From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753098AbZBQEFZ (ORCPT ); Mon, 16 Feb 2009 23:05:25 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751270AbZBQEFL (ORCPT ); Mon, 16 Feb 2009 23:05:11 -0500 Received: from fgwmail6.fujitsu.co.jp ([192.51.44.36]:36349 "EHLO fgwmail6.fujitsu.co.jp" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751064AbZBQEFK (ORCPT ); Mon, 16 Feb 2009 23:05:10 -0500 Date: Tue, 17 Feb 2009 13:03:52 +0900 From: KAMEZAWA Hiroyuki To: balbir@linux.vnet.ibm.com Cc: linux-mm@kvack.org, Sudhir Kumar , YAMAMOTO Takashi , Bharata B Rao , Paul Menage , lizf@cn.fujitsu.com, linux-kernel@vger.kernel.org, KOSAKI Motohiro , David Rientjes , Pavel Emelianov , Dhaval Giani , Rik van Riel , Andrew Morton Subject: Re: [RFC][PATCH 0/4] Memory controller soft limit patches (v2) Message-Id: <20090217130352.4ba7f91c.kamezawa.hiroyu@jp.fujitsu.com> In-Reply-To: <20090217030526.GA20958@balbir.in.ibm.com> References: <20090216110844.29795.17804.sendpatchset@localhost.localdomain> <20090217090523.975bbec2.kamezawa.hiroyu@jp.fujitsu.com> <20090217030526.GA20958@balbir.in.ibm.com> Organization: FUJITSU Co. LTD. X-Mailer: Sylpheed 2.5.0 (GTK+ 2.10.14; i686-pc-mingw32) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 17 Feb 2009 08:35:26 +0530 Balbir Singh wrote: > * KAMEZAWA Hiroyuki [2009-02-17 09:05:23]: > > > On Mon, 16 Feb 2009 16:38:44 +0530 > > Balbir Singh wrote: > > > > > > > > From: Balbir Singh > > > > > > Changelog v2...v1 > > > 1. Soft limits now support hierarchies > > > 2. Use spinlocks instead of mutexes for synchronization of the RB tree > > > > > > Here is v2 of the new soft limit implementation. Soft limits is a new feature > > > for the memory resource controller, something similar has existed in the > > > group scheduler in the form of shares. The CPU controllers interpretation > > > of shares is very different though. We'll compare shares and soft limits > > > below. > > > > > > Soft limits are the most useful feature to have for environments where > > > the administrator wants to overcommit the system, such that only on memory > > > contention do the limits become active. The current soft limits implementation > > > provides a soft_limit_in_bytes interface for the memory controller and not > > > for memory+swap controller. The implementation maintains an RB-Tree of groups > > > that exceed their soft limit and starts reclaiming from the group that > > > exceeds this limit by the maximum amount. > > > > > > This is an RFC implementation and is not meant for inclusion > > > > > > > some thoughts after reading patch. > > > > 1. As I pointed out, cpuset/mempolicy case is not handled yet. > > That should be esy to do with zonelists passed from reclaim path > plz do. > > > 2. I don't like to change usual direct-memory-reclaim path. It will be obstacles > > for VM-maintaners to improve memory reclaim. memcg's LRU is designed for > > shrinking memory usage and not for avoiding memory shortage. IOW, it's slow routine > > for reclaiming memory for memory shortage. > > I don't think I agree here. Direct reclaim is the first indication of > shortage and if order 0 pages are short, memcg's above their soft > limit can be targetted first. > My "slow" means "the overhead seems to be big". The latency will increase. About 0-order In patch 4/4 + did_some_progress = mem_cgroup_soft_limit_reclaim(gfp_mask); + /* should be if (!order) did_some_progress = mem.... I don't want to add any new big burden to kernel hackers of memory management, they work hard to improve memory reclaim. This patch will change the behavior. BTW, in typical bad case, several threads on cpus goes into memory recalim at once and all thread will visit this memcg's soft-limit tree at once and soft-limit will not work as desired anyway. You can't avoid this problem at alloc_page() hot-path. > > 3. After this patch, res_counter is no longer for general purpose res_counter... > > It seems to have too many unnecessary accessories for general purpose. > > Why not? Soft limits are a feature of any controller. The return of > highest ancestor might be the only policy we impose right now. But as > new controllers start using res_counter, we can clearly add a policy > callback. > I think you forget that memroy cgroups is an only controller in which the kernel can reduce the usage of resource without any harmful to users. soft-limit is nonsense for general resources, I think. -Kame