[PATCH 0/2] ASoC: codecs: ak4619: Add PDN pin handling

[PATCH 0/2] ASoC: codecs: ak4619: Add PDN pin handling

[Alexander Koch]

The AK4619 datasheet describes the functionality of the power down (PDN)
pin as follows:

  The AK4619 should be powered up when the PDN pin = "L". The PDN pin
  should be set "H" after all power supplies have ramped up.
  At that point, the reference voltage generator block (REFBLK) and LDO
  (AVDRV) circuit are powered up and the control registers are
  initialized. Control register settings should be no less than 10ms
  after PDN pin = "H". [1]

The current implementation of the driver does not operate the PDN pin
in any way, it seems to rely on external control of the pin.

Improve this by adding a device tree binding to enable the driver to
set the pin 'high' during probe. Make it optional to keep the behaviour
for existing device trees unchanged.

Tested on an i.MX8MM board with the AK4619 connected to GPIO1 and SSI3.


[1] AK4619 datasheet, section 2 'Power-up/down Sequence', page 39


Alexander Koch (2):
  ASoC: codecs: ak4619: Add optional PDN GPIO support
  SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

 .../devicetree/bindings/sound/asahi-kasei,ak4619.yaml     | 8 ++++++++
 sound/soc/codecs/ak4619.c                                 | 7 +++++++
 2 files changed, 15 insertions(+)

-- 
2.55.0

[PATCH 1/2] ASoC: codecs: ak4619: Add optional PDN pin support

[Alexander Koch]

Add optional device tree binding that enables operation of the 'PDN' pin
during initialization:

    codec@10 {
        compatible = "asahi-kasei,ak4619";
        reg = <0x10>;
        pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;  // <--
        #sound-dai-cells = <0>;
    };

If defined, PDN is set high during driver probe, followed by a 10 ms
delay as required before first register access [1].

If no 'pdn-gpios' property is defined, the behaviour is unchanged.

[1] AK4619 datasheet, sec. 2 'Power-up/down Sequence', page 39

Signed-off-by: Alexander Koch <mail@alexanderkoch.net>
---
 sound/soc/codecs/ak4619.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/sound/soc/codecs/ak4619.c b/sound/soc/codecs/ak4619.c
index d9c9f6b20028..f3357c1ee907 100644
--- a/sound/soc/codecs/ak4619.c
+++ b/sound/soc/codecs/ak4619.c
@@ -152,6 +152,7 @@
 struct ak4619_priv {
 	struct regmap *regmap;
 	struct snd_pcm_hw_constraint_list constraint;
+	struct gpio_desc *pdn_gpio;
 	int deemph_en;
 	unsigned int playback_rate;
 	unsigned int sysclk;
@@ -875,6 +876,12 @@ static int ak4619_i2c_probe(struct i2c_client *i2c)
 
 	i2c_set_clientdata(i2c, ak4619);
 
+	ak4619->pdn_gpio = devm_gpiod_get_optional(dev, "pdn", GPIOD_OUT_HIGH);
+	if (IS_ERR(ak4619->pdn_gpio))
+		return PTR_ERR(ak4619->pdn_gpio);
+	if (ak4619->pdn_gpio)
+		msleep(10);
+
 	ak4619->regmap = devm_regmap_init_i2c(i2c, &ak4619_regmap_cfg);
 	if (IS_ERR(ak4619->regmap)) {
 		ret = PTR_ERR(ak4619->regmap);
-- 
2.55.0

[PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[Alexander Koch]

Add documentation for optional property that enables operation of the
'PDN' pin during initialization:

    codec@10 {
        compatible = "asahi-kasei,ak4619";
        reg = <0x10>;
        pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;  // <--
        #sound-dai-cells = <0>;
    };

Signed-off-by: Alexander Koch <mail@alexanderkoch.net>
---
 .../devicetree/bindings/sound/asahi-kasei,ak4619.yaml     | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
index d412531ef9a2..3c7754351e8a 100644
--- a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
+++ b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
@@ -34,6 +34,12 @@ properties:
     $ref: audio-graph-port.yaml#
     unevaluatedProperties: false
 
