Re: [TLS] Still more on Record Layers

["Brian Smith" <[e-mail]>]

JCA wrote:
> If you have encryption you necessarily have MAC,

Or an authentication tag, for AEAD cipher suites

> but you can have MAC without encryption. What does the Record
> Layer length field contain in the latter case? Is it the length of the
> (unencrypted) encapsulated protocol messages alone, or the sum
> of this plus the length of the MAC?

It is treated exactly the same as the stream cipher (e.g. RC4) case, and
that is why it is described in the same section. Think of NULL encryption as
a stream cipher that always produces zeros.

I think many of your questions can be answered by examining the code of one
of the many existing open-source implementations such as GnuTLS or OpenSSL.
(The code of NSS's libssl is perhaps less clear in how it deals with the
record layer as it is kind of obfuscated with optimizations, and also
contains some subtle bugs in how it checks the length of the record against
the various limits that the specification imposes.)

Regards,
Brian