mate-panel crashes at startup #431

bol-van · 2016-04-30T06:27:13Z

Ubuntu 16.04 MATE. mate-panel crashes once after logon if compiz window manager is used instead of marco. Then it restarts. Tested on physical hardware (nvidia) and vmware.
crash is very stable. same crash was reported many times on the web

segfault at 6 ip 00007f155df0f7e5 sp 00007ffc1328bd18 error 4 in libgobject-2.0.so.0.4800.0[7f155dedb000+52000]

--- Want to back this issue? **[Post a bounty on it!](https://www.bountysource.com/issues/33485666-mate-panel-crashes-at-startup?utm_campaign=plugin&utm_content=tracker%2F314880&utm_medium=issues&utm_source=github)** We accept bounties via [Bountysource](https://www.bountysource.com/?utm_campaign=plugin&utm_content=tracker%2F314880&utm_medium=issues&utm_source=github).

monsta · 2016-04-30T09:54:29Z

Is there a report on Launchpad about it? I'm asking because it might contain a backtrace collected via apport (which would be helpful to us).

bol-van · 2016-04-30T10:17:11Z

Since this bug is very easy to reproduce I thought it would not be necessary.
But if you need it I post it.

_usr_bin_mate-panel.1000.crash.zip

monsta · 2016-05-05T10:45:02Z

Well, I don't see a backtrace in that archive. I can't reproduce it either in my VirtualBox.

bol-van · 2016-05-05T11:01:34Z

This archive is standard apport report. It can be unpacked by "apport-unpack" tool. Unpacked directory has CoreDump file. It can be opened by "gdb /usr/bin/mate-panel CoreDump". Then "bt" command in gdb. To see function names install mate-panel-dbg package.
Here is your backtrace :

#0  0x00007f0300a277e5 in g_type_check_instance_is_fundamentally_a () from /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#1  0x00007f0300a07c17 in g_object_unref () from /usr/lib/x86_64-linux-gnu/libgobject-2.0.so.0
#2  0x00000000004685bd in container_child_background_set (source_object=<optimized out>, res=0x10a0dc0, user_data=<optimized out>)
    at mate-panel-applet-frame-dbus.c:240
#3  0x00007f0300cbc447 in g_simple_async_result_complete () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#4  0x00000000004697d0 in set_applet_property_cb (source_object=<optimized out>, res=0x12f38e0, user_data=<optimized out>)
    at mate-panel-applet-container.c:560
#5  0x00007f0300ccdb33 in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#6  0x00007f0300cce1de in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#7  0x00007f0300d1ca2f in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#8  0x00007f0300ccdb33 in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#9  0x00007f0300ccdb69 in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#10 0x00007f030072bfda in g_main_context_dispatch () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#11 0x00007f030072c380 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#12 0x00007f030072c6a2 in g_main_loop_run () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#13 0x00007f0301f816a7 in gtk_main () from /usr/lib/x86_64-linux-gnu/libgtk-x11-2.0.so.0
#14 0x00000000004224cc in main (argc=1, argv=0x7ffeb8552e38) at main.c:170

I suppose reproducing the problem requires working 3d acceleration and working compiz.

monsta · 2016-05-05T11:32:25Z

Thanks, I wasn't aware of apport-unpack. Will check it.

monsta · 2016-05-05T13:14:48Z

Ouch. Looks similar to #214, that's weird.

monsta · 2016-05-05T14:11:46Z

Looks similar to https://bugzilla.redhat.com/1238820 as well.

monsta · 2016-05-24T14:41:28Z

Ok, new mate-panel build with possible fix is uploaded to the official MATE PPA:
https://launchpad.net/~ubuntu-mate-dev/+archive/ubuntu/xenial-mate

Please test if it fixes the crash for you (and that it does not cause any regressions).

bol-van · 2016-05-24T16:51:45Z

Same crash still happens

raveit65 · 2016-05-24T17:12:28Z

Same crash still happens

Whow, people working really hard to fix that bug, and you're not able to tell more than 4 words.......
Please attach a new stacktrace.

bol-van · 2016-05-24T17:56:13Z

I checked bugtrace and it looks identical. Strange thing is EXACTLY the same function addresses (aside ASLR). If codes is changed function offsets usually also change. Double checked version. Manually downloaded deb from PPA, unpacked, compared md5sum to actual /usr/bin/mate-panel.

