Re: [PATCH v2 1/4] PM: device: Introduce platform_resources_managed flag

[Rafael J. Wysocki]

On Mon, Apr 14, 2025 at 8:53 PM Michal Wilczynski
<m.wilczynski@xxxxxxxxxxx> wrote:
>
> Introduce a new dev_pm_info flag - platform_resources_managed, to
> indicate whether platform PM resources such as clocks or resets are
> managed externally (e.g. by a generic power domain driver) instead of
> directly by the consumer device driver.

I think that this is genpd-specific and so I don't think it belongs in
struct dev_pm_info.

There is dev->power.subsys_data->domain_data, why not use it for this?

Also, it should be documented way more comprehensively IMV.

Who is supposed to set it and when?  What does it mean when it is set?

> This flag enables device drivers to cooperate with SoC-specific PM
> domains by conditionally skipping management of clocks and resets when
> the platform owns them.
>
> This idea was discussed on the mailing list [1].
>
> [1] - https://lore.kernel.org/all/CAPDyKFq=BF5f2i_Sr1cmVqtVAMgr=0FqsksL7RHZLKn++y0uwg@xxxxxxxxxxxxxx/
>
> Signed-off-by: Michal Wilczynski <m.wilczynski@xxxxxxxxxxx>
> ---
>  include/linux/device.h | 11 +++++++++++
>  include/linux/pm.h     |  1 +
>  2 files changed, 12 insertions(+)
>
> diff --git a/include/linux/device.h b/include/linux/device.h
> index 79e49fe494b7c4c70d902886db63c4cfe5b4de4f..3e7a36dd874cfb6b98e2451c7a876989aa9f1913 100644
> --- a/include/linux/device.h
> +++ b/include/linux/device.h
> @@ -881,6 +881,17 @@ static inline bool device_async_suspend_enabled(struct device *dev)
>         return !!dev->power.async_suspend;
>  }
>
> +static inline bool device_platform_resources_pm_managed(struct device *dev)

Could this function name be shorter?

> +{
> +       return dev->power.platform_resources_managed;
> +}
> +
> +static inline void device_platform_resources_set_pm_managed(struct device *dev,
> +                                                           bool val)

Ditto?

> +{
> +       dev->power.platform_resources_managed = val;
> +}
> +
>  static inline bool device_pm_not_required(struct device *dev)
>  {
>         return dev->power.no_pm;
> diff --git a/include/linux/pm.h b/include/linux/pm.h
> index f0bd8fbae4f2c09c63d780bb2528693acf2d2da1..cd6cb59686e4a5e9eaa2701d1e44af2abbfd88d1 100644
> --- a/include/linux/pm.h
> +++ b/include/linux/pm.h
> @@ -670,6 +670,7 @@ struct dev_pm_info {
>         bool                    no_pm:1;
>         bool                    early_init:1;   /* Owned by the PM core */
>         bool                    direct_complete:1;      /* Owned by the PM core */
> +       bool                    platform_resources_managed:1;
>         u32                     driver_flags;
>         spinlock_t              lock;
>  #ifdef CONFIG_PM_SLEEP
>
> --
> 2.34.1
>
>