You are not logged in.
Pages: 1
I installed arch linux onto a new pc and everything was working fine, but when i tried to use my GPU, nothing i do works. I have the nvidia drivers installed but almost everything i do says i dont have a gpu.
system information
OS: Arch Linux x86_64
Kernel: 6.3.9-arch1-1
Uptime: 25 mins
Packages: 1353 (pacman)
Shell: bash 5.1.16
Resolution: 1920x1080
DE: Plasma 5.27.6
WM: KWin
Theme: [Plasma], Breeze [GTK2/3]
Icons: [Plasma], breeze [GTK2/3]
Terminal: konsole
CPU: Intel i3-9100F (4) @ 4.200GHz
GPU: NVIDIA GeForce RTX 2060 Rev. A
Memory: 3093MiB / 7878MiB when i type
lspci -k | grep -A 2 -E "(VGA|3D)" it gives back
01:00.0 VGA compatible controller: NVIDIA Corporation TU106 [GeForce RTX 2060 Rev. A] (rev a1)
Subsystem: ASUSTeK Computer Inc. TU106 [GeForce RTX 2060 Rev. A]
Kernel driver in use: nouveau when i type
sudo dmesg i get
[ 138.907738] NVRM: No NVIDIA devices probed.
[ 138.908042] nvidia-nvlink: Unregistered Nvlink Core, major device number 235
[ 140.342878] nvidia-nvlink: Nvlink Core is being initialized, major device number 235
[ 140.342883] NVRM: The NVIDIA probe routine was not called for 1 device(s).
[ 140.343993] NVRM: This can occur when a driver such as:
NVRM: nouveau, rivafb, nvidiafb or rivatv
NVRM: was loaded and obtained ownership of the NVIDIA device(s).
[ 140.343994] NVRM: Try unloading the conflicting kernel module (and/or
NVRM: reconfigure your kernel without the conflicting
NVRM: driver(s)), then try loading the NVIDIA kernel module
NVRM: again.
[ 140.343995] NVRM: No NVIDIA devices probed.
[ 140.344247] nvidia-nvlink: Unregistered Nvlink Core, major device number 235when i type
nvidia-smi -L i get
[lucat@mr-Smith-AR ~]$ nvidia-smi -L
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.i am unsure what to do. please help
Last edited by Ayson_Roberto (2023-06-25 18:43:41)
Offline
Read https://wiki.archlinux.org/title/NVIDIA#Installation and figure which step you skipped when lspci says
Kernel driver in use: nouveau Offline
how did you install the nvidia driver and what configs have you changed because the nvidia driver blacklists the nouveau driver by default but you are clearly using the nouveau driver.
Offline
Paste the output of:
pacman -Q | grep nvidiaAlso, make sure you didn't restrict the usage of your GPU in BIOS settings or anything like that (you can select Discrete vs Hybrid).
Offline
"-Qs nvidia" and like the OP you should maybe read what the OP posted.
The system tells you exactly what the (immediate) problem is - and that kernel module and nvidia-utils /are/ installed and the GPU is responding (so not disabled in the BIOS)
Offline
"-Qs nvidia" and like the OP you should maybe read what the OP posted.
The system tells you exactly what the (immediate) problem is - and that kernel module and nvidia-utils /are/ installed and the GPU is responding (so not disabled in the BIOS)
[lucat@mr-Smith-AR ~]$ pacman -Qs | grep nvidia
local/deepin-nvidia-prime-git 1.0.0.r0.g07d25e0-1
nvidia prime for deepin
local/lib32-nvidia-utils 535.54.03-2
local/mhwd-nvidia 525.60.11-1
mhwd-nvidia pci id
local/nvidia 535.54.03-3
local/nvidia-modprobe-service 2-4
Fixes /oldroot unmount issues by unloading nvidia modules before shutdown / reboot.
local/nvidia-prime 1.0-4
local/nvidia-settings 535.54.03-1
local/nvidia-utils 535.54.03-1
Fetches GPU status from NVIDIA GPUs using nvidia-smi
[lucat@mr-Smith-AR ~]$ Offline
Yeah, as mentioned: that's not it.
Did you read my or jonno2002's posts?
Offline
You could try blacklisting the nouveau driver manually. It's weird because as jonno2002 said, nvidia driver blacklists nouveau out of the box.
Open the terminal and create a new configuration file:
sudo vim /etc/modprobe.d/blacklist-nouveau.confAdd these lines into the file:
blacklist nouveau
options nouveau modeset=0The next step is to update the initial RAM filesystem (initramfs) which is used during the system boot process. Use the following command:
sudo mkinitcpio -PFinally, reboot
Offline
You don't need to blacklist anything specifically, you just need to follow the instructions in the wiki.
Online
The drivers are installed, and everything's fine with your card. Modprobe can't reliably stop nouveau from loading before Nvidia so you have to pass the kernel a parameter that blocks the module all together.
