Remove the capacity inversion detection which is now handled by
util_fits_cpu() returning -1 when we need to continue to look for a
potential CPU with better performance.

This ends up almost reverting patches below except for some comments:
commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection")
commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()")
commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion")

Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
---
 kernel/sched/fair.c  | 84 +++-----------------------------------------
 kernel/sched/sched.h | 19 ----------
 2 files changed, 5 insertions(+), 98 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 074742f107c0..c6c8e7f52935 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4476,17 +4476,9 @@ static inline int util_fits_cpu(unsigned long util,
 	 *
 	 * For uclamp_max, we can tolerate a drop in performance level as the
 	 * goal is to cap the task. So it's okay if it's getting less.
-	 *
-	 * In case of capacity inversion we should honour the inverted capacity
-	 * for both uclamp_min and uclamp_max all the time.
 	 */
-	capacity_orig = cpu_in_capacity_inversion(cpu);
-	if (capacity_orig) {
-		capacity_orig_thermal = capacity_orig;
-	} else {
-		capacity_orig = capacity_orig_of(cpu);
-		capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
-	}
+	capacity_orig = capacity_orig_of(cpu);
+	capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
 
 	/*
 	 * We want to force a task to fit a cpu as implied by uclamp_max.
@@ -9027,82 +9019,16 @@ static unsigned long scale_rt_capacity(int cpu)
 
 static void update_cpu_capacity(struct sched_domain *sd, int cpu)
 {
-	unsigned long capacity_orig = arch_scale_cpu_capacity(cpu);
 	unsigned long capacity = scale_rt_capacity(cpu);
 	struct sched_group *sdg = sd->groups;
-	struct rq *rq = cpu_rq(cpu);
 
-	rq->cpu_capacity_orig = capacity_orig;
+	cpu_rq(cpu)->cpu_capacity_orig = arch_scale_cpu_capacity(cpu);
 
 	if (!capacity)
 		capacity = 1;
 
-	rq->cpu_capacity = capacity;
-
-	/*
-	 * Detect if the performance domain is in capacity inversion state.
-	 *
-	 * Capacity inversion happens when another perf domain with equal or
-	 * lower capacity_orig_of() ends up having higher capacity than this
-	 * domain after subtracting thermal pressure.
-	 *
-	 * We only take into account thermal pressure in this detection as it's
-	 * the only metric that actually results in *real* reduction of
-	 * capacity due to performance points (OPPs) being dropped/become
-	 * unreachable due to thermal throttling.
-	 *
-	 * We assume:
-	 *   * That all cpus in a perf domain have the same capacity_orig
-	 *     (same uArch).
-	 *   * Thermal pressure will impact all cpus in this perf domain
-	 *     equally.
-	 */
-	if (sched_energy_enabled()) {
-		unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
-		struct perf_domain *pd;
-
-		rcu_read_lock();
-
-		pd = rcu_dereference(rq->rd->pd);
-		rq->cpu_capacity_inverted = 0;
-
-		for (; pd; pd = pd->next) {
-			struct cpumask *pd_span = perf_domain_span(pd);
-			unsigned long pd_cap_orig, pd_cap;
-
-			/* We can't be inverted against our own pd */
-			if (cpumask_test_cpu(cpu_of(rq), pd_span))
-				continue;
-
-			cpu = cpumask_any(pd_span);
-			pd_cap_orig = arch_scale_cpu_capacity(cpu);
-
-			if (capacity_orig < pd_cap_orig)
-				continue;
-
-			/*
-			 * handle the case of multiple perf domains have the
-			 * same capacity_orig but one of them is under higher
-			 * thermal pressure. We record it as capacity
-			 * inversion.
-			 */
-			if (capacity_orig == pd_cap_orig) {
-				pd_cap = pd_cap_orig - thermal_load_avg(cpu_rq(cpu));
-
-				if (pd_cap > inv_cap) {
-					rq->cpu_capacity_inverted = inv_cap;
-					break;
-				}
-			} else if (pd_cap_orig > inv_cap) {
-				rq->cpu_capacity_inverted = inv_cap;
-				break;
-			}
-		}
-
-		rcu_read_unlock();
-	}
-
-	trace_sched_cpu_capacity_tp(rq);
+	cpu_rq(cpu)->cpu_capacity = capacity;
+	trace_sched_cpu_capacity_tp(cpu_rq(cpu));
 
 	sdg->sgc->capacity = capacity;
 	sdg->sgc->min_capacity = capacity;
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 1072502976df..3e8df6d31c1e 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1044,7 +1044,6 @@ struct rq {
 
 	unsigned long		cpu_capacity;
 	unsigned long		cpu_capacity_orig;
-	unsigned long		cpu_capacity_inverted;
 
 	struct balance_callback *balance_callback;
 
