From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751079AbXBNE2x (ORCPT ); Tue, 13 Feb 2007 23:28:53 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751177AbXBNE2x (ORCPT ); Tue, 13 Feb 2007 23:28:53 -0500 Received: from ns2.suse.de ([195.135.220.15]:55289 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751079AbXBNE2x (ORCPT ); Tue, 13 Feb 2007 23:28:53 -0500 Date: Wed, 14 Feb 2007 05:28:43 +0100 From: Nick Piggin To: Linux Kernel Mailing List , Linux Memory Management List Subject: Re: [patch] mm: NUMA replicated pagecache Message-ID: <20070214042843.GD7125@wotan.suse.de> References: <20070213060924.GB20644@wotan.suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20070213060924.GB20644@wotan.suse.de> User-Agent: Mutt/1.5.9i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Feb 13, 2007 at 07:09:24AM +0100, Nick Piggin wrote: > > Issues: > - Not commented. I want to change the interfaces around anyway. > - Breaks filesystems that use filemap_nopage, but don't call filemap_mkwrite > (eg. XFS). Fix is trivial for most cases. > - Haven't tested NUMA yet (only tested via a hack to do per-CPU replication) > - Would like to be able to control replication via userspace, and maybe > even internally to the kernel. > - Ideally, reclaim might reclaim replicated pages preferentially, however > I aim to be _minimally_ intrusive. > - Would like to replicate PagePrivate, but filesystem may dirty page via > buffers. Any solutions? (currently should mount with 'nobh'). Hmm, I guess we should be able to do this for pagecache of regular files, as filesystems should not have any business dirtying that.