Add MIG config support when MIG-backed vGPU config

This change adds MIG configuration support via nvidia-mig-parted when
MIG-backed vGPU types are included in the selected vGPU config, before
the vGPU configuration takes place.

Specifically:

- Add CLI flags for MIG configuration options.
- Include the nvidia-mig-parted binary in the container image.
- Parse vGPU config to detect MIG requirements, convert the vGPU config
  to the respective MIG config and configure MIG before vGPUs via
  nvidia-mig-parted.
- nvidia-mig-parted requires NVML. The NVIDIA driver library path is
  searched and used with the `LD_PRELOAD` env var when running
  nvidia-mig-parted commands. When NVML is not available, skip
  MIG configuration and proceed to vGPU configuration. This ensures
  that the vGPU Device Manager is backwards compatible with components
  that do not make NVML available to it.

Signed-off-by: Michail Resvanis <mresvani@redhat.com>

Author: mresvanis

api/spec/v1/spec.go

@@ -20,6 +20,9 @@ import (
 	"encoding/json"
 	"fmt"
 
+	migpartedv1 "github.com/NVIDIA/mig-parted/api/spec/v1"
+	migtypes "github.com/NVIDIA/mig-parted/pkg/types"
+
 	"github.com/NVIDIA/vgpu-device-manager/pkg/types"
 )
 
@@ -160,6 +163,44 @@ func (s *VGPUConfigSpec) UnmarshalJSON(b []byte) error {
 	return nil
 }
 
+func (s VGPUConfigSpecSlice) ToMigConfigSpecSlice() (migpartedv1.MigConfigSpecSlice, error) {
+	var migConfigSpecs migpartedv1.MigConfigSpecSlice
+
+	for _, vgpuSpec := range s {
+		migSpec := migpartedv1.MigConfigSpec{
+			DeviceFilter: vgpuSpec.DeviceFilter,
+			Devices:      vgpuSpec.Devices,
+			MigDevices:   make(migtypes.MigConfig),
+		}
+
+		migEnabled := false
+		for vgpuType := range vgpuSpec.VGPUDevices {
+			vgpu, err := types.ParseVGPUType(vgpuType)
+			if err != nil {
+				return nil, fmt.Errorf("failed to parse vGPU type %s: %w", vgpuType, err)
+			}
+
+			if vgpu.G > 0 {
+				migEnabled = true
+				migProfile := fmt.Sprintf("%dg.%dgb", vgpu.G, vgpu.GB)
+				for _, attr := range vgpu.Attr {
+					if attr == types.AttributeMediaExtensions {
+						migProfile += ".me"
+						break
+					}
+				}
+				migSpec.MigDevices[migProfile] = vgpuSpec.VGPUDevices[vgpuType]
+			}
+		}
+
+		migSpec.MigEnabled = migEnabled
+
+		migConfigSpecs = append(migConfigSpecs, migSpec)
+	}
+
+	return migConfigSpecs, nil
+}
+
 func containsKey(m map[string]json.RawMessage, s string) bool {
 	_, exists := m[s]
 	return exists

api/spec/v1/spec_test.go

@@ -21,6 +21,11 @@ import (
 
 	"github.com/stretchr/testify/require"
 	"sigs.k8s.io/yaml"
+
+	migpartedv1 "github.com/NVIDIA/mig-parted/api/spec/v1"
+	migtypes "github.com/NVIDIA/mig-parted/pkg/types"
+
+	"github.com/NVIDIA/vgpu-device-manager/pkg/types"
 )
 
 func TestSpec(t *testing.T) {
@@ -230,3 +235,190 @@ func TestVGPUConfigSpec(t *testing.T) {
 	}
 
