[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

gdk_property_get() warnings



        Hi,

 Some of you might have noticed gdk_property_get() warnings are
 appearing in their consoles since the latest Gtk package in Debian.
   This is because a warning has been added to spot a common mistake, it
 is not a bug in Gtk.

 I think this problem is 64-bit specific, and is due to an integer
 overflow when gdk_property_get() is called with G_MAXLONG or INT_MAX.
 You should simply not pass more than G_MAXLONG - 3 to
 gdk_property_get().
   I attach a sample patch against Firefox to demonstrate the required
 change.  Another patch against vte is at:
    <http://bugzilla.gnome.org/attachment.cgi?id=48330&action=view>

 This bug is most notably present in vte (and hence exposed to apps
 using libvte) and in mozilla and firefox (and hence exposed to browsers
 based on these).

 I've found the following packages with gdk_property_get calls installed
 on my system:
 classpath gnome-panel helix-player kaffe-pthreads libsablevm-native1
 libvte4 mozilla-browser mozilla-firefox mozilla-thunderbird nautilus
 nautilus-cd-burner unison2.9.1-gtk

 The following bugs are related to this problem:
   315083 315462 316940 317673 317937

 I've checked (and reassigned accordingly) bug reports on:
 epiphany-browser firefox galeon gnome-terminal grip gtkterm

   Bye,

-- 
Loïc Minier <lool@dooz.org>
Come, your destiny awaits!
diff -urN mozilla-firefox-1.0.5.orig/gfx/src/gtk/nsScreenGtk.cpp mozilla-firefox-1.0.5/gfx/src/gtk/nsScreenGtk.cpp
--- mozilla-firefox-1.0.5.orig/gfx/src/gtk/nsScreenGtk.cpp	2004-04-20 19:35:51.000000000 +0200
+++ mozilla-firefox-1.0.5/gfx/src/gtk/nsScreenGtk.cpp	2005-07-17 18:10:27.104328544 +0200
@@ -143,7 +143,7 @@
   if (!gdk_property_get(root_window,
                         gdk_atom_intern ("_NET_WORKAREA", FALSE),
                         cardinal_atom,
-                        0, G_MAXLONG, FALSE,
+                        0, G_MAXLONG - 3, FALSE,
                         &type_returned,
                         &format_returned,
                         &length_returned,

Reply to: