IETF-SSH archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

RE: applying AES-GCM to secure shell: proposed "tweak"



--On Thursday, April 09, 2009 08:36:24 AM -0400 "Igoe, Kevin M." <kmigoe%nsa.gov@localhost> wrote:

Damien Miller writes:

5.1/5.2 - The suggested algorithm names are structurally different to
the other algorithm identifiers used in SSH. In particular, I don't see
any need for the "-ssh" to appear in the algorithm name as the cipher
has not been modified in any substantial way. Why not just "aes128-gcm"
or "aes128-gcm-aead" if you wanted to be particularly verbose?

AEAD has its own convention for naming algorithms.  The names selected
are more in keeping with the AEAD conventions.  I'm not terribly fond of
theses names and would be willing to modify them as needed, but I'd like
to keep an "ssh" tag in the name so that when one looks at the namespace
of all AEAD algorithms, it is clear that these algorithms are intended
for use in secsh.

I don't understand. There is nothing "ssh" about the algorithm; it's just AES128 in GCM mode, no? Any "how to use this in SSH bits" are part of the SSH spec but not part of the algorithm. I'm not sure what value there is in a "namespace of all AEAD algorithms"; is there some registry you are concerned about? Why would an ssh encryption algorithm name ever appear in a context other than ssh algorithm negotiation? Why would you expect the names of, say, SSH encryption algorithms, TLS cipher suites, and Kerberos enctypes not to overlap?

In the context of the namespace in which these values are actually defined, which is that of SSH encryption algorithm names, the "-ssh" adds no value.

-- Jeff



Home | Main Index | Thread Index | Old Index