+  pdn-gpios:
+    maxItems: 1
+    description:
+      GPIO controlling the AK4619 PDN input pin. Used during driver
+      probing to power up the codec.
+
 required:
   - compatible
   - reg
@@ -49,6 +55,8 @@ examples:
             compatible = "asahi-kasei,ak4619";
             reg = <0x10>;
 
+            pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;
+
             clocks = <&rcar_sound>;
             clock-names = "mclk";
 
-- 
2.55.0

Re: [PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] dt-bindings: missing gpio header for GPIO_ACTIVE_HIGH
--

commit 8c4430d8f688f6a14fffa90ba8e9b1227185a45b
Author: Alexander Koch <mail@alexanderkoch.net>

SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

This commit adds documentation for an optional pdn-gpios property that
enables operation of the PDN pin during initialization.

> diff --git a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
> index 2e81947bb8dd9..bc173811a8c46 100644
> --- a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
> +++ b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
[ ... ]
> @@ -49,6 +55,8 @@ examples:
>              compatible = "asahi-kasei,ak4619";
>              reg = <0x10>;
>  
> +            pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;

[Severity: Low]
Does this example block require a #include <dt-bindings/gpio/gpio.h> directive
before using GPIO_ACTIVE_HIGH? 

Running make dt_binding_check extracts the example into a dts file and passes
it through the C preprocessor and dtc. Without the header, the undefined literal
causes dtc to fail with a syntax error because the GPIO_ACTIVE_HIGH macro
remains unexpanded.

> +
>              clocks = <&rcar_sound>;
>              clock-names = "mclk";

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260928114019.1441752-1-mail@alexanderkoch.net?part=2

Re: [PATCH 1/2] ASoC: codecs: ak4619: Add optional PDN pin support

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
- [High] gpio: API semantic violation via incorrect PDN logical polarity
- [High] sound: missing 10ms hardware wake delay in resume path
- [Medium] sound: resource leak due to missing physical power-down
--

commit 8cb28b5bd3bb703c3fa3ae908ee58f9cf02df5c7
Author: Alexander Koch <mail@alexanderkoch.net>

ASoC: codecs: ak4619: Add optional PDN pin support

This commit adds optional device tree binding to support the PDN pin on the
ak4619 codec. If defined, it asserts the PDN pin during driver probe and
waits 10ms before the first register access.

> diff --git a/sound/soc/codecs/ak4619.c b/sound/soc/codecs/ak4619.c
> index 5252b30669d48..776199489db79 100644
> --- a/sound/soc/codecs/ak4619.c
> +++ b/sound/soc/codecs/ak4619.c
> @@ -894,6 +895,12 @@ static int ak4619_i2c_probe(struct i2c_client *i2c)
>  
>  	i2c_set_clientdata(i2c, ak4619);
>  
> +	ak4619->pdn_gpio = devm_gpiod_get_optional(dev, "pdn", GPIOD_OUT_HIGH);

[Severity: High]
Does this code use the incorrect logical polarity for the power-down pin?

If a device tree correctly defines the PDN (power down, active-low) pin with
GPIO_ACTIVE_LOW, requesting GPIOD_OUT_HIGH (logical assert) will translate
to a physical LOW state. This permanently powers down the codec instead of
waking it up, causing probe to fail.

To work around this, the commit message example forces DT authors to
incorrectly define the active-low pin as GPIO_ACTIVE_HIGH.

Should this use GPIOD_OUT_LOW to logically deassert the power-down state,
which correctly yields a physical HIGH when paired with an ACTIVE_LOW DT
definition?

> +	if (IS_ERR(ak4619->pdn_gpio))
> +		return PTR_ERR(ak4619->pdn_gpio);
> +	if (ak4619->pdn_gpio)
> +		msleep(10);
> +

[Severity: High]
Does the addition of this PDN pin and 10ms hardware wake delay also require
an update to the resume path?

If power is cut or the GPIO controller loses context during system suspend,
the codec resets and loses state. Upon resume, the GPIO controller restores
the physical PDN pin to HIGH.

