Re: [TLS] Mail regarding draft-ietf-tls-record-limit

[Martin Thomson <[e-mail]>]

(+tls@)

This is a good question Jim and one that I thought through during
implementation, but failed to capture in the doc.

Basically, there is no way to validate the extension if the client
includes an unknown version of TLS or an extension that it doesn't
understand.  A client can know because it should understand the
protocol as negotiated.

The text currently says "an endpoint MUST NOT send a value higher than
the protocol-defined maximum record size unless explicitly allowed by
such a future version or extension"  I think that we should add "A
server MUST NOT enforce this restriction; a client might advertise a
higher limit that is enabled by an extension or version the server
does not understand."

Does that make sense?


On Mon, Feb 19, 2018 at 5:51 PM, Jim Schaad <[e-mail]> wrote:
> I was looking at this document relative to a specific question for Kathleen,
> and I had one thing that I would like you to look at and see if you think it
> is clear enough.
>
> I have a server that is TLS 1.2, a client that is TLS 1.2 & 1.3.   It sends
> a hello w/ and extension value of 2^14+1.  It is not completely clear to me
> that the server should accept this as a legal value and compute the min of
> it and the maximum 1.2 value as the value to use when sending messages to
> the client rather than producing an error message because the value is too
> large.
>
> Jim
>
>