22.04 black screen after boot/reboots

Support and Help
, , ,
1

I am running out of ideas to try. When I boot my 22.04 Desktop on bare metal, everything looks normal until the end of the boot process, where I would expect to see my GUI login, all I get is a black screen. I can ssh into the machine fine. This machine is running as a KVM host. All the VMs are running ubuntu server 22.04 and are all running fine.

inxi -F (portion) on the host gives me this:

System:
  Host: b550a-rzn Kernel: 5.15.0-164-generic x86_64 bits: 64 compiler: gcc v: 11.4.0
    Console: pty pts/2 Distro: Ubuntu 22.04.5 LTS (Jammy Jellyfish)
Machine:
  Type: Desktop System: ASUS product: N/A v: N/A serial: <superuser required>
  Mobo: ASUSTeK model: ROG STRIX B550-A GAMING v: Rev X.0x serial: <superuser required>
    UEFI: American Megatrends v: 2806 date: 10/27/2022
CPU:
  Info: 8-core model: AMD Ryzen 7 5700G with Radeon Graphics bits: 64 type: MT MCP arch: Zen 3
    rev: 0 cache: L1: 512 KiB L2: 4 MiB L3: 16 MiB
  Speed (MHz): avg: 1396 high: 1397 min/max: 1400/3800 boost: enabled cores: 1: 1397 2: 1396
    3: 1397 4: 1396 5: 1396 6: 1392 7: 1396 8: 1397 9: 1396 10: 1396 11: 1397 12: 1397 13: 1397
    14: 1396 15: 1396 16: 1395 bogomips: 121373
  Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm
Graphics:
  Device-1: AMD Cezanne vendor: ASUSTeK driver: amdgpu v: kernel bus-ID: 0e:00.0
  Display: server: X.org v: 1.21.1.4 with: Xwayland v: 22.1.1 driver: X: loaded: nvidia
    unloaded: fbdev,modesetting,nouveau,vesa gpu: amdgpu tty: 127x47
  Message: GL data unavailable in console. Try -G --display
Audio:
  Device-1: AMD Renoir Radeon High Definition Audio vendor: ASUSTeK driver: snd_hda_intel
    v: kernel bus-ID: 0e:00.1
  Sound Server-1: ALSA v: k5.15.0-164-generic running: yes
  Sound Server-2: PulseAudio v: 15.99.1 running: yes
  Sound Server-3: PipeWire v: 0.3.48 running: yes

In full disclosure. About 2 years ago when I set this machine up, I had purged the gnome full desktop and installed Vanilla gnome desktop. It has run fine all along, I think until a recent update. So, I reversed that and purged the Vanilla desktop and installed the gnome full desktop again. Still no GUI after boot/reboot.

Appreciate any suggestions

2

Did you try adding nomodeset temporarily to Grub?

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash nomodeset"

Perhaps reconfigure gdm3?
sudo dpkg-reconfigure gdm3

Which command did you use to install the desktop?

I have read a few times that you may need this one.
The caret at the end is important.

sudo apt install ubuntu-desktop^
3

The key issue here is that your system has an AMD iGPU, but X.org reports the NVIDIA driver being loaded. That mismatch commonly causes a black screen.

I’d recommend purging any leftover NVIDIA packages:

sudo apt purge 'nvidia*'
sudo apt autoremove

Then reinstall the display stack:

sudo apt install --reinstall ubuntu-desktop gdm3 xorg

If it still fails, try disabling Wayland in /etc/gdm3/custom.conf (WaylandEnable=false).

Since SSH works, this is almost certainly a display driver / GDM issue, not a system boot problem.

