On Wed, 21 Sep 2005, Elwyn Davies wrote:
Dear IPv6 WG Chairs,
I previously sent this mail to the list at the time of the wg meeting in
Paris
but there was no response. Has any decision been taken on how to move
forward
with the IPv6 suite going towards full standard? I believe these items
should
be looked at before RFC2460 goes forward.
==================================================
Resent message from 2 August 2005:
In the course of doing the security overview draft in v6ops, we identified a
number of issues that appear to be errata or issues with the main IPv6
standard.
These should potentially be fixed up as 2460 goes to full standard:
Points for full standard;
? (non-)Processing of Type 0 routing headers in hosts
? (non-)Processing of Type 2 routing headers in routers
This is necessary to be defined.
? Processing Extension Headers in Middleboxes -
o words about where headers are inspected need to be relaxed
o words about processing headers out of order need to be relaxed
o should intermediate nodes be allowed to discard packets with unknown
extension headers?
I think this can be relaxed: Any internmediate system can delete any
packet if policy dictates. This should not be very strict, but somehow
force the IPv6 conformance.
? Requiring that new extension headers use TLV format (and maybe
putting
the length into the fragmentation header) - to simplify skipping
headers
in middleboxes
This would be a good decision.
? Constraining new hop-by-hop options both as to number and function -
h-by-h should be designed either for simple fast path processing
(like
jumbo packets) or to explicitly cause a packet to be dumped to the
slow
path (like router alert).
It would be good if such a distinction could be done.
? Fragment reassembly algorithm - should explicitly forbid overlapped
fragments and possibly require that non-final fragments are (say) at
least 1024 bytes.
According the RFC2460 the minimal MTU must be 1280. So Each fragments -
except the last fragment packet should be at least 1280. - this is
implicitly written in RFC2460 spec: "The lengths of the fragments must be
chosen such that the resulting fragment packets fit within the MTU of the
path to the packets' destination(s)." I understand this, that each
fragment packet should be no smaller than MTU, which is at least 1280.
I think it is wise to make it more explicit.
This my opinion only.
Regards,
Janos Mohacsi
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------