bug#24614: 25.1.50; Exception after moving mouse over over a different window

[Eli Zaretskii]

> From: Alain Schneble <a.s@realize.ch>
> CC: <24614@debbugs.gnu.org>
> Date: Tue, 4 Oct 2016 18:25:34 +0200
> 
> > OK, so what does this print:
> >
> >  (gdb) frame 10
> >  (gdb) p *glyph
> >
> > also these:
> >
> >  (gdb) p *width
> >  (gdb) p w->contents
> >  (gdb) xtype
> >
> > If the last command says it's a buffer, type "xbuffer", otherwise type
> > "xwindow" and show the result.
> 
> (gdb) frame 10
> #10 0x00000004000469f9 in get_window_cursor_type (active_cursor=<synthetic 
> pointer>, width=0xbfd6ac, glyph=0x10c46fd0, w=0x12af8400) at xdisp.c:28096
> 28096       if (glyph != NULL && glyph->type == XWIDGET_GLYPH)
> (gdb) p *width
> $3 = 1
> (gdb) p w->contents
> $4 = 282715685
> (gdb) xtype
> Lisp_Vectorlike
> PVEC_BUFFER
> (gdb) xbuffer
> $5 = (struct buffer *) 0x10d9e620
> (unsigned char *) 0xc661598 "TaskList.org"

What about

 (gdb) p *glyph

?

> > Btw, how come the call to get_window_cursor_type is shown with its
> > argument in reverse order?  In the backtrace:
> 
> I have no clue... Could optimization reorder the arguments?

Maybe the function was inlined.  Or maybe it's a GDB bug.

> > Something really weird went on in that frame #10...
> 
> Should I have fear of something now? ;)

Fear of a crash, of course.

If this happens again, I suggest to build an unoptimized binary and
see if the problem can be reproduced there.