Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 10 Apr 2013 21:18:19 +0200
From: magnum <>
Subject: Re: default_cmp_all

On 10 Apr, 2013, at 20:55 , "jfoug" <> wrote:

> From: magnum [] 
>> So the new-style bitmaps fully replace cmp_all()? I thought they only
> replace it under certain conditions. I >suggest we do absolutely nothing
> until Solar comments this.
> I agree with the wait and see.  However, cmp_all is not called in the
> current unstable either (prior to the bitmaps).  Look at the code in
> cracker.c
> There is an outer if statement, on whether the format is salted or not. If
> not salted, then the else route is taken, and within that block of code,
> cmp_all is never called, and only the bin_hashes and cmp_one/cmp_exact is
> used.  I am not fully sold that is the right way, BUT that is how JtR has
> been running forever (for a long time at least).
> Like I said, it was news to me, and surprised me some.  I thought the code
> always something similar to:
> crypt_all();
> if (hash() && cmp_all()) {
>   foreach: if (cmp_one() && cmp_exact())
>     output found pw
> }
> But it is not that way for non-salted.

Are you saying that if I attack 12M MD5 hashes in unstable, we do 1M calls to cmp_one() for each and every crypt_all()? That simply can not be true. I'll have a look!


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.