@@ -2899,24 +2898,6 @@ static inline unsigned long capacity_orig_of(int cpu)
 	return cpu_rq(cpu)->cpu_capacity_orig;
 }
 
-/*
- * Returns inverted capacity if the CPU is in capacity inversion state.
- * 0 otherwise.
- *
- * Capacity inversion detection only considers thermal impact where actual
- * performance points (OPPs) gets dropped.
- *
- * Capacity inversion state happens when another performance domain that has
- * equal or lower capacity_orig_of() becomes effectively larger than the perf
- * domain this CPU belongs to due to thermal pressure throttling it hard.
- *
- * See comment in update_cpu_capacity().
- */
-static inline unsigned long cpu_in_capacity_inversion(int cpu)
-{
-	return cpu_rq(cpu)->cpu_capacity_inverted;
-}
-
 /**
  * enum cpu_util_type - CPU utilization type
  * @FREQUENCY_UTIL:	Utilization used to select frequency
-- 
2.34.1

On 02/01/23 15:36, Vincent Guittot wrote:
> Remove the capacity inversion detection which is now handled by
> util_fits_cpu() returning -1 when we need to continue to look for a
> potential CPU with better performance.
> 
> This ends up almost reverting patches below except for some comments:

nit: I think this comment must be removed/reworeded though

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 691a2f9c4efa..c6c8e7f52935 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4476,10 +4476,6 @@ static inline int util_fits_cpu(unsigned long util,
         *
         * For uclamp_max, we can tolerate a drop in performance level as the
         * goal is to cap the task. So it's okay if it's getting less.
-        *
-        * In case of capacity inversion, which is not handled yet, we should
-        * honour the inverted capacity for both uclamp_min and uclamp_max all
-        * the time.
         */
        capacity_orig = capacity_orig_of(cpu);
        capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);

> commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection")
> commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()")
> commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
> 
> Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>

Apart from that, LGTM.

Reviewed-by: Qais Yousef <qyousef@layalina.io>


Thanks!

