Merge pull request #11208 from LK4D4/new_libcontainer_api

New libcontainer api

daemon/container.go

@@ -14,6 +14,8 @@ import (
 	"syscall"
 	"time"
 
+	"github.com/docker/libcontainer"
+	"github.com/docker/libcontainer/configs"
 	"github.com/docker/libcontainer/devices"
 	"github.com/docker/libcontainer/label"
 
@@ -259,18 +261,18 @@ func populateCommand(c *Container, env []string) error {
 	pid.HostPid = c.hostConfig.PidMode.IsHost()
 
 	// Build lists of devices allowed and created within the container.
-	userSpecifiedDevices := make([]*devices.Device, len(c.hostConfig.Devices))
+	userSpecifiedDevices := make([]*configs.Device, len(c.hostConfig.Devices))
 	for i, deviceMapping := range c.hostConfig.Devices {
-		device, err := devices.GetDevice(deviceMapping.PathOnHost, deviceMapping.CgroupPermissions)
+		device, err := devices.DeviceFromPath(deviceMapping.PathOnHost, deviceMapping.CgroupPermissions)
 		if err != nil {
 			return fmt.Errorf("error gathering device information while adding custom device %q: %s", deviceMapping.PathOnHost, err)
 		}
 		device.Path = deviceMapping.PathInContainer
 		userSpecifiedDevices[i] = device
 	}
-	allowedDevices := append(devices.DefaultAllowedDevices, userSpecifiedDevices...)
+	allowedDevices := append(configs.DefaultAllowedDevices, userSpecifiedDevices...)
 
-	autoCreatedDevices := append(devices.DefaultAutoCreatedDevices, userSpecifiedDevices...)
+	autoCreatedDevices := append(configs.DefaultAutoCreatedDevices, userSpecifiedDevices...)
 
 	// TODO: this can be removed after lxc-conf is fully deprecated
 	lxcConfig, err := mergeLxcConfIntoOptions(c.hostConfig)
@@ -972,7 +974,7 @@ func (container *Container) Exposes(p nat.Port) bool {
 	return exists
 }
 
-func (container *Container) GetPtyMaster() (*os.File, error) {
+func (container *Container) GetPtyMaster() (libcontainer.Console, error) {
 	ttyConsole, ok := container.command.ProcessConfig.Terminal.(execdriver.TtyTerminal)
 	if !ok {
 		return nil, ErrNoTTY

daemon/execdriver/driver.go

@@ -1,17 +1,22 @@
 package execdriver
 
 import (
+	"encoding/json"
 	"errors"
 	"io"
+	"io/ioutil"
 	"os"
 	"os/exec"
+	"path/filepath"
+	"strconv"
 	"strings"
 	"time"
 
 	"github.com/docker/docker/daemon/execdriver/native/template"
 	"github.com/docker/docker/pkg/ulimit"
 	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/devices"
+	"github.com/docker/libcontainer/cgroups/fs"
+	"github.com/docker/libcontainer/configs"
 )
 
 // Context is a generic key value pair that allows
@@ -42,7 +47,7 @@ type Terminal interface {
 }
 
 type TtyTerminal interface {
-	Master() *os.File
+	Master() libcontainer.Console
 }
 
 // ExitStatus provides exit reasons for a container.
@@ -109,7 +114,7 @@ type Resources struct {
 }
 
 type ResourceStats struct {
-	*libcontainer.ContainerStats
+	*libcontainer.Stats
 	Read        time.Time `json:"read"`
 	MemoryLimit int64     `json:"memory_limit"`
 	SystemUsage uint64    `json:"system_usage"`
@@ -149,8 +154,8 @@ type Command struct {
 	Pid                *Pid              `json:"pid"`
 	Resources          *Resources        `json:"resources"`
 	Mounts             []Mount           `json:"mounts"`
-	AllowedDevices     []*devices.Device `json:"allowed_devices"`
-	AutoCreatedDevices []*devices.Device `json:"autocreated_devices"`
+	AllowedDevices     []*configs.Device `json:"allowed_devices"`
+	AutoCreatedDevices []*configs.Device `json:"autocreated_devices"`
 	CapAdd             []string          `json:"cap_add"`
 	CapDrop            []string          `json:"cap_drop"`
 	ContainerPid       int               `json:"container_pid"`  // the pid for the process inside a container
@@ -161,23 +166,19 @@ type Command struct {
 	AppArmorProfile    string            `json:"apparmor_profile"`
 }
 
-func InitContainer(c *Command) *libcontainer.Config {
+func InitContainer(c *Command) *configs.Config {
 	container := template.New()
 
 	container.Hostname = getEnv("HOSTNAME", c.ProcessConfig.Env)
-	container.Tty = c.ProcessConfig.Tty
-	container.User = c.ProcessConfig.User
-	container.WorkingDir = c.WorkingDir
-	container.Env = c.ProcessConfig.Env
 	container.Cgroups.Name = c.ID
 	container.Cgroups.AllowedDevices = c.AllowedDevices
-	container.MountConfig.DeviceNodes = c.AutoCreatedDevices
-	container.RootFs = c.Rootfs
-	container.MountConfig.ReadonlyFs = c.ReadonlyRootfs
+	container.Readonlyfs = c.ReadonlyRootfs
+	container.Devices = c.AutoCreatedDevices
+	container.Rootfs = c.Rootfs
+	container.Readonlyfs = c.ReadonlyRootfs
 
 	// check to see if we are running in ramdisk to disable pivot root
-	container.MountConfig.NoPivotRoot = os.Getenv("DOCKER_RAMDISK") != ""
-	container.RestrictSys = true
+	container.NoPivotRoot = os.Getenv("DOCKER_RAMDISK") != ""
 	return container
 }
 
@@ -191,7 +192,7 @@ func getEnv(key string, env []string) string {
 	return ""
 }
 
-func SetupCgroups(container *libcontainer.Config, c *Command) error {
+func SetupCgroups(container *configs.Config, c *Command) error {
 	if c.Resources != nil {
 		container.Cgroups.CpuShares = c.Resources.CpuShares
 		container.Cgroups.Memory = c.Resources.Memory
@@ -203,28 +204,98 @@ func SetupCgroups(container *libcontainer.Config, c *Command) error {
 	return nil
 }
 
-func Stats(stateFile string, containerMemoryLimit int64, machineMemory int64) (*ResourceStats, error) {
-	state, err := libcontainer.GetState(stateFile)
-	if err != nil {
-		if os.IsNotExist(err) {
-			return nil, ErrNotRunning
+// Returns the network statistics for the network interfaces represented by the NetworkRuntimeInfo.
+func getNetworkInterfaceStats(interfaceName string) (*libcontainer.NetworkInterface, error) {
+	out := &libcontainer.NetworkInterface{Name: interfaceName}
+	// This can happen if the network runtime information is missing - possible if the
+	// container was created by an old version of libcontainer.
+	if interfaceName == "" {
+		return out, nil
+	}
+	type netStatsPair struct {
+		// Where to write the output.
+		Out *uint64
+		// The network stats file to read.
+		File string
+	}
+	// Ingress for host veth is from the container. Hence tx_bytes stat on the host veth is actually number of bytes received by the container.
+	netStats := []netStatsPair{
+		{Out: &out.RxBytes, File: "tx_bytes"},
+		{Out: &out.RxPackets, File: "tx_packets"},
+		{Out: &out.RxErrors, File: "tx_errors"},
+		{Out: &out.RxDropped, File: "tx_dropped"},
+
+		{Out: &out.TxBytes, File: "rx_bytes"},
+		{Out: &out.TxPackets, File: "rx_packets"},
+		{Out: &out.TxErrors, File: "rx_errors"},
+		{Out: &out.TxDropped, File: "rx_dropped"},
+	}
+	for _, netStat := range netStats {
+		data, err := readSysfsNetworkStats(interfaceName, netStat.File)
+		if err != nil {
+			return nil, err
 		}
+		*(netStat.Out) = data
+	}
+	return out, nil
+}
+
+// Reads the specified statistics available under /sys/class/net/<EthInterface>/statistics
+func readSysfsNetworkStats(ethInterface, statsFile string) (uint64, error) {
+	data, err := ioutil.ReadFile(filepath.Join("/sys/class/net", ethInterface, "statistics", statsFile))
+	if err != nil {
+		return 0, err
+	}
+	return strconv.ParseUint(strings.TrimSpace(string(data)), 10, 64)
+}
+
+func Stats(containerDir string, containerMemoryLimit int64, machineMemory int64) (*ResourceStats, error) {
+	f, err := os.Open(filepath.Join(containerDir, "state.json"))
+	if err != nil {
+		return nil, err
+	}
+	defer f.Close()
+
+	type network struct {
+		Type              string
+		HostInterfaceName string
+	}
+
+	state := struct {
+		CgroupPaths map[string]string `json:"cgroup_paths"`
+		Networks    []network
+	}{}
+
+	if err := json.NewDecoder(f).Decode(&state); err != nil {
 		return nil, err
 	}
 	now := time.Now()
-	stats, err := libcontainer.GetStats(nil, state)
+
+	mgr := fs.Manager{Paths: state.CgroupPaths}
+	cstats, err := mgr.GetStats()
 	if err != nil {
 		return nil, err
 	}
+	stats := &libcontainer.Stats{CgroupStats: cstats}
 	// if the container does not have any memory limit specified set the
 	// limit to the machines memory
 	memoryLimit := containerMemoryLimit
 	if memoryLimit == 0 {
 		memoryLimit = machineMemory
 	}
+	for _, iface := range state.Networks {
+		switch iface.Type {
+		case "veth":
+			istats, err := getNetworkInterfaceStats(iface.HostInterfaceName)
+			if err != nil {
+				return nil, err
+			}
+			stats.Interfaces = append(stats.Interfaces, istats)
+		}
+	}
 	return &ResourceStats{
-		Read:           now,
-		ContainerStats: stats,
-		MemoryLimit:    memoryLimit,
+		Stats:       stats,
+		Read:        now,
+		MemoryLimit: memoryLimit,
 	}, nil
 }

daemon/execdriver/lxc/driver.go

@@ -23,7 +23,9 @@ import (
 	"github.com/docker/docker/utils"
 	"github.com/docker/libcontainer"
 	"github.com/docker/libcontainer/cgroups"
-	"github.com/docker/libcontainer/mount/nodes"
+	"github.com/docker/libcontainer/configs"
+	"github.com/docker/libcontainer/system"
+	"github.com/docker/libcontainer/user"
 	"github.com/kr/pty"
 )
 
@@ -42,7 +44,7 @@ type driver struct {
 }
 
 type activeContainer struct {
-	container *libcontainer.Config
+	container *configs.Config
 	cmd       *exec.Cmd
 }
 
@@ -190,7 +192,7 @@ func (d *driver) Run(c *execdriver.Command, pipes *execdriver.Pipes, startCallba
 	c.ProcessConfig.Path = aname
 	c.ProcessConfig.Args = append([]string{name}, arg...)
 
-	if err := nodes.CreateDeviceNodes(c.Rootfs, c.AutoCreatedDevices); err != nil {
+	if err := createDeviceNodes(c.Rootfs, c.AutoCreatedDevices); err != nil {
 		return execdriver.ExitStatus{ExitCode: -1}, err
 	}
 
@@ -231,11 +233,17 @@ func (d *driver) Run(c *execdriver.Command, pipes *execdriver.Pipes, startCallba
 	}
 
 	state := &libcontainer.State{
-		InitPid:     pid,
-		CgroupPaths: cgroupPaths,
+		InitProcessPid: pid,
+		CgroupPaths:    cgroupPaths,
 	}
 
-	if err := libcontainer.SaveState(dataPath, state); err != nil {
+	f, err := os.Create(filepath.Join(dataPath, "state.json"))
+	if err != nil {
+		return terminate(err)
+	}
+	defer f.Close()
+
+	if err := json.NewEncoder(f).Encode(state); err != nil {
 		return terminate(err)
 	}
 
@@ -245,8 +253,9 @@ func (d *driver) Run(c *execdriver.Command, pipes *execdriver.Pipes, startCallba
 		log.Debugf("Invoking startCallback")
 		startCallback(&c.ProcessConfig, pid)
 	}
+
 	oomKill := false
-	oomKillNotification, err := libcontainer.NotifyOnOOM(state)
+	oomKillNotification, err := notifyOnOOM(cgroupPaths)
 	if err == nil {
 		_, oomKill = <-oomKillNotification
 		log.Debugf("oomKill error %s waitErr %s", oomKill, waitErr)
@@ -265,9 +274,57 @@ func (d *driver) Run(c *execdriver.Command, pipes *execdriver.Pipes, startCallba
 	return execdriver.ExitStatus{ExitCode: exitCode, OOMKilled: oomKill}, waitErr
 }
 
+// copy from libcontainer
+func notifyOnOOM(paths map[string]string) (<-chan struct{}, error) {
+	dir := paths["memory"]
+	if dir == "" {
+		return nil, fmt.Errorf("There is no path for %q in state", "memory")
+	}
+	oomControl, err := os.Open(filepath.Join(dir, "memory.oom_control"))
+	if err != nil {
+		return nil, err
+	}
+	fd, _, syserr := syscall.RawSyscall(syscall.SYS_EVENTFD2, 0, syscall.FD_CLOEXEC, 0)
+	if syserr != 0 {
+		oomControl.Close()
+		return nil, syserr
+	}
+
+	eventfd := os.NewFile(fd, "eventfd")
+
+	eventControlPath := filepath.Join(dir, "cgroup.event_control")
+	data := fmt.Sprintf("%d %d", eventfd.Fd(), oomControl.Fd())
+	if err := ioutil.WriteFile(eventControlPath, []byte(data), 0700); err != nil {
+		eventfd.Close()
+		oomControl.Close()
+		return nil, err
+	}
+	ch := make(chan struct{})
+	go func() {
+		defer func() {
+			close(ch)
+			eventfd.Close()
+			oomControl.Close()
+		}()
+		buf := make([]byte, 8)
+		for {
+			if _, err := eventfd.Read(buf); err != nil {
+				return
+			}
+			// When a cgroup is destroyed, an event is sent to eventfd.
+			// So if the control path is gone, return instead of notifying.
+			if _, err := os.Lstat(eventControlPath); os.IsNotExist(err) {
+				return
+			}
+			ch <- struct{}{}
+		}
+	}()
+	return ch, nil
+}
+
 // createContainer populates and configures the container type with the
 // data provided by the execdriver.Command
-func (d *driver) createContainer(c *execdriver.Command) (*libcontainer.Config, error) {
+func (d *driver) createContainer(c *execdriver.Command) (*configs.Config, error) {
 	container := execdriver.InitContainer(c)
 	if err := execdriver.SetupCgroups(container, c); err != nil {
 		return nil, err
@@ -297,6 +354,87 @@ func cgroupPaths(containerId string) (map[string]string, error) {
 	return paths, nil
 }
 
+// this is copy from old libcontainer nodes.go
+func createDeviceNodes(rootfs string, nodesToCreate []*configs.Device) error {
+	oldMask := syscall.Umask(0000)
+	defer syscall.Umask(oldMask)
+
+	for _, node := range nodesToCreate {
+		if err := createDeviceNode(rootfs, node); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// Creates the device node in the rootfs of the container.
+func createDeviceNode(rootfs string, node *configs.Device) error {
+	var (
+		dest   = filepath.Join(rootfs, node.Path)
+		parent = filepath.Dir(dest)
+	)
+
+	if err := os.MkdirAll(parent, 0755); err != nil {
+		return err
+	}
+
+	fileMode := node.FileMode
+	switch node.Type {
+	case 'c':
+		fileMode |= syscall.S_IFCHR
+	case 'b':
+		fileMode |= syscall.S_IFBLK
+	default:
+		return fmt.Errorf("%c is not a valid device type for device %s", node.Type, node.Path)
+	}
+
+	if err := syscall.Mknod(dest, uint32(fileMode), node.Mkdev()); err != nil && !os.IsExist(err) {
+		return fmt.Errorf("mknod %s %s", node.Path, err)
+	}
+
+	if err := syscall.Chown(dest, int(node.Uid), int(node.Gid)); err != nil {
+		return fmt.Errorf("chown %s to %d:%d", node.Path, node.Uid, node.Gid)
+	}
+
+	return nil
+}
+
+// setupUser changes the groups, gid, and uid for the user inside the container
+// copy from libcontainer, cause not it's private
+func setupUser(userSpec string) error {
+	// Set up defaults.
+	defaultExecUser := user.ExecUser{
+		Uid:  syscall.Getuid(),
+		Gid:  syscall.Getgid(),
+		Home: "/",
+	}
+	passwdPath, err := user.GetPasswdPath()
+	if err != nil {
+		return err
+	}
+	groupPath, err := user.GetGroupPath()
+	if err != nil {
+		return err
+	}
+	execUser, err := user.GetExecUserPath(userSpec, &defaultExecUser, passwdPath, groupPath)
+	if err != nil {
+		return err
+	}
+	if err := system.Setgid(execUser.Gid); err != nil {
+		return err
+	}
+	if err := system.Setuid(execUser.Uid); err != nil {
+		return err
+	}
+	// if we didn't get HOME already, set it based on the user's HOME
+	if envHome := os.Getenv("HOME"); envHome == "" {
+		if err := os.Setenv("HOME", execUser.Home); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
 /// Return the exit code of the process
 // if the process has not exited -1 will be returned
 func getExitCode(c *execdriver.Command) int {

daemon/execdriver/lxc/lxc_init_linux.go

@@ -3,8 +3,6 @@ package lxc
 import (
 	"fmt"
 
-	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/namespaces"
 	"github.com/docker/libcontainer/utils"
 )
 
@@ -12,9 +10,7 @@ func finalizeNamespace(args *InitArgs) error {
 	if err := utils.CloseExecFrom(3); err != nil {
 		return err
 	}
-	if err := namespaces.SetupUser(&libcontainer.Config{
-		User: args.User,
-	}); err != nil {
+	if err := setupUser(args.User); err != nil {
 		return fmt.Errorf("setup user %s", err)
 	}
 	if err := setupWorkingDirectory(args); err != nil {

daemon/execdriver/lxc/lxc_template.go

@@ -11,7 +11,6 @@ import (
 	nativeTemplate "github.com/docker/docker/daemon/execdriver/native/template"
 	"github.com/docker/docker/utils"
 	"github.com/docker/libcontainer/label"
-	"github.com/docker/libcontainer/security/capabilities"
 )
 
 const LxcTemplate = `
@@ -169,7 +168,7 @@ func keepCapabilities(adds []string, drops []string) ([]string, error) {
 	var newCaps []string
 	for _, cap := range caps {
 		log.Debugf("cap %s\n", cap)
-		realCap := capabilities.GetCapability(cap)
+		realCap := execdriver.GetCapability(cap)
 		numCap := fmt.Sprintf("%d", realCap.Value)
 		newCaps = append(newCaps, numCap)
 	}
@@ -180,13 +179,10 @@ func keepCapabilities(adds []string, drops []string) ([]string, error) {
 func dropList(drops []string) ([]string, error) {
 	if utils.StringsContainsNoCase(drops, "all") {
 		var newCaps []string
-		for _, cap := range capabilities.GetAllCapabilities() {
-			log.Debugf("drop cap %s\n", cap)
-			realCap := capabilities.GetCapability(cap)
-			if realCap == nil {
-				return nil, fmt.Errorf("Invalid capability '%s'", cap)
-			}
-			numCap := fmt.Sprintf("%d", realCap.Value)
+		for _, capName := range execdriver.GetAllCapabilities() {
+			cap := execdriver.GetCapability(capName)
+			log.Debugf("drop cap %s\n", cap.Key)
+			numCap := fmt.Sprintf("%d", cap.Value)
 			newCaps = append(newCaps, numCap)
 		}
 		return newCaps, nil

daemon/execdriver/lxc/lxc_template_unit_test.go

@@ -5,11 +5,6 @@ package lxc
 import (
 	"bufio"
 	"fmt"
-	"github.com/docker/docker/daemon/execdriver"
-	nativeTemplate "github.com/docker/docker/daemon/execdriver/native/template"
-	"github.com/docker/libcontainer/devices"
-	"github.com/docker/libcontainer/security/capabilities"
-	"github.com/syndtr/gocapability/capability"
 	"io/ioutil"
 	"math/rand"
 	"os"
@@ -17,6 +12,11 @@ import (
 	"strings"
 	"testing"
 	"time"
+
+	"github.com/docker/docker/daemon/execdriver"
+	nativeTemplate "github.com/docker/docker/daemon/execdriver/native/template"
+	"github.com/docker/libcontainer/configs"
+	"github.com/syndtr/gocapability/capability"
 )
 
 func TestLXCConfig(t *testing.T) {
@@ -53,7 +53,7 @@ func TestLXCConfig(t *testing.T) {
 			Mtu:       1500,
 			Interface: nil,
 		},
-		AllowedDevices: make([]*devices.Device, 0),
+		AllowedDevices: make([]*configs.Device, 0),
 		ProcessConfig:  execdriver.ProcessConfig{},
 	}
 	p, err := driver.generateLXCConfig(command)
@@ -295,7 +295,7 @@ func TestCustomLxcConfigMisc(t *testing.T) {
 	grepFile(t, p, "lxc.cgroup.cpuset.cpus = 0,1")
 	container := nativeTemplate.New()
 	for _, cap := range container.Capabilities {
-		realCap := capabilities.GetCapability(cap)
+		realCap := execdriver.GetCapability(cap)
 		numCap := fmt.Sprintf("%d", realCap.Value)
 		if cap != "MKNOD" && cap != "KILL" {
 			grepFile(t, p, fmt.Sprintf("lxc.cap.keep = %s", numCap))
@@ -359,7 +359,7 @@ func TestCustomLxcConfigMiscOverride(t *testing.T) {
 	grepFile(t, p, "lxc.cgroup.cpuset.cpus = 0,1")
 	container := nativeTemplate.New()
 	for _, cap := range container.Capabilities {
-		realCap := capabilities.GetCapability(cap)
+		realCap := execdriver.GetCapability(cap)
 		numCap := fmt.Sprintf("%d", realCap.Value)
 		if cap != "MKNOD" && cap != "KILL" {
 			grepFile(t, p, fmt.Sprintf("lxc.cap.keep = %s", numCap))

daemon/execdriver/native/create.go

@@ -3,21 +3,24 @@
 package native
 
 import (
+	"errors"
 	"fmt"
-	"os/exec"
+	"net"
 	"path/filepath"
+	"strings"
+	"syscall"
 
 	"github.com/docker/docker/daemon/execdriver"
-	"github.com/docker/libcontainer"
+	"github.com/docker/docker/pkg/symlink"
 	"github.com/docker/libcontainer/apparmor"
+	"github.com/docker/libcontainer/configs"
 	"github.com/docker/libcontainer/devices"
-	"github.com/docker/libcontainer/mount"
-	"github.com/docker/libcontainer/security/capabilities"
+	"github.com/docker/libcontainer/utils"
 )
 
 // createContainer populates and configures the container type with the
 // data provided by the execdriver.Command
-func (d *driver) createContainer(c *execdriver.Command) (*libcontainer.Config, error) {
+func (d *driver) createContainer(c *execdriver.Command) (*configs.Config, error) {
 	container := execdriver.InitContainer(c)
 
 	if err := d.createIpc(container, c); err != nil {
@@ -33,6 +36,13 @@ func (d *driver) createContainer(c *execdriver.Command) (*libcontainer.Config, e
 	}
 
 	if c.ProcessConfig.Privileged {
+		// clear readonly for /sys
+		for i := range container.Mounts {
+			if container.Mounts[i].Destination == "/sys" {
+				container.Mounts[i].Flags &= ^syscall.MS_RDONLY
+			}
+		}
+		container.ReadonlyPaths = nil
 		if err := d.setPrivileged(container); err != nil {
 			return nil, err
 		}
@@ -57,43 +67,52 @@ func (d *driver) createContainer(c *execdriver.Command) (*libcontainer.Config, e
 	if err := d.setupLabels(container, c); err != nil {
 		return nil, err
 	}
-
 	d.setupRlimits(container, c)
-
-	cmds := make(map[string]*exec.Cmd)
-	d.Lock()
-	for k, v := range d.activeContainers {
-		cmds[k] = v.cmd
-	}
-	d.Unlock()
-
 	return container, nil
 }
 
-func (d *driver) createNetwork(container *libcontainer.Config, c *execdriver.Command) error {
+func generateIfaceName() (string, error) {
+	for i := 0; i < 10; i++ {
+		name, err := utils.GenerateRandomName("veth", 7)
+		if err != nil {
+			continue
+		}
+		if _, err := net.InterfaceByName(name); err != nil {
+			if strings.Contains(err.Error(), "no such") {
+				return name, nil
+			}
+			return "", err
+		}
+	}
+	return "", errors.New("Failed to find name for new interface")
+}
+
+func (d *driver) createNetwork(container *configs.Config, c *execdriver.Command) error {
 	if c.Network.HostNetworking {
-		container.Namespaces.Remove(libcontainer.NEWNET)
+		container.Namespaces.Remove(configs.NEWNET)
 		return nil
 	}
 
-	container.Networks = []*libcontainer.Network{
+	container.Networks = []*configs.Network{
 		{
-			Mtu:     c.Network.Mtu,
-			Address: fmt.Sprintf("%s/%d", "127.0.0.1", 0),
-			Gateway: "localhost",
-			Type:    "loopback",
+			Type: "loopback",
 		},
 	}
 
+	iName, err := generateIfaceName()
+	if err != nil {
+		return err
+	}
 	if c.Network.Interface != nil {
-		vethNetwork := libcontainer.Network{
-			Mtu:        c.Network.Mtu,
-			Address:    fmt.Sprintf("%s/%d", c.Network.Interface.IPAddress, c.Network.Interface.IPPrefixLen),
-			MacAddress: c.Network.Interface.MacAddress,
-			Gateway:    c.Network.Interface.Gateway,
-			Type:       "veth",
-			Bridge:     c.Network.Interface.Bridge,
-			VethPrefix: "veth",
+		vethNetwork := configs.Network{
+			Name:              "eth0",
+			HostInterfaceName: iName,
+			Mtu:               c.Network.Mtu,
+			Address:           fmt.Sprintf("%s/%d", c.Network.Interface.IPAddress, c.Network.Interface.IPPrefixLen),
+			MacAddress:        c.Network.Interface.MacAddress,
+			Gateway:           c.Network.Interface.Gateway,
+			Type:              "veth",
+			Bridge:            c.Network.Interface.Bridge,
 		}
 		if c.Network.Interface.GlobalIPv6Address != "" {
 			vethNetwork.IPv6Address = fmt.Sprintf("%s/%d", c.Network.Interface.GlobalIPv6Address, c.Network.Interface.GlobalIPv6PrefixLen)
@@ -107,21 +126,24 @@ func (d *driver) createNetwork(container *libcontainer.Config, c *execdriver.Com
 		active := d.activeContainers[c.Network.ContainerID]
 		d.Unlock()
 
-		if active == nil || active.cmd.Process == nil {
+		if active == nil {
 			return fmt.Errorf("%s is not a valid running container to join", c.Network.ContainerID)
 		}
-		cmd := active.cmd
 
-		nspath := filepath.Join("/proc", fmt.Sprint(cmd.Process.Pid), "ns", "net")
-		container.Namespaces.Add(libcontainer.NEWNET, nspath)
+		state, err := active.State()
+		if err != nil {
+			return err
+		}
+
+		container.Namespaces.Add(configs.NEWNET, state.NamespacePaths[configs.NEWNET])
 	}
 
 	return nil
 }
 
-func (d *driver) createIpc(container *libcontainer.Config, c *execdriver.Command) error {
+func (d *driver) createIpc(container *configs.Config, c *execdriver.Command) error {
 	if c.Ipc.HostIpc {
-		container.Namespaces.Remove(libcontainer.NEWIPC)
+		container.Namespaces.Remove(configs.NEWIPC)
 		return nil
 	}
 
@@ -130,37 +152,38 @@ func (d *driver) createIpc(container *libcontainer.Config, c *execdriver.Command
 		active := d.activeContainers[c.Ipc.ContainerID]
 		d.Unlock()
 
-		if active == nil || active.cmd.Process == nil {
+		if active == nil {
 			return fmt.Errorf("%s is not a valid running container to join", c.Ipc.ContainerID)
 		}
-		cmd := active.cmd
 
-		container.Namespaces.Add(libcontainer.NEWIPC, filepath.Join("/proc", fmt.Sprint(cmd.Process.Pid), "ns", "ipc"))
+		state, err := active.State()
+		if err != nil {
+			return err
+		}
+		container.Namespaces.Add(configs.NEWIPC, state.NamespacePaths[configs.NEWIPC])
 	}
 
 	return nil
 }
 
-func (d *driver) createPid(container *libcontainer.Config, c *execdriver.Command) error {
+func (d *driver) createPid(container *configs.Config, c *execdriver.Command) error {
 	if c.Pid.HostPid {
-		container.Namespaces.Remove(libcontainer.NEWPID)
+		container.Namespaces.Remove(configs.NEWPID)
 		return nil
 	}
 
 	return nil
 }
 
-func (d *driver) setPrivileged(container *libcontainer.Config) (err error) {
-	container.Capabilities = capabilities.GetAllCapabilities()
+func (d *driver) setPrivileged(container *configs.Config) (err error) {
+	container.Capabilities = execdriver.GetAllCapabilities()
 	container.Cgroups.AllowAllDevices = true
 
-	hostDeviceNodes, err := devices.GetHostDeviceNodes()
+	hostDevices, err := devices.HostDevices()
 	if err != nil {
 		return err
 	}
-	container.MountConfig.DeviceNodes = hostDeviceNodes
-
-	container.RestrictSys = false
+	container.Devices = hostDevices
 
 	if apparmor.IsEnabled() {
 		container.AppArmorProfile = "unconfined"
@@ -169,39 +192,52 @@ func (d *driver) setPrivileged(container *libcontainer.Config) (err error) {
 	return nil
 }
 
-func (d *driver) setCapabilities(container *libcontainer.Config, c *execdriver.Command) (err error) {
+func (d *driver) setCapabilities(container *configs.Config, c *execdriver.Command) (err error) {
 	container.Capabilities, err = execdriver.TweakCapabilities(container.Capabilities, c.CapAdd, c.CapDrop)
 	return err
 }
 
-func (d *driver) setupRlimits(container *libcontainer.Config, c *execdriver.Command) {
+func (d *driver) setupRlimits(container *configs.Config, c *execdriver.Command) {
 	if c.Resources == nil {
 		return
 	}
 
 	for _, rlimit := range c.Resources.Rlimits {
-		container.Rlimits = append(container.Rlimits, libcontainer.Rlimit((*rlimit)))
-	}
-}
-
-func (d *driver) setupMounts(container *libcontainer.Config, c *execdriver.Command) error {
-	for _, m := range c.Mounts {
-		container.MountConfig.Mounts = append(container.MountConfig.Mounts, &mount.Mount{
-			Type:        "bind",
-			Source:      m.Source,
-			Destination: m.Destination,
-			Writable:    m.Writable,
-			Private:     m.Private,
-			Slave:       m.Slave,
+		container.Rlimits = append(container.Rlimits, configs.Rlimit{
+			Type: rlimit.Type,
+			Hard: rlimit.Hard,
+			Soft: rlimit.Soft,
 		})
 	}
+}
 
+func (d *driver) setupMounts(container *configs.Config, c *execdriver.Command) error {
+	for _, m := range c.Mounts {
+		dest, err := symlink.FollowSymlinkInScope(filepath.Join(c.Rootfs, m.Destination), c.Rootfs)
+		if err != nil {
+			return err
+		}
+		flags := syscall.MS_BIND | syscall.MS_REC
+		if !m.Writable {
+			flags |= syscall.MS_RDONLY
+		}
+		if m.Slave {
+			flags |= syscall.MS_SLAVE
+		}
+
+		container.Mounts = append(container.Mounts, &configs.Mount{
+			Source:      m.Source,
+			Destination: dest,
+			Device:      "bind",
+			Flags:       flags,
+		})
+	}
 	return nil
 }
 
-func (d *driver) setupLabels(container *libcontainer.Config, c *execdriver.Command) error {
+func (d *driver) setupLabels(container *configs.Config, c *execdriver.Command) error {
 	container.ProcessLabel = c.ProcessLabel
-	container.MountConfig.MountLabel = c.MountLabel
+	container.MountLabel = c.MountLabel
 
 	return nil
 }

daemon/execdriver/native/driver.go

@@ -4,28 +4,28 @@ package native
 
 import (
 	"encoding/json"
-	"errors"
 	"fmt"
 	"io"
 	"io/ioutil"
 	"os"
 	"os/exec"
 	"path/filepath"
+	"strings"
 	"sync"
 	"syscall"
+	"time"
 
 	log "github.com/Sirupsen/logrus"
 	"github.com/docker/docker/daemon/execdriver"
+	"github.com/docker/docker/pkg/reexec"
 	sysinfo "github.com/docker/docker/pkg/system"
 	"github.com/docker/docker/pkg/term"
 	"github.com/docker/libcontainer"
 	"github.com/docker/libcontainer/apparmor"
-	"github.com/docker/libcontainer/cgroups/fs"
 	"github.com/docker/libcontainer/cgroups/systemd"
-	consolepkg "github.com/docker/libcontainer/console"
-	"github.com/docker/libcontainer/namespaces"
-	_ "github.com/docker/libcontainer/namespaces/nsenter"
+	"github.com/docker/libcontainer/configs"
 	"github.com/docker/libcontainer/system"
+	"github.com/docker/libcontainer/utils"
 )
 
 const (
@@ -33,16 +33,12 @@ const (
 	Version    = "0.2"
 )
 
-type activeContainer struct {
-	container *libcontainer.Config
-	cmd       *exec.Cmd
-}
-
 type driver struct {
 	root             string
 	initPath         string
-	activeContainers map[string]*activeContainer
+	activeContainers map[string]libcontainer.Container
 	machineMemory    int64
+	factory          libcontainer.Factory
 	sync.Mutex
 }
 
@@ -59,11 +55,22 @@ func NewDriver(root, initPath string) (*driver, error) {
 	if err := apparmor.InstallDefaultProfile(); err != nil {
 		return nil, err
 	}
+	cgm := libcontainer.Cgroupfs
+	if systemd.UseSystemd() {
+		cgm = libcontainer.SystemdCgroups
+	}
+
+	f, err := libcontainer.New(root, cgm, libcontainer.InitPath(reexec.Self(), DriverName))
+	if err != nil {
+		return nil, err
+	}
+
 	return &driver{
 		root:             root,
 		initPath:         initPath,
-		activeContainers: make(map[string]*activeContainer),
+		activeContainers: make(map[string]libcontainer.Container),
 		machineMemory:    meminfo.MemTotal,
+		factory:          f,
 	}, nil
 }
 
@@ -81,101 +88,141 @@ func (d *driver) Run(c *execdriver.Command, pipes *execdriver.Pipes, startCallba
 
 	var term execdriver.Terminal
 
+	p := &libcontainer.Process{
+		Args: append([]string{c.ProcessConfig.Entrypoint}, c.ProcessConfig.Arguments...),
+		Env:  c.ProcessConfig.Env,
+		Cwd:  c.WorkingDir,
+		User: c.ProcessConfig.User,
+	}
+
 	if c.ProcessConfig.Tty {
-		term, err = NewTtyConsole(&c.ProcessConfig, pipes)
+		rootuid, err := container.HostUID()
+		if err != nil {
+			return execdriver.ExitStatus{ExitCode: -1}, err
+		}
+		cons, err := p.NewConsole(rootuid)
+		if err != nil {
+			return execdriver.ExitStatus{ExitCode: -1}, err
+		}
+		term, err = NewTtyConsole(cons, pipes, rootuid)
 	} else {
-		term, err = execdriver.NewStdConsole(&c.ProcessConfig, pipes)
+		p.Stdout = pipes.Stdout
+		p.Stderr = pipes.Stderr
+		r, w, err := os.Pipe()
+		if err != nil {
+			return execdriver.ExitStatus{ExitCode: -1}, err
+		}
+		if pipes.Stdin != nil {
+			go func() {
+				io.Copy(w, pipes.Stdin)
+				w.Close()
+			}()
+			p.Stdin = r
+		}
+		term = &execdriver.StdConsole{}
 	}
 	if err != nil {
 		return execdriver.ExitStatus{ExitCode: -1}, err
 	}
 	c.ProcessConfig.Terminal = term
 
+	cont, err := d.factory.Create(c.ID, container)
+	if err != nil {
+		return execdriver.ExitStatus{ExitCode: -1}, err
+	}
 	d.Lock()
-	d.activeContainers[c.ID] = &activeContainer{
-		container: container,
-		cmd:       &c.ProcessConfig.Cmd,
-	}
+	d.activeContainers[c.ID] = cont
 	d.Unlock()
-
-	var (
-		dataPath = filepath.Join(d.root, c.ID)
-		args     = append([]string{c.ProcessConfig.Entrypoint}, c.ProcessConfig.Arguments...)
-	)
-
-	if err := d.createContainerRoot(c.ID); err != nil {
-		return execdriver.ExitStatus{ExitCode: -1}, err
-	}
-	defer d.cleanContainer(c.ID)
-
-	if err := d.writeContainerFile(container, c.ID); err != nil {
-		return execdriver.ExitStatus{ExitCode: -1}, err
-	}
-
-	execOutputChan := make(chan execOutput, 1)
-	waitForStart := make(chan struct{})
-
-	go func() {
-		exitCode, err := namespaces.Exec(container, c.ProcessConfig.Stdin, c.ProcessConfig.Stdout, c.ProcessConfig.Stderr, c.ProcessConfig.Console, dataPath, args, func(container *libcontainer.Config, console, dataPath, init string, child *os.File, args []string) *exec.Cmd {
-			c.ProcessConfig.Path = d.initPath
-			c.ProcessConfig.Args = append([]string{
-				DriverName,
-				"-console", console,
-				"-pipe", "3",
-				"-root", filepath.Join(d.root, c.ID),
-				"--",
-			}, args...)
-
-			// set this to nil so that when we set the clone flags anything else is reset
-			c.ProcessConfig.SysProcAttr = &syscall.SysProcAttr{
-				Cloneflags: uintptr(namespaces.GetNamespaceFlags(container.Namespaces)),
-			}
-			c.ProcessConfig.ExtraFiles = []*os.File{child}
-
-			c.ProcessConfig.Env = container.Env
-			c.ProcessConfig.Dir = container.RootFs
-
-			return &c.ProcessConfig.Cmd
-		}, func() {
-			close(waitForStart)
-			if startCallback != nil {
-				c.ContainerPid = c.ProcessConfig.Process.Pid
-				startCallback(&c.ProcessConfig, c.ContainerPid)
-			}
-		})
-		execOutputChan <- execOutput{exitCode, err}
+	defer func() {
+		cont.Destroy()
+		d.cleanContainer(c.ID)
 	}()
 
-	select {
-	case execOutput := <-execOutputChan:
-		return execdriver.ExitStatus{ExitCode: execOutput.exitCode}, execOutput.err
-	case <-waitForStart:
-		break
+	if err := cont.Start(p); err != nil {
+		return execdriver.ExitStatus{ExitCode: -1}, err
 	}
 
-	oomKill := false
-	state, err := libcontainer.GetState(filepath.Join(d.root, c.ID))
-	if err == nil {
-		oomKillNotification, err := libcontainer.NotifyOnOOM(state)
-		if err == nil {
-			_, oomKill = <-oomKillNotification
-		} else {
-			log.Warnf("WARNING: Your kernel does not support OOM notifications: %s", err)
+	if startCallback != nil {
+		pid, err := p.Pid()
+		if err != nil {
+			p.Signal(os.Kill)
+			p.Wait()
+			return execdriver.ExitStatus{ExitCode: -1}, err
 		}
-	} else {
-		log.Warnf("Failed to get container state, oom notify will not work: %s", err)
+		startCallback(&c.ProcessConfig, pid)
 	}
-	// wait for the container to exit.
-	execOutput := <-execOutputChan
 
