phc-discussions - Re: [PHC] Re: Attack on Argon2i?

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: Tue, 12 Jan 2016 10:28:25 -0800
From: Bill Cox <waywardgeek@...il.com>
To: "discussions@...sword-hashing.net" <discussions@...sword-hashing.net>
Subject: Re: [PHC] Re: Attack on Argon2i?

On Tue, Jan 12, 2016 at 9:44 AM, Krisztián Pintér <pinterkr@...il.com>
wrote:

>
> so i'm long since against pseudorandom but fix access patterns. first,
> i don't see the benefit of a pseudorandom but fixed over a simple and
> fixed.

I agree, at least for cache-timing-resistant algorithms.

I just read the paper.  From a theoretical point of view, it is an
interesting.  I found the summary misleading, which says there is a 5X TMTO
against Argon2i without mentioning that to do this, you have to run Argon2i
in a non-recommended 1-pass mode.  They go on to say this:

"The Balloon hash functions are easy to implement and are fast enough to
support hundreds of authentications per second while using
using all of the high-speed (L2) memory on a modern CPU core."

As I've said before, time*memory defense is often proportional to the
square of the memory filling speed, and it is true in this case.  If this
algorithm takes 2ms to fill 256KiB of L2 cache, then it is so slow that it
should be ignored for any real-world password hashing.  We have algorithms
that run more than 32X faster, for a 1024X-ish improvement in time*memory
defense.

While using password-independent addressing, I think they made a mistake in
using the salt.  Salt should not be public.  Also, they lose defense
against important meta-data when using salt, in that an attacker with
cache-timing can determine when a user logs in.

Bill

Content of type "text/html" skipped