Date: Mon, 21 Jan 2008 16:04:05 -0500 From: Jan Harkes <jaharkes@cs.cmu.edu> To: Robert Watson <rwatson@FreeBSD.org> Cc: freebsd-fs@FreeBSD.org Subject: Re: Coda on FreeBSD problem reports? Message-ID: <20080121210405.GB9253@cs.cmu.edu> In-Reply-To: <20080121203838.G4194@fledge.watson.org> References: <20080116085630.GA32361@pappardelle.tekno.chalmers.se> <20080117080359.U51764@fledge.watson.org> <20080118073445.GA30721@pappardelle.tekno.chalmers.se> <20080118095652.GC30721@pappardelle.tekno.chalmers.se> <20080118103952.D18977@fledge.watson.org> <20080118210621.GF7898@cs.cmu.edu> <20080118211556.T46437@fledge.watson.org> <20080119005938.Q53920@fledge.watson.org> <rmi63xn6yu9.fsf@fnord.ir.bbn.com> <20080121203838.G4194@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jan 21, 2008 at 08:40:30PM +0000, Robert Watson wrote: > On Mon, 21 Jan 2008, Greg Troxel wrote: > >> Coda support in NetBSD (-current and netbsd-4 branch) is mostly ok, so >> you may want to glance at that for reference while working on the >> FreeBSD support. > > I was looking at the NetBSD Coda parts over the weekend, but it seemed > that most of the bugs in the FreeBSD code were due to gradual rot as VFS > evolved and our Coda module source didn't remain code up. > > I notice that NetBSD still supports CODA_COMPAT_5 as well -- is there any > reason to keep on doing this, really? I understand originally it was > because Coda 5 remained under the BSD license and Coda 6 was under GPL, > but it's been a while since anyone did anything with Coda 5. Actually it was Coda 4 that remained under the BSD license. The difference between Coda 5 and 6 was the introduction of realms. So instead of going to /coda/usr/jaharkes, I now have to access /coda/coda.cs.cmu.edu/usr/jaharkes. The important kernel difference is that we went from 96-bit file identifiers to 128-bit by adding a 32-bit realm-id. Aside from the inconvenience of having longer pathnames the initial versions with the dynamic root were not neccesarily as stable. I think it took at least 5 or 6 iterations before we got to a point that was somewhat comparable in stability. Same thing with coda-6.9.x clients, which use only write-disconnected operation and although that code has been around for a long time, we were never forced to really to rely on it 100%, so all those little nagging problems that have been around for a long time but were hard to reproduce or pin down are now pretty much unavoidable. Jan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080121210405.GB9253>