From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from sd0109e.au.ibm.com (d23rh905.au.ibm.com [202.81.18.225]) by e23smtp06.au.ibm.com (8.13.1/8.13.1) with ESMTP id m816OtBx014199 for ; Mon, 1 Sep 2008 16:24:55 +1000 Received: from d23av04.au.ibm.com (d23av04.au.ibm.com [9.190.235.139]) by sd0109e.au.ibm.com (8.13.8/8.13.8/NCO v9.0) with ESMTP id m816PeiR257386 for ; Mon, 1 Sep 2008 16:25:40 +1000 Received: from d23av04.au.ibm.com (loopback [127.0.0.1]) by d23av04.au.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id m816Pe0b001513 for ; Mon, 1 Sep 2008 16:25:40 +1000 Message-ID: <48BB8AE3.7070704@linux.vnet.ibm.com> Date: Mon, 01 Sep 2008 11:55:39 +0530 From: Balbir Singh Reply-To: balbir@linux.vnet.ibm.com MIME-Version: 1.0 Subject: Re: [RFC][PATCH] Remove cgroup member from struct page References: <20080831174756.GA25790@balbir.in.ibm.com> <20080901090102.46b75141.kamezawa.hiroyu@jp.fujitsu.com> <48BB6160.4070904@linux.vnet.ibm.com> <20080901130351.f005d5b6.kamezawa.hiroyu@jp.fujitsu.com> <48BB8716.5090805@linux.vnet.ibm.com> <20080901152424.d9adfe47.kamezawa.hiroyu@jp.fujitsu.com> In-Reply-To: <20080901152424.d9adfe47.kamezawa.hiroyu@jp.fujitsu.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux-mm@kvack.org Return-Path: To: KAMEZAWA Hiroyuki Cc: Andrew Morton , hugh@veritas.com, menage@google.com, xemul@openvz.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, "nickpiggin@yahoo.com.au" List-ID: KAMEZAWA Hiroyuki wrote: > On Mon, 01 Sep 2008 11:39:26 +0530 > Balbir Singh wrote: > > >>> The development of lockless-page_cgroup is not stalled. I'm just waiting for >>> my 8cpu box comes back from maintainance... >>> If you want to see, I'll post v3 with brief result on small (2cpu) box. >>> >> I understand and I am not pushing you to completing it, but at the same time I >> don't want to queue up behind it for long. I suspect the cost of porting >> lockless page cache on top of my patches should not be high, but I'll never know >> till I try :) >> > My point is, your patch adds big lock. Then, I don't have to do meaningless effort > to reduce lock. My patch does not add a big lock, it moves the lock from struct page->page_cgroup to struct page_cgroup. The other locking added is the locking overhead associated with inserting entries into the radix tree, true. I ran oprofile along with lockdep and lockstats enabled on my patches. I don't see the radix_tree or page_cgroup->lock showing up, I see __slab_free and __slab_alloc showing up. I'll poke a little further. Please don't let my patch stop you, we'll integrate the best of both worlds and what is good for memcg. -- Balbir -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org