-Syu uninstalls Nvidia driver, nvidia-inst cannot satisfy dependencies

The last few times I have done an -Syu command, it appears to uninstall my Nvidia driver (GPU is a GTX 3050). When I try to use nvidia-inst to re-install the driver, I get the following error:

checking dependencies…
error: failed to prepare transaction (could not satisfy dependencies)
:: removing nvidia-utils breaks dependency ‘nvidia-utils’ required by lib32-nvidia-utils
:: removing nvidia-utils breaks dependency ‘nvidia-utils’ required by nvidia-open-dkms
:: removing nvidia-utils breaks dependency ‘nvidia-utils’ required by nvidia-settings
:: removing nvidia-utils breaks dependency ‘vulkan-driver’ required by steam
[abram@Endeavour-abram ~]$ nvidia inst nvidia-utils
bash: nvidia: command not found
[abram@Endeavour-abram ~]$ nvidia-inst nvidia-utils
2026-01-17 16:04:49: Info: nvidia-inst version 26.1.1-1
2026-01-17 16:04:49: Info: Command line: nvidia-inst nvidia-utils
2026-01-17 16:04:49: Info: Selected mode: nvidia (Nvidia’s open source)
2026-01-17 16:04:49: Info: Removing packages: nvidia-utils

COMMANDS TO RUN:
    pacman -Rs --noconfirm --noprogressbar --nodeps nvidia-utils

==> NOTE: running the commands may take several minutes…

checking dependencies…
error: failed to prepare transaction (could not satisfy dependencies)
:: removing nvidia-utils breaks dependency ‘nvidia-utils’ required by lib32-nvidia-utils
:: removing nvidia-utils breaks dependency ‘nvidia-utils’ required by nvidia-open-dkms
:: removing nvidia-utils breaks dependency ‘nvidia-utils’ required by nvidia-settings
:: removing nvidia-utils breaks dependency ‘vulkan-driver’ required by steam

Is there an easy work-around for this so I can re-install my driver? Also, is there a way to stop -Syu from uninstalling it every time?

In fact, nvidia-open-dkms is the NVidia driver. Are you sure that it doesn’t work with your hardware?

If I check graphical devices, it says I don’t have a driver:

Graphics:
Device-1: NVIDIA GA107 [GeForce RTX 3050 6GB] driver: N/A alternate: nouveau
non-free: 550-580.xx+ status: current (as of 2025-11; EOL~2026-12-xx)
arch: Ampere code: GAxxx process: TSMC n7 (7nm) built: 2020-2023 pcie:
gen: 4 speed: 16 GT/s lanes: 8 link-max: lanes: 16 bus-ID: 01:00.0
chip-ID: 10de:2584 class-ID: 0300

nvidia-open is the correct driver for your hardware.

Can we see what you have installed?

pacman -Q | grep nvidia

lib32-nvidia-utils 590.48.01-1
linux-firmware-nvidia 20260110-1
nvidia-hook 1.5.3-1
nvidia-inst 26.1.1-1
nvidia-open-dkms 590.48.01-2
nvidia-settings 590.48.01-1
nvidia-utils 590.48.01-2

What does cat /proc/cmdline show?

BOOT_IMAGE=/boot/vmlinuz-linux-lts root=UUID=67400523-3634-435b-b294-3f12a474d2e8 rw nowatchdog nvme_load=YES resume=UUID=56cd0eba-5a43-405a-9860-9148cb8f615f loglevel=3

Have you rebooted?

A few times, yes

Can you share journalctl -b --grep nvidia

Jan 17 16:25:14 user systemd-modules-load[248]: Failed to find module ‘nvidia_uvm’
Jan 17 16:25:16 user systemd-modules-load[642]: Failed to find module ‘nvidia_uvm’
Jan 17 16:25:17 user kernel: input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/00>
Jan 17 16:25:17 user kernel: input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/00>
Jan 17 16:25:17 user kernel: input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/00>
Jan 17 16:25:17 user kernel: input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/00>

That seems like maybe the dkms package didn’t apply properly.

