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: <20210624111815.57937-1-odin@uged.al> <E16E71B3-E941-4522-AFF1-ABDF918FED19@linux.vnet.ibm.com>
 <20210701103448.GL3840@techsingularity.net>
In-Reply-To: <20210701103448.GL3840@techsingularity.net>
From:   Odin Ugedal <odin@uged.al>
Date:   Thu, 1 Jul 2021 13:09:33 +0200
Message-ID: <CAFpoUr1mpErE1Soa05p36wL1uTeojQ2mqNLJ1GKVnpJ+x-3itw@mail.gmail.com>
Subject: Re: [PATCH] sched/fair: Ensure _sum and _avg values stay consistent
To:     Mel Gorman <mgorman@techsingularity.net>
Cc:     Sachin Sant <sachinp@linux.vnet.ibm.com>,
        Odin Ugedal <odin@uged.al>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Naresh Kamboju <naresh.kamboju@linaro.org>,
        Ingo Molnar <mingo@redhat.com>,
        open list <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

tor. 1. jul. 2021 kl. 12:34 skrev Mel Gorman <mgorman@techsingularity.net>:
>
> What was HEAD when you checked this? 1c35b07e6d39 was merged in the
> 5.14-rc1 merge window so would not be in 5.13.

From the kernel log it looks like he used commit dbe69e433722
(somewhere in Linus' tree), and that should include my patch. I don't
think the patches from Vincent in the previous thread have been posted
properly, and I think those can fix the edge case you are seeing.

This should mitigate some issues:
https://lore.kernel.org/lkml/20210622143154.GA804@vingu-book/

while
https://lore.kernel.org/lkml/20210623071935.GA29143@vingu-book/
might also help in some cases as well.

Both of those are needed to make sure that *_avg is zero whenever *_sum is zero.

> # git log v5.13..origin/master --pretty=one | grep 1c35b07e
> 1c35b07e6d3986474e5635be566e7bc79d97c64d sched/fair: Ensure _sum and _avg values stay consistent
>
> It's not tagged for stable and lacks a Fixes: tag so I don't think
> it'll be automatically picked up for 5.13-stable unless Odin sends it
> to stable@vger.kernel.org.

Ahh, so the "Fixes" tag is what makes them pick it up...  Thanks! I am
fairly new to this, but it does make sense when I think about it. I
have seen https://www.kernel.org/doc/html/v5.13/process/stable-kernel-rules.html
before, so I guess I can send it to them to get it into -stable. The
assertion is not a part of v5.13 either though, so people will not see
this warning there (but it can still cause fairness issues).

> --
> Mel Gorman
> SUSE Labs

Thanks
Odin