Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 08 Jun 2012 21:07:17 +0200
From: magnum <>
Subject: Re: JtR to process the LinkedIn hash dump

On 06/08/2012 05:45 PM, jfoug wrote:
> I believe this version (patches into magnum-jumbo), fixes the problem.

You mean bleeding-jumbo. It won't work with magnum-jumbo.

While I was offline I played with this too and it's amazing how we ended 
up almost exactly the same.

Instead of adding that key_count, I just used max_keys_per_crypt from 
the format struct. That should be faster ;-)

The other difference is that for some reason I totally overlooked that 
we had the missing bits already calculated in crypt_key[] so I actually 
re-created the whole hash using OpenSSL. Sometimes I'm a moron :)

Lastly, I check all available bits in cmp_one, trying to minimize false 
positives due to partial hash collisions. But that might be totally 

-//	if( ((ARCH_WORD_32*)binary)[0] != crypt_key[x+y*MMX_COEF*5] )
-//		return 0;
+	if( (((ARCH_WORD_32*)binary)[0] & 0x00000fff) != 
(crypt_key[x+y*MMX_COEF*5] & 0x00000fff) )
+		return 0;

The above is fixed endianness. For the non-MMX code I had similar things 
with #if ARCH_LITTLE_ENDIAN .. #else ...


Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.