Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <4555470c-ea4f-244b-ed40-9403df3f5e4f@bytedance.com>
Date:   Fri, 28 Jul 2023 10:34:38 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0)
 Gecko/20100101 Thunderbird/102.13.0
Subject: Re: [External] Re: [RFC PATCH 0/5] mm: Select victim memcg using
 BPF_OOM_POLICY
To:     Alan Maguire <alan.maguire@oracle.com>, hannes@cmpxchg.org,
        mhocko@kernel.org, roman.gushchin@linux.dev, ast@kernel.org,
        daniel@iogearbox.net, andrii@kernel.org
Cc:     bpf@vger.kernel.org, linux-kernel@vger.kernel.org,
        wuyun.abel@bytedance.com, robin.lu@bytedance.com
References: <20230727073632.44983-1-zhouchuyi@bytedance.com>
 <7dbaabf9-c7c6-478b-0d07-b4ce0d7c116c@oracle.com>
From:   Chuyi Zhou <zhouchuyi@bytedance.com>
In-Reply-To: <7dbaabf9-c7c6-478b-0d07-b4ce0d7c116c@oracle.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi,

在 2023/7/27 19:43, Alan Maguire 写道:
> On 27/07/2023 08:36, Chuyi Zhou wrote:
>> This patchset tries to add a new bpf prog type and use it to select
>> a victim memcg when global OOM is invoked. The mainly motivation is
>> the need to customizable OOM victim selection functionality so that
>> we can protect more important app from OOM killer.
>>
> 
> It's a nice use case, but at a high level, the approach pursued here
> is, as I understand it, discouraged for new BPF program development.
> Specifically, adding a new BPF program type with semantics like this
> is not preferred. Instead, can you look at using something like
> 
> - using "fmod_ret" instead of a new program type
> - use BPF kfuncs instead of helpers.
> - add selftests in tools/testing/selftests/bpf not samples.
> 
> There's some examples of how solutions have evolved from the traditional
> approach (adding a new program type, helpers etc) to using kfuncs etc on
> this list - for example HID-BPF and the BPF scheduler series - which
> should help orient you. There are presentations from Linux Plumbers 2022
> that walk through some of this too.
> 
> Judging by the sample program example, all you should need here is a way
> to override the return value of bpf_oom_set_policy() - a noinline
> function that by default returns a no-op. It can then be overridden by
> an "fmod_ret" BPF program.
> 
Indeed, I'll try to use kfuncs & fmod_ret.

Thanks for your advice.
--
Chuyi Zhou
> One thing you lose is cgroup specificity at BPF attach time, but you can
> always add predicates based on the cgroup to your BPF program if needed.
> 
> Alan
> 
>> Chuyi Zhou (5):
>>    bpf: Introduce BPF_PROG_TYPE_OOM_POLICY
>>    mm: Select victim memcg using bpf prog
>>    libbpf, bpftool: Support BPF_PROG_TYPE_OOM_POLICY
>>    bpf: Add a new bpf helper to get cgroup ino
>>    bpf: Sample BPF program to set oom policy
>>
>>   include/linux/bpf_oom.h        |  22 ++++
>>   include/linux/bpf_types.h      |   2 +
>>   include/linux/memcontrol.h     |   6 ++
>>   include/uapi/linux/bpf.h       |  21 ++++
>>   kernel/bpf/core.c              |   1 +
>>   kernel/bpf/helpers.c           |  17 +++
>>   kernel/bpf/syscall.c           |  10 ++
>>   mm/memcontrol.c                |  50 +++++++++
>>   mm/oom_kill.c                  | 185 +++++++++++++++++++++++++++++++++
>>   samples/bpf/Makefile           |   3 +
>>   samples/bpf/oom_kern.c         |  42 ++++++++
>>   samples/bpf/oom_user.c         | 128 +++++++++++++++++++++++
>>   tools/bpf/bpftool/common.c     |   1 +
>>   tools/include/uapi/linux/bpf.h |  21 ++++
>>   tools/lib/bpf/libbpf.c         |   3 +
>>   tools/lib/bpf/libbpf_probes.c  |   2 +
>>   16 files changed, 514 insertions(+)
>>   create mode 100644 include/linux/bpf_oom.h
>>   create mode 100644 samples/bpf/oom_kern.c
>>   create mode 100644 samples/bpf/oom_user.c
>>