1

Hello, complete noob here. I very recently started installing Nvidia the Hard way under advice from Malcolmlewis following some troubles after re-installing suse after I somehow managed to break it. Today we had an update and there was a new kernel update as part of it. I followed the Nvidia Hard Way instructions 3 times and every time after restart I get a wall of noveau errors and a terminal login asking for a login on a black screen, and it’s really small also, looks like only the top left 1080x1920 pixels of the 4K displays. I tried finding a solution and I just don’t know where to even start looking, very soory I am noob at this. I tried running startx and it gives a bunch of warnings and errors about not being able to start server or nonnect to server, sorry I didn’t capture the full readout. Going into artl+alt+F7 only shows a blinking cursor on a black screen.

I have now logged in to a snapshot and used: snapper rollback , to return to a working state. I have also gone into yast>software and turned kernel-default, kernel-default-devel, kernel-devel to Do Not Modify in an attempt to receive updates without updating the kernel until Nvidia releases a new update, even though they just did only a week ago. I am not sure if this is the right thing to do or if I am being a complete noob here. I would just like some quick help to see if I did the right thing or if I am just making things worse, or perhaps how to fix things.

Thank you

2

Hi, confirming that current Nvidia driver doesn’t build for kernel 5.5.x, see this https://lists.opensuse.org/opensuse-factory/2020-02/msg00242.html for instance.
Locking the kernel set may be a workaround, until something else breaks on the system because of other updates (but by then an updated driver might be available).
You can also scroll down the boot menu and boot an older kernel as long as it is still on the system (please ask if you need advice on keeping a specific kernel).
BTW “startx” has been abandoned in TW long ago and should never be attempted: it doesn’t work and might do funny things.

3

Oh I see, thank you very much for the answer.

So you’re saying update everything normally, but then boot into an older kernel at grub, if that’s a better way then yes please if you could let me know how to preserve the previous kernel that would be great!

Thank you again!

4

Hi
Have you downloaded the latest 440.59 driver or using an older one, the new one works fine with the 5.5. kernel.

5

I have now rolled back to yesterday version again just to have a fresh start. Allowing kernel to update normally.

I had a look for those advanced boot options inside grub and I had last week kernel 5.4.14.1, now on 5.4.14.2. And when trying to boot into that it put me in that same black login screen, but this time it was normal size instead of a tiny little box in the corner, but yes still no GUI. Not sure what I am doing wrong there, but I did find the options.

Yes I am using the latest 440.59 downloaded from Nvidia, I was already on it after I downloaded it a few days ago when I was having my Blender/DaVinci Resolve and Xauthority problems. Those have all been fixed with the Hard Way install, which isn’t all that hard. During the installation I did get a quick warning pop up about the installer not being able to find the path to libglvnd and to point it to it manually, but the only options was “ok” and after hitting ok it just installed normally by it self. It did the exact same thing on the old kernel too. Just putting it in here to make sure it’s not what’s causing problems, because it’s the only thing I noticed that was a bit weird. Someone suggested to download libglvnd-devel from yast which I did now also to see if that helps.

Thank you again for the help.

6

Hi
Yes, all you need to do is install the development package, no need to point to any libraries…

For me it’s working fine, but I only use the cuda cores, my qemu machines work fine when I switch to the nvidia card…


pinxi -GSxxz

System:    Kernel: 5.5.1-1-default x86_64 bits: 64 compiler: gcc v: 9.2.1 Desktop: Gnome 3.34.3 wm: gnome-shell dm: GDM 
           Distro: openSUSE Tumbleweed 20200207 
Graphics:  Device-1: Intel Xeon E3-1200 v2/3rd Gen Core processor Graphics driver: i915 v: kernel bus ID: 00:02.0 
           chip ID: 8086:016a 
           Device-2: NVIDIA GK208B [GeForce GT 710] vendor: ZOTAC driver: nvidia v: 440.59 bus ID: 02:00.0 chip ID: 10de:128b 
           Device-3: NVIDIA GK208B [GeForce GT 710] vendor: ZOTAC driver: nvidia v: 440.59 bus ID: 05:00.0 chip ID: 10de:128b 
           Display: x11 server: X.Org 1.20.7 driver: modesetting,nouveau unloaded: fbdev,vesa alternate: intel,nv,nvidia 
           compositor: gnome-shell resolution: 1920x1080~60Hz, 1920x1080~60Hz 
           OpenGL: renderer: Mesa DRI Intel Ivybridge Server v: 4.2 Mesa 19.3.3 compat-v: 3.0 direct render: Yes
7

Thank you for checking. I will run sudo zypper dup again in the morning, then do the Nvidia hard way install again. I did look it up and I saw replies that say I don’t need to uninstall first, is this correct? Just do nomodeset 3 , and run the file. Can’t do right now as it’s midnight here and I have overnight renders that need to go. Will experiment again tomorrow morning and report back.

Thank you for the help.

8

Hi
The nomodeset is not necessary, just the 3 to get to multi-user.target :wink:

9

I am also using Kernel 5.4 with Nvidia Drivers from Repository because 5.5 does not work, I have three Kernels installed and it is not a problem to start a previous one

10

I have redo everything this morning. No error after installing the development package. However after installing the update and Nvidia hard way exactly same results as before. I also tried to log into an older kernel and it’s still the same, just a bunch of nouveao errors, timeout error and a login.

