[PATCH] media: i2c: imx415: Drop redundant runtime PM callbacks

[PATCH] media: i2c: imx415: Drop redundant runtime PM callbacks

[Elgin Perumbilly]

Replace runtime_suspend/resume wrappers by using power helpers
directly with DEFINE_RUNTIME_DEV_PM_OPS().

Signed-off-by: Elgin Perumbilly <elgin.perumbilly@siliconsignals.io>
---
 drivers/media/i2c/imx415.c | 44 +++++++++++++++-----------------------
 1 file changed, 17 insertions(+), 27 deletions(-)

diff --git a/drivers/media/i2c/imx415.c b/drivers/media/i2c/imx415.c
index 0b424c17e880..b7d44f3e165b 100644
--- a/drivers/media/i2c/imx415.c
+++ b/drivers/media/i2c/imx415.c
@@ -1129,8 +1129,12 @@ static void imx415_subdev_cleanup(struct imx415 *sensor)
 	v4l2_ctrl_handler_free(&sensor->ctrls);
 }

-static int imx415_power_on(struct imx415 *sensor)
+static int imx415_power_on(struct device *dev)
 {
+	struct i2c_client *client = to_i2c_client(dev);
+	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
+	struct imx415 *sensor = to_imx415(subdev);
+
 	int ret;

 	ret = regulator_bulk_enable(ARRAY_SIZE(sensor->supplies),
@@ -1161,11 +1165,17 @@ static int imx415_power_on(struct imx415 *sensor)
 	return ret;
 }

-static void imx415_power_off(struct imx415 *sensor)
+static int imx415_power_off(struct device *dev)
 {
+	struct i2c_client *client = to_i2c_client(dev);
+	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
+	struct imx415 *sensor = to_imx415(subdev);
+
 	clk_disable_unprepare(sensor->clk);
 	gpiod_set_value_cansleep(sensor->reset, 1);
 	regulator_bulk_disable(ARRAY_SIZE(sensor->supplies), sensor->supplies);
+
+	return 0;
 }

 static int imx415_identify_model(struct imx415 *sensor)
@@ -1371,7 +1381,7 @@ static int imx415_probe(struct i2c_client *client)
 	 * work when runtime PM is disabled in the kernel. To that end, power
 	 * the sensor on manually here, identify it, and fully initialize it.
 	 */
-	ret = imx415_power_on(sensor);
+	ret = imx415_power_on(sensor->dev);
 	if (ret)
 		return ret;

@@ -1411,7 +1421,7 @@ static int imx415_probe(struct i2c_client *client)
 	pm_runtime_put_noidle(sensor->dev);
 	imx415_subdev_cleanup(sensor);
 err_power:
-	imx415_power_off(sensor);
+	imx415_power_off(sensor->dev);
 	return ret;
 }

@@ -1430,32 +1440,12 @@ static void imx415_remove(struct i2c_client *client)
 	 */
 	pm_runtime_disable(sensor->dev);
 	if (!pm_runtime_status_suspended(sensor->dev))
-		imx415_power_off(sensor);
+		imx415_power_off(sensor->dev);
 	pm_runtime_set_suspended(sensor->dev);
 }

-static int imx415_runtime_resume(struct device *dev)
-{
-	struct i2c_client *client = to_i2c_client(dev);
-	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
-	struct imx415 *sensor = to_imx415(subdev);
-
-	return imx415_power_on(sensor);
-}
-
-static int imx415_runtime_suspend(struct device *dev)
-{
-	struct i2c_client *client = to_i2c_client(dev);
-	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
-	struct imx415 *sensor = to_imx415(subdev);
-
-	imx415_power_off(sensor);
-
-	return 0;
-}
-
-static DEFINE_RUNTIME_DEV_PM_OPS(imx415_pm_ops, imx415_runtime_suspend,
-				 imx415_runtime_resume, NULL);
+static DEFINE_RUNTIME_DEV_PM_OPS(imx415_pm_ops, imx415_power_off,
+				 imx415_power_on, NULL);

 static const struct of_device_id imx415_of_match[] = {
 	{ .compatible = "sony,imx415" },
--
2.34.1

Re: [PATCH] media: i2c: imx415: Drop redundant runtime PM callbacks

[Michael Riesch]

Hi Elgin,

Thanks for the patch but...

On 3/24/26 13:45, Elgin Perumbilly wrote:
> Replace runtime_suspend/resume wrappers by using power helpers
> directly with DEFINE_RUNTIME_DEV_PM_OPS().

...why? What advantage does this refactoring bring?