--
Qais Yousef

> ---
>  kernel/sched/fair.c  | 84 +++-----------------------------------------
>  kernel/sched/sched.h | 19 ----------
>  2 files changed, 5 insertions(+), 98 deletions(-)
> 
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 074742f107c0..c6c8e7f52935 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -4476,17 +4476,9 @@ static inline int util_fits_cpu(unsigned long util,
>  	 *
>  	 * For uclamp_max, we can tolerate a drop in performance level as the
>  	 * goal is to cap the task. So it's okay if it's getting less.
> -	 *
> -	 * In case of capacity inversion we should honour the inverted capacity
> -	 * for both uclamp_min and uclamp_max all the time.
>  	 */
> -	capacity_orig = cpu_in_capacity_inversion(cpu);
> -	if (capacity_orig) {
> -		capacity_orig_thermal = capacity_orig;
> -	} else {
> -		capacity_orig = capacity_orig_of(cpu);
> -		capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
> -	}
> +	capacity_orig = capacity_orig_of(cpu);
> +	capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
>  
>  	/*
>  	 * We want to force a task to fit a cpu as implied by uclamp_max.
> @@ -9027,82 +9019,16 @@ static unsigned long scale_rt_capacity(int cpu)
>  
>  static void update_cpu_capacity(struct sched_domain *sd, int cpu)
>  {
> -	unsigned long capacity_orig = arch_scale_cpu_capacity(cpu);
>  	unsigned long capacity = scale_rt_capacity(cpu);
>  	struct sched_group *sdg = sd->groups;
> -	struct rq *rq = cpu_rq(cpu);
>  
> -	rq->cpu_capacity_orig = capacity_orig;
> +	cpu_rq(cpu)->cpu_capacity_orig = arch_scale_cpu_capacity(cpu);
>  
>  	if (!capacity)
>  		capacity = 1;
>  
> -	rq->cpu_capacity = capacity;
> -
> -	/*
> -	 * Detect if the performance domain is in capacity inversion state.
> -	 *
> -	 * Capacity inversion happens when another perf domain with equal or
> -	 * lower capacity_orig_of() ends up having higher capacity than this
> -	 * domain after subtracting thermal pressure.
> -	 *
> -	 * We only take into account thermal pressure in this detection as it's
> -	 * the only metric that actually results in *real* reduction of
> -	 * capacity due to performance points (OPPs) being dropped/become
> -	 * unreachable due to thermal throttling.
> -	 *
> -	 * We assume:
> -	 *   * That all cpus in a perf domain have the same capacity_orig
> -	 *     (same uArch).
> -	 *   * Thermal pressure will impact all cpus in this perf domain
> -	 *     equally.
> -	 */
> -	if (sched_energy_enabled()) {
> -		unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
> -		struct perf_domain *pd;
> -
> -		rcu_read_lock();
> -
> -		pd = rcu_dereference(rq->rd->pd);
> -		rq->cpu_capacity_inverted = 0;
> -
> -		for (; pd; pd = pd->next) {
> -			struct cpumask *pd_span = perf_domain_span(pd);
> -			unsigned long pd_cap_orig, pd_cap;
> -
> -			/* We can't be inverted against our own pd */
> -			if (cpumask_test_cpu(cpu_of(rq), pd_span))
> -				continue;
> -
> -			cpu = cpumask_any(pd_span);
> -			pd_cap_orig = arch_scale_cpu_capacity(cpu);
> -
> -			if (capacity_orig < pd_cap_orig)
> -				continue;
> -
> -			/*
> -			 * handle the case of multiple perf domains have the
> -			 * same capacity_orig but one of them is under higher
> -			 * thermal pressure. We record it as capacity
> -			 * inversion.
> -			 */
> -			if (capacity_orig == pd_cap_orig) {
> -				pd_cap = pd_cap_orig - thermal_load_avg(cpu_rq(cpu));
> -
> -				if (pd_cap > inv_cap) {
> -					rq->cpu_capacity_inverted = inv_cap;
> -					break;
> -				}
> -			} else if (pd_cap_orig > inv_cap) {
> -				rq->cpu_capacity_inverted = inv_cap;
> -				break;
> -			}
> -		}
> -
> -		rcu_read_unlock();
> -	}
> -
> -	trace_sched_cpu_capacity_tp(rq);
> +	cpu_rq(cpu)->cpu_capacity = capacity;
> +	trace_sched_cpu_capacity_tp(cpu_rq(cpu));
>  
>  	sdg->sgc->capacity = capacity;
>  	sdg->sgc->min_capacity = capacity;
> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index 1072502976df..3e8df6d31c1e 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -1044,7 +1044,6 @@ struct rq {
>  
>  	unsigned long		cpu_capacity;
>  	unsigned long		cpu_capacity_orig;
> -	unsigned long		cpu_capacity_inverted;
>  
>  	struct balance_callback *balance_callback;
>  
> @@ -2899,24 +2898,6 @@ static inline unsigned long capacity_orig_of(int cpu)
>  	return cpu_rq(cpu)->cpu_capacity_orig;
>  }
>  
> -/*
> - * Returns inverted capacity if the CPU is in capacity inversion state.
> - * 0 otherwise.
> - *
> - * Capacity inversion detection only considers thermal impact where actual
> - * performance points (OPPs) gets dropped.
> - *
> - * Capacity inversion state happens when another performance domain that has
> - * equal or lower capacity_orig_of() becomes effectively larger than the perf
> - * domain this CPU belongs to due to thermal pressure throttling it hard.
> - *
> - * See comment in update_cpu_capacity().
> - */
> -static inline unsigned long cpu_in_capacity_inversion(int cpu)
> -{
> -	return cpu_rq(cpu)->cpu_capacity_inverted;
> -}
> -
>  /**
>   * enum cpu_util_type - CPU utilization type
>   * @FREQUENCY_UTIL:	Utilization used to select frequency
> -- 
> 2.34.1
>

On Sat, 4 Feb 2023 at 19:42, Qais Yousef <qyousef@layalina.io> wrote:
>
> On 02/01/23 15:36, Vincent Guittot wrote:
> > Remove the capacity inversion detection which is now handled by
> > util_fits_cpu() returning -1 when we need to continue to look for a
> > potential CPU with better performance.
> >
> > This ends up almost reverting patches below except for some comments:
>
> nit: I think this comment must be removed/reworeded though

