[1239] in linux-net channel archive
Re: ip aliasing support ...
daemon@ATHENA.MIT.EDU (Alan Cox)
Tue Oct 24 23:55:49 1995
From: Alan Cox <alan@cymru.net>
To: linux-vger@wab-tis.rabobank.nl
Date: Tue, 24 Oct 1995 09:42:54 +0100 (BST)
Cc: alan@cymru.net, srb@cuci.nl, linux-net@vger.rutgers.edu
In-Reply-To: <XAA14549@sys3.pe1chl.ampr.org> from "Rob Janssen reading Linux mailinglist" at Oct 23, 95 11:08:21 pm
> > if its gets that big its a problem. I was going to linear search it on the
> > grounds its normally 16 items or less.
> When you are considering linear search anyway, the described hashed
> list will probably be faster. And more versatile at the same time.
I did actually sit and time a few strategies a long time ago. For a small
list the gain by the fact its tight trivial assembler output (move, move,
not xor and) far outweights any advantages - also remember that for the
alias list (which is what this is about) you have addr/mask pairs so a large
number of entries should not be needed.
Alan