Shortly after, the ASoC core invokes ak4619_resume():

sound/soc/codecs/ak4619.c:ak4619_resume() {
    ...
	regcache_cache_only(regmap, false);
	return regcache_sync(regmap);
}

Because ak4619_resume() immediately executes regcache_sync() to burst-write
restored registers over I2C, it violates the mandatory 10ms hardware wake
delay specified by the datasheet.

Could this cause the codec to fail initialization on resume, leading to
NACKed I2C transfers and permanently silent audio?

>  	ak4619->regmap = devm_regmap_init_i2c(i2c, &ak4619_regmap_cfg);
>  	if (IS_ERR(ak4619->regmap)) {
>  		ret = PTR_ERR(ak4619->regmap);
>  		dev_err(dev, "regmap_init() failed: %d\n", ret);
>  		return ret;
>  	}

[Severity: Medium]
Does this error path leak power by failing to physically power down the codec?

If devm_regmap_init_i2c() or devm_snd_soc_register_component() fails during
probe, the devm framework calls gpiod_put() to release the GPIO. By design,
gpiolib frees the descriptor but intentionally leaves the physical pin in its
currently driven HIGH state, keeping the codec powered up.

Should devm_add_action_or_reset() be used to register a custom cleanup action
that calls gpiod_set_value_cansleep() to properly power down the device?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260928114019.1441752-1-mail@alexanderkoch.net?part=1

Re: [PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[Rob Herring (Arm)]

On Mon, 28 Sep 2026 13:40:19 +0200, Alexander Koch wrote:
> Add documentation for optional property that enables operation of the
> 'PDN' pin during initialization:
> 
>     codec@10 {
>         compatible = "asahi-kasei,ak4619";
>         reg = <0x10>;
>         pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;  // <--
>         #sound-dai-cells = <0>;
>     };
> 
> Signed-off-by: Alexander Koch <mail@alexanderkoch.net>
> ---
>  .../devicetree/bindings/sound/asahi-kasei,ak4619.yaml     | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 

My bot found errors running 'make dt_binding_check' on your patch:

yamllint warnings/errors:

dtschema/dtc warnings/errors:
Lexical error: Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.example.dts:25.39-55 Unexpected 'GPIO_ACTIVE_HIGH'
FATAL ERROR: Syntax error parsing input tree
make[2]: *** [scripts/Makefile.dtbs:140: Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.example.dtb] Error 1
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [/builds/robherring/linux-dt-review/Makefile:1714: dt_binding_check] Error 2
make: *** [Makefile:248: __sub-make] Error 2

doc reference errors (make refcheckdocs):

See https://patchwork.kernel.org/project/devicetree/patch/20260928114019.1441752-3-mail@alexanderkoch.net

The base for the series is generally the latest rc1. A different dependency
should be noted in *this* patch.

If you already ran 'make dt_binding_check' and didn't see the above
error(s), then make sure 'yamllint' is installed and dt-schema is up to
date:

pip3 install dtschema --upgrade

Please check and re-submit after running the above command yourself. Note
that DT_SCHEMA_FILES can be set to your schema file to speed up checking
your schema. However, it must be unset to test all examples with your schema.

Re: [PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[Rob Herring]

On Mon, Sep 28, 2026 at 01:40:19PM +0200, Alexander Koch wrote:
> Add documentation for optional property that enables operation of the
> 'PDN' pin during initialization:
> 
>     codec@10 {
>         compatible = "asahi-kasei,ak4619";
>         reg = <0x10>;
>         pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;  // <--
>         #sound-dai-cells = <0>;
>     };
> 
> Signed-off-by: Alexander Koch <mail@alexanderkoch.net>
> ---
>  .../devicetree/bindings/sound/asahi-kasei,ak4619.yaml     | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
> index d412531ef9a2..3c7754351e8a 100644
> --- a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
> +++ b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
> @@ -34,6 +34,12 @@ properties:
>      $ref: audio-graph-port.yaml#
>      unevaluatedProperties: false
>  
> +  pdn-gpios:

powerdown-gpios is a standard name for this function.

> +    maxItems: 1
> +    description:
> +      GPIO controlling the AK4619 PDN input pin. Used during driver
> +      probing to power up the codec.

When a driver does something is not relevant. Would be useful to state 
the active (meaning powered down state based on the naming) state.