This comment has already been removed. That's why I said almost revert
except for some comments in the commit message

>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 691a2f9c4efa..c6c8e7f52935 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -4476,10 +4476,6 @@ static inline int util_fits_cpu(unsigned long util,
>          *
>          * For uclamp_max, we can tolerate a drop in performance level as the
>          * goal is to cap the task. So it's okay if it's getting less.
> -        *
> -        * In case of capacity inversion, which is not handled yet, we should
> -        * honour the inverted capacity for both uclamp_min and uclamp_max all
> -        * the time.
>          */
>         capacity_orig = capacity_orig_of(cpu);
>         capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
>
> > commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection")
> > commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()")
> > commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
> >
> > Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
>
> Apart from that, LGTM.
>
> Reviewed-by: Qais Yousef <qyousef@layalina.io>

Thanks

>
>
> Thanks!
>
> --
> Qais Yousef
>
> > ---
> >  kernel/sched/fair.c  | 84 +++-----------------------------------------
> >  kernel/sched/sched.h | 19 ----------
> >  2 files changed, 5 insertions(+), 98 deletions(-)
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index 074742f107c0..c6c8e7f52935 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -4476,17 +4476,9 @@ static inline int util_fits_cpu(unsigned long util,
> >        *
> >        * For uclamp_max, we can tolerate a drop in performance level as the
> >        * goal is to cap the task. So it's okay if it's getting less.
> > -      *
> > -      * In case of capacity inversion we should honour the inverted capacity
> > -      * for both uclamp_min and uclamp_max all the time.
> >        */
> > -     capacity_orig = cpu_in_capacity_inversion(cpu);
> > -     if (capacity_orig) {
> > -             capacity_orig_thermal = capacity_orig;
> > -     } else {
> > -             capacity_orig = capacity_orig_of(cpu);
> > -             capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
> > -     }
> > +     capacity_orig = capacity_orig_of(cpu);
> > +     capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
> >
> >       /*
> >        * We want to force a task to fit a cpu as implied by uclamp_max.
> > @@ -9027,82 +9019,16 @@ static unsigned long scale_rt_capacity(int cpu)
> >
> >  static void update_cpu_capacity(struct sched_domain *sd, int cpu)
> >  {
> > -     unsigned long capacity_orig = arch_scale_cpu_capacity(cpu);
> >       unsigned long capacity = scale_rt_capacity(cpu);
> >       struct sched_group *sdg = sd->groups;
> > -     struct rq *rq = cpu_rq(cpu);
> >
> > -     rq->cpu_capacity_orig = capacity_orig;
> > +     cpu_rq(cpu)->cpu_capacity_orig = arch_scale_cpu_capacity(cpu);
> >
> >       if (!capacity)
> >               capacity = 1;
> >
> > -     rq->cpu_capacity = capacity;
> > -
> > -     /*
> > -      * Detect if the performance domain is in capacity inversion state.
> > -      *
> > -      * Capacity inversion happens when another perf domain with equal or
> > -      * lower capacity_orig_of() ends up having higher capacity than this
> > -      * domain after subtracting thermal pressure.
> > -      *
> > -      * We only take into account thermal pressure in this detection as it's
> > -      * the only metric that actually results in *real* reduction of
> > -      * capacity due to performance points (OPPs) being dropped/become
> > -      * unreachable due to thermal throttling.
> > -      *
> > -      * We assume:
> > -      *   * That all cpus in a perf domain have the same capacity_orig
> > -      *     (same uArch).
> > -      *   * Thermal pressure will impact all cpus in this perf domain
> > -      *     equally.
> > -      */
> > -     if (sched_energy_enabled()) {
> > -             unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
> > -             struct perf_domain *pd;
> > -
> > -             rcu_read_lock();
> > -
> > -             pd = rcu_dereference(rq->rd->pd);
> > -             rq->cpu_capacity_inverted = 0;
> > -
> > -             for (; pd; pd = pd->next) {
> > -                     struct cpumask *pd_span = perf_domain_span(pd);
> > -                     unsigned long pd_cap_orig, pd_cap;
> > -
> > -                     /* We can't be inverted against our own pd */
> > -                     if (cpumask_test_cpu(cpu_of(rq), pd_span))
> > -                             continue;
> > -
> > -                     cpu = cpumask_any(pd_span);
> > -                     pd_cap_orig = arch_scale_cpu_capacity(cpu);
> > -
> > -                     if (capacity_orig < pd_cap_orig)
> > -                             continue;
> > -
> > -                     /*
> > -                      * handle the case of multiple perf domains have the
> > -                      * same capacity_orig but one of them is under higher
> > -                      * thermal pressure. We record it as capacity
> > -                      * inversion.
> > -                      */
> > -                     if (capacity_orig == pd_cap_orig) {
> > -                             pd_cap = pd_cap_orig - thermal_load_avg(cpu_rq(cpu));
> > -
> > -                             if (pd_cap > inv_cap) {
> > -                                     rq->cpu_capacity_inverted = inv_cap;
> > -                                     break;
> > -                             }
> > -                     } else if (pd_cap_orig > inv_cap) {
> > -                             rq->cpu_capacity_inverted = inv_cap;
> > -                             break;
> > -                     }
> > -             }
> > -
> > -             rcu_read_unlock();
> > -     }
> > -
> > -     trace_sched_cpu_capacity_tp(rq);
> > +     cpu_rq(cpu)->cpu_capacity = capacity;
> > +     trace_sched_cpu_capacity_tp(cpu_rq(cpu));
> >
> >       sdg->sgc->capacity = capacity;
> >       sdg->sgc->min_capacity = capacity;
> > diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> > index 1072502976df..3e8df6d31c1e 100644
> > --- a/kernel/sched/sched.h
> > +++ b/kernel/sched/sched.h
> > @@ -1044,7 +1044,6 @@ struct rq {
> >
> >       unsigned long           cpu_capacity;
> >       unsigned long           cpu_capacity_orig;
> > -     unsigned long           cpu_capacity_inverted;
> >
> >       struct balance_callback *balance_callback;
> >
> > @@ -2899,24 +2898,6 @@ static inline unsigned long capacity_orig_of(int cpu)
> >       return cpu_rq(cpu)->cpu_capacity_orig;
> >  }
> >
> > -/*
> > - * Returns inverted capacity if the CPU is in capacity inversion state.
> > - * 0 otherwise.
> > - *
> > - * Capacity inversion detection only considers thermal impact where actual
> > - * performance points (OPPs) gets dropped.
> > - *
> > - * Capacity inversion state happens when another performance domain that has
> > - * equal or lower capacity_orig_of() becomes effectively larger than the perf
> > - * domain this CPU belongs to due to thermal pressure throttling it hard.
> > - *
> > - * See comment in update_cpu_capacity().
> > - */
> > -static inline unsigned long cpu_in_capacity_inversion(int cpu)
> > -{
> > -     return cpu_rq(cpu)->cpu_capacity_inverted;
> > -}
> > -
> >  /**
> >   * enum cpu_util_type - CPU utilization type
> >   * @FREQUENCY_UTIL:  Utilization used to select frequency
> > --
> > 2.34.1
> >

