[Bug 939594] New: X11 fails to start after second boot
http://bugzilla.suse.com/show_bug.cgi?id=939594 Bug ID: 939594 Summary: X11 fails to start after second boot Classification: openSUSE Product: openSUSE Distribution Version: 42.1 Milestone 1 Hardware: Other OS: Other Status: NEW Severity: Normal Priority: P5 - None Component: X.Org Assignee: xorg-maintainer-bugs@forge.provo.novell.com Reporter: ohering@suse.com QA Contact: xorg-maintainer-bugs@forge.provo.novell.com Found By: --- Blocker: --- Created attachment 642078 --> http://bugzilla.suse.com/attachment.cgi?id=642078&action=edit journalctl-b.txt leap 42 milestone 1 on Dell OptiPlex 980. Right after installing the GNOME desktop came up fine. I was able to poke around in yast2, but I changed nothing except the repos. Then I did a shutdown of the system. A few hours later I powered it on again, and got just a text console on vt7. The attached journalctl -b output indicates some permissions problem on vt7?! Other logs are in bug #939544 -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c1
Stefan Dirsch
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c2
--- Comment #2 from Stefan Dirsch
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c3
--- Comment #3 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c4
--- Comment #4 from Stefan Dirsch
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c5
--- Comment #5 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c6
Egbert Eich
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c7
--- Comment #7 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c8
--- Comment #8 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c9
--- Comment #9 from Egbert Eich
(In reply to Egbert Eich from comment #6)
Hmmm ... and the old systemd-210 should know that gdm now does not work as root anymore?
No, of course not, and what we are seeing here is to be expected. Gnome is such that it relies on the latest and greatest components underneath. If one tries to set a current Gnome on top of two year old components one will run into such issues. There is no reason to be sarcastic. This bug is IHMO mis-assigned. But I'm still waiting for an answer on my question before I do a reassignment. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c10
Olaf Hering
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c11
--- Comment #11 from Olaf Hering
I assume that starting the Xserver as root (runlvl 3 -> log into text console as root -> '/usr/bin/Xorg -retro') works - Olaf?
olaf@probook:~ $ connect.sh root@optiplex Password: Last login: Mon Jul 27 20:23:43 2015 from probook.fritz.box Have a lot of fun... optiplex:~ # X X.Org X Server 1.17.2 Release Date: 2015-06-16 X Protocol Version 11, Revision 0 Build Operating System: openSUSE SUSE LINUX Current Operating System: Linux optiplex 4.1.2-5-desktop #1 SMP PREEMPT Tue Jul 14 13:20:23 UTC 2015 (5ccacb4) x86_64 Kernel command line: BOOT_IMAGE=/boot/vmlinuz-4.1.2-5-desktop root=UUID=32e14674-9f53-40f2-8dd3-8f7622ba3f72 panic=9 sysrq_always_enabled resume=/dev/disk/by-label/WD20_SWAP splash=silent quiet showopts Build Date: 23 July 2015 07:51:40AM Current version of pixman: 0.32.6 Before reporting problems, check http://wiki.x.org to make sure that you have the latest version. Markers: (--) probed, (**) from config file, (==) default setting, (++) from command line, (!!) notice, (II) informational, (WW) warning, (EE) error, (NI) not implemented, (??) unknown. (==) Log file: "/var/log/Xorg.0.log", Time: Tue Jul 28 12:17:02 2015 (==) Using config directory: "/etc/X11/xorg.conf.d" (==) Using system config directory "/usr/share/X11/xorg.conf.d" (II) [KMS] Kernel modesetting enabled. The XKEYBOARD keymap compiler (xkbcomp) reports:
Warning: Type "ONE_LEVEL" has 1 levels, but <RALT> has 2 symbols Ignoring extra symbols Errors from xkbcomp are not fatal to the X server
Yes, and appending -retro gives the wellknown gray pattern. Then (after a cold boot) I did three reboots via the GNOME menu, it always came back just fine. Now I logged in as root and run 'init 6 & exit 0' and got the permission problem. So there is either a race, or there are stale files, or its something else ... Olaf -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c12
Egbert Eich
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c13
--- Comment #13 from Egbert Eich
Olaf, can you please check under which user this is running when the login screen starts successfully?
