No GUI after Kubuntu 18.04 -> 18.10 Upgrade
Upgrading my (MSI GP72) laptop running Kubuntu 18.04 -> 18.10 (via `sudo do-release-
I can login using the terminal and use the laptop normally locally or remotely (i.e. I can SSH into it). I've not found any (obvious) error messages in the following log files:
* /var/log/syslog
* /var/log/boot.log
* /var/log/kern.log
* /var/log/
Checking systemctl status lightdm.service shows it's running. Restarting the service makes no difference. I removed/purged plasma-workspace, plasma-desktop, and nvidia-driver-390 packages (and then autoremoved unused packages), rebooted, and re-installed them, to no effect.
A couple of years ago, I was able to fix a similar issue for 16.04 (https:/
How can I fix this issue?
Thanks in advance.
PS: I've been using Ubuntu-based distros since 2006, and it's a shame that upgrading ALWAYS breaks something, usually on the display side. I would've hoped that this kind of recurring problem would've been addressed at some point in the last 10+ years... :/
Question information
- Language:
- English Edit question
- Status:
- Answered
- For:
- Ubuntu Edit question
- Assignee:
- No assignee Edit question
- Last query:
- Last reply:
Related FAQ:
None Link to a FAQ
Revision history for this message
|
#1 |
Do you use proprietary video drivers, please?
Revision history for this message
|
#2 |
Yes, I do. That's why I mentioned the NVIDIA logs. I took a look at the following log files (among others):
/var/
/var/
/var/
Checking systemctl status lightdm.service shows it's running. Restarting the service makes no difference. I removed/purged the plasma-workspace, plasma-desktop, and nvidia-driver-390 packages (and then sudo apt autoremove -y --purge unused packages), rebooted, and re-installed them, to no effect.
Output from a few commands that may be helpful:
$ ubuntu-drivers devices
== /sys/devices/
modalias : pci:v000010DEd0
vendor : NVIDIA Corporation
model : GM107M [GeForce GTX 960M]
driver : nvidia-driver-390 - distro non-free recommended
$ dkms status
nvidia, 390.87, 4.15.0-36-generic, x86_64: installed
nvidia, 390.87, 4.18.0-10-generic, x86_64: installed
virtualbox, 5.2.18, 4.15.0-36-generic, x86_64: installed
virtualbox, 5.2.18, 4.18.0-10-generic, x86_64: installed
$ lsmod | grep nvidia
nvidia_uvm 786432 0
nvidia_drm 40960 1
nvidia_modeset 1110016 2 nvidia_drm
nvidia 14368768 61 nvidia_
drm_kms_helper 172032 2 nvidia_drm,i915
drm 458752 5 drm_kms_
ipmi_msghandler 102400 2 ipmi_devintf,nvidia
After getting 198 lines back, this is the subset of lines that stood out to me the most; it looks as if it can't find nvidia driver files:
$ grep nvidia /var/log/syslog
Oct 21 14:36:22 red-dragon [485]: failed to execute '/usr/bin/
Oct 21 14:36:22 red-dragon systemd-udevd[456]: Process '/usr/bin/
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon ureadahead[826]: ureadahead:
Oct 21 14:36:22 red-dragon nvidia-
Oct 21 14:36:22 red-dragon nvidia-
Oct 21 14:36:22 red-dragon nvidia-
Oct 21 14:36:22 red-dragon nvidia-
While it could be that the files are being found elsewhere (dkms and lsmod show the drivers loaded), the odd part is that it also complains about not being able to execute /usr/bin/
driver : xserver-
Revision history for this message
|
#3 |
If you uninstall the Nvidia drivers and use the opensource nouveau driver is it OK?
What is the output of:
dpkg -l | grep -i Nvidia
Thanks
Revision history for this message
|
#4 |
I've not tried using the nouveau driver in some time, because the OpenGL-based applications I've been working in the last couple of years won't run with it, among other things. Even if it were to work out, I'd simply be less stuck.
The command's output is as follows:
```
$ dpkg -l | grep nvidia
ii libcuda1-384 390.87-0ubuntu1 amd64 Transitional package for nvidia-headless-390
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii libnvidia-
ii nvidia-
ii nvidia-dkms-390 390.87-0ubuntu1 amd64 NVIDIA DKMS package
ii nvidia-driver-390 390.87-0ubuntu1 amd64 NVIDIA driver metapackage
ii nvidia-headless-390 390.87-0ubuntu1 amd64 NVIDIA headless metapackage
ii nvidia-
ii nvidia-
ii nvidia-
rc nvidia-
rc nvidia-
ii nvidia-prime 0.8.10 all Tools to enable NVIDIA's Prime
ii nvidia-settings 390.77-0ubuntu1 amd64 Tool for configuring the NVIDIA graphics driver
ii nvidia-utils-390 390.87-0ubuntu1 amd64 NVIDIA driver support binaries
ii xserver-
```
PS: Sorry for the poor formatting. I'm checking to see if this supports markdown.
Revision history for this message
|
#5 |
sudo dpkg -P nvidia-
sudo apt-get --purge remove xserver-
sudo apt-get --purge remove nvidia-driver-390
sudo apt-get --purge autoremove
sudo apt-get clean
sudo reboot
Do you get the GUI OK?
Revision history for this message
|
#6 |
No, I don't get the GUI.
