linux-kernel - Re: [PATCH v2] mm: terminate shrink

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: <20171210101709.GB20234@dhcp22.suse.cz>
Date:   Sun, 10 Dec 2017 11:17:09 +0100
From:   Michal Hocko <mhocko@...nel.org>
To:     Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
Cc:     surenb@...gle.com, akpm@...ux-foundation.org, hannes@...xchg.org,
        hillf.zj@...baba-inc.com, minchan@...nel.org,
        mgorman@...hsingularity.net, ying.huang@...el.com,
        linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        timmurray@...gle.com, tkjos@...gle.com
Subject: Re: [PATCH v2] mm: terminate shrink_slab loop if signal is pending

On Sat 09-12-17 17:08:42, Tetsuo Handa wrote:
> Suren Baghdasaryan wrote:
> > On Fri, Dec 8, 2017 at 6:03 AM, Tetsuo Handa
> > <penguin-kernel@...ove.sakura.ne.jp> wrote:
> > >> > >> This change checks for pending
> > >> > >> fatal signals inside shrink_slab loop and if one is detected
> > >> > >> terminates this loop early.
> > >> > >
> > >> > > This changelog doesn't really address my previous review feedback, I am
> > >> > > afraid. You should mention more details about problems you are seeing
> > >> > > and what causes them.
> > 
> > The problem I'm facing is that a SIGKILL sent from user space to kill
> > the least important process is delayed enough for OOM-killer to get a
> > chance to kill something else, possibly a more important process. Here
> > "important" is from user's point of view. So the delay in SIGKILL
> > delivery effectively causes extra kills. Traces indicate that this
> > delay happens when process being killed is in direct reclaim and
> > shrinkers (before I fixed them) were the biggest cause for the delay.
> 
> Sending SIGKILL from userspace is not releasing memory fast enough to prevent
> the OOM killer from invoking? Yes, under memory pressure, even an attempt to
> send SIGKILL from userspace could be delayed due to e.g. page fault.
> 
> Unless it is memcg OOM, you could try OOM notifier callback for checking
> whether there are SIGKILL pending processes and wait for timeout if any.

Hell no! You surely do not want all the OOM livelocks you were pushing
so hard to get fixed, do you?

The whole problem here is that there are two implementations of the OOM
handling and they do not use any synchronization. You cannot be really
surprise they step on each others toes. That is one of the reasons why I
really hated the LMK in the kernel btw.

Stalling shrinkers is a real problem and it should be addressed but
let's not screw an already nasty/fragile code all around that.
-- 
Michal Hocko
SUSE Labs