I'm referring to the Xserver of course ;) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c14
--- Comment #14 from Egbert Eich
http://bugzilla.suse.com/show_bug.cgi?id=939594
Egbert Eich
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c16
--- Comment #16 from Olaf Hering
http://bugzilla.suse.com/show_bug.cgi?id=939594
Egbert Eich
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c17
Jeff Mahoney
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c18
Olaf Hering
http://bugzilla.suse.com/show_bug.cgi?id=939594
Olaf Hering
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c19
Dominique Leuenberger
Let's track that one as GNOME (update triggered) issue.
GNOME is just the messenger here that fails to get access to the display.. the issue is that X is not coming up properly - for not receiving access to vt7 (In reply to Stefan Dirsch from comment #4)
Honestly I don't know whch systemd version openSUSE:42/leap is currently using. Seems to be some 210 release.
Leap 42.1 comes with the SLE12 version of systemd. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c20
--- Comment #20 from Dominique Leuenberger
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c21
--- Comment #21 from Dr. Werner Fink
^^ should not really be a problem as GDM has code to ask plymouth to quit before it starts X
It seems to be a problem. AFAICR I've added code upstream to both sulogin as well as to agetty to wait upto the point where plymouthd does not locked the virtual console anymore. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c22
--- Comment #22 from Dominique Leuenberger
(In reply to Dominique Leuenberger from comment #20)
^^ should not really be a problem as GDM has code to ask plymouth to quit before it starts X
It seems to be a problem. AFAICR I've added code upstream to both sulogin as well as to agetty to wait upto the point where plymouthd does not locked the virtual console anymore.
This goes in very much the same direction as I was just investigating: GDM might well send the quit signal to plymouth, but plymouth might be too slow in acting upon this before gdm fires up X... thus resulting in the racing condition we see (after all, the problem does not appear at every boot... ) -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c23
--- Comment #23 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c26
--- Comment #26 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c28
Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c32
--- Comment #32 from Dr. Werner Fink
At this very moment, my efforts are actually showing that 'something' == 'systemd'. The current theory is that the service type forking (as used by display-manager.service) is causing this;
systemd seems to be calling chown_terminal in this case and this resetting the permissions.
`something' and `seems' isn't sufficient: I'm in doubt that systemd does this by its self. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c36
--- Comment #36 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c37
--- Comment #37 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c38
--- Comment #38 from Dominique Leuenberger
(In reply to Dr. Werner Fink from comment #36)
Ouch, that is really a brown paper back ;) Change the line
if rc_status ; then
into
if ! rc_status ; then
I'm not convinced :) the 'if rc_status is also the case where PIDFILE is being linked.. so I guess this needs to be changed to something like: startproc -p $PIDFILE $DISPLAYMANAGER $XDMOPTIONS || rc_failed if rc_status ; then # when started up successfuly, use DMs PID file as ours ln -snf $PIDFILE $PIDFILELNK else # After a crash or a kill signal we may have # a wrong ownership of /dev/xconsole if test -x $DMDIR/TakeDevices ; then $DMDIR/TakeDevices else chown root:tty /dev/xconsole /dev/tty0 chmod 622 /dev/xconsole /dev/tty0 fi fi In the original case, TakeDevice was done just before linking the PID file... one action is meant when it crashed, the other when it failed... so they can't really be meant to be in the same block, can they? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c39
Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c40
--- Comment #40 from Dominique Leuenberger
http://bugzilla.suse.com/show_bug.cgi?id=939594
Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c58
--- Comment #58 from Dr. Werner Fink
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c59
--- Comment #59 from Egbert Eich
http://bugzilla.suse.com/show_bug.cgi?id=939594
http://bugzilla.suse.com/show_bug.cgi?id=939594#c60
--- Comment #60 from Egbert Eich
Dominique, thanks :)
I wonder if we should update SLE12 as well. Currently we won't need it but maybe in the future. if we don't do it now we will run into this then.
Actually, no the other way around: the xdm script is not used on SLE12. -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com