4

Thank you both for you suggestions. Here is what I did.
@tea-for-one I did all your suggestions except the grub edit. Rebooted each time and no change. Then I went on and performed the suggestions by @MesterPerfect and rebooted each time and no change. Then I last did the grub edit suggested by @tea-for-one and this does give a low resolution (big print) login screen. However prior to getting the login screen, I see these lines post at the top of the screen:

1.570762]  mpt2sas_cm0: overriding NVDATA EEDPTagMode setting
11.297642]  [drm:amdgpu_init [amdgpu]] *ERROR* VGACON disables amdgpu kernel modesetting.
16.6140781  [drm:amdgpu_init [amdgpu]] *ERROR* VGACON disables amdgpu kernel modesetting.
19.647109]  libvirt-guests.sh[4695]: libvirt-guests is configured not to start any guests on boot

Not sure if this output gives any further housekeeping to do.

Thanks again!

5

You received some good advice above. Sometimes on has to try several things to sort out what the best solution is.

A couple of things here.

The error above is triggered by nomodset with some AMD products. Basically, nomodset keeps the amdgpu kernel module from loading. That explains that part, which you can fix by removing the nomodset. Problem is that you’ll be back to blank screen.

We need to solve a deeper problem before removing the nomodset parameter. So far, what I’ve found in research is that the kernel parameter

amdgpu.dc=1

can precipitate the behavior you have noted in some hardware configurations, Cezanne seeming to be a frequent culprit.

That parameter controls AMD’s display core. The default was 0 until some later kernel versions started using 1 as the default.

The solution, if that is the cause, is to (temporarily for testing) set the kernel parameter to

amdgpu.dc=0

and remove nomodset.

This gives a pretty good rundown on temporarily setting a kernel parameter (so you can test), and then making the change permanent if it solves your issue.

Let us know how that turns out.

6

Is the Nvidia GPU essential for running Ubuntu server virtual machines?

7

Thanks @qiii I tested this and it gave screen artifacts during the bootsplash and eventually the screen froze.

The VMs are not in need of a GPU.

Originally, hareware-wise, this machine used to have an Nvidia GT-1030 card in it along with an AMD 5600X CPU. About a year later, I replaced the CPU with an AMD 5700G with onboard video, and I removed & uninstalled the Nvidia card. It all has worked since then for about 1.5 years. Perhaps I am getting close to needing a reinstall.

This install is a root on ZFS with daily snapshots. If I am close to a reinstall, I may try a rollback to about 1 month ago and see if that recovers the display. If not, no biggie, I have good backups.

Thanks again all, for the suggestions!

8

This topic was automatically closed after 30 days. New replies are no longer allowed.

9

I wanted to update on this earlier thread:

I did a zfs rollback of my root system to a snapshot from November 1. Everything booted perfectly as it did before, giving me the Vanilla Gnome login screen. After sudo apt update there were 127 updates available. After sudo apt upgrade the system boot was back to not working again, giving a black screen with no gdm.

I can roll back again if needed, would it help to show a list here of those 127 packages to perhaps identify which one may be breaking?

It’s not urgent, because this system doesn’t really need a desktop environment, it just hosts a couple of VM services.

thanks again

10

New post added to end of this thread, and this post re-opened.

11

I think I found a acceptable fix path here. These were my discovery steps, ultimately the fix was easy.

Rolled back root & boot pools to November 1, 2025 using a zfs snapshot from that time

This returned the system to a successful boot giving the normal login screen for Vanilla Gnome, as it was in November

sudo apt update (130 packages need updating)

Looked through the package list and can only suspect that either the linux-firmware update or the kernel update was sending me into the black hole. I could SSH and do everything else, system was stable and VMs ran just fine, but…

sudo apt upgrade (confirmed that updating the packages once again did give the black screen)

Tried removing Vanilla Gnome and reinstalling the normal Gnome:

sudo apt install ubuntu-desktop gnome-shell-extension-ubuntu-dock
sudo apt purge vanilla-gnome-desktop vanilla-gnome-default-settings
sudo apt autoremove --purge
sudo reboot now

Still gives black screen

Last effort - updated kernel to HWE

sudo apt-get install --install-recommends linux-generic-hwe-22.04

Booted like normal again :slight_smile:

The 5.15 kernel had been working fine on my hardware for a few years, not sure what changed there.

Thanks again everyone

12

This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.