Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 21 Aug 2019 20:37:13 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Qais Yousef <qais.yousef@arm.com>
Cc:     Peng Liu <iwtbavbm@gmail.com>, linux-kernel@vger.kernel.org,
        mingo@redhat.com
Subject: Re: [PATCH] sched/fair: eliminate redundant code in sched_slice()
Message-ID: <20190821183713.GF2349@hirez.programming.kicks-ass.net>
References: <20190816141202.GA3135@iZj6chx1xj0e0buvshuecpZ>
 <20190821151523.lwazjd2d6rp5otdh@e107158-lin.cambridge.arm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190821151523.lwazjd2d6rp5otdh@e107158-lin.cambridge.arm.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Aug 21, 2019 at 04:15:24PM +0100, Qais Yousef wrote:
> On 08/16/19 22:12, Peng Liu wrote:
> > Since sched_slice() is used in high frequency,
> > small change should also make sense.
> > 
> > Signed-off-by: Peng Liu <iwtbavbm@gmail.com>
> > ---
> >  kernel/sched/fair.c | 11 ++++-------
> >  1 file changed, 4 insertions(+), 7 deletions(-)
> > 
> > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > index 1054d2cf6aaa..6ae2a507aac0 100644
> > --- a/kernel/sched/fair.c
> > +++ b/kernel/sched/fair.c
> > @@ -694,19 +694,16 @@ static u64 sched_slice(struct cfs_rq *cfs_rq, struct sched_entity *se)
> >  	u64 slice = __sched_period(cfs_rq->nr_running + !se->on_rq);
> >  
> >  	for_each_sched_entity(se) {
> > -		struct load_weight *load;
> >  		struct load_weight lw;
> >  
> >  		cfs_rq = cfs_rq_of(se);
> > -		load = &cfs_rq->load;
> > +		lw = cfs_rq->load;
> >  
> > -		if (unlikely(!se->on_rq)) {
> > +		if (unlikely(!se->on_rq))
> >  			lw = cfs_rq->load;
> >  
> > -			update_load_add(&lw, se->load.weight);
> > -			load = &lw;
> > -		}
> > -		slice = __calc_delta(slice, se->load.weight, load);
> > +		update_load_add(&lw, se->load.weight);
> > +		slice = __calc_delta(slice, se->load.weight, &lw);
> 
> Unless I misread the diff, you changed the behavior.
> 
> update_load_add() is only called if (unlikely(!se->on_rq)), but with your
> change it is called unconditionally. And lw is set twice.
> 
> I think what you intended is this?

So I'd really rather hold off on this; Rik is poking at getting rid of
all of this hierarchical crud in one go.