Inconsistent bits/bytes in symmetric specification

[usnistgov]

Inconsistent bits/bytes in symmetric specification

[celic]

It should be noted, in the SHA spec, there is no reference to bits or bytes for things like message length. The intended notion is bit length in terms of SHA, but I think this should be discussed. Perhaps we could have a default of bits unless otherwise specified? I'm not sure how much sense this would make in terms of other algorithms however.

[atvassilev]

If you look at SP 800-38C http://nvlpubs.nist.gov/nistpubs/Legacy/SP/nistspecialpublication800-38c.pdf, Appendix A, the specification of AES-CCM is defined in bits with the additional constraint that the bit-length of each input string shall be a multiple of 8. So, it seems that the byte-length inputs are something that CAVS came up with and it is not how the standard defines it.

When in doubt, we should check the standard and if a discrepancy is found, the definition in the standard prevails.

If we use the bit-oriented definitions in SP 800-38C and change the numbers for AES-CCM in the table above the inconsistency would go away and the sub-symmetric spec will be simpler.

[atvassilev]

Resolved with PR #133