DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:sender:in-reply-to:references:from:date
         :x-google-sender-auth:message-id:subject:to:cc:content-type;
        b=NhuTuY8/DnRkMP5x0WAAI4cfTyTGJ12Xj+Fszfa72r6pU9SwH/eCcWsfykRVBF66Nc
         2MmoNGH7vbeHMwRVqPwcYpiT+irCOrF2gltbCx4LSNfgPg8PGxVof6qzU8JHUhipIDN4
         HveY5EQ1Ygslwv6KiroFXjI8MnQTn0FksNaCo=
MIME-Version: 1.0
In-Reply-To: <20110519145147.GA14658@localhost>
References: <BANLkTikofp5rHRdW5dXfqJXb8VCAqPQ_7A@mail.gmail.com>
 <20110514165346.GV6008@one.firstfloor.org> <BANLkTik6SS9NH7XVSRBoCR16_5veY0MKBw@mail.gmail.com>
 <20110514174333.GW6008@one.firstfloor.org> <BANLkTinst+Ryox9VZ-s7gdXKa574XXqt5w@mail.gmail.com>
 <20110515152747.GA25905@localhost> <BANLkTim-AnEeL=z1sYm=iN7sMnG0+m0SHw@mail.gmail.com>
 <20110517060001.GC24069@localhost> <BANLkTi=TOm3aLQCD6j=4va6B+Jn2nSfwAg@mail.gmail.com>
 <BANLkTi=9W6-JXi94rZfTtTpAt3VUiY5fNw@mail.gmail.com> <20110519145147.GA14658@localhost>
From: Andrew Lutomirski <luto@mit.edu>
Date: Thu, 19 May 2011 11:00:24 -0400
Message-ID: <BANLkTim4caSU_uRY1GLMh9D=8gctp34jMA@mail.gmail.com>
Subject: Re: Kernel falls apart under light memory pressure (i.e. linking vmlinux)
To: Wu Fengguang <fengguang.wu@intel.com>
Cc: Minchan Kim <minchan.kim@gmail.com>, Andi Kleen <andi@firstfloor.org>,
        "linux-mm@kvack.org" <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>,
        KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
        KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
        Mel Gorman <mgorman@suse.de>, Johannes Weiner <hannes@cmpxchg.org>,
        Rik van Riel <riel@redhat.com>
Content-Type: text/plain; charset=ISO-8859-1
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1341
Lines: 35

On Thu, May 19, 2011 at 10:51 AM, Wu Fengguang <fengguang.wu@intel.com> wrote:
>> > I had 6GB swap available, so there shouldn't have
>> > been any OOM.
>>
>> Yes. It's strange but we have seen such case several times, AFAIR.
>
> I noticed that the test script mounted a "ramfs" not "tmpfs", hence
> the 1.4G pages won't be swapped?

That's intentional.

I run LVM over dm-crypt on an SSD, and I thought that might be part of
the problem.  I wanted a script that would see if I could reproduce
the problem without stressing that system too much, so I created a
second backing store on dm-crypt over ramfs so that no real I/O will
happen.  The script is quite effective at bringing down my system, so
I haven't changed it.

(I have 6GB of "real" swap on the LVM, so pinning 1500MB into RAM
ought to cause some thrashing but not take the system down.  And this
script with a larger ramfs does not take down my desktop, which is an
8GB Sandy Bridge box.  But whatever the underlying bug is seems to
mainly affect Sandy Bridge *laptops*, so maybe that's expected.)

--Andy

>
> Thanks,
> Fengguang
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/