On Mar 6, 2015, at 4:21 AM, Jari Arkko <jari.arkko@xxxxxxxxx> wrote: >> If there are docs that specify what a host should do with such >> "malchecksummed" packets, I'm happy to put a pointer in there. If not, >> I'd rather avoid discussing it in this draft. > > That may be the best approach. That is, not start specifying protocol in this document, unless it is just a reference. > > FWIW, I re-read 1035 and 2181 and 768 quickly, and did not see such a statement. But I could have missed it. Or it may have been considered to be obvious :-) It is clear that the latter is the case. As Mark points out, if you get a UDP packet with a malformed checksum, you don't even know if it was actually a UDP packet. Given that, I withdraw my request that this document actually specify what to do when one cannot verify a UDP packet, unless there is a statement about that in some (probably early) UDP document. --Paul Hoffman
Attachment:
signature.asc
Description: Message signed with OpenPGP using GPGMail