Totally don't understand how to configure conty with nvidia driver 510.39

[ShulkerSakura]

I'm so sorry but I can't understand the operation in the known issues
I tried to modify the conty.sh but I don't know how to extract utils.tar.gz out of it .
I tried to compile a new conty rootfs and transplant the /usr/lib/nvidia to it.
But when I create the /etc/ld.so.conf.d/nvidia.conf and write "/usr/lib/nvidia" in it, and run ldconfig.
Nothing changed, the nvidia-driver is still not working.
Run glxgears:
X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 152 (GLX)
Minor opcode of failed request: 3 (X_GLXCreateContext)
Value in failed request: 0x0
Serial number of failed request: 28
Current serial number in output stream: 29
Run nvidia-smi:
Failed to initialize NVML: Driver/library version mismatch

Sorry I'm noob in Linux and can't understand how to do the operation in the known issues.

[Kron4ek]

Hello, here is how to install any driver version into Conty:

1. Download the installer of the driver version you are currently using from nvidia.com (in this example i will use NVIDIA-Linux-x86_64-510.39.01.run).

2. Clone the Conty github repo with git:

    $ git clone https://github.com/Kron4ek/Conty
   

3. Obtain Arch Linux bootsrap with the create-arch-bootstrap.sh script (this will create root.x86_64 directory):

    $ sudo ./create-arch-bootstrap.sh
   

4. Copy the driver installer into root.x86_64/opt:

    $ sudo cp ~/Downloads/NVIDIA-Linux-x86_64-510.39.01.run root.x86_64/opt
   

5. Execute the enter-chroot.sh script to chroot into root.x86_64:

    $ sudo ./enter-chroot.sh
   

6. Now when you are inside the chroot, remove nvidia-utils and lib32-nvidia-utils:

    # pacman -Rdd nvidia-utils lib32-nvidia-utils
   

7. Then run the driver installer (while still being in the chroot) and let it install the driver libraries:

    # chmod +x /opt/NVIDIA-Linux-x86_64-510.39.01.run
    # /opt/NVIDIA-Linux-x86_64-510.39.01.run --no-kernel-module
   

8. Exit the chroot:

    # exit
   

9. Run the create-conty.sh script, which will create your custom conty.sh

10. Done!

[ShulkerSakura]

New status:
I extracted all the files in the same version of NVIDIA driver .run and copy it both to /usr/lib and /usr/lib/nvidia (Just to be foolproof) and tested the glxinfo glxgears vulkaninfo and vkcube and they all working in properly.

glxinfo -B :
name of display: :0
display: :0 screen: 0
direct rendering: Yes
Memory info (GL_NVX_gpu_memory_info):
Dedicated video memory: 6144 MB
Total available memory: 6144
Currently available dedicated video memory: 4407 MB
OpenGL vendor string: NVIDIA Corporation
OpenGL renderer string: NVIDIA GeForce RTX 3060 Laptop GPU/PCIe/SSE2
OpenGL core profile version string: 4.6.0 NVIDIA 510.39.01
OpenGL core profile shading language version string: 4.60 NVIDIA
OpenGL core profile context flags: (none)
OpenGL core profile profile mask: core profile

OpenGL version string: 4.6.0 NVIDIA 510.39.01
OpenGL shading language version string: 4.60 NVIDIA
OpenGL context flags: (none)
OpenGL profile mask: (none)

OpenGL ES profile version string: OpenGL ES 3.2 NVIDIA 510.39.01
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20

glxgears :
Running synchronized to the vertical refresh. The framerate should be
approximately the same as the monitor refresh rate.
822 frames in 5.0 seconds = 164.238 FPS
826 frames in 5.0 seconds = 165.006 FPS
.....

vulkaninfo :
It's too long, but it's basically all true.

vkcube :
MESA-INTEL: warning: Performance support disabled, consider sysctl dev.i915.perf_stream_paranoid=0
Selected GPU 0: NVIDIA GeForce RTX 3060 Laptop GPU, type: DiscreteGpu

***But ***
when I try to run any wine programs which need graphics card to calculate, it is said:

X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 152 (GLX)
Minor opcode of failed request: 3 (X_GLXCreateContext)
Value in failed request: 0x0
Serial number of failed request: 290
Current serial number in output stream: 291

I really don't know how to solve these problems, please help me.

[ShulkerSakura]

Hello, here is how to install any driver version into Conty:

1. Download the installer of the driver version you are currently using from nvidia.com (in this example i will use NVIDIA-Linux-x86_64-510.39.01.run).
2. Clone the Conty github repo with git:

    $ git clone https://github.com/Kron4ek/Conty
   

3. Obtain Arch Linux bootsrap with the create-arch-bootstrap.sh script (this will create root.x86_64 directory):

    $ sudo ./create-arch-bootstrap.sh
   

4. Copy the driver installer into root.x86_64/opt:

    $ sudo cp ~/Downloads/NVIDIA-Linux-x86_64-510.39.01.run root.x86_64/opt
   

5. Execute the enter-chroot.sh script to chroot into root.x86_64:

    $ sudo ./enter-chroot.sh
   

6. Now when you are inside the chroot, remove nvidia-utils and lib32-nvidia-utils:

    # pacman -Rdd nvidia-utils lib32-nvidia-utils
   

7. Then run the driver installer (while still being in the chroot) and let it install the driver libraries:

    # chmod +x /opt/NVIDIA-Linux-x86_64-510.39.01.run
    # /opt/NVIDIA-Linux-x86_64-510.39.01.run --no-kernel-module
   

8. Exit the chroot:

    # exit
   

9. Run the create-conty.sh script, which will create your custom conty.sh
10. Done!

Sorry for ingored this cause I'm editing the New Status I'll try it thank you.

[Kron4ek]

I extracted all the files in the same version of NVIDIA driver .run and copy it both to /usr/lib and /usr/lib/nvidia
when I try to run any wine programs which need graphics card to calculate, it is said:
X Error of failed request: BadValue (integer parameter out of range for operation)

I think you also need to copy 32-bit libraries to a proper location. I mean when you extract the nvidia (.run) installer you get the directory named "32", you need to copy files from it to conty's /usr/lib32 directory. For example:

cp NVIDIA-Linux-x86_64-510.39.01/* squashfsroot/usr/lib
cp NVIDIA-Linux-x86_64-510.39.01/32/* squashfsroot/usr/lib32

[ShulkerSakura]

Oops, The driver I extracted doesn't contain a 32 bit driver files ....

[Kron4ek]

They are present in the installers from here.

[p4cm4n]

I tried the solution, but got the same with the glxgears

❯ NVIDIA_FIX=1 ./conty.sh glxgears
Running Conty

X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 152 (GLX)
Minor opcode of failed request: 3 (X_GLXCreateContext)
Value in failed request: 0x0
Serial number of failed request: 36
Current serial number in output stream: 37

[p4cm4n]

The previous version works perfectly, even with the needo of the NVIDIA_FIX parameter

[Kron4ek]

The previous version likely has the same driver version as your host system.

NVIDIA_FIX=1 ./conty.sh glxgears

NVIDIA_FIX has been removed several months ago. And even when it was available, it probably didn't work properly.

[p4cm4n]

Thank you very much!

It's closed for mee, IDK about OP.... :-/

[VHSgunzo]

@Kron4ek
Hi, I am creating a similar project and have solved the problem with Nvidia driver.

To do this, I created a package with fake nvidia-utils to lighten the weight of the container and to mount external nvidia libraries.
https://github.com/VHSgunzo/runimage-fake-nvidia-utils

And created images with nvidia drivers, which are used as additional modules for the container.
https://github.com/VHSgunzo/runimage-nvidia-drivers

see how the functions are works:

• check_nvidia_driver
• get_nvidia_driver_image
• mount_nvidia_driver_image
  https://github.com/VHSgunzo/runimage/blob/main/Run.sh

[VHSgunzo]

@Kron4ek
Do you have any problems adding this fix?

[Kron4ek]

@VHSgunzo The problem is i don't have an Nvidia GPU and i don't want to add functions that i can't test myself. I mean even if it's something known to be working, i still need an ability to test it myself. So no Nvidia-related fixes, at least until i get some Nvidia GPU.

[VHSgunzo]

@Kron4ek
Well, I have conducted various tests of my container on different linux distributions on nvidia, amd and intel video cards and everything works as it should, with this fix. In any case, the source code is open, add this fix when you are ready.