From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751661AbdIRDIk (ORCPT ); Sun, 17 Sep 2017 23:08:40 -0400 Received: from mga01.intel.com ([192.55.52.88]:45348 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751554AbdIRDIi (ORCPT ); Sun, 17 Sep 2017 23:08:38 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.42,410,1500966000"; d="scan'208";a="901191738" Subject: Re: [PATCH 2/3] mm: Handle numa statistics distinctively based-on different VM stats modes To: Michal Hocko Cc: "Luis R . Rodriguez" , Kees Cook , Andrew Morton , Jonathan Corbet , Mel Gorman , Johannes Weiner , Christopher Lameter , Sebastian Andrzej Siewior , Vlastimil Babka , kemi , Dave , Tim Chen , Andi Kleen , Jesper Dangaard Brouer , Ying Huang , Aaron Lu , Proc sysctl , Linux MM , Linux Kernel References: <1505467406-9945-1-git-send-email-kemi.wang@intel.com> <1505467406-9945-3-git-send-email-kemi.wang@intel.com> <20170915115049.vqthfawg3y4r6ogh@dhcp22.suse.cz> From: kemi Message-ID: <26bd25c8-294f-d3cc-8ba2-845a6da33fe5@intel.com> Date: Mon, 18 Sep 2017 11:07:20 +0800 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.2.1 MIME-Version: 1.0 In-Reply-To: <20170915115049.vqthfawg3y4r6ogh@dhcp22.suse.cz> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2017年09月15日 19:50, Michal Hocko wrote: > On Fri 15-09-17 17:23:25, Kemi Wang wrote: > [...] >> @@ -2743,6 +2745,17 @@ static inline void zone_statistics(struct zone *preferred_zone, struct zone *z) >> #ifdef CONFIG_NUMA >> enum numa_stat_item local_stat = NUMA_LOCAL; >> >> + /* >> + * skip zone_statistics() if vmstat is a coarse mode or zone statistics >> + * is inactive in auto vmstat mode >> + */ >> + >> + if (vmstat_mode) { >> + if (vmstat_mode == VMSTAT_COARSE_MODE) >> + return; >> + } else if (disable_zone_statistics) >> + return; >> + >> if (z->node != numa_node_id()) >> local_stat = NUMA_OTHER; > > A jump label could make this completely out of the way for the case > where every single cycle matters. > Could you be more explicit for how to implement it here. Thanks very much.