jetson example (x11) provided by Balena breaking nvidia-smi

Hello,

I tried your example from here : jetson-examples/jetson-orin/Dockerfile at master · balena-io-examples/jetson-examples · GitHub

On my Nvidia Jetson Orin Nano 8GB (SD) Devkit NVME (BalenaOS version 6.4.0).

Here was my docker-compose :

x11:
  restart: always
  privileged: true
  build: 
    context: ./x11

Even if the screen is showing, I see the following error in the terminal:
image

And this is because of this “tegra_drm” module we are not able to “modprobe” that nvidia-smi is not possible anymore (and any GPU related tasks) as you can see below.

image

Of course, if I don’t rmmod tegra_drm, it can’t find the screen. So how can I solve this so that I’m able to execute nvidia-smi (or a cuda sample) while showing something on the screen ?

Edit : this is consequently a regression from this commit : Update Jetson Orin samples to Jetpack 6.2 - L4T 36.4.3 · balena-io-examples/jetson-examples@08e968f · GitHub

Hello, as you may know, nvidia-smi until recently was only available for discrete GPUs but is now available for Jetson devices with integrated GPUs since Jetpack 6.x. In the example repo you referenced, I was able to run nvidia-smi with the expected output:

root@4521c86:/# nvidia-smi
Thu Jul  3 18:55:53 2025       
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 540.4.0                Driver Version: 540.4.0      CUDA Version: 12.6     |
|-----------------------------------------+----------------------+----------------------+
| 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  Orin (nvgpu)                  N/A  | N/A              N/A |                  N/A |

Perhaps if you post your Dockerfile we can see where it differs from the example?