On 02/05/23 18:29, Vincent Guittot wrote:
> On Sat, 4 Feb 2023 at 19:42, Qais Yousef <qyousef@layalina.io> wrote:
> >
> > On 02/01/23 15:36, Vincent Guittot wrote:
> > > Remove the capacity inversion detection which is now handled by
> > > util_fits_cpu() returning -1 when we need to continue to look for a
> > > potential CPU with better performance.
> > >
> > > This ends up almost reverting patches below except for some comments:
> >
> > nit: I think this comment must be removed/reworeded though
> 
> This comment has already been removed. That's why I said almost revert
> except for some comments in the commit message

Oh, my diff cmd had the wrong order then, sorry.

> 
> >
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index 691a2f9c4efa..c6c8e7f52935 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -4476,10 +4476,6 @@ static inline int util_fits_cpu(unsigned long util,
> >          *
> >          * For uclamp_max, we can tolerate a drop in performance level as the
> >          * goal is to cap the task. So it's okay if it's getting less.
> > -        *
> > -        * In case of capacity inversion, which is not handled yet, we should
> > -        * honour the inverted capacity for both uclamp_min and uclamp_max all
> > -        * the time.
> >          */
> >         capacity_orig = capacity_orig_of(cpu);
> >         capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
> >
> > > commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection")
> > > commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()")
> > > commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
> > >
> > > Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
> >
> > Apart from that, LGTM.
> >
> > Reviewed-by: Qais Yousef <qyousef@layalina.io>
> 
> Thanks
> 
> >
> >
> > Thanks!
> >
> > --
> > Qais Yousef
> >
> > > ---
> > >  kernel/sched/fair.c  | 84 +++-----------------------------------------
> > >  kernel/sched/sched.h | 19 ----------
> > >  2 files changed, 5 insertions(+), 98 deletions(-)
> > >
> > > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > > index 074742f107c0..c6c8e7f52935 100644
> > > --- a/kernel/sched/fair.c
> > > +++ b/kernel/sched/fair.c
> > > @@ -4476,17 +4476,9 @@ static inline int util_fits_cpu(unsigned long util,
> > >        *
> > >        * For uclamp_max, we can tolerate a drop in performance level as the
> > >        * goal is to cap the task. So it's okay if it's getting less.
> > > -      *
> > > -      * In case of capacity inversion we should honour the inverted capacity
> > > -      * for both uclamp_min and uclamp_max all the time.
> > >        */
> > > -     capacity_orig = cpu_in_capacity_inversion(cpu);
> > > -     if (capacity_orig) {
> > > -             capacity_orig_thermal = capacity_orig;
> > > -     } else {
> > > -             capacity_orig = capacity_orig_of(cpu);
> > > -             capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
> > > -     }
> > > +     capacity_orig = capacity_orig_of(cpu);
> > > +     capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
> > >
> > >       /*
> > >        * We want to force a task to fit a cpu as implied by uclamp_max.
> > > @@ -9027,82 +9019,16 @@ static unsigned long scale_rt_capacity(int cpu)
> > >
> > >  static void update_cpu_capacity(struct sched_domain *sd, int cpu)
> > >  {
> > > -     unsigned long capacity_orig = arch_scale_cpu_capacity(cpu);
> > >       unsigned long capacity = scale_rt_capacity(cpu);
> > >       struct sched_group *sdg = sd->groups;
> > > -     struct rq *rq = cpu_rq(cpu);
> > >
> > > -     rq->cpu_capacity_orig = capacity_orig;
> > > +     cpu_rq(cpu)->cpu_capacity_orig = arch_scale_cpu_capacity(cpu);
> > >
> > >       if (!capacity)
> > >               capacity = 1;
> > >
> > > -     rq->cpu_capacity = capacity;
> > > -
> > > -     /*
> > > -      * Detect if the performance domain is in capacity inversion state.
> > > -      *
> > > -      * Capacity inversion happens when another perf domain with equal or
> > > -      * lower capacity_orig_of() ends up having higher capacity than this
> > > -      * domain after subtracting thermal pressure.
> > > -      *
> > > -      * We only take into account thermal pressure in this detection as it's
