On Feb 5, 2021, at 10:42 AM, John Mattsson 
<[email protected]> wrote:
> The resumption_master_secret includes the client finished so the client in 
> your handshake with client authentication should not be able to reconnect, if 
> it can it is an OpenSSL bug. Alternatively the server did not ask for client 
> authentication and it is ok that the client reconnects.

  At a high level, if there are two session tickets, then the draft should 
explain *why* one of them is invalid.

>> The packet flows in Figure 2 of draft-14 shows only one exchange of session 
>> tickets, not 2.
> 
> Looks to me that the Figure 2 of draft-14 provisions two tickets...?
> 
>                                                         EAP-Request/
>                                                    EAP-Type=EAP-TLS
>                                               (TLS NewSessionTicket,
>                                                TLS NewSessionTicket,
>                                 <--------          TLS close_notify)
> 

  My reading of the TLS specs && OpenSSL indicates that there are not only two 
tickets, but the tickets are in different messages.

  i.e. the first ticket is in the same packet when the server sends "TLS 
Finished".

  After the server receives the client certs, it goes "whoops", and issues a 
*new* session ticket in the next packet.

  So no packet should have *two* session tickets.

  Alan DeKok.

_______________________________________________
Emu mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/emu

Reply via email to