Subject: Re: New hash algorithm - FNV - for use in the kernel
To: None <tech-perform@netbsd.org, tech-kern@netbsd.org>
From: Thor Lancelot Simon <tls@rek.tjls.com>
List: tech-perform
Date: 11/27/2001 23:04:59
On Wed, Nov 28, 2001 at 12:31:29AM -0000, David Laight wrote:
> >  > But there is no reason why it shouldn't optimise for what it expects to
> >  > see most often....
> > 
> > And what would that be?  Solaris?  NetBSD?  Tru64?
> 
> This is NFS, all the implementations are clones of each other,

Um, that's nonsense.  There are numerous distinct NFS code bases:  Sun's,
which has changed significantly over time; the ancient Wisconsin code that
was available for 4.3BSD; the Macklem implementation in its NFSv2 
incarnation from Berkeley net/2; the Macklem implementation in its NFSv3
incarnation, which behaves significantly differently in some ways than its
ancestor; other Unix vendors' code such as SGI's and IBM's which is
occasionally updated from the Sun "trunk" but has diverged greatly over
time; the Linux implementation, which is probably the most alien and evil,
in its user-space incarnation, its version 2 kernel incarnation, and its
version 3 kernel incarnation, amd's user-space incarnation, and NetApp's
code.

I'm sure I've missed some.

> I'd guess that there is little difference between the implementations.

Again, that's not really true.

Thor