GNOME Bugzilla – Bug 751854
gnome-shell crashes in meta_clutter_init
Last modified: 2021-06-10 11:34:21 UTC
Created attachment 306650 [details] Logs of session closed Downstream bug: https://bugs.archlinux.org/task/44749 Basically the session is randomly closed. I usually see this in the logs after it happens: Jun 29 17:13:18 mateus-arch gnome-session[4491]: WARNING: Lost name on bus: org.gnome.SessionManager Logs from one of the times it happened attached.
Created attachment 306669 [details] Recent gnome-shell coredump My session was closed again, this is one of the coredumps generated around that time.
Another coredump: mateus ~ $ sudo coredumpctl gdb [sudo] senha para mateus: PID: 27048 (gnome-session-f) UID: 120 (gdm) GID: 120 (gdm) Signal: 6 (ABRT) Timestamp: Sex 2015-07-03 01:44:18 BRT (4min 3s ago) Command Line: /usr/lib/gnome-session/gnome-session-failed --debug --allow-logout Executable: /usr/lib/gnome-session/gnome-session-failed Control Group: /user.slice/user-120.slice/session-c1.scope Unit: session-c1.scope Slice: user-120.slice Session: c1 Owner UID: 120 (gdm) Boot ID: bc86fadc76f24e0aa5848c56babaf435 Machine ID: 6f73af040ad64d90ae2bddb5baf67eef Hostname: mateus-arch Coredump: /var/lib/systemd/coredump/core.gnome-session-f.120.bc86fadc76f24e0aa5848c56babaf435.27048.1435898658000000.lz4 Message: Process 27048 (gnome-session-f) of user 120 dumped core. GNU gdb (GDB) 7.9.1 Copyright (C) 2015 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-unknown-linux-gnu". Type "show configuration" for configuration details. For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>. Find the GDB manual and other documentation resources online at: <http://www.gnu.org/software/gdb/documentation/>. For help, type "help". Type "apropos word" to search for commands related to "word"... Reading symbols from /usr/lib/gnome-session/gnome-session-failed...(no debugging symbols found)...done. warning: core file may not match specified executable file. [New LWP 27048] [New LWP 27143] warning: Could not load shared library symbols for linux-vdso.so.1. Do you need "set solib-search-path" or "set sysroot"? [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib/libthread_db.so.1". Core was generated by `/usr/lib/gnome-session/gnome-session-failed --debug --allow-logout'. Program terminated with signal SIGABRT, Aborted.
+ Trace 235228
I have to know what the logs and assertion messages are here.
I think I actually choose the wrong product.
Excerpt of logs: Jul 06 10:13:06 mateus-arch gnome-session[8520]: WARNING: Lost name on bus: org.gnome.SessionManager Jul 06 10:13:14 mateus-arch gnome-session[8520]: WARNING: App 'gnome-settings-daemon.desktop' exited with code 1 Jul 06 10:13:14 mateus-arch gnome-session[8520]: WARNING: App 'gnome-shell.desktop' exited with code 1 Jul 06 10:13:16 mateus-arch bluetoothd[711]: Endpoint unregistered: sender=:1.276 path=/MediaEndpoint/A2DPSource Jul 06 10:13:16 mateus-arch bluetoothd[711]: Endpoint unregistered: sender=:1.276 path=/MediaEndpoint/A2DPSink Jul 06 10:13:27 mateus-arch kernel: traps: gnome-shell[8185] trap int3 ip:7fc638a07b5b sp:7ffdc1605720 error:0 Jul 06 10:13:27 mateus-arch gdm-password][8105]: pam_unix(gdm-password:session): session closed for user mateus Jul 06 10:13:28 mateus-arch systemd-coredump[8371]: Process 8179 (chrome) of user 1000 dumped core. Jul 06 10:13:29 mateus-arch systemd-coredump[8372]: Process 8185 (gnome-shell) of user 1000 dumped core. This coredump might be related: mateus ~ $ coredumpctl gdb PID: 8185 (gnome-shell) UID: 1000 (mateus) GID: 1000 (mateus) Signal: 5 (TRAP) Timestamp: Seg 2015-07-06 10:13:27 BRT (3min 10s ago) Command Line: /usr/bin/gnome-shell Executable: /usr/bin/gnome-shell Control Group: /user.slice/user-1000.slice/session-c7.scope Unit: session-c7.scope Slice: user-1000.slice Session: c7 Owner UID: 1000 (mateus) Boot ID: 52a606ca7ea047fc8c9ca3f3b9edb4bf Machine ID: 6f73af040ad64d90ae2bddb5baf67eef Hostname: mateus-arch Coredump: /var/lib/systemd/coredump/core.gnome-shell.1000.52a606ca7ea047fc8c9ca3f3b9edb4bf.8185.1436188407000000.lz4 Message: Process 8185 (gnome-shell) of user 1000 dumped core. GNU gdb (GDB) 7.9.1 Copyright (C) 2015 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-unknown-linux-gnu". Type "show configuration" for configuration details. For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>. Find the GDB manual and other documentation resources online at: <http://www.gnu.org/software/gdb/documentation/>. For help, type "help". Type "apropos word" to search for commands related to "word"... Reading symbols from /usr/bin/gnome-shell...(no debugging symbols found)...done. [New LWP 8185] [New LWP 8368] warning: Could not load shared library symbols for linux-vdso.so.1. Do you need "set solib-search-path" or "set sysroot"? [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib/libthread_db.so.1". Core was generated by `/usr/bin/gnome-shell'. Program terminated with signal SIGTRAP, Trace/breakpoint trap.
+ Trace 235234
Same issue here, also on Arch. Here is the relevant output from journald: Jul 09 21:07:30 zoidberg gnome-session[915]: WARNING: Lost name on bus: org.gnome.SessionManager Jul 09 21:07:30 zoidberg gnome-session[915]: WARNING: App 'gnome-settings-daemon.desktop' exited with code 1 Jul 09 21:07:30 zoidberg gnome-session[2471]: (gnome-settings-daemon:2546): color-plugin-WARNING **: failed to connect to device: Failed to connect to missing device /org/freedesktop/ColorManager/devices/xrandr_LVDS1_phw_1000 Jul 09 21:07:30 zoidberg polkitd[565]: Unregistered Authentication Agent for unix-session:c2 (system bus name :1.80, object path /org/freedesktop/PolicyKit1/AuthenticationAgent, locale de_DE.utf8) (disconnected from bus) Jul 09 21:07:30 zoidberg gnome-session[915]: WARNING: App 'gnome-shell.desktop' exited with code 1 Jul 09 21:07:30 zoidberg atom.desktop[1681]: /usr/bin/atom: Zeile 99: 1696 Beendet nohup "$ATOM_PATH" --executed-from="$(pwd)" --pid=$$ "$@" > "$ATOM_HOME/nohup.out" 2>&1 Jul 09 21:07:30 zoidberg firefox.desktop[1703]: (firefox:1703): GConf-WARNING **: Got Disconnected from DBus. Jul 09 21:07:30 zoidberg firefox.desktop[1703]: firefox: Fatal IO error 104 (Die Verbindung wurde vom Kommunikationspartner zurückgesetzt) on X server :0. Jul 09 21:07:30 zoidberg atom.desktop[1681]: [1696:0709/202553:ERROR:browser_main_loop.cc(170)] Running without the SUID sandbox! See https://code.google.com/p/chromium/wiki/LinuxSUIDSandboxDevelopment for more information on developing Jul 09 21:07:30 zoidberg atom.desktop[1681]: ATTENTION: default value of option force_s3tc_enable overridden by environment. Jul 09 