[PATCH v3 0/2] media: i2c: imx412: Add regulator control to imx412

[PATCH v3 0/2] media: i2c: imx412: Add regulator control to imx412

[Bryan O'Donoghue]

V3:
- int num_supplies -> unsigned int num_supplied - Saraki
- Move imx412->num_supplies to same line as devm_regulator_bulk_get - Saraki
- Return ret on regulator_bulk_enable() - Saraki
- Remember to regulator_bulk_disable() on !clk_prepare_enable() - Saraki

V2:
- Drops redundant verbage "Definition of the" from each of the newly defined
  regulators - Krzysztof Kozlowski
- Adds Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
- "description: thing described" to the same line in new additions
  following the established way of doing it in this file - bod

V1:
The imx412 has three regulators which depending on platform may need to be
individually enabled.

- dovdd
- avdd
- dvdd

Existing code for an Intel platform doesn't appear to need to enable these
rails directly. On my reference hardware, a Qualcomm Thundercomm RB5 some
of these rails do need to be enabled individually.

Add in the above named rails as optional in the YAML in the first patch
along with code to switch them on in the second patch.

Existing imx412 users should get dummy regulators populated which the
regulator_bulk_enable()/regulator_bulk_disable() code can safely ignore.

Bryan O'Donoghue (2):
  media: dt-bindings: imx412: Add regulator descriptions
  media: i2c: imx412: Add bulk regulator support

 .../bindings/media/i2c/sony,imx412.yaml       |  9 ++++++
 drivers/media/i2c/imx412.c                    | 31 +++++++++++++++++++
 2 files changed, 40 insertions(+)

-- 
2.35.1

[PATCH v3 1/2] media: dt-bindings: imx412: Add regulator descriptions

[Bryan O'Donoghue]

The imx412 like many I2C camera sensors has three voltage rails which
depending on platform may be necessary to switch power onto directly.

Add in as optional rails so as not to break anything for existing users.

Cc: Rob Herring <robh+dt@kernel.org>
Cc: Krzysztof Kozlowski <krzk+dt@kernel.org>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
---
 .../devicetree/bindings/media/i2c/sony,imx412.yaml       | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/Documentation/devicetree/bindings/media/i2c/sony,imx412.yaml b/Documentation/devicetree/bindings/media/i2c/sony,imx412.yaml
index afcf70947f7e..26d1807d0bb6 100644
--- a/Documentation/devicetree/bindings/media/i2c/sony,imx412.yaml
+++ b/Documentation/devicetree/bindings/media/i2c/sony,imx412.yaml
@@ -32,6 +32,15 @@ properties:
     description: Clock frequency 6MHz, 12MHz, 18MHz, 24MHz or 27MHz
     maxItems: 1
 
+  dovdd-supply:
+    description: Interface power supply.
+
+  avdd-supply:
+    description: Analog power supply.
+
+  dvdd-supply:
+    description: Digital power supply.
+
   reset-gpios:
     description: Reference to the GPIO connected to the XCLR pin, if any.
     maxItems: 1
-- 
2.35.1

[PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Bryan O'Donoghue]

Depending on the platform we may need to enable and disable three separate
regulators for the imx412.

- DOVDD
Digital I/O power

- AVDD
Analog power

- DVDD
Digital core power

The addition of these regulators shouldn't affect existing users using
fixed-on/firmware-controlled regulators.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
---
 drivers/media/i2c/imx412.c | 31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)

diff --git a/drivers/media/i2c/imx412.c b/drivers/media/i2c/imx412.c
index be3f6ea55559..27170e641b53 100644
--- a/drivers/media/i2c/imx412.c
+++ b/drivers/media/i2c/imx412.c
@@ -11,6 +11,7 @@
 #include <linux/i2c.h>
 #include <linux/module.h>
 #include <linux/pm_runtime.h>
+#include <linux/regulator/consumer.h>
 
 #include <media/v4l2-ctrls.h>
 #include <media/v4l2-fwnode.h>
@@ -101,6 +102,12 @@ struct imx412_mode {
 	struct imx412_reg_list reg_list;
 };
 
