After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 561306 - hangs on exit
hangs on exit
Status: RESOLVED OBSOLETE
Product: evolution
Classification: Applications
Component: Mailer
2.22.x (obsolete)
Other All
: Normal critical
: ---
Assigned To: evolution-mail-maintainers
Evolution QA team
Depends on:
Blocks:
 
 
Reported: 2008-11-18 04:26 UTC by Mikel Ward
Modified: 2012-12-19 19:02 UTC
See Also:
GNOME target: ---
GNOME version: 2.21/2.22



Description Mikel Ward 2008-11-18 04:26:28 UTC
Steps to reproduce:
Every time I quit Evolution, it hangs rather than exiting cleanly.  I have to kill the process.

Stack trace:


Other information:
Comment 1 André Klapper 2008-11-18 14:22:27 UTC
There's not a single Evolution call in that trace, so it's not helpful. Which distro is this?
Comment 2 Matthew Barnes 2008-11-18 15:59:01 UTC
Might be NetworkManager-related.  Looks like libnm-glib has got at least two or three main loops running (Thread 3 & 5, maybe 4.  I don't know where Thread 2 came from.)  Maybe one of them is doing a blocking poll?

Evolution doesn't appear to be at fault on this one, perhaps other than the fact that we still use libnm-glib.
Comment 3 Mikel Ward 2008-11-18 20:18:43 UTC
It's Fedora 9.

I've got no idea what it's doing at shutdown, but it's happening every time.

Is there some way I can kill just the nm-related thread to see if that helps?
Comment 4 Matthew Barnes 2008-11-18 20:50:31 UTC
I need to review GLib's shutdown semantics when multiple main loops are running.  If you'd like to do a little GDB debugging, I'd be interested in whether a breakpoint on g_main_loop_quit() is ever hit when you try to shutdown, and if so for which thread(s).

Another idea is to try disabling networking through nm-applet when Evolution is hung.  That's more of a longshot but it might jar the NM threads loose.  Of course I'm still not certain this is really NetworkManager-related at all.
Comment 5 Mikel Ward 2008-11-18 22:27:27 UTC
Quitting Evolution then running /etc/init.d/NetworkManager stop had no effect.  I then killed /usr/sbin/nm-system-settings and nm-applet, which also had no effect.

gdb evolution
run
Ctrl+C
break g_main_loop_quit
cont
(quit Evolution)

Does not break.

Maybe there's some way of seeing what resource the __poll is blocked on?

Will also try killing GNOME components (g-s-d, seahorse, bonobo, DBus, etc.) and see if any of those do it.
Comment 6 Mikel Ward 2008-11-18 23:07:07 UTC
A colleague suggested that maybe the window wasn't getting a proper close/quit event.

I had a quick look using xev, and assuming I'm not doing anything wrong, then he's right!

If I use xprop on the Evo window, grab the line matching _NET_WM_USER_TIME_WINDOW(WINDOW), then run xev -id with that hex number, I don't see any DestroyNotify event, whereas I do if I try the same method on a gedit window.

Also, to be more precise, when I say "hangs", then sidebar and toolbar go gray/disabled, the selected item in the mail list has its background go gray, and the mail preview pane stays with its white background.  It is totally unresponsive at this point, except to a TERM signal so far as I can tell.

I'll try again with another window manager just in case (I'm currently using Openbox).
Comment 7 Matthew Barnes 2008-11-19 00:29:36 UTC
Interesting.  When this kind of hang occurs on shutdown it's usually due to Camel going off in the weeds, but that would show up on a stack trace.

If you drag another window over and away from Evolution when it's hung like this, does the Evolution window repaint itself?

Now I'm starting to wonder if this might be a case of your window manager not getting along with Bonobo.
Comment 8 Mikel Ward 2008-12-05 04:40:30 UTC
I think what happened was due to my IMAP server not responding.  I had 18,000 messages in my INBOX, and it was timing out when responding to IMAP commands like EXPUNGE.
Comment 9 André Klapper 2012-08-12 13:10:25 UTC
Is this still an issue nowadays (3.2.3 or 3.4.3), or can this be closed as RESOLVED OBSOLETE?
Comment 10 Tobias Mueller 2012-12-19 19:02:52 UTC
Closing as per comment #9.