Can you share the full output of sudo pacman -S nvidia-open-dkms

This is weird. Getting an error related to the pacman both before and after rebooting and trying the command:

error: failed to init transaction (unable to lock database)
error: could not lock database: File exists
if you’re sure a package manager is not already
running, you can remove /var/lib/pacman/db.lck

I tried to delete /var/lib/pacman/db.lck using the rm command but system is saying it doesn’t exist:

rm: cannot remove ‘/var/lib/pacman/db.lck’: No such file or directory

Tried -Syu again and the command when through without error. Used the sudo pacman -S nvidia-open-dkm command and package started to install, but took forever to install DKMS packages and my computer was screaming the whole time. Posting the install record before restart just in case there are issues:

(1/1) checking keys in keyring [--------------------------------] 100%
(1/1) checking package integrity [--------------------------------] 100%
(1/1) loading package files [--------------------------------] 100%
(1/1) checking for file conflicts [--------------------------------] 100%
:: Running pre-transaction hooks…
(1/1) Remove upgraded DKMS modules
==> dkms remove --no-depmod nvidia/590.48.01 -k 6.18.5-arch1-1
:: Processing package changes…
(1/1) reinstalling nvidia-open-dkms [--------------------------------] 100%
:: Running post-transaction hooks…
(1/5) Arming ConditionNeedsUpdate…
(2/5) Install DKMS modules
==> dkms install --no-depmod nvidia/590.48.01 -k 6.18.5-arch1-1
==> dkms install --no-depmod nvidia/590.48.01 -k 6.12.65-1-lts
==> depmod 6.12.65-1-lts
==> depmod 6.18.5-arch1-1
(3/5) Updating initramfs…
:: Building initramfs for linux-lts (6.12.65-1-lts)
:: Building fallback initramfs for linux-lts (6.12.65-1-lts)
:: Building initramfs for linux (6.18.5-arch1-1)
:: Building fallback initramfs for linux (6.18.5-arch1-1)
(4/5) Check if user should be informed about rebooting after certain system package upgrades.
(5/5) Checking which packages need to be rebuilt

Looks like it worked.

Hopefully all is working after the reboot.

Yep it worked! PC seems to be fine now. Is there anything I can be doing to make this less likely to happen in the future?

1 Like

I can’t say for sure but I would guess that something went wrong when the dkms modules were being built.

This is what it should look like:

No errors there.

joekamprad@KDEDVL$~ [Sa Jan 17 23:08]$ nvidia-inst nvidia-utils
2026-01-17 23:08:49: Info: nvidia-inst version 26.1.1-1
2026-01-17 23:08:49: Info: Command line: nvidia-inst nvidia-utils
2026-01-17 23:08:49: Info: Selected mode: nvidia (Nvidia's open source)
2026-01-17 23:08:49: Info: Installing packages: nvidia-hook
2026-01-17 23:08:49: Info: Removing packages: nvidia-utils vulkan-nouveau
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
COMMANDS TO RUN:
    pacman -Rs --noconfirm --noprogressbar --nodeps nvidia-utils vulkan-nouveau
    pacman -Syuq --noconfirm --noprogressbar --needed nvidia-hook
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

==> NOTE: running the commands may take several minutes...

[sudo] password for joekamprad:

seems like a BUG still ? @manuel

@ZarasGhost you do not run nvidia-inst to update the drivers, they will get updated with common package update over pacman or yay paru etc..

nvidia-inst takes only options, not package names. See nvidia-inst -h
But it should complain about that, will check.

If it’s working again now, then yes – this screams DKMS build failure during a kernel update.
That usually happens when the NVIDIA module didn’t build cleanly for the new kernel, but the system rebooted anyway. Result: broken graphics, black screen, random freezes – pick your poison.

How to make this less likely in the future:

  1. Install an LTS kernel and make it your default
    Seriously. Use LTS as your daily driver and treat the newest kernel as a test bench.
    With NVIDIA + rolling release, you really don’t want to be on the bleeding edge unless you enjoy debugging DKMS at 2 AM.

  2. Always check DKMS output after kernel updates
    After pacman -Syu, make sure DKMS and depmod ran without errors.
    If DKMS fails: do not reboot until it’s fixed. Rebooting into a kernel without a working NVIDIA module is just pain.

  3. Important: nvidia-inst is NOT a driver updater
    nvidia-inst takes options, not package names.
    Drivers are updated via normal system updates (pacman, yay, paru, etc.).
    Running nvidia-inst nvidia-utils is not the intended workflow – if it half-works, that’s a tool UX issue, not a feature.

  4. Keep a fallback kernel
    LTS as default, current kernel as “experimental”.
    If things break, boot LTS and move on with your life.


Personal NVIDIA trauma corner (so you’re not alone):
My GTX 950M is now officially e-waste for gaming under Linux.
The ASUS gaming laptop it lives in is no longer usable for modern games – not because the hardware suddenly broke, but because NVIDIA dropped proper support and kernel progress moved on.

I feel your pain.

At some point you have to accept reality:
NVIDIA + legacy hardware + rolling kernels = guaranteed headaches.

Blunt conclusion:
If you want your system to just work on Linux:
buy AMD or Intel next time and enjoy kernel updates without fear.

NVIDIA works… until it doesn’t.
And when it doesn’t, it fails spectacularly.