From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261752AbULJQwE (ORCPT ); Fri, 10 Dec 2004 11:52:04 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261757AbULJQwE (ORCPT ); Fri, 10 Dec 2004 11:52:04 -0500 Received: from holomorphy.com ([207.189.100.168]:2450 "EHLO holomorphy.com") by vger.kernel.org with ESMTP id S261752AbULJQwB (ORCPT ); Fri, 10 Dec 2004 11:52:01 -0500 Date: Fri, 10 Dec 2004 08:51:50 -0800 From: William Lee Irwin III To: Andrea Arcangeli Cc: Thomas Gleixner , Andrew Morton , marcelo.tosatti@cyclades.com, LKML , nickpiggin@yahoo.com.au Subject: Re: [PATCH] oom killer (Core) Message-ID: <20041210165150.GO2714@holomorphy.com> References: <1101938767.13353.62.camel@tglx.tec.linutronix.de> <20041202033619.GA32635@dualathlon.random> <1101985759.13353.102.camel@tglx.tec.linutronix.de> <1101995280.13353.124.camel@tglx.tec.linutronix.de> <20041202164725.GB32635@dualathlon.random> <20041202085518.58e0e8eb.akpm@osdl.org> <20041202180823.GD32635@dualathlon.random> <1102013716.13353.226.camel@tglx.tec.linutronix.de> <20041202233459.GF32635@dualathlon.random> <20041203022854.GL32635@dualathlon.random> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20041203022854.GL32635@dualathlon.random> Organization: The Domain of Holomorphy User-Agent: Mutt/1.5.6+20040722i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Dec 03, 2004 at 03:28:54AM +0100, Andrea Arcangeli wrote: > Ok, I expect this patch to fix the problem completely. The biggest > difference is that it doesn't affect the exit fast path and it doesn't need > notification. It's not even slower than your patch since you were > really polling too. This schedule properly, if you get any PREEMPT=n > problem I'd really like to know. This is on top of my previous patch so > it does the checks for the watermarks correctly (those are not obsoleted > by whatever thing we do in out_of_memory()). This still make use of the > PF_MEMDIE info but not in kernel/, only in mm/oom_kill.c where it born > and dies there, so the race is somewhat hidden in there. As you said > converting PF_MEMDIE to a set_tsk_thread_flag or some other non-racy > thing is a due change but I'm not doing it in the below patch. I see some real bugfixes in here. I'll start up a fresh thread with some of those broken out and cc: the relevant people shortly. It's not really any "new material", just reorganizing things into smaller patches, in particular the ones I see needing merging ASAP that aren't controversial. -- wli