Re: [PATCH] cpufreq: schedutil: Call sugov_update_next_freq() before check to fast_switch_enabled

[Rafael J. Wysocki]

On Wed, Feb 24, 2021 at 6:44 AM Yue Hu  wrote:
>
> From: Yue Hu 
>
> Note that sugov_update_next_freq() may return false, that means the
> caller sugov_fast_switch() will do nothing except fast switch check.
>
> Similarly, sugov_deferred_update() also has unnecessary operations
> of raw_spin_{lock,unlock} in sugov_update_single_freq() for that case.
>
> So, let's call sugov_update_next_freq() before the fast switch check
> to avoid unnecessary behaviors above. Update the related interface
> definitions accordingly.
>
> Signed-off-by: Yue Hu 
> ---
>  kernel/sched/cpufreq_schedutil.c | 28 ++--
>  1 file changed, 14 insertions(+), 14 deletions(-)
>
> diff --git a/kernel/sched/cpufreq_schedutil.c 
> b/kernel/sched/cpufreq_schedutil.c
> index 41e498b..d23e5be 100644
> --- a/kernel/sched/cpufreq_schedutil.c
> +++ b/kernel/sched/cpufreq_schedutil.c
> @@ -114,19 +114,13 @@ static bool sugov_update_next_freq(struct sugov_policy 
> *sg_policy, u64 time,
> return true;
>  }
>
> -static void sugov_fast_switch(struct sugov_policy *sg_policy, u64 time,
> - unsigned int next_freq)
> +static void sugov_fast_switch(struct sugov_policy *sg_policy, unsigned int 
> next_freq)
>  {
> -   if (sugov_update_next_freq(sg_policy, time, next_freq))
> -   cpufreq_driver_fast_switch(sg_policy->policy, next_freq);
> +   cpufreq_driver_fast_switch(sg_policy->policy, next_freq);
>  }
>
> -static void sugov_deferred_update(struct sugov_policy *sg_policy, u64 time,
> - unsigned int next_freq)
> +static void sugov_deferred_update(struct sugov_policy *sg_policy)
>  {
> -   if (!sugov_update_next_freq(sg_policy, time, next_freq))
> -   return;
> -
> if (!sg_policy->work_in_progress) {
> sg_policy->work_in_progress = true;
> irq_work_queue(_policy->irq_work);
> @@ -368,16 +362,19 @@ static void sugov_update_single_freq(struct 
> update_util_data *hook, u64 time,
> sg_policy->cached_raw_freq = cached_freq;
> }
>
> +   if (!sugov_update_next_freq(sg_policy, time, next_f))
> +   return;
> +
> /*
>  * This code runs under rq->lock for the target CPU, so it won't run
>  * concurrently on two different CPUs for the same target and it is 
> not
>  * necessary to acquire the lock in the fast switch case.
>  */
> if (sg_policy->policy->fast_switch_enabled) {
> -   sugov_fast_switch(sg_policy, time, next_f);
> +   sugov_fast_switch(sg_policy, next_f);
> } else {
> raw_spin_lock(_policy->update_lock);
> -   sugov_deferred_update(sg_policy, time, next_f);
> +   sugov_deferred_update(sg_policy);
> raw_spin_unlock(_policy->update_lock);
> }
>  }
> @@ -456,12 +453,15 @@ static unsigned int sugov_next_freq_shared(struct 
> sugov_cpu *sg_cpu, u64 time)
> if (sugov_should_update_freq(sg_policy, time)) {
> next_f = sugov_next_freq_shared(sg_cpu, time);
>
> +   if (!sugov_update_next_freq(sg_policy, time, next_f))
> +   goto unlock;
> +
> if (sg_policy->policy->fast_switch_enabled)
> -   sugov_fast_switch(sg_policy, time, next_f);
> +   sugov_fast_switch(sg_policy, next_f);
> else
> -   sugov_deferred_update(sg_policy, time, next_f);
> +   sugov_deferred_update(sg_policy);
> }
> -
> +unlock:
> raw_spin_unlock(_policy->update_lock);
>  }
>
> --

Applied as 5.13 material, thanks!

