[SOLVED] Not able to switch to NVIDIA GPU anymore

D3vil0p3r · 2023-10-29 11:53:27

I have a GPU hybrid setup with a iGPU Intel and NVIDIA GPU. In the last period, the switching to NVIDIA GPU stopped to work. If I use the iGPU, it is used correctly. If I switch to NVIDIA, it is not used anymore. My environment is Cinnamon (X11) even if I tested it in different ones.

Not sure if it is due to one of the last kernel versions (I'm using vanilla kernel latest version), but, despite both of my GPUs are detected:

lspci | grep VGA

00:02.0 VGA compatible controller: Intel Corporation CoffeeLake-H GT2 [UHD Graphics 630]
01:00.0 VGA compatible controller: NVIDIA Corporation TU106M [GeForce RTX 2060 Mobile] (rev a1)

if I try to use software like envycontrol or optimus-manager, NVIDIA GPU seems to not be used correctly. For example, when I use envycontrol to switch to NVIDIA mode, after reboot, if I run glxinfo command, I get LLVM (I guess software rendering) as OpenGL renderer. While optimus-manager returns the error:

ERROR: the latest GPU setup attempt failed at Xorg pre-start hook.
Log at /var/log/optimus-manager/switch/switch-20231029T123338.log
Cannot execute command because of previous errors.

and the logs say

[176] INFO: # Xorg pre-start hook
[176] INFO: Previous state was: {'type': 'pending_pre_xorg_start', 'requested_mode': 'integrated', 'current_mode': None}
[176] INFO: Requested mode is: integrated
[250] INFO: Available modules: ['nouveau', 'nvidia', 'nvidia_drm', 'nvidia_modeset', 'nvidia_uvm']
[250] INFO: Unloading modules ['nvidia_drm', 'nvidia_modeset', 'nvidia_uvm', 'nvidia'] (if loaded)
[254] INFO: Cannot unload modules: None
[254] INFO: Waiting 1s and retrying.
[1263] INFO: Cannot unload modules: None
[1263] INFO: Waiting 1s and retrying.
[2268] INFO: Cannot unload modules: None
[2268] INFO: Waiting 1s and retrying.
[3276] INFO: Cannot unload modules: None
[3277] INFO: Waiting 1s and retrying.
[4283] INFO: Cannot unload modules: None
[4283] INFO: Waiting 1s and retrying.
[5292] INFO: Max tries (6) exceeded
[5292] ERROR: Xorg pre-start setup error
Traceback (most recent call last):
  File "/usr/lib/python3.11/site-packages/optimus_manager/kernel.py", line 269, in _unload_modules
    subprocess.check_call(
  File "/usr/lib/python3.11/subprocess.py", line 413, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command 'modprobe -r nvidia_drm nvidia_modeset nvidia_uvm nvidia' returned non-zero exit status 1.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/usr/lib/python3.11/site-packages/optimus_manager/hooks/pre_xorg_start.py", line 51, in main
    setup_kernel_state(config, prev_state, requested_mode)
  File "/usr/lib/python3.11/site-packages/optimus_manager/kernel.py", line 25, in setup_kernel_state
    _nvidia_down(config)
  File "/usr/lib/python3.11/site-packages/optimus_manager/kernel.py", line 104, in _nvidia_down
    _unload_nvidia_modules(available_modules)
  File "/usr/lib/python3.11/site-packages/optimus_manager/kernel.py", line 214, in _unload_nvidia_modules
    _unload_modules(available_modules, ["nvidia_drm", "nvidia_modeset", "nvidia_uvm", "nvidia"])
  File "/usr/lib/python3.11/site-packages/optimus_manager/kernel.py", line 277, in _unload_modules
    raise KernelSetupError(f"Cannot unload modules {modules_to_unload}: {e.stderr}") from e
optimus_manager.kernel.KernelSetupError: Cannot unload modules ['nvidia_drm', 'nvidia_modeset', 'nvidia_uvm', 'nvidia']: None
[5431] INFO: Removing /etc/X11/xorg.conf.d/10-optimus-manager.conf (if present)
[5431] INFO: Writing state {'type': 'pre_xorg_start_failed', 'switch_id': '20231029T123338', 'requested_mode': 'integrated'}

I have installed the following NVIDIA packages:

local/gwe 0.15.6-1
    A system utility for controlling NVIDIA GPUs
local/libvdpau 1.5-2
    Nvidia VDPAU library
local/libxnvctrl 535.113.01-1
    NVIDIA NV-CONTROL X extension
local/nvidia-exec 0.1.2-1
    Run programs in nvidia optimus setups with power management for Xorg and Wayland without log out
local/nvidia-open-dkms 535.113.01-8
    NVIDIA open kernel modules
local/nvidia-settings 535.113.01-1
    Tool for configuring the NVIDIA graphics driver
local/nvidia-utils 535.113.01-2
    NVIDIA drivers utilities
local/nvtop 3.0.2-1
    GPUs process monitoring for AMD, Intel and NVIDIA
local/opencl-nvidia 535.113.01-2
    OpenCL implemention for NVIDIA

What am I missing?

Last edited by D3vil0p3r (2023-10-29 13:14:06)

seth · 2023-10-29 12:59:45

You're using nvidia-open and the dkms version thereof…

dkms status
uname -a
sudo journalctl -b | curl -F 'file=@-' 0x0.st

D3vil0p3r · 2023-10-29 13:13:51

@seth the problem was "nvidia-exec" package (that is used to switch GPU in Wayland environment). I uninstalled it and envycontrol started again to work and recognize the NVIDIA GPU.

Arch Linux

#1 2023-10-29 11:53:27

[SOLVED] Not able to switch to NVIDIA GPU anymore

#2 2023-10-29 12:59:45

Re: [SOLVED] Not able to switch to NVIDIA GPU anymore

#3 2023-10-29 13:13:51

Re: [SOLVED] Not able to switch to NVIDIA GPU anymore

Board footer