Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 19 Sep 2012 00:51:19 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: 1.7.9-jumbo-7

On 18 Sep, 2012, at 21:20 , magnum <john.magnum@...hmail.com> wrote:
> On 18 Sep, 2012, at 20:41 , magnum <john.magnum@...hmail.com> wrote:
>> On 18 Sep, 2012, at 20:28 , magnum <john.magnum@...hmail.com> wrote:
>>> There is a regression problem in hmac-sha1 only found by Test Suite. I will revert the clear_keys() patch (I thought it was to magnum-jumbo only).
>> 
>> Very odd. I only get the problem on OSX, not on Linux (both are 64-bit). The clear_keys() is not in -fixes, there are only very minor changes and an added split() that unifies case. I have no idea what goes wrong!
> 
> It's getting worse: The problem is not in hmacSHA1_fmt.c. If I revert the whole file to the Jumbo-6 version, Test Suite still does not pass. For good measure I tried reverting wordlist.c too, to no avail. I'm stumped! I'll have a break.

Fixed and committed.

The problem was that OSX native llvm use a SHA_BUF_SIZ of 80 because it's faster (this is added since Jumbo-6). Certain SSE2 optimizations in hmac-sha1 did not bring that into the equation.

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.