[Interest] Windows segfault, but not on Linux

[Bo Thorsen]

Hi William,

Sounds like you have quite a nightmare of a bug there.

I don't have a solution for you, but I can tell you my starting points 
for trying to figure this one out.

First is *always* this one: Have you run your application in valgrind?

You can also try and run it in the debugger on Linux and see if 
something weird is picked up.

It's possible that you are hit by compiler differences. To test this, 
you could try and compile with another compiler on Linux - Intel or a 
GCC. You can also try and find some code testing tool to analyze your 
code for potential problems.

It's possible that you have found a bug in the way Qt paints on Windows. 
Or even a compiler bug. But remember that for every 100 times we 
developer claim we found a system bug, only actually is :)

Sorry that I can't be more helpful.

Bo.

Den 21-11-2013 07:13, William Hallatt skrev:
> Good day everybody,
>
> I have a bit of an odd scenario.  At home I have Qt 5.2.x built from
> source, running on an Ubuntu 12.04 distribution.  I also use QtCreator
> 3.x and Clang 3.3 (both built from source) for my Qt development.  Long
> story short, I have a little data-gathering simulation application that
> I can run for hours and hours on the Ubuntu box without any issues (I
> regularly run it overnight), but when compiling the exact same source
> using the Qt 5.1.1 for Windows 32-bit (MinGW 4.8, OpenGL, 666 MB)
> <http://download.qt-project.org/official_releases/qt/5.1/5.1.1/qt-windows-opensource-5.1.1-mingw48_opengl-x86-offline.exe>package
> on Windows 7 (64-bit), I get segfaults within minutes.
>
> I have attached here two disassembler text files, the one is where the
> break occurs and the other is for the only Qt related stack that I can
> find in the ~30 threads that is running at that point.  I have marked
> the breaks with huge ( BREAKS HERE!!! ) markers in the text files (to
> the right of the actual line) and the attached image is also of the
> complete stack for the Qt-related function calls.  I seriously doubt the
> disassemled output would mean anything to anybody, but I include it here
> just in case.
>
> None of the stacks trace back directly to any of my code so I am at a
> bit of a loss, however, my actual problem is why everything works on
> Linux, but falls over on Windows...and please don't say it is because
> it's Windows, I know that already :P
>
> If anyone has any ideas, I'm all ears.
>
> Thanks a lot!
> William.
>
>
> _______________________________________________
> Interest mailing list
> Interest at qt-project.org
> http://lists.qt-project.org/mailman/listinfo/interest
>


-- 
Bo Thorsen, European Engineering Manager, ICS
Integrated Computer Solutions. Delivering World-Class Applications
http://ics.com/services