Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 8 Nov 2011 20:59:06 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: algorithm info wrongly put in format_name

On Tue, Nov 08, 2011 at 10:18:07AM -0600, jfoug wrote:
> I believe what you are seeing is the 'BENCHMARK_FORMAT' value.

You mean BENCHMARK_COMMENT, and you're correct.  The Subject is slightly
wrong then.

In the main tree, my use for the benchmark_comment field (and the
BENCHMARK_COMMENT macros) was to convey info on the tests[] being used
for benchmarking, as opposed to info on the hash/cipher type (should be
in format_name) and on the algorithm (should be in algorithm_name).
In practice, the only places in the main tree where BENCHMARK_COMMENT is
non-empty are these two:

BF_fmt.c:#define BENCHMARK_COMMENT              " (x32)"
BSDI_fmt.c:#define BENCHMARK_COMMENT            " (x725)"

We could similarly use this when benchmarking phpass hashes - to tell
the user that the reported speeds pertain to "$P$9" hashes (with the '9'
rather than some other cost factor).  We could thus use " (x2048)" for
"$P$9" hashes.

> In dynamic_fmt.c, there is this code:
> 
> #ifdef MMX_TYPE
> #define BENCHMARK_COMMENT	MMX_TYPE
> #else
> #define BENCHMARK_COMMENT		""
> #endif
> 
> If we get rid of that code, then I believe that the output will be the way
> you want it.
> 
> Simply changing the above 5 lines to this, I think will do the trick.
> 
> #define BENCHMARK_COMMENT		""

Yes.  I think MMX_TYPE was also included in algorithm_name, so including
it in benchmark_comment was redundant.

> I am putting a patch on the wiki.

Thanks!  What about these? -

> >Only in file 2: Salted SHA(8x):Only one salt
> >Only in file 2: Salted SHA(8x):Many salts
> >Only in file 2: Raw SHA-1(8x):Raw
> >
> >You can try SSE2 and other builds as well.  I think/hope that addressing
> >the above will take care of them all (including those with "SSE2" in
> >place of "MMX"), but this is worth checking.
> >
> >Oh, while you're at it, maybe also remove the leading spaces for md5_gen
> >(or for dynamic now).

Alexander

Powered by blists - more mailing lists

Your e-mail address:

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