21:07:30 zoidberg atom.desktop[1681]: App load time: 5155ms Jul 09 21:07:30 zoidberg atom.desktop[1681]: [1765:0709/202558:INFO:renderer_main.cc(212)] Renderer process started Jul 09 21:07:30 zoidberg atom.desktop[1681]: [1696:0709/202743:INFO:CONSOLE(56)] "Window load time: 95628ms", source: file:///usr/share/atom/resources/app.asar/static/index.js (56) Jul 09 21:07:30 zoidberg atom.desktop[1681]: g_dbus_connection_real_closed: Remote peer vanished with error: Underlying GIOStream returned 0 bytes on an async read (g-io-error-quark, 0). Exiting. Jul 09 21:07:32 zoidberg systemd-coredump[6993]: Process 6987 (gnome-shell) of user 1000 dumped core. Jul 09 21:07:32 zoidberg kernel: traps: gnome-shell[6987] trap int3 ip:7f713b448ac0 sp:7fff37dacca0 error:0 Jul 09 21:07:32 zoidberg gdm-password][815]: pam_unix(gdm-password:session): session closed for user phw Jul 09 21:07:32 zoidberg systemd-logind[523]: Removed session c2. Jul 09 21:07:32 zoidberg systemd[1]: Stopping User Manager for UID 1000... Jul 09 21:07:32 zoidberg systemd[895]: Stopped target Default. Jul 09 21:07:32 zoidberg systemd[895]: Stopping Default. Jul 09 21:07:32 zoidberg systemd[895]: Reached target Shutdown. Jul 09 21:07:32 zoidberg systemd[895]: Starting Shutdown. Jul 09 21:07:32 zoidberg systemd[895]: Starting Exit the Session... Jul 09 21:07:32 zoidberg systemd[895]: Stopped target Basic System. Jul 09 21:07:32 zoidberg systemd[895]: Stopping Basic System. Jul 09 21:07:32 zoidberg systemd[895]: Stopped target Timers. Jul 09 21:07:32 zoidberg systemd[895]: Stopping Timers. Jul 09 21:07:32 zoidberg systemd[895]: Stopped target Paths. Jul 09 21:07:32 zoidberg systemd[895]: Stopping Paths. Jul 09 21:07:32 zoidberg systemd[895]: Stopped target Sockets. Jul 09 21:07:32 zoidberg systemd[895]: Stopping Sockets. Jul 09 21:07:32 zoidberg systemd[895]: Received SIGRTMIN+24 from PID 7004 (kill). Jul 09 21:07:32 zoidberg systemd[897]: pam_unix(systemd-user:session): session closed for user phw Jul 09 21:07:32 zoidberg systemd[1]: Stopped User Manager for UID 1000. Jul 09 21:07:32 zoidberg systemd[1]: Removed slice user-1000.slice. Jul 09 21:07:32 zoidberg systemd[1]: Stopping user-1000.slice. Jul 09 21:07:38 zoidberg kernel: ACPI Warning: \_SB_.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20150204/nsarguments-95) Jul 09 21:07:38 zoidberg kernel: ACPI: \_SB_.PCI0.PEG0.PEGP: failed to evaluate _DSM Jul 09 21:07:38 zoidberg kernel: ACPI Warning: \_SB_.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20150204/nsarguments-95) A lot of things happening there should be only symptoms of the underlying crash. No idea what causes the "Lost name on bus: org.gnome.SessionManager" coredumpctl info 6987 (1) PID: 6987 (gnome-shell) UID: 1000 (phw) GID: 100 (users) Signal: 5 (TRAP) Timestamp: Do 2015-07-09 21:07:31 CEST (12h ago) Command Line: /usr/bin/gnome-shell Executable: /usr/bin/gnome-shell Control Group: /user.slice/user-1000.slice/session-c2.scope Unit: session-c2.scope Slice: user-1000.slice Session: c2 Owner UID: 1000 (phw) Boot ID: d128e5a1f4ff4e3c81b9cfddff6d01c8 Machine ID: fdd6b79f14ce4f32811fb44123beb637 Hostname: zoidberg Coredump: /var/lib/systemd/coredump/core.gnome-shell.1000.d128e5a1f4ff4e3c81b9cfddff6d01c8.6987.143646885100000 Message: Process 6987 (gnome-shell) of user 1000 dumped core.
