Help Needed with Cryptanalysis

[ashish.s...@gmail.com]

Hi,

I need some help with the cryptanalysis of "FastFlex", a new
cryptographic function I have designed before I submit it to AsiaCrypt.

In house cryptanalysis has raised no concerns as of now. The design is
pretty resistant to LC and DC. Its the other attacks which are of
concern (Which I have little experience about!).

Would be helpful if anyone could take a look at it.

Paper : http://fastflex.sourceforge.net/files/FastFlex-SAM.pdf
Website: http://fastflex.sourceforge.net

Also, I'm looking for sponsors and mentors for the project. Any
links/references would be helpful.

Thankyou for your time.

Ashish Sharma
Army Institute of Technology, Pune, India

[con...@matmail.com]

First, your C code has errors: you need to initialise your W array.

I presume that you want it to be all 0's. If that is the case, then
here is
a collision:

First input:
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0
(i.e. sixteen 32-bit words that are all 0)

Second input:
2, 2, 2, 2, 2, 2, 2, 2,
0xae73151f, 0xc904c746, 0x614d0189, 0x62383163,
0x609106b7, 0x40499dd0, 0x3188b2ba, 0xf0ae72cb
(i.e. sixteen 32-bit words, the first 8 are all 0x2, the others are
full 32-bit words)

The hash value in both cases is:
69997dfa 73847e9e 19248276 c43903c2 95772d2d 4eaf85eb ad90eb0e 6ce0ea42

(in hexadecimal)

Scott

[Dave Turner]

heh how did you manage to find a collision in his algo so quickly !?
amusing and impressive!

[ashish.s...@gmail.com]

Ok. I'll fix the code and check the check the collission!
Thanks again...And great work!

[ashish.s...@gmail.com]

Well, Yeah dunno how I missed something like this. Anyways, this is an
example constructon and I'll fix it straight away. FastFlex is still
safe. Will take off FastHash from the page straight away and come up
with something better. Thx.
Ashish Sharma