[PATCH v2 0/2] PM / Domains: Initial improvements of system PM code in genpd

[PATCH v2 0/2] PM / Domains: Initial improvements of system PM code in genpd

[Ulf Hansson]

The system PM code in genpd needs a round of modernization and clean-up. This is
only the first initial part.

I intend to in step by step improve the code, so in a later point enable to do
optimizations. Such as, avoid waking up devices unless it's really needed.

Ulf Hansson (2):
  PM / Domains: Remove redundant pm_runtime_get|put*() in
    pm_genpd_prepare()
  PM / Domains: Drop unnecessary wakeup code from pm_genpd_prepare()

 drivers/base/power/domain.c | 14 +-------------
 1 file changed, 1 insertion(+), 13 deletions(-)

-- 
1.9.1

[PATCH v2 1/2] PM / Domains: Remove redundant pm_runtime_get|put*() in pm_genpd_prepare()

[Ulf Hansson]

The PM core increases and decreases the runtime PM usage count in the
system PM prepare phase. This makes some of the pm_runtime_get|put*()
calls in pm_gendp_prepare() redundant, so let's remove them.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
---

Changes in v2:
	Updated changlog and added Kevin's reviewed by.

---
 drivers/base/power/domain.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index 4ce4ce0..60a3573 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -730,14 +730,11 @@ static int pm_genpd_prepare(struct device *dev)
 	 * at this point and a system wakeup event should be reported if it's
 	 * set up to wake up the system from sleep states.
 	 */
-	pm_runtime_get_noresume(dev);
 	if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
 		pm_wakeup_event(dev, 0);
 
-	if (pm_wakeup_pending()) {
-		pm_runtime_put(dev);
+	if (pm_wakeup_pending())
 		return -EBUSY;
-	}
 
 	if (resume_needed(dev, genpd))
 		pm_runtime_resume(dev);
@@ -751,10 +748,8 @@ static int pm_genpd_prepare(struct device *dev)
 
 	mutex_unlock(&genpd->lock);
 
-	if (genpd->suspend_power_off) {
-		pm_runtime_put_noidle(dev);
+	if (genpd->suspend_power_off)
 		return 0;
-	}
 
 	/*
 	 * The PM domain must be in the GPD_STATE_ACTIVE state at this point,
@@ -776,7 +771,6 @@ static int pm_genpd_prepare(struct device *dev)
 		pm_runtime_enable(dev);
 	}
 
-	pm_runtime_put(dev);
 	return ret;
 }
 
-- 
1.9.1

Re: [PATCH v2 1/2] PM / Domains: Remove redundant pm_runtime_get|put*() in pm_genpd_prepare()

[Ulf Hansson]

On 26 April 2016 at 08:47, Ulf Hansson <ulf.hansson@linaro.org> wrote:
> The PM core increases and decreases the runtime PM usage count in the
> system PM prepare phase. This makes some of the pm_runtime_get|put*()
> calls in pm_gendp_prepare() redundant, so let's remove them.

Huh, I noticed I missed to update pm_gendp_prepare() to pm_genpd_prepare().

Rafael, can you amend the patch or you want me to send a v3?

Kind regards
Uffe

>
> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
> Reviewed-by: Kevin Hilman <khilman@baylibre.com>
> ---
>
> Changes in v2:
>         Updated changlog and added Kevin's reviewed by.
>
> ---
>  drivers/base/power/domain.c | 10 ++--------
>  1 file changed, 2 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
> index 4ce4ce0..60a3573 100644
> --- a/drivers/base/power/domain.c
> +++ b/drivers/base/power/domain.c
> @@ -730,14 +730,11 @@ static int pm_genpd_prepare(struct device *dev)
>          * at this point and a system wakeup event should be reported if it's
>          * set up to wake up the system from sleep states.
>          */
> -       pm_runtime_get_noresume(dev);
>         if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
>                 pm_wakeup_event(dev, 0);
>
> -       if (pm_wakeup_pending()) {
> -               pm_runtime_put(dev);
> +       if (pm_wakeup_pending())
>                 return -EBUSY;
> -       }
>
>         if (resume_needed(dev, genpd))
>                 pm_runtime_resume(dev);
> @@ -751,10 +748,8 @@ static int pm_genpd_prepare(struct device *dev)
>
>         mutex_unlock(&genpd->lock);
>
> -       if (genpd->suspend_power_off) {
> -               pm_runtime_put_noidle(dev);
> +       if (genpd->suspend_power_off)
>                 return 0;
> -       }
>
>         /*
>          * The PM domain must be in the GPD_STATE_ACTIVE state at this point,
> @@ -776,7 +771,6 @@ static int pm_genpd_prepare(struct device *dev)
>                 pm_runtime_enable(dev);
>         }
>
> -       pm_runtime_put(dev);
>         return ret;
>  }
>
> --
> 1.9.1
>

