Re: draft-ietf-ipngwg-p2p-pingpong-00.txt vs RFC4443

[JINMEI Tatuya / 神明達哉 <[e-mail]>]

At Tue, 17 Aug 2010 08:02:07 -0300,
Fernando Gont <[e-mail]> wrote:

> >> It is clear that there is one more action done on the packet with 
> >> RFC4443. But this has no impact on shipping ASIC based routers. It
> >> is difficult to say though if some smaller routers could be
> >> impacted.
> > 
> > This, and what Ole Troan wrote on interface lookup, is interesting.
> > 
> > RFC4443 requires checking that destination address matches the subnet
> >  prefix.  Is this the hot issue?
> > 
> > Note that pingpong-00 document did not have this requirement;
> 
> AFAICT, it does. It says: "....and the destination address on the packet
> seems to be on-link (in terms of Neighbor Discovery) on the
> point-to-point interface". Or am I missing something?

(Aside from what is actually documented in the RFC/draft) the
additional check about on-link-ness or subnet matching was
intentionally added and that's what we implemented for KAME/*BSDs.

See, e.g.,
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/netinet6/ip6_forward.c?rev=1.68&content-type=text/x-cvsweb-markup
(around the comment beginning with "If the incoming interface is equal...")

As commented, we wanted to separate an inevitable loop with a valid
configuration from loops due to
configuration/operational/implementation errors (most typically a
temporary routing loop).  We'd rather catch the latter type of error
easily with usual diagnostic tools such as traceroute so that we can
find and fix it quickly.

---
JINMEI, Tatuya
Internet Systems Consortium, Inc.