Runsc checkpoint works.

This is the first iteration of checkpoint that actually saves to a file.
Tests for checkpoint are included.

Ran into an issue when private unix sockets are enabled. An error message
was added for this case and the mutex state was set.

PiperOrigin-RevId: 200269470
Change-Id: I28d29a9f92c44bf73dc4a4b12ae0509ee4070e93

pkg/sentry/fs/gofer/session.go

@@ -28,8 +28,9 @@ import (
 )
 
 type endpointMap struct {
-	mu sync.RWMutex
-	m  map[device.MultiDeviceKey]unix.BoundEndpoint
+	mu sync.RWMutex `state:"nosave"`
+	// TODO: Make map with private unix sockets savable.
+	m map[device.MultiDeviceKey]unix.BoundEndpoint
 }
 
 // add adds the endpoint to the map.

pkg/sentry/fs/gofer/session_state.go

@@ -22,6 +22,15 @@ import (
 	"gvisor.googlesource.com/gvisor/pkg/unet"
 )
 
+// beforeSave is invoked by stateify.
+//
+// TODO: Make map with private unix sockets savable.
+func (e *endpointMap) beforeSave() {
+	if len(e.m) != 0 {
+		panic("EndpointMap with existing private unix sockets cannot be saved")
+	}
+}
+
 // afterLoad is invoked by stateify.
 func (s *session) afterLoad() {
 	// The restore environment contains the 9p connection of this mount.

runsc/boot/loader.go

@@ -100,7 +100,9 @@ func New(spec *specs.Spec, conf *Config, controllerFD int, ioFDs []int, console
 	}
 
 	// Create VDSO.
-	vdso, err := loader.PrepareVDSO(p)
+	//
+	// Pass k as the platform since it is savable, unlike the actual platform.
+	vdso, err := loader.PrepareVDSO(k)
 	if err != nil {
 		return nil, fmt.Errorf("error creating vdso: %v", err)
 	}

runsc/cmd/checkpoint.go

@@ -15,6 +15,8 @@
 package cmd
 
 import (
+	"os"
+
 	"context"
 	"flag"
 	"github.com/google/subcommands"
@@ -24,6 +26,7 @@ import (
 
 // Checkpoint implements subcommands.Command for the "checkpoint" command.
 type Checkpoint struct {
+	imagePath string
 }
 
 // Name implements subcommands.Command.Name.
@@ -44,6 +47,7 @@ func (*Checkpoint) Usage() string {
 
 // SetFlags implements subcommands.Command.SetFlags.
 func (c *Checkpoint) SetFlags(f *flag.FlagSet) {
+	f.StringVar(&c.imagePath, "image-path", "", "path to saved container image")
 }
 
 // Execute implements subcommands.Command.Execute.
@@ -62,7 +66,18 @@ func (c *Checkpoint) Execute(_ context.Context, f *flag.FlagSet, args ...interfa
 		Fatalf("error loading container: %v", err)
 	}
 
-	if err := cont.Checkpoint(); err != nil {
+	if c.imagePath == "" {
+		Fatalf("image-path flag must be provided")
+	}
+
+	// Create the image file and open for writing.
+	file, err := os.OpenFile(c.imagePath, os.O_CREATE|os.O_EXCL|os.O_RDWR, 0644)
+	if err != nil {
+		Fatalf("os.OpenFile(%q) failed: %v", c.imagePath, err)
+	}
+	defer file.Close()
+
+	if err := cont.Checkpoint(file); err != nil {
 		Fatalf("checkpoint failed: %v", err)
 	}
 

runsc/container/container.go

@@ -339,13 +339,14 @@ func (c *Container) Signal(sig syscall.Signal) error {
 }
 
 // Checkpoint sends the checkpoint call to the container.
-func (c *Container) Checkpoint() error {
+// The statefile will be written to f, the file at the specified image-path.
+func (c *Container) Checkpoint(f *os.File) error {
 	log.Debugf("Checkpoint container %q", c.ID)
 	if c.Status == Stopped {
 		log.Warningf("container %q not running, not checkpointing", c.ID)
 		return nil
 	}
-	return c.Sandbox.Checkpoint(c.ID)
+	return c.Sandbox.Checkpoint(c.ID, f)
 }
 
