phc-discussions - Re: [PHC] hash encryption

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: <20160324122934.GA10265@openwall.com>
Date: Thu, 24 Mar 2016 15:29:34 +0300
From: Solar Designer <solar@...nwall.com>
To: discussions@...sword-hashing.net
Subject: Re: [PHC] hash encryption

On Thu, Mar 24, 2016 at 01:15:22PM +0100, Dmitry Chestnykh wrote:
> What about starting a round counter (which gets hashed) from a different, non-overlapping, value for different key sizes?
> 
> 128-bit key: round# = 0, 1, 2, 3
> 256-bit key: round# = 10, 11, 12, 13, 14, 15, 16

Sure, but for keys that are just 16 vs. 32 bytes the problem doesn't
exist because all of them fit in one SHA-256 block (along with the
cipher half-block, which is another 16 bytes).  Andy was referring to
unusually long k2's, needing at least two SHA-256 blocks.

I am considering hashing in round+keylen.  For one byte, this is almost
as easy and clean as hashing in just the round number.  But this becomes
potentially susceptible to Andy's attack for keylen >= 257.  To deal
with that, we'd be back to encoding the full length or full round+keylen
as a 64-bit number.

Another detail is that 6 least significant bits of keylen don't really
need to be encoded, but this detail doesn't appear to be of any use,
unless we were to encode (round + (keylen >> 6)) and declare keylen's
needing more than 14 bits unsupported.

Alexander