Crash during MIDI import

[GoogleCodeExporter]

After all the updates to the MIDI code, I tried to import the unedited chart 
for "Livin' on a Prayer", but EOF consistently crashes before reaching the end 
of MIDI import.  If I run it in debug mode in Codeblocks, it doesn't crash..  
Sometimes, I get different results based on whether I use F6 to bring up the 
MIDI import dialog instead of just using the mouse, but this hasn't been 
entirely reliable for preventing the crash.

Original issue reported on code.google.com by raynebc on 26 Oct 2010 at 9:29

[GoogleCodeExporter]

The import may be aborting somehow in a way that EOF doesn't realize that the 
chart isn't loaded.  For some reason, EOF believes the F2 key or CTRL+S was 
pressed, initiating a save, which causes a crash.  This is probably due to a 
buffer overflow or something, the call stack that is shown when I attach GDB to 
the crashed process is:

eof_logic() -> eof_read_editor_keys() -> eof_menu_file_save() -> 
eof_save_helper(NULL) -> eof_export_midi()

Original comment by raynebc on 26 Oct 2010 at 9:41

[GoogleCodeExporter]

r507 begins adding checking to track down the cause of this.  I don't think 
that the stack trace reported by attaching once the crash occurs is accurate, 
as live debugging mechanisms (ie. allegro_message() and set_window_title()) 
indicate this occurs in during pass 2 of MIDI import.

Original comment by raynebc on 26 Oct 2010 at 11:30

• Changed state: Started
• Added labels: Priority-Critical
• Removed labels: Priority-Medium

[GoogleCodeExporter]

After I couldn't cause this crash in EOF 1.7RC3 (r500), I decided to delete my 
EOF config file and try again.  After this, it succeeded a bit until I changed 
my EOF configuration and re-opened it.  From here on, it crashed again.  I 
still cannot get it to crash when I launch through a debugger.

Original comment by raynebc on 26 Oct 2010 at 11:43

[GoogleCodeExporter]

After my last commit, I'm now not able to reproduce this crash on demand, the 
unedited MIDI imports without crashing EOF.  I'll continue to look at the 
import logic.

Original comment by raynebc on 27 Oct 2010 at 5:10

• Added labels: Priority-High
• Removed labels: Priority-Critical

[GoogleCodeExporter]

I haven't been able to recreate the crash for a while, so the debug logic can 
remain in place.  It won't compile into the program unless 
EOF_DEBUG_MIDI_IMPORT is defined, and the definition is commented out right 
before the eof_import_midi() function.  Until the problem returns, something 
must have changed to correct the problem and I'll consider it fixed.

Original comment by raynebc on 1 Nov 2010 at 5:29

• Changed state: Fixed