Re: [TLS] Working Group Last Call for draft-ietf-tls-downgrade-scsv-00

[Bodo Moeller <[e-mail]>]

Martin Thomson <[e-mail]>:

The exception in Section 4 implies that there is some interoperability
> reason to avoid including the SCSV in a resumption request.  I think
> that it would be better if the actual interoperability problem.


I think it would be better if the verb :-)


> That
> is, your original session was made to a server instance that only had
> a lower version and is being resumed on an instance that has a higher
> version AND the node with the higher version doesn't treat a resumed
> handshake differently with respect to the handling of the SCSV.
> Alternatively, the server could have been updated since the original
> session was established.
>
> After some consideration, I think that the only issue with resumption
> is that it is somewhat wasteful.  Requiring servers to ignore the SCSV
> on resumption would work equally well.
>

What specifically do you consider wasteful?

Note that ignoring the SCSV *on resumption* wouldn't quite work. The server
would need to ignore the SCSV on resumption *attempts*, successful or not;
I don't think I'd like that.

I have a few minor editorial nits:
>   - the lists in Sections 3 & 4 have a single item, which is a little odd
>   - Section 4 should just say when to use the SCSV, rather than
> announce "This section specifies when to send it."
>

ACK.

Bodo