NVIDIA Drivers not working anymore since last update (08/01/2021)

rjuarezp

Hi everyone!

First of all, happy new year to you all!

The update from yesterday broke my NVIDIA driver. After updating the system, the X-Server was not able to start again. Some hardware module from NVIDIA could not be found. Before the update, the drivers were working without any issue (GeForce GT710).

I just removed all nvidia drivers and now the X-Server is starting without any issue with the noveau drivers.

Does anybody has the same problem?

Thanks!

brent

rjuarezp Responding only for the scientific method so OP can troubleshoot.
I have the nvidia common and 390 packages. Both were updated Friday. No problems.
EDIT// regarding what Staudey said: I did the full update.

Staudey

It sounds like you did a full system update (i.e. updated ALL the packages) is that correct?
Asking because only updating the driver, or only the kernel, will lead to issues.

rjuarezp

Staudey Thanks for the answer! Yes,it was a full update of the system done with the Software Center App.

rjuarezp

brent This was the update I did:

rjuarezp

I also tried to revert the update with sudo eopkg history -t N (N is for me the number of the update) but it didn't help. I can execute the command but I don't see whether anything is happening and after that, the problem persists.

Until now the only way of having the X-Server again is to remove the NVIDIA drivers. I have also tried to install them again with the hardware configuration tool but as soon as I have them installed, the X-Server does not boot again...

I don't know... maybe I am lucky next Friday and the next update fixes the problem...

Staudey

SuperJC710e However, when gathering some info, I noticed that for some reason, at some point, the system updated the nvidia drivers to the current version instead of the 390.x series which actually supports the card in my system.

Clarification: Nvidia removed support for your card from their current driver serious, and unfortunately they don't really announce this sort of thing. Ideally we would have a mechanism in place to automatically detect this and switch to a different branch, but that is currently not possible.

rjuarezp Btw, the GT710 should still be supported just fine by the new driver, according to their Supported Products page, but downgrading to the 390 branch is worth a try anyway.

But first, can you tell me what the command uname -a shows on your system?

P.S. in your eopkg history command you'd have to give the number of the operation BEFORE the latest update, because the command resets to the state AFTER the given operation finished. Which means if you give it the latest one, it will do nothing.

SuperJC710e

I too experienced a similar issue. It's the first time I had to do a rollback. I think it was related to nvidia drivers as well. However, when gathering some info, I noticed that for some reason, at some point, the system updated the nvidia drivers to the current version instead of the 390.x series which actually supports the card in my system.

I noticed it when I ran inxi -Gz

Graphics:  Device-1: Intel 3rd Gen Core processor Graphics driver: i915 v: kernel 
           Device-2: NVIDIA GK107GLM [Quadro K1000M] driver: nvidia v: 455.45.01 
           Device-3: Acer ThinkPad Integrated Camera type: USB driver: uvcvideo 
           Display: x11 server: X.Org 1.20.10 driver: modesetting,nvidia resolution: 1920x1080~60Hz 
           OpenGL: renderer: Quadro K1000M/PCIe/SSE2 v: 4.6.0 NVIDIA 455.45.01

So after the rollback I "downgraded" the drivers.

sudo eopkg install nvidia-390-glx-driver-current nvidia-390-glx-driver-32bit

After that I ran the full upgrade again sudo eopkg upgrade -y, then I rebooted, and voila! It's working properly again. I'm not sure when the system decided to upgrade the nvidia drivers, and strangely the system kept working until that last upgrade...anyway, it's good now. Maybe you need to do something similar?

Graphics:  Device-1: Intel 3rd Gen Core processor Graphics driver: i915 v: kernel 
           Device-2: NVIDIA GK107GLM [Quadro K1000M] driver: nvidia v: 390.141 
           Device-3: Acer ThinkPad Integrated Camera type: USB driver: uvcvideo 
           Display: x11 server: X.Org 1.20.10 driver: modesetting,nvidia resolution: 1920x1080~60Hz 
           OpenGL: renderer: Quadro K1000M/PCIe/SSE2 v: 4.6.0 NVIDIA 390.141

[Edit] Maybe it might be to run doflicky-ui (Hardware Drivers) and see what version it's suggesting?

rjuarezp

Staudey Here your are the output: solus@pedrito ~ $ uname -a Linux pedrito 5.10.1-162.current #1 SMP PREEMPT Wed Dec 16 10:10:41 UTC 2020 x86_64 GNU/Linux

Thanks for the support!

Staudey

rjuarezp 5.10.1

There's your problem. Seems like you are not running the latest kernel for some reason. Might be that your ESP is too small to add the kernel (since I can see that eopkg did indeed update to 5.10.5 in your log, but it isn't applying).
Because the nvidia-drivers and the kernel have to be updated together, this is what is causing your problems.

One other possibility: You did restart after the kernel update, correct? Otherwise the kernel update won't apply obviously.

rjuarezp

Staudey Why I am not running the last kernel? Can I just force the update? I always update the system from the Control Center...

Is there a way to force the kernel update?

Thanks!

By the way, using the 390 branch of the NVIDIA drivers is not helping me.

Staudey

rjuarezp By the way, using the 390 branch of the NVIDIA drivers is not helping me.

Yes that won't work for the same reason.

Do you get any errors when running sudo clr-boot-manager update?

rjuarezp

Staudey Now the NVIDIA drivers work with the kernel 5.10.5.-167!!!!

But it is strange that when I run the command sudo clr-boot-manager list-kernels
I get the following kernels listed:

com.solus-project.current.5.10.5-167
com.solus-project.current.5.10.1-162

But at the boot sequence, I can also select another kernel from the 4.x branch... How can I clean the kernels?

Also the command sudo clr-boot-manager set-timeoutis not setting the timeout from my boot manager. My boot manager ignores it. If I want to change it, I have to press the key t while booting. Exactly the same happens if I want to set a new entry as default boot. I have to press d at the entry I want to have as default boot...

Very strange...

Staudey

rjuarezp How can I clean the kernels?

I'm not exactly sure and hope somebody else can help, before I accidentally break your system 😀

rjuarezp Also the command sudo clr-boot-manager set-timeoutis not setting the timeout from my boot manager.

Did you also do a sudo clr-boot-manager update after setting the timeout?
(because the Help Center Document does it like that)