tobing
Since lately, my game crashes during exit. I've tracked down this to
static void shutdown_system_driver(void)
in system.c, where a destructor call for user_exe_path is called, on already destroyed data. It helps to move the new lines up, like this:
1static void shutdown_system_driver(void)
2{
3 if (active_sysdrv) {
4 ALLEGRO_CONFIG *temp = active_sysdrv->config;
5 if (active_sysdrv->user_exe_path) <<-- new lines
6 al_destroy_path(active_sysdrv->user_exe_path); <<-- must be called before shutdown_system()
7 if (active_sysdrv->vt && active_sysdrv->vt->shutdown_system)
8 active_sysdrv->vt->shutdown_system();
9 active_sysdrv = NULL;
10 /* active_sysdrv is not accessible here so we copied it */
11 al_destroy_config(temp);
12
13 while (!_al_vector_is_empty(&_al_system_interfaces))
14 _al_vector_delete_at(&_al_system_interfaces, _al_vector_size(&_al_system_interfaces)-1);
15 _al_vector_free(&_al_system_interfaces);
16 _al_vector_init(&_al_system_interfaces, sizeof(ALLEGRO_SYSTEM_INTERFACE *));
17 }
18}
Trent Gamblin
Makes sense, but it's not destroying it twice, it's trying to access user_exe_path through a freed pointer.
kazzmir
It doesn't look like the user_exe_path has anything to do with with current driver. Can you explain why your fix is necessary?
Trent Gamblin
My understanding was that active_sysdrv was being freed.
Edit: that comment seems to be erroneuous now.
Elias
Thanks, I committed it to SVN. (Was also me who had added those 2 lines to the wrong place... just wanted to get my speedhack entry working under Windows and wasn't careful enough 
)