phc-discussions - Re: [PHC] Modified pseudo-random distribution in NoelKDF

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]

Message-ID: <20140121021923.GA30971@openwall.com>
Date: Tue, 21 Jan 2014 06:19:23 +0400
From: Solar Designer <solar@...nwall.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] Modified pseudo-random distribution in NoelKDF

On Mon, Jan 20, 2014 at 03:56:13PM -0500, Bill Cox wrote:
> On Mon, Jan 20, 2014 at 9:01 AM, Solar Designer <solar@...nwall.com> wrote:
> > Do you have specific numbers for the original approach above, and what
> > would be high enough (in your opinion)?
> 
> I would like to hurt a guy using only 1/4 of the memory enough that
> his attack is not practical.  I also want to not spend much time in
> the second loop forcing an attacker to show memory locations, so I'd
> like to read only 1% of the blocks.  A guy using only 1/8th should be
> deep into impractical TMTO territory.

How do you implement thread-level parallelism, or is this scheme with
the second loop at 1% only suitable for p=1?

> The average recalculation for 10,000,000 nodes covered by evenly
> spaced pebbles [...]

What if they are not evenly spaced?  You're making the distribution of
lookup indices highly non-uniform, so perhaps an attacker with limited
memory can adjust the spacing accordingly and achieve a lower TMTO
penalty?

Alexander