From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263402AbTJBRXW (ORCPT ); Thu, 2 Oct 2003 13:23:22 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263408AbTJBRXV (ORCPT ); Thu, 2 Oct 2003 13:23:21 -0400 Received: from parcelfarce.linux.theplanet.co.uk ([195.92.249.252]:58243 "EHLO www.linux.org.uk") by vger.kernel.org with ESMTP id S263402AbTJBRXU (ORCPT ); Thu, 2 Oct 2003 13:23:20 -0400 Date: Thu, 2 Oct 2003 18:23:19 +0100 From: viro@parcelfarce.linux.theplanet.co.uk To: Nikita Danilov Cc: Andrew Morton , linux-kernel@vger.kernel.org, linux-mm@kvack.org Subject: Re: 2.6.0-test6-mm2 Message-ID: <20031002172318.GF7665@parcelfarce.linux.theplanet.co.uk> References: <20031002022341.797361bc.akpm@osdl.org> <16252.23200.511369.466054@laputa.namesys.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <16252.23200.511369.466054@laputa.namesys.com> User-Agent: Mutt/1.4.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Oct 02, 2003 at 09:04:32PM +0400, Nikita Danilov wrote: > Andrew Morton writes: > > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.0-test6/2.6.0-test6-mm2/ > > > > . A large series of VFS patches from Al Viro which replace usage of > > file->f_dentry->d_inode->i_mapping with the new file->f_mapping. > > > > This is mainly so we can get disk hot removal right. > > What consequences does this have for (out-of-the-tree) file systems, > beyond s/->f_dentry->d_inode->i_mapping/->f_mapping/g ? None. It only matters for block device inodes. Out-of-tree fs is free to do whatever it does with inodes of regular files/directories/etc. In quite a few cases you can cut down on dereferencing that way, but that's covered by what you've mentioned. If you take a look at the patchset you'll see * change of method prototypes in block devices (aka "you don't need to start with bdev = inode->i_bdev, you get it from arguments" - check RD1--RD6 and you'll see) * a lot of places in mm/* that got aforementioned search-and-replace treatment * very few changes in fs code, most of them of the same variety (same search-and-replace) * couple of helper functions changed their prototypes: generic_write_checks() lost "inode" argument generic_osync_inode(), OTOH, got explicit address_space one. That's it.