[Box Backup] Failing tests

Chris Wilson boxbackup@fluffy.co.uk
Fri, 30 Nov 2007 23:52:18 +0000 (GMT)

Hi Guno,

On Mon, 26 Nov 2007, Guno Heitman wrote:

> Program received signal SIGSEGV, Segmentation fault.
> Timers::Cleanup () at
> /usr/lib/gcc/i486-slackware-linux/4.1.2/../../../../include/c++/4.1.2/bits/stl_vector.h:752
> 752             this->_M_impl._M_finish = this->_M_impl._M_start;
> Happens every time, on two different machines.
> > The test failure immediately before it is interesting, it reminds me 
> > of the problem Matt Brown reported and which we haven't fixed yet. Is 
> > there any way that you could give me remote access by SSH to a system 
> > where this test fails, for debugging?
> I'm sorry, can't do that. However, since it happens on two distinct 
> machines (both running on slackware 12, one of which is almost 
> unmodified after the OS was installed), chances are that a clean install 
> of slackware 12.0 is enough to reproduce the failing tests (and the 
> segfault). In case you have an unused machine lying around somewhere, 
> you could try installing slackware and see if you can reproduce the 
> problem.

Sorry, just tried Slackware 12 under qemu and all tests pass for me. So I 
don't have an obvious way to reproduce the problem. I'm afraid I will need 
your help to fix it.

> And if there are any other tests that I can perform for you, I'd be
> happy to.

I suspect that what's happening is that the C++ objects are being 
deallocated in a bad order, which causes the built-in memory leak tracking 
to fail. To check this, please could you run the test under gdb (using the 
t-gdb script in debug/test/bbackupd) and print a backtrace when it crashes 
(bt command)?

Thanks for your help!

Cheers, Chris.
_____ __     _
\  __/ / ,__(_)_  | Chris Wilson <0000 at qwirx.com> - Cambs UK |
/ (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer |
\ _/_/_/_//_/___/ | We are GNU-free your mind-and your software |