Record Size Limit Extension for TLS
RFC 8449

Thanks to everyone who contributed work towards this document.

---------------------------------------------------------------------------

§4:

>  MUST NOT send a value higher than the protocol-defined maximum record
>  size unless explicitly allowed by such a future version or extension.

Presumably, recipients MUST gracefully accept values higher than the maximum
record size?  That is implied by this text (and the text that follows), but
given how TLS frequently aborts connections at the first sign of any
irregularity, it's probably worth saying explicitly.

---------------------------------------------------------------------------

§4:

>  a DTLS endpoint that
>  receives a record larger than its advertised limit MAY either
>  generate a fatal "record_overflow" alert or discard the record.

I'm concerned about the interaction between the option to discard the record and
protocols that perform retransmission of lost packets over DTLS (e.g., proposals
such as draft-rescorla-quic-over-dtls). In the case that an oversized packet is
simply discarded, retransmissions of that (presumably still oversized) packet
will take a while to time out (I'm not particularly well-versed in QUIC, but
assume it has characteristics similar to TCP's ~nine-minute timeout), which
would result in really bad user experiences.  Is there rationale for this optionality?
It would seem to be cleaner if the response were simply to always send a fatal
error.

If this optionality is retained, please at least consider adding text describing
the impact of discarding oversized packets when used with a reliable protocol.

---------------------------------------------------------------------------

§4.1:

>  In TLS 1.2, block ciphers allow between 1 and 256 octets of padding.

nit: The formulation "between X and Y" is ambiguous as to whether X and Y are
included in the range. Consider rephrasing as "...from 1 to 256...".

I apologize for not catching the point about "negotiated" earlier.


   An Authentication Encryption with Additional Data (AEAD) ciphers (see
   [RFC5116]) API requires that an entire record be present to decrypt
Nit: "An" .. "ciphers" do not agree.



   incremental processing of records minimally exposes endpoints to the
   risk of forged data.
This seems a bit weak. It's just not an acceptable practices to incrementally process.



   Unprotected messages - handshake messages in particular - are not
   subject to this limit.
This text is a bit confusing. Consider a case where a server recognizes the extension and has no limit, but the client offers one.

In that case, the server can either:

Return an extension with a max-size limit
Not return the extension
I think we want the server to conform to the client's limit in any case, but "negotiated" is unclear.


   they have no need to limit the size of records.  This allows servers
   to apply a limit at their discretion.  If this extension is not
   negotiated, endpoints can send records of any size permitted by the
I think by "apply" you mean "request" or "advertise"