[PATCH] sched/uclamp: Fix overzealous type replacement

[PATCH] sched/uclamp: Fix overzealous type replacement

[Valentin Schneider]

Some uclamp helpers had their return type changed from 'unsigned int' to
'enum uclamp_id' by commit

  0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")

but it happens that some *actually* do return an unsigned int value. Those
are the helpers that return a utilization value: uclamp_rq_max_value() and
uclamp_eff_value(). Fix those up.

Note that this doesn't lead to any obj diff using a relatively recent
aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
already figures out that the return value is 11 bits (bits_per(1024)) and
doesn't seem to do anything funny. I'm still marking this as fixing the
above commit to be on the safe side.

Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
---
 kernel/sched/core.c  | 4 ++--
 kernel/sched/sched.h | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 513a4794ff36..71a45025cd2e 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -853,7 +853,7 @@ static inline void uclamp_idle_reset(struct rq *rq, enum uclamp_id clamp_id,
 }
 
 static inline
-enum uclamp_id uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
+unsigned int uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
 				   unsigned int clamp_value)
 {
 	struct uclamp_bucket *bucket = rq->uclamp[clamp_id].bucket;
@@ -918,7 +918,7 @@ uclamp_eff_get(struct task_struct *p, enum uclamp_id clamp_id)
 	return uc_req;
 }
 
-enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
+unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
 {
 	struct uclamp_se uc_eff;
 
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 05c282775f21..280a3c735935 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2300,7 +2300,7 @@ static inline void cpufreq_update_util(struct rq *rq, unsigned int flags) {}
 #endif /* CONFIG_CPU_FREQ */
 
 #ifdef CONFIG_UCLAMP_TASK
-enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id);
+unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id);
 
 static __always_inline
 unsigned int uclamp_util_with(struct rq *rq, unsigned int util,
-- 
2.22.0

Re: [PATCH] sched/uclamp: Fix overzealous type replacement

[Qais Yousef]

On 11/13/19 16:53, Valentin Schneider wrote:
> Some uclamp helpers had their return type changed from 'unsigned int' to
> 'enum uclamp_id' by commit
> 
>   0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> 
> but it happens that some *actually* do return an unsigned int value. Those
> are the helpers that return a utilization value: uclamp_rq_max_value() and
> uclamp_eff_value(). Fix those up.
> 
> Note that this doesn't lead to any obj diff using a relatively recent
> aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
> already figures out that the return value is 11 bits (bits_per(1024)) and
> doesn't seem to do anything funny. I'm still marking this as fixing the
> above commit to be on the safe side.
> 
> Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
> ---

The changelog could be a bit simpler and more explicitly say 0413d7f33e60
wrongly changed the return type of some functions. For a second I thought
something weird is going inside these functions.

But that could be just me :-)

Reviewed-by: Qais Yousef <qais.yousef@arm.com>

Thanks!

--
Qais Yousef

Re: [PATCH] sched/uclamp: Fix overzealous type replacement

[Valentin Schneider]

On 14/11/2019 20:28, Qais Yousef wrote:
> On 11/13/19 16:53, Valentin Schneider wrote:
>> Some uclamp helpers had their return type changed from 'unsigned int' to
>> 'enum uclamp_id' by commit
>>
>>   0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
>>
>> but it happens that some *actually* do return an unsigned int value. Those
>> are the helpers that return a utilization value: uclamp_rq_max_value() and
>> uclamp_eff_value(). Fix those up.
>>
>> Note that this doesn't lead to any obj diff using a relatively recent
>> aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
>> already figures out that the return value is 11 bits (bits_per(1024)) and
>> doesn't seem to do anything funny. I'm still marking this as fixing the
>> above commit to be on the safe side.
>>
>> Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
>> Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
>> ---
> 
> The changelog could be a bit simpler and more explicitly say 0413d7f33e60
> wrongly changed the return type of some functions. For a second I thought
> something weird is going inside these functions.
> 

The first paragraph is exactly that, no? The rest (that starts with "Note
that") is just optional stuff I look into because I was curious.

