From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755957AbZAJWnz (ORCPT ); Sat, 10 Jan 2009 17:43:55 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752879AbZAJWnr (ORCPT ); Sat, 10 Jan 2009 17:43:47 -0500 Received: from mx3.mail.elte.hu ([157.181.1.138]:46979 "EHLO mx3.mail.elte.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752556AbZAJWnq (ORCPT ); Sat, 10 Jan 2009 17:43:46 -0500 Date: Sat, 10 Jan 2009 23:43:23 +0100 From: Ingo Molnar To: Mike Travis Cc: Ingo Molnar , Rusty Russell , Yinghai Lu , Jack Steiner , linux-kernel@vger.kernel.org Subject: Re: [PATCH 0/4] irq: change irq_desc and kstat_irq_legacy to variable sized arrays Message-ID: <20090110224323.GA17917@elte.hu> References: <20090110223818.459493000@polaris-admin.engr.sgi.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20090110223818.459493000@polaris-admin.engr.sgi.com> User-Agent: Mutt/1.5.18 (2008-05-17) X-ELTE-VirusStatus: clean X-ELTE-SpamScore: -1.5 X-ELTE-SpamLevel: X-ELTE-SpamCheck: no X-ELTE-SpamVersion: ELTE 2.0 X-ELTE-SpamCheck-Details: score=-1.5 required=5.9 tests=BAYES_00 autolearn=no SpamAssassin version=3.2.3 -1.5 BAYES_00 BODY: Bayesian spam probability is 0 to 1% [score: 0.0000] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Mike Travis wrote: > > The following patches change irq_desc and kstat_irq_legacy into > variable sized arrays based on nr_cpu_ids when CONFIG_SPARSE_IRQS=y. > > irq: change references from NR_IRQS to nr_irqs > irq: allocate irq_desc_ptrs array based on nr_irqs > irq: initialize nr_irqs based on nr_cpu_ids > kstat: modify kstat_irqs_legacy to be variable sized > > Based on: tip/cpus4096 @ v2.6.28-6140-g36c401a > > (Ingo - I will push these to your tip/cpus4096 branch via my cpus4096-for-ingo > git tree.) Thanks - please send a pull request when you feel good about them. The changes look good to me. > Affects of the SPARSE changes on NR_CPUS values. > > 1 - 128-defconfig (non-SPARSE) > 2 - 4k-defconfig (non-SPARSE) > 3 - 4k-defconfig (SPARSE) > > ====== Data > > .1. .2. .3. ..final.. > 1114112 . -1114112 . -100% irq_desc(.data.cacheline_aligned) > 208896 -69632 -138752 512 -99% irq_cfgx(.data) > 34816 . -34816 . -100% irq_timer_state(.bss) > 17480 . -17480 . -100% per_cpu__kstat(.data.percpu) > 0 . +4096 4096 . irq_desc_legacy(.data.cacheline_aligned) Impressive! If i remember your previous figures correctly we've got about +4MB of memory bloat at 4K CPUs, that is now down to 3MB total? Ingo