lkml.org 
[lkml]   [2023]   [Feb]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 2/2] sched/isolation: Add cpu_is_isolated() API
On Fri, Feb 03, 2023 at 10:53:46PM -0500, Waiman Long wrote:
> On 2/3/23 18:24, Frederic Weisbecker wrote:
> > Provide this new API to check if a CPU has been isolated either through
> > isolcpus= or nohz_full= kernel parameter.
> >
> > It aims at avoiding kernel load deemed to be safely spared on CPUs
> > running sensitive workload that can't bear any disturbance, such as
> > pcp cache draining.
> >
> > Suggested-by: Michal Hocko <mhocko@suse.com>
> > Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
> > ---
> > include/linux/sched/isolation.h | 6 ++++++
> > 1 file changed, 6 insertions(+)
> >
> > diff --git a/include/linux/sched/isolation.h b/include/linux/sched/isolation.h
> > index b645cc81fe01..088672f08469 100644
> > --- a/include/linux/sched/isolation.h
> > +++ b/include/linux/sched/isolation.h
> > @@ -53,4 +53,10 @@ static inline bool housekeeping_cpu(int cpu, enum hk_type type)
> > return true;
> > }
> > +static inline bool cpu_is_isolated(int cpu)
> > +{
> > + return !housekeeping_test_cpu(cpu, HK_TYPE_DOMAIN) ||
> > + !housekeeping_test_cpu(cpu, HK_TYPE_KERNEL_NOISE);
> > +}
> > +
> > #endif /* _LINUX_SCHED_ISOLATION_H */
>
> CPUs in an isolated cpuset partition is similar to HK_TYPE_DOMAIN CPUs as
> load balancing is disabled. I can add an API to access the cpumask and add
> to this API. However, that list is dynamic as it can be changed at run time.
> Will that be a problem? Or should that be used separately?

So that's what I intended first but the dynamic part of cpuset made me
postpone that to better days.

But yes ideally it should look like:

static inline bool cpu_is_isolated(int cpu)
{
return !housekeeping_test_cpu(cpu, HK_TYPE_KERNEL_NOISE) ||
on_null_domain(cpu_rq(cpu));
}

And there should be a hook in something like detach_destroy_domains()
to flush the pcp cache when a CPU is attached to a NULL domain.

All that with proper RCU synchronization:

UPDATE READER
------ ------
rcu_assign_pointer(cpu_rq(cpu)->sd, NULL); rcu_read_lock();
synchronize_rcu(); if (!cpu_is_isolated(cpu))
stock = &per_cpu(memcg_stock, cpu); schedule_work_on(cpu, &stock->work);
flush_work(&stock->work); rcu_read_unlock()

Thanks.

\
 
 \ /
  Last update: 2023-03-27 00:12    [W:0.137 / U:0.068 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site