Rob

Re: [PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[Rob Herring]

On Mon, Sep 28, 2026 at 01:40:19PM +0200, Alexander Koch wrote:
> Add documentation for optional property that enables operation of the
> 'PDN' pin during initialization:

Also, s/Soc/ASoC/ in the subject.

Rob

Re: [PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[Alexander Koch]

>> Add documentation for optional property that enables operation of the
>> 'PDN' pin during initialization:
>>
>>      codec@10 {
>>          compatible = "asahi-kasei,ak4619";
>>          reg = <0x10>;
>>          pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;  // <--
>>          #sound-dai-cells = <0>;
>>      };
>>
>> Signed-off-by: Alexander Koch <mail@alexanderkoch.net>
>> ---
>>   .../devicetree/bindings/sound/asahi-kasei,ak4619.yaml     | 8 ++++++++
>>   1 file changed, 8 insertions(+)
>>
> My bot found errors running 'make dt_binding_check' on your patch:
>
> yamllint warnings/errors:
>
> dtschema/dtc warnings/errors:
> Lexical error: Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.example.dts:25.39-55 Unexpected 'GPIO_ACTIVE_HIGH'
> FATAL ERROR: Syntax error parsing input tree
> make[2]: *** [scripts/Makefile.dtbs:140: Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.example.dtb] Error 1
> make[2]: *** Waiting for unfinished jobs....
> make[1]: *** [/builds/robherring/linux-dt-review/Makefile:1714: dt_binding_check] Error 2
> make: *** [Makefile:248: __sub-make] Error 2
>
> doc reference errors (make refcheckdocs):
>
> See https://patchwork.kernel.org/project/devicetree/patch/20260928114019.1441752-3-mail@alexanderkoch.net
>
> The base for the series is generally the latest rc1. A different dependency
> should be noted in *this* patch.
>
> If you already ran 'make dt_binding_check' and didn't see the above
> error(s), then make sure 'yamllint' is installed and dt-schema is up to
> date:
>
> pip3 install dtschema --upgrade
>
> Please check and re-submit after running the above command yourself. Note
> that DT_SCHEMA_FILES can be set to your schema file to speed up checking
> your schema. However, it must be unset to test all examples with your schema.

Thanks for the review and the clarifications!

I must admit I am a bit new to the concept of YAML based DT bindings 
documentation and did not realize that examples must be syntactically 
complete.

So for a fixed v2 would you prefer

a) the missing #include added above the i2c block or
b) the new property removed from the example since it is optional anyway?


Best regards,

Alex

Re: [PATCH 2/2] SoC: dt-bindings: asahi-kasei,ak4619: Add PDN GPIO support

[Alexander Koch]

>> Add documentation for optional property that enables operation of the
>> 'PDN' pin during initialization:
>>
>>      codec@10 {
>>          compatible = "asahi-kasei,ak4619";
>>          reg = <0x10>;
>>          pdn-gpios = <&gpio1 0 GPIO_ACTIVE_HIGH>;  // <--
>>          #sound-dai-cells = <0>;
>>      };
>>
>> Signed-off-by: Alexander Koch <mail@alexanderkoch.net>
>> ---
>>   .../devicetree/bindings/sound/asahi-kasei,ak4619.yaml     | 8 ++++++++
>>   1 file changed, 8 insertions(+)
>>
>> diff --git a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
>> index d412531ef9a2..3c7754351e8a 100644
>> --- a/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
>> +++ b/Documentation/devicetree/bindings/sound/asahi-kasei,ak4619.yaml
>> @@ -34,6 +34,12 @@ properties:
>>       $ref: audio-graph-port.yaml#
>>       unevaluatedProperties: false
>>   
>> +  pdn-gpios:
> powerdown-gpios is a standard name for this function.

Thanks for the hint, this might actually inverse the logical polarity of 
the GPIO that I had in mind. I will look at other examples and consider 
them when submitting v2.

>> +    maxItems: 1
>> +    description:
>> +      GPIO controlling the AK4619 PDN input pin. Used during driver
>> +      probing to power up the codec.
> When a driver does something is not relevant. Would be useful to state
> the active (meaning powered down state based on the naming) state.

Understood, I will make that more clear in v2.


Best regards,

Alex