Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 16:18, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 03:08:20PM +, Patrick Bellasi wrote: > > > And than we can move this bit into an inline function, something like e.g.: > > > >static inline bool sugov_this_cpu_is_busy() > >{ > >return

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 16:18, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 03:08:20PM +, Patrick Bellasi wrote: > > > And than we can move this bit into an inline function, something like e.g.: > > > >static inline bool sugov_this_cpu_is_busy() > >{ > >return

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 06:00:17 PM Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 04:18:52PM +0100, Rafael J. Wysocki wrote: > > +static bool sugov_cpu_is_busy(struct sugov_cpu *sg_cpu) > > +{ > > + unsigned long idle_calls = tick_nohz_get_idle_calls(); > > + bool not_idle = idle_calls ==

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 06:00:17 PM Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 04:18:52PM +0100, Rafael J. Wysocki wrote: > > +static bool sugov_cpu_is_busy(struct sugov_cpu *sg_cpu) > > +{ > > + unsigned long idle_calls = tick_nohz_get_idle_calls(); > > + bool not_idle = idle_calls ==

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 15:58, Peter Zijlstra wrote: > > On Tue, Mar 21, 2017 at 03:16:19PM +0100, Vincent Guittot wrote: > > On 21 March 2017 at 15:03, Peter Zijlstra wrote: > > > > > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > > >

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 15:58, Peter Zijlstra wrote: > > On Tue, Mar 21, 2017 at 03:16:19PM +0100, Vincent Guittot wrote: > > On 21 March 2017 at 15:03, Peter Zijlstra wrote: > > > > > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > > > > On 21 March 2017 at 14:22, Peter Zijlstra

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 18:00, Vincent Guittot wrote: > On 21 March 2017 at 15:58, Peter Zijlstra wrote: >> >> On Tue, Mar 21, 2017 at 03:16:19PM +0100, Vincent Guittot wrote: >> > On 21 March 2017 at 15:03, Peter Zijlstra

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 18:00, Vincent Guittot wrote: > On 21 March 2017 at 15:58, Peter Zijlstra wrote: >> >> On Tue, Mar 21, 2017 at 03:16:19PM +0100, Vincent Guittot wrote: >> > On 21 March 2017 at 15:03, Peter Zijlstra wrote: >> > >> > > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 04:18:52PM +0100, Rafael J. Wysocki wrote: > +static bool sugov_cpu_is_busy(struct sugov_cpu *sg_cpu) > +{ > + unsigned long idle_calls = tick_nohz_get_idle_calls(); > + bool not_idle = idle_calls == sg_cpu->saved_idle_calls; > + > + sg_cpu->saved_idle_calls =

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 04:18:52PM +0100, Rafael J. Wysocki wrote: > +static bool sugov_cpu_is_busy(struct sugov_cpu *sg_cpu) > +{ > + unsigned long idle_calls = tick_nohz_get_idle_calls(); > + bool not_idle = idle_calls == sg_cpu->saved_idle_calls; > + > + sg_cpu->saved_idle_calls =

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 04:04:03 PM Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 03:46:07PM +0100, Rafael J. Wysocki wrote: > > @@ -207,6 +212,8 @@ static void sugov_update_single(struct u > > if (!sugov_should_update_freq(sg_policy, time)) > > return; > > > > +

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 04:04:03 PM Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 03:46:07PM +0100, Rafael J. Wysocki wrote: > > @@ -207,6 +212,8 @@ static void sugov_update_single(struct u > > if (!sugov_should_update_freq(sg_policy, time)) > > return; > > > > +

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

Seriously people, trim your replies. On Tue, Mar 21, 2017 at 03:08:20PM +, Patrick Bellasi wrote: > And than we can move this bit into an inline function, something like e.g.: > >static inline bool sugov_this_cpu_is_busy() >{ >return this_rq()->rd->overloaded >}

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

Seriously people, trim your replies. On Tue, Mar 21, 2017 at 03:08:20PM +, Patrick Bellasi wrote: > And than we can move this bit into an inline function, something like e.g.: > >static inline bool sugov_this_cpu_is_busy() >{ >return this_rq()->rd->overloaded >}

