linux-kernel - Re: [PATCH v2 3/3] sched/fair: fix unnecessary increase of balance interval

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAKfTPtAtf4WMu-eLALfqn3-sKru58=uH-yo2SFygaDfmVoP0Hg@mail.gmail.com>
Date:   Wed, 19 Dec 2018 17:54:09 +0100
From:   Vincent Guittot <vincent.guittot@...aro.org>
To:     Valentin Schneider <valentin.schneider@....com>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...nel.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Morten Rasmussen <Morten.Rasmussen@....com>
Subject: Re: [PATCH v2 3/3] sched/fair: fix unnecessary increase of balance interval

On Wed, 19 Dec 2018 at 16:54, Valentin Schneider
<valentin.schneider@....com> wrote:
>
> On 19/12/2018 13:29, Vincent Guittot wrote:
> [...]
> >> My point is that AFAICT the LBF_ALL_PINNED flag would cover all the cases
> >> we care about, although the one you're mentioning is the only one I can
> >> think of. In that case LBF_ALL_PINNED would never be cleared, so when we do
> >> the active balance we'd know it's because all other tasks were pinned so
> >> we should probably increase the interval (see last snippet I sent).
> >
> > There are probably several other UC than the one mentioned below as
> > tasks can be discarded for several reasons.
> > So instead of changing for all UC by default, i would prefer only
> > change for those for which we know it's safe
>
> I get your point. Thing is, I've stared at the code for a while and
> couldn't find any other usecase where checking LBF_ALL_PINNED wouldn't
> suffice.

The point is that LBF_ALL_PINNED flag is not set otherwise we would
have jump to out_*_pinned
But conditions are similar

>
> It would be nice convince ourselves it is indeed enough (or not, but then
> we should be sure of it rather than base ourselves on assumptions), because
> then we can have just a simple condition rather than something that
> introduces active balance categories.

this can be part of the larger rework that Peter asked few days ago