Re: [Iris-devel] more on fs problem & segfaults

[Cédric Delfosse]

Hi

It seems that fs problem are coming from sawfish (1.0.1 for me). I have
no more problem with twm. Maybe we do not handle well the wm, but I'm
not a specialist.

With efence and gdb, I finally get a SIGSEGV on create_knot ! 
Here is the backtrace:

(gdb) bt
#0  0x4429b24a in createknot (scaling_factor1=16, scaling_factor2=64, 
    radius1=2, radius2=2, radius3=1) at theme_knot.c:143
#1  0x4429b832 in init () at theme_knot.c:221
#2  0x44290fc3 in iris_init () at iris.c:879
#3  0x08060716 in enable_vis_plugin ()
#4  0x40124b6c in gtk_signal_remove_emission_hook ()
   from /usr/lib/libgtk-1.2.so.0
#5  0x40123fd5 in gtk_signal_set_funcs () from /usr/lib/libgtk-1.2.so.0
#6  0x401220b3 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#7  0x401413ce in gtk_toggle_button_toggled () from
/usr/lib/libgtk-1.2.so.0
#8  0x40141b6f in gtk_toggle_button_toggled () from
/usr/lib/libgtk-1.2.so.0
#9  0x400f51b5 in gtk_marshal_NONE__NONE () from
/usr/lib/libgtk-1.2.so.0
#10 0x40123e6f in gtk_signal_set_funcs () from /usr/lib/libgtk-1.2.so.0
#11 0x401220b3 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#12 0x40093068 in gtk_button_clicked () from /usr/lib/libgtk-1.2.so.0
#13 0x40141aad in gtk_toggle_button_toggled () from
/usr/lib/libgtk-1.2.so.0
#14 0x400f51b5 in gtk_marshal_NONE__NONE () from
/usr/lib/libgtk-1.2.so.0
#15 0x40123e6f in gtk_signal_set_funcs () from /usr/lib/libgtk-1.2.so.0
#16 0x401220b3 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#17 0x40092fa8 in gtk_button_released () from /usr/lib/libgtk-1.2.so.0
#18 0x40093f58 in gtk_button_get_relief () from /usr/lib/libgtk-1.2.so.0
#19 0x400f4e3f in gtk_marshal_BOOL__POINTER () from
/usr/lib/libgtk-1.2.so.0
#20 0x40124013 in gtk_signal_set_funcs () from /usr/lib/libgtk-1.2.so.0
#21 0x401220b3 in gtk_signal_emit () from /usr/lib/libgtk-1.2.so.0
#22 0x40158acb in gtk_widget_event () from /usr/lib/libgtk-1.2.so.0
#23 0x400f4d85 in gtk_propagate_event () from /usr/lib/libgtk-1.2.so.0
#24 0x400f3eee in gtk_main_do_event () from /usr/lib/libgtk-1.2.so.0
#25 0x401a2457 in gdk_wm_protocols_filter () from
/usr/lib/libgdk-1.2.so.0
#26 0x401d54d8 in g_get_current_time () from /usr/lib/libglib-1.2.so.0
#27 0x401d5ae3 in g_get_current_time () from /usr/lib/libglib-1.2.so.0
#28 0x401d5c7c in g_main_run () from /usr/lib/libglib-1.2.so.0
#29 0x400f37e7 in gtk_main () from /usr/lib/libgtk-1.2.so.0
#30 0x0807d1d5 in main ()
#31 0x4031514f in __libc_start_main () from /lib/libc.so.6

Line 143 of theme_knot is for me beginning here:

->        knotobject.Colors[knotobject.numverts].r =
            ((2 / dist) + (0.5 * sin (beta) + 0.4)) / 2.0;
          knotobject.Colors[knotobject.numverts].g =
            ((2 / dist) + (0.5 * sin (beta) + 0.4)) / 2.0;
          knotobject.Colors[knotobject.numverts].b =
            ((2 / dist) + (0.5 * sin (beta) + 0.4)) / 2.0;

          knotobject.numverts++;
        }
    }

I will investigate this tomorrow. I think we go out of bounds somewhere.

'bye

-- Cédric

Le mer 15/05/2002 à 16:36, foser a écrit :
> Cédric Delfosse wrote:
> > 
> > That's very weird ! Maybe we have a memory corruption problem somewhere,
> > but it's not easy to debug a shared lib.
> > 
> 
> No, i've tried that and failed :)
> 
> Oh, see if this helps to have less fs problems : 1. in create_window, 
> set all GLWin X stuff to null (to make sure they get properly reset. 2. 
> in the 'f' switch part it only calls kill_gl stuff right now, make it 
> also do the window destroy stuff from stop_display (except the thread 
> stop of course). With these 2 things added i could not reproduce it at 
> all anymore (before i could occasionaly).
> 
> -fos