Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Fri, 22 Jun 2012 21:38:48 +0200
From: Frank Dittrich <frank_dittrich@...mail.com>
To: john-dev@...ts.openwall.com
Subject: relbench: map old and new format names

Hi all,

the attached patch fixes mapping of old (john-1.7.9-jumbo-5 or official
john) format names to those used in latest git.
This allows to compare benchmark results of different john versions,
even if the format names have been changed.

Side effect: the names reported to exist only in one of the files will
also be converted to the newest format name, even if you compare two
files with benchmark results of older john formats.

You'll find  the mappings used at the end of the relbench script:
old	new

A few john-1.7.9-jumbo-5 format names have not yet been mapped, for
various reasons:

Only in file 1: MYSQL:Raw

MySQL_fast has been renamed to MySQL, the other implementation has been
moved to unused.
That's why, MYSQL_fast (the jumbo-5 name) got mapped to MySQL


Only in file 1: NT v2:Raw

NT should be renamed to NT (label nt) or NT (label nt2), according to
Alexander.

When this is done in latest git, I'll map NT to NT (label nt) and NT v2
to NT (label nt2).

Only in file 1: ssh:Raw

I didn't map the jumbo-5 ssh to current  SSH RSA/DSA, because the test
vector has changed: (one 2048-bit RSA and one 1024-bit DSA key)

Only in file 1: HMAC MD5:Raw

jumbo-5 used just raw, current git distinguishes between only one salt
and many salts.

Hopefully all other mappings are correct, but I'll do a review.
I just wanted to give you the chance to point out any additional
problems I didn't notice (may be more format names that shouldn't be
mapped because the test vectors changed to much, format names I missed,
and so on.
(I don't think, I'll try to map format names from any jumbo older than
john-1.7.9-jumbo-5, but if you think this is necessary, please let me know.)

I know that there is still the issue of "Raw SHA-1" being used as format
name in rawSHA1_fmt_plug.c and rawSHA1_ng_fmt.c (the rawSHA1_ng_fmt.c
supports a smaller maximum password length) which needs to be resolved.

Also, please let me know if you totally disagree with my approach.

Frank

View attachment "0001-relbench-map-old-and-new-format-names.patch" of type "text/x-patch" (3420 bytes)

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.