[Hipsec] next steps/issues in interopability testing

>From our Friday morning unofficial BOF, a few issues that will need
attention from all implementors:

New cookie mechanism needs to be settled and implemented.

New HIT masking for IPv6 APIs needs to be documented and implemented. (HITs will be 127 bits of hash, but when passed accross APIs that want an IPv6 address, only the bottom 112 bits of hash will be passed, and the top bits will be a 16-bit constant assigned from the IPv6 address space by IANA. Inside the HIP implementation, there will need to be a copy of the full HIT kept and a mechanism to find the full hit from just the bottom 112 bits.)

IPv6 TCP & UDP checksum incremental correction mechanism needs to be documented and implemented. (We want the entire HIT to be used in the psuedo header checksum. This can be accomplished by incremental update of the 16-bit TCP/UDP checksum field before handing the packet off to TCP or UDP.)

(Each of these will require further expanded explanation. I just wanted to capture this much to the hipsec list before we adjourned the unofficial BOF.)

Pekka raises one more issue: do we need to do a hipkey thing like ipseckey did? (To be discussed later.)

			-Tim Shepard
			 shep@alum.mit.edu
_______________________________________________