Merge branch 'upstream-add-driver-root-flag' into 'master'

Add a flag to specify the root path of the NVIDIA driver installation

See merge request nvidia/kubernetes/device-plugin!73

Author: klueska

README.md

@@ -142,6 +142,8 @@ a number of customizable values. The most commonly overridden ones are:
   deviceListStrategy:
       the desired strategy for passing the device list to the underlying runtime
       [envvar | volume-mounts] (default "envvar")
+  nvidiaDriverRoot:
+      the root path for the NVIDIA driver installation (typical values are '/' or '/run/nvidia/driver')
 ```
 
 When set to true, the `failOnInitError` flag fails the plugin if an error is

deployments/helm/nvidia-device-plugin/templates/daemonset.yml

@@ -62,6 +62,7 @@ spec:
         - "--pass-device-specs={{ .Values.compatWithCPUManager }}"
         - "--fail-on-init-error={{ .Values.failOnInitError }}"
         - "--device-list-strategy={{ .Values.deviceListStrategy }}"
+        - "--nvidia-driver-root={{ .Values.nvidiaDriverRoot }}"
         securityContext:
         {{- if ne (len .Values.securityContext) 0 }}
           {{- toYaml .Values.securityContext | nindent 10 }}

deployments/helm/nvidia-device-plugin/values.yaml

@@ -3,6 +3,7 @@ compatWithCPUManager: false
 migStrategy: none
 failOnInitError: true
 deviceListStrategy: envvar
+nvidiaDriverRoot: "/"
 
 nameOverride: ""
 fullnameOverride: ""

main.go

@@ -38,7 +38,7 @@ var failOnInitErrorFlag = flag.Bool(
 	true,
 	"fail the plugin if an error is encountered during initialization, otherwise block indefinitely [default: true]\n")
 
-var passDeviceSpecs = flag.Bool(
+var passDeviceSpecsFlag = flag.Bool(
 	"pass-device-specs",
 	false,
 	"pass the list of DeviceSpecs to the kubelet on Allocate()")
@@ -49,6 +49,11 @@ var deviceListStrategyFlag = flag.String(
 	"the desired strategy for passing the device list to the underlying runtime\n"+
 		"[envvar | volume-mounts]")
 
+var nvidiaDriverRootFlag = flag.String(
+	"nvidia-driver-root",
+	"/",
+	"the root path for the NVIDIA driver installation (typical values are '/' or '/run/nvidia/driver')")
+
 func main() {
 	flag.Parse()
 

server.go

@@ -280,8 +280,8 @@ func (m *NvidiaDevicePlugin) Allocate(ctx context.Context, reqs *pluginapi.Alloc
 			response.Envs = m.apiEnvs(m.deviceListEnvvar, []string{deviceListAsVolumeMountsContainerPathRoot})
 			response.Mounts = m.apiMounts(req.DevicesIDs)
 		}
-		if *passDeviceSpecs {
-			response.Devices = m.apiDeviceSpecs(req.DevicesIDs)
+		if *passDeviceSpecsFlag {
+			response.Devices = m.apiDeviceSpecs(*nvidiaDriverRootFlag, req.DevicesIDs)
 		}
 
 		responses.ContainerResponses = append(responses.ContainerResponses, &response)
@@ -348,7 +348,7 @@ func (m *NvidiaDevicePlugin) apiMounts(filter []string) []*pluginapi.Mount {
 	return mounts
 }
 
-func (m *NvidiaDevicePlugin) apiDeviceSpecs(filter []string) []*pluginapi.DeviceSpec {
+func (m *NvidiaDevicePlugin) apiDeviceSpecs(driverRoot string, filter []string) []*pluginapi.DeviceSpec {
 	var specs []*pluginapi.DeviceSpec
 
 	paths := []string{
@@ -362,7 +362,7 @@ func (m *NvidiaDevicePlugin) apiDeviceSpecs(filter []string) []*pluginapi.Device
 		if _, err := os.Stat(p); err == nil {
 			spec := &pluginapi.DeviceSpec{
 				ContainerPath: p,
-				HostPath:      p,
+				HostPath:      filepath.Join(driverRoot, p),
 				Permissions:   "rw",
 			}
 			specs = append(specs, spec)
@@ -374,7 +374,7 @@ func (m *NvidiaDevicePlugin) apiDeviceSpecs(filter []string) []*pluginapi.Device
 			if d.ID == id {
 				spec := &pluginapi.DeviceSpec{
 					ContainerPath: d.Path,
-					HostPath:      d.Path,
+					HostPath:      filepath.Join(driverRoot, d.Path),
 					Permissions:   "rw",
 				}
 				specs = append(specs, spec)