Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [P1619-2] the EME2 ref code vs. byte-oriented pseudocode

Hi Hal,

Using 0x80 vs. 0x01 is a Big-Endian vs. Little-Endian issue.  In a strict Little-Endian world, bit 0 of byte 0 comes first in the stream, which means that the little-endian byte stream 01 00 00 00, represents a binary stream of a 1 followed by zeros, which is identical to the big-endian byte stream of 80 00 00 00, which is also a binary 1 followed by zeros.

Here's an example:

Binary Stream       1  0  0  0  0  0  0  0 -  0  0  0  0  0  0  0  0
LittleEnd BitOrder 00 01 02 03 04 05 06 07 - 08 09 10 11 12 13 14 15
BigEndian BitOrder 15 14 13 12 11 10 09 08 - 07 06 05 04 03 02 01 00
LittleEnd Byte Rep [          01          ] [            00         ]
BigEndian Byte Rep [          80          ] [            00         ]

The confusion comes because in America we represent numbers in big endian format, where the most-significant digits come first.  That's why 01 00 00 00 looks strange, but makes perfect sense when converted to a little endian binary stream.

If the byte stream is in little-endian order, then using 0x01 00 00 as padding is the logical choice over 0x80 00 00 00.

Cheers,
-Matt

On Tue, Jun 17, 2008 at 4:08 PM, Hal Finney <hal.finney@gmail.com> wrote:
Well, either one would work. I did 0x80 because that is how padding is
done in some other algorithms, like SHA. It also leaves open the
possibility of a future extension that would allow for odd bit
lengths. If we had a 513 bit value then we could pad with 0x40 0 0
.... (This assumes that the 513th bit would be 0x80, but we do sort of
assume that anyway in our multByAlpha representation.) So I'd prefer
changing the byte oriented pseudo-code to do padding with 0x80 0 0....

Hal

On Tue, Jun 17, 2008 at 2:17 PM, Shai Halevi <shaih@alum.mit.edu> wrote:
> The current EME2 reference code on sourceforge uses padding with
>
>  0x80 0 0 ...
>
> whenever one needs to pad things to 16-byte boundaries. On the other
> hand, The byte-oriented pseudo-code says to pad with
>
>  1 0 0 ...
>
> So we need to change either the pseudo-code in the spec or the
> reference code.
>
> -- Shai
>



--
Thanks!
Matt Ball, IEEE P1619.x SISWG Chair
M.V. Ball Technical Consulting, Inc.
Phone: 303-469-2469, Cell: 303-717-2717
http://www.mvballtech.com
http://www.linkedin.com/in/matthewvball