I had the brilliant idea of running GDB on Suns java.exe to see if it too showed the heap corruption, and it did. I closed the bug as not worth pursuing at this point. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10919 Øyvind