Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 10 Aug 2013 02:46:49 +0200
From: Dániel Bali <balijanosdaniel@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Daniel's weekly report #8

Hello!


2013/8/9 Dániel Bali <balijanosdaniel@...il.com>
>
>
> 2013/8/9 Lukas Odzioba <lukas.odzioba@...il.com>
>
>>
>> > - I experimented with the kernel initialization process:
>> >     - Tried to remove various lines from the beginning to see if I get
>> error
>> >     - Unfortunately I got an error almost all the time
>>
>> What kind of error is it?
>> What enqueNDRange returns?
>>
>
> I should add a few lines of code that looks at the return value, this is a
> good idea. I will report back when I have that (tomorrow).
>

I checked and the only possible error type that is related is
"CL_INVALID_PROGRAM_EXECUTABLE", so whenever the program is invalid we
would get that error. The rest of the error codes are about invalid kernel,
context, dimensions, etc.

Today I modified the program that creates the kernel so that it passes the
appropriate number/type of arguments for the blowfish kernel.
Then when I tried to copy the assembly generated fromt he kernel, I
realized that it might be wrong.

Could somebody please take a look at this ISA? (I attached it to the mail)
All it does is loading dwords, writing something to the LDS and do a very
little computation. Surely blowfish is more complicated than this. Might
the ISA be wrong? All the ds_write calls and no reads is a bad sign.

The ISA of DES (which also uses DS instructions) looks much better, maybe I
should try to use that instead.

Regards,
Daniel

Content of type "text/html" skipped

Download attachment "bf_kernel.isa" of type "application/octet-stream" (67772 bytes)

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.