On 30-Nov 14:41, Juri Lelli wrote:
> Hi,
>
> On 30/11/17 11:47, Patrick Bellasi wrote:
> > In system where multiple CPUs shares the same frequency domain a small
> > workload on a CPU can still be subject to frequency spikes, generated by
> > the activation of the sugov's kthread.
> >
> > Since the sugov kthread is a special RT task, which goal is just that to
> > activate a frequency transition, it does not make sense for it to bias
> > the schedutil's frequency selection policy.
> >
> > This patch exploits the information related to the current task to silently
> > ignore cpufreq_update_this_cpu() calls, coming from the RT scheduler, while
> > the sugov kthread is running.
> >
> > Signed-off-by: Patrick Bellasi <[email protected]>
> > Reviewed-by: Dietmar Eggemann <[email protected]>
> > Cc: Ingo Molnar <[email protected]>
> > Cc: Peter Zijlstra <[email protected]>
> > Cc: Rafael J. Wysocki <[email protected]>
> > Cc: Viresh Kumar <[email protected]>
> > Cc: [email protected]
> > Cc: [email protected]
> >
> > ---
> > Changes from v2:
> > - rebased on v4.15-rc1
> > - moved at the end of the stack since considered more controversial
> > Changes from v1:
> > - move check before policy spinlock (JuriL)
> >
> > Change-Id: I4d749458229b6496dd24a8c357be42cd35a739fd
> > ---
> > kernel/sched/cpufreq_schedutil.c | 8 ++++++++
> > 1 file changed, 8 insertions(+)
> >
> > diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c
> > index 3eea8884e61b..a93ad5b0c40d 100644
> > --- a/kernel/sched/cpufreq_schedutil.c
> > +++ b/kernel/sched/cpufreq_schedutil.c
> > @@ -270,6 +270,10 @@ static void sugov_update_single(struct update_util_data *hook, u64 time,
> > bool rt_mode;
> > bool busy;
> >
> > + /* Skip updates generated by sugov kthreads */
> > + if (unlikely(current == sg_policy->thread))
> > + return;
> > +
> > sugov_set_iowait_boost(sg_cpu, time, flags);
> > sg_cpu->last_update = time;
> >
> > @@ -356,6 +360,10 @@ static void sugov_update_shared(struct update_util_data *hook, u64 time,
> > unsigned int next_f;
> > bool rt_mode;
> >
> > + /* Skip updates generated by sugov kthreads */
> > + if (unlikely(current == sg_policy->thread))
> > + return;
> > +
> > raw_spin_lock(&sg_policy->update_lock);
> >
> > sugov_get_util(&util, &max, sg_cpu->cpu);
>
> If the DL changes (which I shall post again as soon as tip/sched/core is
> bumped up to 4.15-rc1) get in first, this is going to be useless (as the
> DL kthread gets ignored by the scheduling class itself). But, this looks
> good to me "in the meantime".
Just to better understand, you mean that the DL kthread does not send
out schedutil updates?
If that's the case I agree we can discard this patch... that's also
one of the reasons why I move it at the end of this series.
> Reviewed-by: Juri Lelli <[email protected]>
>
> Best,
>
> Juri
--
#include <best/regards.h>
Patrick Bellasi
From 1585498686390010250@xxx Thu Nov 30 13:42:55 +0000 2017
X-GM-THRID: 1585491510524518851
X-Gmail-Labels: Inbox,Category Forums,HistoricalUnread