john-users - md5_gen(0) broken for ages?

[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Fri, 25 Feb 2011 01:30:35 +0100
From: magnum <rawsmooth@...dband.net>
To: john-users@...ts.openwall.com
Subject: md5_gen(0) broken for ages?

After applying the intrinsics fixes, raw-md5 is broken for me. It claims 
there are a number of salts (for tens of thousands of hashes, it claimed 
some 190 salts, for 10 hashes, it claimed 9 salts... maybe something 
uninitialized?). And, probably as a consequence of this, it's AWFULLY 
slow (while the benchmark show almost a 2x speedup).

It does crack stuff correctly though.

I tried all sorts of things, like adding the tag "md5_gen(0)" to the 
hashes in the files they resided. I also tried adding a $ (or $: or ::) 
in the end of the hash or using a file with nothing but the hash (no 
username, no colons, just the hashes). Nothing make any difference.

Then I started reverting. This is not new at all. The oldest JtR I have 
on disk (1.7.6-jumbo7) have this bug. It's just that until now, we have 
been using the separate raw-md5 format so no-one has noticed md5gen(0) 
is broken. Or is it just me?

This is on AMD x64 as well as Intel x64. Compiling with clang made no 
difference.

magnum