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;
MIME-Version: 1.0
References: <1548236816-18712-1-git-send-email-ufo19890607@gmail.com>
 <20190206171915.GJ17564@hirez.programming.kicks-ass.net> <CAHCio2ggaG9gCB642=GgwfM2iCRm+ZcAp0HGqevJtVrRmnKPUA@mail.gmail.com>
 <CAHCio2hn-r1_4bCEPtTm7=EjZmD=kUAFvLB9KEsEDbLi0mGvEg@mail.gmail.com>
In-Reply-To: <CAHCio2hn-r1_4bCEPtTm7=EjZmD=kUAFvLB9KEsEDbLi0mGvEg@mail.gmail.com>
From:   =?UTF-8?B?56a56Iif6ZSu?= <ufo19890607@gmail.com>
Date:   Mon, 25 Feb 2019 11:29:46 +0800
Message-ID: <CAHCio2g4DR8DcCtRvW69Ct2A3fZCnYaS39QP-VL3BJxVr-OAnA@mail.gmail.com>
Subject: Re: [PATCH] sched/debug: Show intergroup and hierarchy sum wait time
 of a task group
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     mingo@redhat.com,
        =?UTF-8?B?546L6LSH?= <yun.wang@linux.alibaba.com>,
        Wind Yu <yuzhoujian@didichuxing.com>,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

PING


=E7=A6=B9=E8=88=9F=E9=94=AE <ufo19890607@gmail.com> =E4=BA=8E2019=E5=B9=B42=
=E6=9C=8819=E6=97=A5=E5=91=A8=E4=BA=8C =E4=B8=8A=E5=8D=8810:15=E5=86=99=E9=
=81=93=EF=BC=9A
>
> PING
>
> =E7=A6=B9=E8=88=9F=E9=94=AE <ufo19890607@gmail.com> =E4=BA=8E2019=E5=B9=
=B42=E6=9C=8812=E6=97=A5=E5=91=A8=E4=BA=8C =E4=B8=8A=E5=8D=8811:14=E5=86=99=
=E9=81=93=EF=BC=9A
> >
> > Hi, Peter
> > I think hierarchy wait time for task groups is worth accounting
> > despite with a little extra overhead. Because we can evaluate task
> > groups' condition with a more direct metric. We cannot get the real
> > situation just with some general metrics, like idle or loadavg, since
> > their value is decreased with the elapse of time. So, I think general
> > metrics cannot satisify the request for task groups.
> >
> > Thanks
> > Yuzhoujian
> >
> >
> > Peter Zijlstra <peterz@infradead.org> =E4=BA=8E2019=E5=B9=B42=E6=9C=887=
=E6=97=A5=E5=91=A8=E5=9B=9B =E4=B8=8A=E5=8D=881:19=E5=86=99=E9=81=93=EF=BC=
=9A
> > >
> > > On Wed, Jan 23, 2019 at 05:46:56PM +0800, ufo19890607@gmail.com wrote=
:
> > > > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > > > index e2ff4b6..35e89ca 100644
> > > > --- a/kernel/sched/fair.c
> > > > +++ b/kernel/sched/fair.c
> > > > @@ -858,6 +858,19 @@ static void update_curr_fair(struct rq *rq)
> > > >  }
> > > >
> > > >  static inline void
> > > > +update_hierarchy_wait_sum(struct sched_entity *se,
> > > > +                     u64 delta_wait)
> > > > +{
> > > > +     for_each_sched_entity(se) {
> > > > +             struct cfs_rq *cfs_rq =3D cfs_rq_of(se);
> > > > +
> > > > +             if (cfs_rq->tg !=3D &root_task_group)
> > > > +                     __schedstat_add(cfs_rq->hierarchy_wait_sum,
> > > > +                                     delta_wait);
> > > > +     }
> > > > +}
> > > > +
> > > > +static inline void
> > > >  update_stats_wait_end(struct cfs_rq *cfs_rq, struct sched_entity *=
se)
> > > >  {
> > > >       struct task_struct *p;
> > > > @@ -880,6 +893,7 @@ static void update_curr_fair(struct rq *rq)
> > > >                       return;
> > > >               }
> > > >               trace_sched_stat_wait(p, delta);
> > > > +             update_hierarchy_wait_sum(se, delta);
> > > >       }
> > > >
> > > >       __schedstat_set(se->statistics.wait_max,
> > >
> > > The problem I have with this is that it will make schedstats even mor=
e
> > > expensive :/