After those commands, rebooting simply drops me on the terminal login. You don't even hear the GUI greeter in the background "behind" the black screen.
That said, it looks like those commands have left some stuff behind. For example:
$ lsmod | grep nvidia
nvidia_uvm 786432 0
nvidia_drm 40960 0
nvidia_modeset 1110016 1 nvidia_drm
nvidia 14368768 2 nvidia_
drm_kms_helper 172032 2 nvidia_drm,i915
drm 458752 4 drm_kms_
ipmi_msghandler 102400 2 ipmi_devintf,nvidia
$ dpkg -l | grep nvidia
ii libcuda1-384 390.87-0ubuntu1 amd64 Transitional package for nvidia-headless-390
ii libnvidia-
ii nvidia-
ii nvidia-dkms-390 390.87-0ubuntu1 amd64 NVIDIA DKMS package
ii nvidia-headless-390 390.87-0ubuntu1 amd64 NVIDIA headless metapackage
ii nvidia-
ii nvidia-
ii nvidia-
Because of this, I went ahead and tried this to remove nvidia left-overs:
$ sudo apt remove --purge -y nvidia-\* && sudo apt autoremove -y --purge && sudo reboot
Now it gets hangs during boot and won't even get to the console, so I'm looking at the recovery options to try and get it to boot. If you're aware of boot options I could use to get the system to boot, I'd appreciate it. (I was already using pci=nomsi acpi_osi="Linux" due to issues specific to this laptop.)
Thanks for the help, btw.
Revision history for this message
|
#7 |
So, I got the system back to a state where it boots. I re-installed the nvidia-driver-390 package (and its dependencies) after that and I'm now back to "square one" in this question, i.e.
1. No GUI;
2. Hear the greeter in the background "behind" the black screen
After more reading and trying more things, I'm still no closer to getting the GUI back.
If anyone has additional suggestions or requests for info, I'd be happy to provide them.
Thanks.
Revision history for this message
|
#8 |
Have you already tried uninstalling the NVidia driver packages and using the nouveau package?
(This as a temporary test to check whether the xorg environment is functioning at all.)
Revision history for this message
|
#9 |
I have not tried that. I intended to look into that when I was uninstalling the nvidia packages as mentioned above. In addition, the steps that I'm coming across in searches normally focus on doing the opposite, i.e. removing nouveau and using the nvidia ones.
I intended to follow some of the steps backwards to re-enable nouveau, but even this was difficult b/c of differences and often a GUI dependency on tutorials.
Checking with the prime-switch, it shows nouveau as being blacklisted and not loaded, but checking the files under /etc/modprobe.d/ directory doesn't show any matches for nouveau being blacklisted in any of those files, as shown below:
$ cd /etc/modprobe.d/
$ for f in $(ls); do
for> echo "Checking ${f}..."
for> grep nouveau ${f}
for> done
Checking alsa-base.conf...
Checking amd64-microcode
Checking blacklist-
Checking blacklist.conf...
Checking blacklist-
Checking blacklist-
Checking blacklist-
Checking blacklist-
Checking blacklist-
Checking dkms.conf...
Checking intel-microcode
Checking iwlwifi.conf...
Checking vmwgfx-
If you can share steps to replace nvidia drivers for nouveau, particularly on removing its blacklisted status so that they get loaded, I'd appreciate it. The nouveau drivers themselves appear to be installed:
$ dpkg --list | grep nouveau
ii libdrm-
ii libdrm-
ii xserver-
But they're not getting loaded for being blacklisted:
$ sudo prime-switch # reading from /var/log/
[...]
Is nvidia loaded? yes
Was nvidia unloaded? no
Is nvidia blacklisted? no
Is intel loaded? yes
Is radeon loaded? no
Is radeon blacklisted? no
Is amdgpu loaded? no
Is amdgpu blacklisted? no
Is amdgpu versioned? no
Is amdgpu pro stack? no
Is nouveau loaded? no
Is nouveau blacklisted? yes
Is nvidia kernel module available? yes
Is amdgpu kernel module available? no
[...]
Revision history for this message
|
#10 |
What is the output of
grep nouveau /etc/default/grub
Revision history for this message
|
#11 |
There's no output; no matches for nouveau in the file.
Revision history for this message
|
#12 |
Is everyone out of ideas? Do I need to turn this into an bug report or something? I'm starting to think that I should, since this problem appears to be non-trivial AND it was caused by the system upgrade process...
Revision history for this message
|
#13 |
I suggest you report a bug. You do realise that Cosmic (Ubuntu 18.10) is only supported for 9 months whereas Bionic (Ubuntu 18.04) is LTS and supported for 5 years.
Why the upgrade at all?
Revision history for this message
|
#14 |
> I suggest you report a bug.
Can this question/post be converted into a bug? I'd hate to spend the time basically repeating everything...
> You do realise that Cosmic (Ubuntu 18.10) is only supported for 9 months whereas Bionic (Ubuntu 18.04) is LTS and supported for 5 years.
I know.
> Why the upgrade at all?
Because I always like to be on the most recent version. Also, the bug exists regardless of whether I do or don't upgrade.
Revision history for this message
|
#15 |
That's cool. I'm always curious when people update off the LTS release to the next one
Revision history for this message
|
#16 |
see https:/
There are several buttons above the "Message:" box. One of them allows creating a bug report based on this question document.
Can you help with this problem?
Provide an answer of your own, or ask xghost for more information if necessary.