Re: [PATCH 03/14] iio: adc: rzg2l_adc: Simplify the runtime PM code

[Claudiu Beznea <claudiu.beznea@tuxon.dev>]

Hi, Paul,

On 03.12.2024 14:53, Paul Barker wrote:
> Hi Claudiu,
> 
> On 03/12/2024 11:13, Claudiu wrote:
>> From: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com>
>>
>> All Renesas SoCs using the rzg2l_adc driver manage ADC clocks through PM
>> domains. Calling pm_runtime_{resume_and_get, put_sync}() implicitly sets
>> the state of the clocks. As a result, the code in the rzg2l_adc driver that
>> explicitly manages ADC clocks can be removed, leading to simpler and
>> cleaner implementation.
>>
>> Additionally, replace the use of rzg2l_adc_set_power() with direct PM
>> runtime API calls to further simplify and clean up the code.
>>
>> Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com>
>> ---
>>  drivers/iio/adc/rzg2l_adc.c | 100 ++++++++----------------------------
>>  1 file changed, 20 insertions(+), 80 deletions(-)
>>
>> diff --git a/drivers/iio/adc/rzg2l_adc.c b/drivers/iio/adc/rzg2l_adc.c
>> index 7039949a7554..a17690ecbdc3 100644
>> --- a/drivers/iio/adc/rzg2l_adc.c
>> +++ b/drivers/iio/adc/rzg2l_adc.c
>> @@ -8,7 +8,6 @@
>>   */
>>  
>>  #include <linux/bitfield.h>
>> -#include <linux/clk.h>
>>  #include <linux/completion.h>
>>  #include <linux/delay.h>
>>  #include <linux/iio/iio.h>
>> @@ -69,8 +68,6 @@ struct rzg2l_adc_data {
>>  
>>  struct rzg2l_adc {
>>  	void __iomem *base;
>> -	struct clk *pclk;
>> -	struct clk *adclk;
>>  	struct reset_control *presetn;
>>  	struct reset_control *adrstn;
>>  	struct completion completion;
>> @@ -188,29 +185,18 @@ static int rzg2l_adc_conversion_setup(struct rzg2l_adc *adc, u8 ch)
>>  	return 0;
>>  }
>>  
>> -static int rzg2l_adc_set_power(struct iio_dev *indio_dev, bool on)
>> -{
>> -	struct device *dev = indio_dev->dev.parent;
>> -
>> -	if (on)
>> -		return pm_runtime_resume_and_get(dev);
>> -
>> -	return pm_runtime_put_sync(dev);
>> -}
>> -
>>  static int rzg2l_adc_conversion(struct iio_dev *indio_dev, struct rzg2l_adc *adc, u8 ch)
>>  {
>> +	struct device *dev = indio_dev->dev.parent;
>>  	int ret;
>>  
>> -	ret = rzg2l_adc_set_power(indio_dev, true);
>> +	ret = pm_runtime_resume_and_get(dev);
>>  	if (ret)
>>  		return ret;
> 
> Should we check (ret < 0) here instead of just (ret)? According to the
> docs [1], pm_runtime_resume_and_get() can return 1 if the device is
> already active.

The v6.13-rc1 implementation of pm_runtime_resume_and_get() is:

static inline int pm_runtime_resume_and_get(struct device *dev)
{
	int ret;

	ret = __pm_runtime_resume(dev, RPM_GET_PUT);
	if (ret < 0) {
		pm_runtime_put_noidle(dev);
		return ret;
	}

	return 0;
}

It can return zero or negative error number.

Thank you,
Claudiu


> 
> [1]: https://docs.kernel.org/power/runtime_pm.html#runtime-pm-device-helper-functions
> 
> Thanks,
>