Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [day] [month] [year] [list]
Date: Tue, 29 May 2012 22:04:53 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Bleeding-jumbo alignment sort-of-problems

Jim,

Just a heads-up. The latest CVS core john (and therefore also
jumbo-bleeding) adds a couple of checks for alignment in the return from
salt() and binary(). This only happens unless the format uses
format_default_* of course. Anyway this is a good thing (it will force
us to not write code that fail on Sparc) but sometimes it gives false
positives. I fixed the few offending formats (some were real bugs) in
January but changes in "upstream" magnum-jumbo since has lead to a
couple dynamic formats failing now:

Benchmarking: dynamic_17: phpass ($P$ or $H$)  [SSE2i 2x4x3]... FAILED
(salt (alignment))
Benchmarking: PHPass MD5 [SSE2i 2x4x3]... FAILED (salt (alignment))

This does not have to be fixed until Jumbo-1 of next core release, which
is probably not very soon. *Unless* of course it indicates real
problems! I haven't checked.

By the way I currently even have canaries in the self-tests, checking
for buffer overwrites in these buffers, but it's a fugly hack and will
probably be reverted or re-written. BTW to my surprise, that did not
catch one single problem. The TS sorted a LOT of problems a year ago, or
was it two?

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.