phc-discussions - Re: [PHC] MCS_PHS memory cost parameter

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: <CAGiyFdfVSZuzX3esF5q7MQkyiGOcwT+=hJHVRk43e8doirVa0Q@mail.gmail.com>
Date: Tue, 27 May 2014 14:22:36 +0200
From: Jean-Philippe Aumasson <jeanphilippe.aumasson@...il.com>
To: discussions@...sword-hashing.net, 
	Михаил Масленников <mikhailmaslennikov@...dex.ru>
Subject: Re: [PHC] MCS_PHS memory cost parameter

on
https://github.com/bsdphk/PHC/blob/master/MCS_PHS/code/mcs_psw/mcs_psw.cpp#L57
an array tmp[] of m_cost bytes is filled with predictable values, thus
it doesn't need to be stored.

if we look at the specs, on p3 and p4:
* steps 3 initializes a large array T_0, then
* steps 4 computes an iterative hash of T_0 to create a short T_1,
then hashes T_1, etc.

as I understand from the specs of MCS-SHA8 in
http://crypto.systema.ru/mcssha/MCSSHA-8%20(eng).pdf, the function
iterates over blocks of the messages, and the complete message doesn't
have to be kept in memory to be efficiently hashes.

On Tue, May 27, 2014 at 1:39 PM, Mikhail Maslennikov
<mikhailmaslennikov@...dex.ru> wrote:
>
>
> 27.05.2014, 13:32, "Jean-Philippe Aumasson"
> <jeanphilippe.aumasson@...il.com>:
>
> AFAIU from the MCS_PHS specs, the memory cost parameter dCostLen is
> ineffective:
> https://password-hashing.net/submissions/specs/MCS_PHS-v0.pdf
>
>
>
> --
> Dear Jean and all other members of PHC.
>
> I prepare my answer in attachment. I can't understand what's problem. I can
> make some modifications of MCS_PHS, but, please, explain me, what is your
> requirements or simply wishes.
>
>
> Best regards.
>
> Mikhail Maslennikov.