-	return execdriver.ExitStatus{ExitCode: execOutput.exitCode, OOMKilled: oomKill}, execOutput.err
+	oomKillNotification, err := cont.NotifyOOM()
+	if err != nil {
+		oomKillNotification = nil
+		log.Warnf("WARNING: Your kernel does not support OOM notifications: %s", err)
+	}
+	waitF := p.Wait
+	if nss := cont.Config().Namespaces; nss.Contains(configs.NEWPID) {
+		// we need such hack for tracking processes with inerited fds,
+		// because cmd.Wait() waiting for all streams to be copied
+		waitF = waitInPIDHost(p, cont)
+	}
+	ps, err := waitF()
+	if err != nil {
+		if err, ok := err.(*exec.ExitError); !ok {
+			return execdriver.ExitStatus{ExitCode: -1}, err
+		} else {
+			ps = err.ProcessState
+		}
+	}
+	cont.Destroy()
+
+	_, oomKill := <-oomKillNotification
+
+	return execdriver.ExitStatus{ExitCode: utils.ExitStatus(ps.Sys().(syscall.WaitStatus)), OOMKilled: oomKill}, nil
 }
 
-func (d *driver) Kill(p *execdriver.Command, sig int) error {
-	if p.ProcessConfig.Process == nil {
-		return errors.New("exec: not started")
+func waitInPIDHost(p *libcontainer.Process, c libcontainer.Container) func() (*os.ProcessState, error) {
+	return func() (*os.ProcessState, error) {
+		pid, err := p.Pid()
+		if err != nil {
+			return nil, err
+		}
+
+		process, err := os.FindProcess(pid)
+		s, err := process.Wait()
+		if err != nil {
+			if err, ok := err.(*exec.ExitError); !ok {
+				return s, err
+			} else {
+				s = err.ProcessState
+			}
+		}
+		processes, err := c.Processes()
+		if err != nil {
+			return s, err
+		}
+
+		for _, pid := range processes {
+			process, err := os.FindProcess(pid)
+			if err != nil {
+				log.Errorf("Failed to kill process: %d", pid)
+				continue
+			}
+			process.Kill()
+		}
+
+		p.Wait()
+		return s, err
 	}
-	return syscall.Kill(p.ProcessConfig.Process.Pid, syscall.Signal(sig))
+}
+
+func (d *driver) Kill(c *execdriver.Command, sig int) error {
+	active := d.activeContainers[c.ID]
+	if active == nil {
+		return fmt.Errorf("active container for %s does not exist", c.ID)
+	}
+	state, err := active.State()
+	if err != nil {
+		return err
+	}
+	return syscall.Kill(state.InitProcessPid, syscall.Signal(sig))
 }
 
 func (d *driver) Pause(c *execdriver.Command) error {
@@ -183,11 +230,7 @@ func (d *driver) Pause(c *execdriver.Command) error {
 	if active == nil {
 		return fmt.Errorf("active container for %s does not exist", c.ID)
 	}
-	active.container.Cgroups.Freezer = "FROZEN"
-	if systemd.UseSystemd() {
-		return systemd.Freeze(active.container.Cgroups, active.container.Cgroups.Freezer)
-	}
-	return fs.Freeze(active.container.Cgroups, active.container.Cgroups.Freezer)
+	return active.Pause()
 }
 
 func (d *driver) Unpause(c *execdriver.Command) error {
@@ -195,44 +238,31 @@ func (d *driver) Unpause(c *execdriver.Command) error {
 	if active == nil {
 		return fmt.Errorf("active container for %s does not exist", c.ID)
 	}
-	active.container.Cgroups.Freezer = "THAWED"
-	if systemd.UseSystemd() {
-		return systemd.Freeze(active.container.Cgroups, active.container.Cgroups.Freezer)
-	}
-	return fs.Freeze(active.container.Cgroups, active.container.Cgroups.Freezer)
+	return active.Resume()
 }
 
-func (d *driver) Terminate(p *execdriver.Command) error {
+func (d *driver) Terminate(c *execdriver.Command) error {
 	// lets check the start time for the process
-	state, err := libcontainer.GetState(filepath.Join(d.root, p.ID))
-	if err != nil {
-		if !os.IsNotExist(err) {
-			return err
-		}
-		// TODO: Remove this part for version 1.2.0
-		// This is added only to ensure smooth upgrades from pre 1.1.0 to 1.1.0
-		data, err := ioutil.ReadFile(filepath.Join(d.root, p.ID, "start"))
-		if err != nil {
-			// if we don't have the data on disk then we can assume the process is gone
-			// because this is only removed after we know the process has stopped
-			if os.IsNotExist(err) {
-				return nil
-			}
-			return err
-		}
-		state = &libcontainer.State{InitStartTime: string(data)}
+	active := d.activeContainers[c.ID]
+	if active == nil {
+		return fmt.Errorf("active container for %s does not exist", c.ID)
 	}
+	state, err := active.State()
+	if err != nil {
+		return err
+	}
+	pid := state.InitProcessPid
 
-	currentStartTime, err := system.GetProcessStartTime(p.ProcessConfig.Process.Pid)
+	currentStartTime, err := system.GetProcessStartTime(pid)
 	if err != nil {
 		return err
 	}
 
-	if state.InitStartTime == currentStartTime {
-		err = syscall.Kill(p.ProcessConfig.Process.Pid, 9)
-		syscall.Wait4(p.ProcessConfig.Process.Pid, nil, 0, nil)
+	if state.InitProcessStartTime == currentStartTime {
+		err = syscall.Kill(pid, 9)
+		syscall.Wait4(pid, nil, 0, nil)
 	}
-	d.cleanContainer(p.ID)
+	d.cleanContainer(c.ID)
 
 	return err
 
@@ -257,15 +287,10 @@ func (d *driver) GetPidsForContainer(id string) ([]int, error) {
 	if active == nil {
 		return nil, fmt.Errorf("active container for %s does not exist", id)
 	}
-	c := active.container.Cgroups
-
-	if systemd.UseSystemd() {
-		return systemd.GetPids(c)
-	}
-	return fs.GetPids(c)
+	return active.Processes()
 }
 
-func (d *driver) writeContainerFile(container *libcontainer.Config, id string) error {
+func (d *driver) writeContainerFile(container *configs.Config, id string) error {
 	data, err := json.Marshal(container)
 	if err != nil {
 		return err
@@ -289,42 +314,61 @@ func (d *driver) Clean(id string) error {
 }
 
 func (d *driver) Stats(id string) (*execdriver.ResourceStats, error) {
-	return execdriver.Stats(filepath.Join(d.root, id), d.activeContainers[id].container.Cgroups.Memory, d.machineMemory)
-}
-
-type TtyConsole struct {
-	MasterPty *os.File
-}
-
-func NewTtyConsole(processConfig *execdriver.ProcessConfig, pipes *execdriver.Pipes) (*TtyConsole, error) {
-	ptyMaster, console, err := consolepkg.CreateMasterAndConsole()
+	c := d.activeContainers[id]
+	now := time.Now()
+	stats, err := c.Stats()
 	if err != nil {
 		return nil, err
 	}
+	memoryLimit := c.Config().Cgroups.Memory
+	// if the container does not have any memory limit specified set the
+	// limit to the machines memory
+	if memoryLimit == 0 {
+		memoryLimit = d.machineMemory
+	}
+	return &execdriver.ResourceStats{
+		Stats:       stats,
+		Read:        now,
+		MemoryLimit: memoryLimit,
+	}, nil
+}
 
+func getEnv(key string, env []string) string {
+	for _, pair := range env {
+		parts := strings.Split(pair, "=")
+		if parts[0] == key {
+			return parts[1]
+		}
+	}
+	return ""
+}
+
+type TtyConsole struct {
+	console libcontainer.Console
+}
+
+func NewTtyConsole(console libcontainer.Console, pipes *execdriver.Pipes, rootuid int) (*TtyConsole, error) {
 	tty := &TtyConsole{
-		MasterPty: ptyMaster,
+		console: console,
 	}
 
-	if err := tty.AttachPipes(&processConfig.Cmd, pipes); err != nil {
+	if err := tty.AttachPipes(pipes); err != nil {
 		tty.Close()
 		return nil, err
 	}
 
-	processConfig.Console = console
-
 	return tty, nil
 }
 
-func (t *TtyConsole) Master() *os.File {
-	return t.MasterPty
+func (t *TtyConsole) Master() libcontainer.Console {
+	return t.console
 }
 
 func (t *TtyConsole) Resize(h, w int) error {
-	return term.SetWinsize(t.MasterPty.Fd(), &term.Winsize{Height: uint16(h), Width: uint16(w)})
+	return term.SetWinsize(t.console.Fd(), &term.Winsize{Height: uint16(h), Width: uint16(w)})
 }
 
-func (t *TtyConsole) AttachPipes(command *exec.Cmd, pipes *execdriver.Pipes) error {
+func (t *TtyConsole) AttachPipes(pipes *execdriver.Pipes) error {
 	go func() {
 		if wb, ok := pipes.Stdout.(interface {
 			CloseWriters() error
@@ -332,12 +376,12 @@ func (t *TtyConsole) AttachPipes(command *exec.Cmd, pipes *execdriver.Pipes) err
 			defer wb.CloseWriters()
 		}
 
-		io.Copy(pipes.Stdout, t.MasterPty)
+		io.Copy(pipes.Stdout, t.console)
 	}()
 
 	if pipes.Stdin != nil {
 		go func() {
-			io.Copy(t.MasterPty, pipes.Stdin)
+			io.Copy(t.console, pipes.Stdin)
 
 			pipes.Stdin.Close()
 		}()
@@ -347,5 +391,5 @@ func (t *TtyConsole) AttachPipes(command *exec.Cmd, pipes *execdriver.Pipes) err
 }
 
 func (t *TtyConsole) Close() error {
-	return t.MasterPty.Close()
+	return t.console.Close()
 }

daemon/execdriver/native/exec.go

@@ -4,67 +4,77 @@ package native
 
 import (
 	"fmt"
-	"log"
 	"os"
 	"os/exec"
-	"path/filepath"
-	"runtime"
+	"syscall"
 
 	"github.com/docker/docker/daemon/execdriver"
-	"github.com/docker/docker/pkg/reexec"
 	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/namespaces"
+	_ "github.com/docker/libcontainer/nsenter"
+	"github.com/docker/libcontainer/utils"
 )
 
-const execCommandName = "nsenter-exec"
-
-func init() {
-	reexec.Register(execCommandName, nsenterExec)
-}
-
-func nsenterExec() {
-	runtime.LockOSThread()
-
-	// User args are passed after '--' in the command line.
-	userArgs := findUserArgs()
-
-	config, err := loadConfigFromFd()
-	if err != nil {
-		log.Fatalf("docker-exec: unable to receive config from sync pipe: %s", err)
-	}
-
-	if err := namespaces.FinalizeSetns(config, userArgs); err != nil {
-		log.Fatalf("docker-exec: failed to exec: %s", err)
-	}
-}
-
 // TODO(vishh): Add support for running in priviledged mode and running as a different user.
 func (d *driver) Exec(c *execdriver.Command, processConfig *execdriver.ProcessConfig, pipes *execdriver.Pipes, startCallback execdriver.StartCallback) (int, error) {
 	active := d.activeContainers[c.ID]
 	if active == nil {
 		return -1, fmt.Errorf("No active container exists with ID %s", c.ID)
 	}
-	state, err := libcontainer.GetState(filepath.Join(d.root, c.ID))
-	if err != nil {
-		return -1, fmt.Errorf("State unavailable for container with ID %s. The container may have been cleaned up already. Error: %s", c.ID, err)
-	}
 
 	var term execdriver.Terminal
+	var err error
+
+	p := &libcontainer.Process{
+		Args: append([]string{processConfig.Entrypoint}, processConfig.Arguments...),
+		Env:  c.ProcessConfig.Env,
+		Cwd:  c.WorkingDir,
+		User: c.ProcessConfig.User,
+	}
 
 	if processConfig.Tty {
-		term, err = NewTtyConsole(processConfig, pipes)
+		config := active.Config()
+		rootuid, err := config.HostUID()
+		if err != nil {
+			return -1, err
+		}
+		cons, err := p.NewConsole(rootuid)
+		if err != nil {
+			return -1, err
+		}
+		term, err = NewTtyConsole(cons, pipes, rootuid)
 	} else {
-		term, err = execdriver.NewStdConsole(processConfig, pipes)
+		p.Stdout = pipes.Stdout
+		p.Stderr = pipes.Stderr
+		p.Stdin = pipes.Stdin
+		term = &execdriver.StdConsole{}
+	}
+	if err != nil {
+		return -1, err
 	}
 
 	processConfig.Terminal = term
 
-	args := append([]string{processConfig.Entrypoint}, processConfig.Arguments...)
+	if err := active.Start(p); err != nil {
+		return -1, err
+	}
 
-	return namespaces.ExecIn(active.container, state, args, os.Args[0], "exec", processConfig.Stdin, processConfig.Stdout, processConfig.Stderr, processConfig.Console,
-		func(cmd *exec.Cmd) {
-			if startCallback != nil {
-				startCallback(&c.ProcessConfig, cmd.Process.Pid)
-			}
-		})
+	if startCallback != nil {
+		pid, err := p.Pid()
+		if err != nil {
+			p.Signal(os.Kill)
+			p.Wait()
+			return -1, err
+		}
+		startCallback(&c.ProcessConfig, pid)
+	}
+
+	ps, err := p.Wait()
+	if err != nil {
+		exitErr, ok := err.(*exec.ExitError)
+		if !ok {
+			return -1, err
+		}
+		ps = exitErr.ProcessState
+	}
+	return utils.ExitStatus(ps.Sys().(syscall.WaitStatus)), nil
 }

daemon/execdriver/native/info.go

@@ -2,13 +2,6 @@
 
 package native
 
-import (
-	"os"
-	"path/filepath"
-
-	"github.com/docker/libcontainer"
-)
-
 type info struct {
 	ID     string
 	driver *driver
@@ -18,13 +11,6 @@ type info struct {
 // pid file for a container.  If the file exists then the
 // container is currently running
 func (i *info) IsRunning() bool {
-	if _, err := libcontainer.GetState(filepath.Join(i.driver.root, i.ID)); err == nil {
-		return true
-	}
-	// TODO: Remove this part for version 1.2.0
-	// This is added only to ensure smooth upgrades from pre 1.1.0 to 1.1.0
-	if _, err := os.Stat(filepath.Join(i.driver.root, i.ID, "pid")); err == nil {
-		return true
-	}
-	return false
+	_, ok := i.driver.activeContainers[i.ID]
+	return ok
 }

daemon/execdriver/native/init.go

@@ -3,55 +3,40 @@
 package native
 
 import (
-	"encoding/json"
-	"flag"
 	"fmt"
 	"os"
-	"path/filepath"
 	"runtime"
 
 	"github.com/docker/docker/pkg/reexec"
 	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/namespaces"
 )
 
 func init() {
 	reexec.Register(DriverName, initializer)
 }
 
+func fatal(err error) {
+	if lerr, ok := err.(libcontainer.Error); ok {
+		lerr.Detail(os.Stderr)
+		os.Exit(1)
+	}
+
+	fmt.Fprintln(os.Stderr, err)
+	os.Exit(1)
+}
+
 func initializer() {
+	runtime.GOMAXPROCS(1)
 	runtime.LockOSThread()
-
-	var (
-		pipe    = flag.Int("pipe", 0, "sync pipe fd")
-		console = flag.String("console", "", "console (pty slave) path")
-		root    = flag.String("root", ".", "root path for configuration files")
-	)
-
-	flag.Parse()
-
-	var container *libcontainer.Config
-	f, err := os.Open(filepath.Join(*root, "container.json"))
+	factory, err := libcontainer.New("")
 	if err != nil {
-		writeError(err)
+		fatal(err)
+	}
+	if err := factory.StartInitialization(3); err != nil {
+		fatal(err)
 	}
 
-	if err := json.NewDecoder(f).Decode(&container); err != nil {
-		f.Close()
-		writeError(err)
-	}
-	f.Close()
-
-	rootfs, err := os.Getwd()
-	if err != nil {
-		writeError(err)
-	}
-
-	if err := namespaces.Init(container, rootfs, *console, os.NewFile(uintptr(*pipe), "child"), flag.Args()); err != nil {
-		writeError(err)
-	}
-
-	panic("Unreachable")
+	panic("unreachable")
 }
 
 func writeError(err error) {

daemon/execdriver/native/template/default_template.go

@@ -1,14 +1,17 @@
 package template
 
 import (
-	"github.com/docker/libcontainer"
+	"syscall"
+
 	"github.com/docker/libcontainer/apparmor"
-	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
+const defaultMountFlags = syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
+
 // New returns the docker default configuration for libcontainer
-func New() *libcontainer.Config {
-	container := &libcontainer.Config{
+func New() *configs.Config {
+	container := &configs.Config{
 		Capabilities: []string{
 			"CHOWN",
 			"DAC_OVERRIDE",
@@ -25,18 +28,51 @@ func New() *libcontainer.Config {
 			"KILL",
 			"AUDIT_WRITE",
 		},
-		Namespaces: libcontainer.Namespaces([]libcontainer.Namespace{
+		Namespaces: configs.Namespaces([]configs.Namespace{
 			{Type: "NEWNS"},
 			{Type: "NEWUTS"},
 			{Type: "NEWIPC"},
 			{Type: "NEWPID"},
 			{Type: "NEWNET"},
 		}),
-		Cgroups: &cgroups.Cgroup{
+		Cgroups: &configs.Cgroup{
 			Parent:          "docker",
 			AllowAllDevices: false,
 		},
-		MountConfig: &libcontainer.MountConfig{},
+		Mounts: []*configs.Mount{
+			{
+				Device:      "tmpfs",
+				Source:      "shm",
+				Destination: "/dev/shm",
+				Data:        "mode=1777,size=65536k",
+				Flags:       defaultMountFlags,
+			},
+			{
+				Source:      "mqueue",
+				Destination: "/dev/mqueue",
+				Device:      "mqueue",
+				Flags:       defaultMountFlags,
+			},
+			{
+				Source:      "sysfs",
+				Destination: "/sys",
+				Device:      "sysfs",
+				Flags:       defaultMountFlags | syscall.MS_RDONLY,
+			},
+		},
+		MaskPaths: []string{
+			"/proc/kcore",
+		},
+		ReadonlyPaths: []string{
+			"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
+		},
+		Rlimits: []configs.Rlimit{
+			{
+				Type: syscall.RLIMIT_NOFILE,
+				Hard: 1024,
+				Soft: 1024,
+			},
+		},
 	}
 
 	if apparmor.IsEnabled() {

daemon/execdriver/native/utils.go

@@ -2,28 +2,21 @@
 
 package native
 
-import (
-	"encoding/json"
-	"os"
+//func findUserArgs() []string {
+//for i, a := range os.Args {
+//if a == "--" {
+//return os.Args[i+1:]
+//}
+//}
+//return []string{}
+//}
 
-	"github.com/docker/libcontainer"
-)
-
-func findUserArgs() []string {
-	for i, a := range os.Args {
-		if a == "--" {
-			return os.Args[i+1:]
-		}
-	}
-	return []string{}
-}
-
-// loadConfigFromFd loads a container's config from the sync pipe that is provided by
-// fd 3 when running a process
-func loadConfigFromFd() (*libcontainer.Config, error) {
-	var config *libcontainer.Config
-	if err := json.NewDecoder(os.NewFile(3, "child")).Decode(&config); err != nil {
-		return nil, err
-	}
-	return config, nil
-}
+//// loadConfigFromFd loads a container's config from the sync pipe that is provided by
+//// fd 3 when running a process
+//func loadConfigFromFd() (*configs.Config, error) {
+//var config *libcontainer.Config
+//if err := json.NewDecoder(os.NewFile(3, "child")).Decode(&config); err != nil {
+//return nil, err
+//}
+//return config, nil
+//}

daemon/execdriver/utils.go

@@ -5,13 +5,83 @@ import (
 	"strings"
 
 	"github.com/docker/docker/utils"
-	"github.com/docker/libcontainer/security/capabilities"
+	"github.com/syndtr/gocapability/capability"
 )
 
+var capabilityList = Capabilities{
+	{Key: "SETPCAP", Value: capability.CAP_SETPCAP},
+	{Key: "SYS_MODULE", Value: capability.CAP_SYS_MODULE},
+	{Key: "SYS_RAWIO", Value: capability.CAP_SYS_RAWIO},
+	{Key: "SYS_PACCT", Value: capability.CAP_SYS_PACCT},
+	{Key: "SYS_ADMIN", Value: capability.CAP_SYS_ADMIN},
+	{Key: "SYS_NICE", Value: capability.CAP_SYS_NICE},
+	{Key: "SYS_RESOURCE", Value: capability.CAP_SYS_RESOURCE},
+	{Key: "SYS_TIME", Value: capability.CAP_SYS_TIME},
+	{Key: "SYS_TTY_CONFIG", Value: capability.CAP_SYS_TTY_CONFIG},
+	{Key: "MKNOD", Value: capability.CAP_MKNOD},
+	{Key: "AUDIT_WRITE", Value: capability.CAP_AUDIT_WRITE},
+	{Key: "AUDIT_CONTROL", Value: capability.CAP_AUDIT_CONTROL},
+	{Key: "MAC_OVERRIDE", Value: capability.CAP_MAC_OVERRIDE},
+	{Key: "MAC_ADMIN", Value: capability.CAP_MAC_ADMIN},
+	{Key: "NET_ADMIN", Value: capability.CAP_NET_ADMIN},
+	{Key: "SYSLOG", Value: capability.CAP_SYSLOG},
+	{Key: "CHOWN", Value: capability.CAP_CHOWN},
+	{Key: "NET_RAW", Value: capability.CAP_NET_RAW},
+	{Key: "DAC_OVERRIDE", Value: capability.CAP_DAC_OVERRIDE},
+	{Key: "FOWNER", Value: capability.CAP_FOWNER},
+	{Key: "DAC_READ_SEARCH", Value: capability.CAP_DAC_READ_SEARCH},
+	{Key: "FSETID", Value: capability.CAP_FSETID},
+	{Key: "KILL", Value: capability.CAP_KILL},
+	{Key: "SETGID", Value: capability.CAP_SETGID},
+	{Key: "SETUID", Value: capability.CAP_SETUID},
+	{Key: "LINUX_IMMUTABLE", Value: capability.CAP_LINUX_IMMUTABLE},
+	{Key: "NET_BIND_SERVICE", Value: capability.CAP_NET_BIND_SERVICE},
+	{Key: "NET_BROADCAST", Value: capability.CAP_NET_BROADCAST},
+	{Key: "IPC_LOCK", Value: capability.CAP_IPC_LOCK},
+	{Key: "IPC_OWNER", Value: capability.CAP_IPC_OWNER},
+	{Key: "SYS_CHROOT", Value: capability.CAP_SYS_CHROOT},
+	{Key: "SYS_PTRACE", Value: capability.CAP_SYS_PTRACE},
+	{Key: "SYS_BOOT", Value: capability.CAP_SYS_BOOT},
+	{Key: "LEASE", Value: capability.CAP_LEASE},
+	{Key: "SETFCAP", Value: capability.CAP_SETFCAP},
+	{Key: "WAKE_ALARM", Value: capability.CAP_WAKE_ALARM},
+	{Key: "BLOCK_SUSPEND", Value: capability.CAP_BLOCK_SUSPEND},
+}
+
+type (
+	CapabilityMapping struct {
+		Key   string         `json:"key,omitempty"`
+		Value capability.Cap `json:"value,omitempty"`
+	}
+	Capabilities []*CapabilityMapping
+)
+
+func (c *CapabilityMapping) String() string {
+	return c.Key
+}
+
+func GetCapability(key string) *CapabilityMapping {
+	for _, capp := range capabilityList {
+		if capp.Key == key {
+			cpy := *capp
+			return &cpy
+		}
+	}
+	return nil
+}
+
+func GetAllCapabilities() []string {
+	output := make([]string, len(capabilityList))
+	for i, capability := range capabilityList {
+		output[i] = capability.String()
+	}
+	return output
+}
+
 func TweakCapabilities(basics, adds, drops []string) ([]string, error) {
 	var (
 		newCaps []string
-		allCaps = capabilities.GetAllCapabilities()
+		allCaps = GetAllCapabilities()
 	)
 
 	// look for invalid cap in the drop list
@@ -26,7 +96,7 @@ func TweakCapabilities(basics, adds, drops []string) ([]string, error) {
 
 	// handle --cap-add=all
 	if utils.StringsContainsNoCase(adds, "all") {
-		basics = capabilities.GetAllCapabilities()
+		basics = allCaps
 	}
 
 	if !utils.StringsContainsNoCase(drops, "all") {

daemon/stats.go

@@ -18,7 +18,7 @@ func (daemon *Daemon) ContainerStats(job *engine.Job) engine.Status {
 	enc := json.NewEncoder(job.Stdout)
 	for v := range updates {
 		update := v.(*execdriver.ResourceStats)
-		ss := convertToAPITypes(update.ContainerStats)
+		ss := convertToAPITypes(update.Stats)
 		ss.MemoryStats.Limit = uint64(update.MemoryLimit)
 		ss.Read = update.Read
 		ss.CpuStats.SystemUsage = update.SystemUsage
@@ -31,20 +31,21 @@ func (daemon *Daemon) ContainerStats(job *engine.Job) engine.Status {
 	return engine.StatusOK
 }
 
-// convertToAPITypes converts the libcontainer.ContainerStats to the api specific
+// convertToAPITypes converts the libcontainer.Stats to the api specific
 // structs.  This is done to preserve API compatibility and versioning.
-func convertToAPITypes(ls *libcontainer.ContainerStats) *types.Stats {
+func convertToAPITypes(ls *libcontainer.Stats) *types.Stats {
 	s := &types.Stats{}
-	if ls.NetworkStats != nil {
-		s.Network = types.Network{
-			RxBytes:   ls.NetworkStats.RxBytes,
-			RxPackets: ls.NetworkStats.RxPackets,
-			RxErrors:  ls.NetworkStats.RxErrors,
-			RxDropped: ls.NetworkStats.RxDropped,
-			TxBytes:   ls.NetworkStats.TxBytes,
-			TxPackets: ls.NetworkStats.TxPackets,
-			TxErrors:  ls.NetworkStats.TxErrors,
-			TxDropped: ls.NetworkStats.TxDropped,
+	if ls.Interfaces != nil {
+		s.Network = types.Network{}
+		for _, iface := range ls.Interfaces {
+			s.Network.RxBytes += iface.RxBytes
+			s.Network.RxPackets += iface.RxPackets
+			s.Network.RxErrors += iface.RxErrors
+			s.Network.RxDropped += iface.RxDropped
+			s.Network.TxBytes += iface.TxBytes
+			s.Network.TxPackets += iface.TxPackets
+			s.Network.TxErrors += iface.TxErrors
+			s.Network.TxDropped += iface.TxDropped
 		}
 	}
 	cs := ls.CgroupStats

integration-cli/docker_cli_exec_test.go

@@ -60,7 +60,7 @@ func TestExecInteractiveStdinClose(t *testing.T) {
 
 		out, err := cmd.CombinedOutput()
 		if err != nil {
-			t.Fatal(err, out)
+			t.Fatal(err, string(out))
 		}
 
 		if string(out) == "" {
@@ -538,7 +538,6 @@ func TestRunExecDir(t *testing.T) {
 	id := strings.TrimSpace(out)
 	execDir := filepath.Join(execDriverPath, id)
 	stateFile := filepath.Join(execDir, "state.json")
-	contFile := filepath.Join(execDir, "container.json")
 
 	{
 		fi, err := os.Stat(execDir)
@@ -552,10 +551,6 @@ func TestRunExecDir(t *testing.T) {
 		if err != nil {
 			t.Fatal(err)
 		}
-		fi, err = os.Stat(contFile)
-		if err != nil {
-			t.Fatal(err)
-		}
 	}
 
 	stopCmd := exec.Command(dockerBinary, "stop", id)
@@ -564,23 +559,12 @@ func TestRunExecDir(t *testing.T) {
 		t.Fatal(err, out)
 	}
 	{
-		fi, err := os.Stat(execDir)
-		if err != nil {
+		_, err := os.Stat(execDir)
+		if err == nil {
 			t.Fatal(err)
 		}
-		if !fi.IsDir() {
-			t.Fatalf("%q must be a directory", execDir)
-		}
-		fi, err = os.Stat(stateFile)
 		if err == nil {
-			t.Fatalf("Statefile %q is exists for stopped container!", stateFile)
-		}
-		if !os.IsNotExist(err) {
-			t.Fatalf("Error should be about non-existing, got %s", err)
-		}
-		fi, err = os.Stat(contFile)
-		if err == nil {
-			t.Fatalf("Container file %q is exists for stopped container!", contFile)
+			t.Fatalf("Exec directory %q exists for removed container!", execDir)
 		}
 		if !os.IsNotExist(err) {
 			t.Fatalf("Error should be about non-existing, got %s", err)
@@ -603,10 +587,6 @@ func TestRunExecDir(t *testing.T) {
 		if err != nil {
 			t.Fatal(err)
 		}
-		fi, err = os.Stat(contFile)
-		if err != nil {
-			t.Fatal(err)
-		}
 	}
 	rmCmd := exec.Command(dockerBinary, "rm", "-f", id)
 	out, _, err = runCommandWithOutput(rmCmd)

project/vendor.sh

@@ -68,8 +68,7 @@ if [ "$1" = '--go' ]; then
 	mv tmp-tar src/code.google.com/p/go/src/pkg/archive/tar
 fi
 
-# this commit is from docker_1.5 branch in libcontainer, pls delete that branch when you'll update libcontainer again
-clone git github.com/docker/libcontainer 2d3b5af7486f1a4e80a5ed91859d309b4eebf80c
+clone git github.com/docker/libcontainer dd3cb8822352fd4acc0b8b426bd86e47e98f6853
 # see src/github.com/docker/libcontainer/update-vendor.sh which is the "source of truth" for libcontainer deps (just like this file)
 rm -rf src/github.com/docker/libcontainer/vendor
 eval "$(grep '^clone ' src/github.com/docker/libcontainer/update-vendor.sh | grep -v 'github.com/codegangsta/cli')"

vendor/src/github.com/docker/libcontainer/.drone.yml

@@ -1,9 +0,0 @@
-image: dockercore/libcontainer
-script:
-# Setup the DockerInDocker environment.
-  - /dind
-  - sed -i 's!docker/docker!docker/libcontainer!' /go/src/github.com/docker/docker/hack/make/.validate
-  - bash /go/src/github.com/docker/docker/hack/make/validate-dco
-  - bash /go/src/github.com/docker/docker/hack/make/validate-gofmt
-  - export GOPATH="$GOPATH:/go:$(pwd)/vendor" # Drone mucks with our GOPATH
-  - make direct-test

vendor/src/github.com/docker/libcontainer/.gitignore

@@ -0,0 +1 @@
+nsinit/nsinit

vendor/src/github.com/docker/libcontainer/MAINTAINERS

@@ -3,4 +3,5 @@ Rohit Jnagal <jnagal@google.com> (@rjnagal)
 Victor Marmol <vmarmol@google.com> (@vmarmol)
 Mrunal Patel <mpatel@redhat.com> (@mrunalp)
 Alexandr Morozov <lk4d4@docker.com> (@LK4D4)
+Daniel, Dao Quang Minh <dqminh89@gmail.com> (@dqminh)
 update-vendor.sh: Tianon Gravi <admwiggin@gmail.com> (@tianon)

vendor/src/github.com/docker/libcontainer/Makefile

@@ -22,3 +22,10 @@ direct-build:
 
 direct-install:
 	go install -v $(GO_PACKAGES)
+
+local:
+	go test -v
+
+validate:
+	hack/validate.sh
+

vendor/src/github.com/docker/libcontainer/NOTICE

@@ -1,5 +1,5 @@
 libcontainer
-Copyright 2012-2014 Docker, Inc.
+Copyright 2012-2015 Docker, Inc.
 
 This product includes software developed at Docker, Inc. (http://www.docker.com).
 

vendor/src/github.com/docker/libcontainer/PRINCIPLES.md

@@ -8,7 +8,7 @@ In the design and development of libcontainer we try to follow these principles:
 * Less code is better.
 * Fewer components are better. Do you really need to add one more class?
 * 50 lines of straightforward, readable code is better than 10 lines of magic that nobody can understand.
-* Don't do later what you can do now. "//FIXME: refactor" is not acceptable in new code.
+* Don't do later what you can do now. "//TODO: refactor" is not acceptable in new code.
 * When hesitating between two options, choose the one that is easier to reverse.
 * "No" is temporary; "Yes" is forever. If you're not sure about a new feature, say no. You can change your mind later.
 * Containers must be portable to the greatest possible number of machines. Be suspicious of any change which makes machines less interchangeable.

vendor/src/github.com/docker/libcontainer/README.md

@@ -1,48 +1,169 @@
-## libcontainer - reference implementation for containers [![Build Status](https://ci.dockerproject.com/github.com/docker/libcontainer/status.svg?branch=master)](https://ci.dockerproject.com/github.com/docker/libcontainer) 
+## libcontainer - reference implementation for containers [![Build Status](https://jenkins.dockerproject.com/buildStatus/icon?job=Libcontainer Master)](https://jenkins.dockerproject.com/job/Libcontainer%20Master/) 
 
-### Note on API changes:
-
-Please bear with us while we work on making the libcontainer API stable and something that we can support long term.  We are currently discussing the API with the community, therefore, if you currently depend on libcontainer please pin your dependency at a specific tag or commit id.  Please join the discussion and help shape the API.
-
-#### Background
-
-libcontainer specifies configuration options for what a container is.  It provides a native Go implementation for using Linux namespaces with no external dependencies.  libcontainer provides many convenience functions for working with namespaces, networking, and management.  
+Libcontainer provides a native Go implementation for creating containers 
+with namespaces, cgroups, capabilities, and filesystem access controls.
+It allows you to manage the lifecycle of the container performing additional operations
+after the container is created.
 
 
 #### Container
-A container is a self contained execution environment that shares the kernel of the host system and which is (optionally) isolated from other containers in the system.
+A container is a self contained execution environment that shares the kernel of the 
+host system and which is (optionally) isolated from other containers in the system.
 
-libcontainer may be used to execute a process in a container. If a user tries to run a new process inside an existing container, the new process is added to the processes executing in the container.
+#### Using libcontainer
+
+To create a container you first have to initialize an instance of a factory
+that will handle the creation and initialization for a container.
+
+Because containers are spawned in a two step process you will need to provide
+arguments to a binary that will be executed as the init process for the container.
+To use the current binary that is spawning the containers and acting as the parent
+you can use `os.Args[0]` and we have a command called `init` setup.
+
+```go
+root, err := libcontainer.New("/var/lib/container", libcontainer.InitArgs(os.Args[0], "init"))
+if err != nil {
+    log.Fatal(err)
+}
+```
+
+Once you have an instance of the factory created we can create a configuration 
+struct describing how the container is to be created.  A sample would look similar to this:
+
+```go
+config := &configs.Config{
+    Rootfs: rootfs,
+    Capabilities: []string{
+        "CHOWN",
+        "DAC_OVERRIDE",
+        "FSETID",
+        "FOWNER",
+        "MKNOD",
+        "NET_RAW",
+        "SETGID",
+        "SETUID",
+        "SETFCAP",
+        "SETPCAP",
+        "NET_BIND_SERVICE",
+        "SYS_CHROOT",
+        "KILL",
+        "AUDIT_WRITE",
+    },
+    Namespaces: configs.Namespaces([]configs.Namespace{
+        {Type: configs.NEWNS},
+        {Type: configs.NEWUTS},
+        {Type: configs.NEWIPC},
+        {Type: configs.NEWPID},
+        {Type: configs.NEWNET},
+    }),
+    Cgroups: &configs.Cgroup{
+        Name:            "test-container",
+        Parent:          "system",
+        AllowAllDevices: false,
+        AllowedDevices:  configs.DefaultAllowedDevices,
+    },
+
+    Devices:  configs.DefaultAutoCreatedDevices,
+    Hostname: "testing",
+    Networks: []*configs.Network{
+        {
+            Type:    "loopback",
+            Address: "127.0.0.1/0",
+            Gateway: "localhost",
+        },
+    },
+    Rlimits: []configs.Rlimit{
+        {
+            Type: syscall.RLIMIT_NOFILE,
+            Hard: uint64(1024),
+            Soft: uint64(1024),
+        },
+    },
+}
+```
+
+Once you have the configuration populated you can create a container:
+
+```go
+container, err := root.Create("container-id", config)
+```
+
+To spawn bash as the initial process inside the container and have the
+processes pid returned in order to wait, signal, or kill the process:
+
+```go
+process := &libcontainer.Process{
+    Args:   []string{"/bin/bash"},
+    Env:    []string{"PATH=/bin"},
+    User:   "daemon",
+    Stdin:  os.Stdin,
+    Stdout: os.Stdout,
+    Stderr: os.Stderr,
+}
+
+err := container.Start(process)
+if err != nil {
+    log.Fatal(err)
+}
+
+// wait for the process to finish.
+status, err := process.Wait()
+if err != nil {
+    log.Fatal(err)
+}
+
+// destroy the container.
+container.Destroy()
+```
+
+Additional ways to interact with a running container are:
+
+```go
+// return all the pids for all processes running inside the container.
+processes, err := container.Processes() 
+
+// get detailed cpu, memory, io, and network statistics for the container and 
+// it's processes.
+stats, err := container.Stats()
 
 
-#### Root file system
+// pause all processes inside the container.
+container.Pause()
 
-A container runs with a directory known as its *root file system*, or *rootfs*, mounted as the file system root. The rootfs is usually a full system tree.
-
-
-#### Configuration
-
-A container is initially configured by supplying configuration data when the container is created.
+// resume all paused processes.
+container.Resume()
+```
 
 
 #### nsinit
 
-`nsinit` is a cli application which demonstrates the use of libcontainer.  It is able to spawn new containers or join existing containers, based on the current directory.
+`nsinit` is a cli application which demonstrates the use of libcontainer.  
+It is able to spawn new containers or join existing containers.  A root
+filesystem must be provided for use along with a container configuration file.
 
-To use `nsinit`, cd into a Linux rootfs and copy a `container.json` file into the directory with your specified configuration. Environment, networking, and different capabilities for the container are specified in this file. The configuration is used for each process executed inside the container.
+To use `nsinit`, cd into a Linux rootfs and copy a `container.json` file into 
+the directory with your specified configuration. Environment, networking, 
+and different capabilities for the container are specified in this file. 
+The configuration is used for each process executed inside the container.
                                                                                                                                
 See the `sample_configs` folder for examples of what the container configuration should look like.
 
 To execute `/bin/bash` in the current directory as a container just run the following **as root**:
 ```bash
-nsinit exec /bin/bash
+nsinit exec --tty /bin/bash
 ```
 
-If you wish to spawn another process inside the container while your current bash session is running, run the same command again to get another bash shell (or change the command).  If the original process (PID 1) dies, all other processes spawned inside the container will be killed and the namespace will be removed. 
+If you wish to spawn another process inside the container while your 
+current bash session is running, run the same command again to 
+get another bash shell (or change the command).  If the original 
+process (PID 1) dies, all other processes spawned inside the container 
+will be killed and the namespace will be removed. 
 