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 15:46, Rafael J. Wysocki wrote: > On Tuesday, March 21, 2017 02:38:42 PM Patrick Bellasi wrote: > > On 21-Mar 15:26, Rafael J. Wysocki wrote: > > > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > > > On 21 March 2017 at 14:22, Peter Zijlstra

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 15:46, Rafael J. Wysocki wrote: > On Tuesday, March 21, 2017 02:38:42 PM Patrick Bellasi wrote: > > On 21-Mar 15:26, Rafael J. Wysocki wrote: > > > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > > > On Tue,

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 03:46:07PM +0100, Rafael J. Wysocki wrote: > @@ -207,6 +212,8 @@ static void sugov_update_single(struct u > if (!sugov_should_update_freq(sg_policy, time)) > return; > > + sg_policy->overload = this_rq()->rd->overload; > + Same problem as before;

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 03:46:07PM +0100, Rafael J. Wysocki wrote: > @@ -207,6 +212,8 @@ static void sugov_update_single(struct u > if (!sugov_should_update_freq(sg_policy, time)) > return; > > + sg_policy->overload = this_rq()->rd->overload; > + Same problem as before;

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 03:26:06PM +0100, Rafael J. Wysocki wrote: > + if ((flags & SCHED_CPUFREQ_RT_DL) || this_rq()->rd->overload) { > next_f = policy->cpuinfo.max_freq; So this I think is wrong; rd->overload is set if _any_ of the CPUs in the root domain is overloaded. And

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 03:26:06PM +0100, Rafael J. Wysocki wrote: > + if ((flags & SCHED_CPUFREQ_RT_DL) || this_rq()->rd->overload) { > next_f = policy->cpuinfo.max_freq; So this I think is wrong; rd->overload is set if _any_ of the CPUs in the root domain is overloaded. And

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 03:46:07 PM Rafael J. Wysocki wrote: > On Tuesday, March 21, 2017 02:38:42 PM Patrick Bellasi wrote: > > On 21-Mar 15:26, Rafael J. Wysocki wrote: > > > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > > > On 21 March 2017 at 14:22, Peter Zijlstra

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 03:46:07 PM Rafael J. Wysocki wrote: > On Tuesday, March 21, 2017 02:38:42 PM Patrick Bellasi wrote: > > On 21-Mar 15:26, Rafael J. Wysocki wrote: > > > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > > > On 21 March 2017 at 14:22, Peter Zijlstra

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 03:16:19PM +0100, Vincent Guittot wrote: > On 21 March 2017 at 15:03, Peter Zijlstra wrote: > > > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > > >

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 03:16:19PM +0100, Vincent Guittot wrote: > On 21 March 2017 at 15:03, Peter Zijlstra wrote: > > > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > > > For the not overloaded case, it makes sense

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > For the not overloaded case, it makes sense to immediately update to > OPP to be aligned with the new utilization of the CPU even if it was > not idle in

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > For the not overloaded case, it makes sense to immediately update to > OPP to be aligned with the new utilization of the CPU even if it was > not idle in the past couple of ticks

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 02:38:42 PM Patrick Bellasi wrote: > On 21-Mar 15:26, Rafael J. Wysocki wrote: > > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > > On Tue, Mar 21, 2017 at 09:50:28AM

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 02:38:42 PM Patrick Bellasi wrote: > On 21-Mar 15:26, Rafael J. Wysocki wrote: > > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 15:26, Rafael J. Wysocki wrote: > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: > > >> On 20 March 2017 at 22:46, Rafael J.

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 15:26, Rafael J. Wysocki wrote: > On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: > > >> On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > > >

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 15:03, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > For the not overloaded case, it makes sense to immediately update to > > OPP to be aligned with the new

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 15:03, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: > > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > > For the not overloaded case, it makes sense to immediately update to > > OPP to be aligned with the new utilization of the CPU even

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: > >> On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > > >> > To

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tuesday, March 21, 2017 02:37:08 PM Vincent Guittot wrote: > On 21 March 2017 at 14:22, Peter Zijlstra wrote: > > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: > >> On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > > >> > To work around this issue use the observation

