linux-kernel - Re: [PATCH] mm, memcg: reclaim more aggressively before high allocator throttling

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: <20200521122814.GN6462@dhcp22.suse.cz>
Date:   Thu, 21 May 2020 14:28:14 +0200
From:   Michal Hocko <mhocko@...nel.org>
To:     Chris Down <chris@...isdown.name>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Johannes Weiner <hannes@...xchg.org>,
        Tejun Heo <tj@...nel.org>, linux-mm@...ck.org,
        cgroups@...r.kernel.org, linux-kernel@...r.kernel.org,
        kernel-team@...com
Subject: Re: [PATCH] mm, memcg: reclaim more aggressively before high
 allocator throttling

On Thu 21-05-20 12:27:11, Chris Down wrote:
[...]
> Regardless, you're pushing for different reclaim semantics for memory.high
> than memory.max here, which requires evidence that the current approach
> taken for memory.max is wrong or causing issues.

Sorry, I have skipped over this part. Memory high limit reclaim has
historically acted as a best effort action to throttle the
allocation/charge pace. This would work both if the implementation
simply tried to reclaim down to the high limit or if the reclaim is
proportional to the memory consumption by a specific consumer. We do the
later because it is much easier to establish fairness for. If you want
to change that you somehow have to deal with the fairness problem.

And yes, we do not guarantee any fairness for the hard limit or direct
reclaim in general but that behavior is generally problematic and there
should be really strong arguments to move high limit reclaim that
direction IMHO.

-- 
Michal Hocko
SUSE Labs