linux-kernel - Re: [PATCH v1 01/14] mm: introduce bpf struct ops for OOM handling

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] [day] [month] [year] [list]

Message-ID: <871pnd2uor.fsf@linux.dev>
Date: Wed, 08 Oct 2025 10:02:28 -0700
From: Roman Gushchin <roman.gushchin@...ux.dev>
To: Song Liu <liu.song.linuxdev@...il.com>
Cc: Song Liu <song@...nel.org>,  Andrii Nakryiko
 <andrii.nakryiko@...il.com>,  Martin KaFai Lau <martin.lau@...ux.dev>,
  Alexei Starovoitov <alexei.starovoitov@...il.com>,  Kumar Kartikeya
 Dwivedi <memxor@...il.com>,  linux-mm <linux-mm@...ck.org>,  bpf
 <bpf@...r.kernel.org>,  Suren Baghdasaryan <surenb@...gle.com>,  Johannes
 Weiner <hannes@...xchg.org>,  Michal Hocko <mhocko@...e.com>,  David
 Rientjes <rientjes@...gle.com>,  Matt Bobrowski
 <mattbobrowski@...gle.com>,  Alexei Starovoitov <ast@...nel.org>,  Andrew
 Morton <akpm@...ux-foundation.org>,  LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v1 01/14] mm: introduce bpf struct ops for OOM handling

Song Liu <liu.song.linuxdev@...il.com> writes:

> On Tue, Oct 7, 2025 at 7:15 PM Roman Gushchin <roman.gushchin@...ux.dev> wrote:
> [...]
>> >
>> > I am not sure what is the best option for cgroup oom killer. There
>> > are multiple options. Technically, it can even be a sysfs entry.
>> > We can use it as:
>> >
>> > # load and pin oom killers first
>> > $ cat /sys/fs/cgroup/user.slice/oom.killer
>> > [oom_a] oom_b oom_c
>> > $ echo oom_b > /sys/fs/cgroup/user.slice/oom.killer
>> > $ cat /sys/fs/cgroup/user.slice/oom.killer
>> > oom_a [oom_b] oom_c
>>
>> It actually looks nice!
>> But I expect that most users of bpf_oom won't use it directly,
>> but through some sort of middleware (e.g. systemd), so Idk if
>> such a user-oriented interface makes a lot of sense.
>>
>> > Note that, I am not proposing to use sysfs entries for oom killer.
>> > I just want to say it is an option.
>> >
>> > Given attach() can be implemented in different ways, we probably
>> > don't need to add it to bpf_struct_ops. But if that turns out to be
>> > the best option, I would not argue against it. OTOH, I think it is
>> > better to keep reg() and attach() separate, though sched_ext is
>> > using reg() for both options.
>>
>> I'm inclining towards a similar approach, except that I don't want
>> to embed cgroup_id into the struct_ops, but keep it in the link,
>> as Martin suggested. But I need to implement it end-to-end before I can
>> be sure that it's the best option. Working on it...
>
> If we add cgroup_id to the link, I guess this means we need the link
> (some fd in user space) to hold reference on the attachment of this
> oom struct_ops on this is cgroup. Do we also need this link to hold
> a reference on the cgroup?

Not necessarily. I agree that the struct_ops should not hold a reference
to the cgroup, it's better to do the opposite.
This is why the link can have cgroup_id, not cgroup pointer.
I think it's similar to Tejun's approach to embed cgroup_id into the
struct ops, but potentially more flexible.

Thanks!