Help with Fedora 42 and 5090 drivers

Ask Fedora
1

Hello everyone,

I installed Fedora 42 on my new Lenovo Legion pro 7 with 5090. Now I’ve a problems with Nvidia’s drivers loading.

I’ve tried a lot of things including rebuild modules with akmods and dracut, login with Wayland or X11, disable security boot, but nothing works.

So, any help will be appreciated.

nvidia-smi
No devices were found

sudo modinfo nvidia
filename: /lib/modules/6.14.9-300.fc42.x86_64/extra/nvidia/nvidia.ko
import_ns: DMA_BUF
alias: char-major-195-*
version: 575.57.08
supported: external
license: Dual MIT/GPL
firmware: nvidia/575.57.08/gsp_tu10x.bin
firmware: nvidia/575.57.08/gsp_ga10x.bin
srcversion: AA6B6D25AE218AFF04CB762
alias: pci:v000010DEdsvsdbc06sc80i00
alias: pci:v000010DEdsvsdbc03sc02i00
alias: pci:v000010DEdsvsdbc03sc00i00
alias: of:NTCnvidia,tegra234-displayC*
alias: of:NTCnvidia,tegra234-display
depends:
name: nvidia
retpoline: Y
vermagic: 6.14.9-300.fc42.x86_64 SMP preempt mod_unload

sudo dmesg | grep -i nvidia
[ 0.000000] Command line: BOOT_IMAGE=(hd0,gpt2)/vmlinuz-6.14.9-300.fc42.x86_64 root=/dev/mapper/fedora–01-root ro rd.lvm.lv=fedora-01/root rd.luks.uuid=luks-0b1c3a25-5c66-4f3a-9684-6d8b489ab123 rd.driver.blacklist=nouveau,nova_core modprobe.blacklist=nouveau,nova_core nvidia-drm.modeset=1
[ 0.069275] Kernel command line: BOOT_IMAGE=(hd0,gpt2)/vmlinuz-6.14.9-300.fc42.x86_64 root=/dev/mapper/fedora–01-root ro rd.lvm.lv=fedora-01/root rd.luks.uuid=luks-0b1c3a25-5c66-4f3a-9684-6d8b489ab123 rd.driver.blacklist=nouveau,nova_core modprobe.blacklist=nouveau,nova_core nvidia-drm.modeset=1
[ 143.185648] nvidia: loading out-of-tree module taints kernel.
[ 143.186860] nvidia: module verification failed: signature and/or required key missing - tainting kernel
[ 143.303870] nvidia-nvlink: Nvlink Core is being initialized, major device number 509
[ 143.308540] nvidia 0000:02:00.0: enabling device (0000 → 0003)
[ 143.310456] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=none
[ 143.322404] NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64 575.57.08 Release Build (dvs-builder@U22-I3-H04-01-5) Sat May 24 07:03:13 UTC 2025
[ 143.468644] nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64 575.57.08 Release Build (dvs-builder@U22-I3-H04-01-5) Sat May 24 06:53:21 UTC 2025
[ 143.471163] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver
[ 144.206290] nvidia 0000:02:00.0: Enabling HDA controller
[ 144.233111] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:06.0/0000:02:00.1/sound/card0/input17
[ 144.233766] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:06.0/0000:02:00.1/sound/card0/input18
[ 146.500482] [drm:nv_drm_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000200] Failed to allocate NvKmsKapiDevice
[ 146.501950] [drm:nv_drm_register_drm_device [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000200] Failed to register device
[ 147.382461] nvidia 0000:02:00.0: Enabling HDA controller
[ 150.518403] nvidia 0000:02:00.0: Enabling HDA controller
[ 238.558273] nvidia 0000:02:00.0: Enabling HDA controller

3

Thanks @karlsruhe.
I tried it but it doesn’t help. nvidia-smi still can’t detect to card. I see in you case it’s 3060. I’ve RTX 5090 that is pretty new.
Btw, I forgot to mention that I use KDE and I’ve tried both Wayland and X11
I still have same errors on device registration

[ 1191.363021] nvidia: loading out-of-tree module taints kernel.
[ 1191.364201] nvidia: module verification failed: signature and/or required key missing - tainting kernel
[ 1191.495000] nvidia-nvlink: Nvlink Core is being initialized, major device number 509
[ 1191.498866] nvidia 0000:02:00.0: enabling device (0000 → 0003)
[ 1191.499057] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=none
[ 1191.510369] NVRM: loading NVIDIA UNIX Open Kernel Module for x86_64 575.57.08 Release Build (dvs-builder@U22-I3-H04-01-5) Sat May 24 07:03:13 UTC 2025
[ 1191.645135] nvidia-modeset: Loading NVIDIA UNIX Open Kernel Mode Setting Driver for x86_64 575.57.08 Release Build (dvs-builder@U22-I3-H04-01-5) Sat May 24 06:53:21 UTC 2025
[ 1191.647675] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver
[ 1192.391416] nvidia 0000:02:00.0: Enabling HDA controller
[ 1192.414031] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:06.0/0000:02:00.1/sound/card0/input17
[ 1192.414959] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:06.0/0000:02:00.1/sound/card0/input18
[ 1194.692275] [drm:nv_drm_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000200] Failed to allocate NvKmsKapiDevice
[ 1194.693826] [drm:nv_drm_register_drm_device [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000200] Failed to register device
[ 1195.574660] nvidia 0000:02:00.0: Enabling HDA controller

5

https://forums.developer.nvidia.com/t/failed-to-allocate-nvkmskapidevice-failed-to-register-device-geforce-rtx-3070-ubuntu-18-04-6/323492/6 also a report for 5090, multiple distros including Fedora. 4 reports also have the nvidia: module verification failed: signature and/or required key missing .

