linux-kernel - Re: [PATCH 1/3] sched/fair: Add tg_load_contrib cfs

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

Message-ID: <CAKfTPtA5NxDpG194zAF9CjchJ5hm-7fOBOPGJZG4+z4-+32qGw@mail.gmail.com>
Date:   Thu, 27 May 2021 14:37:16 +0200
From:   Vincent Guittot <vincent.guittot@...aro.org>
To:     Odin Ugedal <odin@...d.al>
Cc:     Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Ben Segall <bsegall@...gle.com>, Mel Gorman <mgorman@...e.de>,
        Daniel Bristot de Oliveira <bristot@...hat.com>,
        "open list:CONTROL GROUP (CGROUP)" <cgroups@...r.kernel.org>,
        linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/3] sched/fair: Add tg_load_contrib cfs_rq decay checking

On Thu, 27 May 2021 at 13:04, Odin Ugedal <odin@...d.al> wrote:
>
> > 1st : ensure that cfs_rq->load_sum is not null if cfs_rq-> load_isn't too
> > 2nd : call update_tg_load_avg() during child update so we will be sure
> > to update tg_load_avg_contrib before removing the cfs from the list
>
> Ahh, yeah, with "1st" that would work. Yeah, that was my initial
> implementation of the change, but I thought that it was better to keep
> the logic away from the "hot path". We can verify this in
> update_tg_cfs_load(), and then force update_tg_load_avg() inside

For 1st problem, the way we were updating load_avg and load_sum, we
were losing the sync between both value

> __update_blocked_fair() when avg.load_avg is 0. (Given that this is
> the only place where we can end up in this situation. I can update
> this patch to do that instead.

In fact, the update was already there but not always called (see the
patchset i just sent)


>
> Another solution is to update avg.load_avg
> inside__update_blocked_fair() when load_sum is 0, and then propagate
> that with update_tg_load_avg(). This removes the logic from the hot
> path all together.
>
> Not sure what the preferred way is. I have not found any other places
> where this situation _should_ occur, but who knows..
>
> Odin