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 143851 - metacity stuck in an infinite loop
metacity stuck in an infinite loop
Status: VERIFIED INCOMPLETE
Product: metacity
Classification: Other
Component: general
2.8.x
Other Linux
: High critical
: ---
Assigned To: Metacity maintainers list
Metacity maintainers list
Depends on:
Blocks:
 
 
Reported: 2004-06-07 08:05 UTC by Hubert Figuiere (:hub)
Modified: 2009-08-15 18:40 UTC
See Also:
GNOME target: ---
GNOME version: 2.9/2.10



Description Hubert Figuiere (:hub) 2004-06-07 08:05:30 UTC
I don't know how to reproduce it, but it occurs quite often.
Visible symptoms are :
-CPU to max (in the cpu monitor applet)
-can no longer change the focussed window, or change workspace.

I have to kill -HUP the window manager.

Here is a gdb stack trace:

  • #0 gdk_interp_type_get_type
    from /usr/lib/libgdk_pixbuf-2.0.so.0
  • #1 gdk_interp_type_get_type
    from /usr/lib/libgdk_pixbuf-2.0.so.0
  • #2 gdk_interp_type_get_type
    from /usr/lib/libgdk_pixbuf-2.0.so.0
  • #3 pixops_scale
    from /usr/lib/libgdk_pixbuf-2.0.so.0
  • #4 gdk_pixbuf_scale
    from /usr/lib/libgdk_pixbuf-2.0.so.0
  • #5 gdk_pixbuf_scale_simple
    from /usr/lib/libgdk_pixbuf-2.0.so.0
  • #6 meta_draw_op_free
  • #7 meta_draw_op_free
  • #8 meta_draw_op_free
  • #9 meta_draw_op_list_draw
  • #10 meta_draw_op_free
  • #11 meta_draw_op_list_draw
  • #12 meta_frame_style_draw
  • #13 meta_theme_draw_frame
  • #14 meta_frames_notify_menu_hide
  • #15 meta_frames_notify_menu_hide
  • #16 _gtk_marshal_BOOLEAN__BOXED
    from /usr/lib/libgtk-x11-2.0.so.0
  • #17 g_cclosure_new_swap
    from /usr/lib/libgobject-2.0.so.0
  • #18 g_closure_invoke
    from /usr/lib/libgobject-2.0.so.0
  • #19 g_signal_emit_by_name
    from /usr/lib/libgobject-2.0.so.0
  • #20 g_signal_emit_valist
    from /usr/lib/libgobject-2.0.so.0
  • #21 g_signal_emit
    from /usr/lib/libgobject-2.0.so.0
  • #22 gtk_widget_send_expose
    from /usr/lib/libgtk-x11-2.0.so.0
  • #23 gtk_main_do_event
    from /usr/lib/libgtk-x11-2.0.so.0
  • #24 gdk_window_clear_area_e
    from /usr/lib/libgdk-x11-2.0.so.0
  • #25 gdk_window_process_all_updates
    from /usr/lib/libgdk-x11-2.0.so.0
  • #26 gdk_window_process_all_updates
    from /usr/lib/libgdk-x11-2.0.so.0
  • #27 g_child_watch_add
    from /usr/lib/libglib-2.0.so.0
  • #28 unblock_source
    from /usr/lib/libglib-2.0.so.0
  • #29 g_main_context_dispatch
    from /usr/lib/libglib-2.0.so.0
  • #30 g_main_context_dispatch
    from /usr/lib/libglib-2.0.so.0
  • #31 g_main_loop_run
    from /usr/lib/libglib-2.0.so.0
  • #32 main

I use version 2.8.1-1 from Debian PowerPC

Here is the line in top:
12481 hub       20   0  9156 8348 8260 R 76.6  2.2   7:33.82 metacity
Comment 1 Hubert Figuiere (:hub) 2004-06-07 08:06:41 UTC
WOW. Update. While committing this bug, metacity ended up going out of this
loop. I don't know what caused that.
Comment 2 Michael Knight 2004-06-12 07:27:49 UTC
I also experience this bug after having been in X for usually over a day or so.
I use version 2.8.1 (running Gentoo).

Changing window focus/virtual desktops is possible but it is incredibly slow. I
can see the border around the windows being drawn very slowly. The applications
themselves seem to run OK when this is happening. As stated, a kill -HUP fixes
it (temporarily, for another day or so).
Comment 3 Elijah Newren 2004-07-17 17:02:50 UTC
Oddly enough, the first few frames of the stack trace (from
gdk_interp_type_get_type down to gdk_pixbuf_scale_simple) basically match the
commonly reported (but long since fixed) stack trace found in bug 135224 for
gnome-system-monitor.  That probably doesn't mean anything, but it was unusual
enough that I thought I'd mention it.

