Re: [PATCH 3/3] sched: add_nr_running(): drop tick_nohz_full_cpu() check

[Luiz Capitulino <lcapitulino@redhat.com>]

On Thu, 24 Sep 2015 16:44:18 +0200
Frederic Weisbecker <fweisbec@gmail.com> wrote:

> On Thu, Sep 10, 2015 at 03:58:27PM -0400, Luiz Capitulino wrote:
> > tick_nohz_full_kick_cpu() performs the same check.
> > 
> > Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
> > ---
> >  kernel/sched/sched.h | 20 +++++++++-----------
> >  1 file changed, 9 insertions(+), 11 deletions(-)
> > 
> > diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> > index 68cda11..102eb18 100644
> > --- a/kernel/sched/sched.h
> > +++ b/kernel/sched/sched.h
> > @@ -1323,17 +1323,15 @@ static inline void add_nr_running(struct rq *rq, unsigned count)
> >  #endif
> >  
> >  #ifdef CONFIG_NO_HZ_FULL
> > -		if (tick_nohz_full_cpu(rq->cpu)) {
> > -			/*
> > -			 * Tick is needed if more than one task runs on a CPU.
> > -			 * Send the target an IPI to kick it out of nohz mode.
> > -			 *
> > -			 * We assume that IPI implies full memory barrier and the
> > -			 * new value of rq->nr_running is visible on reception
> > -			 * from the target.
> > -			 */
> > -			tick_nohz_full_kick_cpu(rq->cpu);
> > -		}
> > +		/*
> > +		 * Tick is needed if more than one task runs on a CPU.
> > +		 * Send the target an IPI to kick it out of nohz mode.
> > +		 *
> > +		 * We assume that IPI implies full memory barrier and the
> > +		 * new value of rq->nr_running is visible on reception
> > +		 * from the target.
> > +		 */
> > +		tick_nohz_full_kick_cpu(rq->cpu);
> 
> Nope, we want to keep this because tick_nohz_full_cpu() does a static key check.
> Most users don't care about nohz_full and I really want to keep nohz full off case
> overhead to the bare minimum.

Makes sense.

> 
> Thanks.
> 
> >  #endif
> >  	}
> >  }
> > -- 
> > 2.1.0
> > 
>