add fallback mechanism for allocator init failing and port upstream changes (ROCm#36)

* port latest partition changes from rocm repo

* handle allocator init failure

introduce fallback option in case besteffort allocator fails. we fallback to default kubelet allocation

Author: biluriuday

cmd/k8s-device-plugin/main.go

@@ -32,25 +32,43 @@ import (
 
 var gitDescribe string
 
-func getResourceList(resourceNamingStrategy string) ([]string, error) {
+type ResourceNamingStrategy string
+
+const (
+	StrategySingle ResourceNamingStrategy = "single"
+	StrategyMixed  ResourceNamingStrategy = "mixed"
+)
+
+func ParseStrategy(s string) (ResourceNamingStrategy, error) {
+	switch s {
+	case string(StrategySingle):
+		return StrategySingle, nil
+	case string(StrategyMixed):
+		return StrategyMixed, nil
+	default:
+		return "", fmt.Errorf("invalid resource naming strategy: %s", s)
+	}
+}
+
+func getResourceList(resourceNamingStrategy ResourceNamingStrategy) ([]string, error) {
 	var resources []string
 
 	// Check if the node is homogeneous
 	isHomogeneous := amdgpu.IsHomogeneous()
-	devices, deviceCountMap := amdgpu.GetAMDGPUs()
-	if len(devices) == 0 {
+	partitionCountMap := amdgpu.UniquePartitionConfigCount(amdgpu.GetAMDGPUs())
+	if len(amdgpu.GetAMDGPUs()) == 0 {
 		return resources, nil
 	}
 	if isHomogeneous {
 		// Homogeneous node will report only "gpu" resource if strategy is single. If strategy is mixed, it will report resources under the partition type name
-		if resourceNamingStrategy == "single" {
+		if resourceNamingStrategy == StrategySingle {
 			resources = []string{"gpu"}
-		} else if resourceNamingStrategy == "mixed" {
-			if len(deviceCountMap) == 0 {
+		} else if resourceNamingStrategy == StrategyMixed {
+			if len(partitionCountMap) == 0 {
 				// If partitioning is not supported on the node, we should report resources under "gpu" regardless of the strategy
 				resources = []string{"gpu"}
 			} else {
-				for partitionType, count := range deviceCountMap {
+				for partitionType, count := range partitionCountMap {
 					if count > 0 {
 						resources = append(resources, partitionType)
 					}
@@ -59,10 +77,10 @@ func getResourceList(resourceNamingStrategy string) ([]string, error) {
 		}
 	} else {
 		// Heterogeneous node reports resources based on partition types if strategy is mixed. Heterogeneous is not allowed if Strategy is single
-		if resourceNamingStrategy == "single" {
+		if resourceNamingStrategy == StrategySingle {
 			return resources, fmt.Errorf("Partitions of different styles across GPUs in a node is not supported with single strategy. Please start device plugin with mixed strategy")
-		} else if resourceNamingStrategy == "mixed" {
-			for partitionType, count := range deviceCountMap {
+		} else if resourceNamingStrategy == StrategyMixed {
+			for partitionType, count := range partitionCountMap {
 				if count > 0 {
 					resources = append(resources, partitionType)
 				}
@@ -92,6 +110,11 @@ func main() {
 	flag.StringVar(&resourceNamingStrategy, "resource_naming_strategy", "single", "Resource strategy to be used: single or mixed")
 	// this is also needed to enable glog usage in dpm
 	flag.Parse()
+	strategy, err := ParseStrategy(resourceNamingStrategy)
+	if err != nil {
+		glog.Errorf("%v", err)
+		os.Exit(1)
+	}
 
 	for _, v := range versions {
 		glog.Infof("%s", v)
@@ -117,7 +140,7 @@ func main() {
 		// /sys/class/kfd only exists if ROCm kernel/driver is installed
 		var path = "/sys/class/kfd"
 		if _, err := os.Stat(path); err == nil {
-			resources, err := getResourceList(resourceNamingStrategy)
+			resources, err := getResourceList(strategy)
 			if err != nil {
 				glog.Errorf("Error occured: %v", err)
 				os.Exit(1)

cmd/k8s-node-labeller/main.go

@@ -353,46 +353,36 @@ var labelGenerators = map[string]func(map[string]map[string]interface{}) map[str
 
