Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200421004749.GC26573@shao2-debian> <20200425012306.13516-1-hdanton@sina.com>
In-Reply-To: <20200425012306.13516-1-hdanton@sina.com>
From:   Vincent Guittot <vincent.guittot@linaro.org>
Date:   Sat, 25 Apr 2020 10:42:54 +0200
Message-ID: <CAKfTPtCvnDoDm9J611uk+fpyPqyh=zn1Rbo=L=MrqbD_jMOx5A@mail.gmail.com>
Subject: Re: [LKP] [sched/fair] 6c8116c914: stress-ng.mmapfork.ops_per_sec
 -38.0% regression
To:     Hillf Danton <hdanton@sina.com>
Cc:     Xing Zhengjun <zhengjun.xing@linux.intel.com>,
        kernel test robot <rong.a.chen@intel.com>,
        Tao Zhou <ouwen210@hotmail.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Mel Gorman <mgorman@suse.de>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sat, 25 Apr 2020 at 03:23, Hillf Danton <hdanton@sina.com> wrote:
>
>
> Hi Xing
>
> On Fri, 24 Apr 2020 16:15:45 +0800 Xing Zhengjun wrote:
> >
> > Hi Tao,
> >
> >     Do you have time to take a look at this? Thanks.
> >
> > On 4/21/2020 8:47 AM, kernel test robot wrote:
> > >
> > > Greeting,
> > >
> > > FYI, we noticed a 56.4% improvement of stress-ng.fifo.ops_per_sec due to commit:
> > >
> > >
> > > commit: 6c8116c914b65be5e4d6f66d69c8142eb0648c22 ("sched/fair: Fix condition of avg_load calculation")
> > > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
> > >
> > > in testcase: stress-ng
> > > on test machine: 88 threads Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz with 128G memory
> > > with following parameters:
> > >
> > >     nr_threads: 100%
> > >     disk: 1HDD
> > >     testtime: 1s
> > >     class: scheduler
> > >     cpufreq_governor: performance
> > >     ucode: 0xb000038
> > >     sc_pid_max: 4194304
> > >
>
> Here is a tiny workaround. If it makes a sense to you see if it helps
> cure the regression.
>
> Hillf
>
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -8744,6 +8744,16 @@ find_idlest_group(struct sched_domain *s
>
>         switch (local_sgs.group_type) {
>         case group_overloaded:
> +               /*
> +                * push the task if it helps cut the imbalance in the
> +                * number of running tasks
> +                */
> +               if (local_sgs.sum_h_nr_running >
> +                   idlest_sgs.sum_h_nr_running)
> +                       return idlest;
> +               else
> +                       return NULL;

Taking into account nr_running when overloaded to bypass load
comparison just break the fairness and the spread of load and will
cause regression for other use cases


> +
>         case group_fully_busy:
>                 /*
>                  * When comparing groups across NUMA domains, it's possible for
> --
>