[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <alpine.DEB.2.00.1002041335140.6071@chino.kir.corp.google.com>
Date: Thu, 4 Feb 2010 13:39:08 -0800 (PST)
From: David Rientjes <rientjes@...gle.com>
To: Jiri Kosina <jkosina@...e.cz>
cc: Lubos Lunak <l.lunak@...e.cz>,
Balbir Singh <balbir@...ux.vnet.ibm.com>,
Rik van Riel <riel@...hat.com>, linux-mm@...ck.org,
linux-kernel@...r.kernel.org,
Andrew Morton <akpm@...ux-foundation.org>,
KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>,
Nick Piggin <npiggin@...e.de>
Subject: Re: Improving OOM killer
On Thu, 4 Feb 2010, Jiri Kosina wrote:
> Why does OOM killer care about forkbombs *at all*?
>
Because the cumulative effects of a forkbomb are detrimental to the
system and the badness() heursitic favors large memory consumers very
heavily. Thus, the forkbomb is never really a strong candidate for oom
kill since the parent may consume very little memory itself and meanwhile
KDE or another large memory consumer will get innocently killed instead as
a result.
> If we really want kernel to detect forkbombs (*), we'd have to establish
> completely separate infrastructure for that (with its own knobs for tuning
> and possibilities of disabling it completely).
>
That's what we're trying to do, we can look at the shear number of
children that the parent has forked and check for it to be over a certain
"forkbombing threshold" (which, yes, can be tuned from userspace), the
uptime of those children, their resident set size, etc., to attempt to
find a sane heuristic that penalizes them.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists