Nvidia problems on kernel 6.x

[leaveone]

I am currently running kernel 5.10.0-23-amd64 on Debian 12. It is a pretty old machine (2012 AM2 processor) with a Nvidia GeForce GT 710 card. I can install the install the proprietary Nvidia drivers on this kernel, but when I boot into a new 6.x kernel, X doesnt startup, I can't even get a CLI using Alt+F2, it just blinks, and I can never get to a prompt to login and install the new driver. I can ssh into the machine from another computer and try to install the drivers but I get an error that the nvidia module is already loaded. I have managed to kill my X session via ssh, and then kick off the install but I get an error message about how I should be using the Debian packages, and not the nvidia-installer script. Being that I have a pretty old video card, I've been using the proprietary Nvidia drivers because I haven't had luck with the nvidia drivers providers by Debian. Do they work better with kernel 6? I can still boot into my 5.10.0.23 kernel which has an older Nvidia driver installed, I believe it's of the 390 version. I'm not sure how to check that. Can anyone help me troubleshoot?

[cogitech]

Proprietary Nvidia drivers are available in the bookworm non-free repo. That's the recommended way.

I'd say log in via ssh as you did before, enable bookworm non-free in your sources.list and then do a "sudo apt install nvidia-driver". That's the meta-package that should give you all that you need, including building the right module for the running kernel (6.x).

Note: Take a backup first. If that new module gets built but something else goes wrong and you have to boot the 5.x kernel again - bye bye GUI.

FWIW - Fresh installs are so... well... fresh!

[sunrat]

The Debian driver for GT710 would be nvidia-tesla-470-driver. The driver installed with the nvidia-driver package does not support that card.
For that old low-end card however it may be best to just uninstall the proprietary driver and use the default Nouveau driver. Performance should be quite satisfactory with Nouveau.

[leaveone]

Replying to both of your comments:

So, I am a little confused. Is the nvidia-testla-470-driver the correct package, or the nvidia-driver package? How can I check which package supports which cards? I think I tried the Debian drivers and the Nouveau driver in the past and did not have success. That was probably 5+ years ago though so maybe my card would be supported by now?

Also, is it possible to leave my 5.x kernel alone and keep the proprietary nvidia driver that are currently installed, then booting into my 6.x kernel and installing the Debian drivers via apt?

[cogitech]

Yeah, my bad! I failed to notice your old video card. My advice is to follow @sunrat 's advice on this one - nvidia-testla-470-driver, or better yet nouveau.

Regarding your second question; I don't think so, but maybe there is a way to do that I am unaware of. Trying to boot with drivers that are mismatched with the kernel (built for another kernel) has always resulted in breakage in my experience.

After considering @sunrat 's advice, I would do the following:

1) Clonezilla image

2) Boot the box with kernel 6.x into the messed up state

3) ssh into it like you did before

4) sudo apt purge *nvidia*

5) Reboot

You should have a working GUI with kernel 6.x + nouveau at that point. If not, either continue troubleshooting, try nvidia-tesla-470-driver, or restore your clonezilla image and take a new approach. If you do have a working GUI after the steps above, evaluate the performance of the nouveau driver and decide if you want/need to try out nvidia-tesla-470-driver.

Note: might be best to wait for sunrat to chime in again, as there is quite possibly something my old brain missed, once again. However, if you make sure to do step one, then you have a plan B.

[leaveone]

I'm not familiar with Clonezilla image, however I do have hundreds of gigs of data on my drive so creating an image of it prob isnt feasible? At some point in my life I was a bit more organized but now I've got data all over the place.

The sudo apt purge *nvidia* command would only remove anything with *nvidia* that had been installed through apt or a .deb file correct? The nvidia driver I am currently using was built with the nvidia .run file.. which is apparently not how Debian wants people to install nvidia drivers. To remove the curret proprietary nvidia driver I think I have to use the same .run install file with the --uninstall parameter.

The reason I was asking about loading a different driver in kernel 6.x is because I was trying the .run --uninstall method on that kernel last night and it kept telling me that the driver was not installed. I'm pretty sure if I run the same command on kernel 5.x that it will definitely uninstall the nvidia driver. So is it possible to boot to 6 again, install the nvidia-testla-470-driver or nouveau, and edit my xconfig file to load the new driver? Does that make sense?

[cogitech]

I feel compelled to highly recommend getting yourself a big external drive and backing up all your data. It doesn't have to be an image, but an image (whether via Clonezilla or other more user-friendly app such as Acronis) is extremely convenient when it comes time to restoring a screwed up system. Ideally, you'd use a combination of an image and regular backups of your /home.

Yes, it is a lot of data for an image. Yes, it will take a long time. No, you will never regret it. You can always delete the giant image after you are satisfied that you've got things running the way you want.

With that said - Yes, I imagine you are right that the nvidia script is probably going to be the only way to uninstall that proprietary driver. You've created a difficult situation for yourself by deviating from "the Debian way". I've only moved to Debian recently after many years with many distros, and I have realized that it always pays to do things the recommended way.

If you boot into 5.x and use the script to uninstall the old driver, then boot into 6.x and install nvidia-testla-470-driver, I don't think you'll even need to edit your xconfig. I would expect that the old nvidia uninstall script and/or (especially) the nvidia-testla-470-driver installation script will ensure that xconfig is updated.

Or... maybe now is the time to take the bull by the horns, backup everything, and do a fresh installation of bookworm. This would give you the opportunity to switch to btrfs and you could implement snapshotting so you have a convenient way of recovering when things go wrong. I followed this guide viewtopic.php?t=155802 and I am very happy with the results!

[leaveone]

Looks like I have a little bit more research to do, but I am trying to avoid an entirely new Debian install. I'm kinda curious if maybe some of the hardware is too old and that's also an issue for kernel 6.x

Uninstalling nvidia from 5 and booting into 6 and trying the preferred Debian way sounds like the easiest and safest approach. If I can't get the 6 kernel working, at least I know I can boot to 5 and then re-run the nvidia .run file, or maybe try the debian provided drivers again. It's been several years since I messed with it, but I remember that way and nouveau both didn't work for me. It's definitely an old machine but it has served me well for years. I will def start backing stuff up so I'll be ready when this machine eventually dies.

I am still waiting for @sunrat to chime in and see if he has any other ideas, or knows if I can install the debian drivers into the 6 kernel while leaving my 5 kernel alone.

Thanks for all the advice and tips!

[sunrat]

First and most important - you don't have backups of your data? See my sig.

What cogitech is recommending sounds the best course of action. Or just stick with 5.x kernel. There's no pressing need for 6.x kernel in Bullseye and it will be supported for a few years yet. I'm still actually using Buster on my notebook with a 4.x kernel, works fine and is still supported for another year or so.

[sip]

Looks like I have a little bit more research to do, but I am trying to avoid an entirely new Debian install. I'm kinda curious if maybe some of the hardware is too old and that's also an issue for kernel 6.x

Uninstalling nvidia from 5 and booting into 6 and trying the preferred Debian way sounds like the easiest and safest approach. If I can't get the 6 kernel working, at least I know I can boot to 5 and then re-run the nvidia .run file, or maybe try the debian provided drivers again. It's been several years since I messed with it, but I remember that way and nouveau both didn't work for me. It's definitely an old machine but it has served me well for years. I will def start backing stuff up so I'll be ready when this machine eventually dies.

I am still waiting for @sunrat to chime in and see if he has any other ideas, or knows if I can install the debian drivers into the 6 kernel while leaving my 5 kernel alone.

Thanks for all the advice and tips!

Hi leaveone.
Open a terminal, log in as root and run this command:

Code: Select all

lspci | grep VGA

Code: Select all

sudo lshw -c video

You will find out which video card works by default and which type of graphics is built into your laptop.
Check out this information, it will work for you.
https://wiki.debian.org/NvidiaGraphicsD ... ullseye.22

The Debian 11 installation option is the best for you. I offer the best option below.
Debian 11 "Bullseye"
https://wiki.debian.org/NVIDIA%20Optimu ... _Bumblebee
https://wiki.debian.org/Bumblebee#Installation

[Trihexagonal]

Have you thought of trying this? From one of my Kali boxen

Code: Select all

apt search nvidia-xconfig 
Sorting... Done
Full Text Search... Done
nvidia-xconfig/kali-rolling 525.85.05-1 amd64
  deprecated X configuration tool for non-free NVIDIA drivers

That is how I do it on BSD.

[leaveone]

Have you thought of trying this? From one of my Kali boxen

Code: Select all

apt search nvidia-xconfig 
Sorting... Done
Full Text Search... Done
nvidia-xconfig/kali-rolling 525.85.05-1 amd64
  deprecated X configuration tool for non-free NVIDIA drivers

That is how I do it on BSD.

I got the exact same output. I think that just shows which package is available in the apt repository. I do not have that installed, since I am still using the proprietary nvidia driver.

[leaveone]

Looks like I have a little bit more research to do, but I am trying to avoid an entirely new Debian install. I'm kinda curious if maybe some of the hardware is too old and that's also an issue for kernel 6.x

Uninstalling nvidia from 5 and booting into 6 and trying the preferred Debian way sounds like the easiest and safest approach. If I can't get the 6 kernel working, at least I know I can boot to 5 and then re-run the nvidia .run file, or maybe try the debian provided drivers again. It's been several years since I messed with it, but I remember that way and nouveau both didn't work for me. It's definitely an old machine but it has served me well for years. I will def start backing stuff up so I'll be ready when this machine eventually dies.

I am still waiting for @sunrat to chime in and see if he has any other ideas, or knows if I can install the debian drivers into the 6 kernel while leaving my 5 kernel alone.

Thanks for all the advice and tips!

Hi leaveone.
Open a terminal, log in as root and run this command:

Code: Select all

lspci | grep VGA

Code: Select all

sudo lshw -c video

You will find out which video card works by default and which type of graphics is built into your laptop.
Check out this information, it will work for you.
https://wiki.debian.org/NvidiaGraphicsD ... ullseye.22

The Debian 11 installation option is the best for you. I offer the best option below.
Debian 11 "Bullseye"
https://wiki.debian.org/NVIDIA%20Optimu ... _Bumblebee
https://wiki.debian.org/Bumblebee#Installation

sudo lshw -c video
*-display
description: VGA compatible controller
product: GK208B [GeForce GT 710]
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:02:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
configuration: driver=nvidia latency=0
resources: irq:27 memory:df000000-dfffffff memory:d0000000-d7ffffff memory:dc000000-ddffffff ioport:ec00(size=128) memory:c0000-dffff

lspci | grep VGA
02:00.0 VGA compatible controller: NVIDIA Corporation GK208B [GeForce GT 710] (rev a1)

That is the correct info for the card. Yes, it is an old card, in an old machine. I don't do any sort of gaming so I don't need anything real powerful. I've never had much luck getting good performace or resolution from the card using the nouveau or debian provided nvidia drivers. I'll review the links you posted.. thanks!

[leaveone]

So, I ran my normal upgrades and my system started trying to install some patches for the 6.0.1 kernel I have installed. I'm still booted into my 5 kernel because I get a kernel panic using 6, and I can't get past it. Anybody know what the deal is? It seems like it's trying to build an nvidia patch, but it fails. I do see in the make.log file that it references nvidia-tesla-470 but I don't have that installed on the system because I am currently using the proprietary driver in my kernel 5. Any help??


0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.
4 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
Do you want to continue? [Y/n] Y
Setting up linux-image-6.1.0-18-amd64 (6.1.76-1) ...
/etc/kernel/postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-18-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub
applying patch cc_version_check-gcc5.patch...patching file conftest.sh

applying patch bashisms.patch...patching file conftest.sh

applying patch linux-2.6.34-dev_pm_info-runtime_auto.patch...patching file conftest.sh

applying patch 0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch...patching file nvidia-modeset/nvidia-modeset-linux.c
patching file nvidia/nv.c

applying patch ppc64el.patch...patching file nvidia/nvlink_common.h

applying patch fragile-ARCH.patch...patching file Makefile

applying patch use-kbuild-compiler.patch...patching file Makefile
patching file Kbuild

applying patch use-kbuild-flags.patch...patching file Kbuild
patching file nvidia/nvidia.Kbuild
patching file Makefile
patching file nvidia-modeset/nvidia-modeset.Kbuild

applying patch conftest-verbose.patch...patching file Kbuild

applying patch conftest-prefer-arch-headers.patch...patching file conftest.sh


Building module:
Cleaning build area...
env NV_VERBOSE=1 make -j2 modules KERNEL_UNAME=6.1.0-18-amd64...........................................................................................................................(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.1.0-18-amd64 (x86_64)
Consult /var/lib/dkms/nvidia-tesla-470/470.223.02/build/make.log for more information.
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
dkms: autoinstall for kernel: 6.1.0-18-amd64 failed!
run-parts: /etc/kernel/postinst.d/dkms exited with return code 11
dpkg: error processing package linux-image-6.1.0-18-amd64 (--configure):
installed linux-image-6.1.0-18-amd64 package post-installation script subprocess returned error exit status 1
Setting up linux-headers-6.1.0-18-amd64 (6.1.76-1) ...
/etc/kernel/header_postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-18-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub
applying patch cc_version_check-gcc5.patch...patching file conftest.sh

applying patch bashisms.patch...patching file conftest.sh

applying patch linux-2.6.34-dev_pm_info-runtime_auto.patch...patching file conftest.sh

applying patch 0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch...patching file nvidia-modeset/nvidia-modeset-linux.c
patching file nvidia/nv.c

applying patch ppc64el.patch...patching file nvidia/nvlink_common.h

applying patch fragile-ARCH.patch...patching file Makefile

applying patch use-kbuild-compiler.patch...patching file Makefile
patching file Kbuild

applying patch use-kbuild-flags.patch...patching file Kbuild
patching file nvidia/nvidia.Kbuild
patching file Makefile
patching file nvidia-modeset/nvidia-modeset.Kbuild

applying patch conftest-verbose.patch...patching file Kbuild

applying patch conftest-prefer-arch-headers.patch...patching file conftest.sh


Building module:
Cleaning build area...
env NV_VERBOSE=1 make -j2 modules KERNEL_UNAME=6.1.0-18-amd64............................................................................................................(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.1.0-18-amd64 (x86_64)
Consult /var/lib/dkms/nvidia-tesla-470/470.223.02/build/make.log for more information.
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
dkms: autoinstall for kernel: 6.1.0-18-amd64 failed!
run-parts: /etc/kernel/header_postinst.d/dkms exited with return code 11
Failed to process /etc/kernel/header_postinst.d at /var/lib/dpkg/info/linux-headers-6.1.0-18-amd64.postinst line 11.
dpkg: error processing package linux-headers-6.1.0-18-amd64 (--configure):
installed linux-headers-6.1.0-18-amd64 package post-installation script subprocess returned error exit status 1
dpkg: dependency problems prevent configuration of linux-image-amd64:
linux-image-amd64 depends on linux-image-6.1.0-18-amd64 (= 6.1.76-1); however:
Package linux-image-6.1.0-18-amd64 is not configured yet.

dpkg: error processing package linux-image-amd64 (--configure):
dependency problems - leaving unconfigured
dpkg: dependency problems prevent configuration of linux-headers-amd64:
linux-headers-amd64 depends on linux-headers-6.1.0-18-amd64 (= 6.1.76-1); however:
Package linux-headers-6.1.0-18-amd64 is not configured yet.

dpkg: error processing package linux-headers-amd64 (--configure):
dependency problems - leaving unconfigured
Errors were encountered while processing:
linux-image-6.1.0-18-amd64
linux-headers-6.1.0-18-amd64
linux-image-amd64
linux-headers-amd64
E: Sub-process /usr/bin/dpkg returned an error code (1)

[Borg]

So, I ran my normal upgrades and my system started trying to install some patches for the 6.0.1 kernel I have installed. I'm still booted into my 5 kernel because I get a kernel panic using 6, and I can't get past it. Anybody know what the deal is? It seems like it's trying to build an nvidia patch, but it fails. I do see in the make.log file that it references nvidia-tesla-470 but I don't have that installed on the system because I am currently using the proprietary driver in my kernel 5. Any help??

There was an issue with the new debian point release and nvidia cards:
viewtopic.php?t=158200

If you stay at Kernel 5.x this shouldn't be an issue for you. Just make sure, that Kernel 5.x is your default kernel in the grub menu.

[leaveone]

so I got linux-image-6.1.0-17-amd64 linux-headers-6.1.0-17-amd64 installed and tried to boot to that kernel. It stalls and I have to ssh into this machine from another, and attempt to install the proprietary nvidia drivers since I know they work when the module actualls build properly. However, the build fails because I think my system was trying to boot to the nvidia-tesla-470 driver that was built when I got the packages for linux-headers-6.1.0-17-amd64 installed. I then try to reinstall the nvidia-tesla-470 driver and it completes, but when I reboot I still don't have any GUI. Do I have to manually go change the xorg.conf to use the tesla driver? if so, what do I choose instead of nvidia as the option? The proprietary nvidia driver shouldnt be loaded with kernel 6.x since it can't build, but the boot process stalls, and I'm not sure what changes to make. Currently, I am booted into kernel 5.10.0-23-amd64 using proprietary nvidia drivers that have been installed for a year or more.

[leaveone]

here's a list of what is installed on my system for nvidia stuff:

sudo dpkg -l | grep nvidia
ii glx-alternative-nvidia 1.2.2 amd64 allows the selection of NVIDIA as GLX provider
ii libegl-nvidia-tesla-470-0:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary EGL library (Tesla 470 version)
ii libgl1-nvidia-tesla-470-glvnd-glx:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary OpenGL/GLX library (GLVND variant) (Tesla 470 version)
ii libgles-nvidia-tesla-470-1:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary OpenGL|ES 1.x library (Tesla 470 version)
ii libgles-nvidia-tesla-470-2:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary OpenGL|ES 2.x library (Tesla 470 version)
ii libglx-nvidia-tesla-470-0:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary GLX library (Tesla 470 version)
ii libnvidia-egl-wayland1:amd64 1:1.1.10-1 amd64 Wayland EGL External Platform library -- shared library
ii libnvidia-tesla-470-cbl:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary Vulkan ray tracing (cbl) library (Tesla 470 version)
ii libnvidia-tesla-470-cfg1:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary OpenGL/GLX configuration library (Tesla 470 version)
ii libnvidia-tesla-470-compiler:amd64 470.223.02-4~deb12u1 amd64 NVIDIA runtime compiler library (Tesla 470 version)
ii libnvidia-tesla-470-cuda1:amd64 470.223.02-4~deb12u1 amd64 NVIDIA CUDA Driver Library (Tesla 470 version)
ii libnvidia-tesla-470-eglcore:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary EGL core libraries (Tesla 470 version)
ii libnvidia-tesla-470-encode1:amd64 470.223.02-4~deb12u1 amd64 NVENC Video Encoding runtime library (Tesla 470 version)
ii libnvidia-tesla-470-glcore:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary OpenGL/GLX core libraries (Tesla 470 version)
ii libnvidia-tesla-470-glvkspirv:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary Vulkan Spir-V compiler library (Tesla 470 version)
ii libnvidia-tesla-470-ml1:amd64 470.223.02-4~deb12u1 amd64 NVIDIA Management Library (NVML) runtime library (Tesla 470 version)
ii libnvidia-tesla-470-nvcuvid1:amd64 470.223.02-4~deb12u1 amd64 NVIDIA CUDA Video Decoder runtime library (Tesla 470 version)
ii libnvidia-tesla-470-ptxjitcompiler1:amd64 470.223.02-4~deb12u1 amd64 NVIDIA PTX JIT Compiler library (Tesla 470 version)
ii libnvidia-tesla-470-rtcore:amd64 470.223.02-4~deb12u1 amd64 NVIDIA binary Vulkan ray tracing (rtcore) library (Tesla 470 version)
ii nvidia-detect 525.147.05-7~deb12u1 amd64 NVIDIA GPU detection utility
ii nvidia-egl-common 525.147.05-7~deb12u1 amd64 NVIDIA binary EGL driver - common files
ii nvidia-installer-cleanup 20220217+3~deb12u1 amd64 cleanup after driver installation with the nvidia-installer
ii nvidia-kernel-common 20220217+3~deb12u1 amd64 NVIDIA binary kernel module support files
ii nvidia-modprobe 535.54.03-1~deb12u1 amd64 utility to load NVIDIA kernel modules and create device nodes
ii nvidia-opencl-common 525.147.05-7~deb12u1 amd64 NVIDIA OpenCL driver - common files
ii nvidia-persistenced 525.85.05-1 amd64 daemon to maintain persistent software state in the NVIDIA driver
ii nvidia-settings-tesla-470 470.161.03-1 amd64 tool for configuring the NVIDIA graphics driver (Tesla 470 version)
ii nvidia-support 20220217+3~deb12u1 amd64 NVIDIA binary graphics driver support files
ii nvidia-tesla-470-alternative 470.223.02-4~deb12u1 amd64 allows the selection of NVIDIA as GLX provider (Tesla 470 version)
ii nvidia-tesla-470-driver 470.223.02-4~deb12u1 amd64 NVIDIA metapackage (Tesla 470 version)
ii nvidia-tesla-470-driver-bin 470.223.02-4~deb12u1 amd64 NVIDIA driver support binaries (Tesla 470 version)
ii nvidia-tesla-470-driver-libs:amd64 470.223.02-4~deb12u1 amd64 NVIDIA metapackage (OpenGL/GLX/EGL/GLES libraries) (Tesla 470 version)
ii nvidia-tesla-470-egl-icd:amd64 470.223.02-4~deb12u1 amd64 NVIDIA EGL installable client driver (ICD)
ii nvidia-tesla-470-kernel-dkms 470.223.02-4~deb12u1 amd64 NVIDIA binary kernel module DKMS source (Tesla 470 version)
ii nvidia-tesla-470-kernel-source 470.223.02-4~deb12u1 amd64 NVIDIA binary kernel module source (Tesla 470 version)
ii nvidia-tesla-470-kernel-support 470.223.02-4~deb12u1 amd64 NVIDIA binary kernel module support files (Tesla 470 version)
ii nvidia-tesla-470-opencl-icd:amd64 470.223.02-4~deb12u1 amd64 NVIDIA OpenCL installable client driver (ICD) (Tesla 470 version)
ii nvidia-tesla-470-smi 470.223.02-4~deb12u1 amd64 NVIDIA System Management Interface (Tesla 470 version)
ii nvidia-tesla-470-vdpau-driver:amd64 470.223.02-4~deb12u1 amd64 Video Decode and Presentation API for Unix - NVIDIA driver (Tesla 470)
ii nvidia-tesla-470-vulkan-icd:amd64 470.223.02-4~deb12u1 amd64 NVIDIA Vulkan installable client driver (ICD) (Tesla 470 version)
ii nvidia-vulkan-common 525.147.05-7~deb12u1 amd64 NVIDIA Vulkan driver - common files
ii xserver-xorg-video-nvidia-tesla-470 470.223.02-4~deb12u1 amd64 NVIDIA binary Xorg driver (Tesla 470 version)

[sunrat]

@leaveone what happened when you followed cogitech's advice above? Do that and you should be running nouveau. It should work fine for your graphics card.
It looks like you have bits of 470 and bits of 525, and you said you had installed the .run file from Nvidia site at some stage so basically you have a mess.
The nvidia issue with kernel 6.1.0-18 has been fixed with a new driver now in bookworm-updates non-free section of the repo.
Make sure to do apt upgrade after purging all the nvidia stuff.

[leaveone]

I got side tracked with some other stuff and never completely finished trying the uninstall. I remember I booted into my 5.x kernel and then tried to run the nvidia script to uninstall but it failed because I still had X going. I had trouble killing my X session and that's as far as I got. I'm kinda hesitant to totally remove the nvidia proprietary driver because then I may not be able to boot into anything? I guess the easiest way would be to boot into recovery mode kernel 5, and it won't load X so then I could properly uninstall the proprietary nvidia driver? I've been trying to get 6 working, while not blowing up 5.. is that possible? I agree, I do have a bit of a mess with all that stuff installed.

apt purge nvidia* is that the easiest way to remove all that debian stuff and then reinstall later?

[Dai_trying]

You could boot to a normal desktop session and then switch to another tty (Ctrl + Alt + F1 is what I usually use) and then use systemd to stop the xserver

Code: Select all

systemctl stop lightdm

This works in Xfce although if using plasma/gnome/otherDE you might need to change "lightdm" to "gdm" "sddm" or whichever is relevant to your desktop
After this X will have stopped and you can remove the Nvidia drivers (from nvidia website) and then to restart x

Code: Select all

systemctl start lightdm

again substitute lightdm to suit your system.

Edit:
Forgot to say you should switch back to your original tty (mine is usually tty7) with
Ctrl + Alt + F7

HTH