> > > -      * the only metric that actually results in *real* reduction of
> > > -      * capacity due to performance points (OPPs) being dropped/become
> > > -      * unreachable due to thermal throttling.
> > > -      *
> > > -      * We assume:
> > > -      *   * That all cpus in a perf domain have the same capacity_orig
> > > -      *     (same uArch).
> > > -      *   * Thermal pressure will impact all cpus in this perf domain
> > > -      *     equally.
> > > -      */
> > > -     if (sched_energy_enabled()) {
> > > -             unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
> > > -             struct perf_domain *pd;
> > > -
> > > -             rcu_read_lock();
> > > -
> > > -             pd = rcu_dereference(rq->rd->pd);
> > > -             rq->cpu_capacity_inverted = 0;
> > > -
> > > -             for (; pd; pd = pd->next) {
> > > -                     struct cpumask *pd_span = perf_domain_span(pd);
> > > -                     unsigned long pd_cap_orig, pd_cap;
> > > -
> > > -                     /* We can't be inverted against our own pd */
> > > -                     if (cpumask_test_cpu(cpu_of(rq), pd_span))
> > > -                             continue;
> > > -
> > > -                     cpu = cpumask_any(pd_span);
> > > -                     pd_cap_orig = arch_scale_cpu_capacity(cpu);
> > > -
> > > -                     if (capacity_orig < pd_cap_orig)
> > > -                             continue;
> > > -
> > > -                     /*
> > > -                      * handle the case of multiple perf domains have the
> > > -                      * same capacity_orig but one of them is under higher
> > > -                      * thermal pressure. We record it as capacity
> > > -                      * inversion.
> > > -                      */
> > > -                     if (capacity_orig == pd_cap_orig) {
> > > -                             pd_cap = pd_cap_orig - thermal_load_avg(cpu_rq(cpu));
> > > -
> > > -                             if (pd_cap > inv_cap) {
> > > -                                     rq->cpu_capacity_inverted = inv_cap;
> > > -                                     break;
> > > -                             }
> > > -                     } else if (pd_cap_orig > inv_cap) {
> > > -                             rq->cpu_capacity_inverted = inv_cap;
> > > -                             break;
> > > -                     }
> > > -             }
> > > -
> > > -             rcu_read_unlock();
> > > -     }
> > > -
> > > -     trace_sched_cpu_capacity_tp(rq);
> > > +     cpu_rq(cpu)->cpu_capacity = capacity;
> > > +     trace_sched_cpu_capacity_tp(cpu_rq(cpu));
> > >
> > >       sdg->sgc->capacity = capacity;
> > >       sdg->sgc->min_capacity = capacity;
> > > diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> > > index 1072502976df..3e8df6d31c1e 100644
> > > --- a/kernel/sched/sched.h
> > > +++ b/kernel/sched/sched.h
> > > @@ -1044,7 +1044,6 @@ struct rq {
> > >
> > >       unsigned long           cpu_capacity;
> > >       unsigned long           cpu_capacity_orig;
> > > -     unsigned long           cpu_capacity_inverted;
> > >
> > >       struct balance_callback *balance_callback;
> > >
> > > @@ -2899,24 +2898,6 @@ static inline unsigned long capacity_orig_of(int cpu)
> > >       return cpu_rq(cpu)->cpu_capacity_orig;
> > >  }
> > >
> > > -/*
> > > - * Returns inverted capacity if the CPU is in capacity inversion state.
> > > - * 0 otherwise.
> > > - *
> > > - * Capacity inversion detection only considers thermal impact where actual
> > > - * performance points (OPPs) gets dropped.
> > > - *
> > > - * Capacity inversion state happens when another performance domain that has
> > > - * equal or lower capacity_orig_of() becomes effectively larger than the perf
> > > - * domain this CPU belongs to due to thermal pressure throttling it hard.
> > > - *
> > > - * See comment in update_cpu_capacity().
> > > - */
> > > -static inline unsigned long cpu_in_capacity_inversion(int cpu)
> > > -{
> > > -     return cpu_rq(cpu)->cpu_capacity_inverted;
> > > -}
> > > -
> > >  /**
> > >   * enum cpu_util_type - CPU utilization type
> > >   * @FREQUENCY_UTIL:  Utilization used to select frequency
> > > --
> > > 2.34.1
> > >