Re: [PATCH v2 1/2] PM / Domains: Remove redundant pm_runtime_get|put*() in pm_genpd_prepare()

[Rafael J. Wysocki]

On Tuesday, April 26, 2016 09:08:22 AM Ulf Hansson wrote:
> On 26 April 2016 at 08:47, Ulf Hansson <ulf.hansson@linaro.org> wrote:
> > The PM core increases and decreases the runtime PM usage count in the
> > system PM prepare phase. This makes some of the pm_runtime_get|put*()
> > calls in pm_gendp_prepare() redundant, so let's remove them.
> 
> Huh, I noticed I missed to update pm_gendp_prepare() to pm_genpd_prepare().
> 
> Rafael, can you amend the patch or you want me to send a v3?

No worries, I've fixed it up.

This one and the [2/2] applied, thanks!

Re: [PATCH v2 1/2] PM / Domains: Remove redundant pm_runtime_get|put*() in pm_genpd_prepare()

[Laurent Pinchart]

Hi Ulrich,

Thank you for the patch.

On Tuesday 26 Apr 2016 08:47:17 Ulf Hansson wrote:
> The PM core increases and decreases the runtime PM usage count in the
> system PM prepare phase. This makes some of the pm_runtime_get|put*()
> calls in pm_gendp_prepare() redundant, so let's remove them.
> 
> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
> Reviewed-by: Kevin Hilman <khilman@baylibre.com>

Apart from the gendp typo,

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

> ---
> 
> Changes in v2:
> 	Updated changlog and added Kevin's reviewed by.
> 
> ---
>  drivers/base/power/domain.c | 10 ++--------
>  1 file changed, 2 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
> index 4ce4ce0..60a3573 100644
> --- a/drivers/base/power/domain.c
> +++ b/drivers/base/power/domain.c
> @@ -730,14 +730,11 @@ static int pm_genpd_prepare(struct device *dev)
>  	 * at this point and a system wakeup event should be reported if it's
>  	 * set up to wake up the system from sleep states.
>  	 */
> -	pm_runtime_get_noresume(dev);
>  	if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
>  		pm_wakeup_event(dev, 0);
> 
> -	if (pm_wakeup_pending()) {
> -		pm_runtime_put(dev);
> +	if (pm_wakeup_pending())
>  		return -EBUSY;
> -	}
> 
>  	if (resume_needed(dev, genpd))
>  		pm_runtime_resume(dev);
> @@ -751,10 +748,8 @@ static int pm_genpd_prepare(struct device *dev)
> 
>  	mutex_unlock(&genpd->lock);
> 
> -	if (genpd->suspend_power_off) {
> -		pm_runtime_put_noidle(dev);
> +	if (genpd->suspend_power_off)
>  		return 0;
> -	}
> 
>  	/*
>  	 * The PM domain must be in the GPD_STATE_ACTIVE state at this point,
> @@ -776,7 +771,6 @@ static int pm_genpd_prepare(struct device *dev)
>  		pm_runtime_enable(dev);
>  	}
> 
> -	pm_runtime_put(dev);
>  	return ret;
>  }

-- 
Regards,

Laurent Pinchart

Re: [PATCH v2 1/2] PM / Domains: Remove redundant pm_runtime_get|put*() in pm_genpd_prepare()

[Laurent Pinchart]

Hi Ulf,

Thank you for the patch.

On Tuesday 26 Apr 2016 08:47:17 Ulf Hansson wrote:
> The PM core increases and decreases the runtime PM usage count in the
> system PM prepare phase. This makes some of the pm_runtime_get|put*()
> calls in pm_gendp_prepare() redundant, so let's remove them.
> 
> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
> Reviewed-by: Kevin Hilman <khilman@baylibre.com>

Apart from the gendp typo,

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

> ---
> 
> Changes in v2:
> 	Updated changlog and added Kevin's reviewed by.
> 
> ---
>  drivers/base/power/domain.c | 10 ++--------
>  1 file changed, 2 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
> index 4ce4ce0..60a3573 100644
> --- a/drivers/base/power/domain.c
> +++ b/drivers/base/power/domain.c
> @@ -730,14 +730,11 @@ static int pm_genpd_prepare(struct device *dev)
>  	 * at this point and a system wakeup event should be reported if it's
>  	 * set up to wake up the system from sleep states.
>  	 */
> -	pm_runtime_get_noresume(dev);
>  	if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
>  		pm_wakeup_event(dev, 0);
> 
> -	if (pm_wakeup_pending()) {
> -		pm_runtime_put(dev);
> +	if (pm_wakeup_pending())
>  		return -EBUSY;
> -	}
> 
>  	if (resume_needed(dev, genpd))
>  		pm_runtime_resume(dev);
> @@ -751,10 +748,8 @@ static int pm_genpd_prepare(struct device *dev)
> 
>  	mutex_unlock(&genpd->lock);
> 
> -	if (genpd->suspend_power_off) {
> -		pm_runtime_put_noidle(dev);
> +	if (genpd->suspend_power_off)
>  		return 0;
> -	}
> 
>  	/*
>  	 * The PM domain must be in the GPD_STATE_ACTIVE state at this point,
> @@ -776,7 +771,6 @@ static int pm_genpd_prepare(struct device *dev)
>  		pm_runtime_enable(dev);
>  	}
> 
> -	pm_runtime_put(dev);
>  	return ret;
>  }

