Subject: Re: works in progress: route cache invalidation, RADIX_MPATH
To: David Young <dyoung@pobox.com>
From: Rui Paulo <rpaulo@fnop.net>
List: tech-net
Date: 11/18/2006 11:50:28
On Nov 16, 2006, at 10:18 PM, David Young wrote:
>> * RADIX_MPATH
>
> I don't know when I will get around to debugging and committing this.
> Not very soon, I'm afraid.
>
> RADIX_MPATH fixes a bug, albeit not a very popular one: IIRC, the
> Host Requirements Spec requires that the routing table can hold more
> than one default route, and NetBSD's will not. So we are not strictly
> compliant.
>
> RADIX_MPATH also makes it a lot easier for someone to bootstrap an
> experiment in wireless routing. Wireless routing is a hot topic these
> days. NetBSD is too often overlooked as a platform for that.
>
> I believe RADIX_MPATH will help us do things like operate in the IPv4
> link-local subnet (169.254/16) on more than one ethernet, without
> bridging
> all of the ethernets together. (Yes, I really do think that's
> useful. :-)
Yes, RADIX_MPATH would be a good addition to NetBSD. I've been also
thinking about bringing this code from KAME, but, unfortunately, I
don't have much time or enough machines to do the testing.
--
Rui Paulo