Re: [quicwg/base-drafts] Path validation should be done even for what looks like NAT rebinding (#2579)

[erickinnear <[e-mail]>]

I have a branch out (PR coming soon) that removes this -- this part of the attack is something I think @martinduke was looking at a bit. 

To @DavidSchinazi's point, the change of CID is a really strong signal that the client intends to migrate, but we may want to move that into a note along with the section we have for "server should use heuristics/be smart about what to choose", since the definition in the text above doesn't really change the requirements on the server's behavior and that may confuse people. 

Whether or not we define it as migration really doesn't change a whole lot, since we say (or should say) that the server has to validate the new path anyways. In other words, always validate everything, unless it's something that's previously passed path validation and not the one you're migrating away from. 

@igorlord, note that the attack also relies on the attacker being able to race and beat the original client's packet to the server, but there's a whole section for that sort of attack (although I think we need to add to it a bit, or make a variant for the attacker overloading the NAT).

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/2579#issuecomment-478809319