+static const char * const imx412_supply_names[] = {
+	"dovdd",	/* Digital I/O power */
+	"avdd",		/* Analog power */
+	"dvdd",		/* Digital core power */
+};
+
 /**
  * struct imx412 - imx412 sensor device structure
  * @dev: Pointer to generic device
@@ -128,6 +135,8 @@ struct imx412 {
 	struct media_pad pad;
 	struct gpio_desc *reset_gpio;
 	struct clk *inclk;
+	struct regulator_bulk_data supplies[ARRAY_SIZE(imx412_supply_names)];
+	unsigned int num_supplies;
 	struct v4l2_ctrl_handler ctrl_handler;
 	struct v4l2_ctrl *link_freq_ctrl;
 	struct v4l2_ctrl *pclk_ctrl;
@@ -946,6 +955,16 @@ static int imx412_parse_hw_config(struct imx412 *imx412)
 		return -EINVAL;
 	}
 
+	/* Get optional DT defined regulators */
+	imx412->num_supplies = ARRAY_SIZE(imx412_supply_names);
+	for (i = 0; i < imx412->num_supplies; i++)
+		imx412->supplies[i].supply = imx412_supply_names[i];
+
+	ret = devm_regulator_bulk_get(imx412->dev, imx412->num_supplies,
+				      imx412->supplies);
+	if (ret)
+		return ret;
+
 	ep = fwnode_graph_get_next_endpoint(fwnode, NULL);
 	if (!ep)
 		return -ENXIO;
@@ -1011,10 +1030,19 @@ static int imx412_power_on(struct device *dev)
 	struct imx412 *imx412 = to_imx412(sd);
 	int ret;
 
+	ret = regulator_bulk_enable(imx412->num_supplies,
+				    imx412->supplies);
+	if (ret < 0) {
+		dev_err(dev, "failed to enable regulators\n");
+		return ret;
+	}
+
 	gpiod_set_value_cansleep(imx412->reset_gpio, 1);
 
 	ret = clk_prepare_enable(imx412->inclk);
 	if (ret) {
+		regulator_bulk_disable(imx412->num_supplies,
+				       imx412->supplies);
 		dev_err(imx412->dev, "fail to enable inclk");
 		goto error_reset;
 	}
@@ -1044,6 +1072,9 @@ static int imx412_power_off(struct device *dev)
 
 	clk_disable_unprepare(imx412->inclk);
 
+	regulator_bulk_disable(imx412->num_supplies,
+			       imx412->supplies);
+
 	return 0;
 }
 
-- 
2.35.1

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Sakari Ailus]

Hi Brian,

Thanks for the update. A few minor matters below...

On Thu, Apr 14, 2022 at 01:45:05PM +0100, Bryan O'Donoghue wrote:
> Depending on the platform we may need to enable and disable three separate
> regulators for the imx412.
> 
> - DOVDD
> Digital I/O power
> 
> - AVDD
> Analog power
> 
> - DVDD
> Digital core power
> 
> The addition of these regulators shouldn't affect existing users using
> fixed-on/firmware-controlled regulators.
> 
> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
> ---
>  drivers/media/i2c/imx412.c | 31 +++++++++++++++++++++++++++++++
>  1 file changed, 31 insertions(+)
> 
> diff --git a/drivers/media/i2c/imx412.c b/drivers/media/i2c/imx412.c
> index be3f6ea55559..27170e641b53 100644
> --- a/drivers/media/i2c/imx412.c
> +++ b/drivers/media/i2c/imx412.c
> @@ -11,6 +11,7 @@
>  #include <linux/i2c.h>
>  #include <linux/module.h>
>  #include <linux/pm_runtime.h>
> +#include <linux/regulator/consumer.h>
>  
>  #include <media/v4l2-ctrls.h>
>  #include <media/v4l2-fwnode.h>
> @@ -101,6 +102,12 @@ struct imx412_mode {
>  	struct imx412_reg_list reg_list;
>  };
>  
> +static const char * const imx412_supply_names[] = {
> +	"dovdd",	/* Digital I/O power */
> +	"avdd",		/* Analog power */
> +	"dvdd",		/* Digital core power */
> +};
> +
>  /**
>   * struct imx412 - imx412 sensor device structure
>   * @dev: Pointer to generic device
> @@ -128,6 +135,8 @@ struct imx412 {
>  	struct media_pad pad;
>  	struct gpio_desc *reset_gpio;
>  	struct clk *inclk;
> +	struct regulator_bulk_data supplies[ARRAY_SIZE(imx412_supply_names)];
> +	unsigned int num_supplies;
>  	struct v4l2_ctrl_handler ctrl_handler;
>  	struct v4l2_ctrl *link_freq_ctrl;
>  	struct v4l2_ctrl *pclk_ctrl;
> @@ -946,6 +955,16 @@ static int imx412_parse_hw_config(struct imx412 *imx412)
>  		return -EINVAL;
>  	}
>  
> +	/* Get optional DT defined regulators */
> +	imx412->num_supplies = ARRAY_SIZE(imx412_supply_names);
> +	for (i = 0; i < imx412->num_supplies; i++)
> +		imx412->supplies[i].supply = imx412_supply_names[i];
> +
> +	ret = devm_regulator_bulk_get(imx412->dev, imx412->num_supplies,
> +				      imx412->supplies);
> +	if (ret)
> +		return ret;
> +
>  	ep = fwnode_graph_get_next_endpoint(fwnode, NULL);
>  	if (!ep)
>  		return -ENXIO;
> @@ -1011,10 +1030,19 @@ static int imx412_power_on(struct device *dev)
>  	struct imx412 *imx412 = to_imx412(sd);
>  	int ret;
>  
> +	ret = regulator_bulk_enable(imx412->num_supplies,
> +				    imx412->supplies);

