Gnome-shell v. 3.23.91 - blackscreen - Zesty

ZZ is unusable here too: login loop, X does not start via tty neither.
I've checked .Xauthority and /tmp as usual in such a case, but does not resolve that issue.
Reconfiguring lightdm or switching to gdm/gdm3 also does not work.

What graphics driver are you using?

Can you find any log messages?

I did reboot and successfully test gnome-shell 3.23.91-0ubuntu1 before uploading.

On my system , it was using nvidia-378; i've purged it, then reboot with nouveau (nvidia-375 is buggy & not usable (libEGL missing symlink lp:1662860 )

Do you have any relevant extensions installed?

What is logged :

gnome-session-binary[8003]: WARNING: Could not parse desktop file gnome-settings-daemon.desktop or it references a not found TryExec binary
gnome-session[8003]: gnome-session-binary[8003]: CRITICAL: gsm_manager_set_phase: assertion 'GSM_IS_MANAGER (manager)' failed
gnome-session-binary[8003]: CRITICAL: gsm_manager_set_phase: assertion 'GSM_IS_MANAGER (manager)' failed

" extensions " ?
what could it be at this early booting process ? its quite a genuine ZZ 64 install.

I'm running all this 3.23.91 stuff (mutter+GS) and I can log in and use GS.
I still experience some sudden logouts (it will happen every time when I get no icons on the desktop at startup), but that was already the case in 3.23.90.

I run Nvidia 375.39 from Ubuntu official repository, but I mainly use Intel iGPU.

dino99, I didn't know GNOME Shell on gdm was usable with nvidia proprietary drivers.

Ubuntu GNOME badly need help triaging these bugs
https://launchpad.net/bugs/1559576
https://launchpad.net/bugs/1632322

There's also this new bug:
https://launchpad.net/bugs/1666664

I don't have NVIDIA graphics, just Intel.

GDM works with Nvidia drivers but *I* have to restart my laptop (simply logout/login gives me a login loop).
https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1559576/comments/88

From auth.log:
lightdm: pam_succeed_if(lightdm:auth): requirement "user ingroup nopasswdlogin" not met by user "oem"

@Jeremy

about lp:1559576 : this is a missed xserver-xorg-video-legacy installation
about lp:1632322 : with that case, its an incomplete wayland support https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1632322/comments/24

My PC has Asus Sabertooth Z97 Mk II Motherboard, with Intel Core i7 4790 Processor (Haswell).
I use the graphics driver directly from the newest mesa (17.0.0-1ubuntu1).
I do not have xserver-xorg-video-intel package installed.

The again my setup has not changed lately.
As I stated above, g-s 3.23.90 worked perfectly, and 3.23.91 not at all.

I do not have any g-s extensions installed on my 64-bit setup.

dino99, installing xserver-xorg-legacy doesn't fix LP: #1632322 ?
gdm uses Wayland by default although Ubuntu's gnome-shell does not.

Your gnome-settings-daemon.desktop error seems very odd to me. What session are you logging into?

The RequiredComponents part of /usr/share/gnome-sessions/gnome.session should not have "gnome-settings-daemon" listed any more in zesty's gnome-session 3.23.91-0ubuntu1 but it should have a long list of org.gnome.SettingsDaemon.* items.

I tried with Nouveau Nvidia driver. Same crash in background, but in foreground no logout (VT switch to #2 instead of #1, if I'm not wrong).

Anyway, here is a log report :

18:18:14 gsd-xrandr: gsd-xrandr: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-print-notif: gsd-print-notifications: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-power: gsd-power: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-media-keys: gsd-media-keys: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-keyboard: gsd-keyboard: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-color: gsd-color: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-clipboard: gsd-clipboard: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-wacom: gsd-wacom: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-xsettings: gsd-xsettings: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gsd-a11y-keyboa: gsd-a11y-keyboard: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 dbus-daemon: after 21 requests (21 known processed) with 0 events remaining.
18:18:14 dbus-daemon: XIO: fatal IO error 11 (Resource temporarily unavailable) on X server ":0"
18:18:14 gnome-shell: gnome-shell: Fatal IO error 11 (Ressource temporairement non disponible) on X server :0.