 // State returns the metadata of the container.

runsc/container/container_test.go

@@ -408,6 +408,57 @@ func TestExec(t *testing.T) {
 	}
 }
 
+// TestCheckpoint verifies that calling checkpoint with an image-path flag succeeds.
+// Since there is no current default image path, confirming that calling
+// checkpoint without an image path fails.
+// Checks that there is a file with the name and location given by image path.
+func TestCheckpoint(t *testing.T) {
+	// Container will succeed.
+	spec := testutil.NewSpecWithArgs("sleep", "100")
+
+	rootDir, bundleDir, conf, err := testutil.SetupContainer(spec)
+	if err != nil {
+		t.Fatalf("error setting up container: %v", err)
+	}
+	defer os.RemoveAll(rootDir)
+	defer os.RemoveAll(bundleDir)
+
+	// Create and start the container.
+	cont, err := container.Create(testutil.UniqueContainerID(), spec, conf, bundleDir, "", "")
+	if err != nil {
+		t.Fatalf("error creating container: %v", err)
+	}
+	defer cont.Destroy()
+	if err := cont.Start(conf); err != nil {
+		t.Fatalf("error starting container: %v", err)
+	}
+
+	// Set the image path, which is where the checkpoint image will be saved.
+	imagePath := filepath.Join(os.TempDir(), "test-image-file")
+
+	// Create the image file and open for writing.
+	file, err := os.OpenFile(imagePath, os.O_CREATE|os.O_EXCL|os.O_RDWR, 0644)
+	if err != nil {
+		t.Fatalf("error opening new file at imagePath: %v", err)
+	}
+	defer file.Close()
+
+	// Checkpoint running container; save state into new file.
+	if err := cont.Checkpoint(file); err != nil {
+		t.Fatalf("error checkpointing container to empty file: %v", err)
+	}
+	defer os.RemoveAll(imagePath)
+
+	// Check to see if file exists and contains data.
+	fileInfo, err := os.Stat(imagePath)
+	if err != nil {
+		t.Fatalf("error checkpointing container: %v", err)
+	}
+	if size := fileInfo.Size(); size == 0 {
+		t.Fatalf("failed checkpoint, file still appears empty: %v", err)
+	}
+}
+
 // TestCapabilities verifies that:
 // - Running exec as non-root UID and GID will result in an error (because the
 //   executable file can't be read).
@@ -602,7 +653,7 @@ func TestSpecUnsupported(t *testing.T) {
 }
 
 // TestRunNonRoot checks that sandbox can be configured when running as
-// non-priviledged user.
+// non-privileged user.
 func TestRunNonRoot(t *testing.T) {
 	spec := testutil.NewSpecWithArgs("/bin/true")
 	spec.Process.User.UID = 343

runsc/sandbox/sandbox.go

@@ -441,7 +441,8 @@ func (s *Sandbox) Signal(cid string, sig syscall.Signal) error {
 }
 
 // Checkpoint sends the checkpoint call for a container in the sandbox.
-func (s *Sandbox) Checkpoint(cid string) error {
+// The statefile will be written to f.
+func (s *Sandbox) Checkpoint(cid string, f *os.File) error {
 	log.Debugf("Checkpoint sandbox %q", s.ID)
 	conn, err := s.connect()
 	if err != nil {
@@ -449,7 +450,13 @@ func (s *Sandbox) Checkpoint(cid string) error {
 	}
 	defer conn.Close()
 
-	if err := conn.Call(boot.ContainerCheckpoint, nil, nil); err != nil {
+	opt := control.SaveOpts{
+		FilePayload: urpc.FilePayload{
+			Files: []*os.File{f},
+		},
+	}
+
+	if err := conn.Call(boot.ContainerCheckpoint, &opt, nil); err != nil {
 		return fmt.Errorf("err checkpointing container %q: %v", cid, err)
 	}
 	return nil