-You can identify if a process is running in a container by looking to see if `state.json` is in the root of the directory.
+You can identify if a process is running in a container by 
+looking to see if `state.json` is in the root of the directory.
    
-You may also specify an alternate root place where the `container.json` file is read and where the `state.json` file will be saved.
+You may also specify an alternate root place where 
+the `container.json` file is read and where the `state.json` file will be saved.
 
 #### Future
 See the [roadmap](ROADMAP.md).

vendor/src/github.com/docker/libcontainer/api_temp.go

@@ -1,21 +0,0 @@
-/*
-Temporary API endpoint for libcontainer while the full API is finalized (api.go).
-*/
-package libcontainer
-
-import (
-	"github.com/docker/libcontainer/cgroups/fs"
-	"github.com/docker/libcontainer/network"
-)
-
-// TODO(vmarmol): Complete Stats() in final libcontainer API and move users to that.
-// DEPRECATED: The below portions are only to be used during the transition to the official API.
-// Returns all available stats for the given container.
-func GetStats(container *Config, state *State) (stats *ContainerStats, err error) {
-	stats = &ContainerStats{}
-	if stats.CgroupStats, err = fs.GetStats(state.CgroupPaths); err != nil {
-		return stats, err
-	}
-	stats.NetworkStats, err = network.GetStats(&state.NetworkState)
-	return stats, err
-}

vendor/src/github.com/docker/libcontainer/apparmor/apparmor.go

@@ -24,7 +24,6 @@ func ApplyProfile(name string) error {
 	if name == "" {
 		return nil
 	}
-
 	cName := C.CString(name)
 	defer C.free(unsafe.Pointer(cName))
 

vendor/src/github.com/docker/libcontainer/capabilities_linux.go

@@ -0,0 +1,91 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/syndtr/gocapability/capability"
+)
+
+const allCapabilityTypes = capability.CAPS | capability.BOUNDS
+
+var capabilityList = map[string]capability.Cap{
+	"SETPCAP":          capability.CAP_SETPCAP,
+	"SYS_MODULE":       capability.CAP_SYS_MODULE,
+	"SYS_RAWIO":        capability.CAP_SYS_RAWIO,
+	"SYS_PACCT":        capability.CAP_SYS_PACCT,
+	"SYS_ADMIN":        capability.CAP_SYS_ADMIN,
+	"SYS_NICE":         capability.CAP_SYS_NICE,
+	"SYS_RESOURCE":     capability.CAP_SYS_RESOURCE,
+	"SYS_TIME":         capability.CAP_SYS_TIME,
+	"SYS_TTY_CONFIG":   capability.CAP_SYS_TTY_CONFIG,
+	"MKNOD":            capability.CAP_MKNOD,
+	"AUDIT_WRITE":      capability.CAP_AUDIT_WRITE,
+	"AUDIT_CONTROL":    capability.CAP_AUDIT_CONTROL,
+	"MAC_OVERRIDE":     capability.CAP_MAC_OVERRIDE,
+	"MAC_ADMIN":        capability.CAP_MAC_ADMIN,
+	"NET_ADMIN":        capability.CAP_NET_ADMIN,
+	"SYSLOG":           capability.CAP_SYSLOG,
+	"CHOWN":            capability.CAP_CHOWN,
+	"NET_RAW":          capability.CAP_NET_RAW,
+	"DAC_OVERRIDE":     capability.CAP_DAC_OVERRIDE,
+	"FOWNER":           capability.CAP_FOWNER,
+	"DAC_READ_SEARCH":  capability.CAP_DAC_READ_SEARCH,
+	"FSETID":           capability.CAP_FSETID,
+	"KILL":             capability.CAP_KILL,
+	"SETGID":           capability.CAP_SETGID,
+	"SETUID":           capability.CAP_SETUID,
+	"LINUX_IMMUTABLE":  capability.CAP_LINUX_IMMUTABLE,
+	"NET_BIND_SERVICE": capability.CAP_NET_BIND_SERVICE,
+	"NET_BROADCAST":    capability.CAP_NET_BROADCAST,
+	"IPC_LOCK":         capability.CAP_IPC_LOCK,
+	"IPC_OWNER":        capability.CAP_IPC_OWNER,
+	"SYS_CHROOT":       capability.CAP_SYS_CHROOT,
+	"SYS_PTRACE":       capability.CAP_SYS_PTRACE,
+	"SYS_BOOT":         capability.CAP_SYS_BOOT,
+	"LEASE":            capability.CAP_LEASE,
+	"SETFCAP":          capability.CAP_SETFCAP,
+	"WAKE_ALARM":       capability.CAP_WAKE_ALARM,
+	"BLOCK_SUSPEND":    capability.CAP_BLOCK_SUSPEND,
+	"AUDIT_READ":       capability.CAP_AUDIT_READ,
+}
+
+func newCapWhitelist(caps []string) (*whitelist, error) {
+	l := []capability.Cap{}
+	for _, c := range caps {
+		v, ok := capabilityList[c]
+		if !ok {
+			return nil, fmt.Errorf("unknown capability %q", c)
+		}
+		l = append(l, v)
+	}
+	pid, err := capability.NewPid(os.Getpid())
+	if err != nil {
+		return nil, err
+	}
+	return &whitelist{
+		keep: l,
+		pid:  pid,
+	}, nil
+}
+
+type whitelist struct {
+	pid  capability.Capabilities
+	keep []capability.Cap
+}
+
+// dropBoundingSet drops the capability bounding set to those specified in the whitelist.
+func (w *whitelist) dropBoundingSet() error {
+	w.pid.Clear(capability.BOUNDS)
+	w.pid.Set(capability.BOUNDS, w.keep...)
+	return w.pid.Apply(capability.BOUNDS)
+}
+
+// drop drops all capabilities for the current process except those specified in the whitelist.
+func (w *whitelist) drop() error {
+	w.pid.Clear(allCapabilityTypes)
+	w.pid.Set(allCapabilityTypes, w.keep...)
+	return w.pid.Apply(allCapabilityTypes)
+}

vendor/src/github.com/docker/libcontainer/cgroups/cgroups.go

@@ -3,16 +3,38 @@ package cgroups
 import (
 	"fmt"
 
-	"github.com/docker/libcontainer/devices"
+	"github.com/docker/libcontainer/configs"
 )
 
-type FreezerState string
+type Manager interface {
+	// Apply cgroup configuration to the process with the specified pid
+	Apply(pid int) error
 
-const (
-	Undefined FreezerState = ""
-	Frozen    FreezerState = "FROZEN"
-	Thawed    FreezerState = "THAWED"
-)
+	// Returns the PIDs inside the cgroup set
+	GetPids() ([]int, error)
+
+	// Returns statistics for the cgroup set
+	GetStats() (*Stats, error)
+
+	// Toggles the freezer cgroup according with specified state
+	Freeze(state configs.FreezerState) error
+
+	// Destroys the cgroup set
+	Destroy() error
+
+	// NewCgroupManager() and LoadCgroupManager() require following attributes:
+	// 	Paths   map[string]string
+	// 	Cgroups *cgroups.Cgroup
+	// Paths maps cgroup subsystem to path at which it is mounted.
+	// Cgroups specifies specific cgroup settings for the various subsystems
+
+	// Returns cgroup paths to save in a state file and to be able to
+	// restore the object later.
+	GetPaths() map[string]string
+
+	// Set the cgroup as configured.
+	Set(container *configs.Config) error
+}
 
 type NotFoundError struct {
 	Subsystem string
@@ -32,25 +54,6 @@ func IsNotFound(err error) bool {
 	if err == nil {
 		return false
 	}
-
 	_, ok := err.(*NotFoundError)
 	return ok
 }
-
-type Cgroup struct {
-	Name   string `json:"name,omitempty"`
-	Parent string `json:"parent,omitempty"` // name of parent cgroup or slice
-
-	AllowAllDevices   bool              `json:"allow_all_devices,omitempty"` // If this is true allow access to any kind of device within the container.  If false, allow access only to devices explicitly listed in the allowed_devices list.
-	AllowedDevices    []*devices.Device `json:"allowed_devices,omitempty"`
-	Memory            int64             `json:"memory,omitempty"`             // Memory limit (in bytes)
-	MemoryReservation int64             `json:"memory_reservation,omitempty"` // Memory reservation or soft_limit (in bytes)
-	MemorySwap        int64             `json:"memory_swap,omitempty"`        // Total memory usage (memory + swap); set `-1' to disable swap
-	CpuShares         int64             `json:"cpu_shares,omitempty"`         // CPU shares (relative weight vs. other containers)
-	CpuQuota          int64             `json:"cpu_quota,omitempty"`          // CPU hardcap limit (in usecs). Allowed cpu time in a given period.
-	CpuPeriod         int64             `json:"cpu_period,omitempty"`         // CPU period to be used for hardcapping (in usecs). 0 to use system default.
-	CpusetCpus        string            `json:"cpuset_cpus,omitempty"`        // CPU to use
-	CpusetMems        string            `json:"cpuset_mems,omitempty"`        // MEM to use
-	Freezer           FreezerState      `json:"freezer,omitempty"`            // set the freeze value for the process
-	Slice             string            `json:"slice,omitempty"`              // Parent slice to use for systemd
-}

vendor/src/github.com/docker/libcontainer/cgroups/fs/apply_raw.go

@@ -1,13 +1,14 @@
 package fs
 
 import (
-	"fmt"
 	"io/ioutil"
 	"os"
 	"path/filepath"
 	"strconv"
+	"sync"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 var (
@@ -24,43 +25,65 @@ var (
 	CgroupProcesses = "cgroup.procs"
 )
 
-// The absolute path to the root of the cgroup hierarchies.
-var cgroupRoot string
-
-// TODO(vmarmol): Report error here, we'll probably need to wait for the new API.
-func init() {
-	// we can pick any subsystem to find the root
-	cpuRoot, err := cgroups.FindCgroupMountpoint("cpu")
-	if err != nil {
-		return
-	}
-	cgroupRoot = filepath.Dir(cpuRoot)
-
-	if _, err := os.Stat(cgroupRoot); err != nil {
-		return
-	}
-}
-
 type subsystem interface {
 	// Returns the stats, as 'stats', corresponding to the cgroup under 'path'.
 	GetStats(path string, stats *cgroups.Stats) error
 	// Removes the cgroup represented by 'data'.
 	Remove(*data) error
 	// Creates and joins the cgroup represented by data.
-	Set(*data) error
+	Apply(*data) error
+	// Set the cgroup represented by cgroup.
+	Set(path string, cgroup *configs.Cgroup) error
+}
+
+type Manager struct {
+	Cgroups *configs.Cgroup
+	Paths   map[string]string
+}
+
+// The absolute path to the root of the cgroup hierarchies.
+var cgroupRootLock sync.Mutex
+var cgroupRoot string
+
+// Gets the cgroupRoot.
+func getCgroupRoot() (string, error) {
+	cgroupRootLock.Lock()
+	defer cgroupRootLock.Unlock()
+
+	if cgroupRoot != "" {
+		return cgroupRoot, nil
+	}
+
+	// we can pick any subsystem to find the root
+	cpuRoot, err := cgroups.FindCgroupMountpoint("cpu")
+	if err != nil {
+		return "", err
+	}
+	root := filepath.Dir(cpuRoot)
+
+	if _, err := os.Stat(root); err != nil {
+		return "", err
+	}
+
+	cgroupRoot = root
+	return cgroupRoot, nil
 }
 
 type data struct {
 	root   string
 	cgroup string
-	c      *cgroups.Cgroup
+	c      *configs.Cgroup
 	pid    int
 }
 
-func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
-	d, err := getCgroupData(c, pid)
+func (m *Manager) Apply(pid int) error {
+	if m.Cgroups == nil {
+		return nil
+	}
+
+	d, err := getCgroupData(m.Cgroups, pid)
 	if err != nil {
-		return nil, err
+		return err
 	}
 
 	paths := make(map[string]string)
@@ -70,10 +93,10 @@ func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
 		}
 	}()
 	for name, sys := range subsystems {
-		if err := sys.Set(d); err != nil {
-			return nil, err
+		if err := sys.Apply(d); err != nil {
+			return err
 		}
-		// FIXME: Apply should, ideally, be reentrant or be broken up into a separate
+		// TODO: Apply should, ideally, be reentrant or be broken up into a separate
 		// create and join phase so that the cgroup hierarchy for a container can be
 		// created then join consists of writing the process pids to cgroup.procs
 		p, err := d.path(name)
@@ -81,16 +104,26 @@ func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
 			if cgroups.IsNotFound(err) {
 				continue
 			}
-			return nil, err
+			return err
 		}
 		paths[name] = p
 	}
-	return paths, nil
+	m.Paths = paths
+
+	return nil
+}
+
+func (m *Manager) Destroy() error {
+	return cgroups.RemovePaths(m.Paths)
+}
+
+func (m *Manager) GetPaths() map[string]string {
+	return m.Paths
 }
 
 // Symmetrical public function to update device based cgroups.  Also available
 // in the systemd implementation.
-func ApplyDevices(c *cgroups.Cgroup, pid int) error {
+func ApplyDevices(c *configs.Cgroup, pid int) error {
 	d, err := getCgroupData(c, pid)
 	if err != nil {
 		return err
@@ -98,12 +131,12 @@ func ApplyDevices(c *cgroups.Cgroup, pid int) error {
 
 	devices := subsystems["devices"]
 
-	return devices.Set(d)
+	return devices.Apply(d)
 }
 
-func GetStats(systemPaths map[string]string) (*cgroups.Stats, error) {
+func (m *Manager) GetStats() (*cgroups.Stats, error) {
 	stats := cgroups.NewStats()
-	for name, path := range systemPaths {
+	for name, path := range m.Paths {
 		sys, ok := subsystems[name]
 		if !ok || !cgroups.PathExists(path) {
 			continue
@@ -116,29 +149,48 @@ func GetStats(systemPaths map[string]string) (*cgroups.Stats, error) {
 	return stats, nil
 }
 
+func (m *Manager) Set(container *configs.Config) error {
+	for name, path := range m.Paths {
+		sys, ok := subsystems[name]
+		if !ok || !cgroups.PathExists(path) {
+			continue
+		}
+		if err := sys.Set(path, container.Cgroups); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
 // Freeze toggles the container's freezer cgroup depending on the state
 // provided
-func Freeze(c *cgroups.Cgroup, state cgroups.FreezerState) error {
-	d, err := getCgroupData(c, 0)
+func (m *Manager) Freeze(state configs.FreezerState) error {
+	d, err := getCgroupData(m.Cgroups, 0)
 	if err != nil {
 		return err
 	}
 
-	prevState := c.Freezer
-	c.Freezer = state
+	dir, err := d.path("freezer")
+	if err != nil {
+		return err
+	}
+
+	prevState := m.Cgroups.Freezer
+	m.Cgroups.Freezer = state
 
 	freezer := subsystems["freezer"]
-	err = freezer.Set(d)
+	err = freezer.Set(dir, m.Cgroups)
 	if err != nil {
-		c.Freezer = prevState
+		m.Cgroups.Freezer = prevState
 		return err
 	}
 
 	return nil
 }
 
-func GetPids(c *cgroups.Cgroup) ([]int, error) {
-	d, err := getCgroupData(c, 0)
+func (m *Manager) GetPids() ([]int, error) {
+	d, err := getCgroupData(m.Cgroups, 0)
 	if err != nil {
 		return nil, err
 	}
@@ -151,9 +203,10 @@ func GetPids(c *cgroups.Cgroup) ([]int, error) {
 	return cgroups.ReadProcsFile(dir)
 }
 
-func getCgroupData(c *cgroups.Cgroup, pid int) (*data, error) {
-	if cgroupRoot == "" {
-		return nil, fmt.Errorf("failed to find the cgroup root")
+func getCgroupData(c *configs.Cgroup, pid int) (*data, error) {
+	root, err := getCgroupRoot()
+	if err != nil {
+		return nil, err
 	}
 
 	cgroup := c.Name
@@ -162,7 +215,7 @@ func getCgroupData(c *cgroups.Cgroup, pid int) (*data, error) {
 	}
 
 	return &data{
-		root:   cgroupRoot,
+		root:   root,
 		cgroup: cgroup,
 		c:      c,
 		pid:    pid,

vendor/src/github.com/docker/libcontainer/cgroups/fs/blkio.go

@@ -9,17 +9,32 @@ import (
 	"strings"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 type BlkioGroup struct {
 }
 
-func (s *BlkioGroup) Set(d *data) error {
-	// we just want to join this group even though we don't set anything
-	if _, err := d.join("blkio"); err != nil && !cgroups.IsNotFound(err) {
+func (s *BlkioGroup) Apply(d *data) error {
+	dir, err := d.join("blkio")
+	if err != nil && !cgroups.IsNotFound(err) {
 		return err
 	}
 
+	if err := s.Set(dir, d.c); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (s *BlkioGroup) Set(path string, cgroup *configs.Cgroup) error {
+	if cgroup.BlkioWeight != 0 {
+		if err := writeFile(path, "blkio.weight", strconv.FormatInt(cgroup.BlkioWeight, 10)); err != nil {
+			return err
+		}
+	}
+
 	return nil
 }
 

vendor/src/github.com/docker/libcontainer/cgroups/fs/blkio_test.go

@@ -1,6 +1,7 @@
 package fs
 
 import (
+	"strconv"
 	"testing"
 
 	"github.com/docker/libcontainer/cgroups"
@@ -72,6 +73,35 @@ func appendBlkioStatEntry(blkioStatEntries *[]cgroups.BlkioStatEntry, major, min
 	*blkioStatEntries = append(*blkioStatEntries, cgroups.BlkioStatEntry{Major: major, Minor: minor, Value: value, Op: op})
 }
 
+func TestBlkioSetWeight(t *testing.T) {
+	helper := NewCgroupTestUtil("blkio", t)
+	defer helper.cleanup()
+
+	const (
+		weightBefore = 100
+		weightAfter  = 200
+	)
+
+	helper.writeFileContents(map[string]string{
+		"blkio.weight": strconv.Itoa(weightBefore),
+	})
+
+	helper.CgroupData.c.BlkioWeight = weightAfter
+	blkio := &BlkioGroup{}
+	if err := blkio.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "blkio.weight")
+	if err != nil {
+		t.Fatalf("Failed to parse blkio.weight - %s", err)
+	}
+
+	if value != weightAfter {
+		t.Fatal("Got the wrong value, set blkio.weight failed.")
+	}
+}
+
 func TestBlkioStats(t *testing.T) {
 	helper := NewCgroupTestUtil("blkio", t)
 	defer helper.cleanup()

vendor/src/github.com/docker/libcontainer/cgroups/fs/cpu.go

@@ -7,33 +7,44 @@ import (
 	"strconv"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 type CpuGroup struct {
 }
 
-func (s *CpuGroup) Set(d *data) error {
+func (s *CpuGroup) Apply(d *data) error {
 	// We always want to join the cpu group, to allow fair cpu scheduling
 	// on a container basis
 	dir, err := d.join("cpu")
 	if err != nil {
 		return err
 	}
-	if d.c.CpuShares != 0 {
-		if err := writeFile(dir, "cpu.shares", strconv.FormatInt(d.c.CpuShares, 10)); err != nil {
+
+	if err := s.Set(dir, d.c); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error {
+	if cgroup.CpuShares != 0 {
+		if err := writeFile(path, "cpu.shares", strconv.FormatInt(cgroup.CpuShares, 10)); err != nil {
 			return err
 		}
 	}
-	if d.c.CpuPeriod != 0 {
-		if err := writeFile(dir, "cpu.cfs_period_us", strconv.FormatInt(d.c.CpuPeriod, 10)); err != nil {
+	if cgroup.CpuPeriod != 0 {
+		if err := writeFile(path, "cpu.cfs_period_us", strconv.FormatInt(cgroup.CpuPeriod, 10)); err != nil {
 			return err
 		}
 	}
-	if d.c.CpuQuota != 0 {
-		if err := writeFile(dir, "cpu.cfs_quota_us", strconv.FormatInt(d.c.CpuQuota, 10)); err != nil {
+	if cgroup.CpuQuota != 0 {
+		if err := writeFile(path, "cpu.cfs_quota_us", strconv.FormatInt(cgroup.CpuQuota, 10)); err != nil {
 			return err
 		}
 	}
+
 	return nil
 }
 

vendor/src/github.com/docker/libcontainer/cgroups/fs/cpu_test.go

@@ -2,11 +2,81 @@ package fs
 
 import (
 	"fmt"
+	"strconv"
 	"testing"
 
 	"github.com/docker/libcontainer/cgroups"
 )
 
+func TestCpuSetShares(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	const (
+		sharesBefore = 1024
+		sharesAfter  = 512
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpu.shares": strconv.Itoa(sharesBefore),
+	})
+
+	helper.CgroupData.c.CpuShares = sharesAfter
+	cpu := &CpuGroup{}
+	if err := cpu.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "cpu.shares")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.shares - %s", err)
+	}
+
+	if value != sharesAfter {
+		t.Fatal("Got the wrong value, set cpu.shares failed.")
+	}
+}
+
+func TestCpuSetBandWidth(t *testing.T) {
+	helper := NewCgroupTestUtil("cpu", t)
+	defer helper.cleanup()
+
+	const (
+		quotaBefore  = 8000
+		quotaAfter   = 5000
+		periodBefore = 10000
+		periodAfter  = 7000
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpu.cfs_quota_us":  strconv.Itoa(quotaBefore),
+		"cpu.cfs_period_us": strconv.Itoa(periodBefore),
+	})
+
+	helper.CgroupData.c.CpuQuota = quotaAfter
+	helper.CgroupData.c.CpuPeriod = periodAfter
+	cpu := &CpuGroup{}
+	if err := cpu.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	quota, err := getCgroupParamUint(helper.CgroupPath, "cpu.cfs_quota_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.cfs_quota_us - %s", err)
+	}
+	if quota != quotaAfter {
+		t.Fatal("Got the wrong value, set cpu.cfs_quota_us failed.")
+	}
+
+	period, err := getCgroupParamUint(helper.CgroupPath, "cpu.cfs_period_us")
+	if err != nil {
+		t.Fatalf("Failed to parse cpu.cfs_period_us - %s", err)
+	}
+	if period != periodAfter {
+		t.Fatal("Got the wrong value, set cpu.cfs_period_us failed.")
+	}
+}
+
 func TestCpuStats(t *testing.T) {
 	helper := NewCgroupTestUtil("cpu", t)
 	defer helper.cleanup()

vendor/src/github.com/docker/libcontainer/cgroups/fs/cpuacct.go

@@ -8,6 +8,7 @@ import (
 	"strings"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 	"github.com/docker/libcontainer/system"
 )
 
@@ -21,7 +22,7 @@ var clockTicks = uint64(system.GetClockTicks())
 type CpuacctGroup struct {
 }
 
-func (s *CpuacctGroup) Set(d *data) error {
+func (s *CpuacctGroup) Apply(d *data) error {
 	// we just want to join this group even though we don't set anything
 	if _, err := d.join("cpuacct"); err != nil && !cgroups.IsNotFound(err) {
 		return err
@@ -30,6 +31,10 @@ func (s *CpuacctGroup) Set(d *data) error {
 	return nil
 }
 
+func (s *CpuacctGroup) Set(path string, cgroup *configs.Cgroup) error {
+	return nil
+}
+
 func (s *CpuacctGroup) Remove(d *data) error {
 	return removePath(d.path("cpuacct"))
 }

vendor/src/github.com/docker/libcontainer/cgroups/fs/cpuset.go

@@ -8,17 +8,34 @@ import (
 	"strconv"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 type CpusetGroup struct {
 }
 
-func (s *CpusetGroup) Set(d *data) error {
+func (s *CpusetGroup) Apply(d *data) error {
 	dir, err := d.path("cpuset")
 	if err != nil {
 		return err
 	}
-	return s.SetDir(dir, d.c.CpusetCpus, d.c.CpusetMems, d.pid)
+	return s.ApplyDir(dir, d.c, d.pid)
+}
+
+func (s *CpusetGroup) Set(path string, cgroup *configs.Cgroup) error {
+	if cgroup.CpusetCpus != "" {
+		if err := writeFile(path, "cpuset.cpus", cgroup.CpusetCpus); err != nil {
+			return err
+		}
+	}
+
+	if cgroup.CpusetMems != "" {
+		if err := writeFile(path, "cpuset.mems", cgroup.CpusetMems); err != nil {
+			return err
+		}
+	}
+
+	return nil
 }
 
 func (s *CpusetGroup) Remove(d *data) error {
@@ -29,7 +46,7 @@ func (s *CpusetGroup) GetStats(path string, stats *cgroups.Stats) error {
 	return nil
 }
 
-func (s *CpusetGroup) SetDir(dir, cpus string, mems string, pid int) error {
+func (s *CpusetGroup) ApplyDir(dir string, cgroup *configs.Cgroup, pid int) error {
 	if err := s.ensureParent(dir); err != nil {
 		return err
 	}
@@ -40,17 +57,10 @@ func (s *CpusetGroup) SetDir(dir, cpus string, mems string, pid int) error {
 		return err
 	}
 
-	// If we don't use --cpuset-xxx, the default value inherit from parent cgroup
-	// is set in s.ensureParent, otherwise, use the value we set
-	if cpus != "" {
-		if err := writeFile(dir, "cpuset.cpus", cpus); err != nil {
-			return err
-		}
-	}
-	if mems != "" {
-		if err := writeFile(dir, "cpuset.mems", mems); err != nil {
-			return err
-		}
+	// the default values inherit from parent cgroup are already set in
+	// s.ensureParent, cover these if we have our own
+	if err := s.Set(dir, cgroup); err != nil {
+		return err
 	}
 
 	return nil

vendor/src/github.com/docker/libcontainer/cgroups/fs/cpuset_test.go

@@ -0,0 +1,63 @@
+package fs
+
+import (
+	"testing"
+)
+
+func TestCpusetSetCpus(t *testing.T) {
+	helper := NewCgroupTestUtil("cpuset", t)
+	defer helper.cleanup()
+
+	const (
+		cpusBefore = "0"
+		cpusAfter  = "1-3"
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpuset.cpus": cpusBefore,
+	})
+
+	helper.CgroupData.c.CpusetCpus = cpusAfter
+	cpuset := &CpusetGroup{}
+	if err := cpuset.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "cpuset.cpus")
+	if err != nil {
+		t.Fatalf("Failed to parse cpuset.cpus - %s", err)
+	}
+
+	if value != cpusAfter {
+		t.Fatal("Got the wrong value, set cpuset.cpus failed.")
+	}
+}
+
+func TestCpusetSetMems(t *testing.T) {
+	helper := NewCgroupTestUtil("cpuset", t)
+	defer helper.cleanup()
+
+	const (
+		memsBefore = "0"
+		memsAfter  = "1"
+	)
+
+	helper.writeFileContents(map[string]string{
+		"cpuset.mems": memsBefore,
+	})
+
+	helper.CgroupData.c.CpusetMems = memsAfter
+	cpuset := &CpusetGroup{}
+	if err := cpuset.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamString(helper.CgroupPath, "cpuset.mems")
+	if err != nil {
+		t.Fatalf("Failed to parse cpuset.mems - %s", err)
+	}
+
+	if value != memsAfter {
+		t.Fatal("Got the wrong value, set cpuset.mems failed.")
+	}
+}

vendor/src/github.com/docker/libcontainer/cgroups/fs/devices.go

@@ -1,27 +1,39 @@
 package fs
 
-import "github.com/docker/libcontainer/cgroups"
+import (
+	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
+)
 
 type DevicesGroup struct {
 }
 
-func (s *DevicesGroup) Set(d *data) error {
+func (s *DevicesGroup) Apply(d *data) error {
 	dir, err := d.join("devices")
 	if err != nil {
 		return err
 	}
 
-	if !d.c.AllowAllDevices {
-		if err := writeFile(dir, "devices.deny", "a"); err != nil {
+	if err := s.Set(dir, d.c); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (s *DevicesGroup) Set(path string, cgroup *configs.Cgroup) error {
+	if !cgroup.AllowAllDevices {
+		if err := writeFile(path, "devices.deny", "a"); err != nil {
 			return err
 		}
 
-		for _, dev := range d.c.AllowedDevices {
-			if err := writeFile(dir, "devices.allow", dev.GetCgroupAllowString()); err != nil {
+		for _, dev := range cgroup.AllowedDevices {
+			if err := writeFile(path, "devices.allow", dev.CgroupString()); err != nil {
 				return err
 			}
 		}
 	}
+
 	return nil
 }
 

vendor/src/github.com/docker/libcontainer/cgroups/fs/devices_test.go

@@ -0,0 +1,48 @@
+package fs
+
+import (
+	"testing"
+
+	"github.com/docker/libcontainer/configs"
+)
+
+var (
+	allowedDevices = []*configs.Device{
+		{
+			Path:        "/dev/zero",
+			Type:        'c',
+			Major:       1,
+			Minor:       5,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+	}
+	allowedList = "c 1:5 rwm"
+)
+
+func TestDevicesSetAllow(t *testing.T) {
+	helper := NewCgroupTestUtil("devices", t)
+	defer helper.cleanup()
+
+	helper.writeFileContents(map[string]string{
+		"device.deny": "a",
+	})
+
+	helper.CgroupData.c.AllowAllDevices = false
+	helper.CgroupData.c.AllowedDevices = allowedDevices
+	devices := &DevicesGroup{}
+	if err := devices.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	// FIXME: this doesn't make sence, the file devices.allow under real cgroupfs
+	// is not allowed to read. Our test path don't have cgroupfs mounted.
+	value, err := getCgroupParamString(helper.CgroupPath, "devices.allow")
+	if err != nil {
+		t.Fatalf("Failed to parse devices.allow - %s", err)
+	}
+
+	if value != allowedList {
+		t.Fatal("Got the wrong value, set devices.allow failed.")
+	}
+}

vendor/src/github.com/docker/libcontainer/cgroups/fs/freezer.go

@@ -5,37 +5,42 @@ import (
 	"time"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 type FreezerGroup struct {
 }
 
-func (s *FreezerGroup) Set(d *data) error {
-	switch d.c.Freezer {
-	case cgroups.Frozen, cgroups.Thawed:
-		dir, err := d.path("freezer")
-		if err != nil {
-			return err
-		}
+func (s *FreezerGroup) Apply(d *data) error {
+	dir, err := d.join("freezer")
+	if err != nil && !cgroups.IsNotFound(err) {
+		return err
+	}
 
-		if err := writeFile(dir, "freezer.state", string(d.c.Freezer)); err != nil {
+	if err := s.Set(dir, d.c); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (s *FreezerGroup) Set(path string, cgroup *configs.Cgroup) error {
+	switch cgroup.Freezer {
+	case configs.Frozen, configs.Thawed:
+		if err := writeFile(path, "freezer.state", string(cgroup.Freezer)); err != nil {
 			return err
 		}
 
 		for {
-			state, err := readFile(dir, "freezer.state")
+			state, err := readFile(path, "freezer.state")
 			if err != nil {
 				return err
 			}
-			if strings.TrimSpace(state) == string(d.c.Freezer) {
+			if strings.TrimSpace(state) == string(cgroup.Freezer) {
 				break
 			}
 			time.Sleep(1 * time.Millisecond)
 		}
-	default:
-		if _, err := d.join("freezer"); err != nil && !cgroups.IsNotFound(err) {
-			return err
-		}
 	}
 
 	return nil

vendor/src/github.com/docker/libcontainer/cgroups/fs/memory.go

@@ -8,12 +8,13 @@ import (
 	"strconv"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 type MemoryGroup struct {
 }
 
-func (s *MemoryGroup) Set(d *data) error {
+func (s *MemoryGroup) Apply(d *data) error {
 	dir, err := d.join("memory")
 	// only return an error for memory if it was specified
 	if err != nil && (d.c.Memory != 0 || d.c.MemoryReservation != 0 || d.c.MemorySwap != 0) {
@@ -25,31 +26,42 @@ func (s *MemoryGroup) Set(d *data) error {
 		}
 	}()
 
-	// Only set values if some config was specified.
-	if d.c.Memory != 0 || d.c.MemoryReservation != 0 || d.c.MemorySwap != 0 {
-		if d.c.Memory != 0 {
-			if err := writeFile(dir, "memory.limit_in_bytes", strconv.FormatInt(d.c.Memory, 10)); err != nil {
-				return err
-			}
-		}
-		if d.c.MemoryReservation != 0 {
-			if err := writeFile(dir, "memory.soft_limit_in_bytes", strconv.FormatInt(d.c.MemoryReservation, 10)); err != nil {
-				return err
-			}
-		}
-		// By default, MemorySwap is set to twice the size of RAM.
-		// If you want to omit MemorySwap, set it to '-1'.
-		if d.c.MemorySwap == 0 {
-			if err := writeFile(dir, "memory.memsw.limit_in_bytes", strconv.FormatInt(d.c.Memory*2, 10)); err != nil {
-				return err
-			}
-		}
-		if d.c.MemorySwap > 0 {
-			if err := writeFile(dir, "memory.memsw.limit_in_bytes", strconv.FormatInt(d.c.MemorySwap, 10)); err != nil {
-				return err
-			}
+	if err := s.Set(dir, d.c); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func (s *MemoryGroup) Set(path string, cgroup *configs.Cgroup) error {
+	if cgroup.Memory != 0 {
+		if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Memory, 10)); err != nil {
+			return err
 		}
 	}
+	if cgroup.MemoryReservation != 0 {
+		if err := writeFile(path, "memory.soft_limit_in_bytes", strconv.FormatInt(cgroup.MemoryReservation, 10)); err != nil {
+			return err
+		}
+	}
+	// By default, MemorySwap is set to twice the size of Memory.
+	if cgroup.MemorySwap == 0 && cgroup.Memory != 0 {
+		if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Memory*2, 10)); err != nil {
+			return err
+		}
+	}
+	if cgroup.MemorySwap > 0 {
+		if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.MemorySwap, 10)); err != nil {
+			return err
+		}
+	}
+
+	if cgroup.OomKillDisable {
+		if err := writeFile(path, "memory.oom_control", "1"); err != nil {
+			return err
+		}
+	}
+
 	return nil
 }
 

vendor/src/github.com/docker/libcontainer/cgroups/fs/memory_test.go

@@ -1,6 +1,7 @@
 package fs
 
 import (
+	"strconv"
 	"testing"
 
 	"github.com/docker/libcontainer/cgroups"
@@ -14,6 +15,46 @@ rss 1024`
 	memoryFailcnt          = "100\n"
 )
 
+func TestMemorySetMemory(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		memoryBefore      = 314572800 // 300M
+		memoryAfter       = 524288000 // 500M
+		reservationBefore = 209715200 // 200M
+		reservationAfter  = 314572800 // 300M
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.limit_in_bytes":      strconv.Itoa(memoryBefore),
+		"memory.soft_limit_in_bytes": strconv.Itoa(reservationBefore),
+	})
+
+	helper.CgroupData.c.Memory = memoryAfter
+	helper.CgroupData.c.MemoryReservation = reservationAfter
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.limit_in_bytes - %s", err)
+	}
+	if value != memoryAfter {
+		t.Fatal("Got the wrong value, set memory.limit_in_bytes failed.")
+	}
+
+	value, err = getCgroupParamUint(helper.CgroupPath, "memory.soft_limit_in_bytes")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.soft_limit_in_bytes - %s", err)
+	}
+	if value != reservationAfter {
+		t.Fatal("Got the wrong value, set memory.soft_limit_in_bytes failed.")
+	}
+}
+
 func TestMemoryStats(t *testing.T) {
 	helper := NewCgroupTestUtil("memory", t)
 	defer helper.cleanup()
@@ -132,3 +173,30 @@ func TestMemoryStatsBadMaxUsageFile(t *testing.T) {
 		t.Fatal("Expected failure")
 	}
 }
+
+func TestMemorySetOomControl(t *testing.T) {
+	helper := NewCgroupTestUtil("memory", t)
+	defer helper.cleanup()
+
+	const (
+		oom_kill_disable = 1 // disable oom killer, default is 0
+	)
+
+	helper.writeFileContents(map[string]string{
+		"memory.oom_control": strconv.Itoa(oom_kill_disable),
+	})
+
+	memory := &MemoryGroup{}
+	if err := memory.Set(helper.CgroupPath, helper.CgroupData.c); err != nil {
+		t.Fatal(err)
+	}
+
+	value, err := getCgroupParamUint(helper.CgroupPath, "memory.oom_control")
+	if err != nil {
+		t.Fatalf("Failed to parse memory.oom_control - %s", err)
+	}
+
+	if value != oom_kill_disable {
+		t.Fatalf("Got the wrong value, set memory.oom_control failed.")
+	}
+}

vendor/src/github.com/docker/libcontainer/cgroups/fs/perf_event.go

@@ -2,12 +2,13 @@ package fs
 
 import (
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
 type PerfEventGroup struct {
 }
 
-func (s *PerfEventGroup) Set(d *data) error {
+func (s *PerfEventGroup) Apply(d *data) error {
 	// we just want to join this group even though we don't set anything
 	if _, err := d.join("perf_event"); err != nil && !cgroups.IsNotFound(err) {
 		return err
@@ -15,6 +16,10 @@ func (s *PerfEventGroup) Set(d *data) error {
 	return nil
 }
 
+func (s *PerfEventGroup) Set(path string, cgroup *configs.Cgroup) error {
+	return nil
+}
+
 func (s *PerfEventGroup) Remove(d *data) error {
 	return removePath(d.path("perf_event"))
 }

vendor/src/github.com/docker/libcontainer/cgroups/fs/util_test.go

@@ -10,6 +10,8 @@ import (
 	"io/ioutil"
 	"os"
 	"testing"
+
+	"github.com/docker/libcontainer/configs"
 )
 
 type cgroupTestUtil struct {
@@ -26,7 +28,9 @@ type cgroupTestUtil struct {
 
 // Creates a new test util for the specified subsystem
 func NewCgroupTestUtil(subsystem string, t *testing.T) *cgroupTestUtil {
-	d := &data{}
+	d := &data{
+		c: &configs.Cgroup{},
+	}
 	tempDir, err := ioutil.TempDir("", fmt.Sprintf("%s_cgroup_test", subsystem))
 	if err != nil {
 		t.Fatal(err)

vendor/src/github.com/docker/libcontainer/cgroups/fs/utils.go

@@ -60,3 +60,13 @@ func getCgroupParamUint(cgroupPath, cgroupFile string) (uint64, error) {
 
 	return parseUint(strings.TrimSpace(string(contents)), 10, 64)
 }
+
+// Gets a string value from the specified cgroup file
+func getCgroupParamString(cgroupPath, cgroupFile string) (string, error) {
+	contents, err := ioutil.ReadFile(filepath.Join(cgroupPath, cgroupFile))
+	if err != nil {
+		return "", err
+	}
+
+	return strings.TrimSpace(string(contents)), nil
+}

vendor/src/github.com/docker/libcontainer/cgroups/systemd/apply_nosystemd.go

@@ -6,24 +6,50 @@ import (
 	"fmt"
 
 	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
 )
 
+type Manager struct {
+	Cgroups *configs.Cgroup
+	Paths   map[string]string
+}
+
 func UseSystemd() bool {
 	return false
 }
 
-func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
-	return nil, fmt.Errorf("Systemd not supported")
-}
-
-func GetPids(c *cgroups.Cgroup) ([]int, error) {
-	return nil, fmt.Errorf("Systemd not supported")
-}
-
-func ApplyDevices(c *cgroups.Cgroup, pid int) error {
+func (m *Manager) Apply(pid int) error {
 	return fmt.Errorf("Systemd not supported")
 }
 
