phc-discussions - Re: [PHC] Makwa parameter generation

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Thu, 11 Sep 2014 18:10:03 +0100
From: Samuel Neves <sneves@....uc.pt>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] Makwa parameter generation

On 09/11/2014 02:54 PM, pornin@...et.org wrote:
> The core idea of RSA UFO is the following: if we want a modulus with
> two prime factors of at least k bits, then we generate several random
> integers of 3k bits, and multiply them all together. If we use enough
> such integers, then probability that at least one of them has two
> prime factors of at least k bits can be made arbitrarily close to 1.
> However, the resulting modulus is very large. Sander says that if
> you want 512-bit-or-more primes, and a probability of having at least
> two such factors in the final product of at least 1-2^(-80), then the
> modulus will have size "much greater than 40000 bits".
>
> To keep on with security levels as recommended in 2014, we would
> prefer prime factors of at least 1024 bits, so the modulus size will
> probably need to be more than 80000 bits, in fact probably much more
> than that (I did not do the maths to get the actual size, and I suspect
> that Sander did not either).

Sander's analysis is a little rough; I point to [1] to a more refined analysis, albeit in a slightly different context,
which achieves moduli of 20k bits for the 80-bit security level, and 60k bits for the 128-bit level.

These methods have in common that they make modulus generation very quick: it is just a matter of generating kl
pseudorandom bits, and multiplying them together. If we instead choose to trade generation speed by modulus size, it may
be possible to use trial division and/or ECM to 'tilt' the probabilities down by rejecting unsuitable integers, and get
smaller moduli in return. This requires more careful analysis.

[1] http://link.springer.com/chapter/10.1007%2F978-3-642-38980-1_35