Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On 11/07/17 18:18, Peter Zijlstra wrote: > On Wed, Jul 05, 2017 at 09:59:00AM +0100, Juri Lelli wrote: > > @@ -4065,6 +4067,9 @@ static int __sched_setscheduler(struct task_struct *p, > > } > > > > if (user) { > > + if (attr->sched_flags & SCHED_FLAG_SPECIAL) > > +

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On 11/07/17 18:18, Peter Zijlstra wrote: > On Wed, Jul 05, 2017 at 09:59:00AM +0100, Juri Lelli wrote: > > @@ -4065,6 +4067,9 @@ static int __sched_setscheduler(struct task_struct *p, > > } > > > > if (user) { > > + if (attr->sched_flags & SCHED_FLAG_SPECIAL) > > +

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Wed, Jul 05, 2017 at 09:59:00AM +0100, Juri Lelli wrote: > @@ -4065,6 +4067,9 @@ static int __sched_setscheduler(struct task_struct *p, > } > > if (user) { > + if (attr->sched_flags & SCHED_FLAG_SPECIAL) > + return -EPERM; Should be -EINVAL I

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Wed, Jul 05, 2017 at 09:59:00AM +0100, Juri Lelli wrote: > @@ -4065,6 +4067,9 @@ static int __sched_setscheduler(struct task_struct *p, > } > > if (user) { > + if (attr->sched_flags & SCHED_FLAG_SPECIAL) > + return -EPERM; Should be -EINVAL I

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Thu, Jul 06, 2017 at 08:56:32PM -0700, Joel Fernandes wrote: > > @@ -3998,7 +3998,9 @@ static int __sched_setscheduler(struct task_struct *p, > > } > > > > if (attr->sched_flags & > > - ~(SCHED_FLAG_RESET_ON_FORK | SCHED_FLAG_RECLAIM)) > > +

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Thu, Jul 06, 2017 at 08:56:32PM -0700, Joel Fernandes wrote: > > @@ -3998,7 +3998,9 @@ static int __sched_setscheduler(struct task_struct *p, > > } > > > > if (attr->sched_flags & > > - ~(SCHED_FLAG_RESET_ON_FORK | SCHED_FLAG_RECLAIM)) > > +

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, Jul 7, 2017 at 3:15 PM, Steven Rostedt wrote: > On Fri, 7 Jul 2017 15:07:09 -0700 > Joel Fernandes wrote: > > >> > It is rather long. Although I actually hate the SUGOV, it is easily >> > grepable. Just comment what it stands for. We can always

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, Jul 7, 2017 at 3:15 PM, Steven Rostedt wrote: > On Fri, 7 Jul 2017 15:07:09 -0700 > Joel Fernandes wrote: > > >> > It is rather long. Although I actually hate the SUGOV, it is easily >> > grepable. Just comment what it stands for. We can always change it >> > later. >> >> I was thinking

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, 7 Jul 2017 15:07:09 -0700 Joel Fernandes wrote: > > It is rather long. Although I actually hate the SUGOV, it is easily > > grepable. Just comment what it stands for. We can always change it > > later. > > I was thinking why not just SCHED_FLAG_CPUFREQ. That says

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, 7 Jul 2017 15:07:09 -0700 Joel Fernandes wrote: > > It is rather long. Although I actually hate the SUGOV, it is easily > > grepable. Just comment what it stands for. We can always change it > > later. > > I was thinking why not just SCHED_FLAG_CPUFREQ. That says its for > cpufreq

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Hi, On Fri, Jul 7, 2017 at 2:58 PM, Steven Rostedt wrote: > On Fri, 07 Jul 2017 15:11:45 +0200 > "Rafael J. Wysocki" wrote: > >> On Friday, July 07, 2017 11:53:16 AM Juri Lelli wrote: >> > On 07/07/17 12:46, Thomas Gleixner wrote: >> > > On Fri, 7 Jul

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Hi, On Fri, Jul 7, 2017 at 2:58 PM, Steven Rostedt wrote: > On Fri, 07 Jul 2017 15:11:45 +0200 > "Rafael J. Wysocki" wrote: > >> On Friday, July 07, 2017 11:53:16 AM Juri Lelli wrote: >> > On 07/07/17 12:46, Thomas Gleixner wrote: >> > > On Fri, 7 Jul 2017, Juri Lelli wrote: >> > > > How about

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, 07 Jul 2017 15:11:45 +0200 "Rafael J. Wysocki" wrote: > On Friday, July 07, 2017 11:53:16 AM Juri Lelli wrote: > > On 07/07/17 12:46, Thomas Gleixner wrote: > > > On Fri, 7 Jul 2017, Juri Lelli wrote: > > > > How about SCHED_FLAG_SUGOV then? > > > > > > I know

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, 07 Jul 2017 15:11:45 +0200 "Rafael J. Wysocki" wrote: > On Friday, July 07, 2017 11:53:16 AM Juri Lelli wrote: > > On 07/07/17 12:46, Thomas Gleixner wrote: > > > On Fri, 7 Jul 2017, Juri Lelli wrote: > > > > How about SCHED_FLAG_SUGOV then? > > > > > > I know sugo della carne,

