[SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

brout · 2026-01-22 01:26:29

solution: PSU was faulty and failing when NVIDIA drivers were installed. Replaced it.

CPU i7 7700
GPU gtx 1060 6gb
MOBO asus Z270-A
6.18.6-arch1-1
gnome 49

drivers: I've tested nvidia-580xx-dkms & nvidia-535xx-dkms from AUR, with both I experience the same issue.

The issue:
My system will just reboot. Twice after I decrypted, but most of the time after I login at gdm, or before I get to log in.

The wiki says "Starting from nvidia-utils 560.35.03-5, DRM is enabled by default." and that to verify you run "cat /sys/module/nvidia_drm/parameters/modeset" but for me it returns "no such file" after driver installation.

I have tried manually setting the kernel module parameters in systemd-boot & grub (I've reinstalled a few times trying to get this to work, and have switched between systemd-boot & grub between installations. Im on grub now though)

I don't know what to do at all. I'm very lost about this.

Last edited by brout (2026-01-27 14:34:17)

brout · 2026-01-22 03:17:23

Okay I'm on a new arch install with grub, 6.18.6-arch1-1, gnome 49

I have done

 sudo pacman -S dkms linux-headers

and

yay -S nvidia-580xx-dkms

$ cat /sys/module/nvidia_drm/parameters/modeset
cat: /sys/module/nvidia_drm/parameters/modeset: No such file or directory

but according to the wiki, it should return Y (https://wiki.archlinux.org/title/NVIDIA … de_setting)

$ dkms status
nvidia/580.126.09, 6.18.6-arch1-1, x86_64: installed

Can anyone tell me what to do next? I'll keep my system turned on until then.

Last edited by brout (2026-01-22 03:23:41)

Lone_Wolf · 2026-01-22 11:44:43

Run (as root)

# journalctl -b > journal.txt
# lspci -k > lspci.txt

Do you have internet access from that system ?
If yes, execute

curl -F 'file=@-' 0x0.st < journal.txt
curl -F 'file=@-' 0x0.st < lspci.txt
curl -F 'file=@-' 0x0.st < /etc/mkinitcpio.conf

Each of those commands will output a link, post those 3 links.

brout · 2026-01-22 13:13:42

Lone_Wolf wrote:

Run (as root)
# journalctl -b > journal.txt
# lspci -k > lspci.txt
Do you have internet access from that system ?
If yes, execute
curl -F 'file=@-' 0x0.st < journal.txt
curl -F 'file=@-' 0x0.st < lspci.txt
curl -F 'file=@-' 0x0.st < /etc/mkinitcpio.conf
Each of those commands will output a link, post those 3 links.

http://0x0.st/PPeM.txt
http://0x0.st/PPeu.txt
http://0x0.st/PPeQ.txt

this is before rebooting, but after installing dkms, linux-headers, and then nvidia-580xx-dkms from aur

seth · 2026-01-22 13:59:55

You at least have nouveau in the initramfs, regenerate that to make sure it gets blacklisted.
The kms hook will move the nvidia modules in there instead.

brout · 2026-01-22 14:29:27

seth wrote:

You at least have nouveau in the initramfs, regenerate that to make sure it gets blacklisted.
The kms hook will move the nvidia modules in there instead.

okay done. What next?

also fyi in case anyone checks the mkinitcpio file again, i reran the command lonewolf provided for uploading my mkinitcpio.conf, and instead of giving me a new link, it seems it might have overwritten the old since it just gave me the same link

seth · 2026-01-22 14:31:31

What does an updated journal look like?

brout · 2026-01-22 14:35:17

seth wrote:

What does an updated journal look like?

http://0x0.st/PP24.txt

seth · 2026-01-22 14:38:01

Jan 22 14:41:30 archlinux kernel: NVRM: GPU 0000:01:00.0 is already bound to nouveau.
Jan 22 14:41:30 archlinux kernel: NVRM: The NVIDIA probe routine was not called for 1 device(s).
Jan 22 14:41:30 archlinux kernel: NVRM: This can occur when another driver was loaded and 
                                  NVRM: obtained ownership of the NVIDIA device(s).

nouveau is obviously still in the initramfs, https://wiki.archlinux.org/title/Mkinitcpio

brout · 2026-01-22 14:49:20

seth wrote:

Jan 22 14:41:30 archlinux kernel: NVRM: GPU 0000:01:00.0 is already bound to nouveau.
Jan 22 14:41:30 archlinux kernel: NVRM: The NVIDIA probe routine was not called for 1 device(s).
Jan 22 14:41:30 archlinux kernel: NVRM: This can occur when another driver was loaded and 
                                  NVRM: obtained ownership of the NVIDIA device(s).

nouveau is obviously still in the initramfs, https://wiki.archlinux.org/title/Mkinitcpio

Okay, all I did was

sudo mkinitcpio -P

I assumed that

yay -S nvidia-580xx-dkms

would've edited /etc/mkinitcpio.conf accordingly but I guess I have to manually edit it then.