1 Like
6

My security boot is disabled, so I don’t really need akmod certificate enrolled

mokutil --sb-state
SecureBoot disabled

7

Sounds like we’re stucked with 5090 :frowning:

8

download the 570.153 driver from koji.rpmfusion.org and try that one instead.
You have to set a rpm macro to build the open kernel modules

closed source modules don’t support the 50* GPUs

1 Like
9

Thank @anotheruser . Do you know that 570.153 supports 50* ?

10

yes, 570.* is still the stable driver branch.
Driver Details | NVIDIA

11

@anotheruser I can’t find akmod-nvidia-570xx in koji.rpmfusion.org.
Do you have any idea what is going on there? or it just me?

12

:slight_smile:

before downgrading to 570.x maybe try 575.64 first. It’s available in rpmfusion-nonfree-updates-testing

13

Thanks @anotheruser, indeed it’s just me :slight_smile:

OK, I downgraded it to 570, but I get the same problem

~$ nvidia-smi
No devices were found

~$ modinfo nvidia
filename: /lib/modules/6.14.11-300.fc42.x86_64/extra/nvidia/nvidia.ko
alias: char-major-195-*
version: 570.153.02
supported: external
license: NVIDIA
firmware: nvidia/570.153.02/gsp_tu10x.bin
firmware: nvidia/570.153.02/gsp_ga10x.bin
srcversion: 670F9066D05CD26313CF34E
alias: pci:v000010DEdsvsdbc06sc80i00
alias: pci:v000010DEdsvsdbc03sc02i00
alias: pci:v000010DEdsvsdbc03sc00i00
depends:
name: nvidia
retpoline: Y
vermagic: 6.14.11-300.fc42.x86_64 SMP preempt mod_unload

from dmesg:

[ 0.068947] Kernel command line: BOOT_IMAGE=(hd0,gpt2)/vmlinuz-6.14.11-300.fc42.x86_64 root=/dev/mapper/fedora–01-root ro rd.lvm.lv=fedora-01/root rd.luks.uuid=luks-0b1c3a25-5c66-4f3a-9684-6d8b489ab123 nvidia-drm.modeset=1 rd.driver.blacklist=nouveau modprobe.blacklist=nouveau
[ 50.036157] nvidia: loading out-of-tree module taints kernel.
[ 50.037271] nvidia: module license ‘NVIDIA’ taints kernel.
[ 50.040962] nvidia: module verification failed: signature and/or required key missing - tainting kernel
[ 50.042671] nvidia: module license taints kernel.
[ 50.207710] nvidia-nvlink: Nvlink Core is being initialized, major device number 509
[ 50.210223] nvidia 0000:02:00.0: enabling device (0000 → 0003)
[ 50.211274] nvidia 0000:02:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=none
[ 50.223232] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 570.153.02 Tue May 13 16:34:43 UTC 2025
[ 50.227226] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:06.0/0000:02:00.1/sound/card0/input17
[ 50.228220] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:06.0/0000:02:00.1/sound/card0/input18
[ 50.233213] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
[ 50.315324] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 570.153.02 Tue May 13 16:21:23 UTC 2025
[ 50.317449] [drm] [nvidia-drm] [GPU ID 0x00000200] Loading driver
[ 50.367602] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2c58)
NVRM: installed in this system requires use of the NVIDIA open kernel modules.
[ 50.370774] [drm:nv_drm_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000200] Failed to allocate NvKmsKapiDevice
[ 50.373651] [drm:nv_drm_register_drm_device [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000200] Failed to register device
[ 60.293484] nvidia 0000:02:00.0: Enabling HDA controller
[ 75.230798] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2c58)
NVRM: installed in this system requires use of the NVIDIA open kernel modules.

14

Note the last line in your log above:

[ 75.230798] NVRM: The NVIDIA GPU 0000:02:00.0 (PCI ID: 10de:2c58)
NVRM: installed in this system requires use of the NVIDIA open kernel modules.

I think that the driver packages < 575 don’t autodetect whether the card requires the open-kernel driver.

The 50-series cards only work with the open-kernel driver, so when using them with a driver version < 575, you’ll have to set the flag manually to get the open-kernel flavour:

sudo sh -c 'echo "%_with_kmod_nvidia_open 1" > /etc/rpm/macros.nvidia-kmod'
sudo akmods --kernels $(uname -r) --rebuild

(Source: RPMFusion docs, and also mentioned earlier in the thread by @anotheruser )

15

Finally, I managed to install 570.153, and my 5090 is now found by nvidia-smi.
Honestly, I think it’s a miracle because it required so many tricks and tweaks that I’m not sure I know how to reproduce it :slight_smile:
Nothing straightforward, every step required a massage. I don’t even want to think of how to upgrade later on to the newest driver.
I am very disappointed with Nvidia’s driver quality level and support.
Thank you all guys for your help and support

1 Like
16

You made the mistake of buying the latest hardware to soon, it always takes around six months for new support to become stable.
This applies equally to any manufacturer, new AMD or intel GPU support is equally poor.

1 Like