 }
+
+func TestVGPUConfigSpecSliceToMigConfigSpecSlice(t *testing.T) {
+	testCases := []struct {
+		Description           string
+		VGPUConfigSpecSlice   VGPUConfigSpecSlice
+		ExpectedMigConfigSpec migpartedv1.MigConfigSpecSlice
+		ExpectedError         string
+	}{
+		{
+			"Empty slice",
+			VGPUConfigSpecSlice{},
+			nil,
+			"",
+		},
+		{
+			"Single MIG-backed vGPU type",
+			VGPUConfigSpecSlice{
+				{
+					DeviceFilter: "MODEL",
+					Devices:      "all",
+					VGPUDevices: types.VGPUConfig{
+						"A100-1-5C": 4,
+					},
+				},
+			},
+			migpartedv1.MigConfigSpecSlice{
+				{
+					DeviceFilter: "MODEL",
+					Devices:      "all",
+					MigEnabled:   true,
+					MigDevices: migtypes.MigConfig{
+						"1g.5gb": 4,
+					},
+				},
+			},
+			"",
+		},
+		{
+			"Multiple MIG-backed vGPU types",
+			VGPUConfigSpecSlice{
+				{
+					DeviceFilter: []string{"MODEL1", "MODEL2"},
+					Devices:      []int{0, 1},
+					VGPUDevices: types.VGPUConfig{
+						"A100-1-5C":  2,
+						"A100-2-10C": 1,
+					},
+				},
+			},
+			migpartedv1.MigConfigSpecSlice{
+				{
+					DeviceFilter: []string{"MODEL1", "MODEL2"},
+					Devices:      []int{0, 1},
+					MigEnabled:   true,
+					MigDevices: migtypes.MigConfig{
+						"1g.5gb":  2,
+						"2g.10gb": 1,
+					},
+				},
+			},
+			"",
+		},
+		{
+			"MIG-backed vGPU type with media extensions",
+			VGPUConfigSpecSlice{
+				{
+					Devices: "all",
+					VGPUDevices: types.VGPUConfig{
+						"A100-1-5CME": 2,
+					},
+				},
+			},
+			migpartedv1.MigConfigSpecSlice{
+				{
+					Devices:    "all",
+					MigEnabled: true,
+					MigDevices: migtypes.MigConfig{
+						"1g.5gb.me": 2,
+					},
+				},
+			},
+			"",
+		},
+		{
+			"Non-MIG vGPU type",
+			VGPUConfigSpecSlice{
+				{
+					Devices: "all",
+					VGPUDevices: types.VGPUConfig{
+						"A100-40C": 2,
+					},
+				},
+			},
+			migpartedv1.MigConfigSpecSlice{
+				{
+					Devices:    "all",
+					MigEnabled: false,
+					MigDevices: migtypes.MigConfig{},
+				},
+			},
+			"",
+		},
+		{
+			"Mixed MIG and non-MIG vGPU types",
+			VGPUConfigSpecSlice{
+				{
+					Devices: "all",
+					VGPUDevices: types.VGPUConfig{
+						"A100-40C":  1,
+						"A100-1-5C": 2,
+					},
+				},
+			},
+			migpartedv1.MigConfigSpecSlice{
+				{
+					Devices:    "all",
+					MigEnabled: true,
+					MigDevices: migtypes.MigConfig{
+						"1g.5gb": 2,
+					},
+				},
+			},
+			"",
+		},
+		{
+			"Multiple specs with different configurations",
+			VGPUConfigSpecSlice{
+				{
+					Devices: "all",
+					VGPUDevices: types.VGPUConfig{
+						"A100-1-5C": 4,
+					},
+				},
+				{
+					DeviceFilter: "MODEL",
+					Devices:      []int{0, 1},
+					VGPUDevices: types.VGPUConfig{
+						"A100-40C": 1,
+					},
+				},
+			},
+			migpartedv1.MigConfigSpecSlice{
+				{
+					Devices:    "all",
+					MigEnabled: true,
+					MigDevices: migtypes.MigConfig{
+						"1g.5gb": 4,
+					},
+				},
+				{
+					DeviceFilter: "MODEL",
+					Devices:      []int{0, 1},
+					MigEnabled:   false,
+					MigDevices:   migtypes.MigConfig{},
+				},
+			},
+			"",
+		},
+		{
+			"Invalid vGPU type",
+			VGPUConfigSpecSlice{
+				{
+					Devices: "all",
+					VGPUDevices: types.VGPUConfig{
+						"InvalidType": 1,
+					},
+				},
+			},
+			nil,
+			"failed to parse vGPU type InvalidType:",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.Description, func(t *testing.T) {
+			result, err := tc.VGPUConfigSpecSlice.ToMigConfigSpecSlice()
+			if tc.ExpectedError != "" {
+				require.NotNil(t, err, "Expected failure but got success")
+				require.Nil(t, result, "Expected nil result on failure")
+				require.ErrorContains(t, err, tc.ExpectedError)
+			} else {
+				require.Nil(t, err, "Unexpected failure: %v", err)
+				require.Equal(t, tc.ExpectedMigConfigSpec, result, "Unexpected result")
+			}
+		})
+	}
+}