Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 13 Jan 2013 03:32:41 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: SIGALRM and new 'done' method

On 12 Jan, 2013, at 15:30 , Claudio André <claudioandre.br@...il.com> wrote:
> Em 12-01-2013 12:23, Claudio André escreveu:
>> 
>> The last time i tried it on bull, it failed.
>> 
> 
> Failed now on bull.
> 
> git pull
> make clean;  make *-opencl
> 
> ../run/john -t=0 -pla:1

This was a problem even in unstable: After a wad of patches tonight, dang near all formats honour LWS and GWS environment variables. That means we can now do this for a quick full self-test (with inaccurate speed figures) even for OpenCL and/or CUDA builds:

$ LWS=3 GWS=33 ./john --test=0

BTW I have used this with LWS=1 and GWS as a prime, as well as LWS as a prime and GWS=$((LWS*7)) trying to trigger array bugs, but none were found.

Oh, and the only formats not honouring these environment variables are Sayantan's DES, BF and MScash2 because I don't want to make a mess of them, but I'd really like this support to be added. It is invaluable for testing and for using weaker devices than the formats expect.

But I digress. This revealed that the problem you saw exist in unstable-jumbo too, so I did some more head-scratching and ended up realizing my last patch was the right way of handling it, but it did not support all kinds of systems. It does now, problem solved. No workarounds, just a correctly restarting signal handler.

magnum

Powered by blists - more mailing lists

Your e-mail address:

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