Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Tue, 8 Sep 2009 23:13:52 -0500
From: "JimF" <>
To: <>
Subject: Re: Re: JTR I have compiled...

>> john -test -form=md5-gen "-subform=md5_gen(4)"
> With quote " " I get :
> iMac-de-xxx-xxx:run xxxxxx$ ./john -test -form=md5-gen
> "-subform=md5_gen(4)"
> Unknown option: "-subform=md5_gen(4)"

It appears you will need to use quotes on *nix shells (some at least).  The 
'unknown option' is what you would expect.  It means john is trying to 
figure out what you are telling it to do, but the version of code you built 
this from, did not yet understand what the -subformat= option is.

However, I have just updated the wiki patch page:   I have updated the -2  (v2) of the 
md5-gen code.  This update WILL make the -subformat=list work (or let you 
specify -subformat=md5_gen(4)  to force the vBulletin mode)


works with benchmarking (-test)
Added the -subformat=NAME to allow one to force a specific sub format to be 
Added -subformat=LIST to dump a list of the formats to screen (and exit).
Fixed some bugs (like non salt formats were using salts, and working slow).
Performance gain.  md5_gen(0)  (raw-md5) is as fast as -raw-md5   (in SSE2 
Added phpass to generic md5
Added 'hooks' to allow formats which are already part of gen-md5 to use 
gen-md5 as the working engine.
Ported phpass to use this (gutting all of the 'native' code).  This is NOT 
included with the base gen-md5, as it will be a patch standing alone.

Note, right now, the non-SSE is a little slower than it CAN be (for mostly 
for the raw-md5 format).  That is due to the code pretty much having to do 
full md5 work, where the raw-md5 had code which did not complete, UNLESS 
there was a very likely hit.  The non-SSE code in raw-md5 only computes 1 of 
the 4 32 bit words of each hash, UNLESS that 1 number happens to match the 
first 4 bytes of a hash.  Then, and only then, does a full MD5 get done for 
that word.  Thus, for some formats, it may be faster to keep the native code 
(like the raw-md5 non-sse). However, most should be as fast (and some 
possibly even slightly faster), with the generic code.  Also, if some 
formats have not been optimized with faster code, then it is likely the 
generic code will be faster than what it is replacing.

NOTE there still are some collision issues between this patch and the 
'multi-option-addition' patch, in the options.h (and possibly in options.h)

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.