Re: [PATCH] cpufreq: schedutil: Call sugov_update_next_freq() before check to fast_switch_enabled

[Yue Hu]

On Wed, 24 Feb 2021 11:32:36 +0530
Viresh Kumar  wrote:

> On 24-02-21, 13:42, Yue Hu wrote:
> > From: Yue Hu 
> > 
> > Note that sugov_update_next_freq() may return false, that means the
> > caller sugov_fast_switch() will do nothing except fast switch check.
> > 
> > Similarly, sugov_deferred_update() also has unnecessary operations
> > of raw_spin_{lock,unlock} in sugov_update_single_freq() for that case.
> > 
> > So, let's call sugov_update_next_freq() before the fast switch check
> > to avoid unnecessary behaviors above. Update the related interface
> > definitions accordingly.
> > 
> > Signed-off-by: Yue Hu 
> > ---
> >  kernel/sched/cpufreq_schedutil.c | 28 ++--
> >  1 file changed, 14 insertions(+), 14 deletions(-)
> > 
> > diff --git a/kernel/sched/cpufreq_schedutil.c 
> > b/kernel/sched/cpufreq_schedutil.c
> > index 41e498b..d23e5be 100644
> > --- a/kernel/sched/cpufreq_schedutil.c
> > +++ b/kernel/sched/cpufreq_schedutil.c
> > @@ -114,19 +114,13 @@ static bool sugov_update_next_freq(struct 
> > sugov_policy *sg_policy, u64 time,
> > return true;
> >  }
> >  
> > -static void sugov_fast_switch(struct sugov_policy *sg_policy, u64 time,
> > - unsigned int next_freq)
> > +static void sugov_fast_switch(struct sugov_policy *sg_policy, unsigned int 
> > next_freq)
> >  {
> > -   if (sugov_update_next_freq(sg_policy, time, next_freq))
> > -   cpufreq_driver_fast_switch(sg_policy->policy, next_freq);
> > +   cpufreq_driver_fast_switch(sg_policy->policy, next_freq);  
> 
> I will call this directly instead, no need of the wrapper anymore.

To fix it in next version.

Thank you.

> 
> >  }
> >  
> > -static void sugov_deferred_update(struct sugov_policy *sg_policy, u64 time,
> > - unsigned int next_freq)
> > +static void sugov_deferred_update(struct sugov_policy *sg_policy)
> >  {
> > -   if (!sugov_update_next_freq(sg_policy, time, next_freq))
> > -   return;
> > -
> > if (!sg_policy->work_in_progress) {
> > sg_policy->work_in_progress = true;
> > irq_work_queue(_policy->irq_work);
> > @@ -368,16 +362,19 @@ static void sugov_update_single_freq(struct 
> > update_util_data *hook, u64 time,
> > sg_policy->cached_raw_freq = cached_freq;
> > }
> >  
> > +   if (!sugov_update_next_freq(sg_policy, time, next_f))
> > +   return;
> > +
> > /*
> >  * This code runs under rq->lock for the target CPU, so it won't run
> >  * concurrently on two different CPUs for the same target and it is not
> >  * necessary to acquire the lock in the fast switch case.
> >  */
> > if (sg_policy->policy->fast_switch_enabled) {
> > -   sugov_fast_switch(sg_policy, time, next_f);
> > +   sugov_fast_switch(sg_policy, next_f);
> > } else {
> > raw_spin_lock(_policy->update_lock);
> > -   sugov_deferred_update(sg_policy, time, next_f);
> > +   sugov_deferred_update(sg_policy);
> > raw_spin_unlock(_policy->update_lock);
> > }
> >  }
> > @@ -456,12 +453,15 @@ static unsigned int sugov_next_freq_shared(struct 
> > sugov_cpu *sg_cpu, u64 time)
> > if (sugov_should_update_freq(sg_policy, time)) {
> > next_f = sugov_next_freq_shared(sg_cpu, time);
> >  
> > +   if (!sugov_update_next_freq(sg_policy, time, next_f))
> > +   goto unlock;
> > +
> > if (sg_policy->policy->fast_switch_enabled)
> > -   sugov_fast_switch(sg_policy, time, next_f);
> > +   sugov_fast_switch(sg_policy, next_f);
> > else
> > -   sugov_deferred_update(sg_policy, time, next_f);
> > +   sugov_deferred_update(sg_policy);
> > }
> > -
> > +unlock:
> > raw_spin_unlock(_policy->update_lock);
> >  }  
> 

