Re: And another crash when scrolling around. No xassert this time.

[Kim F. Storm]

David Kastrup <address@hidden> writes:

> Which would be the UNBLOCK_INPUT line in xdisp.c at
>
> #ifdef HAVE_WINDOW_SYSTEM
>   if (FRAME_WINDOW_P (f)
>       && update_window_fringes (w, 0)
>       && !just_this_one_p
>       && (used_current_matrix_p || overlay_arrow_seen)
>       && !w->pseudo_window_p)
>     {
>       update_begin (f);
>       BLOCK_INPUT;
>       if (draw_window_fringes (w, 1))
>       x_draw_vertical_border (w);
>       UNBLOCK_INPUT;
>       update_end (f);
>     }
> #endif /* HAVE_WINDOW_SYSTEM */


I don't see how interrupt_input_blocked got zeroed out in that
context -- unless TOTALLY_UNBLOCK_INPUT was called, and that
only happens if an error is signalled or the X server connection
is lost.  But in those cases, emacs should have longjmp'ed to the
error handler.  So I'm puzzled.


Could this be GTK related?

> In GNU Emacs 22.0.50.6 (i686-pc-linux-gnu, GTK+ Version 2.6.2)
>  of 2005-02-20 on lola.goethe.zz


> #0  abort () at /home/tmp/emacs/src/emacs.c:456
> #1  0x080a5e8b in redisplay_window (window=141773068, just_this_one_p=1)
>     at /home/tmp/emacs/src/xdisp.c:12478

> #6  0x08089f3e in sit_for (sec=0, usec=0, reading=0, display=1, 
>     initial_display=1) at /home/tmp/emacs/src/dispnew.c:6358

> (gdb) xbacktrace
> "sit-for"
> "line-move"

I don't know why calling sit-for there should cause problems.
The only reason you may see the problem _now_ is that calling
sit-for in line-move causes it the be called much more frequently
than before...  so perhaps the error has been there for a long time...

-- 
Kim F. Storm <address@hidden> http://www.cua.dk