From owner-freebsd-current Tue Nov 18 14:20:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id OAA28834 for current-outgoing; Tue, 18 Nov 1997 14:20:03 -0800 (PST) (envelope-from owner-freebsd-current) Received: from usr01.primenet.com (tlambert@usr01.primenet.com [206.165.6.201]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id OAA28764; Tue, 18 Nov 1997 14:19:58 -0800 (PST) (envelope-from tlambert@usr01.primenet.com) Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id PAA26416; Tue, 18 Nov 1997 15:18:34 -0700 (MST) From: Terry Lambert Message-Id: <199711182218.PAA26416@usr01.primenet.com> Subject: Re: cvs commit: src/sys/gnu/ext2fs ... To: phk@critter.freebsd.dk (Poul-Henning Kamp) Date: Tue, 18 Nov 1997 22:18:32 +0000 (GMT) Cc: cvs-committers@freebsd.org, current@freebsd.org In-Reply-To: <17977.879863525@critter.freebsd.dk> from "Poul-Henning Kamp" at Nov 18, 97 03:32:05 pm X-Mailer: ELM [version 2.4 PL23] Content-Type: text Sender: owner-freebsd-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > This commit rather heavyhandedly rectifies a mistake made long ago. > > I've pushed all the crud back over in ext2fs, ufs is our primary > filesystem, and it shouldn't be polluted the way ext2 did. > > I have no idea if ext2fs works after this (I have no way to test it), > but I've done my best. > > I have not studied in detail what the directory structure difference > is between ufs and ext2fs, if somebody looks at it, we may be able to > return to using the ufs_ VOP's with some amount of modification the > the underlying directory tweaking functions (ufs_direnter &c) but > this will be a task for somebody else. I simply don't care enough > for ext2fs to even start looking at it. Thank you for this change! There is (currently) insufficient divorce between the directory and file inode acquisition at this point for the code to be shared with EXT2FS (IMO). This was a good change; it has to get worse before it gets better, and until the original mistakes are corrected, anything on top of them is a kludge at best and a compounding of the mistakes at worst. Now if only we could attack the mistakes introduced by the rushed integration by CSRG, we might be able to then attack filesystem structure vs. content issues, and be able to share the code properly (ie: ONLY to get the POSIX semantics enforcement). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.