> I fixed a race condition in mmsystem.c that could cause heap corruption > (and has done so): The problem is that the proc_PlaySound thread could > exit (in case of error, like sound card not available) before the > assignment wps->hThread = CreateThread(...) was completed. At the end > of proc_PlaySound wps got freed, and the memory of wps->hThread gets > the next free pointer in the heap control area. > > Please review my patch carefully. I am not that much experienced with > Windows and synchronisation, but I think, it should work. Yes, your patch should work. I have another (larger) fix for some other problems (and which shall fix this one too). I'll send it later on today (when I'm back home) A+
