From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030758AbXCSX6G (ORCPT ); Mon, 19 Mar 2007 19:58:06 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030754AbXCSX6G (ORCPT ); Mon, 19 Mar 2007 19:58:06 -0400 Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:40490 "EHLO sunset.davemloft.net" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1030766AbXCSX6E (ORCPT ); Mon, 19 Mar 2007 19:58:04 -0400 Date: Mon, 19 Mar 2007 16:58:03 -0700 (PDT) Message-Id: <20070319.165803.55508162.davem@davemloft.net> To: akpm@linux-foundation.org Cc: clameter@sgi.com, linux-mm@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [QUICKLIST 1/5] Quicklists for page table pages V3 From: David Miller In-Reply-To: <20070319165329.b5830789.akpm@linux-foundation.org> References: <20070319233716.13775.45569.sendpatchset@schroedinger.engr.sgi.com> <20070319165329.b5830789.akpm@linux-foundation.org> X-Mailer: Mew version 5.1.52 on Emacs 21.4 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org From: Andrew Morton Date: Mon, 19 Mar 2007 16:53:29 -0700 > Would it provide a superior solution if we were to a) stop zeroing out the > pte's when doing a fullmm==1 teardown and b) go direct to the page allocator > for these pages? While you could avoid zero'ing them out, you certainly can't avoid reading them into the cpu caches. And for the PGDs you have to initialize these things partially to non-zero values on x86{,_64} on every new PGD you allocate, which is a complete waste of cpu cache dirtying. Avoiding this overhead alone justifies the quicklists I think. It's not just a "zero" thing, so GFP_ZERO cannot help you here. The more I think about it the more I like the quicklists.