-func Freeze(c *cgroups.Cgroup, state cgroups.FreezerState) error {
+func (m *Manager) GetPids() ([]int, error) {
+	return nil, fmt.Errorf("Systemd not supported")
+}
+
+func (m *Manager) Destroy() error {
+	return fmt.Errorf("Systemd not supported")
+}
+
+func (m *Manager) GetPaths() map[string]string {
+	return nil
+}
+
+func (m *Manager) GetStats() (*cgroups.Stats, error) {
+	return nil, fmt.Errorf("Systemd not supported")
+}
+
+func (m *Manager) Set(container *configs.Config) error {
+	return nil, fmt.Errorf("Systemd not supported")
+}
+
+func (m *Manager) Freeze(state configs.FreezerState) error {
+	return fmt.Errorf("Systemd not supported")
+}
+
+func ApplyDevices(c *configs.Cgroup, pid int) error {
+	return fmt.Errorf("Systemd not supported")
+}
+
+func Freeze(c *configs.Cgroup, state configs.FreezerState) error {
 	return fmt.Errorf("Systemd not supported")
 }

vendor/src/github.com/docker/libcontainer/cgroups/systemd/apply_systemd.go

@@ -16,21 +16,38 @@ import (
 	systemd "github.com/coreos/go-systemd/dbus"
 	"github.com/docker/libcontainer/cgroups"
 	"github.com/docker/libcontainer/cgroups/fs"
+	"github.com/docker/libcontainer/configs"
 	"github.com/godbus/dbus"
 )
 
-type systemdCgroup struct {
-	cgroup *cgroups.Cgroup
+type Manager struct {
+	Cgroups *configs.Cgroup
+	Paths   map[string]string
 }
 
 type subsystem interface {
-	GetStats(string, *cgroups.Stats) error
+	// Returns the stats, as 'stats', corresponding to the cgroup under 'path'.
+	GetStats(path string, stats *cgroups.Stats) error
+	// Set the cgroup represented by cgroup.
+	Set(path string, cgroup *configs.Cgroup) error
+}
+
+var subsystems = map[string]subsystem{
+	"devices":    &fs.DevicesGroup{},
+	"memory":     &fs.MemoryGroup{},
+	"cpu":        &fs.CpuGroup{},
+	"cpuset":     &fs.CpusetGroup{},
+	"cpuacct":    &fs.CpuacctGroup{},
+	"blkio":      &fs.BlkioGroup{},
+	"perf_event": &fs.PerfEventGroup{},
+	"freezer":    &fs.FreezerGroup{},
 }
 
 var (
-	connLock              sync.Mutex
-	theConn               *systemd.Conn
-	hasStartTransientUnit bool
+	connLock                        sync.Mutex
+	theConn                         *systemd.Conn
+	hasStartTransientUnit           bool
+	hasTransientDefaultDependencies bool
 )
 
 func newProp(name string, units interface{}) systemd.Property {
@@ -64,6 +81,18 @@ func UseSystemd() bool {
 			if dbusError, ok := err.(dbus.Error); ok {
 				if dbusError.Name == "org.freedesktop.DBus.Error.UnknownMethod" {
 					hasStartTransientUnit = false
+					return hasStartTransientUnit
+				}
+			}
+		}
+
+		// Assume StartTransientUnit on a scope allows DefaultDependencies
+		hasTransientDefaultDependencies = true
+		ddf := newProp("DefaultDependencies", false)
+		if _, err := theConn.StartTransientUnit("docker-systemd-test-default-dependencies.scope", "replace", ddf); err != nil {
+			if dbusError, ok := err.(dbus.Error); ok {
+				if dbusError.Name == "org.freedesktop.DBus.Error.PropertyReadOnly" {
+					hasTransientDefaultDependencies = false
 				}
 			}
 		}
@@ -81,16 +110,14 @@ func getIfaceForUnit(unitName string) string {
 	return "Unit"
 }
 
-func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
+func (m *Manager) Apply(pid int) error {
 	var (
+		c          = m.Cgroups
 		unitName   = getUnitName(c)
 		slice      = "system.slice"
 		properties []systemd.Property
-		res        = &systemdCgroup{}
 	)
 
-	res.cgroup = c
-
 	if c.Slice != "" {
 		slice = c.Slice
 	}
@@ -108,6 +135,11 @@ func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
 		newProp("CPUAccounting", true),
 		newProp("BlockIOAccounting", true))
 
+	if hasTransientDefaultDependencies {
+		properties = append(properties,
+			newProp("DefaultDependencies", false))
+	}
+
 	if c.Memory != 0 {
 		properties = append(properties,
 			newProp("MemoryLimit", uint64(c.Memory)))
@@ -119,20 +151,29 @@ func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
 			newProp("CPUShares", uint64(c.CpuShares)))
 	}
 
-	if _, err := theConn.StartTransientUnit(unitName, "replace", properties...); err != nil {
-		return nil, err
+	if c.BlkioWeight != 0 {
+		properties = append(properties,
+			newProp("BlockIOWeight", uint64(c.BlkioWeight)))
 	}
 
-	if !c.AllowAllDevices {
-		if err := joinDevices(c, pid); err != nil {
-			return nil, err
-		}
+	if _, err := theConn.StartTransientUnit(unitName, "replace", properties...); err != nil {
+		return err
+	}
+
+	if err := joinDevices(c, pid); err != nil {
+		return err
+	}
+
+	// TODO: CpuQuota and CpuPeriod not available in systemd
+	// we need to manually join the cpu.cfs_quota_us and cpu.cfs_period_us
+	if err := joinCpu(c, pid); err != nil {
+		return err
 	}
 
 	// -1 disables memorySwap
-	if c.MemorySwap >= 0 && (c.Memory != 0 || c.MemorySwap > 0) {
+	if c.MemorySwap >= 0 && c.Memory != 0 {
 		if err := joinMemory(c, pid); err != nil {
-			return nil, err
+			return err
 		}
 
 	}
@@ -140,11 +181,11 @@ func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
 	// we need to manually join the freezer and cpuset cgroup in systemd
 	// because it does not currently support it via the dbus api.
 	if err := joinFreezer(c, pid); err != nil {
-		return nil, err
+		return err
 	}
 
 	if err := joinCpuset(c, pid); err != nil {
-		return nil, err
+		return err
 	}
 
 	paths := make(map[string]string)
@@ -158,24 +199,53 @@ func Apply(c *cgroups.Cgroup, pid int) (map[string]string, error) {
 		"perf_event",
 		"freezer",
 	} {
-		subsystemPath, err := getSubsystemPath(res.cgroup, sysname)
+		subsystemPath, err := getSubsystemPath(m.Cgroups, sysname)
 		if err != nil {
 			// Don't fail if a cgroup hierarchy was not found, just skip this subsystem
 			if cgroups.IsNotFound(err) {
 				continue
 			}
-			return nil, err
+			return err
 		}
 		paths[sysname] = subsystemPath
 	}
-	return paths, nil
+
+	m.Paths = paths
+
+	return nil
+}
+
+func (m *Manager) Destroy() error {
+	return cgroups.RemovePaths(m.Paths)
+}
+
+func (m *Manager) GetPaths() map[string]string {
+	return m.Paths
 }
 
 func writeFile(dir, file, data string) error {
 	return ioutil.WriteFile(filepath.Join(dir, file), []byte(data), 0700)
 }
 
-func joinFreezer(c *cgroups.Cgroup, pid int) error {
+func joinCpu(c *configs.Cgroup, pid int) error {
+	path, err := getSubsystemPath(c, "cpu")
+	if err != nil {
+		return err
+	}
+	if c.CpuQuota != 0 {
+		if err = ioutil.WriteFile(filepath.Join(path, "cpu.cfs_quota_us"), []byte(strconv.FormatInt(c.CpuQuota, 10)), 0700); err != nil {
+			return err
+		}
+	}
+	if c.CpuPeriod != 0 {
+		if err = ioutil.WriteFile(filepath.Join(path, "cpu.cfs_period_us"), []byte(strconv.FormatInt(c.CpuPeriod, 10)), 0700); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func joinFreezer(c *configs.Cgroup, pid int) error {
 	path, err := getSubsystemPath(c, "freezer")
 	if err != nil {
 		return err
@@ -188,7 +258,7 @@ func joinFreezer(c *cgroups.Cgroup, pid int) error {
 	return ioutil.WriteFile(filepath.Join(path, "cgroup.procs"), []byte(strconv.Itoa(pid)), 0700)
 }
 
-func getSubsystemPath(c *cgroups.Cgroup, subsystem string) (string, error) {
+func getSubsystemPath(c *configs.Cgroup, subsystem string) (string, error) {
 	mountpoint, err := cgroups.FindCgroupMountpoint(subsystem)
 	if err != nil {
 		return "", err
@@ -207,8 +277,8 @@ func getSubsystemPath(c *cgroups.Cgroup, subsystem string) (string, error) {
 	return filepath.Join(mountpoint, initPath, slice, getUnitName(c)), nil
 }
 
-func Freeze(c *cgroups.Cgroup, state cgroups.FreezerState) error {
-	path, err := getSubsystemPath(c, "freezer")
+func (m *Manager) Freeze(state configs.FreezerState) error {
+	path, err := getSubsystemPath(m.Cgroups, "freezer")
 	if err != nil {
 		return err
 	}
@@ -226,11 +296,14 @@ func Freeze(c *cgroups.Cgroup, state cgroups.FreezerState) error {
 		}
 		time.Sleep(1 * time.Millisecond)
 	}
+
+	m.Cgroups.Freezer = state
+
 	return nil
 }
 
-func GetPids(c *cgroups.Cgroup) ([]int, error) {
-	path, err := getSubsystemPath(c, "cpu")
+func (m *Manager) GetPids() ([]int, error) {
+	path, err := getSubsystemPath(m.Cgroups, "cpu")
 	if err != nil {
 		return nil, err
 	}
@@ -238,7 +311,26 @@ func GetPids(c *cgroups.Cgroup) ([]int, error) {
 	return cgroups.ReadProcsFile(path)
 }
 
-func getUnitName(c *cgroups.Cgroup) string {
+func (m *Manager) GetStats() (*cgroups.Stats, error) {
+	stats := cgroups.NewStats()
+	for name, path := range m.Paths {
+		sys, ok := subsystems[name]
+		if !ok || !cgroups.PathExists(path) {
+			continue
+		}
+		if err := sys.GetStats(path, stats); err != nil {
+			return nil, err
+		}
+	}
+
+	return stats, nil
+}
+
+func (m *Manager) Set(container *configs.Config) error {
+	panic("not implemented")
+}
+
+func getUnitName(c *configs.Cgroup) string {
 	return fmt.Sprintf("%s-%s.scope", c.Parent, c.Name)
 }
 
@@ -253,7 +345,7 @@ func getUnitName(c *cgroups.Cgroup) string {
 // Note: we can't use systemd to set up the initial limits, and then change the cgroup
 // because systemd will re-write the device settings if it needs to re-apply the cgroup context.
 // This happens at least for v208 when any sibling unit is started.
-func joinDevices(c *cgroups.Cgroup, pid int) error {
+func joinDevices(c *configs.Cgroup, pid int) error {
 	path, err := getSubsystemPath(c, "devices")
 	if err != nil {
 		return err
@@ -267,26 +359,26 @@ func joinDevices(c *cgroups.Cgroup, pid int) error {
 		return err
 	}
 
-	if err := writeFile(path, "devices.deny", "a"); err != nil {
-		return err
-	}
-
-	for _, dev := range c.AllowedDevices {
-		if err := writeFile(path, "devices.allow", dev.GetCgroupAllowString()); err != nil {
+	if !c.AllowAllDevices {
+		if err := writeFile(path, "devices.deny", "a"); err != nil {
+			return err
+		}
+	}
+	for _, dev := range c.AllowedDevices {
+		if err := writeFile(path, "devices.allow", dev.CgroupString()); err != nil {
 			return err
 		}
 	}
-
 	return nil
 }
 
 // Symmetrical public function to update device based cgroups.  Also available
 // in the fs implementation.
-func ApplyDevices(c *cgroups.Cgroup, pid int) error {
+func ApplyDevices(c *configs.Cgroup, pid int) error {
 	return joinDevices(c, pid)
 }
 
-func joinMemory(c *cgroups.Cgroup, pid int) error {
+func joinMemory(c *configs.Cgroup, pid int) error {
 	memorySwap := c.MemorySwap
 
 	if memorySwap == 0 {
@@ -305,7 +397,7 @@ func joinMemory(c *cgroups.Cgroup, pid int) error {
 // systemd does not atm set up the cpuset controller, so we must manually
 // join it. Additionally that is a very finicky controller where each
 // level must have a full setup as the default for a new directory is "no cpus"
-func joinCpuset(c *cgroups.Cgroup, pid int) error {
+func joinCpuset(c *configs.Cgroup, pid int) error {
 	path, err := getSubsystemPath(c, "cpuset")
 	if err != nil {
 		return err
@@ -313,5 +405,5 @@ func joinCpuset(c *cgroups.Cgroup, pid int) error {
 
 	s := &fs.CpusetGroup{}
 
-	return s.SetDir(path, c.CpusetCpus, c.CpusetMems, pid)
+	return s.ApplyDir(path, c, pid)
 }

vendor/src/github.com/docker/libcontainer/config.go

@@ -1,154 +0,0 @@
-package libcontainer
-
-import (
-	"github.com/docker/libcontainer/cgroups"
-	"github.com/docker/libcontainer/mount"
-	"github.com/docker/libcontainer/network"
-)
-
-type MountConfig mount.MountConfig
-
-type Network network.Network
-
-type NamespaceType string
-
-const (
-	NEWNET  NamespaceType = "NEWNET"
-	NEWPID  NamespaceType = "NEWPID"
-	NEWNS   NamespaceType = "NEWNS"
-	NEWUTS  NamespaceType = "NEWUTS"
-	NEWIPC  NamespaceType = "NEWIPC"
-	NEWUSER NamespaceType = "NEWUSER"
-)
-
-// Namespace defines configuration for each namespace.  It specifies an
-// alternate path that is able to be joined via setns.
-type Namespace struct {
-	Type NamespaceType `json:"type"`
-	Path string        `json:"path,omitempty"`
-}
-
-type Namespaces []Namespace
-
-func (n *Namespaces) Remove(t NamespaceType) bool {
-	i := n.index(t)
-	if i == -1 {
-		return false
-	}
-	*n = append((*n)[:i], (*n)[i+1:]...)
-	return true
-}
-
-func (n *Namespaces) Add(t NamespaceType, path string) {
-	i := n.index(t)
-	if i == -1 {
-		*n = append(*n, Namespace{Type: t, Path: path})
-		return
-	}
-	(*n)[i].Path = path
-}
-
-func (n *Namespaces) index(t NamespaceType) int {
-	for i, ns := range *n {
-		if ns.Type == t {
-			return i
-		}
-	}
-	return -1
-}
-
-func (n *Namespaces) Contains(t NamespaceType) bool {
-	return n.index(t) != -1
-}
-
-// Config defines configuration options for executing a process inside a contained environment.
-type Config struct {
-	// Mount specific options.
-	MountConfig *MountConfig `json:"mount_config,omitempty"`
-
-	// Pathname to container's root filesystem
-	RootFs string `json:"root_fs,omitempty"`
-
-	// Hostname optionally sets the container's hostname if provided
-	Hostname string `json:"hostname,omitempty"`
-
-	// User will set the uid and gid of the executing process running inside the container
-	User string `json:"user,omitempty"`
-
-	// WorkingDir will change the processes current working directory inside the container's rootfs
-	WorkingDir string `json:"working_dir,omitempty"`
-
-	// Env will populate the processes environment with the provided values
-	// Any values from the parent processes will be cleared before the values
-	// provided in Env are provided to the process
-	Env []string `json:"environment,omitempty"`
-
-	// Tty when true will allocate a pty slave on the host for access by the container's process
-	// and ensure that it is mounted inside the container's rootfs
-	Tty bool `json:"tty,omitempty"`
-
-	// Namespaces specifies the container's namespaces that it should setup when cloning the init process
-	// If a namespace is not provided that namespace is shared from the container's parent process
-	Namespaces Namespaces `json:"namespaces,omitempty"`
-
-	// Capabilities specify the capabilities to keep when executing the process inside the container
-	// All capbilities not specified will be dropped from the processes capability mask
-	Capabilities []string `json:"capabilities,omitempty"`
-
-	// Networks specifies the container's network setup to be created
-	Networks []*Network `json:"networks,omitempty"`
-
-	// Routes can be specified to create entries in the route table as the container is started
-	Routes []*Route `json:"routes,omitempty"`
-
-	// Cgroups specifies specific cgroup settings for the various subsystems that the container is
-	// placed into to limit the resources the container has available
-	Cgroups *cgroups.Cgroup `json:"cgroups,omitempty"`
-
-	// AppArmorProfile specifies the profile to apply to the process running in the container and is
-	// change at the time the process is execed
-	AppArmorProfile string `json:"apparmor_profile,omitempty"`
-
-	// ProcessLabel specifies the label to apply to the process running in the container.  It is
-	// commonly used by selinux
-	ProcessLabel string `json:"process_label,omitempty"`
-
-	// RestrictSys will remount /proc/sys, /sys, and mask over sysrq-trigger as well as /proc/irq and
-	// /proc/bus
-	RestrictSys bool `json:"restrict_sys,omitempty"`
-
-	// Rlimits specifies the resource limits, such as max open files, to set in the container
-	// If Rlimits are not set, the container will inherit rlimits from the parent process
-	Rlimits []Rlimit `json:"rlimits,omitempty"`
-
-	// AdditionalGroups specifies the gids that should be added to supplementary groups
-	// in addition to those that the user belongs to.
-	AdditionalGroups []int `json:"additional_groups,omitempty"`
-}
-
-// Routes can be specified to create entries in the route table as the container is started
-//
-// All of destination, source, and gateway should be either IPv4 or IPv6.
-// One of the three options must be present, and ommitted entries will use their
-// IP family default for the route table.  For IPv4 for example, setting the
-// gateway to 1.2.3.4 and the interface to eth0 will set up a standard
-// destination of 0.0.0.0(or *) when viewed in the route table.
-type Route struct {
-	// Sets the destination and mask, should be a CIDR.  Accepts IPv4 and IPv6
-	Destination string `json:"destination,omitempty"`
-
-	// Sets the source and mask, should be a CIDR.  Accepts IPv4 and IPv6
-	Source string `json:"source,omitempty"`
-
-	// Sets the gateway.  Accepts IPv4 and IPv6
-	Gateway string `json:"gateway,omitempty"`
-
-	// The device to set this route up for, for example: eth0
-	InterfaceName string `json:"interface_name,omitempty"`
-}
-
-type Rlimit struct {
-	Type int    `json:"type,omitempty"`
-	Hard uint64 `json:"hard,omitempty"`
-	Soft uint64 `json:"soft,omitempty"`
-}

vendor/src/github.com/docker/libcontainer/configs/cgroup.go

@@ -0,0 +1,57 @@
+package configs
+
+type FreezerState string
+
+const (
+	Undefined FreezerState = ""
+	Frozen    FreezerState = "FROZEN"
+	Thawed    FreezerState = "THAWED"
+)
+
+type Cgroup struct {
+	Name string `json:"name"`
+
+	// name of parent cgroup or slice
+	Parent string `json:"parent"`
+
+	// If this is true allow access to any kind of device within the container.  If false, allow access only to devices explicitly listed in the allowed_devices list.
+	AllowAllDevices bool `json:"allow_all_devices"`
+
+	AllowedDevices []*Device `json:"allowed_devices"`
+
+	// Memory limit (in bytes)
+	Memory int64 `json:"memory"`
+
+	// Memory reservation or soft_limit (in bytes)
+	MemoryReservation int64 `json:"memory_reservation"`
+
+	// Total memory usage (memory + swap); set `-1' to disable swap
+	MemorySwap int64 `json:"memory_swap"`
+
+	// CPU shares (relative weight vs. other containers)
+	CpuShares int64 `json:"cpu_shares"`
+
+	// CPU hardcap limit (in usecs). Allowed cpu time in a given period.
+	CpuQuota int64 `json:"cpu_quota"`
+
+	// CPU period to be used for hardcapping (in usecs). 0 to use system default.
+	CpuPeriod int64 `json:"cpu_period"`
+
+	// CPU to use
+	CpusetCpus string `json:"cpuset_cpus"`
+
+	// MEM to use
+	CpusetMems string `json:"cpuset_mems"`
+
+	// Specifies per cgroup weight, range is from 10 to 1000.
+	BlkioWeight int64 `json:"blkio_weight"`
+
+	// set the freeze value for the process
+	Freezer FreezerState `json:"freezer"`
+
+	// Parent slice to use for systemd TODO: remove in favor or parent
+	Slice string `json:"slice"`
+
+	// Whether to disable OOM Killer
+	OomKillDisable bool `json:"oom_kill_disable"`
+}

vendor/src/github.com/docker/libcontainer/configs/config.go

@@ -0,0 +1,145 @@
+package configs
+
+import "fmt"
+
+type Rlimit struct {
+	Type int    `json:"type"`
+	Hard uint64 `json:"hard"`
+	Soft uint64 `json:"soft"`
+}
+
+// IDMap represents UID/GID Mappings for User Namespaces.
+type IDMap struct {
+	ContainerID int `json:"container_id"`
+	HostID      int `json:"host_id"`
+	Size        int `json:"size"`
+}
+
+// Config defines configuration options for executing a process inside a contained environment.
+type Config struct {
+	// NoPivotRoot will use MS_MOVE and a chroot to jail the process into the container's rootfs
+	// This is a common option when the container is running in ramdisk
+	NoPivotRoot bool `json:"no_pivot_root"`
+
+	// ParentDeathSignal specifies the signal that is sent to the container's process in the case
+	// that the parent process dies.
+	ParentDeathSignal int `json:"parent_death_signal"`
+
+	// PivotDir allows a custom directory inside the container's root filesystem to be used as pivot, when NoPivotRoot is not set.
+	// When a custom PivotDir not set, a temporary dir inside the root filesystem will be used. The pivot dir needs to be writeable.
+	// This is required when using read only root filesystems. In these cases, a read/writeable path can be (bind) mounted somewhere inside the root filesystem to act as pivot.
+	PivotDir string `json:"pivot_dir"`
+
+	// Path to a directory containing the container's root filesystem.
+	Rootfs string `json:"rootfs"`
+
+	// Readonlyfs will remount the container's rootfs as readonly where only externally mounted
+	// bind mounts are writtable.
+	Readonlyfs bool `json:"readonlyfs"`
+
+	// Mounts specify additional source and destination paths that will be mounted inside the container's
+	// rootfs and mount namespace if specified
+	Mounts []*Mount `json:"mounts"`
+
+	// The device nodes that should be automatically created within the container upon container start.  Note, make sure that the node is marked as allowed in the cgroup as well!
+	Devices []*Device `json:"devices"`
+
+	MountLabel string `json:"mount_label"`
+
+	// Hostname optionally sets the container's hostname if provided
+	Hostname string `json:"hostname"`
+
+	// Namespaces specifies the container's namespaces that it should setup when cloning the init process
+	// If a namespace is not provided that namespace is shared from the container's parent process
+	Namespaces Namespaces `json:"namespaces"`
+
+	// Capabilities specify the capabilities to keep when executing the process inside the container
+	// All capbilities not specified will be dropped from the processes capability mask
+	Capabilities []string `json:"capabilities"`
+
+	// Networks specifies the container's network setup to be created
+	Networks []*Network `json:"networks"`
+
+	// Routes can be specified to create entries in the route table as the container is started
+	Routes []*Route `json:"routes"`
+
+	// Cgroups specifies specific cgroup settings for the various subsystems that the container is
+	// placed into to limit the resources the container has available
+	Cgroups *Cgroup `json:"cgroups"`
+
+	// AppArmorProfile specifies the profile to apply to the process running in the container and is
+	// change at the time the process is execed
+	AppArmorProfile string `json:"apparmor_profile"`
+
+	// ProcessLabel specifies the label to apply to the process running in the container.  It is
+	// commonly used by selinux
+	ProcessLabel string `json:"process_label"`
+
+	// Rlimits specifies the resource limits, such as max open files, to set in the container
+	// If Rlimits are not set, the container will inherit rlimits from the parent process
+	Rlimits []Rlimit `json:"rlimits"`
+
+	// AdditionalGroups specifies the gids that should be added to supplementary groups
+	// in addition to those that the user belongs to.
+	AdditionalGroups []int `json:"additional_groups"`
+
+	// UidMappings is an array of User ID mappings for User Namespaces
+	UidMappings []IDMap `json:"uid_mappings"`
+
+	// GidMappings is an array of Group ID mappings for User Namespaces
+	GidMappings []IDMap `json:"gid_mappings"`
+
+	// MaskPaths specifies paths within the container's rootfs to mask over with a bind
+	// mount pointing to /dev/null as to prevent reads of the file.
+	MaskPaths []string `json:"mask_paths"`
+
+	// ReadonlyPaths specifies paths within the container's rootfs to remount as read-only
+	// so that these files prevent any writes.
+	ReadonlyPaths []string `json:"readonly_paths"`
+}
+
+// Gets the root uid for the process on host which could be non-zero
+// when user namespaces are enabled.
+func (c Config) HostUID() (int, error) {
+	if c.Namespaces.Contains(NEWUSER) {
+		if c.UidMappings == nil {
+			return -1, fmt.Errorf("User namespaces enabled, but no user mappings found.")
+		}
+		id, found := c.hostIDFromMapping(0, c.UidMappings)
+		if !found {
+			return -1, fmt.Errorf("User namespaces enabled, but no root user mapping found.")
+		}
+		return id, nil
+	}
+	// Return default root uid 0
+	return 0, nil
+}
+
+// Gets the root uid for the process on host which could be non-zero
+// when user namespaces are enabled.
+func (c Config) HostGID() (int, error) {
+	if c.Namespaces.Contains(NEWUSER) {
+		if c.GidMappings == nil {
+			return -1, fmt.Errorf("User namespaces enabled, but no gid mappings found.")
+		}
+		id, found := c.hostIDFromMapping(0, c.GidMappings)
+		if !found {
+			return -1, fmt.Errorf("User namespaces enabled, but no root user mapping found.")
+		}
+		return id, nil
+	}
+	// Return default root uid 0
+	return 0, nil
+}
+
+// Utility function that gets a host ID for a container ID from user namespace map
+// if that ID is present in the map.
+func (c Config) hostIDFromMapping(containerID int, uMap []IDMap) (int, bool) {
+	for _, m := range uMap {
+		if (containerID >= m.ContainerID) && (containerID <= (m.ContainerID + m.Size - 1)) {
+			hostID := m.HostID + (containerID - m.ContainerID)
+			return hostID, true
+		}
+	}
+	return -1, false
+}

vendor/src/github.com/docker/libcontainer/configs/config_test.go

@@ -1,12 +1,11 @@
-package libcontainer
+package configs
 
 import (
 	"encoding/json"
+	"fmt"
 	"os"
 	"path/filepath"
 	"testing"
-
-	"github.com/docker/libcontainer/devices"
 )
 
 // Checks whether the expected capability is specified in the capabilities.
@@ -19,13 +18,13 @@ func contains(expected string, values []string) bool {
 	return false
 }
 
-func containsDevice(expected *devices.Device, values []*devices.Device) bool {
+func containsDevice(expected *Device, values []*Device) bool {
 	for _, d := range values {
 		if d.Path == expected.Path &&
-			d.CgroupPermissions == expected.CgroupPermissions &&
+			d.Permissions == expected.Permissions &&
 			d.FileMode == expected.FileMode &&
-			d.MajorNumber == expected.MajorNumber &&
-			d.MinorNumber == expected.MinorNumber &&
+			d.Major == expected.Major &&
+			d.Minor == expected.Minor &&
 			d.Type == expected.Type {
 			return true
 		}
@@ -34,7 +33,7 @@ func containsDevice(expected *devices.Device, values []*devices.Device) bool {
 }
 
 func loadConfig(name string) (*Config, error) {
-	f, err := os.Open(filepath.Join("sample_configs", name))
+	f, err := os.Open(filepath.Join("../sample_configs", name))
 	if err != nil {
 		return nil, err
 	}
@@ -45,6 +44,34 @@ func loadConfig(name string) (*Config, error) {
 		return nil, err
 	}
 
+	// Check that a config doesn't contain extra fields
+	var configMap, abstractMap map[string]interface{}
+
+	if _, err := f.Seek(0, 0); err != nil {
+		return nil, err
+	}
+
+	if err := json.NewDecoder(f).Decode(&abstractMap); err != nil {
+		return nil, err
+	}
+
+	configData, err := json.Marshal(&container)
+	if err != nil {
+		return nil, err
+	}
+
+	if err := json.Unmarshal(configData, &configMap); err != nil {
+		return nil, err
+	}
+
+	for k := range configMap {
+		delete(abstractMap, k)
+	}
+
+	if len(abstractMap) != 0 {
+		return nil, fmt.Errorf("unknown fields: %s", abstractMap)
+	}
+
 	return container, nil
 }
 
@@ -59,11 +86,6 @@ func TestConfigJsonFormat(t *testing.T) {
 		t.Fail()
 	}
 
-	if !container.Tty {
-		t.Log("tty should be set to true")
-		t.Fail()
-	}
-
 	if !container.Namespaces.Contains(NEWNET) {
 		t.Log("namespaces should contain NEWNET")
 		t.Fail()
@@ -101,11 +123,6 @@ func TestConfigJsonFormat(t *testing.T) {
 				t.Fail()
 			}
 
-			if n.VethPrefix != "veth" {
-				t.Logf("veth prefix should be veth but received %q", n.VethPrefix)
-				t.Fail()
-			}
-
 			if n.Gateway != "172.17.42.1" {
 				t.Logf("veth gateway should be 172.17.42.1 but received %q", n.Gateway)
 				t.Fail()
@@ -119,18 +136,12 @@ func TestConfigJsonFormat(t *testing.T) {
 			break
 		}
 	}
-
-	for _, d := range devices.DefaultSimpleDevices {
-		if !containsDevice(d, container.MountConfig.DeviceNodes) {
+	for _, d := range DefaultSimpleDevices {
+		if !containsDevice(d, container.Devices) {
 			t.Logf("expected device configuration for %s", d.Path)
 			t.Fail()
 		}
 	}
-
-	if !container.RestrictSys {
-		t.Log("expected restrict sys to be true")
-		t.Fail()
-	}
 }
 
 func TestApparmorProfile(t *testing.T) {
@@ -154,8 +165,8 @@ func TestSelinuxLabels(t *testing.T) {
 	if container.ProcessLabel != label {
 		t.Fatalf("expected process label %q but received %q", label, container.ProcessLabel)
 	}
-	if container.MountConfig.MountLabel != label {
-		t.Fatalf("expected mount label %q but received %q", label, container.MountConfig.MountLabel)
+	if container.MountLabel != label {
+		t.Fatalf("expected mount label %q but received %q", label, container.MountLabel)
 	}
 }
 
@@ -170,3 +181,69 @@ func TestRemoveNamespace(t *testing.T) {
 		t.Fatalf("namespaces should have 0 items but reports %d", len(ns))
 	}
 }
+
+func TestHostUIDNoUSERNS(t *testing.T) {
+	config := &Config{
+		Namespaces: Namespaces{},
+	}
+	uid, err := config.HostUID()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if uid != 0 {
+		t.Fatalf("expected uid 0 with no USERNS but received %d", uid)
+	}
+}
+
+func TestHostUIDWithUSERNS(t *testing.T) {
+	config := &Config{
+		Namespaces: Namespaces{{Type: NEWUSER}},
+		UidMappings: []IDMap{
+			{
+				ContainerID: 0,
+				HostID:      1000,
+				Size:        1,
+			},
+		},
+	}
+	uid, err := config.HostUID()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if uid != 1000 {
+		t.Fatalf("expected uid 1000 with no USERNS but received %d", uid)
+	}
+}
+
+func TestHostGIDNoUSERNS(t *testing.T) {
+	config := &Config{
+		Namespaces: Namespaces{},
+	}
+	uid, err := config.HostGID()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if uid != 0 {
+		t.Fatalf("expected gid 0 with no USERNS but received %d", uid)
+	}
+}
+
+func TestHostGIDWithUSERNS(t *testing.T) {
+	config := &Config{
+		Namespaces: Namespaces{{Type: NEWUSER}},
+		GidMappings: []IDMap{
+			{
+				ContainerID: 0,
+				HostID:      1000,
+				Size:        1,
+			},
+		},
+	}
+	uid, err := config.HostGID()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if uid != 1000 {
+		t.Fatalf("expected gid 1000 with no USERNS but received %d", uid)
+	}
+}

vendor/src/github.com/docker/libcontainer/configs/device.go

@@ -0,0 +1,52 @@
+package configs
+
+import (
+	"fmt"
+	"os"
+)
+
+const (
+	Wildcard = -1
+)
+
+type Device struct {
+	// Device type, block, char, etc.
+	Type rune `json:"type"`
+
+	// Path to the device.
+	Path string `json:"path"`
+
+	// Major is the device's major number.
+	Major int64 `json:"major"`
+
+	// Minor is the device's minor number.
+	Minor int64 `json:"minor"`
+
+	// Cgroup permissions format, rwm.
+	Permissions string `json:"permissions"`
+
+	// FileMode permission bits for the device.
+	FileMode os.FileMode `json:"file_mode"`
+
+	// Uid of the device.
+	Uid uint32 `json:"uid"`
+
+	// Gid of the device.
+	Gid uint32 `json:"gid"`
+}
+
+func (d *Device) CgroupString() string {
+	return fmt.Sprintf("%c %s:%s %s", d.Type, deviceNumberString(d.Major), deviceNumberString(d.Minor), d.Permissions)
+}
+
+func (d *Device) Mkdev() int {
+	return int((d.Major << 8) | (d.Minor & 0xff) | ((d.Minor & 0xfff00) << 12))
+}
+
+// deviceNumberString converts the device number to a string return result.
+func deviceNumberString(number int64) string {
+	if number == Wildcard {
+		return "*"
+	}
+	return fmt.Sprint(number)
+}

vendor/src/github.com/docker/libcontainer/configs/device_defaults.go

@@ -0,0 +1,137 @@
+package configs
+
+var (
+	// These are devices that are to be both allowed and created.
+	DefaultSimpleDevices = []*Device{
+		// /dev/null and zero
+		{
+			Path:        "/dev/null",
+			Type:        'c',
+			Major:       1,
+			Minor:       3,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+		{
+			Path:        "/dev/zero",
+			Type:        'c',
+			Major:       1,
+			Minor:       5,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+
+		{
+			Path:        "/dev/full",
+			Type:        'c',
+			Major:       1,
+			Minor:       7,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+
+		// consoles and ttys
+		{
+			Path:        "/dev/tty",
+			Type:        'c',
+			Major:       5,
+			Minor:       0,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+
+		// /dev/urandom,/dev/random
+		{
+			Path:        "/dev/urandom",
+			Type:        'c',
+			Major:       1,
+			Minor:       9,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+		{
+			Path:        "/dev/random",
+			Type:        'c',
+			Major:       1,
+			Minor:       8,
+			Permissions: "rwm",
+			FileMode:    0666,
+		},
+	}
+	DefaultAllowedDevices = append([]*Device{
+		// allow mknod for any device
+		{
+			Type:        'c',
+			Major:       Wildcard,
+			Minor:       Wildcard,
+			Permissions: "m",
+		},
+		{
+			Type:        'b',
+			Major:       Wildcard,
+			Minor:       Wildcard,
+			Permissions: "m",
+		},
+
+		{
+			Path:        "/dev/console",
+			Type:        'c',
+			Major:       5,
+			Minor:       1,
+			Permissions: "rwm",
+		},
+		{
+			Path:        "/dev/tty0",
+			Type:        'c',
+			Major:       4,
+			Minor:       0,
+			Permissions: "rwm",
+		},
+		{
+			Path:        "/dev/tty1",
+			Type:        'c',
+			Major:       4,
+			Minor:       1,
+			Permissions: "rwm",
+		},
+		// /dev/pts/ - pts namespaces are "coming soon"
+		{
+			Path:        "",
+			Type:        'c',
+			Major:       136,
+			Minor:       Wildcard,
+			Permissions: "rwm",
+		},
+		{
+			Path:        "",
+			Type:        'c',
+			Major:       5,
+			Minor:       2,
+			Permissions: "rwm",
+		},
+
+		// tuntap
+		{
+			Path:        "",
+			Type:        'c',
+			Major:       10,
+			Minor:       200,
+			Permissions: "rwm",
+		},
+	}, DefaultSimpleDevices...)
+	DefaultAutoCreatedDevices = append([]*Device{
+		{
+			// /dev/fuse is created but not allowed.
+			// This is to allow java to work.  Because java
+			// Insists on there being a /dev/fuse
+			// https://github.com/docker/docker/issues/514
+			// https://github.com/docker/docker/issues/2393
+			//
+			Path:        "/dev/fuse",
+			Type:        'c',
+			Major:       10,
+			Minor:       229,
+			Permissions: "rwm",
+		},
+	}, DefaultSimpleDevices...)
+)

vendor/src/github.com/docker/libcontainer/configs/mount.go

@@ -0,0 +1,21 @@
+package configs
+
+type Mount struct {
+	// Source path for the mount.
+	Source string `json:"source"`
+
+	// Destination path for the mount inside the container.
+	Destination string `json:"destination"`
+
+	// Device the mount is for.
+	Device string `json:"device"`
+
+	// Mount flags.
+	Flags int `json:"flags"`
+
+	// Mount data applied to the mount.
+	Data string `json:"data"`
+
+	// Relabel source if set, "z" indicates shared, "Z" indicates unshared.
+	Relabel string `json:"relabel"`
+}

vendor/src/github.com/docker/libcontainer/configs/namespaces.go

@@ -0,0 +1,109 @@
+package configs
+
+import (
+	"fmt"
+	"syscall"
+)
+
+type NamespaceType string
+
+const (
+	NEWNET  NamespaceType = "NEWNET"
+	NEWPID  NamespaceType = "NEWPID"
+	NEWNS   NamespaceType = "NEWNS"
+	NEWUTS  NamespaceType = "NEWUTS"
+	NEWIPC  NamespaceType = "NEWIPC"
+	NEWUSER NamespaceType = "NEWUSER"
+)
+
+// Namespace defines configuration for each namespace.  It specifies an
+// alternate path that is able to be joined via setns.
+type Namespace struct {
+	Type NamespaceType `json:"type"`
+	Path string        `json:"path"`
+}
+
+func (n *Namespace) Syscall() int {
+	return namespaceInfo[n.Type]
+}
+
+func (n *Namespace) GetPath(pid int) string {
+	if n.Path != "" {
+		return n.Path
+	}
+	return fmt.Sprintf("/proc/%d/ns/%s", pid, n.file())
+}
+
+func (n *Namespace) file() string {
+	file := ""
+	switch n.Type {
+	case NEWNET:
+		file = "net"
+	case NEWNS:
+		file = "mnt"
+	case NEWPID:
+		file = "pid"
+	case NEWIPC:
+		file = "ipc"
+	case NEWUSER:
+		file = "user"
+	case NEWUTS:
+		file = "uts"
+	}
+	return file
+}
+
+type Namespaces []Namespace
+
+func (n *Namespaces) Remove(t NamespaceType) bool {
+	i := n.index(t)
+	if i == -1 {
+		return false
+	}
+	*n = append((*n)[:i], (*n)[i+1:]...)
+	return true
+}
+
+func (n *Namespaces) Add(t NamespaceType, path string) {
+	i := n.index(t)
+	if i == -1 {
+		*n = append(*n, Namespace{Type: t, Path: path})
+		return
+	}
+	(*n)[i].Path = path
+}
+
+func (n *Namespaces) index(t NamespaceType) int {
+	for i, ns := range *n {
+		if ns.Type == t {
+			return i
+		}
+	}
+	return -1
+}
+
+func (n *Namespaces) Contains(t NamespaceType) bool {
+	return n.index(t) != -1
+}
+
+var namespaceInfo = map[NamespaceType]int{
+	NEWNET:  syscall.CLONE_NEWNET,
+	NEWNS:   syscall.CLONE_NEWNS,
+	NEWUSER: syscall.CLONE_NEWUSER,
+	NEWIPC:  syscall.CLONE_NEWIPC,
+	NEWUTS:  syscall.CLONE_NEWUTS,
+	NEWPID:  syscall.CLONE_NEWPID,
+}
+
+// CloneFlags parses the container's Namespaces options to set the correct
+// flags on clone, unshare. This functions returns flags only for new namespaces.
+func (n *Namespaces) CloneFlags() uintptr {
+	var flag int
+	for _, v := range *n {
+		if v.Path != "" {
+			continue
+		}
+		flag |= namespaceInfo[v.Type]
+	}
+	return uintptr(flag)
+}