> But that could be just me :-)
> 
> Reviewed-by: Qais Yousef <qais.yousef@arm.com>
> 
> Thanks!
> 
> --
> Qais Yousef
> 

Re: [PATCH] sched/uclamp: Fix overzealous type replacement

[Vincent Guittot]

On Wed, 13 Nov 2019 at 17:55, Valentin Schneider
<valentin.schneider@arm.com> wrote:
>
> Some uclamp helpers had their return type changed from 'unsigned int' to
> 'enum uclamp_id' by commit
>
>   0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
>
> but it happens that some *actually* do return an unsigned int value. Those
> are the helpers that return a utilization value: uclamp_rq_max_value() and
> uclamp_eff_value(). Fix those up.
>
> Note that this doesn't lead to any obj diff using a relatively recent
> aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
> already figures out that the return value is 11 bits (bits_per(1024)) and
> doesn't seem to do anything funny. I'm still marking this as fixing the
> above commit to be on the safe side.
>
> Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
> ---
>  kernel/sched/core.c  | 4 ++--
>  kernel/sched/sched.h | 2 +-
>  2 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 513a4794ff36..71a45025cd2e 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -853,7 +853,7 @@ static inline void uclamp_idle_reset(struct rq *rq, enum uclamp_id clamp_id,
>  }
>
>  static inline
> -enum uclamp_id uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
> +unsigned int uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
>                                    unsigned int clamp_value)
>  {
>         struct uclamp_bucket *bucket = rq->uclamp[clamp_id].bucket;
> @@ -918,7 +918,7 @@ uclamp_eff_get(struct task_struct *p, enum uclamp_id clamp_id)
>         return uc_req;
>  }
>
> -enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
> +unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
>  {
>         struct uclamp_se uc_eff;
>

And static inline enum uclamp_id uclamp_none(enum uclamp_id clamp_id) ?

Should it be fixed as well as it can return SCHED_CAPACITY_SCALE ?



> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index 05c282775f21..280a3c735935 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -2300,7 +2300,7 @@ static inline void cpufreq_update_util(struct rq *rq, unsigned int flags) {}
>  #endif /* CONFIG_CPU_FREQ */
>
>  #ifdef CONFIG_UCLAMP_TASK
> -enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id);
> +unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id);
>
>  static __always_inline
>  unsigned int uclamp_util_with(struct rq *rq, unsigned int util,
> --
> 2.22.0
>

Re: [PATCH] sched/uclamp: Fix overzealous type replacement

[Qais Yousef]

On 11/15/19 00:05, Valentin Schneider wrote:
> On 14/11/2019 20:28, Qais Yousef wrote:
> > On 11/13/19 16:53, Valentin Schneider wrote:
> >> Some uclamp helpers had their return type changed from 'unsigned int' to
> >> 'enum uclamp_id' by commit
> >>
> >>   0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> >>
> >> but it happens that some *actually* do return an unsigned int value. Those
> >> are the helpers that return a utilization value: uclamp_rq_max_value() and
> >> uclamp_eff_value(). Fix those up.
> >>
> >> Note that this doesn't lead to any obj diff using a relatively recent
> >> aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
> >> already figures out that the return value is 11 bits (bits_per(1024)) and
> >> doesn't seem to do anything funny. I'm still marking this as fixing the
> >> above commit to be on the safe side.
> >>
> >> Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> >> Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
> >> ---
> > 
> > The changelog could be a bit simpler and more explicitly say 0413d7f33e60
> > wrongly changed the return type of some functions. For a second I thought
> > something weird is going inside these functions.
> > 
> 
> The first paragraph is exactly that, no? The rest (that starts with "Note
> that") is just optional stuff I look into because I was curious.

The way it read to me is that the function was returning uclamp_id as unsigned
int, hence my confusion/comment. Anyway, it's not a big deal. It's not really
a problem.

Thanks

--
Qais Yousef

Re: [PATCH] sched/uclamp: Fix overzealous type replacement

[Qais Yousef]

On 11/15/19 09:12, Vincent Guittot wrote:
> On Wed, 13 Nov 2019 at 17:55, Valentin Schneider
> <valentin.schneider@arm.com> wrote:
> >
> > Some uclamp helpers had their return type changed from 'unsigned int' to
> > 'enum uclamp_id' by commit
> >
> >   0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> >
> > but it happens that some *actually* do return an unsigned int value. Those
> > are the helpers that return a utilization value: uclamp_rq_max_value() and
> > uclamp_eff_value(). Fix those up.
> >
> > Note that this doesn't lead to any obj diff using a relatively recent
> > aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
> > already figures out that the return value is 11 bits (bits_per(1024)) and
> > doesn't seem to do anything funny. I'm still marking this as fixing the
> > above commit to be on the safe side.
> >
> > Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
> > Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
> > ---
> >  kernel/sched/core.c  | 4 ++--
> >  kernel/sched/sched.h | 2 +-
> >  2 files changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > index 513a4794ff36..71a45025cd2e 100644
> > --- a/kernel/sched/core.c
> > +++ b/kernel/sched/core.c
> > @@ -853,7 +853,7 @@ static inline void uclamp_idle_reset(struct rq *rq, enum uclamp_id clamp_id,
> >  }
> >
> >  static inline
> > -enum uclamp_id uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
> > +unsigned int uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
> >                                    unsigned int clamp_value)
> >  {
> >         struct uclamp_bucket *bucket = rq->uclamp[clamp_id].bucket;
> > @@ -918,7 +918,7 @@ uclamp_eff_get(struct task_struct *p, enum uclamp_id clamp_id)
> >         return uc_req;
> >  }
> >
> > -enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
> > +unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
> >  {
> >         struct uclamp_se uc_eff;
> >
> 
> And static inline enum uclamp_id uclamp_none(enum uclamp_id clamp_id) ?
> 
> Should it be fixed as well as it can return SCHED_CAPACITY_SCALE ?

