Re: Testing XTEA

Wolfgang Ehrhardt wrote:
Various libraries have xtea source in them. LibTomCrypt (libtom.org)
is one of them.
The OP should be aware that there different implementations of XTEA
regarding the endianness. LibTomCrypt uses little-endian code, in
contrast for example Botan (and Crypto++ IIRC).

So if there are rubbish test vector result, try them with the opposite
endianness.

My code conforms to various random snippets as found on wiretapped.net.

It should be noted that XTEA is not a standard nor are there standard
vectors to work from (they never published any in the 2 page paper).
So neither of us is "right" or "wrong." :-)

That said, in response to the OPs followup, consider using Skipjack.
It's at least a standard and MEANT for 8-bit MCUs.

Tom

.