Bug 627188 – gdbus-non-socket test occasionally fails

Fix logical bug in test case to avoid race condition between the
client and the server.

Signed-off-by: David Zeuthen <davidz@redhat.com>
This commit is contained in:
David Zeuthen 2010-08-18 13:07:25 -04:00
parent 6c340c0b03
commit 7d6a6ca57b

View File

@ -237,9 +237,20 @@ test_non_socket (void)
stream = my_io_stream_new_for_fds (read_fd, write_fd);
guid = g_dbus_generate_guid ();
error = NULL;
/* We need to delay message processing to avoid the race
* described in
*
* https://bugzilla.gnome.org/show_bug.cgi?id=627188
*
* This is because (early) dispatching is done on the IO thread
* (method_call() isn't called until we're in the right thread
* though) so in rare cases the parent sends the message before
* we (the child) register the object
*/
connection = g_dbus_connection_new_sync (stream,
guid,
G_DBUS_CONNECTION_FLAGS_AUTHENTICATION_SERVER,
G_DBUS_CONNECTION_FLAGS_AUTHENTICATION_SERVER |
G_DBUS_CONNECTION_FLAGS_DELAY_MESSAGE_PROCESSING,
NULL, /* GDBusAuthObserver */
NULL,
&error);
@ -250,9 +261,6 @@ test_non_socket (void)
/* make sure we exit along with the parent */
g_dbus_connection_set_exit_on_close (connection, TRUE);
/* btw, no need to delay message processing since method
* invocations are delivered via the main loop
*/
error = NULL;
g_dbus_connection_register_object (connection,
"/pokee",
@ -263,6 +271,9 @@ test_non_socket (void)
&error);
g_assert_no_error (error);
/* and now start message processing */
g_dbus_connection_start_message_processing (connection);
g_main_loop_run (loop);
g_assert_not_reached ();