Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20150607103457.GA4394@openwall.com>
Date: Sun, 7 Jun 2015 13:34:57 +0300
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: poor man's fuzzer

Hi Kai,

On Sun, Jun 07, 2015 at 05:44:28PM +0800, Kai Zhao wrote:
> > (ulimit -v 2097152; time ~/j/fuzz.pl &> fuzz.log)
> 
> I execute as you did, why the fuzz time are different?
> Some takes little time, some takes much time.
> 
> $ ulimit -v 2097152; time ./fuzz_solar.pl &> fuzz.log
> 
> real    0m7.820s

This is time until the original process terminates.  There are most
likely child processes still running in the background at this point.
I tried to minimize the discrepancy in the different processes' running
times with "| shuf |", but not to much avail.  I think a different
approach to (re-)distributing the work across child processes is needed.
This fuzz.pl is just a quick hack, with lots of room for improvement.

shuf is also why things differ between runs.

> This leads to different results. Some have fuzz-err.log while some did not.
> Some fuzz-err.log have serval hashes while some have few.
> Why the results are different ?

If you let all of the child processes run to completion, the results
should be the same - except they'd be in different order.

Alexander

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.