On Wednesday 04 of September 2013 14:29:31 Dan Williams wrote: > On Wed, 2013-09-04 at 18:18 +0200, Jirka Klimes wrote: > > On Tuesday 03 of September 2013 13:36:49 Dan Williams wrote: > > > This part of the logs shows the problem: > > I didn't debug much further, but what's interesting is also > > GLib-CRITICAL **: g_hash_table_iter_next: assertion `ri->version == > > ri->hash_table->version' failed > > > > It comes from remove_agent(): > > g_hash_table_iter_init (&iter, priv->requests); > > while (g_hash_table_iter_next (&iter, NULL, &data)) > > > > request_remove_agent ((Request *) data, agent); > > > > But I do not know why, because iter is not manipulated here. > > See http://ur1.ca/fdkca > > > > > NetworkManager[13479]: dbus_g_proxy_cancel_call: assertion > > > `!DBUS_G_PROXY_DESTROYED (proxy)' failed NetworkManager[13479]: > > > dbus_g_proxy_begin_call: assertion `!DBUS_G_PROXY_DESTROYED (proxy)' > > > failed > > > > > > *** this is likely because the applet's D-Bus connection was closed, and > > > the dbus-glib DBusGProxy object got a "destroyed" signal from D-Bus, and > > > NM 0.9.8 doesn't have the fixes to avoid this error message. These two > > > messages should be coming from nm_secret_agent_cancel_secrets(). The > > > fix for this would be to attach to the "destroyed" signal of the agent's > > > DBusGProxy and then maybe set a priv->destroyed variable in > > > NMSecretAgent, and then not call these two functions in > > > nm_secret_agent_cancel_secrets() if priv->destroyed == TRUE. > > > > For the proxy stuff, I cherry-picked and updated > > 802d4cdad24f9d1a2bded3c4ae5680084ba9542e master commit - see > > jklimes/098-proxy-destroy > > Branch looks good, please merge to 098. > Pushed: a10dea328309d6fb3d3c6c3f543c1aeb189c4411
Jirka
|