Add --create-device-nodes option to toolkit config

This change adds a --create-device-nodes option to the toolkit config CLI.
Most noteably, this allows the creation of control devices to be skipped
when CDI spec generation is enabled.

Currently values of "", "node", and "control" are supported and can be set
via the command line flag or the CREATE_DEVICE_NODES environment variable.

The default value of CREATE_DEVICE_NODES=control will trigger the creation
of control device nodes. Setting this envvar to include the (comma-separated)
strings of "" or "none" will disable device node creation regardless of
whether other supported strings are included.

Signed-off-by: Evan Lezar <elezar@nvidia.com>

tools/container/toolkit/toolkit.go

@@ -70,6 +70,8 @@ type options struct {
 	cdiVendor    string
 	cdiClass     string
 
+	createDeviceNodes cli.StringSlice
+
 	acceptNVIDIAVisibleDevicesWhenUnprivileged bool
 	acceptNVIDIAVisibleDevicesAsVolumeMounts   bool
 
@@ -224,6 +226,13 @@ func main() {
 			Hidden:      true,
 			Destination: &opts.ignoreErrors,
 		},
+		&cli.StringSliceFlag{
+			Name:        "create-device-nodes",
+			Usage:       "(Only applicable with --cdi-enabled) specifies which device nodes should be created. If any one of the options is set to '' or 'none', no device nodes will be created.",
+			Value:       cli.NewStringSlice("control"),
+			Destination: &opts.createDeviceNodes,
+			EnvVars:     []string{"CREATE_DEVICE_NODES"},
+		},
 	}
 
 	// Update the subcommand flags with the common subcommand flags
@@ -252,6 +261,29 @@ func validateOptions(c *cli.Context, opts *options) error {
 	opts.cdiVendor = vendor
 	opts.cdiClass = class
 
+	if opts.cdiEnabled && opts.cdiOutputDir == "" {
+		log.Warning("Skipping CDI spec generation (no output directory specified)")
+		opts.cdiEnabled = false
+	}
+
+	isDisabled := false
+	for _, mode := range opts.createDeviceNodes.Value() {
+		if mode != "" && mode != "none" && mode != "control" {
+			return fmt.Errorf("invalid --create-device-nodes value: %v", mode)
+		}
+		if mode == "" || mode == "none" {
+			isDisabled = true
+			break
+		}
+	}
+	if !opts.cdiEnabled && !isDisabled {
+		log.Info("disabling device node creation since --cdi-enabled=false")
+		isDisabled = true
+	}
+	if isDisabled {
+		opts.createDeviceNodes = *cli.NewStringSlice()
+	}
+
 	return nil
 }
 
@@ -330,6 +362,13 @@ func Install(cli *cli.Context, opts *options) error {
 		log.Errorf("Ignoring error: %v", fmt.Errorf("error installing NVIDIA container toolkit config: %v", err))
 	}
 
+	err = createDeviceNodes(opts)
+	if err != nil && !opts.ignoreErrors {
+		return fmt.Errorf("error creating device nodes: %v", err)
+	} else if err != nil {
+		log.Errorf("Ignoring error: %v", fmt.Errorf("error creating device nodes: %v", err))
+	}
+
 	return generateCDISpec(opts, nvidiaCTKPath)
 }
 
@@ -677,27 +716,37 @@ func createDirectories(dir ...string) error {
 	return nil
 }
 
-// generateCDISpec generates a CDI spec for use in managemnt containers
+func createDeviceNodes(opts *options) error {
-func generateCDISpec(opts *options, nvidiaCTKPath string) error {
+	modes := opts.createDeviceNodes.Value()
-	if !opts.cdiEnabled {
+	if len(modes) == 0 {
-		return nil
-	}
-	if opts.cdiOutputDir == "" {
-		log.Info("Skipping CDI spec generation (no output directory specified)")
 		return nil
 	}
 
-	log.Infof("Creating control device nodes at %v", opts.DriverRootCtrPath)
 	devices, err := nvdevices.New(
 		nvdevices.WithDevRoot(opts.DriverRootCtrPath),
 	)
 	if err != nil {
 		return fmt.Errorf("failed to create library: %v", err)
 	}
-	if err := devices.CreateNVIDIAControlDevices(); err != nil {
-		return fmt.Errorf("failed to create control device nodes: %v", err)
-	}
 
+	for _, mode := range modes {
+		log.Infof("Creating %v device nodes at %v", mode, opts.DriverRootCtrPath)
+		if mode != "control" {
+			log.Warningf("Unrecognised device mode: %v", mode)
+			continue
+		}
+		if err := devices.CreateNVIDIAControlDevices(); err != nil {
+			return fmt.Errorf("failed to create control device nodes: %v", err)
+		}
+	}
+	return nil
+}
+
+// generateCDISpec generates a CDI spec for use in managemnt containers
+func generateCDISpec(opts *options, nvidiaCTKPath string) error {
+	if !opts.cdiEnabled {
+		return nil
+	}
 	log.Info("Generating CDI spec for management containers")
 	cdilib, err := nvcdi.New(
 		nvcdi.WithMode(nvcdi.ModeManagement),