18:18:14 gnome-session-b: CRITICAL: We failed, but the fail whale is dead. Sorry....
18:18:14 gnome-session-b: gnome-session-binary[1249]: CRITICAL: We failed, but the fail whale is dead. Sorry....
18:18:14 gnome-session-b: Unrecoverable failure in required component org.gnome.SettingsDaemon.Sharing.desktop
18:18:14 gnome-session-b: WARNING: Application 'org.gnome.SettingsDaemon.Sharing.desktop' failed to register before timeout
18:18:14 gnome-session-b: gnome-session-binary[1249]: WARNING: Application 'org.gnome.SettingsDaemon.Sharing.desktop' failed to register before timeout
18:18:13 systemd-resolve: Processing query...
18:17:55 update-notifier: GtkDialog mapped without a transient parent. This is discouraged.
18:17:54 update-notifier: /var/lib/dpkg/lock:
18:17:54 pkexec: ft: Executing command [USER=root] [TTY=unknown] [CWD=/home/ft] [COMMAND=/usr/lib/update-notifier/package-system-locked]
18:17:54 pkexec: pam_unix(polkit-1:session): session opened for user root by (uid=1000)

(I was running the Intel GPU.)

i always login with gnome-shell

AS i can see, after reconfiguring gnome-session-daemon, "The RequiredComponents part of /usr/share/gnome-sessions/gnome.session does not have "gnome-settings-daemon" listed, but a lot of org.gnome.SettingsDaemon.* as you said

Could someone try filing their bug directly with GNOME?

https://wiki.ubuntu.com/Bugs/Upstream/GNOME

Because gnome-shell with mutter 3.23.91 works for me.

https://bugzilla.gnome.org/show_bug.cgi?id=779488

Thank you dino99 for the Gnome (bugzilla) bug report.
You can confirm the bug (gnome-shell or mutter or both - 3.23.91) with nvidia hardware.
I can confirm the same thing with Intel Haswell GPU hardware.
It would be interesting to know on which hardware Jeremy manages to work with v. 3.23.91 (both mutter and gnome-shell).
Gnome-shell 3.23.91 is still in proposed (zesty).

There is now a patched version of GS in proposed. At the moment, I run non-patched GS with Skylake Xeon mobile, Intel GPU (switchable Nvidia graphics, nouveau driver).
One of the patches is this one :
https://git.gnome.org/browse/gnome-shell/commit/?id=252dce1ec2aa0ead84d14379a624bdc0768bd286

I do have some crashes but after successful login. The crash has not the same effect with Nvidia proprietary and Nouveau. With Nvidia (running iGPU), it's a direct logout. With Nouveau (running iGPU), it gives a running terminal (VT1) and switches GS to VT2 (I found that when I tried to recover).

Anyway, the background bug is the same.

My system is using a nvidia nv117 (maxwell) card (gtx750)
With the latest updated gnome-shell 3.23.91-ubuntu2 :
- the boot process now slow down when hitting 'starting lxc container ...' then utmp, and stop there.
- hitting 'enter' show directly the text login (like tty do)
- startx seems working in the first step: see a flashing nvidia logo, but then get a blackscreen

I've purged *lxc* & libertine*, then ran autoremove and reboot : still get the same boot issue.
Get the same trouble with 'nouveau' or selecting 4.10 or 4.8 kernels.

If the actual upgrade ends with a critical booting process (even dont get lightdm login now), the problem has started when gnome-session have swicthed from 3.22 to 3.23 : the login process was apparently fine at first glance, but always return back to the login screen after 10 or 20 seconds (suggesting a kind of race), then the opened gnome-shell session was stable. Neither journalctl nor logs have logged something usefull about that sadly.

So that let me thinking about the graphic drivers needing some tweaks to support such hardware with the 3.24 gnome installation: maybe ubuntu-drivers-common , and also 'nouveau' and the nvidia-xxx ones.

Feedback:
the newest mutter (...ubuntu1.1) has been upgraded, and a cold boot made : problem not resolved (no login screen, blackscreen)

I get the bug with latest GS packages too. With Nouveau, each time I get this bug, I'm not disconnected but the session's TTY is increased ! I run in TTY #6 now...

I don't have any particular expertise with gnome-shell or drivers. You may want to file your own bugs with GNOME if you want to have a better chance of getting these issues addressed.

For those with NVIDIA issues, maybe try talking to whoever packages NVIDIA drivers for Ubuntu.