-- 
Regards,

Laurent Pinchart

[PATCH v2 2/2] PM / Domains: Drop unnecessary wakeup code from pm_genpd_prepare()

[Ulf Hansson]

As the PM core already have wakeup management during the system PM phase,
it seems reasonable that genpd and its users should be able to rely on
that. Therefore let's remove this from pm_genpd_prepare().

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
---

Changes in v2:
	None.

---
 drivers/base/power/domain.c | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index 60a3573..de23b64 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -730,12 +730,6 @@ static int pm_genpd_prepare(struct device *dev)
 	 * at this point and a system wakeup event should be reported if it's
 	 * set up to wake up the system from sleep states.
 	 */
-	if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
-		pm_wakeup_event(dev, 0);
-
-	if (pm_wakeup_pending())
-		return -EBUSY;
-
 	if (resume_needed(dev, genpd))
 		pm_runtime_resume(dev);
 
-- 
1.9.1

Re: [PATCH v2 2/2] PM / Domains: Drop unnecessary wakeup code from pm_genpd_prepare()

[Laurent Pinchart]

Hi Ulf,

Thank you for the patch.

On Tuesday 26 Apr 2016 08:47:18 Ulf Hansson wrote:
> As the PM core already have wakeup management during the system PM phase,
> it seems reasonable that genpd and its users should be able to rely on
> that. Therefore let's remove this from pm_genpd_prepare().

It's probably because I'm not familiar enough with the PM implementation, but 
I haven't been able to locate the wakeup management code in the PM core.

> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
> ---
> 
> Changes in v2:
> 	None.
> 
> ---
>  drivers/base/power/domain.c | 6 ------
>  1 file changed, 6 deletions(-)
> 
> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
> index 60a3573..de23b64 100644
> --- a/drivers/base/power/domain.c
> +++ b/drivers/base/power/domain.c
> @@ -730,12 +730,6 @@ static int pm_genpd_prepare(struct device *dev)
>  	 * at this point and a system wakeup event should be reported if it's
>  	 * set up to wake up the system from sleep states.
>  	 */
> -	if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
> -		pm_wakeup_event(dev, 0);
> -
> -	if (pm_wakeup_pending())
> -		return -EBUSY;
> -
>  	if (resume_needed(dev, genpd))
>  		pm_runtime_resume(dev);

-- 
Regards,

Laurent Pinchart

Re: [PATCH v2 2/2] PM / Domains: Drop unnecessary wakeup code from pm_genpd_prepare()

[Ulf Hansson]

On 26 April 2016 at 09:31, Laurent Pinchart
<laurent.pinchart@ideasonboard.com> wrote:
> Hi Ulf,
>
> Thank you for the patch.
>
> On Tuesday 26 Apr 2016 08:47:18 Ulf Hansson wrote:
>> As the PM core already have wakeup management during the system PM phase,
>> it seems reasonable that genpd and its users should be able to rely on
>> that. Therefore let's remove this from pm_genpd_prepare().
>
> It's probably because I'm not familiar enough with the PM implementation, but
> I haven't been able to locate the wakeup management code in the PM core.

It's done in __device_suspend() (drivers/base/power/main.c).

So it means it will be moved from the system PM prepare phase (as
currently in genpd) to the system PM suspend phase. I think this
should be okay for genpd users.

Kind regards
Uffe

>
>> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
>> ---
>>
>> Changes in v2:
>>       None.
>>
>> ---
>>  drivers/base/power/domain.c | 6 ------
>>  1 file changed, 6 deletions(-)
>>
>> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
>> index 60a3573..de23b64 100644
>> --- a/drivers/base/power/domain.c
>> +++ b/drivers/base/power/domain.c
>> @@ -730,12 +730,6 @@ static int pm_genpd_prepare(struct device *dev)
>>        * at this point and a system wakeup event should be reported if it's
>>        * set up to wake up the system from sleep states.
>>        */
>> -     if (pm_runtime_barrier(dev) && device_may_wakeup(dev))
>> -             pm_wakeup_event(dev, 0);
>> -
>> -     if (pm_wakeup_pending())
>> -             return -EBUSY;
>> -
>>       if (resume_needed(dev, genpd))
>>               pm_runtime_resume(dev);
>
> --
> Regards,
>
> Laurent Pinchart
>