codecs/wsa883x.c

Comments

Mark Brown June 29, 2022, 10:58 a.m. UTC | #1
On Wed, Jun 29, 2022 at 10:06:43AM +0100, Srinivas Kandagatla wrote:

> +static int wsa_dev_mode_put(struct snd_kcontrol *kcontrol,
> +			    struct snd_ctl_elem_value *ucontrol)
> +{
> +	struct snd_soc_component *component = snd_soc_kcontrol_component(kcontrol);
> +	struct wsa883x_priv *wsa883x = snd_soc_component_get_drvdata(component);
> +
> +	if (wsa883x->dev_mode == ucontrol->value.enumerated.item[0])
> +		return 0;
> +
> +	wsa883x->dev_mode = ucontrol->value.enumerated.item[0];
> +
> +	return 0;
> +}

This isn't returning 1 when the value changes so will miss generating
events, please test the driver with mixer-test.

> +	switch (event) {
> +	case SND_SOC_DAPM_POST_PMU:
> +		if (wsa883x->dev_mode == RECEIVER) {
> +			snd_soc_component_write_field(component, WSA883X_CDC_PATH_MODE,
> +						      WSA883X_RXD_MODE_MASK,
> +						      WSA883X_RXD_MODE_HIFI);
> +			snd_soc_component_write_field(component, WSA883X_SPKR_PWM_CLK_CTL,
> +						      WSA883X_SPKR_PWM_FREQ_SEL_MASK,
> +						      WSA883X_SPKR_PWM_FREQ_F600KHZ);
> +			snd_soc_component_write_field(component, WSA883X_DRE_CTL_0,
> +						       WSA883X_DRE_PROG_DELAY_MASK, 0x0);
> +		} else if (wsa883x->dev_mode == SPEAKER) {

This looks like it'd be better written as a switch statement.

> +static const struct snd_kcontrol_new wsa883x_snd_controls[] = {
> +	SOC_SINGLE_RANGE_TLV("PA Gain", WSA883X_DRE_CTL_1, 1,
> +			     0x0, 0x1f, 1, pa_gain),

Volume controls should end in Volume, mixer-test should also have caught
this.
Pierre-Louis Bossart June 29, 2022, 3:03 p.m. UTC | #2
> +/* 4 ports */
> +static struct sdw_dpn_prop wsa_sink_dpn_prop[WSA883X_MAX_SWR_PORTS] = {
> +	{
> +		/* DAC */
> +		.num = 1,
> +		.type = SDW_DPN_SIMPLE,
> +		.min_ch = 1,
> +		.max_ch = 1,
> +		.simple_ch_prep_sm = true,
> +		.read_only_wordlength = true,
> +	}, {

nit-pick: it's unusual to see such opening brackets for structure
initialization, usually there are on a new line.

> +		/* COMP */
> +		.num = 2,
> +		.type = SDW_DPN_SIMPLE,
> +		.min_ch = 1,
> +		.max_ch = 1,
> +		.simple_ch_prep_sm = true,
> +		.read_only_wordlength = true,
> +	}, {
> +		/* BOOST */
> +		.num = 3,
> +		.type = SDW_DPN_SIMPLE,
> +		.min_ch = 1,
> +		.max_ch = 1,
> +		.simple_ch_prep_sm = true,
> +		.read_only_wordlength = true,
> +	}, {
> +		/* VISENSE */
> +		.num = 4,
> +		.type = SDW_DPN_SIMPLE,
> +		.min_ch = 1,
> +		.max_ch = 1,
> +		.simple_ch_prep_sm = true,
> +		.read_only_wordlength = true,
> +	}
> +};

> +static int wsa883x_update_status(struct sdw_slave *slave,
> +				 enum sdw_slave_status status)
> +{
> +	struct wsa883x_priv *wsa883x = dev_get_drvdata(&slave->dev);
> +
> +	if (status == SDW_SLAVE_ATTACHED && slave->dev_num > 0)

do you actually need to test if slave->dev_num is > 0?

if I look at drivers/soundwire/bus.c, update_status cannot really be
invoked with dev_num == 0.

> +		wsa883x_init(wsa883x);
> +
> +	return 0;
> +}
> +

> +static int __maybe_unused wsa883x_runtime_resume(struct device *dev)
> +{
> +	struct sdw_slave *slave = dev_to_sdw_dev(dev);
> +	struct regmap *regmap = dev_get_regmap(dev, NULL);
> +	struct wsa883x_priv *wsa883x = dev_get_drvdata(dev);
> +	int ret;
> +
> +	ret = regulator_enable(wsa883x->vdd);
> +	if (ret) {
> +		dev_err(dev, "Failed to enable vdd regulator (%d)\n", ret);
> +		return ret;
> +	}
> +
> +	gpiod_direction_output(wsa883x->sd_n, 1);
> +
> +	wait_for_completion_timeout(&slave->initialization_complete,
> +				    msecs_to_jiffies(WSA883X_PROBE_TIMEOUT));

check for success? You don't want to enable regmap sync below if the
device never successfully attached and initialized.

> +
> +	usleep_range(20000, 20010);
> +	regcache_cache_only(regmap, false);
> +	regcache_sync(regmap);
> +
> +	return 0;
> +}
Pierre-Louis Bossart June 29, 2022, 4:42 p.m. UTC | #3
>>> +static int wsa883x_update_status(struct sdw_slave *slave,
>>> +                 enum sdw_slave_status status)
>>> +{
>>> +    struct wsa883x_priv *wsa883x = dev_get_drvdata(&slave->dev);
>>> +
>>> +    if (status == SDW_SLAVE_ATTACHED && slave->dev_num > 0)
>>
>> do you actually need to test if slave->dev_num is > 0?
>>
> Few years back I think it was you who asked me to add this check.. :-)
> 
> https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg2073074.html

Oops!

My comment was valid in general but at the bus level. With the benefit
of hindsight, I don't think this comment is valid in this callback.

update_status is either called with UNATTACHED, or with ATTACHED/ALERT
after programming dev_num to a value > 0.

It's not wrong to leave the code as is, but it's likely to be an
always-true condition.