Indeed. That should return unsigned int too as it's returning the uclamp value.

Thanks

--
Qais Yousef

Re: [PATCH] sched/uclamp: Fix overzealous type replacement

[Valentin Schneider]

On 15/11/2019 08:12, Vincent Guittot wrote:
> 
> And static inline enum uclamp_id uclamp_none(enum uclamp_id clamp_id) ?
> 
> Should it be fixed as well as it can return SCHED_CAPACITY_SCALE ?
> 

Right, thanks for staring at this. I'll go double check the diff and fix up
any strays.

[tip: sched/urgent] sched/uclamp: Fix overzealous type replacement

[tip-bot2 for Valentin Schneider]

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

Commit-ID:     16d6bbe7982e99f7f326ec6f088c0aaf85efa69d
Gitweb:        https://git.kernel.org/tip/16d6bbe7982e99f7f326ec6f088c0aaf85efa69d
Author:        Valentin Schneider <valentin.schneider@arm.com>
AuthorDate:    Wed, 13 Nov 2019 16:53:34 
Committer:     Ingo Molnar <mingo@kernel.org>
CommitterDate: Fri, 15 Nov 2019 11:02:19 +01:00

sched/uclamp: Fix overzealous type replacement

Some uclamp helpers had their return type changed from 'unsigned int' to
'enum uclamp_id' by commit:

  0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")

but it happens that some *actually* do return an unsigned int value. Those
are the helpers that return a utilization value: uclamp_rq_max_value() and
uclamp_eff_value(). Fix those up.

Note that this doesn't lead to any obj diff using a relatively recent
aarch64 compiler (8.3-2019.03). The current code of e.g. uclamp_eff_value()
already figures out that the return value is 11 bits (bits_per(1024)) and
doesn't seem to do anything funny. I'm still marking this as fixing the
above commit to be on the safe side.

Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Qais Yousef <qais.yousef@arm.com>
Cc: Dietmar.Eggemann@arm.com
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: patrick.bellasi@matbug.net
Cc: qperret@google.com
Cc: surenb@google.com
Cc: tj@kernel.org
Cc: vincent.guittot@linaro.org
Fixes: 0413d7f33e60 ("sched/uclamp: Always use 'enum uclamp_id' for clamp_id values")
Link: https://lkml.kernel.org/r/20191113165334.14291-1-valentin.schneider@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 kernel/sched/core.c  | 4 ++--
 kernel/sched/sched.h | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 44123b4..a4f76d3 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -853,7 +853,7 @@ static inline void uclamp_idle_reset(struct rq *rq, enum uclamp_id clamp_id,
 }
 
 static inline
-enum uclamp_id uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
+unsigned int uclamp_rq_max_value(struct rq *rq, enum uclamp_id clamp_id,
 				   unsigned int clamp_value)
 {
 	struct uclamp_bucket *bucket = rq->uclamp[clamp_id].bucket;
@@ -918,7 +918,7 @@ uclamp_eff_get(struct task_struct *p, enum uclamp_id clamp_id)
 	return uc_req;
 }
 
-enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
+unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id)
 {
 	struct uclamp_se uc_eff;
 
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index c8870c5..49ed949 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2309,7 +2309,7 @@ static inline void cpufreq_update_util(struct rq *rq, unsigned int flags) {}
 #endif /* CONFIG_CPU_FREQ */
 
 #ifdef CONFIG_UCLAMP_TASK
-enum uclamp_id uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id);
+unsigned int uclamp_eff_value(struct task_struct *p, enum uclamp_id clamp_id);
 
 static __always_inline
 unsigned int uclamp_util_with(struct rq *rq, unsigned int util,