The following commit has been merged into the sched/core branch of tip:

Commit-ID:     a2e90611b9f425adbbfcdaa5b5e49958ddf6f61b
Gitweb:        https://git.kernel.org/tip/a2e90611b9f425adbbfcdaa5b5e49958ddf6f61b
Author:        Vincent Guittot <vincent.guittot@linaro.org>
AuthorDate:    Wed, 01 Feb 2023 15:36:28 +01:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Sat, 11 Feb 2023 11:18:09 +01:00

sched/fair: Remove capacity inversion detection

Remove the capacity inversion detection which is now handled by
util_fits_cpu() returning -1 when we need to continue to look for a
potential CPU with better performance.

This ends up almost reverting patches below except for some comments:
commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection")
commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()")
commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion")

Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230201143628.270912-3-vincent.guittot@linaro.org
---
 kernel/sched/fair.c  | 84 ++-----------------------------------------
 kernel/sched/sched.h | 19 +----------
 2 files changed, 5 insertions(+), 98 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 074742f..c6c8e7f 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4476,17 +4476,9 @@ static inline int util_fits_cpu(unsigned long util,
 	 *
 	 * For uclamp_max, we can tolerate a drop in performance level as the
 	 * goal is to cap the task. So it's okay if it's getting less.
-	 *
-	 * In case of capacity inversion we should honour the inverted capacity
-	 * for both uclamp_min and uclamp_max all the time.
 	 */
-	capacity_orig = cpu_in_capacity_inversion(cpu);
-	if (capacity_orig) {
-		capacity_orig_thermal = capacity_orig;
-	} else {
-		capacity_orig = capacity_orig_of(cpu);
-		capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
-	}
+	capacity_orig = capacity_orig_of(cpu);
+	capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu);
 
