phc-discussions - Re: [PHC] RE: Password hashing as a self-overwriting Turing machine

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] [day] [month] [year] [list]

Message-ID: <CAOLP8p5bqBBn6JuAChunp2RjQ8+OOjZ3MHM0ZVOgPCDcrJtxLA@mail.gmail.com>
Date: Thu, 2 Jul 2015 08:41:32 -0700
From: Bill Cox <waywardgeek@...il.com>
To: "discussions@...sword-hashing.net" <discussions@...sword-hashing.net>
Subject: Re: [PHC] RE: Password hashing as a self-overwriting Turing machine

On Wed, Jul 1, 2015 at 1:15 PM, Marsh Ray <maray@...rosoft.com> wrote:

>  Denis Bider:
>
> > I consider formal proof a nice thing to have, but less than crucial.
>
> > We don't have formal proof for ECC, DH, or RSA, either.
>
> What is the basis for your claim that a randomized sequence of operations
> from the set you have chosen results in a irreducible function?
>

He uses ARX (add, rotate, xor) instructions randomly generated by a
pseudo-random-oracle (SHA-512).  This has been proven Turing complete.


>
>
> I'm not suggesting you can't come up with a good answer for this. I'm just
> saying I think this is the place to start with further analysis of your
> function.
>

Agreed.  However, it looks like he got the algorithm almost-right, just
based on intuition.


> > Focusing solely on formal proof seems to me like the case of
>
> > building a fence around a house where most of the planks are
>
> > six feet, but one of the planks is sky-high.
>
>
>
> Don't worry, we're all very practically minded here.
>

I wish that were true!  The number of entries that did not consider speed
important in a memory-hard hash function were surprisingly high!  Basic
memory*time defense typically goes as the square of memory-filling speed.


> > If you don't have CPU AES support, then the way I see it, if you rely
>
> > solely on AES, you're spending proportionally more time operating
>
> > on small blocks of data, instead of exercising what the CPU does
>
> > well, which is fetching stuff from memory.
>
>
>
> I think you would enjoy reading through the list archives. Topics such as
> this are discussed in considerable depth.
>

I think he would also have enjoyed helping create those archives... The
timing of his entry is simply wrong.

Bil

Content of type "text/html" skipped