When adding nodes to a nodepool in a dual-stack HCP scenario, nodes doesn't get the CSR automatically approved and even if manually approved the nodes has a taint to avoid pods running.
Checking "kubevirt-cloud-controller-manager" pod messages like these appears:
E0115 15:09:42.974850       1 node_controller.go:236] error syncing 'temp-xjl6k': failed to get node modifiers from cloud provider: provided node ip for node "temp-xjl6k" is not valid: failed to parse node IP "10.x.x.x,2001:x:x:x:x:x:x:x": dual-stack not supported in this configuration, requeuing

The node object clearly states the annotation:
    alpha.kubernetes.io/provided-node-ip: 10.x.x.x,2001:x:x:x:x:x:x:x

Checking the error log in the code i manage to get to the function in:
https://github.com/openshift/cloud-provider-kubevirt/blob/3f4542ecd17fb0e47da4c6d9bceb076b98fb314b/vendor/k8s.io/component-helpers/node/util/ips.go#L81

which dualStack activation depends on a featureGate called "CloudDualStackNodeIPs" [1] which is not present anymore in openshift in 4.19 and onwards. This makes the verification to always fail in dual stack environments and holds the nodes from being Ready on the cluster.

In recent versions of this kubernetes package the dualstack variable is set to True [2] always.

[1] https://github.com/openshift/cloud-provider-kubevirt/blob/3f4542ecd17fb0e47da4c6d9bceb076b98fb314b/vendor/k8s.io/cloud-provider/controllers/node/node_controller.go#L744

[2]  https://github.com/openshift/cloud-provider-kubevirt/blob/3f4542ecd17fb0e47da4c6d9bceb076b98fb314b/vendor/k8s.io/component-helpers/node/util/ips.go#L81

Hosting cluster: 4.19.14
Hosted cluster: 4.20.10

Always

    1. Install a Hub cluster with dual stack
    2. Install HCP cluster with kubevirt
    3. Add nodes to nodepool
    4. monitor "kubevirt-cloud-controller-manager" pod logs

Nodes cannot be added to the hosted cluster

"kubevirt-cloud-controller-manager" doesn't fail the verification and nodes are added as expected.

As a possible workaround we can pause the HCP reconciliation by adding:
  pausedUntil: "true"
in the spec of the hostedCluster and then modify the deployment "kubevirt-cloud-controller-manager" by adding the feature-gate parameter, eg.:

...
    containers:
    - args:
      - --cloud-provider=kubevirt
      - --cloud-config=/etc/cloud/cloud-config
      - --kubeconfig=/etc/kubernetes/kubeconfig/kubeconfig
      - --authentication-skip-lookup
      - --cluster-name=tc-temp
      - --feature-gates CloudDualStackNodeIPs=true
      command:
      - /bin/kubevirt-cloud-controller-manager
...