Nvidia gfxG04 with 6.x kernel not working.

I’ve had my kernel and nvidia drivers locked via zypper for a long time since this is always a issue…but I’d like to update to the new 6x kernel. Last night after referencing the following pages I decided to unlock the kernel and nvidia driver as it seems like it was patched. Everything else was up to date just fyi…so I unlocked the kernel and nvidia packages and ran dup. On a reboot I was dumped into a shell after it looked like it was trying to start. I’m on a laptop with a external monitor attached via hdmi. The laptop screen acted like for a second it was going to start which is normal as it usually starts first the the monitor starts; after about a sec it just dumped me into a shell on the laptop. I was able to roll back fine but man I hate this nvidia ****. Doesn’t this page seem to indicate it’s ready? First question is anybody running the newer driver with the 6.x kernel? Second, what’s the best way people are resolving this manually if drivers are not patched…is there a document that shows how to patch this ourselves? And, is there a way to use zypper to install a patched driver we manually patched? I hate manually installing things on linux, I’ve done this in my younger years but I’d rather have the package manager handle the packages on my OS to eliminate cruft and other junk left behind from manual patching.

Here’s the nvidia card…it’s a old one. “NVIDIA Corporation GF114M [GeForce GTX 670M] (rev a1)”

This seems to indicate that the gfxG04 driver version 390.154-18.1 would work with the 6.0.0-1.2 kernel

https://build.opensuse.org/package/view_file/X11:Drivers:Video/nvidia-gfxG04/nvidia-gfxG04.changes?expand=1

https://download.nvidia.com/opensuse/tumbleweed/x86_64/

As of current I’m running on kernel 5.17.4-1-default with 390.147 but want to upgrade to the 6.0.0-1.2 kernel which shows to be available with the new 390.154-18.1 driver.

il | nvidia-computeG04 | package | 390.147-13.1 | x86_64 | (System Packages)
vl | nvidia-computeG04 | package | 390.154-18.1 | x86_64 | NVIDIA

il | kernel-default | package | 5.14.14-1.3 | x86_64 | (System Packages)
vl | kernel-default | package | 6.0.0-1.2 | x86_64 | Main Repository (OSS)

Hi
You need to boot with the 6.0.n kernel into runlevel 3 and force the re-install of the nvidia driver packages installed, then reboot.

See https://forums.opensuse.org/showthread.php/576526-X-won-t-start-after-upgrade-to-latest-kernel-6-0-0-1?p=3165889#post3165889

Thanks, I’ll give it a try. I wonder why this is…generally dup will install and recompile the driver?

Hi
Since you locked the kernel, it won’t rebuild, the force re-install should fix it.

Hmm, maybe I didn’t convey that I unlocked both the kernel and all nividia packages…then I ran a dup. Are you saying despite removing all locks and then running a dup it still didn’t install it because it was previously locked? To be clear I ran a zypper rl for all locked packages.

sudo zypper rl kernel-de* nvidia* x11-video-nvidiaG04*

Then I ran zypper dup.

I’ll try to force it when I get back to the computer but I’m not understanding what your saying about it not rebuilding since the kernel was locked which it was not because I pulled in both a new kernel and new drivers with the dup.

Hi
Ahh ok, so when you rebooted it was the new kernel and drivers? I would force the re-install, just to make sure. Then it would be Factory Mailing list or bug report…

I’ve tried the suggested …force install of nvidia drivers will running the new linux-6.0.1-1 kernel and it does the same thing. I even rollback to my 5.x kernel which was working and then updated the kernel to the new 6.x one by itself…rebooted and then from the shell ran dup once more to pick up the new nvidia drivers. Same thing no matter which way I install it. Xserver says there’s no screen configured. While lsmod does show some nvidia stuff loaded it looks different than the one I’m booted into now which is my old kernel and the old nvidia drivers.

Working one:

"uname -a && lsmod |grep nv
Linux localhost.localdomain 5.17.4-1-default #1 SMP PREEMPT Wed Apr 20 07:43:03 UTC 2022 (75e9961) x86_64 x86_64 x86_64 GNU/Linux

nvidia_drm 57344 1
nvidia_modeset 1060864 80 nvidia_drm
nvidia 15896576 2631 nvidia_modeset
drm_kms_helper 348160 1 nvidia_drm
ipmi_msghandler 126976 2 ipmi_devintf,nvidia
drm 626688 4 drm_kms_helper,nvidia_drm

Not working one with the 6.x kernel and new drivers.

nvidia_drm 57344 0
nvidia_modeset 1060864 1 nvidia_drm
nvidia 15884288 1 nvidia_modeset
ipmi_msghandler 81920 2 ipmi_devintf,nvidia

It doesn’t appear all the modules are either compiled or running.

Here is the dup for the new drivers while successfully running the new kernel.

https://pastebin.com/3Pnyr5N9

Here is the xserver log after installing…duping the new nvidia drivers and then rebooting. It says it can’t start the gpu.

https://pastebin.com/r16mWd0Y

Didn’t really see anything in journalctl …but it’s full of junk so I might be missing something. If anybody knows what to grep for let me know. I think this is a issue with the nvidia drivers not compiling and installing all the modules per the lsmod output above.

Hi
Sounds like you need to create a bug report… openSUSE:Submitting bug reports - openSUSE