[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: another padding question ...
Thanks for the note. I'm not completely clear about the
alignment requirement in the IPv6 context, and I think previous
messages on this list suggest that the lack of clarity is
widespread. Can you cite a specific IPv6 document that
indicates the granularity at which 64-bit alignment must be
maintained? We all understand the benefits of making the
length of AH be zero mod 8, because it affects the alignment of
the subsequent payload. However, in the ESP context, we're
just talking about the position of the Auth Data field at the
end of the packet. In some sense, it's just another field
within ESP, and not all the ESP (or AH) fields can be 8-byte
aligned.
Yup -- no higher-layer protocol than ESP will ever see the data.
The question of interest is whether or not a 64-bit machine will
be able to process or generate that field more quickly if it's
64-bit aligned. I strongly suspect that it doesn't matter -- it's
not a lot of data that needs copying, and the cost of calculating
the Auth Data field far dwarfs the cost of storing or retrieving
the result, even if unaligned.