linux-kernel - Re: [PATCH 2/5] Revert "oom: give the dying task a higher priority"

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] [day] [month] [year] [list]

Date:	Tue, 29 Mar 2011 11:46:33 +0900 (JST)
From:	KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
Cc:	kosaki.motohiro@...fujitsu.com,
	"Luis Claudio R. Goncalves" <lclaudio@...g.org>,
	Minchan Kim <minchan.kim@...il.com>,
	linux-kernel@...r.kernel.org,
	Andrew Morton <akpm@...ux-foundation.org>,
	David Rientjes <rientjes@...gle.com>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Rik van Riel <riel@...hat.com>,
	Oleg Nesterov <oleg@...hat.com>, linux-mm <linux-mm@...ck.org>,
	Andrey Vagin <avagin@...nvz.org>,
	Hugh Dickins <hughd@...gle.com>,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
Subject: Re: [PATCH 2/5] Revert "oom: give the dying task a higher priority"

Hi

> > I mean, in the context of SCHED_OTHER tasks would it really help the dying
> > task to be scheduled sooner to release its resources? 
> 
> That very much depends on how all this stuff works, I guess if everybody
> serializes on OOM and only the first will actually kill a task and all
> the waiting tasks will try to allocate a page again before also doing
> the OOM thing, and the pending tasks are woken after the OOM target task
> has completed dying.. then I don't see much point in boosting things,
> since everybody interested in memory will block and eventually only the
> dying task will be left running.

Probably I can answer this question. When OOM occur, kernel has very a
few pages (typically 10 - 100). but not 0. therefore bloody page-in vs
page-out battle (aka allocation vs free battle) is running.

IOW, While we have multiple cpu or per-cpu page queue, we don't see
page cache become completely 0.

Therefore, not killed task doesn't sleep completely. page-out may have
very small allocation successful chance. (but almostly it's fail. pages
are stealed by another task)

Before Luis's patch, kernel livelock on oom may be solved within 30min,
but after his patch, it's solved within 1 second. that's big different
for human response time. That's the test result.

Thanks.


--
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/