vendor/src/github.com/docker/libcontainer/configs/network.go

@@ -0,0 +1,66 @@
+package configs
+
+// Network defines configuration for a container's networking stack
+//
+// The network configuration can be omited from a container causing the
+// container to be setup with the host's networking stack
+type Network struct {
+	// Type sets the networks type, commonly veth and loopback
+	Type string `json:"type"`
+
+	// Name of the network interface
+	Name string `json:"name"`
+
+	// The bridge to use.
+	Bridge string `json:"bridge"`
+
+	// MacAddress contains the MAC address to set on the network interface
+	MacAddress string `json:"mac_address"`
+
+	// Address contains the IPv4 and mask to set on the network interface
+	Address string `json:"address"`
+
+	// Gateway sets the gateway address that is used as the default for the interface
+	Gateway string `json:"gateway"`
+
+	// IPv6Address contains the IPv6 and mask to set on the network interface
+	IPv6Address string `json:"ipv6_address"`
+
+	// IPv6Gateway sets the ipv6 gateway address that is used as the default for the interface
+	IPv6Gateway string `json:"ipv6_gateway"`
+
+	// Mtu sets the mtu value for the interface and will be mirrored on both the host and
+	// container's interfaces if a pair is created, specifically in the case of type veth
+	// Note: This does not apply to loopback interfaces.
+	Mtu int `json:"mtu"`
+
+	// TxQueueLen sets the tx_queuelen value for the interface and will be mirrored on both the host and
+	// container's interfaces if a pair is created, specifically in the case of type veth
+	// Note: This does not apply to loopback interfaces.
+	TxQueueLen int `json:"txqueuelen"`
+
+	// HostInterfaceName is a unique name of a veth pair that resides on in the host interface of the
+	// container.
+	HostInterfaceName string `json:"host_interface_name"`
+}
+
+// Routes can be specified to create entries in the route table as the container is started
+//
+// All of destination, source, and gateway should be either IPv4 or IPv6.
+// One of the three options must be present, and ommitted entries will use their
+// IP family default for the route table.  For IPv4 for example, setting the
+// gateway to 1.2.3.4 and the interface to eth0 will set up a standard
+// destination of 0.0.0.0(or *) when viewed in the route table.
+type Route struct {
+	// Sets the destination and mask, should be a CIDR.  Accepts IPv4 and IPv6
+	Destination string `json:"destination"`
+
+	// Sets the source and mask, should be a CIDR.  Accepts IPv4 and IPv6
+	Source string `json:"source"`
+
+	// Sets the gateway.  Accepts IPv4 and IPv6
+	Gateway string `json:"gateway"`
+
+	// The device to set this route up for, for example: eth0
+	InterfaceName string `json:"interface_name"`
+}

vendor/src/github.com/docker/libcontainer/configs/validate/config.go

@@ -0,0 +1,93 @@
+package validate
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+
+	"github.com/docker/libcontainer/configs"
+)
+
+type Validator interface {
+	Validate(*configs.Config) error
+}
+
+func New() Validator {
+	return &ConfigValidator{}
+}
+
+type ConfigValidator struct {
+}
+
+func (v *ConfigValidator) Validate(config *configs.Config) error {
+	if err := v.rootfs(config); err != nil {
+		return err
+	}
+	if err := v.network(config); err != nil {
+		return err
+	}
+	if err := v.hostname(config); err != nil {
+		return err
+	}
+	if err := v.security(config); err != nil {
+		return err
+	}
+	if err := v.usernamespace(config); err != nil {
+		return err
+	}
+	return nil
+}
+
+// rootfs validates the the rootfs is an absolute path and is not a symlink
+// to the container's root filesystem.
+func (v *ConfigValidator) rootfs(config *configs.Config) error {
+	cleaned, err := filepath.Abs(config.Rootfs)
+	if err != nil {
+		return err
+	}
+	if cleaned, err = filepath.EvalSymlinks(cleaned); err != nil {
+		return err
+	}
+	if config.Rootfs != cleaned {
+		return fmt.Errorf("%s is not an absolute path or is a symlink", config.Rootfs)
+	}
+	return nil
+}
+
+func (v *ConfigValidator) network(config *configs.Config) error {
+	if !config.Namespaces.Contains(configs.NEWNET) {
+		if len(config.Networks) > 0 || len(config.Routes) > 0 {
+			return fmt.Errorf("unable to apply network settings without a private NET namespace")
+		}
+	}
+	return nil
+}
+
+func (v *ConfigValidator) hostname(config *configs.Config) error {
+	if config.Hostname != "" && !config.Namespaces.Contains(configs.NEWUTS) {
+		return fmt.Errorf("unable to set hostname without a private UTS namespace")
+	}
+	return nil
+}
+
+func (v *ConfigValidator) security(config *configs.Config) error {
+	// restrict sys without mount namespace
+	if (len(config.MaskPaths) > 0 || len(config.ReadonlyPaths) > 0) &&
+		!config.Namespaces.Contains(configs.NEWNS) {
+		return fmt.Errorf("unable to restrict sys entries without a private MNT namespace")
+	}
+	return nil
+}
+
+func (v *ConfigValidator) usernamespace(config *configs.Config) error {
+	if config.Namespaces.Contains(configs.NEWUSER) {
+		if _, err := os.Stat("/proc/self/ns/user"); os.IsNotExist(err) {
+			return fmt.Errorf("USER namespaces aren't enabled in the kernel")
+		}
+	} else {
+		if config.UidMappings != nil || config.GidMappings != nil {
+			return fmt.Errorf("User namespace mappings specified, but USER namespace isn't enabled in the config")
+		}
+	}
+	return nil
+}

vendor/src/github.com/docker/libcontainer/console.go

@@ -0,0 +1,15 @@
+package libcontainer
+
+import "io"
+
+// Console represents a pseudo TTY.
+type Console interface {
+	io.ReadWriter
+	io.Closer
+
+	// Path returns the filesystem path to the slave side of the pty.
+	Path() string
+
+	// Fd returns the fd for the master of the pty.
+	Fd() uintptr
+}

vendor/src/github.com/docker/libcontainer/console/console.go

@@ -1,128 +0,0 @@
-// +build linux
-
-package console
-
-import (
-	"fmt"
-	"os"
-	"path/filepath"
-	"syscall"
-	"unsafe"
-
-	"github.com/docker/libcontainer/label"
-)
-
-// Setup initializes the proper /dev/console inside the rootfs path
-func Setup(rootfs, consolePath, mountLabel string) error {
-	oldMask := syscall.Umask(0000)
-	defer syscall.Umask(oldMask)
-
-	if err := os.Chmod(consolePath, 0600); err != nil {
-		return err
-	}
-
-	if err := os.Chown(consolePath, 0, 0); err != nil {
-		return err
-	}
-
-	if err := label.SetFileLabel(consolePath, mountLabel); err != nil {
-		return fmt.Errorf("set file label %s %s", consolePath, err)
-	}
-
-	dest := filepath.Join(rootfs, "dev/console")
-
-	f, err := os.Create(dest)
-	if err != nil && !os.IsExist(err) {
-		return fmt.Errorf("create %s %s", dest, err)
-	}
-
-	if f != nil {
-		f.Close()
-	}
-
-	if err := syscall.Mount(consolePath, dest, "bind", syscall.MS_BIND, ""); err != nil {
-		return fmt.Errorf("bind %s to %s %s", consolePath, dest, err)
-	}
-
-	return nil
-}
-
-func OpenAndDup(consolePath string) error {
-	slave, err := OpenTerminal(consolePath, syscall.O_RDWR)
-	if err != nil {
-		return fmt.Errorf("open terminal %s", err)
-	}
-
-	if err := syscall.Dup2(int(slave.Fd()), 0); err != nil {
-		return err
-	}
-
-	if err := syscall.Dup2(int(slave.Fd()), 1); err != nil {
-		return err
-	}
-
-	return syscall.Dup2(int(slave.Fd()), 2)
-}
-
-// Unlockpt unlocks the slave pseudoterminal device corresponding to the master pseudoterminal referred to by f.
-// Unlockpt should be called before opening the slave side of a pseudoterminal.
-func Unlockpt(f *os.File) error {
-	var u int32
-
-	return Ioctl(f.Fd(), syscall.TIOCSPTLCK, uintptr(unsafe.Pointer(&u)))
-}
-
-// Ptsname retrieves the name of the first available pts for the given master.
-func Ptsname(f *os.File) (string, error) {
-	var n int32
-
-	if err := Ioctl(f.Fd(), syscall.TIOCGPTN, uintptr(unsafe.Pointer(&n))); err != nil {
-		return "", err
-	}
-
-	return fmt.Sprintf("/dev/pts/%d", n), nil
-}
-
-// CreateMasterAndConsole will open /dev/ptmx on the host and retreive the
-// pts name for use as the pty slave inside the container
-func CreateMasterAndConsole() (*os.File, string, error) {
-	master, err := os.OpenFile("/dev/ptmx", syscall.O_RDWR|syscall.O_NOCTTY|syscall.O_CLOEXEC, 0)
-	if err != nil {
-		return nil, "", err
-	}
-
-	console, err := Ptsname(master)
-	if err != nil {
-		return nil, "", err
-	}
-
-	if err := Unlockpt(master); err != nil {
-		return nil, "", err
-	}
-
-	return master, console, nil
-}
-
-// OpenPtmx opens /dev/ptmx, i.e. the PTY master.
-func OpenPtmx() (*os.File, error) {
-	// O_NOCTTY and O_CLOEXEC are not present in os package so we use the syscall's one for all.
-	return os.OpenFile("/dev/ptmx", syscall.O_RDONLY|syscall.O_NOCTTY|syscall.O_CLOEXEC, 0)
-}
-
-// OpenTerminal is a clone of os.OpenFile without the O_CLOEXEC
-// used to open the pty slave inside the container namespace
-func OpenTerminal(name string, flag int) (*os.File, error) {
-	r, e := syscall.Open(name, flag, 0)
-	if e != nil {
-		return nil, &os.PathError{Op: "open", Path: name, Err: e}
-	}
-	return os.NewFile(uintptr(r), name), nil
-}
-
-func Ioctl(fd uintptr, flag, data uintptr) error {
-	if _, _, err := syscall.Syscall(syscall.SYS_IOCTL, fd, flag, data); err != 0 {
-		return err
-	}
-
-	return nil
-}

vendor/src/github.com/docker/libcontainer/console_linux.go

@@ -0,0 +1,147 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"syscall"
+	"unsafe"
+
+	"github.com/docker/libcontainer/label"
+)
+
+// newConsole returns an initalized console that can be used within a container by copying bytes
+// from the master side to the slave that is attached as the tty for the container's init process.
+func newConsole(uid, gid int) (Console, error) {
+	master, err := os.OpenFile("/dev/ptmx", syscall.O_RDWR|syscall.O_NOCTTY|syscall.O_CLOEXEC, 0)
+	if err != nil {
+		return nil, err
+	}
+	console, err := ptsname(master)
+	if err != nil {
+		return nil, err
+	}
+	if err := unlockpt(master); err != nil {
+		return nil, err
+	}
+	if err := os.Chmod(console, 0600); err != nil {
+		return nil, err
+	}
+	if err := os.Chown(console, uid, gid); err != nil {
+		return nil, err
+	}
+	return &linuxConsole{
+		slavePath: console,
+		master:    master,
+	}, nil
+}
+
+// newConsoleFromPath is an internal fucntion returning an initialzied console for use inside
+// a container's MNT namespace.
+func newConsoleFromPath(slavePath string) *linuxConsole {
+	return &linuxConsole{
+		slavePath: slavePath,
+	}
+}
+
+// linuxConsole is a linux psuedo TTY for use within a container.
+type linuxConsole struct {
+	master    *os.File
+	slavePath string
+}
+
+func (c *linuxConsole) Fd() uintptr {
+	return c.master.Fd()
+}
+
+func (c *linuxConsole) Path() string {
+	return c.slavePath
+}
+
+func (c *linuxConsole) Read(b []byte) (int, error) {
+	return c.master.Read(b)
+}
+
+func (c *linuxConsole) Write(b []byte) (int, error) {
+	return c.master.Write(b)
+}
+
+func (c *linuxConsole) Close() error {
+	if m := c.master; m != nil {
+		return m.Close()
+	}
+	return nil
+}
+
+// mount initializes the console inside the rootfs mounting with the specified mount label
+// and applying the correct ownership of the console.
+func (c *linuxConsole) mount(rootfs, mountLabel string, uid, gid int) error {
+	oldMask := syscall.Umask(0000)
+	defer syscall.Umask(oldMask)
+	if err := label.SetFileLabel(c.slavePath, mountLabel); err != nil {
+		return err
+	}
+	dest := filepath.Join(rootfs, "/dev/console")
+	f, err := os.Create(dest)
+	if err != nil && !os.IsExist(err) {
+		return err
+	}
+	if f != nil {
+		f.Close()
+	}
+	return syscall.Mount(c.slavePath, dest, "bind", syscall.MS_BIND, "")
+}
+
+// dupStdio opens the slavePath for the console and dup2s the fds to the current
+// processes stdio, fd 0,1,2.
+func (c *linuxConsole) dupStdio() error {
+	slave, err := c.open(syscall.O_RDWR)
+	if err != nil {
+		return err
+	}
+	fd := int(slave.Fd())
+	for _, i := range []int{0, 1, 2} {
+		if err := syscall.Dup2(fd, i); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// open is a clone of os.OpenFile without the O_CLOEXEC used to open the pty slave.
+func (c *linuxConsole) open(flag int) (*os.File, error) {
+	r, e := syscall.Open(c.slavePath, flag, 0)
+	if e != nil {
+		return nil, &os.PathError{
+			Op:   "open",
+			Path: c.slavePath,
+			Err:  e,
+		}
+	}
+	return os.NewFile(uintptr(r), c.slavePath), nil
+}
+
+func ioctl(fd uintptr, flag, data uintptr) error {
+	if _, _, err := syscall.Syscall(syscall.SYS_IOCTL, fd, flag, data); err != 0 {
+		return err
+	}
+	return nil
+}
+
+// unlockpt unlocks the slave pseudoterminal device corresponding to the master pseudoterminal referred to by f.
+// unlockpt should be called before opening the slave side of a pty.
+func unlockpt(f *os.File) error {
+	var u int32
+	return ioctl(f.Fd(), syscall.TIOCSPTLCK, uintptr(unsafe.Pointer(&u)))
+}
+
+// ptsname retrieves the name of the first available pts for the given master.
+func ptsname(f *os.File) (string, error) {
+	var n int32
+	if err := ioctl(f.Fd(), syscall.TIOCGPTN, uintptr(unsafe.Pointer(&n))); err != nil {
+		return "", err
+	}
+	return fmt.Sprintf("/dev/pts/%d", n), nil
+}

vendor/src/github.com/docker/libcontainer/container.go

@@ -1,8 +1,53 @@
-/*
-NOTE: The API is in flux and mainly not implemented. Proceed with caution until further notice.
-*/
+// Libcontainer provides a native Go implementation for creating containers
+// with namespaces, cgroups, capabilities, and filesystem access controls.
+// It allows you to manage the lifecycle of the container performing additional operations
+// after the container is created.
 package libcontainer
 
+import (
+	"github.com/docker/libcontainer/configs"
+)
+
+// The status of a container.
+type Status int
+
+const (
+	// The container exists and is running.
+	Running Status = iota + 1
+
+	// The container exists, it is in the process of being paused.
+	Pausing
+
+	// The container exists, but all its processes are paused.
+	Paused
+
+	// The container does not exist.
+	Destroyed
+)
+
+// State represents a running container's state
+type State struct {
+	// ID is the container ID.
+	ID string `json:"id"`
+
+	// InitProcessPid is the init process id in the parent namespace.
+	InitProcessPid int `json:"init_process_pid"`
+
+	// InitProcessStartTime is the init process start time.
+	InitProcessStartTime string `json:"init_process_start"`
+
+	// Path to all the cgroups setup for a container. Key is cgroup subsystem name
+	// with the value as the path.
+	CgroupPaths map[string]string `json:"cgroup_paths"`
+
+	// NamespacePaths are filepaths to the container's namespaces. Key is the namespace type
+	// with the value as the path.
+	NamespacePaths map[configs.NamespaceType]string `json:"namespace_paths"`
+
+	// Config is the container's configuration.
+	Config configs.Config `json:"config"`
+}
+
 // A libcontainer container object.
 //
 // Each container is thread-safe within the same process. Since a container can
@@ -12,67 +57,88 @@ type Container interface {
 	// Returns the ID of the container
 	ID() string
 
-	// Returns the current run state of the container.
+	// Returns the current status of the container.
 	//
-	// Errors:
+	// errors:
 	// ContainerDestroyed - Container no longer exists,
-	// SystemError - System error.
-	RunState() (*RunState, Error)
+	// Systemerror - System error.
+	Status() (Status, error)
+
+	// State returns the current container's state information.
+	//
+	// errors:
+	// Systemerror - System erroor.
+	State() (*State, error)
 
 	// Returns the current config of the container.
-	Config() *Config
+	Config() configs.Config
 
-	// Start a process inside the container. Returns the PID of the new process (in the caller process's namespace) and a channel that will return the exit status of the process whenever it dies.
+	// Returns the PIDs inside this container. The PIDs are in the namespace of the calling process.
 	//
-	// Errors:
+	// errors:
+	// ContainerDestroyed - Container no longer exists,
+	// Systemerror - System error.
+	//
+	// Some of the returned PIDs may no longer refer to processes in the Container, unless
+	// the Container state is PAUSED in which case every PID in the slice is valid.
+	Processes() ([]int, error)
+
+	// Returns statistics for the container.
+	//
+	// errors:
+	// ContainerDestroyed - Container no longer exists,
+	// Systemerror - System error.
+	Stats() (*Stats, error)
+
+	// Set cgroup resources of container as configured
+	//
+	// We can use this to change resources when containers are running.
+	//
+	// errors:
+	// Systemerror - System error.
+	Set() error
+
+	// Start a process inside the container. Returns error if process fails to
+	// start. You can track process lifecycle with passed Process structure.
+	//
+	// errors:
 	// ContainerDestroyed - Container no longer exists,
 	// ConfigInvalid - config is invalid,
 	// ContainerPaused - Container is paused,
-	// SystemError - System error.
-	Start(config *ProcessConfig) (pid int, exitChan chan int, err Error)
+	// Systemerror - System error.
+	Start(process *Process) (err error)
 
 	// Destroys the container after killing all running processes.
 	//
 	// Any event registrations are removed before the container is destroyed.
 	// No error is returned if the container is already destroyed.
 	//
-	// Errors:
-	// SystemError - System error.
-	Destroy() Error
-
-	// Returns the PIDs inside this container. The PIDs are in the namespace of the calling process.
-	//
-	// Errors:
-	// ContainerDestroyed - Container no longer exists,
-	// SystemError - System error.
-	//
-	// Some of the returned PIDs may no longer refer to processes in the Container, unless
-	// the Container state is PAUSED in which case every PID in the slice is valid.
-	Processes() ([]int, Error)
-
-	// Returns statistics for the container.
-	//
-	// Errors:
-	// ContainerDestroyed - Container no longer exists,
-	// SystemError - System error.
-	Stats() (*ContainerStats, Error)
+	// errors:
+	// Systemerror - System error.
+	Destroy() error
 
 	// If the Container state is RUNNING or PAUSING, sets the Container state to PAUSING and pauses
 	// the execution of any user processes. Asynchronously, when the container finished being paused the
 	// state is changed to PAUSED.
 	// If the Container state is PAUSED, do nothing.
 	//
-	// Errors:
+	// errors:
 	// ContainerDestroyed - Container no longer exists,
-	// SystemError - System error.
-	Pause() Error
+	// Systemerror - System error.
+	Pause() error
 
 	// If the Container state is PAUSED, resumes the execution of any user processes in the
 	// Container before setting the Container state to RUNNING.
 	// If the Container state is RUNNING, do nothing.
 	//
-	// Errors:
+	// errors:
 	// ContainerDestroyed - Container no longer exists,
-	// SystemError - System error.
-	Resume() Error
+	// Systemerror - System error.
+	Resume() error
+
+	// NotifyOOM returns a read-only channel signaling when the container receives an OOM notification.
+	//
+	// errors:
+	// Systemerror - System error.
+	NotifyOOM() (<-chan struct{}, error)
 }

vendor/src/github.com/docker/libcontainer/container_linux.go

@@ -0,0 +1,306 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"sync"
+	"syscall"
+
+	log "github.com/Sirupsen/logrus"
+	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
+)
+
+type linuxContainer struct {
+	id            string
+	root          string
+	config        *configs.Config
+	cgroupManager cgroups.Manager
+	initPath      string
+	initArgs      []string
+	initProcess   parentProcess
+	m             sync.Mutex
+}
+
+// ID returns the container's unique ID
+func (c *linuxContainer) ID() string {
+	return c.id
+}
+
+// Config returns the container's configuration
+func (c *linuxContainer) Config() configs.Config {
+	return *c.config
+}
+
+func (c *linuxContainer) Status() (Status, error) {
+	c.m.Lock()
+	defer c.m.Unlock()
+	return c.currentStatus()
+}
+
+func (c *linuxContainer) State() (*State, error) {
+	c.m.Lock()
+	defer c.m.Unlock()
+	return c.currentState()
+}
+
+func (c *linuxContainer) Processes() ([]int, error) {
+	pids, err := c.cgroupManager.GetPids()
+	if err != nil {
+		return nil, newSystemError(err)
+	}
+	return pids, nil
+}
+
+func (c *linuxContainer) Stats() (*Stats, error) {
+	var (
+		err   error
+		stats = &Stats{}
+	)
+	if stats.CgroupStats, err = c.cgroupManager.GetStats(); err != nil {
+		return stats, newSystemError(err)
+	}
+	for _, iface := range c.config.Networks {
+		switch iface.Type {
+		case "veth":
+			istats, err := getNetworkInterfaceStats(iface.HostInterfaceName)
+			if err != nil {
+				return stats, newSystemError(err)
+			}
+			stats.Interfaces = append(stats.Interfaces, istats)
+		}
+	}
+	return stats, nil
+}
+
+func (c *linuxContainer) Set() error {
+	c.m.Lock()
+	defer c.m.Unlock()
+	return c.cgroupManager.Set(c.config)
+}
+
+func (c *linuxContainer) Start(process *Process) error {
+	c.m.Lock()
+	defer c.m.Unlock()
+	status, err := c.currentStatus()
+	if err != nil {
+		return err
+	}
+	doInit := status == Destroyed
+	parent, err := c.newParentProcess(process, doInit)
+	if err != nil {
+		return newSystemError(err)
+	}
+	if err := parent.start(); err != nil {
+		// terminate the process to ensure that it properly is reaped.
+		if err := parent.terminate(); err != nil {
+			log.Warn(err)
+		}
+		return newSystemError(err)
+	}
+	process.ops = parent
+	if doInit {
+
+		c.updateState(parent)
+	}
+	return nil
+}
+
+func (c *linuxContainer) newParentProcess(p *Process, doInit bool) (parentProcess, error) {
+	parentPipe, childPipe, err := newPipe()
+	if err != nil {
+		return nil, newSystemError(err)
+	}
+	cmd, err := c.commandTemplate(p, childPipe)
+	if err != nil {
+		return nil, newSystemError(err)
+	}
+	if !doInit {
+		return c.newSetnsProcess(p, cmd, parentPipe, childPipe), nil
+	}
+	return c.newInitProcess(p, cmd, parentPipe, childPipe)
+}
+
+func (c *linuxContainer) commandTemplate(p *Process, childPipe *os.File) (*exec.Cmd, error) {
+	cmd := &exec.Cmd{
+		Path: c.initPath,
+		Args: c.initArgs,
+	}
+	cmd.Stdin = p.Stdin
+	cmd.Stdout = p.Stdout
+	cmd.Stderr = p.Stderr
+	cmd.Dir = c.config.Rootfs
+	if cmd.SysProcAttr == nil {
+		cmd.SysProcAttr = &syscall.SysProcAttr{}
+	}
+	cmd.ExtraFiles = []*os.File{childPipe}
+	cmd.SysProcAttr.Pdeathsig = syscall.SIGKILL
+	if c.config.ParentDeathSignal > 0 {
+		cmd.SysProcAttr.Pdeathsig = syscall.Signal(c.config.ParentDeathSignal)
+	}
+	return cmd, nil
+}
+
+func (c *linuxContainer) newInitProcess(p *Process, cmd *exec.Cmd, parentPipe, childPipe *os.File) (*initProcess, error) {
+	t := "_LIBCONTAINER_INITTYPE=standard"
+	cloneFlags := c.config.Namespaces.CloneFlags()
+	if cloneFlags&syscall.CLONE_NEWUSER != 0 {
+		if err := c.addUidGidMappings(cmd.SysProcAttr); err != nil {
+			// user mappings are not supported
+			return nil, err
+		}
+		// Default to root user when user namespaces are enabled.
+		if cmd.SysProcAttr.Credential == nil {
+			cmd.SysProcAttr.Credential = &syscall.Credential{}
+		}
+	}
+	cmd.Env = append(cmd.Env, t)
+	cmd.SysProcAttr.Cloneflags = cloneFlags
+	return &initProcess{
+		cmd:        cmd,
+		childPipe:  childPipe,
+		parentPipe: parentPipe,
+		manager:    c.cgroupManager,
+		config:     c.newInitConfig(p),
+	}, nil
+}
+
+func (c *linuxContainer) newSetnsProcess(p *Process, cmd *exec.Cmd, parentPipe, childPipe *os.File) *setnsProcess {
+	cmd.Env = append(cmd.Env,
+		fmt.Sprintf("_LIBCONTAINER_INITPID=%d", c.initProcess.pid()),
+		"_LIBCONTAINER_INITTYPE=setns",
+	)
+
+	if p.consolePath != "" {
+		cmd.Env = append(cmd.Env, "_LIBCONTAINER_CONSOLE_PATH="+p.consolePath)
+	}
+
+	// TODO: set on container for process management
+	return &setnsProcess{
+		cmd:         cmd,
+		cgroupPaths: c.cgroupManager.GetPaths(),
+		childPipe:   childPipe,
+		parentPipe:  parentPipe,
+		config:      c.newInitConfig(p),
+	}
+}
+
+func (c *linuxContainer) newInitConfig(process *Process) *initConfig {
+	return &initConfig{
+		Config:  c.config,
+		Args:    process.Args,
+		Env:     process.Env,
+		User:    process.User,
+		Cwd:     process.Cwd,
+		Console: process.consolePath,
+	}
+}
+
+func newPipe() (parent *os.File, child *os.File, err error) {
+	fds, err := syscall.Socketpair(syscall.AF_LOCAL, syscall.SOCK_STREAM|syscall.SOCK_CLOEXEC, 0)
+	if err != nil {
+		return nil, nil, err
+	}
+	return os.NewFile(uintptr(fds[1]), "parent"), os.NewFile(uintptr(fds[0]), "child"), nil
+}
+
+func (c *linuxContainer) Destroy() error {
+	c.m.Lock()
+	defer c.m.Unlock()
+	status, err := c.currentStatus()
+	if err != nil {
+		return err
+	}
+	if status != Destroyed {
+		return newGenericError(fmt.Errorf("container is not destroyed"), ContainerNotStopped)
+	}
+	if !c.config.Namespaces.Contains(configs.NEWPID) {
+		if err := killCgroupProcesses(c.cgroupManager); err != nil {
+			log.Warn(err)
+		}
+	}
+	err = c.cgroupManager.Destroy()
+	if rerr := os.RemoveAll(c.root); err == nil {
+		err = rerr
+	}
+	c.initProcess = nil
+	return err
+}
+
+func (c *linuxContainer) Pause() error {
+	c.m.Lock()
+	defer c.m.Unlock()
+	return c.cgroupManager.Freeze(configs.Frozen)
+}
+
+func (c *linuxContainer) Resume() error {
+	c.m.Lock()
+	defer c.m.Unlock()
+	return c.cgroupManager.Freeze(configs.Thawed)
+}
+
+func (c *linuxContainer) NotifyOOM() (<-chan struct{}, error) {
+	return notifyOnOOM(c.cgroupManager.GetPaths())
+}
+
+func (c *linuxContainer) updateState(process parentProcess) error {
+	c.initProcess = process
+	state, err := c.currentState()
+	if err != nil {
+		return err
+	}
+	f, err := os.Create(filepath.Join(c.root, stateFilename))
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+	return json.NewEncoder(f).Encode(state)
+}
+
+func (c *linuxContainer) currentStatus() (Status, error) {
+	if c.initProcess == nil {
+		return Destroyed, nil
+	}
+	// return Running if the init process is alive
+	if err := syscall.Kill(c.initProcess.pid(), 0); err != nil {
+		if err == syscall.ESRCH {
+			return Destroyed, nil
+		}
+		return 0, newSystemError(err)
+	}
+	if c.config.Cgroups != nil && c.config.Cgroups.Freezer == configs.Frozen {
+		return Paused, nil
+	}
+	return Running, nil
+}
+
+func (c *linuxContainer) currentState() (*State, error) {
+	status, err := c.currentStatus()
+	if err != nil {
+		return nil, err
+	}
+	if status == Destroyed {
+		return nil, newGenericError(fmt.Errorf("container destroyed"), ContainerNotExists)
+	}
+	startTime, err := c.initProcess.startTime()
+	if err != nil {
+		return nil, newSystemError(err)
+	}
+	state := &State{
+		ID:                   c.ID(),
+		Config:               *c.config,
+		InitProcessPid:       c.initProcess.pid(),
+		InitProcessStartTime: startTime,
+		CgroupPaths:          c.cgroupManager.GetPaths(),
+		NamespacePaths:       make(map[configs.NamespaceType]string),
+	}
+	for _, ns := range c.config.Namespaces {
+		state.NamespacePaths[ns.Type] = ns.GetPath(c.initProcess.pid())
+	}
+	return state, nil
+}

vendor/src/github.com/docker/libcontainer/container_linux_test.go

@@ -0,0 +1,200 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"fmt"
+	"os"
+	"testing"
+
+	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
+)
+
+type mockCgroupManager struct {
+	pids  []int
+	stats *cgroups.Stats
+	paths map[string]string
+}
+
+func (m *mockCgroupManager) GetPids() ([]int, error) {
+	return m.pids, nil
+}
+
+func (m *mockCgroupManager) GetStats() (*cgroups.Stats, error) {
+	return m.stats, nil
+}
+
+func (m *mockCgroupManager) Apply(pid int) error {
+	return nil
+}
+
+func (m *mockCgroupManager) Set(container *configs.Config) error {
+	return nil
+}
+
+func (m *mockCgroupManager) Destroy() error {
+	return nil
+}
+
+func (m *mockCgroupManager) GetPaths() map[string]string {
+	return m.paths
+}
+
+func (m *mockCgroupManager) Freeze(state configs.FreezerState) error {
+	return nil
+}
+
+type mockProcess struct {
+	_pid    int
+	started string
+}
+
+func (m *mockProcess) terminate() error {
+	return nil
+}
+
+func (m *mockProcess) pid() int {
+	return m._pid
+}
+
+func (m *mockProcess) startTime() (string, error) {
+	return m.started, nil
+}
+
+func (m *mockProcess) start() error {
+	return nil
+}
+
+func (m *mockProcess) wait() (*os.ProcessState, error) {
+	return nil, nil
+}
+
+func (m *mockProcess) signal(_ os.Signal) error {
+	return nil
+}
+
+func TestGetContainerPids(t *testing.T) {
+	container := &linuxContainer{
+		id:            "myid",
+		config:        &configs.Config{},
+		cgroupManager: &mockCgroupManager{pids: []int{1, 2, 3}},
+	}
+	pids, err := container.Processes()
+	if err != nil {
+		t.Fatal(err)
+	}
+	for i, expected := range []int{1, 2, 3} {
+		if pids[i] != expected {
+			t.Fatalf("expected pid %d but received %d", expected, pids[i])
+		}
+	}
+}
+
+func TestGetContainerStats(t *testing.T) {
+	container := &linuxContainer{
+		id:     "myid",
+		config: &configs.Config{},
+		cgroupManager: &mockCgroupManager{
+			pids: []int{1, 2, 3},
+			stats: &cgroups.Stats{
+				MemoryStats: cgroups.MemoryStats{
+					Usage: 1024,
+				},
+			},
+		},
+	}
+	stats, err := container.Stats()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if stats.CgroupStats == nil {
+		t.Fatal("cgroup stats are nil")
+	}
+	if stats.CgroupStats.MemoryStats.Usage != 1024 {
+		t.Fatalf("expected memory usage 1024 but recevied %d", stats.CgroupStats.MemoryStats.Usage)
+	}
+}
+
+func TestGetContainerState(t *testing.T) {
+	var (
+		pid                 = os.Getpid()
+		expectedMemoryPath  = "/sys/fs/cgroup/memory/myid"
+		expectedNetworkPath = "/networks/fd"
+	)
+	container := &linuxContainer{
+		id: "myid",
+		config: &configs.Config{
+			Namespaces: configs.Namespaces{
+				{Type: configs.NEWPID},
+				{Type: configs.NEWNS},
+				{Type: configs.NEWNET, Path: expectedNetworkPath},
+				{Type: configs.NEWUTS},
+				{Type: configs.NEWIPC},
+			},
+		},
+		initProcess: &mockProcess{
+			_pid:    pid,
+			started: "010",
+		},
+		cgroupManager: &mockCgroupManager{
+			pids: []int{1, 2, 3},
+			stats: &cgroups.Stats{
+				MemoryStats: cgroups.MemoryStats{
+					Usage: 1024,
+				},
+			},
+			paths: map[string]string{
+				"memory": expectedMemoryPath,
+			},
+		},
+	}
+	state, err := container.State()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if state.InitProcessPid != pid {
+		t.Fatalf("expected pid %d but received %d", pid, state.InitProcessPid)
+	}
+	if state.InitProcessStartTime != "010" {
+		t.Fatalf("expected process start time 010 but received %s", state.InitProcessStartTime)
+	}
+	paths := state.CgroupPaths
+	if paths == nil {
+		t.Fatal("cgroup paths should not be nil")
+	}
+	if memPath := paths["memory"]; memPath != expectedMemoryPath {
+		t.Fatalf("expected memory path %q but received %q", expectedMemoryPath, memPath)
+	}
+	for _, ns := range container.config.Namespaces {
+		path := state.NamespacePaths[ns.Type]
+		if path == "" {
+			t.Fatalf("expected non nil namespace path for %s", ns.Type)
+		}
+		if ns.Type == configs.NEWNET {
+			if path != expectedNetworkPath {
+				t.Fatalf("expected path %q but received %q", expectedNetworkPath, path)
+			}
+		} else {
+			file := ""
+			switch ns.Type {
+			case configs.NEWNET:
+				file = "net"
+			case configs.NEWNS:
+				file = "mnt"
+			case configs.NEWPID:
+				file = "pid"
+			case configs.NEWIPC:
+				file = "ipc"
+			case configs.NEWUSER:
+				file = "user"
+			case configs.NEWUTS:
+				file = "uts"
+			}
+			expected := fmt.Sprintf("/proc/%d/ns/%s", pid, file)
+			if expected != path {
+				t.Fatalf("expected path %q but received %q", expected, path)
+			}
+		}
+	}
+}

vendor/src/github.com/docker/libcontainer/container_nouserns_linux.go

@@ -0,0 +1,13 @@
+// +build !go1.4
+
+package libcontainer
+
+import (
+	"fmt"
+	"syscall"
+)
+
+// not available before go 1.4
+func (c *linuxContainer) addUidGidMappings(sys *syscall.SysProcAttr) error {
+	return fmt.Errorf("User namespace is not supported in golang < 1.4")
+}

vendor/src/github.com/docker/libcontainer/container_userns_linux.go

@@ -0,0 +1,26 @@
+// +build go1.4
+
+package libcontainer
+
+import "syscall"
+
+// Converts IDMap to SysProcIDMap array and adds it to SysProcAttr.
+func (c *linuxContainer) addUidGidMappings(sys *syscall.SysProcAttr) error {
+	if c.config.UidMappings != nil {
+		sys.UidMappings = make([]syscall.SysProcIDMap, len(c.config.UidMappings))
+		for i, um := range c.config.UidMappings {
+			sys.UidMappings[i].ContainerID = um.ContainerID
+			sys.UidMappings[i].HostID = um.HostID
+			sys.UidMappings[i].Size = um.Size
+		}
+	}
+	if c.config.GidMappings != nil {
+		sys.GidMappings = make([]syscall.SysProcIDMap, len(c.config.GidMappings))
+		for i, gm := range c.config.GidMappings {
+			sys.GidMappings[i].ContainerID = gm.ContainerID
+			sys.GidMappings[i].HostID = gm.HostID
+			sys.GidMappings[i].Size = gm.Size
+		}
+	}
+	return nil
+}

vendor/src/github.com/docker/libcontainer/devices/defaults.go

@@ -1,159 +0,0 @@
-package devices
-
-var (
-	// These are devices that are to be both allowed and created.
-
-	DefaultSimpleDevices = []*Device{
-		// /dev/null and zero
-		{
-			Path:              "/dev/null",
-			Type:              'c',
-			MajorNumber:       1,
-			MinorNumber:       3,
-			CgroupPermissions: "rwm",
-			FileMode:          0666,
-		},
-		{
-			Path:              "/dev/zero",
-			Type:              'c',
-			MajorNumber:       1,
-			MinorNumber:       5,
-			CgroupPermissions: "rwm",
-			FileMode:          0666,
-		},
-
-		{
-			Path:              "/dev/full",
-			Type:              'c',
-			MajorNumber:       1,
-			MinorNumber:       7,
-			CgroupPermissions: "rwm",
-			FileMode:          0666,
-		},
-
-		// consoles and ttys
-		{
-			Path:              "/dev/tty",
-			Type:              'c',
-			MajorNumber:       5,
-			MinorNumber:       0,
-			CgroupPermissions: "rwm",
-			FileMode:          0666,
-		},
-
-		// /dev/urandom,/dev/random
-		{
-			Path:              "/dev/urandom",
-			Type:              'c',
-			MajorNumber:       1,
-			MinorNumber:       9,
-			CgroupPermissions: "rwm",
-			FileMode:          0666,
-		},
-		{
-			Path:              "/dev/random",
-			Type:              'c',
-			MajorNumber:       1,
-			MinorNumber:       8,
-			CgroupPermissions: "rwm",
-			FileMode:          0666,
-		},
-	}
-
-	DefaultAllowedDevices = append([]*Device{
-		// allow mknod for any device
-		{
-			Type:              'c',
-			MajorNumber:       Wildcard,
-			MinorNumber:       Wildcard,
-			CgroupPermissions: "m",
-		},
-		{
-			Type:              'b',
-			MajorNumber:       Wildcard,
-			MinorNumber:       Wildcard,
-			CgroupPermissions: "m",
-		},
-
-		{
-			Path:              "/dev/console",
-			Type:              'c',
-			MajorNumber:       5,
-			MinorNumber:       1,
-			CgroupPermissions: "rwm",
-		},
-		{
-			Path:              "/dev/tty0",
-			Type:              'c',
-			MajorNumber:       4,
-			MinorNumber:       0,
-			CgroupPermissions: "rwm",
-		},
-		{
-			Path:              "/dev/tty1",
-			Type:              'c',
-			MajorNumber:       4,
-			MinorNumber:       1,
-			CgroupPermissions: "rwm",
-		},
-		// /dev/pts/ - pts namespaces are "coming soon"
-		{
-			Path:              "",
-			Type:              'c',
-			MajorNumber:       136,
-			MinorNumber:       Wildcard,
-			CgroupPermissions: "rwm",
-		},
-		{
-			Path:              "",
-			Type:              'c',
-			MajorNumber:       5,
-			MinorNumber:       2,
-			CgroupPermissions: "rwm",
-		},
-
-		// tuntap
-		{
-			Path:              "",
-			Type:              'c',
-			MajorNumber:       10,
-			MinorNumber:       200,
-			CgroupPermissions: "rwm",
-		},
-
-		/*// fuse
-		   {
-		    Path: "",
-		    Type: 'c',
-		    MajorNumber: 10,
-		    MinorNumber: 229,
-		    CgroupPermissions: "rwm",
-		   },
-
-		// rtc
-		   {
-		    Path: "",
-		    Type: 'c',
-		    MajorNumber: 254,
-		    MinorNumber: 0,
-		    CgroupPermissions: "rwm",
-		   },
-		*/
-	}, DefaultSimpleDevices...)
-
-	DefaultAutoCreatedDevices = append([]*Device{
-		{
-			// /dev/fuse is created but not allowed.
-			// This is to allow java to work.  Because java
-			// Insists on there being a /dev/fuse
-			// https://github.com/docker/docker/issues/514
-			// https://github.com/docker/docker/issues/2393
-			//
-			Path:              "/dev/fuse",
-			Type:              'c',
-			MajorNumber:       10,
-			MinorNumber:       229,
-			CgroupPermissions: "rwm",
-		},
-	}, DefaultSimpleDevices...)
-)

vendor/src/github.com/docker/libcontainer/devices/devices.go

@@ -7,14 +7,12 @@ import (
 	"os"
 	"path/filepath"
 	"syscall"
-)
 