Re: [PATCH] cpufreq: schedutil: Call sugov_update_next_freq() before check to fast_switch_enabled

[Viresh Kumar]

On 24-02-21, 13:42, Yue Hu wrote:
> From: Yue Hu 
> 
> Note that sugov_update_next_freq() may return false, that means the
> caller sugov_fast_switch() will do nothing except fast switch check.
> 
> Similarly, sugov_deferred_update() also has unnecessary operations
> of raw_spin_{lock,unlock} in sugov_update_single_freq() for that case.
> 
> So, let's call sugov_update_next_freq() before the fast switch check
> to avoid unnecessary behaviors above. Update the related interface
> definitions accordingly.
> 
> Signed-off-by: Yue Hu 
> ---
>  kernel/sched/cpufreq_schedutil.c | 28 ++--
>  1 file changed, 14 insertions(+), 14 deletions(-)
> 
> diff --git a/kernel/sched/cpufreq_schedutil.c 
> b/kernel/sched/cpufreq_schedutil.c
> index 41e498b..d23e5be 100644
> --- a/kernel/sched/cpufreq_schedutil.c
> +++ b/kernel/sched/cpufreq_schedutil.c
> @@ -114,19 +114,13 @@ static bool sugov_update_next_freq(struct sugov_policy 
> *sg_policy, u64 time,
>   return true;
>  }
>  
> -static void sugov_fast_switch(struct sugov_policy *sg_policy, u64 time,
> -   unsigned int next_freq)
> +static void sugov_fast_switch(struct sugov_policy *sg_policy, unsigned int 
> next_freq)
>  {
> - if (sugov_update_next_freq(sg_policy, time, next_freq))
> - cpufreq_driver_fast_switch(sg_policy->policy, next_freq);
> + cpufreq_driver_fast_switch(sg_policy->policy, next_freq);

I will call this directly instead, no need of the wrapper anymore.

>  }
>  
> -static void sugov_deferred_update(struct sugov_policy *sg_policy, u64 time,
> -   unsigned int next_freq)
> +static void sugov_deferred_update(struct sugov_policy *sg_policy)
>  {
> - if (!sugov_update_next_freq(sg_policy, time, next_freq))
> - return;
> -
>   if (!sg_policy->work_in_progress) {
>   sg_policy->work_in_progress = true;
>   irq_work_queue(_policy->irq_work);
> @@ -368,16 +362,19 @@ static void sugov_update_single_freq(struct 
> update_util_data *hook, u64 time,
>   sg_policy->cached_raw_freq = cached_freq;
>   }
>  
> + if (!sugov_update_next_freq(sg_policy, time, next_f))
> + return;
> +
>   /*
>* This code runs under rq->lock for the target CPU, so it won't run
>* concurrently on two different CPUs for the same target and it is not
>* necessary to acquire the lock in the fast switch case.
>*/
>   if (sg_policy->policy->fast_switch_enabled) {
> - sugov_fast_switch(sg_policy, time, next_f);
> + sugov_fast_switch(sg_policy, next_f);
>   } else {
>   raw_spin_lock(_policy->update_lock);
> - sugov_deferred_update(sg_policy, time, next_f);
> + sugov_deferred_update(sg_policy);
>   raw_spin_unlock(_policy->update_lock);
>   }
>  }
> @@ -456,12 +453,15 @@ static unsigned int sugov_next_freq_shared(struct 
> sugov_cpu *sg_cpu, u64 time)
>   if (sugov_should_update_freq(sg_policy, time)) {
>   next_f = sugov_next_freq_shared(sg_cpu, time);
>  
> + if (!sugov_update_next_freq(sg_policy, time, next_f))
> + goto unlock;
> +
>   if (sg_policy->policy->fast_switch_enabled)
> - sugov_fast_switch(sg_policy, time, next_f);
> + sugov_fast_switch(sg_policy, next_f);
>   else
> - sugov_deferred_update(sg_policy, time, next_f);
> + sugov_deferred_update(sg_policy);
>   }
> -
> +unlock:
>   raw_spin_unlock(_policy->update_lock);
>  }