This is my new /etc/mkinitcpio.conf, does it look correct?
http://0x0.st/PP2l.txt

This is the journal after regenerating
http://0x0.st/PP2G.txt

seth · 2026-01-22 15:02:44

That journal still starts at

Jan 22 14:05:27 archlinux kernel: Linux version 6.18.6-arch1-1 (linux@archlinux) (gcc (GCC) 15.2.1 20260103, GNU ld (GNU Binutils) 2.45.1) #1 SMP PREEMPT_DYNAMIC Sun, 18 Jan 2026 00:34:07 +0000

You'll have to reboot to have the updated initramfs have any effect and hopefully get nouveau out of the kernel.

Edit: you don't need the MODULES entries w/ the kms hook and maybe want to remove the latter altogether (to not bloat the initramfs w/ blacklisted modules)

Last edited by seth (2026-01-22 15:04:06)

brout · 2026-01-22 15:07:22

seth wrote:

That journal still starts at
Jan 22 14:05:27 archlinux kernel: Linux version 6.18.6-arch1-1 (linux@archlinux) (gcc (GCC) 15.2.1 20260103, GNU ld (GNU Binutils) 2.45.1) #1 SMP PREEMPT_DYNAMIC Sun, 18 Jan 2026 00:34:07 +0000
You'll have to reboot to have the updated initramfs have any effect and hopefully get nouveau out of the kernel.
Edit: you don't need the MODULES entries w/ the kms hook and maybe want to remove the latter altogether (to not bloat the initramfs w/ blacklisted modules)

Okay, I'll reboot.

Before that, just to confirm, do I remove the kms hook or the modules entries?

seth · 2026-01-22 15:13:21

Editing the file itself won't do anything, but you'd likely want to remove the kms hook and keep the module entries to not accidentally re-add the nouveau module at a later point.

brout · 2026-01-22 15:17:53

seth wrote:

Editing the file itself won't do anything, but you'd likely want to remove the kms hook and keep the module entries to not accidentally re-add the nouveau module at a later point.

Ok, I removed kms hook, then regenerated.

Rebooted.

Now my system reboots by itself right after

Loading Linux linux ...
Loading initial ramdisk ...

seth · 2026-01-22 15:22:03

"system reboots itself" means either of
- underpowered
- overheated
- broken cpu
- bad RAM
and is not a software problem.

Can you boot the multi-user.target (2nd link below)?
Did you forget to attach the dedicated power supply to the GPU?
Is the GPU properly seated in a PEG slot?

StarWolf3000 · 2026-01-22 15:25:18

Actually, when nvidia-580xx-utils was installed, it should've blacklisted nouveau on it's own, according to the PKGBUILD: https://aur.archlinux.org/cgit/aur.git/ … utils#n286

brout · 2026-01-22 15:37:05

seth wrote:

"system reboots itself" means either of
- underpowered
- overheated
- broken cpu
- bad RAM
and is not a software problem.
Can you boot the multi-user.target (2nd link below)?
Did you forget to attach the dedicated power supply to the GPU?
Is the GPU properly seated in a PEG slot?

PSU is 700w
the case has good cooling and the gpu fans spin
cpu - i use it daily without issue on my other installation with igpu instead of the gtx 1060
ram - well its only 2 years old, i guess i could run memtest

yeah the gpu is connected with power & is properly seated. I checked earlier.

seth wrote:

Can you boot the multi-user.target (2nd link below)?

I'll get back to you on this. I have to read the manual, because I have no idea what that's about (linux noob)

seth · 2026-01-22 15:38:34

Yes, the problem is that the nouveau modules were in the initramfs but the blacklisting entry wasn't.
It's necessary to regenerate the initramfs to at least get the blacklisting config in there as well (or replace nouveau w/ nvidia)

brout · 2026-01-22 15:44:11

seth wrote:

Can you boot the multi-user.target (2nd link below)?

Okay, not sure if I did this right but I added

systemd.unit=multi-user.target

at the end of the linux line in my grub boot entry, but I get the same issue. It reboots after trying to load initial ramdisk

I still have the installation medium to chroot or reinstall

seth · 2026-01-22 15:52:29

You can probably still boot by adding "nomodeset" or "module_blacklist=nvidia" to the kernel parameters, but regardless of that - the immediate reboot most likely is because the GPU starves the CPU when it initializes (the PSU output is pretty irrelevant, what matters is what you end up distributing to the devices. Even kernel panics would not trigger a reboot.

