On 19-sep-2007, at 15:56, Brian Dickson wrote:
Iljitsch van Beijnum wrote:
I haven't read all your messages on the subject yet, but how do
you handle the universal/local bit that's in the 16 bits that you
want to reclaim?
The U/L bit is currently being munged on the high bit of the
EUI-64, which is the high bit of the OUI part of the MAC-48 address.
No, it's the 7th highest bit (bit 6 according to IETF bit numbering,
which is different from what we learn in software engineering school)
of the EUI-64. In ethernet MAC addresses, the first bit on the wire
is the group bit and the second the u/l bit. However, the bits are
sent most significant byte first, least significant bit first, so the
bit order within each byte is the opposite of what you would expect.
The 16 bits I want to reclaim, are in the *middle* of the EUI-64,
not at the top end. I want to reclaim the "FFFE" fixed value bits.
Well, ok. But the effect is the same: the u/l bit is no longer bit 70
in the IPv6 address but bit 86.
So, in the /80 case, the top bit of OUI gets set to 0 or 1
regardless of what it was before, *exactly* like it does in the /64
for EUI-64 use.
Yes, but now if I have a 128-bit IPv6 address I don't know where the
u/l bit is. I guess this isn't an issue for 128-bit addresses but
only when dealing with the lower 64 or 48 bits used to number a
subnet, and then the subnet size will tell you what you need to know.
This still breaks the notion that once you do duplicate address
detection for your MAC-derived link-local address you can assume that
any other MAC-derived addresses are also unique.
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------