Yes it is. I wrote my original implementation from a description
in "Unix Internals: The New Frontiers" (sorry, can't remember
author of the book).
I'm not a member of USENIX, so thought I couldn't get a copy of
Bonwick's paper (wasn't on Sun's site either).
I found it last week, on a database in NZ, but haven't had chance
to read it.
My computer and I are a 'distance' apart at the moment. When we
are re-united (hopefully, this weekend), I'll place the code on
an ftp server somewhere - a patch against 2.0.0.
BTW: The two major users of objects in the kernel seem to be inodes
and buffer headers. Linus has (of course) done a v. good job of
implementing the allocation for these objects, so to use the
slab allocator for these seems pointless. Oh, well...
markhe