Aren't there some errors before this log output? It seems like the session is just going down. Could anyone please ensure that they have debug versions installed? Previous comment indicates a clutter problem, but need way more info before a developer can do anything with it.
For me there is not much before that output. There is one line 6 minutes earlier, but I think that's unrelated, and else its nearly half an hour earlier: Jul 09 20:28:31 zoidberg gnome-shell[2369]: GNOME Shell started at Thu Jul 09 2015 20:28:23 GMT+0200 (CEST) Jul 09 20:28:34 zoidberg kernel: ACPI Warning: \_SB_.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], Jul 09 20:28:34 zoidberg kernel: ACPI: \_SB_.PCI0.PEG0.PEGP: failed to evaluate _DSM Jul 09 20:28:34 zoidberg kernel: ACPI Warning: \_SB_.PCI0.PEG0.PEGP._DSM: Argument #4 type mismatch - Found [Buffer], Jul 09 20:30:17 zoidberg kernel: EXT4-fs (dm-0): error count since last fsck: 9 Jul 09 20:30:17 zoidberg kernel: EXT4-fs (dm-0): initial error at time 1414670750: mb_free_blocks:1450: block 1081856 Jul 09 20:30:17 zoidberg kernel: EXT4-fs (dm-0): last error at time 1435817574: ext4_mb_generate_buddy:757 Jul 09 20:39:43 zoidberg systemd[1]: Starting Cleanup of Temporary Directories... Jul 09 20:39:45 zoidberg systemd[1]: Started Cleanup of Temporary Directories. Jul 09 21:01:01 zoidberg CROND[6240]: (root) CMD (run-parts /etc/cron.hourly) Jul 09 21:07:30 zoidberg gnome-session[915]: WARNING: Lost name on bus: org.gnome.SessionManager Jul 09 21:07:30 zoidberg gnome-session[915]: WARNING: App 'gnome-settings-daemon.desktop' exited with code 1 Jul 09 21:07:30 zoidberg gnome-session[2471]: (gnome-settings-daemon:2546): color-plugin-WARNING **: failed to connec Jul 09 21:07:30 zoidberg polkitd[565]: Unregistered Authentication Agent for unix-session:c2 (system bus name :1.80, Jul 09 21:07:30 zoidberg gnome-session[915]: WARNING: App 'gnome-shell.desktop' exited with code 1 ... Rest see above.
Basically gnome-shell crashes somewhere in clutter. We need debug symbols installed and a good stack trace. Which version of clutter does Arch use? Thanks for taking the time to report this. Without a stack trace from the crash it's very hard to determine what caused it. Can you get us a stack trace? Please see https://wiki.gnome.org/Community/GettingInTouch/Bugzilla/GettingTraces for more information on how to do so. When pasting a stack trace in this bug report, please reset the status of this bug report from NEEDINFO to its previous status. Thanks in advance!
Same crash here. 1.22.4 The issue is that Arch Linux strips debug symbols. It happens a few times a day when I open a new application.
Same issue on Ubuntu https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1425782 gnome-shell is crashing in meta_clutter_init.
Possibly the same bug too https://bugzilla.redhat.com/show_bug.cgi?id=1178304
None of the linked bugs have a decent stack trace, so until I can seen one there's not much I can do.
(In reply to Emmanuele Bassi (:ebassi) from comment #13) > None of the linked bugs have a decent stack trace, so until I can seen one > there's not much I can do. In which package(s)? Will clutter be enough?
Could you say which packages need to be rebuilt with debugging symbols? I currently plan on rebuilding gtk3, gvfs, glib2, clutter and mutter, are those enough?
You will need to rebuild: - Glib (gobject, gio) - Cogl - Clutter - Mutter - GNOME-Shell It would be great to have a Mesa build with debug symbols as well, if you're using an open source driver.
