linux-kernel - Re: [PATCH] sched/fair: Ensure _sum and

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]

Date:   Thu, 24 Jun 2021 14:16:43 +0200
From:   Vincent Guittot <vincent.guittot@...aro.org>
To:     Odin Ugedal <odin@...d.al>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Sachin Sant <sachinp@...ux.vnet.ibm.com>,
        Naresh Kamboju <naresh.kamboju@...aro.org>,
        Ingo Molnar <mingo@...hat.com>,
        Juri Lelli <juri.lelli@...hat.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Ben Segall <bsegall@...gle.com>, Mel Gorman <mgorman@...e.de>,
        Daniel Bristot de Oliveira <bristot@...hat.com>,
        linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] sched/fair: Ensure _sum and _avg values stay consistent

On Thu, 24 Jun 2021 at 13:21, Odin Ugedal <odin@...d.al> wrote:
>
> The _sum and _avg values are in general sync together with the PELT
> divider. They are however not always completely in perfect sync,
> resulting in situations where _sum gets to zero while _avg stays
> positive. Such situations are undesirable.
>
> This comes from the fact that PELT will increase period_contrib, also
> increasing the PELT divider, without updating _sum and _avg values to
> stay in perfect sync where (_sum == _avg * divider). However, such PELT

_sum is always synced and updated with PELT contrib and _avg is only
updated when crossing the 1024us period boundary. The problem here is
that the contribution to _sum can be null (not running or sleeping
state) whereas the formula "_avg * divider" assumes that all
contributions in the current period are not null. So "_avg * divider"
overestimates _sum.

Another solution would be to underestimate _sum and use _avg
*LOAD_AVG_MAX - 1024" when subtracting some _sum and keep using
LOAD_AVG_MAX - 1024 + avg->period_contrib when adding _sum.  Note,
that this doesn't make any real difference at the end for the patch
below because we don't save any multiplication operation anyway

So after updating the commit message

Reviewed-by: Vincent Guittot <vincent.guittot@...aro.org>

> change will never lower _sum, making it impossible to end up in a
> situation where _sum is zero and _avg is not.
>
> Therefore, we need to ensure that when subtracting load outside PELT,
> that when _sum is zero, _avg is also set to zero. This occurs when
> (_sum < _avg * divider), and the subtracted (_avg * divider) is bigger
> or equal to the current _sum, while the subtracted _avg is smaller than
> the current _avg.
>
> Reported-by: Sachin Sant <sachinp@...ux.vnet.ibm.com>
> Reported-by: Naresh Kamboju <naresh.kamboju@...aro.org>
> Signed-off-by: Odin Ugedal <odin@...d.al>
> ---
>
> Reports and discussion can be found here:
>
> https://lore.kernel.org/lkml/2ED1BDF5-BC0C-47CD-8F33-9A46C738F8CF@linux.vnet.ibm.com/
> https://lore.kernel.org/lkml/CA+G9fYsMXELmjGUzw4SY1bghTYz_PeR2diM6dRp2J37bBZzMSA@mail.gmail.com/
>
>  kernel/sched/fair.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index bfaa6e1f6067..def48bc2e90b 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -3688,15 +3688,15 @@ update_cfs_rq_load_avg(u64 now, struct cfs_rq *cfs_rq)
>
>                 r = removed_load;
>                 sub_positive(&sa->load_avg, r);
> -               sub_positive(&sa->load_sum, r * divider);
> +               sa->load_sum = sa->load_avg * divider;
>
>                 r = removed_util;
>                 sub_positive(&sa->util_avg, r);
> -               sub_positive(&sa->util_sum, r * divider);
> +               sa->util_sum = sa->util_avg * divider;
>
>                 r = removed_runnable;
>                 sub_positive(&sa->runnable_avg, r);
> -               sub_positive(&sa->runnable_sum, r * divider);
> +               sa->runnable_sum = sa->runnable_avg * divider;
>
>                 /*
>                  * removed_runnable is the unweighted version of removed_load so we
> --
> 2.32.0
>