 		return createLabels("cu-count", counts)
 	},
-}
-
-var labelProperties = make(map[string]*bool, len(labelGenerators))
-
-func generatePartitionLabels() map[string]string {
-	_, deviceCountMap := amdgpu.GetAMDGPUs()
-	isHomogeneous := amdgpu.IsHomogeneous()
-	IsComputePartitionSupported := amdgpu.IsComputePartitionSupported()
-	IsMemoryPartitionSupported := amdgpu.IsMemoryPartitionSupported()
-
-	labels := make(map[string]string)
-
-	if isHomogeneous {
-		// Iterate through deviceCountMap and find the partition type with count > 0
-		for partitionType, count := range deviceCountMap {
-			if count > 0 {
-				labels[partitionTypeLabel] = partitionType
-				break
+	"compute-memory-partition": func(gpus map[string]map[string]interface{}) map[string]string {
+		partitionCountMap := amdgpu.UniquePartitionConfigCount(gpus)
+		isHomogeneous := amdgpu.IsHomogeneous()
+		if isHomogeneous {
+			for partitionType, count := range partitionCountMap {
+				if count > 0 {
+					pfx := createLabelPrefix("compute-memory-partition", false)
+					return map[string]string{pfx: partitionType}
+				}
 			}
 		}
-	}
-
-	if IsComputePartitionSupported {
-		labels[computePartitioningSupportedLabel] = "true"
-	} else {
-		labels[computePartitioningSupportedLabel] = "false"
-	}
-
-	if IsMemoryPartitionSupported {
-		labels[memoryPartitioningSupportedLabel] = "true"
-	} else {
-		labels[memoryPartitioningSupportedLabel] = "false"
-	}
-
-	return labels
+		return map[string]string{}
+	},
+	"compute-partitioning-supported": func(gpus map[string]map[string]interface{}) map[string]string {
+		val := strconv.FormatBool(amdgpu.IsComputePartitionSupported())
+		pfx := createLabelPrefix("compute-partitioning-supported", false)
+		return map[string]string{pfx: val}
+	},
+	"memory-partitioning-supported": func(gpus map[string]map[string]interface{}) map[string]string {
+		val := strconv.FormatBool(amdgpu.IsMemoryPartitionSupported())
+		pfx := createLabelPrefix("memory-partitioning-supported", false)
+		return map[string]string{pfx: val}
+	},
 }
 
+var labelProperties = make(map[string]*bool, len(labelGenerators))
+
 func generateLabels(lblProps map[string]*bool) map[string]string {
 	results := make(map[string]string, len(labelGenerators))
-	gpus, _ := amdgpu.GetAMDGPUs()
+	gpus := amdgpu.GetAMDGPUs()
 
 	for l, f := range labelGenerators {
 		if !*lblProps[l] {
@@ -403,13 +393,6 @@ func generateLabels(lblProps map[string]*bool) map[string]string {
 			results[k] = v
 		}
 	}
-
-	// Add the new GPU labels
-	gpuLabels := generatePartitionLabels()
-	for k, v := range gpuLabels {
-		results[k] = v
-	}
-
 	return results
 }
 

cmd/k8s-node-labeller/main_test.go

@@ -10,29 +10,32 @@ import (
 
 var (
 	expectedAllLabelKeys = map[string]bool{
-		"amd.com/gpu.family":                     true,
-		"amd.com/gpu.driver-version":             true,
-		"amd.com/gpu.driver-src-version":         true,
-		"amd.com/gpu.firmware":                   true,
-		"amd.com/gpu.device-id":                  true,
-		"amd.com/gpu.product-name":               true,
-		"amd.com/gpu.vram":                       true,
-		"amd.com/gpu.simd-count":                 true,
-		"amd.com/gpu.cu-count":                   true,
-		"amd.com/compute-partitioning-supported": true,
-		"amd.com/memory-partitioning-supported":  true,
-		"amd.com/compute-memory-partition":       true,
+		"amd.com/gpu.family":                         true,
+		"amd.com/gpu.driver-version":                 true,
+		"amd.com/gpu.driver-src-version":             true,
+		"amd.com/gpu.firmware":                       true,
+		"amd.com/gpu.device-id":                      true,
+		"amd.com/gpu.product-name":                   true,
+		"amd.com/gpu.vram":                           true,
+		"amd.com/gpu.simd-count":                     true,
+		"amd.com/gpu.cu-count":                       true,
+		"amd.com/gpu.compute-memory-partition":       true,
+		"amd.com/gpu.compute-partitioning-supported": true,
+		"amd.com/gpu.memory-partitioning-supported":  true,
 	}
 	expectedAllExperimentalLabelKeys = map[string]bool{
-		"beta.amd.com/gpu.family":             true,
-		"beta.amd.com/gpu.driver-version":     true,
-		"beta.amd.com/gpu.driver-src-version": true,
-		"beta.amd.com/gpu.firmware":           true,
-		"beta.amd.com/gpu.device-id":          true,
-		"beta.amd.com/gpu.product-name":       true,
-		"beta.amd.com/gpu.vram":               true,
-		"beta.amd.com/gpu.simd-count":         true,
-		"beta.amd.com/gpu.cu-count":           true,
+		"beta.amd.com/gpu.family":                         true,
+		"beta.amd.com/gpu.driver-version":                 true,
+		"beta.amd.com/gpu.driver-src-version":             true,
+		"beta.amd.com/gpu.firmware":                       true,
+		"beta.amd.com/gpu.device-id":                      true,
+		"beta.amd.com/gpu.product-name":                   true,
+		"beta.amd.com/gpu.vram":                           true,
+		"beta.amd.com/gpu.simd-count":                     true,
+		"beta.amd.com/gpu.cu-count":                       true,
+		"beta.amd.com/gpu.compute-memory-partition":       true,
+		"beta.amd.com/gpu.compute-partitioning-supported": true,
+		"beta.amd.com/gpu.memory-partitioning-supported":  true,
 	}
 )
 