(In reply to Emmanuele Bassi (:ebassi) from comment #16) > You will need to rebuild: > > - Glib (gobject, gio) > - Cogl > - Clutter > - Mutter > - GNOME-Shell > > It would be great to have a Mesa build with debug symbols as well, if you're > using an open source driver. Okay, I rebuilt all these packages, including Mesa. I thought I should check the backtrace of the latest coredump again to see if anything changed. Here it is: mateus ~ $ coredumpctl gdb 29510 PID: 29510 (gnome-shell) UID: 1000 (mateus) GID: 1000 (mateus) Signal: 5 (TRAP) Timestamp: Sex 2015-07-17 23:00:41 BRT (15h ago) Command Line: /usr/bin/gnome-shell Executable: /usr/bin/gnome-shell Control Group: /user.slice/user-1000.slice/session-c4.scope Unit: session-c4.scope Slice: user-1000.slice Session: c4 Owner UID: 1000 (mateus) Boot ID: 183d4a4d1e7b49b5a6b1fd01504fe221 Machine ID: 6f73af040ad64d90ae2bddb5baf67eef Hostname: mateus-arch Coredump: /var/lib/systemd/coredump/core.gnome-shell.1000.183d4a4d1e7b49b5a6b1fd01504fe221.29510.1437184841000000.lz4 Message: Process 29510 (gnome-shell) of user 1000 dumped core. GNU gdb (GDB) 7.9.1 Copyright (C) 2015 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-unknown-linux-gnu". Type "show configuration" for configuration details. For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>. Find the GDB manual and other documentation resources online at: <http://www.gnu.org/software/gdb/documentation/>. For help, type "help". Type "apropos word" to search for commands related to "word"... Reading symbols from /usr/bin/gnome-shell...done. [New LWP 29510] [New LWP 29518] warning: .dynamic section for "/usr/lib/libcogl.so.20" is not at the expected address (wrong library or version mismatch?) warning: .dynamic section for "/usr/lib/libcogl-path.so.20" is not at the expected address (wrong library or version mismatch?) warning: Could not load shared library symbols for linux-vdso.so.1. Do you need "set solib-search-path" or "set sysroot"? [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib/libthread_db.so.1". Core was generated by `/usr/bin/gnome-shell'. Program terminated with signal SIGTRAP, Trace/breakpoint trap.
+ Trace 235283
Is that any useful? Should I wait for a new coredump?
Probably useful. It mentions the line number https://github.com/GNOME/mutter/blob/master/src/backends/meta-backend.c#L629
(In reply to Jasper St. Pierre from comment #3) > I have to know what the logs and assertion messages are here. I think I should file a different bug report for it, anyway I got this: mateus ~ $ sudo coredumpctl gdb PID: 29668 (gnome-session-f) UID: 120 (gdm) GID: 120 (gdm) Signal: 6 (ABRT) Timestamp: Sáb 2015-07-18 18:22:33 BRT (1min 48s ago) Command Line: /usr/lib/gnome-session/gnome-session-failed --debug --allow-logout Executable: /usr/lib/gnome-session/gnome-session-failed Control Group: /user.slice/user-120.slice/session-c1.scope Unit: session-c1.scope Slice: user-120.slice Session: c1 Owner UID: 120 (gdm) Boot ID: 8ed0a27a57a548afaaf8d41783a6e0b1 Machine ID: 6f73af040ad64d90ae2bddb5baf67eef Hostname: mateus-arch Coredump: /var/lib/systemd/coredump/core.gnome-session-f.120.8ed0a27a57a548afaaf8d41783a6e0b1.29668.1437254553000000.lz4 Message: Process 29668 (gnome-session-f) of user 120 dumped core. GNU gdb (GDB) 7.9.1 Copyright (C) 2015 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-unknown-linux-gnu". Type "show configuration" for configuration details. For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>. Find the GDB manual and other documentation resources online at: <http://www.gnu.org/software/gdb/documentation/>. For help, type "help". Type "apropos word" to search for commands related to "word"... Reading symbols from /usr/lib/gnome-session/gnome-session-failed...(no debugging symbols found)...done. warning: core file may not match specified executable file. [New LWP 29668] [New LWP 29669] warning: Could not load shared library symbols for linux-vdso.so.1. Do you need "set solib-search-path" or "set sysroot"? [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib/libthread_db.so.1". Core was generated by `/usr/lib/gnome-session/gnome-session-failed --debug --allow-logout'. Program terminated with signal SIGABRT, Aborted.
