[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 22 Sep 2013 23:38:01 +0100
From: Dániel Bali <balijanosdaniel@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Daniel's reports #13, #14
Hello!
2013/9/21 Solar Designer <solar@...nwall.com>
> Here's a simple task for you: please make a numbered release of gcnasm,
> in a tarball like gcnasm-0.1.tar.gz that you'll upload to our wiki. It
> should include some brief documentation inside it, as well as links to
> the corresponding wiki page and to your GitHub repository.
I did this and put a download link here:
http://openwall.info/wiki/john/development/GCN-ISA
And the readme has a link to the GitHub repository
> > 2013/9/19 D?niel Bali <balijanosdaniel@...il.com>
> > > I'm sorry for not writing status reports lately. I now realize it might
> > > have looked like I didn't care about my project. I was really ashamed
> about
> > > not making any progress/not having sufficient time.
>
> Understood. I hope Lukas explained to you that you should have been
> posting the weekly reports anyway, even if you had no progress to report.
Yes, he explained it and it's my fault that I didn't send them every week.
> > > I will be trying to get the CRC32 kernel working with gcnasm.
>
> Yes, please - but please make the gcnasm release first. In general,
> both of these tasks - gcnasm release and a working CRC-32 JtR format
> written using gcnasm - are intended to let someone else conveniently
> continue with the project, whether along with you or on their own.
> As such, your completion of these tasks adds much value to the code
> you've written during the summer.
>
I'm still working on the CRC32, hopefully I can finish it by tomorrow.
>
> > > Here is the current version:
> > > https://gist.github.com/balidani/079634dae9b993fa6f8b
> > > With this test kernel:
> > > https://gist.github.com/balidani/13f8520b0bcbfdf2584b
> > >
> > > CRC32 will only have 8 as ciphertext length, but even with 16 it
> fails. I
> > > get the error:
> > > "FAILED (get_hash[4](0))" - but when I print what the corresponding
> hash
> > > is it's 0xaaaaaaaa as expected.
> > >
> > > I'll try to debug this and get it working with 8 as the ciphertext
> length.
>
> After a couple of minutes of staring at your code^Whack, I think I see
> the bug: you set BINARY_SIZE to 2, whereas it should be 4 (or even 8, if
> you're for some totally weird reason experimenting with 64-bit hash
> values for now, even though CRC-32 will only use 32-bit).
>
BINARY_SIZE started out as 4, when I was testing and didn't modify
everything from the keccak format.
But I believe in the end the BINARY_SIZE should be 1, because a CRC32 is
32-bit.
For example raw-md5-opencl has:
#define CIPHERTEXT_LENGTH 32
#define DIGEST_SIZE 16
#define BINARY_SIZE 4
Regards,
Daniel
Content of type "text/html" skipped
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.
