On Sun, 23 Sep 2001, Matt wrote: > 3. On win32, I can consistently crash FreeSCI with the following test > case: > a) start lsl2 > b) save the game, then quit > c) start lsl2, load the game, then quit > > This results in a crash (depending upon OS) because the pointer for > gamestate is invalid somehow when it's free'd in main.c > > You may notice some weird messages about restoring invalid handles and > things. Let me know if you don't -- that might mean it's win32-specific > (though I have no idea why that would be -- sigh). (now cc'ing the mailing list) Actually, after tracing through it completely, nothing happens to the pointer for gamestate. It's not being free'd twice as far as I can tell. This leads me to believe we are dealing with general memory corruption. Can you try the testcase above with dmalloc enabled? Can you even reproduce the problem on your machine(s)? Can anyone on the list reproduce FreeSCI crashing upon exit using the steps above? -- http://www.clock.org/~matt
