Subject: Re: Tunnel fs idea
To: Assar Westerlund <email@example.com>
From: Todd Vierling <firstname.lastname@example.org>
Date: 10/03/1997 07:59:36
On 3 Oct 1997, Assar Westerlund wrote:
: The good news: I have almost implemented this twice.
: The bad news: the first one works under SunOS and is only partially
: ported to NetBSD. If you are really interested and willing to hack on
: it, send me mail and I can send you the code.
I'd like to look at it. I may tear it down to the bare bones and rebuild it
for NetBSD, but if I decide to use it, you'll certainly get copyright
: The second one works under HP/UX and you can't have it (yet). Sorry.
<cough> Excuse me. <grin> I guess I'm just a BSD person at heart.
: They both work by communicating with a daemon over a character device
: and keeping a small cache of vnodes in the kernel that's serviced by
: the daemon.
The perspective I had was similar, but with removal of the character device
completely from the picture--using a unix domain socket instead. This
difference severely reduces security concerns, since the socket is never
exposed to userland programs.
It wouldn't be using a "cache" of vnodes; it would use _real_ vnodes, with
the vnode "data pointer" as an opaque 'void *' that only the userland
program would understand.
== Todd Vierling (Personal email@example.com; Business firstname.lastname@example.org)
== I know you like the Internet, Bobby. Now go eat your Frosted Flakes.