[v2,2/7] pm: runtime: Add pm_runtime_put_mark_busy_autosusp() helper

Add pm_runtime_put_mark_busy_autosusp() helper function for users that
wish to set the last_busy timestamp to current time and put the
usage_count of the device and set the autosuspend timer.

Essentially calling pm_runtime_suspend_mark_busy_autosusp() equal to
calling first pm_runtime_mark_last_busy() and then
pm_runtime_put_autosuspend().

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
---
 include/linux/pm_runtime.h | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

Hi Sakari,

Thank you for the patch.

On Fri, Nov 17, 2023 at 01:14:28PM +0200, Sakari Ailus wrote:
> Add pm_runtime_put_mark_busy_autosusp() helper function for users that
> wish to set the last_busy timestamp to current time and put the
> usage_count of the device and set the autosuspend timer.
> 
> Essentially calling pm_runtime_suspend_mark_busy_autosusp() equal to
> calling first pm_runtime_mark_last_busy() and then
> pm_runtime_put_autosuspend().

The vast majority if the pm_runtime_put_autosuspend() users call
pm_runtime_mark_last_busy() right before. Let's make the
pm_runtime_put_autosuspend() function do that by default, and add a
__pm_runtime_put_autosuspend() (name to be bikshedded) for the minority
of cases where updating the last busy timestamp isn't desired. We want
to simplify the API, not make it more complex.

> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> ---
>  include/linux/pm_runtime.h | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> diff --git a/include/linux/pm_runtime.h b/include/linux/pm_runtime.h
> index 13cd526634c1..4afe7b0b9d7e 100644
> --- a/include/linux/pm_runtime.h
> +++ b/include/linux/pm_runtime.h
> @@ -495,6 +495,23 @@ static inline int pm_runtime_put_autosuspend(struct device *dev)
>  	    RPM_GET_PUT | RPM_ASYNC | RPM_AUTO);
>  }
>  
> +/**
> + * pm_runtime_put_mark_busy_autosusp - Update the last access time of a device
> + *				       and drop device usage counter and queue
> + *				       autosuspend if 0.
> + * @dev: Target device.
> + *
> + * Update the last access time of @dev using pm_runtime_mark_last_busy(), then
> + * decrement the runtime PM usage counter of @dev and if it turns out to be
> + * equal to 0, queue up a work item for @dev like in pm_request_autosuspend().
> + */
> +static inline int pm_runtime_put_mark_busy_autosusp(struct device *dev)
> +{
> +	pm_runtime_mark_last_busy(dev);
> +
> +	return pm_runtime_autosuspend(dev);
> +}
> +
>  /**
>   * pm_runtime_put_sync - Drop device usage counter and run "idle check" if 0.
>   * @dev: Target device.

On Sat, Nov 18, 2023 at 6:49 PM Laurent Pinchart
<laurent.pinchart@ideasonboard.com> wrote:
>
> Hi Sakari,
>
> Thank you for the patch.
>
> On Fri, Nov 17, 2023 at 01:14:28PM +0200, Sakari Ailus wrote:
> > Add pm_runtime_put_mark_busy_autosusp() helper function for users that
> > wish to set the last_busy timestamp to current time and put the
> > usage_count of the device and set the autosuspend timer.
> >
> > Essentially calling pm_runtime_suspend_mark_busy_autosusp() equal to
> > calling first pm_runtime_mark_last_busy() and then
> > pm_runtime_put_autosuspend().
>
> The vast majority if the pm_runtime_put_autosuspend() users call
> pm_runtime_mark_last_busy() right before. Let's make the
> pm_runtime_put_autosuspend() function do that by default, and add a
> __pm_runtime_put_autosuspend() (name to be bikshedded) for the minority
> of cases where updating the last busy timestamp isn't desired. We want
> to simplify the API, not make it more complex.

I would also prefer it to be done this way if not too problematic.

Hi Rafael,

On Sat, Nov 18, 2023 at 10:20:46PM +0100, Rafael J. Wysocki wrote:
> On Sat, Nov 18, 2023 at 6:49 PM Laurent Pinchart wrote:
> > On Fri, Nov 17, 2023 at 01:14:28PM +0200, Sakari Ailus wrote:
> > > Add pm_runtime_put_mark_busy_autosusp() helper function for users that
> > > wish to set the last_busy timestamp to current time and put the
> > > usage_count of the device and set the autosuspend timer.
> > >
> > > Essentially calling pm_runtime_suspend_mark_busy_autosusp() equal to
> > > calling first pm_runtime_mark_last_busy() and then
> > > pm_runtime_put_autosuspend().
> >
> > The vast majority if the pm_runtime_put_autosuspend() users call
> > pm_runtime_mark_last_busy() right before. Let's make the
> > pm_runtime_put_autosuspend() function do that by default, and add a
> > __pm_runtime_put_autosuspend() (name to be bikshedded) for the minority
> > of cases where updating the last busy timestamp isn't desired. We want
> > to simplify the API, not make it more complex.
> 
> I would also prefer it to be done this way if not too problematic.

I'm glad you agree :-) The change will probably be a bit painful, but I
think it's for the best. Sakari, please let me know if I can help.