From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754690Ab1H2SmG (ORCPT ); Mon, 29 Aug 2011 14:42:06 -0400 Received: from mx1.redhat.com ([209.132.183.28]:12199 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754323Ab1H2SmC (ORCPT ); Mon, 29 Aug 2011 14:42:02 -0400 Date: Mon, 29 Aug 2011 20:37:43 +0200 From: Oleg Nesterov To: KOSAKI Motohiro Cc: rientjes@google.com, akpm@linux-foundation.org, torvalds@linux-foundation.org, roland@hack.frob.com, tj@kernel.org, dvlasenk@redhat.com, matt.fleming@linux.intel.com, linux-kernel@vger.kernel.org, avagin@openvz.org, fhrbata@redhat.com, yinghan@google.com Subject: Re: mm->oom_disable_count is broken Message-ID: <20110829183743.GA15216@redhat.com> References: <20110727163159.GA23785@redhat.com> <20110727163610.GJ23793@redhat.com> <20110727175624.GA3950@redhat.com> <20110728154324.GA22864@redhat.com> <20110729141431.GA3501@redhat.com> <20110730143426.GA6061@redhat.com> <20110730152238.GA17424@redhat.com> <4E369372.80105@jp.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4E369372.80105@jp.fujitsu.com> User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 08/01, KOSAKI Motohiro wrote: > > > And this reminds me. mm->oom_disable_count looks absolutely broken. > > IIRC, I already complained but nobody replied. > > > > [...snip...] > > IIRC, I did pointed out this issue. But nobody replied. > I think ->oom_disable_count is currently broken. but now I have no time to > audit this stuff. So, I'd suggest to revert this code if nobody don't fix it. I tend to agree, of course we can fix oom_disable_count but I don't really understand why do we want it. David, could you please explain? I mean, CLONE_VM (without CLONE_THREAD) is not that common, I think. Does this counter really help in practice? And. personally I dislike it because ->oom_disable_count is just another proof that ->oom_score_adj should be in ->mm, not per-process. IIRC, you already explained me why we can't do this, but - sorry - I forgot. May be something with vfork... Could you explain this again? Oleg.