Add this to the parameters: modprobe.blacklist=nouveau nvidia_drm.modeset=1
You're probably using GRUB, so just edit the /etc/default/grub file, add those parameters at the end of GRUB_CMDLINE_LINUX_DEFAULT=, save the file, and run: grub-mkconfig -o /boot/grub/grub.cfg
And by the way, can the Wiki page be updated? This issue has been here for ages, and no one seems to have noticed that the modprobe method doesn't work.
Last edited by denexter3 (2023-06-27 16:37:12)
Offline
And by the way, can the Wiki page be updated?
No, because that's all nonsense.
The problem is down to the makeup of the initramfs, not more, not less.
And the wiki does EXPLICITLY cover that situation.
And the kms hook isn't default since "ages" either.
Offline
And by the way, can the Wiki page be updated?
No, because that's all nonsense.
The problem is down to the makeup of the initramfs, not more, not less.
And the wiki does EXPLICITLY cover that situation.
And the kms hook isn't default since "ages" either.
Which part of preventing that singular un-needed module from starting do you consider nonsense?
Maybe it's more sane to remove an entire Kernel function instead of specifically a single module, but in that case I want to be crazy.
Also, not one of you told the guy what the problem was, and instead just told him to read the Wiki page. You all knew what the problem was, you were just angry at him.
Last edited by denexter3 (2023-06-27 20:17:12)
Offline
The nonsense parts are "Modprobe can't reliably stop nouveau from loading before Nvidia" and "This issue has been here for ages, and no one seems to have noticed that the modprobe method doesn't work." and nobody is angry at anyone and actually the nvidia kernel module told what the /problem/ is and we pointed towards the solution.
Offline
Also, not one of you told the guy what the problem was
figure which step you skipped when lspci says
Kernel driver in use: nouveau
the nvidia driver blacklists the nouveau driver by default but you are clearly using the nouveau driver.
we told him what the problem was and i asked a question that was ignored, seth told him to read the wiki page and loook for the step he missed with reguard to the problem "nouvea in use", all we got back was a "pacman -Qs" output with NOTHING else to help at all, you get back what you give basically.
EDIT: oh yea and there is one part of the wiki that should be removed and thats this part:
https://wiki.archlinux.org/title/NVIDIA … figuration
Last edited by jonno2002 (2023-06-27 21:35:46)
Offline
The nonsense parts are "Modprobe can't reliably stop nouveau from loading before Nvidia" and "This issue has been here for ages, and no one seems to have noticed that the modprobe method doesn't work." and nobody is angry at anyone and actually the nvidia kernel module told what the /problem/ is and we pointed towards the solution.
As far as I know his issue is as simple as Initramfs image starting up KMS, which in turn calls for the Nouveau driver.
This completely circumvents the /etc/modprobe.conf since it only works after the root is mounted. By default modprobe.conf settings are not added to the Initial Ram Disk images.
Removing kms works as well, but that's probably gonna bite in the future. Kernel parameters are more difficult to set, but are a safer bet.
Offline
Initramfs image starting up KMS, which in turn calls for the Nouveau driver
That's backwards logic, the modules alias the HW ID and get loaded asap, so within the initramfs if they're included.
This completely circumvents the /etc/modprobe.conf since it only works after the root is mounted. By default modprobe.conf settings are not added to the Initial Ram Disk images.
And that is the complete nonsense part. It's factually wrong.
Adding modprobe.conf entries *after* building the initramfs can of course not magically apply them to the initramfs, but they'll get picked up when the initramfs is built by mkinitcpio.
Of course it's also pointless to add a module to the initramfs to then only blacklist it and that entire situation is explicitly addressed in the linked section of the nvidia wiki page.
Offline
Initramfs image starting up KMS, which in turn calls for the Nouveau driver
That's backwards logic, the modules alias the HW ID and get loaded asap, so within the initramfs if they're included.
This completely circumvents the /etc/modprobe.conf since it only works after the root is mounted. By default modprobe.conf settings are not added to the Initial Ram Disk images.
And that is the complete nonsense part. It's factually wrong.
Adding modprobe.conf entries *after* building the initramfs can of course not magically apply them to the initramfs, but they'll get picked up when the initramfs is built by mkinitcpio.
Of course it's also pointless to add a module to the initramfs to then only blacklist it and that entire situation is explicitly addressed in the linked section of the nvidia wiki page.
Just checked everything again, the modprobe configs, mkinitcpio... You were right, I was wrong.
Well, mistakes make us stronger, and mom didn't raise a weak person.
Offline
Should've gone team red.
Hello There.
Offline
Should've gone team red.
1. this doesnt help in any way
2. the brand/make of card has nothing to do with this problem, its a case of not following instructions
3. "team red" still requires instructions to be followed with just as much chance of failure
Offline
Pages: 1