Baffled by a strange bug - please help!!!

Question

Hi all,
I am baffled by a bug in a program I am working on that I just
can't understand. Here is the code that causes a SIGSEGV crash:
268      int k = wfall_pixbuf.rowstride;
269      pix[0] = pix[ -k ];
270      pix[0] = pix[ -wfall_pixbuf.rowstride ];
Output of gdb:
Program received signal SIGSEGV, Segmentation fault.
0x00000000004152b2 in Display_Waterfall () at display.c:270
270          pix[0] = pix[ -wfall_pixbuf.rowstride];
wfall_pixbuf is a struct that encapsulates a GdkPixbuf together
with some variables that hold the values of some properties like
rowstride above.
I added lines 268 and 269 in despair and the program
works well if I comment out line 270, which does (I think)
the same job as 269. I can't see how wfall_pixbuf.rowstride
can change value from line 268 to 270???
Please help!!!

Answer 1

danielsoft wrote:
hello,
I think it's a type issue: what type is wfall_pixbuf.rowstride ?
in the case
int k = wfall_pixbuf.rowstride;
pix[0] = pix[ -k ];
you are making a negative value out of type "int" - precisely signed int
when wfall_pixbuf.rowstride is unsigned, the result is undefined...
what does
pix[0] = pix[ -(int)wfall_pixbuf.rowstride ];
do? should work the same as lines 268 and 269
Ahh, thank you very much! rowstride is of type guint, which is
a GTK2 type definition, essentially I think it boils down to
unsigned int. So that was the issue :-)
I am surprised that I didn't make a mistake of this sort till now
but its never too late.... ;-)