Cannot ,Activate' the NVidia driver (470.256.02)

FlexQ4 · 2026-01-09 18:42

Hello,
on Q4OS 4.12 i did:
- apt install linux-headers-amd64
- add to /etc/apt/sources.list: deb http://deb.debian.org/debian/ bullseye main contrib non-free
- apt update
- apt install nvidia-detect
- apt install nvidia-driver firmware-misc-nonfree
without any problems (like the last and successful time).

Then a reboot, but it still uses the Nouveau driver:

# sudo dkms status
nvidia-current, 470.256.02: added

Controlled the /etc/modprobe.d/nvidia-blacklists-nouveau.conf file:

# You need to run "update-initramfs -u" after editing this file.
# see #580894
blacklist nouveau

Did the "sudo update-initramfs -u" again & rebooted again. But the Nvidia driver is still NOT loaded (just 'added').
Inxi tells:

Graphics:
  Device-1: NVIDIA GK208B [GeForce GT 730] driver: N/A
  Display: x11 server: X.Org 1.20.11 driver: loaded: nouveau,vesa

instead of:

Graphics:
  Device-1: NVIDIA GK208B [GeForce GT 730] driver: nvidia v: 470.256.02
  Display: x11 server: X.Org 1.20.11 driver: loaded: nvidia

(like on the old machine).
The latest Nvidia driver is on the system:

# apt install nvidia-driver firmware-misc-nonfree
...
firmware-misc-nonfree ist schon die neueste Version (20210315-3).
nvidia-driver ist schon die neueste Version (470.256.02-2).
0 aktualisiert, 0 neu installiert, 0 zu entfernen.

Whats's wrong here, Who can help?

q4osteam · 2026-01-09 22:13

We would recommend you to use nvidia-xconfig https://packages.debian.org/trixie/nvidia-xconfig

FlexQ4 · 2026-01-10 09:40

But that package is for Trixie (driver version 550.163.01-1) and here is still Bullseye (driver version 470.256.02-2) ?

hchiper · 2026-01-10 11:09

Also available for Q4OS 4 (which I still use):

$ apt list nvidia-xconfig
nvidia-xconfig/oldoldstable 470.103.01-1~deb11u1 amd64
nvidia-xconfig/oldoldstable 470.103.01-1~deb11u1 i386

FlexQ4 · 2026-01-10 14:20

Thank you, hchiper, maybe we can figure out the issue...
Here too:

# apt list nvidia-xconfig
Auflistung… Fertig
nvidia-xconfig/oldoldstable 470.103.01-1~deb11u1 amd64

BUT:

# sudo dkms status
nvidia-current, 470.256.02: added

wich seems to differ a bit... however, in the installation process (see above) i didn't specify a driver version?!

So: (how) is it possible to change nvidia-xconfig/oldoldstable "470.103.01-1~deb11u1 amd64" to the correct "470.256.02-2" (which oviously was added)?

Last edited by FlexQ4 (2026-01-10 14:31)

hchiper · 2026-01-11 10:00

Is 470.103.01-1 not working?
Does this help https://codelucky.com/dkms-command-linu … S_Commands ?

FlexQ4 · 2026-01-11 12:09

Thanks again hchiper!
"My DKMS situation": (referring to https://codelucky.com/dkms-command-linu … S_Commands)

codelucky wrote:

1. Adding a Module to DKMS:
Example:
sudo dkms add -m nvidia -v 470.141.03
Expected output:
Creating symlink /var/lib/dkms/nvidia/470.141.03/source -> /usr/src/nvidia-470.141.03

Here i have a symlink /var/lib/dkms/nvidia-current/470.256.02/source -> /usr/src/nvidia-current/470.256.02

Folders without the "-current" do not exist here!

codelucky wrote:

2. Building a Module
After adding a module, you need to build it for your current kernel:
Example:
sudo dkms build -m nvidia -v 470.141.03

And hence, in my case i cannot build the module for the kernel:

# sudo dkms build -m nvidia -v 470.256.02
Error! Could not find module source directory.
Directory: /usr/src/nvidia-470.256.02 does not exist.

So, i have a /usr/src/nvidia-current-470.256.02 folder here (size: 47.8 MB), but not the /usr/src/nvidia-470.256.02 !

How can that be??

Edit: On my Q4-4.12 backup-ssd, where the test-instalation process went well a couple of months ago, i also have only the
"/usr/src/nvidia-current-470.256.02" folder.

dkms --version here is: 2.8

Last edited by FlexQ4 (2026-01-11 12:26)

FlexQ4 · 2026-01-11 18:36

Well, despite the fact that my system has kernel is version 5.10.0-28-amd64

# uname -r
5.10.0-28-amd64

But it obviously did install kernel-heders for version 5.10.0-37 into /usr/src:

drwxr-xr-x 4 root root 4096  9. Jan 15:54 linux-headers-5.10.0-37-amd64
drwxr-xr-x 4 root root 4096  9. Jan 15:54 linux-headers-5.10.0-37-common
drwxr-xr-x 9 root root 4096  9. Jan 23:20 nvidia-current-470.256.02
lrwxrwxrwx 1 root root   24 11. Dez 18:55 linux-kbuild-5.10 -> ../lib/linux-kbuild-5.10

Hmm.
i tried to install the suitable kernel-heders - but i get an ERROR:

# sudo apt install linux-headers-$(uname -r)
Paketlisten werden gelesen… Fertig
Abhängigkeitsbaum wird aufgebaut… Fertig
Statusinformationen werden eingelesen… Fertig
E: Paket linux-headers-5.10.0-28-amd64 kann nicht gefunden werden.
E: Mittels des Musters »linux-headers-5.10.0-28-amd64« konnte kein Paket gefunden werden.

it can't find the suitable linux-headers!

