Re: [quicwg/base-drafts] Avoid attack on address validation during connection migration (#746)

[Martin Thomson <[e-mail]>]

I missed something... If the attacker can reliably win the race, then increasing the rigour of the source address validation won't fix that problem.  You need to have to have the client in the loop somehow, and you can't just tell the client that it moved.  Otherwise, this forwarding attack that you describe looks no different to a NAT rebinding (which is a situation we want to permit).

The primary concern here is that the attacker just put themselves on-path, right?  It's tempting to shrug and not worry about this one.  The performance of that path is clearly uniformly better than that of the old path and the address can't be spoofed or the bidirectional forwarding wouldn't work.

-- 
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/pull/746#issuecomment-324221300