Well i have now tried the latest gnome-shell 3.23.91-0ubuntu2 and mutter 3.23.91-0ubuntu1.1.
Those did not help, still blackscreen.
I also found tty in tty6.
Downgraded to 3.23.90 and all is well again.
This bug definitely persists with Intel Haswell GPU hardware.

Francois, Replying to comment #20 gdm uses vt1 by default since Ubuntu 16.10 and the first user to log in will get the first available vt which should usually be vt2.

Jeremy,

I tested new mutter 3.23.91 alone and that is working fine.
So this bug is only about g-s 3.23.91.

About the vt numbering.
Yes gdm uses vt1 and the desktop is in vt2, console is in vt6.

Here is some more info revealed during the blackscreen and taken from the vt6 after loggin in there:

Command "journalctl"

org.gnome.Shell.desktop[744]: (EE)
org.gnome.Shell.desktop[744]: Fatal server error
org.gnome.Shell.desktop[744]: (EE)
gnome-session-binary[726]: Unrecoverable failure in requested component org.gnome.Shell.desktop
gnome-session-binary[726]: Warning: app ’org.gnome.Shell.desktop’ exited with code 1
gnome-session[726]: gnome-session-binary: Warning: app ’org.gnome.Shell.desktop’ exited with code 1
gdm-launch-environment[704]: Pam_unix (gdm-launch-environment: session) session closed for user gdm

The above is there 5 times. Five consecutive trials to launch gdm during the the same booting operation.
There are no other error messages.

This bug may have something to with gdm3 3.23.91 too.

Harry, the latest comment on https://bugzilla.gnome.org/779488 suggests the problem is from gdm 3.23.91. Do you want to try to install gdm3 3.23.4-0ubuntu1 as a workaround?

Jeremy,
Yes I did that test with the gdm3_3.23.4-0ubuntu1, but no avail.
Resulting exactly to the same error and blackscreen.

@Jeremy,

if gnome's dev seems blaming 'gdm', that point out an other unexpected mess : when lightdm is reconfigured to use 'lightdm' and/or lightdm is purged then reinstalled, why and how 'gdm' can disturb the X start ?

Seems to me that X/wayland/mir/... relationship is messy: if the user choose to login with lightdm, then none of the other DE should disturb the process.

dino99, please ask GNOME that.

Jeremy,

i'll do; but i was first thinking about the specific ubuntu packaging (ubuntu-driver-common ? , mir/wayland ?)

I don't think mir has anything to do with this.

For those with NVIDIA hardware, it's possible that something needs to be done better in the NVIDIA drivers so you're welcome to talk to the NVIDIA driver packager about that.

If you want to confirm that it's not distro-specific, you could try another distro. The Fedora development release often gets GNOME beta packages so that the easier non-Ubuntu I would recommend to test GNOME betas.

There's a new gdm3 release. I guess it probably won't help much for the issues reported here:

https://launchpad.net/ubuntu/+source/gdm3/3.23.91.1-0ubuntu1

Feedback after the gdm3 upgrade (and all other upgrades too):
- login with lightdm: still get the login loop after validation; but no black screen
- login with gdm3: X load quickly abort, and drop to a text prompt

bugzilla.gnome dev have made that comment:
" The GDM thing I was thinking of is bug 779338 and probably unrelated to this bug. "

dino99, thanks for reporting the bug to GNOME.

Could the other people experiencing this bug also report bugs upstream to GNOME? The developers don't know about this bug unless someone tells them. And they might not all be the same bug.

That's especially the case for those who experience a problem triggered by this update on a computer without NVIDIA hardware.

Harry, since you're the only one (beside maybe Francois ?) that has only Intel hardware, is gir1.2-geoclue-2.0 installed?

https://launchpad.net/ubuntu/+source/gnome-shell/3.23.91-0ubuntu3

The problem is resolved by adding gir1.2-geoclue-2.0

gnome-shell (3.23.91-0ubuntu3) zesty; urgency=medium

  * debian/control.in:
    - Depend on gir1.2-geoclue-2.0 (LP: #1671823)

 -- Jeremy Bicha <email address hidden> Fri, 10 Mar 2017 09:19:54 -0500

I can confirm this bug has been fixed now.
Marking it "fix released".

I am marking this a duplicate of bug 1671823 since the original reporter confirmed that the fix for that bug fixed his issue here.

If anyone else is still having problems with gnome-shell/gdm/mutter 3.23.91 after installing gnome-shell 3.23.91-0ubuntu3, please file a new bug.