If you keep nvidia installed but add "modprobe.blacklist=nvidia,nvidia_drm" to the kernel paramers can you
1. boot the multi-user.target
2. then there explicitly "modprobe nvidia_drm" w/o crashing the system?

brout · 2026-01-22 15:58:42

seth wrote:

You can probably still boot by adding "nomodeset" or "module_blacklist=nvidia" to the kernel parameters, but regardless of that - the immediate reboot most likely is because the GPU starves the CPU when it initializes (the PSU output is pretty irrelevant, what matters is what you end up distributing to the devices. Even kernel panics would not trigger a reboot.
If you keep nvidia installed but add "modprobe.blacklist=nvidia,nvidia_drm" to the kernel paramers can you
1. boot the multi-user.target
2. then there explicitly "modprobe nvidia_drm" w/o crashing the system?

I added

modprobe.blacklist=nvidia,nvidia_drm systemd.unit=multi-user.target

and it allowed me to reach the decryption and then user login in tty1

but the moment i ran

sudo modprobe nvidia_drm

my system crashed

brout · 2026-01-22 16:05:45

btw after the system crashes and starts to reboot, it reboots like 2-10 times (the fans spinning for 1-3 seconds each reboot) before being able to reach grub again. Not sure if this information helps indicate something

seth · 2026-01-22 19:50:04

There's going to be some surge or electric leak or something - does some live distro that ships the nvida drivers work?
Or OtherOS™ (windows)?

Maybe it helps to direct more power to the CPU, do you have control over such in the BIOS/UEFI (do not undervolt or overclock the CPU and disable XMP)?
Does adding

intel_idle.max_cstate=1 i915.enable_dc=0 ahci.mobile_lpm_policy=1 pcie_aspm=off

to the kernel parameters have any impact?

brout · 2026-01-23 03:32:08

seth wrote:

does some live distro that ships the nvida drivers work?
Or OtherOS™ (windows)?

No, actually not. I tested. Same shutdown issue with the nvidia drivers, regardless of OS (tested arch, fedora, windows 11. The crashes kept occuring on windows after I removed the "nvidia graphics driver" & "nvidia hd audio driver" in control panel, and plugged in the hdmi to the motherboard instead of the gpu, and set primary display to IGFX instead of PEG in bios)

seth wrote:

Maybe it helps to direct more power to the CPU, do you have control over such in the BIOS/UEFI (do not undervolt or overclock the CPU and disable XMP)?

The cpu is locked, so I don't think so. XMP was & is disabled.

seth wrote:

Does adding intel_idle.max_cstate=1 i915.enable_dc=0 ahci.mobile_lpm_policy=1 pcie_aspm=off to the kernel parameters have any impact?

No, without the "modprobe.blacklist=nvidia,nvidia_drm systemd.unit=multi-user.target" or "module_blacklist=nvidia" it currently crashes upon "Loading initial ramdisk" before being able to see the decrypt prompt, even with those parameters. I'm still on the same arch install as earlier.

There's going to be some surge or electric leak or something

I suppose. Really strange though. My PSU should be enough for my system. My system with dGPU running nouveau drivers only draws about 20-30 watts during regular desktop or browsing tasks, i tested it with a power station like a year ago. I find it kinda crazy that the NVIDIA drivers would spike that to the point of crashing the system.

But if I were to look at a new PSU, what would be the thing to look for, if output is pretty irrelevant?

seth · 2026-01-23 10:18:52

But if I were to look at a new PSU, what would be the thing to look for, if output is pretty irrelevant?

The crashes kept occuring on windows after I removed the "nvidia graphics driver" & "nvidia hd audio driver" in control panel, and plugged in the hdmi to the motherboard instead of the gpu, and set primary display to IGFX instead of PEG in bios

Can you trigger it w/ https://wiki.archlinux.org/title/Stress_testing#stress ?
Or the script in https://bbs.archlinux.org/viewtopic.php … 9#p2260219 ?

Arch Linux

#1 2026-01-22 01:26:29

[SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#2 2026-01-22 03:17:23

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#3 2026-01-22 11:44:43

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#4 2026-01-22 13:13:42

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#5 2026-01-22 13:59:55

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#6 2026-01-22 14:29:27

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#7 2026-01-22 14:31:31

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#8 2026-01-22 14:35:17

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#9 2026-01-22 14:38:01

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#10 2026-01-22 14:49:20

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#11 2026-01-22 15:02:44

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#12 2026-01-22 15:07:22

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#13 2026-01-22 15:13:21

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#14 2026-01-22 15:17:53

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#15 2026-01-22 15:22:03

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#16 2026-01-22 15:25:18

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#17 2026-01-22 15:37:05

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#18 2026-01-22 15:38:34

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#19 2026-01-22 15:44:11

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#20 2026-01-22 15:52:29

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#21 2026-01-22 15:58:42

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#22 2026-01-22 16:05:45

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#23 2026-01-22 19:50:04

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#24 2026-01-23 03:32:08

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

#25 2026-01-23 10:18:52

Re: [SOLVED] legacy nvidia drivers not working for me (keeps rebooting)

Board footer