+ Trace 235284
(gdb) quit
Just got this coredump a few minutes ago, the backtrace seems to be the same as comment #17 though. mateus ~ $ coredumpctl gdb PID: 13173 (gnome-shell) UID: 1000 (mateus) GID: 1000 (mateus) Signal: 5 (TRAP) Timestamp: Dom 2015-07-19 20:30:35 BRT (1min 30s ago) Command Line: /usr/bin/gnome-shell Executable: /usr/bin/gnome-shell Control Group: /user.slice/user-1000.slice/session-c3.scope Unit: session-c3.scope Slice: user-1000.slice Session: c3 Owner UID: 1000 (mateus) Boot ID: 8ed0a27a57a548afaaf8d41783a6e0b1 Machine ID: 6f73af040ad64d90ae2bddb5baf67eef Hostname: mateus-arch Coredump: /var/lib/systemd/coredump/core.gnome-shell.1000.8ed0a27a57a548afaaf8d41783a6e0b1.13173.1437348635000000.lz4 Message: Process 13173 (gnome-shell) of user 1000 dumped core. GNU gdb (GDB) 7.9.1 Copyright (C) 2015 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-unknown-linux-gnu". Type "show configuration" for configuration details. For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>. Find the GDB manual and other documentation resources online at: <http://www.gnu.org/software/gdb/documentation/>. For help, type "help". Type "apropos word" to search for commands related to "word"... Reading symbols from /usr/bin/gnome-shell...done. [New LWP 13173] [New LWP 13207] warning: Could not load shared library symbols for linux-vdso.so.1. Do you need "set solib-search-path" or "set sysroot"? [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib/libthread_db.so.1". Core was generated by `/usr/bin/gnome-shell'. Program terminated with signal SIGTRAP, Trace/breakpoint trap.
+ Trace 235286
I added CLUTTER_BACKEND=x11 to /etc/environment and restarted gdm then my user session. No crashes so far today but that may just be luck.
Husaam: any update on whether it works around the bug for you? But more generally: why would a gnome-shell crash cause the whole session to go down? I experience similar forced log outs (Bug 752722), and I'm quite confused about these symptoms...
(Oops, sorry, I typed too fast, please s/Husaam/Hussam.)
(In reply to Milan Bouchet-Valat from comment #22) > Husaam: any update on whether it works around the bug for you? > > But more generally: why would a gnome-shell crash cause the whole session to > go down? I experience similar forced log outs (Bug 752722), and I'm quite > confused about these symptoms... I updated to cogl 1.21.2 / clutter 1.23.6 (mutter and everything are still at gnome 3.16) and added CLUTTER_BACKEND=x11 to /etc/environment. No crashes since. This is the only change I did on gnome's side. On systend's side, I stopped doing 'systemctl daemon-reexec' after updating some service without logging off and exiting gdm first. No sudden session exists since then either. I have a feeling that was causing some my sudden gdm/user session exits. I know for sure gdm session was restarting too. The main thing 'systemctl daemon-reexec' does is reexec main systemd process and reload alsa state. Perhaps pulseaudio or Xorg did not like that?