#3  0x00007fc7843a8447 in g_simple_async_result_complete () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#4  0x00000000004697d0 in set_applet_property_cb (source_object=<optimized out>, res=0xb041d0, user_data=<optimized out>)
    at mate-panel-applet-container.c:560
#5  0x00007fc7843b9b33 in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#6  0x00007fc7843ba1de in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#7  0x00007fc784408a2f in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#8  0x00007fc7843b9b33 in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#9  0x00007fc7843b9b69 in ?? () from /usr/lib/x86_64-linux-gnu/libgio-2.0.so.0
#10 0x00007fc783e17fda in g_main_context_dispatch () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#11 0x00007fc783e18380 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#12 0x00007fc783e186a2 in g_main_loop_run () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#13 0x00007fc78566d6a7 in gtk_main () from /usr/lib/x86_64-linux-gnu/libgtk-x11-2.0.so.0
#14 0x00000000004224cc in main (argc=1, argv=0x7ffe103c8a78) at main.c:170

monsta · 2016-06-02T15:10:28Z

Weird... ok, I'll merge the commits to master for now, since they apparently do no harm. Will investigate more later.

monsta · 2016-06-10T10:09:30Z

Well, I'm not really sure if that would change things, but MATE 1.14 is now available for Ubuntu 16.04 via their official PPA: https://ubuntu-mate.org/blog/mate-desktop-114-for-xenial-xerus/

You can try it and see if anything changes...

monsta · 2016-06-27T15:18:15Z

@flexiondotorg: can you please update mate-panel to 1.14.1 in PPA?

flexiondotorg · 2016-06-27T23:29:29Z

@monsta Thanks for the reminder. Will do tomorrow.

monsta · 2016-09-13T13:29:52Z

@bol-van: can you update MATE to 1.14.x from PPA and check if mate-panel still crashes?

bol-van · 2016-09-13T14:37:45Z

1.14.2 vmware see no crash
cannot test now with nvidia on real hardware. not sure when i'll be at that place. Possibly on 22-sep may be later

joakim-tjernlund · 2016-09-19T16:31:21Z

On MATE 14 (mate-panel 1.14.2) I see SEGV (mate-panel) in dmesg when I logout.

did a mate-panel --replace an got this:
(mate-panel:3458): GLib-CRITICAL **: g_hash_table_remove_internal: assertion 'hash_table != NULL' failed

(mate-panel:3458): GLib-GObject-WARNING **: invalid unclassed pointer in cast to 'MatePanelAppletFrameDBus'

(mate-panel:3458): GLib-CRITICAL **: g_hash_table_remove_internal: assertion 'hash_table != NULL' failed

(mate-panel:3458): GLib-GObject-WARNING **: invalid unclassed pointer in cast to 'MatePanelAppletFrameDBus'

But now I don't see a SEGV :(

joakim-tjernlund · 2016-09-19T16:32:52Z

Forgot, on Gentoo using mozo and gtk2(where possible)

joakim-tjernlund · 2016-09-19T16:35:14Z

A bit more info from the kernel:
kernel: mate-panel[3802]: segfault at 6 ip 00007fdaed0a2b15 sp 00007ffdc2f97528 error 4 in libgobject-2.0.so.0.4600.2[7fdaed06f000+52000]

joakim-tjernlund · 2016-09-19T17:14:09Z

So new idea, generate a core file and a backtrace from core after the fact:

(gdb) bt
#0  0x00007f0a97da8fd9 in g_type_check_instance_is_fundamentally_a (type_instance=type_instance@entry=0x6, fundamental_type=fundamental_type@entry=80) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gobject/gtype.c:4030
#1  0x00007f0a97d890c8 in g_object_unref (_object=0x6) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gobject/gobject.c:3076
#2  0x000000000046b0fe in container_child_background_set (source_object=<optimized out>, res=0x9d9ca0, user_data=<optimized out>) at panel-applet-frame-dbus.c:239
#3  0x00007f0a9803fd57 in g_simple_async_result_complete (simple=0x9d9ca0) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gio/gsimpleasyncresult.c:801
#4  0x000000000046c395 in set_applet_property_cb (source_object=<optimized out>, res=0x9c8d00, user_data=<optimized out>) at panel-applet-container.c:560
#5  0x00007f0a980513db in g_task_return_now (task=0x9c8d00) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gio/gtask.c:1106
#6  0x00007f0a98051ab6 in g_task_return (task=0x9c8d00, type=<optimized out>) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gio/gtask.c:1164
#7  0x00007f0a980a13f9 in g_dbus_connection_call_done (source=<optimized out>, result=0x9c8dd0, user_data=0x9c8d00) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gio/gdbusconnection.c:5702
#8  0x00007f0a980513db in g_task_return_now (task=0x9c8dd0) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gio/gtask.c:1106
#9  0x00007f0a98051411 in complete_in_idle_cb (task=0x9c8dd0) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/gio/gtask.c:1120
#10 0x00007f0a97a8425e in g_main_dispatch (context=0x7dba30) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/glib/gmain.c:3154
#11 g_main_context_dispatch (context=context@entry=0x7dba30) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/glib/gmain.c:3769
#12 0x00007f0a97a84638 in g_main_context_iterate (context=0x7dba30, block=block@entry=1, dispatch=dispatch@entry=1, self=<optimized out>) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/glib/gmain.c:3840
#13 0x00007f0a97a84982 in g_main_loop_run (loop=0x9667d0) at /var/tmp/portage/dev-libs/glib-2.46.2-r3/work/glib-2.46.2/glib/gmain.c:4034
#14 0x00007f0a99314d17 in gtk_main () from /usr/lib64/libgtk-x11-2.0.so.0
#15 0x0000000000422bfc in main (argc=1, argv=0x7ffcc0d05548) at main.c:170

joakim-tjernlund · 2016-09-19T17:21:23Z

Maybe this can be helpful(missing a g_object_ref somewhere):
https://git.gnome.org/browse/evolution/commit/?id=b0ebeeb

monsta · 2016-09-20T13:39:44Z

Please use GitHub's formatting for code when posting backtraces - otherwise lots of unrelated reports get mentioned in the post...

joakim-tjernlund · 2016-09-20T17:39:06Z

Finally managed to find a way to quote the BT, can't say it is easy to use when you don't known MD
beforehand.

joakim-tjernlund · 2016-09-21T16:51:40Z

Just a hunch: Changed panel-applet-contatiner.c:560
g_simple_async_result_complete (result); ->g_simple_async_result_complete_in_idle (result);

Managed to login/logout a few times without SEGV but this SEGV comes and goes anyway

don't mess with additional references to a GCancellable, instead keep just a const pointer to the internal operation data and use it as a key when it's needed to cancel the operation. fixes #214 fixes #431

monsta · 2018-01-04T20:38:25Z

@joakim-tjernlund: check #701, it should fix the crash.

@bol-van: are you still able to reproduce the crash? I can provide the patched packages via my PPA to test.

joakim-tjernlund · 2018-01-04T22:19:03Z

@monsta, my crash is gone and has been for a while no idea why though(or I haven't paid
attention lately)

monsta · 2018-01-04T22:24:42Z

I think it's normal, it's a kind of random crash which can happen all the time for some users but don't happen for others. Could be a race condition between all these asynchronous callbacks.

joakim-tjernlund · 2018-01-04T22:24:56Z

Anyhow, applied it to latest mate-panel 1.18, see what happens ...

joakim-tjernlund · 2018-01-04T22:27:06Z

Starts Ok :) My TOK theme is causing some complaints(but no ill effects)

mate-panel --replace

** (mate-panel:8190): WARNING **: unsupported panel image background such as -gtk-gradient


** (mate-panel:8190): WARNING **: use an image file or a standard css gradient


** (mate-panel:8190): WARNING **: unsupported panel image background such as -gtk-gradient


** (mate-panel:8190): WARNING **: use an image file or a standard css gradient


** (mate-panel:8190): WARNING **: unsupported panel image background such as -gtk-gradient


** (mate-panel:8190): WARNING **: use an image file or a standard css gradient


** (mate-panel:8190): WARNING **: unsupported panel image background such as -gtk-gradient


** (mate-panel:8190): WARNING **: use an image file or a standard css gradient

bol-van · 2018-01-04T22:27:47Z

From my point of view it was fixed long time ago. I havent experienced similar crashes since when

joakim-tjernlund · 2018-01-04T22:33:41Z

@monsta, I hope mate-screensaver is on your list next

lukefromdc · 2018-01-05T04:31:28Z

I put the warnings there in a1445bc "Make panel image background in GTK theme work again. " Prior to that commit, image backgrounds for the panel did not work for the GTK theme in GTK 3 (at least in 3.22) while they had worked just fine in the GTK2 days.

