0

My final goal is to make use of a Geforce GTX 550 Ti on a modern Linux Gnome desktop. Since recent Ubuntu versions stopped supporting the Nvidia 390 driver that is required for this I instead chose Debian Sid (a.k.a. Unstable), which offers seemingly straightforward install instructions for it.

After trying to install them I'm left with a white "Something has gone wrong" low-resolution GUI error screen instead of the login screen (no buttons, no cursor), and switching to a console doesn't give me a login prompt, but only black screen and non-responsive caret at top left. I then need to restart and do a safe-boot to access the console. Update: After a full-upgrade I now end up with software rendering, which is technically better than having to reboot, but not much.

What I've done

  1. Installed Debian Testing amd64 weekly DVD build. This was working just fine in Gnome desktop using nouveau and Wayland.
  2. Filled my sources.list with the appropriate content since it was completely empty apart from a reference to the install media (USB iso). Before that I couldn't even do an apt update. I don't know why I was left with an essentially empty sources.list immediately following a clean install.
  3. Enabled unstable by adding one line in sources.list, according to Debian's instructions to enable unstable on Testing.
  4. Followed the instructions for Nvidia 390 driver for Sid in the above linked wiki page. Yes, I forgot to install the kernel headers before doing this.
  5. Ended up with a non-working system as described above, so I uninstalled the two driver packages again using safe-boot (nvidia-legacy-390xx-driver firmware-misc-nonfree) and removed all three nvidia symlinks in etc/modprobe.d (one of which was blacklisting nouveau). I got a working system back.
  6. I realized my mistake and installed the kernel headers (linux-headers-amd64) according to instructions.
  7. I again tried installing the driver (nvidia-legacy-390xx-driver firmware-misc-nonfree) but ended up with the same result as before after a reboot.
  8. I didn't see much actual build-action going on when installing the driver packages, so I tried to --reinstall the nvidia-legacy-390xx-kernel-dkms package, which finally actually menioned "building" something.
  9. I still ended up in the same situation, i.e. no login screen and no access to console during normal boot.

I'm unsure if I accidentally messed something up and didn't fix it properly, or what the problem could be at this point. What would be the best course of action to troubleshoot this situation?

I also did a full-upgrade for good measure to make sure all packages were up to date.

Observations

  • When the system is working (with nouveau), Gnome desktop works fine on Wayland but gives a fullscreen "Something has gone wrong" GUI error screen when logging in with Gnome on Xorg. It has a button to "log out" which I can click with the space bar (no mouse cursor). MATE desktop works fine though, and it also uses Xorg. I'm fine with Gnome on Xorg temporarily not working — it's Testing + Unstable after all, but I would like to think that at least the login screen (gdm) and console should work if the graphics drivers work (just as with nouveau)?
  • One reason I have specifically chosen Debian Testing, i.e. Debian 13 Trixie, instead of Debian 12 is that I needed to ensure I had the very latest nouveau possible. This is because even on the most up-to-date Ubuntu 24.10 the nouveau driver will invariably freeze the entire system after some time (I have determined this experimentally; Ubuntu is locking up only when nouveau is loaded and never when running only on Intel integrated graphics). I'm trying to maximize my chances that nouveau will work (Windows 10 does not freeze, so I attribute this to the nouveau drivers, and not to the hardware).

Additional info

sources.list

I had to reconstruct this file immediately after OS install, for some reason, because it only contained the first line (now commented out).

# deb cdrom:[Debian GNU/Linux testing _Trixie_ - Official Snapshot amd64 DVD Binary-1 with firmware 20250310-03:29]/ trixie contrib main non-free-firmware

# unstable
deb http://deb.debian.org/debian unstable main contrib non-free non-free-firmware


deb http://deb.debian.org/debian testing main contrib non-free non-free-firmware
deb-src http://deb.debian.org/debian testing main contrib non-free non-free-firmware

deb http://deb.debian.org/debian-security/ testing-security main contrib non-free non-free-firmware
deb-src http://deb.debian.org/debian-security/ testing-security contrib non-free main non-free-firmware

deb http://deb.debian.org/debian testing-updates contrib non-free main non-free-firmware
deb-src http://deb.debian.org/debian testing-updates contrib non-free main non-free-firmware

Xorg log errors

These might not be relevant since I can't even get to the login screen and console, and since MATE works on Xorg, but I include these for completeness until I know they can be dismissed.

$ grep EE /var/log/Xorg.0.log
[    74.735] Current Operating System: Linux debian 6.12.17-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.12.17-1 (2025-03-01) x86_64
    (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[    75.166] (EE) systemd-logind: failed to take device /dev/dri/card0: Invalid argument
[    77.033] (EE) Failed to load module "nv" (module does not exist, 0)
[    77.552] (EE) [drm] Failed to open DRM device for pci:0000:01:00.0: -19
[    77.552] (EE) open /dev/dri/card0: Invalid argument
[    77.552] (EE) open /dev/dri/card0: Invalid argument
[    77.562] (EE) Unable to find a valid framebuffer device
[    77.563] (EE) Screen 0 deleted because of no matching config section.
[    77.563] (EE) Screen 0 deleted because of no matching config section.
[    77.653] (II) Initializing extension MIT-SCREEN-SAVER
[    77.667] (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
[    78.331] (EE) systemd-logind: failed to take device /dev/dri/card0: Invalid argument

$ grep EE /var/log/Xorg.0.log.old
[    23.670] Current Operating System: Linux debian 6.12.17-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.12.17-1 (2025-03-01) x86_64
    (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[    23.970] (EE) systemd-logind: failed to take device /dev/dri/card0: Invalid argument
[    25.852] (EE) Failed to load module "nv" (module does not exist, 0)
[    26.471] (EE) [drm] Failed to open DRM device for pci:0000:01:00.0: -19
[    26.471] (EE) open /dev/dri/card0: Invalid argument
[    26.472] (EE) open /dev/dri/card0: Invalid argument
[    26.490] (EE) Unable to find a valid framebuffer device
[    26.491] (EE) Screen 0 deleted because of no matching config section.
[    26.491] (EE) Screen 0 deleted because of no matching config section.
[    26.522] (II) Initializing extension MIT-SCREEN-SAVER
[    26.535] (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
[    27.025] (EE) systemd-logind: failed to take device /dev/dri/card0: Invalid argument

Feel free to ask for anything else that seems relevant and I'll add it.

Improve this question

1 Answer 1

Reset to default
1

The system did not have an xorg.conf file, and it had to be created manually.

Just like how the system did not have a proper sources.list I suppose.

I investigated this after Nvidia settings told me to run the nvidia-xconfig command, which — of course — also did not exist. A pattern is emerging.

Update

I've since learned that the nvidia-xconfig command is available, but in its own package (with the same name) which is not specified as a dependency or even a recommendation of the Nvidia Settings package that explicitly tells you to "just run" the command (!)

Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.