From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753364Ab2HGPWy (ORCPT ); Tue, 7 Aug 2012 11:22:54 -0400 Received: from lobo.ruivo.org ([173.14.175.98]:60528 "EHLO lobo.ruivo.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753130Ab2HGPWu (ORCPT ); Tue, 7 Aug 2012 11:22:50 -0400 Date: Tue, 7 Aug 2012 11:22:30 -0400 From: Aristeu Rozanski To: Tejun Heo Cc: Hugh Dickins , Aristeu Rozanski , linux-kernel@vger.kernel.org, Li Zefan , Hillf Danton , lennart@poettering.net Subject: Re: [PATCH v3 0/3] cgroup: add xattr support Message-ID: <20120807152230.GE20962@cathedrallabs.org> References: <20120702142925.795007114@napanee.usersys.redhat.com> <20120717204126.GE24336@google.com> <20120718221044.GJ24336@google.com> <20120720175959.GA32741@cathedrallabs.org> <20120720180459.GJ32763@google.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20120720180459.GJ32763@google.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 (Added Lennart to the discussion so he can help explaining how systemd will use the xattrs) On Fri, Jul 20, 2012 at 11:04:59AM -0700, Tejun Heo wrote: > Hello, Hugh, Aristeu. > > On Fri, Jul 20, 2012 at 01:59:59PM -0400, Aristeu Rozanski wrote: > > hm, not sure that's what Tejun meant. tmpfs uses anonymous memory for the file > > contents, so reuse that infrastructure to allocate space for the extended > > attributes the same way, instead of using kmem. > > > > First thing I can think of is to use whole pages for it to prevent further > > complexity. Shouldn't make much difference considering the usecases we have > > now (systemd and containers), right? > > Yeah, that's what I meant. The internal fragmentation is ugly but I > think that should do for now at least. while talking to Lennart, it seems there're a lot of ways the xattrs can be used in systemd. If we're concerned with the amount of memory used, why not create a new namespace just for this, limiting by definition the maximum value size? -- Aristeu