Re: [Cfrg] OCB test vectors reusing nonces

[Matt Caswell <[e-mail]>]

On 29 January 2014 01:06, Manger, James <[e-mail]> wrote:
> One solution is to change the key for each tag length, giving 9 distinct keys for the 9 cases.
>
>    K = num2str(TAGLEN, KEYLEN)    // eg 00..0040 when TAGLEN == 64
>
> I then get:
>
>    AEAD_AES_128_OCB_TAGLEN128 Output: 67E944D23256C5E0B6C61FA22FDF1EA2
>    AEAD_AES_192_OCB_TAGLEN128 Output: F673F2C3E7174AAE7BAE986CA9F29E17
>    AEAD_AES_256_OCB_TAGLEN128 Output: D90EB8E9C977C88B79DD793D7FFA161C
>    AEAD_AES_128_OCB_TAGLEN96 Output : 77A3D8E73589158D25D01209
>    AEAD_AES_192_OCB_TAGLEN96 Output : 05D56EAD2752C86BE6932C5E
>    AEAD_AES_256_OCB_TAGLEN96 Output : 5458359AC23B0CBA9E6330DD
>    AEAD_AES_128_OCB_TAGLEN64 Output : 192C9B7BD90BA06A
>    AEAD_AES_192_OCB_TAGLEN64 Output : 0066BC6E0EF34E24
>    AEAD_AES_256_OCB_TAGLEN64 Output : 7D4EA5D445501CBE
>

I have successfully verified these.

>> Also - a more minor nit - your notation for incrementing the nonce
>> seems odd to me. The nonce is defined as a string of bytes - whereas
>> in your notation it is treated as an integer which can be incremented.
>>
>> Matt
>
> I think "N = N + 1" is more obvious than "N = num2str(3*i + 1, 96)", even if it does mix byte strings and arithmetic.

Agreed. One possibility is just to use the convention used elsewhere
in the draft and define a function for this purpose, e.g. "N =
incrementstr(N)".

Matt