Well, I tried to analyze coredump where I got this stacktrace (I'm running ArchLinux so no debug symbols but I think there are useless here if one can read a bit of assembly): $ bt
+ Trace 235398
Interesting stuff appears here: 0x00007f004ba7a86a <+42>: call 0x7f004ba76700 <clutter_init@plt> 0x00007f004ba7a86f <+47>: cmp eax,0x1 0x00007f004ba7a872 <+50>: je 0x7f004ba7a890 <meta_clutter_init+80> 0x00007f004ba7a874 <+52>: lea rdx,[rip+0x748e5] # 0x7f004baef160 0x00007f004ba7a87b <+59>: lea rdi,[rip+0x7c74e] # 0x7f004baf6fd0 0x00007f004ba7a882 <+66>: mov esi,0x4 0x00007f004ba7a887 <+71>: xor eax,eax 0x00007f004ba7a889 <+73>: call 0x7f004ba73090 <g_log@plt> Which corresponds to source meta_create_backend (); if (clutter_init (NULL, NULL) != CLUTTER_INIT_SUCCESS) g_error ("Unable to initialize Clutter.\n"); in backends/meta-backend.c. It is supposed to log "Unable to initialize Clutter.", string I get in $rdx, but for some reason nothing related to Clutter appears in my systemd logs. I think I missed some logging/verbosity options (because clutter_init is supposed to log error messages when everything goes wrong, too), and when I'll know them I should be able to investigate further. I got this with gnome-shell 3.16.3, clutter 1.22.4 and mutter 3.16.3.
I forgot to mention it (sorry), this code happens in meta_clutter_init
I am experiencing the same bug on my Arch Linux computer. The gnome display manager is started as a systemd service. The display manager sometimes starts up properly and sometimes displays the "Oh no! Something has went wrong." screen. It seems unpredictable whether or not I will get an error the next time I will launch gdm. I am able to reproduce by running # systemctl restart gdm several times until the problem occurs. I have rebuild glib, cogl, clutter, mutter and gnome-shell with debug symbols enabled. I use the proprietary nvidia driver. Here is the information about the last gnome-shell core dump: $ coredumpctl info 1445 PID: 1445 (gnome-shell) UID: 120 (gdm) GID: 120 (gdm) Signal: 5 (TRAP) Timestamp: lun. 2015-06-15 08:16:03 CEST (3 months 18 days ago) Command Line: gnome-shell --mode=gdm --wayland --display-server Executable: /usr/bin/gnome-shell Control Group: /user.slice/user-120.slice/session-c1.scope Unit: session-c1.scope Slice: user-120.slice Session: c1 Owner UID: 120 (gdm) Boot ID: 49d5c11f1e674ea8a931f9de4842e67d Machine ID: 31df6e902f11496684ce14a75a7c7afa Hostname: pc-guillaume Message: Process 1445 (gnome-shell) of user 120 dumped core. PID: 1445 (gnome-shell) UID: 120 (gdm) GID: 120 (gdm) Signal: 5 (TRAP) Timestamp: sam. 2015-10-03 13:51:52 CEST (30min ago) Command Line: gnome-shell --mode=gdm --wayland --display-server Executable: /usr/bin/gnome-shell Control Group: /user.slice/user-120.slice/session-c7.scope Unit: session-c7.scope Slice: user-120.slice Session: c7 Owner UID: 120 (gdm) Boot ID: 2ed85b2dbff8477a82bc96896a283db0 Machine ID: 31df6e902f11496684ce14a75a7c7afa Hostname: pc-guillaume Coredump: /var/lib/systemd/coredump/core.gnome-shell.120.2ed85b2dbff8477a82bc96896a283db0.1445.1443873112000000.lz4 Message: Process 1445 (gnome-shell) of user 120 dumped core. and the backtrace:
+ Trace 235543
Here are the relevant parts of my journal: oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: DEBUG(+): GsmManager: starting phase DISPLAY_SERVER oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: DEBUG(+): GsmManager: starting app '/org/gnome/SessionManager/App1' oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: DEBUG(+): Starting app: /org/gnome/SessionManager/App1 oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: DEBUG(+): GsmAutostartApp: starting gnome-shell-wayland.desktop: command=gnome-shell --mode=gdm --wayland --display-server startup-id=10c90dd4e2215bdc54144387311270096600000014380000 oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: DEBUG(+): GsmAutostartApp: started pid:1445 oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: (gnome-shell:1445): Clutter-CRITICAL **: Unable to initialize Clutter: Failed to connected to any renderer: oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: Couldn't initialize EGL oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: (gnome-shell:1445): mutter-ERROR **: Unable to initialize Clutter. oct. 03 13:51:52 pc-guillaume kernel: traps: