From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752943Ab3ABQaP (ORCPT ); Wed, 2 Jan 2013 11:30:15 -0500 Received: from mail-qa0-f41.google.com ([209.85.216.41]:61353 "EHLO mail-qa0-f41.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752709Ab3ABQaN (ORCPT ); Wed, 2 Jan 2013 11:30:13 -0500 Date: Wed, 2 Jan 2013 11:30:10 -0500 From: Tejun Heo To: Vivek Goyal Cc: lizefan@huawei.com, axboe@kernel.dk, containers@lists.linux-foundation.org, cgroups@vger.kernel.org, linux-kernel@vger.kernel.org, ctalbott@google.com, rni@google.com Subject: Re: [PATCH 23/24] cfq-iosched: collect stats from dead cfqgs Message-ID: <20130102163010.GC11220@mtj.dyndns.org> References: <1356726946-26037-1-git-send-email-tj@kernel.org> <1356726946-26037-24-git-send-email-tj@kernel.org> <20130102162415.GA4306@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20130102162415.GA4306@redhat.com> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hey, Vivek. On Wed, Jan 02, 2013 at 11:24:15AM -0500, Vivek Goyal wrote: > On Fri, Dec 28, 2012 at 12:35:45PM -0800, Tejun Heo wrote: > > To support hierarchical stats, it's necessary to remember stats from > > dead children. Add cfqg->dead_stats and make a dying cfqg transfer > > its stats to the parent's dead-stats. > > Hi Tejun, > > Why not directly transfer stats to cfqg->stats. IOW, what's the advantage > of maintaining dead_stats separately. Backward compatibility? The existing stat cgroupfs files expect to see non-recursive stats. > > + cfqg_stats_merge(&parent->dead_stats, &cfqg->stats); > > + cfqg_stats_merge(&parent->dead_stats, &cfqg->dead_stats); > > + cfqg_stats_reset(&cfqg->stats); > > + cfqg_stats_reset(&cfqg->dead_stats); > > Anyway group will be marked offline and later freed. So resetting stats > might not be required. Yeah, it isn't strictly necessary. I tried to transfer the residual IOs between offline and release first so the resetting. Kinda like them there tho. > In fact if we have a realiable way of resetting status then online/offline > infrastructure might not be required? I think per cpu stats will be a > problem though and that's why we probably require logic to online/offline > the group? Hmmm? What do you mean? Thanks. -- tejun