phc-discussions - Re: [PHC] omegacrypt and timing

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: Wed, 17 Sep 2014 11:36:57 -0500 (CDT)
From: Steve Thomas <steve@...tu.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] omegacrypt and timing

> On September 17, 2014 at 4:30 AM Bill Cox <waywardgeek@...hershed.org> wrote:
>
> However, AntCrypt, OmegaCrypt, and Schvrch all tried to introduce data
> based branching for GPU defense, and "cyclomatic complexity". I am
> not sure we've seen this idea implemented well yet, and data-based
> branching has been considered a no-no for years apparently. However,
> when 3 authors all invent the same thing, I think we should take a
> closer look.
>

Adobe did this for encrypting PDFs
(http://esec-lab.sogeti.com/post/The-undocumented-password-validation-algorithm-of-Adobe-Reader-X).
This is why we see these submissions.

P.S. Schvrch is not in this category. As it is not an actual branch. In x86 SIMD
you do a compare then xor (compare sets it to all 1s or all 0s). With GPUs, they
would set a conditional variable (compare) then conditionally run bitwise invert
(one instruction). Note that a branch in x86 takes two instructions compare and
conditional jump. Both SIMD and GPUs use two instructions with neither being a
jump.