[Issue]: Update DevceConfig driver version to supported driver version for each release

[LaVLaS]

Problem Description

When specifying a bare minimum DeviceConfig on OpenShift that enables the driver but does not specify the version, the operator will set .spec.driver.version = 6.2.2. This is no longer a supported version for building on OCP v4.20.z clusters (k8s v1.33) causing the driver will fail to build due to the

The recommendation is to update the default driver version to match what is the currently supported version for each release unless you want to make the driver.version mandatory

$ oc version
Server Version: 4.20.8
Kubernetes Version: v1.33.6

Operating System

OCP 4.20.8

CPU

N/A

GPU

N/A

ROCm Version

6.2.2

ROCm Component

No response

Steps to Reproduce

1. Deploy a DeviceConfig that does not specify .spec.driver.version

apiVersion: amd.com/v1alpha1
kind: DeviceConfig
metadata:
  namespace: openshift-amd-gpu
  name: test-deviceconfig
spec:
  devicePlugin:
    devicePluginImage: docker.io/rocm/k8s-device-plugin:latest
    nodeLabellerImage: docker.io/rocm/k8s-device-plugin:labeller-latest
  driver:
    enable: true
    #version: 6.2.2  #<--- Not required for successful DeviceConfig deployment
  selector:
    feature.node.kubernetes.io/amd-gpu: 'true'

2. Once deployed, the module build will pass DRIVERS_VERSION=6.2.2 as the KMM build argument.

test-deviceconfig-build-wnj8d-build-docker-build.log

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

No response

Additional Information

No response

[yansun1996]

Thanks for reporting this behavior @LaVLaS , currently we have documented in most of all sections to indicate users to put a specific driver version value in spec.driver.version so that the driver version is under their control.

We will push the fix for the default driver version and pick up the fix in the next major release.