Re: [PATCH] i2c: qcom-geni: Use devm_pm_runtime_enable() for PM management

[Konrad Dybcio]

On 7/10/26 2:13 PM, Mukesh Kumar Savaliya wrote:
> The current implementation manually calls pm_runtime_enable() in probe()
> and pm_runtime_disable() in remove() and error paths. This pattern is
> error-prone and requires careful cleanup in all failure paths. Using the
> devres-managed variant eliminates this complexity.
> 
> Migrate from manual pm_runtime_enable()/pm_runtime_disable() calls to
> the devres-managed devm_pm_runtime_enable() API. This simplifies the
> driver by automatically handling runtime PM cleanup when the device is
> removed or probe fails. This helps  with Simplified error handling and
> Automatic cleanup.
> 
> Changes:
> - Replace pm_runtime_enable() with devm_pm_runtime_enable() in probe()
> - Remove pm_runtime_disable() from remove() function
> - Remove pm_runtime_disable() from probe() error paths after
>   geni_i2c_init() and i2c_add_adapter()
> - Use 'dev' pointer consistently instead of 'gi2c->se.dev' for PM APIs

Drop this paragraph, describing changes like is unnecessary since
we can just infer this from the diff

> 
> Signed-off-by: Mukesh Kumar Savaliya <mukesh.savaliya@xxxxxxxxxxxxxxxx>
> ---
>  drivers/i2c/busses/i2c-qcom-geni.c | 25 +++++++++++--------------
>  1 file changed, 11 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-qcom-geni.c b/drivers/i2c/busses/i2c-qcom-geni.c
> index 96dbf04138be..6d6c2fa287ca 100644
> --- a/drivers/i2c/busses/i2c-qcom-geni.c
> +++ b/drivers/i2c/busses/i2c-qcom-geni.c
> @@ -1126,27 +1126,25 @@ static int geni_i2c_probe(struct platform_device *pdev)
>  	gi2c->adap.dev.of_node = dev->of_node;
>  	strscpy(gi2c->adap.name, "Geni-I2C", sizeof(gi2c->adap.name));
>  
> -	pm_runtime_set_suspended(gi2c->se.dev);
> -	pm_runtime_set_autosuspend_delay(gi2c->se.dev, I2C_AUTO_SUSPEND_DELAY);
> -	pm_runtime_use_autosuspend(gi2c->se.dev);
> -	pm_runtime_enable(gi2c->se.dev);
> +	ret = devm_pm_runtime_enable(dev);
> +	if (ret)
> +		return ret;

This shifts the enabling of runpm> +
> +	pm_runtime_set_suspended(dev);
> +	pm_runtime_set_autosuspend_delay(dev, I2C_AUTO_SUSPEND_DELAY);
> +	pm_runtime_use_autosuspend(dev);


..which is wrong since:

/**
 * pm_runtime_set_suspended - Set runtime PM status to "suspended".
 * @dev: Target device.
 *
 * Set the runtime PM status of @dev to %RPM_SUSPENDED and ensure that
 * dependencies of it will be taken into account.
 *
 * It is not valid to call this function for devices with runtime PM enabled.
 */


Konrad