Best regards,
Michael

> 
> Signed-off-by: Elgin Perumbilly <elgin.perumbilly@siliconsignals.io>
> ---
>  drivers/media/i2c/imx415.c | 44 +++++++++++++++-----------------------
>  1 file changed, 17 insertions(+), 27 deletions(-)
> 
> diff --git a/drivers/media/i2c/imx415.c b/drivers/media/i2c/imx415.c
> index 0b424c17e880..b7d44f3e165b 100644
> --- a/drivers/media/i2c/imx415.c
> +++ b/drivers/media/i2c/imx415.c
> @@ -1129,8 +1129,12 @@ static void imx415_subdev_cleanup(struct imx415 *sensor)
>  	v4l2_ctrl_handler_free(&sensor->ctrls);
>  }
> 
> -static int imx415_power_on(struct imx415 *sensor)
> +static int imx415_power_on(struct device *dev)
>  {
> +	struct i2c_client *client = to_i2c_client(dev);
> +	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
> +	struct imx415 *sensor = to_imx415(subdev);
> +
>  	int ret;
> 
>  	ret = regulator_bulk_enable(ARRAY_SIZE(sensor->supplies),
> @@ -1161,11 +1165,17 @@ static int imx415_power_on(struct imx415 *sensor)
>  	return ret;
>  }
> 
> -static void imx415_power_off(struct imx415 *sensor)
> +static int imx415_power_off(struct device *dev)
>  {
> +	struct i2c_client *client = to_i2c_client(dev);
> +	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
> +	struct imx415 *sensor = to_imx415(subdev);
> +
>  	clk_disable_unprepare(sensor->clk);
>  	gpiod_set_value_cansleep(sensor->reset, 1);
>  	regulator_bulk_disable(ARRAY_SIZE(sensor->supplies), sensor->supplies);
> +
> +	return 0;
>  }
> 
>  static int imx415_identify_model(struct imx415 *sensor)
> @@ -1371,7 +1381,7 @@ static int imx415_probe(struct i2c_client *client)
>  	 * work when runtime PM is disabled in the kernel. To that end, power
>  	 * the sensor on manually here, identify it, and fully initialize it.
>  	 */
> -	ret = imx415_power_on(sensor);
> +	ret = imx415_power_on(sensor->dev);
>  	if (ret)
>  		return ret;
> 
> @@ -1411,7 +1421,7 @@ static int imx415_probe(struct i2c_client *client)
>  	pm_runtime_put_noidle(sensor->dev);
>  	imx415_subdev_cleanup(sensor);
>  err_power:
> -	imx415_power_off(sensor);
> +	imx415_power_off(sensor->dev);
>  	return ret;
>  }
> 
> @@ -1430,32 +1440,12 @@ static void imx415_remove(struct i2c_client *client)
>  	 */
>  	pm_runtime_disable(sensor->dev);
>  	if (!pm_runtime_status_suspended(sensor->dev))
> -		imx415_power_off(sensor);
> +		imx415_power_off(sensor->dev);
>  	pm_runtime_set_suspended(sensor->dev);
>  }
> 
> -static int imx415_runtime_resume(struct device *dev)
> -{
> -	struct i2c_client *client = to_i2c_client(dev);
> -	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
> -	struct imx415 *sensor = to_imx415(subdev);
> -
> -	return imx415_power_on(sensor);
> -}
> -
> -static int imx415_runtime_suspend(struct device *dev)
> -{
> -	struct i2c_client *client = to_i2c_client(dev);
> -	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
> -	struct imx415 *sensor = to_imx415(subdev);
> -
> -	imx415_power_off(sensor);
> -
> -	return 0;
> -}
> -
> -static DEFINE_RUNTIME_DEV_PM_OPS(imx415_pm_ops, imx415_runtime_suspend,
> -				 imx415_runtime_resume, NULL);
> +static DEFINE_RUNTIME_DEV_PM_OPS(imx415_pm_ops, imx415_power_off,
> +				 imx415_power_on, NULL);
> 
>  static const struct of_device_id imx415_of_match[] = {
>  	{ .compatible = "sony,imx415" },
> --
> 2.34.1
> 

Re: [PATCH] media: i2c: imx415: Drop redundant runtime PM callbacks

[Sakari Ailus]

Hi Michael,

On Tue, Mar 24, 2026 at 08:42:31PM +0100, Michael Riesch wrote:
> Hi Elgin,
> 
> Thanks for the patch but...
> 
> On 3/24/26 13:45, Elgin Perumbilly wrote:
> > Replace runtime_suspend/resume wrappers by using power helpers
> > directly with DEFINE_RUNTIME_DEV_PM_OPS().
> 
> ...why? What advantage does this refactoring bring?

It looks like patch removes two redundant functions, doesn't it? :-)

-- 
Regards,

Sakari Ailus

Re: [PATCH] media: i2c: imx415: Drop redundant runtime PM callbacks

[Michael Riesch]

Hi Elgin, Sakari,

On 3/25/26 13:28, Sakari Ailus wrote:
> Hi Michael,
> 
> On Tue, Mar 24, 2026 at 08:42:31PM +0100, Michael Riesch wrote:
>> Hi Elgin,
>>
>> Thanks for the patch but...
>>
>> On 3/24/26 13:45, Elgin Perumbilly wrote:
>>> Replace runtime_suspend/resume wrappers by using power helpers
>>> directly with DEFINE_RUNTIME_DEV_PM_OPS().
>>
>> ...why? What advantage does this refactoring bring?
> 
> It looks like patch removes two redundant functions, doesn't it? :-)

Well yes. 10 lines saved. Not sure whether this was the biggest issue
with this driver -- surely it was not a big issue for me.

Elgin, I was wondering whether this change is within a certain context
or whether there is something I fail to recognize.

Anyway, if you absolutely want this change: I feel it would be nicer if
you left the imx415_runtime_{resume,suspend} and moved the bit of code
from imx415_power_{on,off} to them. Reasons for that may be mostly
aesthetical -- I claim better readability that way.

Best regards,
Michael