Please explain the difference between TLS re-negotiation and resumption.
My understanding is as follows: in the context of TLS 1.2, re-negotiation and resumption look the same, but the former can be started with either a HelloRequest/ClientHello from server/client while the latter can start with just with ClientHello from the client. In both cases, a client-server dialog followed by an abbreviated handshake occurs (or not ?), security parameters do not change, and new keying materials are recalculated using an "old" pre-master secret with new random numbers.
From the record layer point of view, all re-negotiation/resumption messages should be processed using "current record state"-- which means they are encrypted/authenticated using "current" keys.
Re-negotiation may happen in the middle of an existing session, a "session ticket" or "session Id" are not required to find "current state parameters", just 5 tuple may be used to look for proper running session context. ( is it true ?)
Moreover, resumption may happen even after actual connection has been closed (is it true? ) and we need sessionTicket or sessionId to look for security parameters and DO decryption, BUT the ClientHello message itself is encrypted. How is this solved for TLS1.2 ?
In the context of TLS1.3, renegotiation is done by using KeyUpdate message, and is clearly described by draft, but resumption has the same issues. How is resumption solved for TLS1.3 ?