No X after upgrade to 20.10 / Failed to open authorization file

[pivert]

Hi,

I just upgraded from Kubuntu 20.04 LTS to 20.10 with do-release-upgrade.
But after reboot I have the Kubuntu logo or black screen.

Xorg is shutting down after the: «Failed to open authorization file "/var/run/sddm/{22c5f5dc-25f3-4262-91e6-721ffae1c6d6}": No such file or directory» error.

The /var/run/sddm folder does not exist. I created it, but this did not help.

Surprisingly, sddm service is still running.

Any idea ?

Additional info:

root@pivert-Z390:~# grep '] (EE)' -B3 -A3 /var/log/Xorg.0.log
[ 80887.934] (--) NVIDIA(GPU-0): Philips PHL BDM4037U (DFP-3): Internal DisplayPort
[ 80887.934] (--) NVIDIA(GPU-0): Philips PHL BDM4037U (DFP-3): 2660.0 MHz maximum pixel clock
[ 80887.934] (--) NVIDIA(GPU-0):
[ 80887.934] (EE) Failed to open authorization file "/var/run/sddm/{22c5f5dc-25f3-4262-91e6-721ffae1c6d6}": No such file or directory
[ 80887.956] (II) event2 - Power Button: device removed
[ 80887.981] (II) event1 - Power Button: device removed
[ 80888.005] (II) event0 - Sleep Button: device removed



pivert@pivert-Z390:~$ inxi -b -c 0
System: Host: pivert-Z390 Kernel: 5.8.0-26-generic x86_64 bits: 64 Console: tty 0 Distro: Ubuntu 20.10 (Groovy Gorilla)
Machine: Type: Desktop System: Gigabyte product: Z390 I AORUS PRO WIFI v: N/A serial: <superuser/root required>
Mobo: Gigabyte model: Z390 I AORUS PRO WIFI-CF serial: <superuser/root required> UEFI: American Megatrends v: F2
date: 10/11/2018
CPU: Info: 6-Core Intel Core i5-9600K [MCP] speed: 800 MHz min/max: 800/4600 MHz
Graphics: Device-1: NVIDIA TU104 [GeForce RTX 2080 SUPER] driver: N/A
Display: server: X.org 1.20.9 driver: nvidia tty: 139x22
Message: Advanced graphics data unavailable in console. Try -G --display
Network: Device-1: Intel Wireless-AC 9560 [Jefferson Peak] driver: iwlwifi
Device-2: Intel Ethernet I219-V driver: e1000e
Drives: Local Storage: total: 1.36 TiB used: 976.57 GiB (69.9%)
Info: Processes: 218 Uptime: 35m Memory: 15.57 GiB used: 496.6 MiB (3.1%) Init: systemd runlevel: 5 Shell: Bash
inxi: 3.1.07

[claydoh]

Were you using any video driver PPAs for the Nvidia card before the upgrade? Not sure if related, but it well could be, as the PPA's maintainer(s) recommend reverting back to stock before upgrading to a new *buntu release, because of file conflicts, and because Nvidia is a pita .

One thing to try is to reconfigure SDDM:
sudo dpkg-reconfigure sddm

Another is to try starting plasma without the login manager, which will give much better error messaging if there is a failure with drivers as opposed to SDDM

ctrl-alt-f2 to get to a tty, log in, and then:

startx

And record errors, if any.

if you get to the desktop, then the issue is likely with sddm itself, and my first suggestion may fix it.

[pivert]

Hi Claydoh,

I did try to reconfigure sddm package already, but this did not help.

I do not have related PPA in my config:

root@pivert-Z390:/etc/apt# grep -ri '^[^#].*ppa' /etc/apt/
/etc/apt/sources.list.d/kdenlive-ubuntu-kdenlive-stable-focal.list.distUpgrade:deb http://ppa.launchpad.net/kdenlive/kdenlive-stable/ubuntu focal main
root@pivert-Z390:/etc/apt#


