linux-kernel - Re: [RFC PATCH] mm: memcg/slab: Stop reparented obj

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20201103213010.GA1938922@carbon.dhcp.thefacebook.com>
Date:   Tue, 3 Nov 2020 13:30:10 -0800
From:   Roman Gushchin <guro@...com>
To:     Michal Hocko <mhocko@...e.com>
CC:     Johannes Weiner <hannes@...xchg.org>,
        Michal Koutný <mkoutny@...e.com>,
        Richard Palethorpe <rpalethorpe@...e.com>,
        <ltp@...ts.linux.it>, Andrew Morton <akpm@...ux-foundation.org>,
        Shakeel Butt <shakeelb@...gle.com>,
        Christoph Lameter <cl@...ux.com>, Tejun Heo <tj@...nel.org>,
        Vlastimil Babka <vbabka@...e.cz>, <linux-mm@...ck.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [RFC PATCH] mm: memcg/slab: Stop reparented obj_cgroups from
 charging root

On Tue, Nov 03, 2020 at 02:22:21PM +0100, Michal Hocko wrote:
> Hi,
> I am sorry but I was quite busy at the time this has been discussed.
> Now that I got to this thread finally I am wondering whether this
> resp. other root cgroup exceptions have any follow up.

I'll address a feedback I've got from Johannes and will post and updated
version soon.

> 
> I have seen the issue is fixed in Linus tree by 8de15e920dc8 ("mm:
> memcg: link page counters to root if use_hierarchy is false"). Thanks
> for taking care of that! I do agree that this approach is the most
> reasonable immediate fix.

Thanks!

> 
> Btw. we have been carrying a warning about non hierarchical hierarchies
> for years in our distribution kernels and haven't heard of any actual
> bug reports (except for LTP driven ones). So we might be really close to
> simply drop this functionality completely. This would simplify the code
> and prevent from future surprises.

Just sent an rfc patchset. Your feedback will be greatly appreciated!

Thanks!