Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 11 Apr 2019 14:16:33 +0200
From:   Michal Hocko <mhocko@kernel.org>
To:     Rik van Riel <riel@surriel.com>
Cc:     Suren Baghdasaryan <surenb@google.com>, akpm@linux-foundation.org,
        dancol@google.com, jannh@google.com, minchan@kernel.org,
        penguin-kernel@I-love.SAKURA.ne.jp, kernel-team@android.com,
        rientjes@google.com, linux-kernel@vger.kernel.org,
        willy@infradead.org, linux-mm@kvack.org, hannes@cmpxchg.org,
        shakeelb@google.com, jrdr.linux@gmail.com,
        yuzhoujian@didichuxing.com, joel@joelfernandes.org,
        timmurray@google.com, lsf-pc@lists.linux-foundation.org,
        guro@fb.com, christian@brauner.io, ebiederm@xmission.com
Subject: Re: [Lsf-pc] [RFC 0/2] opportunistic memory reclaim of a killed
 process
Message-ID: <20190411121633.GV10383@dhcp22.suse.cz>
References: <20190411014353.113252-1-surenb@google.com>
 <e1fc2c84f5ef2e1408f6fee7228a52a458990b31.camel@surriel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <e1fc2c84f5ef2e1408f6fee7228a52a458990b31.camel@surriel.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu 11-04-19 07:51:21, Rik van Riel wrote:
> On Wed, 2019-04-10 at 18:43 -0700, Suren Baghdasaryan via Lsf-pc wrote:
> > The time to kill a process and free its memory can be critical when
> > the
> > killing was done to prevent memory shortages affecting system
> > responsiveness.
> 
> The OOM killer is fickle, and often takes a fairly
> long time to trigger. Speeding up what happens after
> that seems like the wrong thing to optimize.
> 
> Have you considered using something like oomd to
> proactively kill tasks when memory gets low, so
> you do not have to wait for an OOM kill?

AFAIU, this is the point here. They probably have a user space OOM
killer implementation and want to achieve killing to be as swift as
possible.

-- 
Michal Hocko
SUSE Labs