The warnings are so theme developers would see that mate-panel background images containing -gtk-gradient or anything else only GTK understands and Cairo does not would not work and would yield a color background only. The background is drawn using Cairo and gdk_window_set_background_pattern rather than by GTK.

On top of all else, the underlying code in GTK to use -gtk-gradient is deprecated inside GTK:
https://developer.gnome.org/gtk3/stable/gtk3-GtkGradient.html
The only place I saw -gtk-gradient set on a panel background in an official MATE theme was in TOK and traditional-green. These should be updated to use a normal CSS gradient such as linear-gradient or radial-gradient(latter not likely for a panel).

don't mess with additional references to a GCancellable, instead keep just a const pointer to the internal operation data and use it as a key when it's needed to cancel the operation. fixes #214 fixes #431 (squashed 180ba4a + 7f711b8 for 1.18 branch)

monsta mentioned this issue Jan 4, 2018

rework the process of cancelling the background change operation #701

Merged

lukefromdc closed this as completed in #701 Jan 5, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

mate-panel crashes at startup #431

mate-panel crashes at startup #431

bol-van commented Apr 30, 2016 •

edited by flexiondotorg

monsta commented Apr 30, 2016

bol-van commented Apr 30, 2016

monsta commented May 5, 2016

bol-van commented May 5, 2016 •

edited

monsta commented May 5, 2016

monsta commented May 5, 2016

monsta commented May 5, 2016

monsta commented May 24, 2016

bol-van commented May 24, 2016

raveit65 commented May 24, 2016

bol-van commented May 24, 2016 •

edited

monsta commented Jun 2, 2016

monsta commented Jun 10, 2016

monsta commented Jun 27, 2016

flexiondotorg commented Jun 27, 2016

monsta commented Sep 13, 2016 •

edited

bol-van commented Sep 13, 2016 •

edited

joakim-tjernlund commented Sep 19, 2016

joakim-tjernlund commented Sep 19, 2016

joakim-tjernlund commented Sep 19, 2016

joakim-tjernlund commented Sep 19, 2016 •

edited

joakim-tjernlund commented Sep 19, 2016

monsta commented Sep 20, 2016

joakim-tjernlund commented Sep 20, 2016

joakim-tjernlund commented Sep 21, 2016

monsta commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

monsta commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

bol-van commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

lukefromdc commented Jan 5, 2018

mate-panel crashes at startup #431

mate-panel crashes at startup #431

Comments

bol-van commented Apr 30, 2016 • edited by flexiondotorg

monsta commented Apr 30, 2016

bol-van commented Apr 30, 2016

monsta commented May 5, 2016

bol-van commented May 5, 2016 • edited

monsta commented May 5, 2016

monsta commented May 5, 2016

monsta commented May 5, 2016

monsta commented May 24, 2016

bol-van commented May 24, 2016

raveit65 commented May 24, 2016

bol-van commented May 24, 2016 • edited

monsta commented Jun 2, 2016

monsta commented Jun 10, 2016

monsta commented Jun 27, 2016

flexiondotorg commented Jun 27, 2016

monsta commented Sep 13, 2016 • edited

bol-van commented Sep 13, 2016 • edited

joakim-tjernlund commented Sep 19, 2016

joakim-tjernlund commented Sep 19, 2016

joakim-tjernlund commented Sep 19, 2016

joakim-tjernlund commented Sep 19, 2016 • edited

joakim-tjernlund commented Sep 19, 2016

monsta commented Sep 20, 2016

joakim-tjernlund commented Sep 20, 2016

joakim-tjernlund commented Sep 21, 2016

monsta commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

monsta commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

bol-van commented Jan 4, 2018

joakim-tjernlund commented Jan 4, 2018

lukefromdc commented Jan 5, 2018

bol-van commented Apr 30, 2016 •

edited by flexiondotorg

bol-van commented May 5, 2016 •

edited

bol-van commented May 24, 2016 •

edited

monsta commented Sep 13, 2016 •

edited

bol-van commented Sep 13, 2016 •

edited

joakim-tjernlund commented Sep 19, 2016 •

edited