-const (
-	Wildcard = -1
+	"github.com/docker/libcontainer/configs"
 )
 
 var (
-	ErrNotADeviceNode = errors.New("not a device node")
+	ErrNotADevice = errors.New("not a device node")
 )
 
 // Testing dependencies
@@ -23,45 +21,20 @@ var (
 	ioutilReadDir = ioutil.ReadDir
 )
 
-type Device struct {
-	Type              rune        `json:"type,omitempty"`
-	Path              string      `json:"path,omitempty"`               // It is fine if this is an empty string in the case that you are using Wildcards
-	MajorNumber       int64       `json:"major_number,omitempty"`       // Use the wildcard constant for wildcards.
-	MinorNumber       int64       `json:"minor_number,omitempty"`       // Use the wildcard constant for wildcards.
-	CgroupPermissions string      `json:"cgroup_permissions,omitempty"` // Typically just "rwm"
-	FileMode          os.FileMode `json:"file_mode,omitempty"`          // The permission bits of the file's mode
-	Uid               uint32      `json:"uid,omitempty"`
-	Gid               uint32      `json:"gid,omitempty"`
-}
-
-func GetDeviceNumberString(deviceNumber int64) string {
-	if deviceNumber == Wildcard {
-		return "*"
-	} else {
-		return fmt.Sprintf("%d", deviceNumber)
-	}
-}
-
-func (device *Device) GetCgroupAllowString() string {
-	return fmt.Sprintf("%c %s:%s %s", device.Type, GetDeviceNumberString(device.MajorNumber), GetDeviceNumberString(device.MinorNumber), device.CgroupPermissions)
-}
-
 // Given the path to a device and it's cgroup_permissions(which cannot be easilly queried) look up the information about a linux device and return that information as a Device struct.
-func GetDevice(path, cgroupPermissions string) (*Device, error) {
+func DeviceFromPath(path, permissions string) (*configs.Device, error) {
 	fileInfo, err := osLstat(path)
 	if err != nil {
 		return nil, err
 	}
-
 	var (
 		devType                rune
 		mode                   = fileInfo.Mode()
 		fileModePermissionBits = os.FileMode.Perm(mode)
 	)
-
 	switch {
 	case mode&os.ModeDevice == 0:
-		return nil, ErrNotADeviceNode
+		return nil, ErrNotADevice
 	case mode&os.ModeCharDevice != 0:
 		fileModePermissionBits |= syscall.S_IFCHR
 		devType = 'c'
@@ -69,36 +42,33 @@ func GetDevice(path, cgroupPermissions string) (*Device, error) {
 		fileModePermissionBits |= syscall.S_IFBLK
 		devType = 'b'
 	}
-
 	stat_t, ok := fileInfo.Sys().(*syscall.Stat_t)
 	if !ok {
 		return nil, fmt.Errorf("cannot determine the device number for device %s", path)
 	}
 	devNumber := int(stat_t.Rdev)
-
-	return &Device{
-		Type:              devType,
-		Path:              path,
-		MajorNumber:       Major(devNumber),
-		MinorNumber:       Minor(devNumber),
-		CgroupPermissions: cgroupPermissions,
-		FileMode:          fileModePermissionBits,
-		Uid:               stat_t.Uid,
-		Gid:               stat_t.Gid,
+	return &configs.Device{
+		Type:        devType,
+		Path:        path,
+		Major:       Major(devNumber),
+		Minor:       Minor(devNumber),
+		Permissions: permissions,
+		FileMode:    fileModePermissionBits,
+		Uid:         stat_t.Uid,
+		Gid:         stat_t.Gid,
 	}, nil
 }
 
-func GetHostDeviceNodes() ([]*Device, error) {
-	return getDeviceNodes("/dev")
+func HostDevices() ([]*configs.Device, error) {
+	return getDevices("/dev")
 }
 
-func getDeviceNodes(path string) ([]*Device, error) {
+func getDevices(path string) ([]*configs.Device, error) {
 	files, err := ioutilReadDir(path)
 	if err != nil {
 		return nil, err
 	}
-
-	out := []*Device{}
+	out := []*configs.Device{}
 	for _, f := range files {
 		switch {
 		case f.IsDir():
@@ -106,7 +76,7 @@ func getDeviceNodes(path string) ([]*Device, error) {
 			case "pts", "shm", "fd", "mqueue":
 				continue
 			default:
-				sub, err := getDeviceNodes(filepath.Join(path, f.Name()))
+				sub, err := getDevices(filepath.Join(path, f.Name()))
 				if err != nil {
 					return nil, err
 				}
@@ -117,16 +87,14 @@ func getDeviceNodes(path string) ([]*Device, error) {
 		case f.Name() == "console":
 			continue
 		}
-
-		device, err := GetDevice(filepath.Join(path, f.Name()), "rwm")
+		device, err := DeviceFromPath(filepath.Join(path, f.Name()), "rwm")
 		if err != nil {
-			if err == ErrNotADeviceNode {
+			if err == ErrNotADevice {
 				continue
 			}
 			return nil, err
 		}
 		out = append(out, device)
 	}
-
 	return out, nil
 }

vendor/src/github.com/docker/libcontainer/devices/devices_test.go

@@ -6,7 +6,7 @@ import (
 	"testing"
 )
 
-func TestGetDeviceLstatFailure(t *testing.T) {
+func TestDeviceFromPathLstatFailure(t *testing.T) {
 	testError := errors.New("test error")
 
 	// Override os.Lstat to inject error.
@@ -14,13 +14,13 @@ func TestGetDeviceLstatFailure(t *testing.T) {
 		return nil, testError
 	}
 
-	_, err := GetDevice("", "")
+	_, err := DeviceFromPath("", "")
 	if err != testError {
 		t.Fatalf("Unexpected error %v, expected %v", err, testError)
 	}
 }
 
-func TestGetHostDeviceNodesIoutilReadDirFailure(t *testing.T) {
+func TestHostDevicesIoutilReadDirFailure(t *testing.T) {
 	testError := errors.New("test error")
 
 	// Override ioutil.ReadDir to inject error.
@@ -28,13 +28,13 @@ func TestGetHostDeviceNodesIoutilReadDirFailure(t *testing.T) {
 		return nil, testError
 	}
 
-	_, err := GetHostDeviceNodes()
+	_, err := HostDevices()
 	if err != testError {
 		t.Fatalf("Unexpected error %v, expected %v", err, testError)
 	}
 }
 
-func TestGetHostDeviceNodesIoutilReadDirDeepFailure(t *testing.T) {
+func TestHostDevicesIoutilReadDirDeepFailure(t *testing.T) {
 	testError := errors.New("test error")
 	called := false
 
@@ -54,7 +54,7 @@ func TestGetHostDeviceNodesIoutilReadDirDeepFailure(t *testing.T) {
 		return []os.FileInfo{fi}, nil
 	}
 
-	_, err := GetHostDeviceNodes()
+	_, err := HostDevices()
 	if err != testError {
 		t.Fatalf("Unexpected error %v, expected %v", err, testError)
 	}

vendor/src/github.com/docker/libcontainer/devices/number.go

@@ -20,7 +20,3 @@ func Major(devNumber int) int64 {
 func Minor(devNumber int) int64 {
 	return int64((devNumber & 0xff) | ((devNumber >> 12) & 0xfff00))
 }
-
-func Mkdev(majorNumber int64, minorNumber int64) int {
-	return int((majorNumber << 8) | (minorNumber & 0xff) | ((minorNumber & 0xfff00) << 12))
-}

vendor/src/github.com/docker/libcontainer/docs/man/nsinit.1.md

@@ -0,0 +1,38 @@
+% nsinit User Manual
+% docker/libcontainer
+% JAN 2015
+
+NAME:
+   nsinit - A low-level utility for managing containers.
+	    It is used to spawn new containers or join existing containers.
+
+USAGE:
+   nsinit [global options] command [command options] [arguments...]
+
+VERSION:
+   0.1
+
+COMMANDS:
+	config	display the container configuration 
+	exec	execute a new command inside a container
+	init	runs the init process inside the namespace
+	oom	display oom notifications for a container
+	pause	pause the container's processes
+	stats	display statistics for the container
+	unpause	unpause the container's processes
+	help, h	shows a list of commands or help for one command
+
+EXAMPLES:
+
+Get the <container_id> of an already running docker container.
+`sudo docker ps` will return the list of all the running containers.
+
+take the <container_id> (e.g. 4addb0b2d307) and go to its config directory
+`/var/lib/docker/execdriver/native/4addb0b2d307` and here you can run the nsinit
+command line utility.
+
+e.g. `nsinit exec /bin/bash` will start a shell on the already running container.
+   
+# HISTORY
+Jan 2015, Originally compiled by Shishir Mahajan (shishir dot mahajan at redhat dot com)
+based on nsinit source material and internal work.	

vendor/src/github.com/docker/libcontainer/error.go

@@ -1,5 +1,7 @@
 package libcontainer
 
+import "io"
+
 // API error code type.
 type ErrorCode int
 
@@ -8,29 +10,52 @@ const (
 	// Factory errors
 	IdInUse ErrorCode = iota
 	InvalidIdFormat
-	// TODO: add Load errors
 
 	// Container errors
-	ContainerDestroyed
+	ContainerNotExists
 	ContainerPaused
+	ContainerNotStopped
+	ContainerNotRunning
+
+	// Process errors
+	ProcessNotExecuted
 
 	// Common errors
 	ConfigInvalid
 	SystemError
 )
 
+func (c ErrorCode) String() string {
+	switch c {
+	case IdInUse:
+		return "Id already in use"
+	case InvalidIdFormat:
+		return "Invalid format"
+	case ContainerPaused:
+		return "Container paused"
+	case ConfigInvalid:
+		return "Invalid configuration"
+	case SystemError:
+		return "System error"
+	case ContainerNotExists:
+		return "Container does not exist"
+	case ContainerNotStopped:
+		return "Container is not stopped"
+	case ContainerNotRunning:
+		return "Container is not running"
+	default:
+		return "Unknown error"
+	}
+}
+
 // API Error type.
 type Error interface {
 	error
 
-	// Returns the stack trace, if any, which identifies the
-	// point at which the error occurred.
-	Stack() []byte
-
 	// Returns a verbose string including the error message
 	// and a representation of the stack trace suitable for
 	// printing.
-	Detail() string
+	Detail(w io.Writer) error
 
 	// Returns the error code for this error.
 	Code() ErrorCode

vendor/src/github.com/docker/libcontainer/error_test.go

@@ -0,0 +1,20 @@
+package libcontainer
+
+import "testing"
+
+func TestErrorCode(t *testing.T) {
+	codes := map[ErrorCode]string{
+		IdInUse:            "Id already in use",
+		InvalidIdFormat:    "Invalid format",
+		ContainerPaused:    "Container paused",
+		ConfigInvalid:      "Invalid configuration",
+		SystemError:        "System error",
+		ContainerNotExists: "Container does not exist",
+	}
+
+	for code, expected := range codes {
+		if actual := code.String(); actual != expected {
+			t.Fatalf("expected string %q but received %q", expected, actual)
+		}
+	}
+}

vendor/src/github.com/docker/libcontainer/factory.go

@@ -1,7 +1,10 @@
 package libcontainer
 
-type Factory interface {
+import (
+	"github.com/docker/libcontainer/configs"
+)
 
+type Factory interface {
 	// Creates a new container with the given id and starts the initial process inside it.
 	// id must be a string containing only letters, digits and underscores and must contain
 	// between 1 and 1024 characters, inclusive.
@@ -11,22 +14,31 @@ type Factory interface {
 	//
 	// Returns the new container with a running process.
 	//
-	// Errors:
+	// errors:
 	// IdInUse - id is already in use by a container
 	// InvalidIdFormat - id has incorrect format
 	// ConfigInvalid - config is invalid
-	// SystemError - System error
+	// Systemerror - System error
 	//
 	// On error, any partially created container parts are cleaned up (the operation is atomic).
-	Create(id string, config *Config) (Container, Error)
+	Create(id string, config *configs.Config) (Container, error)
 
-	// Load takes an ID for an existing container and reconstructs the container
-	// from the state.
+	// Load takes an ID for an existing container and returns the container information
+	// from the state.  This presents a read only view of the container.
 	//
-	// Errors:
+	// errors:
 	// Path does not exist
 	// Container is stopped
 	// System error
-	// TODO: fix description
-	Load(id string) (Container, Error)
+	Load(id string) (Container, error)
+
+	// StartInitialization is an internal API to libcontainer used during the rexec of the
+	// container.  pipefd is the fd to the child end of the pipe used to syncronize the
+	// parent and child process providing state and configuration to the child process and
+	// returning any errors during the init of the container
+	//
+	// Errors:
+	// pipe connection error
+	// system error
+	StartInitialization(pipefd uintptr) error
 }

vendor/src/github.com/docker/libcontainer/factory_linux.go

@@ -0,0 +1,262 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"fmt"
+	"io/ioutil"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"regexp"
+
+	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/cgroups/fs"
+	"github.com/docker/libcontainer/cgroups/systemd"
+	"github.com/docker/libcontainer/configs"
+	"github.com/docker/libcontainer/configs/validate"
+)
+
+const (
+	stateFilename = "state.json"
+)
+
+var (
+	idRegex  = regexp.MustCompile(`^[\w_]+$`)
+	maxIdLen = 1024
+)
+
+// InitArgs returns an options func to configure a LinuxFactory with the
+// provided init arguments.
+func InitArgs(args ...string) func(*LinuxFactory) error {
+	return func(l *LinuxFactory) error {
+		name := args[0]
+		if filepath.Base(name) == name {
+			if lp, err := exec.LookPath(name); err == nil {
+				name = lp
+			}
+		}
+		l.InitPath = name
+		l.InitArgs = append([]string{name}, args[1:]...)
+		return nil
+	}
+}
+
+// InitPath returns an options func to configure a LinuxFactory with the
+// provided absolute path to the init binary and arguements.
+func InitPath(path string, args ...string) func(*LinuxFactory) error {
+	return func(l *LinuxFactory) error {
+		l.InitPath = path
+		l.InitArgs = args
+		return nil
+	}
+}
+
+// SystemdCgroups is an options func to configure a LinuxFactory to return
+// containers that use systemd to create and manage cgroups.
+func SystemdCgroups(l *LinuxFactory) error {
+	l.NewCgroupsManager = func(config *configs.Cgroup, paths map[string]string) cgroups.Manager {
+		return &systemd.Manager{
+			Cgroups: config,
+			Paths:   paths,
+		}
+	}
+	return nil
+}
+
+// Cgroupfs is an options func to configure a LinuxFactory to return
+// containers that use the native cgroups filesystem implementation to
+// create and manage cgroups.
+func Cgroupfs(l *LinuxFactory) error {
+	l.NewCgroupsManager = func(config *configs.Cgroup, paths map[string]string) cgroups.Manager {
+		return &fs.Manager{
+			Cgroups: config,
+			Paths:   paths,
+		}
+	}
+	return nil
+}
+
+// New returns a linux based container factory based in the root directory and
+// configures the factory with the provided option funcs.
+func New(root string, options ...func(*LinuxFactory) error) (Factory, error) {
+	if root != "" {
+		if err := os.MkdirAll(root, 0700); err != nil {
+			return nil, newGenericError(err, SystemError)
+		}
+	}
+	l := &LinuxFactory{
+		Root:      root,
+		Validator: validate.New(),
+	}
+	InitArgs(os.Args[0], "init")(l)
+	Cgroupfs(l)
+	for _, opt := range options {
+		if err := opt(l); err != nil {
+			return nil, err
+		}
+	}
+	return l, nil
+}
+
+// LinuxFactory implements the default factory interface for linux based systems.
+type LinuxFactory struct {
+	// Root directory for the factory to store state.
+	Root string
+
+	// InitPath is the absolute path to the init binary.
+	InitPath string
+
+	// InitArgs are arguments for calling the init responsibilities for spawning
+	// a container.
+	InitArgs []string
+
+	// Validator provides validation to container configurations.
+	Validator validate.Validator
+
+	// NewCgroupsManager returns an initialized cgroups manager for a single container.
+	NewCgroupsManager func(config *configs.Cgroup, paths map[string]string) cgroups.Manager
+}
+
+func (l *LinuxFactory) Create(id string, config *configs.Config) (Container, error) {
+	if l.Root == "" {
+		return nil, newGenericError(fmt.Errorf("invalid root"), ConfigInvalid)
+	}
+	if err := l.validateID(id); err != nil {
+		return nil, err
+	}
+	if err := l.Validator.Validate(config); err != nil {
+		return nil, newGenericError(err, ConfigInvalid)
+	}
+	containerRoot := filepath.Join(l.Root, id)
+	if _, err := os.Stat(containerRoot); err == nil {
+		return nil, newGenericError(fmt.Errorf("Container with id exists: %v", id), IdInUse)
+	} else if !os.IsNotExist(err) {
+		return nil, newGenericError(err, SystemError)
+	}
+	if err := os.MkdirAll(containerRoot, 0700); err != nil {
+		return nil, newGenericError(err, SystemError)
+	}
+	return &linuxContainer{
+		id:            id,
+		root:          containerRoot,
+		config:        config,
+		initPath:      l.InitPath,
+		initArgs:      l.InitArgs,
+		cgroupManager: l.NewCgroupsManager(config.Cgroups, nil),
+	}, nil
+}
+
+func (l *LinuxFactory) Load(id string) (Container, error) {
+	if l.Root == "" {
+		return nil, newGenericError(fmt.Errorf("invalid root"), ConfigInvalid)
+	}
+	containerRoot := filepath.Join(l.Root, id)
+	state, err := l.loadState(containerRoot)
+	if err != nil {
+		return nil, err
+	}
+	r := &restoredProcess{
+		processPid:       state.InitProcessPid,
+		processStartTime: state.InitProcessStartTime,
+	}
+	return &linuxContainer{
+		initProcess:   r,
+		id:            id,
+		config:        &state.Config,
+		initPath:      l.InitPath,
+		initArgs:      l.InitArgs,
+		cgroupManager: l.NewCgroupsManager(state.Config.Cgroups, state.CgroupPaths),
+		root:          containerRoot,
+	}, nil
+}
+
+// StartInitialization loads a container by opening the pipe fd from the parent to read the configuration and state
+// This is a low level implementation detail of the reexec and should not be consumed externally
+func (l *LinuxFactory) StartInitialization(pipefd uintptr) (err error) {
+	var (
+		pipe = os.NewFile(uintptr(pipefd), "pipe")
+		it   = initType(os.Getenv("_LIBCONTAINER_INITTYPE"))
+	)
+	// clear the current process's environment to clean any libcontainer
+	// specific env vars.
+	os.Clearenv()
+	defer func() {
+		// if we have an error during the initialization of the container's init then send it back to the
+		// parent process in the form of an initError.
+		if err != nil {
+			// ensure that any data sent from the parent is consumed so it doesn't
+			// receive ECONNRESET when the child writes to the pipe.
+			ioutil.ReadAll(pipe)
+			if err := json.NewEncoder(pipe).Encode(newSystemError(err)); err != nil {
+				panic(err)
+			}
+		}
+		// ensure that this pipe is always closed
+		pipe.Close()
+	}()
+	i, err := newContainerInit(it, pipe)
+	if err != nil {
+		return err
+	}
+	return i.Init()
+}
+
+func (l *LinuxFactory) loadState(root string) (*State, error) {
+	f, err := os.Open(filepath.Join(root, stateFilename))
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil, newGenericError(err, ContainerNotExists)
+		}
+		return nil, newGenericError(err, SystemError)
+	}
+	defer f.Close()
+	var state *State
+	if err := json.NewDecoder(f).Decode(&state); err != nil {
+		return nil, newGenericError(err, SystemError)
+	}
+	return state, nil
+}
+
+func (l *LinuxFactory) validateID(id string) error {
+	if !idRegex.MatchString(id) {
+		return newGenericError(fmt.Errorf("Invalid id format: %v", id), InvalidIdFormat)
+	}
+	if len(id) > maxIdLen {
+		return newGenericError(fmt.Errorf("Invalid id format: %v", id), InvalidIdFormat)
+	}
+	return nil
+}
+
+// restoredProcess represents a process where the calling process may or may not be
+// the parent process.  This process is created when a factory loads a container from
+// a persisted state.
+type restoredProcess struct {
+	processPid       int
+	processStartTime string
+}
+
+func (p *restoredProcess) start() error {
+	return newGenericError(fmt.Errorf("restored process cannot be started"), SystemError)
+}
+
+func (p *restoredProcess) pid() int {
+	return p.processPid
+}
+
+func (p *restoredProcess) terminate() error {
+	return newGenericError(fmt.Errorf("restored process cannot be terminated"), SystemError)
+}
+
+func (p *restoredProcess) wait() (*os.ProcessState, error) {
+	return nil, newGenericError(fmt.Errorf("restored process cannot be waited on"), SystemError)
+}
+
+func (p *restoredProcess) startTime() (string, error) {
+	return p.processStartTime, nil
+}
+
+func (p *restoredProcess) signal(s os.Signal) error {
+	return newGenericError(fmt.Errorf("restored process cannot be signaled"), SystemError)
+}

vendor/src/github.com/docker/libcontainer/factory_linux_test.go

@@ -0,0 +1,125 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"io/ioutil"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/docker/libcontainer/configs"
+)
+
+func newTestRoot() (string, error) {
+	dir, err := ioutil.TempDir("", "libcontainer")
+	if err != nil {
+		return "", err
+	}
+	if err := os.MkdirAll(dir, 0700); err != nil {
+		return "", err
+	}
+	return dir, nil
+}
+
+func TestFactoryNew(t *testing.T) {
+	root, rerr := newTestRoot()
+	if rerr != nil {
+		t.Fatal(rerr)
+	}
+	defer os.RemoveAll(root)
+	factory, err := New(root, Cgroupfs)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if factory == nil {
+		t.Fatal("factory should not be nil")
+	}
+	lfactory, ok := factory.(*LinuxFactory)
+	if !ok {
+		t.Fatal("expected linux factory returned on linux based systems")
+	}
+	if lfactory.Root != root {
+		t.Fatalf("expected factory root to be %q but received %q", root, lfactory.Root)
+	}
+}
+
+func TestFactoryLoadNotExists(t *testing.T) {
+	root, rerr := newTestRoot()
+	if rerr != nil {
+		t.Fatal(rerr)
+	}
+	defer os.RemoveAll(root)
+	factory, err := New(root, Cgroupfs)
+	if err != nil {
+		t.Fatal(err)
+	}
+	_, err = factory.Load("nocontainer")
+	if err == nil {
+		t.Fatal("expected nil error loading non-existing container")
+	}
+	lerr, ok := err.(Error)
+	if !ok {
+		t.Fatal("expected libcontainer error type")
+	}
+	if lerr.Code() != ContainerNotExists {
+		t.Fatalf("expected error code %s but received %s", ContainerNotExists, lerr.Code())
+	}
+}
+
+func TestFactoryLoadContainer(t *testing.T) {
+	root, err := newTestRoot()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(root)
+	// setup default container config and state for mocking
+	var (
+		id             = "1"
+		expectedConfig = &configs.Config{
+			Rootfs: "/mycontainer/root",
+		}
+		expectedState = &State{
+			InitProcessPid: 1024,
+			Config:         *expectedConfig,
+		}
+	)
+	if err := os.Mkdir(filepath.Join(root, id), 0700); err != nil {
+		t.Fatal(err)
+	}
+	if err := marshal(filepath.Join(root, id, stateFilename), expectedState); err != nil {
+		t.Fatal(err)
+	}
+	factory, err := New(root, Cgroupfs)
+	if err != nil {
+		t.Fatal(err)
+	}
+	container, err := factory.Load(id)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if container.ID() != id {
+		t.Fatalf("expected container id %q but received %q", id, container.ID())
+	}
+	config := container.Config()
+	if config.Rootfs != expectedConfig.Rootfs {
+		t.Fatalf("expected rootfs %q but received %q", expectedConfig.Rootfs, config.Rootfs)
+	}
+	lcontainer, ok := container.(*linuxContainer)
+	if !ok {
+		t.Fatal("expected linux container on linux based systems")
+	}
+	if lcontainer.initProcess.pid() != expectedState.InitProcessPid {
+		t.Fatalf("expected init pid %d but received %d", expectedState.InitProcessPid, lcontainer.initProcess.pid())
+	}
+}
+
+func marshal(path string, v interface{}) error {
+	f, err := os.Create(path)
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+	return json.NewEncoder(f).Encode(v)
+}

vendor/src/github.com/docker/libcontainer/generic_error.go

@@ -0,0 +1,74 @@
+package libcontainer
+
+import (
+	"fmt"
+	"io"
+	"text/template"
+	"time"
+
+	"github.com/docker/libcontainer/stacktrace"
+)
+
+var errorTemplate = template.Must(template.New("error").Parse(`Timestamp: {{.Timestamp}}
+Code: {{.ECode}}
+{{if .Message }}
+Message: {{.Message}}
+{{end}}
+Frames:{{range $i, $frame := .Stack.Frames}}
+---
+{{$i}}: {{$frame.Function}}
+Package: {{$frame.Package}}
+File: {{$frame.File}}@{{$frame.Line}}{{end}}
+`))
+
+func newGenericError(err error, c ErrorCode) Error {
+	if le, ok := err.(Error); ok {
+		return le
+	}
+	gerr := &genericError{
+		Timestamp: time.Now(),
+		Err:       err,
+		ECode:     c,
+		Stack:     stacktrace.Capture(1),
+	}
+	if err != nil {
+		gerr.Message = err.Error()
+	}
+	return gerr
+}
+
+func newSystemError(err error) Error {
+	if le, ok := err.(Error); ok {
+		return le
+	}
+	gerr := &genericError{
+		Timestamp: time.Now(),
+		Err:       err,
+		ECode:     SystemError,
+		Stack:     stacktrace.Capture(1),
+	}
+	if err != nil {
+		gerr.Message = err.Error()
+	}
+	return gerr
+}
+
+type genericError struct {
+	Timestamp time.Time
+	ECode     ErrorCode
+	Err       error `json:"-"`
+	Message   string
+	Stack     stacktrace.Stacktrace
+}
+
+func (e *genericError) Error() string {
+	return fmt.Sprintf("[%d] %s: %s", e.ECode, e.ECode, e.Message)
+}
+
+func (e *genericError) Code() ErrorCode {
+	return e.ECode
+}
+
+func (e *genericError) Detail(w io.Writer) error {
+	return errorTemplate.Execute(w, e)
+}

vendor/src/github.com/docker/libcontainer/generic_error_test.go

@@ -0,0 +1,14 @@
+package libcontainer
+
+import (
+	"fmt"
+	"io/ioutil"
+	"testing"
+)
+
+func TestErrorDetail(t *testing.T) {
+	err := newGenericError(fmt.Errorf("test error"), SystemError)
+	if derr := err.Detail(ioutil.Discard); derr != nil {
+		t.Fatal(derr)
+	}
+}

vendor/src/github.com/docker/libcontainer/hack/validate.sh

@@ -0,0 +1,13 @@
+#!/usr/bin/env bash
+set -e
+
+# This script runs all validations
+
+validate() {
+    sed -i 's!docker/docker!docker/libcontainer!' /go/src/github.com/docker/docker/hack/make/.validate
+    bash /go/src/github.com/docker/docker/hack/make/validate-dco
+    bash /go/src/github.com/docker/docker/hack/make/validate-gofmt
+}
+
+# run validations
+validate

vendor/src/github.com/docker/libcontainer/init_linux.go

@@ -0,0 +1,253 @@
+// +build linux
+
+package libcontainer
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"strings"
+	"syscall"
+
+	log "github.com/Sirupsen/logrus"
+	"github.com/docker/libcontainer/cgroups"
+	"github.com/docker/libcontainer/configs"
+	"github.com/docker/libcontainer/netlink"
+	"github.com/docker/libcontainer/system"
+	"github.com/docker/libcontainer/user"
+	"github.com/docker/libcontainer/utils"
+)
+
+type initType string
+
+const (
+	initSetns    initType = "setns"
+	initStandard initType = "standard"
+)
+
+type pid struct {
+	Pid int `json:"pid"`
+}
+
+// network is an internal struct used to setup container networks.
+type network struct {
+	configs.Network
+
+	// TempVethPeerName is a unique tempory veth peer name that was placed into
+	// the container's namespace.
+	TempVethPeerName string `json:"temp_veth_peer_name"`
+}
+
+// initConfig is used for transferring parameters from Exec() to Init()
+type initConfig struct {
+	Args     []string        `json:"args"`
+	Env      []string        `json:"env"`
+	Cwd      string          `json:"cwd"`
+	User     string          `json:"user"`
+	Config   *configs.Config `json:"config"`
+	Console  string          `json:"console"`
+	Networks []*network      `json:"network"`
+}
+
+type initer interface {
+	Init() error
+}
+
+func newContainerInit(t initType, pipe *os.File) (initer, error) {
+	var config *initConfig
+	if err := json.NewDecoder(pipe).Decode(&config); err != nil {
+		return nil, err
+	}
+	if err := populateProcessEnvironment(config.Env); err != nil {
+		return nil, err
+	}
+	switch t {
+	case initSetns:
+		return &linuxSetnsInit{
+			config: config,
+		}, nil
+	case initStandard:
+		return &linuxStandardInit{
+			config: config,
+		}, nil
+	}
+	return nil, fmt.Errorf("unknown init type %q", t)
+}
+
+// populateProcessEnvironment loads the provided environment variables into the
+// current processes's environment.
+func populateProcessEnvironment(env []string) error {
+	for _, pair := range env {
+		p := strings.SplitN(pair, "=", 2)
+		if len(p) < 2 {
+			return fmt.Errorf("invalid environment '%v'", pair)
+		}
+		if err := os.Setenv(p[0], p[1]); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// finalizeNamespace drops the caps, sets the correct user
+// and working dir, and closes any leaked file descriptors
+// before execing the command inside the namespace
+func finalizeNamespace(config *initConfig) error {
+	// Ensure that all non-standard fds we may have accidentally
+	// inherited are marked close-on-exec so they stay out of the
+	// container
+	if err := utils.CloseExecFrom(3); err != nil {
+		return err
+	}
+	w, err := newCapWhitelist(config.Config.Capabilities)
+	if err != nil {
+		return err
+	}
+	// drop capabilities in bounding set before changing user
+	if err := w.dropBoundingSet(); err != nil {
+		return err
+	}
+	// preserve existing capabilities while we change users
+	if err := system.SetKeepCaps(); err != nil {
+		return err
+	}
+	if err := setupUser(config); err != nil {
+		return err
+	}
+	if err := system.ClearKeepCaps(); err != nil {
+		return err
+	}
+	// drop all other capabilities
+	if err := w.drop(); err != nil {
+		return err
+	}
+	if config.Cwd != "" {
+		if err := syscall.Chdir(config.Cwd); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// joinExistingNamespaces gets all the namespace paths specified for the container and
+// does a setns on the namespace fd so that the current process joins the namespace.
+func joinExistingNamespaces(namespaces []configs.Namespace) error {
+	for _, ns := range namespaces {
+		if ns.Path != "" {
+			f, err := os.OpenFile(ns.Path, os.O_RDONLY, 0)
+			if err != nil {
+				return err
+			}
+			err = system.Setns(f.Fd(), uintptr(ns.Syscall()))
+			f.Close()
+			if err != nil {
+				return err
+			}
+		}
+	}
+	return nil
+}
+
+// setupUser changes the groups, gid, and uid for the user inside the container
+func setupUser(config *initConfig) error {
+	// Set up defaults.
+	defaultExecUser := user.ExecUser{
+		Uid:  syscall.Getuid(),
+		Gid:  syscall.Getgid(),
+		Home: "/",
+	}
+	passwdPath, err := user.GetPasswdPath()
+	if err != nil {
+		return err
+	}
+	groupPath, err := user.GetGroupPath()
+	if err != nil {
+		return err
+	}
+	execUser, err := user.GetExecUserPath(config.User, &defaultExecUser, passwdPath, groupPath)
+	if err != nil {
+		return err
+	}
+	suppGroups := append(execUser.Sgids, config.Config.AdditionalGroups...)
+	if err := syscall.Setgroups(suppGroups); err != nil {
+		return err
+	}
+	if err := system.Setgid(execUser.Gid); err != nil {
+		return err
+	}
+	if err := system.Setuid(execUser.Uid); err != nil {
+		return err
+	}
+	// if we didn't get HOME already, set it based on the user's HOME
+	if envHome := os.Getenv("HOME"); envHome == "" {
+		if err := os.Setenv("HOME", execUser.Home); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// setupNetwork sets up and initializes any network interface inside the container.
+func setupNetwork(config *initConfig) error {
+	for _, config := range config.Networks {
+		strategy, err := getStrategy(config.Type)
+		if err != nil {
+			return err
+		}
+		if err := strategy.initialize(config); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func setupRoute(config *configs.Config) error {
+	for _, config := range config.Routes {
+		if err := netlink.AddRoute(config.Destination, config.Source, config.Gateway, config.InterfaceName); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func setupRlimits(config *configs.Config) error {
+	for _, rlimit := range config.Rlimits {
+		l := &syscall.Rlimit{Max: rlimit.Hard, Cur: rlimit.Soft}
+		if err := syscall.Setrlimit(rlimit.Type, l); err != nil {
+			return fmt.Errorf("error setting rlimit type %v: %v", rlimit.Type, err)
+		}
+	}
+	return nil
+}
+
+// killCgroupProcesses freezes then iterates over all the processes inside the
+// manager's cgroups sending a SIGKILL to each process then waiting for them to
+// exit.
+func killCgroupProcesses(m cgroups.Manager) error {
+	var procs []*os.Process
+	if err := m.Freeze(configs.Frozen); err != nil {
+		log.Warn(err)
+	}
+	pids, err := m.GetPids()
+	if err != nil {
+		m.Freeze(configs.Thawed)
+		return err
+	}
+	for _, pid := range pids {
+		if p, err := os.FindProcess(pid); err == nil {
+			procs = append(procs, p)
+			if err := p.Kill(); err != nil {
+				log.Warn(err)
+			}
+		}
+	}
+	if err := m.Freeze(configs.Thawed); err != nil {
+		log.Warn(err)
+	}
+	for _, p := range procs {
+		if _, err := p.Wait(); err != nil {
+			log.Warn(err)
+		}
+	}
+	return nil
+}

vendor/src/github.com/docker/libcontainer/integration/exec_test.go

@@ -1,34 +1,50 @@
 package integration
 
 import (
+	"bytes"
+	"io/ioutil"
 	"os"
 	"strings"
 	"testing"
 
 	"github.com/docker/libcontainer"
+	"github.com/docker/libcontainer/configs"
 )
 
 func TestExecPS(t *testing.T) {
+	testExecPS(t, false)
+}
+
+func TestUsernsExecPS(t *testing.T) {
+	if _, err := os.Stat("/proc/self/ns/user"); os.IsNotExist(err) {
+		t.Skip("userns is unsupported")
+	}
+	testExecPS(t, true)
+}
+
+func testExecPS(t *testing.T, userns bool) {
 	if testing.Short() {
 		return
 	}
-
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
 	defer remove(rootfs)
-
 	config := newTemplateConfig(rootfs)
-	buffers, exitCode, err := runContainer(config, "", "ps")
-	if err != nil {
-		t.Fatal(err)
+	if userns {
+		config.UidMappings = []configs.IDMap{{0, 0, 1000}}
+		config.GidMappings = []configs.IDMap{{0, 0, 1000}}
+		config.Namespaces = append(config.Namespaces, configs.Namespace{Type: configs.NEWUSER})
 	}
 
+	buffers, exitCode, err := runContainer(config, "", "ps")
+	if err != nil {
+		t.Fatalf("%s: %s", buffers, err)
+	}
 	if exitCode != 0 {
 		t.Fatalf("exit code not 0. code %d stderr %q", exitCode, buffers.Stderr)
 	}
-
 	lines := strings.Split(buffers.Stdout.String(), "\n")
 	if len(lines) < 2 {
 		t.Fatalf("more than one process running for output %q", buffers.Stdout.String())
@@ -45,7 +61,7 @@ func TestIPCPrivate(t *testing.T) {
 		return
 	}
 
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -76,7 +92,7 @@ func TestIPCHost(t *testing.T) {
 		return
 	}
 
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -88,7 +104,7 @@ func TestIPCHost(t *testing.T) {
 	}
 
 	config := newTemplateConfig(rootfs)
-	config.Namespaces.Remove(libcontainer.NEWIPC)
+	config.Namespaces.Remove(configs.NEWIPC)
 	buffers, exitCode, err := runContainer(config, "", "readlink", "/proc/self/ns/ipc")
 	if err != nil {
 		t.Fatal(err)
@@ -108,7 +124,7 @@ func TestIPCJoinPath(t *testing.T) {
 		return
 	}
 
