state of nullfs?
Andrew Atrens
atrens at nortelnetworks.com
Fri Feb 20 13:01:14 PST 2004
Hi Folks,
I'd like to use nullfs as a template for a new fs - actually for wrapping a linux fs
(clearcase mvfs to be exact). I've been warned off this a couple of times, but thought
it might be interesting, even if it turns out to be just as a learning exercise.
I suppose this could be generalized to allow support for any linux fs as a loadable
module, but since the linux and bsd vfs's differ greatly, that's not my initial goal.
Clearcase Mvfs (multi-version filesystem) is a bit special in that the backend is a
database of sorts, so nullfs seemed like a decent fit. What's changed since the last
time I considered this, is that the most recent version of Clearcase, along with
providing a binary mvfs.o module also provides in source code form something called
vnode.o. Also a module, but encapsulating all of their linux kernel interfaces. So
they keep their proprietary stuff in one module that gets distributed in binary form,
along with a second compileable module that can be rebuilt against the target kernel's
header files.
Anyway I'm rambling, back to nullfs. The main reason I'm asking about the state of
nullfs is the warning at the bottom of the mount_null man page -
BUGS
THIS FILESYSTEM TYPE IS NOT YET FULLY SUPPORTED (READ: IT DOESN'T WORK)
AND USING IT MAY, IN FACT, DESTROY DATA ON YOUR SYSTEM. USE AT YOUR OWN
RISK. BEWARE OF DOG. SLIPPERY WHEN WET.
This code also needs an owner in order to be less dangerous - serious
hackers can apply by sending mail to <hackers at xxxxxxxxxxx> and announcing
their intent to take it over.
HISTORY
The mount_null utility first appeared in 4.4BSD.
DragonFly 1.0 May 1, 1995 DragonFly 1.0
So, does anyone know if it's still in a useable state ?
Cheers,
Andrew.
More information about the Kernel
mailing list