Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 15 Jul 2012 12:21:52 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: New core (?) LM fails alignment

On Sun, Jul 15, 2012 at 10:12:45AM +0200, magnum wrote:
> ...but when building a non-jumbo from latest CVS, the problem does not
> appear. But I am starting to believe there actually is a problem in
> core, that just happens not to show up. Was there a recent change that
> could have introduced this?

Oh, indeed.  I changed BINARY_SIZE from 32-bit to 64-bit, because we now
derive full LM hash halves (for writing to john.pot) from the binaries.
This is actually two 32-bit words, not one 64-bit word.  But indeed the
self-test does not know that, and tries to insist on a 64-bit alignment.

I think I will just drop the alignment check, and instead revise the
loader so that alignment of the pointer returned by binary() would not
be expected anymore.

Thank you for reporting this - very timely (just before I proceeded with
other changes, which would make producing clear per-change patches more
difficult).

Alexander

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.