Gnome-session crashed by various applications
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
gnome-session (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: gnome-session
Since I upgraded to Ubuntu 11.04 a couple of weeks ago my gnome-session crashed several times when I was starting some application or opening a new window. At first I was not sure whether it wasn't always related to Mozilla Thunderbird but today my session crashed when I double-clicked a PDF file in Nautilus, expecting it to open in Evince. It is like in bug #282939 where Gnusound propagates a signal to the whole process group, which includes the running gnome-session.
The effect is, ofcourse, that the session ends and the log-in mask is presented.
Could it be that the X11 server generates (and propagates) such a signal on creation of a new window? Only I see that the X server, which is up even before log-in, is not running under the same process group as my gnome-session.
I suppose the killing of the session prevents any attempts to gather useful data about the crash. Log files show nothing suspicious (to me), except maybe this in .xsession-errors:
(process:3099): DEBUG: desktop-
It just happened again: When I opened calibre and clicked on the "convert" icon my GNOME session crashed, leaving me with the login screen. Linux was fine and even preserved my open DSL connection.
I attach my auth.log and syslog, which seem to contain a few suspicious lines, such as: authority= local): Unregistered Authentication Agent for unix-session: /org/freedeskto p/ConsoleKit/ Session2. .. worker[ 4652]: WARNING: Unable to load file '/etc/gdm/ custom. conf'.. . greeter[ 4649]: Gtk-WARNING: /build/ buildd/ gtk+2.0- 2.24.4/ gtk/gtkwidget. c:5687: widget not within a GtkWindow worker[ 4652]: GLib-GObject- CRITICAL: g_value_ get_boolean: assertion `G_VALUE_ HOLDS_BOOLEAN (value)' failed
* polkitd(
* gdm-session-
* gdm-simple-
* gdm-session-
Especially the assertion in the last line does not seem very reassuring to me but I have no idea what may have caused it. Grepping older logs shows, however, that this assertion was also logged on days when Gnome did not crash.