Fits on the same line.

> +	if (ret < 0) {
> +		dev_err(dev, "failed to enable regulators\n");
> +		return ret;
> +	}
> +
>  	gpiod_set_value_cansleep(imx412->reset_gpio, 1);
>  
>  	ret = clk_prepare_enable(imx412->inclk);
>  	if (ret) {
> +		regulator_bulk_disable(imx412->num_supplies,
> +				       imx412->supplies);

As the function already has an error handling section using labels, this
should go there as well.

>  		dev_err(imx412->dev, "fail to enable inclk");
>  		goto error_reset;
>  	}
> @@ -1044,6 +1072,9 @@ static int imx412_power_off(struct device *dev)
>  
>  	clk_disable_unprepare(imx412->inclk);
>  
> +	regulator_bulk_disable(imx412->num_supplies,
> +			       imx412->supplies);

Fits on the same line.

> +
>  	return 0;
>  }
>  

-- 
Sakari Ailus

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Bryan O'Donoghue]

On 14/04/2022 14:00, Sakari Ailus wrote:
>>   	ret = clk_prepare_enable(imx412->inclk);
>>   	if (ret) {
>> +		regulator_bulk_disable(imx412->num_supplies,
>> +				       imx412->supplies);
> As the function already has an error handling section using labels, this
> should go there as well.
> 
Are you asking to move regulator_bulk_disable() to error_reset ?

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Sakari Ailus]

On Thu, Apr 14, 2022 at 02:44:00PM +0100, Bryan O'Donoghue wrote:
> On 14/04/2022 14:00, Sakari Ailus wrote:
> > >   	ret = clk_prepare_enable(imx412->inclk);
> > >   	if (ret) {
> > > +		regulator_bulk_disable(imx412->num_supplies,
> > > +				       imx412->supplies);
> > As the function already has an error handling section using labels, this
> > should go there as well.
> > 
> Are you asking to move regulator_bulk_disable() to error_reset ?

No. You'll need another label.

-- 
Sakari Ailus

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Bryan O'Donoghue]

On 14/04/2022 14:56, Sakari Ailus wrote:
> On Thu, Apr 14, 2022 at 02:44:00PM +0100, Bryan O'Donoghue wrote:
>> On 14/04/2022 14:00, Sakari Ailus wrote:
>>>>    	ret = clk_prepare_enable(imx412->inclk);
>>>>    	if (ret) {
>>>> +		regulator_bulk_disable(imx412->num_supplies,
>>>> +				       imx412->supplies);
>>> As the function already has an error handling section using labels, this
>>> should go there as well.
>>>
>> Are you asking to move regulator_bulk_disable() to error_reset ?
> 
> No. You'll need another label.
> 

Hmm.

I think another label is not required, have a look at V4.

