[Tuxmath-devel] Fix for new Windows TuxMath bug

[Kento L. <me...@kl...>]

Hey all,

I was bored so I cross-compiled TuxMath for Windows, but the game kept 
quitting when I tried to start a mission. Seg Fault! 
http://pastie.org/3759844

After mucking around in gdb, I found and fixed the problem... I attached 
the diff git gave me since I don't have write access to the git repo 
(diff for the head of the tuxmath t4k-0.1.1 branch). I haven't used C 
since high-school, so it might not be the ideal way...

Basically the problem is the MC_MathGame instance's question pointers 
are never explicitly initialized to null. The contents of the memory 
block malloc returns are implementation dependent, and on Windows 
apparently that can mean a block of memory filled with 0xbaadf00d [1]. 
Since MC_Initialize doesn't initialize the question pointers, we're left 
with a bunch of not-null pointers to invalid memory... obviously a 
problem. So, the simple fix is to explicitly initialize them to null.

Note: this isn't currently a problem on linux - which zeroes memory that 
wasn't previously allocated to the program (for security) - nor is there 
a problem with the Windows executable on the site - which was built back 
when MC_MathQuestion were globals that were explicitly initialized to null.

Cheers!
Kento Locatelli

[1] 
https://en.wikipedia.org/wiki/Magic_number_(programming)#Magic_debug_values 
<https://en.wikipedia.org/wiki/Magic_number_%28programming%29#Magic_debug_values>