Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 30 Oct 2013 12:27:34 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: descrypt-opencl "section 0" fix

On Tue, Oct 29, 2013 at 07:51:16PM +0100, magnum wrote:
> On 2013-10-29 19:20, magnum wrote:
> >I do not get the difference between the above vs. having a built-in test
> >vector using the same data? In both cases, we call set_salt("VT"),
> >set_key(0) and finally crypt_all(1).
> 
> OK, it's because during self-test, salt passed to crypt_all() is NULL so 
> we don't really test that functionality yet, right?

I think so, yes.  Although the bug was in host code, it was related to
on-GPU hash comparisons.  To figure it out, I had to read the OpenCL
kernel code, and in particular the cmp() function, and only then return
to host code to find and fix the bug.

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