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Sakari Ailus]

On Thu, Apr 14, 2022 at 03:04:10PM +0100, Bryan O'Donoghue wrote:
> On 14/04/2022 14:56, Sakari Ailus wrote:
> > On Thu, Apr 14, 2022 at 02:44:00PM +0100, Bryan O'Donoghue wrote:
> > > On 14/04/2022 14:00, Sakari Ailus wrote:
> > > > >    	ret = clk_prepare_enable(imx412->inclk);
> > > > >    	if (ret) {
> > > > > +		regulator_bulk_disable(imx412->num_supplies,
> > > > > +				       imx412->supplies);
> > > > As the function already has an error handling section using labels, this
> > > > should go there as well.
> > > > 
> > > Are you asking to move regulator_bulk_disable() to error_reset ?
> > 
> > No. You'll need another label.
> > 
> 
> Hmm.
> 
> I think another label is not required, have a look at V4.

Ah, yes, indeed. There's just a single location where this will be needed.

On another note, gpiod_set_value_cansleep() seems to enable reset in
resume and disable it in suspend. I.e. the polarity is wrong.

-- 
Sakari Ailus

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Bryan O'Donoghue]

On 14/04/2022 15:16, Sakari Ailus wrote:
> On Thu, Apr 14, 2022 at 03:04:10PM +0100, Bryan O'Donoghue wrote:
>> On 14/04/2022 14:56, Sakari Ailus wrote:
>>> On Thu, Apr 14, 2022 at 02:44:00PM +0100, Bryan O'Donoghue wrote:
>>>> On 14/04/2022 14:00, Sakari Ailus wrote:
>>>>>>     	ret = clk_prepare_enable(imx412->inclk);
>>>>>>     	if (ret) {
>>>>>> +		regulator_bulk_disable(imx412->num_supplies,
>>>>>> +				       imx412->supplies);
>>>>> As the function already has an error handling section using labels, this
>>>>> should go there as well.
>>>>>
>>>> Are you asking to move regulator_bulk_disable() to error_reset ?
>>>
>>> No. You'll need another label.
>>>
>>
>> Hmm.
>>
>> I think another label is not required, have a look at V4.
> 
> Ah, yes, indeed. There's just a single location where this will be needed.
> 
> On another note, gpiod_set_value_cansleep() seems to enable reset in
> resume and disable it in suspend. I.e. the polarity is wrong.
> 

Agreed, the polarity looks wrong - in my DTS right now I have 
ACTIVE_HIGH for the relevant GPIO.

For example if I do this

@@ -1363,7 +1363,7 @@ camera@1a {
                 compatible = "sony,imx412";
                 reg = <0x1a>;

-               reset-gpios = <&tlmm 78 GPIO_ACTIVE_HIGH>;
+               reset-gpios = <&tlmm 78 GPIO_ACTIVE_LOW>;
                 pinctrl-names = "default", "suspend";
                 pinctrl-0 = <&cam2_default>;
                 pinctrl-1 = <&cam2_suspend>;
diff --git a/drivers/media/i2c/imx412.c b/drivers/media/i2c/imx412.c
index a9cdf4694d58..1442b416f5aa 100644
--- a/drivers/media/i2c/imx412.c
+++ b/drivers/media/i2c/imx412.c
@@ -1036,7 +1036,7 @@ static int imx412_power_on(struct device *dev)
                 return ret;
         }

-       gpiod_set_value_cansleep(imx412->reset_gpio, 1);
+       gpiod_set_value_cansleep(imx412->reset_gpio, 0);

         ret = clk_prepare_enable(imx412->inclk);
         if (ret) {
@@ -1049,7 +1049,7 @@ static int imx412_power_on(struct device *dev)
         return 0;

  error_reset:
-       gpiod_set_value_cansleep(imx412->reset_gpio, 0);
+       gpiod_set_value_cansleep(imx412->reset_gpio, 1);
         regulator_bulk_disable(imx412->num_supplies, imx412->supplies);

         return ret;
@@ -1068,7 +1068,7 @@ static int imx412_power_off(struct device *dev)

         clk_disable_unprepare(imx412->inclk);

-       gpiod_set_value_cansleep(imx412->reset_gpio, 0);
+       gpiod_set_value_cansleep(imx412->reset_gpio, 1);

Seems like changing the logic would negatively affect the Intel people. 
Might have to churn ACPI to change that logic..

