presence: Do not auto-start the GNOME Shell screensaver service

As is, we are requesting the screensaver proxy to be started too early
in the login process. One issue that this is causing that login may hang
for a long time, other possible issues appear to be that the screensaver
service is not running at all in the end.

Note that gnome-settings-daemon will request the startup later on.

Thanks to Xiaoguang Wang <xwang@suse.com> for finding this issue and
suggesting a fix!

Closes: #88

gnome-session/gsm-presence.c

@@ -229,6 +229,34 @@ on_screensaver_dbus_signal (GDBusProxy  *proxy,
         }
 }
 
+static void
+screensaver_get_active_cb (GDBusProxy  *screensaver_proxy,
+                           GAsyncResult *res,
+                           GsmPresence *presence)
+{
+        g_autoptr(GVariant) data = NULL;
+        g_autoptr(GError) error = NULL;
+        gboolean is_active;
+
+        data = g_dbus_proxy_call_finish (screensaver_proxy, res, &error);
+        if (data) {
+                if (error) {
+                        g_warning ("Could not retrieve current screensaver active state: %s",
+                                   error->message);
+                } else {
+                        g_warning ("Could not retrieve current screensaver active state!");
+                }
+
+                return;
+        }
+
+        g_variant_get (data, "(b)", &is_active);
+        if (presence->priv->screensaver_active != is_active) {
+                presence->priv->screensaver_active = is_active;
+                set_session_idle (presence, is_active);
+        }
+}
+
 static void
 on_screensaver_name_owner_changed (GDBusProxy  *screensaver_proxy,
                                    GParamSpec  *pspec,
@@ -240,9 +268,19 @@ on_screensaver_name_owner_changed (GDBusProxy  *screensaver_proxy,
         if (name_owner == NULL) {
                 g_debug ("Detected that screensaver has left the bus");
 
-                presence->priv->screensaver_proxy = NULL;
                 presence->priv->screensaver_active = FALSE;
                 set_session_idle (presence, FALSE);
+        } else {
+                g_debug ("Detected that screensaver has aquired the bus");
+
+                g_dbus_proxy_call (presence->priv->screensaver_proxy,
+                                   "GetActive",
+                                   NULL,
+                                   G_DBUS_CALL_FLAGS_NO_AUTO_START,
+                                   1000,
+                                   NULL,
+                                   (GAsyncReadyCallback) screensaver_get_active_cb,
+                                   presence);
         }
 
         g_free (name_owner);
@@ -327,8 +365,12 @@ gsm_presence_constructor (GType                  type,
                 g_warning ("Unable to register presence with session bus");
         }
 
+        /* This only connects to signals and resolves the current name owner
+         * synchronously. It is important to not auto-start the service!
+         */
         presence->priv->screensaver_proxy = g_dbus_proxy_new_sync (presence->priv->connection,
-                                                                   G_DBUS_PROXY_FLAGS_NONE,
+                                                                   G_DBUS_PROXY_FLAGS_DO_NOT_AUTO_START |
+                                                                   G_DBUS_PROXY_FLAGS_DO_NOT_LOAD_PROPERTIES,
                                                                    NULL,
                                                                    GS_NAME,
                                                                    GS_PATH,
@@ -442,6 +484,7 @@ gsm_presence_finalize (GObject *object)
 
         g_clear_pointer (&presence->priv->status_text, g_free);
         g_clear_object (&presence->priv->idle_monitor);
+        g_clear_object (&presence->priv->screensaver_proxy);
 
         G_OBJECT_CLASS (gsm_presence_parent_class)->finalize (object);
 }

[Marc-Antoine Perennou]

@bberg shouldn't this be "!data"?

[Benjamin Berg]

Ugh. you are right. How could I miss that?

[Benjamin Berg]

Overall pretty harmless btw. We'll just print a warning, and the initial sync should not be important anyway.

[Marc-Antoine Perennou]

Won't that cause a problem in the case where data is NULL, since afterwards you do g_variant_get (data, "(b)", &is_active);?

[Benjamin Berg]

Of course :-)

But it should never happen in the first place. And if it does, then you'll "just" get a critical warning and the screensaver state might be misreported by gnome-session.

Unless of course you compile with assertions disabled, then it'll just crash …

[Joshua Strobl]

acquired*