On Thu, 26 Apr 2001, Christoph Reichenbach wrote:
> > >> if (rect->x < clipzone.x) {
> > Reading uninitialized memory: rect->x
> > Stack trace where the error occurred:
> > _gfxop_clip() D:\src\freesci\src\gfx\operations.c, 94
> > _gfxop_buffer_propagate_box() D:\src\freesci\src\gfx\operations.c, 1219
> > _gfxop_full_pointer_refresh() D:\src\freesci\src\gfx\operations.c, 1330
> [...]
>
> Is there a similar report from earlier on? The same rect was clipped in
> 1324. (If there is, it might be caused by the gfx driver not initializing
> its pointer coordinates (IIRC, none of the drivers does this)).
Doesn't appear that it's triggered before that. There are several nested
if's that are hard for me to follow, but there may be a codepath where
these aren't initialized. Should the gfx drivers do it?
> When does this happen?
This only appears to happen when first running, but I haven't tried to see
if it comes up again. I think this tool only reports errors once, anyways.
> > Freeing dangling pointer.
> ???
> Whoah... while this looks critical, I'm at a loss to explain it. The code,
> while somewhat ancient in style, should not cause the problem _as far as I
> can tell_ - is there some way you can determine the value of
> compressionMethod when this happens, or the point where the pointer was
> freed first?
You're right, this is weird. I can't find any codepaths where the pointer
is freed or manipulated. I'll try to be more careful when reporting next
time and check things more thoroughly.
--
http://www.clock.org/~matt