Easier probably to leave as is and define as ACTIVE_HIGH in DTS

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Sakari Ailus]

Hi Bryan,

On Thu, Apr 14, 2022 at 03:50:50PM +0100, Bryan O'Donoghue wrote:
> On 14/04/2022 15:16, Sakari Ailus wrote:
> > On Thu, Apr 14, 2022 at 03:04:10PM +0100, Bryan O'Donoghue wrote:
> > > On 14/04/2022 14:56, Sakari Ailus wrote:
> > > > On Thu, Apr 14, 2022 at 02:44:00PM +0100, Bryan O'Donoghue wrote:
> > > > > On 14/04/2022 14:00, Sakari Ailus wrote:
> > > > > > >     	ret = clk_prepare_enable(imx412->inclk);
> > > > > > >     	if (ret) {
> > > > > > > +		regulator_bulk_disable(imx412->num_supplies,
> > > > > > > +				       imx412->supplies);
> > > > > > As the function already has an error handling section using labels, this
> > > > > > should go there as well.
> > > > > > 
> > > > > Are you asking to move regulator_bulk_disable() to error_reset ?
> > > > 
> > > > No. You'll need another label.
> > > > 
> > > 
> > > Hmm.
> > > 
> > > I think another label is not required, have a look at V4.
> > 
> > Ah, yes, indeed. There's just a single location where this will be needed.
> > 
> > On another note, gpiod_set_value_cansleep() seems to enable reset in
> > resume and disable it in suspend. I.e. the polarity is wrong.
> > 
> 
> Agreed, the polarity looks wrong - in my DTS right now I have ACTIVE_HIGH
> for the relevant GPIO.
> 
> For example if I do this
> 
> @@ -1363,7 +1363,7 @@ camera@1a {
>                 compatible = "sony,imx412";
>                 reg = <0x1a>;
> 
> -               reset-gpios = <&tlmm 78 GPIO_ACTIVE_HIGH>;
> +               reset-gpios = <&tlmm 78 GPIO_ACTIVE_LOW>;
>                 pinctrl-names = "default", "suspend";
>                 pinctrl-0 = <&cam2_default>;
>                 pinctrl-1 = <&cam2_suspend>;
> diff --git a/drivers/media/i2c/imx412.c b/drivers/media/i2c/imx412.c
> index a9cdf4694d58..1442b416f5aa 100644
> --- a/drivers/media/i2c/imx412.c
> +++ b/drivers/media/i2c/imx412.c
> @@ -1036,7 +1036,7 @@ static int imx412_power_on(struct device *dev)
>                 return ret;
>         }
> 
> -       gpiod_set_value_cansleep(imx412->reset_gpio, 1);
> +       gpiod_set_value_cansleep(imx412->reset_gpio, 0);
> 
>         ret = clk_prepare_enable(imx412->inclk);
>         if (ret) {
> @@ -1049,7 +1049,7 @@ static int imx412_power_on(struct device *dev)
>         return 0;
> 
>  error_reset:
> -       gpiod_set_value_cansleep(imx412->reset_gpio, 0);
> +       gpiod_set_value_cansleep(imx412->reset_gpio, 1);
>         regulator_bulk_disable(imx412->num_supplies, imx412->supplies);
> 
>         return ret;
> @@ -1068,7 +1068,7 @@ static int imx412_power_off(struct device *dev)
> 
>         clk_disable_unprepare(imx412->inclk);
> 
> -       gpiod_set_value_cansleep(imx412->reset_gpio, 0);
> +       gpiod_set_value_cansleep(imx412->reset_gpio, 1);
> 
> Seems like changing the logic would negatively affect the Intel people.
> Might have to churn ACPI to change that logic..
> 
> Easier probably to leave as is and define as ACTIVE_HIGH in DTS

It's still wrong and should be fixed. It seems there are no boards in the
DT source in the kernel using this sensor. These changes seem fine to me.

I'm not really worried about ACPI: it's unlikely the GPIO is even declared
for the sensor, and instead is controlled in AML. There can of course be
bugs in ACPI tables, too...

-- 
Sakari Ailus

Re: [PATCH v3 2/2] media: i2c: imx412: Add bulk regulator support

[Bryan O'Donoghue]

On 14/04/2022 22:58, Sakari Ailus wrote:
> It's still wrong and should be fixed. It seems there are no boards in the
> DT source in the kernel using this sensor. These changes seem fine to me.
> 
> I'm not really worried about ACPI: it's unlikely the GPIO is even declared
> for the sensor, and instead is controlled in AML. There can of course be
> bugs in ACPI tables, too...

Hmm.

OK well I'll add the GPIO patch in for V5 but with a putative guard in 
for ACPI, inverting the logic.

We can always pull the guard if it turns out not be necessary.

---
bod