on demand for clone and restore

Author: CMGS

cmd/core/helpers.go

@@ -318,6 +318,8 @@ func CloneVMConfigFromFlags(cmd *cobra.Command, snapCfg *types.SnapshotConfig) (
 		return nil, err
 	}
 
+	onDemand, _ := cmd.Flags().GetBool("on-demand")
+
 	cfg := &types.VMConfig{
 		Name:          vmName,
 		CPU:           cpu,
@@ -329,6 +331,7 @@ func CloneVMConfigFromFlags(cmd *cobra.Command, snapCfg *types.SnapshotConfig) (
 		Network:       network,
 		NoDirectIO:    noDirectIO,
 		Windows:       snapCfg.Windows,
+		OnDemand:      onDemand,
 	}
 	if err := cfg.Validate(); err != nil {
 		return nil, err
@@ -350,6 +353,9 @@ func RestoreVMConfigFromFlags(cmd *cobra.Command, vm *types.VM, snapCfg *types.S
 	result.Memory = memBytes
 	result.Storage = storBytes
 
+	onDemand, _ := cmd.Flags().GetBool("on-demand")
+	result.OnDemand = onDemand
+
 	return &result, nil
 }
 

cmd/vm/commands.go

@@ -109,6 +109,7 @@ func Command(h Actions) *cobra.Command {
 	restoreCmd.Flags().Int("cpu", 0, "boot CPUs (0 = keep current)")
 	restoreCmd.Flags().String("memory", "", "memory size (empty = keep current)")
 	restoreCmd.Flags().String("storage", "", "COW disk size (empty = keep current)")
+	restoreCmd.Flags().Bool("on-demand", false, "use UFFD on-demand memory loading for faster restore (CH only; snapshot file must remain on disk)")
 
 	debugCmd := &cobra.Command{
 		Use:   "debug [flags] IMAGE",
@@ -176,4 +177,5 @@ func addCloneFlags(cmd *cobra.Command) {
 	cmd.Flags().String("network", "", "CNI conflist name (empty = inherit from source VM)")
 	cmd.Flags().String("bridge", "", "use TAP-on-bridge instead of CNI (value is bridge device, e.g. cni0)")
 	cmd.Flags().Bool("no-direct-io", false, "disable O_DIRECT on writable disks (inherit from snapshot if not set)")
+	cmd.Flags().Bool("on-demand", false, "use UFFD on-demand memory loading for faster clone (CH only; snapshot file must remain on disk)")
 }

hypervisor/cloudhypervisor/clone.go

@@ -134,6 +134,7 @@ func (ch *CloudHypervisor) cloneAfterExtract(ctx context.Context, vmID string, v
 		cpu:                 vmCfg.CPU,
 		diskQueueSize:       vmCfg.DiskQueueSize,
 		noDirectIO:          vmCfg.NoDirectIO,
+		onDemand:            vmCfg.OnDemand,
 	}); err != nil {
 		return nil, err
 	}
@@ -162,6 +163,7 @@ type cloneResumeOpts struct {
 	cpu                 int
 	diskQueueSize       int
 	noDirectIO          bool
+	onDemand            bool
 }
 
 func (ch *CloudHypervisor) restoreAndResumeClone(
@@ -176,7 +178,7 @@ func (ch *CloudHypervisor) restoreAndResumeClone(
 		}
 	}()
 
-	if err = restoreVM(ctx, sockPath, runDir); err != nil {
+	if err = restoreVM(ctx, sockPath, runDir, opts.onDemand); err != nil {
 		return fmt.Errorf("vm.restore: %w", err)
 	}
 	hc := utils.NewSocketHTTPClient(sockPath)

hypervisor/cloudhypervisor/helper.go

@@ -82,10 +82,28 @@ func snapshotVM(ctx context.Context, sockPath, destDir string) error {
 	return err
 }
 
-func restoreVM(ctx context.Context, sockPath, sourceDir string) error {
-	body, err := json.Marshal(map[string]string{
-		"source_url": "file://" + sourceDir,
-	})
+// chMemoryRestoreMode controls how CH restores guest memory from a snapshot.
+type chMemoryRestoreMode string
+
+const (
+	// chMemoryRestoreOnDemand uses userfaultfd (UFFD) to lazily page in
+	// guest memory from the snapshot file, avoiding a full upfront copy.
+	chMemoryRestoreOnDemand chMemoryRestoreMode = "OnDemand"
+)
+
+type chRestoreConfig struct {
+	SourceURL         string              `json:"source_url"`
+	MemoryRestoreMode chMemoryRestoreMode `json:"memory_restore_mode,omitempty"`
+}
+
+func restoreVM(ctx context.Context, sockPath, sourceDir string, onDemand bool) error {
+	cfg := chRestoreConfig{
+		SourceURL: "file://" + sourceDir,
+	}
+	if onDemand {
+		cfg.MemoryRestoreMode = chMemoryRestoreOnDemand
+	}
+	body, err := json.Marshal(cfg)
 	if err != nil {
 		return fmt.Errorf("marshal restore request: %w", err)
 	}

hypervisor/cloudhypervisor/restore.go

@@ -135,7 +135,7 @@ func (ch *CloudHypervisor) restoreAfterExtract(ctx context.Context, vmID string,
 		}
 	}()
 
-	if err = restoreVM(ctx, sockPath, rec.RunDir); err != nil {
+	if err = restoreVM(ctx, sockPath, rec.RunDir, vmCfg.OnDemand); err != nil {
 		return nil, fmt.Errorf("vm.restore: %w", err)
 	}
 	hc := utils.NewSocketHTTPClient(sockPath)

types/vm.go

@@ -38,6 +38,7 @@ type VMConfig struct {
 
 	NoDirectIO bool `json:"no_direct_io,omitempty"` // disable O_DIRECT on writable disks
 	Windows    bool `json:"windows,omitempty"`      // Windows guest: UEFI boot, kvm_hyperv=on, no cidata
+	OnDemand   bool `json:"-"`                      // use UFFD on-demand memory restore (CH only); transient, not persisted
 
 	// Transient cloud-init credentials — carried in-memory from CLI to cidata
 	// generation, never serialized to JSON or persisted in the VM record.