On 15 Apr 2009, at 19:13, Dino Farinacci wrote:
That's unclear to me. One of the arguments for dropping the layer 3
checksum in IPv6 is that modern layer 2 solutions come with error
detection. Making the layer 4 checksum compulsory strengthened
that argument, of course. But I don't remember discussion of
why the pseudo-header checksum is essential. In fact, when the IPNG
WG
was chartered in 11/94, the charter said:
TCP/UDP: The IPng Working Group will specify the procedures for
hosts to
compute and verify TCP/UDP pseudo-headers. Any other changes to TCP
beyond making TCP work with IPng are out of scope of the working
group
and should be dealt with by a TCPng Working Group.
So it was taken for granted that we needed it.
I recall the IPng directorate wanted to make the fewest changes to the
transport layers. The only changes that were required was to lengthen
the address and that is what turned out.
The same reason why the low-order 64-bits were not used as a socket-id
for TCP for tcb lookups.
All of the above is consistent with my recollections.
More is the pity, since decoupling the IPv6 addresses from
the upper-layer pseudo-header calculations would have done
IPv6 users a tremendous amount of good.
Mea culpa. AH for IPv6 should not have included the IPv6
Routing Prefixes either (although including the IPv6 IID
would have been sensible).
Ran
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
