From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ww0-f44.google.com ([74.125.82.44]:37124 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754583Ab0G2I3n convert rfc822-to-8bit (ORCPT ); Thu, 29 Jul 2010 04:29:43 -0400 Received: by wwj40 with SMTP id 40so71553wwj.1 for ; Thu, 29 Jul 2010 01:29:42 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: References: Date: Thu, 29 Jul 2010 10:29:38 +0200 Message-ID: Subject: Re: arm1136 with 2.6.29.5 running out of memory with 32k cache only, 16k and below is fine From: koen velle Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT Sender: linux-arm-msm-owner@vger.kernel.org List-ID: To: linux-arm-msm@vger.kernel.org Dear all, sorry for posting this to the wrong list, should have gone to linux-arm-kernel, please ignore. Koen. On Thu, Jul 29, 2010 at 10:14 AM, koen velle wrote: > Hi, > > I've got a (simulated) system where I can select my cache sizes.  Boot > goes fine with 4, 8, and 16k cache, but goes boink with 32k cache. > While debugging, I'm able to find out that in __do_page_fault, I end > up at the label out_of_memory, as handle_mm_fault returns > VM_FAULT_OOM. > > This does not happen with caches <32k.  Otherwise, both my system and > my linux config are identical, only the cache size is different. > > Is there anything that can explain this ? Do I need to modify the > kernel configuration somehow if I want to use 32k caches ? Does the > kernel consume more memory when using 32k caches due to dealing with > cache coloring ? > > In short, any easy answers ? > > Thanks, > > Koen. >