From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752597Ab0IHNaV (ORCPT ); Wed, 8 Sep 2010 09:30:21 -0400 Received: from one.firstfloor.org ([213.235.205.2]:41981 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750840Ab0IHNaT (ORCPT ); Wed, 8 Sep 2010 09:30:19 -0400 From: Andi Kleen To: Aaron Sowry Cc: linux-kernel@vger.kernel.org, Pierre Ossman Subject: Re: Help decoding oom-killer output References: <4C8744D1.1010405@cendio.se> Date: Wed, 08 Sep 2010 15:30:17 +0200 In-Reply-To: <4C8744D1.1010405@cendio.se> (Aaron Sowry's message of "Wed, 08 Sep 2010 10:09:53 +0200") Message-ID: <87r5h4wefq.fsf@basil.nowhere.org> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Aaron Sowry writes: > After running in to problems with the amount of vmalloc memory on a SLED > 11 SP1 system (due to a large number of ncp mounts), we have increased > this with the kernel parameter 'vmalloc=512M'. This means the 32bit kernel has only about 400MB low mem virtual address space left, large parts of it being used by the mem_map arrays. So whole kernel has to run in 100-200MB or so which is likely not enough. > > Since this change, oom-killer seems to think we are running out of > memory, Yes you will run out of address space all the time for kernel operations. Real fix is to use 64bit instead. 32bit is so 90ies. -Andi -- ak@linux.intel.com -- Speaking for myself only.