Description of problem: For the 32 bit i686 version login just returns immediately to the login screen for the x64 version login hangs in a gray screen. Version-Release number of selected component (if applicable): gnome-session-wayland-session-3.14.0-1.fc21.x86_64 How reproducible: Every time Steps to Reproduce: 1.Select Wayland at the session login option 2. 3. Actual results: See description above. Expected results: Additional info: The setup for trying this was using Fedora 20 as the host computer and running libvirt VM clients. Some of the clients had been Fedora 20 clients upgraded using the fedup process and we also did a fresh Fedora 21 Beta install.
Hi, while I could reproduce the issue on a freshly installed F21 GA on KVM (that was due to XWayland failing to start), it works just fine after updating F21 to current. Do you still see this issue after updating F21?
I did try it on a system we did a fresh Fedora 21 install and ran into the same issue we saw with the virtual host. I will do more testing to answer your question. Do we need to start a service for XWayland?
Still does not work for us. We are running Fedora 21 fresh install fully patched non virtual 64 bit. Provide password and session get immediately terminated. Maybe this information is relevant from journalctl -rax: Feb 09 18:59:32 hemlock org.a11y.atspi.Registry[29634]: ** (at-spi2-registryd:30095): WARNING **: AT-SPI: Cannot open default display Feb 09 18:59:32 hemlock org.a11y.atspi.Registry[29634]: SpiRegistry daemon is running with well-known name - org.a11y.atspi.Registry Feb 09 18:59:32 hemlock org.a11y.Bus[29518]: Successfully activated service 'org.a11y.atspi.Registry' Feb 09 18:59:32 hemlock org.a11y.atspi.Registry[29634]: ** (at-spi2-registryd:30095): WARNING **: Could not open X display Feb 09 18:59:32 hemlock gnome-session[29509]: ** (gnome-settings-daemon:30087): WARNING **: Unable to initialize GTK+ Feb 09 18:59:32 hemlock org.a11y.Bus[29518]: Activating service name='org.a11y.atspi.Registry' Feb 09 18:59:32 hemlock gdm-session-worker[29427]: <3>GLib: Source ID 52 was not found when attempting to remove it Feb 09 18:59:32 hemlock gnome-session[29509]: ** (gnome-settings-daemon:30087): WARNING **: Could not open X display Feb 09 18:59:32 hemlock gdm-Xorg-:0[1472]: (II) intel(0): switch to mode 2560x1440 on DP1 using pipe 0, position (0, 0), rotation normal, reflection none Feb 09 18:59:32 hemlock gdm-Xorg-:0[1472]: (II) AIGLX: Resuming AIGLX clients af ter VT switch Feb 09 18:59:32 hemlock gdm-password][29427]: pam_unix(gdm-password:session): session closed for user dhighley Feb 09 18:59:32 hemlock gnome-session[29509]: gnome-session[29509]: WARNING: App 'gnome-settings-daemon.desktop' exited with code 1 Feb 09 18:59:32 hemlock gnome-session[29509]: WARNING: App 'gnome-settings-daemon.desktop' exited with code 1 Feb 09 18:59:32 hemlock gnome-session[29509]: Unrecoverable failure in required component gnome-shell-wayland.desktop Feb 09 18:59:32 hemlock gnome-session[29509]: WARNING: Application 'gnome-shell-wayland.desktop' killed by signal 11 Feb 09 18:59:32 hemlock gnome-session[29509]: gnome-session[29509]: WARNING: Application 'gnome-shell-wayland.desktop' killed by signal 11 Feb 09 18:59:32 hemlock gnome-session[29509]: (gnome-settings-daemon:29827): Gdk-WARNING **: gnome-settings-daemon: Fatal IO error 2 (No such file or directory) on X server :1. Feb 09 18:59:32 hemlock gnome-session[29509]: (gnome-settings-daemon:29827): power-plugin-WARNING **: Could not create GnomeRRScreen: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.gnome.Mutter.DisplayConfig was not provided by any .service files Feb 09 18:59:32 hemlock org.a11y.atspi.Registry[29634]: after 23 requests (23 kn own processed) with 0 events remaining. Feb 09 18:59:32 hemlock org.a11y.atspi.Registry[29634]: XIO: fatal IO error 11 (Resource temporarily unavailable) on X server ":1" Feb 09 18:59:32 hemlock org.gnome.Caribou.Daemon[29518]: (caribou:29711): Gdk-WARNING **: caribou: Fatal IO error 11 (Resource temporarily unavailable) on X server :1. Feb 09 18:59:32 hemlock gnome-session[29509]: (evolution-alarm-notify:29887): Gdk-WARNING **: evolution-alarm-notify: Fatal IO error 11 (Resource temporarily unavailable) on X server :1. Feb 09 18:59:32 hemlock bluetoothd[851]: Endpoint unregistered: sender=:1.290 path=/MediaEndpoint/A2DPSink Feb 09 18:59:32 hemlock bluetoothd[851]: Endpoint unregistered: sender=:1.290 path=/MediaEndpoint/A2DPSource Feb 09 18:59:32 hemlock gnome-session[29509]: (org.gnome.Software:29888): Gdk-WARNING **: org.gnome.Software: Fatal IO error 11 (Resource temporarily unavailable) on X server :1. Feb 09 18:59:32 hemlock gnome-session[29509]: seapplet: Fatal IO error 11 (Resource temporarily unavailable) on X server :1. Feb 09 18:59:32 hemlock gnome-session[29509]: seapplet: Fatal IO error 11 (Resource temporarily unavailable) on X server :1. Feb 09 18:59:32 hemlock gnome-session[29509]: (gnome-settings-daemon:29827): wacom-plugin-WARNING **: Failed to create GnomeRRScreen: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.gnome.Mutter.DisplayConfig was not provided by any .service files Feb 09 18:59:32 hemlock gnome-session[29509]: (gnome-settings-daemon:29827): common-plugin-WARNING **: Failed to construct RR screen: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.gnome.Mutter.DisplayConfig was not provided by any .service files Feb 09 18:59:32 hemlock gnome-session[29509]: (EE) Feb 09 18:59:32 hemlock gnome-session[29509]: (EE) failed to dispatch Wayland events: Connection reset by peer Feb 09 18:59:32 hemlock gnome-session[29509]: Fatal server error: Feb 09 18:59:32 hemlock gnome-session[29509]: (EE) Feb 09 18:59:32 hemlock polkitd[28504]: Unregistered Authentication Agent for unix-session:105 (system bus name :1.285, object path /org/freedesktop/PolicyKit1/AuthenticationAgent, locale en_US.UTF-8) (disconnected from bus) Feb 09 18:59:32 hemlock gnome-session[29509]: abrt-applet: Lock file '/var/tmp/abrt/ccpp-2015-02-09-18:59:31-29574.new/.lock' is locked by process 29844 Feb 09 18:59:32 hemlock gnome-session[1521]: (gnome-settings-daemon:1547): color-plugin-WARNING **: failed to set screen _ICC_PROFILE: Failed to open file '/home/dhighley/.local/share/icc/edid-0fbd8696ae19de3d58d5127aab3c9f45.icc': Permission denied Feb 09 18:59:31 hemlock dbus[885]: [system] Successfully activated service 'org.freedesktop.locale1' Feb 09 18:59:31 hemlock org.gnome.Caribou.Daemon[29518]: ** (caribou:29711): WARNING **: AT-SPI: Error in GetItems, sender=org.freedesktop.DBus, error=The name :1.4 was not provided by any .service files
(In reply to David Highley from comment #3) > Do we need to start a service for XWayland? No, it's started by mutter itself in meta_xwayland_start(): https://git.gnome.org/browse/mutter/tree/src/wayland/meta-xwayland.c#n425 (In reply to David Highley from comment #4) > Maybe this information is relevant from journalctl -rax: [...] The journal is reversed, so the most important bits is at the bottom, i.e.: gnome-session[29509]: (EE) gnome-session[29509]: (EE) failed to dispatch Wayland events: Connection reset by peer gnome-session[29509]: Fatal server error: gnome-session[29509]: (EE) So Xwayland was started but it could not connect to the Wayland server (mutter in this case) and bailed out. All the rest is cascading failures because of this (gnome-settings-daemon failing to start, all XIO errors, etc.) Unfortunately, the reason as of why Xwayland could not connect to mutter is not visible from the excerpt of the log available in comment 4, my take would be that mutter itself has vanished shortly after launching Xwayland or something else is coming into play (SElinux, access rights, etc.). Could you provide the log for an entire sessions and not just a portion of it? I am a bit surprised to see a permission denied on the ICC file located in the home directory, is that a network mounted home directory (e.g. NFS)? Is this user an existing user, was the home directory migrated from a previous install, or a newly created user on that particular machine? Does the same issue occurs for any user (including a newly created user) or just this one in particular?
Created attachment 990338 [details] Journal text from failed session login Attempted another Wayland login with selinux in permissive mode. Failed as before. Home directories are NFS auto mounted from another system. Checked for any new selinux avc's after the login failure and found none. Home directory does not have world permissions which is normal for home directories.
Created attachment 990343 [details] Another journal log Disabled Gnome tracker and changed to another user account. Still fails the login session. We think many of the whines about the ICC file and possibly others might have been due to the fact that the other user was logged into another system. Anyway we do not see those complaints in this journal log.
Ah there you go: Feb 10 18:46:13 hemlock kernel: gnome-shell[7043]: segfault at 0 ip 00007fcccb8db606 sp 00007fffed7d0510 error 4 in libmutter.so.0.0.0[7fcccb828000+103000] So, as suspected, gnome-shell/mutter have crashed (so at least it's neither Wayland nor Xwayland) This message just before the crash is most likely related: Feb 10 18:46:13 hemlock gnome-session[6977]: (gnome-shell:7043): Clutter-CRITICAL **: clutter_input_device_get_pointer_stage: assertion 'CLUTTER_IS_INPUT_DEVICE (device)' failed It's a "g_return_val_if_fail()" that fails, it returns NULL and the segfault is a NULL pointer dereference, so most likely the returned value from clutter_input_device_get_pointer_stage() is not tested against NULL.
A quick look at the code does not reveal an obvious issue with the clutter_input_device_get_pointer_stage() returned value (only place where it's used is appropriately tested against NULL). Could you please capture the backtrace of gnome-shell when it fails? I think abrt would notify of the crash, so the backtrace should be available there with abrt (you may need to install the relevant debuginfo packages though).
I ran abrt and it gave me the information that this issue is already reported under bug id 1159585. So if you agree I guess we link this one as a duplicate to the other report.
Assuming the backtrace is identical, then yes, it's a dupe of bug 1159585
*** This bug has been marked as a duplicate of bug 1159585 ***