Crash in Win7 using fastloop

Hello :)

My apologies in advance that I am not good with bug reports, but I released a demo recently for MMF and Windows 7 users had been complaining that their computer crashes at a certain point in the demo where a file is loaded and a large number of objects are created and destroyed in a single event loop. I don't know if this is a known issue or not, but it only seems to occur on computers running Windows 7.

To reproduce the bug, I've attached a source file, but it depends on some external files found only in the demo, so please download the demo first and run the source file to reproduce the bug accurately. (No crash will occur if the file the program looks for is not found.)

The demo:
http://pouet.net/prod.php?which=54271

The frame in particular which reproduces the crash:
http://mfa.aquadasoft.com/view/1262916247-nulltro_crashtest


Warning: while the source file to reproduce the crash is SFW, the demo itself contains some juvinile humor and swearing, so be aware, thanks :]

Hopefully, this bug is easy to fix. (I don't have Windows 7 so I can't reproduce the crash, personally...)


Edit: Using build r248 on windows xp sp3

sorry for the bump, but any new developments on this issue? Even an acknowledgment that the bug is reproducible?

You say you use the build 248, have to tried to rebuild your demo with the latest version of the build 249? That would be the first thing to do.

hi again, sorry for not getting around to this sooner.

I assumed r248 was the last stable version, which is why I still use it, but I tried r249 b10 today and recompiled it. According to the person who tested it, the demo still crashes on Windows 7 after being recompiled under the latest build. Whatever is causing it to crash seems to have not changed between the two builds.

Probably an obvious question, but have they tried running in compatibility mode for an older Windows?

I can confirm that it crashes in my Windows 7. According to my debugger, it's an access violation reading from a pointer that looks valid, suggesting an "access after free" style error.

EDIT: After installing the Windows Debugging Symbols, I can now see some of the callstack. It's in free() at the time, so it must be a double-free error, or a free on an offset pointer. I would need debug symbols for stdrt.exe to find out more.

yes. I'm pretty sure that all the 'obvious' stuff has been isolated out now. However, I still don't have Windows 7 (and don't want it!) So I guess this will just have to be something Win7 users will have to live with. Hopefully it isn't a consistent bug with >x events per loop or >x Actives per loop being loaded causing the runtime to crash on 7, cause I can imagine other programs doing this from time to time (if the programs are reasonably complex)...

I can reproduce the problem. Could you send me the MFA file so that I can build it and try it with the debug version of MMF2?