MIME-Version: 1.0
References: <20211217021610.12801-1-yajun.deng@linux.dev> <YbyTuRWkB0gYbn7x@linutronix.de>
In-Reply-To: <YbyTuRWkB0gYbn7x@linutronix.de>
From:   Daniel Vacek <neelx.g@gmail.com>
Date:   Fri, 17 Dec 2021 22:58:06 +0100
Message-ID: <CAA7rmPEbRbztCHzanF148UVnL=Q_iAsA28yPJiu51aA=BhCwag@mail.gmail.com>
Subject: Re: [PATCH v3] lib/raid6: Reduce high latency by using migrate
 instead of preempt
To:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc:     Yajun Deng <yajun.deng@linux.dev>, song@kernel.org,
        masahiroy@kernel.org, williams@redhat.com, pmenzel@molgen.mpg.de,
        LKML <linux-kernel@vger.kernel.org>,
        linux-rt-users <linux-rt-users@vger.kernel.org>,
        linux-raid@vger.kernel.org, stable@vger.kernel.org,
        Thomas Gleixner <tglx@linutronix.de>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Dec 17, 2021 at 8:09 PM Sebastian Andrzej Siewior
<bigeasy@linutronix.de> wrote:
>
> On 2021-12-17 10:16:10 [+0800], Yajun Deng wrote:
> > We found an abnormally high latency when executing modprobe raid6_pq, the
> > latency is greater than 1.2s when CONFIG_PREEMPT_VOLUNTARY=y, greater than
> > 67ms when CONFIG_PREEMPT=y, and greater than 16ms when CONFIG_PREEMPT_RT=y.
> >
> > How to reproduce:
> >  - Install cyclictest
> >      sudo apt install rt-tests
> >  - Run cyclictest example in one terminal
> >      sudo cyclictest -S -p 95 -d 0 -i 1000 -D 24h -m
> >  - Modprobe raid6_pq in another terminal
> >      sudo modprobe raid6_pq
> >
> > This is caused by ksoftirqd fail to scheduled due to disable preemption,
> > this time is too long and unreasonable.
> >
> > Reduce high latency by using migrate_disabl()/emigrate_enable() instead of
> > preempt_disable()/preempt_enable(), the latency won't greater than 100us.
> >
> > This patch beneficial for CONFIG_PREEMPT=y or CONFIG_PREEMPT_RT=y, but no
> > effect for CONFIG_PREEMPT_VOLUNTARY=y.
>
> Why does it matter? This is only during boot-up/ module loading or do I
> miss something?
> The delay is a jiffy so it depends on CONFIG_HZ. You do benchmark for
> the best algorithm and if you get preempted during that period then your
> results may be wrong and you make a bad selection.
>
> You can either enable one algorithm and or disable
> CONFIG_RAID6_PQ_BENCHMARK. I don't see the need for this patch not to
> mention the stable tree.

Exactly. We should not touch this. I've just sent a verbose
explanation in the original report thread.

--nX

> Sebastian