On Wed, 9 Jul 2025 02:11:52 +0300
Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx> wrote:
> pm_runtime_put_autosuspend(), pm_runtime_put_sync_autosuspend(),
> pm_runtime_autosuspend() and pm_request_autosuspend() now include a call
> to pm_runtime_mark_last_busy(). Remove the now-reduntant explicit call to
> pm_runtime_mark_last_busy().
>
> Signed-off-by: Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx>
> Reviewed-by: Linus Walleij <linus.walleij@xxxxxxxxxx>
Some comments for the future as more about what can be improved on the back
of this than what you have here.
>
> diff --git a/drivers/iio/adc/rcar-gyroadc.c b/drivers/iio/adc/rcar-gyroadc.c
> index cc326f21d398..3a17b3898bf6 100644
> --- a/drivers/iio/adc/rcar-gyroadc.c
> +++ b/drivers/iio/adc/rcar-gyroadc.c
> @@ -163,12 +163,10 @@ static int rcar_gyroadc_set_power(struct rcar_gyroadc *priv, bool on)
> {
> struct device *dev = priv->dev;
>
This is a very clear example of where the *_set_power() pattern is a bad idea.
There are two calls of this function, one with it hard coded as on and one with it
hard coded as off. We can just push the pm_runtime_resume_and_get()
to the on case etc.
I don't mind that much if we do so as a follow up patch so this one can
be the mechanical change and then we follow up with the enabled simplification.
> - if (on) {
> + if (on)
> return pm_runtime_resume_and_get(dev);
> - } else {
> - pm_runtime_mark_last_busy(dev);
> - return pm_runtime_put_autosuspend(dev);
> - }
> +
> + return pm_runtime_put_autosuspend(dev);
> }
>
> static int rcar_gyroadc_read_raw(struct iio_dev *indio_dev,
>> diff --git a/drivers/iio/adc/ti-ads1015.c b/drivers/iio/adc/ti-ads1015.c
> index 48549d617e5f..f2a93c63ca14 100644
> --- a/drivers/iio/adc/ti-ads1015.c
> +++ b/drivers/iio/adc/ti-ads1015.c
> @@ -374,12 +374,10 @@ static int ads1015_set_power_state(struct ads1015_data *data, bool on)
> int ret;
> struct device *dev = regmap_get_device(data->regmap);
>
> - if (on) {
> + if (on)
> ret = pm_runtime_resume_and_get(dev);
> - } else {
> - pm_runtime_mark_last_busy(dev);
> + else
> ret = pm_runtime_put_autosuspend(dev);
> - }
>
> return ret < 0 ? ret : 0;
So this one has a stub version which only brings benefits because
we have checks on the pm_runtime_put_autosuspend() path failing
(which it does if we have !CONFIG_PM).
I think the right option there is check the return value is < 0
for the resume_and_get() and don't check the _put_autosuspend()
return value at all. Then we can just push this down to the
call sites as all of them hard code the bool value.
> }