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -120,7 +136,7 @@ func TestIPCJoinPath(t *testing.T) {
 	}
 
 	config := newTemplateConfig(rootfs)
-	config.Namespaces.Add(libcontainer.NEWIPC, "/proc/1/ns/ipc")
+	config.Namespaces.Add(configs.NEWIPC, "/proc/1/ns/ipc")
 
 	buffers, exitCode, err := runContainer(config, "", "readlink", "/proc/self/ns/ipc")
 	if err != nil {
@@ -141,14 +157,14 @@ func TestIPCBadPath(t *testing.T) {
 		return
 	}
 
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
 	defer remove(rootfs)
 
 	config := newTemplateConfig(rootfs)
-	config.Namespaces.Add(libcontainer.NEWIPC, "/proc/1/ns/ipcc")
+	config.Namespaces.Add(configs.NEWIPC, "/proc/1/ns/ipcc")
 
 	_, _, err = runContainer(config, "", "true")
 	if err == nil {
@@ -161,7 +177,7 @@ func TestRlimit(t *testing.T) {
 		return
 	}
 
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -172,38 +188,289 @@ func TestRlimit(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	if limit := strings.TrimSpace(out.Stdout.String()); limit != "1024" {
-		t.Fatalf("expected rlimit to be 1024, got %s", limit)
+	if limit := strings.TrimSpace(out.Stdout.String()); limit != "1025" {
+		t.Fatalf("expected rlimit to be 1025, got %s", limit)
 	}
 }
 
-func TestPIDNSPrivate(t *testing.T) {
+func newTestRoot() (string, error) {
+	dir, err := ioutil.TempDir("", "libcontainer")
+	if err != nil {
+		return "", err
+	}
+	if err := os.MkdirAll(dir, 0700); err != nil {
+		return "", err
+	}
+	return dir, nil
+}
+
+func waitProcess(p *libcontainer.Process, t *testing.T) {
+	status, err := p.Wait()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !status.Success() {
+		t.Fatal(status)
+	}
+}
+
+func TestEnter(t *testing.T) {
 	if testing.Short() {
 		return
 	}
+	root, err := newTestRoot()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(root)
 
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
 	defer remove(rootfs)
 
-	l, err := os.Readlink("/proc/1/ns/pid")
-	if err != nil {
-		t.Fatal(err)
-	}
-
 	config := newTemplateConfig(rootfs)
-	buffers, exitCode, err := runContainer(config, "", "readlink", "/proc/self/ns/pid")
+
+	factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	if exitCode != 0 {
-		t.Fatalf("exit code not 0. code %d stderr %q", exitCode, buffers.Stderr)
+	container, err := factory.Create("test", config)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer container.Destroy()
+
+	// Execute a first process in the container
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
 	}
 
-	if actual := strings.Trim(buffers.Stdout.String(), "\n"); actual == l {
-		t.Fatalf("pid link should be private to the container but equals host %q %q", actual, l)
+	var stdout, stdout2 bytes.Buffer
+
+	pconfig := libcontainer.Process{
+		Args:   []string{"sh", "-c", "cat && readlink /proc/self/ns/pid"},
+		Env:    standardEnvironment,
+		Stdin:  stdinR,
+		Stdout: &stdout,
+	}
+	err = container.Start(&pconfig)
+	stdinR.Close()
+	defer stdinW.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+	pid, err := pconfig.Pid()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Execute another process in the container
+	stdinR2, stdinW2, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	pconfig2 := libcontainer.Process{
+		Env: standardEnvironment,
+	}
+	pconfig2.Args = []string{"sh", "-c", "cat && readlink /proc/self/ns/pid"}
+	pconfig2.Stdin = stdinR2
+	pconfig2.Stdout = &stdout2
+
+	err = container.Start(&pconfig2)
+	stdinR2.Close()
+	defer stdinW2.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	pid2, err := pconfig2.Pid()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	processes, err := container.Processes()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	n := 0
+	for i := range processes {
+		if processes[i] == pid || processes[i] == pid2 {
+			n++
+		}
+	}
+	if n != 2 {
+		t.Fatal("unexpected number of processes", processes, pid, pid2)
+	}
+
+	// Wait processes
+	stdinW2.Close()
+	waitProcess(&pconfig2, t)
+
+	stdinW.Close()
+	waitProcess(&pconfig, t)
+
+	// Check that both processes live in the same pidns
+	pidns := string(stdout.Bytes())
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	pidns2 := string(stdout2.Bytes())
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if pidns != pidns2 {
+		t.Fatal("The second process isn't in the required pid namespace", pidns, pidns2)
+	}
+}
+
+func TestProcessEnv(t *testing.T) {
+	if testing.Short() {
+		return
+	}
+	root, err := newTestRoot()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(root)
+
+	rootfs, err := newRootfs()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer remove(rootfs)
+
+	config := newTemplateConfig(rootfs)
+
+	factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	container, err := factory.Create("test", config)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer container.Destroy()
+
+	var stdout bytes.Buffer
+	pconfig := libcontainer.Process{
+		Args: []string{"sh", "-c", "env"},
+		Env: []string{
+			"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+			"HOSTNAME=integration",
+			"TERM=xterm",
+			"FOO=BAR",
+		},
+		Stdin:  nil,
+		Stdout: &stdout,
+	}
+	err = container.Start(&pconfig)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Wait for process
+	waitProcess(&pconfig, t)
+
+	outputEnv := string(stdout.Bytes())
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Check that the environment has the key/value pair we added
+	if !strings.Contains(outputEnv, "FOO=BAR") {
+		t.Fatal("Environment doesn't have the expected FOO=BAR key/value pair: ", outputEnv)
+	}
+
+	// Make sure that HOME is set
+	if !strings.Contains(outputEnv, "HOME=/root") {
+		t.Fatal("Environment doesn't have HOME set: ", outputEnv)
+	}
+}
+
+func TestFreeze(t *testing.T) {
+	if testing.Short() {
+		return
+	}
+	root, err := newTestRoot()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(root)
+
+	rootfs, err := newRootfs()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer remove(rootfs)
+
+	config := newTemplateConfig(rootfs)
+
+	factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	container, err := factory.Create("test", config)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer container.Destroy()
+
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	pconfig := libcontainer.Process{
+		Args:  []string{"cat"},
+		Env:   standardEnvironment,
+		Stdin: stdinR,
+	}
+	err = container.Start(&pconfig)
+	stdinR.Close()
+	defer stdinW.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	pid, err := pconfig.Pid()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	process, err := os.FindProcess(pid)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if err := container.Pause(); err != nil {
+		t.Fatal(err)
+	}
+	state, err := container.Status()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := container.Resume(); err != nil {
+		t.Fatal(err)
+	}
+	if state != libcontainer.Paused {
+		t.Fatal("Unexpected state: ", state)
+	}
+
+	stdinW.Close()
+	s, err := process.Wait()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !s.Success() {
+		t.Fatal(s.String())
 	}
 }

vendor/src/github.com/docker/libcontainer/integration/execin_test.go

@@ -1,62 +1,70 @@
 package integration
 
 import (
+	"bytes"
+	"io"
 	"os"
-	"os/exec"
 	"strings"
-	"sync"
 	"testing"
+	"time"
 
 	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/namespaces"
 )
 
 func TestExecIn(t *testing.T) {
 	if testing.Short() {
 		return
 	}
-
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
 	defer remove(rootfs)
-
 	config := newTemplateConfig(rootfs)
-	if err := writeConfig(config); err != nil {
-		t.Fatalf("failed to write config %s", err)
-	}
-
-	containerCmd, statePath, containerErr := startLongRunningContainer(config)
-	defer func() {
-		// kill the container
-		if containerCmd.Process != nil {
-			containerCmd.Process.Kill()
-		}
-		if err := <-containerErr; err != nil {
-			t.Fatal(err)
-		}
-	}()
-
-	// start the exec process
-	state, err := libcontainer.GetState(statePath)
+	container, err := newContainer(config)
 	if err != nil {
-		t.Fatalf("failed to get state %s", err)
+		t.Fatal(err)
 	}
-	buffers := newStdBuffers()
-	execErr := make(chan error, 1)
-	go func() {
-		_, err := namespaces.ExecIn(config, state, []string{"ps"},
-			os.Args[0], "exec", buffers.Stdin, buffers.Stdout, buffers.Stderr,
-			"", nil)
-		execErr <- err
-	}()
-	if err := <-execErr; err != nil {
-		t.Fatalf("exec finished with error %s", err)
+	defer container.Destroy()
+
+	// Execute a first process in the container
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	process := &libcontainer.Process{
+		Args:  []string{"cat"},
+		Env:   standardEnvironment,
+		Stdin: stdinR,
+	}
+	err = container.Start(process)
+	stdinR.Close()
+	defer stdinW.Close()
+	if err != nil {
+		t.Fatal(err)
 	}
 
+	buffers := newStdBuffers()
+	ps := &libcontainer.Process{
+		Args:   []string{"ps"},
+		Env:    standardEnvironment,
+		Stdin:  buffers.Stdin,
+		Stdout: buffers.Stdout,
+		Stderr: buffers.Stderr,
+	}
+	err = container.Start(ps)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := ps.Wait(); err != nil {
+		t.Fatal(err)
+	}
+	stdinW.Close()
+	if _, err := process.Wait(); err != nil {
+		t.Log(err)
+	}
 	out := buffers.Stdout.String()
-	if !strings.Contains(out, "sleep 10") || !strings.Contains(out, "ps") {
+	if !strings.Contains(out, "cat") || !strings.Contains(out, "ps") {
 		t.Fatalf("unexpected running process, output %q", out)
 	}
 }
@@ -65,76 +73,244 @@ func TestExecInRlimit(t *testing.T) {
 	if testing.Short() {
 		return
 	}
-
-	rootfs, err := newRootFs()
+	rootfs, err := newRootfs()
 	if err != nil {
 		t.Fatal(err)
 	}
 	defer remove(rootfs)
-
 	config := newTemplateConfig(rootfs)
-	if err := writeConfig(config); err != nil {
-		t.Fatalf("failed to write config %s", err)
-	}
-
-	containerCmd, statePath, containerErr := startLongRunningContainer(config)
-	defer func() {
-		// kill the container
-		if containerCmd.Process != nil {
-			containerCmd.Process.Kill()
-		}
-		if err := <-containerErr; err != nil {
-			t.Fatal(err)
-		}
-	}()
-
-	// start the exec process
-	state, err := libcontainer.GetState(statePath)
+	container, err := newContainer(config)
 	if err != nil {
-		t.Fatalf("failed to get state %s", err)
+		t.Fatal(err)
 	}
+	defer container.Destroy()
+
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	process := &libcontainer.Process{
+		Args:  []string{"cat"},
+		Env:   standardEnvironment,
+		Stdin: stdinR,
+	}
+	err = container.Start(process)
+	stdinR.Close()
+	defer stdinW.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+
 	buffers := newStdBuffers()
-	execErr := make(chan error, 1)
-	go func() {
-		_, err := namespaces.ExecIn(config, state, []string{"/bin/sh", "-c", "ulimit -n"},
-			os.Args[0], "exec", buffers.Stdin, buffers.Stdout, buffers.Stderr,
-			"", nil)
-		execErr <- err
-	}()
-	if err := <-execErr; err != nil {
-		t.Fatalf("exec finished with error %s", err)
+	ps := &libcontainer.Process{
+		Args:   []string{"/bin/sh", "-c", "ulimit -n"},
+		Env:    standardEnvironment,
+		Stdin:  buffers.Stdin,
+		Stdout: buffers.Stdout,
+		Stderr: buffers.Stderr,
+	}
+	err = container.Start(ps)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := ps.Wait(); err != nil {
+		t.Fatal(err)
+	}
+	stdinW.Close()
+	if _, err := process.Wait(); err != nil {
+		t.Log(err)
 	}
-
 	out := buffers.Stdout.String()
-	if limit := strings.TrimSpace(out); limit != "1024" {
-		t.Fatalf("expected rlimit to be 1024, got %s", limit)
+	if limit := strings.TrimSpace(out); limit != "1025" {
+		t.Fatalf("expected rlimit to be 1025, got %s", limit)
 	}
 }
 
-// start a long-running container so we have time to inspect execin processes
-func startLongRunningContainer(config *libcontainer.Config) (*exec.Cmd, string, chan error) {
-	containerErr := make(chan error, 1)
-	containerCmd := &exec.Cmd{}
-	var statePath string
-
-	createCmd := func(container *libcontainer.Config, console, dataPath, init string,
-		pipe *os.File, args []string) *exec.Cmd {
-		containerCmd = namespaces.DefaultCreateCommand(container, console, dataPath, init, pipe, args)
-		statePath = dataPath
-		return containerCmd
+func TestExecInError(t *testing.T) {
+	if testing.Short() {
+		return
 	}
+	rootfs, err := newRootfs()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer remove(rootfs)
+	config := newTemplateConfig(rootfs)
+	container, err := newContainer(config)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer container.Destroy()
 
-	var containerStart sync.WaitGroup
-	containerStart.Add(1)
-	go func() {
-		buffers := newStdBuffers()
-		_, err := namespaces.Exec(config,
-			buffers.Stdin, buffers.Stdout, buffers.Stderr,
-			"", config.RootFs, []string{"sleep", "10"},
-			createCmd, containerStart.Done)
-		containerErr <- err
+	// Execute a first process in the container
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	process := &libcontainer.Process{
+		Args:  []string{"cat"},
+		Env:   standardEnvironment,
+		Stdin: stdinR,
+	}
+	err = container.Start(process)
+	stdinR.Close()
+	defer func() {
+		stdinW.Close()
+		if _, err := process.Wait(); err != nil {
+			t.Log(err)
+		}
 	}()
-	containerStart.Wait()
+	if err != nil {
+		t.Fatal(err)
+	}
 
-	return containerCmd, statePath, containerErr
+	unexistent := &libcontainer.Process{
+		Args: []string{"unexistent"},
+		Env:  standardEnvironment,
+	}
+	err = container.Start(unexistent)
+	if err == nil {
+		t.Fatal("Should be an error")
+	}
+	if !strings.Contains(err.Error(), "executable file not found") {
+		t.Fatalf("Should be error about not found executable, got %s", err)
+	}
+}
+
+func TestExecInTTY(t *testing.T) {
+	if testing.Short() {
+		return
+	}
+	rootfs, err := newRootfs()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer remove(rootfs)
+	config := newTemplateConfig(rootfs)
+	container, err := newContainer(config)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer container.Destroy()
+
+	// Execute a first process in the container
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	process := &libcontainer.Process{
+		Args:  []string{"cat"},
+		Env:   standardEnvironment,
+		Stdin: stdinR,
+	}
+	err = container.Start(process)
+	stdinR.Close()
+	defer stdinW.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	var stdout bytes.Buffer
+	ps := &libcontainer.Process{
+		Args: []string{"ps"},
+		Env:  standardEnvironment,
+	}
+	console, err := ps.NewConsole(0)
+	copy := make(chan struct{})
+	go func() {
+		io.Copy(&stdout, console)
+		close(copy)
+	}()
+	if err != nil {
+		t.Fatal(err)
+	}
+	err = container.Start(ps)
+	if err != nil {
+		t.Fatal(err)
+	}
+	select {
+	case <-time.After(5 * time.Second):
+		t.Fatal("Waiting for copy timed out")
+	case <-copy:
+	}
+	if _, err := ps.Wait(); err != nil {
+		t.Fatal(err)
+	}
+	stdinW.Close()
+	if _, err := process.Wait(); err != nil {
+		t.Log(err)
+	}
+	out := stdout.String()
+	if !strings.Contains(out, "cat") || !strings.Contains(string(out), "ps") {
+		t.Fatalf("unexpected running process, output %q", out)
+	}
+}
+
+func TestExecInEnvironment(t *testing.T) {
+	if testing.Short() {
+		return
+	}
+	rootfs, err := newRootfs()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer remove(rootfs)
+	config := newTemplateConfig(rootfs)
+	container, err := newContainer(config)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer container.Destroy()
+
+	// Execute a first process in the container
+	stdinR, stdinW, err := os.Pipe()
+	if err != nil {
+		t.Fatal(err)
+	}
+	process := &libcontainer.Process{
+		Args:  []string{"cat"},
+		Env:   standardEnvironment,
+		Stdin: stdinR,
+	}
+	err = container.Start(process)
+	stdinR.Close()
+	defer stdinW.Close()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	buffers := newStdBuffers()
+	process2 := &libcontainer.Process{
+		Args: []string{"env"},
+		Env: []string{
+			"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+			"DEBUG=true",
+			"DEBUG=false",
+			"ENV=test",
+		},
+		Stdin:  buffers.Stdin,
+		Stdout: buffers.Stdout,
+		Stderr: buffers.Stderr,
+	}
+	err = container.Start(process2)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := process2.Wait(); err != nil {
+		out := buffers.Stdout.String()
+		t.Fatal(err, out)
+	}
+	stdinW.Close()
+	if _, err := process.Wait(); err != nil {
+		t.Log(err)
+	}
+	out := buffers.Stdout.String()
+	// check execin's process environment
+	if !strings.Contains(out, "DEBUG=false") ||
+		!strings.Contains(out, "ENV=test") ||
+		!strings.Contains(out, "HOME=/root") ||
+		!strings.Contains(out, "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin") ||
+		strings.Contains(out, "DEBUG=true") {
+		t.Fatalf("unexpected running process, output %q", out)
+	}
 }

vendor/src/github.com/docker/libcontainer/integration/init_test.go

@@ -1,76 +1,27 @@
 package integration
 
 import (
-	"encoding/json"
 	"log"
 	"os"
 	"runtime"
 
 	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/namespaces"
-	_ "github.com/docker/libcontainer/namespaces/nsenter"
+	_ "github.com/docker/libcontainer/nsenter"
 )
 
 // init runs the libcontainer initialization code because of the busybox style needs
 // to work around the go runtime and the issues with forking
 func init() {
-	if len(os.Args) < 2 {
+	if len(os.Args) < 2 || os.Args[1] != "init" {
 		return
 	}
-	// handle init
-	if len(os.Args) >= 2 && os.Args[1] == "init" {
-		runtime.LockOSThread()
-
-		container, err := loadConfig()
-		if err != nil {
-			log.Fatal(err)
-		}
-
-		rootfs, err := os.Getwd()
-		if err != nil {
-			log.Fatal(err)
-		}
-
-		if err := namespaces.Init(container, rootfs, "", os.NewFile(3, "pipe"), os.Args[3:]); err != nil {
-			log.Fatalf("unable to initialize for container: %s", err)
-		}
-		os.Exit(1)
+	runtime.GOMAXPROCS(1)
+	runtime.LockOSThread()
+	factory, err := libcontainer.New("")
+	if err != nil {
+		log.Fatalf("unable to initialize for container: %s", err)
 	}
-
-	// handle execin
-	if len(os.Args) >= 2 && os.Args[0] == "nsenter-exec" {
-		runtime.LockOSThread()
-
-		// User args are passed after '--' in the command line.
-		userArgs := findUserArgs()
-
-		config, err := loadConfigFromFd()
-		if err != nil {
-			log.Fatalf("docker-exec: unable to receive config from sync pipe: %s", err)
-		}
-
-		if err := namespaces.FinalizeSetns(config, userArgs); err != nil {
-			log.Fatalf("docker-exec: failed to exec: %s", err)
-		}
-		os.Exit(1)
+	if err := factory.StartInitialization(3); err != nil {
+		log.Fatal(err)
 	}
 }
-
-func findUserArgs() []string {
-	for i, a := range os.Args {
-		if a == "--" {
-			return os.Args[i+1:]
-		}
-	}
-	return []string{}
-}
-
-// loadConfigFromFd loads a container's config from the sync pipe that is provided by
-// fd 3 when running a process
-func loadConfigFromFd() (*libcontainer.Config, error) {
-	var config *libcontainer.Config
-	if err := json.NewDecoder(os.NewFile(3, "child")).Decode(&config); err != nil {
-		return nil, err
-	}
-	return config, nil
-}

vendor/src/github.com/docker/libcontainer/integration/template_test.go

@@ -3,19 +3,25 @@ package integration
 import (
 	"syscall"
 
-	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/cgroups"
-	"github.com/docker/libcontainer/devices"
+	"github.com/docker/libcontainer/configs"
 )
 
+var standardEnvironment = []string{
+	"HOME=/root",
+	"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+	"HOSTNAME=integration",
+	"TERM=xterm",
+}
+
+const defaultMountFlags = syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
+
 // newTemplateConfig returns a base template for running a container
 //
 // it uses a network strategy of just setting a loopback interface
 // and the default setup for devices
-func newTemplateConfig(rootfs string) *libcontainer.Config {
-	return &libcontainer.Config{
-		RootFs: rootfs,
-		Tty:    false,
+func newTemplateConfig(rootfs string) *configs.Config {
+	return &configs.Config{
+		Rootfs: rootfs,
 		Capabilities: []string{
 			"CHOWN",
 			"DAC_OVERRIDE",
@@ -32,41 +38,60 @@ func newTemplateConfig(rootfs string) *libcontainer.Config {
 			"KILL",
 			"AUDIT_WRITE",
 		},
-		Namespaces: libcontainer.Namespaces([]libcontainer.Namespace{
-			{Type: libcontainer.NEWNS},
-			{Type: libcontainer.NEWUTS},
-			{Type: libcontainer.NEWIPC},
-			{Type: libcontainer.NEWPID},
-			{Type: libcontainer.NEWNET},
+		Namespaces: configs.Namespaces([]configs.Namespace{
+			{Type: configs.NEWNS},
+			{Type: configs.NEWUTS},
+			{Type: configs.NEWIPC},
+			{Type: configs.NEWPID},
+			{Type: configs.NEWNET},
 		}),
-		Cgroups: &cgroups.Cgroup{
+		Cgroups: &configs.Cgroup{
+			Name:            "test",
 			Parent:          "integration",
 			AllowAllDevices: false,
-			AllowedDevices:  devices.DefaultAllowedDevices,
+			AllowedDevices:  configs.DefaultAllowedDevices,
 		},
-
-		MountConfig: &libcontainer.MountConfig{
-			DeviceNodes: devices.DefaultAutoCreatedDevices,
+		MaskPaths: []string{
+			"/proc/kcore",
 		},
+		ReadonlyPaths: []string{
+			"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
+		},
+		Devices:  configs.DefaultAutoCreatedDevices,
 		Hostname: "integration",
-		Env: []string{
-			"HOME=/root",
-			"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
-			"HOSTNAME=integration",
-			"TERM=xterm",
+		Mounts: []*configs.Mount{
+			{
+				Device:      "tmpfs",
+				Source:      "shm",
+				Destination: "/dev/shm",
+				Data:        "mode=1777,size=65536k",
+				Flags:       defaultMountFlags,
+			},
+			{
+				Source:      "mqueue",
+				Destination: "/dev/mqueue",
+				Device:      "mqueue",
+				Flags:       defaultMountFlags,
+			},
+			{
+				Source:      "sysfs",
+				Destination: "/sys",
+				Device:      "sysfs",
+				Flags:       defaultMountFlags | syscall.MS_RDONLY,
+			},
 		},
-		Networks: []*libcontainer.Network{
+		Networks: []*configs.Network{
 			{
 				Type:    "loopback",
 				Address: "127.0.0.1/0",
 				Gateway: "localhost",
 			},
 		},
-		Rlimits: []libcontainer.Rlimit{
+		Rlimits: []configs.Rlimit{
 			{
 				Type: syscall.RLIMIT_NOFILE,
-				Hard: uint64(1024),
-				Soft: uint64(1024),
+				Hard: uint64(1025),
+				Soft: uint64(1025),
 			},
 		},
 	}

vendor/src/github.com/docker/libcontainer/integration/utils_test.go

@@ -2,15 +2,15 @@ package integration
 
 import (
 	"bytes"
-	"encoding/json"
 	"fmt"
 	"io/ioutil"
 	"os"
 	"os/exec"
-	"path/filepath"
+	"strings"
+	"syscall"
 
 	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/namespaces"
+	"github.com/docker/libcontainer/configs"
 )
 
 func newStdBuffers() *stdBuffers {
@@ -27,31 +27,19 @@ type stdBuffers struct {
 	Stderr *bytes.Buffer
 }
 
-func writeConfig(config *libcontainer.Config) error {
-	f, err := os.OpenFile(filepath.Join(config.RootFs, "container.json"), os.O_CREATE|os.O_RDWR|os.O_TRUNC, 0700)
-	if err != nil {
-		return err
+func (b *stdBuffers) String() string {
+	s := []string{}
+	if b.Stderr != nil {
+		s = append(s, b.Stderr.String())
 	}
-	defer f.Close()
-	return json.NewEncoder(f).Encode(config)
+	if b.Stdout != nil {
+		s = append(s, b.Stdout.String())
+	}
+	return strings.Join(s, "|")
 }
 
-func loadConfig() (*libcontainer.Config, error) {
-	f, err := os.Open(filepath.Join(os.Getenv("data_path"), "container.json"))
-	if err != nil {
-		return nil, err
-	}
-	defer f.Close()
-
-	var container *libcontainer.Config
-	if err := json.NewDecoder(f).Decode(&container); err != nil {
-		return nil, err
-	}
-	return container, nil
-}
-
-// newRootFs creates a new tmp directory and copies the busybox root filesystem
-func newRootFs() (string, error) {
+// newRootfs creates a new tmp directory and copies the busybox root filesystem
+func newRootfs() (string, error) {
 	dir, err := ioutil.TempDir("", "")
 	if err != nil {
 		return "", err
@@ -79,17 +67,51 @@ func copyBusybox(dest string) error {
 	return nil
 }
 
+func newContainer(config *configs.Config) (libcontainer.Container, error) {
+	factory, err := libcontainer.New(".",
+		libcontainer.InitArgs(os.Args[0], "init", "--"),
+		libcontainer.Cgroupfs,
+	)
+	if err != nil {
+		return nil, err
+	}
+	return factory.Create("testCT", config)
+}
+
 // runContainer runs the container with the specific config and arguments
 //
 // buffers are returned containing the STDOUT and STDERR output for the run
 // along with the exit code and any go error
-func runContainer(config *libcontainer.Config, console string, args ...string) (buffers *stdBuffers, exitCode int, err error) {
-	if err := writeConfig(config); err != nil {
+func runContainer(config *configs.Config, console string, args ...string) (buffers *stdBuffers, exitCode int, err error) {
+	container, err := newContainer(config)
+	if err != nil {
 		return nil, -1, err
 	}
-
+	defer container.Destroy()
 	buffers = newStdBuffers()
-	exitCode, err = namespaces.Exec(config, buffers.Stdin, buffers.Stdout, buffers.Stderr,
-		console, config.RootFs, args, namespaces.DefaultCreateCommand, nil)
+	process := &libcontainer.Process{
+		Args:   args,
+		Env:    standardEnvironment,
+		Stdin:  buffers.Stdin,
+		Stdout: buffers.Stdout,
+		Stderr: buffers.Stderr,
+	}
+
+	err = container.Start(process)
+	if err != nil {
+		return nil, -1, err
+	}
+	ps, err := process.Wait()
+	if err != nil {
+		return nil, -1, err
+	}
+	status := ps.Sys().(syscall.WaitStatus)
+	if status.Exited() {
+		exitCode = status.ExitStatus()
+	} else if status.Signaled() {
+		exitCode = -int(status.Signal())
+	} else {
+		return nil, -1, err
+	}
 	return
 }

vendor/src/github.com/docker/libcontainer/mount/init.go

@@ -1,209 +0,0 @@
-// +build linux
-
-package mount
-
-import (
-	"fmt"
-	"os"
-	"path/filepath"
-	"syscall"
-
-	"github.com/docker/libcontainer/label"
-	"github.com/docker/libcontainer/mount/nodes"
-)
-
-// default mount point flags
-const defaultMountFlags = syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
-
-type mount struct {
-	source string
-	path   string
-	device string
-	flags  int
-	data   string
-}
-
-// InitializeMountNamespace sets up the devices, mount points, and filesystems for use inside a
-// new mount namespace.
-func InitializeMountNamespace(rootfs, console string, sysReadonly bool, mountConfig *MountConfig) error {
-	var (
-		err  error
-		flag = syscall.MS_PRIVATE
-	)
-
-	if mountConfig.NoPivotRoot {
-		flag = syscall.MS_SLAVE
-	}
-
-	if err := syscall.Mount("", "/", "", uintptr(flag|syscall.MS_REC), ""); err != nil {
-		return fmt.Errorf("mounting / with flags %X %s", (flag | syscall.MS_REC), err)
-	}
-
-	if err := syscall.Mount(rootfs, rootfs, "bind", syscall.MS_BIND|syscall.MS_REC, ""); err != nil {
-		return fmt.Errorf("mouting %s as bind %s", rootfs, err)
-	}
-
-	if err := mountSystem(rootfs, sysReadonly, mountConfig); err != nil {
-		return fmt.Errorf("mount system %s", err)
-	}
-
-	// apply any user specified mounts within the new mount namespace
-	for _, m := range mountConfig.Mounts {
-		if err := m.Mount(rootfs, mountConfig.MountLabel); err != nil {
-			return err
-		}
-	}
-
-	if err := nodes.CreateDeviceNodes(rootfs, mountConfig.DeviceNodes); err != nil {
-		return fmt.Errorf("create device nodes %s", err)
-	}
-
-	if err := SetupPtmx(rootfs, console, mountConfig.MountLabel); err != nil {
-		return err
-	}
-
-	// stdin, stdout and stderr could be pointing to /dev/null from parent namespace.
-	// Re-open them inside this namespace.
-	if err := reOpenDevNull(rootfs); err != nil {
-		return fmt.Errorf("Failed to reopen /dev/null %s", err)
-	}
-
-	if err := setupDevSymlinks(rootfs); err != nil {
-		return fmt.Errorf("dev symlinks %s", err)
-	}
-
-	if err := syscall.Chdir(rootfs); err != nil {
-		return fmt.Errorf("chdir into %s %s", rootfs, err)
-	}
-
-	if mountConfig.NoPivotRoot {
-		err = MsMoveRoot(rootfs)
-	} else {
-		err = PivotRoot(rootfs)
-	}
-
-	if err != nil {
-		return err
-	}
-
-	if mountConfig.ReadonlyFs {
-		if err := SetReadonly(); err != nil {
-			return fmt.Errorf("set readonly %s", err)
-		}
-	}
-
-	syscall.Umask(0022)
-
-	return nil
-}
-
-// mountSystem sets up linux specific system mounts like mqueue, sys, proc, shm, and devpts
-// inside the mount namespace
-func mountSystem(rootfs string, sysReadonly bool, mountConfig *MountConfig) error {
-	for _, m := range newSystemMounts(rootfs, mountConfig.MountLabel, sysReadonly) {
-		if err := os.MkdirAll(m.path, 0755); err != nil && !os.IsExist(err) {
-			return fmt.Errorf("mkdirall %s %s", m.path, err)
-		}
-		if err := syscall.Mount(m.source, m.path, m.device, uintptr(m.flags), m.data); err != nil {
-			return fmt.Errorf("mounting %s into %s %s", m.source, m.path, err)
-		}
-	}
-	return nil
-}
-
-func createIfNotExists(path string, isDir bool) error {
-	if _, err := os.Stat(path); err != nil {
-		if os.IsNotExist(err) {
-			if isDir {
-				if err := os.MkdirAll(path, 0755); err != nil {
-					return err
-				}
-			} else {
-				if err := os.MkdirAll(filepath.Dir(path), 0755); err != nil {
-					return err
-				}
-				f, err := os.OpenFile(path, os.O_CREATE, 0755)
-				if err != nil {
-					return err
-				}
-				f.Close()
-			}
-		}
-	}
-	return nil
-}
-
-func setupDevSymlinks(rootfs string) error {
-	var links = [][2]string{
-		{"/proc/self/fd", "/dev/fd"},
-		{"/proc/self/fd/0", "/dev/stdin"},
-		{"/proc/self/fd/1", "/dev/stdout"},
-		{"/proc/self/fd/2", "/dev/stderr"},
-	}
-
-	// kcore support can be toggled with CONFIG_PROC_KCORE; only create a symlink
-	// in /dev if it exists in /proc.
-	if _, err := os.Stat("/proc/kcore"); err == nil {
-		links = append(links, [2]string{"/proc/kcore", "/dev/kcore"})
-	}
-
-	for _, link := range links {
-		var (
-			src = link[0]
-			dst = filepath.Join(rootfs, link[1])
-		)
-
-		if err := os.Symlink(src, dst); err != nil && !os.IsExist(err) {
-			return fmt.Errorf("symlink %s %s %s", src, dst, err)
-		}
-	}
-
-	return nil
-}
-
-// TODO: this is crappy right now and should be cleaned up with a better way of handling system and
-// standard bind mounts allowing them to be more dynamic
-func newSystemMounts(rootfs, mountLabel string, sysReadonly bool) []mount {
-	systemMounts := []mount{
-		{source: "proc", path: filepath.Join(rootfs, "proc"), device: "proc", flags: defaultMountFlags},
-		{source: "tmpfs", path: filepath.Join(rootfs, "dev"), device: "tmpfs", flags: syscall.MS_NOSUID | syscall.MS_STRICTATIME, data: label.FormatMountLabel("mode=755", mountLabel)},
-		{source: "shm", path: filepath.Join(rootfs, "dev", "shm"), device: "tmpfs", flags: defaultMountFlags, data: label.FormatMountLabel("mode=1777,size=65536k", mountLabel)},
-		{source: "mqueue", path: filepath.Join(rootfs, "dev", "mqueue"), device: "mqueue", flags: defaultMountFlags},
-		{source: "devpts", path: filepath.Join(rootfs, "dev", "pts"), device: "devpts", flags: syscall.MS_NOSUID | syscall.MS_NOEXEC, data: label.FormatMountLabel("newinstance,ptmxmode=0666,mode=620,gid=5", mountLabel)},
-	}
-
-	sysMountFlags := defaultMountFlags
-	if sysReadonly {
-		sysMountFlags |= syscall.MS_RDONLY
-	}
-
-	systemMounts = append(systemMounts, mount{source: "sysfs", path: filepath.Join(rootfs, "sys"), device: "sysfs", flags: sysMountFlags})
-
-	return systemMounts
-}
-
-// Is stdin, stdout or stderr were to be pointing to '/dev/null',
-// this method will make them point to '/dev/null' from within this namespace.
-func reOpenDevNull(rootfs string) error {
-	var stat, devNullStat syscall.Stat_t
-	file, err := os.Open(filepath.Join(rootfs, "/dev/null"))
-	if err != nil {
-		return fmt.Errorf("Failed to open /dev/null - %s", err)
-	}
-	defer file.Close()
-	if err = syscall.Fstat(int(file.Fd()), &devNullStat); err != nil {
-		return fmt.Errorf("Failed to stat /dev/null - %s", err)
-	}
-	for fd := 0; fd < 3; fd++ {
-		if err = syscall.Fstat(fd, &stat); err != nil {
-			return fmt.Errorf("Failed to stat fd %d - %s", fd, err)
-		}
-		if stat.Rdev == devNullStat.Rdev {
-			// Close and re-open the fd.
-			if err = syscall.Dup2(int(file.Fd()), fd); err != nil {
-				return fmt.Errorf("Failed to dup fd %d to fd %d - %s", file.Fd(), fd, err)
-			}
-		}
-	}
-	return nil
-}

vendor/src/github.com/docker/libcontainer/mount/mount.go

@@ -1,109 +0,0 @@
-package mount
-
-import (
-	"fmt"
-	"os"
-	"path/filepath"
-	"syscall"
-
-	"github.com/docker/docker/pkg/symlink"
-	"github.com/docker/libcontainer/label"
-)
-
-type Mount struct {
-	Type        string `json:"type,omitempty"`
-	Source      string `json:"source,omitempty"`      // Source path, in the host namespace
-	Destination string `json:"destination,omitempty"` // Destination path, in the container
-	Writable    bool   `json:"writable,omitempty"`
-	Relabel     string `json:"relabel,omitempty"` // Relabel source if set, "z" indicates shared, "Z" indicates unshared
-	Private     bool   `json:"private,omitempty"`
-	Slave       bool   `json:"slave,omitempty"`
-}
-
-func (m *Mount) Mount(rootfs, mountLabel string) error {
-	switch m.Type {
-	case "bind":
-		return m.bindMount(rootfs, mountLabel)
-	case "tmpfs":
-		return m.tmpfsMount(rootfs, mountLabel)
-	default:
-		return fmt.Errorf("unsupported mount type %s for %s", m.Type, m.Destination)
-	}
-}
-
-func (m *Mount) bindMount(rootfs, mountLabel string) error {
-	var (
-		flags = syscall.MS_BIND | syscall.MS_REC
-		dest  = filepath.Join(rootfs, m.Destination)
-	)
-
-	if !m.Writable {
-		flags = flags | syscall.MS_RDONLY
-	}
-
-	if m.Slave {
-		flags = flags | syscall.MS_SLAVE
-	}
-
-	stat, err := os.Stat(m.Source)
-	if err != nil {
-		return err
-	}
-
-	// FIXME: (crosbymichael) This does not belong here and should be done a layer above
-	dest, err = symlink.FollowSymlinkInScope(dest, rootfs)
-	if err != nil {
-		return err
-	}
-
-	if err := createIfNotExists(dest, stat.IsDir()); err != nil {
-		return fmt.Errorf("creating new bind mount target %s", err)
-	}
-
-	if err := syscall.Mount(m.Source, dest, "bind", uintptr(flags), ""); err != nil {
-		return fmt.Errorf("mounting %s into %s %s", m.Source, dest, err)
-	}
-
-	if !m.Writable {
-		if err := syscall.Mount(m.Source, dest, "bind", uintptr(flags|syscall.MS_REMOUNT), ""); err != nil {
-			return fmt.Errorf("remounting %s into %s %s", m.Source, dest, err)
-		}
-	}
-
-	if m.Relabel != "" {
-		if err := label.Relabel(m.Source, mountLabel, m.Relabel); err != nil {
-			return fmt.Errorf("relabeling %s to %s %s", m.Source, mountLabel, err)
-		}
-	}
-
-	if m.Private {
-		if err := syscall.Mount("", dest, "none", uintptr(syscall.MS_PRIVATE), ""); err != nil {
-			return fmt.Errorf("mounting %s private %s", dest, err)
-		}
-	}
-
-	return nil
-}
-
-func (m *Mount) tmpfsMount(rootfs, mountLabel string) error {
-	var (
-		err  error
-		l    = label.FormatMountLabel("", mountLabel)
-		dest = filepath.Join(rootfs, m.Destination)
-	)
-
-	// FIXME: (crosbymichael) This does not belong here and should be done a layer above
-	if dest, err = symlink.FollowSymlinkInScope(dest, rootfs); err != nil {
-		return err
-	}
-
-	if err := createIfNotExists(dest, true); err != nil {
-		return fmt.Errorf("creating new tmpfs mount target %s", err)
-	}
-
-	if err := syscall.Mount("tmpfs", dest, "tmpfs", uintptr(defaultMountFlags), l); err != nil {
-		return fmt.Errorf("%s mounting %s in tmpfs", err, dest)
-	}
-
-	return nil
-}