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 15:03, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: >> On 21 March 2017 at 14:22, Peter Zijlstra wrote: > >> For the not overloaded case, it makes sense to immediately update to >> OPP to

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 15:03, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 02:37:08PM +0100, Vincent Guittot wrote: >> On 21 March 2017 at 14:22, Peter Zijlstra wrote: > >> For the not overloaded case, it makes sense to immediately update to >> OPP to be aligned with the new utilization of the

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 14:22, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: >> On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > >> > To work around this issue use the observation that, from the >> > schedutil

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21 March 2017 at 14:22, Peter Zijlstra wrote: > On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: >> On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > >> > To work around this issue use the observation that, from the >> > schedutil governor's perspective, it does not make

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: > On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > To work around this issue use the observation that, from the > > schedutil governor's perspective, it does not make sense to decrease > > the frequency

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 09:50:28AM +0100, Vincent Guittot wrote: > On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > To work around this issue use the observation that, from the > > schedutil governor's perspective, it does not make sense to decrease > > the frequency of a CPU that doesn't

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 7:40 AM, Viresh Kumar wrote: > On 20-03-17, 22:46, Rafael J. Wysocki wrote: >> Index: linux-pm/kernel/sched/cpufreq_schedutil.c > >> +static void sugov_update_commit(struct sugov_cpu *sg_cpu, >> + struct sugov_policy

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On Tue, Mar 21, 2017 at 7:40 AM, Viresh Kumar wrote: > On 20-03-17, 22:46, Rafael J. Wysocki wrote: >> Index: linux-pm/kernel/sched/cpufreq_schedutil.c > >> +static void sugov_update_commit(struct sugov_cpu *sg_cpu, >> + struct sugov_policy *sg_policy, >> +

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 09:50, Vincent Guittot wrote: > On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > From: Rafael J. Wysocki > > > > The way the schedutil governor uses the PELT metric causes it to > > underestimate the CPU utilization in some cases.

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 21-Mar 09:50, Vincent Guittot wrote: > On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > > From: Rafael J. Wysocki > > > > The way the schedutil governor uses the PELT metric causes it to > > underestimate the CPU utilization in some cases. > > > > That can be easily demonstrated by

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 20-Mar 22:46, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The way the schedutil governor uses the PELT metric causes it to > underestimate the CPU utilization in some cases. > > That can be easily demonstrated by running kernel compilation on > a Sandy

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 20-Mar 22:46, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The way the schedutil governor uses the PELT metric causes it to > underestimate the CPU utilization in some cases. > > That can be easily demonstrated by running kernel compilation on > a Sandy Bridge Intel processor,

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The way the schedutil governor uses the PELT metric causes it to > underestimate the CPU utilization in some cases. > > That can be easily demonstrated by running

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 20 March 2017 at 22:46, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki > > The way the schedutil governor uses the PELT metric causes it to > underestimate the CPU utilization in some cases. > > That can be easily demonstrated by running kernel compilation on > a Sandy Bridge Intel

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 20-03-17, 22:46, Rafael J. Wysocki wrote: > Index: linux-pm/kernel/sched/cpufreq_schedutil.c > +static void sugov_update_commit(struct sugov_cpu *sg_cpu, > + struct sugov_policy *sg_policy, > + u64 time, unsigned int next_freq) > { >

Re: [RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

On 20-03-17, 22:46, Rafael J. Wysocki wrote: > Index: linux-pm/kernel/sched/cpufreq_schedutil.c > +static void sugov_update_commit(struct sugov_cpu *sg_cpu, > + struct sugov_policy *sg_policy, > + u64 time, unsigned int next_freq) > { >

[RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

From: Rafael J. Wysocki The way the schedutil governor uses the PELT metric causes it to underestimate the CPU utilization in some cases. That can be easily demonstrated by running kernel compilation on a Sandy Bridge Intel processor, running turbostat in parallel

[RFC][PATCH v2 2/2] cpufreq: schedutil: Avoid decreasing frequency of busy CPUs

From: Rafael J. Wysocki The way the schedutil governor uses the PELT metric causes it to underestimate the CPU utilization in some cases. That can be easily demonstrated by running kernel compilation on a Sandy Bridge Intel processor, running turbostat in parallel with it and looking at the