← Back to team overview

touch-packages team mailing list archive

[Bug 1065169] Re: segfault when used in plugin load, unload, load cycle

 

quantal has seen the end of its life and is no longer receiving any
updates. Marking the quantal task for this ticket as "Won't Fix".

** Changed in: indicator-messages (Ubuntu Quantal)
       Status: Fix Committed => Won't Fix

-- 
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to indicator-messages in
Ubuntu.
https://bugs.launchpad.net/bugs/1065169

Title:
  segfault when used in plugin load, unload, load cycle

Status in The Messaging Menu:
  Fix Released
Status in indicator-messages package in Ubuntu:
  Fix Released
Status in indicator-messages source package in Quantal:
  Won't Fix

Bug description:
  Impact:
  libmessaging-menu using apps can segfault when unloading and reloading code

  Test Case:
  - download the file from comment #1
  - gcc test-app.c -o test `pkg-config --cflags --libs messaging-menu`
  - run ./test

  with the fixed version it shouldn't segfault

  Regression Potential:
  Limited, check that the usecase is correctly supported

  ---
  I've been experimenting with adding messaging-menu support to an app through a plugin, and I've noticed errors/crashing when recreating the MessagingMenuApp object after having g_object_unref'd it, as in the situation where the user unloads the plugin, then loads it again.

  This is the error, which follows from pressing 'Plugin Load', 'Plugin
  Unload', 'Plugin Load' in the test application:

  ** (test-app:3805): WARNING **: unable to export action group: An
  object is already exported for the interface org.gtk.Actions at
  /com/canonical/indicator/messages/test_app_desktop

  (test-app:3805): GLib-GIO-CRITICAL **:
  g_dbus_connection_register_object: assertion `error == NULL || *error
  == NULL' failed

  ** (test-app:3805): WARNING **: unable to export menu: (null)

  I would include a stack trace but it is not consistent, so I suspect
  some sort of memory corruption (sometimes it doesn't crash at all,
  just spins at 100% and brings down compiz). I'm going to take a look
  at indicator-messages and see if I notice anything but I thought I'd
  post the bug first, maybe someone could immediately tell what the
  issue is, or suggest a fix in my code if I am not freeing the object
  correctly before recreating it.

To manage notifications about this bug go to:
https://bugs.launchpad.net/indicator-messages/+bug/1065169/+subscriptions