Re: [quicwg/base-drafts] Receiver's behavior on key update (#2791)

[MikkelFJ <[e-mail]>]

> An endpoint should not call it a violation when it receives a packet carrying an awfully outdated key phase, because such behavior would allow an on-path attacker to disrupt a connection, by buffering a and injecting a very old packet. That also means that there's no reason to define when an old key is dropped.

My point is the arrival of a packet encrypted with a very new key in a very old packet number. Exactly when that packet number is too old is the question. It relates to the sender MUST not send with an older key. Ultimately it comes down to not observing keys with overlapping ranges. We cannot say that a packet should always be dropped since it is valid to close in some cases, it is just hard to specify.

My mention of unprotect relates to my own text which uses your term "authenticate and decrypt" repetitively in order to be precise.

-- 
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/2791#issuecomment-503459242