gnome-shell[1445] trap int3 ip:7f3461d82b5b sp:7ffcfe55d9d0 error:0 oct. 03 13:51:52 pc-guillaume gnome-session[1438]: DEBUG(+): GsmAutostartApp: (pid:1445) done (signal:5) oct. 03 13:51:52 pc-guillaume gnome-session[1438]: WARNING: Application 'gnome-shell-wayland.desktop' killed by signal 5 oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: DEBUG(+): GsmAutostartApp: (pid:1445) done (signal:5) oct. 03 13:51:52 pc-guillaume /usr/lib/gdm/gdm-wayland-session[1435]: gnome-session[1438]: WARNING: Application 'gnome-shell-wayland.desktop' killed by signal 5 oct. 03 13:51:52 pc-guillaume gnome-session[1438]: Unrecoverable failure in required component gnome-shell-wayland.desktop and later: oct. 03 13:51:53 pc-guillaume systemd-coredump[1459]: Process 1445 (gnome-shell) of user 120 dumped core.
Try rebuilding gdm (In reply to Guillaume Bonnet from comment #27) > I am experiencing the same bug on my Arch Linux computer. try rebuilding gdm with --disable-wayland-support if you are running the nvidia proprietary xorg driver.
With --disable-wayland-support, gdm starts properly after reboot, but fails with its "Oh no! Something has went wrong." error screen whenever I try to restart it with # systemctl restart gdm However, I think it is a different error because there is no core dump. The following part of my journal is maybe relevant, but probably unrelated to the clutter bug: oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (II) UnloadModule: "libinput" oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (II) systemd-logind: releasing fd for 13:68 oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (II) NVIDIA: Freed GPU:0 (GPU-b4017b8d-2568-3fd6-7514-3a99e0bb04aa) @ oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (II) NVIDIA: PCI:0000:01:00.0 oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (II) NVIDIA(GPU-0): Deleting GPU-0 oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (WW) xf86CloseConsole: KDSETMODE failed: Input/output error oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (WW) xf86CloseConsole: VT_GETMODE failed: Input/output error oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: Fatal server error: oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) xf86CloseConsole: VT_ACTIVATE failed: Input/output error oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: Please consult the The X.Org Foundation support oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: at http://wiki.x.org oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: for help. oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information. oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (WW) xf86CloseConsole: KDSETMODE failed: Input/output error oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (WW) xf86CloseConsole: VT_GETMODE failed: Input/output error oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: FatalError re-entered, aborting oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE) xf86CloseConsole: VT_ACTIVATE failed: Input/output error oct. 03 15:03:05 pc-guillaume /usr/lib/gdm/gdm-x-session[952]: (EE)
maybe remove xwayland as well and even rebuild mutter without wayland support?
Here is what I discoverd before comment 30 being posted: gdm compiled with --enable-wayland-support and WaylandEnable=false in the [daemon] section of /etc/gdm/custom.conf seems, logically enough, to produce similar effects than gdm with --disable-wayland-support. It may be a temporary workaround for people using an X11-only driver. It would be interesting to know if this bug affects people using wayland-compatible drivers too.
The bug from comment 29 seems already reported in gdm: https://bugzilla.gnome.org/show_bug.cgi?id=751308
GNOME is going to shut down bugzilla.gnome.org in favor of gitlab.gnome.org. As part of that, we are mass-closing older open tickets in bugzilla.gnome.org which have not seen updates for a longer time (resources are unfortunately quite limited so not every ticket can get handled). If you can still reproduce the situation described in this ticket in a recent and supported software version of clutter, then please follow https://wiki.gnome.org/GettingInTouch/BugReportingGuidelines and create a ticket at https://gitlab.gnome.org/GNOME/clutter/-/issues/ Thank you for your understanding and your help.