Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 14 Aug 2013 16:57:42 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Katja's weekly report #9

Katja, Yaniv -

On Wed, Aug 14, 2013 at 12:49:08PM +0200, Katja Malvoni wrote:
> On Wed, Aug 14, 2013 at 12:10 AM, Frank Dittrich <frank_dittrich@...mail.com
> > wrote:
> 
> > Are you sure it's not a hardware error?
> 
> I don't know... So far it passed all tests with password.lst and
> pw-fake-unix (1 hour and 20 minutes) but sometimes it doesn't crack all
> passwords when I generate 3000 random strings and 1000 hashes (tests take
> around 20 minutes)

Maybe you can test on the other Parallella prototype board?

> I tried running official matrix multiplication test but it doesn't work -
> host stalls in polling loop when checking done signal from cores.

It worked with the previous eSDK.  Now the old eSDK's matmul no longer
works (the prebuilt one misses libraries, and rebuilding this old source
code with the new eSDK fails), and the new eSDK's fails as you describe.

I _guess_ the new eSDK's matmul example uses "the new E_DMA_MSGMODE
control mode on E_DMA_DWORD transfers", which requires an updated
BOOT.BIN as Yaniv described here:

http://forums.parallella.org/viewtopic.php?f=13&t=422

I may try that.

I wonder if our inconsistent eSDK and BOOT.BIN versions may be causing
any trouble for the bcrypt code as well.  Probably not.

Alexander

Powered by blists - more mailing lists

Your e-mail address:

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