From owner-freebsd-hackers Fri Apr 30 10:37:52 1999 Delivered-To: freebsd-hackers@freebsd.org Received: from ns1.yes.no (ns1.yes.no [195.204.136.10]) by hub.freebsd.org (Postfix) with ESMTP id 8794114BD2; Fri, 30 Apr 1999 10:37:48 -0700 (PDT) (envelope-from eivind@bitbox.follo.net) Received: from bitbox.follo.net (bitbox.follo.net [195.204.143.218]) by ns1.yes.no (8.9.1a/8.9.1) with ESMTP id TAA09130; Fri, 30 Apr 1999 19:37:44 +0200 (CEST) Received: (from eivind@localhost) by bitbox.follo.net (8.8.8/8.8.6) id TAA11676; Fri, 30 Apr 1999 19:37:43 +0200 (MET DST) Date: Fri, 30 Apr 1999 19:37:43 +0200 From: Eivind Eklund To: "Daniel C. Sobral" Cc: W Gerald Hicks , adrian@FreeBSD.ORG, freebsd-hackers@FreeBSD.ORG, wghicks@wghicks.bellsouth.net Subject: Re: Adding desktop support Message-ID: <19990430193743.B11334@bitbox.follo.net> References: <199904290441.AAA02012@bellsouth.net> <37281E66.7AAF71A3@newsguy.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.95.1i In-Reply-To: <37281E66.7AAF71A3@newsguy.com>; from Daniel C. Sobral on Thu, Apr 29, 1999 at 05:55:02PM +0900 Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Thu, Apr 29, 1999 at 05:55:02PM +0900, Daniel C. Sobral wrote: > W Gerald Hicks wrote: > > > > > I find it really curious that layered filesystems seem to never be > > > considered for the jobs they were created for. > > > > Filesystem layering is not in very good shape right now, although > > Eivind has been spotted making good progress. > > AFAIK, he is making the existing fs (more specifically nullfs and > maybe unionfs) work, not correcting any existing flaw in the > layering code. I'm fixing bugs in the layering code. I've also fixed a couple of bugs in NULLFS itself (seems to be "bit-rot", actually, due to people doing changes elsewhere without propagating them), but the main part of the work is on the infrastructure. Main problems so far: * v_object is attached to the vnode as a property, instead of having a call to get hold of it. In most cases, this is wrong. * The locking protocol for vnodes is NOT being followed. It seems to be more or less completely ignored. (That's a little unfair, but there is a whole host of bugs). My present state is that I have patches that seems to fix the first issue (but I'm not certain they're correct - they need me to re-review them, and to test them in an environment where locking actually works); these are at http://www.freebsd.org/~eivind/FixNULL.patch These give a locked vnode problem if run with the present kernel; I'm not sure if this is due to bugs in the locking in the kernel or bugs in the patches. In order to clean out the locking problems of the kernel (so I can work on problems that are only in my patches, instead of the interaction between a buggy kernel and my patches) I also have ported vnode_if.sh to perl and made it emit locking assertions; this is at http://www.freebsd.org/~eivind/vnode_if_asserts.pl I'm presently working privately on getting a kernel with locking assertions enabled to actually boot. I have not gotten far enough to be able to mount a read-write filesystem; I don't know how much longer it will take. As I said, there seems to be a lot of bugs. Eivind. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message