[Scummvm-devel] Re: MI2 / IJ4 memory trash

J.Brown (Ender/Amigo) ender at enderboi.com
Fri Nov 15 11:34:04 CET 2002


Nope.

I've basically got it, it seems to be a null pointer access. Not sure why
yet, but looking into it:

Loading room 74
==31010==
==31010== Invalid read of size 2
==31010==    at 0x805EDF0: Scumm::initRoomSubBlocks(void) (common/scummsys.h:265)
==31010==    by 0x805EA71: Scumm::startScene(int, Actor *, int) (scumm/scummvm.cpp:659)
==31010==    by 0x808C1DF: Scumm::o5_loadRoom(void) (scumm/script_v1.cpp:1288)
==31010==    by 0x8088D0F: Scumm::executeScript(void) (scumm/script.cpp:288)
==31010==    Address 0x0 is not stack'd, malloc'd or free'd

The crash only seems to occur if the whole intro sequence runs. Escaping
it seems to prevent the problem. I would say it's corruption from
something during the intro, however valgrind doesn't seem to find any
errors during THAT portion.. I'm not sure.

Looking into it - if anyone else has any ideas, let me know :)

 - Ender

On Fri, 15 Nov 2002, Ruediger Hanke wrote:

> Date: Fri, 15 Nov 2002 20:59:19 +0100
> From: Ruediger Hanke <tomjoad at muenster.de>
> To: "J.Brown (Ender/Amigo)" <ender at enderboi.com>
> Cc: scummvm-devel at lists.sourceforge.net
> Subject: Re: MI2 / IJ4 memory trash
>
> > Is this a reproducable crash, or does it only happen randomly at these
> > points?
> >
> > Eg, does ScummVM -always- crash in that intro sequence?
>
> Yes, the point at where it crashes is always the same (when it switches
> to the big view of the statue in Indy's hand) and it crashes every time.
> The error message is totally random, though.
>
> I had a similar crash in Monkey 2 on Booty Island although this is not
> always reproducable.
>
> I suppose it's an Endian bug? (Maybe Max could try on OS X ...?)
>
> RĂ¼diger
> --
> "...wherever there's somebody fightin' for a place to stand
> Or a decent job or a helping hand
> Wherever somebody's struggling to be free
> Look in their eyes Mom you'll see me."
> - Bruce Springsteen, "The Ghost of Tom Joad"
>





More information about the Scummvm-devel mailing list