# RSA signature attack check on lowest 16 bytes - Implementation

I am referring to solution given in this question

I need help to implement Poncho hint in this comment:

As the first question i have an RSA signature of 1024 bit applied on an hash (MD5) of the message.

I am trying to get a valid result coding it in Java but i don't get any result from my iterations.

Can someone help me to find out why i don't get results as mentioned by @Poncho?

Example of hash: 0xCC844F9AF903BCB3CB0A7F0433AF70D3 Decimal representation: 271849508914011436342484527734062018771

Thanks a lot.

if(checkValue.compareTo(resultOfModulus) == 0)


Here's your problem; you're checking if $y^3 = (md5BigInteger \bmod 2^i)$; that's wrong (as $md5BigInteger \bmod 2^i$ won't typically be a cube).

Instead, what you want is to check if $y^3 \equiv md5BigInteger \pmod{2^i}$; or, in other words (thinking less like a mathematician, and more like a programmer) if $(y^3 \bmod 2^i) = (md5BigInteger \bmod 2^i)$

Also, don't forget to check if $md5BigInteger$ is odd (or more generally, $8^k w$ for integer $k$ and odd $w$); if it's not of that form, no such simple cube will exist.

• Thanks a lot for your clarification. I updated my code and i am sure that md5BigInteger is odd but again i didn't get results. I start to work with y from 1 and then increment it at each iteration. Is that right? Can you check my code and tell me where i am wrong please? – Seed3Key Aug 3 '15 at 17:29
• That's wrong as well. If the $y^3$ and $hash$ agree in the lower $i$ bits, then you don't need to update $y$ at all (because you know that the lower $i$ bits are correct. If they don't agree, then you know you need to flip bit $i-1$ of $y$ (and since you initialize that bit to 0, that's the same as adding two,pow(i-1) to it. – poncho Aug 3 '15 at 17:52
• I misstyped my edit and now i think that the code is correct. Right? Again i don't get a valid result. I added for further experiments the sample hash i am working on. Thanks again for your help. – Seed3Key Aug 3 '15 at 18:08
• @Seed3Key: look at what I wrote; not y = y.add(two) but y = y.add(two.pow(i-1)) – poncho Aug 3 '15 at 18:32