From: Brad Boyer Subject: Re: [RFC][PATCH 0/7] Optimized MPEG file system Date: Wed, 15 Mar 2006 20:33:51 -0800 Message-ID: <20060316043351.GA5838@cynthia.pants.nu> References: <11424781042886-git-send-email-me@bobcopeland.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-fsdevel@vger.kernel.org Return-path: Received: from adsl-216-102-214-42.dsl.snfc21.pacbell.net ([216.102.214.42]:35512 "EHLO cynthia.pants.nu") by vger.kernel.org with ESMTP id S932442AbWCPFN2 (ORCPT ); Thu, 16 Mar 2006 00:13:28 -0500 To: Bob Copeland Content-Disposition: inline In-Reply-To: <11424781042886-git-send-email-me@bobcopeland.com> Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Wed, Mar 15, 2006 at 10:01:44PM -0500, Bob Copeland wrote: > - the disk has a structure that doesn't represent a file or a directory > (a continuation of the extent table for extremely fragmented files) > which otherwise operates the same as an inode. Could/should it be an > inode even if it wouldn't go into the dcache, i.e. just for > mark_inode_dirty? Right now it has its own write path. This sounds a lot like HFS/HFS+. Each file has a fixed number of extents in the main catalog record, and extras go into the extents overflow tree. If you look at fs/{hfs,hfsplus}/btree.c, you can find the code handling the catalog and extents overflow trees for HFS and HFS+. The initialization happens in the function hfs_btree_open(). In particular, it fetches an inode out of the regular inode cache and then the regular page cache routines are used to read the data. Just as a warning, I wrote the original version of the HFS+ code, but I don't maintain it anymore and haven't touched it in ages. I'm also commenting based on a slightly out of date tree, but I don't think the parts I've mentioned have changed much recently. Brad Boyer flar@allandria.com