Alma 10.2 + Nvidia Driver + RTX 3050 = stuck at 800x600 resolution

General Discussion & Chat
1

Hardware:

Minisforum MS-01 with i5-12600H

Yeston low profile RTX 3050

Problem:

After installing the Nvidia drivers on a fresh install of Alama 10.2, the desktop is stuck at 800x600 resolution. No other resolution is presented in Gnome display settings.

Before the Nvidia drivers were installed, the desktop ran fine at 1080 resolution.

Steps to Reproduce:

  1. Install the Nvidia drivers per the wiki:
  1. Reboot.

Troubleshooting:

Nvidia driver loads succesfully

$ lsmod |grep -i nvidia
nvidia_uvm 2498560 0
nvidia_drm 163840 0
nvidia_modeset 1925120 2 nvidia_drm
nvidia 17960960 65 nvidia_uvm,nvidia_modeset
drm_ttm_helper 16384 1 nvidia_drm
video 81920 1 nvidia_modeset

Nvidia-smi output looks normal to me

$ nvidia-smi
Wed May 27 15:51:59 2026
Ā±----------------------------------------------------------------------------------------+
| NVIDIA-SMI 610.43.02 KMD Version: 610.43.02 CUDA UMD Version: 13.3 |
Ā±----------------------------------------Ā±-----------------------Ā±---------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 3050 Off | 00000000:01:00.0 On | N/A |
|100% 35C P8 8W / 70W | 16MiB / 6144MiB | 0% Default |
| | | N/A |
Ā±----------------------------------------Ā±-----------------------Ā±---------------------+

Ā±----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 2838 C+G /usr/bin/gnome-control-center 5MiB |
Ā±----------------------------------------------------------------------------------------+

Modeset appears to be properly enabled:

$ sudo cat /sys/module/nvidia_drm/parameters/modeset
Y

Now this might be the cause:

[ 2.310565] nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64 610.43.02 Release Build (mockbuild@) Wed May 27 08:26:33 UTC 2026
[ 2.311959] [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
[ 2.312066] nvidia 0000:01:00.0: Direct firmware load for nvidia/610.43.02/gsp_ga10x.bin failed with error -2
[ 2.312069] NVRM: RmFetchGspRmImages: No firmware image found
[ 2.312071] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x61:0x56:2025)
[ 2.312073] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0
[ 2.312119] [drm:nv_drm_dev_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000100] Failed to allocate NvKmsKapiDevice

However the firmware file appears to be properly installed on my system

$ ls -lh /usr/lib/firmware/nvidia/610.43.02/gsp_ga10x.bin
-rw-rā€“rā€“. 1 root root 81M May 19 05:20 /usr/lib/firmware/nvidia/610.43.02/gsp_ga10x.bin

I am at a loss as to why the firmware binary cannot load.

Additional Information

If I completely uninstall the Alama Nvidia driver, and then install akmod-nvidia from rpmfusion-nonfree, everything works as expected.

However, the rpmfusion driver is the older 580 series. From the Nvidia website, the RTX 3050 should work with the latest 595 driver.

If all else fails, Iā€™ll just stick with the older driver in rpmfusion, but I thought Iā€™d ask here first, in case someone has any ideas on how to troubleshoot this further.

2

According to the ā€œNVIDIA Driver Packagesā€ page on the AlmaLinux Wiki, AlmaLinux includes NVIDIA driver packages based on NVIDIAā€™s open-source drivers.

These packages are Secure Boot-signed and built using kernel kmods, so it appears that DKMS compilation is not required during installation or updates.

Reference:

The installation commands listed in the Wiki are as follows:

sudo dnf install almalinux-release-nvidia-driver
sudo dnf install nvidia-open-kmod nvidia-driver

In this case, the NVIDIA module appears to be loaded, but the following error occurs:

Direct firmware load for nvidia/610.43.02/gsp_ga10x.bin failed with error -2

Since the firmware file is located under /usr/lib/firmware, I would check whether it is included in the initramfs:

lsinitrd /boot/initramfs-$(uname -r).img | grep -F 'nvidia/610.43.02/gsp_ga10x.bin'

If it is not present, it is possible that nvidia-drm was initialized too early for KMS before the firmware became available from the root filesystem.