-- 
viresh

[PATCH] cpufreq: schedutil: Call sugov_update_next_freq() before check to fast_switch_enabled

[Yue Hu]

From: Yue Hu 

Note that sugov_update_next_freq() may return false, that means the
caller sugov_fast_switch() will do nothing except fast switch check.

Similarly, sugov_deferred_update() also has unnecessary operations
of raw_spin_{lock,unlock} in sugov_update_single_freq() for that case.

So, let's call sugov_update_next_freq() before the fast switch check
to avoid unnecessary behaviors above. Update the related interface
definitions accordingly.

Signed-off-by: Yue Hu 
---
 kernel/sched/cpufreq_schedutil.c | 28 ++--
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c
index 41e498b..d23e5be 100644
--- a/kernel/sched/cpufreq_schedutil.c
+++ b/kernel/sched/cpufreq_schedutil.c
@@ -114,19 +114,13 @@ static bool sugov_update_next_freq(struct sugov_policy 
*sg_policy, u64 time,
return true;
 }
 
-static void sugov_fast_switch(struct sugov_policy *sg_policy, u64 time,
- unsigned int next_freq)
+static void sugov_fast_switch(struct sugov_policy *sg_policy, unsigned int 
next_freq)
 {
-   if (sugov_update_next_freq(sg_policy, time, next_freq))
-   cpufreq_driver_fast_switch(sg_policy->policy, next_freq);
+   cpufreq_driver_fast_switch(sg_policy->policy, next_freq);
 }
 
-static void sugov_deferred_update(struct sugov_policy *sg_policy, u64 time,
- unsigned int next_freq)
+static void sugov_deferred_update(struct sugov_policy *sg_policy)
 {
-   if (!sugov_update_next_freq(sg_policy, time, next_freq))
-   return;
-
if (!sg_policy->work_in_progress) {
sg_policy->work_in_progress = true;
irq_work_queue(_policy->irq_work);
@@ -368,16 +362,19 @@ static void sugov_update_single_freq(struct 
update_util_data *hook, u64 time,
sg_policy->cached_raw_freq = cached_freq;
}
 
+   if (!sugov_update_next_freq(sg_policy, time, next_f))
+   return;
+
/*
 * This code runs under rq->lock for the target CPU, so it won't run
 * concurrently on two different CPUs for the same target and it is not
 * necessary to acquire the lock in the fast switch case.
 */
if (sg_policy->policy->fast_switch_enabled) {
-   sugov_fast_switch(sg_policy, time, next_f);
+   sugov_fast_switch(sg_policy, next_f);
} else {
raw_spin_lock(_policy->update_lock);
-   sugov_deferred_update(sg_policy, time, next_f);
+   sugov_deferred_update(sg_policy);
raw_spin_unlock(_policy->update_lock);
}
 }
@@ -456,12 +453,15 @@ static unsigned int sugov_next_freq_shared(struct 
sugov_cpu *sg_cpu, u64 time)
if (sugov_should_update_freq(sg_policy, time)) {
next_f = sugov_next_freq_shared(sg_cpu, time);
 
+   if (!sugov_update_next_freq(sg_policy, time, next_f))
+   goto unlock;
+
if (sg_policy->policy->fast_switch_enabled)
-   sugov_fast_switch(sg_policy, time, next_f);
+   sugov_fast_switch(sg_policy, next_f);
else
-   sugov_deferred_update(sg_policy, time, next_f);
+   sugov_deferred_update(sg_policy);
}
-
+unlock:
raw_spin_unlock(_policy->update_lock);
 }
 
-- 
1.9.1