GNOME Bugzilla – Bug 665752
Crash on startup
Last modified: 2011-12-09 21:41:53 UTC
Created attachment 203026 [details] debug log I am consistently seeing the following crash on startup on OS X. The bockbuild environment used: https://github.com/DavidNielsen/bockbuild/commits/master This is using gtk-sharp master from the 2.12 branch, mono 2.10.7 and otherwise uses components we have used to build Banshee on OS X for the past couple of months or what Xamarin uses to build their official packages. at (wrapper managed-to-native) Gtk.Widget.gtksharp_widget_base_expose_event (intptr,intptr) <IL 0x00026, 0xffffffff> at Gtk.Widget.OnExposeEvent (Gdk.EventExpose) [0x00000] in /Users/gnomeuser/Projects/bockbuild/profiles/banshee/build-root/gtk-sharp-2-12-branch/_build/mono-gtk-sharp-ae2a95c/gtk/generated/Widget.cs:2000 at Banshee.Sources.Gui.SourceView.OnExposeEvent (Gdk.EventExpose) [0x000b4] in /Users/gnomeuser/Projects/banshee/src/Core/Banshee.ThickClient/Banshee.Sources.Gui/SourceView.cs:343 at Gtk.Widget.exposeevent_cb (intptr,intptr) [0x0000d] in /Users/gnomeuser/Projects/bockbuild/profiles/banshee/build-root/gtk-sharp-2-12-branch/_build/mono-gtk-sharp-ae2a95c/gtk/generated/Widget.cs:1979 at (wrapper native-to-managed) Gtk.Widget.exposeevent_cb (intptr,intptr) <IL 0x00020, 0xffffffff> at (wrapper managed-to-native) Gtk.Application.gtk_main () <IL 0x00022, 0xffffffff> at Gtk.Application.Run () [0x00000] in /Users/gnomeuser/Projects/bockbuild/profiles/banshee/build-root/gtk-sharp-2-12-branch/_build/mono-gtk-sharp-ae2a95c/gtk/Application.cs:135 at Banshee.Gui.GtkBaseClient.Run () [0x00013] in /Users/gnomeuser/Projects/banshee/src/Core/Banshee.ThickClient/Banshee.Gui/GtkBaseClient.cs:208 at Banshee.Gui.GtkBaseClient.Startup () [0x00000] in /Users/gnomeuser/Projects/banshee/src/Core/Banshee.ThickClient/Banshee.Gui/GtkBaseClient.cs:82 at Hyena.Gui.CleanRoomStartup.Startup (Hyena.Gui.CleanRoomStartup/StartupInvocationHandler) [0x00044] in /Users/gnomeuser/Projects/banshee/src/Hyena/Hyena.Gui/Hyena.Gui/CleanRoomStartup.cs:54 at Banshee.Gui.GtkBaseClient.Startup<T> () [0x00024] in /Users/gnomeuser/Projects/banshee/src/Core/Banshee.ThickClient/Banshee.Gui/GtkBaseClient.cs:77 at Banshee.Gui.GtkBaseClient.Startup<T> (string[]) [0x0004d] in /Users/gnomeuser/Projects/banshee/src/Core/Banshee.ThickClient/Banshee.Gui/GtkBaseClient.cs:67 at Nereid.Client.Main (string[]) [0x00000] in /Users/gnomeuser/Projects/banshee/src/Clients/Nereid/Nereid/Client.cs:54 at (wrapper runtime-invoke) <Module>.runtime_invoke_void_object (object,intptr,intptr,intptr) <IL 0x00050, 0xffffffff> Is this gtk, gtk-sharp or mono?
Adding the amazing John Ralls for a comment on weither or not this is in gtk. The relevant gtk+ version and patches applied can be found here: https://github.com/DavidNielsen/bockbuild/blob/master/packages/gtk+.py
Well, it looks like the eventual crash is from Pango -- but that's just where the NULL pointer finally got dereferenced. The source of the NULL pointer is the actual problem, and there isn't enough information about that. Build debug versions (-O0 -ggdb3 is what I use) of cairo, pango, and gtk and see if you can find where that NULL pointer came from.
Created attachment 203167 [details] trace This is a trace as caught by the Apple crash reporter from a bundle built with -O0 -ggdb3. Hopefully it will prove useful.
Ah, OK. The problem is in this patch: http://bugzilla-attachments.gnome.org/attachment.cgi?id=201356 (from bug 647969). Something's wrong with dereferencing fontmap_class->load_fontset. I suggest that you close this bug and take up the problem with Kris on #647969.
(In reply to comment #4) > Ah, OK. > > The problem is in this patch: > http://bugzilla-attachments.gnome.org/attachment.cgi?id=201356 (from bug > 647969). > > Something's wrong with dereferencing fontmap_class->load_fontset. ? But a fontmap_class pointer is not even at play in the particular function (pango_core_text_font_map_load_fonset) ? I get the impression the final fallback, to load the font "Sans" fails, resulting in pango_core_text_fontset_new() to return NULL and the subsequent call to pango_core_text_fontset_get_key() to fail. > I suggest that you close this bug and take up the problem with Kris on #647969. Yes, let's close this bug and discuss the problem in #647969. *** This bug has been marked as a duplicate of bug 647969 ***