I’m very sorry if I am just being a noob, I just don’t know what to do next. What it reads out is something along the lines of, doing this from memory so might be wrong:

7.243252] nouvoau 0000:09:00.0: gr: 51a200 - done 00000300
7.243259] nouvoau 0000:09:00.0: gr: 51a200 - stat 00000000 00000000 00000000

9.617529] nouvoau 0000:09:00.0: DRM: core notifier timeout
11.617612] nouvoau 0000:09:00.0: DRM: base-0 timeout

Welcome to opensuse tumbleweed. . . . login

And the readout is super tiny, I can’t even take a photo of it, camera will not pick it up. Just a white line. Don’t know fi there’s a way to save this output or if it even means anything. Looks like the 0000:09:00.0 is my master Graphics card that all the monitors are plugged into. I tried splitting the monitors up to run off both cards before but I just could not get it to work either. I am on the read only snapshot now from yesterday, I haven’t rolled back in case there is a solution and that update was huge and take a very long time to redo over and over.

Thank you

11

Hi
Sounds like nouveau is not blacklisted… Does the file exist /etc/modprobe.d/50-nvidia.conf? Or a file blacklisting nouveau?

12

Hello,

I have Nvidia.conf, 50-nvidia-default.conf.rpmsave and there are no files with the name Nouveau

Nvidia did complete the blacklisting step during install, unfortunate that it didn’t apply. What do I have to create to make this work? I have created a blacklist file before, I have an asus motherboard and used to always get a watchdog hardware error and made a blacklist file in this folder. So I assume same must be done with the Nouveau

edit: I did look inside the nvidia.conf file and it just says: blacklist nouveau.

however this is on the working snapshot. I will have to log out and back into the broken one to see if it’s there. Or shuld I run the hard way steps for manually blacklisting nouveau?

13

Hi
So there are two files one called nvidia.conf and one called 50-nvidia-default.conf.rpmsave? if only 50-nvidia-default.conf.rpmsave, rename this as 50-nvidia-default.conf, boot to the newest kernel at runlevel 3 and run the command mkinitrd, then reboot into the new kernel again and install the nvidia run file as per normal…

14

Hello

I have logged into the new broken version and there was no nvidia.conf. I ran the first step of the nvidia installer which creates the nvidia.conf with the blacklist inside. rebooted and re-installed Nvidia driver. after install I checked and the installer now removed the nvidia.conf file. I tried again and added the file myself manually, still no good.

15

I’m sure many on here know a lot more about this than I do, but I had a similar problem after the last update. The latest nvidia version is now in the regular repo. I removed then reinstalled the kernel, removed and reinstalled nvidia-gfxG05-kmp-default, and then rebooted, and everything is working now. Your results may vary, but this worked for me using versions all now in the regular repos. Did all this from within the broken version. Hope this helps.

16

Hi
Did you run the mkinitrd command after the file was added? Is it present in the initrd?


lsinitrd |grep nvidia

-rw-r--r--   1 root     root           18 Oct 13 12:40 etc/modprobe.d/50-nvidia.conf
17

Hello, I am bad at forums, the replies are all out of order.

I was running mkinitrd after the install

I have now followed your instruction with renaming, I had to look up commands for how to find and rename files via the terminal. I have then follower the rest of the instructions and the GUI loaded me in but things are very broken. I have about 5-10 fps only. everything is very laggy. My desktop backgrounds are black instead of blurred and cannot be changed. Video playback is barely playing. And I cannot open Blender. When trying to open blender. I also can’t apply my Nvidia fan and power limit script, walls of errors for no permissions. When trying to run blender via the terminal I get this:

Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Received X11 Error:
        error code:   2
        request code: 151
        minor code:   34
        error text:   BadValue (integer parameter out of range for operation)
Error! Unsupported graphics card or driver.
A graphics card and driver with support for OpenGL 3.3 or higher is required.
The program will now close.

I have tried restarting twice just to see if it fixes itself as it sometimes does, but nothing, everything is a bit broken and weird. Nvidia settings says Driver 440.59 and all the monitors are at the correct resolutions and forced composition pipeline options enabled. xorg.conf appears fine.

I also checked while in the etc folder, the nvidia.conf is missing again. It just vanishes after installing the driver, should I re-add in manually, could this be the problem?

Thank you for all the help and sorry to be such a pest, I would just really like it all to work again so I can get back to work.

Edit:

-rw-r–r-- 1 root root 1224 Feb 5 19:01 etc/modprobe.d/50-nvidia-defualt.conf
-rw-r–r-- 1 root root 18 Feb 11 12:51 etc/modprobe.d/nvidia.conf

This is the first part of the lsinitrd |grep nvidia

But as I said above, the nvidia.conf is no longer there, the installer removes it.

Untill I received your instruction I was always running mkinitrd after the install as per hard way instructions.

edit edit:
I just noticed in the bottom right(system tray?) There is a warning that the software renderer is in use.

18

Hi
Delete the 50-nvidia-default.conf file it is probablythe cause of your issue, it must have a lot of data in it due to the size. Then rename nvidia.conf to 50-nvidia.conf and run mkinitrd again.


 ls -la /etc/modprobe.d/50-nvidia.conf
-rw-r--r-- 1 root root 18 Oct 13 12:40 /etc/modprobe.d/50-nvidia.conf

Check the lsinitrd output should just have output like mine? This all needs to be done at runlevel 3 with the newest kernel.

19

There is no nvidia.conf , should I be adding it manually with just the lines inside: blacklist nouveau ?

20

Hi
What about the file called 50-nvidia-default.conf in your output? It needs to be gone from the directory /etc/modprobe.d…

Yes, but call it 50-nvidia.conf and the contents as indicated blacklist nouveau.