From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753629Ab0CCENU (ORCPT ); Tue, 2 Mar 2010 23:13:20 -0500 Received: from hera.kernel.org ([140.211.167.34]:41061 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752828Ab0CCENT (ORCPT ); Tue, 2 Mar 2010 23:13:19 -0500 Message-ID: <4B8DE479.1030308@kernel.org> Date: Wed, 03 Mar 2010 13:24:25 +0900 From: Tejun Heo User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.5) Gecko/20091130 SUSE/3.0.0-1.1.1 Thunderbird/3.0 MIME-Version: 1.0 To: werner@guyane.dyn-o-saur.com CC: akpm@linux-foundation.org, linux-kernel@vger.kernel.org Subject: Re: 2.6.33 problems References: <1267587210.10602@guyane.dyn-o-saur.com> In-Reply-To: <1267587210.10602@guyane.dyn-o-saur.com> X-Enigmail-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (hera.kernel.org [127.0.0.1]); Wed, 03 Mar 2010 04:13:14 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hello, On 03/03/2010 12:33 PM, werner@guyane.dyn-o-saur.com wrote: > dmesg isn't complete. That's a general problem. When we get 1 GB > dmesg buffer in the kernel config ???? Kernel log can be incomplete for many reasons. In your case klogd wasn't logging debug messages so I couldn't see pcpu debug info. dmesg w/ sufficiently large -s combined with large kernel buffer usually works. > Linux version 2.6.33-git7 (root@werner) (gcc version 4.3.3 (GCC) ) #1 SMP PREEMPT Tue Mar 2 06:15:19 GFT 2010 ... > setup_percpu: NR_CPUS:32 nr_cpumask_bits:32 nr_cpu_ids:2 nr_node_ids:1 > PERCPU: Embedded 14 pages/cpu @c3c00000 s35876 r0 d21468 u2097152 > pcpu-alloc: s35876 r0 d21468 u2097152 alloc=1*4194304 > pcpu-alloc: [0] 0 1 Here, the unit size is 2M and the chunk size 4M for the two possible cpus. This isn't excessive at all and i386 shouldn't have much problem accomodating it. Have you tried percpu_alloc=page? Thanks. -- tejun