@@ -91,6 +94,26 @@ func TestRemoveOldNodeLabels(t *testing.T) {
 				"dummyLabel2": "2",
 			},
 		},
+		{
+			inputNode: &corev1.Node{
+				ObjectMeta: metav1.ObjectMeta{
+					Labels: map[string]string{
+						"amd.com/cpu":    "true",
+						"amd.com/gpu":    "true",
+						"amd.com/mi300x": "true",
+						"dummyLabel1":    "1",
+						"dummyLabel2":    "2",
+					},
+				},
+			},
+			expectLabels: map[string]string{
+				"amd.com/cpu":    "true",
+				"amd.com/gpu":    "true",
+				"amd.com/mi300x": "true",
+				"dummyLabel1":    "1",
+				"dummyLabel2":    "2",
+			},
+		},
 	}
 
 	for _, tc := range testCases {

docs/conf.py

@@ -4,20 +4,29 @@
 # list see the documentation:
 # https://www.sphinx-doc.org/en/master/usage/configuration.html
 
-# configurations for PDF output by Read the Docs
-project = "k8s-device-plugin Documentation"
-author = "Advanced Micro Devices, Inc."
-copyright = "Copyright (c) 2024 Advanced Micro Devices, Inc. All rights reserved."
-version = "6.2.0"
-release = "6.2.0"
 
-external_toc_path = "./sphinx/_toc.yml"
+external_projects_local_file = "projects.yaml"
+external_projects_remote_repository = ""
+#external_projects = ["k8s-device-plugin"]
+external_projects = []
+external_projects_current_project = "k8s-device-plugin"
+
+project = "AMD Kubernetes Device Plugin Documentation"
+version = "1.3.1"
+release = version
+html_title = f"Device Plugin Documentation {version}"
+author = "Advanced Micro Devices, Inc."
+copyright = "Copyright (c) 2025 Advanced Micro Devices, Inc. All rights reserved."
 
+# Required settings
+html_theme = "rocm_docs_theme"
+html_theme_options = {
+    "flavor": "instinct"
+}
 extensions = ["rocm_docs"]
 
-external_projects_current_project = "rocm"
+external_toc_path = "./sphinx/_toc.yml"
 
-html_theme = "rocm_docs_theme"
-html_theme_options = {"flavor": "rocm-docs-home"}
+extensions = ["rocm_docs"]
 
-html_title = project
+exclude_patterns = ['.venv']

docs/contributing/development.md

@@ -0,0 +1,52 @@
+# Development Guidelines
+
+## Getting Started
+
+### Prerequisites
+
+Before you begin development, ensure you have the following prerequisites:
+
+- A working installation of Go (version 1.23 or later).
+- Access to a Kubernetes cluster for testing.
+- Familiarity with Git and GitHub workflows.
+
+### Cloning the Repository
+
+To contribute to the project, start by cloning the repository:
+
+```bash
+git clone https://github.com/ROCm/k8s-device-plugin.git
+cd k8s-device-plugin
+```
+
+### Branching Strategy
+
+When working on a new feature or bug fix, create a new branch from the `main` branch:
+
+```bash
+git checkout -b feature/my-new-feature
+```
+
+Make sure to name your branch descriptively to reflect the changes you are making.
+
+## Development Workflow
+
+1. **Make Changes**: Implement your changes in the codebase.
+2. **Testing**: Ensure that your changes are covered by tests. Run existing tests and add new ones as necessary.
+3. **Commit Changes**: Commit your changes with a clear and concise commit message:
+
+   ```bash
+   git commit -m "Add feature X to improve Y"
+   ```
+
+4. **Push Changes**: Push your branch to the remote repository:
+
+   ```bash
+   git push origin feature/my-new-feature
+   ```
+
+5. **Create a Pull Request**: Navigate to the GitHub repository and create a pull request. Provide a detailed description of your changes and why they are necessary.
+
+## Code Review Process
+
+All contributions will undergo a code review process. Reviewers will assess the quality of the code, adherence to coding standards, and the completeness of tests. Be open to feedback and ready to make adjustments as needed.