Questions on Microsoft C Exceptions, Gcc, and Gdb

[Ángel González]

On 09/02/16 22:34, Thomas Dineen wrote:
> Gentle People:
>
>       I am working on a rather large application based on GTK / Cairo 
> using Mingw GCC for the MS Windows platform.
> Debugging has gone well for some time finding and fixing a significant 
> number of bugs including crash bugs.
>
>      However given a fairly complex multi - thread architecture I am 
> having trouble with GDB finding the specific
> code location of code crash bugs. In many cases Mingw GDB dose not 
> seem usable. (due to the multi - threaded
> architecture?). or my lack of GDB skill?
>
>      Now the question? Given the Mingw GCC / GDB environment running 
> on Windows: Is there a simple way to
> find the crash point in the code? MS claims an exception! Is there a 
> way to attach an exception handler to print
> out more information about the crash point? Please note that I am 
> aware of C Exceptions throw and catch which
> require implementation in the code.
>
>      MS dose provide a Exception Offset when the crash occurs. Is 
> there any way to use gdb or other tool to find
> the crash point in the code. Disassembler? Do the exception addresses 
> match the offsets found in a tool such as a
> disassembler?
>
> Thanks for help or ideas.
> Thomas Dineen

First of all, I expect you are compiling with debugging symbols (-ggdb 
or at least -g).

And then, when you have a debugger attached to the running program, once 
it crashes (no matter on which thread), the debugger will point you 
exactly to the point where it happened. At most you may need to browse 
the backtrace (bt) to another stack frame (up) to figure out the exact 
point in your code (as it may have crashed inside a library you called).

Knowing why it crashed should be relatively straightforward (eg. a null 
pointer reference inside the linked list). A different problem may be to 
determine why the environment was that way (why there was a null pointer 
where none was expected).

Regards