Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761364AbXEPKTo (ORCPT ); Wed, 16 May 2007 06:19:44 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1758202AbXEPKTg (ORCPT ); Wed, 16 May 2007 06:19:36 -0400 Received: from mx1.redhat.com ([66.187.233.31]:55769 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757429AbXEPKTf (ORCPT ); Wed, 16 May 2007 06:19:35 -0400 Organization: Red Hat UK Ltd. Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SI4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 3798903 From: David Howells In-Reply-To: <20070318233008.GA32597093@melbourne.sgi.com> References: <20070318233008.GA32597093@melbourne.sgi.com> To: David Chinner Cc: lkml , linux-mm , linux-fsdevel Subject: Re: [PATCH 1 of 2] block_page_mkwrite() Implementation V2 X-Mailer: MH-E 8.0; nmh 1.1; GNU Emacs 22.0.50 Date: Wed, 16 May 2007 11:19:29 +0100 Message-ID: <18993.1179310769@redhat.com> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1790 Lines: 46 David Chinner wrote: > + ret = block_prepare_write(page, 0, end, get_block); As I understand the way prepare_write() works, this is incorrect. The start and end points passed to block_prepare_write() delimit the region of the page that is going to be modified. This means that prepare_write() doesn't need to fill it in if the page is not up to date. It does, however, need to fill in the region before (if present) and the region after (if present). Look at it like this: +---------------+ | | | | <-- Filled in by prepare_write() | | to-> |:::::::::::::::| | | | | <-- Filled in by caller | | offset->|:::::::::::::::| | | | | <-- Filled in by prepare_write() | | page-> +---------------+ However, page_mkwrite() isn't told which bit of the page is going to be written to. This means it has to ask prepare_write() to make sure the whole page is filled in. In other words, offset and to must be equal (in AFS I set them both to 0). With what you've got, if, say, 'offset' is 0 and 'to' is calculated at PAGE_SIZE, then if the page is not up to date for any reason, then none of the page will be updated before the page is written on by the faulting code. You probably get away with this in a blockdev-based filesystem because it's unlikely that the page will cease to be up to date. However, if someone adds a syscall to punch holes in files, this may change... David - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/