And what can i do now?

hchiper · 2026-01-11 22:02

Seems inconsistent. Q4OS 4 presently uses kernel 5.10.0-37-amd64 (default boot) and 5.10.0-36-amd64.
If you have not already installed Update Manager from the Software Centre, install it and let it update your system (or run $ q4os-pkui).

Also, you could boot into a live session and try $ sudo apt install nvidia-xconfig to see what happens.

FlexQ4 · 2026-01-12 00:26

hchiper wrote:

If you have not already installed Update Manager from the Software Centre, install it and let it update your system.

Many thanks for your assistance hchiper, but no, i will not update my system.

Here is why: (see also https://www.q4os.org/forum/viewtopic.php?id=5041)

In 2024 i installed 4.12.on a first ssd. During installation it updated the kernel (from 5.10.0-22) to 5.10.0-28.
This happenes during the desktop profiling procedure.

Later (early 2025) i installed 4.12.on a second backup/testing ssd. Again desktop profiling procedure updated the kernel, that time to 5.10.0-33.
Thereafter an update-grub did not find the two windows partitions (by os-prober) any more.

Unfortunately the issue couldn't be resolved (since it was probably a Debian issue).

Since such issues have the potential to ruin the complete system i decided to never-ever update again - in contrary:
re-installed 4.12 on the second ssd and ,freezed' it to 5.10.0-22.

Funny enough: i tested the nvidia procedure (in early 2025, with a lent card) on that 5.10.0-22 ssd - and it went flawlessly.
About a year later now (with an x-mas gift GT730 installed) the procedure fails with 5.10.0-28 kernel, and appropriate linux-headers-5.10.0-28 seem not available any more.

Good night.

FlexQ4 · 2026-01-15 13:31

FlexQ4 wrote:

Many thanks for your assistance hchiper, but no, i will not update my system.

I actually did try an update, on an old (3.5") disk: installed q4-4.12, updated (to 4.13 / 5.10.0-37):
Now booting up, i have a screen full of small Copyright-signs for 6 seconds, together with a whimper/Whistle (of the MB spules?).

I think another user had the same issue, but i can' find the thread?
Could i be resolved?

Edit: i found it, here: https://www.q4os.org/forum/viewtopic.php?id=5853

Last edited by FlexQ4 (2026-01-15 15:13)

FlexQ4 · 2026-01-17 15:44

hchiper wrote:

Does this help https://codelucky.com/dkms-command-linu … S_Commands ?

Thanks hchiper, i'll now try building/installing the nvidia module manually via dkms by following the instrucktions of the link above.

Please keep in mind, that my current kernel version is 5.10.0-28:

#uname -r
5.10.0-28-amd64

So first i removed the 5.10.0-37 from my system:

sudo apt-get purge linux-headers-5.10.0-37

and instead copied the appropriate kernel-headers from a healthy backup disk to /usr/src:

me@flex7:~\> ls -l /usr/src
insgesamt 12
drwxr-xr-x 4 root root 4096 15. Jan 22:25 linux-headers-5.10.0-28-amd64
drwxr-xr-x 4 root root 4096 15. Jan 22:25 linux-headers-5.10.0-28-common
lrwxrwxrwx 1 root root   24 11. Dez 18:55 linux-kbuild-5.10 -> ../lib/linux-kbuild-5.10
drwxr-xr-x 9 root root 4096 15. Jan 22:34 nvidia-current-470.256.02

But the headers ar not known to apt:

apt list --installed | grep header
# apt list --installed | grep header

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

Empty! So, is it possible to tell apt that the correct kernel-headers are located again in /usr/src ??
(I found nothing on the web...)
Thank you!

hchiper · 2026-01-17 17:44

FlexQ4 wrote:

So, is it possible to tell apt that the correct kernel-headers are located again in /usr/src ??

To install anything that you want the apt system to recognize, you must use the `apt install` command. Simply copying files is not enough.
The only way I know is to `sudo apt install linux-image-5.10.0-28-amd64` but the package seems to be no more available and you don't want to install a kernel.
Your system appears to be corrupted in some way: `#uname -r` gives 5.10.0-28-amd64 but corresponding linux headers were missing in /usr/src.
Unfortunately I have no more idea to help.

Q4OS Forum

#1 2026-01-09 18:42

Cannot ,Activate' the NVidia driver (470.256.02)

#2 2026-01-09 22:13

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#3 2026-01-10 09:40

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#4 2026-01-10 11:09

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#5 2026-01-10 14:20

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#6 2026-01-11 10:00

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#7 2026-01-11 12:09

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#8 2026-01-11 18:36

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#9 2026-01-11 22:02

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#10 2026-01-12 00:26

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#11 2026-01-15 13:31

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#12 2026-01-17 15:44

Re: Cannot ,Activate' the NVidia driver (470.256.02)

#13 2026-01-17 17:44

Re: Cannot ,Activate' the NVidia driver (470.256.02)

Board footer