Here are the installed drivers:

root@pivert-Z390:/etc/apt# apt search nvidia | grep installed

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

gimp-normalmap/groovy,now 1.2.3-0ubuntu6 amd64 [installed]
libnvidia-compute-450/groovy,now 450.80.02-0ubuntu1 amd64 [installed,automatic]
libvdpau1/groovy,now 1.4-2 amd64 [installed,automatic]
libxnvctrl0/groovy,now 440.82-0ubuntu1 amd64 [installed,automatic]
numba-doc/groovy,groovy,now 0.50.1-2 all [installed,automatic]
nvidia-utils-440/groovy,now 450.80.02-0ubuntu1 amd64 [installed]
nvidia-utils-450/groovy,now 450.80.02-0ubuntu1 amd64 [installed,automatic]
python3-numba/groovy,now 0.50.1-2 amd64 [installed,automatic]
ubuntu-drivers-common/groovy,now 1:0.8.6.1 amd64 [installed,automatic]
vdpau-driver-all/groovy,now 1.4-2 amd64 [installed,automatic]
vdpauinfo/groovy,now 1.4-1 amd64 [installed]
xserver-xorg-video-nouveau/groovy,now 1:1.0.16-1 amd64 [installed,automatic]
root@pivert-Z390:/etc/apt#


However, the startx did work. Thanks I forgot that this command still exists... It seems that it switches to fbdev, after an error loading nvidia driver, strange as nvidia driver seems to load properly when booting the computer:

root@pivert-Z390:/etc/apt# grep '] (EE)' -B2 -A1 /var/log/Xorg.0.log
[ 7678.285] (II) LoadModule: "ramdac"
[ 7678.285] (II) Module "ramdac" already built-in
[ 7678.294] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[ 7678.294] (EE) NVIDIA: system's kernel log for additional error messages and
[ 7678.294] (EE) NVIDIA: consult the NVIDIA README for details.
[ 7678.294] (EE) No devices detected.
[ 7678.430] (==) Matched nouveau as autoconfigured driver 0
--
[ 7678.440] (II) FBDEV: driver for framebuffer: fbdev
[ 7678.440] (II) VESA: driver for VESA chipsets: vesa
[ 7678.573] (EE) [drm] Failed to open DRM device for pci:0000:01:00.0: -19
[ 7678.573] (EE) open /dev/dri/card0: No such file or directory
[ 7678.573] (WW) Falling back to old probe method for modesetting
[ 7678.573] (EE) open /dev/dri/card0: No such file or directory
[ 7678.573] (II) Loading sub module "fbdevhw"
--
[ 7678.574] (**) FBDEV(1): claimed PCI slot 1@0:0:0
[ 7678.574] (II) FBDEV(1): using default device
[ 7678.574] (EE) Screen 0 deleted because of no matching config section.
[ 7678.574] (II) UnloadModule: "modesetting"


I'll remove all the nvidia & nouveau drivers, and install the latest "short term" drivers directly from Nvidia download site...

[pivert]

Thanks Claydoh.

Problem fixed. Solution:

• Remove nvidia related packages
• Download and install the latest Nvidia drivers (from Nvidia web site)

In my case, this was:

apt remove libnvidia-compute-450/groovy nvidia-utils-440/groovy nvidia-utils-450/groovy xserver-xorg-video-nouveau/groovy
apt autoremove
sh NVIDIA-Linux-x86_64-455.38.run

[claydoh]

Probably a reinstall, or reconfigure of the package (similar to what was tried for sddm) would have fixed things as well. Likely mismatches between the 20.04 kerne/driver combo and the new ones. Because Nvidia sucks sometimes

Just remember, you probably will have to reinstall the drivers with each kernel update, or at least many of them. The self installed driver does NOT rebuild the kernel module when this happens, the distro provided packages do.

[Snowhog]

As to kernel module rebuilding. Wouldn't/doesn't dkms do this if it is installed?

[claydoh]

As to kernel module rebuilding. Wouldn't/doesn't dkms do this if it is installed?

Not on a manual driver install, ie not using deb files. Unless it is possible to manually set up dkms to do this. Which I am sure it probably is.
I know these used to have a dkms install option, see a 415 readme
but this is missing from more recent releases


oh poop I just lied https://download.nvidia.com/XFree86/Linux-x86_64/455.38/README/installdriver.html#RegisteringTheNda02d

The option must be specified when installing, however.

The option did not always work, from, what I can tell, and from past (semi ancient) history, but that does not mean it is not worth trying, for sure.