From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933335Ab0J0OkM (ORCPT ); Wed, 27 Oct 2010 10:40:12 -0400 Received: from hera.kernel.org ([140.211.167.34]:52883 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932607Ab0J0OkK (ORCPT ); Wed, 27 Oct 2010 10:40:10 -0400 Message-ID: <4CC83994.4000208@kernel.org> Date: Wed, 27 Oct 2010 16:39:16 +0200 From: Tejun Heo User-Agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.2.11) Gecko/20101013 Lightning/1.0b2 Thunderbird/3.1.5 MIME-Version: 1.0 To: Eric Dumazet CC: Peter Zijlstra , Brian Gerst , x86@kernel.org, linux-kernel@vger.kernel.org, torvalds@linux-foundation.org, mingo@elte.hu Subject: Re: [PATCH] x86-32: Allocate irq stacks seperate from percpu area References: <1288158182-1753-1-git-send-email-brgerst@gmail.com> <1288159670.2652.181.camel@edumazet-laptop> <1288173442.15336.1490.camel@twins> <1288186405.2709.117.camel@edumazet-laptop> <4CC82C2F.1020707@kernel.org> <1288187870.2709.128.camel@edumazet-laptop> <4CC83067.5000009@kernel.org> <1288189461.2709.144.camel@edumazet-laptop> In-Reply-To: <1288189461.2709.144.camel@edumazet-laptop> X-Enigmail-Version: 1.1.1 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.3 (hera.kernel.org [127.0.0.1]); Wed, 27 Oct 2010 14:39:17 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 10/27/2010 04:24 PM, Eric Dumazet wrote: > Le mercredi 27 octobre 2010 à 16:00 +0200, Tejun Heo a écrit : > >> Heh, interesting table. What does the same code say on 64bit? Is it >> the same? >> > > Yes this is the same Weird, then why did the percpu code interleaved cpus 16-31 between node 0 and 1? Percpu layout code tries pretty hard to group cpus into percpu units according to NUMA mapping but if the nodes are too unbalanced that doing so would result in too big waste of address space, it gives up. I _think_ that's what happened with the weird 24:8 NUMA split on 32bit. So, the interesting part is cpus 16-31 not 0-15. -- tejun