Runtime Error with AMD GPU Helm Chart Installation in Kubernetes

I am experiencing a runtime error while trying to install the AMD GPU Helm Chart (link: AMD GPU Helm Chart) in my Kubernetes cluster. The pod spawned by the daemonset fails to run, and the error log indicates a segmentation violation (SIGSEGV), perhaps related to some permission issues?

I0113 15:54:58.192567       1 main.go:305] AMD GPU device plugin for Kubernetes
I0113 15:54:58.192606       1 main.go:305] ./k8s-device-plugin version v1.18.1-27-g5eb0a0f
I0113 15:54:58.192608       1 main.go:305] hwloc: _VERSION: 2.9.2, _API_VERSION: 0x00020800, _COMPONENT_ABI: 7, Runtime: 0x00020800
I0113 15:54:58.192613       1 manager.go:42] Starting device plugin manager
I0113 15:54:58.192617       1 manager.go:46] Registering for system signal notifications
I0113 15:54:58.192738       1 manager.go:52] Registering for notifications of filesystem changes in device plugin directory
panic: runtime error: invalid memory address or nil pointer dereference
    panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x38 pc=0x5333f9]

goroutine 1 [running]:
github.com/fsnotify/fsnotify.(*Watcher).Close(0x0)
    /go/src/github.com/RadeonOpenCompute/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:75 +0x19
panic({0x8e5fc0, 0xd3af30})
    /usr/local/go/src/runtime/panic.go:884 +0x212
github.com/fsnotify/fsnotify.(*Watcher).isClosed(...)
    /go/src/github.com/RadeonOpenCompute/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:66
github.com/fsnotify/fsnotify.(*Watcher).Add(0x0, {0x986f88?, 0xc000116000?})
    /go/src/github.com/RadeonOpenCompute/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:94 +0x6b
github.com/kubevirt/device-plugin-manager/pkg/dpm.(*Manager).Run(0xc00006be70)
    /go/src/github.com/RadeonOpenCompute/k8s-device-plugin/vendor/github.com/kubevirt/device-plugin-manager/pkg/dpm/manager.go:55 +0x226
main.main()
    /go/src/github.com/RadeonOpenCompute/k8s-device-plugin/cmd/k8s-device-plugin/main.go:331 +0x4bb

Steps to Reproduce:

Followed the installation instructions for the AMD GPU Helm Chart from Artifact Hub.
Deployed the chart to my Kubernetes cluster.
Observed the pod failing to start, logging the above error.

Expected Behavior: The AMD GPU device plugin should install without errors and run successfully in the Kubernetes cluster.

Actual Behavior: The pod crashes immediately after starting, with the log indicating a segmentation fault.

Extra information

Kubernetes cluster version: Server Version: v1.27.6+k3s1
AMD CPU: AMD Ryzen 9 7940hs

I have similar error with latest version

I0525 16:11:36.754627       1 main.go:305] AMD GPU device plugin for Kubernetes
I0525 16:11:36.754694       1 main.go:305] ./k8s-device-plugin version v1.25.2.7-7-g813f150
I0525 16:11:36.754701       1 main.go:305] hwloc: _VERSION: 2.10.0, _API_VERSION: 0x00020800, _COMPONENT_ABI: 7, Runtime: 0x00020800
I0525 16:11:36.754709       1 manager.go:42] Starting device plugin manager
I0525 16:11:36.754719       1 manager.go:46] Registering for system signal notifications
I0525 16:11:36.754985       1 manager.go:52] Registering for notifications of filesystem changes in device plugin directory
panic: runtime error: invalid memory address or nil pointer dereference
    panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x38 pc=0x53bbf3]

goroutine 1 [running]:
github.com/fsnotify/fsnotify.(*Watcher).isClosed(...)
    /go/src/github.com/ROCm/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:66
github.com/fsnotify/fsnotify.(*Watcher).Close(0x0)
    /go/src/github.com/ROCm/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:75 +0x13
panic({0x8cf480?, 0xd53320?})
    /usr/local/go/src/runtime/panic.go:914 +0x21f
github.com/fsnotify/fsnotify.(*Watcher).isClosed(...)
    /go/src/github.com/ROCm/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:66
github.com/fsnotify/fsnotify.(*Watcher).Add(0x0, {0x977505?, 0xc000204010?})
    /go/src/github.com/ROCm/k8s-device-plugin/vendor/github.com/fsnotify/fsnotify/inotify.go:94 +0x57
github.com/kubevirt/device-plugin-manager/pkg/dpm.(*Manager).Run(0xc000033550)
    /go/src/github.com/ROCm/k8s-device-plugin/vendor/github.com/kubevirt/device-plugin-manager/pkg/dpm/manager.go:55 +0x21f
main.main()
    /go/src/github.com/ROCm/k8s-device-plugin/cmd/k8s-device-plugin/main.go:331 +0x4e9

EDIT: Installing latest ROCm and hard restart fixed the issue. https://repo.radeon.com/amdgpu-install/latest/ubuntu/jammy/amdgpu-install_6.1.60101-1_all.deb

ROCm / k8s-device-plugin

Runtime Error with AMD GPU Helm Chart Installation in Kubernetes #48