[PATCH] sched/core: fix sched_rt_global_validate

[PATCH] sched/core: fix sched_rt_global_validate

[Juri Lelli]

Don't compare sysctl_sched_rt_runtime against sysctl_sched_rt_period if
the former is equal to RUNTIME_INF, otherwise disabling -rt bandwidth
management always fails.

Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Juri Lelli <juri.lelli@gmail.com>
---
 kernel/sched/core.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 210a12a..5c0a304 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -7477,7 +7477,8 @@ static int sched_rt_global_validate(void)
 	if (sysctl_sched_rt_period <= 0)
 		return -EINVAL;
 
-	if (sysctl_sched_rt_runtime > sysctl_sched_rt_period)
+	if ((sysctl_sched_rt_runtime != RUNTIME_INF) &&
+		(sysctl_sched_rt_runtime > sysctl_sched_rt_period))
 		return -EINVAL;
 
 	return 0;
-- 
1.7.9.5

Re: [PATCH] sched/core: fix sched_rt_global_validate

[Henrik Austad]

On Fri, Jan 31, 2014 at 05:43:27PM +0100, Juri Lelli wrote:
> Don't compare sysctl_sched_rt_runtime against sysctl_sched_rt_period if
> the former is equal to RUNTIME_INF, otherwise disabling -rt bandwidth
> management always fails.
> 
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Signed-off-by: Juri Lelli <juri.lelli@gmail.com>
> ---
>  kernel/sched/core.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 210a12a..5c0a304 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -7477,7 +7477,8 @@ static int sched_rt_global_validate(void)
>  	if (sysctl_sched_rt_period <= 0)
>  		return -EINVAL;
>  
> -	if (sysctl_sched_rt_runtime > sysctl_sched_rt_period)
> +	if ((sysctl_sched_rt_runtime != RUNTIME_INF) &&
> +		(sysctl_sched_rt_runtime > sysctl_sched_rt_period))
>  		return -EINVAL;

Won't this be caught by the test above?

#define RUNTIME_INF ((u64)~0ULL)

which means that if sysctl_sched_rt_runtime is set to RUNTIME_INF, it will 
trigger on the previous test, and the first part of this test will always 
be true.

Or have I suffered catastrophic monday-morning braindamage?

-- 
Henrik Austad

Re: [PATCH] sched/core: fix sched_rt_global_validate

[Juri Lelli]

On 02/03/2014 10:53 AM, Henrik Austad wrote:
> On Fri, Jan 31, 2014 at 05:43:27PM +0100, Juri Lelli wrote:
>> Don't compare sysctl_sched_rt_runtime against sysctl_sched_rt_period if
>> the former is equal to RUNTIME_INF, otherwise disabling -rt bandwidth
>> management always fails.
>>
>> Cc: Ingo Molnar <mingo@redhat.com>
>> Cc: Peter Zijlstra <peterz@infradead.org>
>> Signed-off-by: Juri Lelli <juri.lelli@gmail.com>
>> ---
>>  kernel/sched/core.c |    3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
>> index 210a12a..5c0a304 100644
>> --- a/kernel/sched/core.c
>> +++ b/kernel/sched/core.c
>> @@ -7477,7 +7477,8 @@ static int sched_rt_global_validate(void)
>>  	if (sysctl_sched_rt_period <= 0)
>>  		return -EINVAL;
>>  
>> -	if (sysctl_sched_rt_runtime > sysctl_sched_rt_period)
>> +	if ((sysctl_sched_rt_runtime != RUNTIME_INF) &&
>> +		(sysctl_sched_rt_runtime > sysctl_sched_rt_period))
>>  		return -EINVAL;
> 
> Won't this be caught by the test above?
> 
> #define RUNTIME_INF ((u64)~0ULL)
> 
> which means that if sysctl_sched_rt_runtime is set to RUNTIME_INF, it will 
> trigger on the previous test, and the first part of this test will always 
> be true.
> 
> Or have I suffered catastrophic monday-morning braindamage?
> 

As I understand it. When you do

 echo -1 > /proc/sys/kernel/sched_rt_runtime_us

sysctl_sched_rt_runtime is actually set to -1 (being an int).
But then you compare it against and unsigned int, so the cast converts it to
actually be RUNTIME_INF, and thus greater than sysctl_sched_rt_period (so the
function returns -EINVAL, while you'd want it to return 0, as you are disabling
-rt throttling).

Makes sense?

Thanks,

- Juri

Re: [PATCH] sched/core: fix sched_rt_global_validate

[Henrik Austad]

[-- Attachment #1: Type: text/plain, Size: 2087 bytes --]

On Mon, Feb 03, 2014 at 11:14:05AM +0100, Juri Lelli wrote:
> On 02/03/2014 10:53 AM, Henrik Austad wrote:
> > On Fri, Jan 31, 2014 at 05:43:27PM +0100, Juri Lelli wrote:
> >> Don't compare sysctl_sched_rt_runtime against sysctl_sched_rt_period if
> >> the former is equal to RUNTIME_INF, otherwise disabling -rt bandwidth
> >> management always fails.
> >>
> >> Cc: Ingo Molnar <mingo@redhat.com>
> >> Cc: Peter Zijlstra <peterz@infradead.org>
> >> Signed-off-by: Juri Lelli <juri.lelli@gmail.com>
> >> ---
> >>  kernel/sched/core.c |    3 ++-
> >>  1 file changed, 2 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> >> index 210a12a..5c0a304 100644
> >> --- a/kernel/sched/core.c
> >> +++ b/kernel/sched/core.c
> >> @@ -7477,7 +7477,8 @@ static int sched_rt_global_validate(void)
> >>  	if (sysctl_sched_rt_period <= 0)
> >>  		return -EINVAL;
> >>  
> >> -	if (sysctl_sched_rt_runtime > sysctl_sched_rt_period)
> >> +	if ((sysctl_sched_rt_runtime != RUNTIME_INF) &&
> >> +		(sysctl_sched_rt_runtime > sysctl_sched_rt_period))
> >>  		return -EINVAL;
> > 
> > Won't this be caught by the test above?
> > 
> > #define RUNTIME_INF ((u64)~0ULL)
> > 
> > which means that if sysctl_sched_rt_runtime is set to RUNTIME_INF, it will 
> > trigger on the previous test, and the first part of this test will always 
> > be true.
> > 
> > Or have I suffered catastrophic monday-morning braindamage?
> > 
> 
> As I understand it. When you do
> 
>  echo -1 > /proc/sys/kernel/sched_rt_runtime_us
> 
> sysctl_sched_rt_runtime is actually set to -1 (being an int).

Yes.

> But then you compare it against and unsigned int, so the cast converts it to
> actually be RUNTIME_INF, and thus greater than sysctl_sched_rt_period (so the
> function returns -EINVAL, while you'd want it to return 0, as you are disabling
> -rt throttling).

Ah, yes, it comes down to my early-monday brain hemorrhage, mixing up 
rt_period and rt_runtime. My apologies!

> Makes sense?

Yes it does.

-- 
Henrik Austad

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]