Michael: Can you verify that you get the same stack trace as Hubert so that we
can be certain you are experiencing the same bug and not just another similar one?
Comment 4 Hubert Figuiere (:hub) 2004-07-19 14:49:50 UTC
Strangely I haven't gotten this problem for a while. I'm at version 2.8.1-3
(Debian), so unless they did apply a patch sent to upstream, it is the same code.
Comment 5 Michael Knight 2004-09-08 09:15:48 UTC
Here's my gdb stack trace. I'm a little worried, as it doesn't really look
anything like Hubert's one. I'm using gdb 6.2:

(gdb) where
  • #0 ??
  • #1 ??
  • #2 ??
  • #3 ??
  • #4 poll
    from /lib/libc.so.6
  • #5 g_main_loop_get_context
    from /usr/lib/libglib-2.0.so.0
  • #6 ??
  • #7 ??
  • #8 ??
  • #9 g_main_context_query
    from /usr/lib/libglib-2.0.so.0
  • #10 ??
  • #11 ??
  • #12 ??
  • #13 g_idle_remove_by_data
    from /usr/lib/libglib-2.0.so.0
  • #14 ??
  • #15 ??
  • #16 ??
  • #17 ??
  • #18 ??
  • #19 pthread_mutex_unlock
    from /lib/libpthread.so.0
  • #20 g_main_context_acquire
    from /usr/lib/libglib-2.0.so.0
  • #21 ??
  • #22 g_thread_use_default_impl
    from /usr/lib/libglib-2.0.so.0
  • #23 g_ascii_table
    from /usr/lib/libglib-2.0.so.0
  • #24 _g_debug_flags
    from /usr/lib/libglib-2.0.so.0
  • #25 ??
  • #26 ??
  • #27 ??
    from /usr/lib/libglib-2.0.so.0
  • #28 ??
  • #29 ??
  • #30 ??
  • #31 g_main_loop_run
    from /usr/lib/libglib-2.0.so.0
  • #32 ??
    from /usr/lib/libglib-2.0.so.0
  • #33 ??
  • #34 ??
  • #35 ??
  • #36 ??
  • #37 _dl_rtld_di_serinfo
    from /lib/ld-linux.so.2
  • #38 main

Here's the line in top:
31957 mike      15   0 16712 7892  13m S  1.0  0.8   3:06.49 metacity

Also interesting is the line in top for the X process:
31912 root      25   0  434m 189m 146m R 88.5 18.7  67:37.05 X

Metacity never really gets high CPU usage when this happens, it seems to only be
X. I'm running X.org 6.7.0 (xorg-x11-6.7.0-r2 on Gentoo). This and the stack
trace may indicate it is not the exact same bug as Hubert.

Sorry for the delay, I was not running Gnome for a while, but am back now. It's
still the same metacity (2.8.1), and this is the first time the slowdown has
happened since resuming use (a couple of days ago).
Comment 6 Elijah Newren 2004-09-08 14:54:26 UTC
Michael: That stack trace is unfortunately useless.  Can you recompile without
optimization flags, include the -g flag for debugging symbols, and make sure to
use FEATURES=nostrip in /etc/make.conf?
Comment 7 Elijah Newren 2004-09-09 05:12:50 UTC
Also, is there any chance someone can get us a verbose log?  To do so:

1. Reduce your desktop to as few windows as possible to reproduce the bug
2. Run METACITY_VERBOSE=1 METACITY_USE_LOGFILE=1 metacity --replace
3. On stdout metacity will print the name of the logfile
4. Reproduce the bug as quickly as possible
5. Kill the metacity you started above to stop the logfile from growing any longer
6. Compress the logfile and attach it to this bug report

I'm really curious (besides a log which would be the most helpful) about any
options you both might have set (choice of theme, click/sloppy/mouse focus,
autoraise, reduced resources, etc.).

Also, bug 152169 may be related.
Comment 8 Michael Knight 2004-09-09 10:54:00 UTC
Hi Elijah!

Thanks for your tips. I have set up metacity as instructed and will hopefully
get a useful stack trace next time it occurs.

I don't have any non-default metacity options, save for
action_double_click_titlebar = toggle_maximize, 6 virtual desktops and the
Nuvola theme.
Comment 9 Luis Villa 2005-01-25 03:58:37 UTC
I'm going to mark this bug NEEDINFO; it appears the maintainer can't do anything
else without more debug information.
Comment 10 Hubert Figuiere (:hub) 2005-01-25 19:21:16 UTC
I haven't had the problem anymore on my side. I'm tempted to mark it as "can not
reproduce". sorry for the noise.
Comment 11 Michael Knight 2005-01-26 00:08:36 UTC
After having updated Metacity and various other possibly related components (X,
etc.) I also have not seen this bug occur.
Comment 12 Elijah Newren 2005-01-26 00:14:09 UTC
/me wonders if it was just a buggy theme that was fixed in newer versions

Okay, well thanks for following up, I'll go ahead and mark this as closed.  Let
us know if you run into any other problems, though.