As a test and workaround:

sudo tee /etc/dracut.conf.d/nvidia-gsp.conf >/dev/null <<'EOF'
install_items+=" /usr/lib/firmware/nvidia/610.43.02/gsp_ga10x.bin "
EOF

sudo dracut -f /boot/initramfs-$(uname -r).img $(uname -r)
sudo reboot

After rebooting, please check the following:

journalctl -b -k | grep -Ei 'nvidia|gsp|firmware|NvKms|RmInitAdapter'
cat /sys/module/nvidia_drm/parameters/modeset
xrandr
nvidia-smi

Thank you.

3

Thank you for the response.

The binary firmware was not present in intird. The following returned nothing:

$ lsinitrd /boot/initramfs-$(uname -r).img | grep -F ā€˜nvidia/610.43.02/gsp_ga10x.binā€™

Creating nvidia-gsp.conf and rebuilding the initrd image did indeed solve the problem.

The desktop functions normally now.

Below is the requested output. Note that xrandr is no longer available, likely because it does not work in wayland.

$ journalctl -b -k | grep -Ei ā€˜nvidia|gsp|firmware|NvKms|RmInitAdapterā€™

May 28 07:56:43 mythtv-lab kernel: Loaded X.509 cert ā€˜Nvidia GPU OOT signing 001: 55e1cef88193e60419f0b0ec379c49f77545acf0ā€™
May 28 07:56:43 mythtv-lab kernel: Loaded X.509 cert ā€˜NVIDIA Jetson SoC OOT signing 001: 7e60d156a289e05b18f70ac4a2d8296c5c8ab797ā€™
May 28 07:56:43 mythtv-lab kernel: Loaded X.509 cert ā€˜NVIDIA Bluefield DPU OOT signing 001: e251cec14b15b5d4c0e452770ca269261ea466bcā€™
May 28 07:56:43 mythtv-lab kernel: Loaded X.509 cert ā€˜AlmaLinux NVIDIA Module Signing: 4a47544e27f990e063664d7d0dd9c2158954567aā€™
May 28 07:56:45 mythtv-lab systemd[1]: systemd-hibernate-clear.service - Clear Stale Hibernate Storage Info was skipped because of an unmet condition check (ConditionPathExists=/sys/firmware/efi/efivars/HibernateLocation-8cf2644b-4b0b-428f-9387-6d876050dc67).
May 28 07:56:45 mythtv-lab kernel: nvidia: loading out-of-tree module taints kernel.
May 28 07:56:45 mythtv-lab kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 235
May 28 07:56:45 mythtv-lab kernel: nvidia 0000:01:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=none
May 28 07:56:45 mythtv-lab kernel: NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64 610.43.02 Release Build (mockbuild@) Wed May 27 08:26:47 UTC 2026
May 28 07:56:45 mythtv-lab kernel: input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input15
May 28 07:56:45 mythtv-lab kernel: input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input16
May 28 07:56:45 mythtv-lab kernel: input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input17
May 28 07:56:45 mythtv-lab kernel: input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input18
May 28 07:56:45 mythtv-lab kernel: nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64 610.43.02 Release Build (mockbuild@) Wed May 27 08:26:33 UTC 2026
May 28 07:56:45 mythtv-lab kernel: [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
May 28 07:56:45 mythtv-lab kernel: mt7921e 0000:5a:00.0: WM Firmware Version: ____000000, Build Time: 20260224103233
May 28 07:56:47 mythtv-lab kernel: [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 1
May 28 07:56:47 mythtv-lab kernel: nvidia 0000:01:00.0: vgaarb: deactivate vga console
May 28 07:56:47 mythtv-lab kernel: fbcon: nvidia-drmdrmfb (fb0) is primary device
May 28 07:56:47 mythtv-lab kernel: nvidia 0000:01:00.0: [drm] fb0: nvidia-drmdrmfb frame buffer device

I donā€™t have a need for secure boot, but I assume this workaround would not work in that environment due to the tainted kernel warning, which is now present.

