Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Sun, 17 Feb 2013 18:46:10 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: Strange Benchmark Issue

On 17 Feb, 2013, at 0:28 , magnum <john.magnum@...hmail.com> wrote:
> On 16 Feb, 2013, at 23:49 , Solar Designer <solar@...nwall.com> wrote:
> 
>> On Sat, Feb 16, 2013 at 11:46:44PM +0100, magnum wrote:
>>> I get it with native gcc (llvm) too. I think it's the libs. Sometime it works, sometimes not. It looks like a non-thread-safe problem except it appears without OMP :-)
>> 
>> What you wrote above suggests it's an uninitialized variable and ASLR.
> 
> 
> Yes, and I have looked for that. Can't see any problem. As soon as I try to debug it, the bug hides under a rock until I give up.

It turns out there is a krb5_is_thread_safe() call that can be used at runtime. We now do, and OMP is indeed turned off on my OSX. However, it still fails (sometimes).

It's odd that it only fails during a full test run, not (or very occasionally) when tested itself. I spent a lot of time on this earlier but I have given up. If I register the krb5 db formats before all plugs, the problem goes away. Maybe I should commit such a "fix"? I'm afraid it just hides a bug.

magnum

Powered by blists - more mailing lists

Your e-mail address:

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