Nvidia 550 drivers causing hard crashing during updates

Just to confirm, you cannot downgrade and run 545 drivers with the mainline kernel, I did a few days ago and had to chroot in and revert.

Have not tried the LTS kernel.

1 Like

nvidia driver is kernel agnostic, if you downgrade you need to downgrade the kernel too to the same used to build the driver packages.

anyone tried this?

  • Add zswap.enabled=0 and numa=off to the kernel parameter
  • Changed the nvidia to nvidia-open-dkms

*) NUMA (non-uniform memory access)

I would expect that switching drivers would fix it but for anyone who needs the proprietary driver, that probably isnā€™t an option.

No clue what exactly will not work using open modules ?

How does one select the open source driver instead of normal though? I used nvidia-inst -o to install the open source driver and I added those kernel parameters but upon reboot I get the same freeze. Is that not how you install it?

That is one way to do it. Alternatively it can be installed manually with
sudo pacman -S nvidia-open-dkms

you will also need to rebuild initramfs images ā€¦

I got the same problem (I fixed it later)
I tried to reinstall nvidia drivers (mine was on version 550.78), I saw that it was not installed correctly and I got problem with dracut[E]: Module ā€˜plymouthā€™ cannot be installed. ( I didnt remove Plymouth correctly from Dracut configuration :slight_smile: :broken_heart: ), then I rebuilt initramfs images
then reinstalling kernel and then reinstalling nvidia drivers.

[edit] still the same problem Iā€™ll try downgrade it to open-dkms 545

I was facing lot of issues and random freezes with Nvidia 550 driver on my Asus TUF A15 Laptop with AMD Ryzen 7 and Nvidia RTX 4050. Switching to Nvidia 535 driver and LTS kernel has resolved all issues.

Here is how I downgraded Nvidia driver from 550 to 535:

pacman -R nvidia-dkms nvidia-settings nvidia-utils 
yay -S nvidia-535xx-dkms nvidia-535xx-utils nvidia-535xx-settings
3 Likes

Hi All - I am on a Thinkpad P50, and I canā€™t remember which one, but one nvidia update completely broke my device. I had to do the --overwrite ā€œ*ā€ with Pacman, and then my system kept freezing after a more recent update (like others have reported). I am on KDE+X11 - wayland just completely freezes and is unusable.

I tried downgrading to 535, but somehow the installation would fail from Yay with the ā€œmissing headersā€ error. Anyhow, so I went and installed nouveau. On reboot, sddm wouldnā€™t start with nouveau. Unsure if itā€™s because I have nvidia modeset kernet parameter set.

Fast forward - I was too lazy to fix nouveau. So I installed the latest nvidia from Pacman (550.78-6) on the latest Kernel (6.9.2). And guess what? My system suddenly feels perfectly fine. It feels way faster (it was very sluggish - which was a surprise - I didnā€™t realize how much this bug was affecting my system). System now seems to be back to what it used to be.

21h up, no freezes so far (I was getting freezes every 30 minutes to an hour or so).

So if anyone is wondering if thereā€™s a TL;DR: Try installing nouveau and then rebooting and then reinstalling latest Nvidia (all this is on arch btw).

Iā€™m not seeing this behaviour on my GTX 1050Ti card running the latest EOS build.

It looks like I am running in X11.

Got it yesterday on shutdown with the new 555 closed source driver.

Arch Linux with XFCE on Xorg.

kernel: BUG: unable to handle page fault for address: 000000000003f249
kernel: #PF: supervisor read access in kernel mode
kernel: #PF: error_code(0x0000) - not-present page
kernel: PGD 0 P4D 0 
kernel: Oops: 0000 [#2] PREEMPT SMP NOPTI
kernel: CPU: 1 PID: 4184 Comm: chromiu:disk$3 Tainted: P S   UD    OE      6.9.7-zen1-1.1-zen #1 ada401ad7257f75841c37008401a305b20155156
kernel: Hardware name: LENOVO 20FW003PGE/20FW003PGE, BIOS R07ET97W (2.37 ) 11/25/2022
kernel: RIP: 0010:refill_obj_stock+0x56/0x1c0
kernel: Code: 81 e4 00 02 00 00 48 8d a8 00 20 03 00 48 8b 45 10 48 39 f8 0f 84 94 00 00 00 48 89 ef e8 12 f0 ff ff 49 89 c6 e8 da 72 d0 ff <49> 8b 07 a8 03 0f 85 fb 00 00 00 65 >
kernel: RSP: 0018:ffffa2b74b3337e0 EFLAGS: 00010002
-- Boot 6b1a6a9c7815461da6d831f863d11604 --

1 month later
image
May open a new thread you can link to this one.

2 Likes