No graphics in aarch64 (SolidRun Honeycomb LX2K) Ubuntu 22.04 using NVidia drivers

NVIDIA Open GPU Kernel Modules Version

515.76

Does this happen with the proprietary driver (of the same version) as well?

Yes

Operating System and Version

Ubuntu 22.04.1 LTS

Kernel Release

5.15.0-50-generic #56-Ubuntu SMP Tue Sep 20 13:23:50 UTC 2022 aarch64

Hardware: GPU

GPU 0: NVIDIA GeForce GTX 1660 Ti

Describe the bug

I have a SolidRun Honeycomb LX2K ARM64 system and recently installed a GTX1660Ti GPU on it. I get no video on Desktop (monitor stays blank as there is no signal to it).

The initial install of Ubuntu 22.04 worked fine due to it's use of Noveau drivers. I then installed the NVidia drivers (stock) and also open drivers and lost the GUI (Gnome).

The install steps I took are listed below:

• Downloaded 515.76 drivers from https://www.nvidia.com/Download/driverResults.aspx/193096/en-us/

• Installed it without Kernel modules (sudo ./NVIDIA-Linux-aarch64-515.76.run --no-kernel-modules)

• Cloned this repo and built according to instructions (make modules/make modules_install). Got this error in the modules_install but the drivers were installed. Build the driver with the patch I got after talking to Solidrun support https://gist.github.com/jnettlet/226c0787ae9744082493accc512a9a6b.

• Blacklisted Noveau (otherwise Ubuntu would boot with it)

• On first boot, I saw these errors in Kernel logs:

10.0.20.203: kern: warning: [2022-09-02T19:03:16.922855838Z]: NVRM: Open nvidia.ko is only ready for use on Data Center GPUs.
10.0.20.203: kern: warning: [2022-09-02T19:03:16.923935838Z]: NVRM: To force use of Open nvidia.ko on other GPUs, see the
10.0.20.203: kern: warning: [2022-09-02T19:03:16.924966838Z]: NVRM: 'OpenRmEnableUnsupportedGpus' kernel module parameter described
10.0.20.203: kern: warning: [2022-09-02T19:03:16.926121838Z]: NVRM: in the README.
10.0.20.203: kern: warning: [2022-09-02T19:03:17.235131838Z]: NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x63:0x0:1905)
10.0.20.203: kern: warning: [2022-09-02T19:03:17.236812838Z]: NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number

Also nvidia-smi showed a message that no devices were found.

• Created /etc/modprobe.d/nvidia.conf and added options nvidia NVreg_OpenRmEnableUnsupportedGpus=1 to it. Run sudo update-initramfs -u and rebooted.

• Linux detected the GPU bus still no graphics (nothing, my monitor stays in standby as before like there is no signal to it). nvidia-smi shows:

I could successfully run some ML workloads like Pytorch but no graphics. I tried each and every setting to X11 I found with no success until now.

Sometimes I see these messages in Kernel logs:

[ 1741.048380] [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00040100] Flip event timeout on head 0
[ 1741.252831] nvidia-modeset: ERROR: GPU:0: Timed out waiting for core channel idle.

To Reproduce

Steps in the description above.

Bug Incidence

Always

nvidia-bug-report.log.gz

nvidia-bug-report.log.gz

More Info

No response

Start by reviewing the attached nvidia-bug-report.log.gz and the kernel errors from nvidia.ko and nvidia_drm, then inspect the module configuration in /etc/modprobe.d/nvidia.conf. Reproduce the Ubuntu 22.04 aarch64 setup with the GTX 1660 Ti and determine why the display remains blank and flip events time out; done means the desktop produces video output without those failures.