Take the 2-minute tour ×
Cryptography Stack Exchange is a question and answer site for software developers, mathematicians and others interested in cryptography. It's 100% free, no registration required.

We know that exposing $p$, or $q$ or $\phi(n)$ results in trivial attacks on RSA since they allow us to factor $n$ and to compute the private exponent $d$.

In OpenSSL (and most RSA implementations) decryption is done via the CRT, so we store $n$ and $d$, but also store $d \bmod (p-1)$ and $d \bmod (q-1)$ as CRT exponents. Suppose these leaked... is there a practical attack?

share|improve this question
    
What is 'd' that you have mentioned? –  Ashwin May 14 '12 at 9:37
    
The $d$ is the RSA private exponent. Have a look at any exposition on RSA encryption for more details. –  Fixee May 15 '12 at 2:34
add comment

1 Answer

up vote 10 down vote accepted

Yes, there is a practical attack. Leaking those (or even just one of those) allows us to factor the modulus quite efficiently.

Suppose the attacker knows the values $n$, $e$ (the public exponent) and the value of $d \bmod (p-1)$ (which we will call $dp$).

Then, the attacker selects a value $m$, and then computes:

$gcd( n, m ^ {e \cdot dp-1} - 1 \bmod n)$

To see why this is likely to give us a nontrivial factor of $n$, let us consider the value of $m ^ {e \cdot dp-1} \bmod p$ (which we can do, even if we don't know the value of p). We see that this is:

$m ^ {e \cdot dp-1} \bmod p = 1$

This is because $e \cdot dp = 1 \mod (p-1)$ (and Fermat's Little Theorem).

In contrast, this same value modulo q is (with high probability, assuming $m$ was selected randomly) not 1, and so $m ^ {e \cdot dp-1} - 1 \bmod n$ will have $p$ as a factor, but not $q$. And thus,

$gcd( n, m ^ {e \cdot dp-1} - 1 \bmod n) = p$

with high probability.

share|improve this answer
    
Can you explain how you arrived at this (nice) solution? What was the motivation? –  Fixee Feb 22 '12 at 16:54
1  
@Fixee: well, the standard way to factor n given the values d and e is to consider the value $m^{e \cdot d - 1} \bmod n$. Now, given the full d, that's 1 (and so for that attack given the full d, we actually look at $m^{(e \cdot d - 1) / k}$ where k are various powers of 2). However, given a partial d ($dp$), we can still consider how $m ^ {e \cdot dp - 1}$ behaves; once we consider that, the attack becomes obvious. –  poncho Feb 22 '12 at 17:44
    
Thanks. I had never seen this attack on $n$ when we know $e$ and $d$. –  Fixee Feb 22 '12 at 23:05
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.