$cat /sys/module/nvidia_drm/parameters/modeset
Y

$ nvidia-smi
Thu May 28 08:24:20 2026
Ā±----------------------------------------------------------------------------------------+
| NVIDIA-SMI 610.43.02 KMD Version: 610.43.02 CUDA UMD Version: 13.3 |
Ā±----------------------------------------Ā±-----------------------Ā±---------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 3050 On | 00000000:01:00.0 On | N/A |
|100% 35C P8 10W / 70W | 107MiB / 6144MiB | 0% Default |
| | | N/A |
Ā±----------------------------------------Ā±-----------------------Ā±---------------------+

Ā±----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 1957 G /usr/bin/gnome-shell 77MiB |
| 0 N/A N/A 2373 G /usr/bin/Xwayland 2MiB |
Ā±----------------------------------------------------------------------------------------+

Should I report this at bugs.almalinux.org?

4

The kernel module built by AlmaLinux is signed with same certificate as the rest of AlmaLinux kernel and therefore loads like the rest of the kernel.

If one rather uses package from NVidiaā€™s repo, then one in practice builds the kernel module self, with dkms. The dkms does autogenerate a certificate on first use and does then use it to sign the modules it builds. One can import that certificate to UEFI and then Secure Boot does allow loading the built modules. The import starts with:

mokutil --import /var/lib/dkms/nvidia/mok.pub
5

Same thing happend to me..it should work out of the box but it seems kernel modules failed to load after reboot..so i did: dnf install kernel-modules-$(uname -r) kernel-modules-extra-$(uname -r) -y

and then after: echo video | sudo tee /etc/modules-load.d/video.conf
sudo grubby --update-kernel=ALL --args=ā€œnvidia-drm.modeset=1 rd.driver.blacklist=nouveau modprobe.blacklist=nouveauā€
sudo dracut -f --kver ā€œ$(uname -r)ā€
sudo reboot and after reboot drivers were up and running..

6

Note: the package kernel should pull in module packages:

# dnf -q rq --requires --resolve kernel | grep kernel
kernel-core-0:6.12.0-211.7.3.el10_2.x86_64_v2
kernel-modules-0:6.12.0-211.7.3.el10_2.x86_64_v2
kernel-modules-core-0:6.12.0-211.7.3.el10_2.x86_64_v2
kernel-modules-extra-0:6.12.0-211.7.3.el10_2.x86_64_v2

while the kernel-core does not:

# dnf -q rq --requires --resolve kernel | grep kernel
### kernel-core ###
kernel-modules-core-0:6.12.0-211.7.3.el10_2.x86_64_v2

(I have heard of installs, where the package kernel was not there ā€“ actual kernel is in kernel-core.)

On minimal system (no NVidia) the ā€˜videoā€™ goes into initramfs:

# lsinitrd | grep video.ko
-rw-r--r--   1 root     root        20792 Jan 30 02:00 usr/lib/modules/6.12.0-211.7.3.el10_2.x86_64_v2/kernel/drivers/acpi/video.ko.xz

Why would the NVidia packages block that?

1 Like
7

Well i dont know, i am no expert:D i did one more reinstall just to make sure i f i did something wrong but not..this time kernel modules were loaded..but still it boots in low resolution. I saw drm error message on boot. And without this commands: echo video | sudo tee /etc/modules-load.d/video.conf
sudo grubby --update-kernel=ALL --args=ā€œnvidia-drm.modeset=1 rd.driver.blacklist=nouveau modprobe.blacklist=nouveauā€
sudo dracut -f --kver ā€œ$(uname -r)ā€ i wasnt able to make nvidia working..nvidia-smi shows gpu but it dosent bring me in native resolution. I firstly used instructions from: AlmaLinux OS - Forever-Free Enterprise-Grade Operating System