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Friday, July 07, 2017 11:53:16 AM Juri Lelli wrote: > On 07/07/17 12:46, Thomas Gleixner wrote: > > On Fri, 7 Jul 2017, Juri Lelli wrote: > > > How about SCHED_FLAG_SUGOV then? > > > > I know sugo della carne, but what's sugo V? > > > > Right.. can't really help not thinking about the same

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Friday, July 07, 2017 11:53:16 AM Juri Lelli wrote: > On 07/07/17 12:46, Thomas Gleixner wrote: > > On Fri, 7 Jul 2017, Juri Lelli wrote: > > > How about SCHED_FLAG_SUGOV then? > > > > I know sugo della carne, but what's sugo V? > > > > Right.. can't really help not thinking about the same

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On 07/07/17 12:46, Thomas Gleixner wrote: > On Fri, 7 Jul 2017, Juri Lelli wrote: > > How about SCHED_FLAG_SUGOV then? > > I know sugo della carne, but what's sugo V? > Right.. can't really help not thinking about the same (especially close to lunch) :) But the abbreviation stands for

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On 07/07/17 12:46, Thomas Gleixner wrote: > On Fri, 7 Jul 2017, Juri Lelli wrote: > > How about SCHED_FLAG_SUGOV then? > > I know sugo della carne, but what's sugo V? > Right.. can't really help not thinking about the same (especially close to lunch) :) But the abbreviation stands for

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, 7 Jul 2017, Juri Lelli wrote: > How about SCHED_FLAG_SUGOV then? I know sugo della carne, but what's sugo V? Thanks, tglx

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On Fri, 7 Jul 2017, Juri Lelli wrote: > How about SCHED_FLAG_SUGOV then? I know sugo della carne, but what's sugo V? Thanks, tglx

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Hi, On 06/07/17 20:56, Joel Fernandes wrote: > Hi Juri, > > On Wed, Jul 5, 2017 at 1:59 AM, Juri Lelli wrote: [...] > > if (attr->sched_flags & > > - ~(SCHED_FLAG_RESET_ON_FORK | SCHED_FLAG_RECLAIM)) > > + ~(SCHED_FLAG_RESET_ON_FORK | >

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Hi, On 06/07/17 20:56, Joel Fernandes wrote: > Hi Juri, > > On Wed, Jul 5, 2017 at 1:59 AM, Juri Lelli wrote: [...] > > if (attr->sched_flags & > > - ~(SCHED_FLAG_RESET_ON_FORK | SCHED_FLAG_RECLAIM)) > > + ~(SCHED_FLAG_RESET_ON_FORK | > > +

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On 05-07-17, 09:59, Juri Lelli wrote: > diff --git a/kernel/sched/cpufreq_schedutil.c > b/kernel/sched/cpufreq_schedutil.c > index f2494d1fc8ef..ba6227625f24 100644 > --- a/kernel/sched/cpufreq_schedutil.c > +++ b/kernel/sched/cpufreq_schedutil.c > @@ -424,7 +424,16 @@ static void

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

On 05-07-17, 09:59, Juri Lelli wrote: > diff --git a/kernel/sched/cpufreq_schedutil.c > b/kernel/sched/cpufreq_schedutil.c > index f2494d1fc8ef..ba6227625f24 100644 > --- a/kernel/sched/cpufreq_schedutil.c > +++ b/kernel/sched/cpufreq_schedutil.c > @@ -424,7 +424,16 @@ static void

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Hi Juri, On Wed, Jul 5, 2017 at 1:59 AM, Juri Lelli wrote: > Worker kthread needs to be able to change frequency for all other > threads. > > Make it special, just under STOP class. > > Signed-off-by: Juri Lelli > Cc: Peter Zijlstra

Re: [RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Hi Juri, On Wed, Jul 5, 2017 at 1:59 AM, Juri Lelli wrote: > Worker kthread needs to be able to change frequency for all other > threads. > > Make it special, just under STOP class. > > Signed-off-by: Juri Lelli > Cc: Peter Zijlstra > Cc: Ingo Molnar > Cc: Rafael J. Wysocki > Cc: Viresh

[RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Worker kthread needs to be able to change frequency for all other threads. Make it special, just under STOP class. Signed-off-by: Juri Lelli Cc: Peter Zijlstra Cc: Ingo Molnar Cc: Rafael J. Wysocki Cc:

[RFC PATCH v1 3/8] sched/cpufreq_schedutil: make worker kthread be SCHED_DEADLINE

Worker kthread needs to be able to change frequency for all other threads. Make it special, just under STOP class. Signed-off-by: Juri Lelli Cc: Peter Zijlstra Cc: Ingo Molnar Cc: Rafael J. Wysocki Cc: Viresh Kumar Cc: Luca Abeni Cc: Claudio Scordino --- Changes from RFCv0: - use a high