From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758144AbYJJLsY (ORCPT ); Fri, 10 Oct 2008 07:48:24 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753844AbYJJLsP (ORCPT ); Fri, 10 Oct 2008 07:48:15 -0400 Received: from e1.ny.us.ibm.com ([32.97.182.141]:47190 "EHLO e1.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753813AbYJJLsO (ORCPT ); Fri, 10 Oct 2008 07:48:14 -0400 Date: Fri, 10 Oct 2008 04:48:11 -0700 From: "Paul E. McKenney" To: Lai Jiangshan Cc: Andi Kleen , linux-kernel@vger.kernel.org, mingo@elte.hu, rjw@sisk.pl, dipankar@in.ibm.com, tglx@linuxtronix.de Subject: Re: [PATCH] rudimentary tracing for Classic RCU Message-ID: <20081010114811.GB6633@linux.vnet.ibm.com> Reply-To: paulmck@linux.vnet.ibm.com References: <20081007071544.GC20740@one.firstfloor.org> <20081007152629.GH6384@linux.vnet.ibm.com> <20081007154939.GN20740@one.firstfloor.org> <20081007163401.GJ6384@linux.vnet.ibm.com> <20081007210947.GP20740@one.firstfloor.org> <20081007212215.GN6384@linux.vnet.ibm.com> <20081009010846.GA10188@linux.vnet.ibm.com> <48EDA2A6.9070707@cn.fujitsu.com> <20081009065529.GC24560@one.firstfloor.org> <48EDAD55.2000502@cn.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <48EDAD55.2000502@cn.fujitsu.com> User-Agent: Mutt/1.5.15+20070412 (2007-04-11) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Oct 09, 2008 at 03:05:57PM +0800, Lai Jiangshan wrote: > Andi Kleen wrote: > >> In this case, why not define it as: > >> static char buf[20*NR_CPUS + 100]; > > > > Actually you should near never use NR_CPUS now but always num_possible_cpus() > > (or even num_online_cpus()) Using NR_CPUS can lead to extreme waste > > of memory on kernels which are compiled for 4096 CPUs for example. > > > > And with num_possible_cpus() kmalloc is needed. > > > > -Andi > > I thought the default value of NR_CPUS is 32. > > if NR_CPUS is too large, I think using seq_file is good idea, > and rcuclassic_trace_mutex is not need too. I never have used seq_file, to be honest. I would welcome a patch from you applying it to my rcucb and rcudata patch set. Thanx, Paul