Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 28 Feb 2012 19:48:00 -0300
From: Claudio André <claudioandre.br@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Re: OpenCL errors

Hi, below (and attached) more information.

The error only happens while testing with ./john --test or ./john --test=0.

It works fine if i test only one format.
- ./john -test -fo:raw-sha1-opencl or ./john -test=0 -fo:raw-sha1-opencl

JtR-TestSuite fails too, but i'm not sure what king of test was done by 
jtrts

Kernel: Linux ceudutra-HP-G42-Notebook-PC 3.0.0-16-generic #28-Ubuntu 
SMP Fri Jan 27 17:44:39 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux
GCC: gcc 4.6.1 (Ubuntu/Linaro 4.6.1-9ubuntu3)
Compile: make clean linux-x86-64-opencl
OpenCL: see clinfo.txt

Sources: i get the sources today and nothing was changed. Attached the 
md5 for *.c e *.cl files.
Catalyst: 11.11, driver: 8.911, more detais on the image.

Claudio


Em 28-02-2012 16:39, Samuele Giovanni Tonon escreveu:
> hello claudio,
> i don't think the problem is related to the poor
> or not GPU, but that's something troubling because i
> can't reproduce your bug.
>
> could you tell me which version of john you are using
> and could you do a md5sum of all the .c and cl kernel
> of the jtr you are using ?
>
> does the problem come out from time to time or it always
> happens?
> does it happens if you go with
> ./john -test -fo:raw-sha1-opencl
> or
> ./john -test=0 -fo:raw-sha1-opencl
> and if you try to crack some hashes instead of using
> -test ?
>
> - unfortunately sem_wait bug is not only related to amd+nvidia,
> as many reported but there's something more which i can't find,
> also distro are starting to put opencl as package and that is where most
> have problem at the moment, that is why i think 2.6 at the moment is to
> be avoided or used with precautions
>
> - i'm aware amd 2.6 is now AMDAPPSDKROOT, i'm keeping
> atistreamsdkroot because many have not yet upgraded to 2.6.
> i'm planning on adding some check on the makefile about that
> but also i'm delaying it due to the point above.
>
>
> Many thanks for your test
> Samuele
>

View attachment "clinfo.txt" of type "text/plain" (5993 bytes)

View attachment "md5" of type "text/plain" (8462 bytes)

View attachment "md5cl" of type "text/plain" (506 bytes)

Download attachment "catalyst_1.jpg" of type "image/jpeg" (37008 bytes)

Powered by blists - more mailing lists

Your e-mail address:

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