vendor/src/github.com/docker/libcontainer/mount/mount_config.go

@@ -1,28 +0,0 @@
-package mount
-
-import (
-	"errors"
-
-	"github.com/docker/libcontainer/devices"
-)
-
-var ErrUnsupported = errors.New("Unsupported method")
-
-type MountConfig struct {
-	// NoPivotRoot will use MS_MOVE and a chroot to jail the process into the container's rootfs
-	// This is a common option when the container is running in ramdisk
-	NoPivotRoot bool `json:"no_pivot_root,omitempty"`
-
-	// ReadonlyFs will remount the container's rootfs as readonly where only externally mounted
-	// bind mounts are writtable
-	ReadonlyFs bool `json:"readonly_fs,omitempty"`
-
-	// Mounts specify additional source and destination paths that will be mounted inside the container's
-	// rootfs and mount namespace if specified
-	Mounts []*Mount `json:"mounts,omitempty"`
-
-	// The device nodes that should be automatically created within the container upon container start.  Note, make sure that the node is marked as allowed in the cgroup as well!
-	DeviceNodes []*devices.Device `json:"device_nodes,omitempty"`
-
-	MountLabel string `json:"mount_label,omitempty"`
-}

vendor/src/github.com/docker/libcontainer/mount/msmoveroot.go

@@ -1,20 +0,0 @@
-// +build linux
-
-package mount
-
-import (
-	"fmt"
-	"syscall"
-)
-
-func MsMoveRoot(rootfs string) error {
-	if err := syscall.Mount(rootfs, "/", "", syscall.MS_MOVE, ""); err != nil {
-		return fmt.Errorf("mount move %s into / %s", rootfs, err)
-	}
-
-	if err := syscall.Chroot("."); err != nil {
-		return fmt.Errorf("chroot . %s", err)
-	}
-
-	return syscall.Chdir("/")
-}

vendor/src/github.com/docker/libcontainer/mount/nodes/nodes.go

@@ -1,57 +0,0 @@
-// +build linux
-
-package nodes
-
-import (
-	"fmt"
-	"os"
-	"path/filepath"
-	"syscall"
-
-	"github.com/docker/libcontainer/devices"
-)
-
-// Create the device nodes in the container.
-func CreateDeviceNodes(rootfs string, nodesToCreate []*devices.Device) error {
-	oldMask := syscall.Umask(0000)
-	defer syscall.Umask(oldMask)
-
-	for _, node := range nodesToCreate {
-		if err := CreateDeviceNode(rootfs, node); err != nil {
-			return err
-		}
-	}
-	return nil
-}
-
-// Creates the device node in the rootfs of the container.
-func CreateDeviceNode(rootfs string, node *devices.Device) error {
-	var (
-		dest   = filepath.Join(rootfs, node.Path)
-		parent = filepath.Dir(dest)
-	)
-
-	if err := os.MkdirAll(parent, 0755); err != nil {
-		return err
-	}
-
-	fileMode := node.FileMode
-	switch node.Type {
-	case 'c':
-		fileMode |= syscall.S_IFCHR
-	case 'b':
-		fileMode |= syscall.S_IFBLK
-	default:
-		return fmt.Errorf("%c is not a valid device type for device %s", node.Type, node.Path)
-	}
-
-	if err := syscall.Mknod(dest, uint32(fileMode), devices.Mkdev(node.MajorNumber, node.MinorNumber)); err != nil && !os.IsExist(err) {
-		return fmt.Errorf("mknod %s %s", node.Path, err)
-	}
-
-	if err := syscall.Chown(dest, int(node.Uid), int(node.Gid)); err != nil {
-		return fmt.Errorf("chown %s to %d:%d", node.Path, node.Uid, node.Gid)
-	}
-
-	return nil
-}

vendor/src/github.com/docker/libcontainer/mount/nodes/nodes_unsupported.go

@@ -1,13 +0,0 @@
-// +build !linux
-
-package nodes
-
-import (
-	"errors"
-
-	"github.com/docker/libcontainer/devices"
-)
-
-func CreateDeviceNodes(rootfs string, nodesToCreate []*devices.Device) error {
-	return errors.New("Unsupported method")
-}

vendor/src/github.com/docker/libcontainer/mount/pivotroot.go

@@ -1,34 +0,0 @@
-// +build linux
-
-package mount
-
-import (
-	"fmt"
-	"io/ioutil"
-	"os"
-	"path/filepath"
-	"syscall"
-)
-
-func PivotRoot(rootfs string) error {
-	pivotDir, err := ioutil.TempDir(rootfs, ".pivot_root")
-	if err != nil {
-		return fmt.Errorf("can't create pivot_root dir %s, error %v", pivotDir, err)
-	}
-
-	if err := syscall.PivotRoot(rootfs, pivotDir); err != nil {
-		return fmt.Errorf("pivot_root %s", err)
-	}
-
-	if err := syscall.Chdir("/"); err != nil {
-		return fmt.Errorf("chdir / %s", err)
-	}
-
-	// path to pivot dir now changed, update
-	pivotDir = filepath.Join("/", filepath.Base(pivotDir))
-	if err := syscall.Unmount(pivotDir, syscall.MNT_DETACH); err != nil {
-		return fmt.Errorf("unmount pivot_root dir %s", err)
-	}
-
-	return os.Remove(pivotDir)
-}

vendor/src/github.com/docker/libcontainer/mount/ptmx.go

@@ -1,30 +0,0 @@
-// +build linux
-
-package mount
-
-import (
-	"fmt"
-	"os"
-	"path/filepath"
-
-	"github.com/docker/libcontainer/console"
-)
-
-func SetupPtmx(rootfs, consolePath, mountLabel string) error {
-	ptmx := filepath.Join(rootfs, "dev/ptmx")
-	if err := os.Remove(ptmx); err != nil && !os.IsNotExist(err) {
-		return err
-	}
-
-	if err := os.Symlink("pts/ptmx", ptmx); err != nil {
-		return fmt.Errorf("symlink dev ptmx %s", err)
-	}
-
-	if consolePath != "" {
-		if err := console.Setup(rootfs, consolePath, mountLabel); err != nil {
-			return err
-		}
-	}
-
-	return nil
-}

vendor/src/github.com/docker/libcontainer/mount/readonly.go

@@ -1,11 +0,0 @@
-// +build linux
-
-package mount
-
-import (
-	"syscall"
-)
-
-func SetReadonly() error {
-	return syscall.Mount("/", "/", "bind", syscall.MS_BIND|syscall.MS_REMOUNT|syscall.MS_RDONLY|syscall.MS_REC, "")
-}

vendor/src/github.com/docker/libcontainer/mount/remount.go

@@ -1,31 +0,0 @@
-// +build linux
-
-package mount
-
-import "syscall"
-
-func RemountProc() error {
-	if err := syscall.Unmount("/proc", syscall.MNT_DETACH); err != nil {
-		return err
-	}
-
-	if err := syscall.Mount("proc", "/proc", "proc", uintptr(defaultMountFlags), ""); err != nil {
-		return err
-	}
-
-	return nil
-}
-
-func RemountSys() error {
-	if err := syscall.Unmount("/sys", syscall.MNT_DETACH); err != nil {
-		if err != syscall.EINVAL {
-			return err
-		}
-	} else {
-		if err := syscall.Mount("sysfs", "/sys", "sysfs", uintptr(defaultMountFlags), ""); err != nil {
-			return err
-		}
-	}
-
-	return nil
-}

vendor/src/github.com/docker/libcontainer/namespaces/create.go

@@ -1,10 +0,0 @@
-package namespaces
-
-import (
-	"os"
-	"os/exec"
-
-	"github.com/docker/libcontainer"
-)
-
-type CreateCommand func(container *libcontainer.Config, console, dataPath, init string, childPipe *os.File, args []string) *exec.Cmd

vendor/src/github.com/docker/libcontainer/namespaces/exec.go

@@ -1,229 +0,0 @@
-// +build linux
-
-package namespaces
-
-import (
-	"encoding/json"
-	"io"
-	"os"
-	"os/exec"
-	"syscall"
-
-	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/cgroups"
-	"github.com/docker/libcontainer/cgroups/fs"
-	"github.com/docker/libcontainer/cgroups/systemd"
-	"github.com/docker/libcontainer/network"
-	"github.com/docker/libcontainer/system"
-)
-
-const (
-	EXIT_SIGNAL_OFFSET = 128
-)
-
-// TODO(vishh): This is part of the libcontainer API and it does much more than just namespaces related work.
-// Move this to libcontainer package.
-// Exec performs setup outside of a namespace so that a container can be
-// executed.  Exec is a high level function for working with container namespaces.
-func Exec(container *libcontainer.Config, stdin io.Reader, stdout, stderr io.Writer, console, dataPath string, args []string, createCommand CreateCommand, startCallback func()) (int, error) {
-	var err error
-
-	// create a pipe so that we can syncronize with the namespaced process and
-	// pass the state and configuration to the child process
-	parent, child, err := newInitPipe()
-	if err != nil {
-		return -1, err
-	}
-	defer parent.Close()
-
-	command := createCommand(container, console, dataPath, os.Args[0], child, args)
-	// Note: these are only used in non-tty mode
-	// if there is a tty for the container it will be opened within the namespace and the
-	// fds will be duped to stdin, stdiout, and stderr
-	command.Stdin = stdin
-	command.Stdout = stdout
-	command.Stderr = stderr
-
-	if err := command.Start(); err != nil {
-		child.Close()
-		return -1, err
-	}
-	child.Close()
-
-	wait := func() (*os.ProcessState, error) {
-		ps, err := command.Process.Wait()
-		// we should kill all processes in cgroup when init is died if we use
-		// host PID namespace
-		if !container.Namespaces.Contains(libcontainer.NEWPID) {
-			killAllPids(container)
-		}
-		return ps, err
-	}
-
-	terminate := func(terr error) (int, error) {
-		// TODO: log the errors for kill and wait
-		command.Process.Kill()
-		wait()
-		return -1, terr
-	}
-
-	started, err := system.GetProcessStartTime(command.Process.Pid)
-	if err != nil {
-		return terminate(err)
-	}
-
-	// Do this before syncing with child so that no children
-	// can escape the cgroup
-	cgroupPaths, err := SetupCgroups(container, command.Process.Pid)
-	if err != nil {
-		return terminate(err)
-	}
-	defer cgroups.RemovePaths(cgroupPaths)
-
-	var networkState network.NetworkState
-	if err := InitializeNetworking(container, command.Process.Pid, &networkState); err != nil {
-		return terminate(err)
-	}
-	// send the state to the container's init process then shutdown writes for the parent
-	if err := json.NewEncoder(parent).Encode(networkState); err != nil {
-		return terminate(err)
-	}
-	// shutdown writes for the parent side of the pipe
-	if err := syscall.Shutdown(int(parent.Fd()), syscall.SHUT_WR); err != nil {
-		return terminate(err)
-	}
-
-	state := &libcontainer.State{
-		InitPid:       command.Process.Pid,
-		InitStartTime: started,
-		NetworkState:  networkState,
-		CgroupPaths:   cgroupPaths,
-	}
-
-	if err := libcontainer.SaveState(dataPath, state); err != nil {
-		return terminate(err)
-	}
-	defer libcontainer.DeleteState(dataPath)
-
-	// wait for the child process to fully complete and receive an error message
-	// if one was encoutered
-	var ierr *initError
-	if err := json.NewDecoder(parent).Decode(&ierr); err != nil && err != io.EOF {
-		return terminate(err)
-	}
-	if ierr != nil {
-		return terminate(ierr)
-	}
-
-	if startCallback != nil {
-		startCallback()
-	}
-
-	ps, err := wait()
-	if err != nil {
-		if _, ok := err.(*exec.ExitError); !ok {
-			return -1, err
-		}
-	}
-	// waiting for pipe flushing
-	command.Wait()
-
-	waitStatus := ps.Sys().(syscall.WaitStatus)
-	if waitStatus.Signaled() {
-		return EXIT_SIGNAL_OFFSET + int(waitStatus.Signal()), nil
-	}
-	return waitStatus.ExitStatus(), nil
-}
-
-// killAllPids itterates over all of the container's processes
-// sending a SIGKILL to each process.
-func killAllPids(container *libcontainer.Config) error {
-	var (
-		procs   []*os.Process
-		freeze  = fs.Freeze
-		getPids = fs.GetPids
-	)
-	if systemd.UseSystemd() {
-		freeze = systemd.Freeze
-		getPids = systemd.GetPids
-	}
-	freeze(container.Cgroups, cgroups.Frozen)
-	pids, err := getPids(container.Cgroups)
-	if err != nil {
-		return err
-	}
-	for _, pid := range pids {
-		// TODO: log err without aborting if we are unable to find
-		// a single PID
-		if p, err := os.FindProcess(pid); err == nil {
-			procs = append(procs, p)
-			p.Kill()
-		}
-	}
-	freeze(container.Cgroups, cgroups.Thawed)
-	for _, p := range procs {
-		p.Wait()
-	}
-	return err
-}
-
-// DefaultCreateCommand will return an exec.Cmd with the Cloneflags set to the proper namespaces
-// defined on the container's configuration and use the current binary as the init with the
-// args provided
-//
-// console: the /dev/console to setup inside the container
-// init: the program executed inside the namespaces
-// root: the path to the container json file and information
-// pipe: sync pipe to synchronize the parent and child processes
-// args: the arguments to pass to the container to run as the user's program
-func DefaultCreateCommand(container *libcontainer.Config, console, dataPath, init string, pipe *os.File, args []string) *exec.Cmd {
-	// get our binary name from arg0 so we can always reexec ourself
-	env := []string{
-		"console=" + console,
-		"pipe=3",
-		"data_path=" + dataPath,
-	}
-
-	command := exec.Command(init, append([]string{"init", "--"}, args...)...)
-	// make sure the process is executed inside the context of the rootfs
-	command.Dir = container.RootFs
-	command.Env = append(os.Environ(), env...)
-
-	if command.SysProcAttr == nil {
-		command.SysProcAttr = &syscall.SysProcAttr{}
-	}
-	command.SysProcAttr.Cloneflags = uintptr(GetNamespaceFlags(container.Namespaces))
-
-	command.SysProcAttr.Pdeathsig = syscall.SIGKILL
-	command.ExtraFiles = []*os.File{pipe}
-
-	return command
-}
-
-// SetupCgroups applies the cgroup restrictions to the process running in the container based
-// on the container's configuration
-func SetupCgroups(container *libcontainer.Config, nspid int) (map[string]string, error) {
-	if container.Cgroups != nil {
-		c := container.Cgroups
-		if systemd.UseSystemd() {
-			return systemd.Apply(c, nspid)
-		}
-		return fs.Apply(c, nspid)
-	}
-	return map[string]string{}, nil
-}
-
-// InitializeNetworking creates the container's network stack outside of the namespace and moves
-// interfaces into the container's net namespaces if necessary
-func InitializeNetworking(container *libcontainer.Config, nspid int, networkState *network.NetworkState) error {
-	for _, config := range container.Networks {
-		strategy, err := network.GetStrategy(config.Type)
-		if err != nil {
-			return err
-		}
-		if err := strategy.Create((*network.Network)(config), nspid, networkState); err != nil {
-			return err
-		}
-	}
-	return nil
-}

vendor/src/github.com/docker/libcontainer/namespaces/execin.go

@@ -1,132 +0,0 @@
-// +build linux
-
-package namespaces
-
-import (
-	"encoding/json"
-	"fmt"
-	"io"
-	"os"
-	"os/exec"
-	"path/filepath"
-	"strconv"
-	"syscall"
-
-	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/apparmor"
-	"github.com/docker/libcontainer/cgroups"
-	"github.com/docker/libcontainer/label"
-	"github.com/docker/libcontainer/system"
-)
-
-// ExecIn reexec's the initPath with the argv 0 rewrite to "nsenter" so that it is able to run the
-// setns code in a single threaded environment joining the existing containers' namespaces.
-func ExecIn(container *libcontainer.Config, state *libcontainer.State, userArgs []string, initPath, action string,
-	stdin io.Reader, stdout, stderr io.Writer, console string, startCallback func(*exec.Cmd)) (int, error) {
-
-	args := []string{fmt.Sprintf("nsenter-%s", action), "--nspid", strconv.Itoa(state.InitPid)}
-
-	if console != "" {
-		args = append(args, "--console", console)
-	}
-
-	cmd := &exec.Cmd{
-		Path: initPath,
-		Args: append(args, append([]string{"--"}, userArgs...)...),
-	}
-
-	if filepath.Base(initPath) == initPath {
-		if lp, err := exec.LookPath(initPath); err == nil {
-			cmd.Path = lp
-		}
-	}
-
-	parent, child, err := newInitPipe()
-	if err != nil {
-		return -1, err
-	}
-	defer parent.Close()
-
-	// Note: these are only used in non-tty mode
-	// if there is a tty for the container it will be opened within the namespace and the
-	// fds will be duped to stdin, stdiout, and stderr
-	cmd.Stdin = stdin
-	cmd.Stdout = stdout
-	cmd.Stderr = stderr
-	cmd.ExtraFiles = []*os.File{child}
-
-	if err := cmd.Start(); err != nil {
-		child.Close()
-		return -1, err
-	}
-	child.Close()
-
-	terminate := func(terr error) (int, error) {
-		// TODO: log the errors for kill and wait
-		cmd.Process.Kill()
-		cmd.Wait()
-		return -1, terr
-	}
-
-	// Enter cgroups.
-	if err := EnterCgroups(state, cmd.Process.Pid); err != nil {
-		return terminate(err)
-	}
-
-	// finish cgroups' setup, unblock the child process.
-	if _, err := parent.WriteString("1"); err != nil {
-		return terminate(err)
-	}
-
-	if err := json.NewEncoder(parent).Encode(container); err != nil {
-		return terminate(err)
-	}
-
-	if startCallback != nil {
-		startCallback(cmd)
-	}
-
-	if err := cmd.Wait(); err != nil {
-		if _, ok := err.(*exec.ExitError); !ok {
-			return -1, err
-		}
-	}
-	return cmd.ProcessState.Sys().(syscall.WaitStatus).ExitStatus(), nil
-}
-
-// Finalize expects that the setns calls have been setup and that is has joined an
-// existing namespace
-func FinalizeSetns(container *libcontainer.Config, args []string) error {
-	// clear the current processes env and replace it with the environment defined on the container
-	if err := LoadContainerEnvironment(container); err != nil {
-		return err
-	}
-
-	if err := setupRlimits(container); err != nil {
-		return fmt.Errorf("setup rlimits %s", err)
-	}
-
-	if err := FinalizeNamespace(container); err != nil {
-		return err
-	}
-
-	if err := apparmor.ApplyProfile(container.AppArmorProfile); err != nil {
-		return fmt.Errorf("set apparmor profile %s: %s", container.AppArmorProfile, err)
-	}
-
-	if container.ProcessLabel != "" {
-		if err := label.SetProcessLabel(container.ProcessLabel); err != nil {
-			return err
-		}
-	}
-
-	if err := system.Execv(args[0], args[0:], os.Environ()); err != nil {
-		return err
-	}
-
-	panic("unreachable")
-}
-
-func EnterCgroups(state *libcontainer.State, pid int) error {
-	return cgroups.EnterPid(state.CgroupPaths, pid)
-}

vendor/src/github.com/docker/libcontainer/namespaces/init.go

@@ -1,331 +0,0 @@
-// +build linux
-
-package namespaces
-
-import (
-	"encoding/json"
-	"fmt"
-	"io/ioutil"
-	"os"
-	"strings"
-	"syscall"
-
-	"github.com/docker/libcontainer"
-	"github.com/docker/libcontainer/apparmor"
-	"github.com/docker/libcontainer/console"
-	"github.com/docker/libcontainer/label"
-	"github.com/docker/libcontainer/mount"
-	"github.com/docker/libcontainer/netlink"
-	"github.com/docker/libcontainer/network"
-	"github.com/docker/libcontainer/security/capabilities"
-	"github.com/docker/libcontainer/security/restrict"
-	"github.com/docker/libcontainer/system"
-	"github.com/docker/libcontainer/user"
-	"github.com/docker/libcontainer/utils"
-)
-
-// TODO(vishh): This is part of the libcontainer API and it does much more than just namespaces related work.
-// Move this to libcontainer package.
-// Init is the init process that first runs inside a new namespace to setup mounts, users, networking,
-// and other options required for the new container.
-// The caller of Init function has to ensure that the go runtime is locked to an OS thread
-// (using runtime.LockOSThread) else system calls like setns called within Init may not work as intended.
-func Init(container *libcontainer.Config, uncleanRootfs, consolePath string, pipe *os.File, args []string) (err error) {
-	defer func() {
-		// if we have an error during the initialization of the container's init then send it back to the
-		// parent process in the form of an initError.
-		if err != nil {
-			// ensure that any data sent from the parent is consumed so it doesn't
-			// receive ECONNRESET when the child writes to the pipe.
-			ioutil.ReadAll(pipe)
-			if err := json.NewEncoder(pipe).Encode(initError{
-				Message: err.Error(),
-			}); err != nil {
-				panic(err)
-			}
-		}
-		// ensure that this pipe is always closed
-		pipe.Close()
-	}()
-
-	rootfs, err := utils.ResolveRootfs(uncleanRootfs)
-	if err != nil {
-		return err
-	}
-
-	// clear the current processes env and replace it with the environment
-	// defined on the container
-	if err := LoadContainerEnvironment(container); err != nil {
-		return err
-	}
-
-	// We always read this as it is a way to sync with the parent as well
-	var networkState *network.NetworkState
-	if err := json.NewDecoder(pipe).Decode(&networkState); err != nil {
-		return err
-	}
-	// join any namespaces via a path to the namespace fd if provided
-	if err := joinExistingNamespaces(container.Namespaces); err != nil {
-		return err
-	}
-	if consolePath != "" {
-		if err := console.OpenAndDup(consolePath); err != nil {
-			return err
-		}
-	}
-	if _, err := syscall.Setsid(); err != nil {
-		return fmt.Errorf("setsid %s", err)
-	}
-	if consolePath != "" {
-		if err := system.Setctty(); err != nil {
-			return fmt.Errorf("setctty %s", err)
-		}
-	}
-
-	if err := setupNetwork(container, networkState); err != nil {
-		return fmt.Errorf("setup networking %s", err)
-	}
-	if err := setupRoute(container); err != nil {
-		return fmt.Errorf("setup route %s", err)
-	}
-
-	if err := setupRlimits(container); err != nil {
-		return fmt.Errorf("setup rlimits %s", err)
-	}
-
-	label.Init()
-
-	if err := mount.InitializeMountNamespace(rootfs,
-		consolePath,
-		container.RestrictSys,
-		(*mount.MountConfig)(container.MountConfig)); err != nil {
-		return fmt.Errorf("setup mount namespace %s", err)
-	}
-
-	if container.Hostname != "" {
-		if err := syscall.Sethostname([]byte(container.Hostname)); err != nil {
-			return fmt.Errorf("unable to sethostname %q: %s", container.Hostname, err)
-		}
-	}
-
-	if err := apparmor.ApplyProfile(container.AppArmorProfile); err != nil {
-		return fmt.Errorf("set apparmor profile %s: %s", container.AppArmorProfile, err)
-	}
-
-	if err := label.SetProcessLabel(container.ProcessLabel); err != nil {
-		return fmt.Errorf("set process label %s", err)
-	}
-
-	// TODO: (crosbymichael) make this configurable at the Config level
-	if container.RestrictSys {
-		if err := restrict.Restrict("proc/sys", "proc/sysrq-trigger", "proc/irq", "proc/bus"); err != nil {
-			return err
-		}
-	}
-
-	pdeathSignal, err := system.GetParentDeathSignal()
-	if err != nil {
-		return fmt.Errorf("get parent death signal %s", err)
-	}
-
-	if err := FinalizeNamespace(container); err != nil {
-		return fmt.Errorf("finalize namespace %s", err)
-	}
-
-	// FinalizeNamespace can change user/group which clears the parent death
-	// signal, so we restore it here.
-	if err := RestoreParentDeathSignal(pdeathSignal); err != nil {
-		return fmt.Errorf("restore parent death signal %s", err)
-	}
-
-	return system.Execv(args[0], args[0:], os.Environ())
-}
-
-// RestoreParentDeathSignal sets the parent death signal to old.
-func RestoreParentDeathSignal(old int) error {
-	if old == 0 {
-		return nil
-	}
-
-	current, err := system.GetParentDeathSignal()
-	if err != nil {
-		return fmt.Errorf("get parent death signal %s", err)
-	}
-
-	if old == current {
-		return nil
-	}
-
-	if err := system.ParentDeathSignal(uintptr(old)); err != nil {
-		return fmt.Errorf("set parent death signal %s", err)
-	}
-
-	// Signal self if parent is already dead. Does nothing if running in a new
-	// PID namespace, as Getppid will always return 0.
-	if syscall.Getppid() == 1 {
-		return syscall.Kill(syscall.Getpid(), syscall.SIGKILL)
-	}
-
-	return nil
-}
-
-// SetupUser changes the groups, gid, and uid for the user inside the container
-func SetupUser(container *libcontainer.Config) error {
-	// Set up defaults.
-	defaultExecUser := user.ExecUser{
-		Uid:  syscall.Getuid(),
-		Gid:  syscall.Getgid(),
-		Home: "/",
-	}
-
-	passwdPath, err := user.GetPasswdPath()
-	if err != nil {
-		return err
-	}
-
-	groupPath, err := user.GetGroupPath()
-	if err != nil {
-		return err
-	}
-
-	execUser, err := user.GetExecUserPath(container.User, &defaultExecUser, passwdPath, groupPath)
-	if err != nil {
-		return fmt.Errorf("get supplementary groups %s", err)
-	}
-
-	suppGroups := append(execUser.Sgids, container.AdditionalGroups...)
-
-	if err := syscall.Setgroups(suppGroups); err != nil {
-		return fmt.Errorf("setgroups %s", err)
-	}
-
-	if err := system.Setgid(execUser.Gid); err != nil {
-		return fmt.Errorf("setgid %s", err)
-	}
-
-	if err := system.Setuid(execUser.Uid); err != nil {
-		return fmt.Errorf("setuid %s", err)
-	}
-
-	// if we didn't get HOME already, set it based on the user's HOME
-	if envHome := os.Getenv("HOME"); envHome == "" {
-		if err := os.Setenv("HOME", execUser.Home); err != nil {
-			return fmt.Errorf("set HOME %s", err)
-		}
-	}
-
-	return nil
-}
-
-// setupVethNetwork uses the Network config if it is not nil to initialize
-// the new veth interface inside the container for use by changing the name to eth0
-// setting the MTU and IP address along with the default gateway
-func setupNetwork(container *libcontainer.Config, networkState *network.NetworkState) error {
-	for _, config := range container.Networks {
-		strategy, err := network.GetStrategy(config.Type)
-		if err != nil {
-			return err
-		}
-
-		err1 := strategy.Initialize((*network.Network)(config), networkState)
-		if err1 != nil {
-			return err1
-		}
-	}
-	return nil
-}
-
-func setupRoute(container *libcontainer.Config) error {
-	for _, config := range container.Routes {
-		if err := netlink.AddRoute(config.Destination, config.Source, config.Gateway, config.InterfaceName); err != nil {
-			return err
-		}
-	}
-	return nil
-}
-
-func setupRlimits(container *libcontainer.Config) error {
-	for _, rlimit := range container.Rlimits {
-		l := &syscall.Rlimit{Max: rlimit.Hard, Cur: rlimit.Soft}
-		if err := syscall.Setrlimit(rlimit.Type, l); err != nil {
-			return fmt.Errorf("error setting rlimit type %v: %v", rlimit.Type, err)
-		}
-	}
-	return nil
-}
-
-// FinalizeNamespace drops the caps, sets the correct user
-// and working dir, and closes any leaky file descriptors
-// before execing the command inside the namespace
-func FinalizeNamespace(container *libcontainer.Config) error {
-	// Ensure that all non-standard fds we may have accidentally
-	// inherited are marked close-on-exec so they stay out of the
-	// container
-	if err := utils.CloseExecFrom(3); err != nil {
-		return fmt.Errorf("close open file descriptors %s", err)
-	}
-
-	// drop capabilities in bounding set before changing user
-	if err := capabilities.DropBoundingSet(container.Capabilities); err != nil {
-		return fmt.Errorf("drop bounding set %s", err)
-	}
-
-	// preserve existing capabilities while we change users
-	if err := system.SetKeepCaps(); err != nil {
-		return fmt.Errorf("set keep caps %s", err)
-	}
-
-	if err := SetupUser(container); err != nil {
-		return fmt.Errorf("setup user %s", err)
-	}
-
-	if err := system.ClearKeepCaps(); err != nil {
-		return fmt.Errorf("clear keep caps %s", err)
-	}
-
-	// drop all other capabilities
-	if err := capabilities.DropCapabilities(container.Capabilities); err != nil {
-		return fmt.Errorf("drop capabilities %s", err)
-	}
-
-	if container.WorkingDir != "" {
-		if err := syscall.Chdir(container.WorkingDir); err != nil {
-			return fmt.Errorf("chdir to %s %s", container.WorkingDir, err)
-		}
-	}
-
-	return nil
-}
-
-func LoadContainerEnvironment(container *libcontainer.Config) error {
-	os.Clearenv()
-	for _, pair := range container.Env {
-		p := strings.SplitN(pair, "=", 2)
-		if len(p) < 2 {
-			return fmt.Errorf("invalid environment '%v'", pair)
-		}
-		if err := os.Setenv(p[0], p[1]); err != nil {
-			return err
-		}
-	}
-	return nil
-}
-
-// joinExistingNamespaces gets all the namespace paths specified for the container and
-// does a setns on the namespace fd so that the current process joins the namespace.
-func joinExistingNamespaces(namespaces []libcontainer.Namespace) error {
-	for _, ns := range namespaces {
-		if ns.Path != "" {
-			f, err := os.OpenFile(ns.Path, os.O_RDONLY, 0)
-			if err != nil {
-				return err
-			}
-			err = system.Setns(f.Fd(), uintptr(namespaceInfo[ns.Type]))
-			f.Close()
-			if err != nil {
-				return err
-			}
-		}
-	}
-	return nil
-}

vendor/src/github.com/docker/libcontainer/namespaces/nsenter/nsenter.c

@@ -1,245 +0,0 @@
-// +build cgo
-//
-// formated with indent -linux nsenter.c
-
-#include <errno.h>
-#include <fcntl.h>
-#include <linux/limits.h>
-#include <linux/sched.h>
-#include <signal.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <sys/prctl.h>
-#include <sys/types.h>
-#include <unistd.h>
-#include <getopt.h>
-
-#define pr_perror(fmt, ...) fprintf(stderr, "nsenter: " fmt ": %m\n", ##__VA_ARGS__)
-
-static const kBufSize = 256;
-static const char *kNsEnter = "nsenter";
-
-void get_args(int *argc, char ***argv)
-{
-	// Read argv
-	int fd = open("/proc/self/cmdline", O_RDONLY);
-	if (fd < 0) {
-		pr_perror("Unable to open /proc/self/cmdline");
-		exit(1);
-	}
-	// Read the whole commandline.
-	ssize_t contents_size = 0;
-	ssize_t contents_offset = 0;
-	char *contents = NULL;
-	ssize_t bytes_read = 0;
-	do {
-		contents_size += kBufSize;
-		contents = (char *)realloc(contents, contents_size);
-		bytes_read =
-		    read(fd, contents + contents_offset,
-			 contents_size - contents_offset);
-		if (bytes_read < 0) {
-			pr_perror("Unable to read from /proc/self/cmdline");
-			exit(1);
-		}
-		contents_offset += bytes_read;
-	}
-	while (bytes_read > 0);
-	close(fd);
-
-	// Parse the commandline into an argv. /proc/self/cmdline has \0 delimited args.
-	ssize_t i;
-	*argc = 0;
-	for (i = 0; i < contents_offset; i++) {
-		if (contents[i] == '\0') {
-			(*argc)++;
-		}
-	}
-	*argv = (char **)malloc(sizeof(char *) * ((*argc) + 1));
-	int idx;
-	for (idx = 0; idx < (*argc); idx++) {
-		(*argv)[idx] = contents;
-		contents += strlen(contents) + 1;
-	}
-	(*argv)[*argc] = NULL;
-}
-
-// Use raw setns syscall for versions of glibc that don't include it (namely glibc-2.12)
-#if __GLIBC__ == 2 && __GLIBC_MINOR__ < 14
-#define _GNU_SOURCE
-#include <sched.h>
-#include "syscall.h"
-#ifdef SYS_setns
-int setns(int fd, int nstype)
-{
-	return syscall(SYS_setns, fd, nstype);
-}
-#endif
-#endif
-
-void print_usage()
-{
-	fprintf(stderr,
-		"nsenter --nspid <pid> --console <console> -- cmd1 arg1 arg2...\n");
-}
-
-void nsenter()
-{
-	int argc, c;
-	char **argv;
-	get_args(&argc, &argv);
-
-	// check argv 0 to ensure that we are supposed to setns
-	// we use strncmp to test for a value of "nsenter" but also allows alternate implmentations
-	// after the setns code path to continue to use the argv 0 to determine actions to be run
-	// resulting in the ability to specify "nsenter-mknod", "nsenter-exec", etc...
-	if (strncmp(argv[0], kNsEnter, strlen(kNsEnter)) != 0) {
-		return;
-	}
-#ifdef PR_SET_CHILD_SUBREAPER
-	if (prctl(PR_SET_CHILD_SUBREAPER, 1, 0, 0, 0) == -1) {
-		pr_perror("Failed to set child subreaper");
-		exit(1);
-	}
-#endif
-
-	static const struct option longopts[] = {
-		{"nspid", required_argument, NULL, 'n'},
-		{"console", required_argument, NULL, 't'},
-		{NULL, 0, NULL, 0}
-	};
-
-	pid_t init_pid = -1;
-	char *init_pid_str = NULL;
-	char *console = NULL;
-	while ((c = getopt_long_only(argc, argv, "n:c:", longopts, NULL)) != -1) {
-		switch (c) {
-		case 'n':
-			init_pid_str = optarg;
-			break;
-		case 't':
-			console = optarg;
-			break;
-		}
-	}
-
-	if (init_pid_str == NULL) {
-		print_usage();
-		exit(1);
-	}
-
-	init_pid = strtol(init_pid_str, NULL, 10);
-	if ((init_pid == 0 && errno == EINVAL) || errno == ERANGE) {
-		pr_perror("Failed to parse PID from \"%s\" with output \"%d\"",
-			  init_pid_str, init_pid);
-		print_usage();
-		exit(1);
-	}
-
-	argc -= 3;
-	argv += 3;
-
-	if (setsid() == -1) {
-		pr_perror("setsid failed");
-		exit(1);
-	}
-	// before we setns we need to dup the console
-	int consolefd = -1;
-	if (console != NULL) {
-		consolefd = open(console, O_RDWR);
-		if (consolefd < 0) {
-			pr_perror("Failed to open console %s", console);
-			exit(1);
-		}
-	}
-	// blocking until the parent placed the process inside correct cgroups.
-	unsigned char s;
-	if (read(3, &s, 1) != 1 || s != '1') {
-		pr_perror("failed to receive synchronization data from parent");
-		exit(1);
-	}
-	// Setns on all supported namespaces.
-	char ns_dir[PATH_MAX];
-	memset(ns_dir, 0, PATH_MAX);
-	snprintf(ns_dir, PATH_MAX - 1, "/proc/%d/ns/", init_pid);
-
-	int ns_dir_fd;
-	ns_dir_fd = open(ns_dir, O_RDONLY | O_DIRECTORY);
-	if (ns_dir_fd < 0) {
-		pr_perror("Unable to open %s", ns_dir);
-		exit(1);
-	}
-
-	char *namespaces[] = { "ipc", "uts", "net", "pid", "mnt" };
-	const int num = sizeof(namespaces) / sizeof(char *);
-	int i;
-	for (i = 0; i < num; i++) {
-		// A zombie process has links on namespaces, but they can't be opened
-		struct stat st;
-		if (fstatat(ns_dir_fd, namespaces[i], &st, AT_SYMLINK_NOFOLLOW)
-		    == -1) {
-			if (errno == ENOENT)
-				continue;
-			pr_perror("Failed to stat ns file %s for ns %s",
-				  ns_dir, namespaces[i]);
-			exit(1);
-		}
-
-		int fd = openat(ns_dir_fd, namespaces[i], O_RDONLY);
-		if (fd == -1) {
-			pr_perror("Failed to open ns file %s for ns %s",
-				  ns_dir, namespaces[i]);
-			exit(1);
-		}
-		// Set the namespace.
-		if (setns(fd, 0) == -1) {
-			pr_perror("Failed to setns for %s", namespaces[i]);
-			exit(1);
-		}
-		close(fd);
-	}
-	close(ns_dir_fd);
-
-	// We must fork to actually enter the PID namespace.
-	int child = fork();
-	if (child == -1) {
-		pr_perror("Unable to fork a process");
-		exit(1);
-	}
-	if (child == 0) {
-		if (consolefd != -1) {
-			if (dup2(consolefd, STDIN_FILENO) != 0) {
-				pr_perror("Failed to dup 0");
-				exit(1);
-			}
-			if (dup2(consolefd, STDOUT_FILENO) != STDOUT_FILENO) {
-				pr_perror("Failed to dup 1");
-				exit(1);
-			}
-			if (dup2(consolefd, STDERR_FILENO) != STDERR_FILENO) {
-				pr_perror("Failed to dup 2\n");
-				exit(1);
-			}
-		}
-		// Finish executing, let the Go runtime take over.
-		return;
-	} else {
-		// Parent, wait for the child.
-		int status = 0;
-		if (waitpid(child, &status, 0) == -1) {
-			pr_perror("nsenter: Failed to waitpid with error");
-			exit(1);
-		}
-		// Forward the child's exit code or re-send its death signal.
-		if (WIFEXITED(status)) {
-			exit(WEXITSTATUS(status));
-		} else if (WIFSIGNALED(status)) {
-			kill(getpid(), WTERMSIG(status));
-		}
-
-		exit(1);
-	}
-
-	return;
-}

vendor/src/github.com/docker/libcontainer/namespaces/nsenter/nsenter.go

@@ -1,10 +0,0 @@
-// +build linux
-
-package nsenter
-
-/*
-__attribute__((constructor)) init() {
-	nsenter();
-}
-*/
-import "C"

vendor/src/github.com/docker/libcontainer/namespaces/utils.go

@@ -1,45 +0,0 @@
-// +build linux
-
-package namespaces
-
-import (
-	"os"
-	"syscall"
-
-	"github.com/docker/libcontainer"
-)
-
-type initError struct {
-	Message string `json:"message,omitempty"`
-}
-
-func (i initError) Error() string {
-	return i.Message
-}
-
-var namespaceInfo = map[libcontainer.NamespaceType]int{
-	libcontainer.NEWNET:  syscall.CLONE_NEWNET,
-	libcontainer.NEWNS:   syscall.CLONE_NEWNS,
-	libcontainer.NEWUSER: syscall.CLONE_NEWUSER,
-	libcontainer.NEWIPC:  syscall.CLONE_NEWIPC,
-	libcontainer.NEWUTS:  syscall.CLONE_NEWUTS,
-	libcontainer.NEWPID:  syscall.CLONE_NEWPID,
-}
-
-// New returns a newly initialized Pipe for communication between processes
-func newInitPipe() (parent *os.File, child *os.File, err error) {
-	fds, err := syscall.Socketpair(syscall.AF_LOCAL, syscall.SOCK_STREAM|syscall.SOCK_CLOEXEC, 0)
-	if err != nil {
-		return nil, nil, err
-	}
-	return os.NewFile(uintptr(fds[1]), "parent"), os.NewFile(uintptr(fds[0]), "child"), nil
-}
-
-// GetNamespaceFlags parses the container's Namespaces options to set the correct
-// flags on clone, unshare, and setns
-func GetNamespaceFlags(namespaces libcontainer.Namespaces) (flag int) {
-	for _, v := range namespaces {
-		flag |= namespaceInfo[v.Type]
-	}
-	return flag
-}