 	/*
 	 * We want to force a task to fit a cpu as implied by uclamp_max.
@@ -9027,82 +9019,16 @@ static unsigned long scale_rt_capacity(int cpu)
 
 static void update_cpu_capacity(struct sched_domain *sd, int cpu)
 {
-	unsigned long capacity_orig = arch_scale_cpu_capacity(cpu);
 	unsigned long capacity = scale_rt_capacity(cpu);
 	struct sched_group *sdg = sd->groups;
-	struct rq *rq = cpu_rq(cpu);
 
-	rq->cpu_capacity_orig = capacity_orig;
+	cpu_rq(cpu)->cpu_capacity_orig = arch_scale_cpu_capacity(cpu);
 
 	if (!capacity)
 		capacity = 1;
 
-	rq->cpu_capacity = capacity;
-
-	/*
-	 * Detect if the performance domain is in capacity inversion state.
-	 *
-	 * Capacity inversion happens when another perf domain with equal or
-	 * lower capacity_orig_of() ends up having higher capacity than this
-	 * domain after subtracting thermal pressure.
-	 *
-	 * We only take into account thermal pressure in this detection as it's
-	 * the only metric that actually results in *real* reduction of
-	 * capacity due to performance points (OPPs) being dropped/become
-	 * unreachable due to thermal throttling.
-	 *
-	 * We assume:
-	 *   * That all cpus in a perf domain have the same capacity_orig
-	 *     (same uArch).
-	 *   * Thermal pressure will impact all cpus in this perf domain
-	 *     equally.
-	 */
-	if (sched_energy_enabled()) {
-		unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
-		struct perf_domain *pd;
-
-		rcu_read_lock();
-
-		pd = rcu_dereference(rq->rd->pd);
-		rq->cpu_capacity_inverted = 0;
-
-		for (; pd; pd = pd->next) {
-			struct cpumask *pd_span = perf_domain_span(pd);
-			unsigned long pd_cap_orig, pd_cap;
-
-			/* We can't be inverted against our own pd */
-			if (cpumask_test_cpu(cpu_of(rq), pd_span))
-				continue;
-
-			cpu = cpumask_any(pd_span);
-			pd_cap_orig = arch_scale_cpu_capacity(cpu);
-
-			if (capacity_orig < pd_cap_orig)
-				continue;
-
-			/*
-			 * handle the case of multiple perf domains have the
-			 * same capacity_orig but one of them is under higher
-			 * thermal pressure. We record it as capacity
-			 * inversion.
-			 */
-			if (capacity_orig == pd_cap_orig) {
-				pd_cap = pd_cap_orig - thermal_load_avg(cpu_rq(cpu));
-
-				if (pd_cap > inv_cap) {
-					rq->cpu_capacity_inverted = inv_cap;
-					break;
-				}
-			} else if (pd_cap_orig > inv_cap) {
-				rq->cpu_capacity_inverted = inv_cap;
-				break;
-			}
-		}
-
-		rcu_read_unlock();
-	}
-
-	trace_sched_cpu_capacity_tp(rq);
+	cpu_rq(cpu)->cpu_capacity = capacity;
+	trace_sched_cpu_capacity_tp(cpu_rq(cpu));
 
 	sdg->sgc->capacity = capacity;
 	sdg->sgc->min_capacity = capacity;
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 1072502..3e8df6d 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1044,7 +1044,6 @@ struct rq {
 
 	unsigned long		cpu_capacity;
 	unsigned long		cpu_capacity_orig;
-	unsigned long		cpu_capacity_inverted;
 
 	struct balance_callback *balance_callback;
 
@@ -2899,24 +2898,6 @@ static inline unsigned long capacity_orig_of(int cpu)
 	return cpu_rq(cpu)->cpu_capacity_orig;
 }
 
-/*
- * Returns inverted capacity if the CPU is in capacity inversion state.
- * 0 otherwise.
- *
- * Capacity inversion detection only considers thermal impact where actual
- * performance points (OPPs) gets dropped.
- *
- * Capacity inversion state happens when another performance domain that has
- * equal or lower capacity_orig_of() becomes effectively larger than the perf
- * domain this CPU belongs to due to thermal pressure throttling it hard.
- *
- * See comment in update_cpu_capacity().
- */
-static inline unsigned long cpu_in_capacity_inversion(int cpu)
-{
-	return cpu_rq(cpu)->cpu_capacity_inverted;
-}
-
 /**
  * enum cpu_util_type - CPU utilization type
  * @FREQUENCY_UTIL:	Utilization used to select frequency