[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 05 Apr 2015 01:36:05 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: [GSoC] John the Ripper support for PHC finalists
On 2015-04-05 00:01, Frank Dittrich wrote:
> Whether it is intended behavior or a bug to filter self tests according
> to --cost parameters during self test, but not in a real cracking run,
> I don't know.
> I think magnum has to decide it (and may be improve the documentation).
We often say "self-test" and mean "benchmark", but they are two separate
events not even put in the same source file. "--test" means "self-test,
then benchmark" while a real crack run is "self-test, then crack".
The benchmark may filter test-vectors according to the --cost options,
but the self-test ignores it. This is intentional, although we could
re-think that if a format like Pomelo would benfit from it.
The self-test ideally takes no time but finds all problems. The
benchmark ideally takes no time yet shows a speed figure on par with the
speed of a real run.
We do something like this for --test
- self-test (formats.c)
-- fmt_init()
-- fmt_reset(NULL)
-- tests
- speed test (bench.c)
In this case, if we want to use small LWS/GWS for a quick self-test, we
also get benchmark figures not reflecting speed of a real-life run.
Maybe we want another reset() between tests and benchmark? Or maybe we
require formats to set up inital buffers in init() that are suitable for
self-test, and move the existing reset() to between test and benchmark?
For a real cracking run, it's currently like this:
- load db (loader.c)
- self-test
-- fmt_init()
-- fmt_reset(NULL)
-- tests
- fmt_reset(db) (from john.c)
- cracking mode starts
In this case, with or without moving reset(NULL) we may use very small
LWS/GWS figures for the self-test and do the auto-tune, with actual
costs and possibly including a re-compile of kernel, only before
cracking mode starts.
magnum
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.
