On Sat, Dec 01, 2012 at 11:25:02AM -0800, kevin360 wrote:
> Chris,
> Thanks for looking at the log files for me. I've had hardware
> problems before in other machines, but every time I've ever had
> those issues then there are multiple programs that will segfault
> and/or the machine will mysteriously lock-up. Beforfe I posted this
> thread I ran memtest86 to see if there was something up the memory
> but it came back ok. Here it's just enfuse 4.1. enfuse 4.0 with
> openmp compliled in works fine which is why I was thinking it was
> something in the enfuse code, but maybe it's a change in the code
> that's working some other lib and exposing a problem there.
In the past I've been searching for a bug in enblend. I got mad at the
fact that nobody would fix it, so I debugged and fixed it myself.
This type of bug, with for instance things like accessing memory
outside of the array bounds, or freeing memory just before the last
access to that memory is sensitive to the compiler and environment.
The one I tracked down happened with one compiler and not with
another. But in the end, it was a blatent programming error that
simply didn't crash on the other compiler.
Anyway, these errors are difficult to track down, as the place where
the crash happens is not related to the place where the bug is.
It would be too easy if you'd get a crash at:
z = x [i];
and found out that i==1000, while the array x runs from 0 to 999. In
practise, this namely simply continues with a "random" value in the z
variable. No crash. Then later on, things might turn into a crash,
because of the unreasonable value of z (easy), or because the
calculated value is placed back into the x array. Now some "random"
memory has been overwritten with a value that doesn't belong
there. That might cause a crash much further down the line.
Anyway, debugging this remotely with a day or two turnaround is
undoable. It takes several tens or hundreds of debugging runs to find
such a bug. So a developer with interest in the bug has to have access
to a system where it can be reproduced.
Roger.
> I'm running Slackware64 with gcc version 4.7.1. If I download the source
> code and the build script for the various libraries, compile with -g and
> link with -g, and then ran enfuse with valgrind, do you think that might
> help in identifying what library is causing the problem?
>
> Kevin
>
> On Saturday, December 1, 2012 10:25:11 AM UTC-5, cspiel wrote:
> >
> > Kevin -
> >
> > Am Freitag, 30. November 2012 10:24:16 UTC+1 schrieb kevin360:
> > > Also got it to crash using DRD and helgrind, here's the commandline that
> > I used for each:
> > > valgrind --tool=drd --read-var-info=yes
> > > valgrind --tool=helgrind
> > > The log file for each is up at:
> > > http://www.bluelavalamp.net/hugin/enfuse-openmp-segfault/
> >
> > IMHO, the logs only prove that you did
> > not run into a heisenbug. Otherwise, the
> > faults' addresses vary, which again points to
> > trouble outside of Enfuse's code base.
> >
> > As you are the only one who ever reported a
> > problem on Enfuse w/OpenMP and I could not
> > reproduce the error with three different tool
> > chains, I suspect a problem with your setup:
> >
> > compiler <-> libpthread <-> lib[g]omp
> >
> > I'd not even dare to exclude your hardware or
> > your OS in conjunction with the three items above.
> >
> > You could change your tools and/or your machine
> > or simply fall back to disabling OpenMP.
> >
> >
> > /Chris
> >
> >
>
> --
> You received this message because you are subscribed to the Google Groups
> "Hugin and other free panoramic software" group.
> A list of frequently asked questions is available at:
> http://wiki.panotools.org/Hugin_FAQ
> To post to this group, send email to [email protected]
> To unsubscribe from this group, send email to
> [email protected]
> For more options, visit this group at http://groups.google.com/group/hugin-ptx
--
** [email protected] ** http://www.BitWizard.nl/ ** +31-15-2600998 **
** Delftechpark 26 2628 XH Delft, The Netherlands. KVK: 27239233 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
The plan was simple, like my brother-in-law Phil. But unlike
Phil, this plan just might work.
--
You received this message because you are subscribed to the Google Groups
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at:
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx
