phc-discussions - Re: [PHC] attacking and hardening escrypt

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: Mon, 10 Mar 2014 01:24:08 -0700
From: Larry Bugbee <bugbee@....com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] attacking and hardening escrypt

On Mar 10, 2014, at 12:29 AM, Solar Designer <solar@...nwall.com> wrote:

> Also, I've been seriously considering using 32-bit lanes.  (This has
> some pros and cons.  Among the pros is better compatibility with Salsa20,
> where it'd let us ignore SIMD shuffling of 32-bit words.  escrypt
> currently has some extra complexity because of this shuffling, yet
> having its new sub-block mixing work on 64-bit lanes.)  I think that
> with careful design and with use of the variable S-boxes, 32-bit lanes
> would be OK in terms of issues described above, but they'd provide a
> smaller safety margin.  (Luckily, we're not talking cryptographic
> security here, but just attacks that would allow for computation of the
> hash with somewhat less resources than intended.)

A naïve question perhaps, but would 64-bit lanes incur an [unnecessary?] performance penalty for defender's implementations on 32-bit processors like ARM?  ...or could/should this be an adjustable parameter the site manager/sysadmin could set?